Split and Splitless Injection for Quantitative Gas Chromatography: Concepts, Processes, Practical Guidelines, Sources of Error (4th, Completely Revised Edition)

~WI LEY-VCH

Konrad Grab

Split and Splitless Injection for Quantitative Gas Chromatography Concepts, Processes, . Practical Guidelines, Sources of Error Fourth, completely revised edition

Konrad Grob

Split and Splitless Injection for Quantitative Gas Chromatography

~WILEY-VCH

KonradGrob

Split and Splitless Injection for Quantitative Gas Chromatography

Concepts, Processes, Practical Guidelines, Sources of Error Fourth, completely revised edition

Dr. Konrad Grob Kantonales Labor Fehrenstr. 15 CH-8032 ZUrich Switzerland

This book was carefully produced. Nevertheless, author and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustra­ tions, procedural details or other items may inadvertently be inaccurate.

4th, completely revised edition 2001 Including CD-ROM The Transparent Injector by Maurus Biedermann l st reprint 2003

Die Deutsche Bibliothek - CIP-Cataloguing-in-Publication-data A catalogue record for this publication is available from Die Deutsche Bibliothek ISBN 3-527-29879-7 © WILEY-VCH Verlag GmbH, 0-69469 Weinheim (Federal Republic of Germany), 2001

Printed on acid-free paper All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form ~ by photoprinting, microfilm, or any other means - nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such are not to be considered unprotected by law.

Printed in the Federal Republic of Germany

iju3.08'1& c;,~~3

Preface

v

Preface In the scientific literature and in commercial catalogs, methods are almost invariably described as "easy"; there seem to be no limitations and problems. If original papers reflect the eupho­ ria of the inventors, this is understandable. That catalogs of instrument manufacturers do not mention weaknesses of a product might be attributed to the rules of business. Even review papers, however, tend to neglect problems, maybe because authors do not want to risk good relationships or have insufficient experimental support for criticism. As a result of this, there is a frightening discrepancy between the rose-colored descriptions and the reality in laboratories. Published work, for instance, reports relative standard devia­ tions that are far lower than commonly obtained in reality - errors by a factor of two are rather frequent, and probably more frequent than recognized. The frustration of the analyst is under­ standable. His position in relation to his boss, who might have never gone through the reality of chromatography, is weak, because he seems to be an especially incapable analyst. For new techniques, a few chromatograms are usually provided as a proof that they work. Inventors cannot be blamed for not having tested them with all possible samples and under all conceivable conditions - an impossible task. Techniques routinely used by tens of thou­ sands of users should, however, be investigated rather comprehensively to enable under­ standing of the mechanisms involved and systematic discovery of the critical samples and conditions. This means, primarily, investigation of possible imperfections - not out of ma­ levolence towards the inventor or instrument manufacturer, but to prevent failures during applications involving particularly unfortunate conditions. The user should know about the problems so they can be foreseen or, if they occur nevertheless, to avoid his spending days in search of the source, finally to discover he was looking in the wrong place. Instruments are usually evaluated by means of a few injections of some alkanes in a simple solvent. Such quick tests resemble Russian roulette: whether an instrument is shot or escapes alive is primarily a matter of luck. Real evaluation is far more demanding. Even today instru­ ments differ significantly in their essential parts, which is why the critical details of injector design are a subject of this book. The book also concentrates on weaknesses of the techniques because it is assumed that prob­ lems are the reason why the analyst takes a look at it. Overemphasis of problems bears a danger, however, that a reader starts wondering why reasonable results were ever obtained or why capillary GC has not been abandoned altogether. He must be reminded that most problems are important only for certain types of sample and conditions. There is no doubt that capillary GC in general and injection in particular are demanding tech­ niques. They are full of pitfalls, but also rich in possibilities for a creative analyst - and cer­ tainly never boring. The better an analyst masters it and the more he knows, the more he is likely to be fascinated and the better he realizes how much more could be made of it. Hopefully many will pick up problems and incomplete concepts, work on the subject, and contribute to the further development of capillary GC. Around 200,000 people use capillary

VI

Preface

GC and could, therefore, profit from such contributions. It is my impression that GC injection techniques are still far from being optimized to the point which could be reached. Thousands of analysts go through the same trouble and lose weeks of work because known problems have not been solved. Apart from the frustration, this results in unnecessary costs. The basic problem seems to be that nobody is willing to carry the burden of perfecting these tech­ niques. We are all paid for specific work (my job is in governmental food control), rather than to help others. Because offering an improved split/splitless injector does not seem to be a way of improving sales of instruments, instrument manufacturers hesitate to invest in this direc­ tion. This book was started as a revision of "Split and Splitless Injection in Capillary GC", pub­ lished by Hirthlq (Heidelberg) in 1993, which in turn was an update of "Classical Split and Splitless Injection" from 1986. The new material, primarily on sample evaporation, necessi­ tated, however, a new structure and finally a large part of the book was rewritten. The CD­ ROM, produced by Maurus Biedermann, was added because the videos on the processes occurring in devices imitating injectors cannot be replaced by a description. Programmed temperature vaporizing (PTV) injection, on the other hand, has grown into a field requiring more space than is available in this book. I wish to thank Ian Davies, Cambridge, UK, for converting Swinglish (Swiss English) into a more proper language, and Jonas Grob, one of my sons, for turning more than one million letters and many figures into attractive pages. Fehraltorf, October 2000

Koni Grob

Survey of Injection Techniques

VII

Survey of Injection Techniques Is splitless injection a procedure during which you never touch the split outlet valve? If there is a danger of such confusion, please have a look at the following list of short definitions. Injection into GC capillary columns can be confusing, because there are so many different techniques. And if you ask why this is so, the answer is that each of these techniques is better than all others in some respects and has features some analysts do not want to do without. The following table provides a survey of the main injection techniques. It does not mention numerous others which have never become popular or have lost their importance, such as injection through a loop, capsule injection, and moving needle or other solid injection tech­ niques.

Injection into Capillary Columns Classical vaporizing injection Split/

I

\irect Splitless

Programmed temperature vaporizing (PTV) injection Split / / \ " Direct Splitless Solvent-split

On-column injection

I ""- Precolumn Classical /' (small volume) solvent splittin~ Retention gap technique

Short definitions might be as follows:

Classical vaporizing injection. Sample evaporation in a permanently hot vaporizing chamber

before transfer into the column.

Split injection. Only a small part of the vapor enters the column, the rest being vented. The

technique of choice for rather concentrated samples, as well as for gas and headspace analy­

sis.

Splitless injection. Nearly all of the sample vapor is transferred from the injector into the

column; the technique is performed with a split injector. Trace analysis of contaminated sam­

ples.

Direct injection. All the vapor is transferred into the column; performed with an injector

without a split outlet. Trace analysis, usually involving instruments converted from packed

column GC.

Programmed temperature vaporizing (PTV) injection. Injection into a cool chamber which

is subsequently heated to vaporize the sample. Newer technique to replace classical vaporiz­

ing injection.

Solvent splitting. Most of the solvent vapor is vented; the solute material is transferred into

the column in splitless mode. Usually used for large volume injection in trace analysis.

:VIII

Survey of Injection Techniques

Ion-column injection. Injection of the sample liquid into the column inlet or an oven-thor­ Imostatted capillary precolumn. Technique providing the best results, but not suitable for highly contaminated samples. Retention gap technique. Use of an uncoated precolumn to overcome band broadening resulting from sample liquid flooding the column inlet. Most important for large volume on­ column injection and on-line coupled LC-GC. Precolumn solvent splitting. Injection into a precolumn connected to a vapor exit through which most of the solvent vapor is released. Used for large volume injection or on-line trans­ fer.

Contents

IX

Contents A Syringe Injection into Hot Vaporizing Chambers 1. Introduction ....•.••••...••.••.•.•.....•..•.••......•...•••.•.••.•...•...••••.••••.•...••.•.••.....••.•.....•.•••. 1

1. 1. Syringe Injection .•........•..........•.•... 1

1.2. Sample Evaporation inside the Needle 2

1.2.1. Inaccurate Sample Volume 2

1.2.2. Discrimination against High Boilers 3

1.2.3. Poor Reproducibility 4

1.2.4. Degradation of Labile Solutes 4

1.3. Conclusions ,.................................•...................................... 4

1.3.1. Fast Autosampler? 4

1.3.2. Suppressing Evaporation inside the Needle 5

1.3.3. Thermospray 5

2. Syringes •.•.•••........••..........••.•••............•......•••.•.•....•.•.•••.•.......•.•.•..•••..•.••.......•••••• 6

"T

2.1. P/unger-in-Barrel Syringes ....................................................................•.... 6

2.1.1. Plungers 6

2.1.2. Plunger Guides 7

2.2. P/unger-in-Needle Syringes .................•...........•....•.............•.......•.............. 9

2.3. Syringe Needles .......................•.•.........................•..................................... 9

2.3.1. Dimensions 9

2.3.2. Needle Tips 10

2.3.3. Fixed versus Removable Needles 11

2.4. Cleaning of Syringes .......•............•.........................................•................. 11

2.4.1. Basic Rules 11

2.4.2. Cleaning Procedures 12

2.4.3: Plugged Needles 14

2.4.4. Blocked Plungers 14

I

3. Evaporation Inside the Needle •.••••...•..•••.•.....................••••..........••...•....•.••••.... 15

3. 1. The Three-Step Model ...................................•......................................... 15

3.2. Models of Evaporation inside the Needle 17

3.2.1. Distillation from the Needle 17

3.2.2. Gas Chromatography in the Needle 17

3.2.3. Ejection from the Needle 18

3.3. Conclusions Regarding Optimized Injection ...............•........................... 19

4. How Much is Really Injected? .......••....•......••••.•.••..•••.••.••....•.•.•..•••..•.•.•.••.•...•.•. 20

4.1. Interpretations of "Sample Volume" ..........................•........................... 20

4.2. Communicating "Sample Volumes" 21

4.3. Effects on Quantitative Analy"is 21

X

Contents

5. Syringe Needle Handling Minimizing Discrimination

22

5.1. Definitions of Techniques 22

5.2. Experimental Determination of Losses in the Needle •............................ 24

5.2.1. Method with Two Instruments 24

25

5.2.2. Experiment with a Single Instrument 5.2.3. Test During Routine Analysis 26

5.3. Comparison of Needle Handling Techniques 26

5.3.1. Filled Needle Injection 27

28

5.3.2. Slow Injection 5.3.3. Cool Needle Injection 28

29

5.3.4. Hot Needle Injection 5.3.5. Solvent Flush Injection 32

37

5.3.6. Air Plug Injection 5.3.7. Sandwich Injection 37

5.4. Heating the Needle after Injection? ...•.....•..•............•.............................. 37

5.5. Effect of Injecting Air..........•..............•..••.............................................•.. 38

5.5.1. Concerns Regarding the Column 39

5.5.2. Detectors 39

5.5.3. Oxidized Sample 39

6. Dependence of Discrimination on Sample Volume

41

6. 1. Experimental Results ............................................................•.................. 41

6.2. Discussion of Mechanism ..................................•.........................•........... 42.

6.3. Conclusions .................................................................................•.•......... 43

7. Solvent and Solutes ......•...••...•...•••.•......••••...•..•........••...............•.•.••...•..•.•......• 43

7. 1. 7.2. 7.3. 7.4.

Volatility of the Solvent ...............................•.......................................... 44

Type of Solute ...........................•................•.................................•.......... 44

Adsorption in the Syringe Needle 45

"Memory Effects" Arising from the Syringe 46

8. Injector Temperature ...•.••.•.......................................•..........•.....••••••••...•••.•...... 47

8. 1. Imposed Temperature ................................................................•............. 47

8.2. Temperature Gradient Towards the Septum ............................•..............• 48

8.2.1. Critical Rear of Needle 49

. 8.2.2. Actual Temperature Profiles 50

8.2.3. Effect on Discrimination 51

53

8.2.4. Quantitative Results Differing from One Injector to Another 8.2.5. Conclusions 54

8.3. Thermostability of Septa ............................................•.•..........•................ 55

8.3.1. Upper Temperature Limit 55

56

8.3.2. Some lips 9. Plunger-in-Needle Syringes .......•............••••...•.•..•.....••.•.•......•....••.••.•.•...•..•...•.• 57

9. 1. Accuracy of Sample Volume .........•..........................•...•,..•...•.................. 57

9.2. Premature Expulsion ....•...........................................•....•.•........•.............. 57

10. Possibilities of Avoiding Evaporation in the Needle ••.•..•••.•...••.•••.......•......... 59

10. 1. High Boiling Sample Matrix .. ....•.•....•...•.•.........•.....•........•.•........•.........., 59

10.1.1. Injector Temperature versus Solvent Boiling Point 59

10.1.2. Practical Aspects : 61

r Contents 10.2. 10.3. 10.4. 10.5.

XI

Cooled Septum ... ..•.•.....•.•.••.....•...•.•.•.....•.•.•.•............•.....•.•...•.•...•.......... 62

Cooled Needle Technique .......•.•.•.......•.•........•.••...•........•....................•.. 62

Fast Injection by Autosampler ...........•.•.......••..•..............................•..... 62

Evaporation in the Injector .•....•.......•.•.•.•.......................:•.•.•.•....•.•..•••.... 63

11. Summarizing Guidelines •...•.•••••.......•••.•..••••••••.••••••••••••••••...•.•.•.•...••.•.•.....••••. 64

References A •.••.......•...........•••..•••••••••........••••••.•••••••••••••••••••••••....•.•.••...•.•••.••....•••• 66

B Sample Evaporation in the Injector 1. Introduction •..••••.•.........•...•..•.••.....•••••...•...•.••.•••........•.............•.......•.......••••...• 69

1. 1. Problems Caused by Incomplete Evaporation 2. Solvent Evaporation - Heat Transfer

70

71

2.1. Available Evaporation Time ••••...••••••.•.•.••••••••••••••••••••••••••••••••.•.•••.••.•.••••••• 71

72

2.1.1. Band of Liquid 2.1.2. Nebulized Sample : 72

2.1.3. Deposition on Surfaces 73

2.2. Amount of Heat Required ....................................................•........•...•.•.... 73

2.3. Sources of Heat ..............................................•.....................•.•....•.•.•.....•. 73

2.3.1. Carrier Gas 74

2.3.2. Packed Injector Liners 74

75

2.3.3. Heat from Liner Wall 2.4. Time Required for Heat Transfer 75

2.4.1. Transfer Within the Liner Wall 75

76

2.4.2. Transfer Through the Gas Phase 2.4.3. Residence Time Required for Evaporation 77

2.5. Conclusions ............•....••.•.•..•.•......•....•.••.....•.•.....•..•.....•...•......•..........•..... 78

2.6. Experimental Results ...•••..•.............................•...............•...............•.•...•.. 79

2.6.1. Calculated and Measured Temperature Drop 79

80

2.6.2. Measurement of Evaporation Time via Split Flow Rate 3. Solvent Evaporation - Visual Observation

81

3. 1. Experimental ..•.•.......•....•.......•...•.........................................•.•................. 81

3.2. Liquid Exiting the Syringe Needle 83

3.2.1. Injection through a Cool Needle 83

84

3.2.2. Injection through a Hot Needle 3.3. Three Scenarios of Evaporation in an Empty Vaporizing Chamber 87

3.3.1. Scenario 1- Nebulization 87

88

3.3.2. Scenario 2 - Band of Liquid 90

3.3.3. Scenario 3 - Liquid Splashing on the Liner Wall 3.4. First Conclusions ......••.•............•.........•.....................•.....•................•.•...•. 92

3.4.1. Fate of Sample Liquid "Shot" to the Bottom of the Liner 93

3.5. Stopping the Sample Liquid .•.•..•......•....•.•..•....•.••.•....•.....•.•..•.•....•........... 96

3.5.1. Liner with Baffles 96

96

3.5.2. Cup or "Jennings" Liner 3.5.3. Glass Bead Liner 98

3.5.4. Cycloliner 98

: 99

3.5.5. Laminar Liner

XII

Contents

3.5.6. Metal Liner 3.5.7. Summary - Stopping Liquid with Obstacles 3.5.8. Wool 3.5.9. Glass Frits 3.5.10. Carbofrit 3.5.11. Column Packing Material 3.6. Other Criteria for Evaluating Obstacles ............................•.•.•...........•... 3.1. Duration of Solvent Evaporation

99 100 101 102 103 103 103

105

4. Solute Evaporation .•.•....••......••.•...••.......••..••.•.....•.••..••.••....••.•....•...••••............ 106

4. ,. Evaporation in the Gas Phase

101

4.1.1. Some Key Terms 107 109 4.1.2. Dilution with Carrier Gas in an Empty Liner 4.1.3. Solute Concentrations in the Injector 110 4.1.4. Glass Wool Improving Evaporation? 112 112 4.1.5. Evaporation from Contaminants 4.1.6. Prevention of Column Contamination 114 4.2. Evaporation from Surfaces ........•...•...........................................•.•......... 116

4.2.1. The Iodine Experiment 117 118 4.2.2. Dilution in Carrier Gas 119 4.2.3. GC Retentive Power of a Surface 4.2.4. Experimental Data 120 4.3. Conclusions on Injector Temperature 121 4.3.1. Thermospray Injection 121 . 4.3.2. Deposition on a Surface 121

5. Sample Degradation in the Injector

122

5.1. Degradation in the Injector or in the Column? 5.1.1. Methods for Distinction 5.2. Mechanisms of Solute Degradation ..•................................................... 5.3. Countermeasures against Solute Degradation 5.4. Examples ........................................•...................................................... 5.4.1. Divinylcyclobutane 5.4.2. Carbamate Insecticides 5.4.3. Oxygenated Dibenzothiophenes 5.4.4. Mustard Oils 5.4.5. Chlorohydrin in a Drug Substance 5.4.6. Drugs Requiring an Empty Liner 5.4.7. Empty Liner for Methyl Esters of Hydroxy Fatty Acids : 5.4.8. Brominated Alkanes 6. Retention and Adsorption in the Vaporizing Chamber

122 123

124

125 126

126 126 127 127 127 127 128 128

129

6. 1. Adsorption in the Injector •.....................•....................•...............•......'" 129

6.1.1. Split Injection 6.1.2. Splitless Injection 6.1.3. Column or Injector? 6.1.4. Experimentally Observed Adsorption 6.1.5. Variability of Adsorption 6.2. Retention in the Injector

129 129 129 131 131

'"

.•......

.•....

132

Contents

7. Deactivation of Liners and Packing Materials

7.1. 7.2. 7.3. 7.4.

Deactivation of the Liners? Deactivation of Commercial Wool Application-Related Testing for Inertne••............................................. More Comprehen.ive Te.ting Procedure

7.4.1. Design of the Test 7.4.2. Goals of the Test 7.4.3. Results

7.5. Silylation of Liners 7.5.1. Background 7.5.2. Wettability? 7.5.3. Method Recommended for Silylation of Liners 7.6. Silylation of G/a•• and Quartz Wool

7.7. Packing. Coated with Stationary Pha.e 7.8. Deactivation by Sample Material 7.8.1. 7.8.2. 7.8.3. 7.8.4.

Unstable Deactivation Heating Injector Overnight and at Weekends? Carrier Gas Overnight? Tests with Sample

8. Cleaning of Injector Liners

8. 1. Wa.hing with Strong Acid. or Ba.es 8.2. Burning the Contaminant.8.3. Gentle Cleaning

References B

XIII

133

133

133

134

134

135

135

136

138

138

139

139

140

140

141

141

142

142

142

143

143

144 144

145

C Split Injection 1. Introduction 1.1. Principle. of Split Injection 1.1.1. Basic Injector Design 1.2. Purpo.e. of Sample Splitting 1.2.1. Injection of Concentrated Samples 1.2.2. Splitting to Generate Sharp Initial Bands 1.3. The Two Principle. of Ga. Supply 1.4. Hi.toric Background of Split Injection

2. The Split Ratio

2. 1. Definition 2.2. Adju.tment!Determination of the Split Ratio

149

149

149

150

150

151

151

152

155

155

2.2.1. Determination ofthe Column Flow Rate 2.2.2. Adjustment ofthe Split Flow Rate

156

157

161

3. Sample Concentrations Suitable for Split Injection

163

3. 1. Split Ratio. Commonly Applied 163

3.2. Range of Suitable Concentration•......................................................... 163

4. Initial Band Widths 4.1. Band Width. in Space and Time

164

164

XIV

Contents

4.2. Factors Dete,mining Initial Band Widths 4.3. Expe,imental Observation of Initial Band Shapes ...•..............•.....•...•.•.. 4.3.1. Description of the Experiment 4.3.2. Subjects to Study 4.3.3. Some Results 4.4. Effect on the Final Peak Width....... ..••.............•.......•....•......•.•.....•.... 4.4.1. Isothermal Runs 4.4.2. Chromatography Involving Temperature Increase

164

165

166

166

168

169

169

171

5. Split Injection for Fast Analysis .•.••••.•.•.•••••.••••.............•.•..•••.••..••.....•.•••••...•.• 171

5. 1. 5.2. 5.3. 5.4.

P,e,equisites fo, Fast Analysis Maximum Tole,able Initial Band Widths Limits to the Sha,pness of Initial Bands Examples of Fast Analyses ................•.....•.......•........................•............

172

173

173

174

6. Analysis Requiring Maximum Sensitivity .•.•.•.......•...•••..••.•••••••••...•.•.•..•.•.•.••.• 176

6. 1. Sha,p Bands at Low Split Ratios .•..••............ .....•......... 176

6.1.1. Headspace Analysis 176

177

6.1.2. Rapid Isothermal Runs at Elevated Column Temperature 6.2. Optimized Split Flow Rate ..................•...........•........................•............. 177

6.2.1. Peaks Growing Broad instead of High 178

178

6.2.2. Dilution in the Injector 178

6.2.3. Dilution in the Column 6.3. Maximum Vapo, Concent,ation in the Injecto, •.......•..•...........•............. 179

6.3.1. Sample Volume 179

181

6.3.2. Optimum Liner Volume 6.3.3. Position of the Column Entrance 182

6.3.4. Injection Point 182

183

6.3.5. Syringe Needles 6.4. Column Flow Rate .............................................................•................... 183

6.4.1. Low Split Ratios Resulting from High Column Flow Rates 184

6.4.2. Selection of the Carrier Gas 184

184

6.4.3. Selection of the Column 6.5. Summary: Maximum Sensitivity f,om Split Injection 185

7. High Split Ratios for Reducing the Sample Size 7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 7.7.

185

Diluent as a Hypothetical Sample 186

The Maximum Split Flow Rate 186

Small Sample Volumes .•..... .•............ 187

Low Column Flow Rate. .•..... ...•........... 189

High Column Capacity - Thick Films ....................................•................ 189

Length of the Sy,inge Needle 190

Summa,izing Guidelines ........................................................•....•.......... 191

8. Problems Concerning the Split Ratio .....•.•..•..•.....••.•••..................••••............. 192

8. 1. Pu,poseful Sea,ch fo, E"o,s •.. .••.•..... .••.... 8.1.1. Systematic Errors 8.1.2. Message from Standard Deviations 8.2. "P,e-Set" ve,sus "T,ue" Split Ratio ....................................•..........•...... 8.3. Mechanisms Causing the Split Ratio to Deviate 8.3.1. The Pressure Wave : 8.3.2. Dependence of the Pressure Wave on Gas Regulation

192

192

193

194

195

195

196

Contents

XV

8.3.3. Recondensation in the Column Inlet 198

8.3.4. Incomplete Evaporation 200

8.3.5. Cool Split Line 200

8.3.6. Charcoal Filters 201

8.3.7. Buffer Volumes 201

8.4. Minimizing the Deviation from the Pre-Set Split Ratio •...••.•.•.•..•.......... 202

8.4.1. Wide Injector Liner 202

8.4.2. Long Distance between Needle Exit and Column Entrance 202

8.4.3. Small Sample Volumes 203

8.4.4. Volatile Solvents 204

8.4.5. Packed Liner 204

8.5. Experimental Results .......•..............•.•.....••........•...•.•....•....•.................... 205

8.5.1. Results Concerning Pressure Wave 205

8.5.2. Course of the Pressure Wave 207

8.5.3. Data on True Split Ratios 208

8.6. Working Rules to Prevent Systematic Errors 209

8.6.1. No Quantitation on the Basis of the Pre-Set Split Ratio 209

8.6.2. Use of the Internal Standard Method 209

8.6.3. Apply the External Standard Method with Caution 210

9. Problems Concerning Linearity of Splitting

213

9. 1. uLinear" Splitting •........•.....................................................•.•..•...•......... 213

9.2. First Cause of Non-Linear Splitting: Diffusion Speeds 214

9.2.1. Isokinetic Splitting 215

9.2.2. Insufficient Experimental Evidence 216

9.2.3. Conclusion 217

9.3. Second Cause: Incomplete Sample Evaporation 217

9.3.1. Vapors and Droplets Split at Different Ratios 217

9.3.2. Neat Samples 217

9.3.3. Dilute Solutions in Solvents 219

9.3.4. Conclusion 220

9.4. Third Cause: Fluctuating Split Ratio •......•....•.•.............•........................ 220

9.4.1. Variation of the Split Ratio 220

9.4.2. Pre-Separation of the Sample in the Injector 221

9.4.3. Cognac as an Example 222

9.5. Danger of Systematic Errors .•.................................•....•.•....•.•............... 223

10. Techniques for Improving Quantitative Analysis

225

10. 1. Systematic Search for the Sest Conditions 225

10.1.1. Strategy: Minimized Deviation 225

10.1.2. Determination of the Correct Result 226

10.2. Flash Evaporation .........•.........••.•...............................•......•.................. 226

10.2.1. Concept 226

10.2.2. Selection of Conditions 227

10.2.3. Problems Arising from Aerosol Formation 229

10.2.4 Stop Flow Split Injection 230

10.2.5. An Experimental Result: Determination of Sucrose 230

10.2.6. Evaluation of Flash Evaporation 231

10.3. Evaporation in Packed Liners 231

10.3.1. Deposition of the Sample 232

10.3.2. Injector Packings , 233

10.3.3. Optimization of Conditions 234

XVI

Contents 10.3.4. Elution from the Packed Bed 234

10.3.5. PAHs as an Example 235

237

10.3.6. "Ghost" Peaks as a Result of Packing Bleed 10.3.7. Matrix Effects 237

10.4. High-Boiling Samples ••.•.•..........•.•..•.•.....•..•....•.......•............................ 239

10.4.1. Optimization of Conditions 239

10.4.2. Experiments by Schomburg 242

242

10.4.3. Application to Herbicide Analysis 10.5. Homogenization of Vapor Across the Liner 243

10.5.1. Obstacles Promoting Homogeneous Distribution 244

10.5.2. Chromatographic Experiment with Two Columns 244

245

10.5.3. Fatty Acid Methyl Esters 10.6. Two Case Studies •..........•....................................•......•........•............... 246

10.6.1. About a Dispute: the Methanol/2-Ethyl-1-Hexanol Mixture 246

248

10.6.2. Analysis of Alcoholic Beverages

11. General Evaluation of Split Injection

251

References C

254

o Splitless Injection 1. Introduction 1.1. Concept 1.2. Historical Background

2. How to Perform Splitless Injection 2.1. Basic Steps of Splitless Injection 2.2. Closing the Split Exit 2.2.1. Mechanical Pressure Regulation 2.2.2. Flow/Back Pressure Regulation 2.3. Purging the Injector 2.3.1. Duration of the Splitless Period 2.3.2. Purge Flow Rate Required 2.4. Septum Purge 2.5. "Ghost" Peaks from Septum Material 2.6. Septum Purge During the Splitless Period 2.6.1. Arguments in Favor of Closing 2.6.2. Sample Material Entering the Carrier Gas Supply Line 2.6.3. Reasons to Leave the Septum Purge Open 3. Sample Volumes Suitable for Splitless Injection 3.1. Calculated Volumes of Solvent Vapor 3.2. Determination of Injector Capacity 3.2.1. Determination from Peak Sizes 3.2.2. Detection of Solvent in the Septum Purge 3.2.3. Measurement of Losses through the Septum Purge 3.3. Results 3.3.1. Pressure Wave versus Diffusion 3.3.2. Volume of the Vaporizing Chamber

257

257

257

261

261

261

262

263

263

264

265

265

265

266

266

267

268

269

269

270

270

270

272

272

273

273

Contents 3.3.3. Length of the Syringe Needle 3.3.4. Inlet Pressure 3.3.5. Solvent Recondensation 3.3.6. Volume of Vapor from Solvent 3.3.7. Liners with a Constriction at the Top? 3.3.8. Valve to prevent Backflow 3.4. Pressure Increase during Splitless Injection 3.4.1. Auto-Regulation? 3.5. Slow Injection 7 3.6. Conclusions 4. Injection of Large Volumes 4. 1. Overflow Technique 4.1.1. Evaporation from Cool Surfaces 4.1.2. Injection Rate 4.1.3. Keeping the Liquid in Place 4.1.4. Retention of Volatile Components 4.1.5. Desorption of Solute Material 4.1.6. Instrumental Requirements 4.1.7. Syringe Needles 4.1.8. Flow Rate through the Septum Purge 4.1.9. Column Temperature During Injection 4.1.10. Examples 4.2. Precolumn Solvent Splitting 4.3. Evaluation 4.3.1. Overflow Technique 4.3.2. Solvent Splitting 5. Sample Transfer into the Column 5. 1. Spreading in the Vaporizing Chamber 5.1.1. Observations with the Iodine Experiment 5.2. The Transfer Process 5.3. Flow Rate and Duration of the Splitless Period 5.3.1. Carrier Gas Flow Rates 5.3.2. Liner Bore 5.3.3. Diffusion Speeds 5.4. Accelerated Transfer by Pressure Increase 5.4.1. Principles 5.4.2. Advantages 5.4.3. Extent of Pressure Increase 5.4.4. Duration of the Pressure Pulse 5.4.5. Accentuated Solvent Recondensation 5.4.6. Recommendations 5.5. Accelerated Transfer by Solvent Recondensation 5.5.1. Efficiency of the Recondensation Effect 5.5.2. Experimental Results 5.6. Tests on Completeness of Sample Transfer 5.6.1. Rapid Check via Accentuated Transfer Conditions 5.6.2. Check via On-Column Injection 5.7. Fast GC/Narrow Bore Columns 5.8. Splitless Injection for SPME .. ,

XVII

274

275

275

275

277

278

279

279

280

281

282

282

282

284

284

285

286

287

287

287

288

288

289

291

291

291

292

292

292

293

294

295

296

297

298

298

299

300

302

303

304

304

305

305

306

307

307

308

310

XVIII

Contents

5.9. Conclusions 5.9.1. Diameter of the Vaporizing Chamber 5.9.2. Duration of the Splitless Period

6. Problems with Quantitative Analysis 6.1. List of Problems Discussed in Other Parts 6.1.1. Selective Evaporation from the Syringe Needle 6.1.2. Poor Sample Evaporation 6.1.3. Injector Overloading 6.1.4. Incomplete Transfer of Sample Vapor 6.1.5. Adsorption and Retention in the Vaporizing Chamber 6.1.6. Degradation of Labile Solutes 6.2. Enhancing Matrix Effects 6.2.1. Definition 6.2.2. Descri ption of the Effect 6.2.3. Effect on Quantitative Analysis 6.2.4. Proposed Solutions 6.3. Reducing Matrix Effects 6.3.1. Contaminants Simulated with DC-200 6.3.2. Triglycerides in the Sample Matrix 6.3.3. Interpretation of the Experimental Results 6.3.4. Effects on Quantitative Analysis 6.3.5. Minimizing the Matrix Effect 6.3.6. Glass Wool in the Liner?

7. Reconcentration of Initial Bands 7.1. Distinction between the Two Band Broadening Effects 7.1.1 in Space 7.1.2 in lime 7.2. Band Broadening in Time 7.2.1. Shape of the Band 7.3. Reconcentration by Cold Trapping 7.3.1. Principle 7.3.2. Reconcentrating Power 7.3.3. Reconcentration Required 7.3.4. Practice of Cold Trapping 7.3.5. Problems with Disturbed Baselines 7.3.6. "Ghost" Peaks 7.3.7. Application of Cold Trapping 7.4. Reconcentration by Solvent Effects 7.4.1. Recondensation of Solvent 7.4.2. Requirements for Solvent Effects 7.4.3. Effects on Retention limes 7.5. Band Broadening in Space 7.5.1. Shape of the Initial Band 7.5.2. Extent of Peak Distortion 7.5.3. Avoidance of Peak Distortion 7.6. Uncoated Precolumns - Retention Gap Techniques 7.6.1. Reconcentration of Bands Broadened in Space 7.6.2. Uncoated Precolumn as Waste Bin 7.6.3. Press-Fit Connections

311

312

312

313

313

313

314

314

314

316

316

316

316

317

318

320

323

323

325

327

328

330

332

334

334

334

335

337

337

339

339

340

340

341

344

346

350

350

351

354

356

357

357

361

365

365

366

368

372

Contents

7.7. Examples of the Use of Reconcentration Effects 7.7.1. Dioctyl Phthalate 7.7.2. Traces of Tetrachloroethylene 7.7.3. Extraction of Water with Pentane 7.7.4. Semivolatiles in Cigarette Smoke 7.7.5. Solvent Residues in Pharmaceutical Preparations 7.7.6. Headspace Analysis 7.7.7. Solvent Effects at Elevated Column Temperatures 8. Related Injection Methods

8. 1. Direct Injection 8.1.1. Injector Design 8.1.2. On-Column Injection? 8.1.3. Injection of Large Volumes 8.1.4. Evaluation of Direct Injection 8.2. Solid Injection 8.2.1. Moving Needle Injection 8.2.2. Direct Sample Introduction 8.3. Injector-Internal Headspace Analysis

9. General Evaluation of Splitless Injection

XIX

374

375

375

375

376

377

377

378

379

379

379

381

382

383

385

385

386

388

391

9.1. Data on Precision from the Literature ..........•..•.••......•........•.................. 391

9.1.1. Limited Utility of Literature Data 394

9.1.2. Message to a Lawyer 394

9.2. Comparison with Alternative Techniques 395

9.2.1. On-Column Injection 395

9.2.2. Splitless Injection for Analysis of "Dirty" Samples 396

9.2.3. PTV Splitless Injection 397

9.2.4. Outlook 397

References D

398

E Injector Design 1. Vaporizing Chamber

1. 1. Classical Teaching 1.1.1. Longitudinal Axis 1.1.2. Internal Diameter for Splitless Injection 1.1.3. Internal Diameter for Split Injection 1.1.4. Conclusions 1.1.5. Column Installation 1.2. Newer Developments 1.2.1. Pressure and Flow Programming 1.2.2. Fast Autosampler .., 1.3. Room for Improvement7 1.3.1. Preference for Thermospray or Band Formation? 1.3.2. Optimized Thermospray 1.3.3. Optimized Injection with Band Formation

406

406

407

.408

.409

410

410

412

412

412

413

414

414

414

XX

Contents

2. Surroundings of the Vaporizing Chamber

416

2.1. Seal between Liner and Injector Body? ..•.•..............•............................. 416

2.2. Accessible Volumes around the Vaporizing Chamber .•••.•••.••.•••....••.....• 420

2.2.1. Reversed Split Flow? 422

2.3. Septum ...................•......•.....•..•.................•.•...•.....................•................ 423

2.3.1. Required lightness 423

424

2.3.2. Septum Bleed 425

2.3.3. Effect of Particles on Sample Evaporation 2.3.4. Recommendations 426

427

2.3.5. Merlin Microseal 2.4. Heating of the Injector ................•.•....................................................... 427

2.4.1. Injector Head 428

429

2.4.2. Base of the Injector 3. Autosamplers .....•....•••••.....•......••............•.......•.....•...•....•......•...•....••..•........... 429

3.1. Injection Speed.......•••..•••..••..•....•••.•....•...••.•.•..••...••••.....•...••••....••..•••...••• 429

3.1.1. Injection Rate 430

430

3.1.2. Adjustable Depth of the Needle 4. The Gas Regulation Systems ........................•...•......•.....•........•......•.....•..•.•.... 431

4.1. Mechanical Pressure Regulation/Flow Restriction 431

4.1.1. Pressure Regulators 432

434

4.1.2. Manometers 4.2. Mechanical Flow/Backpressure Regulation 435

4.2.1. Comparison of the Two Systems 436

4.3. Electronic Regulation Systems•...••.•••.•..•.••••.•..•..................•.••••.•.•......... 437

4.3.1. Flow/Backpressure Regulation 437

4.3.2. Pressure Regulation/Flow Restriction 438

4.4. Charcoal Filters in the Split Outlet 439

4.4.1. Adva ntages 439

4.4.2. Drawbacks 440

4.4.3. Suitable Size 440

4.5. Septum Purge ..•••..•.............•.....•......•..•.•......•••..•••.....••...••.....•....••...•...••. 441

References E .•••...•.•......•......•••..................•..•...•.•.••...•......•.•......•....•.......•....•••...•• 443

Appendix 1 .••..••........•........•...••............••...•.••••..••...•.•••.......•............•.•.••.•....•.•.•..• 445

Selection of the Injection Technique .........•.......................•........................... 445

Appendix 2 ..••....•...............••...••.....•••....•.••.....•...•..•....••...•.....•••....•..•..•.•..••.......... 446

Selection of Conditions for Classical Split and Splitless Injection ....•.......... 446

Appendix 3 •.....•....•.....................................••....••..•..••....•.......•....•.•.•..••..•.....•..••.. 448

Glossary of the Most Important Terms Used in the Text •••.•••..•••.••••••..•••....••• 448

Subject Index •...•.•.•..•.•...••.•.....•.....•.•.•..•••....••••...••..•.•.......•.....••...•...•..•...••.......•.. 453

1.1. Syringe Injection

1

A Syringe Injection into Hot Vaporizing

Chambers

1. Introduction 1.1. Syringe Injection

There are several reasons for the general success of the syringe for sample introduction in chromatography: the flexibility with which the sample volume can be ad­ justed; the possibility of releasing the sample in a predetermined region of the vaporizing chamber; withdrawal of the device after depositing the sample; easy cleaning of the sampling device; easy construction of autosamplers - the sample can be picked from the vial closed by a septum using the same device. This does not mean, however, that the syringe only has ad­ vantages, as will be shown below, but the alternatives en­ gender just as many problems and inconveniences.

Alternatives

In fact, in the past, some alternatives have been tested, but none has become a serious competitor with the syringe. Systems based on rotating switching valves, similar to those used in HPLC, have been proposed several times (e.g. [1)). They are widely used for gaseous samples, but not for the liquids commonly analyzed. Samples have been placed in small capsules which were opened in the vaporizing chamber. For solid (solvent-free) injection solutions were placed in glass tubes of ca. 15 x 0.7 mm i.d., from where the solvent was evaporated in a manifold that could be evacu­ ated. These tubes were then dropped into the vaporizing chamber from a rotating block situated above the chamber.

2

A 1. Introduction

Complex Process

At first sight, the concept of syringe injection into the classi­

cal vaporizing injector seems to be obvious - the needle re­

leases a liquid sample into the hot vaporizing chamber, where

the Iiq uid quickly evaporates such that only vapors reach the

column entrance. On closer inspection, the process is more

complicated.

1 The sample solvent (normally more than 99 % of the sam­

ple consists of volatile solvent) evaporates at least par­ tially inside the needle because the latter enters a zone at a temperature far above the solvent boiling point. Fast autosampler injection is an exception to this. 2 Evaporation inside the needle produces a spray effect that largely determines sample evaporation inside the vaporizing chamber. It is, in fact, the prerequisite for sample evaporation inside empty injector liners.

Neglected Subject?

The problem of syringe injection into vaporizing injectors has long been neglected, although some analysts, mostly working with packed columns, have been aware of it since the sixties. Perhaps the complexity of the problem was the reason, hindering the discovery of simple, generally valid solutions. The discussion of how to inject a liquid sample also has a touch of awkwardness, comparable perhaps with teach­ ing an adult how to eat Italian spaghetti without smearing the red tomato sauce over his face and tie. Evaporation in­ side the needle is, however, often the major source of error in quantitative analysis, and it might well turn out that introducing a sample in a volatile matrix into a hot injec­ tor is even more difficult than eating spaghetti properly in front of a very important person.

1 .2. Sample Evaporation inside the Needle

It is tempting to think of sample introduction into the injec­ tor as a purely mechanical process executed by depressing the plunger ofthe syringe - an injection as in medicine or liquid chromatography. In cold on-column injection this is indeed the case, but in vaporizing injection it is the excep­ tion rather than the rule. Partial evaporation in the needle causes two main problems.

1.2. 1. Inaccurate Sample Volume

Sample (solvent) evaporation in the syringe needle renders the amount of sample delivered into the injector unreliable (Figure A 1). Syringes are conceived to inject an amount of liquid that corresponds to the volume read on the barrel of the syringe. The liquid inside the needle is not measured by the commonly used plunger-in-barrel syringes (of, e.g., 10 Ill) because it is supposed to remain there at the end of the injection. If a solution in a volatile solvent is introduced into an injec­ tor at 250 to 300°C, it is difficult to prevent some liquid evapo­ rating and emptying the needle largely. Because of this, the

amount of sample injected is greater than that meas­ ured. Because the volume inside the needle is 0.6-1 III and

1.2. Sample Evaporation inside the Needle

3

Syringe

Injector insert

Sample liquid Evaporating solvent + volatile solutes Layer of high boilers

~.~

:.:.:::':

Ejected sample liquid

..~;:

Figure A1 Basic problems caused by syringe injection of samples in volatile matrices into hot injectors. a) Some of the sample material which should remain in the syringe needle at the end of the injection is expelled. in­ creasing the volume of sample actually introduced above that measured on the barrel. b) Part of the high-boiling solute material remains on the internal wall of the needle and is finally taken out of the injector with the syringe, resulting in a distortion of the sample composition (discrimination).

the sample size commonly injected between 1 and 2 ~L, the needle volume is anything but negligible. Injection of a vol­ ume below ca. 0.6 ~L is not possible if the needle volume is emptied. 1.2.2. Discrimination against High Boilers

Discrimination resulting from selective elution from the syringe needle is often even more troublesome. When the analyst withdraws the plunger after an injection, he might find little liquid hanging on the tip of the plunger. It is tempt­ ing to conclude that most of the needle volume has been transferred into the injector and that a nominal injection of, e.g., 1 ~L in reality introduced 1.6-2JlL. While this conclusion may be correct for the solvent and the most volatile solutes, components with an elevated boiling point are likely to be only transferred partially; of these an equivalent of only, e.g., 1JlL was injected - the exact amount cannot be determined visually. Thus, high-boiling sample components enter the vaporizing chamber in amounts which are too low relative to the others, and hence are "discriminated" against compared with the volatile material.

Overdosage of Volatiles

It may be objected that one should speak of "overdosage" of the volatile components rather than "discrimination"

4

A 7. Introduction

against the high boilers because, in fact, too much of the volatile material is analyzed. However, such terminology has not become popular. Samples of Broad Range of Volatility

Discrimination by selective elution from the needle is a severe problem for samples containing components of a wide range of volatility, particularly when some have elevated boiling points; it is mostly negligible when all solutes are volatile, and absent if gases are injected (including headspace analysis). Discrimination is one ofthe main reasons why the volatility of internal standards should be similar to that of the solutes of interest.

1.2.3. Poor Reproduc­ ibility

Deviations because of partial elution from the syringe nee­ dle call for compensation by means of calibrated correction factors (often wrongly termed "response factors"). The de­ viations are, however, usually poorly reproducible both within a series of injections of the same solution (random error) and between injections of different solutions, such as the calibration mixture and the samples. This results in in­ creased standard deviations and possibly systematic errors.

1.2.4. Degradation of Labile Solutes

Degradation of labile solutes on the hot metallic needle' surface or on the layer of contaminants deposited on the internal wall of the needle may be another problem. GC in­ struments are constructed such that the sample does not make contact with metal surfaces, but if a component evapo­ rates from the needle wall, such contact is intense.

1.3. Conclusions

As sample evaporation inside the vaporizing chamber is linked with that inside the needle, Sections A and B are inter­ related and are directed towards the following conclusions.

1.3. 1. Fast Autosampler?

In the second half of the nineteen eighties, Hewlett-Packard introduced the fast autosampler which avoided sample evaporation inside the needle. For some time this seemed to be the solution of the problem, although it meant that manual injection was no longer equivalent - the autosampler was no longer an automated version of manual injection, but a different technique often producing significantly different results.

Handicapped Evaporation in the Injector

This conclusion was questioned again when it became obvi­ ous that the fast autosampler not only solved a problem, but also created a new one - it rendered sample evaporation in­ side the vaporizing chamber more difficult (Qian et al. [2]). Figure A2 anticipates the conclusions of Sections A and B; there is a dilemma - performance regarding syringe intro­ duction is traded against evaporation performance inside the vaporizing chamber.

1.3. Conclusions

5

Evaporation inside the needle nebulizes the liquid

Sample liquid forming a band that must be stopped

.~i~lR} Microdroplets ;l~~ evaporating in the -~~~ :\~?:. gas phase

Packing

Injection suppressing evaporation inside the needle

Thermospray resulting

from partial evaporation

In the needle

Figure A2

The dilemma regarding sample evaporation: fast autosamp­

lers avoid evaporation inside the n_dle, but render vapori­

zation inside the liner difficult. Slower injection causes par­

tial evaporation inside the n_dle, which improves vaporiza­

tion inside the liner by production of a thermospray.

1.3.2. Suppressing Evaporation inside the Needle

With regard to the accuracy of the sample volume injected and the composition of the sample analyzed, the best tech­ niques for introducing the sample into a hot chamber are those preventing sample evaporation inside the syringe needle. This can be achieved by injection at a velocity such that heating and evaporation of the sample inside the syringe needle is avoided (fast autosampler), injection of samples in high-boiling solvents, or injection through a short needle. Programmed temperature vaporization (PTV) and on-column injection are also solutions to this problem.

Band Formation

Injection suppressing evaporation in the needle causes the sample liquid to leave the needle as a band (jet). As this band moves at high velocity and covers long distances in hot cham­ bers, it must be stopped by a packing (such as deactivated glass wool) or by obstacles (Section B). This may lead to losses of high-boiling, adsorptive, or labile solute material.

1.3.3. Thermospray

The most gentle sample evaporation inside the vapor­ izing cha.mber is obtained when some solvent evaporation inside the needle nebulizes the sample liquid at the needle

6

A 1. Introduction exjt. The resulting microdroplets readily evaporate while suspended in the carrier gas. This avoids contact with adsorptive or contaminated surfaces. Because vaporization inside the needle often causes uncon­ trolled elution, the technique must be optimized such that transfer from the needle is as complete as possible. Sample volumes will be too large, however, and discrimina­ tion against high boilers cannot be totally avoided.

2. Syringes Here syringes suitable for vaporizing injection are described. Catalogs of syringe suppliers provide useful further infor­ mation. A summary of the subject has been published by Hinshaw [31.

2.1. Plunger-in-Barrel Syringes

Figure A3 shows the front of the most commonly used microsyringe with a fixed needle. The needle is sealed into the glass barrel by means of a droplet of epoxy glue. The· sample volume to be injected is measured in the barrel of the syringe and does not include the liquid inside the needle. Measurement assumes that the needle remains filled with liquid. Seal with glue

Plunger

I \~.----.~~------+{

Needle

I

I

/

Sample volume

Glass barrel

Figure A3 The most commonly used syringe with fixed needle end steel plunger.

2.1.1. Plungers

Steel plungers seal against the glass barrel by closely fitting dimensions: clearance between the plunger and the barrel is approximately 0.5 urn. Because the glass barrels and steel wires cannot be fabricated with the appropriate precision, plungers are adjusted individually by immersion in acid. This explains why plungers should not be exchanged from one syringe to another (if they seem to fit, they might not be tight).

PTFE Tips

Plungers with a PTFE tip have been less successful. They enable the production of syringes with exchangeable plung­ ers at lower cost, but tightness usually becomes a problem after prolonged use.

2.1. Plunger-in-Barrel Syringes

7

Tightness of the Plunger in the Barrel

Moderately high pressures are encountered when the nee­ dle is inserted into the injector and the carrier gas inlet pres­ sure is high. Far higher pressures can, however, be reached during depression of the plunger, because the cross section of the latter is only ca. 0.2 mm 2• Force on the plunger corre­ sponding to 100 g (which is clearly more than normally ap­ plied) relates to 50 bar or 5 MPa.

Maximum 80 % Withdrawal of Plunger

lightness of steel plungers without PTFE tips depends on the position inside the barrel - the further the plunger is withdrawn, the shorter is the tight section. This is why it is sometimes recommended that the plunger is not withdrawn by more than about 80 % of the syringe capacity. This means that in a 10 IJ.L syringe, the tip of the plunger should not be behind the 8IJ.L mark.

Viscosity of the Sample

lightness also depends on the viscosity of the medium between the plunger and the barrel- seals are tight up to far higher pressures when there is a film of liquid instead of gas; the type of liquid (usually the solvent) also has a strong influence. Syringes with capacities of 50-500 IJ.L are available with steel plungers fitting tightly in the barrel (as for 10 IJ.L syringes), as well as with "gas-tight" plungers equipped with PTFEtips. Steel plungers are more reliable because they are not de­ formed during prolonged use, as are PTFE tips. If they are used for injection of gases, however, tightness is critical be­ cause of the low viscosity of the gas.

Test of Tightness

In case of doubt, the tightness of the fit of the plunger in the barrel should be tested. For injection of liquids a solvent of low viscosity, such as hexane, is picked up and pulled backwards out of the needle into the barrel. The needle is inserted into an injector with a high gas pressure inside. If there is leakage, the meniscus of the liquid moves upwards and liquid accumulates in the region where the plunger leaves the barrel. The test becomes sensitive when the plunger is inserted a short distance only into the syringe and when waiting for a time longer than during a normal injection. The most sensitive test involves a dry syringe. The plunger is pulled out of the barrel and allowed to dry. The needle is introduced into an injector, causing a stream of carrier gas to flow backwards through the syringe and dry it. The plunger is then re-introduced to the level to be tested and a drop of a solvent of low viscosity (such as hexane) is placed around the plunger where it enters the glass barrel (Figure A4). Some liquid flows into the narrow gap between the plunger and the barrel. Escaping gas (leakage) is sensitively detected by visual observation.

2.1.2. Plunger Guides

With manual injection, death of syringes most frequently results from deformation of the plunger - when not de­

8

A 2. Syringes

Figure A4 Test of the tightness of the plunger by application of a drop of liquid in the region where gas would leave.

pressed concentrically, the steel wire is bent. Plungers can­ not be re-straightened properly, because there remains a deformation that rubs on the glass wall. This hinders fast depression (as required for hot needle injection). Grayish sludge containing fines from the plunger and the glass soon further hinders the movement of the plunger. The plunger guide was introduced as a solution to this problem. The, plunger guide can also be of advantage for the injection of samples in highly volatile matrices, because warm­ ing of the barrel by the fingers can be avoided. Elongated Barrel

SGE elongated the glass barrel by adding a region of wider bore in which a thicker rear part of the (also elongated) plunger moves (Figure A5). Only this robust thicker section leaves the barrel. Hamilton produces removable metal plunger guides working on the same principle. One draw­ back is that the syringe is heavier and more difficult to han­ dle with one hand only. Measuring section with fine plunger

Plunger guide with more robust plunger

I Figure A5 Syringe with plunger guide.

5 ul: Syringes

As prices of syringes decreased, fewer 10 ~L syringes with plunger guides are used. For 5 ~L syringes, however, the use of a guide is recommended. Their plunger has only half the cross section and is bent correspondingly easily.

Reinforced Plunger Neck

Because a high proportion of all plungers are bent when they reach the zero position (they are pushed excentrically into the barrel), SGE produces syringes of standard length, but with reinforcement of the last section of the plunger that enters a specially designed nut at the rear of the barrel. The plunger button is reinforced also. This facilitates fast depres­ sion of the plunger as needed forthe "hot needle" technique.

Flexible Plunger

SGE also offers a syringe with an elastic plunger which can­ not snap off or be deformed permanently.

2.2. Plunger-in-Needle Syringes

2.2. Plunger-in-Needle Syringes

9

Plunger-in-needle syringes keep the sample inside the needle. The plunger is equipped with a thin wire protruding into the needle to displace the liquid (Figure A6). The barrel of the syringe indicates the position of the wire inside the needle, but does not make contact with the sample. All the liquid is displaced. Measured sample volume

Needle

Figure A6

Plunger in needle syringe.

1ul: Syringe

Plunger-in-needle syringes of 0.5 to 25 III capacity are avail­ able commercially, but only the 1 III syringe has found wide­ spread use. It enables accurate measurement of ten times small:er sample volumes than standard 10 III syringes, i.e. as little as 0.05 Ill, and suggests itself for the injection of non-diluted samples. Standard needles are 56 or 70 mm long. 56 mm needles have a 90° cut at the outlet; the internal and external diameters are 0.15 and 0.70 mm, respectively. 70 mm needles have a 17° tip; internal and external diameters are 0.15 and 0.47 mm, respectively.

Problems

There are several problems connected with injection into hot chambers; they will be discussed in Section A9. Cleaning is more difficult and there is no visual control of whether or not air bubbles are included in the sample plug.

No Withdrawal of Plunger

One should resist the temptation to take a look at the fine tungsten wire serving as the plunger - after the plunger has been fully withdrawn from the syringe, it is extremely dif­ ficult to insert it again.

2.3. Syringe Needles

Needle diameters are standardized by "gauge". Those most important for GC are listed in Table A1. The intemal diameter is kept as small as possible to minimize the inner volume of the needle (extra volume being transferred when the needle is heated). On the other hand, the needle should not cause build up of an excessive pressure drop, because this hinders sucking up the sample liquid, particularly when volatile solvents are involved. The outer diameter is a compromise between robust­ ness and a minimized effect on the septum.

2.3.1. Dimensions

Length

Standard syringes are equipped with needles 51 mm long (2 inches, including the section glued into the glass barrel). As will be shown later, split injection at low split ratios and splitless injection often require longer needles, commonly

A 2. Syringes

10

Table A1 Diameters of the most important syringe needles and internal volumes for needles of 51 mm length.

Gauge

22 225 23 235 25S 26 26S

Diameters (mm) internal external 0.41 0.15 0.64 0.64 0.15 0.26 0.13

0.72 0.72 0.34 0.15 0.52 0.46 0.47

Internal volume (Ill) 6.73 0.90

Main use

0.90

Headspace Autosampler Autosampler Autosampler Autosampler

0.68

Manual injection

71 mm (3 inches). For injection with band formation, 3.7 mm (1.5 inches) needles are most suitable. Gas syringes for headspace analysis should have an 80 mm needle with a side port hole. Syringes with needles of custom length are available at a small extra cost.

2.3.2. Needle Tips Beveled Tips

The standard style needle tip for injection through a septum, the beveled point, is polished at an angle of 17·20°. The tip is bent slightly inward, i.e. towards the center of the tubing, for better displacement of the septum and to reduce the chance of the needle cutting away a particle of the septum material. The tip is easily bent, e.g. after the syringe is dropped on the floor. The deformation is more easily felt by sliding the fingers over the needle tip than seen by eye. It affects the way the liquid exits the needle (see Section 83.2) and scrapes a hole into the septum. Needle tips should, therefore, be regu­ larly checked.

Conical Style

Syringes for autosamplers, in particular, are often equipped with conical style needles - cut squarely, but polished to a cone with an 8° angle. If they always pierce the septum at the same position, they are supposed to reduce septum coring (and resulting deposition of particles inside the va­ porizing chamber).

Side Port Hole

The tip of needles with a side port hole is closed to a rounded point. About 1 mm back, there is a small hole in the side wall. This needle style practically rules out cutting of septum particles and is unlikely to be plugged, which is particularly suitable for headspace syringes, because other needles tend to be plugged. They have, nevertheless, never become popular. For injection of liquid samples, release through the side port influences sample evaporation and distribution within the vaporizing chamber- sometimes advantageously, some­ times not.

2.3. Syringe Needles

2.3.3. Fixed versus Removable Needles

11

Most manufacturers offer syringes with fixed needles, ce­ mented into the barrel at a position corresponding to the zero graduation, or removable needles, tightened against the barrel with a small PTFEferrule. When a fixed needle is dam­ aged, the entire syringe must be replaced; this is probably the only argument in favor of the removable needle.

Problems with Removable Needles

Prices of syringes with removable needles are substantially higher, and this investment is justified only when the needle is ruined rather frequently. Furthermore, connection of the needle to the barrel can be a problem, firstly, because it usu­ ally retains some air and encourages bubble formation, like a boiling stone, when picking up volatile solvents. Sec­ ondly, some connections have significant dead volume ­ sample material enters this by diffusion, particularly when the syringe is lying around after the injection with sample liquid remaining in the critical region. Because rinsing with solvent does not clean dead volumes, this readily generates "memory effects".

2.4. Cleaning of Syringes

Before investing much effort in sophisticated procedures for cleaning syringes, it is useful to consider some general rules which help minimize the effort required. Such rules might even become parts of validated methods, because the reliability of the results easily depends on them.

2.4.1. Basic Rules

When performing series of analyses, it is usually sufficient to remove 99 % of the material from the previous sample, because solute concentrations vary by less than a factor of 10. Such cleaning is readily achieved.

At the opposite end of the scale of difficulty, a syringe might

first be used to prepare a standard solution, measuring a

neat substance. Afterwards it is used for injection in trace

analysis, in which picogram quantities of the same compo­

nent, levels maybe 100,000,000 times less, are detected.

Cleaning the syringe to remove 99.9999999 % of the mate­

rial is virtually impossible.

Classify the Cleaning Re­ quired

Use the Same Syringe

Use the same syringe throughout a series of analyses (as autosamplers inevitably do). This renders the requirements more transparent. It rules out introduction of materials from other sources (such as from the preparation of a standard solution). It also ensures that the sample always leaves the needle in the same way - small deformations of the needle tip may have a strong effect on the evaporation process (e.g. through a spray effect).

Estimate the Required Cleaning Effect

lfthe samples contain the solutes of interest in amounts vary­ ing by not more than one order of magnitude (e.g. analysis of the fatty acid composition of edible oils), 99 % cleaning is sufficient. In the analysis of pesticide residues, a high con­ centration' might be 100 times above the lower detection limit

12

A 2. Syringes of ,the method, i.e. 99.9 % cleaning guarantees that subse­

quent samples will be free from residues, i.e. there is no

memory effect.

Efficiency of 99.9 % is probably about the limit of reli­

able syringe cleaning by autosamplers or manual injection

without special precautions. If higher efficiency is required,

a blank must be run after the analysis of every sample.

Beware of Concentrated Samples

A common experience, e.g., in residue analysis, is that a

highly concentrated solution of a standard is injected to find

the peak of interest (setting up the method). The samples

analyzed subsequently are all positive. As the analyst recog­

nizes that his results are puzzling, he runs a blank and con­

firms the carry-over. It is concluded that injection of highly

concentrated solutions should be avoided and that blanks

must be analyzed before running the first analysis.

Separate Syringes for Adding Standard

Although addition of standards by use of a 10 III syringe is

not highly accurate, it is frequently used in the interest of

working with small sample sizes and vials. The danger of

this procedure is that the same syringe is subsequently used

for injection of the sample. As the standard solutions are

usually 100-10,000 times more concentrated, cleaning

is demanding.

Label Syringes

It is convenient and advantageous for the reliability of the

results to use different syringes for different purposes.

It might be necessary to label them to rule out confusion,

e.g. by use of colored rings at the top of the glass barrel. Alternatively, syringes with especially short needles can be used for purposes other than injection.

Silylation Reagents

If samples contain high concentrations of derivatization rea­ gents, such as for silylation or acylation, the syringe must be cleaned immediately after injection, since hydrolysis by humidity from the air easily plugs the needle otherwise.

2.4.2. Cleaning Proce­ dures

The most simple cleaning procedure is moving the plunger up and down. The effectiveness of this procedure is limited by the volume of liquid inside the needle, which is moved up and down without really being replaced - it is merely mixed with the solvent or the subsequent sample. Because turbulence caused by transition from the narrow­ bore needle into the wider barrel provides most of the mix­ ing, the liquid should be withdrawn as fast as possible. Rapid suction also prevents all the material deposited on the sy­ ringe wall dissolving in the first small amount of liquid en­ tering the needle; this is most difficult to remove afterwards. Use of autosamplers shows that reliable 99.9 % cleaning is achieved in this manner.

Sample material between the plunger and the barrel is

not efficiently removed. The amount is, however, small - if

there is a 1 11m gap between the barrel and the plunger, this

Movement of the Plunger

2.4. Cleaning of Syringes

13

volume amounts to ca. 0.7 % ofthe whole internal volume of the barrel and 99.9 % cleaning efficiency is, hence, hardly endangered. Solvent or Subsequent Sample?

Often syringes are not cleaned with solvent, but with the subsequent sample. Whether or not this is acceptable is de­ termined by the tolerable carry-over. Material from the first sample corresponding to a volume of about 1 ~L might be transferred into the following sample. If it is assumed that the component of interest was present at a concentration 100 times higher in the first sample and that the volume of the second sample is 10 mL, contamination reaches 1 %. If the sample volume is only 1 mL (autosampler vial), contami­ nation reaches 10 %. Cleaning with the subsequent sample is, hence, acceptable if 90-99 % cleaning efficiency is sufficient.

Discharge Backwards

When performed manually, cleaning efficiency can be sub­ stantially improved by passage of a plug of liquid backwards out of the syringe. Some 5 ~L of liquid is sucked into the syringe and the plunger is removed from the barrel. At this moment, the 5 ~L are in the upper region of the barrel. The syringe is then shaken sharply such that most of the liq­ uid leaves the barrel. In this way, the poorly exchanged plug is removed and the whole channel in the barrel is rinsed. The plunger can be immersed in solvent to clean its outside before it is brought back into the syringe.

Vacuum

A source of vacuum can be used to suck solvent through the syringe. The plunger is pulled out ofthe barrel, the nee­ dle immersed in a suitable solvent, and the vacuum applied. Soft rubber or silicone tubing connecting to the vacuum is suitable - if the rear of the syringe is pressed against it, sufficient tightness is obtained. The plunger is again rinsed before being re-inserted. A weak vacuum is preferable, par­ ticularly for a volatile solvent, because a strong vacuum causes evaporation instead of rinsing.

Pressurized Solvent

Syringe cleaners are available consisting of a solvent con­ tainer connected to a source of pressurized gas. They are equipped with a septum through which the syringe needle is introduced. The plunger is removed, opening the way for the solvent to rinse the syringe needle and the barrel. The solvent may need frequent replacement, not least because septum particles tend to accumulate and release silicone components which show up in the chromatograms.

Drying the Syringe by Vacuum

Other syringe cleaners (e.g. Hamilton, SGE) heat the needle and evacuate it. The needle is introduced through a septum into a chamber that can be heated to 380°C. The plunger can be moved backwards and forwards to move the vapors, or removed completely to allow passage of a stream of air.

14

A 2. Syringes

Such a device cannot eliminate high-boiling or involatile material- on the contrary, once lacquered at the high tem­ perature, it can no longer be removed even by use of sol­ vent, as observed for injector liners. Hence the syringe should be rinsed with solvent before introduction into this type of cleaner. The device is particularly effective for plunger-in-needle syringes, because the whole part in contact with sample liquid is heated. Drying in the Injector

An equally efficient method simply uses a normal vaporiz­ ing injector. The plunger is removed and the syringe needle is inserted through the septum. With a low pressure in the injector a stream of carrier gas purges the volatile mate­ rial from the syringe. Because the syringe is purged outwards, no material enters the injector.

2.4.3. Plugged Needles

Needles may become plugged, e.g., after injection of silylated ortrifluoroacetylated samples containing high concentrations of residual reagent. When the syringe is left for some time, hydrolysis forms a plug near the tip of the needle. Plugged needles should not be cleared by applying high pres­ sure to the plunger because all too easily the barrel cracks (pressures exceeding 100 bar are easily reached).

Cleaning Wire

New syringes sometimes contain thin wires in the needle which can be used to unblock the needle. Hamilton and SGE supply thin tungsten wires for the same purpose.

Heating

A rapid method involves warming of the needle at the site where blockage is assumed. Some solvent is placed in the barrel from the rear - by removing the plunger and introduc­ ing solvent by means of another syringe with a long, thin needle. Modest pressure is then applied to the plunger while the needle is warmed gently in a yellow flame (e.g. cigarette lighter). The plug softens and is displaced by the solvent, which flushes the needle. The needle must not reach high temperatures, however; otherwise it turns permanently soft.

2.4.4. Blocked Plungers

The plunger moves with difficulty if gray sludge contain­ ing the fines of abraded glass and metal accumulates between the plunger and the barrel. Solvent usually does not remove it. Although against the advice of syringe manufacturers, pull­ ing the plunger through the fingers removes such material rather efficiently and can solve the problem if repeated. If this does not help, the sludge must be removed by use of hydrochloric or phosphoric acid. Immediately afterwards, the syringe and the plunger must be thoroughly rinsed with water and a solvent, such as ethanol or acetone. Alkali must be avqided because it attacks the glass.

3.1. The Three-Step Model

15

3. Evaporation Inside the Needle When the plunger of the syringe is pulled upwards after a manual injection of a solution in a commonly used solvent, hardly any liquid is seen clinging to its tip (plunger-in-barrel syringe). This implies that not only the volume of sample read on the barrel of the syringe was injected, but also that which should have been left inside the needle. What we see by eye, however, is the sample solvent, which is the sample component of least interest. Manual Injection

This section deals with sample (solvent) evaporation inside the needle and transfer of solute material as it occurs with manual injection or with autosamplers which imitate this. It does not apply to autosampler injection at such a speed that evaporation inside the needle is suppressed.

3.1. The Three-Step Model

At first, the problem seems to be the fate of the liquid re­ maining inside the needle after the plunger was depressed. A closer look reveals that things might be more complicated.

Assumptions

Below we consider the injection of 1 J!L of liquid measured on the barrel of a syringe equipped with a 71 mm needle of 1 J!L internal volume. We assume that the liquid is with­ drawn into the barrel of the syringe before introduction of the needle into the hot injector.

1. Evaporation of the First Liquid

The first 1 I!L of liquid injected, which is actually that left in the needle and not that observed when measuring the sam­ ple volume, encounters a needle wall which has been heated above the solvent boiling point, primarily during passage through the septum. Violent evaporation is initiated ­ vapors formed along the needle wall push some ofthe liquid out of the needle. Overpressure is built up, increasing the boiling point of the solvent; when the liquid leaves the nee­ dle, it explodes into small droplets, driven apart by the va­ pors (thermospray). The evaporating solvent leaves high-boiling material on the needle wall, because the temperature of its environment does not exceed the (pressure-corrected) boiling point of the lat­ ter (left in Figure A7).

2. Cooling of the Needle Wall

Consumption of heat by the evaporating solvent cools the surface of the needle wall. When its temperature falls to the solvent boiling point, the sample liquid wets the wall and the following liquid passes without evaporation (center

16

A 3. Evaporation Inside the Needle

2

1

•

•

Sample

evapor ation

in hot needl e;

no welting of

Ihewall

3

Syringe needl e

Liquid wets the needle cooled by solvent eva poration

: ~{

J

!

• '"

Remaining liquid eluted by partial evaporation

•

Figure A7 Three steps during injection through a hot syringe needle.

in Figure A7) . The liquid might even re-dissolve the material previously deposited on the needle wall and carry it into the injector. This picture obviously simpl ifies - only continuing solvent evaporation keeps the needle temperature at the boiling point. When the surface is wetted again after the formation of some vapor, however, the essential point is still achieved: transfer without loss of high-boiling material. 3. Expulsion of the Remain ­ ing Liquid

When the plunger reaches the bottom, the syr inge needle is filled with the second microliter of liquid (that observed in the barrel). Before the needle can be withdrawn, its surface is again heated above the solvent boiling point, caus­ ing the content to undergo partial evaporation; a mixture of vapor and droplets is ejected into the vaporizing chamber. Again high-boiling solute material from the evaporating liq­ uid is left on the needle wall.

Effect on Discrimination

Steps 1 and 3 in Figure A7 result in loss of high-boiling sol­ ute material as a result of incomplete sample evaporation on the needle surface. If the internal wall is sufficiently cooled to enable step 2, however, losses occurring during the first step are recovered .

Sufficient Cooling for Wet­ ting?

It is largely speculation whether cooling is sufficient for a step 2. If we assume that the plunger is depressed at a veloc­ ity of 1 mis, liquid enters the needle during a period of 15 ms. Partial evaporation of 2 III of liquid absorbs a consider­ able amount of heat, but the heat capacity of the needle far exceeds the heat consumed (the mass of the needle exceeds that of the sample by a factor of about 25). The sample can,

3.7. The Three-Step Model

77

therefore, cool a thin surface layer at best and, because of the high thermal conductivity of the metal, the cooling process must be very rapid if it is to be quicker than the rate at which heat is supplied. This also means that the tempera­ ture increase in step 3 is rapid - too rapid to give us a chance of (manually) withdrawing the needle before evaporation starts again. Experimental data on losses in the needle suggest that liq­ uid does wet the needle wall if the sample volume exceeds a certain minimum and depression of the plunger is fast. The videos on the CD, on the other hand, do not support this since a band of liquid should then be expected to leave the needle. 3.2. Models of Evapora­ tion inside the Needle

Losses of high-boiling solute material depend on the spe­ cific nature of the injection. It is helpful to consider the three models below which describe how the solutes can leave the needle. First we concentrate on the liquid remaining inside the needle after depression of the plunger.

3.2. ,. Distillation from

If the sample evaporates fully, only vapor leaves the needle. Vapor is expelled because of the expansion in volume ac­ companying evaporation (a factor of 100-500). According to the most simple model, transfer should be almost com­ plete, as the volume of vapor remaining in the needle is less than 1 % of the original liquid content (0.6-1 ~L of the 100­ 500 ~L of vapor formed). This assumes that all of the sample is vaporized at once. If a needle temperature of 200°C is assumed (in an injector thermostatted at 250°C), the distillation model would pre­ dict that of the n-alkanes only those with a molecular weight below that of n-undecane should reach the injector. It is, how­ ever, obvious that this does not accord with common experi­ ence.

the Needle

Theoretical Treatment

Guha [4] studied the effects of sample evaporation inside the needle both theoretically and for some test mixtures, as­ suming complete evaporation and a distillation-like model. He used basic gas laws to calculate the effect of needle size, injector temperature, carrier gas inlet pressure, and sample volatility on the amount injected. The conclusion was that representative sampling could be achieved only by use of plunger-in-needle syringes without dead volume in the nee­ dle.

3.2.2. Gas Chromatogra­ phy in the Needle

The above distillation model is inadequate, because it is not generally necessary that the solute vapor reach at­ mospheric pressure to leave the needle. In particular, the material deposited on the needle wall near the exit of the needle is well flushed out of the needle by the passage of the vapor of the volatiles (solvent).

18

A 3. Evaporation Inside the Needle

Small Vapor Pressure Suffices

Transfer of high-boiling material through and finally out of the syringe needle resembles a gas chromatographic proc­ ess - the needle is the capillary column, the condensed sam­ ple material and the contaminants from previous injections on the needle wall are the stationary phase, and the vapor of the sample (solvent) evaporating in the rear of the needle is the carrier gas (see upper scenario in Figure AS).

Chromatography in the Needle Chromatography of solutes .:~::.'

.:.:.

...:.:.....•.'.

Stream of vapor serving as carrier gas

.:~

"

Dirt layer acting as retaining

stationary phase

Ejection from the Needle

~=F

Expanding vapor bubbles

build up pressure

Figure AS Two models describing the elution of the sample from the syringe needle at the end of the injection.

The components are partitioned between the gas (vapor) phase and the liquid phase on the needle wall in accordance with their vapor pressure. A small amount evaporates. This vapor is immediately removed by the stream of sol­ vent vapor, which prompts more solute material to evapo­ rate, etc. This model correctly predicts that the solutes eluted may include components which boil at temperatures far above that of the needle.

3.2.3. Ejection f,om the

Needle

The above models require fairly gentle evaporation condi­ tions inside the needle, in particular an amount of time which is usually not available. This gives rise to a third mechanism, which again is not realistic in the extreme form. Rapid depression of the plunger might introduce the plug of liquid into the syringe needle at a speed such that no signifi­ cant evaporation occurs until the plug is fully introduced. Violent evaporation on the needle wall then forms bubbles of rapidly expanding vapor, building up high pressure and discharging the liquid through the center of the needle.

Ejected liquid carries all dissolved sample material out of the needle. irrespective of volatility. Losses and dis­ crirnination are restricted to the amount of solution evapo­ rated on the needle wall.

3.3. Conclusions Regarding Optimized Injection

19

3.3. Conclusions Regard­ ing Optimized Iniection

The above models enable us to draw the following conclu­ sions about how best to perform syringe injection into a hot injector.

Ejection rather than Evapora­ tion

Injection must be performed in such a way that ejection of sample liquid is privileged over evaporation. This means that evaporation should be rendered violent, the needle wall should be as hot as possible to build up maximum pressure with a minimum evaporation, and the liquid should be moved as rapidly as possible. This results in the recommendation to inject by the "hot nee­ dle" technique: see below.

Losses of Solute at the Rear of the Needle

It has been confirmed experimentally that most of the loss of solute material occurs at the rear of the syringe needle (Figure A9). Elution of sample components deposited on the needle wall near the needle exit is relatively easy, be­ cause this front section is flushed by the largest volume of vapor (the vapors of the solvent evaporated behind this polnt), At the rear of the needle, the opposite is true: only a small volume of vapor passes over the sample material, and the latter must move a large distance to reach the nee­ dle exit. The solute material must have a high vapor pres­ sure if it is to evaporate in the presence of only a small amount of solvent. Point at which evaporation

~ _ro:,: 10' h,•

).,f'

---"»,.. ~-m~$;~'

Solute to be flushed through a long distance by a small volume of vapor. A high vapor pressure is needed.

Solute near the needle exit. flushed forward by a large volume of vapor. A small vapor pressure is sufficient.

Figure A9 Elution from the syringe needle, considered for two small piles of solute material at the front and rear of the needle. 1 ng of material forms some 0.0001 j.1.L of vapor, i.e. the solute material cannot leave the needle without being carried by solvent vapors.

High Temperature at the Rear of the Needle

If high-boiling solute material is to have a chance of leaving the rear of the needle, the temperature must be higher there than at the front. Usually the opposite is true - because the rear of the needle is heated by the head of the injector (septum area), its temperature might well be more than 1000 below that regulated in the center of the injector (see Sec­ tion A8.2).

Contaminated Needles

It is well known that injections performed with different sy­ ringes can furnish different quantitative results. There are probably many reasons for this. One is a layer of contami­ nants on the inner wall of the needle which retains

20

A 3. Evaporation Inside the Needle

the solute material in the same manner as would a sta­ tionary phase in a capillary column. It is normal to care about the cleanliness of the injector liner - because "dirt" is visible there. Analysts hardly go peering into syringe needles, however, although we can safely as­ sume that internal wall of many needles is coated with a thick, dark-brown layer of contaminants. The solutes are, further­ more, in more intimate contact with this layer on the needle wall than with that on the liner wall. If significant quantities of high-boiling solutes are lost inside the needle, involatile sample by-products would be expected to remain on the needle wall to an even greater extent. They form a lacquer-like layer which can often no longer be washed out - at least not by solvents such as hexane, as we learn from cleaning injector liners. Matrix Effects as a Result of Expulsion from the Needle?

By "matrix effects" we understand effects of the sample on the quantitative results for the solutes of interest. The re­ sults depend on the sample by-products. It would not be surprising if injection of a "dirty" sample were to result in greater losses inside the syringe needle than a mixture of standards in pure solvent - a relatively thick layer of retaining material on the needle wall retains the high-boiling components more strongly. No data are, . however, available on this point.

4. How Much is Really Injected? The following discussion on the practical implications of sam­

ple evaporation in the syringe needle applies to 5 and 10

ilL syringes; plunger-in-needle (one-microliter) syringes are

not recommended for the analysis of samples in a volatile

matrix (Section A9).

If evaporation inside the syringe needle interferes, the vol­

ume injected is poorly defined. There is, in fact, little benefit

in having syringes graduated with an accuracy usually guar­

anteed at ±1%.

4.1. Interpretations of "Sample Volume"

The sample volume really injected is often shrouded in con­ fusion. The following three definitions are common.

Calibrated Volume

The first analyst means the volume marked on the bar­ reI. i.e. the volume he adjusted by depressing the plunger after having withdrawn a volume in excess of that required. He ne~lects the volume eluted from the needle and uses the syringe in the way it was conceived.

4.1. Interpretations of "Sample Volume"

2 1

Absolute Measurement

The second analyst is more knowledgeable. He withdraws the plunger before and after the injection and determines the total volumes of liquid present in the syringe. His sample volume corresponds to the difference between these two absolute measurements. He is probably closer to the truth, but, if the sample is a solution, he actually only knows the volume of solvent injected.

Adding Needle Volume

The third analyst adds the internal needle volume to the vol­ ume read on the barrel without having a closer look at it, saying that the volume of liquid remaining in the syringe is of little importance in relation to the other uncertainties in­ volved.

4.2. Communicating "Sample Volumes"

Because there is no convincing way of accurately determin­ ing sample volumes, there is a need to define what is meant by, e.g., "sample volume, 1 Ill". Was injection performed by a fast autosampler which avoided transfer of the needle vol­ ume or by a technique causing the needle volume to be emptied? Is the 1 III mentioned or read in a method just the needle volume? Is it 1 III measured on the barrel of the sy­ ringe plus the needle volume? It is, of course, important to know whether effectively 1 or 2 III were injected.

Proposed Definition

Distinction must be made between injections with and with­

out sample evaporation inside the needle.

1 If there is no evaporation (including PTV and on-col­

umn injection), the volume is written in the normal way, i.e. without quotation marks. 2 If there is evaporation, we suggest that the volume cor­ responding to the graduation on the barrel is written between quotation marks, i.e. as a quotation from the syringe. It indicates that the real vol ume is or was larger. 3 The read-off sample volume plus the internal volume of the syringe needle is written without quotation marks, because, for volatile components at least, it is closer to the truth.

Example

If, for instance, a needle volume of sample is manually in­ jected into a hot injector and the syringe needle has an inter­ nal volume of 0.9 Ill, the sample volume is 0.9 III or "0 Ill". For the same syringe, "1 Ill" is equivalent to 1.9 ul.,

4.3. Effects on Quantita­ tive Analysis

The problem of measuring an accurate sample volume when evaporation occurs inside the needle leads to the following three working rules:

1. Absolute Quantitation only for Volatiles

It is possible to measure the sample volume accurately ifthe solutes leave the syringe in the same proportion as the sol­ vent. When the injector temperature is 250°C and the septum cap well heated, this is usually a reasonable approximation for components up to, e.g., the n-alkane C20 or the methyl

22

A 4. How Much is Really Injected? ester of the C'6 fatty acid. For higher boiling solutes, the quantitation procedure must be independent of the sample volume because the amount injected must be re­ garded as unknown.

2. External Standard Method - Constant Sample Volume

If the external standard method is used, the sample volume must be kept constant, because an essentially unknown volume cannot be changed by a known proportion. It is, for instance, impossible to double a sample volume accu­ rately (e.g. to increase sensitivity). "1 ~L" is certainly far less than twice "0.5 ~L" (1.9 and 1.4 ~L, assuming a needle vol­ ume of 0.9 ~L), but for high-boiling solutes, 2.8 ~L is also more than twice 1.4 ~L, because the proportion of material expelled increases with the sample volume injected. In split injection, the sample volume must also be kept constant to keep the true split ratio constant.

3. Internal Standard Method - Less Critical

For analyses involving internal standards, the sample vol­ ume is less of a problem because a wrong volume is wrong to the same extent for the solutes as for the internal stand­ ard. Since discrimination arising from losses inside the sy­ ringe needle depends on the volume of sample injected, however, it is nevertheless advisable to keep the sample volume constant.

5. Syringe Needle Handling Minimizing Discrimina­

tion Discrimination against high-boiling compounds is usually more troublesome than assessment of the amount of sam­ ple material injected. Maximizing Transfer from the Needle

If the composition of the sample entering the injector is to be identical with that in the vial, all solutes must leave the syringe needle in the same proportion. For a mix­ ture containing components with a wide range of volatility, this can only be fulfilled if transfer is either totally suppressed or approaches completeness. Here we consider the second option and search for the needle handling technique which performs best.

5.1. Definitions of Tech­ niques

Discussion of ways of handling the syringe during the injec­ tion is facilitated if some methods are referred to by name.

Filled Needle Injection

Injection by the filled needle method means that the needle is full of sample liquid when inserted into the injector, i.e. the sample is not withdrawn from the needle into the

5.1. Definitions of Techniques

23

barrel. The plunger is depressed as soon as the needle is fully introduced. Autosamplers classically apply this method. Cool (Cold) Needle Injection

The measured sample is withdrawn into the barrel of the syringe (usually by more than is necessary to pull the sam­ ple liquid out of the needle); the empty needle is inserted into the injector and the plunger depressed immediately. "Cool" refers to the needle temperature at the moment the plunger is depressed.

Hot Needle Injection

The sample is withdrawn into the barrel of the syringe as with cool needle injection, but the inserted needle is pre­ heated in the injector for 3-5 s so that its temperature ap­ proaches that of the injector. Depression of the plunger oc­ curs as fast as possible; to achieve this it is advisable to with­ draw the sample 1-2 cm behind the needle entrance.

Slow Injection

Slow injection is used to release the sample at a rate such that the vapors generated are transferred to the column more or less concurrently. The sample liquid is withdrawn into the barrel just above the entrance of the needle. After the needle is inserted, the plunger is pushed down slowly (some 5-10 s per microliter of sample).

Wet Needle Injection

This is a method for the introduction of extremely small sam­ ple volumes by making use of 5 or 10 III syringes. Shortly before injection, the sample is withdrawn into the barrel of the syringe. The needle is introduced into the injector as rap­ idly as possible and withdrawn again after 2-3 s with­ out depressing the plunger. Only the liquid coating the needle wall is injected; this corresponds to 20-100 nt, de­ pending on the viscosity of the sample, the speed at which the sample plug is pulled out ofthe needle (rapid withdrawal leaves a large amount in the needle), and the wettability of the needle wall by the sample.

Solvent Flush Injection

The syringe needle is filled with pure solvent followed by ca. 0.2 III air and then the sample. The solvent is sup­ posed to rinse the sample from the needle. Other conditions are not specified, but solvent flush injection is usually per­ formed according to the cool needle method.

Air Plug Injection

Approximately 3 III of air are sucked into the syringe, followed by the sample. The air is supposed to push the plug of sample liquid through the needle. The sample must be withdrawn into the barrel of the syringe, because a slight expansion of the air plug (solvent vapor, warming of the sy­ ringe by the fingers) would otherwise push some sample liquid out of the needle. The injection speed is not defined.

Sandwich Injection

The syringe needle is filled with pure solvent, followed by ca. 0.2Ill'air, the sample, another plug of air and about 1 III of pure solvent. Again, no further details are stipulated.

UNIVERSIDAD DE ANTIoqUlJ'

BmLIOTBCA Ch"'NTRAL

24

A 5. Syringe Needle Handling Minimizing Discrimination

Fast Injection

Some authors describe their (manual) injection as "fast". They usually mean a cool needle technique performed at high speed. Other times they mean a filled needle injection.

5.2. Experimental Deter­ mination of Losses in the Needle

Comparison of different needle handling techniques requires a method for determining the loss of high-boiling solutes inside the syringe needle. It must, in particular, be possible to differentiate between losses in the needle and losses in the injector (such as insufficient sample transfer in splitless injection and non-linear splitting in split injection).

Principle

A mixture containing solutes with a wide range of volatilities is injected by the technique to be tested and peaks areas are integrated. Solvent is sucked into the syringe needle and this "needle rinse" is injected into the same or a second instru­ ment. The peak areas obtained from the two injections are compared.

5.2.1. Method with Two Instruments

A test sample consisting, e.g., of equal amounts of n-alkanes from C,o to C44 or the sample currently being analyzed is in­ jected by the split or the splitless method into the first instru­ ment. Peak areas are calculated as percentage of the most volatile component. If the latter is sufficiently vola­ tile (such as n-C,o), it can be considered to be fully eluted from the needle, i.e. in the same proportion as the solvent.

Alkanes - Equal Response

Because alkanes give practically equal FlO response per unit weight, peak areas can be interpreted directly. The results can be represented by a so-called discrimination curve, as shown in Figure A10 (lower curve, "first chromatogram"). At this stage, however, discrimination resulting from selec­ tive losses inside the syringe needle cannot yet be distin­ guished from other types of discrimination.

Needle Rinse Injection on Second Instrument

A second instrument is used to determine the solute mate­ rial left inside the syringe needle by means of a "needle rinse injection" [5]. Conditions are selected such that dis­ crimination is minimized - high injector temperature (e.g. 350°C for alkanes), hot needle injection of '" Ill". Absolute peak areas obtained from this instrument are calibrated by injecting the test sample. After injection into the first instrument, a "'Ill" volume of solvent (the same as used for the calibration of the second instrument) is drawn into the syringe without moving the plunger up and down, accepting the air plug behind the liq­ uid (otherwise solute material is lost). After giving the sol­ vent some time to dissolve the material on the needle wall, the liquid is slowly withdrawn into the barrel of the syringe and injected.

Correction for Liquid Left in the Needle

The peak area obtained for the most volatile solute in the needle rinse analysis can be assumed to correspond to the

5.2. Experimental Determination of Losses in the Needle relative

peak

25

area

80

60

40

first

chromatogram

20

10'2

14

22

28

34

40

44

alleane C.

Figur. A10 Exp.rim.ntal d.t.rmination of 10•••• in the .yringe n••dl. for cool n••dl••plit inj.ction of a n••dl. volum. (0.9 Illl. Inj.ctor t.mp.rature, 250°C. Th. low.r curv• •how. the m.anp.ak ar.a. obtain.d from r.peat.d .plit inj.ction of the t.st .ampl. (C,G-44 ...alkan••, 100 ppm .ach in h.xan.l. Th. m.an 10•••• in the n••dl. (hatch.d ar.al, d.t.rmin.d on a ••cond instrum.nt, w.r. add.d to the p.ak ar.a. of the first inj.ction. (From [5]).

amount dissolved in the sample liquid which remained at the tip of the plunger after the first injection, i.e. to the por­ tion of which nothing entered the injector (it was inside the cool section of the needle fitted into the barrel). It is sub­ tracted from the areas of all the higher alkanes. Plotting Needle Rinse Results

After subtracting the area e.g. n-C,o, the peak areas obtained from the needle rinse injections are expressed as percent­ ages of the areas obtained by calibration and added to the top of the discrimination curve from the first injection, as shown by the hatched area in Figure A10. If there is no dis­ crimination during the first injection other than that arising from the syringe needle, the combined peak areas should approach 100 % - they do so sometimes but not always.

5.2.2. Experiment with" Single Instrument

A similar experiment can be performed with a single instru­ ment only, injecting the needle rinse after the first analy­ sis. Unless highly volatile solutes are included in the mix­ ture, solute losses from the needle during the time taken for the first analysis are negligible (e.g. n-C,o did not leave the needle to a significant extent during a period of one hour). Quantitation can be simplified - if the same volume of the same solvent is used for the needle rinse injection as for the first injection, peak areas obtained from the two injections can be compared directly (otherwise the split ratio might be different). It remains, however, unknown how much mate­ rial still remains in the needle after the second injection.

A 5. Syringe Needle Handling Minimizing Discrimination

26

5.2.3. Test During Rou­ tine Analysis

5.3. Comparison of Needle Handling Tech­ niques

In everyday GC it is often difficult to ascribe strong dis­ crimination and/or excessively high standard devia­ tions to the correct cause - there are too many possible causes of similar phenomena. In such circumstances it is helpful to obtain a quick estimate of losses inside the nee­ dle. The experiment is performed as described above forthe one-instrument method; after analysis of a sample a needle rinse is injected using the sample solvent or a solvent ensur­ ing complete dissolution of critical sample components. The result immediately tells the analyst whether losses in the syringe needle contributed significantly to his problems.

Figure A11 compares discrimination by split injection with different needle handling techniques. A needle volume ("0 ul," or 0.9 ut.) of n-alkanes between Cg and C44 in hexane was injected (Carlo Erba Mod. 4160 instrument). Results from on-column injection show the "true" (undiscriminated) com­ position of the sample -the peak areas follow the 100 % line, as might be expected. All methods of needle handling resulted in discrimi­ nation starting just beyond ",C20 , even though the injec­ tor temperature (the temperature in the center of the injec­ tor, not that of the septum cap) was as high as 350 DC. Filled needle injection can be regarded as a double injec­ tion. During insertion of the needle, the volume of sample contained therein is expelled. This is followed by the rneas­

peak area

normalized for Cg (=100)

':;:--'­

100

--

---

50

25

9

12

16

20

26

Figure A11

Discrimination against high-boiling sample components as a result of losses inside the sy­

ringe needle for different injection techniques. Equal concentrations of Ce-CoM n-alkanes in

hexane; peak areas normalized relative to n-C e (average values from at least six injections).

Needle rinse analyses confirmed that discrimination was virtually exclusively a consequence of selective elution from the needle. (From [6]).

5.3. Comparison of Needle Handling Techniques

5.3. t. Filled Needle Injection

Loss Through the Septum Purge

High Losses

Speed of Needle Introduc­ tion

Poor Reproducibility

2 7

ured sample volume when the plunger is depressed. Injec­ tion must be performed rapidly to avoid peak splitting. The results in Figure A 11 were obtained by injection of a needle volume only, i.e. injection consisted of the first step only. A full needle was introduced; the plunger was kept at the "0 Ill" position throughout the injection. The injection point is ill-defined. Evaporation and elution usually start as soon as the needle pierces the hot septum. Material eluted while the needle tip passes through the injector head is vented through the septum purge. losses of high-boiling components relative to n-Cg were ex­ tremely high: they amounted to almost 10 % for the rather volatile n-C20 , 40 % for n-C32 , and 80 % for n-C 44 (Figure A 11). Whether sample material was lost through the septum purge could not be determined, because absolute areas var­ ied too much from one needle technique to another. The results shown in Figure A 11 are somewhat arbitrary av­ erages because the discrimination was highly dependent on the speed of needle introduction. If the needle en­ tered smoothly and rapidly, the results were far better than average; occasionally they even approached those obtained by hot needle injection. During such exceptionally success­ ful injections the sample apparently remained in the needle until the latter was fully inserted and most was then ejected as a liquid by the sudden onset of violent evaporation (simi­ lar to delayed evaporation). losses were particularly high when a new septum had to be pierced. possibly with a long and already somewhat distorted syringe needle. In such instances it was often found that vir­ tually no peaks of n-C36 -44 were detected. Slow introduc­ tion of the needle provides much time for sample evaporation from the needle. The reproducibility of results obtained by filled needle injec­ tion depends on the ability of the chromatographer always to inject with the same rhythm, but also on factors not under the control of the analyst. The presence of nuclei initiating evaporation, e.g. of porous material on the needle wall behaving as a boiling stone, is expected to influence the results. Sometimes an air bubble included in the sample plug expands rapidly upon warming and pushes the liquid out of the needle before violent evapo­ ration starts. Such mechanisms lead to high standard devia­ tions and frequent runaway results (which seem inexpli­ cable and all too often the analyst assumes it was his fault). No statistical data are given here because such results are representative of the person rather than the techn ique. Maybe candidates for a job in GC could be evaluated by the stand­ ard deviation they achieve for such injections.

28

A 5. Syringe Needle Handling Minimizing Discrimination

Autosamplers

5.3.2. Slow Injection

Extreme Discrimination

Visual Observation of Evaporation

5.3.3. Cool Needle Injection

Classical autosamplers apply the filled needle method, a ba­ sically poor technique. Introduction of the needle is, how­ ever, rapid and highly reproducible, which eliminates the important weaknesses of manual filled needle injection. Autosamplers are, in fact, known to be considerably more reproducible than manual injection (often even when the latter involves the best needle handling techniques). Slow depression of the plunger has been recommended for splitless injection to reduce the vapor cloud to a size that could be temporarily stored in the vaporizing chamber. If injected at a rate enabling the vapors to be concurrently trans­ ferred into the column, large samples can be introduced into small chambers. Depending on the carrier gas flow rate and the volume of vapor created by a given volume of liquid, 1 JlL of sample would have to be injected within 5-20 s. This idea was a flop because it usually led to severe losses of high-boiling compounds inside the needle - it favors evaporation over ejection. n-C24 , for instance, was lost almost completely inside the needle (injector temperature, 250°C). Results from "dirty" samples were even worse, ow­ ing to the additional retentive power of the by-products ac­ cumulated at the point of sample evaporation in the needle. No results from slow injection are shown in Figure A 11, be­ cause they are highly dependent on whether the plunger was moved in small steps (each causing some violent evapora­ tion) or continuously and smoothly.

Wang et al. [71 observed evaporation from a 0.37 mm i.d. glass capillary heated in an oil bath, imitating a syringe nee­ dle. Water containing a red dye was introduced at a ve­ locity of 1 cm/s (corresponding approximately to slow injec­ tion). At 120°C, water was expelled from the capillary with a vigorous pop, transporting the dye. At 155 and 200 °C, the water evaporated at the rear. The dye was deposited near the center of the capillary and colorless water left the device. Plugs of liquid sporadically moved the dye from the rear to the center of the heated capillary. The red dye is a substitute for the high-boiling sample com­ ponents and showed how slow injection can cause totalloss of solute material. Cool (or cold) needle injection, in earlier times the most com­ monly used method of manual injection, afforded consider­ ably better results than manual filled needle injection - only ca. 40 % of n-C44 was lost (Figure A 11). The liquid seems to be eluted more rapidly and violently, moderately favoring non-discriminatory ejection. The reproducibility of results obtained by manual cool nee­ dle injection is often rather poor. Losses in the needle are highly dependent on the speed of the injection process, be­

29

5.3. Comparison of Needle Handling Techniques

Reproduction of Injection Speed

Results Depend on the Analyst

Relative Standard Deviations

cause the latter determines the needle temperature at the moment of sample introduction. If an easy-going and a nervous analyst both perform injec­ tions by this technique, the needle temperature at the mo­ ment of importance will be different and so will the results. With the slower procedure, the results are better, because ejection from the warmer needle is more violent. This ex­ plains why two persons injecting the same sample often obtain different results. The dependence of discrimination on the speed of injection is probably the main reason why an experienced analyst in­ jecting with a reproducible rhythm obtains the best re­ producibility. It might, however, happen even to him that the needle enters the septum with more difficulty than usual, or that the plunger does not move as freely as it should and he obtains runaway results. The relative standard deviations given in Table A2 were ob­ tained, from manual injections by an analyst who knew about the factors affecting reproducibility. There were, furthermore, no special problems, such as a new septum. Thus results could .easilv be worse. Needle volumes (7.1 cm needle, 0.9 Ill) were injected with an injector temperature of 350°C. Table A2 Relative standard deviations (%) for peak areas of n-alkanes normalized on n-C•• Split injections of needle volumes using different needle handling (n .. 8). (From [8]).

5.3.4. Hot Needle Injec­ tion

Injection method

C20

C32

C36

C44

Vaporizer cool needle hot needle solvent flush air plug On-column

1.7 1.0 2.1 2.9 0.7

12.5 1.2 4.7 3.0 0.9

11.0 1.1 5.7 3.0 0.7

10.6 1.1 7.6 3.4 0.8

Hot needle injection differs from the cool needle method in that the needle is pre-heated to the injector temperature be­ fore the plunger is depressed. The temperature of the nee­ dle approaches that of the surrounding injector within ca. 3 s [6]. Pre-heating for 3-5 s is, therefore, recommended. The plunger should be depressed as rapidly as possi­ ble, because a minimum amount of evaporation along the needle wall should eject as much of the sample as possible as a liquid. Explosive evaporation, particularly important in the rear of the needle, should expel most of the sample in the liquid state.

30

A 5. Syringe Needle Handling Minimizing Discrimination

Low Discrimination

Optimum Reproducibility

Speed of the Plunger

Hot needle injections resulted in the lowest discrimination of the methods tested. 23 % of the n-C44 was recovered by re-injection (needle rinse), which, although far from negligi­ ble, was the best that could be achieved for a needle volume and the given injector (temperature gradient to the septum cap). Reproducibility was far better than for the alternative needle handling techniques tested (Table A2). This is because of the reproducible needle temperature when the plunger is depressed. The speed of needle introduction is not impor­ tant. The results are, therefore, more easily reproduced by different persons. The only parameter found to contribute substantially to the standard deviation of the results was the speed with which the plunger was depressed. Optimum results cannot be obtained by use of syringes with distorted plungers which move only with difficulty. It has been found [8] that losses in the needle were smaller if the sample plug was withdrawn further back into the barrel than just out of the needle, obviously because the liquid en­ tered the needle at a higher speed when introduced with a run-ue. en

• •

<.>

z en

...." II

Z

J

z UI

<.>

5

~

S,

o-~

B

B

La..

2be).

1l1L

-'

'­

164

Progr. 5°/min 100°

Figure A12 Analysis of mustard oils and a related nitrile from radishes. 15 m x 0.3 mm i.d. glass capillary column coated with an 0.08 ~m film of Pluronic L 64 (a polyglycoll; AFID in the N­ mode... 1 ~L" of a dichloromethane solution injected splitless by the hot needle technique into an injector at 225°C. S, internal standard; B, degradation products. With cold on­ column injection, no degradation products were observed. Vaporizing injection required careful optimization to find the injector temperature providing an acceptable compromise between high discrimination/standard deviation and severe degradation. (From [9], see also [10)).

5.3. Comparison of Needle Handling Techniques

Minimum Degradation of Labile Solutes

Pre-Peaks

3 1

The hot needle method minimizes degradation of labile solutes inside the needle. At first this might seem sur­ prising, because "hot needle" is inevitably associated with pyrolysis or roasting in a frying pan. The opening statement was, however, the result of a study on the optimization of a method for injecting mustard oils and related compounds from radishes into a vaporizing injector (see Figure A12). These sulfur-containing compounds degrade rapidly on hot metal surfaces, such as hot syringe needles. Hot needle injections proved superior to the alternatives, be­ cause most sample material leaves the needle in the liquid phase. Material ejected as a solution in a volatile sol­ vent neither touches the hot needle wall, nor is it heated above the solvent boiling point. The concept of hot needle injection assumes that an empty syringe needle is introduced into the injector. This is a sim­ plification, because a layer of sample liquid remains on the needle wall after the sample plug has been withdrawn into the barrel of the syringe. In reality, the hot needle tech­ niqueinvolves a double injection: the sample left in the nee­ dle, 0.02-0.1 ul., evaporates from the needle on insertion into the injector. After 3-5 s, when the plunger is depressed, it is followed by the bulk of the sample. A small pre-peak should, therefore, be expected some 3-5 s before the main peak (Fig­ ure A13). Main peak from depressing the plunger

Pre-peak from inserting the needle

J ~~'--------" '---------' Needle pre-heating, e.g. 3-5 s

Figure A13 Pre-peaks arising from hot needle injection: isothermal run after split injection.

When is it Observed?

Although initial bands are always split, pre-peaks are sel­ dom actually observed. 1 If the peak is broader than about 7 s, the small pre-peak is contained by the main peak (see last peak in Figure A13). 2 If temperature-programming is used, the two bands are recombined by cold trapping.

32

A 5. Syringe Needle Handling Minimizing Discrimination 3

In splitless injection reconcentration is needed to focus the broad initial band. This also combines the band re­ sulting from needle introduction with that from the main injection. In isothermal runs, solvent effects are used. Thus peak splitting is observed solely in the early part of isothermal runs after split injection.

Countermeasures

5.3.5. Solvent Flush Injection

Air Plug to Avoid Mixing

Concept Requires Cool Needle ...

Most applications showing this kind of peak splitting involve volatile components for which discrimination is anyway no problem. Thus pre-heating of the needle can be reduced or might be totally unnecessary. Pre-peaks can also be prevented by picking up ca. 0.2 ilL of pure solvent after the sample. When the liquid is withdrawn into the barrel, it is this pure solvent that coats the needle wall. Some autosamplers provide the option of injecting in this way (e.g. AS800 from CE Instruments). There are three arguments favoring the use of solvent flush injection. 1 . The name "solvent flush" suggests that at the end of the injection the needle is rinsed with solvent, trans­ ferring into the vaporizing chamber what would other­ . wise remain on the needle wall. 2 The plug of pure solvent elongates the plunger' such that the whole sample is pushed through the nee­ dle. Just solvent is left in the needle when the plunger reaches the bottom, i.e. final evaporation from the nee­ dle involves solvent only. 3 Only the volume of sample measured on the barrel of the syringe is injected, enabling introduction with accu­ rate measurement. The plugs of sample and solvent must be separated by a small plug of air to prevent mixing of the two liquids (Fig­ ure A14). Such mixing would occur primarily as a result of turbulence when the liquid passes from the narrow bore needle into the wider barrel. Keeping the plug of air short (0.2 Ill) facilitates withdrawal of the correct volume of sam­

ple (small elastic gas volume).

Roeraade (11) showed that, despite the presence of the air

plug, rapid injection results in considerable mixing of the

liquids (some 15 % when a 1 III air plug separates 1 III plugs

of liquid): the film of sample in the air plug section carries

sample material into the flushing solvent. The sample mate­

rial was mixed with more than 1 III of a 31ll volume offlush­

ing solvent.

At first, the advantages of the solvent flush method seem beyond doubt, but a closer look revealed problems. Results remained puzzling [5,6). There are two options: 1 The solvent following the sample must touch the needle wall in order to rinse solute material from it,

5.3. Comparison of Needle Handling Techniques

33

Solvent flush injection Air plug Plunger

:

\ Flushing solvent

'SamPle

Air plug injection Sample Plunger

i

:

Air plug

Sandwich method Sample Plunger

( Flushing s"olvent

Figure A14 Arrangements of the plugs of sample. air. and flushing sol­ vent f~Jr three needle handling techniques.

but this is as difficult as cleaning a hot cooking plate with water. Wetting the needle wall with solvent presup­ poses that its temperature is reduced to the solvent boil­ ing point. 2 The sample does not evaporate during passage through the needle. Both require that the needle temperature is not far above the solvent boiling point; this calls for an injection performed as rapidly as possible - cool needle solvent flush injection.

...but Hot Needle Gives Better Results

Counterproductive Air Bubble

Volume of Flushing Solvent

Experiments with a mixture of n-alkanes (CWC44 ) in hexane produced the opposite result, however. When 1 ~L of sam­ ple and a needle volume (0.9 ~L) of solvent were injected, losses were ca. 40 % lower with the hot needle solvent flush method (injector at 250°C). The result suggests that the needle is too hot to be wetted even when injection is performed as rapidly as possible. Another puzzling result obtained with the same test sample was that the losses in the needle were nearly halved when there was no air bubble separating the sample from the flushing solvent - even though tests with a colored liq­ uid confirmed that then the two liquids were considerably mixed. This again casts doubt on the validity of the concept of the solvent flush method. If 1 ~L of sample and a needle volume of solvent were in­ jected into an injector at 250°C (a in Figure A15), nearly 40 % of the n-C40 and n-C44 was left inside the syringe needle ­ eventhough it was a hot needle solvent flush injection. The result in (b) shows that the losses were almost halved

34

A 5. Syringe Needle Handling Minimizing Discrimination peak oreo [%]

peokall!O[%] 100

50

to 12

16

22

28

peak area[%)

40

44

alkane C.

peak areo [%] 100

50

50

to 12

16

22

28

34

40

44

alkane C.

Figure A15 losses of solute material in the syringe needle. determined for the same mixture of ... alkanes as used for Figure A10. Split injection (20:11; injector at 250°C. lower curves: peak areas from the first injection normalized to ",C,o. Hatched area: solute materia. recovered from the syringe needle by r8-injection. Averages from five experiments. ' al Hot needle solvent flush injection of 1 III and a needle volume (0.9 Illl of solvent. bl As al. but with 2 III of flushing solvent. cl Hot needle injection of 2 III of sample without added solvent. dl As b), but with the order of the plugs reversed: the sample flushed the pure solvent out of the needle! (From [5]).

when 2 III of flushing solvent were injected behind the 1 ul, of sample. As a rather typical side effect, the increased injection volume resulted in increased discrimination by other mechanisms (the upper curve now being up to 30 % below the 100 % line), such that the analytical result (that of the first injection) hardly improved. Sample Flushing out Solvent

Any remaining esteem in which the method might have been held was lost as a consequence of the result shown in (dl: the losses in the needle were practically identical whether the 2 III of flushing solvent was behind (b) or ahead (d) of the sample, i.e. were not increased when the sample flushed the solvent out instead of vice versa. One might speculate that this (strong) improvement com­ pared with the hot needle injection of merely a needle vol­ ume resulted from cooling of the needle surface by the evaporation ofthe preceding solvent. The sample might have entered the needle more easily, to be violently ejected shortly afterwards. The sandwich injection method uses solvent ahead of the sample. Is this the reason?

Solvent Flush versus Hot Needle

FigureA15c shows that hot needle injection of a volume of sample equal to the total volume of liquid introduced by solvent flush injection (1.9 IlLI results in lower losses in­

5.3. Comparison of Needle Handling Techniques

35

side the needle than the solvent flush method (a). This is no longer surprising, bearing in mind that all the features of the solvent flush method had no effect, or even negative ef­ fects, and the order of solvent and sample plugs could be reversed without adverse consequences. In fact, hot needle injection was superior to hot needle solvent flush injection because the latter was still performed with an air plug. Rodriguez et al. [12] found that with an injector at 200°C losses in the needle were substantially greater when using the hot needle technique, but a 1 III injection was compared with solvent flush injection of twice as much liquid. Solvent Flush at Low Injector Temperatures?

It is possible that the solvent flush method is advantageous at low injector temperatures or for relatively high-boiling solvents: evaporation inside the needle might be of insufficient violence for effective hot needle injection; the solvent flush method is more likely to fulfil expecta­ tions, because the sample passes through the nee­ dle with less evaporation and the solvent might re­ ally have some flushing action. Try and check by the needle rinse method!

Reproducibility

Relative standard deviations of results from injections per­ formed by the classical cool needle solvent flush method were several times higher than those from the hot needle method (Table A2). This can be explained by the poorly reproducible needle temperature at the moment of injection and the (poorly understood) influence of the air bubble.

Poor Suitability for Splitless Injection

When evaluating the solvent flush technique, it is important to consider also other aspects. Splitless injection is used if high sensitivity is important. The volume of liquid which can be introduced without overfilling the injector liner is limited. If the solvent flush method is used, the flushing solvent takes away at least half of it (and of the sensitivity other­ wise achieved). Another drawback of the solvent flush method in splitless injection is the tendency of the vapor of the flushing sol­ vent to displace the sample vapor (which enters the in­ jector first) from the column entrance into the upper part of the injector. If transfer of the sample vapor to the column is incomplete, it is paradoxical to introduce the vapor of the flushing solvent into the column preferentially. If the injector is overfilled, it is, moreover, primarily the sample which is flushed through the septum purge.

Split Injection: Circumvented Dilution

When using split injection, the analyst is usually not inter­ ested in high sensitivity. Even the opposite may apply - he wants to inject as little as possible. For reasons discussed in Section A9, plunger-in-needle syringes should be avoided for samples tending to evaporate inside the needle and sam­ ples containing high-boiling components. The use of 10 III

36

A 5. Syringe Needle Handling Minimizing Discrimination syringes with the hot needle method implies, however, work­ ing with sample volumes corresponding at least to the nee­ dle volume and preferably to "0.5 ~L" or even" 1 ~L", i.e. ca. 1.5-2 ~L (Section A6). The solvent flush technique enables injection of sample vol­ umes as small as 0.2 III (limited only by the readability of the scale on the syringe) behind a larger plug of flushing solvent. Instead of adding more solvent to the sample in the vial, the additional solvent is picked up with the syringe.

High Boiling Flushing Solvents

High boiling solvents provide more efficient flushing of the needle, because they wet the needle surface more eas­ ily. The possibility of using these depends, however, on the sample and the solutes to be separated from the solvent. Recondensation in the column inlet might, furthermore, se­ verely affect the split ratio. Distinction must be made between high-boiling solvents which are still expelled from the syringe needle and even higher-boiling solvents which essentially remain therein. The former rinse the solute material from the, needle wall into the injector, whereas the latter have a flushing effect only if their volume exceeds that of the nee­ dle.

n-Octsne Compared with Pentane

As shown in Figure A 16, discrimination arising from losses' in the syringe needle was nearly halved by use of n-octane (b.p. 126 ec) rather than pentane (b.p, 36 "C) as a flushing solvent behind a sample dissolved in hexane (injector tem­ perature, 250°C; split injection, needle volume of flushing solvent).

Problems with High-Boiling Sample Solvents

Schomburg et al. [13] advocated the use of high-boiling sol­ vents to reduce or eliminate discrimination. This advice was rarely followed, presumably for practical reasons: high-boil­ ing solvents are seldom sufficiently clean and of limited suit­ ability for sample preparation, because sample reconcen­ tration by solvent evaporation is impossible. When used solely as a flushing solvent, however, no reconcentration is needed. peak area [%J

peak area [% J

lool---aZZU';'7;"

50

50

to 12

16

22

28

34

40

alkane

44

e.

Figure A16 Cool needle solvent flush injections of n-alkanes in hexane (1 ~LI, comparing n-pentane (al and n-octane (bl as flushing solvents (needle volumes, 0.9 ~LI. (From [5]).

5.3. Comparison of Needle Handling Techniques

37

5.3.6. Ai' Plug Injection

It is a severe drawback of the solvent flush method that al­ most 1 ul, of solvent (and the impurities therein) must be injected merely to reduce losses inside the needle. If the plug of flushing solvent is just an elongation of the plunger into the needle, an obvious idea would be to use a plug of air instead of solvent (Figure A 14). Air has no rinsing effect, but rinsing of a hot needle wall with solvent seems, anyway, illusory.

Method

Air plug injection is performed by starting to suck sample liquid into the syringe with the plunger at a position of ca. 1 ul., i.e. with some 2 ul, of air between the plunger and the sample. The plunger is not moved up and down, as is other­ wise normal. The needle is withdrawn from the liquid when the rear meniscus reaches the required mark. The liquid must be withdrawn from the needle into the bar­ rel of the syringe, because otherwise it is easily lost. Upon warming, the air plug (saturated with solvent vapor) expands. The plunger is depressed as rapidly as possible to mini­ mize evaporation during passage of the sample through the syringe needle.

Severe Discrimination

The discrimination determined with the air plug method was greater than that with cool needle injection. After in­ jection, some 0.1-0.2 ul, of liquid were usually observed to cling to the (initially dry) plunger, suggesting that the air plug was not sufficiently effective at moving the sample. Evapo­ ration during passage through the syringe needle might cre­ ate such pressure that the air plug is compressed and liquid driven back to the plunger.

5.3.7. Sandwich Injec­ tion

The sandwich method is well known in packed column GC. The sample plug is located between two plugs of solvent, each separated from the sample plug by a short plug of air (Figure A 14). On injection, a plug of solvent passes through the needle first, maybe to cool the needle surface, fol­ lowed by the sample. The second plug of solvent is supposed to serve as an elongation of the plunger into the needle and for flushing the needle.

Large Volume of Liquid

Adaptation of the sandwich method to capillary GC is hin­ dered by the inevitably large volume of liquid which must, as a consequence, be injected. The two plugs of solvent usu­ ally contribute 2 ut, to the total sample volume, which is al­ ready nearthe maximum compatible with splitless injection. Considering the volume of the vapor cloud, larger sample volumes are tolerable for split injection. It is, however, gen­ erally observed that other deviations are then accentuated.

5.4. Heating the Needle after Injection?

Is it expedient to leave the syringe needle in the injec­ tor for a certain time after the plunger has been depressed to improve the elution of the solute material from the nee­

38

A 5. Syringe Needle Handling Minimizing Discrimination

dle? The answers generally given range from "important" to "useless" or even "bad". No Improvement for Volatile Solvents...

The discrimination curves in Figure A 11 were little affected. When the filled needle technique was used, losses of solute material were slightly reduced when the needle was with­ drawn after waiting for 1 s, but no such differences were observed for the cool or the hot needle method. Cool needle injection resulted in the same relatively strong discrimination even when the needle was left in the injector throughout the whole run, giving the high-boiling solutes more than 15 min to be eluted and recombined with the bulk of the solute material by cold trapping .

...but for High-Boiling Solvents

With some higher boiling solvents and/or lower injector tem­ peratures, however, noticeably reduced losses of high-boil­ ing solute material were observed when the needle was heated for 1-3 s after injection. The evaporation process ob­ viously takes longer.

Solvent Vapor as Carrier

Solutes cannot leave the needle by expansion of their vol­ ume during evaporation. 10 ng of solute material, for in­ stance, forms a vapor cloud which is a thousand times smaller than the internal volume of the needle. Trou- ' blesome solutes, furthermore, evaporate to a small extent only (low vapor pressure at the needle temperature). Solute vapor must, therefore, be flushed from the needle by other vapor serving as a carrier.

Until the End of Solvent Evaporation

The most abundant carrier available is vaporized solvent ­ even if merely 0.1 III of solvent evaporates behind the de­ posited solute, it creates a volume of vapor which flushes the needle volume ca. 10-50 times. When solvent evapo­ ration in the needle ceases, however, because the sol­ vent is exhausted or because some of it remains in the cool rear of the needle, elution of solute material essentially ends and there is no longer any reason to keep the needle in the injector.

5.5. Effect of Injecting Air

Most injection methods proposed involve withdrawal of the sample liquid into the barrel of the syringe, which means sucking up several microliters of air and injecting it with the sample. A further (maybe even larger) amount of air is dissolved in the sample liquid. Surprisingly little has been done to check possible negative effects of this air.

Concentration in the Carrier Gas

Mixed with the carrier gas inside the injector, the air injected together with a liquid sample reaches a concentration of 0.2­ 0.5 %. This is far more than generally considered tolerable in the carrier gas (a few ppm). During the first part of the analysis, the sample is in contact with maybe 10 ml of carrier gas. Injected in splitless mode the air carried into the column with the sample builds up an

5.5. Effect of Injecting Air

39

average concentration of about 200-500 ppm, about 100 times more than in high-grade carrier gas. 2 III of air con­ tains some 600 ng oxygen, i.e. theoretically more than enough to oxidize the components of interest. Is it reasonable to invest into traps in the carrier gas supply line to remove the last traces of oxygen and humidity when the sample introduces 10-100 times more? Headspace Analysis

Classical headspace analysis introduces even 100-1000 times more air into the injector. As injection usually involves splitting, 10-100 III of air reaches the column, representing concentrations in the percentage range. Fortunately tem­ peratures are usually low - injector temperatures seldom need to be above 100°C and column temperatures during injection are even lower.

5.5. 1. Concern. Regard­ ing the Column

Air can damage the stationary phase of the column. Every beginner knows this and is careful about selecting a suffi­ ciently clean carrier gas. Again, little has been done to deter­ mine the real requirements. Probably large sums of money are paid for a purity which is of no use.

Carbowax

Carbowax or FFAP columns are sensitive to air even at mod­ erately high temperatures. When filled with air, e.g. while lying around on an instrument, a few hours at 40-50 °C, par­ ticularly in the light, may be sufficient to cause extremely high bleed. This bleed indicates the beginning of a self-sup­ porting oxidation process that cannot usually be stopped; there is then no way to save the column. In contrast with this it should be stated that nobody has reported that headspace injections at around 100°C have demeged a Carbowax column. Nor have injections of liquid samples at higher temperatures. This does not rule out accelerated aging, but there is no experimental evidence identifying the various factors causing normal column dete­ rioration.

Silicone Stationary Phases

Columns with silicone stationary phases of low polarity have been used with air as carrier gas (actually non-purified, humid air from a compressor) for several months (for test­ ing purposes - air is as poorly suited as nitrogen for capil­ lary GC). The column was heated to 350°C over dozens of hours. The bleed was higher and the stationary phase be­ came lacquered somewhat sooner than usual, but no spec­ tacular column degradation was observed.

5.5.2. Detector.

Detectors like ECD or MS are sensitive to air. In fact, the per­ formance of ECDs deteriorated somewhat after frequent use for headspace analysis. particularly when the split ratio was low. Nevertheless, no special precautions usually are taken.

5.5.3. Oxidized Sample

In contrast with the column, little attention has been paid to oxidation of the sample. In fact, triglyceride analysis with

40

A 5. Syringe Needle Handling Minimizing Discrimination air as carrier gas was not hindered by column degradation, but by oxidation of unsaturated fatty acids. Trilinoleate was totally lost.

Trilinoleete

The experimental set-up enabled switching between air and hydrogen as carrier gas. Most interestingly, little trilinoleate was lost when the carrier gas consisted of air during analy­ sis up to 350°C as long as hydrogen was used during injec­ tion, but most trilinoleate was degraded when air was the carrier gas during the 1-2 min of injection and solvent evaporation. It was introduced on-column as a solution in tridecane at a column temperature of 200°C. These are gen­ tle conditions compared with the alternative, split injection at an injector temperature approaching 400 °C.

Dependence on Concentra­ Oxidation is a radical chain reaction. The solvent vapors tion supported this reaction and caused the mixture to "catch fire". During chromatography, the concentration ofthe readily oxidizable material is too low (silicone resists oxidation well), except when a sufficiently large amount of solute material was injected. Check for Solute Losses

This experiment generated more questions than it answered. The real danger seemed to be loss of solute material by oxi- . dation in a vapor phase supporting radical reactions. The

vaporizing chamber of hot injectors is a particularly peak area ["10] 100

~~~

50

50

to 12

t6

22

28

40

44

alkare CJ peak area ["!oj

peak area ["!o) lOOt-<>~~_~_

lOO.--_-'Qo___

50

50

to

'2

16

22

28

40

44

alkane CJ

Figure A17 Comperison of solute losses in the needle for ..0 ~L" (e end blend ..2 ~L" (c end d) injections using the cool needle (e end c) end hot needle (b end dl techniques. Injector et 250°C. Differ­ ences between needle hendling methods become smell if the semple volume exceeds ebout ••1 ~L". (From ref. [5]).

5.5. Effect of Injecting Air

4 1

dangerous zone - together with the air there is a high con­ centration of solvent and solute material and the tempera­ ture is high. Solvents such as isoalkanes support radical re­ actions more than others (e.g. toluene), i.e. the sample sol­ vent could well be even more important than the tempera­ ture. Peroxides?

No such effect is mentioned in literature - either because it never occurs or, more likely, because nobody has investi­ gated it. There is a related subject that has not been deeply investigated: peroxides in solvents. Diethyl ether and tetra­ hydrofuran often contain high concentrations of peroxides. Alkanes contain far less, but concentrations in hexane might still approach 1 ppm. Considering the nature of the chain reaction, this could be sufficient to severely affect the sam­ ple.

6. Dependence of Discrimination on Sample Volume The results discussed above were obtained by injecting nee­ dle volumes of sample. They are pessimistic because losses and discrimination are considerably reduced upon in­ jection of larger sample volumes. 6.1. Experimental Re­ sults

Figure A17 compares "0 Ill" and "2 Ill" cool and hot nee­ dle split injections of ClO-C44 n-alkanes in hexane (71 mm needle, 0.9 III internal volume). When only a needle volume was injected, the loss of n-C22 inside the needle by cool nee­ dle injection (ca. 50 %) was at least four times higher than by hot needle injection. For the C3o-C 44 n-alkanes, the difference amounted to a factor of two.

Losses Decrease with Increasing Volume

As the sample volume was roughly doubled from needle volume to "1 u],", the relative loss of solute material inside the syringe needle decreased substantially. For hot needle injection, loss of n-C22 decreased from 13 % ("0 Ill", Figure A17b) to 3 % ("1 Ill", Figure A15c), that of n-C34 from 42 to 12 %, and that of n-C40 from 50 to 28 %. The relative loss of n­ C44 remained about constant, however. As shown by the results for "2 Ill" hot needle injection in Figure A17d, increasing the sample volume from "1 j.lL" to "2 u]," no longer resulted in significant improve­ ment.

Difference between Cool and Hot Needle

Figure A 17 also shows that differences between cool and hot needle injection are no longer significant when "2 j.lL" are injected. Cooling of the needle surface by pas­

42

A 6. Dependence of Discrimination on Sample Volume sage of the liquid might be so effective that temperature dif­ ferences at the beginning of the injection are no longer im­ portant.

Sample Volumes of "0.5"-"1 j1L"

The greatest reduction of losses in the needle, i.e. of dis­ crimination, is achieved by increasing the sample volume from "0" to "0.5 IiL". An increase to "1 IiL" provides a fur­ ther significant improvement. Beyond "1 IiL", the gain is small and is often overridden by other, rapidly growing prob­ lems. In Figure A17, involving split injection, discrimination resulting from mechanisms other than losses inside the needle reached 20-30 % for "2 IiL" sample volumes; for hot needle injection, this almost offset the improvement brought about by reduction of losses in the needle.

6.2. Discussion of Mechanism

Semiquantitative estimates enable us to obtain further infor­ mation about what happens inside the syringe needle. As a working model, it is assumed that a "1 IiL" injection com­ prises two steps, viz. mechanical displacement of a 1 IiL vol­ ume through the needle and introduction of a similar vol­ ume into the needle followed by ejection or evaporation therefrom.

Calculated Losses for each Step

The losses of solute material during passage through the needle (i.e. forthe first microliter) can be estimated from the results obtained by solvent flush injection of a 1 IiL volume of sample ahead of a needle volume (ca. 1 IiL) of solvent (Figure A 15a). The losses during the second step are com­ pared with those observed on injection of a needle volume (Figure A17). To facilitate the comparison, relative losses are transformed into absolute amounts. In Table A3 they are expressed in terms of sample volumes containing a corresponding amount of solute. The sum of the first two columns, the losses expected from the two steps, far exceed those actually ob­ served for the" 1 IiL" injection (last column).

Smaller Loss for Second Step

There is no reason to assume that losses during the first step (passage through the syringe needle) should be different, depending upon whether the sample is followed by pure solvent (solvent flush injection) or by more sample ("1 IiL" injection). The difference between observed and expected losses can, therefore, be explained only in terms of smaller losses during the second step, i.e. the elution from the needle after the plunger reached the bottom. According to the data, up to n-C34 the second step proceeds without any losses. In absolute terms, losses during the "1 IiL" hot needle injection are equal to (actually insignificantly lower than) those during the "0 u]," injection.

Temporary Cooling of the Needle Surface

The small losses during the second step of injection could be explained by temporary cooling of the internal surface of the needle by partial evaporation of the most advanced sam­

6.2. Discussion of Mechanism Table A3 Losses of solute material, expressed as sample volumes amounts.

(~L)

43

containing the corresponding

n-Alkane

1 ~L = ,,0 ~L" Hot needle

1 ~L + 1 ~L Solvent flush

2 ~L = ,,1 ~L" Expected

Hot needle Observed

C22 C28 C34 C40 C44

0.10 0.22 0.42 0.50 0.45

0.10 0.20 0.22 0.35 0.35

0.20 0.40 0.64 0.85 0.80

0.08 0.20 0.22 0.55 0.45

pie material. Such cooling could enable the following liquid to pass through or enter the needle without much evapora­ tion; a moment later, re-heating of the needle surface results in violent expulsion of most of the sample in the liquid phase. Arguments in Favor of Solvent ahead of Sample

If losses occur primarily during entrance of the first liquid into the needle, this favors positioning of solvent ahead of the sample plug and is evidence in favor of the sandwich rather than the solvent flush method. The same logic might also explain why "solvent flush" injection with the sequence of solvent and sample plug reversed (Figure A 15d) produced results similar to those obtained by the regular solvent flush method.

6.3. Conclusions

The experiments suggest two conclusions. 1 Unless evaporation inside the needle can be suppressed, at least "1 ~L" should be injected; the gain from increas­ ing to "2 ~L" is small. 2 As losses occur primarily when the first of the sample liquid enters the needle, solvent ahead of the sample liquid (i.e. drawn into the syringe after the sample) might be more useful than solvent behind the sample ("sol­ vent flush"). As injection of more than ca. "2 ~L" of liquid should gener­ ally be avoided, however, the possibilities of co-inject­ ing solvent are severely restricted.

7. Solvent and Solutes The solvent influences the evaporation process inside the syringe. Should it be volatile or high boiling? At this point we do not consider solvents boiling so high that evapora­ tion in the needle can be suppressed.

44

A 7. Solvent and Solutes

7.1. Volatility of the Solvent

As. shown in Table A4, hot needle injection produced re­ sults with no significant difference whether the sample was dissolved in n-pentane (b.p. 36°C), n-heptane (b.p. 99°C), or n-nonane (b.p. 151°C). With the filled needle technique, however, the higher boiling solvent clearly resulted in less discrimination (although still morethan hot needle injection). Table A4 Relative peak areas of alkanes, normalized with respect to n-nonane or n-dodecane (%1, comparing various solvents; hot needle or filled needle injections of needle volumes; injec­ tor at 300 °C. (From ref. [6]).

Hot Needle Solvent C20 n-Pentane n-Heptane n-Nonane

C44

80.2 83.6 84.1

67.9 75.1 72.3

Sample component C36 C32

C44

99.7 102.6 101.3

Filled Needle Solvent C20 n-Pentane n-Heptane n-Nonane

Sample component C32 C36

94.3 96.4 100.2

84.0 85.1 82.7

68.6 75.1 97.1

62.4 61.8 72.3

21.2 26.1 53.2

Tentative Explanation

Losses in the syringe needle depend on the ratio of evapo­ rated and expelled sample. Volatile solvents evaporate more readily and should be inferior in this respect; they do, how­ ever, also evaporate more violently, with a tendency to expel a larger proportion of the sample. Apparently, the two fac­ tors just about balance each other such that there is surpris­ ingly little difference between solvents of different volatility. Experience from practice largely confirms this, but there were also exceptions.

7.2. Type of Solute

All experiments discussed above were performed with alkanes as solutes. Other types of solute of comparable volatility are lost in different proportions, surprisingly to an extent which is often considerably lower. Losses of triglycerides (tri-C,o to tri-C'B) were between 20 and 28 %, in another instrument only 12 to 17 % (injection of needle vol­ umes; hot needle method at an injector temperature of 400 °C [14]). These losses amounted to hardly half those expected for n-alkanes of similar boiling point. No systematic comparison has been made, nor are there ex­ planations or conclusions.

7.3. Adsorption in the Syringe Needle

45

7.3. Adsorption in the Syringe Needle

If peaks are too small and the possibility of adsorption in the column is excluded, the injector is commonly blamed for the losses. Often this is true, but the losses arise from adsorp­ tion inside the syringe needle more often than is recognized. The sample comes into closer contact with the needle wall than, e.g., the wall of the injector liner.

Influence of Solvent Polarity - a Case Story

Le Bel and Williams [151 thoroughly studied a case which is probably relevant to a wide range of analyses. They encoun­ tered problems in the quantitative analysis of dimethoete and beta-phosphamidon (two pesticides) when injecting solutions in hexane by the solvent flush method. When a few nanograms ofthe compounds were introduced, only 64 and 43 % were recovered at the detector. The losses de­ creased when larger amounts were injected - behavior typical of adsorption.

Needle Rinse with Hexane and Acetone

After injection of a hexane solution, a needle rinse injec­ tion with hexane revealed no solute material from the need_e. When the needle rinse injection was repeated with acetone, however, using the same or a different column from that used for the sample, the missing solute material was recovered (Figure A18). When the sample in hexane was followed by acetone using another, clean syringe, no solute material was detected. This tells us that needle rinse analy­ sis produces a false negative result ifthe solvent is unable to desorb the solute material from the wall (or a layer of con­ taminants on it).

...

c

8

Ul

z

A

0

e,

...a: Ul

a:

...u ...w... 0

0

'"

~

10

5

o

10

5

RETENTION TIME

0

10

0

(min)

Figure A1S 4 ng dimethoate and S ng beta-phosphamidon in hexane in­ jected on to a packed column by the solvent flush method. A: hexane as flushing solvent; only about half of the solute material is detected. B: needle rinse injection with hexane does not recover the missing material. C: subsequent needle rinse injection with acetone. (From Le Beland Williama(5)).

46

A 7. Solvent and Solutes

Sample Solvent Avoiding Adsorption

This example demonstrates the importance of adsorption inside the syringe and choosing an appropriate solvent. This choice is, however, often restricted by sample preparation. LeBel and Williams solved their problem by a solvent flush injection with acetone as flushing solvent behind the sam­ ple still in hexane.

Adsorption Suppressors ­ Another Case Story

Brotel/[16] described problems with the analysis of tertiary amines (packed column GC).The pentafluoro ethers of pen­ tazocine and ketobemidone were almost completely lost when injected after the syringe had been washed with metha­ nol. He assumed the source of the problem was adsorption of the compounds on the glass surface of the syringe barrel.

Silylating the Syringe?

Silylation of the glass barrel resulted in only slight improve­ ment. Numerous other analysts attempted to deactivate the syringe, but it seems that it has never been successful. The efficiency with which a syringe can be silylated is, how­ ever, anyway limited, because syringes should not be heated above 60-80 "C, Deactivation would, furthermore, be effec­ tive for the glass barrel only, while the needle is the more probable site of adsorption or degradation.

Sandwich Injection

losses were prevented by adding a large excess of an ad~ sorption suppressor, desipramine. The sandwich injection method was used with 2 III plugs of n-heptane containing 0.2 % desipramine on each side of a 5 III plug of sample (packed column GCI).

Injection Speed

Brotell also observed that without an adsorption surpressor, rapid and slow (10 s) injection resulted in higher losses than injection performed in ca. 4 s (0.5 s/Ill). When the adsorption suppressor was used for non-adsorptive solutes, no such effects were noted. He cites other authors who made similar observations. It remained unclear, however, whether the effects resulted from the syringe, the injector, or even the detector.

7.4. "Memory Effects" Arising from the Syringe

Some solutes are strongly adsorbed by or otherwise retained inside the syringe. This not only causes losses of solute ma­ terial for the on-going analysis, but is also responsible for "memory effects" and unsatisfactory blanks, because part of the lost material is likely to be transferred to subsequent injections. Horning's group [17] reported an analysis in which a set of syringes had to be used for different concentra­ tion ranges, because the adsorbed material could not be sat­ isfactorily removed. We had a similar experience with triphenylphosphine oxide or 2-oxazolidinone.

Misleading Blanks

"Memory effects" create the danger of false-positive r. suits. Blank tests are misleading when "memory effects" are absent or negligibly small upon injection of pure solvent, but strong for the real sample injected subsequently. Sam­

7.4. "Memory Effects" Arising from the Syringe

47

pie material obviously displaces adsorbed material better than does the solvent. This is related to the effect mentioned above, i.e. needle rinse with hexane did not re­ veal loss inside the needle whereas acetone caused the ma­ terial to be transferred. Addition of a few percent of a polar solvent or a component with adsorptive functional groups to the sample might help the problem. Needle Attachment as a Cause of "Memory Effects"

Solute material may diffuse into dead volumes or fittings in the zone in which either fixed needles are fitted to the barrel of the syringe with cement or removable needles are sealed with plastic material. The extent of such diffusion depends on contact time - large amounts of solute material can dif­ fuse into the needle attachment if the syringe still contains sample when put aside after an injection. To overcome such problems, it is recommended that sol­ vent be sucked into the syringe and positioned in the region of needle attachment immediately after an injec­ tion. In this manner, solute material diffuses out of, rather than into, dead volumes or plastic components.

8. Injector Temperature 8.1. Imposed Tempera­ ture

The optimum injector temperature is determined by many factors. Here we restrict ourselves to aspects concerning transfer from the needle and to samples and conditions for which evaporation in the syringe needle cannot be prevented. By "imposed" injector temperature we mean the ther­ mostatted temperature given in the readout of the instru­ ment. As injectors are cooler at the top (septum cap) and toward the oven, real temperatures can differ substantially, particularly in the region surrounding the needle.

High Temperatures Substantially Reduce Discrimination ...

Table A5 shows the influence of the imposed injector tem­ perature on discrimination of the previously discussed mix­ ture of C9-C44 n-alkanes (Carlo Erba 4150 instrument). Dis­ crimination is strongly reduced when the injector tempera­ ture is increased.

... but Do Not Eliminate It

The results also show, on the other hand, that substantial discrimination remains even when extremely high injector temperatures are used. Even at 400°C, 6 % of n-C26 was lost compared with n-Cg (not shown in Table A5), and 23 % of n­ C44 • Discrimination against n-C26 was not much greater at only 275°C (8 %1. whereas that against n-C44 increased to 72 %. The data suggest that for each compound there is an upper temperature limit beyond which discrimination cannot be further reduced.

48

A 8. Injector Temperature Table A5

Influence of the imposed injector temperature on discrimi­

netion egainst higher boiling ... elkenes. Hot needle split in­

jection of needle volumes (0.9 Ill). Peek erees normelized to

n-C. (= 100 %1. (From ref. [6]).

Injector temp. [OC] 250 275 300 350 400

Relative peak areas C32 C36 C20 100 102.4 101.2 102.7 100.5

65.9 85.9 85.8 86.0 86.2

41.6 72.1 83.6 83.5 85.1

C44 14.2 27.6 67.9 75.0 77.2

Effect on Reproducibility

The effect on reproducibility of increasing the injector tem­ perature is even more pronounced. The reproducibility of the results might be largely determined by the reproduc­ ibility of discriminative losses in the syringe needle. If they are reduced by increasing the injector temperature, they also contribute less to the standard deviation of the results. Increasing the injector temperature also reduces the impor­ tance of accidental influences, e.g. air bubbles or sites behaving like boiling stones, on evaporation process in the needle.

Some Data

As shown in Table A6, the relative standard deviation of normalized n-C44 peak areas was reduced from 50 % at 250 °C to 7.6 % at 300 °C and to 1 % at 350 °C. Probably more sur­ prising than that, high injector temperatures reduced rela­ tive standard deviations even for relatively volatile solutes, such as n-C20, indicating that the evaporation characteris­ tics of the sample as a whole and of the solvent in particu­ lar are just as important as the volatilization of the individual solutes.

Conclusion

The injector temperature should be set as high as the sample components tolerate. Even peak areas of rather volatile solutes are optimally reproduced only at high injec­ tor temperatures. n-C 20 , for instance, is analyzed more reproducibly at an injector temperature of at least 300 °C. The injection of samples in high-boiling solvents is an ex­ ception to this because evaporation inside the needle can be avoided when the injector temperature is kept low - pre­ supposing that this also suits the components to be analyzed.

8.2. Temperature Gradi­ ent Towards the Septum

The syringe needle must either remain cool enough to pre­ vent sample evaporation or be as hot as possible to approach complete transfer even of relatively high-boiling solutes. The former requires an injector head as cool as possible, the lat­ ter a high temperature along the entire length of the needle, including the rear part close to the barrel.

8.2. Temperature Gradient Towards the Septum

49

Table A6

Relative standard deviations obtained by hot needle split in­

jection of needle volumes at different injector temperatures,

calculated from peak areas normalized relative to n-C•. (From

ref. [6]).

Injector temp.

Relative standard deviation [%]

lOCI

C20

C32

C36

C44

250

4.6 2.8 2.6 0.9

18

5.6 4.4

27 9.3 6.6

50 20

1.2

1.1

1.1

0.2

0.4

0.8

0.8

275 300 350 400

B.2. 1. Critical Rear of

Needle

7.6

The syringe needle is a dead volume. A higher vapor pres­ sure is, therefore, needed to elute the solute material from the rear of the needle than is required for evaporation inside the injector. If sample evaporation occurs inside the needle,

the temperature of the region of the injector respon­ sible for heating the rear part of the needle, hence of the injector head (Figure A19), is particularly critical. Seen from this angle, the temperature of the top of the injector should even exceed that of its center. In reality, profiles are the other way around, however.

1

Syringe

Septum

Temperature determining the elution from the syringe needle

~

Dead volume

Temperature determining the evaporation of the sample

Rinsed volume

Column

II

Figure A19 Temperatures at the top and in the center of the injector determine different processes.

Cooler Septa

The call for a hot injector head conflicts with restrictions concerning the septum. Essentially consisting of silicone rubber, septa become soft at temperatures above 250-300 °C.

50

A 8. Injector Temperature For this reason, earlier instruments had septum caps equipped with cooling shells to accentuate the temperature drop towards the septum.

8.2.2. Actual Tempera­ ture Profiles

The actual temperature in different regions of the injector depends on the construction, in particular on the design of the heating block. The injector body, made of stainless steel, a rather poor thermal conductor, is seated in a heating block which contains the heating cartridges and the thermocouple serving for thermoregulation (Figure A20).

Syringe

Healing bloc Healing cartridge

Figure A20

Typical heating system for vaporizing injectors.

Thermocouple

The thermocouple also measures the "injector tempera­ ture" given on the display. It must be positioned close to the heating cartridges to prevent overheating and oscillat­ ing temperatures, but this has two consequences: 1 On heating the injector, the vaporizing chamber and par­ ticularly the injector head reach their temperatures only substantially after the readout shows the set value. Hence more time is needed than is suggested by the display. 2 The readout gives a higher temperature than is reached by the parts projecting from the heating block.

Two Examples

Hewlett-Packard [181 reported the temperature profile along the axis of the vaporizing injector of their 5890 gas chroma­ tograph for a setting of 350 DC. The septum cap was at merely 140 DC, and only the needle tip reached into the zone thermo­ statted at the temperature indicated by the display (Figure E11). This injector was designed for the fast autosampler.

8.2. Temperature Gradient Towards the Septum

5 1

Figure A21 shows analogous profiles through the injector with, probably, the most constant temperature (CE In­ struments, 8000 series to TRACE). the heating block also set at 350°C. The injector temperature is close to the regulated value for a length of over 5 cm. This is achieved by use of a heating block which is as long as the injector body. 150

200

250

300

350

top heated

E oS

top unheated

E 20 :J

C. CI)

(/J

E 40

g

CI)

0

c:

nI

iii 60

0

80

100

150

200

250

300

350

Measured temperature 1°C)

Figure A21 Temperature profile through the vaporizing injector of the CE Instrument thermoBtatted at 350°C; temperatures corre­ spond to the positions in the injector drawn at the right. The top region of the injector can be heated by an extension of the heating block and additional insulation.

Variable Heating of the Injector Head

The temperature of the injector head can be varied. If the septum cap is exposed to ambient air, the septum tempera­ ture is at ca. 220°C (imposed injector temperature, 350 "C). Alternatively, the heating block can be extended to the top of the injector and the septum cap insulated, which increases the septum temperature to nearly 300°C. This de­ sign maintains the septum below 300 °C when the injector is thermostatted at an extremely high temperature, whereas a rather straight temperature profile can be achieved at lower settings.

8.2.3. Effect on Discrimi­ nation

The effect on discrimination of various temperature drops towards the septum resulting from losses in the nee­ dle was studied with an injector heated by an aluminum heat­ ing block only 3 cm high and positioned some 5 mm above

52

A 8. Injector Temperature

the bottom ofthe vaporizing chamber [191.The temperature of the top of the injector was varied by the following means: Test Configurations

A)

the injector body above the heating block exposed to ambient air without any insulation; B) insulation with a PTFE ring, 4 cm high and 8 mm thick, but the injector head still exposed to ambient air; C) the whole injector, including the septum cap, packed in a large amount of glass wool; D) the top of the injector heated to the injector tempera­ ture by means of heating tape. On setting the injector temperature to 300°C, the profiles given in Figure A22 were obtained. Because the tempera­ ture inside the injector could not be determined for configu­ rations (C) and (D), only that of the septum is given.

300

__0

_ ------:---­

200

100

........--

'---

.

i"

,--~-----.------~

2

3

5

7.5

[em] from septum Figure AZZ Temperature gradients in a vaporizing injector for which dis­ crimination curves are shown in Figure AZ3. Positions meas­ ured from the septum cap downwards into the vaporizing chamber as far as the distance reached bV the svringe nee­ dle (71 mm). The heating block, thermostatted at 300 °e, was located between 6.5 and 9.5 cm below the septum. (From ref. (19]).

Resulting Discrimination Curves

Discrimination was determined by means of a test mixture containing equal amounts of CWC44 n-alkanes in hexane. Needle volumes were injected by the split technique, the hot needle method, and a 71 mm needle. The mean relative peak areas obtained are shown in Figure A23. Needle rinse injections confirmed that discrimination was almost exclusively a result of losses inside the needle. Temperature profile A with a septum at about 80°C resulted in 27 % loss for n-Cn and a loss of 90 % for ,..C44 . Profile

8.2. Temperature Gradient Towards the Septum

53

B (septum at 157°C) led to greatly improved transfer from the needle. With the injector (and the syringe needle) ther­ mostatted at 300°C over its whole length (D), merely some 10 % of n-C28 and 20 % of n-C44 were lost. relative peak area 100

80

B

60

40

20 A

() 12

16

22

28

34

40 " alkane

C.

Figure A23

Peak areas normalized relative to n-C,o' obtained bV hot nee­

dle split injection into an injector set at 300°C; results for

the temperature profiles A-D in the upper part of the injec­

tor according to Figure A22). (From ref. [19]).

Independent of Needle Length

With cool needle injection, differences were even more pro­ nounced than shown for the hot needle method (19). losses inside the needle were similar, irrespective of whether the length of the syringe needle was 38,51, or 71 mm (con­ stant injection volume of 0.9 Ill). This confirms that most of the losses occurred in the rear, i.e. the part heated inside the injector cap.

8.2.4. Quantitative Results Differing from One Injector to Another

The reader might have noticed that discrimination against n-C44 differed considerably between Figure A 17 and Table A5, although conditions, including the nominal injector tem­ perature (250°C), were identical. The two injectors (of iden­ tical geometry, Carlo Erba/CE Instruments 2000 and 4000 Series) were heated by differently shaped heating blocks. The 85 % loss (Table A5) was obtained with a block only about 3 cm in height, positioned near the bottom of the vaporizing chamber. With the extended heating block, the loss was reduced to 50 %.

Difference between Injectors

Some heating blocks have a height of merely 3-4 cm whereas others are as long as the injector body, minimizing tempera­ ture gradients towards the top and the bottom of the injec­ tor. This is one reason why injectors from different manufac­ turers or in instruments of different series will seldom have

54

A 8. Injector Temperature

the same real temperature - no surprise that they pro­ duce different results!

8.2.5. Conclusions

2

Losses inside the needle can be substantially re­ duced by thoroughly heating the injector up to the septum. Relative standard deviations are lowered even more dramatically. Even with an optimally heated injector, losses inside the needle cannot be completely avoided.

What is Really Optimized?

Some analysts carefully optimize injector temperatures for given applications - but which temperature do they really optimize: that in the vaporizing chamber orthat atthe rear of the syringe needle? This determines whether the optimum found has a more general value, e.g. is transferable to other instruments. When it is taken into account that experimen­ tal optimization of the injector temperature is. in re­ ality. often an optimization of the needle temperature, strange situations are likely to be frequent.

High Injector Temperature to Achieve Warm Septum Cap

Perhaps an injector temperature of 400°C is found to pro­ duce the best results. The analyst might not be aware that this extremely high temperature was needed simply because this was the only way to heat the injector head to, say, 200°C. It is still more disconcerting to realize that often cool needle injection is applied, not even fully exploiting the 200°C achieved with such difficulty, l.e. that the 400 °C "injector temperature" was required to achieve a needle temperature of maybe 150°C. A hot needle injection into an injector homogeneously heated at 150 °C might. therefore. provide equally good results.

Reporting Injector Tempera­ tures

In the description of analytical methods, the injector tem­ perature is regarded as important information. Reported in­ jector temperatures are of limited usefulness, however, be­ cause they refer to the center of the vaporizing chamber only. It would be useful to add at least the easily measurable tem­ perature of the septum. As the temperature profile of the injector cannot be described in a method, the injector must be specified. Commonly the instrument manufacturer and the model ofthe gas chro­ matograph are indicated. This is not fully satisfactory, how­ ever, because, on the one hand, the same injector is usually mounted on several instrument models, and, on the other hand, parts like the heating system might be changed dur­ ing production of a model without the user being aware of it.

8.2. Temperature Gradient Towards the Septum

On the Illusion of Rapid Evaluation of GC Instru­ ments

55

Many analysts have attempted to evaluate GC instru­

ments. e.g••wt'len tIleyJntended to purchase one. by means

of a few injections. There is usually 8m8zing disagr. .

~bout the conclusions - one analyst finds far less

discrimination With ins .yment 1.whereas another greatly

prefersil'lstrument 2 reotly for the same reason. What

did they really test7\~n it turns out to be the t ..........

tare profile of the injector.

If (me l:l,na!yst h~~ned to use.. a.~e~t s~ll1plein. aJ~ther

hiSt'l-boi!i.l'lg. a pool septum cap of in~rument 1

ll1ight~a~~pre d evaporation Inside tile needle (.pro­ videdheinjecte with the cool needle mathodl. At the

same nominal injector temperature on instrument 2. the

same sample might. however. have partially evaporated

in the needle. because the injector was more intensely

heated towards the top.

The other anaIYfit might have used a slightly more vola­

tile solvent or injected somewhat moreslgwly. During in­

jection into instrument 1 his test sample started evapo­

ratingin.the nee.dle. It was poorly transferred (cool injec­

tor head) and di$Crimination was terrible. He found that

the more intensely heated injector of instrument 2 pro­

vided better results.

Testing and comparison of instruments is,gf course.• im­

portant. but the rElsultsare usually valid only for the test

sample. the injectionyolume. the injector temperature.

the needle handling technique. and all the other condi­

tionsapPUl:ld. which Wpyld be obvious ifthe background

were understood.

The story should. nevertheless, not be put aside with a

smile about ignorant instrument testers. Serious account

must be taken of the observation that a certain temPElra­

ture distribution along the axis of the injector is prefer­

able for one application, but a drawback for another.

me"

8.3. Thermostability of Septa

As mentioned above, an injector with a perfect temperature profile is of limited practical usefulness if low discrimination is obtained at the expense of severe septum problems. What are the problems and where are the limits? Other aspects concerning septa will be discussed in Section E2.3.

8.3. 1. Upper Tempera­

In the past. injectors were designed to keep the septum cool to avoid excessive septum bleed -volatile components from the septum produced complex patterns of peaks and drift­ ing baselines, particularly in temperature-programmed chromatograms. after bleed material had been accumulated in the column inlet while the oven temperature was low. Septum bleed was a severe problem for classical packed col­ umn inlets. Modern injectors for capillary GC. however. are equipped with a septum purge, i.e. a permanent stream of gas removes the vapors released by the septum.

ture Limit Septum Bleed

56

A 8. Injector Temperature

Excessive Softening

Today the temperature limit of a septum is primarily deter­ mined by softening. Excessive heating causes the septum to flow like honey into the narrow channel towards the liner. After a few injections, such material is torn away from the septum and drops into the vaporizing chamber. The septum also becomes baked on to the metal parts such that it must be scraped off in small pieces when it needs to be replaced.

Advertized Thermostability

When septa are advertized as resisting high temperatures, the temperatures indicated usually refer to those set, not to the real temperatures of the septum cap. This renders comparison of data rather confusing - a septum might well resist "400°C" on one injector yet melt on another at "300 °C".

8.3.2. Some Tips

We are used to high injector temperatures - encouraged by data such as those shown above. We should keep in mind, however, that often these were necessary simply to achieve a modest temperature at the injector head. With a more constant temperature distribution, lower temperatures are sufficient; an injector thoroughly heated at 250°C might well perform better than another at 350 °C.

Lower Injector Temperatures

Tightening Septum Cap

Leakage upon Cooling

If high temperatures are applied, the septum cap must be tightened especially carefully. Septa expand upon heat­ ing. If the cap is tightened at ambient temperature, heating of the injector builds up a pressure on the septum that squeezes it into all corners, including the channel through which the syringe needle should pass. The needle will enter with difficulty and frequently cut off pieces of septum. The septum cap should be tightened only when the in­

jector temperature has been reached.

To avoid leakage, the carrier gas must be switched on

only then.

Apply minimum force, just sufficient to prevent leakage,

because the septum will last longer and there will be

fewer septum particles in the vaporizing chamber.

Septa contract when cooled, wh ich means that there is a risk of leakage after cooling of the injector. The injector should, therefore, remain heated when the instrument is kept in stand-by (consuming 30-100 WI. It also means that the carrier gas must be switched off before the injector is cooled. Some septum caps are equipped with a spring, providing a more constant pressure on the septum when the tempera­ ture changes. This also prevents leakage upon cooling after an electric power breakdown.

9.1. Accuracy of Sample Volume

57

9. Plunger-in-Needle Syringes No Needle Problems?

One-microliter syringes have been claimed to be a solution to the needle problems described above - the plunger is supposed to displace the whole of the measured sam­ ple mechanically, i.e. to prevent volatilization of liquid re­ maining inside the needle after the injection. The sample volume should, therefore, be accurate and the introduction non-discriminating. One must differentiate between samples evaporating inside the needle and those for which this can be avoided. If evapo­ ration can be prevented. plunger-in-needle syringes fulfil the above expectations. Otherwise, however, they have several serious shortcomings rendering their use ques­ tionable.

9.1. Accuracy of Sample Volume

One-microliter syringes are graduated to dispense an accu­ rate volume of liquid under conditions which exclude evapo­ ration inside the needle. There is, however, an annular space between the internal needle wall and the plunger which is also filled with sample liquid. When evaporating, it leaves the needle in addition to the measured volume. In this way, the actual volume injected exceeds that measured by easily 0.1-0.2 p.L (in many cases corresponding to an in­ crease of 100 %). The sample material between the needle and the plunger often, moreover, causes problems with sy­ ringe cleaning.

Reproducibility

Transfer of the material from the space between the needle and the plunger tends to be poorly reproducible. Optimum reproducibility presupposes that it is complete. This might require the syringe needle to be kept inside the injector for 5-10 s.

9.2. Premature Expulsion

In contrast with plunger-in-barrel (5-10 Ill) syringes. plunger­ in-needle syringes offer no possibility of withdrawing the sample into the barrel when the needle is inserted into the hot injector. Thus injection is always performed by the filled needle technique.

Evaporation before the Plunger is Depressed

Mechanical displacement ofthe sample liquid from the nee­ dle presupposes that the liquid is still there when the plunger is depressed, but this is rather unlikely for samples dissolved in the most commonly used solvents. At the moment the analyst pushes the plunger - to "inject" • as he fondly imagines - the needle is already empty.

58

A 9. Plunger-in-Needle Syringes

It is not completely empty, as there is still high-boiling sol­ ute material on the internal needle wall. The plunger is, of course, unable to scrape these nanograms of solute mate­ rial from the wall - 10 ng forms a layer about 2 nm thick. Depression of the plunger locks it up between the needle and the plunger, where, maybe, it acts as a lubricant. Hence injection does not proceed as intended. Losses through the Septum Purge

The maximum sample volume of 1 III fills the 5-cm needle. A volume of 0.1 III corresponds, therefore, to a plug 5 mm in length, and when the plunger is left at the "0.1 Ill" mark, the sample is located in the front 5 mm of the needle. When the syringe needle penetrates the septum, the sample starts evaporating and is expelled into the septum or the area just below it (Figure A24). From there it is likely to be flushed out of the injector by the septum purge. The first 1-2 cm of the syringe needle, the content of which might be transferred into the stream of the septum purge, contains 0.2-0.4 III of sample. This might be the total amount of sample to be in­ jected.

10-20 mm

Figure A24

Premature elution from plunger-in-needle syringes during in­

jection of samples with volatile matrices. Material leaving

the syringe needle before its exit reaches the vaporizing cham­

ber iii vented through the septum purge.

Switching off the septum purge during injection is of little help because the sample vapors eluted into the injector head are not efficiently transported into the vaporizing chamber. The needle must reach the vaporizing chamber be­ fore it releases the first solute material. The sample must, therefore, be withdrawn by at least a corresponding distance (0.2-0.4 ul, read on the barrel) before the needle is inserted into the septum. This means that only part of the capacity of the syringe can be exploited. In conclusion, plunger-in-needle syringes might well create more new problems than they solve.

10.1. High Boiling Sample Matrix

59

10. Possibilities of Avoiding Evaporation in the Needle This section lists suggestions and possibilities for avoiding evaporation inside the syringe needle. 10.1. High Boiling Sam­ ple Matrix

In quality control of industrial production, many samples are injected without prior dilution. Often they are in a matrix of sufficiently high-boiling point that evaporation inside the syringe needle can be avoided. For other samples it is possi­ ble to use high-boiling solvents, such as alkanes between octane and tetradecane, alkylated benzenes, dimethylfor­ mamide (DMF), glycol ethers, or esters of fatty acids with intermediate chain lengths.

10.1. 1. Injector Tempera­ ture venus Solvent Boiling Point

In the 'design of a method exploiting this concept, the differ­ ence between the sample (solvent) boiling point and the in­ jectortemperature plays a key role. The sample (solvent) must not evaporate inside the needle, but should be vaporized within a reasonably short time inside the vaporizing chamber.

Minimum Injector Tempera­ ture

A low injector temperature helps prevent evaporation inside the needle. Hence, as a first step of method development, an estimate is made of the minimum injector temperature required for volatilization of the solutes. This tempera­ ture is usually substantially below the boiling point of the solutes, but depends on tolerable initial band widths (required speed of evaporation) and possible adsorption effects in the vaporizing chamber.

Solvent Boiling Point

The second step is selection of the solvent. Its boiling point must lie above the temperature the syringe needle reaches in the injector, but is preferably below that of the injector. This might, at first, seem an impossible task, but several fac­ tors contribute to rendering it rather easy.

Rapid Injection

Injection must be performed by the cool needle technique and as rapidly as possible, so that the syringe needle is with­ drawn before it approaches the injector temperature. The faster the injection, the lower may be the solvent boiling point compared with the injector temperature. Injection speeds are usually high and highly reproducible if autosamplers are used; in manual injection, the design of a method has to bear in mind that some injections will be slower than others.

60

A 10. Possibilities of Avoiding Evaporation in the Needle

Short Syringe Needles

Rapid introduction of the syringe needle through the

septum is greatly facilitated when it is short. Short needles

protrude, furthermore, less far into the well heated part

of the injector.

Temperature Gradient towards Septum

A cool septum is advantageous, because it transfers less

Testing

Whether or not the sample evaporates inside the needle can

easily checked by withdrawing the plunger after an in­

jection. A volume of liquid corresponding to the needle vol­

ume should have remained there.

If injection is performed manually, the test injection is car­

ried out slightly more slowly than normal to ensure that the

liquid remains inside the needle even when the needle en­

ter.s with more difficulty (e.g. because a new septum has been

installed) or the plunger does not move as freely as it should.

An Experimental Result

Schomburg et al. [13] investigated discrimination against n­ alkanes when manually injecting solutions in ...octane'

(b.p. 126°C) or ...dodecane (b.p. 216°C) at injectortem­ peratures of 210 or 310°C using the cool needle technique (see also Section C10.4.2). Injection of the octane solution at 210°C resulted in severe discrimination against the high-boiling compounds, indicat­ ing that evaporation inside the needle could not be sup­ pressed by use of a solvent boiling 84° below the injector temperature. The dodecane solution, on the other hand, af­ forded perfectly linear results even at an injector tempera­ ture of 310°C. Thus there was no evaporation inside the needle despite a solvent boiling point 94° below the (imposed) injector temperature. The difference probably resulted from the steeper temperature drop towards the septum cap when the injector temperature was high.

Thermal Conductivity of the Carrier Gas

The maximum difference between the solvent boiling point and the injector temperature also depends on factors d.

heat to the needle. Furthermore, the needle tip preferably

does not reach into the truly thermostatted region of the in­

jector, calling for a short needle. In fact, whether or not evapo­

ration inside the needle should be as complete as possible

or totally avoided requires opposite conditions.

termining how rapidly the syringe needle is heated, such as the contact with hot surfaces and the thermal con­ ductivity of the carrier gas, as shown by Schomburg et al. [20]. Reglera et al. [21] optimized split injection for a test mixture of C,,-C 20 n-alkane standards. Hydrogen as carrier gas re­ sulted in relative standard deviations of absolute peak areas of 25-35 % (I) (sample volume, "0.4 ul," ); they were 8 % with helium and 4-5 % with nitrogen. The effects were probably the result of variable transfer from the needle, depending on the thermal conductivity of the carrier gas.

10.1. High Boiling Sample Matrix

61

Although such results are impressive, they do not furnish sufficient arguments for use of nitrogen as carrier gas; hy­ drogen remains first choice in almost all situations.

Conclusion

There is no simple rule about how much the sample or sol­ vent boiling point should be below the injector temperature, because there too many interfering factors, the temperature of the injector head being the most important. In practice, solvents with boiling points up to some 1000 below the injector temperature were successful. The experimental test is simple!

10.1.2. Practical Aspects

High boiling solvents are seldom suitable for the analysis of volatile solutes, because the latter tend to be obscured by the often broad peaks of the solvent and its impurities. Components eluted before the solvent form peaks distorted (broadened or even split) primarily by partial solvent trap­ ping [22,23]. Injection at high split ratios is an exception to this.

Not for Volatile Solutes

Split Injection

In split injection, three factors determine the upper limit of the solvent boiling point. Volatilization of the sample in the injector may become so slow that the initial bands of the solutes are ex­ cessively broadened. This is important in isothermal runs; temperature-programmed analysis is more toler­ ant because cold trapping reconcentrates the initial bands. 2 If the oven temperature during injection is well below the solvent boiling point, the solvent strongly recondenses in the column inlet. This increases the flow rate into the column and affects the split ratio (see Section C8.3.3). It might, moreover, cause non-linear splitting iftransport into the column is enhanced for vola­ tile solutes only. 3 Depending on the instrument, the solvent may recondense in the split outlet line, which again af­ fects the split ratio. The split outlet line, including the restrictor determining the split flow rate, should, there­ fore, be short and heated. 4 As a result of all these effects, the reproducibility of absolute and, to a lesser extent, relative peak areas tends to be poor.

Solvent Trapping at Elevated Oven Temperatures

Besides solving the needle problems, in splitless injection the use of high-boiling solvents can accelerate the analy­ sis, because solvent effects can be achieved at elevated col­ umn temperatures and less cooling of the oven is required. This might even enable the analyst to perform GC isother­ mally at elevated temperatures.

Solvent Purity

The purity of high-boiling solvents tends to be a limiting fac­ tor. Efficient redistillation in the laboratory is usually diffi­

62

A 10. Possibilities of Avoiding Evaporation in the Needle cult. In practice, use of high-boiling solvents for splitless in­ jection is often restricted to applications employing selective detectors.

10.2. Cooled Septum

Intimate contact with the septum efficiently transfers heat to the syringe needle. Pankow et al. [24] investigated the possi­ bility of preventing evaporation inside the needle by active cooling of the septum cap. Unfortunately, improved repeatability of the results was reported rather than the ex­ tent of evaporation in the needle. The injector temperature of 330°C was, furthermore, too high to give hexane (used as solvent for the test mixture) a real chance of remaining unevaporated inside the syringe needle.

10.3. Cooled Needle Technique

In 1984, Schomburg et al. [25] described an accessory for cooling the syringe needle inside the vaporizing cham­ ber: a double tube was introduced with coolant flowing through the outer tube to keep the inner tube at a low tem­ perature. The syringe needle passed through the cooled in­ ner tube and only its tip protruded into the hot chamber. A modification described later [26] included better insulation of the cooling mantle against the liner and the use of a wider vaporizing chamber.

Absence of Discrimination

For split injection of a mixture of ClO-C32 n-alkanes in a sol­ vent not specified, discrimination of the high-boiling com­ ponents could be eliminated completely, indicating that evaporation in the needle was successfully suppressed. The same test mixture showed absence of discrimination also for splitless injection. These data were used to support the opinion that no high-boiling solute material condensed on the outer wall of the cold finger reaching into the vaporiz­ ing chamber. As the sample vapors could not expand up­ wards (the space being filled by the cooling finger), there seemed to be no room for them.

10.4. Fast Injection by Autosampler

In 1985 Hewlett-Packard introduced an autosampler (HP 7673A) injecting at a speed suppressing evaporation in the syringe needle. The maximum needle dwell time, i.e. the maximum residence time of the needle inside the injector not initiating sample evaporation inside the needle, was de­ termined [27]. As is shown in Figure A25 for an injector set at 350°C and a sample in hexane, there was no discrimi­ nation against n-C40 with needle dwell times up to 300 ms. Since results were similar for pentane as solvent, evaporation inside the needle could be ruled out for all the commonly used solvents. Typical fast manual injection is es­ timated to take 1-2 s.

Cool Injector Head

The results in the above figure are valid only for the injector used, because they are influenced by the temperature pro­ file in the injector head. In fact, the temperature of the injec­

10.4. Fast Injection by Autosampler

63

1.2

1.1

~ U

7673A Fest Injsctlon (100 ms, 1.0

!

l!

•

!

c

0.9

Typical menusl Injection

0.8

0.7 0.0

1.0

2.0

Needls dwell time (seconds'

Figure A25 Dependence of the r8tio of the pe8k 8re8S, n-C.Jn-C,o' on needle dwell time in the injector, determined using 8n 8utosampler with progr8mm8ble injection speed. An 8re8 ratio of unity indic8tes no discrimin8tion 8g8inst n-C4Q' i.e, th8t the s8mple did not eV8por8te inside the syringe n_dle. (From· Snyder [27]).

tor head was low: the injector temperature set at 350 "C brought the septum to hardly 140 °C. Such fast injection has several advantages: it eliminates the most important source of discrimination against high­

Advantages

boiling components, improves reproducibility by virtually eliminating a poorly reproducible deviation, enables the in­ jection of sample volumes smaller than that of the needle, and provides peak areas proportional to a well defined injec­ tion volume. Dilemma

There are hardly objections against a low septum tempera­ ture as long as injection is performed exclusively by means of the fast autosampler, thus suppressing evapo­ ration in the needle. If, however, the same injector is used for manual injection of samples in volatile solvents, the low temperature of the injector head results in severe dis­ crimination against high-boiling components. For samples sensitive to discrimination, performance with manual injec­ tion is rather poor. It is probably impossible to perform autosampler injection at a speed preventing evaporation inside the needle with an injector well heated up to its top. Hence the temperature pro­ file is either optimized forthe fast autosampler or for manual injection.

10.5. Evaporation in the Injector

When samples are injected by a technique preventing evapo­ ration in the needle (high-boiling solvents or fast auto-

UNIVERSIPAP DE ANTIOQUlA

BlBLIOTBCA CENTRAL

64

A 10. Possibilities of Avoiding Evaporation in the Needle

sampler), the evaporation process in the vaporizing chamber is also influenced. As shown in more detail in the next section, evaporation inside the needle nebulizes the sample liquid upon leaving the needle, which greatly facili­ tates further evaporation. Nebulization is, in fact, the prereq­ uisite for sample vaporization in an empty If the sample leaves the needle as a band of liquid, it must be stopped by means such as glass wool or obstacles built into the liner. A large proportion of the sample, at least, evaporates from surfaces and often suffers from adsorption or retention by non-evapo­ rated material from previous injections. One Problem Traded Against Another

Hence partial evaporation in the needle causes the prob­ lems outlined in this section. but avoids those de­ scribed in the next.

Autosampler Injection is not Automated Manual Injection

It is tempting to think of autosampler injection as automa­ tion of manual injection. Some autosamplers really imitate manual injection, enabling hot needle injection to be per­ formed with the addition of some solvent ahead or behind the sample plug (Section E3). A fast autosampler, however, fundamentally changes the evaporation process. Methods must specify the kind of a sample introdu'ction for which they are written, and they must be r.validated when chang­ ing from manual injection to fast autosampling or vice versa.

11. Summarizing Guidelines At the risk of making simplifications which are not always appropriate, the following working rules can serve as con­ cluding guidelines. They refer to the injection of liquid sam­ ples. Injection of gases is not considered here because there are no problems related to those discussed in this chapter. Strategy

1) Decide whether to go for suppression of evaporation in­ side the needle or a transfer of the needle content which is as complete as possible.

Suppressed Evaporation inside Needle

2) If suppression is chosen: a) Inject with a fast autosampler, select a high-boiling sample matrix, or use an extremely short syringe needle (20 mm or longer needle inserted only par­ tially). b) Unless injecting with a fast autosampler, use the low­ est injector temperature ensuring satisfactory evapo­ ration of the solutes.

11. Summarizing Guidelines

65

c) A cool injector head helps to keep the needle tem­ perature low. d) Pack the injector liner with deactivated glass wool or use a liner with a suitable obstacle to stop the band of liquid and promote its vaporization. 3) If partial evaporation inside the syringe needle is pre­ ferred or cannot be avoided, render elution from the nee­ dle as complete as possible. The following remarks ap­ ply to samples containing solutes of intermediate to high elution temperatures (volatility below that of, e.g., n­ pentadecane). a) Do not use plunger-in-need/e (one microliter) sy­ ringes. b) Use a hot injector head, i.e. septum caps without cool­ ing fins, and insulate the top of the injector. c) Use injector temperatures as high as can be toler­ ated by the solutes. d) Use the hot needle method, i.e. withdraw the sample into the barrel of the syringe, pre-heat the needle in the injector for 3-5 s, and depress the plunger as rapidly as possible. e). At injector temperatures below ca. 200°C, use the solvent flush technique. f) Inject sample volumes exceeding the needle volume: "0.5" to "1 ~L". If smaller sample volumes are to be introduced, use the solvent flush technique. g) Do not quantitate on the basis of the sample volume. The accurate amount of injected solute material must be considered unknown, although reproducible. h) Do not change the volume injected between calibra­ tion and analysis of the sample - even relative peak areas might depend on the injection volume. i) Use the same syringe throughout a quantitation pro­ cedure to rule out systematic deviations resulting from contamination of the needle wall. k) Use solvents of a polarity sufficient to avoid adsorp­ tion in the syringe (perhaps as pure solvent behind the sample plug, applying the hot needle solvent flush method). I) Test the losses of solute material in the needle by needle rinse injections, using a strong solvent if adsorptive solutes are involved. m) Use long syringe needles and empty liners (accord­ ing to the requirements of thermospray injection).

66

A References

References A

2

3 4 5 6 7

8 9 10

11 12 13

14

15

16 17

18 19 20

21

22

D. Tong, AM. Barnes, K.D. Bartle, and AA Clifford, "Valve Injection for Gas Chromato­ graphic Analysis on Small-Bore Open Tubular Columns", J. Microcolumn Separations 8 (1996) 353. J. Qian, C.E. Polymeropulos, and R. Ulisse, "Liquid Jet Evolution from a GC Injector", J. Chromatogr. 609 (1992) 269. J.v. Hinshaw, "Syringes", LC-GC International 1 (4) (1988) 24. O.K. Guns, "Effect of Injection Needle Dimensions in GC", J. Chromatogr. 292 (1984) 57. K. Grob and S. Rennhard, "Evaluation of Syringe Handling Techniques for Injections into Vaporizing GC Injectors", HRC & CC 3 (1980) 627. K. Grob and H.P. Neukam, "The Influence of the Syringe Needle on the Precision and Accuracy of Vaporizing GC Injections", HRC & CC 2 (1979) 15. F.-S. Wang, H. Shan field, and A,. Zlstkis, "Injection Temperature Effects Using On-Column and Split Sampling in Capillary GC", HRC & CC 6 (1983) 471. P. Hilling (personal communication) K. Grob and K. Grob, "On-Column Injection on to Glass Capillary Columns", J. Chroma­ togr. 151 (1978) 311. K. Grob, and Ph. Mati/e, "Capillary GC of Glucosinolate-Derived Horseradish Constitu­ ents", Phytochemistry 19 (1980) 1789. J. Roeraade, "Factors Affecting Sample Transfer from Microlitre Syringes", J. Chroma­ togr.441 (1988) 367. P.A Rodriguez, C.L. Eddy, G.M. Ridder, and C.R. Culbertson, "Automated Quartz Injection Trap for Fused-Silica Capillary Columns", J. Chromatogr. 236 (1982) 39. G. Schomburg, R. Dielmann, H. Borwitzky, and H. Husmann, "Capillary GC of Compounds of Low Volatility", in: G. Schomburg and L. Rohrschneider(Eds.), Proc. 12th Int. Symp. on Chromatography, Baden-Baden, 1976, Elsevier, Amsterdam (1978) P157. K. Grab, "Evaluation of Injection Techniques for Triglycerides in Capillary GC", J. Chro­ matogr. 178 (1979) 387. G.L. Le Bel, and D. T. Williams, "Effect of Injection Solvent on GC Quantitation of some Polar Organophosphorus Pesticides", J. Assoc. Off. Anal. Chem. 62 (1979) 1353. H. Brotell, "Syringe and Column Adsorption of Tertiary Amines in GC", J. Chromatogr. 196 (1980) 489. D.I. Carroll, I. Dzidic, R.N. Stillwell, M.G. Horning, and E.G. Horning, "Subpicogram De­ tection System for Gas Phase Analysis Based upon Atmospheric Pressure Ionisation (API) Mass Spectrometry", Anal. Chem. 46 (1974) 706. M.S. Klee, "GC Inlets - An Introduction", Hewlett-Packard Co., Avondale (1991) 42. K. Grob and H.P. Neukam, "Should the Septum Part of Vaporizing Injectors be Kept at Lower Temperature?" J. Chromatogr. 198 (1980) 64. G. Schomburg, H. Husmann, and R. Rittmann, "Direct (On-Column) Sampling into Glass Capillary Columns. Comparative Investigations on Split, Splitless and On-Column Sam­ pling", J. Chromatogr. 204 (1981) 85. G. Reglero, M. Herraiz, and M.D. Cabezudo, "Sampling in Capillary GC: A Comparison

between the Split and the PTV-Split Procedures", Chromatographia 22 (1986) 333.

K. Grob, "Broadening of Peaks Eluted Before the Solvent in Capillary GC. Part 1: The Role

of Solvent Trapping", Chrornatoqraphia 17 (1983) 357.

A References

67

23 K. Grob and B. Schilling, "Broadening of Peaks Eluted Before the Solvent in Capillary GC. Part 2: The Role of Phase Soaking", Chromatographia 17 (1983) 361. 24 J. F. Pankow, ~E. Asher, and L.M. Isabelle, "Reduction of GC Needle Volatilization and Septum Bleed with Active Septum Cooling", Anal. Chem. 55 (1983) 1451. 25 G. Schomburg, U. Hiiusig, H. Husmann, and H. Bebleu, "Sampling onto Capillary Col­ umns. Difficulties with Various Types of Samples. A Simple Accessory to Split Injectors for Avoidance of Discrimination", Chromatographia 19 (1984) 29. 26 G. Schomburg and U. Hiiusig, "Application ofthe 'Cooled Needle' Technique to Split and Splitless Sampling onto Capillary Columns", HRC & CC 8 (1985) 572. 27 ~D. Snyder, "Automatic Sample Handling. Fast Injection with the HP 7673A Automatic Injector: Chemical Performance", Technical Paper Nr. 108, Hewlett-Packard (1985).

1. Introduction

69

B Sample Evaporation in the Injector

1. Introduction According to the classical concept of split and splitless injec­ tion, the sample liquid leaving the syringe needle must be evaporated before it reaches the column entrance. Then the vapor is either split or transferred into the colum n in splitless mode (Figure B1). For many years, there was little solid knowledge about the vaporization process, but much specu­ lation. It is, in fact, difficult to tell what has happened inside Syringe

Septum

Injector liner

Vaporizing chamber

~ Split outlet

Sample leaving the syringe needle with or without partial evaporation inside the needle (subject of Section A) The sample must evaporate before reaching the column entrance (subject of Section B)

Split or splitless transfer into the column (Section C and 0)

Column

Figure 81

The steps of the injection process.

70

B 1. Introduction

the injector merely by looking at chromatograms and com­ paring peak areas. Visual observations made in recent years, discussed here and shown on the CD-ROM, have im­ proved matters substantially.

1.1. Problems Caused by Incomplete Evaporation

Incomplete evaporation does not produce obvious effects in the chromatograms, such as broadened or distorted peaks, nor are there other simple indicators of whether or not a sam­ ple was completely vaporized before it reached the column. Incomplete evaporation can severely affect quantitative re­ sults, but not all problems in quantitative analysis are re­ lated to sample evaporation. Hence the subject is complex and requires the attention of the analyst.

Inaccurate Split Ratio

In split injection, non-evaporated sample material is usually split by an incorrect ratio (Section C8.3.4), because the prob­ ability of larger droplets or a band of liquid hitting or pass­ ing the column entrance is independent ofthe gas flow rates, i.e. of the adjusted split ratio. As the liquid has a poorly reproducible flight path, the amount of sample material entering the column is not under control. This is not neces­ sarilytrue for small droplets or particles, since aerosols might behave like vapors.

Non-Linearity of Splitting

The other problem, discussed in more detail in Section C9.3, is the "Iinearity of splitting". Incomplete evaporation is one reason why the composition of the sample material entering the column can differ from that injected. If the volatile components evaporate, while the rest passes the split point as droplets or as a band of liquid, vapor and liquid (and the sample components in these two phases) are likely to be split in different ratios.

Splitless Injection

In splitless injection, incomplete evaporation causes sample material to be "shot" to the bottom of the vaporizing chamber, from where it may not return to the column en­ trance. Particularly if septum particles or other material ab­ sorb the higher-boiling sample components, they are lost to the analysis. As a result, peak areas are too small and high­ boiling components likely to be affected more severely than the volatiles. Again, the process is not under control.

Poor Quantitative Results

Problems associated with incomplete sample evaporation be­ come apparent from experiments on the accuracy of results - peak areas differ from what they should be. Detection of deviations from the correct peak size is not always easy, since it presupposes comparison with a reference. Conclu­ sions are further complicated by other mechanisms result­ ing in similar deviations.

High Standard Deviations

The phenomenon most likely to be noticed is poor reproduc­ ibility of absolute and relative peak areas. High standard de­

1.1. Problems Caused by Incomplete Evaporation

71

viations are the result of poor reproducibility of the de­ viations from the peak areas required. Unfortunately there are, again, other reasons for high standard deviations.

2. Solvent Evaporation - Heat Transfer Some information on the vaporization process can be drawn from the heat consumed by sample evaporation and the time available for heat transfer. The estimates presented below are semi-quantitative, aimed at painting a picture of the situation. Assumptions are not sufficiently accurate to en­ able practical conclusions to be drawn, as is also shown by the visual observations discussed subsequently. Solvent Evaporation as First Obstacle

Solvent evaporation is the first serious problem. It seems simple to evaporate 2 III of a solvent with a boiling point below 100°C in a vaporizing chamber at, e.g., 250°C. The facts presented below show, however, that it is not. As long as the solvent is not fully evaporated, the tempera­ ture of the droplets is the solvent boiling point. Only ex­ tremely volatile solutes can then evaporate; all the compo­ nents eluted after the solvent peak remain in the liquid phase. Hence, because solutes evaporate after the solvent, va­ porization of the solvent is the first obstacle to overcome.

2.1. Available Evapora­ tion Time

The time available for sample evaporation depends on the movement of the sample liquid in the vaporizing chamber. Three scenarios will be considered, two of which are shown in Figure 82.

Carrier gas supply Injector insert

Split outlet

Scenario 1 Band of liquid

Scenario 2 Nebulized liquid

Figure 82 Two scenarios of sample evaporation. (From ref. [1]).

72

B 2. Solvent Evaporation - Heat Transfer

2 3

The sample liquid forms a narrow band "shot" through an empty chamber. The sample liquid is nebulized, forming a fog driven by the carrier gas. The liquid is deposited on to the liner wall or a pack­ ing material.

2. 1. 1. Band of Liquid

The sample liquid leaves the syringe needle as a narrow band

of liquid and is "shot" through the chamber, as observed

when the sample is "injected" into ambient air.

Speed of the Sample Liquid

The sample liquid leaves the needle at a velocity determined

by the speed of the plunger and the type of syringe used. In

manual injection, the plunger is normally depressed at 1-2

m/s. The liquid is further accelerated when it passes from

the wide-bore glass barrel into the narrower-bore syringe

needle. For 10 III syringes and 26S gauge needles, accelera­

tion corresponds to a factor of 15, i.e. the liquid leaves the

needle at a speed of 15-30 mls (50-100 krn/h). Since friction

is low, the band continues traveling at this speed for quite

some distance, certainly to the base of the injector.

Time for Evaporation

The band of liquid covers a 4-cm distance to the column

entrance in ca. 2 ms. In this situation, the time available '

for evaporation is primarily determined by the duration of

the injection. If the plunger is depressed at 1 mis, a 1 cm

plug of liquid is introduced in 10 ms. Visual experiments

(video on the CD-ROM) confirmed that the liquid mostly left

the needle in a single frame of the film (40 ms) even when

5 III were injected.

2.1.2. Nebulized Sample

As a Scenario 2 we assume that the sample liquid is nebulized

upon leaving the needle (thermospray, see below). Initially

the speed of the small droplets is the same as that estimated

above for the band or even higher (the vapors formed in the

needle act like gunpowder), but friction between the drop­

lets and the carrier gas rapidly slows the former to the speed

of the gas. To a first approximation, they cover the distance

between the needle exit and the column entrance at the

speed of the carrier gas.

Speed of the Carrier Gas

Some calculated carrier gas velocities in the injector are given

in Table 81. A 1 mlzmln flow rate is typical for splitless in­

jection; 20 and 100 rnt/mln refer to the flow rates in split

injection.

In split injection into a 2 mm i.d. liner, it takes the carrier gas

75 ms to 1.5 s to cover a 4 cm distance from the needle exit

to the column entrance, and if the nebulized liquid travels at

the same speed, this is also the time available for evapora­

tion. As the gas in a 4 mm i.d, liner is four times slower than

in a 2 mm i.d. liner, correspondingly more time is available

for evaporation.

2.1. Available Evaporation Time

73

Table 81

Gas velocities in 2 or 4 mm i.d. liners for given gas flow

rates. and the resulting time available for sample evapora­

tion if the nebulized liquid moves at the gas velocity.

Liner i.d. [mml

Flow rate [mLJminl

Gas velocity lcrn/s]

Evap. time [sl

2mm

20 100 1 (splitless) 20 100

10 53 0.1 2.6 53

0.4 0.075 30 1.5 0.3

4mm

In splitless injection, residence times in the vaporizing cham­ ber are long.

2. 1.3. Deposition on Surfaces

The third scenario assumes that the sample liquid is trans­ ferred. to the liner wall or on to a packing material. As the liquid remains stationary, evaporation from this surface may be slow. The time available for sample evaporation is theoretically unlimited.

2.2. Amount of Heat Required

The limiting factor in rapid evaporation is the transfer of heat to the sample liquid, usually primarily consisting of solvent. Table 82 lists amounts of heat required to evapo­ rate 1 III of solvents and to heat it up to an injector tempera­ ture of 250 "C. As a simplification, calculations use heat ca­ pacities for the solvent vapor. Table 82 Heat [meal] required to evaporate 1 III of solvent and bring it to 250°C ([2.3]).

Solvent Hexane Toluene Diethyl ether 1-Propanol Methanol Water

2.3. Sources of Heat

Heating

Evaporation

Total

89 61 78 77 72 220

80 86 85 166 263 539

169 147 163 243 335 759

The energy required to vaporize and heat 1 III of solvent varies by a factor of almost five. It is approximately five times lower for the solvents normally used than for the most energy-consuming solvent, i.e. water. To evaporate the sample in a short time, heat must be read­ ily available; there is insufficient time for heat transfer over longer distances.

74

B 2. Solvent Evaporation - Heat Transfer

Cooling of the Injector

Before estimating the amount of heat available from differ­

ent sources, it should be remembered that absorption of

heat will always cool the source. Sample evaporation

does not cool the whole heating block, only the parts of the

vaporizing chamber with which the sample is in contact. The

amount of heat available depends on the extent to which

cooling is acceptable - is, for instance, cooling to 230°C tol­

erable if the injector temperature is set at 250 °C?

2.3. 1. Carrier Gas

The heat most readily available to the sample is that con­

tained in the carrier gas. Little more than the carrier gas be­

tween the needle exit and the column entrance is in contact

with the sample, whether a nebulized sample moves with

the gas or a band of liquid is shot through it.

Assuming a 2 or a 4 mm i.d. liner and a 40 mm distance

between the needle tip and the column inlet, heat can be

extracted from 130 or 500 ~L carrier gas, respectively.

Less than 1 % of the Heat Required

The heat capacity of hydrogen is 3.4 cal!g K, hence the heat

available from 130 III of hydrogen is 4 x 10-5 cal per de­

gree of cooling (3 and 4 x 10-5 cal/K for helium and nitro­

gen, respectively). Even if we allow the carrier gas to be

cooled from 250 to 200°C, only 2 x 10-3 cal become avail­

able, i.e. not even 1 % of the heat required if 2 III of a tvpical :

solvent is injected. Expressed differently, injection of some

0.04 ~L of solvent is sufficient to cool the carrier gas in the vaporization zone to a temperature near ambi­

ent. Mixing with Carrier Gas does not Help

As the amount of heat extractable from the carrier gas is negligible compared with that required, mixing of the sam­ ple with carrier gas is an inefficient means of supporting sample evaporation. We need richer sources of heat.

2.3.2. Packed Injector

A current idea is to increase the thermal capacity of the vaporizing chamber by packing it with solid material, e.g. glass wool. Transfer of heat to the sample from this packing material will again reduce the temperature, but the larger the mass of material cooled, the larger is the amount of heat available.

Liners

2 J.lL Hexane, Cooling by 20°

As an example, we assume a 2 III injection of an "easy" solvent, such as hexane, and accept cooling of the vaporiza­ tion zone by 20° (e.g., from 250 to 230°C). Evaporation and heating of 2 III hexane to 230 °C requires 0.322 cal. With a heat capacity of 0.2 cal!g K for glass (glass wool, silica, or column packing materials), heat must be extracted from 80 mg packing material.

Glass Wool

Even if densely packed, it is difficult to introduce 10 mg glass wool into a vaporizing chamber 4 em long and of 2 mm i.d.; approximately 30 mg of glass wool can be packed into the same length of a 4 mm i.d. liner. This is, how­

2.3. Sources of Heat

75

ever, nothing like sufficient. When a 2 mm i.d. liner packed with 8 mg wool is used, the sample cools the packing to a temperature near the solvent boiling point. Even this is un­ realistic because the sample extracts the heat merely from the top layer of the packing. Column Packing Material

The materials used for packed columns have densities higher than that of glass wool (a rough average of, maybe, 0.5 g/mL), and thus provide a substantially larger reservoir of heat. The 2 and 4 mm l.d. liners could be packed with some 65 or 200 mg material, respectively. Theoretically this fulfills the above requirements, but if heat is extracted from the top 5 mm only, the heat capacity is still insufficient.

2 ul: of Water

The results are still worse if the above estimates are repeated for water, the solvent consuming the largest amount of heat. Evaporation and heating of 2 IJ.L water to 230°C requires ca. 1.5 cal, cooling 355 mg of silica-type material by 20°. Hence the packing material which can be brought into con­ tact with the sample delivers at best 1 % of the heat required.

High Injector Temperatures?

It seems plausible to solve the problem by setting a higher injector temperature. If 350°C is chosen instead of 250 °C, and 230°C is still the temperature required, cooling can ex­ ploit a temperature difference of 120°, which means that 6 times as much heat is available. This is, however, still far from what is needed.

2.3.3. Heat from Liner Wall

The heat consumed is ultimately provided by the heating block of the injector, but transfer from the heating cartridges to the vaporizing chamber is too slow to be relevant. Only the liner wall is important as a heat reservoir. It provides the heat not only for injections into empty chambers, but also that needed to replenish the heat consumed from the pack­ ing material.

Thickness of the Cooled Glass Layer

Is has been shown that evaporation and heating of 2 IJ.L hexane requires an amount of heat which can be provided by cooling 80 mg glass by 20°. Within the vaporizing zone of the 2 mm i.d.liner, 80 mg glass corresponds to a layer 0.32 mm thick (Figure 83). The analogous calculation for water results in cooling of a 1.4 mm layer. The wall thicknesses of conventional liners vary between 0.7 and 1.5 mm.

2.4. Time Required for Heat Transfer

Two steps of heat transfer must be considered, i.e. transfer within the liner to its surface and that from its surface to the sample. Rough estimation of the heat flow rates is sufficient, because we merely want to obtain an idea of what is possi­ ble.

2.4. 1. Transfer Within the

The temperature gradient within the liner wall has an ex­ ponentialprofile rather than the step profile suggested by

Liner Wall

76

B 2. Solvent Evaporation - Heat Transfer

~Sy<;~i • • : ; ' : ' . ' ~ : ; ~ I

•

•••

Heat transfer from insert wall to the evaporating sample

Column inlet

Figure B3 Sample evaporation in an empty liner. Because there is not enough heat in the gas phase or in a packing. most of the heat must by supplied by a thin surface layer of the liner.

Figure B3. For rough estimation we assume a linear gradient with a 20° temperature drop over the extracted layer. Flow of Heat

Because the thermal conductivity of glass is 2.7x10-3 cat/ern s K [4], the heat flow rate within a 4 cm section (distance between the needle tip and the column entrance) ofthe 2 mm i.d, liner is about 4 calls. The 0.3 cal required for evaporation and heating of 2 ul, hexane can, therefore, be brought to the surface within approximately 140 m8. The larger amount of heat consumed by the aqueou8 8am­ pie (1.5 cal) must be transferred over a longer distance and within a flatter temperature gradient, causing the flow of heat to drop below 1 cal/s: now the transfer takes ca. 3 8.

2.4.2. Transfer Through the Gas Phase

From the surface of the liner, the heat must be brought ei­ ther to the liquid moving through an empty tube or to that deposited on a packing. If the liquid is deposited on to packing material, the rate of heat transfer is still largely determined by transport through the gas phase, because the thermal conductivity of glass is low and the cross section of the fibers is small. Packings might even reduce transport by hindering turbulence. The length of packing holding the sample liquid is 4 mm rather than 4 ern, which further reduces heat transport. When the sample is deposited on to the liner well, no heat transfer through the gas phase is necessary.

Low Thermal Conductivity

Thermal conductivities of gases are low - those of hydro­ gen, helium, and nitrogen are ca. 5,7, or 40 times lower than that of glass (which is, e.g., nearly 300 times lower than that of copper). As the distances involved in transfer through the

2.4. TIme Required for Heat Transfer

77

gas phase are greater than those to be covered within the liner wall, transfer through the gas is normally the rate-de­ termining step. Turbulence?

The rate of transfer of heat through the gas phase cannot be calculated accurately. Firstly, heat is not transported by con­ ductivity exclusively. Turbulence could considerably in­ crease it. Secondly, the sample leaving the needle is not spread uniformly over the cross-section ofthe liner, i.e. heat consumption is unequally distributed.

Heat Transport to the Center of the Liner

First we assume that the heat must be brought to the center of the chamber, as required when a band of liquid is shot through the center of an empty tube. As friction is low, we neglect heat exchange by turbulence. When hydrogen is the carrier gas and a temperature drop of 20° is accepted, the rate of transfer of heat to the center of a 4 cm section of the liner is ca. 0.3 cal/s. It is independent of the liner diameter, because the heat-delivering surface grows as rapidly as the distance over which the heat must be trans­ ported. The thermal flow rates for helium and nitrogen as carrier gases are given in Table 83. Table 83 Rate of transfer of heat from the liner wall to a sample pass­ ing through the center of the liner or to a nebulized sample (fog) distributed throughout the cross section; 4 cm section of the liner; accepted temperature drop, 20°.

Carrier gas

Hydrogen Helium Nitrogen

Rate of heat transfer [calls] Band of liquid Fog 0.28 0.21 0.036

0.92 0.68 0.12

Nebulized Samples

The heat needs to be transported less far if the sample is nebulized and uniformly distributed over the cross-section of the liner. It can be assumed that all the heat is consumed at a distance from the liner wall which encloses half of the liner cross-section. Results are shown in the right column of Table B3.

2.4.3. Residence Time Required for Evaporation

From the calculated thermal flow rates and the amounts of heat consumed by the 2 III samples dissolved in hexane or water, 0.32 and 1.5 cal, respectively, the time required for evaporation can be calculated (Table 84).

Packed Beds

If the sample is deposited on to a packing material, the evapo­ ration zone is shorter and heat transfer approximately eight times slower. Hence even for hexane, times range

78

B 2. Solvent Evaporation - Heat Transfer Table B4 Residence time in the vaporizing chamber required to trans­ port the heat consumed by evaporation of 2 J.1L hexane or water and heating them to 230 °C (20 0 below the set injector temperature).

Carrier gas

Required residence times [s] 2 J.1L Water Fog Band Fog Band

Hydrogen Helium Nitrogen

0.47

2 J.1L Hexane

0.35

2.7

1.2 1.6 8.9

1.6

5.3

2.2

7.3

12.5

41.3

between 1.5 and 10 s. This is certainly not the type of flash evaporation one tends to assume!

2.5. Conclusions

The above data were not presented to provide a description of the evaporation process inside a vaporizing injector, but to enable the following conclusions to be drawn.

Empty Liner - Only after Nebulization

In an empty liner, complete sample evaporation (all material reaches a temperature close to that set for the injector) seems possible in favorable circumstances. For "easy" solvents and using hydrogen or helium as carrier gas, the time re­ quired for heat transfer ranges from 0.3 to 1.5 s. This is avail­ able if the liquid is nebulized, the split flow rate is low, and a 4 mm i.d, liner is used. Nitrogen creates problems, because the heat transfer takes 7.5 times longer than with hydrogen.

This result is in accord with the common experience

that relatively small volumes of samples in the solvents most

commonly used can be analyzed with an empty liner with­

out noticeable problems arising from incomplete evapora­

tion.

No Chance for Samples Moving as a Band of Liquid

There is no possibility of completely evaporating the sample if the latter is "shot" through the vaporizing chamber as a band of liquid. Residence times in the vaporizing zone are in the order of a few tens of milliseconds, i.e. 10-100 times

shorter than required. Wool does not Improve the Availability of Heat

The heat capacity of glass wool packed into the liner is too small to contribute a significant amount of heat to the evapo­ ration process. Hence, if a packed liner is used, the heat must be supplied from the liner wall. Owing to the short section of the liner providing heat, evaporation is several times slower than within an empty tube.

Stopping the Liquid Sample

The only means of prolonging the residence time of the sam­ ple in the injector is to stop the liquid. This means catching

2.5. Conclusions

79

it on a 8urface or within a trap and evaporating it from there.

Temperature in the Vaporiz­ ing Chamber

There is no doubt that the injector temperature indicated on the instrument is correct before injection. During the only moment of importance, i.e. during sample evaporation, how­ ever, it is probably far lower than we think (and largely out of control). An injector temperature of, e.g., 350°C may be required, not to heat the sample to this temperature, but to prevent the temperature in the vaporizing chamber from falling below, say, 150°C. Through the cooling effect, the 80lvent and the 8ample volume could well turn out to be more impor­ tant than the injector temperature 8et.

2.6. Experimental Re­ 8ult8

Kaufman and Polymeropoulos [5] derived an equation for calculating the rate of vaporization. They came to the conclusion that a 5 III injection of hexane into an injector at 200°C would cool the vaporization zone by 65° (i.e. to 135 DC) during ca. 0.1 8. They assumed that turbulence would increase the transfer of heat by a factor of five (open tubular 4 mm i.d. liner). This was estimated from the observed pressure wave, but seems high. If it increased heat transfer by a factor of 2.5 only, the injection zone would have been cooled to a temperature be­ low the solvent boiling point.

2.6.1. Calculated and

Measured Temperature Drop

Measurement by Thermo­ couple

The same authors measured the temperature drop during injection by means of a thermocouple of ca. 0.3 mm diameter. The resulting curve (Figure 84) was indicative of cooling by 80me 40° during 0.3 8. They emphasized that the heat capacity of the thermocouple was excessive for accurately following the rapid change - in fact it was not far below that of the sample, i.e. the thermocouple heated the sample nearly as much as the sample cooled it. The curve, nevertheless, provides interesting information. The dampening effect ofthe mass ofthe thermocouple broad­

,

480

.

;.; II:

470 480

:::>

450

~

440

~

...i!

430 420

v

410 0.0

0.2

0.4

0.6

0.8

1.0

, .2

nilE, •

Figure B4 Estimated (solid Iinel and measured (dotted Iinel tempera­ ture drop on injecting 5 III hexane into a 4 mm i.d. liner at 200 °C (473 °KI. Split flow rate, 500 mUmin. (From Kaufman and Polymeropoulos [5]).

80

B 2. Solvent Evaporation - Heat Transfer ened the curve by a factor in excess of five, because the resi­ dence time of the sample in the vaporization zone must have been short. The split flow rate was 500 rnt/rnln and the fast autosampler from Hewlett-Packard released the sample in less than 30 ms. This dampening effect correspondingly re­ duced the depth of the valley, i.e. the temperature drop. If it corresponded to a factor of at least five, the temperature in

the vaporization zone dropped near to the boiling point of hexane. In fact, it is most unlikely that the sample was completely vaporized.

2.6.2. Measurement of Evaporation Time via Split Flow Rate

Bowermaster [6) determined the duration of sample evapo­ ration from the expansion of the vapors by monitoring the increase in the split flow rate during generation of the vapors. He attached a floating ball flow meter (Rotameter) to the split exit and video-taped the movement of the floating particle.

Flow-Regulated Gas Supply

Because a flow/backpressure regulation system was used, the, volume of the vapors was added to a constant carrier gas flow rate. The split flow rates were found to increase by more than a factor of two. The integrated area under this curve corresponded to the volume of vapor expected to be produced from the volume of solvent injected, indicating' complete evaporation of the sample.

Less than 1 s

Figure 85 shows the increase in the split flow rate after in­ jection of 1 ~L of four different solvents at injector tempera­ tures of 150, 200, and 250 DC. A Hewlett-Packard 7673A fast autosampler was used, and the liner contained glass wool. 125

c

115

E E

<, '05

• •5

~

o

L;:

85

:!:

Q.

III

75

Acetone

Heptane

c: 115

E

<, 105

E

. '5

2"·

~

o

ii:

85

.2

.3

.4

.5

.6

nme. I.e

.7

.6

.0

•

.1

.2

.3

.4

.5

.8

.7

.8

.9

1

Time, lee

Figure 85 Flow rates at the split exit during injection and evaporation of 1 Jll of the four solvents at the injector temperatures indicated. (From Bowermaster [6]).

2.6. Experimental Results

81

For pentane, increase in the split flow rate lasted 0.2-0.3 s; for heptane at 250°C the time exceeded 0.5 s (250°C injector temperature). The results are indicative of evaporation times in the range predicted by the above estimations. The latter were based on the assumption of 2 IJ.L injections, and the times included re-heating of the zone to a temperature 20° below the regulated value. In this experiment, merely the solvent had to evaporate, i.e. temperatures could drop much further.

3. Solvent Evaporation - Visual Observation In complex systems, such as split injection, models and theo­ ries usually involve broad simplifications. They easily omit the decisive aspects. In the end, sometimes complicated theories are produced about rather irrelevant factors. Visual observation of sample evaporation in a device imitating a vaporizing injector, described below and on the CD-ROM in­ cluded with this volume, reduces such errors. 3.1. Experimental

Direct visual observation of 1-51J.L of liquid is difficult. A dye could be added, but for the small amounts of liquid to be observed, color is weak. Fluorescence is preferable because it is more intense when observed in a dark room.

Perylene

Perylene, a polycyclic aromatic hydrocarbon eluted from thin film apolar GC columns at ca. 260°C, has excellent proper­ ties. It has intense yellowish fluorescence upon UV irra­ diation at 366 nm. Fluorescence of solutions is strong, whereas that of crys­ tals is weak; vapor is not visible, maybe because of ex­ cessively low vapor pressure. The melting point, 278°C, is too high to be relevant. Thus, as long as strong fluo­ rescence is visible. sample evaporation is incom­ plete; fluorescence is an indicator of non-evaporated solvent.

Video Taping

In 1992, perylene experiments were performed with direct observation by eye [7]. This was demanding because of the speed. Furthermore, the process seemed different whenever an experiment was repeated, i.e. results could not be checked by collecting observations from many injections. In 1998/99, Biedermann repeated and extended the experi­ ments, using a digital video camera [8]. Some of the videos are shown and discussed on the CD-ROM included with this book. Most ofthe observations cited in the following text are shown there.

82

B 3. Solvent Evaporation - Visual Observation The CD tells the story from the beginning to the end and can be viewed before or after reading the following chapters. As both the CD and the book are, in their way, complete, some duplication could not be avoided.

"Transparent Injector"

The evaporation process was observed in a "transparent in­ jector", i.e. a system imitating the essential parts of a real vaporizing injector. For many experiments, a Pasteur pipette was bent to form a "U" shape (Figure 86). Injections were performed into the wide bore (5 mm i.d.) part, which will, therefore, be called "vaporizing chamber". The narrower bore tip of the pipette served as the outlet ("split outlet"). Flowmeter Syringe

, Soft silicone tube

Pressure regulator

Narrow bore tubin

/-\

Vaporizmq cham r

Split outlet

Figure B6

..Transparent injector" enabling visual observation of sam­

ple evaporation. (From ref. [7]).

Nitrogen was introduced through soft silicone rubber tubing which also served as the septum. The tip of the pipette was attached to thin plastic tubing, which served as a resistance and was connected to a flow meter of the floating 'particle type. There was no column, but the flow rate into the col­ umn is small and hardly influences processes in the injector. The pipette was immersed in a silicone oil bath and heated to the "injector temperature", usually ca. 200°C. The set-up was installed in a dark room and irradiated at 366 nm with a powerful UV lamp. Test Solutions

Perylene solutions were prepared in solvents the important physical properties of which are listed in Table 85. Solu­ tions were usually highly concentrated (ca. 100-1000 ~g/mL). As aqueous solutions are not fluorescent, the experiment is not suitable for the observation of their behavior.

3.2. Liquid Exiting the Syringe Needle

83

Table 85 Physical properties of the solvents used for visual experi­ ments

Solvent

Standard boiling point 1°C]

Surface tension

36 80 40 61 56 78 110 153

18 26 27 27 24 23 29

n-Pentane Cyclohexane Dichloromethane Chloroform Acetone Ethanol Toluene Dimethylformamide

Heat of evaporation lcal/g]

ldvn/cml

85 87 41 59 125 205 86

3.2. Liquid Exiting the Syringe Needle

As a first step, the manner in which the sample liquids left the syringe needle was observed, because of its strong in­ fluence on the evaporation process in the vaporizing chamber.

3.2. 1. Injection through a

When solvent is injected through a cool needle, the liquid leaves the needle as a band (Figure B7) as water runs from the tap. This band formation does not depend on whether injection occurs into ambient air or into a heated chamber. The band had a width of 0.3 rnrn, which exceeds the internal diameter of the needle (0.11 mm) and suggests that it actu­ ally consists of a series of droplets. Sometimes small drop­ lets are split away, as observed for water from the tap when

Cool Needle

Spreading of the liquid

Droplets split from main stream Needle tip

./ .JO"

{~!~;" ~

\

"'\; F.;~ ~~ .... "

,.

4

:':<~.. ,

";"'oIfIIlIl!!:.v'--'

o:lIIt;:,.-=.­, ..•';.-: ~

"';~

Band of liquid

Figure 87 Liquid leaving the syringe needle at ambient temperature. The drawing is not to scale. as the band easily covers a dis­ tance of 50 cm. and the spreading zone has a length of 10­ 20 cm. (From ref. [7]).

the flow rate exceeds a certain minimum. Further ahead, typi­ cally several tens of centimeters from the needle tip, the stream of liquid splits up over a rather short distance; in­ creased resistance in the air slows the droplets. After an­ other maybe 10-20 ern, fluorescence disappears.

84

B 3. Solvent Evaporation - Visual Observation

Mechanical Spray Effect

A small deformation of the needle tip (hardly visible by

eye, but felt by running a finger over it) is sufficient to split

the band of liquid into droplets and to divert their direction.

An example is shown on the CD-ROM. In this way, a minor

irregularity can completely change the behavior of the

sample liquid. If two syringes do not provide the same re­

sults, this is a possible reason. Conical style needles might

be more robust in this respect.

Perhaps optimized deformation of the needle outlet could

achieve a reliable mechanical spray effect, nebulizing the

sample liquid without the help of solvent evaporation inside

the needle (see below).

Fast Autosampler Injection

The most important technique releasing a band of liquid in­

volves the combination of a fast autosampler with a cool in­

jector head, suppressing evaporation inside the needle.

It was introduced to eliminate the drawbacks related to par­

tial evaporation inside the needle (Section A), but created

problems associated with stopping a band of liquid in the

vaporizing chamber.

Injection of samples in high-boiling matrices or through a

short needle combined with a cool injector head can also

result in band formation.

3.2.2. Injection through a Hot Needle

If the liquid is injected through a hot needle into ambient air,

a fog is observed instead of a band; the liquid is nebulized

at the needle tip. A core of bright fluorescence, usually 1­

5 mm long, leaves the needle. Further away, the fluorescence

rapidly loses its intensity. It forms a cone, sometimes with

rather clear boundaries at the side, other times diffuse. Of­

ten it seemed that fine droplets were squirted away for a

distance of 5-20 mm (a kind of spreading jet).

Cone of Diffuse Fluorescence

Shapes and dimensions of the cone vary. 2 III of a chloro­

form solution injected into ambient air on occasion produced

a broad cone, visible up to 2 cm from the needle exit and

3 cm wide at the front, another time a sharper cone, visible

for 4 cm and only 4 mm wide at that point. 5 III of the chloro­

form solution covered a distance of 8 cm and reached a width

of 4-6 cm. Solutions in other solvents behaved similarly.

The fog of nebulized perylene solution was visible for 50 ms

at most. Its disappearance suggests rapid solvent evapora­

tion, consuming heat from ambient air.

Extrapolation to what happens inside the vaporizing cham­

ber might not be straight-forward - there is little room for

expansion, i.e. less gas from which to extract heat, but the

temperature is higher.

Thermospray Effect

The sample enters a pre-heated syringe needle at a high ve­

locity (it passes through it in ca. 5 rns), Bubbles of solvent

vapor on the needle surface build up high pressure and ex­

pel the liquid through the center of the needle (Figure B8).

3.2. Liquid Exiting the Syringe Needle

85

Pressure inside the needle causes the liquid to overheat (in­ creased boiling point). On leaving the needle, the liquid ex­ plodes and fragments into small droplets. Partial evapora­

tion inside the needle turns the solvent into a propel­ lant and produces a thermospray effect. Needle tip inside hot injector

Vapor formed along the internal needle wall Vapors pressurize the sample liquid in the center of the needle

Overheated sample liquid boils violently

Figure 88

Thermospray of the sample liquid after hot needle injection.

(From ref. [8]).

Slowed Droplets

Friction with the gas rapidly slows the resulting particles to the gas velocity. While suspended in the gas, there is enough time for the solvent to evaporate. When solvent evaporation is complete, the temperature of the droplets in­ creases to that of the injector and the solutes evaporate.

Limits to the Nebulization

Under critical conditions the first part of the sample liquid is nebulized whereas the rest leaves the needle as a band. A 5 III injection of a DMF solution through a short needle heated to 250°C is an example. Solvent evaporation extracts so much heat from the needle that the temperature drops to the solvent boiling point, which stops nebulization.

Thermal Capacity of the Needle

A 51 mm 26S gauge needle has a thermal capacity of ca. 7 mcal/degree. Complete evaporation of 2 III hexane con­ sumes an amount of heat cooling the needle by 50°. For va­ porization of the same amount of water, cooling by 217° is calculated - but the needle is hardly sufficiently hot and evaporation remains incomplete. Such a calculation does not properly reflect reality, but shows that heat extraction by solvent evaporation is substantial and sometimes sufficient to stop nebulization.

86

B 3. Solvent Evaporation - Visual Observation

Experimental Data

Table 86 summarizes observations for examples considered either typical or borderline. Injection of solutions in the sol­ vents commonly used (b.p. up to 100°C) through a 51 mm needle at 220°C always nebulized the liquid (volume tested, 5 Ill). Toluene (b.p. 110°C, injection no. 4) was nebulized, but not DMF (153°C, no. 5). Atthis point it should be remem­ bered that in real injectors the needle temperature often re­ mains below the regulated injectortemperature, because the head of the injector is cooler than its center (Section A8.2). If the real needle temperature only reaches, e.q., 130°C (rapid injection, short needle, cool injector head), even manual injection can lead to band formation.

Length of Heated Needle

Heating the needle to 250 °C over a length of merely 1.5 cm still nebulized 51ll of the low-boiling solvents, but 5 III tolu­ ene left as a band (no. 10); 300°C resulted in the borderline situation (no. 11): initial nebulization turned into band for­ mation.

Thickness of Needle Wall

Some experiments were performed with a 32 gauge nee­ dle (for manual on-column injection). Being of 0.23 mm o.d.

Table B6 Sample liquid exiting a heated needle under different conditions. Length of the needle heated in the metal block; times the band or spray was visible (number of 4G-ms-frames); observed spray (with length of the cone) or band formation. (From ref. [8]).

No. Solvent

Volume [Ill]

Needle temp. lOCI

length heated [mm]

lime visible [frames]

Spray or band, length of cone

1 2 3 4 5 6 7 8 9 10 11 12 13

26S gauge needle Chloroform Chloroform Chloroform Toluene DMF Hexane Dichloromethane Chloroform Ethanol Toluene Toluene DMF DMF

2 2 5 5 5 5 5 5 5 5 5 5 2

220 220 220 220 220 250 250 250 250 250 300 300 350

50 50 50 50 50 15 15 15 15 15 15 15 15

1 1 2 2 2 2 2 2 3 2 2 2 1

Spray 4 cm Spray 2 cm Spray 8 cm Spray 5 cm Band Spray 4 cm Spray 2.5 cm Spray 4 cm Spray 8 cm Band Spray/band Band Band

14 15 16 17

32 gauge needle Hexane Hexane Chloroform Chloroform

2 5 2 3

250 250 250 250

15 15 15 15

1 2 1 1

Spray 8 cm Spray/band Spray 5 mm Band

3.3. Three Scenarios of Evaporation

87

and 0.11 mm l.d., its thermal capacity is five times lower than that of the 265 gauge needle. 5 ~L of hexane cooled it to such an extent that part of the liquid left as a band (15 mm section heated to 250°C), approximately 3 ~L being the limit for complete nebulization.lfthe whole needle wall was cooled to the boiling point of hexane, the amount of heat extracted was sufficient to vaporize almost half the hexane. This illus­ trates the importance of the thermal capacity of the needle.

3.3. Three Scenarios of Evaporation in an Empty Vaporizing Chamber

Observations related to sample evaporation in an empty liner are described in three scenarios, as this enables us to bring the many observations together in simple models. The real process often does not strictly follow one scenario, but maybe a mixture of two.

3.3. t. Scenario'­

When injected through a hot needle, the sample liquid in a volatile matrix is nebulized and spreads across the liner bore (Figure 89). The microdroplets are rapidly slowed down to the speed of the carrier gas and move like fog in the wind.

Nebulization

Nebulization of liquid Droplets slowing down

~':I+f+;~-

Fog of small droplets turning into vapors (disappearing fluorescence)

Figure 89 Scenario 1: flash evaporation after nebulization of the liquid at the needle exit.

Evaporation in the Gas Phase

Fluorescence disappears 2-15 mm from the needle exit. This is indicative of complete solvent evaporation. Vaporization occurs from droplets suspended in the gas phase, essentially without contact with the surfaces of the vaporizing chamber.

Stable Fog

When perylene concentrations are high, the fog remains vis­ ible beyond the end of solvent evaporation. In the absence of gas flow it is stationary for many minutes whereas a gas flow discharges it like wind blows away clouds or smoke. No fluorescent material remains on the liner wall, confirming the stability of the aerosol.

88

B 3. Solvent Evaporation - Visual Observation

During solvent evaporation the fine droplets of dilute solu­ tions shrink by a factor of ca. 10 (0.1 % solute material in the solution) and form an aerosol. These particles are too .mall to ••ttl. by gravity and too larg. to diffu•• towards surfaces where they might condense. Transfer as an Aerosol

The results of the visual observations are realistic in so far as perylene is widely analyzed in GC, e.g. as internal stand­ ard for the determination of benzopyrene. N.buliz.d p.ry­ I.n. can b. introduc.d into the column under condi­ tions far from providing complete evaporation, namely as aerosol. In fact, thermospray injection does not presuppose complete evaporation in the injector. Transfer as an aerosol at low temperature is attractive for the analysis of thermola­ bile compounds, as further discussed in the context of sol­ ute evaporation (Section 84.1).

3.3.2. Scenario 2 - Band of Liquid

Scenario 2 is observed in the ab •• nee of .olv.nt .vapo­ ration in.id. the n••dl., such as after fast autosampler injection, For the visual experiments vaporization in the nee­ dle was suppressed by introducing the needle into the heated zone by merely 1 ern,

V-Shaped Pipette

In the U-shaped Pasteur pipette, the band of liquid rushed' through the "vaporizing chamber", around the bend of the tube, and upwards through the "split outlet" (see videos) at a speed near the limit of what can be followed by eye. Often the liquid left the "split outlet" without any .ign. of .vapo­ ration, i.e. passed unaffected through a 16 cm long bent tube at 200°C. The liquid did not touch the liner surface. On other occasions the liquid was slowed in the narrow-bore "split outlet", accumulated to form an oblong droplet, and fell back into the bend of the pipette, where it contracted to a round droplet. Sometimes some liquid was blasted out, or the whole ball jumped back into the vaporizing chamber by up to 4 cm. This part of the proc••• was out of control. It hardly ever repeated itself. Evaporation took 1-3 s (CD-ROM), but this duration also varied substantially when similar in­ jections were repeated. The transfer of the process into a real vaporizing injector is shown in Figur. 810: the sample liquid is "shot" past the column entrance.

Negligible Dependence on Gas Flow Rate

The gas (split) flow rate had no visible influence on the move­ ment of the liquid. Ev.n und.r ...plitl..... condition., i.e. virtually without gas flow, the stream of fluorescence of­ ten passed at high speed through the "split outlet". This is no longer a surprise when velocities are compared, because the liquid leaving the needle moves at least 100 times faster than carrier gas at a high split flow rate, i.e. it makes little difference whether the liquid is injected into a gas stream or a stationary gas.

3.3. Three Scenarios of Evaporation

89

Syringe needle

Band of liquid Cushion of vapor repelling liquid from wall

Split outlet ) Column

Figure 810 Scenario 2: the sample liquid is shot through the vaporizing chamber as a band of liquid.

Extra Flow Rate Through Split Outlet

Solvent evaporation results in rapid expansion of vapor. In the experimental set-up used, the pressure wave gener­ ated discharged this extra volume through the split outlet and the flow meter, the vapor primarily pushing gas ahead of itself. This extra flow was monitored by the flow meter, as shown by Bowermaster [61. After nebulization of the samples the floating particle knocked audibly against the top of the flow meter, but when the liq­ uid left the needle as a band, the response of the flow meter was small, confirming that hardly any liquid had been vaporized. Moving at about 20 mis, the liquid passed through the pipette in such a short time that the amount of heat transferred corresponded to a few percent of that re­ quired for complete solvent evaporation. Hardly any sol­ ute material was vaporized, since solutes boiling above the solvent boiling point are retained by the cool liquid.

Visual Experiments by Qian et al.

In a specially designed apparatus, Qian et et. [91 observed how heptane and water (5 Ill) left the needle of a Hewlett­ Packard 7673A fast autosampler. The needle was intro­ duced in 100 ms. Because the sample liquid is not withdrawn into the barrel of the syringe (as is usual in manual injection), it leaves the needle as soon as the plunger starts moving. Injection lasted 61 rns. During the first 5 ms the velocity ofthe liquid increased to 10.5 m/s. It then decreased to 5.3 mls during the following 15 rns, and then remained constant for 38 ms. This is slower than normal manual injection. During the acceleration period later expelled liquid caught up with the earlier liquid and formed a drop up to 1 mm in diameter'. Subsequently the sample formed a band of Iiq­

90

B 3. Solvent Evaporation - Visual Observation

uid with an oscillatory motion, because of vibration of the

needle during depression of the plunger.

After covering some 8 ern, the band disintegrated into drop­

lets. Those of water were larger than those of heptane (ca.

370 and 250 11m, respectively). In the following step, growth

of the droplets by coalescence counteracted evaporation; at

265 "C, water droplets grew, whereas droplets of heptane

became smaller.

For hexane injections at 300 "C, the band of liquid disinte­

grated after ca. 3 mm. After 7 mm it formed a jet ca. 3 mm

wide and the droplets disappeared 1 cm from the needle tip.

In this experiment, the whole environment of the inserted

needle was heated to 300 °e, i.e. much above the real tem­

perature of the injector head. The process corresponded to

the nebulization described above and confirms that fast

autosamplers can suppress solvent evaporation in the nee­

dle only if the injector head is rather cool.

3.3.3. Scenario 3 - Liquid Splashing on the Liner

Wall

Injection of a dimethylformamide (DMF) solution revealed a

third mechanism. Owing to the high-boiling point, there was

no evaporation inside the needle and so the liquid left as a

band even when the needle was hot (200-220 "C), The liq­

uid flew to the liner wall. splashed on its surface, and

spread to a spot several millimeters in diameter (Figure B11)..

From there, the solvent evaporated smoothly and rather

slowly (1-3 s).

Depending on the direction ofthe "shot", the spot to which

the liquid was transferred was located 1-3 cm below the tip

of the syringe needle. In the 5 mm i.d, Pasteur pipette used,

occasionally even a "shot" to the bottom of the injector was

observed.

Syringe needle

Liquid deposited on surface Slow evaporation from surface

Only vapors reach column entrance

Split outlet ) Column

Figure 811 Scenario 3: The sample liquid is transferred to the liner wall and smoothly evaporates from there.

3.3. Three Scenarios of Evaporation

91

Repulsion of Liquid from the Hot Surface

Scenario 3 is what is often expected to commonly occur in­ side the vaporizing chamber even with solutions in volatile solvents. This neglects the difficulty the liquids have touch­ ing hot surfaces. Repulsion from the hot wall was observed for the bands of liquid negotiating the bend at the bottom of our U-shaped device without touching the wall (CD-ROM). Even with "shots" directed perpendicularly on to the wall (by use of a syringe with a side port hole), no contact with the liner wall could be provoked (with the exception of ethanol and, of course, DMF).

Vapor Cushion

Liquids do not touch a solid as long as the surface tempera­ ture of the latter exceeds their boiling point because a cush­ ion of vapor, formed between the hot surface and the liquid, repels them (Figure 812). The water droplets "dancing" on a hot cooking plate illustrates the phenomenon nicely. The closer the liquid approaches the hot surface, the more heat is transferred, the more vapor is formed, and the stronger is the repulsion.

JJ1= .

Heal

~

Evaporation of solvent accelerated upon approaching hot surface

Cushion of solvent vapor

repels the droplet

Figure 812 Sample liquid repelled from hot surfaces by the vapor formed between the surface and the liquid. (From ref. [10])

Cooling the Surface

To touch a hot surface, a liquid must cool it to a temperature corresponding to its boiling point. The cooled layer can be thin. Once the liquid wets the surface, cooling is no longer a problem as evaporation extracts heat efficiently.

Prerequisites for Scenario 3

This explains why scenario 3 is observed only 1 at relatively low injector temperatures, 2 with rather high-boiling samples, and/or 3 with solvents consuming a large amount of heat on evaporation, i.e. strongly cooling the surface.

Solvents

Dimethylformamide (b.p. 153°C) wetted the liner wall because cooling of a thin surface layer by 50-70° was obvi­ ously no problem. Also ethanol often touched a liner at about 210°C, hence cooled it by 130°. This is probably a conse­ quence of its large heat of vaporization, i.e. the consumption of large amounts of heat per unit volume of liquid evapo­ rated. For toluene, cooling by 100° would have been suffi­ cient, but no wetting was observed.

92

B 3. Solvent Evaporation - Visual Observation

Kinetic Energy Directed toward the Liner Wall

Whether or not the liquid touches the liner wall also depends

on the angle of injection and the kinetic energy of the

liquid "shot" against a surface (speed and mass): if the

angle is acute, relatively weak repulsion is sufficient to pre­

vent contact; if, on the other hand, a large droplet approaches

the surface at high speed and perpendicularly, the liquid has

a better chance of overcoming the hindrance to wetting of

the surface.

Incomplete Transfer

More often than not all the sample liquid was transferred to

the same spot; other times it was split into several portions

and evaporated from several sites. Sometimes small drop­

lets became detached from the bulk. Below a critical

mass they are unable to cool the wall surface and are re­

pelled. In these instances, transfer of the liquid to the liner

remained incomplete.

Maximum Sample Volume

Generally, 1 III of a liquid such as DMF wetted a spot on the

liner wall without further movement. With 2 Ill, the liquid

accumulated on the lower end of the spot and formed a tear,

without moving significantly. With 5 Ill, the tear flowed

down the wall and normally passed the position of the col­

umn entrance, i.e. covered more than 3-4 cm.

For an empty liner, the volume of sample that can safely be .

retained on the wall is limited to 2-3 !J.L. It has not been in­

vestigated whether wettability of the surface is an important

issue.

3.4. First Conclusions

When a GC method is being developed, the solute is the

center of interest. This neglects the role of solvent evapo­

ration:

1 it is the first obstacle to be overcome (no solute evapo­ ration before the solvent is vaporized); 2 it determines how the sample liquid leaves the syringe needle; 3 it determines the movement of the liquid within the va­ porizing chamber; and 4 the site of solute evaporation.

Solvent Evaporation as Critical Step

High Boiling Samples Not Necessarily More Difficult

High-boiling samples, i.e. undiluted or in high-boiling sol­ vents, are not necessarily difficult to evaporate, because they tend to be transferred to the liner wall and to evaporate slowly and smoothly from there. Solvents often considered easy may be more difficult matrixes since the movement through the vaporizing cham­ ber is sometimes violent and hard to control.

Needle Temperature

For solutions in volatile solvents, the most common type of sample, the needle temperature during injection is the most important factor. Hot needles cause nebulization, which facilitates further evaporation. Use of cool needles eliminates the problems of evaporation inside the needle, but creates those related to the rapidly moving band of liquid. Hence

3.4. First Conclusions

93

the processes inside the needle are linked with those in the vaporizing chamber. Autosampler ;t Automated Manual Injection

Injection by means of a fast autosampler is more different from manual injection than previously thought. As a conse­ quence, methods must be re-optimized and re-validated when changing from manual to automatic injection or vice versa.

3.4. 1. Fate of Sample Liquid "Shot" to the Bottom of the Liner

From the observations described above we must assume that samples injected into empty liners by fast autosamplers are "shot" to the bottom of the vaporizing chamber. This is par­ ticularly important for splitless injection, because the use of an empty liner is still state of the art (use of packings fre­ quently creates severe problems). The consequences are not obvious and depend of numerous factors.

Shape of the Base of the Chamber

The shape of the bottom of the vaporizing chamber largely determines how the liquid reacts on hitting the surface. Fig­ ure 813 shows three possibilities. When the liquid hits a horizontal surface, it is rejected or stopped, as will be dis­ cussed later. If the shape is conical, the liquid is likely to pass on into the column attachment area or the split outlet. This funnel shape can be in the metallic parts (drawing 2) or in a liner with a constriction at the bottom ("goose neck", drawing 3). If the liquid passes on, it is lost to the analysis, because it has no chance of returning to the column entrance. For this reason, only the horizontal surface is considered fur­ ther.

1

2

3

Horizontalbottom, Funnel-shaped Liner with narrow orifice bottom, wide orifice constriction Figure 813

Shape of the bottom of the vaporizing chamber; three possi­

bilities.

Three Scenarios

To imitate a horizontal bottom, a metal disk was introduced into a glass tube which was closed below it. There was no gas flow leaving the chamber, reflecting splitless injection. The observations shown on the CD-ROM confirmed the three scenarios 'shown in Figure 814. It is assumed that the col­

94

B 3. Solvent Evaporation - Visual Observation

umn entrance is positioned ca. 5 mm above the bottom of the chamber.

....~ \I ~\ ·.\"1

Syringe needle

'/1;~ Rejected liquid returns above column entrance

<;

s~~~~e Cn outlet~'~ Split

Injection according to scenario 2

,.:.... __

'ii,:

3 Liquid absorbed by septum particles. Only solvent vapor returns above column entrance

2 Liquid evaporates at bottom of vaporizing chamber. Only vapor of solvent and volatiles returns to column entrance

Figure 814

Three possibilities of sample evaporation after the liquid is

"shot" to the bottom of the vaporizing chamber. (From ref.

[10).

1 Rejection

As a first possibility, the liquid is rejected back towards the center of the vaporizing chamber, often as high as 5 ern, i.e. close to the upper end of the liner. When it explodes there (delayed evaporation) or splits into smaller droplets for other reasons, the small droplets may evaporate there before they drop back to the bottom, suspended by the gas, or driven upwards by expanding vapor. This gives all sam­ ple components (including high-boilers) a chance of enter­ ing the column. Larger droplets fall back to the bottom. Sol­ utes might, however, have a second chance upon another explosion.

2 Evaporation at the Bottom

During other injections, the liquid remained at the base of the vaporizing chamber and formed a ball, often rotat­ ing at a high speed (2 in Figure 814). It danced nervously on the hot surface, as if it were hurt, and evaporated over a pe­ riod of several seconds. Under such conditions, only the vapor of solvent and vola­ tile solutes expands backwards above the column entrance such that it has a chance of being swept into the column; the high-boiling material (e.g. perylene) does not evaporate from the ball of liquid since the temperature of the latter is the

3.4. First Conclusions

95

solvent boiling point. When vaporized at the end, the vapor volume is too small to expand above the column en­ trance, i.e. at best a small proportion of it has a chance of being analyzed. 3 Absorption in Septum Particles

A third scenario involves a mechanism which also leads to the complete loss of the solutes. Septum particles or other dust accumulated at the base of the chamber are readily cooled to the solvent boiling point and absorb the sample liquid (as observed for glass wool, see below). This prevents rejection into the vaporizing chamber. The vapor from the solvent and the volatile solutes is re­ leased first and expands to a volume corresponding to maybe half of the vaporizing chamber. In splitless injection, it ex­ pands far above the column entrance. Finally the higher-boil­ ing solutes also evaporate, but their vapor volume is small and so they remain at the bottom of the chamber, again with little chance of entering the column. High-boiling sol­ utes do not evaporate at all from strongly retaining material, such as septum particles (consisting of silicone rubber, simi­ lar to stationary phases).

4 Low Column Entrance

Losses in Scenarios 2 and 3 are related to solute evaporation below the column entrance. It might seem preferable to in­ stall the column lower, such that the solutes evaporate above its entrance. Surfaces of glass are preferred to those of metal because at the final stage of evaporation, the droplets containing only higher-boiling material are deposited on to them, Often a goose neck liner is used with the column entrance posi­ tioned within the constriction (Figure 815). The band of liq­ uid was rejected as described above and whirled through the chamber. At the last stage of evaporation, droplets formed which tended to glide towards the orifice of the constriction and probably occasionally dropped into it. Sometimes some liquid was driven straight into the column or past the col­ umn into the outlet channel.

1. liquid jumping around

2. Droplels gliding towards orifice, falling into or past the column

3. Solutes filtered through septum particles

Figure B15 Band of liquid hitting the bottom of a goose neck liner, and two subsequent scenarios.

96

B 3. Solvent Evaporation - Visual Observation Another problem concerns the dust and septum particles (3 in Figure 815). Now all the solute material must pass through this possibly highly retentive material. High-boilers are likely to remain there.

Wetting Column from Outside

When a piece of fused silica capillary was mounted in the vaporizing chamber, imitating the column inlet protruding into the chamber, sample liquid sometimes wetted it from outside, perhaps because its thermal mass is small. In the same way as any other high-boiling solute, the perylene de­ posited on the polyimide had no chance of entering the column since this would have presupposed upwards flow.

3.5. Stopping the Sample Liquid

To improve sample evaporation text books commonly rec­ ommend mixing of the sample liquid with carrier gas. Since the introduction of the fast autosampler, however, the first requirement has been to stop the sample liquid at the appropriate site within the chamber.

Retention of Liquids on Surfaces

The. most reliable method of stopping a liquid and holding it firmly involves deposition on to a surface. This is difficult, however, as anyone knows who has seen water hovering above a hot cooking plate. Liquids are rejected from hot sur­ faces, divide into smaller droplets, or nervously glide around with hardly any resistance, maybe a millimeter above the surface, carried by a cushion of vapor.

Cooling the Surface

Deposition presupposes cooling of the surface to the boiling point of the liquid. As described above, the liner wall is touched only if the boiling point of the liquid is not too far below the injector temperature; a solvent such as hexane is unable to make contact with a surface of massive glass at, e.g., 200°C. The same applies to surfaces positioned in the way of the liquid (obstacles) to stop its movement. Numerous liners with obstacles have been suggested in the literature. The claimed improvement was usually demon­ strated with reproducibility data, but only visual observations really enable their classification.

3.5. f. Liner with Baffles

Baffles protruding well beyond the center of the tube (Figure B16) should preclude direct passage of the liquid from the needle to the column; the liquid is supposed to splash against a surface. This is not true, however. A band of fluorescence rushed around the baffles almost as easily as through a straight tube. The beam of light curving around the baffles, performing perfect slalom without touching the surfaces, is quite spectacular (see CD-ROM).

3.5.2. Cup or Jennings" Liner

In 1975, Jennings [11) proposed a split injector "in which injection, vaporization, mixing, expansion, and splitting oc­ cur in a glass insert", which tells us that it was not really conceived for the purpose in question here. A schematic drawing of a present-day version of the liner is shown in

II

3.5. Stopping the Sample Liquid

;t

97

Septum

Injector insert

Baffles Funnel

Cup

II Column entrance Liner with baffles Figure 816 Liners with baffles and cup.

Cup liner

Figure 816. The liner is often called an "inverted cup liner", although the cup does not seem to be inverted. The cup forces the gas flow to change its direction twice: after entering it must flow upwards before returning again to the normal direction of flow.

Performance

When the corresponding liner from Hewlett-Packard/Agilent was tested, most of the liquid bounced against the hori­ zontal surface of the funnel and usually rebounded a dis­ tance of up to a few centimeters. The larger droplets, at least, fell back and then "danced" around the center cavity without touching the surface, usually rotating at high speed. This could last between a fraction of a second and several seconds. Sooner or later some liquid dropped into the cup and again formed a ball, performing the same dance in the hot environment. Often this droplet exploded. Liquid splashed against the bottom surface of the funnel, escaped the device, and dropped to the bottom of the liner, flying past the column. With a (split) flow rate of 40 mLJmin, some liquid regularly broke through the device driven by the first thrust of the in­ jection. With a small gas flow (splitless injection), on the other hand, breakthrough was only occasional.

Vaporization of Perylene

The perylene was released at the end of solvent evapora­ tion. This was sometimes observed as bluish light, probably from perylene adsorbed on the glass surface. The (cool) liq­

uid retained the perylene up to the end of solvent evaporation, analogous to solvent trapping. Release occurred 0.5-3 s after injection. This is important for understanding phenomena related to the deviation of the true split ratio from that pre-set, and to the linearity of splitting. It tells us, for instance, that changes of flow rates during sol­ vent evaporation may not affect the transfer of solutes such as perylene.

98

B 3. Solvent Evaporation - Visual Observation

3.5.3. Glass Bead Liner

Bayer and Liu [121 proposed a liner with a glass ball posi­ tioned above three baffles, as shown in Figure 817. The ball, positioned in the top section of the liner to leave ample mixing space below the restriction, leaves three narrow chan­ nels for the sample to pass. This liner was proposed for use with stop flow split injection (Section C10.2.4.).

A

t

A- A'

Figure 817 Liner with a glass bead positioned on three baffles. (From Bayer and Liu [12])

Liu and Xin [131 provided data on the effect of different posi­ tions of the bead in the liner. There is a shortage of space for housing the expanding sample vapors, because there is no split flow. Positioning the bead at the top of the liner pro­ vides the smallest room for expansion and resulted in some loss of volatiles as a result of backflow into the gas supply lines or through the septum purge. Positioning nearer to the bottom, however, caused the standard deviations of the re­ sults to increase several fold. The authors recommended that the bead should be positioned half way up the liner and that the maximum sample size be 0.5 !!L. 3.5.4. eye/oliner

The "Cyclosplitter" from Restek (Bellefonte, USA, Figure 818) is a 4 mm i.d. glass tube which contains a spiral, 20 mm

Funnel First site of evaporation

Inverted cup Center tube

Main site of evaporation -

Cycloliner

laminar liner

Figure 818

Cycloliner and Laminar liner from Restek.

3.5. Stopping the Sample Liquid

99

high and with 6 turns, similar to winding stairs, made of glass.

The sample is forced to pass through a channel of maybe

1 mrnz cross section and at least 5 cm long.

The liquid usually found little difficulty curving through

the spiral and dropped to the bottom of the chamber [14].

Sometimes a droplet remained hanging in the spiral, pre­

sumably because it touched a particle deposited there. Chlo­

roform solutions often behaved differently: most of the liq­

uid remained hanging on the wall as small droplets.

3.5.5. Laminar Liner

The "laminar liner" from Restek could be described as an

"inverted cup" liner, with the cup, or rather a Champagne

glass, really being inverted. The sample passes through an

orifice resembling a funnel and then through the narrow

space (less than 0.5 mm) between the cup and the liner wall

(length, 35 mrn), At the bottom of the liner, it changes direc­

tion and flows upwards through the narrow space between

the inner surface of the cup and the center tube. After return­

ing to half the height of the liner, it is re-directed through the

center tube to the column.

Test Results

According to the perylene experiment, the "laminar liner"

was the most successful of the liners with obstacles. It re­

sulted in complete evaporation even for a sample vol­

ume of 5 ~L.

Frequently the sample was stopped after passage through

the "funnel" and a ball of liquid "danced" on the bottom of

the inverted cup ("first evaporation site" in Figure 818). En­

trance of the liquid into the narrow space between the in­

verted cup and the liner wall is hindered by vapor formation.

After most injections part of the liquid made its way to the

bottom of the liner into what could be termed the "main

evaporation chamber". Usually the liquid accumulated there

as a ball and moved around without touching surfaces until

the solvent had evaporated. No liquid was ever observed to

go further. Apparently there was insufficient force to move

the liquid into the next narrow space.

3.5.6. Metal Liner

Restek introduced the first metal liners, enabled by their pro­

prietary procedure for deactivating metal surfaces ("Silco­

steel"). It contains a restriction at the bottom and two obsta­

cles which look like short pieces of bolt (Figure 819). These

force the sample through a narrow channel along the inter­

nal liner wall. The top surface of the obstacle is designed

such that the sample is likely to be rebound.

Coiled channel

01---­ Bottom

Figure 819 Deactivated steel liner from Restek.

Top

100

B 3. Solvent Evaporation - Visual Observation

New Aspects

Metal liners introduce a number of new features which have not yet been evaluated. The liner is unbreakable (important for especially wild chromatographers). 2 It can be sealed against the injector body by use of a metal ferrule, circumventing problems presently expe­ rienced with graphite ferrules (Section E2.2). 3 Metal surfaces might behave differently during thermospray injection of" dirty" samples. If nebulization separates charges, the charged droplets more easily cre­ ate counter charges, which might attract the droplets to the wall more strongly. 4 Metal enables fabrication of narrower channels than glass. 5 The stability of the deactivation against attack by sam­ ple by-products is unknown. 6 The liner is not transparent, depriving us of the possi­ bility of visually checking whether the liner is clean or a packing well positioned. (7 . The liner cannot be tested by the perylene experiment...)

3.5.7. Summary - Stop­ ping Liquid with Obsta­ cles

In the past, liners with obstacles were conceived to improve

evaporation by changing flow directions; it was assumed

that the liquid would splash against surfaces. Visual experi­

ments showed, however, that a band of liquid readily passes

through many curves and even performs slalom.

Horizontal Surfaces

When the band of liquid hits a surface perpendicular to its

direction of flight, it is often stopped and remains there, typi­

cally contracting to a ball which dances on the hot surface.

More frequently it is rejected and flies backwards by several

centimeters. Hence, a horizontal surface can effectively

stop the liquid. but not hold it in place and ensure a

reproducible process.

Narrow Channels

The most effective obstacles involve narrow channels. Con­

trary to previous expectations, success is not based on achiev­

ing evaporation as a result of short distances for heat trans­

fer (e.g. in the cycloJiner) - the velocity of the liquid is so

high that heat transfer remains insignificant. Their effective­

ness originates from stopping the sample liquid [141.

Vapors repel the liquid, because entering a narrow channel

accelerates evaporation.

Shape of the Orifice

Much depends on the geometry of the orifice to the narrow

zone. If it has the shape of a funnel, approaching liquid can­

not return and is driven deeper into the channel. To be re­

jected, the liquid must be given an easy route of escape

by gliding away sideways. Hence the narrow channel should

protrude slightly into a wider space.

Well-shaped narrow orifices have the effect of a sieve: gas

and vapor pass, whereas droplets are stopped.

3.5. Stopping the Sample Liquid Trapping the Liquid

101

The most successful liner is designed as a trap for liquid. It invites the band of liquid into a narrow zone by means of a funnel-shaped entrance, then offers an open space for evapo­ ration (the trap), and prevents escape by hindering access to the narrow channels forming routes forward or backward (Figure 820). The laminar liner is close to putting this con­ cept into practice.

Inviting

entrance

Narrow

channel

Rejecting entrances to narrow channels Space for evaporation

Figure 820 Section of a hypothetical liner trapping the sample liquid in a cavity between narrow channels with rejecting entrances.

3.5.8. Wool

In 1977, Schomburg et al. [151 proposed packing the liner with glass wool to improve split injection. No explanation on the effects involved were given. Only in 1991 has the perylene experiment provided the answer: after injection with band formation, wool is probably the most effective means of holding the liquid.

Sucking up Liquid

A small amount of glass or fused silica wool is sufficient to change the evaporation process fundamentally. The liquid seems to be sucked up by the nearest fibers (irrespective of the solvent) and is firmly held there. A zone of maybe 3 mm in diameter remained intensely fluorescent until the solvent had evaporated. No liquid or fog left the plug at the bottom end, i.e. retention was complete.

Low Thermal Mass

The sample is in contact with a few milligrams of wool. The key is not the amount of heat the wool is able to transfer to the sample (as was usually suggested) but the contrary: the thermal mass of the fibers is so small that a small amount of liquid immediately cools them to the sample boiling point. Vapor cushions formed in the hotter neighborhood guide the rest of the liquid to the same spot. The liquid re­ mains hanging in the framework, which gives it virtually unlimited time for evaporation and suppresses the forma­ tion of an 'aerosol.

102

B 3. Solvent Evaporation - Visual Observation

Required Size of the Plug

5 ~L of sample was spread within a region of a few millimeters in diameter, i.e. not necessarily over the whole cross section of the plug, and no deeper than 4 mm. A plug ca. 5 mm high is all it takes for reliable retention of up to 5 ~L of liquid. The wool does not need to be dense, but there should not be large gaps through which the band of liquid could be shot. Any additional wool merely aggravates problems of adsorp­ tion or degradation of labile solutes.

Evaporation from a Surface

The drawbacks of using glass or quartz wool are high adsorptivity and chemical activity, even if deactivated. Activity was often related to a large surface area. The fibers (ca. 5 urn in diameter) are not porous, however. 1 mg of wool consists of fibers with a surface area of hardly 2 cm 2 . This is less than the surface of the liner wall - that of a 4 cm section of a 4 mm i.d. liner is 5 cm 2. Non-deactivated faces of broken fibers are not relevant either. Because adsorption in a liner containing deactivated wool can far exceed that in an empty, raw glass liner, another ex­ planation must be found. The perylene experiment suggests that the closeness of contact with the surface is the important factor. As solvent vapors repel the liquid, samples in volatile solvents do not touch the liner wall or the surfaces of massive obstacles, whereas surfaces of wool and other : light packings are wetted.

Wool for Nebulized Sam­ ples?

The effect of the wool is entirely different when the sample is nebulized above it. Sprayed liquid passes through the wool. Evaporation is not improved. nor are particles retained. This also suggests thatthere are no relevant interactions with the solutes, i.e. the packing neither causes adsorption nor degrades labile solutes. This explains established experience that the negative effects of wool strongly depend on how the sample is injected.

3.5.9. Glass Frits

Wool packings are not homogeneous and not well repro­ duced - at least optically. Frits seem preferable in this re­ spect. They are made from glass particles sintered at high temperature.

Excessive Thermal Mass

As shown on the CD, the behavior of liquid approaching a frit is different. The liquid dances above the frit and wets it at best towards the end of solvent evaporation. Actually it often performed wild, non-reproducible movements, some­ times rebounding up to the top of the liner, as if the frit sur­ face were a solid wall. Almost half of the chloroform solutions even went through the frit. Droplets left the bottom of the frit on the first thrust of the injection, i.e. they must have found channels through which they could pass. In contrast with a loose plug of wool, the more dense frit is not tight enough to stop the liquid.

3.5. Stopping the Sample Liquid

103

The poor performance of a sintered bed must be the result of its high thermal mass: the particles have a diameter of a few hundred micrometers. There is no chromatographic data comparing frits with other packings. Marshall and Crowe (16) found, however, clearly increased standard deviations when they replaced glass wool by glass beads. 3.5.10. Carbo'rit

Carbofrit (Restek), a filigree network of a carbon-type mate­ rial treated at high temperature, behaved more like glass wool, sucking up the liquid without visible resistance. No liquid whirled around. With concentrated solutions, a fine cloud was observed to leave the bottom of the plug, which suggested that some liquid was nebulized upon hitting the hot surfaces before the latter were cooled.

3.5.11. Column Packing Material

Packing materials for GC columns, like deactivated Chromo­ sorb, tend to be more inert than glass or fused silica wool. They must be supported by wool, but, as mentioned above, wool has a less detrimental effect ifthe vapor passes through it instead of its being the surface from which the solutes evaporate.

Stirred Packing

When an 8 mm plug of Chromosorb was used above a 6 mm plug of dense wool, the band of liquid stirred up the packing to a depth of at least 5 mm. The liquid dug into the packing and propelled particles upwards to a height of 2-3 cm. After about 100 ms the packing re-settled and fell on to the part soaked with sample liquid. Hence the sample ended up evaporating in the middle of the bed. There was no rejec­ tion of liquid, i.e. the thermal mass of the (porous) parti­ cles is sufficiently low. Mixing into the packing might be advantageous in so far as the samples make contact with more fresh material. Stirring up the packing could be avoided by placing a plug of wool on its surface. If this is done, however, the packing material below could just as well be eliminated because the wool would determine the nature of the evaporation process.

3.6. Other Criteria for Evaluating Obstacles

So far we have concentrated on the behavior ofthe liquid, as well as the effect of obstacles and packings on the latter. This does not adequately reflect work done previously. The need to stop a band of liquid shot through the vaporizing cham­ ber was not an important criterion then, because fast autosamplers releasing a band of liquid were introduced in the later eighties only. Older concepts of split injectors distinguished a vaporiza­

tion zone, a mixing chamber, an expansion volume, and a splitting zone. In some of the injectors proposed, these zones were physically separated. They were, however, difficult to clean and were never widely used.

UNMlUiIDAD DB~~~UIA

BmLIOTBCA Ciw,,, . . ~

104

B 3. Solvent Evaporation - Visual Observation

Mixing with Carrier Gas

Teaching texts on split injection stress the importance of mixing the sample vapors with carrier gas to render results more reproducible. The subject is, however, not so self­ explanatory that no further comment is required. The fol­ lowing arguments were proposed. Support of sample evaporation by mixing with hot gas. This is unrealistic, as shown in Section 82.3.1. More homogeneous distribution of the vapors across the vaporizing chamber to improve the reproducibility of splitting. Dilution of the sample vapors with carrier gas to im­ prove splitting. Dilution of the vapors to support evaporation of high­ boilers.

Improved Heat Transfer as a Result of Turbulence

Heat flow rates estimated above were based on thermal con­ ductivity in a laminar gas flow. This is a pessimistic assump­ tion because vaporization of the sample disturbs the gas flow and increases the transport of heat by mixing of hot and cooled gas and vapor. It is difficult to predict the increase in the rate of heat transfer which can be achieved.

Uniform Distribution of Sample Vapor

In split injection, the solute vapor should be uniformly spread across the liner before it reaches the split point (not necessary for splitless injection). Such homogenization was the primary goal for the design of most liners with obsta­ cles. The gas and vapor are forced through a narrow chan­ nel and spread into the wider bore tube below. The extent to which inhomogeneity across the liner is really a problem remains an open question (Section Cl0.5). Nebulization at the needle exit forms a plug of vapor which largely displaces the carrier gas. The perylene aerosol (con­ centrated solution or solution containing contaminants imi­ tating edible oil) seemed to fill the cross section of the 5 mm i.d, liner homogeneously within 1-2 frames ofthe video (less than 40-80 rns, injections without gas flow). This suggests that mixing by thermospray injection is fast compared with the process of sample splitting, which takes some 500­ 1000 ms when the split flow rate is 60 mUmin. Maybe there is turbulence as a result of the high initial velocity ofthe sam­ ple material and the violent evaporation.

Relevant for Packed Liners?

Evaporation from a plug of wool might be different. Some­ times the liquid is deposited near the liner wall and the sol­ utes evaporate into the gas stream flowing along the wall. Since they evaporate after the solvent, there could be quite a laminar gas flow carrying them largely past the column en­ trance. The need to improve solute distribution seems to be con­ firmed by the numerous analysts who use a cup below wool. Little is known, however, about the experimental back­ ground. When they used solely wool, it might have been placed close to the column entrance, leaving little room for

3.6. Other Criteria for Evaluating Obstacles

105

spreading across the liner. Results might have been differ­ entwith the packing placed in the upper part of the liner (com­ bined with the use of a short needle). 3.7. Duration of Solvent Evaporation

As derived above by quantitative estimation, solvent evapo­ ration is slower than often assumed because of the large amount of heat to be transported. The duration was estimated visually, by noting the moment when bright fluorescence disappeared. ..

High Variability

Table 87 lists results for injections with band formation into chambers at 200°C and with a split flow rate of 40 mLlmin. Results varied widely, from 200 ms to more than 5 s. Under most conditions they were poorly reproducible (often vary­ ing by a factor of three, i.e. by up to 3 s in absolute terms). Usually there was no clear proportionality to the sample vol­ ume, or between duration and the heat of evaporation of a given solvent. Table '-7

Duration of evaporation; injections with band formation; 190­

200°C; gas flow rate, 40 mUmin. (From ref. [17])

Liner Empty

Solvent

Time lrnsl

5 III chloroform 2 III chloroform 5 III hexane Cup liner 1 III chloroform 2 III chloroform 2 III dichloromethane Cyclo liner 2 III ethanol 1 III chloroform 2 III hexane 2 III chloroform laminar liner 2 III hexane 5 III hexane 2 III chloroform Minilam liner 1 III hexane 5 III hexane Glass wool 2 III hexane 5 III hexane Frit 1 III hexane 3 III ethanol 3 III chloroform Carbofrit 2 III chloroform 5 III chloroform Chromosorb 5 III hexane 5 III chloroform * with some breakthrough through the obstacle

3500 880 600 280 600 1000 320* 640* 1500 800 350 480 500 800 400 1500 3900 480 240 720 4200 5600 440 1500

106

8 3. Solvent Evaporation - Visual Observation

Heat Supply

Evaporation tended to be rapid when a large droplet jumped backwards into the empty vaporizing chamber and exploded there, because this extracted heat from a large part of the chamber. It was, on the other hand, rather slow when most of the liquid remained sitting at the bottom of the chamber or in a wool packing, because it extracted the heat from a small area. This was nicely shown by the large difference between the liners containing wool and a frit. Evaporation from glass wool was the only smooth process of reproducible duration.

Consequences

The consequences of variable duration of sample evapora­ tion after injection with band formation have not been ex­ plored. The most obvious effect is that chromatography is started with a delay corresponding to this duration and that retention times vary according to the variability of the latter. This is relevant only for fast GC, however. Other effects might affect quantitative analysis with split in­ jection. Because the split ratio tends to fluctuate during the injection process, the split ratio and also the linearity of splitting might depend on the moment when splitting oc­ curs, i.e. on the duration of solvent evaporation (Sections C8 and C9).

4. Solute Evaporation Up to this point we have considered, nearly exclusively, the evaporation of the sample solvent - although, of course, we want to analyze the sample components. After the Solvent

In fact, most samples analyzed by GC are dilute solutions and solvent evaporation is the first obstacle to be overcome. As long as the solvent is not fully vaporized, the solute material cannot be heated above the solvent boiling point and vaporized, irrespective of the injector temperature.

Choice of Injector Tempera­ ture

Little information has been collected on solute evaporation. Analysts tend to work pragmatically, by trial and error. Some basic considerations could, nevertheless, be helpful, because they provide some ideas on what to try and what else has little promise. There is, however, no formula for calculat­ ing the required injector temperature. There are numerous ideas on how the injector temperature should be selected. In earlier times, when analysis was still limited to rather volatile compounds, a common opinion was that the injector temperature should be at least as high as the boiling point of the least volatile compound. As the range of components amenable to GC widened, the rule was modi­ fied: a temperature 100° below the boiling point was consid­

4.1. Evaporation in the Gas Phase

107

ered sufficient. Even this could, of course, no longer be main­ tained when the boiling points of the solutes analyzed ex­ ceeded 400-500 -c, 4.1. Evaporation in the Gas Phase

Requirements for solute evaporation are fundamentally dif­ ferent, depending on whether evaporation occurs after nebulization or from a surface. Here we look at the process after thermospray formation by partial evaporation in­ side the needle.

Steps of the Evaporation Process

Evaporation from droplets suspended in the gas phase passes through the steps shown in Figure 821. 1 The solvent evaporates, which keeps the droplets and their close environment near the solvent boiling point. 2 At the end of solvent evaporation, the particles or drop­ lets consisting of solute material are heated to the injec­ tor temperature. 3 During or after heating, the solutes evaporate, possibly leaving micro particles of non-volatile by-products.

.

~

' " 01

'yo",' needle

~~:~t. ~-:';"'OO Solvent

evaporation

01 the sample

Droplets at solvent boiling point Solvent vapor

!

Solute evaporation

..

Solute droplets warm up

Solute vapor

Figure 821 Evaporation of a nebulized sample in an empty liner. The vertical axis represents time. not necessarily distance from the needle: evaporation often proceeds largely at the same site (depending on the gas flowl.

4. 1. 1. Some Key Terms

Because there are no forces attracting the solute molecules to a surface (adsorption, partitioning), it should be relatively easy to model the volatilization process. Solute evaporation is a matter of vapor pressure and vapor concentration in the gas phase.

Vapor Pressure

By analogy with distillation we are used to assuming that vaporization requires a temperature around the boiling point. The boiling point is the temperature at which the vapor pres­

108

B 4. Solute Evaporation sure of a compound equals the pressure ofthe environment.

At the standard boiling point, the vapor pressure is 1 bar. For

instance, at 100 DC, the vapor pressure of water is sufficient

to form bubbles within the liquid and to expand against the

pressure of the air above the water.

We also know that the vapor pressure is never really zero

even at low temperatures. Water evaporates at tempera­

tures far below its boiling point and snow slowly evapo­

rates even at temperatures well below freezing point. If hu­

midity is low (i.e. if the atmosphere is not saturated with

water), the air can pick up water vapor even at low tempera­

ture. A small amount of vapor will, however, be sufficient to

saturate the air and, hence, a large amount of air is needed

to vaporize a small block of ice.

Dilution

If there is a large volume of gas, a solute can be vaporized

even if its boiling point is well above the temperature of the

gas. In other words, at high dilution, a small vapor pres­

sure, i.e. a low injector temperature, is sufficient to

evaporate high-boiling material. In GC we usually deal with

highly diluted gaseous phases.

Saturation

At a given temperature, a component may be completely

vaporized if its concentration in the gas phase (in the injec­

tor consisting of carrier gas and vapor, primarily of solvent)

does not reach saturation. The concentration of a substance

corresponding to saturation is determined by its vapor pres­

sure compared with the total pressure, i.e. the partial vapor

pressure. For instance, if a component has a vapor pres­

sure of 0.2 kPa and the inlet pressure is 200 kPa, saturation is

reached at a volumetric concentration of 0.1 %.

At 0 DC, the vapor pressure of water is 6 mbar, i.e. up to 0.6 %

(v/v) of the air can consist of water vapor. At -30 DC, it is about

0.1 mbar and saturation is reached at about 100 ppm - which can still be high in GC.

Pressure-Dependent Boiling Point

Because a compound boils when its vapor pressure equals

the pressure of its environment, reducing the pressure low­

ers the boiling point. Table 88 shows the boiling point of

some substances typically analyzed by GC at ambient pres­

sure (100 kPa), 1 kPa (10 rnbar). and 1 Pa (0.01 mbar). As is

known from vacuum distillation, boiling points at 1 kPa are

reduced by 100-150°, those at "high vacuum" (1 Pal by

200-300°.

Dew Point

Dilution with carrier gas is analogous to reduction of pres­

sure insofar as a limit is also reached below which the par­

tial vapor pressure (equivalent to the concentration in the

gas) drops below the vapor pressure of a compound, prompt­

ing evaporation of the latter.

The dew point is the temperature at which a given concen­

tration of a component in a gas phase corresponds to satu­

ration. At or above the dew point the component is fully va­

4.1. Evaporation in the Gas Phase

109

Table B8

Boiling points of some components at ambient (100 kPa)

and reduced pressure; reduction of the boiling point com­

pared with that 8t 8mbient pressure.

Boiling points at Compound

100 kPa 1 kPa

n-C12 n-C18 n-C28 Butylnaphthalene Fluorene Butyl stearate Dioctyl phthalate Eicosanol

216 316 430 287 295 307 341 355

85 167 263 139 137 201 231 213

Reduction

1 Pa -5 61 136 44

48 100 122 119

1 kPa 1 Pa 131 149 167 148 158 106 110 142

221 255 294 243 247 207 219 236

porized. Below this temperature volatilization remains incom­

plete or the component recondenses.

The lower the solute concentration, the lower the dew point,

i.e. the lower is the injector temperature required for com­ plete vaporization. The minimum injector temperature

corresponds to the dew point of the most critical sol­ ute; this can be the one with the highest boiling point or one of lower boiling point present at a higher concentration. 4.1.2. Dilution with Carrier Gas in an Empty

Liner

The required injector temperature depends on the solute con­

centration in the vaporizing chamber. Dilute solutions can

be injected at lower temperatures. In other words, re­

duction of the amount injected is an alternative to increas­

ing the injector temperature. This may be useful for avoid­

ing thermal stress.

Achievable Dilution

The extent of the dilution achievable in a conventional split!

splitless injector depends, first of all, on whether evapora­

tion occurs in the gas phase or from a surface. When the

sample evaporates through nebulization in the gas phase,

the possibilities of enhancing dilution are limited, because

the sample vapor cannot be mixed with more carrier

gas than is available in the injection zone. Two sce­

narios can be considered.

1. Droplets "Shot" through Gas

If droplets travel some distance through the vaporizing cham­

ber, the vapor is mixed with the plug (volume) of carrier gas

through which it passes (Figure 822). Spreading of the liq­

uid is so fast that the concurrent gas flow is negligible. If the

liquid is nebulized in 50 ms, even gas flow at a rate of 100

mLjmin replaces a volume of 80 III only.

In a 4 mm i.d, liner, droplets traveling up to 2 cm from the

needle tip are mixed with a gas volume of ca. 250 ut., If the

sample generates 500 III of vapor, primarily of solvent, dilu­

tion is less than 1:1.

110

B 4. Solute Evaporation

Negligible gas~

MiXing!

zone

+

" Syringe needle

~

••:

Droplets "shot"

through carrier gas

Sample material evaporating into carrier gas present in zone

II

Column entrance

Figure 822 Mixing after nebulization of the sample liquid in an empty liner. 2. Vapor Displaces Carrier Gas

If nebulization and evaporation occur near the tip of the sy­ ringe needle, expanding vapor displaces the carrier gas in both directions, backwards towards the gas supply and for­ wards into the split outlet (Figure 823). In the center of the vapor cloud there is little dilution, whereas there is increas­ ing mixing with gas towards the edge, as a result of turbu­ lence and diffusion. Still, overall dilution is weak.

Large Vapor Cloud in Small Chamber

When the sample evaporates in the gas phase, the potential for mixing of the vapor with carrier gas is small because a

large volume of vapor is mixed with a small voluma of gas. This also means that obstacles promoting turbulence mix the sample vapor within itself rather than with carrier gas. Basically, more extensive mixing would be possible if the chamber was enlarged. The effect is, however, too small to be important. If it were enlarged, e.g., by a factor of four, which means doubling the internal diameter, the vapor might be four times more dilute. This, however, makes transfer dif­ ficult or impossible in splitless injection, and in split injec­ tion the same effect is achieved more easily by diluting the sample liquid four times.

4. 1.3. Solute Concentra­ tions in the Injector

To obtain some idea about solute concentrations in the mix­ ture of carrier gas and sample (primarily solvent) vapor in the vaporization zone, we estimate them for two examples. They enable estimation of dew points, i.e. the minimum in­ jector temperature.

Split Injection

A 1 ~L sample containing a component at a concentration of 0.1 %, i.e. 1 ~g of the component is injected. If we assume negligible mixing with the carrier gas, complete evaporation of this compound results in dilution of the 1 ~g of material in

4.1. Evaporation in the Gas Phase

111

Syringe needle

Pressure wave displaces carrier gas

Center of evaporation; more or less undiluted vapors

Column entrance

Figure 123 Nebulization at the needle tip and subsequent evaporation displaces the carrier gas and forms a plug of vapor with hardly any dilution at its center.

ca. 200 III of vapor. This dilution by a factor of 5000:1 (v/v) results in a concentration in the gas phase of 200 ppm. In split injection, most common concentrations range from ten times lower to ten times higher than this. If the inlet pressure is 1 bar (2 bar absolute), a 200 ppm con­ centration in the gas phase corresponds to a vapor pressure of 0.4 mbar. From vacuum distillation we know that at 0.4 mbar (40 Pal the boiling point of a compound is reduced by approximately 150-200°. In GC, the compound does not evaporate into a vacuum, but into the carrier gas, i.e. we speak about the dew point, and the minimum injector temperature is this far below the standard boiling point. Such theory implies that with split injection C25 is the last n­ alkane to be fully vaporized at an injector temperature of 250°C. Splitless Injection

In splitless injection, vapor clouds can be handled only if di­ lution with carrier gas is kept to a minimum. Hence even high-boiling material must be evaporated with limited dilu­ tion. If a component is injected splitless and recorded by a spe­ cific detector or a mass spectrometer, ca. 1 ng/Ill is an aver­ age solute concentration in the sample solution. If the com­ ponent evaporates into a 1 ml cloud of vapor and carrier gas, it is diluted to roughly 0.1 ppm (vM. and a vapor pressure of about 10-4 mbar (10 mPa) is sufficient. This vapor pressure is usually reached at temperatures more than 300° below the standard boiling point.

Practical Experience

In practice, results often seem more encouraging: with an empty liner the n-alkane C34 can be successfully analyzed at

112

B 4. Solute Evaporation an injector temperature of 250°C, elution from the syringe

needle being more of a limitation than transfer from the va­

porizing chamber into the column. Peaks are even obtained

for triglycerides after injection at an injector temperature of

300°C. This might be the result of introducing an aerosol

into the column rather than vapor.

Experience would also suggest that positive results presup­

pose "clean" solutions, i.e. the absence of substantial

amounts of involatile material.

4. 1.4. Glass Woollm­ proving Eliaporation 7

It is often believed that a plug of glass wool would improve

solute evaporation. Either wool is assumed to bring heat to

the particles or to retain aerosol particles on the fibers until

the solutes are vaporized from there (evaporation into a larger

volume of gas).

Visual observation did not confirm this. The cloud of

peryfene particles formed after hot needle injection was

driven through a 15 mm long, dense plug of glass wool with­

out noticeable loss offluorescence. The wool neither retained

the.perylene particles nor helped the vaporization. Of course,

aerosols pass even more easily through liners with obsta­

cles.

4. 1.5. Evaporation from Contaminants

For the thermospray injection of contaminated samples, two'

more problems must be considered:

1 There is an additional step in the vaporization process:

separation from contaminants. 2 The injector should perform like a filter: vaporizing the solutes of interest while retaining the high-boiling or involatile by-products to prevent column contamination.

Steps of the Evaporation Process

After nebulization, the evaporation of a contaminated sam­ ple liquid might pass through the following steps, the first two being the same as for clean (fully evaporating) samples. 1 The volatile material (i.e. primarily the solvent) evapo­ rates from the droplets; droplets cannot touch the liner wall at this stage. 2 When solvent evaporation is complete, the temperature of the droplets increases and more solute material evaporates. For the next step, there are the possibilities shown in Fig­ ure 824. 3a The solutes evaporate from the droplets suspended in the gas. 3b The increased boili ng point of the droplets enables con­ tact with the liner wall. When the solutes are carried along with the droplets, they must evaporate from a layer consisting of non-evaporating material from this and previous injections. 3c The solutes reach the column embedded in particles of nebulized matrix material and must evaporate from the latter in the column inlet only.

4.1. Evaporation in the Gas Phase

113

~ By';"". needle Injector

liner

~ ~

• 1) Solutes evap­ orate from flying

"dirt" droplets

Solvent evaporates, droplet at solvent boiling point The solvent is evaporated; the droplet warms up

2) Droplets adhere

to insert wall;

solutes evaporate from surface

Column inlet

Figure 824

Solute evaporation from nebulized involatile matrix mate­

rial in an empty vaporizing chamber.

Evaporation from Moving Droplets

Evaporation from suspended droplets may be difficult. The matrix material exerts retentive power in the same way as the stationary phase in a column (except that it is moving) and hinders solute vaporization. At least initially, the temperature of the droplet is be­ low that set for the injector because of cooling by sol­ vent evaporation. Often time is short since most of the droplets are rap­ idly transferred to the liner wall.

Stable Fog in 20 mm i.d. Tube

Important information about the behavior of droplets and a test solute (perylene) dissolved therein was again obtained from visual experiments. Droplets were created by addition of edible oil (virtually involatile at 200°C) to the perylene solution. Thermospray injection into a 20 mm i.d. heated centrifuge tube produced a fog which remained stable over many min­ utes (CD-ROM). The fog could be sucked from this chamber through a fused silica capillary tube, which demonstrates that it could be transferred into a column almost completely. The wall of the tube remained fairly clean, i.e. little oil and perylene was deposited there.

Transfer to Wall, 5 mm i.d. Liner

When the same solution was injected into a 5 mm Ld, tube, more realistically imitating an injector, fluorescence was transferred to the wall within a single frame of the video. The oil did not form a patch, as when a larger droplet splashes against a surface, but appeared as a cloud over the surface, indicating that small droplets were transferred. Together with the oil, virtually all the perylene was transferred to the liner

114

B 4. Solute Evaporation wall. Perylene is eluted from GC columns at ca. 250°C and

thus well represents components of intermediate to high­

boiling point.

This extremely fast movement to the liner wall presupposes

strong forces which also overcome repulsion by solvent

vapor. Only electrostatic effects can explain this. They could

originate from charge separation during thermospray forma­

tion.

Poor Control

The mechanisms involved in the attraction to the liner wall

are not understood and the extent of solute transfer with

the matrix material to the liner wall is largely out of control.

Most probably it is influenced by the particle or droplet size:

nebulized pervlerm injected as a clean solution (0.1 %, com­

pared with 1 % oil) was not transferred.

Matrix Effects

Problems in quantitative analysis arise if solutes are trans­

ferred to the liner wall to a variable extent, as has been dras­

tically demonstrated for perylene:

The nebulized perylene of the clean solution was swept into the column as if it had been vaporized - at 200°C, 80° below its melting point. With the oil, it was almost completely transferred to the liner wa!1. The oil film formed there remained fluo­ rescent for a long time, indicating that the perylene was unable to evaporate. Hardly any perylene reached the column under these conditions. Hence results turn out different depending on whether, e.g., a clean calibration mixture or a matrix-loaded sample is in­ jected. This is a source of systematic errors. This subject will be further described in the context of splitless injection under the heading of "reducing matrix effects" in Section D6.3.

4.1.6. Prevention of Column Contemination

Linked with the above matrix problems are those of column contamination. Nebulized non-evaporating sample matrix material should be retained in the injector, preventing a smoke-like cloud of "dirt" being swept into the column, where the particles soon stick to the column wall or station­ ary phase surface.

Effects on Column Perform­ ance

Involatile matrix material in the column inlet can have the following effects. 1 It builds up retention power. Retention is greater in the oven-thermostatted column than in the injector be­ cause temperature is lower. It results in tailing peaks (delayed release of solute material) or loss if the release occurs too late. 2 It causes adsorption. If, for instance, acidic material is deposited in the inlet, basic solutes form a salt and can no longer be vaporized. More rarely, contaminants de­ activate adsorptive sites, improving column perform­ ance.

4.1. Evaporation in the Gas Phase 3 4

115

It turns the inlet catalytically active, degrading labile compounds. Aggressive sample by-products (e.g. alkali) degrade the stationary phase in the column inlet and produce bleed (baseline drift in temperature programs).

Transfer to the Liner Wall

As described above, in a 20 mm l.d, tube nebulized oil formed a stable fog which was carried into a fused silica capillary in the same way as vapor. This is frightening, because it means that all the contaminants are introduced into the column in­ let. From practical experience we know, however, that this is not what happens in real injectors. When the experiment was repeated with a 5 mm i.d, liner, the process was entirely different: particles seemed to be attracted to the wall by strong forces. With a gas flow rate of 40 mLJmin, a small amount of the fog was driven on­ wards and left the chamber. It seemed that droplets not immediately deposited on to the wall would remain sus­ pended. With a flow rate of only 2 mLJmin, as in splitless injection, deposition was virtually complete.

"Dirt" Ring on Liner Wall

From practical experience it is known that numerous therrnosprav injections of severely contaminated samples can be performed before signs of column contamination are observed. The liner shows the fate of most of the non-evapo­ rating material: there is a thick, brownish-black ring on its wall, reaching from ca. 5 mm behind the tip of the in­ serted needle to 10-15 mm ahead. This confirms that with the oily perylene solution the most common process was observed.

Improving the Retention of Contaminants

If transfer to the liner wall is insufficient, the question arises how we could improve the retention of involatile material in the injector. Retaining aerosols might be as difficult as preventing smoke from leaving a chimney. Initially the droplets cannot touch the liner wall because they contain solvent. By the time the solvent has evaporated fully, the droplets may have reached or passed the column entrance (split injection). In splitless injection there is more time, but the movement of the particles is so slow that the probability of hitting the liner wall is negligible.

Wool filtering out contami­ nants?

The effectiveness of a dense plug of glass wool at filtering out nebulized matrix material was tested in the 20 mm i.d. tube which did not attract the fog to the wall. A perylene solution containing 5 % oil was sprayed into a tube packed with a dense 3 cm plug of wool. The nebulized edible oil slowly passed through the wool without noticeable re­ tention. Fluorescence above and below the wool was not noticeably different (CD-ROM), nor was fluorescing material deposited on the wool.

116

B 4. Solute Evaporation

Prevention at the Beginning

If an aerosol is formed, and the particles are not attracted to surfaces in the vaporizing chamber, their retention seems almost impossible. The problem is reliably solved only by preventing aerosol formation. This requires the deposition ofthe sample solution on to a light packing material. Evapo­ ration from a surface yields clean vapors. The best method for transferring sample liquid on to a suit­ able surface (such as wool) involves band formation, hence injection by means of a fast autosampler or in a high-boiling solvent.

Injection into Wool

When partial evaporation in the needle cannot be suppressed (manual injection of solutions in volatile solvents), the best retention of non-evaporating material is achieved when the wool is positioned at the tip of the inserted needle (Figure 825). Far less fog left the plug than when the sam­ ple was nebulized above the wool (CD-ROM). The concen­ trated droplets might consume a sufficient amount of heat to cool the fibers which then receive the liquid. Maybe elec­ trostatic effects again playa role. Syringe needle Injector liner

Wool

Sample liquid transferred to nearest fibers

II Column inlet Figure 825 Retention of "dirt" after nebulization of the sample liquid by wool positioned at the exit of the inserted needle.

Conclusion

Injection into wool helps retain contaminants, and in split­ less injection of contaminated test samples the reducing matrix effects were largely eliminated (Sections D6.3.6). On the other hand, the advantages of gas phase evaporation are lost. In the past thermospray injections were usually performed into empty liners, because this was the way chromate­ graphers were taught. There is not sufficient experimental data to support this recommendation.

4.2. Evaporation from Surfaces

Solute evaporation from surfaces follows injection with band formation. A liquid sample in a volatile solvent is immedi­ ately deposited on the packing when a low thermal mass material, such as glass wool, is used. When trapped within obstacles, it is suspended in the gas phase until solvent evaporation is complete and is then transferred to a surface.

4.2. Evaporation from Surfaces

117

Evaporation may also proceed from surfaces after thermo­ spray injection into a packing or when matrix material trans­ fers the solutes to the liner wall. General Characteristics

Evaporation from surfaces differs from that in the gas phase in three important respects. 1 The time available for evaporation is long, because the solutes are sitting on the surface. The solute material can be carried to the column by a far larger volume of carrier gas and a lower vapor pressure is sufficient. 2 The solute material is retained on the surfaces by a variety offorces which reduce its vapor pressure. A high temperature is, therefore, required. 3 The vapor is clean, i.e. there is little risk of column con­ tamination (except when the volatility of the contami­ nants is such that they are transferred into the column, but do not pass through it, e.g. fat and waxes).

Steps of the Evaporation

Evaporation from a packing material proceeds through the following steps. 1 The solvent evaporates, keeping the region at its boiling point. 2 When cooling by solvent evaporation ceases, the tem­ perature increases. 3 The solutes evaporate into the carrier gas stream during or after heating of the support material to the injector temperature.

4.2.1. The Iodine Experi­ ment

Perylene is not suitable for the visual observation of solute vapor. Driessen and Lugtenberg [18] described compounds which fluoresce strongly in the vapor phase also. We used iodine. It is volatile and behaves like a GC solute.

Experimental

Concentrated solutions of iodine were injected into a "trans­ parent injector" consisting of a glass tube with a flame-sealed bottom heated in a silicone oil bath. No carrier gas was sup­ plied, but after each injection, the iodine vapors were re­ moved by means of a gas stream. Acetone was the pre­ ferred solvent because this enabled the preparation of the most concentrated solution. Because iodine degrades the acetone, however, forming a black precipitate, it was neces­ sary to prepare fresh solutions before each experiment. Old syringes were used since iodine severely attacks the needle surface.

Empty Tube

After injection into an empty 5 mm i.d, tube, a large part of the solution usually jumped backwards, easily as far as 4 ern, and then dropped back to the bottom before it evaporated. The iodine commonly evaporated from the bottom. Its vapor looked like smoke. It seldom moved as a plug filling the bore (A in Figure B26), but more often formed a trail curv­ ing through the chamber as shown in B. Other times the movement had an angular momentum, which produced

118

B 4. Solute Evaporation

A B C Plug Trail Spiral Figure 826

Expanding iodine vapor in an empty liner.

something like a smoke spiral (e). This expansion process was not reproducible, which has consequences mainly for splitless injection (Section 05.1.1). Wool

A small amount of glass wool at the bottom of the tube fun­ damentally changed the spreading process. No longer was there a well audible sizzle of evaporating liquid. and the io­ dine vapor was not driven upwards by violently expanding acetone vapor. The solution entered the glass wool and dur­ ing the evaporation of the acetone hardly any iodine left the site. The first substantial amounts of iodine vapor were produced 1-2 s after injection and formed a smell cloud in the region of the wool.

Dilution of the Vapor

Visual experiments with a "transparent injector" and a split flow enabled estimation of the dilution of the sample vapor with carrier gas. During evaporation of the acetone, the wetted wool was cooled. Evaporation of the iodine started 1 s later and took 3 s. This caused the iodine to be diluted by the gas passing the site of evaporation during this period. At the split flow rate of 100 mLJmin (1.7 rnt/s) applied the io­ dine was diluted with about 5 mL carrier gas, compared with 0.5 mL when evaporation occurred in the gas phase of an empty tube. Hence evaporation from a surface promotes di­ lution more efficiently than, e.g., enlarging the injector.

4.2.2. Dilution in Carrier

Evaporation from a surface results in dilution of the sample vapor not only by the carrier gas present in the vaporizing zone, but by the gas passing the evaporation site possi­ bly over an extended period of time. Dilution is limited only by the volume of the initial band tolerated by chromatog­ raphy (peak broadening).

Gas

Dilution Self-Adjusting to Needs

Dilution depends on the rates of evaporation. A vola­ tile solute is vaporized rather rapidly and ends up as a con­ centrated vapor whereas a slowly evaporating, high-boiling

4.2. Evaporation from Surfaces

119

component evaporates during, maybe, many seconds and is diluted by a correspondingly larger volume of carrier gas passing the site of evaporation. The same applies to different amounts of solute. A small amount of a compound might evaporate within a short time and, therefore, be diluted by a smaller volume of carrier gas than a larger amount of the same component. The concen­ tration of the vapor in the gas phase could turn out to be constant, corresponding to the vapor pressure of the sub­ stance. As a means of achieving dilution of the solute material with carrier gas, evaporation from a surface is elegant, since it is automatically adjusted to the vapor pressure and the amount of solute injected and provides the fastest possi­ ble evaporation by saturating the passing gas. Split Injection

In split injection with a long temperature program, the initial bands ofthe high-boiling components, i.e. ofthe late peaks, are reconcentrated in the column inlet by cold trapping. Sol­ ute material entering the column after a delay is still recom­ bined with the bulk. If the split flow rate is 100 mLJmin and the component starts chromatography only 10 min after in­ jection, the solute material may be diluted by as much as 1000 mL of gas. The concentration will be ca. 0.1 ppm and the vapor pressure 10-4 rnbar, as calculated above for the splitless injection of a trace component into an empty liner.

Splitless Injection

In splitless injection, the dilution cannot be increased by deposition on to surfaces because the volume of gas and vapor that can be transferred into the column remains in the range of a few milliliters. Hence evaporation needs to be completed almost instantly.

4.2.3. GC Retentive

The forces acting against evaporation originate from solva­ tion in a liquid (e.g. non-evaporated sample material) or ad­ sorption on a possibly contaminated surface. The larger these forces. the higher is the temperature required for volatilization.

Power of a Surface

Estimated Retentive Power

The gas chromatographic retentive power of a surface can vary over a wide range. It can be estimated in terms of the phase ratio. The region of evaporation has a volume of 50­ 100 IlL If 100 I.1g non-evaporating material is accumulated there, the phase ratio is equivalent to that of a thin-film cap­ illary column, i.e. the temperature for desorption should be similar to that required for the analysis. This presupposes, however, an exchange of the gas comparable with that in chromatography, which is not the case if injection occurs in splitless mode. According to practical experience, this estimate is too opti­ mistic: usually the injector temperature must exceed the elution temperature of the last component of interest.

120

B 4. Solute Evaporation

4.2.4. Experimental Data

Bowermaster [6] described an experiment which enabled him to determine the speed of evaporation of the n-alkanes C,o·C30 • An additional pressure source was installed in the gas supply line. The normal split flow rate was set at 50 mLl min. The additional source of carrier gas increased it to 350 mLlmin. At a given time after injection. the additional gas supply was switched on, removing the remaining material at a higher split ratio. Peak areas were plotted against the time spent at the low split ratio. Results for a 4 mm i.d. straight liner tightly packed with a 1 cm plug of glass wool are shown in Figure 827. Injection was performed with the Hewlett-Packard 7673A fast auto­ sampler.

3E04

15

20

25

30

Alkane Carbon Number

Figure 827 Peak areas of n-alkanes measured after different periods [min] at a low split ratio. (From Bowerme.ter [6]).

Injector at 200°C

At an injector temperature of 200 °C, n-alkanes up to Cn (b.p. 303°C) were completely evaporated within 1.2 s, but within this time little was evaporated beyond n-C24 • 2.4 s after the injection, nearly all the n-C21 had evaporated. After 38 s, the n-C 2 7 (standard boiling point. 421°C) had va­ porized. Assuming that it took 8 s to bring the evaporation region back to 200°C, evaporation of n-C27 took 30 s and the vapor. were diluted in 25 mL of gas. Because 200 ng of this compound was injected. its concentration in the gas phase was around 2x10-4 mbar. At 10-2 mbar, the boiling point of n-C27 is 137°C. which indicates that evaporation was hin­ dered by adsorption effects.

4.3. Conclusions on Injector Temperature

121

4.3. Conclusions on Injector Temperature

The literature creates the impression that the injector tem­

perature is of paramount importance. Methods tend to leave

out many essential specifications, but seldom the injector

temperature. Detailed studies have been performed to de­

termine whether, e.g., 270°C is preferable to 250 °C. Practi­

cal experience, on the other hand, often shows that perform­

ance on another instrument is different and these indications

are of limited usefulness. Considering the injector tempera­

ture as the key parameter is probably a misjudgment of

priorities.

Thermospray or Deposition on to Surfaces

The most important factor determining sample evaporation

is the manner the liquid leaves the syringe needle, i.e.

whether or not there is a thermospray. The key factors here

are:

1 manual or autosampler injection;

2 fast autosampler or autosampler imitating manual in­

jection; the boiling point of the sample matrix in relation to the 3 injector temperature.

4.3.1. Thermospray Injection

The key factors in thermospray injection, thus partial solvent evaporation inside the needle, are: hot needle injection; sample volume; solvent properties; needle length. Whether 1 or 2 III is injected probably has more influence on the real temperature in the chamber than whether 200 or 300°C is set as injector temperature, and with a long needle 200 °C may be sufficient whereas 300°C is needed if a short needle is used.

Role of Injector Temperature

The injector temperature must be optimized, primarily by con­ sidering nebulization of the sample liquid. The temperature should be fairly high (ca. 250°C) even when exclusively volatile components are analyzed. Extremely high temperatures can be avoided when sol­ ute transfer as an aerosol is acceptable. The temperature of the injector head, determining that ofthe rear part ofthe syringe needle, should be high (at a given temperature setting this easily varies by more than 100° from one injector to another). Best use ofthe injector temperature is made by the hot needle technique. When the solutes are pulled to the liner wall by large amounts of high-boiling or involatile matrix material, or a packing is used to retain contaminants, higher temperatures are re­ quired.

4.3.2. Deposition on a Surface

Injection with a fast autosampler generally requires a packed liner or a liner with obstacles capable of stopping the band of liquid. The key factors are then:

122

B 4. Solute Evaporation

the retentive power and adsorptivity of the packing ma­

terial;

amount of non-evaporating material accumulated at the

evaporation site as a result of previous injections;

amount of non-evaporating by-products in the sample

injected;

amount of solute material injected.

The effect ofthe last point is rather unpredictable. When con­ sidering vapor pressures and gas volumes necessary for evaporation, increasing amounts require higher tempera­ tures. More often, however, larger amounts are better at over­ coming adsorption and even enable to reduce injector tem­ peratures. Role of Injector Temperature

After deposition of the sample on to a surface, the solutes must be vaporized by generation of sufficient vapor pres­ sure above the retaining or adsorbing material on the surface. This process is, primarily, determined by the reten­ tion power, secondly by the volume of gas passing, and, fi­ nally, by the temperature. The retentive power of the surface easily changes by an equivalent of 1000 temperature differ­ ence.

5. Sample Degradation in the Injector Many of the common problems in GC analysis are related to insufficient inertness of the vaporizing chamber: hindered evaporation because of retention or adsorp­ tion; Joss of material as a result of degradation; formation of artifacts from chemical transformations. Residence Time in the Injector

Solute degradation in the injector is particularly important in splitless injection, because of the long residence time in the hot vaporizing chamber. In split injection, cooling dur­ ing solvent evaporation and rapid discharge through the in­ jector clearly result in milder conditions.

5.1. Degradation in the Injector or in the Col­ umn?

If solutes are degraded, it must first be distinguished whether this occurred in the vaporizing chamber, in the column, or both.

Injector-Labile Substances

Practice shows that some components are readily degraded during injection, but hardly during chromatography. Such behavior is plausible ifthe injector temperature substantially exceeds the column temperature used for chromatography. Often, however, elution temperatures are almost as high as the injector temperature, suggesting that properties of

5.1. Degradation in the Injector or in the Column

123

surfaces play the key role: the surfaces in the injector are more active than those in the column. Many carbamate in­ secticides are examples of injector-labile substances [191. Column-Labile Compounds

Other compounds decay during chromatography, but resist vaporizing injection rather well - particularly thermospray injection, for which even non-deactivated (empty) liners can be used. Many phenylurea herbicides belong to this group

[201. If the degradation products are known and are eluted in GC, the chromatogram should be checked forthe corresponding peaks.

5.1.1. Methods for Distinction

Degradation products formed in the injector start chro­

matography as sharp initial bands (together with the rest

ofthe sample) and produce, in principle, perfectly shaped

peaks (A in Figure 828).

Degradation in the column forms products over an ex­

tended period of time, and maybe throughout the col­

umn. They start chromatography as initial bands which

are broad in time and space, and well shaped peaks can­

not be expected.

Peak Shapes

Partial degradation inside the column usually results in char­ acteristic peak deformation. If the degradation product is chromatographed more rapidly than the educt (part of the molecule is lost) the peak tails towards earlier retention times. If the product moves more slowly through the column the tail is eluted after the main peak and is more difficult to distinguish from that resulting from adsorption. In isothermal chromatography, degradation should occur at a constant rate and produce a peak as shown in Figure 8280, with a "foot" stretching from the retention time of the degradation product (material degraded in

Tailing Resulting from Degradation in the Column

Component to be analyzed

B

A Degradation product eluted at regular retention time

-.

Degraded In the Injector

Time

Intact

component

Degraded in

the center of

the column

Degra~ed the Inletin ~

~

Degraded primarily In column Inlet

c

D

Degraded near column outlet Degraded near column inlet

\

Retention tim of degradatio product

~

Degraded throughout column Temperatureprogrammed

Isothermal elution

Figure 828 Schematic diagram of peaks of degraded components for a degradation product eluted be­ fore the intact component.

124

B 5. Sample Degradation in the Injector

the column inlet) to that of the intact material (degrada­ tion near column outlet or no degradation). In temperature-programmed runs the rate of degra­ dation increases with column temperature and the de­ formation looks as shown in C. Sometimes cold trap­ ping of early-degraded material produces peaks with a shape between 8 and C. Exceptions

In practice, results are often less clear. Degradation products formed in the injector can be po­ lar and strongly acidic or basic and, thus, form tailing peaks or not be eluted from the column. Chemical reactions often occur primarily in a column inlet contaminated with aggressive material. Then nev­ ertheless fairly normal peaks, such as 8 in Figure 828, are observed.

Test with On-Column Injec­ tion

An easy and reliable way of checking the influence of sam­ ple introduction is a test run performed by on-column injec­ tion. Direct deposition in the cool column inlet rules out degradation by injection. Observed degradation mustthen be attributed to the column. On-column injection for test purposes can also be performed without an on-column in­ jector (see Section D5.6.2).

5.2. Mechanisms of Solute Degradation

Degradation in the vaporizing chamber is an obvious prob­ lem in view of the high injector temperatures normally used: chemistry at 250-350 °C is rapid and functional groups usually considered stable become reactive. Cooling by sol­ vent evaporation causes, on the other hand, the real tem­ peratures to be below those set, reactions in the vapor phase are slower than in concentrated liquids, and residence times in the injector may be less than 1 s.

Evaporation Process

The manner in which the solutes are vaporized is im­ portant. Evaporation from fine droplets in the gas phase oc­ curs under fundamentally different conditions than that from a surface, and is again different if the surface is coated with a thick layer of involatile material, particularly if it contains aggressive residues.

Optimization of the Vaporiza­ tion Process

Seen from this point of view, it should not be a surprise if labile components are degraded or rearranged inside the injector. We should also be prepared to see phenomena which are variable and difficult to understand, because losses de­ pend on many factors, such as the sample matrix (impuri­ ties, solvent), the injection volume, or the needle handling technique, to name but few. A component might be severely degraded when injected in one matrix, but hardly at all in another, or when injected by use of an autosampler rather than manually. Hence optimization must be achieved by cor­ rect choice of all the factors affecting the vaporization proc­ ess.

5.2. Mechanisms of Solute Degradation

125

Activity of Matrix Material

Often the activity of matrix material deposited on the liner

wall or a packing material is crucial. As an obvious exam­

ple, traces of alkali (e.g. extracted from an aqueous potas­

sium hydroxide solution by a solvent such as ether) have a

strong impact on esters: these are likely to be saponified,

particularly when they are high-boiling, i.e. evaporate only

slowly from this surface.

If samples are nebulized, solutes often do not make contact

with the liner surface, but with the alkali in the small parti­

cles formed after solvent evaporation. Then degradation is

different from one sample to another and, particularly, from

that observed for a clean mixture of standards.

If sample impurities are the cause, deactivation ofthe liner is

useless. More effective clean-up is required and/or more fre­

quent replacement of the liner.

Syringe Needle

If solutes are affected similarly by split and splitless injec­

tion, they are likely to be degraded primarily on the hot metal

surface of the syringe needle (thermospray injection).

Metal is the probable site of degradation if the solute mol­

ecules undergo reactions involving electron transfer.

Purely Thermal Degradation

Finally, many solutes undergo purely thermal degradation

or rearrangement, i.e. without catalytic support. For such

spontaneous reactions the temperature of the injector and

the residence time in the vaporizing chamber are the only

factors which can be adjusted.

5.3. Countermeasures against Solute Degrada­ tion

If degradation in the vaporizing chamber is not too severe,

the following measures help to reduce it to an acceptable

level:

reduction of the injector temperature; thermospray: use of hot needle injection and an empty liner; pressure pulse to accelerate sample transfer (possibly at a lower injector temperature); accelerated transfer into the column by strong solvent recondensation (splitless injection); split injection with a low split ratio instead of splitless injection; better (or different) deactivation of the liner and possi­ ble packing; improved sample clean-up (removal of aggressive by­ products); use of a fast autosampler to avoid contact with a hot syringe needle; and addition of a rather high-boiling active solvent (e.g. 5 %) for temporary deactivation (matrix effect, see below); Such optimization may be tiresome, because effects are sometimes unstable (unreliable) and countermeasures ag­ gravate other problems. Quantitation with splitless injection is, for instance, less precise at lower injector temperatures (severe discrimination and high standard deviations).

126

B 5. Sample Degradation in the Injector

On-column or PTV Splitless Injection

If degradation of labile solutes is observed, the best advice

is to use on-column injection.

PTV splitless injection with an empty liner is often some­

what better than classical splitless injection, because the

~Q\at\\e 'l.Q\ute'l. \eave the va\lorizing chamber before the up­

per temperature is reached, and transfer from the far smaller vaporizing chamber into the column is faster. This technique should be considered when on-column injection is hindered by excessively high concentrations of sample by-products.

5.4. Examples

The literature contains numerous reports of problems con­

cerning solute degradation. Some are mentioned below,

because the experience could be of help in similar situations.

5.4.1. Divinylcyc/obutane

Schomburg et al. [15] described the thermal rearrangement

of divinylcyclobutane into 1,5-cyclooctadiene and 4-vinyl-1­

cyclohexene as a sample for testing injection techniques.

In splitless injection with the injector at 290°C conversion

was nearly complete.

In split injection the rate of degradation was highly de­

pendent on the residence time in the injector, i.e. on the split

flow rate: at 200 ml/min, 54 % of the divinylcyclobutane sur­

vived, at 100 ml/min it was 39 %, and at 5 ml/min merely

2.6 %. This is probably an example of a purely thermally induced reaction.

5.4.2. Carbamata Insecti­ cides

Wiiest and Meier [19] determined the loss of carbamate in­ secticides in the vaporizing chamber, using on-column in­ jection as a reference. Some ofthe losses were a consequence of poor transfer from the syringe needle, but the major pro­ portion of the missing material was degraded (as evidenced occasionally by peaks from the degradation products). At an injector temperature of 250°C, losses of 30 to 40 Ufo were typical; they reached 70 % for methomyl and over 90 % for aldicarb. With split injection at the same injector temperature, losses were considerably lower: methomyl was degraded by only 25 % and aldicarb by 45 %.

Degradation or Discrimina­ tion?

The data illustrate a dilemma often encountered. When the injector was kept at 175 instead of 250°C, degradation of the labile components was substantially reduced (for the two compounds named above it was reduced from 70 and 90 % to ca. 45 %). At the same time, however, transfer from the syringe needle became a severe problem, as manifested by strong discrimination against the higher-boiling solutes and increased standard deviations.

Comparison of Injection Techniques

Miiller and Stan [21] compared the degradation of carbamate pesticides which resulted from different injection techniques: whereas no degradation was observed with on-column injection, splitless injection at an injector temperature of

5.4. Examples

127

220°C resulted in nearly complete decomposition of dimethoate. aminocarb. bendiocarb. dioxacarb. and carbaryl to phenols and methyl isocyanate. With PTV split­ less injection degradation was modest as long as the liner was not packed with glass wool. 5.4.3. Oxygenated Dibenzothiophenes

Vignier et al. [221 investigated the effect of injection tech­ nique on the analysis of oxygenated dibenzothiophenes.

With splitless injection. degradation was found to be considerably worse than with moving needle injection (sol­ vent-free injection after deposition of the sample on to a glass needle, Section 08.2.1), with the latter, in turn, being clearly worse than on-column injection. Anderson [231 partly contradicted this. stating that the diox­ ide is thermostable whereas the monoxide is labile. Degra­ dation was highly dependent on the volume of sample in­ jected (it became worse as the amount injected was reduced). 5.4.4. Mustard Oils

Degradation of some mustard oils from radishes during thermospray injection was described in Section A5.3.4. It is regarded as an example of chemical reaction on the inte r­ nal wall of the syringe needle. Use of hot needle injec­ tion considerably improved results compared with other sy­ ringe handling techniques. which was interpreted in terms of greater proportion of the sample being violently ejected from the needle in the liquid phase.

5.4.5. Chlorohydrin in a Drug Substance

Klick [241 described struggles to optimize a quantitative method for an epoxide (p-cyano(epoxypropoxy)benzene) and the corresponding chlorohydrin. The chlorohydrin tended to form the epoxy compound by elimination of HCI and the epoxy compound tended to be lost in a manner not identi­ fied. When splitless injection with the injector at 200°C was used, 26 % of the chlorohydrin was converted; at 230°C it was 37 %. Removal of the glass wool from the liner did not im­ prove the result. Increasing the inlet pressure from 0.71 bar to 2.2 bar for 30 s reduced the loss to 9 %. On-col­ umn injection was the technique finally selected, because this eliminated degradation.

5.4.6. Drugs Requiring an Empty Liner

Vo/mut et al. [251 studied quantitative results obtained after split injection of a mixture of anti epileptic drug standards in toluene-methanol. 96:4. at an injector temperature of 240°C. 2IJ.L volumes were injected manually by the hot nee­ dle solvent flush technique. The liner was silanized with di­ methyldichlorosilane in toluene.

Accuracy and Precision

Carbamazepine was found to be degraded. irrespective of whether the liner was packed with phosphoric acid-treated or silanized glass wool, or with a packing material "dedi­ cated" to anti epileptic drugs.

128

B 5. Sample Degradation in the Injector The reproducibility of the results was best with an empty liner, because relative standard deviations of both normal­ ized and absolute peak areas were reduced by a factor of more than two. In fact, no significant degradation was ob­ served when an empty liner was used.

Dependence on Split Ratio

The extent of the degradation was reduced by a factor of 3 when the split ratio was increased from 20:1 to 80:1, pre­ sumably because of the reduced residence time in the injector.

5.4.7. Empty Liner for Methyl Esters of Hydroxy Fatty Acids

Husmann et al. [26] came to the conclusion that fatty acids from bacteria should be injected into an empty liner, other­ wise 3-hydroxy acids formed aldehydes by loss of ace­ tic acid. The results showed conversion to be nearly com­ plete. Nothing was specified about the injector temperature or the type of glass wool employed. No aldehydes were pro­ duced when on-column injection was used.

5.4.8. Brominated Alkanes

Cardoso and Afonso [27] described problems encountered during analysis of a brominated alkane (pristane) by split and splitJess injection. Even when empty glass liners were used hydrogen bromide was eliminated, producing the corre­ sponding alkene. With splitless injection up to 90 % of the bromoalkane was converted; with split injection it was still 75 %. Variation of the injector temperature between 160 and 260°C had little effect. The metal surface of the syringe needle was not involved, as shown by an experiment with a fused silica syringe needle.

Metal Surface at the Base of the Injector

Kurt Grob observed similar degradation of chloroalkanes to alkenes in splitless injection and assumed the metal sur­ faces at the base of the injector to be the cause. There must have been numerous other examples of the detrimental ef­ fect of contact with these metal surfaces, because later in­ jectors were offered in which exposed surfaces were coated with noble metal. If sample material is degraded at the base of the injector, the main problem is poor evaporation, l.e. the sample being "shot" there, rather than the chemical activity ofthe surface. No sample material is supposed to enter the region below the column entrance.

6.1. Adsorption in the Injector

129

6. Retention and Adsorption in the Vaporizing Chamber Retention or adsorption on the surfaces of the vaporizing

chamber hinders transfer to the column, which can result in

loss of solute material.

Retention or Adsorption?

Delayed release often results from retention or adsorption.

Retention is understood as a gas chromatographic phenom­

enon, i.e. a layer of material, such as contaminants, behaves

like a stationary phase and retains solutes by partitioning.

Retention is, in principle, independent ofthe amount injected

and, thus, results can still be linear.

Adsorption means adherence to a site as a result of strong

attractive forces. Because adsorptive sites tend to be rela­

tively few in number, at some time they are saturated, pro­

ducing non-linear performance.

6.1. Adsorption in the Injector

Split injection is relatively tolerant of delayed release, par­

ticularly when the analysis involves wide-range temperature

programming. Retention or adsorption typically affects the

components eluted last, maybe 30 min after injection. Be­

cause chromatography of this material might start only

20 min after injection, components released from the

adsorptive site up to that time will be included in the peak.

They still enter the column and are combined with the rest

of the material that awaits chromatography at the column

entrance. The volume of gas passing the site is large.

When the oven temperature is increased, the split ratio in­

creases because the column flow rate decreases. Hence

material released late produces too small a peak area.

The situation is less favorable in fast GC in which the com­

pounds of interest might be eluted after 1 min and material

released 10 s after injection is no longer part ofthe integrated

signal.

6. 1. 1. Split Injection

6. 1.2. Split/ess Injection

Splitless injection is sensitive to retention or adsorption in

the liner because during the transfer period the volume of

the vaporizing chamber is transferred to the column a few

times only. A small delay is sufficient for release of material

from a retentive or adsorptive site only when the split

outlet is re-opened (when the flow increases) causing

nearly complete loss (actually the material is split by the ra­

tio of the split and column flow rates).

6.1.3. Column or Injec­ tor?

It may be difficult to distinguish whether adsorption occurs

inside the 'injector, in the column, or in both. The symptoms

130

B 6. Retention and Adsorption in the Vaporizing Chamber are similar: non-linearity of the response, i.e. peak area is not proportional to the amount of solute material injected. Some solute material is adsorbed by the active sites, tempo­ rarily deactivating the system for the following material. Adsorption inside the column can be betrayed by peak tail­ ing (but not always), whereas adsorption in the liner seldom results in peak distortion.

Comparison with On-Column Injection

Comparison with on-column injection is usually the best con­ trol. Plotczyk [281 gave a "school book" example. Areas of two pharmaceutical compounds were compared with the area of a non-adsorbed compound (n-hexadecane). In split­ less injection, the relative areas decreased substantially on reduction ofthe amount injected, indicating adsorption (Fig­ ure B29). Because no such decrease of the relative areas was observed after on-column injection, loss must have oc­ curred in the injector. ElhosUllimide 060

ON·COLUMN - - - - - - - - - - - - -..

0.50

04

0.3

Relative Response

~ n I'

SPLITLESS

0.1

Phenobarbital

g.60

e so ON·COLUMN 040 030 020 0.10

SPLITLESS 2

4

10

20

40

100

AmountIng)

Figure 829 Response to adsorptive drug components relative to that of an inert compound (n-hexadecane), on varying the amount injected (1-100 ng). The constant relative areas after on-col­ umn injection prove that adsorption in splltless injection occurred in the injector. (From Plotczyk. [28]).

Dependence on Temperature

Rapid information about adsorption in the column is obtained by testi ng the dependence of peak area on column tem­ perature. If the elution temperature of the critical solutes is increased (lower gas flow rate and/or faster programming), adsorption is reduced, i.e. peak areas should increase. Ad­ sorption in the vaporizing chamber is, of course, independ­

6.1. Adsorption in the Injector

131

ent of elution temperature. Increasing the injector tempera­ ture should reduce adsorption in the injector.

6. 1.4. Experimentally Observed Adsorption

Figure B30 shows chromatograms obtained after split in­ jection into a contaminated, empty liner at 200°C. All peaks in chromatogram A are highly tailed; broadening at their base corresponds to 15 to 20 s. The chromatograms were obtained by temperature programming, i.e. cold trapping had reconcentrated even broader bands. The solute material eluted last, in the tails of the peaks, must have entered the column 30-40 s after the bulk of the material. Adsorption was strong as the carrier gas flow rate through the vaporizing chamber (split flow rate) was 50 mLjmin.

Dependence on Split Flow Rate

When the split flow rate was increased to 200 rnt/rnin, the residence time of the most strongly retained material in the injector was reduced accordingly (chromatogram B) and the peak shapes improved. If, however, the flow rate had not been multiplied by four, but reduced by a factor of 20, to a flow typical of splitless injection, only a small proportion of the solute material would have reached the column entrance during the splitless period.

6.1.5. Variability of Adsorption

Chromatographers tend to be opportunists. Results are ac­ cepted when similar values are obtained from three injec­ tions. Peak areas which are too small, e.g. owing to adsorp­ tion, are of little concern because losses are considered to be taken into account by calibration of "response fac­ tors" (which should then rather be called "correction fac­ tors").

Deactivating By-Products

Even if calibration is performed by use of several different concentrations, such quantitation procedures are danger-

A

B

dirty Injector regular flow 10

C

dirty mjector fourfold flow 10

clean Injector regular flow 10 01

P 01 P III P

12

A

P

12 A

A

-

xl

Figure B30 Retention of solutes in a contaminated empty liner, observed upon split injection of the stand­ ard column teat mixture [29] at 30°C; temperature program, 2.5°/min to 80°C. In B, the flow rate during injection was increased fourfold by raising the inlet pressure; the pre-set split ratio (30:1) was not affected by this. Peaks, 10: n-decane, 01: 1-octanol, P: 2,6-dimethylphenol, 5: ethylhexanoic acid, A: 2.6-dimethylaniline, 12: n-dodecane. (From Grob and Grob [30)).

132

B 6. Retention and Adsorption in the Vaporizing Chamber

ous: adsorptivity in the injector often changes, sometimes

from one injection to the next.

For instance, the surface of a freshly cleaned liner is likely to

be rather active. Standards in pure solvent will be subject to

considerable losses. Subsequent introduction of the sample

may introduce by-products which deactivate the liner, with

the result that injections of equal amounts produce consid­

erably larger peaks. If the clean solution was used for cali­

bration, the results were excessively high.

Even if the calibration is repeated at the end of the sequence,

the effects of short-term deactivation (e.g. by water) cannot

be imitated.

Activating By-Products

The reverse situation is just as likely to occur: a freshly cleaned

or silylated liner has no adsorptivity, yet contaminants intro­

duced with the sample act as an adsorbent and render previ­

ous assessment of injector activity obsolete or misleading.

Methods are reliable only when losses by adsorption

in the injector are small.

6.2. Retention in the Injector

Quantitative sample transfer in splitless injection becomes

difficult when solute material is retained on the liner or on a

packing material. Retention might result from a layer of

sample by-products, e.g. fat from food extracts. Partition"

ing in this layer has the result that, e.g., only 10 % of a solute

is in the gas phase and the gas must be replaced many times

to achieve complete extraction from the layer.

Retention is more or less independent ofthe amount injected.

Retentive power cannot be saturated by solute material in

the same way as adsorptive sites.

Liner as Chromatographic (Pre-)Column

The liner can be regarded as a chromatographic column kept

isothermally at the injector temperature. During the splitless

period the carrier gas flow rate is a few mLlmin, correspond­

ing to a gas velocity of merely a few rnrn/s, Retention corre­

sponding to a distribution constant, K, of unity doubles the

volume of gas required to elute the solute or, in our case, to

transfer it to the column entrance.

Contamination as a Retain­ ing Stationary Phase

100 Jl9 involatile material (2 ~L, 5 % in the solution) de­

posited on the liner wall in the injection zone produces a

layer with an average thickness of 0.3 um, This is similar to

the film thickness of a standard capillary column, but because

the bore of the liner is much wider, the retentive power

(phase ratio) is roughly 100 times lower. This estimate

shows that large amounts of material must be deposited on

to the liner wall if significant retentive power is to be built

up.

7.1. Deactivation of the Liners

133

7. Deactivation of Liners and Packing Materials 7.1. Deactivation of the Liners?

There is general agreement that injector packings, such as glass wool, must be deactivated, but it is less obvious whether this is also necessary for liners. Although no statistics have been reported, it would seem that in the past packing materials were nearly always deacti­ vated, whereas most applications were performed us­ ing untreated liners. Silanized glass wool and deactivated materials for packed column GC have been available for a long time, but silanized liners became common only in the nineteen nineties. Few applications have been reported which cannot be per­ formed with an untreated liner but provide satisfactory re­ sults if it is deactivated. Most users of deactivated liners have never .tested whether raw glass would have been equally suitable.

Little Contact

At first it does not seem plausible that liners made of ordi­ nary glass should be less critical than deactivated packings of similar or purer material, but practical experience shows it to be so. The difference lies in the more intimate contact with a packing material (primarily true when the liquid is deposited on to it after injection with band formation). In general, there is little contact with the liner. If there is con­ tact, the solutes are carried there by matrix material, which then determines the activity of the surface.

Drifting Results

The first injections into new or freshly cleaned liners often produce results which differ from those obtained later - usu­ ally results improve. Use of a deactivated liner might reduce initial activity, but results nevertheless drift and 3-5 injec­ tions are still required to check the stability of the perform­ ance.

7.2. Deactivation of

Probably the only - but important - argument against pack­ ing the injector with a material such as glass, quartz, or fused silica wool is adsorptivity and/or chemically activity. Com­ mercially available "deactivated" wool is more active than the rest of the analytical system. In this sense it is of insufficient quality.

Commercial Wool

Wool Turning Gray

New silylated glass wool is bright white. After heating in an injector to maybe 250°C for, say, 30 min, it often turns slightly gray, even if nothing is injected. This reminds us of experi­

134

B 7. Deactivation of Liners and Packing Materials ence gained when straightening the ends of glass capillary columns by use of a small flame [311. When the ends were coated with stationary phase and no air (oxygen) was passed through them, they acquired a slightly gray appearance which was just visible when the end section was viewed more or less along its axis. Hardly any compounds could be eluted from such columns; even alkanes were completely adsorbed. Heating apparently produced extremely adsorptive carbon­ ized material. This might happen to wool also.

7.3. Application-Related Testing for Inertness

Treatment of packings or liners must be tested to check the success. For a long time little testingwas performed and there was justified doubt that the deactivation was of much use, in particular that it resisted the temperature in the injec­ tor for more than a short time.

Phenols

Many analysts tested liners or liner packings with the com­ ponents in which they were interested, but few published the results. Kalman [32] reported adsorption of phenols on silanized glass wool and degradation of nitrated phenols at injector temperatures above 225°C. He concluded that quan­ titative analysis of such components is possible only after careful optimization of conditions and "rigorous attention to system activity". Phenols are primarily lost on basic sites and by formation of charge transfer complexes with certain ions.

Methyl Esters

Bayer and Liu [33] investigated the analysis of C,o-C 20 fatty acid methyl esters with a liner containing different kinds of glass wool. No esters were detected when silanized glass wool was used at an injector temperature of 300°C, nor when glass wool washed with chromic acid and water was used at 200 °C. Reasonable results were, however, obtained when the latter material was used at 300°C. When glass wool was washed with hydrochloric acid and water, fair results were achieved at 200 °C, but there was less discrimination against high-boilers at 300°C. The poor results are surprising, because it is common prac­ tice to analyze fatty acid methyl esters with such packings. It confirms that the method used for silylation of glass wool is rather critical.

Endrin

Endrin, an organochlorine pesticide, is the only test compo­ nent being used more widely (e.g. Wylie et al. [3411. It is an injector-labile compound, and hence a good test for chemi­ cal activity. Chemically active surfaces also tend to be ad­ sorptive (maybe because the same silanol groups and salts are responsible for the activity), but there is no direct rela­ tionship between the two [35].

7.4. More Comprehen­ sive Testing Procedure

A more general testing procedure for injector liners and liner packing materials has been described [36]. It determines

7.4. More Comprehensive Testing Procedure

135

7.4.1. Design of the Test

adsorptivity. acid/base behavior. and degradation of labile components. Since the mechanisms involved are rather complex, it is impossible to standardize the test strictly. Con­ ditions must be adjusted according to the goals of the test.

Adsorptivity

Split injection and rapid isothermal chromatography (producing sharp peaks) enable sensitive determination of adsorption and retention. Delayed release of the test com­ ponents from the injector into the column is apparent by broadened or tailing peaks. Adsorption over longer periods of time is detected as reduced peak sizes. Use of a relatively low injector temperature (200°C) accentuates these results.

Degradation

Degradation is detected as reduced peak sizes or by the ap­ pearance of degradation products. It depends on the resi­ dence time in the injector and, hence, on whether split or splitless injection is applied. Effects are accentuated at high temperatures. In split injection, cooling by solvent evapo­ ration might also need to be considered, because it reduces the injector temperature below the level thermostatted.

Nebulization, Evaporation from Surfaces

Tests depend on the mode of sample evaporation. If vapori­ zation involves nebulization, vapors pass through the liner and packing materials, which results in weaker contact than if the solutes must evaporate from these surfaces after depo­ sition (e.g. by use of a fast autosampler).

Dependence on Amounts

Injector inertness, particularly adsorptlvltv, is highly depend­ ent on the amount of solute material injected. Glass or quartz wool are widely used for split injection of concentrated or even undiluted samples. If 100 Ilg of solute material (2 III of a 5 % solution) is distributed over 5 mg of wool, it forms a layer ca. 50 nm thick, i.e. only a few mol­ ecules are in direct contact with the surface. Adsorptivity by the surface will be a minor problem. At the other extreme, in trace analysis barely 1 pg of a com­ ponent is injected splitless and sensitively detected by ECD or MS. It merely covers a small part of the surface with a monomolecular layer, i.e. probably just the few most active sites. Problems will be severe.

7.4.2 Goals of the Test

The test can be used to obtain two types of information: Performance under the conditions used for a given analy­ sis, possibly with the intention of optimization. This can be used to identify the most suitable solvent, whether the sample liquid is nebulized, and whether split or splitless injection should be used. Comparison of different deactivation procedures or of materials from different suppliers. This presupposes the use of standardized conditions, irrespective of the con­ ditions used for an application.

136

B 7. Deactivation of Liners and Packing Materials

Test Components

Three test mixtures of different compositions are used. 1 The adsorption of polar functional groups is tested with cholesterol, that of aromatic groups (charge-transfer in­ teraction) with perylene. Peaks areas are compared with that from an inert compound, n-triacontane, to enable the detection of a loss. 2 Basic and acidic sites are determined by use of diethy­ lstilbestrol, with two phenol groups, and N,N'-diphenyl­ 1,4-phenylenediamine, with fairly basic aromatic amino groups. 3 Chemical activity is evaluated with endrin (epoxy group sensitive to acidic sites). DDT. and a fatty acid silylester (testing for hydrolytic activity, in particular free silanol groups).

7.4.3. Results

Figure B31 compares results obtained from different glass and fused silica wools. Some 30 mg of wool ii.e. 10 times more than necessary) were inserted into the vaporizing cham­ ber. 300 ng per test component were injected manually by the split method (FlO). Raw borosilicate glass wool had strong retentive power, as shown by the fact that broadened peaks were obtained for all the components (including the hydrocarbon). Chemi­ cal activity was fairly high, maybe for the same reason. Hy­ drolysis of the silyl ester was almost complete. Deactivation greatly improved the performance. The reduction in reten­ tion power suggests that some retaining material coating the fibers was removed. Raw fused silica wool provided far better results than the glass wool. For perylene the raw wool even performed bet­ ter than deactivated borosilicate wool, although chemical activity was somewhat higher (5 % hydrolysis of the silyl ester). Deactivation (by use of a polymeric material) improved chemical inertness (bottom row). but also increased reten­ tive power, as recognized by the broadening of the peaks. High retentive power hinders the release of high-boiling sol­ utes and, particularly, transfer in splitless injection. It neces­ sitates the use of higher injector temperatures.

Adsorptivity with Smaller Amounts

Some tests were repeated by injection of 30-1000 times less test material and detecting by MS. They confirmed that adsorptivity is highly dependent on the amount injected. No peak was detected for cholesterol when a 10 ng/IlL so­ lution was injected, either with an empty liner or into liners packed with raw or deactivated fused silica wool. Hence a higher injector temperature was needed to overcome adsorp­ tivity. With a 1 ng/IlL solution of perylene, an empty, raw liner resulted in a badly tailing peak. Surprisingly, use ofthe same liner packed with raw or deactivated fused silica wool re­ sulted in a substantially better peak shape.

7.4. More Comprehensive Testing Procedure

137

DES (10 ng/~L) did not form a peak when injected into a packing of raw fused silica wool, but produced a fair peak when deactivated wool was used.

DDT

Adsorption

Acid/Base

I

Degradation

Pery 30

EK Sil

End

DDT

DPD

30 Chele

DES

Pery

DDE

\ EK Sil

DES

End DDT

DPD Pery

30 Chele

Ac DES

as

,g U5

-g en

::J

Sil

EK End DDT

DPD Pery

30 Chele

U.

~CD

o

EK

Figure 831

Test results on raw and deactivated borosilicate and fused silica wool from Restek. The dura­

tion of each chromatogram was ca. 2.5 min.

• Adsorptivity test: perylene (Pery), ...triacontane (30), cholesterol (Chole). • Acidlbase test: diethylstilbestrol (DES, mixture of cis and trans), N,N-diphenyl-1,4-phe­ nylenediamine (DPD). • Degradation test: trimethylsilyl heptadecanoate (SiI), endrin (End), DDT. Degradation products: endrin aldehyde (EA), endrin ketone (EK), DDD, DDE, free C n acid (Ac). Split injection at 200°C injector temperafUre; packings of 30 mg wool. For further experi­ mental details, see [36].

138

B 7. Deactivation of Liners and Packing Materials

Degradation

Residence Time

Degradation of endrin and DDT was hardly dependent on

the amount injected. When ECDwas used with 10,000-100,000

times more dilute samples, raw fused silica wool resulted in

small peak areas for the degradation products. Injection into

the empty liner yielded nearly identical results.

Degradation depended on the injection technique applied.

Splitless injection prolongs the residence time in the va­

porizing chamber and, thus, increases thermal stress. Deg­

radation products now approached 10 % of the endrin and

DDT injected.

7.5. Silylation of Liners 7.5. 1. Background

Silylation of Laboratory Glassware

Deactivation of glass surfaces has been extensively investi­

gated for the preparation of glass capillary columns. The tech­

niques have been summarized by Kurt Grob [37).

The most efficient and most commonly applied deactivation

involves silylation, but not all silylation techniques are suit­

able: rapid methods of silylation are quite useless.

For critical applications, laboratory glassware is silylated by

immersion in a 10-15 % solution of trimethylchlorosilane

or dimethyldichlorosilane in toluene and heating to the

boiling point, 110 DC, for ca. 30 min. Similar conditions have

been used for silylating silica gel for HPLC. Fenimore et al.

[38) proposed vapor phase silylation with hexamethyl­

disilazane (HMDS) in an evacuated oven at 200 DC.

Muller and Stan [39) described an procedure for PTV liners.

Possibly packed with wool, they were immersed in 10 %

dimethyldichlorosilane in toluene overnight. Immediately

before mounting, liners were rinsed with toluene, methanol,

and again toluene. They were heated to 300 DC in an injector

to which an old column was connected to regulate a gas flow

purging the liner. Finally toluene (the solvent used for the

analysis) was injected a dozen times before re-adjusting the

injector temperature and starting analysis.

Experience Gained Produc­ In the late sixties and early seventies, low-temperature

ing Glass Capillary Columns silylation methods were used to produce glass capillary GC

columns, but inertness was unstable: after brief heating

above 200 DC. activity returned. The silyl groups bonded

to the surface probably disappeared into the bulk ofthe glass

as silanols beneath the surface picked them away.

Hinshaw(40) stated that silylated liners should not be heated

above 250 DC, as silylation is "reversed" at higher tempera­

tures - probably because of the same effect.

The breakthrough in the preparation of inert glass capillary

columns came as a result of two modifications of the proce­

dure.

1 The glass surface needed leaching, i.e. extraction

with 18 % hydrochloric acid at 160-170 DC overnight, to remove salts and to open strained siloxane groups. Such leaching results in a gel rich in silanols which must be handled with care.

7.5. Si!ylation of Liners

139

Subsequent silylation of the silanol groups had to be performed at temperatures of 350-400 DC [41-431. The procedure described below for achieving thermostable deactivation of glass or quartz liners is derived from these methods. 2

7.5.2. Wettability?

Nothing is known about whether wettability of liners by the sample liquid is relevant. Lack of wettability could prevent spreading and retention of certain sample liquids on the liner wall (high-boiling matrix). Wettability of packing materials is probably less of a problem, because liquids are retained by means other than surface interaction.

"Phesi!" Surfaces

Raw glass is wetted by most liquids; at elevated tempera­ ture, maybe even by water. Trimethylsilylated surfaces, on the other hand, have a low critical surface energy [44-461 and are not even wetted by aromatic and chlorinated sol­ vents, acetone, or ethyl acetate. To improve wettability, uncoated capillary pre-columns are commonly deactivated by phenyldimethyl silylation, with diphenvltetrarnethvldisilazane (DPTMDS) as reagent. These "phesil" surfaces are wetted by almost all organic sol­ vents: toluene and methanol are atthe critical limit, but water does not wet them.

7.5.3. Method Recom­ mended for Silylation of Liners

Leaching. Liners are immersed in 18 % hydrochloric acid

in a pressure-resisting glass tube. The tube is flame-sealed

and heated to 150 DC for several hours.

Rinsing. The liners are gently rinsed with 1 % hydrochlo­

ric acid (the labile silica on the surface must not be stripped

off).

Drying. They are then heated to 250 °C for 15 min for

drying. This should occur immediately after rinsing to pre­

vent ions migrating back to the surface from lower lay­

ers. The most simple means of heating is to insert them

into an injector without the septum cap.

SilylatiClQ. Liners are placed, one on top of the other, in

a glass tube with an internal diameter only slightly ex­

ceeding the outer diameter of the liners and the inlet of

the tube is pulled out for later flame sealing. Now for every

liner 100 J.1L diphenyltetramethyldisilazane (DPTMDS,

Fluka) is introduced into the tube by means of a syringe,

leaving the entrance clean. The (clean) tube is flame-sealed

under vacuum. It is rotated until the liners are wetted with

reagent. Silylation is performed by heating to 400 °C over­

night in a GC oven, after wrapping some glass wool

around the parts of the tube which would otherwise come

into contact with the GC oven (to avoid tension and pos­

sible breakage).

Washing. The silylated liners are washed with toluene,

methanol, and ether.

140

B 7. Deactivation of Liners and Packing Materials

Conditioning. Freshly prepared liners release oligomeric silylation material during the first injections, possibly as a result of reaction with hydroxyl compounds (water?) in the samples. For this reason, the first sample should be injected at least twice to rule out "ghost peaks".

7.6. Silylation of Glass and Quartz Wool

It seems that no thoroughly optimized procedure for deacti­

vating wool has been described in the literature. The discus­

sion below is, therefore, not conclusive.

Cleaning

Raw wool must be cleaned to remove residual organic ma­

terial, e.g. by warming in dimethylformamide. Such impuri­

ties were probably the reason for the high retentive power

observed in Figure B31.

Leaching, Silylation

Leaching of glass wool cannot be performed by the above

procedure because the fibers are attacked too deeply and

fragment into short pieces.

Quartz wool resists leaching better. Because less salt must

be extracted, leaching can be performed in the gas phase.

The acid is added to the bottom of the tube in which the

wool is heated. Quartz wool is, however, brittle and read­

ily forms small pieces which might be carried into the col-:

umn or the split line by the carrier gas. Silylation is performed

as described above for liners.

7.7. Packings Coated with Stationary Phase

Although it is certainly an exaggeration to describe packed

columns as inert - capillary columns are usually more so ­

they are more inert than the commonly used silylated

wool.

Deactivation by Stationary Phase

In packed column GC it is often assumed that only surfaces

coated with stationary phase can be inert. Because the sol­

ute material diffuses through the stationary phase down to

the solid surface, the stationary phase does not prevent con­

tact - there is no "shielding effect" in this sense. Stationary

phases, particularly polysiloxanes, do, however, react with

silanols on the silica by mechanisms summarized by Welsch

and Teichmann [471. As stationary phases are permanently

present, they are always ready to react with newly gener­

ated silanols.

Retentive Power of Coated Packing Materials

Column packing materials have a large surface area and a

substantial coating of stationary phase is required to pro­

vide the best inertness. This results in high retentive power,

or that high-boiling compounds are released only at high

injector temperatures.

"Ghost Peaks"

Packings coated with stationary phase often produce "ghost

peaks" in the chromatograms. Degradation materials

(bleed) entering the column at low temperatures are recon­

7.7. Packings Coated with Stationary Phase

141

centrated there and form sharp initial bands. Upon tempera­ ture programming they are eluted as sharp peaks. The main bleed components from methyl silicones, hexamethyl­ cyclotrisiloxane (D3) and octamethylcyclotetrasiloxane (D4), are eluted somewhat below 250 °C from most common thin film apolar columns. Particles Entering Column

Almost inevitably short glass wool fibers or fragments (dust) of packing materials are driven into the oven-thermostatted column by the carrier gas. There have been no reports on their effect on chromatography, but it is expected that they will be more adsorptive in the column than in the injector, because the column temperature is usually lower. From on-line coupled LC-GC it is known that silica gel trans­ ferred to GC with the LC fraction adsorbs polar compounds rather strongly, almost irrespective of whether raw or derivatized silica gel is involved.

7.8. Deactivation by Sample Material

Injection of samples containing high concentrations of involatile by-products or active components can improve the injector performance just as it may create addi­ tional problems.

"Priming" of Packed Col­ umns

Raw glass liners are often efficiently deactivated after a few injections. The effect of such "priming", well known in packed column GC, is manifested in the often substantial improve­ ment in results during the course of continuing meas­ urements.

7.8. 1. Unstable Deactiva­

Often such deactivation is unstable (especially if injector tem­ peratures are high), as we learn from the experience that the first injection in the morning frequently produces results differing from those obtained either the day before or during subsequent analyses. A change in injector performance is certainly one, although hardly the only, explanation of this phenomenon (column performance is another).

tion

Many Possible Mechanisms

Changing performance might arise from loss of deactiva­ tion during waiting periods, e.g., evaporation of the mate­ rial shielding the active site (such as water from the humid­ ity of samples). There may be a thermal decay of deactivat­ ing material and evaporation of the resulting fragments, or degradation promoted by air penetrating the injector. Loss of inertness could also be the result of newly created activ­ ity, e.g. the formation of adsorbents similar to charcoal, or as a result of older adsorbents, from which previously in­ troduced material has been thermally desorbed and which have again become active. Inertness can also improve overnight, e.g. as a result of chemical transformation of a layer of deactivating material. Contaminants on the liner wall turn black, but also become more solid, less solvent-soluble, and less accessible to sam­

142

B 7. Deactivation of Liners and Packing Materials pie components. Polymerization is involved, but also dehydroxylation, possibly removing adsorbing functional groups.

7.8.2. Heating Injector Overnight and at Week­ ends?

Such considerations suggest that the first results obtained in a series of analyses depend on how the instrument was switched to stand-by, e.g., overnight and at week­ ends. The injector usually remains heated overnight and at week­ ends because re-heating in the morning is time-consuming (thermostatting takes longer than the readout of the instru­ ment suggests, since the thermocouple is located in the heat­ ing block near the heating cartridges). Furthermore, cooling and heating causes the septum, the ferrule of the column attachment, and a possible ferrule tightening the liner against the injector body to contract and expand, which reduces the reliability of the seal. Whether or not the injector remains heated influences the injector performance in a direction which can be positive or. negative.

7.8.3. Carrier Gas Over­

To prevent oxidative degradation of the deactivation in the liner, it seems plausible that the injector should be kept under inert gas, i.e. that some carrier gas inlet pressure should be maintained whilst the instrument is not in use. Exchange of the column could be a problem, because the carrier gas is switched off while the injector remains hot. Such effects, although plausible, have not been experi­ mentally confirmed. In the cases tested the first results next morning were equally good or bad whether or not the carrier gas was completely switched off overnight. For Carbowax-type columns, some carrier gas must remain run­ ning because of the sensitivity of these stationary phases to oxygen, but not for columns coated with silicones.

night?

2-3 Injections Before Baking Out the Column

A simple method seems to be effective. When the instru­ ment is switched on in the morning or after a column or liner has been newly installed, a sample is injected 2-3 times, with an interval of maybe one or a few minutes. Only then is the column baked out. This procedure costs hardly any extra time, but clearly improves the accuracy of the first results.

7.8.4. Tests with Sample

Sample by-products can also increase activity in the vapor­ izing chamber. Extended studies on the feasibility of analyzing certain components and the probable accuracy of the results are quite useless when performed with "clean" solutions of standards. After injection of the first real sample, all the "beauty" might have vanished - or shows up only then. Re­ sponse factors, recoveries of labile components, and repro­ ducibilities should thus be determined by use of spiked sam­ ples.

8.1. Washing with Strong Acids or Bases

143

8. Cleaning of Injector Liners Cleaning normally refers to the liner wall, and maybe obsta­ cles therein. Packing materials are replaced. The treatment should remove the contaminants, but not the deacti­ vation, neither should it produce new active sites. Extent of Cleaning Required

Whether or not results are affected by the cleanliness of the liner depends on the nature of the sample and its evapora­ tion. For some samples, liners need frequent cleaning. With other samples, good results are obtained from liners containing brown deposits of contaminants - some­ times results are even better than from new or freshly cleaned llners.It does not always seem to be desirable to work with absolutely clean liners.

8.1. Washing with Strong Acids or Bases

Drastic cleaning methods include washing with hot hydrox­ ide solutions, or with nitric, sulfuric, or even chromic acids.

Mineral Acids

Chromic acid should be avoided as it introduces highly adsorptive and chemically active chromic ions into the silica surface. Once they are introduced, it is almost impossible to remove them. Immersion in warm mineral acids (hydrochloric, sulfuric, or nitric acids) for, say, 15 min, efficiently removes some con­ taminant, but also generates silanol groups in the silica surface, either by opening siloxane bonds or by removing chemically bonded or adsorbed deactivating material (leach­ ing process).

Hydroxide Solutions

Warmed hydroxide solutions afford more radical cleaning (e.g. by saponification of polymerized fat), but they also at­ tack silica more strongly, dissolving a surface layer ofthe glass. From the preparation of capillary columns it is known that alkali leaves behind extremely adsorptive surfaces [481, maybe by formation of vicinal silanols. The treatment with alkali must be followed by immersion in warm hydro­ chloric acid (e.g. 18 %) and rinsing with 0.1 % hydrochloric acid. If polymerized fat is the main contaminant, warming in a solution of 5 % sodium methoxide in methanol (5 min, 50°C) is more gentle and at least equally effective: transmethyla­ tion depolymerizes the material.

Re-Deactivation

Cleaned liners can be dried in an open hot injector for a few minutes. Sometimes they can be appropriately re-deactivated

UNIVERSIDAD DE ANTIOQUlA

BIBLIOTECA CENTRAL

144

B 8. Cleaning of Injector Liners

by repeated injection of a sample (the sample is injected several times without baking out the column). Even "condi­ tioning" (simple heating) overnight has a surprisingly posi­ tive effect. It is, nevertheless, often a disappointing experi­ ence to discover that a freshly cleaned liner performs worse than it did before cleaning. Cleaned liners can be re-silylated. After rather gentle clean­ ing with acid it is sufficient to silylate the surface by treat­ ment in hot toluene (immersion in a 10-15 % solution of trimethylchlorosilane or dimethyldichlorosilane in toluene and heating to the boiling point, 110°C, for 30 min).

8.2. Burning the Con­ taminants

Organic contaminants can be oxidized in a stream of air. The liner is held in the blue flame of a Bunsen burner with an old pair of tweezers (as a result of annealing, the tips of the tweezers become soft). The liner is held steeply inclined, with the lower end protruding from the flame. In this way, air is sucked through the tube. After cooling, it may be necessary to rinse the liner with water to remove salts (alkali metal ox­ ides).

Danger of Bending the Liner

The flame method can only be recommended for liners which do not fit tightly in the injector. As they are all too easily slightly bent, such liners still enter the injector, but can no longer be removed (or only after tapping them to powder with a hammer and a screw driver of suitable di­ mensions). Oxidative cleaning is highly efficient and leaves behind a surface of rather low activity. Silylation of the original deactivation is, of course, removed.

8.3. Gentle Cleaning

Gentle cleaning with solvent (methanol, acetone) and some scratching is usually sufficient even when it is visually ap­ parent that the process is incomplete. The resulting surfaces become less active. Silylation is not attacked, nor is the deactivation obtained by reaction ofthe sample material with the surface. The method is, of course, applicable to open­ tubular liners only.

Procedure with Acetone

A ball of crumpled tissue paper is pushed into the liner and compressed, e.g. with the tip of a Pasteur pipette, against a finger at the exit of the tube. Paper compressed in this way also presses against the liner wall. It is then soaked with sol­ vent (e.g. acetone) and moved up and down by means of the pipette. Sometimes some additional scraping is necessary, e.g. by using a small spatula with a bent tip.

B References

145

References B K. Grob, "Sample Evaporation in Conventional GC Split/Splitless Injectors; Part 1: Some Quantitative Estimates Concerning Heat Consumption during Evaporation", HRC 15 (1992) 190. 2 Handbook of Chemistry and Physics, 66th edition, R.C.Weast (Ed.), CRCPress, Boca Raton (1985) D-173. 3 Landolt-Bornstein, Physikalisch-Chemische Tabellen II, Springer (1923) 1434. 4 Taschenbuch fUr Chemiker und Physiker, 2nd edition, J.D'Ans and E. Lax (Eds) Springer, Berlin (1949) 1130. 5 A.E. Kaufman and C.E. Po/ymeropou/os, "Study of the Injection Process in a GC Split Injection Port", J. Chromatogr. 454 (1988) 23. J. Bowermaster, "The Influence of Evaporation Dynamics on Precision in Split Capillary 6 GC", HRC & CC 11 (1988) 802. 7 K. Grob and M. De Martin, "Sample Evaporation in Conventional GC Split/Splitless Injec­ tors; Part 2: Three Scenarios Visually Observed in Empty Injector Inserts Using Perylene", HRC 15 (1992) 335. 8 K. Grob and M. Biedermann, "Video-Taped Sample Evaporation in Hot Chambers Simu­ lating GC Split/Splitless Injectors; Part 1,Thermospray Injection", J. Chromatogr. (in press). 9 J. Dian, C.E. Po/ymeropu/os, and R. Ulisse, "Liquid Jet Evolution from a GC Injector", J. Chromatogr. 609 (1992) 269. 10K. Grob and M. De Martin, "Sample Evaporation in Conventional Split/Splitless GC Injec­ tors; Part 3: Retaining the Liquid in the Vaporizing Chamber", HRC 15 (1992) 399. 11 ~G. Jennings, "Glass Inlet Splitter for GC", J. Chromatogr. Sci. 13 (1975) 185. 12 E. Bayer and G.H. Liu, "New Split Injection Technique in Capillary Column GC", J. Chromatogr. 256 (1983) 201. 13 G. Liu and Z. Xin, "The Glass Insert in Stop-Flow Split Injection", Chromatographia 29 (1990) 385. 14 K. Grob and C. Wagner, "Inserts Promoting Sample Evaporation in Vaporizing GC Injec­ tors: Effectiveness and Inertness", HRC 16 (1993) 429. 15 G. Schomburg, H. Beh/au, R. Die/mann, F. Weeke, and H. Husmann, "Sampling Tech­ niques in Capillary GC", J. Chromatogr. 142 (1977) 87. 16 J.L. Marshall and B. Crowe, "Evaluation and Optimization of a Splitter Injector for Capil­ lary Chromatography", Chromatographia 19 (1984) 335. 17 K. Grob and M. Biedermann, "Video-Taped Sample Evaporation in Hot Chambers Simu­ lating GC Split/Splitless Injectors; Part 2, Injection with Band Formation", J. Chromatogr. A (2000). 18 O. Driessen and J. Lugtenberg, "Determination of Adsorption by Fluorescence in GC", HRC & CC 3 (1980) 405. 19 O. Wiiestand W. Meier, "Bestimmung von sieben insecticiden Carbamaten auf Fruchten und Gernusen mit Capillar-Saulen GC und AFID", Z. Lebensm. Unters. Forsch.177 (1983) 25. 20 K.Grob, "Evaluation of Capillary GC for Thermolabile Phenylurea Herbicides; Compari­ son of different Columns Including Fused Silica", J. Chromatogr. 208 (1981) 217.

146

B References

21 H.-M. Muller and H.-J. Stan, "Thermal Degradation Observed with Different Injection Techniques: Quantitative Estimation by the Use of Thermolabile Carbamate Pesticides", HRC 13 (1990) 759. 22 V. Vignier, F. Berthou, and D. Picert, "Influence of Injection Systems on the GC Analysis of Oxygenated Dibenzothiophenes", HRC & CC 6 (1983) 661. 23 J. T. Anderson, "Stability of Dibenzothiophene Oxides Under GC Conditions", HRC & CC 7 (1984) 334. 24 S. Klick, "Evaluation of Different Injection Techniques in the GC Determination of Ther­ molabile Trace Impurities in a Drug Substance", J. Chromatogr. 689 (1995) 69. 25 J. votmut, E. Matisova, and P.T. Hs, "Influence of Injector Liner Packing on the Analysis of Antiepileptic Drugs by Capillary GC", HRC 12 (1989) 760. 26 H. Husmann, G. Schomburg, K.-D. Muller, H.P. Nsiik, and G. von Recklinghausen, "GC FAME Analysis for Characterization of Bacteria: Formation of Aldehydes from 3-Hydroxy Fatty Acid Methyl Esters During Sample Introduction", HRC 13 (1990) 780. 27 J.N. Cardoso and J.C. Afonso, "Behavior of Branched Bromo-alkanes in Vaporizing In­ jectors", HRC &CC 11 (1988) 537. 28 L.L. Plotczyk, "Application of Fused-Silica Capillary GC to the Analysis of Underivatized Drugs", J. Chromatogr. 240 (1982) 349. 29 K. Grob, G. Grob, and K. Grob, "Comprehensive, Standardized Quality test for Glass Capillary Columns", J. Chrornatoqr, 156 (1978) 1. 30 K. Grob and G. Grob, "Practical Capillary GC - a Systematic Approach", HRC & CC 2 (1979) 109. 31 K. Grob, G. Grob, B. Brechbuehler, and P. Pichler, "Straightening of the Ends of Glass Capillary Columns", J. Chromatogr. 205 (1981) 1. 32 D. Kalman, "Optimized Injection for Determination of Free Phenols by GC using Fused Silica Columns", HRC & CC 6 (1983) 564. 33 E. Bayer and G.H. Liu, "New Split Injection Technique in Capillary Column GC", J. Chromatogr. 256 (1983) 201. 34 P.L. Wylie, K.J. Klein, M.a. Thompson, and B. vv. Hermann, "Using Electronic Pressure Programming to Reduce the Decomposition of Labile Compounds in Splitless Injection", HRC 15 (1992) 763. 35 K. Grob, "Evaluation of Capillary GC for Thermolabile Phenylurea Herbicides; Compari­ son of Different Columns Including Fused Silica", J. Chromatogr. 208 (1981) 217. 36 K. Grob und Ch. Wagner, "Procedure for Testing Inertness of Inserts and Insert Packing Materials for GC Injectors", HRC 16 (1993) 464. 37 Kurt Grob, "Making and Manipulating Capillary Columns for GC", Huethig, Heidelberg, 1986. 38 D.C. Fenimore, C.M. Chester, J.H. Whitford, and C. A. Harrington, "Vapor Phase Silylation of Laboratory Glassware", Anal. Chern. 48 (1976) 2289. 39 H.-M. Muller and H.-J. Stan, "Pesticide Residue Analysis in Food with CGC - Study of the Long-Term Stability by the Use of Different Injection Techniques", HRC 13 (1990) 697. 40 J. V. Hinshaw, "GC Troubleshooting, Detectors", LC-GC Int. 1/3 (1988) 24. 41 Th. Welsch, vv. Engewald, and Ch. Klaucke, Chromatographia 10 (1977) 22. 42 K. Grob, G. Grob, and K. Grob, "Deactivation of Glass Capillary Columns by Silylation", HRC & CC 2 (1979) 31. 43 K. Grob, G. Grob, vv. Blum, and vv. Walther, "Preparation of Inert Glass Capillary Col­ umns for GC", J. Chromatogr. 244 (1982) 197. 44 M. vv. Ogden and H.M. McNair, "Characteristics of Fused Silica Capillary Tubing by Con­ tact Angle Measurement", J. Chromatogr. 354 (1986) 7. 45 T. Welsch, R. Muller, vv. Engewald, and G. Werner, "Surface Modification of Glass Capil­ laries by High Temperature Silylation", J. Chromatogr. 241 (1982) 41. 46 K. Grob, "On-Line Coupled LC-GC", Huthiq, Heidelberg, p. 186.

B References

147

47 Th. Welsch and U. Teichmann, "The Thermal Immobilization of Hydroxy-Terminated Silicone Phases in High-Temperature-Silylated Glass Capillaries. A Study of Reaction Mechanisms." HRC 14 (1991) 153. 48 K. Grab and Th. Varburger, "Testing the Polarity and Adsorptivity of Nondeactivated GC Capillary Surfaces", HRC 19 (1996) 27.

1.1. Principles of Split Injection

149

C Split Injection

1. Introduction 1.1. Principles of Split Injection

During split injection the vaporized sample is divided into two unequal parts: a small proportion ofthe sample is trans­ ferred into the column by the flow of carrier gas, whereas the major part is vented through the split exit. The pro­ portion of sample material entering the column is usually between 0.3 and 20 % of the material introduced into the injector.

Different Types of Split Injection

There are three types of split injection: 1 injection into a permanently hot chamber ("conven­ tional vaporizing injection"), treated here; 2 Programmed temperature vaporizing (PTV) split injec­ tion; 3 On-column split injection, which involves splitting of the sample after vaporization in an uncoated pre-col­ umn [1,2).

1. 1. 1. Basic Injector Design

Figure C1 shows the design of a split injector schematically. The carrier gas is introduced at the top of the vaporizing chamber housed in a glass tube ("injector insert" or "Iiner"). It drives the sample vapor towards the column en­ trance, which is usually positioned ca. 5 mm above the base of the chamber. At the split point, the vapor is divided into a small stream entering the column and a main stream vented through the split outlet. The latter flows around the bot­ tom of the liner and between the outer wall of the liner and the injector body upwards into the outlet line.

150

C 1. Introduction Syringe Septum

.crc:==- Septum r

purge outlet

Carrier I

gas -'iiiiii~~.

.-;:o""'"...........~~

Split outlet

e.g. 40 mUmln

Vaporizing chamber-

I'--- Sample vapors

Injector insert­

,.,••,....... Split point

••

Column

e.g. 2 mUmin

Figure C1 Basic design of the split injector. A small proportion of the sample vapor enters the column whereas most of the mat. rial is removed through the split outlet. With a column flow rate of 2 mLJmin and a split flow rate of 40 mLJmin, the split ratio is 20:1, i.e. ca. 5 % of the material injected is analyzed:

1.2. Purposes of Sample Splitting

Sample splitting is used for two totally different purposes, i.e. to reduce the amount of material reaching the column and to achieve sharp initial bands. This is one reason split injection is used for widely differing applications.

1.2.1. Injection of Con­ centrated Samples

Splitting is used because there is no device enabling the han­ dling of appropriately small volumes of sample material. The smallest amount reliably introduced by means of a syringe is ca. 1 Ill, corresponding to approximately 1 mg. This dras­ tically overloads capillary columns, because it is similar to the total amount of stationary phase in the column. Capaci­ ties of standard capillary columns (referring to a single component) range between 20 and 500 ng, i.e. 2000-50,000 times less than can be injected. To reduce the amounts of solute material to this level, the sample is either strongly di­ luted or injected with splitting. Split injection enables the introduction of relatively concen­ trated samples, which is primarily of interest for samples which cannot be diluted or where dilution is incon­ venient.

Avoidance of Solvent Effects

In the opinion of the originators, splitting of the sample was the only way to keep the sample volumes sufficiently small not to overtax the small "sample capacity" of capillary columns (understood as capacity for the total amount of sam­ ple material}. The statement that capillary columns have a small sample capacity does not hold true. Today several

1.2. Purposes of Sample Splitting

151

milliliters of sample are introduced by on-line lC-GC, as highly diluted solutions, of course. Split injection is, how­ ever, the only sample introduction technique which elimi­ nates significant effects of the sample on its own chro­ matography, i.e. the so-called solvent effects. 1.2.2. Splitting to Gener­ ate Sharp Initial Bands

Chromatography requires sharp initial bands. The solute ma­ terial must be introduced into the column such that it is dis­ tributed over a column section which is clearly shorter than the section over which the material is spread when leaving the column. Otherwise, injection causes peak broadening and reduces the separation efficiency of the system.

Only Split Injection Produces Sharp Bands

As will be discussed in more detail in Section C4, only split injection generally produces initial bands which are suffi­ ciently sharp to suit capillary GC. For splitless or on-column injection, reconcentration effects (cold trapping, solvent effects, retention gap techniques) must be exploited to re­ sharpen broad initial bands. In the absence of reconcentration effects, the width of the initiai bands corresponds to the time during which sample is introduced into the column. If splitting is not used, the cloud of sample vapor resulting from an injection of, e.g., 2 III of liquid, ca. 0.5-2 rnl., enters the column in some 30-90 s, which is far beyond what can be accepted if peaks of only a few seconds width are expected.

Column Picking out Small Portion

Because of the split flow rate, the sample vapor moves rapidly past the column entrance. Sample material en­ ters the column during a short period of time only. The col­ umn picks up a small portion ofthe cloud passing rapidly by its entrance. Initial bands are correspondingly sharp. As mentioned above, several techniques can be used to reconcentrate initial bands. Split injection must be used, how­ ever, if these reconcentration techniques fail (e.g. injection of headspace or gases) or if they are inconvenient (e.g. be­ cause they entail cooling of the oven).

Loss of Sensitivity

The sharp initial bands achieved by splitting are paid for in terms of reduced sensitivity. This is a problem in trace analysis, where the loss of sample material might be too high a price to pay for obtaining sharp initial bands.

1.3. The Two Principles of Gas Supply

Two fundamentally different carrier gas regulation systems are used for split injection. They are described just briefly, because we shall return to them in later sections.

Pressure Regulator/Flow Resistance Systeme

The carrier gas is fed through a pressure regulator located upstream of the injector (Figure C2). The inlet pressure de­ termines the column flow rate. The split flow rate is adjusted by means of a variable restriction, e.g. a needle valve or an electronlcflow regulator (proportional valve).

152

C 1. Introduction I ----...:;r:-----.~ Septum Ipurge

Vaporizing chamber

Needle valve

f--'-

SO'"

I

o",~1 ~

Flow meter

Column Figure C2 Carrier gas supply by means of a pressure regulator/flow resistance system.

Flow/Backpressure Regulator System

HewleU-PackardiAgilent introduced a system which feeds the carrier gas to the injector through a flow regulator (Fig­ ure C3) [3] adjusting the sum of the flow rates through the split exit, into the column, and through the septum purge. The column flow rate is still pressure-regulated, but the pres­ sure regulator is situated in the split outlet line. It acts like a dam, keeping pressure behind itself at the level required to obtain the desired column flow rate. l"""----l.~ . Septum purge .1------::-1 Total flow rate

Column

Figure C3

Flow!backpressure regulator system.

Backpressure Regulation

A backpressure regulator differs from a standard regulator: whereas standard pressure regulators release an amount of gas which establishes a predetermined pressure downstream (the pressure behind the regulator is not of importance), a backpressure regulator releases an amount of gas which es­ tablishes a predetermined pressure upstream of the regula­ tor.

1.4. Historic Background of Split Injection

The early history, in particular, was reported in a lively man­ ner by Ettre [4]. Split injection was the first method of sam­ ple introduction in capillary GC, introduced in 1958 by Desty [5]. It remained practically the only injection technique avail­

1.4. Historic Background of Split Injection

153

able for more than 10 years. Its concept was largely derived from theory, which led to the astonishing fact that splitless injection was only studied seriously ca. 10 years after the introduction of split injection. Extrapolation from Packed Column GC

Split injection was extrapolated from packed column GC. As derived by Ettre, a typical packed column contains ca. 230 mg stationary phase - it is merely about 5 mg in the average capillary column. In a capillary column, the effective volume of a theoretical plate is three orders of magnitude smaller than in a packed column, which means the sample capac­ ity is at least two orders of magnitude lower. It was assumed that only extremely small plugs of sample vapor could provide sufficiently sharp initial bands. Because these are smaller than those resulting from the volumes handled by syringes, splitting seemed to be the only answer.

Mineral Oi/lndustry

Typical samples analyzed by early capillary GC were hydro­ carbons from refineries and related industries (hardly any other compounds would have passed through the very long and highly adsorptive steel capillaries in use at that time). Undiluted samples were injected at extremely high split ratios..

Problems with Quantitative Analysis

The quantitative results obtained by split injection were of­ ten poor, which gave capillary GC the reputation of being efficient in separation but poor in accuracy and precision. During the nineteen sixties several groups investigated the problems and tried to improve the technique. At Perkin-Elmer there was the group of Condon [6], Ettre and Averill [7], in Germany that of Bruderreck, Halasz, and Schneider [8]. Dis­ cussions primarily centered on the linearity of splitting, i.e. the problem of achieving the same split ratio for all com­ ponents of an injected mixture. The complicated injectors proposed never became widely used, however, and the suc­ cess of all these efforts was limited.

Development in the Seven­ ties

The work was continued in the nineteen seventies, primarily by pragmatic workers interested in the analysis of their sam­ ples rather than the systematic development of new devices and techniques. The range of samples analyzed by capillary GC grew rapidly (also, in part, as a result of wastly improved columns). Now highly diluted solutions were commonly injected at low split ratios (or by splitless injection, which became popular in the same period). Important names of that time were the Hornings [9], Schomburg [10], and Jennings [11]. No working group achieved comprehensive explanation of the sources of often massive errors, and the transfer of con­ ditions found suitable for one type of sample to another usu­ ally led to disappointment. In a collaborative study on a min­ erai oil fraction, performed in 1978/79, some laboratories came to the conclusion that the last alkane present in the

154

C 1. Introduction mixture was octacosane (C2a), whereas others even found

alkanes beyond tetracontane (C40 ) . This showed that prob­

lems were not just limited to increased standard deviations.

Since poor results are seldom published, little was made

public, however.

At the end of the seventies, no breakthrough to a general

solution of the problems of quantitative analysis had been

achieved, and because of the excessive complexity of the

problems, such a breakthrough was no longer even consid­

ered possible.

Non-Linearity and Needle Problems

The distorted composition of the sample entering the col­

umn ("discrimination" against those components which

entered the column in proportions which were too low) was

attributed to "non-linearity" of splitting, i.e. unequal split­

ting of different components. It was, therefore, considered

to be a phenomenon peculiar to split injection. Only in 1979

was it fully realized that the predominant source of error was

often the selective elution of different components from the

syringe needle [12), which also affects splitless injection.

Alternative Injection Tech­ niques

At the end of the nineteen seventies, efforts were invested in

the development of better alternatives rather than the con­

tinued pursuit of a hopeless issue. On-column injection and

programmed temperature vaporizing (PlY) injection were

proposed. Around 1980, most would have agreed that clas­

sical split injection would soon be replaced for most applica­

tions.

Development in the Nineteen Eighties and Nineties

Leading analysts immediately accepted on-eolumn and PTV

injection as far superior to conventional vaporizing tech­

niques. To determine the "true" result, on-column injection

became the standard method. To the great surprise (and dis­

appointment) of all involved, however, most GC analysts

continued using split and splitless injection. Numer­

ous on-column injectors were sold, but seldom used for rou­

tine analysis. The impact of the PlY injector was even lower.

In the methods developed in the nineteen nineties conven­

tional split and splitJess injection were still by far the most

commonly used injection techniques.

Analysts have proved to be extremely conservative. Although capillary GC became broadly applied only in the eighties, introduction of better injection techniques seemed to arrive too late. Newcomers usually started with the most difficult of all methods, i.e. split injection. If results were un­ satisfactory, they changed to HPLC, rather than trying better injection methods. Instrument manufacturers invested little in improved in­ jector designs, and new instruments were mostly equipped with the old devices; not even rather obvious deficiencies were removed. At the end of the eighties, electronic carrier gas regula­ tion was introduced by Hewlett-Packard and was further

1.4. Historic Background of Split Injection

155

developed into rather complex systems enabling computer­ control of all flow rates. Rather than introducing a basic im­ provement, this was a contribution to the "Quality Man~ agement" which became fashionable. Huge resources were invested in "quality" and sophisticated documentation, but little attempt was made to get the critical parameters under better control, maybe because this management originated from the office rather than from the laboratory. Commercial Interests

In the nineteen eighties, GC turned from being largely a sci­ ence into a commercial issue. The influence of the manufac­ turers grew rapidly, and the commercial catalogues be­ came the most important source of information in many laboratories. Several items (e.g. 0.53 mm i.d. col­ umns) became a commercial success which was not under­ standable scientifically. Nothing should prevent salesmen advertising their goods, but the voice of science should be able to correct developments. This voice, however, grew si­ lent. ManyOtelt that the takeover by commercial interests was the beginning of a decay of the culture of GC. Methods and instrumentation were bought in and no longer developed in-house. The average worker's knowledge of the techniques began to diminish; inexpensive people without sufficient education replaced the skilled workers who had introduced GC, and if they did not succeed, the tendency was to discard GC altogether. As this happened in most laboratories, the decline remained less obvious. Capillary GC did not lose in numbers, however. In the late nineties, between 15,000 and 20,000 instruments were sold annually. More than 200,000 instruments are now operating with capillary columns, and more than 100,000 analysts are regularly using the technique. For half of these, split injection might be the main method of sample introduction.

2. The Split Ratio The proportion of the sample entering the column is deter­ mined by the split ratio. This split ratio is deduced from (or set by adjustment of) the carrier gas flow rates at the column entrance (the "split point"). First we must define "split ra­ tio" and what we mean by "large" and "small" split ratios.

2.1. Definition

Most frequently the split ratio is defined as the ratio of the flow rates passing by the column and entering it (Fig­ ure Cl). If expressed this way, a large split flow rate results in a large split ratio, because a split ratio of, e.g., 100:1 corre­

156

C 2. The Split Ratio sponds to a large number. A "large" split ratio means that the amount of sample material entering the column is calcu­ lated from that injected by dividing by a large number; a "large" split ratio is usually manifested as small peaks. Sensitive analysis requires a "small" or "low" split ratio.

Deviating Definitions

There are, however, also other definitions. Quite frequently split ratios are given as, e.g., 1:100. In this instance, a "large" split ratio would mean introducing a large amount of mate­ rial into the column, because, e.q., 1:1 (=1) is a larger number than 1:100 (=0.01). Nevertheless many would call 1:100 larger than 1:1, adding to the confusion. Hinshaw [131 considers it normal to define the split ratio as the ratio of the gas flow rate entering the vaporizing cham­ ber (split flow rate plus column flow rate) to that entering the column. Unless the split ratio is small, the difference be­ tween this and the above definition is small. If, however, a split ratio is 10:1 by our definition, he would call it 11:1.

2.2. Acljustment/Determi­ nation of the Split Ratio

Instruments with electronic regulation of pressure and flow rates enable entry of the split ratio without further consid­ eration of this parameter. With older instruments, the col­ umn and the split flow rate must be adjusted/determined by use of a flow meter. The split ratio is rarely determined with great accuracy. Firstly, the split ratio seldom needs to be accurate since the peak areas of the components analyzed are nearly always compared with those of known amounts or concentrations obtained at the same split ratio. Secondly, the sample is of­ ten split by a ratio which differs considerably from that of the gas flow rates adjusted before injection (a subject dis­ cussed in Section C8). This means that painstaking determi­ nation of the split ratio is of limited value.

Which Flow Rates are Determined?

The flow rates in the injector are not the same as those measured at the column or split exit because the gas pres­ sure and temperature differ (Figure C4). The higher tem­ perature causes expansion of the gas, correspondingly in­ creasing the flow rates, whereas the inlet pressure com­ presses it and has the opposite effect. The deviations from the correct values is, however, the same for both flow rates. Hence, although we determine "wrong" flow rates, the resulting split ratio is accurate. The same applies to elec­ tronic regulation, because flow rates are calibrated for vol­ umes under ambient conditions. Problems occur only if the column temperature interferes (see below).

Influence of Column Tem­ perature

The split ratio depends on column temperature, because the predominantly used regulation with constant inlet pressure renders the column flow rate dependent on the oven temperature: when measured at the column exit under ambient conditions it decreases by a factor of two between 25 and 300°C. Thus, if the split flow rate is constant, the split

2.2. Adjustment/Determination of the Split Ratio

157

ratio increases with increasing column temperature. For cor­ rect determination of the split ratio, the column flow rate must be determined at the oven temperature used during injection. Carrier gas supply ~==il

Measured flow rates, 20 DC, 0 bar

Needle valve Flow rates of Interest, e.g. at 250°C,1 bar

11 ~

Split flow rate

c o,umn

Tflow rate

Figure C4 The flow rates of the hot and compressed carrier gas in the vaporizing chamber, which determine the split ratio, cannot be measured directly. The ratio of these flow rates is, how­ ever, identical with that of the flow rates measured at the split and column exit.

2.2.1. Determination ot the Column Flow Rate Calculation

Measurement at the Column Outlet

Today the column flow rate is most easily determined by

calculation. Most integration and data handling software

contains a program enabling calculation of the carrier gas

flow rate for hydrogen or helium on the basis of the column

dimensions, the column exit pressure (vacuum in GC-MS)

and the column temperature during injection.

Results are likely to be as accurate as those obtained by meas­

urement; the commonly observed variations in column di­

mensions (e.g. ± 0.003 mm for a 0.25 mm l.d, column) have

no significant effect on the determination of the split ratio

(although a deviation in the diameter enters by the fourth

power!).

If a precolumn of larger internal diameter is used, its contri­

bution to the pressure drop is usually negligible and calcula­

tion can disregard it.

If the column exit pressure is ambient, the column flow rate

can be measured at the outlet by use of a flow meter.

Some instruments (e.g. those from CE Instruments) enable

access to the detector base from outside and thus en­

able measurement at the detector base block without dis­

mounting the column outlet. The oven can be kept closed

158

C 2. The Split Ratio and the column temperature regulated during measurement.

The detector head is removed and an adapter attached, to

which the flow meter is attached, or the flow meter is con­

nected directly to the flame jet. The detector gases must be

switched off. When using an ECD, the flow rate is most eas­

ily determined atthe detector outlet after turning off the make

up gas.

For other instruments, measurement ofthe column flow rate

presupposes dismantling of the column outlet from the

detector. If the column temperature of interest is above

ambient, the capillary outlet must be made accessible out­

side the oven through a closed door. As it is difficult to find a

plastic tube small enough to fit the capillary, it is easier to

connect it via a press-fit connector. Alternatively, soft sili­

cone tubing is fitted to the flow meter, closed at its outlet,

and the column is pushed through its wall.

Soap Bubble Meters

The soap bubble meter is the oldest means of measuring the

column flow rate. It consists of a calibrated glass tube

(available in different sizes, e.g. 1 and 10 mL graduated vol­

ume) with a side-arm at the bottom enabling entry of the gas

stream to be measured. At the bottom is a rubber bulb filled

with a fairly concentrated soap solution. Compressing

the bulb moves the solution up to reach into the gas stream.

The gas forms a soap bubble that is driven upwards into the

graduated tube. Pressure on the bulb is then released. The

transit time between two points of the tube, representing a

known volume, is measured by means of a stop watch. The

transit volume divided by the time difference yields the flow

rate.

The common experience is that the soap bubble collapses

shortly before arriving at the intended second measurement

point. Corresponding annoyance can be reduced by rinsing

the wall of the tube with the soap solution shortly before

starting measurement. It might also be advantageous to form

several soap bubbles and to use the last one for measure­

ment, the others preparing the way.

Errors from Gas Type

The soap bubble meter is not only a rather inelegant tool, it

also easily yields incorrect results for rapidly diffusing

carrier gases, particularly hydrogen. The flow rate deter­

mined is too low if carrier gas is on one side of the soap

bubble and air on the other, because the carrier gas diffuses

more rapidly outwards than air inwards (Figure C5). This

deviation becomes particularly important if the soap bubble

moves slowly, i.e. if the gas flow rate is small and/or the cali­

brated tube of the meter is wide. When the flow is stopped,

the soap bubble might even recede (14).

Accurate measurement presupposes that the gases on the

two sides of the soap bubble are identical. If flow rates are

low, air should be introduced below the soap bubble by using a connecting tube with a rather large internal vol­ ume purged with air. This way, the carrier gas pushes air

2.2. Adjustment/Determination of the Split Ratio

159

into the flow meter (Figure C5, right). For higher flow rates, it is preferable to rinse the whole flow meter with carrier gas. Presence of air above the membrane is prevented by con­ necting the exit of the meter to a narrow tube that allows the gas to exit, but hinders access of air. Incorrect

Correct

Fromcolumn

I Air

Soap membrane Air

Figure C5 Soap bubble meters can yield flow rates which are too low if carrier gas is located below the soap bubble and air above it (left). Measurements are correct if the carrier gas pushes air ahead of itself (right).

Errors as a Result of Tem­ perature and Humidity

Results obtained by use of the soap bubble meter depend on temperature. The tube is usually calibrated for a tempera­ ture of 20°C. At higher temperatures, the gas will expand and the measured flow rate increase correspondingly. At 30 °C, for instance, the difference amounts to 3.4 %. The gas is, furthermore, saturated with humidity from the soap solution (carrier gases are dry). At 20°C, the vapor pres­ sure of water is 23 mbar, adding 2.3 % voluma and increas­ ing the flow rate by the same proportion (73 mbar at 40°C).

Acoustic Flow Meters

J & W introduced an alectronic flow meter for flow rates of 0.1-1000 mt/rnin with continuous readout [15). It meas­ ures the velocity of a movable diaphragm. The gas flows past an acoustic-type displacement transducer, resembling a loud­ speaker, and through a valve of low resistance that is nor­ mally open. Closure initiates a pressure wave that is meas­ ured: gas is accumulated and displaces the diaphragm. Although the meter does a perfect job with mechanically regulated gas flow rates, it can disturb electronically con­ trolled gas pressures or flow rates if there is insufficient damping between the regulator and the flow meter (no prob­ lem if a carrier gas flow rate is measured at the column exit). The acoustic flow meter measures flow rates irrespective of the type of gas. It is particularly useful for determina­

160

C 2. The Split Ratio tion of flow rates of gas mixtures or gases containing sol­ vent vapor. Flow rates are, however, pressure-dependent. The exit must, therefore, be at ambient pressure. The device does not resist increased pressure.

Measurement bi' Thermal Conductivity

There are, alternatively, electronic flow meters based on the determination of thermal conductivity, i.e. the cooling effect of the passing gas. Because the cooling depends on the type of gas, the gas must be identified; they are not suit­ able for gas mixtures. They measure mass flow rates and, thus, provide correct data also at pressures above or below ambient. They can, for instance, be inserted into the carrier gas line to the injector or into a line to a vacuum source. The flow rates measured will still be in units of gas volumes at ambient pressure.

Calculation from Measured Gas Velocity

The flow rate can be calculated from the average linear gas velocity through the column obtained from the gas hold-up time. This is possible at any column temperature, without dismounting the column outlet and disturbing the

detector. Some gas is injected which is unretained by the column and recorded by the detector in use. For FlO, methane from the fuel gas of a Bunsen burner or some gas from a cigarette lighter is used; for ECO, air is suitable. Even if the fuel gas is diluted with air, 1-31!L is sufficient to generate a large signal. The syringe should be clean and dry, which is most easily achieved by removing the plunger and inserting the needle into a hot injector - hot carrier gas will remove the volatile material. If a column with a thick (>0.5 urn) film of stationary phase is used, the column temperature should be somewhat above ambient to overcome possible retention in the col­ umn, especially jf gas (propane or butane) from a cigarette lighter is used. Conversion of Gas Velocities into Flow Rates

The column length divided by the dead time yields the gas velocity as the average of the relatively slow gas flow at the column inlet and the higher speed of the expanded gas at the column exit. This linear gas velocity is converted into a volumetric gas flow rate in terms of gas volume at ambient pressure and temperature. The following equation can be used [16]:

Fe

= 1zrj2298

tM Teoli

where Fe = corrected column flow rate (rnt/mln) L = column length (cm) r= column radius (not diameter!) tern) 1M = dead time of the column (min) Teal = absolute column temperature j =compressibility correction factor for the carrier gas

2.2. Adjustment/Determination of the Split Ratio The compressibility correction factor,

l. is calculated

161 as:

j =~.:J.

2p3-1 where p is the ratio of the inlet pressure to that at the outlet (absolute pressures). A table of pressure-drop correction fac­ tors is available [16]. 2.2.2. Adjustment of the Split Flow Rate

Approximate split flow rates can be directly adjusted if the carrier gas supply involves the flow/backpressure regulator system: it corresponds to the total flow rate minus the col­ umn and the septum purge flow rate. With classical pres­ sure regulation the split flow rate must be adjusted by meas­ uring the flow rate leaving the split exit.

Flow Meters with Floating Particles

Flow rates through the split outlet are usually high enough to be measured by use of flow meters with a floating particle (minimum flow rate 5-10 mLJmin). These are preferable to soap bubble meters, not only because of convenience, but also because they enable continuous reading of the flow rate and rapid adjustment of the needle valve. They consist of an accurately ground glass tube with an internal diameter which increases slightly from the bottom to the top (Figure C6). The gas flow drives a particle upwards to a position where the force driving upwards, the friction of the gas on the particle, corresponds to the weight of the particle. Fric­ tion depends on the velocity of the gas, i.e. on the distance between the wall of the tube and the particle. The higher the gas flow rate, the larger must be this distance for the equilib­ rium position and the higher is the particle. gj Ol

Conical glass tube

Qi

.~

c:

o

g> •

Floating particle

I

Gas stream pushing particle upwards

'5

T

c:

o

~

,g "iii

o

Figure C6

Flow meter with floating particle.

Calibration

Because floating particle meters essentially determine fric­ tion, measurements depend on the viscosity of the gas.

762

C 2. The Split Ratio Nearly three times as much hydrogen as air passes the par­ ticle at a given position. This has two consequences: firstly, the calibration must be performed with the carrier gas used or, because these flow meters are often calibrated for air, the values read must be corrected. Secondly, it must be ensured that the gas pass­ ing the floating particle really consists of carrier gas. When the flow meter is connected to the split exit, initial values read on the meter must be disregarded until the tube has been well purged with carrier gas. Some flow meters are available with floating particles manufactured from different materials: the differences between their weights enable the same tube to be used for different ranges of gas flow rate.

Maintenance

Flow meters with floating particles are sensitive to con­ tamination. If dust or sample material from the split outlet adheres to the surface of the glass tube or the floating parti­ cle, the particle tends to stick to the wall and to jump rather than to float smoothly. Rinsing with a solvent, e.g. metha­ nol and dichloromethane, solves this problem. A relatively long silicone rubber tube connecting the flow meter to the split exit retains sample material, because it exerts such strong retentive power that all except highly vola­ tile sample components dissolve in the wall of the tube (the material is essentially the same as the stationary phase in the column, but of extremely high "film thickness"). This not only helps to keep the meter clean, it also prevents contami­ nation of the laboratory atmosphere.

Permanent Installation

Floating particle flow meters should be kept permanently in­ stalled at the split outlet. Most of the carrier gas consumed is vented through this exit. Sometimes the flow rate becomes excessive, e.g. if the carrier gas inlet pressure is increased for a rapid purging of the column. Under other conditions, carrier gas can be saved by reducing a high split flow rate ca. 20 s after injection. This presupposes convenient adjustment of the flow rate.

Slight Overpressure

Flow meters with floating particles generate slight back­ pressure at the split exit and correspondingly reduce the split flow rate. This can be calculated from the weight of the particle and its cross section, and is significant at low inlet pressures only. This is another reason for leaving the flow meter permanently installed: when removed, the split flow rate becomes slightly higher than was measured.

3.1. Split Ratios Commonly Applied

163

3. Sample Concentrations Suitable for Split Injection The possibility of varying the split ratio provides unique flex­ ibility with regard to adjustment of the sampling tech­ nique to the solute concentration in the sample: the split ratio can be adjusted instead of the dilution of the sam­ ple. This is of particular importance in capillary GC because the range of solute concentrations providing a signal of suf­ ficient size while avoiding overloading ofthe column is rather narrow. 3.1. Split Ratios Com­ monly Applied

There is no straightforward answerto how much the sample should be diluted. Suitable concentrations depend on the split ratio and on the detector. It is, furthermore, more appropri­ ate to ask about the range of applicable split flow rates ­ split ratios will then result from the split and column flow rates used.

Lower Limit for Split Flow Rates

At low split flow rates, the residence time of the sample material in the injector is long, i.e. sample enters the col­ umn during an extended period of time. This causes more material to reach the column, and increases sensitivity, but also broadens the initial bands. Beyond a certain limit, peaks will be broadened. If high sensitivity is required, try split flow rates as low as 10 mllmin, but be aware of the possibil­ ity of peak broadening.

Upper Limit

Split flow rates should not be increased beyond some 300-500 mlJmin. Reasons for this will be discussed in more detail in Section C7: gas consumption becomes excessive, the pressure drop within the gas lines is a problem, and the carrier gas becomes an efficient coolant of the vaporizing chamber. Flow-regulated systems usually limit the total flow rate to 500 mllmin.

3.2. Range of Suitable

An estimate of the range of concentrations appropriate for split injection can be based on the rule that FIDs require a minimum of ca. 1 ng to produce a clear signal and columns of standard film thickness are overloaded when more than ca. 50 ng of a compound are chromatographed. If we as­ sume 20 ng reaches the column, a convenient split flow rate of 50 mllmin, and a column flow rate of 2 mllmin, 25 times more solute material passes by the column than enters it, i.e. 500 ng. If a total of 520 ng is injected in 1 III solvent, its concentration corresponds to 520 Ilg/mL.

Concentrations An Example

164

C 3. Sample Concentrations Suitable for Split Injection

Recommendations

Table C1 suggests some ranges of concentration per com­ ponent suitable for a typical thin-film column and FlO or ECO (linearity determining the upper limit for the latter). Concen­ trations of 10.000 ~g/mL (1 Ufo) are high and require a fairly high split ratio. For FlO, 10 Ilg/mL (10 ppm) is at the other limit where a non-splitting method (splitless or on-col­ umn injection) becomes preferable. Table C1 Recommended concentrations per component (llg/mL or ppm).

Split ratio 10:1 100:1

FlO

ECO

10-1000 100-10,000

0.1-5 1-50

4. Initial Band Widths The purpose of sample splitting is not only to reduce the amount of sample material entering the column. This might even be an undesired side effect of the other objective, viz. formation of a narrow initial band.

4.1. Band Widths in Space and lime

In conventional liquid chromatography with glass columns, the length of the band of, e.q., chlorophyll can be observed directly. On moving through the column, this green band gradually becomes broader. In this case, band width is ex­ pressed in terms of length or, considering also the column diameter, of volume or space. In GC, direct observation of the band in the column is impos­ sible. We detect the band when it leaves the column - as a peak on the chart paper. The width of the peak repre­ sents time - the time required by the band to leave the col­ umn. The width in terms of length ofthe capillary over which the solute is spread can be determined only by calculation. In capillary GC, we deal with band widths in terms of time and space, depending on the problem considered. For the initial bands resulting from split injection. band widths in time clearly better suit the purpose. In splitless injection, both types of band width are important, as will be discussed in Section 07.1.

4.2. Factors Determining

Unless the solute band is reconcentrated before the start of chromatography, the initial band width (in time) is equal to the time during which sample material enters the

Initial Band Widths

4.2. Factors Determining Initial Band Widths

165

column. The faster the split flow rate purges the sample vapor through the split exit, the sharper is the initial band (see Figure Initial band widths in time are equal for all solutes. Band lengths, however, vary. The first molecules of a rapidly mi­ grating component travel relatively far into the column be­ fore the last follow from the injector, whereas the most ad­ vanced material of slower components remains further back.

en

Syringe needle

,~~~

. ~~::

;11!111111111 Cloud of sample vapor

; imW' :

11 1II !!l

driven by the carrier gas

j

... .

Slit exit line

~ Initial band in

column inlet

Figure C7 The column inlet picks some sample from the vapor cloud while the latter is driven into the split outlet. The initial band width (in time) corresponds to the time required by the sam­ ple vapor to pass the column entrance.

Transfer Time

The transfer time, i.e. the initial band width, is influenced by three factors: 1 the split flow rate, which determines the residence time of the sample vapor in the injector; 2 the sample volume and the nature of the solvent, which determine the volume of the vapor cloud; and 3 the size and geometry of the vaporizing chamber and other factors influencing the dilution of the sample vapor with carrier gas.

4.3. Experimental Obser­

Because initial bands cannot be observed directly, an experi­ ment was designed such that the initial band essentially de­ termines the shape of the peak. It can be used to study proc­ esses in the injector or to optimize conditions for critical ap­ plications. Results are valid for isothermal analysis only, because a temperature increase causes the bands to be recon­ centrated by cold trapping (see below). Several options are described, depending on the material readily available and the direction of the investigations.

vation of Initial Band Shapes

166

C 4. Initial Band Widths

4.3. 1. Description of the

Experiment

Several hundred microliters of a gas. diluted in air. are injected to simulate the behavior of a regular gaseous sample or of the vapor of a liquid sample (2 ~L of liquid pro­ duce some 400-800 ~L of vapor). As the test component is not retained by the column, its peak is minimally broadened during passage through the column and gives a clear pic­ ture of the initial band.

Sample

About 200 ~L methane, e.g. fuel gas from the tap, propane from a cigarette lighter, or butane from a torch are diluted in ca. 100 mL of air. A volatile solvent (1 ~L of liquid) can be used provided the column temperature is increased to at least 50-100°C (depending on the retentive power/film thick­ ness of the stationary phase). Dilution in air can be a prob­ lem if a Carbowax-type column is used at temperatures above 40°C - for such columns an inert gas (e.g. argon) is more appropriate.

Sample Container

A flask fitted with a septum is well suited. Alternatively, a !ilass bottle with a narrow neck (e.g. a measuring flask) will do if closed by a plastic stopper through which the sy­ ringe pierces a small hole (loss by diffusion through this hole is slow).

Column

A short column contributes little to the final peak width. It must create sufficient pressure drop to enable easy regula­ tion of the carrier gas, which is why a 2 m x 0.15 mm i.d. capillary is well suited. Older devices may, however, be too slow for proper recording of the resulting sharp signals. For this reason, but also to approach more realistic analytical conditions, a standard 15 m column of 0.2-0.32 mm i.d. might be preferable or, to obtain specific information for a given application, even the column in use. The stationary phase is not of importance as long as it does not have sig­ nificant retentive power for the test solute; even an uncoated capillary, e.g. raw fused silica tubing, serves the purpose.

Injection Speed

Gas volumes of 100-1000 ~L are injected as rapidly as pos­ sible. As the release of the sample from the syringe needle into the injector easily contributes to the initial band width, the injection time should be estimated and taken into con­ sideration when looking at very sharp peaks.

4.3.2. Subjects to Study

The split flow rate is varied to study its effect on peak width. At low split flow rates, broad and tailing peaks are ob­ served, providing a picture of the transfer of the vapor from the vaporizing chamber into the column. When the split flow rate is increased, a point is reached beyond which the peak no longer gets sharper, i.e. the initial band width becomes negligible compared with band broadening during passage through the column. The corresponding split flow rate is the minimum acceptable under the given conditions.

Dependence of Peak Width on the Split Flow Rate

4.3. Experimental Observation of Initial Band Shapes

167

Maximum Peak Height

The split flow rate is determined at which a maximum peak height is obtained. It provides maximum sensitivity for a real analysis. As will be shown below, maximum sensitivity is not achieved at the lowest split flow rates.

Sample Volume

A split flow rate is selected that creates clear peak broaden­ ing for an injection volume of, e.g., 100 JlL. The sample vol­ ume is then increased and the additional band broadening observed. If this additional broadening is not proportional to the injection volume, it indicates reduced mixing with car­ rier gas. By varying the sample volume and the split flow rate, the combination producing maximum peak height at mini­ mum peak width can be determined. This optimizes the con­ ditions such that the maximum concentration of sample vapor enters the column. This is of importance in split injec­ tion requiring high sensitivity, the subject of Section CG.

Length of Syringe Needle

The length of the syringe needle is varied, by using either different syringes or one syringe with a long needle inserted to different depths. The different distances between the nee­ dle exit and the column entrance influence the dilution of the sample with carrier gas and affect peak heights and peak widths, particularly at relatively low split flow rates.

Position of Column Entrance

The position of the column entrance in the vaporizing cham­ ber is varied by pushing the column further up into the injec­ tor, nearer to the exit of the syringe needle. This has an ef­ fect similar to varying the length of the syringe nee­ dle. In addition, less injector volume can be used (material driven below the column entrance cannot reach the column).

Type of Liner

The size, shape, or packing of the liner is varied to alter the extent of mixing of the sample with carrier gas. Mixing is considered important for the homogeneous distribution of the sample material across the liner and for reproducible split ratios. Claims can be checked that certain liners, e.g. with built-in obstacles, provide improved mixing. Mixing causes the vapor cloud to be enlarged and the initial band to be broadened - a 1:1 mixing would double the peak width.

Influence of Diffusion Speed

Different test components are compared to investigate the influence of diffusion speed. For instance, methane is compared with heptane (which presupposes a column tem­ perature of at least 100°C, unless an uncoated capillary is used). Faster diffusion produces a more rapidly growing vapor cloud and a broader peak. Effects related to diffusion should also be visible upon changing the carrier gas: diffu­ sion speeds are ca. three times lower in nitrogen than in hy­ drogen or helium. The injector temperature is another factor affecting diffusion (low temperatures should produce sharper signals)..

168

C 4. Initial Band Widths

4.3.3. Some Results

Figure C8 shows four methane peaks obtained at different split flow rates. 400 III volumes were injected in about 0.5 s by means of a syringe equipped with a 5 cm needle. The injector at 200°C contained an 80 x 4 mm i.d. liner, with the column entrance 5 mm above the base. The distance between the needle exit and the column entrance was 45 mm. A 15 m x 0.31 mm i.d. glass capillary column was used with a col­ umn flow rate of 1.8 mllmin. 20 .1/.1n

2.4

o .1I.1n S

(spIttlessl 5.05

0.75 s

inj.

inj.

inj.

inj.

LJ

j

j

j

Figure C8 Methane peaks obtained by injection at the split flow rates indicated; constant attenuation.

Splitless Injection

The broad peak at the right resulted from an injection with­ out splitflow (splitless injection with permanently closed split exit). As the vaporizing chamber was emptied solely through the small flow rate into the column, transfer took a long time: after a minute it was still incomplete. This indicates that the 400 III sample was diluted to a cloud of more than 2 mL. The relatively slow upslope of the peak is the result of using too short a syringe needle. The vapor cloud formed upon injection was positioned considerably above the column entrance. Carrier gas, then diluted sample at the front edge of the vapor plug entered the column before the more con­ centrated vapor arrived.

Low Split Flow Rate

At a split flow rate of 8 rnt/mln (i.e. with a flow rate of ca. 10 rnt/rnln through the injector), the peak width at half height was 5 s, that at the bottom ca. 15 s. At this flow rate transfer of 400 III is calculated to take 4 s, which suggests that most of the material entered the column without extensive mix­ ing, because time for diffusion was short. The peak has a sharp top and does not resemble the rectan­ gular shape of a sample plug with sharp edges passing by

4.3. Experimental Observation of Initial Band Shapes

169

the column entrance (as shown in Figure C7). This shows that diffusion or convective mixing "erodes" the edges of the plug. Diffusion of methane at the injector tempera­ ture (200°C) is, in fact, rapid (faster than that of typical sam­ ple components). Sharp Peak

At a split flow rate of 70 mLJmin, the peak width at half height was 0.75 s. This is probably near the width arising as a result of longitudinal diffusion during passage through the column (chromatographic peak broadening), but injection also took 0.5 s, and it cannot be ruled out that the recorder further contributed to the peak width. The width at a split flow rate of 150 ml/min hardly differed and its rather poor repeatability was indicative of the large effect of injection speed. Thus the split flow rate was no longer the determin­ ing factor.

4.4. Effect on the Final Peak Width

The width of the final peak is determined by the sum of chro­ matographic band broadening and extra-column contribu­ tions, of which injection is the most important. The subject of extra-column contributions has been discussed, mostly in theoretical terms, by numerous authors, including Guiochon [17], Sternberg [18], Gaspar et al. [19], Bartle [20], Hinshaw [21], and Bemgard and Colmsjo [22].

4.4. 1. Isothermal Runs

Almost all theory is based on isothermal runs at the injec­ tion temperature. This was the most common mode of packed column GC, but is rather rarely used in capillary GC. The ef­ fect of temperature increase (cold trapping) will be consid­ ered later.

Square Relationship

The width of the band arising from extra-column contribu­ tions, such as the initial band width, and the amount of band broadening during chromatography, are not directly addi­ tive: contributions are added as squares (Sternberg [18]). If they are assumed to be Gaussian in shape, with a width expressed as the standard deviation of this curve, s, the final peak width, Sp, is calculated from individual contributions, s; as:

This relationship causes the effect of the initial band on the peak width to be weaker than might at first be ex­ pected. It contributes significantly only if it approaches the magnitude of band broadening during chromatography, i.e. if it becomes comparable with the peak width under ideal conditions. Calculated Examples

A more specific impression of maximum tolerable initial band widths can be obtained by calculation. First we assume a rather rapidly eluted peak with a width resulting from chro­

170

C 4. Initial Band Widths

Table C2

Influence of initial band width on final peak width.

Initial band width, ss, [sl Chromatogr. peak broadening, Sch' lsl Calculated final peak width, lsl Peak broadening by initial band [%1

0.2 1 1.02 2

1 1 1.4 40

1

2 2.2 10

1 10 10.05 0.5

3

10 10.4 4

matography of 2 s. If the initial band has a width of 1 s, this peak is broadened by 0.23 s, i.e. by some 11 %. Separation efficiency in terms of resolution (separation numbers, TZ) is reduced accordingly, and the number of theoretical plates calculated for this peak is reduced by 17 %. A more strongly retained peak of 10 s width is broadened by a negligible 0.5 %. Some further examples are given in Table C2, indicating that initial bands 1-3 s wide can usually be tolerated.

More Realistic Band Shapes

The initial band is, of course, not really Gaussian in shape. If a plug of vapor with somewhat diluted zones at its edges passes by the column entrance, it is expected to have the shape shown at the left in Figure C9. Added to a band of Gaussian shape resulting from chromatography, this pro­ duces broadening of the final peak which is greater at the top than at the bottom. The effect on a critical sepa­ ration depends on the situation. For nearly completely sepa­ rated peaks, peak widths at the bottom are decisive whereas broadening at the top is important if peaks are poorly sepa­ rated.

Initial band

Chromatographic band broadening

Final peak

Figure C9 Because the initial band is unlikely to be Gaussian in shape. broadening is usually different at the bottom and top of a peak.

Minimum Split Flow Rate Required

From the maximum tolerable initial band length we can esti­ mated the minimum split flow rate required. To achieve an initial band width of. e.g.• 1 s, the sample vapor must pass the column entrance within 1 s. 500 IJ.L of a gaseous sample (e.g. heads pace) or 2 IJ.L of liquid form a vapor cloud of ca. 1 mL volume (vapor mixed with gas). Passage by the column entrance in 1 s means a flow rate through the injec­ tor of 60 mUmin. At a column flow rate of, e.g., 2 rnt/min, this results in a split ratio of 30:1. In other words, we must accept a split ratio of 30:1 and loss of 97 % of the sample material injected to obtain an initial band as sharp as 1 s.

4.4. Effect on the Final Peak Width

171

4.4.2. Chromatography Involving Temperature Increase

Assumption of isothermal chromatography at the injection temperature paints a pessimistic picture. Most capillary GC involves temperature programming, for which initial band widths are less critical.

Reconcentration by Cold Trapping

When a component is eluted at a column temperature above that set during injection, cold trapping reduces its initial band width. During injection, the component moves slowly or hardly at all. Cold trapping, described in more detail in Sec­ tion 07.3, reduces the initial band width by a factor of ap­ proximately two for each 15° temperature increase. As a re­ sult of this, initial band widths become negligible for compounds eluted at temperatures 40-50° or more above the column temperature during injection.

Reconcentration by Solvent Effects

Solvent effects, actively used for non-splitting injection tech­ niques (Section 07.4), also reconcentrate bands in split in­ jection. If the column temperature during injection is at least ca. 20° below the boiling point of the solvent (or of the sam­ ple matrix, if there is no solvent), recondensation in the oven-thermostatted column inlet produces a short, thick layer of liquid which strongly retains solute material until the liq­ uid evaporates again (solvent trapping). Retention of solvent in the stationary phase (phase soaking) tem­ porarily builds up increased retentive power even at column temperatures slightly exceeding the solvent boiling point. Both solvent effects hinder the most advanced solute mate­ rial from moving further into the column. Solvent effects are particularly effective when the split ra­ tio is low, i.e. exactly when the otherwise broad initial bands need reconcentration. They explain to a significant extent why experiments dealing with broad initial bands frequently result in peaks which are sharper than expected.

5. Split Injection for Fast Analysis GC analyses are often more time-consuming than necessary. Many analyses could be performed in a few minutes by use of columns a few meters long without buying any special equipment. Loss of separation efficiency is smaller than usu­ ally expected, as shown, e.g., by van Lieshout et al. [23]. It has been the dream of many to perform GC runs in less than a minute, such that the analyst injects and waits for the result at the instrument. This changes work fundamen­ tally, because an analysis can be completed in one stroke; an autosampler is no longer needed. This, however, mostly pre­ supposes dedicated equipment.

172

C 5. Split Injection for Fast Analysis

5.1. Prerequisites for Fast Analysis

Isothermal Analysis

When using conventional GC ovens, the speed of analy­ sis is often limited by the achievable rate oftemperature pro­ gramming and cooling. High programming rates necessitate powerful heating or more direct heating of the column. The very hot air produced at the heating filaments of conven­ tional ovens may no longer be mixed well enough with the cooler air in the oven, particularly along the oven walls where most of the heat is consumed. As a consequence, the col­ umn is hit alternately by cooler then hot air. The fluctuating temperature in the column leads to expansion/contraction of the carrier gas and rapid changes in retentive power, caus­ ing "jumping" chromatography of the solutes and distorted peaks ("Christmas tree effect"). Hence fast GC with con­ ventional instruments is restricted to isothermal runs or short temperature programs at rates not exceeding ca. 50 o/min.

Short, Narrow Bore Column

Fast analysis employs short columns. The separation effi­ ciency (and inertness) of a 2-4 m section of a normal capil­ lary column is always a surprise. Narrow-bore columns (0.1­ 0.2 mm i.d.) produce maximum separation efficiency per unit retention time because they enable the use of very high carrier gas velocities. Their length is primarily limited by the high inlet pressure.

Only Split Injection

Split injection is the technique of choice for fast analysis: 1 It is the only technique enabling injection at any col­ umn temperature (i.e. at the temperature of analysis). For on-column injection, the column must be below the solvent boiling point and splitless injection usually re­ quires cooling to achieve cold trapping or solvent ef­ fects. 2 It is a rapid technique, not requiring, e.q., a time-con­ suming transfer period. 3 Provided the split flow rate is sufficiently high, it creates sharp initial bands.

No Trace Analysis

Sharp initial bands for isothermal runs require high split flow rates. At the same time, the flow rates through narrow bore columns are low (even at high gas velocities). As a result, split ratios are high. If, for instance, a split flow rate of 200 mLJmin is combined with a column flow rate of 1 mLJmin (about 2 m/s in a 0.1 mm i.d. column), the split ratio is 200:1 and only 0.5 % of the material injected reaches the column. The sharp and high signals obtained in fast GC only partly compensate of this loss of sensitivity. The opinion that narrow bore columns can only be used for concentrated samples was questioned by van Ysacker et al. [24], who found that splitless injection is possible even for 0.1 mm i.d, columns if a liner of merely 1 mm l.d, is used. The high inlet pressure compresses the sample vapor such that it is still possible to inject 1 Ill. Such injection usually prolongs analysis times, however.

5.2. Maximum Tolerable Initial Band Widths 5.2. Maximum Tolerable Initial Band Widths

5.3. Limits to the Sharp­ ness of Initial Bands

173

Maximum tolerable initial band widths can be estimated from expected peak widths, assuming that no reconcentration ef­ fects interfere. They help us to define the limits of fast GC. Table C3 shows peak widths calculated for a short column with a separation efficiency corresponding to 20,000 theo­ retical plates. A peak with a retention time of 10 s has a width (chromatographic band broadening) of 0.16 s. If an initial band width corresponding to half the final peak width is con­ sidered acceptable (reducing separation efficiency by some 10 %), the initial band must be as narrow as 0.08 s. For a peak with a 1 min retention time, an initial band of 0.5 s width can be tolerated. Table C3

Calculated widths [s] of peaks with short retention times [s];

maximum tolerable initial band widths [s].

Retention time

Peak width

Max. init. band width

10 30 60

0.16 0.5 1

0.08 0.25 0.5

It is not possible to generally define the smallest initial band which can be achieved with a conventional split injector. Several factors are involved.

Split Flow Rate

If we assume a cloud of (diluted) sample vapor of 1 ml vol­ ume (from, e.g., a 2 III injection into a 4 mm i.d. liner), a split flow rate of 300 mUmin is required to carry this cloud past the column entrance during a period of 200 ms. Under strictly optimized conditions. somewhat better results are obtained: only 1 III is injected (hardly more than the volume of the needle if the latter is emptied); a solvent forming a relatively small volume of vapor is chosen (e.g. hexane - about 150 III for 1 III of liquid); use of a 2 mm i.d. liner restricts mixing such that the vapor cloud has a volume of around 250 III only. Then the same split flow rate results in an initial band of 50 ms, or a split flow rate of 75 mL./min is sufficient for the 200 ms band.

Injection Speed

Manually inserting the needle through the septum easily takes more than 500 ms. Even if the bulk of the sample liquid is withdrawn into the barrel of the syringe, some liquid re­ mains as a film on the needle wall and is eluted into the in­ jector during introduction of the needle. This results in a small pre-peak (Section A5.3.4). It can be avoided by rinsing the needle backwards into the barrel of the syringe with a small amount of pure solvent. Autosamplers insert the needle some ten times faster than humans.

174

C 5. Split Injection for Fast Analysis Depression of the plunger takes 30-80 ms. For this opera­ tion manual injection tends to be faster than that of auto­ samplers.

Elution from the Needle

Sample introduction into a hot vaporizing injector usually involves two steps. First, some liquid is pushed through the syringe needle into the injector, which is a fast process (the 30-80 ms to depress the plunger). This may be followed by evaporation or ejection of sample material from the nee­ dle, which easily occurs with some delay when the internal needle surface is cooled by the liquid passing in the first step. It is unknown how much time is required for this step. Fast autosamplers avoid sample evaporation inside the syringe needle.

Sample Evaporation

Sample evaporation in the injector often turns out to be the most time-consuming step. Visual experiments suggest the time taken is between 0.2 and 4 s (1 Ill, see Section 83), with hot needle injection (thermospray) being faster than deposition of the liquid on to packing or into a trap.

Intermediate Cold Trapping

If injection cannot produce the extremely sharp initial bands required by high speed GC, bands must be focused after­ wards. Ewels and Sacks [251 described a system based on cold trapping and subsequent high-speed thermal desorption of trapped components. The trap was situated below a stand­ ard vaporizing injector and consisted of an 0.5 mm i.d, steel tubing, cooled by nitrogen and heated resistively by pulses of current. An improved version of the trap was described shortly afterwards by Lanning et al. [261 who showed a chro­ matogram with nine peaks eluted in less than 2.5 s. Rijks et al. [271 built a similar system and reported a number of chromatograms of 20-50 s duration, some of which involved temperature programming atthe fastest possible rate. A sys­ tem temporarily trapping in a metal tube or in a coated nar­ row bore fused silica capillary was recently described by Bogerding and Wilkerson (281.

5.4. Examples of Fast Analyses

Husek et al. [291 described a method for analyzing protein amino acids in four minutes (not .including cooling of the oven). 3-5 m x 0.1 mm l.d. columns were used with the fastest temperature program possible with the instrument (40 a/min). Samples were derivatized with ethyl chloroformate and injected by the split method.

Protein Amino Acids in Four Minutes

Determination of Enzyme Activity

In 1978, the activity of the enzyme myrosinase from horse radish roots was routinely determined by measuring the rate of liberation of allylisothiocyanate from the corresponding glucosinolate sinigrin (301. The substrate solution, including the internal standard (butylisothiocyanate) and hexane, were added to cell preparations containing the enzyme. The sam­ ple was shaken every 60 sand 1 III of the hexane phase was

5.4. Examples of Fast Analyses

175

injected by the split method. Isothermal chromatography lasted 2 min. Between the solvent peak and the two peaks of interest another sample could be injected, resulting in over­ lepping chrometogrems. An 8 m x 0.29 mm i.d. glass cap­ illary was used, and the split flow rate was 150 ml/min.

GC Analysis in 18 s

Figure C10 shows the chromatogram obtained from an ex­ perimental analysis lasting 18 s. It was performed in 1982 with conventionel equipment (Carlo Erba/CE Instruments, Mod. 2900 GC)with a carrier gas pressure regulator for higher pressure. A 3 m x 0.03 mm i.d. capillary column coated with a 0.05 urn film of SE-30 was used at 14 bar inlet pressure. Neither the sample volume injected nor the split ratio were given. Since the column flow rate was about 0.03 ml/min, the split ratio must have been more than 3000: 1 (the sample was undiluted). Peak widths varied between 0.5 and 1 s. From the shape of the peaks it can be estimated that the widths of the initial bands were between 0.3 and 0.6 s. This might be the best achievable with conventional split injectors. A similar result was published by Purcell (32). A mixture of hydrocarbons including benzene and heptane was analyzed in 30 s by use of a 15 m x 0.25 mm i.d. column with a gas velocity of 185 cm/s (utilizing 22 % of the theoretical column efficiency). 7

8 C)

10

5 (, 1

3

.­

0\l I

0

S

\

'­

10

L....

15

20

-

SEC

Figure C10 GC analysis taking 18 s. 1 = ...pentane; 2 = 4-methyl-1­ pentene; 3 = 2,3-dimethylbutane; 4 .. 2-methyl-1-pentene; 5 .. ...hexane; 6 .. methylcyclopentane; 7 .. 2,4-dimethyl­ pentane; 8 .. benzene; 9 .. cyclohexane; 10 .. ..-heptane. (From Schutjes fit .1. [31]).

176

C 6. Analysis Requiring Maximum Sensitivity

6. Analysis Requiring Maximum Sensitivity

It is impossible to give specific advice appropriate to all the

many widely differing types of application involving split injection. As is typical of practically oriented work, each application is its own special case with particular technical requirements. Instead, two extreme applications will be described, one requiring the introduction of as much of the sample as possible into the column under conditions still providing acceptably narrow initial bands (minimized split­ ting), the other aiming for a maximum split ratio for analysis of undiluted samples (Section C7). The different philosophies involved illustrate the span of possible requirements.

6.1. Sharp Bands at Low Split Ratios

For some analyses, split injection is used exclusively to achieve narrow initial bands. Venting of sample material

through the split outlet is an unwanted side effect and should, therefore, be minimized. The maximum amount of sample possible should be introduced into the column while maintaining a tolerable initial band width. The toler­ able initial band width dictates the time during which sam­ ple material may enter the column. Two types of sample re­ quiring such conditions are described briefly.

6. 1. 1. Headspace Analy­ sis

Headspace analysis is used to determine volatile components from higher-boiling or involatile matrices, such as tetrachloro­ ethylene in egg yolk. It is an elegant technique to avoid the injection of high-boiling or involatile sample by-products. Two versions are distinguished, the static headspace tech­ nique, involving analysis of the gas phase above a sample, and dynamic headspace analysis (also called purge and trap), in which the components are purged from the sample by means of a gas stream and the components of interest are enriched on a trap. Both are well described in literature, e.g., by Kolb [33-35].

Simple Version of Static Headspace Analysis

Headspace analysis can be extremely simple. A small amount of sample material (e.g. egg yolk) is enclosed in a vial with a septum cap. Usually the amount does not even need to be weighed, because equilibrium concentrations, not absolute amounts, are measured (VOCs in water are the ex­ ception because the proportion of evaporated material may be high); 0.5 g sample material in a 5-10 mL vial is suitable. Often the septum should be protected by aluminum foil to avoid loss of solute material. The vial is left to stand until the gas phase is equilibrated with the sample. This takes less

6.1. Sharp Bands at Low Split Ratios

177

than 30 min if the sample is a liquid, but might take 24 h if

diffusion in a solid sample is hindered.

Equilibration Temperature

Warming of the sample vial increases sensitivity (by ap­

proximately a factor of 2 for each 15-20° temperature in­

crease), but is also recommended if the sample can be melted

(such as chocolate), because this accelerates equilibration.

Care is required to prevent cooling of the vial during sam­

pling, since this would move the equilibrium back to lower

concentrations in the gas phase. Furthermore, the higher the

temperature of the sample, the more critical becomes recon­

densation of vapor on surfaces inside the syringe. This

primarily calls for a clean syringe, i.e. surfaces free from thin

layers of a liquid behaving as stationary phase picking up

solute material. If partitioning with such retaining material is

ruled out, the syringe may be at a temperature several tens

of degrees below that of the sample without risking loss of

solute material.

Some 500 ul, of gas is taken from the vial and injected by the

split method. Quantitation is based on the method of stand­

ard addition, i.e. a sample is analyzed with and without

addition of a known amount of the component(s) of interest.

Reconcentration usually Impossible

Static headspace analysis is not very sensitive unless the

components of interest are highly volatile. Sensitivity is,

therefore, often a problem and split injection is chosen merely

because of the need for sharp initial bands.

Non-splitting injection techniques cannot usually be used,

because no simple GC-internal reconcentration technique can

be applied. Because heads pace analysis is usually performed

with a column at or near ambient temperature, there is no

scope for reducing the column temperature during the injec­

tion to obtain a cold trapping effect (unless special traps,

e.g. cooled with liquid nitrogen, are utilized). If additional solvent is introduced to obtain solvent effects, important solute peaks are often obscured.

6.1.2. Rapid Isothermal Runs at Elevated Column

Because split injection of liquid samples is compatible with fully isothermal runs, its use often enables more rapid analy­ sis than splitless or on-column injection. This is of particular advantage if the sample components of interest are eluted within a narrow range of retention times. Bearing this ad­ vantage in mind, split injection might be preferable even if the sensitivity achieved is at the limit.

Temperature

6.2. Optimized Split Flow Rate

If sensitivity is a problem in split injection, ones first incli­ nation is to reduce the split flow rate, assuming that an increase in the proportion of the sample material entering the column would help. Down to a certain split flow rate this is certainly true. Below this limit, however, the analyst might be disappointed to see that the peaks grow "fat" rather than "tall".

178

C 6. Analysis Requiring Maximum Sensitivity

6.2.1. Peaks Growing Broad instead of High

The effects of the split ratio on peak height and peak width are shown in Figure CB for 400 III injections of methane di­ luted with air, i.e. injections analogous to headspace analy­ sis. When injected without a split flow, i.e. splitless, the meth­ ane peak was broad and relatively low. At a split flow rate of B mLJmin, the peak was narrower and higher. Further in­ creasing the split flow rate caused these changes to continue: at a split flow rate of 20 mt/min. the peak was again sharper and higher. Material vented through the split exit was, there­ fore, being taken away from the sides and not from the top of the peak. Beyond a certain split flow rate, the peak again started losing height, of course.

Sensitivity: Peak Height, not Area

No doubt, splitting at a higher ratio reduces peak areas since it reduces the amount of sample material entering the col­ umn. This does not, however, necessarily mean that it also reduces sensitivity. A sharper signal may be higher even though its area is smaller. Sensitivity is determined by peak height, not by peak area. The optimum split flow rate is that producing maximum peak height.

6.2.2. Dilution in the Injector

The increase in peak height observed on increasing the proportion of sample material vented through the split exit resulted from the increased carrier gas velocity in the' injector. The needles of standard gas syringes are 5 cm long. In our experiment, this left a distance of 4.5 ern between the needle exit and the column entrance. During the time required to cover this distance, the vapor was diluted with carrier gas. The higher the split flow rate, i.e. the shorter this transfer time, the more concentrated they remained. Use of a longer needle would have reduced this effect.

Peak Height = Vapor Concen­ tration

As long as peak broadening during passage through the col­ umn is negligible (as for the two broadest methane peaks in Figure CB), peak height represents the amount of sample material entering the column per unit time. Because the flow rate into the column was the same for all four injections, this amount was determined by the concentration of the sam­ ple vapor in the carrier gas. Increasing the split flow rate meant feeding the column with sample vapor for a shorter time (shorter initial band), but also with more concentrated vapor.

6.2.3. Dilution in the Column

For the methane injections discussed above, the split flow rate producing a maximum peak height was 40 mU min. It is higher than for real analyses because of the sharp­ ness of the non-retained peak. The more a component is re­ tained, the more its band is broadened; chromatography di­ lutes the solute material with carrier gas (as did passage through the injector liner). A broader peak means that a broader initial band can be tolerated.

6.2. Optimized Split Flow Rate

179

Maximum Peak Height = "Height" of Initial Band

The maximum possible peak height corresponds to the "height" of the initial band, as would be determined by a detector positioned in the column inlet. It is reduced by band broadening (dilution) during chromatography. Since the height of an already broad initial band suffers less than that of a sharp one, the split flow rate providing maximum sensitivity is lower the more strongly a solute peak is retained. Optimized splitting provides peak heights not far below that of the initial band. This means that the concentra­ tions of the solutes in the carrier gas leaving the column are not far below those in the initial band.

Splitting Optimized for Solute of Interest

Splitting must, hence, be optimized for the solute of interest. The optimum split flow rate is lower the more the com­ pound is retained. It should not be a worry that earlier eluted compounds then form clearly broadened peaks.

6.3. Maximum Vapor Concentration in the Injector

As a next step we should consider factors determining the concentration of solute in the gas entering the column. Dur­ ing the time available, e.g., 1 s, and with maybe 30 III gas/ vapor 'entering during this time (column flow rate 1.8 mL/ min), a maximum amount of solute material should be trans­ ferred. Because the maximum concentration is that in the sample, we should minimize dilution with carrier gas.

6.3. 1. Sample Volume

We are used to the idea that increasing the volume injected enhances sensitivity. Split injection with a split flow rate mini­ mized as described above is different. At first sight, the sam­ ple volume has no effect on sensitivity. If an initial band of a given width can be tolerated, an increased injection vol­ ume requires a correspondingly increased split flow rate to move the larger volume past the column entrance within the same period of time.

Adjustment of Sample Volume to Liner Volume

This is not fully true, however, because the injection volume influences the dilution of the sample with carrier gas. Dilu­ tion can be minimized by adjustment of the volume of sam­ ple to that of the liner: if the sample nearly fills the vapo.... izing chamber, it largely displaces the carrier gas and the center ofthe sample plug is highly concentrated. Correct and inappropriate ways are shown in Figure C11.

Optimum Sample Volume for Gaseous Samples

To estimate the optimum sample volume, a static situation is assumed: at small split flow rates, the sample vapor ex­ pands far more rapidly than it is discharged through the split outlet or into the column. Mixing with carrier gas might double the size of the cloud (a rough estimate depending on injector geom­ etry, injection speed, turbulence, etc.). Heating to the injector temperature causes expansion, e.g. by approximately 50 % at 150°C (a lower injector

180

C 6. Analysis Requiring Maximum Sensitivity

I

A

B

c

I D

Figure C11 Split injection maximizing sensitivity. A Sample size well adiusted to the vaporizing chamber, leading to minimum dilution with carrier gas. B Excessive sample volume for the given liner, causing sample material to be lost through the septum purge and/or to penetrate the gas supply system. C Syringe needle too short. The volume of the vaporizing chamber is not used optimally and the vapor is diluted on its way to the column entrance. D Excessively large vaporizing chamber or insufficient sample size, resulting in strong dilution. temperature is usually sufficient as the sample is gase­

ous even at ambient temperature).

The carrier gas inlet pressure compresses the cloud,

e.g., to half its size at 1 bar inlet pressure.

As a rule of thumb, the appropriate sample volume can be estimated as the internal volume of the vapor­ izing chamber divided by two. Liquid Samples

For liquid samples, analogous estimates of optimum sam­ ple volumes are more complex as the vapor volume pro­ duced per unit volume of liquid depends on the na­ ture of the latter, which is usually mainly solvent. For in­ stance, a given volume of condensed water produces ap­ proximately seven times as much vapor as hexane. This sub­ ject is discussed in Section D3 (splitless injection). Table A4lists recommended injection volumes for vaporiz­ ing chambers of different sizes. 1 III of liquid is assumed to be turned into 200 III of vapor (volumes of liquid understood in absolute terms, i.e. including that possibly eluted from the syringe needle).

Discharge of Vapor during Injection

The following estimate should verify the extent to which the above assumption, i.e. that no relevant part ofthe sample is discharged during injection, is realistic. 2 III of a liquid sam­ ple (forming some 400 III vapor) is evaporated in ca. 0.5-2 s.

6.3. Maximum Vapor Concentration in the Injector

181

TableA4

Sample volumes recommended for split injection with minimum split ratio for liners of differ­

ent sizes.

Liner Length [mm]

Internal width [mm]

80

120

Sample volume [~I] Liquid samples Gaseous samples

2

0.6

3.5 5

2

2

3.5 5

3.5 1 3 5

130 400 700 200 600 1200

Injection of 400 ~L of headspace gas also takes about 0.5 s. 400 ~L of vapor introduced or formed within 0.5 s corresponds to a rate of 48 mLJmin. Vapors Replacing Carrier Gas

If the flow rate through the injector (split flow rate plus column flow rate) is 48 mLJmin, the sample is carried to­ wards the column entrance at the same rate as it leaves the needle and none of the vapor expands backwards towards the septum. Under these conditions, unlimited volumes of sample can be injected without overloading the injec­ tor. Basically, the sample vapor is not diluted, because the carrier gas is replaced by sample. At a split flow rate of 25 mLJmin, half of the injected vapor is still carried forward during its introduction while the other half fills the injector, enabling us to inject twice the amount recommended above. Hence the recommended injection volumes are upper limits only if split flow rates are below 20 ml/min. This is, however, mostly the range of interest for our application.

6.3.2. Optimum Liner Volume

A large vaporizing chamber enables the injection of relatively large volumes. This was shown to be of limited usefulness, because a large liner calls for a large sample volume, which in turn calls for an increased split flow rate. There is, how­ ever, a minimum size.

Volume of Vapor Transferred into the Column

The volume of the vaporizing chamber must exceed that of the gas and vapor transferred to the column and can be esti­ mated as in the following example. If the tolerable initial band width is 3 s (optimum for peaks of ca. 6 s width) and the column flow rate 3 mlzmln, 150 III of vapor/gas mixture enters the column. In an extreme case, a 5 s initial band width is acceptable and the column flow rate is 6 ml/min, i.e. 500 III of (dilute) sample vapor could reach the column.

Purge Flow

If the internal volume of the vaporizing chamber corresponds exactly to the volume acceptable for transfer into the col­

182

C 6. Analysis Requiring Maximum Sensitivity umn, theoretically no split flow would be needed and the injection become splitless. Some purge flow is, however, required to remove residual vapor (to avoid tailing peaks) and to reduce dilution with carrier gas. It is probably 5-10 mUmin.

Recommended Liners

Liners of 3-4 mm Ld, with an internal volume of 560-1000 III are recommended, enabling injection of 300-500 III of gaseous samples or 1.5-2.5 III of liquid (including that eluted from the syringe needle). At split flow rates of 25-40 rnl./rnin, these sample volumes can be doubled.

6.3.3. Position of the Column Entrance

The column should be installed in the injector such that its entrance is positioned ca. 5 mm above the bottom of the liner (Figure C12). This is a compromise resulting from consideration of two factors. As the carrier gas flows from the top to the bottom of the vaporizing chamber, only the volume above the column entrance can be used to store sam­ ple material between injection and transfer into the column. Material "shot" below the column entrance does not partici­ pate in the splitting process. Gas and Liner

IU Accumulated

~:- ~~'I> ACti~e

'~rt"

metal surfaces

Column inlet Figure C12

Bottom section of the vaporizing injector. The column en­

trance should be positioned as low as possible. but clearly

above the contaminants usually accumulated there.

Accumulation of Particles

The other concern is the septum particles and other invola­ tile material accumulated at the bottom of the vaporizing chamber, and the metal surfaces which degrade labile com­ pounds. The column entrance should be about 5 mm above this "garbage bin". In this way, material reaching the bottom of the vaporizing chamber leaves through the split outlet; adsorption by contaminants or degradation on metal surfaces have no influence.

6.3.4. Injection Point

Diffusion speeds of the vaporized sample components in the vaporizing chamber are high, resulting in rapid dilution with carrier gas. To achieve rapid transfer to the column, the sam­ ple must be released from the syringe needle near the col­

6.3. Maximum Vapor Concentration in the Injector

183

umn entrance. For gases (headspace analysis), the injection point should be ca. 10 mm above the column entrance, taking into account some expansion of the sample cloud to­ wards the column. 6.3.5. Syringe Needles

Forthe most common vaporizing chambers of 80 mm length, needles of gas syringes should be 85 mm long, whereas standard 71 mm needles suit the injection of liquids. For the injector of CEInstruments/ThermoQuest with a length of 120 rnrn, it is recommended thatthe liner with the restric­ tion at the bottom is used. This reduces the length of the vaporizing chamber to 80 mm - excessively long needles would be needed otherwise. Standard gas syringes are equipped with 51 mm nee­ dies, even though this is not appropriate for any commer­ cial injector. Syringes with longer needles must, therefore, be specially ordered. Most autosamplers work with short needles and do not allow the use of longer ones.

Length of the Needle

Needles with Side Port Holes

When :syringes with special needles are ordered, another problem should be solved at the same time. Especially when hard (high-temperature) septa are used, standard 22-gauge needles of gas syringes frequently cut out rather large pieces, causing severe leakage after a few injections only. Softer, more elastic septa are superior in this respect (appropriate also because injector temperatures are low), but needles are all too often plugged by septum particles. This is sel­ dom observed when liquids are injected, not because the needle is less frequently plugged, but because violent evapo­ ration inside the needle removes the particle (found later at the bottom of the vaporizing chamber). Both problems are solved by use of needles with a side port hole (Section A2.3.2).

Plungers with PTFE TIp?

Plungers of standard "gas tight" syringes are equipped with PTFE tips. They are usually a tighter fit than steel plungers. At temperatures up to ca. 150°C, PTFE behaves as a solid and does not pick up solute material. Small amounts of con­ taminants coating the PTFE surface or entering the pores are, however, sufficient to turn it into a stationary phase absorbing sample material. Steel plungers are kept clean more easily and might therefore be preferable. For reasons of tightness, the syringe should be filled only partially and the plunger should not withdrawn beyond ca. 80 % of the graduation (e.g. the 400 III mark for a 500 III syringe). Tight­ ness can be tested by applying a drop of a low viscosity liq­ uid, such as hexane, at the point where the plunger leaves the barrel: escaping gas forms bubbles.

6.4. Column Flow Rate

As shown above, a maximum amount of sample material should enter the column within a time corresponding to the maximum tolerable initial band width. At a given (high) con­ centrationof the sample material, this amount is determined

~lDJ\DDEANTIOQUlA

BlBLIOTBCA CENTRAL

184

C 6. Analysis Requiring Maximum Sensitivity by the carrier gas flow rate into the column. Increasing the column flow rate proportionately increases the amount of sample material carried into the column per unit time.

Mass- or Concentration­ Dependent Detectors

Increasing the carrier gas flow rate increases the amount of material entering the detector in unit time, but does not in­ crease the concentration of the solute material in the carrier gas. Hence the response of mass-dependent detectors increases. but not that of concentration-dependent detectors. Does this affect the above statement? Flame ionization detectors (FlO, NPO or AFIO) and mass spectrometers are mass-dependent. Basically electron cap­ ture detectors (ECD) and flame photometric detectors (FPD) are concentration-dependent, but since the column flow rate represents only a small fraction of a more or less constant total flow of detector gases, an increase in the column flow rate also causes an increase in the sample concentration in the detector. It is, therefore, more appropriate to consider them as mass-dependent also.

6.4. 1. Low Split Ratios Resulting from High Column Flow Rates

If the column flow rate is, e.g., 10 mUmin, a split flow rate of 10 rnt/rnln means a split ratio of just 1:1 or that half the sample enters the column. As the flow rate through the injector is 20 rnt/rnin, the initial band width is ca. 2-3 s, enabling reasonable chromatography if the peaks of interest are not eluted very rapidly. In this way, the sensitivity ob­ tained by split injection is inferior to splitless injection by merely a factor of two.

6.4.2. Selection of the Carrier Gas

Chromatography must be optimized for high column flow rates. Hydrogen enables the use of the highest flow at a given separation efficiency. For short columns (inlet pres­ sures below 1 bar), helium is nearly equivalent. Nitrogen is not suitable. To obtain comparable column performance, gas flow rates must be kept ca. three times lower owing to slow diffusion in this gas.

6.4.3. Selection of the Column

At first sight it seems that wider bore columns enable the use of higher carrier gas flow rates. We are not, however, looking for the column resulting in a high gas flow rate at a low inlet pressure, but that providing maximum sensitiv­ ity at a high separation efficiency.

0.53 mm or 0.32 mm i.d. Columns?

At a given gas velocity, the volumetric gas flow rate through a 0.53 mm i.d. column is 2.7 times greater than that through a 0.32 mm i.d. column, which seems to favor the use of wide bore columns. Increasing the column diameter results in loss of separation efficiency, however, and optimum gas veloci­ ties are lower. Comparison of the separation efficiencies (in TZ) of columns of different diameter and length shows that 0.32 mm i.d, col­

6.4. Column Flow Rate

185

umns are superior even at extremely high flow rates (e.g. 20 ml/min) [36). A 5 m x 0.32 mm i.d. column usually provides almost the same separation as a 10 m x 0.53 mm i.d. col­ umn. Columns of 0.25 mm i.d. would be yet more efficient, but these must be short owing to the high inlet pressures required (an important drawback in the practice of headspace analysis, because high retentive power is needed). For such reasons, 0.32 mm i.d. columns are the most appropriate.

6.5. Summary: Maximum Sensitivity from Split Injection

1 2 3

4 5 6 7

Do not use nitrogen as carrier gas; hydrogen is some­ what better than helium. Use 0.32 mm i.d. columns at high flow rates. For heads pace analysis of volatile components, high gas flow rates necessitate the use of columns with rather thick (e.g. 1 urn) films and maybe increased length (25-30 m) to obtain sufficient retention of the solutes. Use 3-4 mm i.d. liners. For gaseous samples use open tubular (empty) liners. Keep injeetortemperatures low (for gaseous samples, 80-100 °C is sufficient). Inject sample volumes as recommended in Table C4, using a syringe equipped with a long needle. Reduce the split flow rate until the first peak of inter­ est starts to become broad and/or no longer grows higher.

7. High Split Ratios for Reducing the Sample Size At the other extreme of applications involving split injection are those requiring high split ratios to enable injection of highly concentrated or undiluted samples. Some sam­ ples cannot be diluted because the solvent peak would ob­ scure important peaks. For others, e.g. from quality control in an industrial process, dilution is omitted for the sake of simplicity. Solvent Promoting Evapora­ Before beginning this topic it should be remembered that tion samples are often diluted not only to avoid the use of high split ratios, but also because of the improved precision and accuracy of results. Solvents facilitate evaporation in the injector, particularly with the hot needle technique, where they act as a propellant nebulizing the sample liquid.

186

C 7. High Split Ratios for Reducing the Sample Size

Viscous Samples

Viscous samples cannot be injected accurately without prior dilution, because they tend to be difficult to separate from the needle tip. Instead of being transferred into the vaporizing chamber, they remain hanging on the needle, are withdrawn together with the latter, and are usually wiped off on the septum.

7.1. Diluent as a Hypo­ thetical Sample

To render our discussion more specific, we assume that the sample consists of a diluent composed of four solvents, each representing 25 % of the total and that dilution with another solvent is impossible.

Split Ratio Required

Because the column capacity for each component might be 50 ng (intermediate film thickness), a total of 200 ng of sam­ ple may enter the column. If we inject 1 ul, (1 mg) into the vaporizing chamber, a split ratio of 5000:1 is required. With a 0.32 mm i.d. column and a carrier gas flow rate of 3 mLJmin, this requires a split flow rate of 15.000 mlJmin - or 900 IIh... This is obviously impossible and the question arises whether this kind of injection feasible at all. Yes it is, but only after optimization.

7.2. The Maximum Split Flow Rate

For reasons to be listed below, split flow rates of ca. 1000 mlJmin should not be exceeded. This is 15 times less than required for our hypothetical sample. Flow-regulated gas supply usually restricts the split flow rate to 500 mLJmin.

Gas Consumption

With a split flow rate of 1000 mLJmin, a large 50 L gas cyl­ inder is emptied in 7 days. It is unreasonable to exceed such consumption. The split flow rate can be turned down shortly after injection since the high flow rate is only needed during sample evaporation. Automated reduction ofthe split flow rate (gas saver) is available on some instruments.

Resistance in the Split Line

The most obvious problem with high split flow rates is the resistance to gas flow in the split line. Even when the split valve is fully opened (pressure regulation/flow resistance system), it may still be impossible to achieve a split flow rate of 1000 mLJmin (depending on inlet pressure). With mechanical backpressure-regulated systems, the col­ umn inlet pressure goes out of control. It will be too high because of the pressure drop across the filter, and the user is hardly likely to notice this. Electronic systems avoid the problem, because the pressure sensor is positioned in the gas supply line or septum purge outlet.

Pre-Heating of the Carrier Gas

The incoming carrier gas must be pre-heated before it reaches the vaporizing chamber - otherwise it acts as a coolant in exactly the zone where the sample should be evaporated. The efficiency of such pre-heating is limited. In fact, the tem­ perature in the upper part of a vaporizing chamber ther­

7.2. The Maximum Split Flow Rate

187

mostatted at 350°C was found to drop to 320 °C when the

split flow rate was set at 100 mt/min.

Cooling as a result of high carrier gas flow rates is particu­

larly critical for the rear part of the inserted syringe nee­

dle, where losses of high-boiling sample material primarily

occur and sensitively react to too Iowa temperature.

Time for Sample Evaporation

The sample liquid leaving the syringe needle should be

evaporated or converted to a fine aerosol before it reaches

the column entrance (split point). For evaporation in the gas

phase after nebulization at the needle exit, the time avail­

able depends on the gas velocity in the vaporizing chamber,

i.e. on the split flow rate. If the liquid is deposited on to a packing material, the flow rate is less critical, because the liquid remains there until evaporation is complete.

7.3. Small Sample Vol­ umes

If we assume a maximum reasonable split flow rate of 1000

mLjmin, the analysis of our undiluted hypothetical mixture

of four components requires the optimization of other vari­

ables to remove the discrepancy of a factor of 15. Here we

look at the possibilities of reducing the sample volume be­

low 1 ul.,

Evaporation from the Needle

With injection performed manually or by use of autosamplers

imitating manual injection, elution of the sample contained

in the syringe needle can hardly be avoided. Even if a sam­

ple volume of "0 Ill" is read on the barrel ofthe syringe, the

0.5-0.7 ~L inside the needle are transferred.

It is not possible to back up a smaller amount of sample by a

plug of solvent (solvent flush injection) because the latter

would produce a solvent peak disturbing the chromatogram.

Hence 0.5-0.7 III is the smallest volume that can be intro­

duced in this way.

Fast Autosamplers

Fast autosamplers are designed to avoid evaporation inside

the needle and, thus, eliminate this problem. They enable

reasonably precise and accurate injection of sample volumes

down to about 0.3-0.4 ~L. Smaller amounts are likely to be

incompletely separated from the syringe needle. This can

easily be checked by injecting 0.1, 0.2, and 0.5 III and com­

paring absolute peak areas.

High-Boiling Samples

Evaporation inside the needle can be avoided for undiluted

higher-boiling samples. When using 10 III syringes, volumes

down to 0.2-0.3 ~L can be measured with reasonable accu­

racy. With 5 III syringes, this lower limit is ca. 0.1-0.2 Ill, but

routine work with their thin plunger is delicate.

Separation from the Needle Tip

When small volumes are injected, the accuracy of abso­

lute peak areas, i.e. of sample volumes, tends to be poor

because the liquid is not accurately separated from that re­

maining inside the needle. On one occasion, additional liq­

188

C 7. High Split Ratios for Reducing the Sample Size uid is pulled from the needle, easily adding 0.1-0.2 ~L to the

volume read on the barrel, whereas on the other occasion

0.1-0.2 ~L remain hanging at the needle tip, particularly if

samples are viscous or the plunger is depressed less rap­

idly. Partial evaporation from liquid remaining on the needle

tip distorts the sample composition and results in poor

accuracy even when an internal standard is used.

Plunger-in-Needle (1 J.lL) Syringes

Plunger-in-needle syringes enable accurate measurement of

volumes as small as 0.05-0.1 ~L (Section A2.2). Problems arise

if samples are volatile (as is our hypothetical sample) or dis­

solved in a volatile solvent (Section A9). Upon passage

through the septum the needle is rapidly heated, initiating

premature evaporation inside the needle. The first vapor

might be lost through the septum purge. Evaporation ofthe

liquid contained in the narrow space between the nee­

dle and the plunger means that an additional 0.2-0.3 ~L of

sample enters the injector. Hence the smallest possible

sample size is ca. 0.3 J.l.L and can only be determined by

calibration with a larger, accurately known volume. The use

of an internal standard helps to overcome poor reproduc­

ibility resulting from inaccurate sample size.

For higher-boiling samples not evaporating within the

needle, separation of small amounts of liquid from the nee"

die tip is again the problem, here accentuated by the rather

low velocity of the liquid.

Wet Needle Injection

A simple trick, the "wet needle", technique enables injection

of very small sample volumes even with standard 10 J.l.L

syringes. The syringe needle is filled with sample liquid.

Shortly before injection, the plunger is withdrawn, which pulls

the bulk ofthe liquid into the barrel ofthe syringe, but leaves

a film of sample on the inner needle wall. The needle is

now inserted into the injector and withdrawn after 2-3 s with­

out depressing the plunger. This results in evaporation and

transfer of only the sample liquid coating the needle wall,

ca. 0.03-0.06 J.l.L. The amount depends on the viscosity of

the sample and the speed the plunger is withdrawn: slow

withdrawal leaves behind a thinner film.

Wet needle injection is well suited to analyze volatile sam­

ples, such as the solvent mixture discussed above, but high­

boiling components are likely to remain in the needle. Abso­

lute peak areas tend to be poorly reproducible and quantita­

tive analysis requires the use of an internal standard.

Conclusion

For volatile samples, the wet needle technique is most ap­

propriate. It enables injection of our hypothetical mixture with

a split flow rate of about 500 mLJmin before other factors are

optimized.

For samples boiling above some 150 °C or those con­

taining high-boiling solutes, there is no possibilty to reduce

the sample volume below 0.3-0.5 ~L.

7.4. Low Column Flow Rate

189

7.4. Low Column Flow Rate

The split flow rate of 15,000 mLJmin for our hypothetical sam­ ple was calculated by multiplying a column flow rate of 3 mLJmin by the split ratio of 5000:1. Reducing the column flow rate enables proportional reduction of the split flow rate. This situation is just the opposite of the applica­ tion requiring maximum sensitivity, where a maximum amount of solute material must be rinsed into the column in a given time.

Narrow Bore Columns

In Section C5.1, short, narrow bore columns were suggested for rapid, isothermal analysis, assuming sufficiently sharp initial bands could be produced by use of high split ratios. This is the application we are now studying from another angle. A carrier gas velocity of 25 crn/s through a 0.10 mm i.d. column produces a flow rate of around 0.15 mU min, which is at least ten times lower than for the same gas velocity through a 0.32 mm i.d. column.

Reduced Capacity of Narrow­ Bore Columns

The lower column capacity must be taken into account. For the same film thickness, the narrow bore column has a ca­ pacity maybe 5 times lower (owing to the sharper peaks, capacities decrease more rapidly than column diameters). This must be accommodated either by increasing the split ratio or by use of a thicker film of stationary phase.

Slow Carrier Gas: Nitrogen

A carrier gas velocity of 25 cm/s is within the optimum for hydrogen and, if the column is short, for helium. Helium can be used at 15-20 crn/s, but nitrogen even as low as 8-12 cm/s. If a narrow bore column is used with nitrogen, the column flow rate can be kept as low as 0.05 mUmin. By this means, the split ratio required for our hypothetical sam­ ple is obtained with a split flow rate of merely 250 mLJmin.

7.5. High Column Capac­ ity - Thick Films

Our calculation assumed a column capacity of 50 ng for each of the four compounds. Column capacity increases more than in proportion to the amount of stationary phase per unit length of the column or film thickness. In 0.25 or 0.32 mm i.d, columns, film thicknesses of methyl silicones can be in­ creased up to 5 11m, those of stationary phases of intermedi­ ate polarity up to ca. 111m. This provides a capacity approach­ ing or even exceeding 1 Ilg per component, i.e. the mix­ ture could be analyzed by using a split ratio 20 times lower than assumed.

Increased Retentive Power

A thicker film also increases retentive power, i.e. retention times and/or elution temperatures. If compounds are chrom­ atographed with comparable retention times in columns with a tenfold thicker film, elution temperatures are nearly 50° higher. This is an advantage for the analysis of volatile and adsorptive compounds, but a drawback for those eluted at elevated temperatures. Increased film thicknesses lead to somewhat reduced separation efficiency, partly because dif­

190

C 7. High Split Ratios for Reducing the Sample Size fusion in the stationary phase becomes time-consuming. The latter requires reduced carrier gas velocities and prefer­ ably "slow" carrier gases, such as nitrogen [37,38]. This is, however, compatible with our interest in keeping split flow rates low.

7.6. Length of the Sy­ ringe Needle

long syringe needles were recommended for split injection aiming at high sensitivity, firstly to release the sample close to the column entrance and, secondly, to prevent expansion of sample vapor backwards out of the chamber. Neither ar­ gument remains valid if split flow rates are high.

No BackfJow at Split Flow Rates Exceeding 75 ml./min

If 1 ul, of liquid could be injected and evaporated within 0.2 s (which is considered rapid), the vapor generated (about 0.25 ml) would expand at a rate of 75 mLlmin. If the flow rate through the injector is 75 mt/min, the vapor just replaces the carrier gas and there is no risk of sample vapor flowing backwards out of the chamber.

Needle Entering Liner by 10mm

Short syringe needles release the sample far away from the column entrance, which prolongs the time available for vaporization and achieving homogeneous distribution across the liner. Further it enables heat to be absorbed from a long section of the liner. The syringe just needs to deposit the sample safely in the vaporizing chamber, i.e. enter it by ca. 1 cm. This means that syringes with needles ca. 25 mm long are best suited (depending on the design of the injec­ tor head).

Reduced Problems Arising from Evaporation inside the Syringe Needle

If the sample evaporates within the syringe needle, the vol­ ume unintentionally transferred from a 25 mm needle is re­ duced to about 0.25 ~L and the smallest amount which can be injected (reading 0 ul, on the barrel) becomes correspond­ ingly small. With short syringe needles, even manual injection can often prevent evaporation inside the needle, because the short, less pliable needle can be introduced more rapidly and merely reaches into the upper region of the injector which is, or can be chosen to be, substantially cooler than the center of the injector.

Partial Introduction of the Needle

The length of the shortest needle considered standard is 37 mm. Shorter needles can be ordered for 5 or 10 ul, syringes, but not for 1 ul, syringes since the volume inside the needle would then become too small. Instead of using special syringes, needles of standard sy­ ringes can be introduced only partially. For some auto­ samplers the depth of the injection point can be adjusted by programming the movement of the syringe. For manual injection, a spacer tube is useful so that the depth is reproducible (see Figure C13).lf used frequently, it is convenient to glue the spacer tube to the bottom of the

7.6. Length of the Syringe Needle

191

barrel of the syringe. It should have an internal diameter closely fitting the outer diameter of the needle, to hinder bending of the needle during penetration of the septum. A piece of 0.5 mm l.d, x 1/16 inch o.d. stainless steel tubing is suitable. The drawback of this makeshift is that the rear of the needle which remains outside the injector is also warmed, and there may still be evaporation if injection is not performed sufficiently rapidly.

Glue Septum

Spacer

=n n110

11=*6== Septum purge

Carrier gas

mm

II I IInjector liner FigureC13 Split injection at high split flow retes: the syringe needle should enter the liner by ca. 10 mm only. If syringe needles are longer than needed, a spacer tube can be used to repro­ duce positioning of the needle outlet.

7.7. SummBrizing Guide­ lines

2 3

4

5

Use a narrow bore capillary column (0.1-0.2 mm i.d.) at a low flow rate, possibly with nitrogen as carrier gas. Choose a column with a thick film of stationary phase for high capacity. Use a wide bore liner containing well deactivated glass or quartz wool (but an empty liner if the sample contains labile components). Inject the smallest volume of sample possible. If evaporation in the needle can be avoided, 1 or 10 ~L syringes can be used. If the sample is volatile, try the wet needle injection technique. Quantitate by use of an internal standard. Apply the smallest split flow rate not causing over­ loading of the column.

192

C 8. Problems Concerning the Split Ratio

8. Problems Concerning the Split Ratio 8.1. Purposeful Search for Errors

Split injection is a complex process in which many things may go wrong. Because there are no simple procedures for eliminating all problems, the analyst must regularly check the accuracy of his results. This is not always simple, be­ cause the common determination of reproducibility, e.g. by injecting a mixture of standards three times, does not rule out serious errors.

8. 1. 1. Systematic Errors

Before beginning a discussion ofthe mechanics of what might happen to distort the results, the perspective of the analyst should be considered. He injects and obtains peak areas ­ anvthinq in between is a black box. It takes several experi­ ments to find out whether these peak areas correspond to what they should be and whether the results are sufficiently accurate. It takes several more experiments to locate the source of the problems if there are any.

Puzzling Deviations

Quantitative analysis with split injection is often demand­ ing, because there can be deviations which are difficult to define and of which we have nothorough grasp. Results may, for instance, be different from one day to another, or upon returning to a method after some time, even though we be­ lieve we have reproduced all the conditions. There is, of course, no magic behind it, but the lacking insight into the processes occurring during injection confronts the ana­ lyst with tough riddles.

The Fault of the Analyst or the Technique?

Literature and reports at conferences generally tend to paint a rosy picture. Even in personal contact problems are mostly played down, perhaps because people do not want to admit their trouble. All too often poor results resulting from imper­ fections inherent in a technique are mistaken as personal incapability. Analysts usually commit fewer "silly" mis­ takes than they think and they overestimate the pe.... fection of a technique. Many analysts play around until results get better. Often they go through an optimization the background of which they do not understand.

Insidious Systematic Errors

Among the deviations, those causing systematic errors are most troublesome. Systematic errors occur, for instance, if a deviation is different for the calibration mixture and the sam­ ples. Systematic errors are insidious, because they are diffi­ cult to recognize. How can systematic errors be detected?

8.1. Purposeful Search for Errors

193

8.1.2. Message from Standard Deviations

Experiments to detect errors frequently concentrate on the

determination of standard deviations, i.e. of random errors,

considering standard deviations as indicative of accuracy.

Such statistical thinking probably originates from purely

physical processes, where the results are commonly scat­

tered around the correct value and deviations obey statisti­

cal rules.

Standard deviations are, indeed, a measure for probable ac­

curacy if systematic errors can be ruled out, for instance when

(almost) identical solutions are compared under same con­

ditions and the system performance does not drift. In chro­

matography this seldom occurs, however.

Maybe standard deviations are as indicative of errors as are

body temperatures for illness - high standard deviations are

an alarm signal, but low standard deviations do not rule out

a malignant illness.

Larger Differences between Laboratories

It is well known, for instance, that deviations between the

results reported by different laboratories usually far exceed

the standard deviations of the results obtained by each

participant (reproducibility and repeatability). This tells us

that systematic errors are involved, often larger than the ran­

dom errors determined by the measurement of standard

deviations. Large deviations are common even if all labora­

tories seem to apply the same method, showing that not all

the relevant factors are under control.

Random Deviations

Standard deviations are the result of deviations from the

correct result. When shooting at a target, deviations in all

directions might be of similar probability. The mean value

would then be close to the target (true result) and the stand­

ard deviation a good measure for the precision and accu­

racy of shooting. In this example, random deviation predomi­

nates and the standard deviation is an appropriate measure

of the deviation.

Systematic Deviations

In GC, however, deviations tend to move in a given di­

rection. The averaged results of repeated analyses can be

far away from the true value (systematic error). The stand­

ard deviation is then a measure of the reproducibility of

the deviation from the correct result and might be far

smaller than the systematic error, i.e. provide a misleadingly

encouraging evaluation of the method.

The expansion of the sample (solvent) vapor upon sample

evaporation, for instance, changes the split ratio. The devia­

tion always has the same direction, the extent depending on

instrument design and conditions. The standard deviation

measures the reproducibility of this deviation - but the ana­

lyst wanted to know whether or not there is a devia­

tion, not whether it is reproducible.

In chromatography, quality assurance by repeating analy­

ses many times gives the feeling of "having done some­

194

C 8. Problems Concerning the Split Ratio thing", but easily turns out to be an alibi rather than an ef­ fective search for problems.

Knowledge about Systematic Errors

The detection of reproducible deviations requires more pur­ poseful experiments. Comparison with results obtained by an independent method would, of course, be most desirable, but this is often impossible because none is available. Fur­ thermore, when the results obtained from two methods are different, it is usually difficult to determine which is more correct. Thus there is usually no alternative to searching for errors within the method applied. Purposeful recognition of systematic errors presupposes knowledge about them to enable the critical checks to be performed and the application of methods of quantltation which are immune to the errors.

Overemphasis of Problems

Throughout this treatise on injection techniques, but particu­ larly in the following sections, we concentrate on problems and stress how badly they can distort the results. We do so assuming that these chapters are consulted when problems occur. Of course, for many analyses there are no such diffi­ culties, and we hope that the reader is sufficiently experi­ enced to maintain a balanced evaluation of the tech­

nique. The Three Types of Problem

The three main problems in quantitative analysis by split in­ jection are:

sample evaporation; the split ratio, which often deviates from that adjusted; the composition of the sample material entering the column, which can easily be different from that in the injector (non-linear splitting). The first was treated in Part B, the last two are the subjects of this and the following chapters.

8.2. "Pre-Set" versus "True" Split Ratio

An important source of error is directly or indirectly related to alteration of the split ratio by the injection. The propor­

tion of the sample material entering the column does not correspond to that expected from the ratio of the gas flow rates. Deviations frequently amount to a factor of 1.5-5, but factors as high as 30 have been observed.

The "Pre-Set" Split Ratio

The "pre-set" split ratio [39] corresponds to that adjusted as described in Section 2. It is determined by the ratio of the gas flow rates directed to the split outlet and entering the column, both measured before injection. It could also be called the "adjusted" or "expected" split ratio.

The "True" Split Ratio

The "true" split can only be determined after the analysis by calculating the proportion of sample which actually entered the column. A split/ess or on-column injection is performed to assess the peak area of a solute if all material entered the column. The peak area obtained from the split

8.2. "Pre-Set" versus "True" Split Ratio

195

injection divided by that from the non-splitting method yields

the true split ratio.

Actually the true split ratio is the average effective split

ratio resulting from a ratio which changes during sample

introduction.

8.3. Mechanisms Caus­ ing the Split Ratio to Deviate

There are several reasons why the true split ratio deviates

from that pre-set. Their description helps to identify the

source of the problems so that appropriate action can be

taken.

8.3.1. The Pressure Wave

Rapid sample evaporation in the vaporizing chamber pro­

duces a relatively large volume of vapor. In displacing a cor­

responding volume of carrier gas it produces a pressure

wave. This expansion occurs within a very short time - a

process resembling an explosion. It upsets the adjusted split

ratio, often exactly at the time the sample is being split.

Volume of Vapor

The volume of vapor produced from the 1-2 III of liquid sam­

ple commonly injected is easily underestimated: 2 III of liq­

uid creates a vapor cloud with a volume from 0.3 to more

than 2 ml (primarily depending on the solvent). The internal

volume of a 2 mm i.d. liner is ca. 0.25 rnl., that of a 4 mm i.d,

chamber 1 mL. Hence the vapor volume is comparable with

the volume of the injector itself.

Pressure Increase

If the sample vapor has, e.g., a volume corresponding to that

of the vaporizing chamber - which is already filled with car­

rier gas - the amount of gas to be housed in this chamber is

doubled, which should cause the absolute pressure to

double. If the inlet pressure is, e.g., 100 kPa (200 kPa abso­

lute pressure), the pressure should increase to 300 kPa (400

kPa absolute). This does not happen because the chamber is

open, both backwards towards the carrier gas supply and

forwards to the split outlet. The expanding sample vapor dis­

places carrier gas, pushing it into the tubings and cavities

accessible around the chamber. Because these volumes are

several times larger than the vaporizing chamber, pressure

increases by much less than a factor of two. If, for instance,

the volumes around the vaporizing chamber are nine times

larger, the pressure increase is calculated as 10 % (rela­

tive to the absolute pressure).

Effect on the Column Flow Rate

Expanding sample vapor generates a pressure wave propa­

gating from the center of evaporation. For hot needle injec­

tions it is located just below the exit of the syringe needle.

The pressure wave affects all the gas flow rates includ­

ing those into the column and towards the split outlet. Flow

rates change in a complex manner and the effect on the split

ratio is difficult to predict.

In Figure C14, the changes in the pressure at the column

entrance and the flow rate into the column are shown

schematically for the pressure regulator/flow resistance type

196

C 8. Problems Concerning the Split Ratio carrier gas supply system. A rapid increase in pressure in­ creases the flow rate into the column inlet far more than in proportion to the pressure, because it compresses the gas in the inlet rather than increasing the flow rate through the whole column (period A). As the pressure starts to decrease (period B), the flow rate into the column drops and, if the column inlet pressure exceeds the (reduced) pressure in the injector, there even might be some flow backwards out of the column.

Time

'..........I.....f - - - - - - - - · · I.....I - - ­

A

8

c

Figure C14 Qualitative diagram of pressure and flow changes caused by explosive sample evaporation in the injector.

Different Effects on Split Flow Rate at Different Times

The split flow is affected by a similar wave, but whereas the pressure wave is sharp and hard at the column entrance, it loses much if its thrust when entering the split outlet. As a result, the flow rates entering the column and leaving through the split outlet change by different amounts at differenttimes and the split ratio fluctuates to values above and below the pre-set value.

Splitting of the Sample

Consideration of the moment at which the sample vapor is split shows the situation to be even more complicated. When the syringe needle releases the sample near the col­ umn entrance, the pressure wave primarily pushes sample vapor into the column. If the sample is released further from the column entrance, the vapor cloud might reach the col­ umn entrance at the time when the pressure decreases and the flow rate into the column is low or even reversed. The moment at which the vapor reaches the split point also de­ pends on the split flow rate and the size of the vaporizing chamber.

8.3.2. Dependence of the Pressure Wave on Gas Regulation

The two carrier gas regulation systems described in Section C1.3 react to the pressure wave in different ways. The man­ ner and speed in which they compensate for the extra volume added by the injection influence the build up of the pressure wave and the changes of the flow rates.

8.3. Mechanisms Causing the Split Ratio to Deviate Pressure Regulator/Flow Resistance

197

The pressure regulator/flow resistance system (Figure C15) reacts to a pressure increase by stopping further gas supply to the injector. The split flow rate increases in the pro­ portion to the pressure increase. If, for instance, pres­ sure increases from 100 to 105 kPa (which is more than is normally observed), it increases by 5 %. The pressure regulator re-opens after a volume correspond­ ing to that of the sample vapor has been released. If, for in­ stance, 1 mL of vapor is created and the split flow rate is 60 ml/min, discharge of the extra volume takes 1 s. Hence this kind of system re-establishes normal conditions only after a period of time which is long compared with the splitting process.

Regulator stops gas supply Evaporating sample produces expanding vapor

~l

S lit outlet Resistance (e.g. needle valve) keeps relatively constant split flow rate

Column

Figure C15 Reaction of the pressure regulatorlflow resistance system towards expanding sample vapor during sample evaporation.

Flow/Backpressure Regula­ tion

The system with the flow regulator upstream of the injector introduces the carrier gas at a constant rate irrespective of pressure changes in the vaporizing chamber (Figure C16). The extra volume generated by sample evaporation is re­ leased by an increase in the split flow rate. Upon pressure increase, the backpressure regulator opens fully to re­ establish the column head pressure - it controls pressure, not really the split flow rate.

Constant carrier gas supply into injector

Evaporating

sample produces

expanding vapor

Split outlet

Column

Regulator opens until column head pressure is reduced back to normal

Figure C16

Reaction of flow!backpressure regulation to expanding sam­

ple vapor.

198

C 8. Problems Concerning the Split Ratio

Changed Flow Rates

With backpressure regulation, pressure waves are weaker than with the pressure regulator/flow resistance system. An extra volume of gas corresponding to the vapor volume cre­ ated upon sample evaporation is released immediately. How­ ever, suppression of the pressure wave solves the prob­ lem from the wrong end: because the column flow rate remains fairly constant, the split ratio increases sharply. It is not the pressure wave which is of concern, but the change of flow rates.

Pressure regulation with flow resistance is preferable, because it keeps variations in the split flow rate small. The relatively pronounced pressure increase causes the split and column flow rates to change in the same direction, even though hardly by the same extent. 8.3.3. Recondensation in the Column Inlet

Recondensation of sample vapor in the column inlet is prob­ ably the mechanism with the largest potential to cause deviations between the true and the pre-set split ratio.

Contraction upon Recondensation

A~ low

oven temperatures, extensive recondensation of sam­ ple vapor can occur in the column inlet just below the at­ tachment to the hot injector (Figure C17). Recondensation results in a drastic reduction in volume (a factor of 150­ 1500). As incoming vapor practically disappears, the pres­ sure in the column inlet drops and causes additional vapor to be sucked into the column.

Injector insert

Oven roof

Partially recondensed sample Column entrance

Figure e17

Volume contraction resulting from recondensation of sam­

ple vapor in the column inlet causes additional vapor to be

sucked into the column, with the effect that more material

is chromatographed than expected from the pre-set split re­

tio.

Reduced Split Ratio

As a result, the flow rate into the column inlet far exceeds the column flow rate imposed. As the split flow rate remains unaffected, the split ratio is reduced and more sample material enters than is expected from the pre-set split ra­ tio.

8.3. Mechanisms Causing the Split Ratio to Deviate

199

Solvent or Non-Diluted Sample

To effect a noticeable change in the split ratio, a substantial proportion of the gas/vapor mixture must recondense. For dilute solutions, this presupposes recondensation of solvent vapor, i.e. a column temperature clearly below the boil­ ing point of the solvent. If the sample does not contain solvent, a large amount of solute material must recondense.

Experimental Data

Figure C18 shows peak areas obtained from 2 ~L split in­ jections of n-octadecane solutions (200 ppm) in different sol­ vents at different column temperatures. The injector tempera­ ture was 300 °C. A 2 mm l.d, liner accentuated the effect be­ cause it restricted dilution of vapor with carrier gas. It con­ tained some glass wool to ensure complete evaporation of the higher-boiling solutions also. The distance between the needle exit and the column entrance was 2 cm (71 mm nee­ dle). At 25 °C, the pre-set split ratio was 100:1. The split flow rate was kept constant for all injections. As the column flow rate decreases with increasing column temperature, the pre-set split ratio increased somewhat with increasing column tem­ perature. The peak areas to be expected from the pre-set split ratio were calculated by dividing the area obtained from a splitless injection by the pre-set split ratio (broken line). The peak areas obtained varied by a factor of nearly 10. At column temperatures above the solvent boiling point, about twice as much sample material entered the column than was expected from the pre-set split ratio. This might have been the result of the pressure wave. The deviation was, in fact, somewhat greater for the violently evaporating pen­ tane than for the octane solution. peak area

-~~~!!.~--!~~-------------" 20

60

100

150

column temperature

200 during injection (00

Figure C18 Dependence of peak area on the sample solvent and on the column temperature during injection. The peak area to be expected from the pre-set split ratio (Ncalculated area") is indicated by a broken line. (From ref. [40]).

200

C 8. Problems Concerning the Split Ratio

Recondensation 10-30° Below the Solvent Boiling Point

When the injection was performed at column temperatures ca. 10-30° below the solvent boiling point, the amount of sample entering the column increased rather dramatically. This agrees with the experience in splitless injection that effective solventtrapping, based on solvent recondensation, presupposes a column temperature at least 20-30° below the solvent boiling point.

Dew Point

Recondensation presupposes a column temperature below the dew point of the vapor/carrier gas mixture. For instance, 25°C below the standard boiling point, the vapor pressure is roughly 0.5 bar. At an inlet pressure of 0.5 bar ti.e. 1.5 bar absolute pressure), recondensation occurs if the vapor con­ centration in the gas phase entering the column exceeds 33 'Yo. Then vapor recondenses until its concentration in the carrier gas is reduced to 33 % (the partial vapor pressure corresponding to saturation). More dilute vapor recondenses only when the column temperature is lower.

8.3.4. Incomplete Evapo­ ration

I~ the sample liquid is "shot" towards the column entrance as a band of liquid and passes the latter before being evapo­ rated, the split ratio is determined by the probability of the liquid hitting the column entrance rather than by the gas flow rates. This probability is influenced by the ratio of the cross sections of the vaporizing chamber and the col­ umn entrance, but also by possible repulsion ofthe liquid by the hot column wall. If the sample vapor is not homogeneously distributed across the liner, similar problems are expected, although with less drastic effects.

Poor Reproducibility of Absolute Peak Areas

For incompletely vaporized samples it is typically observed that absolute peak areas are poorly reproducible - "large" chromatograms are followed, in an unpredictable manner, by others with small peaks. In most instances, variations are within 30 'Yo, but under extreme conditions, they can reach a factor of 10-20.

Sample in n-Octane

Peak areas obtained from a solution in n-octane were usu­ ally 2-5 times smaller than expected from the pre-set split ratio [401. indicating that most of the liquid passed by the column entrance (empty 2 mm i.d, liner). In one chro­ matogram out of about ten, however, peaks were eas­ ily ten times larger. Not knowing the background, most analysts would probably assume a misplaced sample vial or another silly error. The large peaks are, however, merely a sign of having scored a direct hit of the column entrance.

8.3.5. Cool Split Line

In many instruments the tubing of the split outlet line imme­ diately leaves the heated zone of the injector and passes through cool regions of the instrument towards the restrictor determining the split flow rate. If sample vapor reconden­

8.3. Mechanisms Causing the Split Ratio to Deviate

201

ses within this tube, its volume is reduced and the flow rate into the split line increases, just as described above for recondensation in the column inlet. Because this increases the split flow rate, the split ratio is increased and the peaks obtained are too small. For high-boiling solvents, deviations by up to a factor of five have been observed. Cool Needle Valve

Recondensation can also occur in the flow restrictor, such as a needle valve, at the end of a warmed split outlet line. The resulting contraction in volume again causes the split ratio to increase. The condensed liquid may also block the valve for a short time, with the opposite effect. Because hot split valves are inconvenient and electric regu­ lators cannot be heated, there is no perfect solution to the problem; the restrictor should, however, be warmed to at least ca. 50°C to rule out the possibility of recondensation of the solvents most commonly used.

8.3.6. Charcoal Filters

Many split injectors are equipped with a charcoal filter situ­ ated between the vaporizing chamber and the restric­ tor in'the split line. Originally they served to absorb sol­ vent vapor and to prevent vapor with a viscosity different from that of the carrier gas changing the flow rate through the restriction (needle valve). Today, the main function ofthe charcoal filter is to prevent contamination ofthe electric valve regulating the split flow rate. If the charcoal really absorbs the solvent vapor (and this can be expected only if it is regularly replaced), it has a nega­ tive effect stronger than the intended positive effect. When it removes the solvent vapor, it removes part of the incoming flow and causes an increase of the split flow rate similar to that resulting from the recondensation effect (re­ moval might be even more complete).

8.3.7. Buffer Volumes

Buffer volumes (empty cavities) with capacities of several milliliters, inserted between the vaporizing chamber and the restrictor at the split outlet, were suggested as a means of delaying passage of the sample vapor through the split valve. The intention was similar to that of charcoal filters. Vapor with a viscosity different from that of the carrier gas should flow through the restrictor only after all the solute material has passed the split point.

Pressure Wave Increases the Split Flow Rate

Sometimes it has been claimed that a buffer volume also solves the problem of the pressure wave, because the vol­ ume of vapor generated would be smaller relative to the larger space available. Additional readily accessible space prevents a high pressure wave, indeed, but if the extra vol­ ume of vapor is dissipated in the buffer volume, the pres­ sure wave is flattened by increasing the split flow rate.

202

C 8. Problems Concerning the Split Ratio

8.4. Minimizing the Deviation from the Pre­ Set Split Ratio

8.4. 1. Wide Injector Liner

8.4.2. Long Distance between Needle Exit and Column Entrance

Experimental Results

There is no way of completely avoiding deviations from the pre-set split ratio. Here the major factors are listed to help the analyst to avoid large deviations. Methods must be made immune to deviating split ratios by other means.

A wide-bore (e.g. 4 mm l.d.) liner results in a smaller pres­ sure wave, because the volume of vapor generated is smaller relative to the space available. The recondensation effect is weakened by dilution of the vapor with carrier gas. Dilution has two positive effects. It reduces the dew point of the vapor/gas mixture, i.e. recondensation occurs only at a lower column temperature. Secondly, recondensation from a diluted vapor phase results in less of a pressure reduction and less additional vapor be­ ing sucked into the column. There are three main reasons for releasing the sample from the syringe needle a long distance from the column entrance. 1 The pressure wave pushes carrier gas rather than sample material into the column. 2 Mixing of vapor with more carrier gas reduces the effi­ ciency of the recondensation effect. 3 The sample is given more time for evaporation and homogenization across the liner. A long distance is achieved by using short syringe nee­ dles. Short needles can be used if the split flow rate is high enough (50-100 mL/min) to ensure that the vapor does not expand backwards out of the liner. It is also achieved by us­ ing a long vaporizing chamber - which is no longer a vari­ able, however, once the instrument has been bought.

Table C5 lists experimentally determined effects of sol­ vent recondensation on the split ratio in terms of fac­ tors by which the peak areas obtained with a column tem­ perature of 30°C were larger than those obtained at 200°C (column temperatures well below and well above the boilTable C5 Effect of solvent recondensation on the split ratio. Depend­ ence on the internal diameter of the vaporizing chamber and on the distance between the needle tip and the column en­ trance. Factors by which peak areas Obtained at an oven temperature well below the solvent boiling point exceeded those at an oven temperature well above it. (From ref. [40]).

Vaporizing chamber i.d. [mm]

Recondensation Effects Needle-column distance 24mm 58mm

3.6

14 1.5

5.5

2

2

1.1 1.5 1.3

8.4. Minimizing the Deviation from the Pre-Set Split Ratio

203

ing point ofthe solvent, n-octane, 125 DC). These results (cor­ rected for the lower column flow rate at high oven tempera­ tures) do not consider the effect of the pressure wave - the true split ratio might have deviated from that pre-set by a factor larger than listed. 1 III (plus ca. 1 III eluted from the needle) of a 200 ppm solution of n-pentadecane in n-octane was injected with a pre-set split ratio of 100:1. Syringes with needles 71 or 37 mm long were used, leaving 24 and 58 rnrn, respectively, between the needle exit and the column entrance. The injector was used with liners of 2 or 3.6 mm i.d., or without a liner (5.5 mm i.d. cavity, possible with the injector design of the Carlo Erbal Fisons GC Mod. 4160). Conclusions

2

Injections with the long syringe needle result in a clear reduction of the recondensation effect when the narrow bore liner was replaced by the 3.6 mm i.d. tube. Further enlargement to 5.5 mm i.d, had little effect. Reduction of the recondensation effect upon injection with a short needle is substantial.

8.4.3. Small Sample Volumes

Small sample volumes reduce the effects on the split ratio. They facilitate evaporation, keep the pressure wave low, and weaken recondensation by stronger dilution with carrier gas.

Dependence of the Recondensation Effect on Sample Volume

Table C61ists absolute peak areas obtained by injecting dif­ ferent volumes of a 200 ppm solution of n-pentadecane in n­ octane. At a column temperature of 30 DC, the pre-set split ratio was 100:1. Use of a 2 mm Ld.liner and a 71 mm syringe needle accentuated the recondensation effect. Table C6 Dependence of peak area [integrator area counts x 10.3] on sample volume and column temperature during injection.

Temperature [OCI

30 60 100 140 200

Sample volume

1 III

2 III

3 III

35 27 21 15 11

510 360 180 26 19

640 490 330 48 28

200°C Column Temperature

Peak areas at a column temperature of 200 DC were approxi­ mately proportional to the volume of sample injected. Accurate determination of the sample volume was a prob­ lem: the volume expelled from the needle had to be added to that read on the barrel of the syringe.

1J1L Injection

At 30 DC, injection of 1 III resulted in a peak area ca. three times larger than at 200 DC. Nearly half of this difference arises

204

C B. Problems Concerning the Split Ratio

from the reduction in the column flow rate at the higher tem­ perature, i.e. from the increase in the split ratio. The other part of the deviation was probably a result of recondensation.

2 ul: Injection

At 30°C, doubling the injection volume resulted in a peak area increased by a factor of almost 15. The area was about 15 times too large in comparison with that ob­ tained after injection of 1 III at 200°C (bearing in mind the changing split ratio). The large increase in peak area was observed when the column temperature was reduced from 140 to 100 °C (to 25° below the solvent boiling point). The recondensation effect became still stronger, however, when the temperature was reduced from 60 to 30°C, because the lower vapor pressure resulted in more complete reconden­ sation. The drastic recondensation effect observed with 2 III injec­ tions shows the importance ofthe concentration of vapor in the carrier gas: the more complete displacement of the carrier gas by the larger vapor cloud produced a gaseous phase which recondensed (disappeared) almost completely in the cool column inlet.

Larger Injection Volumes

Peak areas obtained with the column at 30°C were far from proportional to sample volume. After the dramatic increase from 1 to 2 Ill, areas increased by merely 35 % when the sample volume was doubled again to 4 ul., This is not sur­ prising, considering that the true split ratio obtained with the 2 I!L injection was less than 7: 1 instead of 100: 1 and the split ratio was no longer determined by the gas flow rates adjusted before injecting.

Poor Reproducibility

Absolute peak areas strongly influenced by the reconden­ sation effect tend to be poorly reproducible, because the dis­ tribution of sample in the injector, i.e. the concentration of vapor in the gas entering the column. is not under control. Small errors in reading the sample volume might also be amplified and have a disproportionately strong ef­ fect. If, for instance, 1.6 III is injected instead of 1.5 III (a difference of 6 %), the peak area could be more than dou­ bled, owing to reinforced recondensation.

8.4.4. Volatile Solvents

For analyses starting at low column temperatures, a volatile solvent should be used. Pentane is, for instance, preferable to hexane if injection is performed at a column temperature near ambient.

8.4.5. Packed Liner

If oven temperatures are sufficiently high to rule out recon­ densation in the column inlet, large variations in absolute peak areas are an indication of incomplete sample evapora­ tion. Methods for improving sample evaporation were dis­ cussed in Section B. Use of a deactivated glass wool packing was one of these.

II !

8.8.4. Minimizing the Deviation from the Pre-Set Split Ratio

205

Retention of Solutes until the Split Ratio is Restored?

Packings could also be useful in another respect. They might retain the solute material above the column entrance until the adjusted split ratio is restored, i.e. keep them out of the turbulent processes during solvent evaporation. Deposition on a packing presupposes release of the sample liquid from the needle as a band (no thermospray). It cools a small area of the packing to the solvent boiling point. The temperature returns to that of the injector only after sol­ vent evaporation is complete, which easily takes a few sec­ onds. High-boiling solutes will be vaporized only then. In the meantime the pressure wave has been dissipated and car­ rier gas free from solvent vapor enters the column and the split outlet, ruling out recondensation effects. No data are available on this approach. Success might de­ pend on the volatility of the sample material. Severe dis­ crimination could occur if the sample contains solutes which are volatile enough to follow the solvent vapor and others evaporating with significant delay. The volatile solutes will then be split by the deviating ratio whereas for the high-boil­ ing components the split ratio could be close to that adjusted.

8.5. Experimental Re­ sults

Kaufman and Polymeropoulos [41] investigated the pressure wave and its dependence on several factors. Because a Hewlett-Packard 5890 instrument with a HP 7673A fast auto­ sampler and flowlbackpressure regulation was used, the pressure wave was rather weak and normal pressure re-es­ tablished rapidly - although at the expense of an accentu­ ated increase in the split ratio. Pressure measurements were performed by means of a pres­ sure transducer with a response time of less than 1 rns, in­ stalled in the septum purge line 15 cm from the injection port. The adjusted split flow rate was usually 100 mLlmin.

8.5. 1. Results Concern­ ing Pressure Wave

Sample Volume, Injector Temperature

The plot on the left in Figure C19 shows the pressure waves measured for injections of 1-5 III hexane (injector with a

76

0.2

0.4

0.6 TillE, •

0.8

1.0

1.2

1.4 00

0.2

0.4

0.6 TIllE, •

0.8

1.0

1.2

Figure C19 Pressure waves generated by split injection in dependence of the sample volume (left) and the injector temperature (right). 3 III injections. (From Kaufman and Polymeropoulos [41]).

206

C 8. Problems Concerning the Split Ratio

4 mm i.d. liner at 200°C). Waves clearly increased with larger sample volumes. For the 1 III injection, the pressure increase was less than 2 %, for 5 III it reached ca. 10 % ofthe inlet pressure. Waves initiated by 3 III injections of hexane at injector tem­ peratures of 100-400 "C are shown at the right. Increasing the injector temperature caused the waves to become sharper and higher, because evaporation was faster.

Duration of Evaporation

Ifthe pressure increase can be interpreted as the consequence of expanding solvent vapor, the curves suggest that solvent evaporation took ca. 0.1 s at an injector temperature of 400°C and ca. 0.2 s at 200°C. This is short compared with the visual observations described in Section B.

Sample Solvent

Isooctane generated a weaker pressure wave than hexane, which is explained by the slower evaporation. Sol­ vents which evaporate with difficulty or form a vapor cloud larger than that from hexane were not tested. Methanol, which produces three times more vapor per volume of liq­ uid than hexane, is expected to produce a higher wave; wa­ ter, with six times more vapor, would be extreme.

Split Flow Rate

The plot on the left in Figure C20 shows the effect of the split flow rate on the pressure wave initiated by a 1 III injec­ tion of hexane (injector temperature, 200°C; 4 mm i.d. liner). As expected from the discharge of the vapor, the highest flow rate generated the sharpest wave (some 50 ms width at half height); the pressure increase persisted for ca. 200 ms when the split flow rate was only 15 mLJmin. The pressure decrease below the value regulated is probably a result of delayed closure of the regulator valve.

Injector Liner

As expected, the narrow-bore liner resulted in a stronger pres­ sure wave than did the wider bore inlet. Results from the

71

r - - - - - - - - - - - - - - - - - - - - , .---.,------:-..,..-------------,

.. SOOmUmin

'.

'.

:\2 mmi.d.

:" / mm i. d., Cup '. 4 mmi.d. ldw-"".J... / ".,••:::..:::::.:•....;;'

......

~

87 +-_---.-~-r~~_r_----.-~-r~-.-----l.t-~__,_-~_,_--_r_~--,--~-,-~-_,J 1.0 0.8 0.8 1.2 0.4 1.0 1.2 0.0 0.2 0.8 0.8 0.2 0.4 0.0 TIllE, • TIME••

Figure C20 Pressure waves generated by 1 III injections of hexane at an injector temperature of 200 °C. left, effect of split flow rate; right, effect of liner. (From Kaufman and Polymeropou/oB [41]).

8.5. Experimental Results

207

packed liner were not significantly different from those from the open tubular liner of the same diameter (4 mm i.d.).

8.5.2. Cou,se of the P,essu,e Wave

Kaufman and Polymeropoulos used their data to model the course of the pressure wave. The data refer to a fast auto­ sampler 1 III injection of a solution in hexane into a 4 mm l.d, liner at 200°C (conditions causing a relatively weak pres­ sure wave). The pre-set split ratio was 100:1.

Fluctuation of the Split Ratio

As shown in Figure C21, the flow rate leaving the split out­ let increased far more rapidly than the column flow rate, with the effect that the split ratio rose by a factor exceeding five. After 100 rns, it dropped to half the pre-set split ratio and resumed the normal level 200 ms after injection. Thus total variation exceeded a factor of ten. 600

500

0

400

i= c II:

!::

.... Q.

Split ratio set at 100: I

300.

III

200

100

V

a 0.0

0.2

0.4

0.6 TIME,s

0.8

1.0

1.2

Figure C21 Predicted split ratio after a 1 III injection of hexane using a flow/backpressure regulator system. (From Kaufman and Po'ymeropou'oB [41]).

Effect on the True Split Ratio

The authors concluded that the error in the average split ratio was merely 6 %. They assumed that the 1 ml vapor­ izing chamber was homogeneously filled with sample vapor and that splitting took 1 s (the time to replace the volume of the chamber twice). The data can be interpreted differently, however. 1 III hexane produces about 150 III of vapor. If this is diluted 1:1 with carrier gas, a 300 III volume is split. At a split flow rate of 150 mLJmin (split ratio of 100:1 at a column flow rate of 1.5 mLJmin), splitting takes 0.12 s, which corresponds to the period of worst deviation and is plausible because sample evaporation and discharge of the vapor is the cause of the deviation. Then the deviation exceeds a factor of two.

Delayed Splitting of Sample

The effect of the fluctuating split flow rate on quantitative analysis is influenced by the moment when the solutes are split. Passage of the solvent and the most volatile solutes by

208

C 8. Problems Concerning the Split Ratio the column entrance is delayed by the gas volume between the vapor cloud formed by sample evaporation and the column entrance. If a 51 mm syringe needle releases the sample into packing placed at the needle tip (Figure C22), this volume is about 440 III (4 mm i.d. liner). For the 1 III hexane injection, this might be just about enough to delay the arrival of the first vapor at the column entrance until the main wave has been dissipated.

Height of septum cap

Depth of the packing

Length of syringe needle

Figure C22

Position of the center of sample evaporation within the va­

porizing chamber.

If 2 III of a solution in methanol are injected and the 1.5 ml of vapor are diluted to a cloud of ca. 3 rnl., this buffer volume is too small and the delay is no longer significant. As the wave of high split flow rates is then far broader, almost all vapor from the solvent and co-evaporated solutes will be split during the wave. It is another question whether the sol­ utes reach the split point with an additional delay.

8.5.3. Data on True Split Ratios

Bannon et al. [421 claimed that large deviations of the true split ratio from that pre-set were a result of "poorly designed injector inserts, either in the form of empty tubes or tubes packed with only a small amount of glass wool" .In fact, when fatty acid methyl esters in isooctane were injected at more or less isokinetic split flow rates (Section C9.2.1, split ratios of 180:1 to 470:1), true split ratios seldom deviated by more than 10 Ofo from the pre-set value. Experiments were performed on a Hewlett-Packard instru­ ment with backpressure regulation. Liners were either packed with glass wool or contained a "Jennings cup" and a pack­ ing both of glass wool and of Gas Chrom Q coated with 10 % SE-30. The column temperature was 160 "C, which ruled out recondensation effects.

8.6. Working Rules to Prevent Systematic Errors

209

8.6. Working Rules to Prevent Systematic Errors

The measures mentioned above help avoid drastic deviations from the true split ratio, but do not eliminate them completely. There are two ways of avoiding systematic errors: the true split ratio is frequently checked, necessitating comparison with peak areas obtained by a non-splitting method, or the quantitation procedure is selected such that it is im­ mune to deviations in the split ratio. Because the latter is the more convenient approach, it is usually the way to proceed.

8.6.1. No Quantitation on the Basis of the Pre­ Set Split Ratio

The first rule: do not calculate amounts or concentrations on the basis of the (pre-set) split ratio; assume that we cannot tell how much solute material enters the column. If we need to know the absolute amount, and this happens rarely, a non-splitting method must be applied, preferably on-column injection.

Hypothetical Story

This rule is illustrated by a story, which has certainly never happened. It is common practice to demonstrate the sensi­ tivity ,of a GC-MS system by split injection of a solution of methyl stearate. The salesman may propose the following procedure: to introduce 1 ng of the compound, 1 III of a 50 ng/ III solution is injected at a (pre-set) split ratio of 50:1. In this way, 1 ng of methyl stearate should reach the ion source. If he injects this sample as a solution in octane (justifying the solvent in terms of its preventing evaporation from the vial) and uses a low column temperature during injection, the true split ratio might well be 5:1. Thus 10 ng methyl stearate en­ ter the system, and the resulting impressive peak should ef­ ficiently promote the sale. Apart from the problem concerning the split ratio, how was the 1 III volume determined? Did it include the sample ma­ terial inside the syringe needle completely, partly, or not at all? This error cannot, however, greatly exceed a factor of two.

8.6.2. Use of the Intemal Standard Method

The internal standard method is independent of the split ratio, because the deviation from the true split ratio is equal for the internal standard and for the solutes of inter­ est. Nor is reproducibility of absolute peak areas important. A known amount of a compound not present in significant quantity in the sample is added as an internal standard. The concentrations or amounts of the components of interest are calculated from the ratio of the peak areas, possibly taking into account response/correction factors.

Method of Standard Addition

Analysis by the method of standard addition also becomes independent of the split ratio when another peak in the chromatogram is used as a reference (as a kind of inter­ nal standard). In the first run, the area ratio of the compo­ nent of interest and the reference compound is determined. A known quantity of the component of interest is then added

210

C 8. Problems Concerning the Split Ratio to the sample and the analysis is repeated. The change in the area ratio corresponds to the known amount added, whence the amount present in the original sample can be calculated. This method does not require knowledge of the identity of the reference component, nor its concentration in the sample.

8.6.3. Apply the Extemal Standard Method with Caution

The external standard method involves comparison of abso­ lute peak areas from a calibration run and the analysis. A solution containing known amounts or concentrations ofthe compounds of interest is analyzed to calibrate absolute peak areas. Then the sample is analyzed and the resulting peak areas are compared.

Avoid Whenever Possible

The external standard method is widely applied for quanti­ tation with packed column GC or LC. In capillary GC, how­ ever, it tends to be avoided, particularly if split injection is involved, because it can easily lead to large systematic errors.

Selective Detectors

The external standard method is difficult to circumvent if a highly selective detector is used since the choice of com­ pounds available as internal standards is limited. In other cases, there is simply no empty space in the chroma" togram, into which an internal standard peak would fit.

Constant Deviation of the True Split Ratio

The external standard method does not require that the true split ratio be equal to that pre-set, but the true split ratio obtained on injecting the sample must be equal to that obtained for the calibration mixture; if a series of samples is analyzed, the true split ratio must be equal for them all. In other words, the deviation of the split ratio must be constant. This implies that all factors affecting the split ratio must be reproduced.

Sample Volume

The sample volume must be kept constant because it influences the pressure wave, the completeness of sample evaporation, and (via the concentration in the carrier gas) the recondensation effect in the column inlet. Incomplete elution of sample material from the syringe needle usually, however, renders accurate increase of the sample volume impossible anyway. If, for instance, a sample is so dilute that a well integratable peak cannot be obtained, the problem must not be solved by increasing the injection volume of the sample alone. The amount of material entering the column often increases out of proportion. Calibration must also be repeated.

Split Ratio

The column and the split flow rate must remain constant, because they influence sample evaporation, the pressure wave, and the recondensation effect in many ways. The ef­ fects of the pressure wave and of recondensation are usu­ ally stronger at lower split flow rates.

8.6. Working Rules to Prevent Systematic Errors Sample Matrix

211

The most common systematic error arises from dif­ ferences between the matrices of the calibration mix­ ture and the sample. By sample matrix we understand those parts of the sample which influence the evaporation characteristics (including the amount of evaporation and the pressure wave), the recondensation effect, and possibly other mechanisms affecting the split ratio. For dilute samples, this is primarily the solvent; for others, the split ratio is influ­ enced by elevated concentrations of sample impurities. For undiluted samples, the characteristics of the whole sample are important.

Examples

2

3

Contaminants, e.g. tetrachloroethylene, in drinking water can be analyzed as extracts in pentane. Suit­ able initial column temperatures vary between 25 and 45°C. With the intention of avoiding rapid evaporation of the solvent, the external standard mixture might be prepared in hexane or even heptane. Because such an external standard mixture is likely to recondense whereas the water extract does not, the concentrations determined will be excessively low. For the determination of benzene in gasoline, gaso­ line is often injected at a high split ratio to avoid the need for dilution. Because it is difficult to find an inter­ nal standard fitting into such a complex mixture, the analyst might prefer the external standard method. To facilitate the separation of benzene from the solvent peak, he might prepare the solution in pentane. This will lead to excessively high benzene concentrations, be­ cause gasoline is likely to recondense, whereas pentane does not. No solvent imitates gasoline except gasoline itself and, hence, there is no reliable way of quantitation by this approach. The analysis could be performed by stand­ ard addition. If the benzene concentration is expected to be around 5 %, 10 % benzene is added and the sam­ ple is re-analyzed, possibly by use of a reference peak as described for the internal standard method. "rhioglycollic acid was to be determined in a hair perm preparation. Direct injection with glass wool in the liner produced a reasonable peak. To circumvent the determination of a response factor, the external stand­ ard method was chosen. The composition of the sam­ ple matrix was unknown. Water as solvent for the exter­ nal standard resulted in a thioglycollic acid concentra­ tion of 12 %; when a solution in ethanol was used, the peak of the external standard was larger and only 7 % was calculated. Which result is correct? The problem was solved by diluting the sample 1:10 with ethanol, such that the sample matrix was domi­ nated by a known solvent. Now the concentration meas­ ured Was near 5 %, which happened to be close to the correct result.

212

C 8. Problems Concerning the Split Ratio

A Case History: Tetrachlo­ roethylene in Waste Water

Tetrachloroethylene was to be determined in a waste water from a textile printers. From the strong smell, the tetrachlo­ roethylene concentration was estimated to be ca. 100 mg/l, which can easily be analyzed by split injection of the aque­ ous sample and use of FID. The injector temperature was 270°C; a 3.6 mm i.d. liner and a 71 mm syringe needle were used. The pre-set split ratio was 50:1. Because no water-soluble internal standard could be found on the shelf, it was decided to apply the external standard method. It turned out to be impossible, however, to prepare tetrachloroethylene solutions in pure water at concentrations in excess of ca. 20 mg/l (addition of some detergent would have solved this problem), and so the external standard solution was prepared in hexane. The following results were obtained (integrator area counts): Waste water External standard (100 mg/l in hexane)

12,300 7,700

Concentration in waste water

160 mg/l

How accurate was this result? Special care was required, be­ cause it was likely that the results would be used in a court of law. Normal procedure: determination of relative stand­ ard deviations. Waste water, RSD Standard solution, RSD

8%

4%

From this it was concluded that the results should be accu­

rate to within some 10 %. Are standard deviations really suit­

able for assessing accuracy? Certainly not!

To confirm the above results, 50 ml waste water was ex­

tracted with 50 ml hexane, resulting in a virtually equal con­

centration in the extract as originally in the sample. This ex­

tract was injected under the same conditions.

Hexane extract area counts

2,600

Concentration in waste water

34 mg/l

As this result was way beyond the range of the standard deviation, a third experiment was performed to determine which of the two results was correct. 0.5 III of a 3 mg/l solu­ tion of tetrachloroethylene in water was injected on-column (on to a column coated with Emulphor, a polyglycol type sta­ tionary phase), then the same volume of the 1:10 diluted sam­ ple. The resulting concentration was 37 mg/l, confirming that the result was ca. 35 mg/l, not 160 ± 16 mg/L. The above error was primarily a result of extensive recon­ densation in the column inlet. As the column temperature was ambient, both water and hexane recondensed, with the

8.6. Working Rules to Prevent Systematic Errors

213

benefit that large peaks were observed because the true split ratio was well below 50:1. Because recondensation of water was much greater than that of hexane (different sizes of vapor clouds), the split ratio was ca. five times smaller. Maybe the true split ratio for the standard solution in hexane was 15:1, whereas that for the sample was near 3:1, i.e. a third of the sample entered the column despite the fairly high pre-set split ratio.

9. Problems Concerning Linearity of Splitting 9.1. "Linear" Splitting

"Linear" splitting, a term introduced by Ettre and Averill [7) in 1961, implies that all sample components are split by the same ratio. This also means that the small portion of material entering the column has the same composition as the sample in the injector. Linearity of splitting does not presuppose that the true split ratio is equal to the pre-set split ratio - such deviations are hardly ever noticed anyway.

Discrimination

The other important expression frequently used in this con­ text is "discrimination". When related to sample splitting, "discriminative" is synonymous with "non-linear". "Discriminative" is, however, more broadly used: several other mechanisms are discriminative, e.g. evaporation in the syringe needle, adsorption in the injector or the column, and unstable detector sensitivity [43). The part of the sample "discriminated" against produces peaks which are too small compared with the part of the sample represented by the larger peaks (understood, of course, as peak areas relative to the amounts present in the sample). If discrimination arises as a result of non-linear splitting, the part of the sample discriminated against is split by a higher ratio than is the rest of the sample. This leaves open whether the small peaks result from losses (i.e. whether they are really too small) or whether too much of the other sample material entered the column.

Discrimination Caused by Evaporation in the Syringe Needle

Non-linearity of splitting has been discussed since the early sixties. Until ca. 1980, the importance of discrimination caused by selective losses in the syringe needle had not been recognized; losses of high-boiling material in the nee­ dle were wrongly attributed to an increased split ra­ tio for these components. The commonly observed reduction in peak sizes towards el­ evated elution temperatures, considered to be the major prob­ lem arising from non-linear splitting, is mostly a result of the correspondinq material not even having entered the injec­

214

C 9. Problems Concerning Linearity of Splitting tor. To find out whether or not non-linear splitting contrib­ utes to such discrimination the material left in the syringe needle must be re-injected ("needle rinse injection", Section A5.2).

Non-Linearity: An Unre­ solved Problem

The problem of non-linear splitting is complex. There is no generally linear splitting injector, nor are there gener­ ally applicable concepts or working rules describing how to ensure linear splitting. Below we describe the causes of non­ linearity, but it must be left to the analyst to find the most appropriate solution to his problem, maybe on the basis of the strategies discussed in SeetionC10.

9.2. First Cause of Non­ Linear Splitting: Diffusion Speeds

The mechanisms of three causes of non-linear splitting will be described below. All were conceived in the minds of ana­ lysts trying to explain puzzling results. The experimental support is usually weak because of interferences from other deviations and limited opportunities for direct obser­ vation. In 1967, Bruderreck et al. [81 studied the influence ofthe dif­ ferent rates of diffusion of the sample components on the linearity of splitting. Atthe column entrance, vapor is forced to change direction, because the linear velocity of the car­ rier gas entering the column usually differs from that pass­ ing it by (Figure C23).

% Deviating vapor

lUll L

Disturbed Flow by column end face

Split flow )

Capillary column Column Flow

Figure C23

At the column entrance. sample vapor is diverted: molecules

of different sizes differ in mobility and will deviate by differ­

ent amounts.

Traffic Jam

The situation at the column entrance can be compared with a traffic jam. If the velocity of the gas phase entering the column is smaller than that passing it by, the column acts as a hindrance and solute molecules must change direction. Small molecules, with a higher speed of diffusion, are more easily diverted than heavy molecules, or (to stay with the picture of the traffic jam) the large molecules make their way

9.2. First Cause of Non-Linear Splitting: Diffusion Speeds

215

into the column pushing the small ones aside, as do trucks in a stream of cars and bicycles. The system resembles the jet separator of a GC-MS interface for packed column GC. Hence, if the column acts as a hindrance, the high­ molecular-weight solutes are enriched in the material analyzed and the volatiles are discriminated against. Bruderreck et al. [81 proposed a splitting system which would compensate for at least some of the differences in diffusion speed. Although excellent data were reported, the system never became popular, and this aspect of discrimination seems to have been forgotten. 9.2.1. Isokinetic Splitting

In 1982, Purcell [321 listed (among other prerequisites for lin­ ear splitting) the "isokinetic behavior" ofthe gas/vapor mix­ ture at the column entrance. Isokinetic splitting means that the linear velocities of the gas and vapor enter­ ing the column and passing it by are equal. No further explanation was given, and this was probably derived from the work cited above (although it was not cited).

Sacrifice of Flexibility

Isokinetic splitting requires sacrifice of an important advan­ tage of split injection: the flexibility of adjusting the split ra­ tio to the concentration of the components in the sample. The split flow rate is now determined by the gas velocity into the column. Table C7 lists the split flow rates providing isokinetic split­ ting at different column flow rates. They depend on the col­ umn and liner diameters. An equation for calculating the split flow rate from the dead time of the column and the column inlet pressure was given by Bannon et al. [441. Results show that the split flow rates required are rather high. Table C7 Isoklnetic split flow rates for different column diameters, column flow rates, and liner diameters.

Liner diameter

lrnrnl

Column flow rate lrnt/rninl

[rnm]

Isokinetic split flow rate Imt/rninl

0.20

1

0.32

2

2 4 2 4 2 4

100 400 77 310 193 770

Column diameter

5

Liners Adjusted to the Split Ratio

Isokinetic splitting at low split ratios presupposes narrow­ bore liners. This contradicts another requirement: for low split ratios, a large liner volume is needed in order to retain the sample vapor in the vaporizing chamber and to keep devia­ tions of the split ratio as low as possible.

216

C 9. Problems Concerning Linearity of Splitting This dilemma could be avoided by the use of liners with constrictions in the split area (Figure C24), enabling in­ stallation of the column entrance at the position where the ratio ofthe cross sections ofthe liner and column correspond to the split ratio required.

-

Room to store vapor

Isokinetic splitting with column entrance at suitable height

Column

Split outlet

Figure C24 Liner enabling isokinetic splitting to be achieved at low split flow rates.

9.2.2. Insufficient Experi­ mental Evidence

If taken seriously, isokinetic splitting would have far reach­ ing consequences on split injection. Experimental evi­ dence supporting the need for it is, however, not convincing. Purcell did not publish experimental results. Bannon et al. found results to be optimum when the split flow rate "was at or not too far away" from that appropriate for isokinetic sam­ pling (deviations by a factor of less than two). They did not, however, stress it as an important point, nor is their evidence overwhelming. Marshall and Crowe (66) obtained linear split­ ting at a split flow rate half that required by isokinetic split­ ting, but severe discrimination was not demonstrated when deviations were larger.

End Face of the Column

If the disturbance of the flow at the column entrance were of such importance, the effect of the column end face would have to be considered. Seen in the direction of the gas flow, the face of a fused silica capillary column end consists of a ring with a surface area slightly exceeding that ofthe bore of the capillary. When glass capillaries were used (and most of the tests on split injection were performed with glass capil­ laries), the area of the face of the column end was more than 6 times larger than the cross-sectional area of the column bore. At the face of the column end, the gas velocity is zero and deviation ofthe vapor is complete. Because the column face is circular, the deviations of vapor inwards into the column bore and outwards towards the split outlet are different, which should greatly disturb the linearity of the splitting proc­ ess.

9.2. First Cause of Non-Linear Splitting: Diffusion Speeds

217

9.2.3. Conclusion

From the experience that linear splitting is possible when the gas velocities into and by the column entrance differ by a factor of 10, at least, it seems that the different diffusion speeds of the sample components are not an important source of non-linear splitting; this is particularly so in comparison with the two other mechanisms to be described below. Unless there is convincing evidence ofthe need for isokinetic splitting, the ease of use and flexibility of adjusting the split ratio should not be sacrificed.

9.3. Second Cause: Incomplete Sample Evaporation

Sample evaporation in empty liners or in liners containing obstacles (baffles, "inverted cup", or cup liner) is often in­ complete, particularly if a fast autosampler is used or the sample is in a high-boiling matrix.

9.3. 1. Vapors and Drop­ lets Split at Different Ratios 2 3

A jet of liquid or relatively large droplets is "shot" to­ wards the column and split at a ratio more or less deter­ mined by whether or not it hits the column en­ trance, not by the adjusted gas flow rates (Figure C25). Vapor and small droplets are transported by the carrier gas and split by the ratio of the gas flow rates. When droplets and vapor are split at different ratios, components primarily located in the droplets will be split by a ratio different from that for the gas phase. Syringe needle

:":. r; '

Sample vapor trans­

ported by carrier gas,

split by the ratio of gas flow rates

liquid sample "shot" towards the column, split at a ratio determined by whether or not it hits the column entrance

..-;.•. ..:,~~

t:l~,"m" entrance

Figure C25

Splitting of incompletely evaporated sample.

9.3.2. Neat Samples

The problem of non-linear splitting is particularly severe

for undiluted samples containing components with a wide

range of boiling points, such as mineral oil fractions, typi­

cally ranging from C4 to C40 alkanes.

Immediately after leaving the syringe needle, the most vola­

tile components start evaporating whereas the higher­

boiling components remain in the non-evaporated liquid. The

218

C 9. Problems Concerning Linearity of Splitting

temperature of the liquid (its boiling point) rises and further solute material evaporates, shifting the limit between the vaporized components and those still left in the liquid phase towards higher-boiling points. When such partially evapo­ rated sample material reaches the column entrance, the volatiles in the gas phase are likely to be split by a ratio dif­ ferent from that experienced by the high-boilers in the liq­ uid. Chaotic Discrimination Pattern

At first sight, results from such mixtures often seem to be chaotic. A range of components is partly lost (discrimina­ tion), but neither the range of the components discriminated against nor the extent of the discrimination is reproducible. With a second effort, it can at least be explained why results turn out like this.

Discrimination against Volatiles or High Boilers?

Incomplete sample evaporation may cause discrimination against either the high-boiling or volatile components. If the liquid is split at a higher ratio than the vapor, the high­ boilers are discriminated against compared with the volatiles. The reverse situation is possible, but more sel­ dom observed. If the split ratio determined by the gas flow rates is low (a large proportion of the vapor enters the column), the prob­ ability of non-evaporated material "falling" into the column is generally lower, i.e. the solutes in the liquid phase are split by a higher ratio and will be discriminated against. If, on the other hand, the gas phase is split at a high ratio or a large proportion of the liquid happens to reach the column, this is likely to discriminate against the volatile components. At the intermediate split ratios most commonly used, it seems that discrimination against the high-boiling compo­ nents is more frequent. The probability of the liquid hit­ ting the column is small, as is confirmed by the finding that the true split ratio for incompletely evaporated samples is usually higher than the pre-set split ratio (see Section C8.3.4).

Discrepancy between Split Ratios

The resulting discrimination depends on the split ratios for the vapor and the liquid. The first is fairly reproducible whereas the second is usually not. The discrepancy between the two split ratios determines the extent of the discrimination. The reproducibility of splitting the liquid phase deter­ mines the reproducibility of the deviation.

Extent of Sample Evapora­ tion

Discrimination affects solutes up to or beyond a certain limit. Assuming that the liquid is split at a higher ratio than the vapor, the higher-boiling compounds are discriminated against, i.e. the chromatogram shows that for the compo­ nents eluted above a given elution temperature the peaks are too small. This limit (elution temperature) is a function of the extent of solute evaporation when the sample reaches

9.3. Second Cause: Incomplete Sample Evaporation

219

the split point. When evaporation is more advanced. higher-boiling components are vaporized and split by the ratio of the gas flow rates. Poor Reproducibility

When sample evaporation is incomplete, the reproducibility of absolute and relative peak areas is usually poor. This is obvious when the described mechanisms are considered. The split ratio for the liquid is poorly reproduc­ ible, because the liquid is unevenly distributed across the liner, whereas the split ratio for the vapor is fairly constant. This accords with the common experience that chromatograms furnish fairly constant peak areas for the volatile compounds, but more widely varying areas for the high boilers (although there are also other rea­ sons for this). The proportion of the sample which vaporizes is not reproducible. With a "shot" straight down the liner, a smaller amount of sample material evaporates than if at least part of the sample is nebulized. Hence the limit beyond which the sample is primarily split in the liquid phase is not reproducible.

9.3.3. Dilute Solutions in Solvents

Incomplete evaporation is, however, not a required condi­ tion for non-linear splitting. It causes non-linearity only if the components of interest are distributed differently in the vapor and droplet phases. In particular, splitting remains linear if the sample material is completely retained by the droplets - and this is a situation frequently encountered for samples diluted in solvent.

Negligible Solute Evapora­ tion during Solvent Evapora­ tion

The samples most frequently analyzed are strongly diluted with a dominating solvent. The temperature of the droplets or jet of liquid is, hence, near the solvent boiling point. Such cool liquid retains all the higher-boiling components, i.e. those eluted after the solvent peak, the solvent acting like a gas chromatographic stationary phase (even though it moves). Thus the components of interest are likely to be lo­ cated in the liquid phase and are split linearly - although often at a poorly reproducible split ratio and possibly far from that pre-set.

Complete Evaporation of Parts

Partial evaporation of the sample components can resulot from 1 a substantial vapor pressure over the unevaporated sam­ ple liquid, i.e. partitioning between the droplet and the gas phase, or 2 complete evaporation of part of the liquid. For in­ stance, small droplets split away from the main jet of liquid and are likely to be fully evaporated earlier than the rest of the liquid. Whereas the first scenario inevitably results in different amounts of the solute material in the gas and droplet phases, the effects of the second scenario are less obvious.

220

C 9. Problems Concerning Linearity of Splitting Complete evaporation of some droplets releases the volatile

and high-boiling components into the gas phase, with the

effect that the composition of the solute material in the gas

and in the remaining liquid is the same; splitting should still

be linear. The released solute material, or maybe only the

high-boiling components, might, however, be picked up again

by the remaining droplets.

9.3.4. Conclusion

The effects of incomplete sample evaporation on quantita­

tive results are complex. The reasons for this were outlined

above. There is no need for a more detailed description, be­

cause it is preferable to cure diseases than to concentrate on

their diagnosis.

Improvement of sample evaporation was a subject of

Part B. Here it should be remembered that the injector tem­

perature is not the main factor influencing sample evapora­

tion. What is more important is whether the sample is dis­

solved in solvent, how it is injected (fast autosampler?), and

how much is injected (cooling effect).

9.4. Third Cause: Fluctu­ ating Split Ratio

Splitting also becomes non-linear if the following two situa­

tions occur in combination [451:

1 the split ratio changes (fluctuates) during splitting of

the sample; and 2 the sample is pre-separated before reaching the split point, causing different parts of the sample to reach the split point at different times and, hence, at different split ratios.

9.4.1. Variation of the Split Ratio

In the preceding section, the true split ratio was frequently found to deviate from that pre-set by factors up to ten. True split ratios considered there represented average effective split ratios, including, e.g., a small deviation at the begin­ ning of the splitting process, a large deviation in the middle, and again a small one at the end. The deviation of the true split ratio from that pre-set is not constant during the split­ ting process, and the maximum deviation from the pre­ set ratio is likely to substantially exceed that from the average effective split ratio. Figure C21 showed the calculated course of the split ratio during an injection initiat­ ing a weak wave. Nevertheless, the split ratio increased five­ fold.

A Hypothetical Course of Events

Figure C26 shows the possible course of changes in the split ratio during an injection, assuming a gas supply sys­ tem of the pressure regulator/flow resistance type, and that the pressure wave overlaps with recondensation in the column inlet. The pressure wave initiated by injection in­ creases the flow rate into the column. As the first vapor recondenses in the column inlet, further vapor is sucked into the column, farther reducing the split ratio. Dilution of the remaining vapor in the injector soon weakens the recon­

9.4. Third Cause: Fluctuating Split Ratio

221

densation effect and, finally, the decrease in the pressure in the injector reduces the flow rate into the column even be­ low normal.

-A­

Injection

1

Pressure wave ~

/

/

Recondensation Average true

- - - 1 , - - - - - - - - - 1 - - - - - split ratio

Split ratio

I

I 8,m", eo'ering column

~.Time

Figure C26 Changes in the split ratio during injection. assuming that first the pressure wave and then recondensation reduce the split ratio. The true split ratio. as manifested by the size of the peaks. corresponds to the effective average split ratio over the splitting period - and might well be different for different solutes. because a high-boiling component tends to be vaporized later or split over a longer period. 9.4.2. Pre-Separation of the Sample in the Injector

When all sample components are split at the same moment

(reach the column entrance simultaneously), a varying split

ratio causes all the peaks to be too large or too small, but the

sample composition (relative peak areas) is correct and split­

ting is linear.

A fluctuating split ratio affects the linearity of splitting only if

the sample is pre-separated before reaching the column en­

trance. For instance, one component is mainly split at a mo­

ment of high split ratio, whereas most of another is split when

the split ratio is low. Several processes result in fractiona­

tion of the sample inside the injector.

Transfer from the Syringe Needle

When sample material evaporates from the syringe needle,

transfer occurs in order of increasing boiling point. High­

boiling solutes may leave the needle when most of the

sample has left and its internal surface is warmed up again,

provided there is still solvent evaporating at the rearto carry

it out of the needle.

If commonly used volatile solvents are used, this delay can­

not be more than a few hundred milliseconds, because

rapid withdrawal of the syringe needle does not increase

losses of high-boiling compounds.

222

C 9. Problems Concerning Linearity of Splitting

Faster Droplets

Droplets containing the high-boiling sample material move faster than the vapor comprising the volatiles (50-200 km/h compared with 300 rn/h at a split flow rate of 60 mt/rnln). This enriches the high-boiling compounds at the front of the vapor cloud.

Fractionating Evaporation

The most important fractionation occurs during sample evaporation from surfaces, such as packing materials. The site is first cooled to the solvent boiling point and cooling continues until solvent evaporation is complete, ab­ sorbing heat from a larger region. Then temperature in­ creases again, releasing higher-boiling solutes in order of increasing boiling point. Solvent evaporation from a glass wool packing easily takes several seconds and it probably takes another few seconds to bring temperature back to that thermostatted.

Adsorption and Retention on Contaminants

Adsorption on active surfaces in the injector delays split­ ting of some ofthe solute material easily by several seconds or even a few minutes. Retention in a layer of non-evapo­ rated material deposited during previous injections might have the same effect.

Deviation from Linearity

The extent of distortion of the composition of the sam­ ple entering the column (the deviation from linearity) is de­ termined by the amplitude of the fluctuation of the split ratio during splitting and the extent to which the sample is frac­ tionated in the injector.

Risk Factors

It is difficult to define when to expect severe non-linearity of splitting from fluctuating split ratios. Maybe listing some fac­ tors contributing to strong deviation of the split ratio and fractionation of the sample is useful. Strong pressure wave: large sample volume; solvent producing large vapor cloud; narrow liner. Recondensation of the solvent (column at least 20° be­ low solvent boiling point) or of a non-diluted sample. Packed liners. Fast autosamplers. Adsorptive and high-boiling sample components. Samples ranging from components co-evaporating with the solvent to high boilers.

9.4.3. Cognac as an Example

Vanillin and syringealdehyde had to be measured in a Cognac to determine whether its yellowish color was really derived from storage in an oak barrel or from addition of caramel and flavor. Oak releases vanillin, but vanillin must be added when the color is derived from caramel. Syringeal­ de hyde has no influence on the flavor and originates from oak. Its concentration should reach a certain level and be similar to that of vanillin.

9.4. Third Cause: Fluctuating Split Ratio

223

Recondensation to Increase Sensitivity

Concentrations should be just below 1 mg/L. An attempt was made to analyze them by split injection of the sample itself. Because sensitivity was critical (FlO), the recondensation effect was exploited by use of a long syringe needle, a nar­ row liner, and a column temperature during injection of 50°C. The split flow rate was 30 mLJmin, providing a pre-set split ratio of ca. 12:1. Two reasonable peaks were, indeed, ob­ served, confirming that the proportion of the material enter­ ing the column far exceeded that to be expected from the pre-set split ratio.

Replacement of Liner Caused Peaks to Disappear

Reproducibility of relative and absolute peak areas was, however, unsatisfactory. Because adsorption was expected to occur in the liner, a new liner was installed. Now vanillin and syringealdehyde disappeared completely from the chromatogram, whereas most ofthe other peaks were unaf­ fected.

Adsorption on the Liner Delayed Splitting

Did the two compounds of interest remain in the injector and accumulate there? Of course not! Theyalso passed the split point, but with some delay. Adsorption on the liner surface

retained them during the period when recondensation was causing the large flow into the column. When the two components of interest were released, seconds later, the split ratio had returned to the much higher pre-set level. At this increased split ratio, peaks became undetectable in the rather complex mixture. Deactivation of the Liner

The problem was solved rather crudely: red wine was in­ jected four times at intervals of 2 min, keeping the split valve permanently closed (splitless injection), the septum purge widely open, and the column flow rate low (in order to in­ crease the residence time of the wine in the injector). This treatment deactivated the liner and brought the two peaks back.

Reduced Split Flow Rate

To obtain better reproducibility, the split flow rate was reduced to 5 mLJmin. This, however, resulted in a strongly tailing, broad "solvent" peak and a high, disturbed baseline, the latter mainly as a result of degradation of other material in the injector. It could be avoided by increasing the split flow rate to 50 mLJmin 30 s after injection, because the deg­ radation products generated later then entered the column only at a high split ratio.

9.5. Danger of System­ atic Errors

If discrimination is poorly reproducible, the results are im­ precise. This is a nuisance, but at least the problem is easily recognized. More dangerous are systematic errors, as they are difficult to detect. Of course, this warning once more relates to more "difficult" samples, but how does the ana­ lyst recognize "difficult" samples?

UNlVERSlPAD DE ANTIOQUlA

SlDUOTBCA CENTRAL

224

C 9. Problems Concerning Linearity of Splitting

The Problem of Calibration

Effects of discrimination are commonly compensated for by use of correction factors (often erroneously called "re­ sponse factors"). A mixture of the components of interest in pure solvent is injected and the correction factors thus de­ termined are applied to the sample. It is assumed that the correction required for the sample is the same as for the cali­ bration, i.e. that the mixture of standards imitates the devia­ tions affecting the analysis of the sample. But how should we reproduce mechanisms causing discrimination if we do not know them? All too easily the processes are differ­ ent and the correction factors turn out wrong.

Example

Evaporation of a clean mixture of standards is easier than that of a sample containing high-boiling or involatile material. These sample by-products inevitably form drop­ lets which retain sample components. Evaporation of the calibration mixture and the sample is, therefore, different. Hence discrimination is likely to be different and the correc­ tion factors do not really correct the deviation.

Sample Matrix more Impor­ tant than Solutes

During method development one usually concentrates on the solutes of interest; the solution of standards is prepared in a solvent selected rather carelessly. The process of split­ ting depends, however, at least as much on the solvent and the sample by-products as on the solutes.

Misleading Tests with Simple Standard Mixtures

The literature is full of papers on the feasibility of certain analyses and the accuracy obtained. Unfortunately all too often the test exclusively involves solutions of standards in solvent, although the real problems are generated by the by­ products. The low standard deviations determined can be misleading. Many authors seem to have found fewer prob­ lems than were reported by others, wrote a paper on this, and only after despatching the paper looked at real samples. Then the follow-up paper on the application of the method to samples, promised in the conclusion, usually never ap­ pears!

Tests by Standard Addition

Calibration and optimization of conditions, such as in­ jector temperature or packing of the injector, should be per­ formed with samples to which standards have been added. Requirements resemble those discussed for the calibration of absolute peaks areas by the external standard method (Section C8.6.3). Reproduction of discrimination phenomena tends to be even more difficult than that of split ratios, al­ though the errors hardly reach a factor of five.

10.1. Systematic Search for the Best Conditions

225

10. Techniques for Improving Quantitative Analysis Previous sections have discussed the basic conditions for quantitative analysis by split injection and the problems which can arise. They have demonstrated that there are no simple recipes ensuring precise and accurate results. This section summarizes the subject from the point of view of coherent strategies. Only liquid samples are considered, be­ cause it is the vaporization step which causes most of the problems. Pragmatic Approach

In practice, the experimental results are relevant, and ifthey are satisfactory, nobody would spend much effort in finding out why. Concepts become import8!nt only if results do not meet the needs or expectations. They help find­ ing the origin of the problem, and which of the many factors involved needs improvement.

10.1. Systematic Search for the Best Conditions

If results are unsatisfactory, there is hardly an alternative to the trial and error approach, varying conditions to see whether or not results improve. This, however, easily ends up as lengthy groping around in the dark. The search must be purposeful.

10.1.1. Strategy: Mini­ mized Deviation

Chromatographers tend to trust in an almost unlimited ef­ fectiveness of correction factors, believing that all the problems possibly occurring, such as non-linear splitting, loss inside the needle, adsorption on surfaces, or varying split ratios, can be offset by correction factors. This is not true. The deviations are often poorly reproduc­ ible and easily introduce systematic errors if they depend on factors which vary from one sample to another or from the calibration to the sample. If a result deviates from what it should be by, e.g., a factor of ten, even a modest variation of this deviation strongly affects precision.

The Three Steps

A more systematic approach involves the following three steps. 1 The "correct" result (peak area or area ratio) is esti­ mated to determine the extent of the deviation currently obtained. 2 Conditions are optimized in the direction which mini­ mizes the deviation. 3 Only the minimized deviation is corrected by cali­ brated factors.

226

C 70. Techniques for Improving Quantitative Analysis Reproducibility usually improves automatically when the dif­ ference between the actual and the correct result diminishes, and the danger of systematic errors is reduced.

10.1.2. Determination of the Correct Result Estimated Peak Area

Sometimes the correct result can be estimated from the peak area to be expected in comparison with another, more easily chromatographed component, such as a hydrocarbon of in­ termediate molecular weight. This presupposes that the de­ tector response is known or can be estimated. An estimate of the response is often possible for the flame ionization detector (FlO), because the response of the hy­ drocarbon part of the molecule tends to be constant. The proportion of heteroatoms and highly oxidized carbon in the molecular weight is estimated and considered as not con­ tributing to the FID response.

On-Column Injection

A test by on-column injection is often most indicative. The sample is diluted by a factor corresponding to the split ratio to furnish data on absolute peak areas (checking the split ratio) and on relative areas (linearity of splitting, discrimina­ tive losses in the needle, or adsorption). To obtain directly comparable peak areas, on-column injec­ tion must be performed on the same instrument (detector) as used for the analysis. Ideally, the instrument is equipped with an on-column injector, enabling a simple change from one injector to the other. Otherwise on-column injection must be performed without such an injector, i.e. directly into the column inlet dismounted from the split in­ jector for the duration of the injection. Such a procedure is described in Section D5.6.2.

Degraded Components, Artifacts

Analysis by on-column injection also enables discovery of components missing from the chromatogram, because of degradation in the injector, or of artifact peaks formed by degradation of other sample material. For instance, high concentrations of furfurol were found in a brandy. This was interpreted as a sign of excessively high temperatures during distillation ofthe beverage and the prod­ uct was rejected. It turned out, however, that most of this furfurol was not present in the sample, but was "synthesized" in the injector!

10.2. Flash Evaporation

The strategies presented below should assist the optimiza­ tion process. After a short description of the concept, the critical parameters are discussed, with the intention of sug­ gesting directions in which to experiment.

10.2.1. Concept

The classical means of obtaining linear splitting involves rapid ("flash") sample vaporization and mixing with car­ rier gas before the sample material reaches the split point. Fast evaporation and rapid passage through the injector are prerequisites for obtaining sharp initial 'bands.

10.2. Flash Evaporation

227

Thermospray

The observations made with the perylene experiment indi­

cate that flash evaporation presupposes nebulization of the

sample liquid at the needle exit.

It is impossible to achieve flash evaporation after injection

with band formation, e.g. by using a fast autosampler. Af­

ter deposition on to a packing or trapping between obsta­

cles, evaporation of the solvent alone easily takes several

seconds and leaves behind a cool zone which must return to

the injector temperature before the high-boiling solutes can

be vaporized. This does not correspond to flash evaporation.

Hot Needle Injection

Efficient and reliable nebulization at the needle exit requires

partial solvent evaporation inside the syringe needle. Opti­

mization ofthe syringe handling technique leaves little room

for variation. The best thermospray and the least discrimi­

nation as a result of losses inside the needle are obtained by

hot needle injection.

Several modern autosamplers are capable of withdrawing

the sample liquid from the needle and injecting after an ad­

justable needle preheating time, i.e. of performing hot nee­

dle injection.

10.2.2. Selection of Conditions

Some of the variables discussed below can be optimized from

basic considerations. Others need testing and experimental

adjustment for the particular application.

Empty liner

Flash evaporation occurs in the gas phase of the empty space

in the vaporizing chamber. Hence an empty liner serves the

purpose. This minimizes contacts with active surfaces and

prevents fractionation of solutes according to volatility, which

has been described above as a source of non-linear split­

ting.

Mixing Devices

It has been claimed that obstacles improving the homoge­

neous distribution of the vapor across the vaporizing cham­

ber prevent that a stream of concentrated vapor hits the col­

umn entrance during one injection whereas during the next

it largely bypasses it. They should improve the reproducibil­

ity of absolute and relative peak areas. Mixing could be es­

pecially important when wide bore liners are used. The cup

and laminar liners are the most promising (see Section C10.5).

Reported positive experience with mixing devices contrasts

with findings that there is no significant improvement. In fact,

violent nebulization and expansion of the vapor might well

form a plug filling the cross section of the liner and thus en­

suring homogeneous distribution without special effort.

Visual observation suggests this.

Injection of Diluted Samples

Nebulization presupposes a volatile sample matrix that

serves as a propellant for the thermospray. This agrees well

with the experience that dilute solutions tend to provide bet­

ter results. Dilution also ensures that the droplets remaining

228

C 10. Techniques for Improving Quantitative Analysis after vaporization of the volatiles are smaller and more uni­

formly dispersed in the gas phase.

This means that samples should be diluted as much as pos­

sible. To achieve the required sensitivity, injection of more

dilute solutions means that the split ratios must be re­

duced. Indeed, experience seems to confirm that it is pref­

erable to inject, e.g., a 10 times more dilute sample at a 10

times lower split ratio.

Readily Evaporating Solvent

In general, the solvent has a greater influence on the

evaporation process than the components to be ana­

lyzed. Solvents should not be selected merely on the basis

of solubility. It is often advantageous to dilute a sample dis­

solved in a poorly evaporating solvent by a better solvent.

Solvents of low surface tension spray the sample into fine droplets from which the solutes evaporate relatively easily. They include the alkanes (maybe with the ex­ ception of cyclohexane) and ether. Solvents consuming a large amount of heat on evapo­ ration (e.g. alcohols) are more difficult to spray. The boiling point of the solvent should be at least 1000 below the injector temperature.

High Injector Temperature

The operating condition probably first thought of in the con­ text of flash evaporation is the injector temperature: the higher this temperature, the faster the sample evaporates and the finer is the spray. Visual experiments confirm this, but the importance should not be overemphasized. The length of the heated needle and the injection volume have a greater effect than the injector temperature. A high injector temperature also improves the elution of the high-boiling solutes from the syringe needle. The upper limit of the injector temperature is determined by degradation of the components of interest or the formation of "ghost" peaks by degradation of sample by-products. In general, compounds subjected to flash evaporation usually tolerate surprisingly high temperatures, prob­ ably because the sample does not actually reach this tem­ perature and there is hardly any contact with surfaces.

Small Sample Volumes

The smaller the volume of sample injected, the less heat is consumed and the less the temperature of the needle wall and the vaporizing chamber drops. A 2 III volume injected at 300 DC might, for instance, result in evaporation at a lower temperature than a 1 III injection at 250 DC.

Length of Syringe Needle

It seems plausible that release of the sample a long distance from the column entrance, i.e. the use of short syringe nee­ dles, would be preferable for split injection. The longer dis­ tance provides more time for evaporation and spreading across the liner. Interestingly, (hot needle) injection with long (71 mm) needles often provides better results than that with 51 mm needles. This might be because of bet­

10.2. Flash Evaporation

229

ter nebulization: a long needle reaches into a hotter zone, and the higher resistance against ejection from a longer nee­ dle builds up higher pressure, more strongly overheating the liquid.

10.2.3. Problems Arising from Aerosol Formation

Flash evaporation has numerous outstanding features, but there are also two problems which can be severe when sam­ ples contain high loads of non-evaporating material.

Contamination of the Col­ umn Inlet?

Nebulization tends to form aerosols. The droplets leaving the syringe needle are reduced in size by evaporation oftheir volatile constituents, such as the solvent, until the high-boil­ ing and involatile material remains, in the form of small par­ ticles. These particles move with the carrier gas and are split more or less like the vapor. When they are swept into the column, the advantage is that difficult sample components are safely carried along, but there is also the drawback that non-evaporating material contaminate the column inlet.

Particles Attracted to the Liner Wall

The behavior of particles is not sufficiently well understood. Visual experiments have shown that some particles form a stable fog and move with the gas (e.g. perylene from con­ centrated solutions) whereas others (e.g. those from 1-5 % edible oil in a solution) were rapidly and quite quantitatively attracted to the liner wall and deposited there. As discussed in Section B4, column contamination is less severe than might be expected, which suggests that larger particles are transferred to the liner wall rather efficiently.

Matrix Effects

Attraction to the liner wall severely affects the fate of the higher-boiling solutes. Because the particles carry them along and "glue" them to the liner wall (or surfaces of packings or obstacles), they can no longer enter the column as an aerosol. Now they must evaporate from a layer of con­ taminants. If this occurs with some delay, they are split later and maybe at another ratio. If evaporation is severely de­ layed, they can even be lost almost completely. This leads to non-linear splitting, discrimination phenomena, an'd usually poor reproducibility.

Danger of Systematic Errors

Ifthe solutes behave differently when injected as a clean cali­ bration mixture rather than as a sample loaded with non­ evaporating by-products, the analysis suffers from matrix effects, i.e. for a given composition, absolute and relative peak areas for the calibration mixture and the samples are different. Hence, the response factors determined by calibration are no longer applicable to the samples. Results will thus be systematically wrong. These effects are discussed in more detail for splitless injection in Section D6.3.

Conclusions

The most obvious solution to these matrix problems is sup­ pression of aerosol formation by sample evaporation from surfaces. This entails the use of fast autosampler injec­

230

C 10. Techniques for Improving Quantitative Analysis tion into a packed liner or manual injection directly into a dense plug of glass wool. No involatile material will then reach the column; the process will be the same for calibra­ tion and analysis of the samples. The advantages of gentle evaporation in the gas phase will, however, be lost. Experimental results only partly confirm these expectations. In particular, matrix effects are seldom eliminated (see Section C10.3.7). They tend to change from reducing to en­ hancing effects (Section 06.2 and 06.3).

10.2.4 Stop Flow Split Injection

At this point, some work merits brief mention, even though it has not been followed up. Bayer and Liu [46] came to the conclusion that incomplete sample evaporation was prima­ rily a consequence of insufficient residence time in the injector and demonstrated reduced discrimination upon reduction of the flow rate through the injector. To distinguish whether the flow rate must be low in the evapo­ ration zone or at the split point, they kept that at the evapo­ ration site low while the split flow rate remained high, by introducing most of the gas through a hole in the liner wall below the evaporation zone. Because results improved, they concluded that the flow rate in the vaporizing zone should be low. This led to the idea of stopping the flow through the injector at the moment of sample evaporation.

Procedure

The method does not require modification ofthe instrument. The liner contained a glass bead in its upper half (Section B3.5.4). Some 10 s before injection, the split valve was closed to reduce the flow rate through the injector to the column flow rate. The sample was then introduced and the valve opened again 1-2 s later. This procedure resembles splitless injection, but no sample material enters the column during this short splitless period because the vapor does not reach the column entrance before the split valve is re-opened.

FurtherOpumuauon

In a later paper, Liu and Xin [47] studied the influence of vapor viscosity on discrimination. Using a system with a needle valve in the split outlet they reasoned that passage of the vapor/gas mixture through this restriction changes the flow rate, i.e. the split ratio. Arrival of the vapor of different components at the split point at different moments causes non-linear splitting. Hydrogen is preferable to helium as car­ rier gas because its viscosity is nearer to that of the solvent vapor. High injector temperatures are recommended if sub­ stantial pre-separation is to be avoided. A further paper dealt with the delayed release of high-boiling components, and the discrimination against the later-eluted components which results [48].

10.2.5. An Experimental Result: Determination of Sucrose

In 1979, Nurok and Reardon [49] optimized the split injection of trimethylsilyl sucrose for the analysis of sugar cane-juice and sugar-factory products. Using an empty liner they ob­

10.2. Flash Evaporation

231

tained a threefold (I) increase in the sucrose peak area when

the injector temperature was increased from 220 to 390°C.

This suggests that an injector temperature as high as 390 °C

was needed to achieve nebulization ofthe high-boiling sam­

ple matrix (silylation reagent).

390°C is above the decomposition temperature oftrimethyl­

silyl derivatives, but the sample might never have really

reached this temperature and the residence time in this hot

environment was extremely short.

Priming of the Injector

The authors reported results with standard deviations be­

tween 0.02 and 0.10 % after first priming the injector. A sam­

ple was injected a few times in rapid succession before com­

mencing a series of analyses.

Hot Needle Injection

The syringe handling technique was optimized. Injections

were performed manually and it was concluded that "the

times for inserting the needle into the septum, leaving it fully

inserted, and withdrawing it from the septum are approxi­

mately equal. The plunger should be rapidly depressed with

a momentary delay before and after. The total process takes

about 6 s".

As the needle temperature is the relevant factor determining

whether nebulization is achieved, heating of the needle to

the injector temperature is essential. There is no need to heat

the injector to 390°C if rapid injection brings the needle tem­

perature to, e.q., 250°C only.

10.2.6. Evaluation of Flash Evaporation

The important characteristics of flash evaporation can be

summarized as follows:

1 Samples must be diluted in volatile solvents.

2 Because the sample makes hardly any contact with

injector surfaces, deactivation and cleanliness of the surfaces are rather uncritical. This renders flash evapo­ ration suitable for the analysis of labile and high-boil­ ing components. 3 Flash evaporation produces sharper initial bands than other evaporation techniques. 4 Nebulization produces aerosols which can carry substan­ tial amounts of involatile material into the column inlet. 5 Matrix effects may be encountered for samples con­ taining elevated concentrations of non-evaporating ma­ terial and bring about the danger of systematic errors.

10.3. Evaporation in Packed Liners

A fundamentally different concept involves deposition of the sample liquid on to a packing material (such as deactivated glass or fused silica wool) which is locally cooled to the sol­ vent boiling point. Sample evaporation proceeds from the surface of this packing material.

Characteristics

Evaporation from packing material has the following char­ acteristics.

232

C 10. Techniques for Improving Quantitative Analysis

1 2

3 4

5

6

It releases clean vapor, practically eliminating the for­ mation of droplets and aerosol particles. Non-evaporating materials are retained on the pack­ ing material and contaminate neither the column inlet nor the split outlet. There is ample time for transfer of heat to the evaporat­ ing liquid. The vapor of high-boiling components are diluted by a large volume of carrier gas. This facilitates the com­ plete evaporation of high-boiling components, but also results in broad initial bands. Solute evaporation from surfaces must overcome inter­ actions with the surface and is particularly difficult for high-boiling compounds. It suffers from adsorption or degradation on active sites and retention in layers of contaminants. Samples are fractionated: volatile compounds evapo­ rate together with or just after the solvent, whereas evaporation of high-boiling material often occurs many seconds later. This can have negative effects on the lin­ earity of splitting.

Packed Column GC

There is general agreement that packed column GC tends to provide more reproducible results than capillary GC, especially if the latter involves conventional vaporizing in­ jection. As the problem is largely related to injection, this experience may be used as an argument in favor of convert­ ing the vaporizing injector into a small packed GC (pre-lcol­ umn.

10.3.1. Deposition of the Sample

The sample liquid is deposited on to a hot packing material. This presupposes cooling of the packing to the sample (solvent) boiling point, because the vapor would otherwise prevent contact (Leidenfrost phenomenon). Nebulization at the needle exit renders such transfer quite impossible.

Fast Autosampler

The fast autosampler is an excellent tool for the purpose. A band of liquid is shot to the packing, cools the fibers it hits, and is sucked into the space between them. If necessary, the band of liquid travels over long distances through a hot en­ vironment without touching the liner wall. Hence it repro­ ducibly transfers the sample material to the packing. The packing can be placed anywhere between the needle exit and the column entrance. In the interest of homogene­ ous spreading of the vapor across the liner, a position far from the column entrance, i.e. near the needle exit is, how­ ever, advantageous.

Manual Injection, Conven­ tional Autosampler

If the sample is dissolved in one ofthe volatile solvents com­ monly used, manual injection with standard syringes can­ not be performed at a speed preventing evaporation inside the needle. The same applies to conventional (non-fast)

10.3. Evaporation in Packed Liners

233

autosamplers. A band of liquid can, however, be achieved

for most solvents when the needle is inserted merely ca.

15 mm and the injector head is at a temperature below ca.

170°C (measure the septum temperature by means of a ther­

mocouple).

When evaporation inside the needle cannot be avoided, the

best transfer to surfaces is achieved when the needle tip

enters a dense plug of packing, such that the droplets hit

the surfaces while still concentrated and efficiently cool that

region. For optimum placement of the packing, the posi­

tion of the needle tip in the liner must be determined. The

distance from the top ofthe liner is equal to the length ofthe

syringe needle minus the height of the septum cap and

septum purge area (Figure C22). The packing is then situ­

ated from a few millimeters above this point to ca. 5 mm

below it.

10.3.2. Injector Packing.

The use of column-packing material in liners was proposed

for split injection of, e.g., steroids in biological samples by

German and Horning in 1973 [50]. A 14 cm x 3.4 mm Ld.liner

was packed with a plug of 10 % SE-30 (a dimethylpoly­

siloxane) on Gas Chrom P (100-120 mesh) followed by a

longer plug of the same support coated with 1 % SE-30. The

whole packing might have been 8 em long.

Hartigan and Ettre (51] supported the use of such systems,

but neither group provided evidence that such massive pack­

ing of the injector is of advantage.

Low Thermal Mass

A packing material of low thermal mass is required to en­

sure that the sample liquid is able to cool and wet it. Deacti­

vated glass, quartz, or fused silica wool is suitable, as

are column packing materials, possibly coated with sta­

tionary phase. The thermal mass of frits is too high (Section

83.5.10).

Amount of Packing Material

The amount of, e.g., glass or quartz wool required is small.

The wool must form a network sufficiently dense to pre­ vent droplets being shot through it without touching the fibers. Visual experiments indicated that even rather loose packing fulfills this requirement, but there must be no major gap. The liquid does not penetrate the packing material by more than 5 mm. Any additional material (longer plug of packing) just adds to the risk of adsorption or degra­ dation.

Special Gas-Regulation System?

Long and densely packed beds, as originally proposed, can create a significant pressure drop, particularly at high split flow rates. This prompted German and Horning to modify the carrier gas regulation system. Their concept was later applied in the flowlbackpressure regulation system of Hewlett-Packard. When, however, the packed bed is reduced

234

C 10. Techniques for Improving Quantitative Analysis

to the length really necessary, no significant pressure drop occurs and this aspect no longer determines the design of the gas regulation system. Critical Inertness of the Packing

The inertness of the packing is the major weak point of the concept. As discussed in Section 87, no packing material of really satisfactory inertness is available. The most convenient packing, glass or fused silica wool, is not as inert as it basically could be, e.g. compared with the surface of a deactivated uncoated precolumn. At least part of the reason is that the procedures developed for deactivating glass or fused silica columns cannot be applied to wool. Column pack­ ing materials are superior in this respect, but they must be kept in place (at least supported at the bottom) by wool. Carbofrit (Restek) is an alternative that does not necessi­ tate the use of wool. It is a fine network of a glassy carbon type material in the shape of a plug which keeps itself in position.

10.3.3. Optimization of Conditions

The injector temperature is primarily determined by the needs of solute evaporation from the surfaces. It should be somewhat above the column temperature during elution of the last component of interest, to overcome the retentive power and adsorptivity of the packing and possibly of involatile sample material deposited at the same site by pre­ vious injections. Transfer on to a surface is favored by a low injector tempera­ ture, but the thermal mass of the packing and the manner of depositing the liquid are more importantfactors. With manual injection or autosamplers injecting at a similar speed, trans­ fer from the syringe needle is also promoted by a high injec­ tor temperature.

Injector Temperature

Short Syringe Needle

There is no need for long syringe needles, because a band of liquid covers the distance to the packing also when it is long. Short needles help avoid solvent evaporation in­ side the needle. Among the needles of standard size, those of 1.5 inch (37 mm) are best suited.

Hot Needle Injection

If solvent vaporization inside the needle cannot be sup­ pressed, injection should be performed by the hot needle technique to minimize losses of high boilers. The result­ ing intense nebulization must be accepted as an unwanted side effect.

Poorly Nebulized Liquids

Most conditions should be optimized in the direction oppo­ site to that discussed for flash evaporation. This means that more concentrated samples in solvents which are less effec­ tively nebulized (high-boiling point, large evaporation energy) are preferable.

10.3.4. Elution from the Packed Bed

Retention on surfaces delays the release of higher-boiling or adsorbed solute material and the initial bands are broader

10.3. Evaporation in Packed Liners

235

than estimated from the size of the vapor cloud formed upon rapid evaporation in the gas phase. Short Packed Column

Packed liners behave like short packed columns. Higher boil­ ing solutes are eluted with characteristic retention and (im­ agining there were a detector at the end of the packing) with a particular peak width. The volume ofthe peak eluted from the packed bed divided by the split ratio yields the volume of the initial bands in the inlet of the capillary column. A vola­ tile component leaves the packing with the band width of an unretained peak. Problems can arise for the noticeably re­ tained components.

Cut-Off for High-Boiling Compounds

Because the separation efficiency of packings is poor and the injector is an isothermal system, transition from sharp, negligibly retained to strongly retained, broad bands is rapid. Accordingly, the typical symptom to be expected is rapidly increasing peak broadening beyond a certain elution tem­ perature, followed by a baseline with perhaps some shallow waves or no peaks at all. High-boiling components are es­ sentially removed from the analysis.

Importance of Cold Trapping

Peak broadening as a result of slow release from the packed liner is most drastic in isothermal GC runs and for rap­ idly eluted peaks. Isothermal GC, in fact, enables rapid de­ tection of retention and band broadening in the injector. Tem­ perature programming of the capillary column reconcentrates broadened initial bands by the cold trapping effect, i.e. tran­ sition from sharp to broadened peaks occurs at elution tem­ peratures a few tens of degrees higher.

Retentive Power of Packing Materials

The retentive power of silanized glass wool is low, but that of column packing materials must not be underestimated. Even deactivated packing material free from stationary phase has a retentive power similar to that of a capillary column. Because optimum and stable deactivation presupposes coat­ ing with stationary phase, the retentive power usually substantially exceeds that of the capillary column. It must be overcome by use of a temperature exceeding the maximum oven temperature by several tens of degrees. Some attention must be paid to the real temperature of the packing in the liner. Most injectors are accurately thermostat­ ted somewhere in the center or lower half. The top and the bottom of the injector are often far cooler.

10.3.5. PAHs as an Example

Munari and Trestianu [52] described results obtained for poly­ cyclic aromatic hydrocarbons (PAH) with and without silan­ ized glass wool in the liner. They demonstrated the range of components for which the glass wool packing was of advantage and when it became a drawback. The results might be rather optimistic, as the experiments were per­ formed with standards in hexane. PAHs in' environmental extracts usually suffer more strongly from discrimination

236

C 10. Techniques for Improving Quantitative Analysis against high-boiling compounds, owing to the presence of matrix material which retains the components of interest.

Incomplete Evaporation

As shown in Figure C27, the mixture with equal concentra­ tions of PAHs produced peaks of constant area when injected on-column. Although the injector was at 360°C, split injec­ tion with an empty liner (5 cm needle, 4 mm i.d. glass liner, "2 ~L" sample volume) resulted in significant discrimination even for a component such as benzo[b]fluorene, commonly eluted at ca. 210°C. With glass wool, this compound was no longer noticeably discriminated against, which suggests that the loss in the empty liner was a result of incomplete evaporation and a smaller proportion of aerosol particles entering the column than would have been expected from the split ratio set. DISCRIMINAnON FACTOR

"

.• SPLITTING INj. (1:100) . . . \~ glass wool

'"... 0.9

'\

'.

".'..\. ,

SPLITIING INJ. (1:100)

~ '-':\

/

no glass wool

"\ 0.1

\~,.

.,~

0.7

~

-,

'.~

"'.,., SPLITLESS INJ.

u 110 200 220 240 2&0 280 300 MOLECULAR WEIGHT (u.am.)

Figure C27

Discrimination against PAHs resulting from the use of dif­

ferent liners. Group 1: phenanthrene, anthracene; group 2:

pyrene, benzo[b]fluorene, benzo[a]anthracene, chrysene;

group 3: 3-methylcholanthrene, dibenzo[a,i]acridine,

benzo[g,h,ilperylene, dibenzo[a,h]anthracene; group 4:

coronene, dibenzo[a,ilpyrene. (From Munari and Trestianu

[77]).

Adsorption on Glass Wool

The five-ring PAHs, including the benzopyrenes, were affected similarly with or without glass wool in the liner. For the larger molecules, the results were the opposite of those reported above: the discrimination observed with glass wool exceeded that without. Evaporation from the glass wool surface was, apparently, exceedingly slow for these compo­ nents. In terms of retentive power, wool at 360 °C should have readily released a compound such as benzopyrene. Thus the ob­ served losses must have resulted from adsorption.

10.3. Evaporation in Packed Liners

237

Discrimination arising from losses of high-boiling mate­ rial inside the syringe needle was not specified, but was, of course, constant throughout the experiment. 10.3.6. "Ghost" Peaks as

a Result of Packing Bleed

Only with Temperature Programs

Degradation of stationary phaBe. silylation material. or high-boiling sample by-products in the injector can pro­ duce volatile compounds at a more or less constant rate. Such bleed from the liner often introduces "ghost" peaks into the chromatogram. "Ghost" peaks of degraded sample by-prod­ ucts can also be observed after flash evaporation in empty liners, but they are usually much smaller.

When the capillary column is cooled. degradation prod­ ucts are accumulated at the column entrance and form sharp peaks during subsequent temperature-programmed chromatography. "Ghost" peaks often tail somewhat, ow­ ing to material introduced into the column after temperature programming has started. The size of the "ghost" peaks resulting from material con­ tinuously released from the vaporizing chamber depends on the duration of oven cooling. This provides us with an easy test of their origin - if they are from the injector, varying the cooling period must strongly influence their size.

No such "ghost" peaks are expected in isothermal GC, because the constant stream of bleed into the column merely raises the baseline slightly and permanently. This provides us with another test for the origin of the "ghost" peaks. "Ghost" Peaks Arising from Aggressive Sample By­ Products

Some samples cause "ghost" peaks even in isothermal runs. Aggressive sample by-products accelerate degrada­ tion in the injector and produce artifacts primarily during the evaporation process. If there is no obvious peak broaden­ ing, it is more difficult to distinguish between peaks of real sample components and such "ghost" peaks. In­ jection of pure solvent or of a mixture of standards may pro­ duce no such Hghost" peaks. Such "blanks" easily mislead, because the "ghost" peaks are then mistaken for real sam­ ple components. An on-column injection immediately pro­ vides the correct picture.

10.3.7. Matrix Effects

Packed liners are of particular interest for accurate analysis of samples in difficult matrices; otherwise flash evapo­ ration is preferable. This raises the question of whether ma­ trix effects are really absent. Rather little has been published on this subject, and work presented by Ferreira et et. [53], summarized here, does not confirm expectations.

Experimental Conditions

The experiments were performed with flavor components injected as standards in pure solvent !dichloromethane or pentane) or in solvent artificially contaminated with an extract from red wine. A Hewlett-Packard 5890 instrument with an HP 7673 fast autosempler was used and a 1 III

238

C 10. Techniques for Improving Quantitative Analysis volume was introduced into an injector at 250°C with a liner either empty or containing a Jennings cup and Chromosorb. The pre-set split ratio was 40:1. Areas corresponding to the pre-set split ratio were determined by on-column injec­ tion of a 40-fold diluted solution.

Clean Mixture

Table C8 shows results, calculated in terms of true split ra­ tios, for the mixture in dichloromethane injected into empty and packed liners. Split ratios were ca. 25 % lower than those adjusted, and varied between 29 and 36 for the empty liner and between 28 and 32 for the packed liner. In the chromatograms the components split at a higher ratio are regarded discriminated against. Discrimination reached 24 and 14 % for the empty and the packed liners, respec­ tively. With the empty liner, 1-hexanol and ethyllaurate were most strongly discriminated against, whereas these same components gave large peaks when the packed liner was used.

Contaminated Mixture

With the contaminated mixture and .an empty liner, true split ratios were between 27 and 30 and, as shown at the right in the table, 5 to 21 % lower than for the clean mix­ ture. Hence more solute material entered the column, corre­ sponding to an enhancing matrix effect. The narrower range of true split ratios indicates weaker discrimination. With the packed liner, the split ratios ranged between 27 and 40, in general being closer to the pre-set value than with the clean mixture. Up to 40 % less material entered the col­ umn; phenylethyl acetate was an exception: 15 % more ma­ terial was analyzed. Thus there was a rather strong reduc­ ing matrix effect. Discriminative effects were correspond­ ingly strong. Relative standard deviations (n=5) were between 1 and 3 %. Results with pentane as solvent were similar.

Table C8 True split ratios for flavor components injected into empty and packed liners (left). Pre-set split ratio, 40: 1. The increase or decrease of the true split ratio for the contaminated solution is given as percentage (matrix effect; columns at the right). (From Ferre;ra et al. [53]).

Liner 1-Propanol Ethyl butyrate 1-Pentanol 1-Hexanol Ethyl oetanoate Phenyl ethyl acetate Ethyl laurate 2-Phenylethanol

True split ratio Clean sample Empty Packed 30 29 32 36 34 34 36 34

28 28 30 29 28 32 29 29

Changes [%) for contaminated sample Empty Packed -6 -5 -16 -21 -17 -16 -18 -12

+39 +37 +25 +32 +25 -15 +9 +5

10.3. Evaporation in Packed Liners

239

Conclusion?

The results provide a disturbing picture. Up to 40 % ma­ trix effect for rather easy, volatile components is not what might be expected from the process involved. In terms of matrix effects, the results from the empty liner were supe­ rior, despite the use of a fast autosampler.

10.4. High-Boiling Sam­ ples

The third approach to improving results from split injection applies to high-boiling samples. The high-boiling point re­ fers to the sample matrix: a high-boiling solvent or an un­ diluted mixture of high-boiling materials. "High-boiling" means a boiling point not far below the injector tem­ perature or higher.

Smooth Evaporation

Samples with high-boiling matrices can be deposited on surfaces in the injector from where they evaporate rela­ tively slowly (scenario 3 in Section 83.3.3). If the boiling point of the sample exceeds the injector temperature, evapo­ ration is slow because the vapor must be diluted with carrier gas. The latter is saturated, and maybe a large volume must flow past the site of evaporation to transport the sample vapor. Only evaporation from a surface provides the conditions required: relatively long evaporation time and passage of a large volume of carrier gas.

Advantages

Smooth and possibly slow evaporation from surfaces avoids the drawbacks of rapid evaporation. No pressure wave, because (at least in pressure-con­ trolled systems) the sample vapor replaces carrier gas rather than expanding as a result of its own vapor pres­ sure. No large volumes of vapor are formed in such short times that they require intermediate storage. No formation of aerosols. No evaporation inside the needle.

10.4. I. Optimization of Conditions

In contrast with the samples discussed above, high-boiling samples do not rely on cooling of the surface before they can be deposited on to them. They can be placed on to surfaces with a large thermal mass behind, such as the wall of the liner. So far, no packing is needed in the liner.

Packing ls Unnecessary

Deactivated Liners

The advantage of using no packing material is in the inert­ ness of the system. Even non-deactivated liners seem to be less adsorptive or catalytically active than" deactivated" wool. Liners can, furthermore, be deactivated by the same efficient procedures as are used for capillary columns.

Narrow-Bore Liners

Slowly evaporating samples produce vapor at a rate posing no problem regarding their discharge. Furthermore, high­ boiling samples are usually highly concentrated and, there­ fore, injected at high split flow rates. Hence the volume of

240

C 10. Techniques for Improving Quantitative Analysis the vaporizing chamber does not need to be large enough for intermediate storage of vapor. Liners of 1-2 mm i.d. can be used, which helps to ensure complete transfer of the liquid on to their wall.

Short Needles

Short syringe needles (37 mm) should be used in order to leave a long distance between the needle exit and the col­ umn entrance. If this distance is 4-6 ern, a straight "shot" of the liquid through the liner is highly improbable (as can easily be tested with a normal syringe and a short glass tube). Maybe electric charges help to pull the liquid to the liner wall. A long distance (and a narrow bore) also aid ho­ mogenization of the distribution of vapor across the liner.

Sideport Hole?

Needles with a sideport hole seem ideally suited to reliable transfer the liquid to the liner wall. There is no experimental experience contradicting this - but standard needles appear to be equally suitable.

Fast Injection

The plunger ofthe syringe must be depressed at high speed to separate the liquid from the needle tip. High-boiling liquids tend to be rather viscous and to remain hanging on the needle tip. When the needle is withdrawn, they are wiped off on to the septum, sometimes accompanied by the release of something looking like smoke. If fast depression of the plunger is not sufficient, a packing is needed into which the needle enters, such that the liquid is wiped off on this mate­ rial.

Wettability of the Liner Surface

Deposition of sample liquid on the liner wall and evapora­ tion from a film of liquid resembles on-column injection, in which the liquid is deposited in the column inlet or an uncoated pre-column. It spreads out until a mechanically more or less stable film, 10-30 urn thick, has been produced. A problem well known in on-column injection could also be relevant for a liner. If the sample liquid does not wet the surface, it cannot form a film, but merely a few droplets here and there, like water on a window pane. Such droplets do not adhere to the surface and are likely to be driven to the bottom of the liner by a fast gas flow. The situation in the liner is better than in on-column injection because higher temperatures reduce surface tension and, hence, facilitate wetting. As known from on-column injection, trimethylsilylated sur­ faces have low critical surface energy and are wetted by a minority of the solvents only. Phenyldimethyl silylation is far better in this respect and should probably also be used for liners.

Small Sample Volumes

The capacity of the liner wall to retain liquid is limited. If ex­ cessive amounts are injected, the liquid flows down the liner wall as a tear and when it has passed the column entrance it

10.4. High-Boiling Samples

241

is lost for the analysis. Even in 1 mm l.d. liners the liquid

does not form a plug filling the liner bore and, hence, does

not make use of all the surface available. It spreads as a band

maybe 2 mm wide.

Visual experiments indicated that 2 III of liquid were reli­

ably retained, whereas 5lJ.L mostly flowed more than 4 cm

down the liner wall and, thus, passed the position of the col­

umn entrance.

Difference between Injector Temperature and Solvent Boiling Point

The maximum injector temperature is determined by three

factors.

The needle temperature must remain low enough to rule out sample evaporation inside the needle. It not only depends on the regulated injector temperature, but also on the injection speed and the septum tempera­ ture (a cool injector head is preferable). The use of a fast autosampler totally eliminates this problem. Transfer of the sample to the liner wall without repul­ sion by vapor. This seems possible for surfaces which are up to 50° above the boiling point of the sample. The solute material must be evaporated. These requirements can create a dilemma: although the in­ jector temperature should not far exceed the solvent boiling point, even high-boiling and adsorptive components must be evaporated and transferred to the column. Be­ cause evaporation occurs from a surface, the temperatures required tend to be rather high. If the injector temperature must be increased, however, this also means using a higher­ boiling solvent. Hence the method is restricted to compo­ nents with a rather narrow volatility range.

Problems with Solvent Purity

Despite the convincing concept and encouraging results, in­

jection in high-boiling solvents has not been widely adopted.

This is probably because of practical problems. Many high­

boiling solvents are not available in sufficient purity, and

redistillation is difficult. High-boiling solvents are, further­

more, expensive.

Problems with impurities can often be overcome by prepar­

ing highly concentrated samples which are then split at

high ratios and injected in small volumes (as the needle con­

tent is not transferred, it is possible to inject, e.g., 0.5 lJ.L).

No Sample Reconcentration

A further problem is encountered during sample prepara­

tion. Because high-boiling solvents evaporate only with dif­

ficulty, solutions cannot be reconcentrated. Often it is prefer­

able to work up a sample with volatile solvents and exchange

them for a high-boiling solvent at the end of the sample­

preparation procedure.

Broad Initial Bands

High-boiling liquids evaporate slowly, because the heat re­

quired for evaporation must be transferred in the absence of

a large temperature gradient and the vapor' strongly diluted

242

C 10. Techniques for Improving Quantitative Analysis by the carrier gas. Deposition on the liner wall, on the other hand, brings the sample into direct contact with the main source of heat. Broad initial bands are most likely to become visible in the chromatogram when components are eluted isothermally at the injection temperature, particularly when retention times are short. Temperature programming starting 30-50° below the elution temperature solves this problem (cold trapping).

Broadened Peaks Eluted Before the Solvent

The analysis of volatile components eluted before the sol­ vent peak is usually difficult, because peaks are often broad­ ened or even fused to the solvent peak, particularly when the split ratio is modest or low. The worst broadening is ob­ served for peaks eluted shortly before the solvent. Broadening is primarily caused by phase soaking, a solvent effect involving co-chromatography of the solute with the overloading solvent band in the coated column [55,561. The only efficient way of reducing such peak broadening is to reduce the amount of sample entering the column, i.e. to reduce the volume injected and/or increase the split ratio.

Recondensation Effects

Vapor of high-boiling samples has a strong tendency to recondense in parts which are substantially cooler than the injector, i.e. the column inlet and the split outlet. Recon­ densation can affect the split ratio, but also plug the split outlet line after a long series of analyses.

10.4.2. Experiments by Schomburg

In 1977 Schomburg et al. showed that the use of high-boil­ ing solvents substantially reduced, or even eliminated, dis­ crimination in split injection [101. Further examples were given later [57-591. Typical solvents investigated were Ce-C'2 n-alkanes and C6-C'2 fatty acid methyl esters.

Low Injector Temperatures

Linked with the use of high-boiling solvents, they recom­ mended the use of injectortemperatures not too far above the solvent boiling point and showed that a mixture con­ taining n-alkanes up to C40 could be analyzed by conventional split injection without discrimination (Figure C28). The sol­ vent was dodecane (b.p, 216°C); injector temperatures were 210 or 310 DC. It is, in fact, remarkable that a perfect peak could be obtained from a component such as C40 (b.p, 540 DC) with the injector at only 210°C. It was certainly important that the test sample was free from involatile material. The samples were injected manually by a rapid cool needle tech­ nique to help avoid sample evaporation in the needle.

10.4.3. Application to Herbicide Analysis

Li showed the advantages of high-boiling solvents for split injection of herbicides [60J. With dichloromethane as solvent, relative standard deviations were clearly higher than those obtained in packed column GC (ca. 2 % compared with 0.3 %); with butyl acetate (b.p, 127°C), however, they were

10.4. High-Boiling Samples

243

I

I, ~

I

I I I I

A

I II

B

Figure C28

Split injection (50:1) of a broad mixture of alkanes (C,o-C 40,

as indicated in chromatogram C), using high-boiling solvents.

(From Schomburg et ",. [57]).

Solvent A B C D

n-Octane n-Dodecane

Injector temp.

Area C,JC 40

310°C 210°C 310°C 210°C

0.13 0.43 1.02 1.02

lower. The injector temperature was 220 DC; injection was performed with a Hewlett-Packard 7671 A autosampler.

10.5. Homogenization of Vapor Across the Liner

Correct splitting presupposes homogeneous distribution of the vapor across the liner. A reproducible split ratio (re­ producible absolute peak areas) cannot be expected if the main stream of vapor hits the column inlet during one injec­ tion, but passes it by on another. This seems obvious, but there are few data enabling one to conclude whether or not this is a significant problem in practice. The videos suggested that homogenization of a fog across the liner is rapid (1-2 frames, i.e. 40-80 rns), particularly if the

244

C 10. Techniques for Improving Quantitative Analysis diameter is small. The splitting process takes some ten times longer. Narrow bore (e.g. 2 mm i.d.) liners promote, on the other hand, a strong pressure wave and recondensation in the column inlet if oven temperatures are low. Both contrib­ ute to increased standard deviations, maybe offsetting the advantage of homogenization.

10.5.1. Obstacles Pro­ moting Homogeneous Distribution

When mixing is aimed primarily at homogenizing the vapor/ gas mixture across the liner, the cup liner is promising, be­ cause it forces the gas phase through a narrow passage and then spreads it equally along the liner wall (Figure C29). The obstacle must be well above the column entrance be­ cause there is a dead space just below the cup. The laminar liner (Section B3.5.6) might perform even bet­ ter because the column is situated in the narrow bore center tube, where little mixing is enough to achieve homogeneity.

~,

-_.""'~-­ .;:~~ "!'o. ",.,

-j.:f!

;-

Mixing in the narrow funnel Spreading along the liner wall

Dead space

Figure C29

Homogenization of vapor and carrier gas across the vaporiz­

ing chamber by use of the cup liner.

10.5.2. Chromatographic Experiment with Two Columns

Bowermaster [611 described a simple experiment for check­ ing the homogeneity of the concentration of vapor across the liner. Two columns were placed side by side at the same height in the injector. If the concentration of vapor over the short distance between the two column entrances (0.20 mm) is constant, peak areas in a constant ratio are obtained from the two columns over several runs. The ratio is that of the flow rates into the two columns.

Densely Packed Straight Liner

The results reproduced in Figure C30 are surprising. For a straight 4 mm l.d, liner at 250°C, tightly packed with a 1 cm plug of glass wool (22 mg), the area ratios for a mixture of C10-C30 alkanes were stable within ca. 1 %. A 1 ul, volume of a hexane solution was injected by means of the HP-7673A fast autosampler.

10.5. Homogenization of Vapor Across the Liner 1.35

245

r----------­

1.3

Inverted Cup Liner. 250 0 !

!

,

!

,

,

15

20

25

15

20

25

Alkane Carbon Number

Alkane Carbon Number

Figure C30

Area ratios for n-alkanes eluted from two columns installed side by side in a split injector.

Straight liner with a tight plug of glass wool, and a liner with a cup below some loose glass

wool. (From Bowermaster[61]).

Cup Liner

With a cup liner loosely packed with 9 mg glass wool above the cup, variation of the area ratios was increased by a factor of at least ten. Area ratios varied from run to run, i.e. vapor of different concentration reached the columns (horizontal lines). Often the area ratios also varied from com­ ponent to component (slanted lines), indicative of material of different composition entering the two columns. As gas­ phase equilibration over such short distances is rapid, Bowermasterassumed unequal aerosol distribution was the source of the problem.

Decane as Solvent

With the same liner, but decane as solvent, the area ratios were almost as stable as with the straight, densely packed liner (not shown). Bowermaster explained this as elimina­ tion of aerosol formation. Boiling at 150 °C, the solution evaporated more slowly and more smoothly.

10.5.3. Fatty Acid Methyl Esters

Bannon et al. [42] searched for the liner providing the best results for split injection of fatty acid methyl esters dissolved in hexane, i.e. for a rather "easy" sample, but one calling for high accuracy. Injections were performed with the fast Hewlett-Packard autosampler. Three liners were compared: a packed with glass wool; b packed with 10 % SE-30 on Gas Chrom Q and glass wool above an "inverted cup"; c glass wool above two cups (double obstacle). In general, the results obtained were good, and, therefore, differences were rather small.

Importance of the "Inverted Cup"

Compared with liner a, use of liner b also generated highly accurate results under conditions further from the optimum (e.g. larger sample volume). It is, however, difficult to con­ clude whether this improvement was the result of the cup or the packing material. Liner c, i.e. that containing two cups and no packing material, provided still better results, sug­ gesting that the cup was the decisive part.

246

C 10. Techniques for Improving Quantitative Analysis

10.6. Two Case Studies

Two analyses studied in some detail are summarized below in order to illustrate how difficult it can be to draw simple conclusions.

10.6.1. About a Dispute: the Methanol/2-Ethyl-1­ Hexanol Mixture

In 1977, Schomburg et al. [101 published a figure (Figure C31), which many took as a generally valid evaluation of liners. The liners were tested by injection of a solvent-free 1:1 mixture of methanol and 2-ethyl-1-hexanol. Only relative standard deviations of the area ratios were given. With an empty glass tube, the relative standard deviation was extremely high - 34 'Yo. A small amount of glass wool in the liner slightly improved it to 25 'Yo. With an liner contain­ ing baffles, a cup liner, or a packing of chromatographic sup­ port material, relative standard deviations dropped to 2-3 'Yo. Finally, a long, tight packing of glass wool resulted in repro­ ducibilitv within less than 1 %.

5

a

c b d e 9 25 <.1 2,1 25 2.0 27 % standard deviation Figure C31 Comparison of different liners. Relative standard deviations obtained for the ratio of the peak areas from a 1: 1 mixture of methanol and 2-ethyl-1-hexanol. a, empty liner; b, short glass wool plug near split point; e. short glass wool plug near needle tip; d, long and tight glass wool plug; e, cup liner; f, liner with baffles; g, chromatographic support pack­ ing. (From Schomburg st al. (10)).

34

On the Difficulty of Repeat­ ing these Results

Because these results were not in accord with our experi­ ence, we repeated the experiment and a different picture emerged [621. Some of our results are summarized in Table C9. We determined the correct area ratio by on-column in­ jection of a diluted mixture. For split injection, the injector temperature, the internal diameter of the liner, the length of the syringe needle, the needle handling technique (cool nee­ dle/hot needle), the split ratio, and the sample volume were varied, always comparing the results from the empty liner with those from a liner densely packed with glass wool.

10.6. Two Case Studies

247

Table C9 Average area ratios and relative standard deviations (RSD. o/Dl for a 1:1 mixture of methanol and 2-ethyl-1-hexanol. Results for two iniector temperatures. for 2 and 4 mm i.d. liners. cool and hot needle inieetion. and for syringes with short and long needles. (Selected resu"s from ref. [62]).

Injection Liner

Technique

Needle

0.397

On-column Split

200°C 200 °C 350°C 350 °C 350°C 350 °C

Peak area ratio (RSD) Without wool With wool

2mm 4mm 2mm 4mm 4mm 4mm

hot needle hot needle hot needle hot needle hot needle cool needle

38mm 38mm 38mm 38mm 71 mm 71 mm

0.435 (4.3%) 0.477 (10%) 0.460 (10%) 0.392 (1.7%) 0.407 (3.8%) 0.425 (1.2%)

0.475 (3.2%) 0.533 (6.7%) 0.295 (11%) 0.450 (2.2%) 0.391 (2.9%) 0.409 (4.0%)

Standard Deviations Inde­ pendent of Use of Glass Wool

The standard deviations of our results never exceeded 11 %, and this maximum was obtained with a liner packed with glass wool. There was no significant difference between the reproducibility of results obtained with and without glass wool. Direct comparison with the results of Schomburg et al. was not possible because information about the condi­ tions used were either vague or missing.

Deviating Area Ratios

The area ratios deviated rather strongly from the correct value, but there was no tendency towards more correct results when glass wool was used. If the injector tem­ perature was high and the injection volume small, results were closer to those expected. The length of the needle, the needle handling technique (cold or hot needle injection). the split ratio, and presence of glass wool seemed to be of sec­ ondary importance.

Evaporation from the Liner Surface

The dispute did not end with a clear result. We do not ques­ tion the results published by Schomburg et el., but we doubt that their conclusions are of general validity. For us the study ended with the same confusion as many other (un­ published) investigations aimed at clarifying the importance of one injection condition or another. Maybe their injector had a cooler injector head, and this helped suppress evapo­ ration inside the needle, whereas we used thermospray in­ jection.

Fluctuating Split Ratio

When the whole complexity ofthe process is taken into con­ sideration, there is not really much surprise about inexplica­ ble results. Packed liners, for instance, provide conditions promoting complete sample evaporation. They also, how­ ever, cause pre-separation of the sample, and then a fluc­ tuating split ratio is bound to render splitting non-linear. Pack­

248

C 10. Techniques for Improving Quantitative Analysis ing of the liner can solve one problem (that of incomplete

evaporation), but create another.

Methanol might have evaporated rapidly and been split at a

ratio deviating from the pre-set value because of the pres­

sure wave and recondensation in the column inlet (the col­

umn temperature during injection was, in fact, ambient). 2­

Ethyl-1-hexanol was vaporized at a later stage and more

slowly, hence split at another ratio.

Such a process depends on numerous factors. On changing,

e.g., the length of the syringe needle, conditions may be­

come such that even methanol reaches the split point only

after the pressure wave has dissipated (both components

being split after the split ratio is re-established). Another time

both components might be equally affected by the fluctuat­

ing split ratio. An increase or decrease in the injector tem­

perature influences not only the amplitude of the pressure

wave, but also whether the liquid touches the liner wall, when

a component is split, and which material is lost inside the

needle.

10.6.2. Analvsis of Alcoholic Beverages

The analysis of the so-called higher alcohols in alcoholic bev­

erages is an interesting example, because it involves a diffi­

cult sample matrix and components in the most critical

range of volatility.

No Alternative to Split Injection

Alcoholic beverages, primarily distillates, were analyzed by

split injection of the undiluted sample [631. Neither splitless

nor on-column injection is suitable for this analysis because

of partial solvent trapping effects [641 which cause the

methanol peak to be fused to the ethanol peak and most of

the early peaks to be broadened or split.

1-1.5IJ.L("0-0.5") of sample was injected manually at a split

flow rate of 40 mLJmin. The capillary column was coated with

Carbowax 400 or a 4 IJ.m film of a 5 % phenylmethylpoly­

sifoxane stationary phase. Dioxane was added as internal

standard.

Difficult Evaporation

Evaporation of the matrix, water and ethanol, requires an

extremely large amount of heat. In addition many alcoholic

beverages contain 15-30 % sugar and other in volatile by­

products, rendering complete vaporization of the compo­

nents of interest even more difficult (caramelized sugar par­

ticles retain solute material).

Volatile Solutes

Another problem is related to the high volatility of some of

the components of interest. Because methanol, acetaldehyde,

and ethyl acetate (poor solvation) are more volatile than the

sample matrix, some of the solutes are evaporated before,

others after the matrix. This easily results in non-linear split­

ting.

10.6. Two Case Studies

249

Strong Recondensation Effect

The high volatility of some components means that the analy­ sis must be started at ambient temperature. Under such con­ ditions, water and ethanol tend to recondense in the column inlet and to pull far more sample material into the column than is expected from the pre-set split ratio. Volatile compo­ nents are likely to be split during a period in which the split ratio fluctuates strongly; this also has the potential to cause non-linear splitting.

Results

Quantitative determinations were based on peak areas rela­ tive to dioxan as internal standard. Selection of the inter­ nal standard is important, because it influences the accuracy and precision with which a component is determined. Table C10 lists concentrations (results from the second and third injection) of selected components of a Williams distillate in­ jected under different conditions. Concentrations (mg/100 mL abs. alcohol) were calculated by use of the same correction factors throughout the experiment.

Hot Injector, Glass Wool

The results in the two top lines were obtained with a short needle and a hot injector (300°C) packed with glass wool. Results obtained for n-propanol, the butanols, pentanols, and methanol were highly reproducible whereas those for ethyl acetate and ethyl lactate varied by more than 10 %.

Day to Day Variation

One week later, the same sample was re-analyzed on the same instrument and under the same conditions. Now 10 % more methanol and three times more ethyl lactate were found. Analysis performed with on-column injection showed the true concentration of ethyl lactate to be 130 mg/100 mL abs. alcohol, i.e. between the two results obtained by split injection.

Table C10

Concentrations of some components of a Williams distillate. (From Altorte, [65]).

meOH. methanol; et ac. ethyl acetate; n-prop. n-propanol; 3-me bu, 3-methyl-1-butanol; et lac.

ethyl lactate; diox. dioxan (for which the absolute area is given). Split ratio. 30:1.

Injector temp.

Glass wool

Needle length

meOH

300°C

yes

37mm

1241 1237 1362 1360 1316 1350 1314 1312 1443 1589

one week later

50°C

yes

71 mm

no

37mm

no

51 mm

n-prop 3-me bu et lac et ac [mg/100 mL abs. alcohol]

117 100 105 110 117 121 115 133 157 129

264 266 289 290 356 356 303 369 294 290

83 83 97 96 84 85 90 89 98 86

43 57 146 153 130 132 115 116 126 147

diox area 512 508 966 1213 2470 2228 1373 1557 1947 1256

250

C 10. Techniques for Improving Quantitative Analysis

Results do not, of course, really depend on the date or the day of the week. Such deviations under apparently identical conditions indicate that we are not aware of all the deci­ sive factors. Long Syringe Needle

Results obtained with a long syringe needle (leaving a 25 mm distance to the column entrance) were clearly more re­ producible, and when the analysis was repeated the fol­ lowing week (together with those produced by the shorter needle), the results remained similar (7 % more methanol). The concentrations of the pentanols and ethyl lactate corre­ sponded exactly to those obtained by on-column injection.

Glass Wool

Results obtained with glass wool hardly differed from those without, either the actual values found (using the same re­ sponse factors) or the reproducibility.

Injector Temperature

At an injector temperature of 370 "C, with a packed liner, and using a short needle, results (not listed in Table e10) were quite reproducible, but concentrations of ethyl lactate were some 20 % too high (which could also mean that the peaks of dioxan and most other components were 20 % too small). At an injector temperature of 230 "C, other conditions being equal, results were more reproducible than at 300 "C,

Cool Injector

A surprising observation: when the injector heater was switched off, the temperature in the vaporizing chamber was ca. 50 "C, Results obtained with an empty liner and a 5 cm needle were, nevertheless, quite accurate. Standard devia­ tions clearly increased, however. At such a low temperature, the sample was probably transferred to the liner wall.

Absolute Peak Areas

Absolute peak areas (in the table given for dioxan) varied by a factor of five, which means that the split ratio varied that much. They were low when a packed liner was used with a short syringe needle and high when a long needle was used with an empty liner.

Dioxan as Internal Standard

Dioxan seems to behave like the sample components of in­ termediate volatility, as concluded from the reproducible area ratios for the butanols and pentanols and the poorly reproducible ratios for methanol, acetaldehyde, ethyl acetate, and ethyl lactate. The most volatile components, primarily ethyl acetate and acetaldehyde, evaporate before the inter­ nal standard and, therefore, tend to behave differently.

Explanation of the Results?

After many years of performing this analysis, we are skep­ tical about the explanation of these results. The feel­ ing prevails that by repeating the experiments we would obtain considerably different results again. Too many intel­ lectually brilliant interpjetations have been found for all sorts of phenomena, and too many of these have proven wrong the following day.

10.6. Two Case Studies

251

Concepts are useful for the optimization of the conditions in rather simple applications, but can fall short when mecha­ nisms are more complex.

11. General Evaluation of Split Injection Comprehensive evaluation of split injection is made difficult by the extremely wide range of samples analyzed. This also explains why some users are fully satisfied with their results whereas others conclude that quantitative analysis involving split injection resembles a football game in a mine field (or simply a waste of time). Headspace and Gaseous Samples

Headspace and other gaseous samples are analyzed almost exclusively by split injection, because this is the only easy way of obtaining sharp initial bands. Splitting of gaseous samples creates few problems because most deviations discussed above result from sample evaporation.

Convenience

No other injection technique is as simple to handle as split injection. Unless very low split flow rates are used it ensures sharp initial bands, irrespective of the column tempera­ ture during injection and the volatility and polarity of the solvent. It is, in fact, the only technique enabling injection at any oven temperature. For some applications it is important that split injection hardly ever leads to disturbing sol­ vent effects (partial solvent trapping) or band broadening in space. This ease of handling is reminiscent of injection on to packed columns. The possibility of running fully isothermal chro­ matograms at elevated column temperatures facilitates rapid analysis.

Qualitative Analysis

Split injection is widely used to check the approximate com­ position of reaction products, the purity of a substance, or for fingerprinting of flavors, essential oils, cigarette smoke, and other complex mixtures. Fine chromatograms can be obtained with minimum understanding of the injection technique.

Quantitative Analysis

For quantitative analysis of difficult samples it is often more art than science to fi~ conditions producing sufficiently accurate and reliable results. Some generalizations might serve as guidelines for predic­ tion of how much of a problem a certain sample might cre­ ate. Relative standard deviations below 10 % can be achieved by use of an external standard if the components are of me­ dium to high volatility and are dissolved in an readily evapo­

252

C 11. General Evaluation of Split Injection rating matrix. Use of external standards can, however, eas­ ily introduce large systematic errors (which are often diffi­ cult to detect). Methods employing internal standards or standard addition are clearly preferable. Standard devia­ tions are lower, the volatility of the solute(s) and the ease of evaporation of the matrix are less critical, and possible sys­ tematic errors are smaller. Results are worse for samples containing components with a wide range of volatility and/or in a matrix which is difficult or impossible to evaporate fully. Repeated in­ jections of mixtures of standards frequently give results with a misleadingly small standard deviation. Calibration of absolute areas or correction factors often pro­ vides values which change from one day to another, or from one operator or one syringe to another, often for reasons which never become apparent. Correction factors may not, furthermore, really be applicable to the sample, because the deviations occurring during the analysis of the calibration mixture differ from those affecting the sample. Poor reliabil­ ity of results has been confirmed by rnanv collaborative stud­ ies revealing serious deviations between the results obtained by different working groups. Quantitative methods involving split injection require especially careful validation and frequent control.

Superior Alternative?

For the quantitative analysis of mixtures containing a wide variety of solutes, split injection should, whenever possible, be replaced by a more reliable technique. such as on­ column injection. For example, sorbitol has been deter­ mined in foods for diabetics; xylitol was used as internal standard. When the acetylated sample was analyzed by split injection, response factors varied between about 1.1 and 1.4. Frequent calibration runs were required, because the cor­ rection factors changed even when all the conditions had (apparently) been reproduced perfectly. There also remained uncertainty about whether or not the response/correction factors determined by use of the calibration mixture really applied to the sample. The sample matrix, dominated by acetic anhydride and pyridine, is, of course, relatively diffi­ cult for split injection. With on-column injection, a single run of the acetylated sample yielded the result, because the re­ sponse factor was practically unity and stable (relative stand­ ard deviations of around 2 %).

On-Column Injection?

The principal problems occurring in classical split injection are probably known, but their complexity leaves little hope that comprehensive solutions could ever be found. Will the slowly on-going process of replacement continue until the classical split injector can be discarded? To remain down-to-earth, the alternatives must also be con­ sidered critically. On-column injection has taken over a range of applications, particularly those requiring high accuracy or

11. General Evaluation of Split Injection

253

dealing with labile solutes. There remain, however, at least four types of application for which on-column injection cannot replace split injection: gaseous samples and headspace analysis; samples which cannot be diluted with solvent because solute peaks would be obscured; if solvent effects disturb the chromatography of solutes (partial solvent trapping); samples containing high concentrations of involatile ma­ terial. PTV Injection

Programmed temperature vaporizing (PTV) injection is another alternative with the potential to partly usurp clas­ sical vaporizing injection. In the split mode it does not, how­ ever, completely eliminate discrimination and cannot com­ pete with the accuracy of on-column injection. The injector is too small for the introduction of gaseous samples or for headspace analysis. Finally, PTV injection does not produce sharp initial bands and is, therefore, not suited to rapid iso­ thermal analysis. Hence, PTV split injection also is suitable for replacing conventional split injection only partly. Whereas split injection probably will (and should) lose still more of its original ground. there is no technique in sight which would render it generally obsolete.

254

B References

References C

2 3

4 5 6 7

8

9 10 11 12 13 14 15

16 17

18

19

20 21

22

G. Schomburg, H. Husmann, and F. Schultz, "On-Column Injection with Split/Splitless Sampling onto GC Capillary Columns", HRC & CC 5 (1982) 565. C. Bicchi, A D'Amato, A Galli, and M. Galli, "Cold On-Column/Sample Split Injection by a Three-Way Press-Fit Device", HRC 13 (1990) 649. D.H. Smith, P.F. Bente III, R.R. Freeman, and J.E. Cusack, "A New Multi-Mode Inlet Sys­ tem for High Resolution GC", Technical Paper 74, Hewlett-Packard (1978). L.S. Ettre, "Sample Introduction into Open-Tubular Columns - History", in "Sample In­ troduction in Capillary GC", P. Sandra (Ed.), Hirthiq, Heidelberg, (1985) 1. D.H. Desty, Abh. Deut. Akad. Wiss. Berlin, KI. Chern. Geol. BioI. 9 (1959) 176. R.D. Condon, "Design Considerations for a GC System Employing High Efficiency Golay Columns", Anal. Chern. 31 (1959) 1717. L.S. Ettre and W Averill, "Investigation of the Linearity of a Stream Splitter for Capillary GC", Anal. Chern. 33 (1961) 680. H. Bruderreck, W Schneider, and I. Halasz, "Quantitative GC Analysis of Hydrocarbons with Capillary Columns and FID. IV. Principle of a New Splitting System", J. Gas Chromatogr. 5 (1967) 91. AL. German and E.e. Horning, "Capillary Column Inlet System for the GC of Biological Samples", Anal. Letters 5 (1972) 619. G. Schomburg, H. Behlau, R. Dielmann, F. Weeke, and H. Husmann, "Sampling Tech­ niques in Capillary GC", J. Chromatogr. 142 (1977) 87. WG. Jennings, "Glass Inlet Splitter for GC", J. Chromatogr. Sci. 13 (1975) 185. K. Grob and H.-P. Neukam, "The Influence of the Syringe Needle on the Precision and Accuracy of Vaporizing GC Injections", HRC & CC 2 (1979) 15. J. V. Hinshaw, "Setting Up an Inlet Splitter", LC-GC Int. 2(5) (1990) 26. T.W Card, Z. Y. AI-Saigh, and P. Munk, "Diffusion in the Bubble Flow Meter in Inverse GC Experiments", J. Chromatogr. 301 (1984) 261. D. Rittenhouse and T. Teit, "A New Technology for More Accurate Gas Flow Rate Meas­ urement", Am. Lab., Sept. 1992. W Jennings, "GC with Glass Capillary Columns", 2nd edn., Academic Press, I\lew York, London, Toronto, Sidney, San Francisco (1980) 75. G. Guiochon, "Influence of Injection lime on the Efficiency of GC Columns", Anal. Chern. 35 (1963) 399. J.C. Sternberg, "Extra Column Contributions to Chromatographic Band Broadening", in: J.e. Giddings and R.A Keller (Eds), Advances in Chromatography, Vol. 2, Marcel Dekker, New York (1966) 205. G. Gaspar, R. Arpino, Vidal-Madjar, and G. Guiochon, "Influence of Instrumental Con­ tributions on the Apparent Column Efficiency in High Speed GC", Anal. Chern. 50 (1978) 1512. K.D. Bartle, in: M.L. Lee, F.J. Yang, and K.D. Bartle, "Open Tubular Column GC", John Wiley, New York, Chichester, Brisbane, Toronto, Singapore (1984) 42. J. V. Hinshaw, "Split injection", in: P. Sandra (Ed.) Sample Introduction in Capillary GC, Volume 1, Huthiq, Heidelberg, (1985) 42. AX Bemgard and AL. Cotmsio, "Influence of Extra-Column Effects in Capillary GC", HRC 13 (1990) 689. '

e.

B References

255

23 M. van Lieshout. R. Derks, H.-G. Janssen, and C.A Cramers, "Fast Capillary GC: Compari­ son of Different Approaches", HRC 21 (1998) 583. 24 P. van Ysscket; H.M. Snijders, H.-G. Janssen, and C.A Cramers, "The Use of Non-Split­ ting Injection Techniques for Trace Analysis in Narrow-Bore Capillary GC", HRC 21 (1998), 491. 25 B.A Ewels and R.D. Sacks, "Electrically Heated Cold Trap Inlet System for High-Speed GC", Anal. Chem. 57 (1985) 2774. 26 L.A Lanning, R.D. Sacks, R.F. Mouradian, S.P. Levine, and J.A Foulke, "Electrically Heated Cold Trap Inlet System for Computer-Controlled High-Speed GC", Anal. Chem. 60 (1988) 1994. 27 J.P.EM. Rijks, H.M.J. Snijders, AJ. Botnbeeck, HEM van Leuken, and J.A. Rijks, "Recent Advances in Cryofocussed Sample Introduction and Applicability of High Speed Capil­ lary GC", Proc. 13th Int. Symp. on Capillary Chromatography, Riva del Garda, 1991, P. Sandra (Ed.l, Hlithig, Heidelberg (1991) 35. 28 AJ. Borgerding and C.W Wilkerson Jr; "A Comparison of Cryofocusing Injectors for Gas Sampling and Analysis in Fast GC", Anal. Chem. 68 (1996) 2874. 29 P.Husek and C.C. Sweeley, "GC Separation of Protein Amino Acids in Four Minutes", HRC 14 (1991) 751. 30 K. Grob and Ph. Matile, "Comparmentation of Ascorbic Acid in Vacuoles of Horseradish Root Cells. Note on Vacuolar Peroxidase", Z. Pflanzenphysiol. 98 (1(j~O) 235. 31 C.P.M. Schutjes, EA Vermeer, J.A Rijks, and C.A Cramers, "Increased Speed of Analysis in Isothermal and Temperature-Programmed Capillary GC by Reduction of the Column Inner Diameter", J. Chromatogr. 253 (1982) 1. 32 JE Purcell, "Quantitative Capillary GC Analysis", Chromatographia 15 (1982) 546. 33 B. Kolb, "Analysis of Food Contaminants by Headspace GC", in: J. Gilbert (Ed.), Analysis of Food Contaminants, Elsevier, Amsterdam (1984) 127. 34 B. Kolb (Ed.), Applied Headspace GC, Heyden, London (1980). 35 B. Kolb, "Headspace GC with Capillary Columns", in: Sample Introduction in Capillary GC, P. Sandra (Ed.) Hlithig, Heidelberg (1985) 191. 36 K. Grob and P. Frech, "0.53 mm i.d, vs 0.32 mm i.d, GC Columns: Are Large Diameter Columns Necessary?", Am. Lab., Aug. 1988, 19. 37 K. Grob and G. Grob, "Capillary Columns with Very Thick Coatings", HRC & CC 6 (1983) 133. 38 F. David, M. Proot, and P. Sandra, "On the Efficiency of Capillary Columns Coated with Medium Polarity Thick Films", HRC & CC 8 (1985) 551. 39 K. Grob and H.P. Neukam, "Pressure and Flow Changes in Vaporizing GC Injectors during Injections and their Impact on Split Ratio and Discrimination of Sample Components", HRC & CC 2 (1979) 563. 40 K. Grob and H.P. Neukam, "Dependence of the Splitting Ratio on Column Temperature in Split Injection Capillary GC", J. Chromatogr. 236 (1982) 297. 41 AE Kaufman and CE Polymeropoulos: "Study of the Injection Process in a GC Split Injection Port", J. Chromatogr. 454 (1988) 23. 42 C.D. Bannon, J.D. Craske, D.L. Felder, and L.M. Norman, "An Investigation of Pre-Set and Actual Splitting Ratios in Split Injection Capillary GC", J. Chromatogr. 404 (1987) 340. 43 K. Grob, "Stability of the FID Sensitivity during an Analysis in Capillary GC", HRC & CC 3 (1980) 286. 44 C.D. Bannon, J.D. Crsske, D.L. Felder, I.J. Garland, and L.M. Norman, "Analysis of Fatty Acid Methyl Esters with High Accuracy and Reliability. VI. Rapid Analysis by Split Injec­ tion Capillary GC." J. Chromatogr. 407 (1987) 231. 45 K. Grob, "Split Injection in Capillary GC", in: RE Kaiser (Ed.], Proc. 4th Int. Symp. on Capillary Chromatography, Hindelang 1981, Hlithig, Heidelberg (1981) 185.

256

B References

E. Bayer and G.H. iiu, "New Split Injection Technique in Capillary Column GC", J. Chromatogr. 256 (1983) 201. 47 G. Liu and Z Xin, "Basic Aspects of Stop Flow Split Injection", Chromatographia 28 (1989) 385. 48 G. Liu and Z Xin, "Sample Plug Tailing and Cold Trap Effect in Hot Split Sampling", Chromatographia 28 (1989) 639. 49 D. Nurok and T.J. Reardon, "Operating Conditions for the Determination of Sucrose by Capillary GC", Anal. Chern. 50 (1978) 855. 50 A.L. German and E.C. Horning, J. Chromatogr. Sci. 11 (1973) 76. 51 M.J. Hartigan and L.S. Ettre, "Questions Related to GC Systems with Glass Open-Tubular Columns", J. Chromatogr. 119 (1976) 187. 52 F. Munari and S. Trestianu, "Comparison of some Quantitative Results Obtained with Non-Vaporizing Cold On-Column and Vaporizing Split-Splitless Injection Techniques", in: R.E. Kaiser (Ed.), Proc. 4th Int. Symp. on Capillary Chromatography, Hindelang 1981, Huthiq, Heidelberg, Basel, New York (1981) 349. 53 \I. Ferreira, A. Escudero, J. Sa/afranca, P. Fernandez, and J. Cacho, "Matrix Effects and Solute Discrimination when Injecting Dirty Samples in Capillary Columns. Comparative Study between Classical Split and Splitless Injections", J. Chromatogr. A 655 (1993) 257. 54 K. Grob and B. Schilling, "Observation of a Peak under the Action of 'Phase Soaking', a GC Solvent Effect, during Passage through a Capillary Column", J. Chromatogr. 259 (1983) 37. 55 K. Grob and B. Schilling, "Broadening of Peaks Eluted Before the Solvent in Capillary GC", Chromatographia 17 (1983) 357 and 361. 56 K. Grob, "On-Column Injection in Capillary GC", Huthiq, Heidelberg (1987) 245. 57 G. Schomburg, R. Die/mann, H. Borwitzky, and H. Husmann, "Capillary GC of Compounds of Low Volatility", in: G. Schomburg and L. Rohrschneider (Eds.], Proceedings 12th Int. Symp. on Chromatography, Baden-Baden, 1976, Elsevier, Amsterdam (1978) P 157. 58 G. Schomburg, "Practical Limitations of Capillary GC", HRC & CC, 2 (1979) 461. 59 M.G. Proske, M. Bender, G, Schomburg, and E. Hiibinger, "Routine Quantitative GCAnaly­ sis of Pesticides for Quality and Production Control Using Capillary Columns and On­ Column (Syringe) Sampling", J. Chromatogr. 240 (1982) 95. 60 R.T. "Ouantitative Determination of Herbicides by Capillary GC", HRC & CC 6 (1983) 680. 61 J. Bowermaster, "The Influence of Evaporation Dynamics on Precision in Split Capillary GC", HRC & CC 11 (1988) 802. 62 K. Grob, H.P. Neukom, and P. Hilling, "Glass Wool in the Inserts of Split Injectors for Capillary GCr, HRC & CC 4 (1981) 203. 63 K. Grob, H.P. Neukom and H. Kaderli, "Higher Alcohols in Alcoholic Beverages by Direct Analysis on Glass Capillary Columns", HRC & CC 1 (1978) 98. 64 K. Grob, "Solvent Effects in Capillary GC", J. Chromatogr. 279 (1983) 225. 65 M. A/torler, Kantonales Labor Zurich, personal communication.

46

u:

1.1. Concept

257

D Splitless Injection

1. Introduction 1.1. Concept

"Splitless" injection means injection without split flow during injection, i.e. without splitting of the sample and venting of (usually a large) part of the sample material through the split outlet. "Splitless" further implies that it is performed with an injector also suited for split injection. Before the sample is introduced, the split exit is closed. The carrier gas now transfers (nearly) all the sample vapor into the column, because this is the only exit of the vaporiz­ ing chamber.

Trace Analysis

Splitting is avoided to achieve higher sensitivity than with split injection. Splitless injection is the classical technique for trace analysis. Appropriate solute concentrations are be­ tween 0.1 and 50 ppm (ng/Ill), the former being the limit for the FlO with injections of 1-3 ul., With more sensitive detec­ tors, far lower detection limits can be reached.

1.2. Historical Back­ ground

Splitless injection was introduced in the late nineteen six­ ties/early seventies, i.e. in the early days of trace analy­ sis. Concentrations in the range of ppb and ppt could be routinely analyzed for the first time, with the breathtaking new experience that, in principle "everything" could be de­ tected. It was the time of the discovery that anthropogenic substances such as DDT were detectable throughout the world. The need for a non-splitting injection technique as a means of enhancing the detectability of trace solutes was obvious.

258

0 1. Introduction

Theory Preventing Experi­ ments

Despite spectacular results from trace analysis and the great need for a kind of non-splitting injection technique, the his­ tory of splitless injection does not start with a brilliant con­ cept or an ingenious idea. On the contrary, (inadequate) theo­ retical considerations suggested that something like splitless injection could not work, and for years this was sufficient to prevent serious experimentation in the area. The widely accepted opinion was that the "sample capac­ ity" of capillary GC was low (sample capacity being under­ stood as the capacity for the total amount of sample mate­ rial injected). To produce sharp initial bands, it seemed obvi­ ous that the cloud of sample vapor entering the column should be small and form a plug occupying only a short sec­ tion of the column inlet.

Start by Accident

The development of splitless injection was the result of an accident. Early one morning in 1968, the split valve was accidentally left closed when my father, Kurt Grab, per­ formed an injection which was supposed to be a normal split injection. The split valve was only opened somewhat later when he noticed the error. All the peaks were, of course, much taller than normally; more surprising than that, however, was that the peaks were well shaped and sharp, not broadened as everyone had predicted. (byproduct of solvent f''.

c 10 p.1 0.01"1. solution injected withoul splitting, after 20 ",. splitting '·20

B 101'1 001"1. solution no splitting at any hme

Figure D1 Basic working instructions as demonstrated by Grob (1] in 1969: A, split injection, serving as a reference. B, splitless injection of a 50 times more dilute solution with the split valve permanently closed. C, splitless inj.ction with the split valve closed for 20 s. C.-C.alkanones; injection at 25°C, then temperature-programmed at 5 °/min.

1.2. Historical Background

259

Reconcentration of Broad Initial Bands

The first lecture by my father which I heard, probably in 1969, still reflected the "impossibility" of splitless injection. With the split exit closed, the transfer of a 0.5-1 mL vapor cloud takes 20-60 s, and the vapor (of volatile components) can start leaving the column before the last vapor enters it­ convincingly explaining why splitless injecting would pro­ duce peaks far worse than those obtained from a packed column. Because this concept of the initial band width was, of course, correct, it had to be concluded that the success of splitless injections was a result of reconcentration effects which sharpened the solute bands before the separation proc­ ess started.

Early Description of the Technique

An initial, yet already detailed, description of splitless injec­ tion was published in two papers by my parents, Kurt and Gertrud Grob in 1969 [1I. As shown in Figure 01, they dem­ onstrated that the split valve must be re-opened when sam­ ple transfer into the column was virtually complete in order to purge a relatively small amount of solvent vapor from the injector and feed pure carrier gas into the column. They showed, furthermore, the effectiveness of reconcen­ tration by what was later to be termed "cold trapping" (see Figure 02), today known to be one of two reconcentration techniques which can be used. The other means of recon­ centrating the solute bands, the "solvent effect", was out­ lined in 1974 [21. It enables isothermal analysis at the col­ umn temperature used for injection. A "solvent effect" is obtained if the column temperature during sample introduc­ tion is sufficiently far below the solvent boiling point that recondensation of the sample solvent occurs in the column inlet. An example is shown in Figure 03. It took another nine years to elucidate the real nature of the "solvent effect"

[31. Injector Suitable for Splitless Injection

The fact that the same injector could be used for split and splitless injection created the misleading impression that any split injector would be suitable for splitless injec­ tion, and because splitless injection did not have a spec­ tacular introduction, instrument manufacturers showed lit­ tle interest. In 1977, Kurt Grob designed an injector on the basis of his experience that a vaporizing injector optimized for splitless injection was also suitable for split injection, but not necessarily vice versa. It was commercialized by Carlo Erba (now CE InstrumentslThermoQuest) and remained the prototype for others for a long time.

Late Investigation of Critical Parameters

Splitless injection was rapidly adopted by chemists per­ forming trace analysis, but the leading chromatographers did not accept it for nearly a decade, with the result that little further development work was performed and papers on splitless injection published up to the early eighties created confusion, rather than contributing to the evaluation and optimization of the technique.

260

D 1. Introduction C injection during 20 sec at 25"C 3'

,

-

isOlh.--

_""'11

~.

B injection during 20sec at 1S"C

5.7 ..e.

iso'h

c.

Cs

cr

5.ZIK

~.

c.

31

'"

II

A injection during cs sac. at 1S" C

1

lsoth.15­

Figure D2 Reconcentration of solute bands by what was later to be termed "cold trapping" [2]: A, split injection and isothermal elution at 75 DC; B, slow split injection over a period of 20 s with conditions as in A: peaks are broadened owing to the slow introduction of the sample into the column; C, split injection as in B, but at 25 DC column temperature, followed by ballistic temperature increase to 75 DC. 2 III of gaseous alkanones Cbutanone to nonanone I, split ratio 10:1. Reduc­ ing the column temperature during transfer of the sample from the injector into the column reconcentrated the initial bands of the solutes.

For many years, the lack of systematic investigation elevated splitless injection to the status of a miraculous te~hnique. Because discrimination was exclusively attributed to non­ linear splitting, it seemed logical that splitless injection would be free from it; discrimination arising as a result of selective elution from the syringe needle (Part Al was not yet recog­ nized. It should also be added that integrators were expen­ sive and rather unreliable devices at that time and quantita­ tive analysis was clearly less important than today. After 1980 broader studies on quantitation by splitless injection were reported. These concentrated on listing the reproducibilities obtained from particular test mixtures, rather than on investigating the sources of the problems. Most re­ sults were embarrassingly poor, driving the reputation of splitless injection to the other extreme. A closer study of these reports reveals, however, that many of the injectors used had fundamental shortcomings and the conditions ap­ plied (carrier gas flow rate, length of syringe needle, dura­ tion of the split/ess period) were far removed from those re­ quired. On the other hand, owing to the use of synthetic,

1.2. Historical Background

c

B

261

A

, Figure D3 Documentary evidence from the time (19741 it was discov­ ered that recondensation of solvent in the column inlet re­ sulted in reconcentration of solute bands. The use of this "solvent effect" enabled isothermal analysis at the tempera­ ture of injection. Ce-C g n-alkanes; splitless period, 25 s; col­ umn

clean test mixtures, severe problems, such as the matrix ef­ fects, were not even recognized. Description ofthe basic prob­ lems and of working rules to avoid massive errors followed only later.

2. How to Perform Splitless Injection 2.1. Basic Steps of Splitless Injection

Splitless injection is performed in three steps as shown in

Figure D4. 1

2

3

2.2. Closing the Split Exit

The split exit is closed and the sample injected into an essentially closed vaporizing chamber, where sample vapor expands rapidly. Vapor must be temporarily stored in this chamber because the transfer into the col­ umn is slow. During the following "splitless period", i.e. the time the split exit is closed, the carrier gas transfers the sam­ ple from the vaporizing chamber into the column. Its duration is commonly between 40 and 80 s. The split exit is re-opened to purge the remaining sam­ ple material from the injector. The recommended split (purge) flow rate is 10 to 20 rnt/rnin,

During the splitless period, the split outlet must be closed. Closure must be reliable to prevent losses of sample mate­ rial. With a leak of 1-2 ml./min more than half of the sample material would usually be lost. As discussed in Part E, two different concepts are used for regulation of the carrier gas and, furthermore, electronic sys­ tems differ from mechanical regulation. This also influences the closure of the split outlet.

262

0 2. How to Perform Splitless Injection

.'.'

Residual vapors discharged through split

outlet Split outlet closed

InJecllon

Sample transfer (splltless period)

Injector purge

Figure D4 Steps of splitless injection.

2.2. ,. Mechanical Pres­ sure Regulation

The mechanical system with pressure regulation in the gas

supply adjusts the flow rate through the split outlet by means

of a needle valve (see Figure C2). The gas flow can be

stopped by manually closing the needle valve or by means

of a separate (manual or automated) on/off valve.

Closure of the Needle Valve?

Needle valves are designed for regulating a gas flow, not for

shutting it off. They consist of a hardened steel needle which

fits accurately into a metallic seat. Both are subject to wear if

the valve is closed many times. When the needle is driven

into the seat with force (to ensure tightness), the seat is de­

formed, which damages fine regulation of the gas flow rate.

As wear progresses, still more force is required to close the

valve.

On/Off Valve

An on/off valve positioned downstream of the needle valve

is clearly preferable. For manual closure, a simple screw

cap at the exit of the needle valve with a soft O-ring serves

the purpose; tightening or untightening by half a turn is suf­

ficient. Such closure also obviates the need to re-adjust the

split flow rate after each injection.

Automated closure usually involves a pneumatic on/off

valve switched by pressurized air from a remote electric valve.

Pneumatic valves prevent contact of solvent vapor with elec­

tric devices.

Reproducibility of the Splitless Period

Automatic on/off valves are expedient, but not essential,

because there is no need for highly accurate timing of

the splitless period. The splitless period must be sufficiently

long to transfer practically all sample material into the col­

umn; variation by a few seconds should have no noticeable

effect on quantitative data because exponentially decreas­

ing amounts of material are transferred towards the end of

the splitless period.

2.2. Closing the Split Exit

263

Electronic Regulation

Electronic gas supply and split flow regulation systems, de­ scribed in Section E4, switch the split flow rate on and off by means of electric valves and the user no longer interferes.

2.2.2. FlowlBack Pres­

The mechanical flow/back pressure regulation system shown in Figure C3, introduced by Hewlett-Packard, does not toler­ ate closure of the split exit, because the flow rate delivered to the injector must be discharged - the inlet pressure would otherwise rise to the level upstream of the flow regulator. Instead of being stopped, the split flow is diverted around the vaporizing chamber. A solenoid valve situated upstream of the back pressure regulator re-routes most of the gas flow into the septum purge line and a link to the split outlet (Fig­ ure D5). It has not been claimed that this system had any advantages; it is merely the consequence of the design cho­ sen for split injection.

sure Regulation

c::::J!'-

,.........;;~ Septum purge

Row'during splhless period

COlumn~

Split outlet

Figure D5 During the splitless period, the mechanical flow!back pres­ sure regulator system from Hew/eft-Packard diverts the split flow through a bypass.

2.3. Purging the Injector

Why cannot we leave the split outlet permanently closed?

This would simplify the injection process and spare us the

problem of loss of sample material which results from pre­

mature opening. Such injection, performed with an injector

having no split outlet, is actually used. It is called "direct"

injection, and will be discussed in Section 08.1.

Purpose of Purging

After transfer of most of the sample material into the col­

umn, the injector is purged in order to feed clean carrier

gas.

To some extent the transfer of sample vapor from the injec­

tor to the column is a dilution process, i.e. it is never really

complete. There are, furthermore, primarily below the col­

umn entrance, dead volumes from which sample vapor

slowly diffuses to the column if the split outlet remains closed.

Hence the purge is needed to remove sample material which

is too difficult to transfer into the column.

Extremely Broad Solvent Peak

A permanently closed split outlet would seldom become

apparent through tailing solute peaks. If, however, detectors,

e.g. the FlO, are sensitive to the solvent, the far larger sol­ vent peak is a sensitive indicator for delayed transport into

UNNERSIDAD DE ANTIOQUlA

BIBLJOTECA CENTRAL

264

D 2. How to Perform Splitless Injection the column, because only its "foot" is observed in the chro­ matogram (Figure D6). When the split outlet is erroneously left closed, the solvent peak returns very slowly from full deflection after more than half an hour (see also Figure D1).

\ . - - - Split exil opened

Visible on chart paper

Figure D6

The injector must be purged after most of the solute mate­

rial has been transferred into the column: the small propor­

tion of the solvent left in the injector can cause the pen to

remain at full deflection for a long time. Opening of the split

exit removes the tail.

2.3. f. Duration of the Splitless Period

The duration of the splitless period is a compromise between two opposing interests. . 1 A narrow solvent peak requires a splitless period as short as possible. 2 The sample components must be transferred into the column almost completely. For high-boiling compo­ nents, in particular, long splitless periods produce more precise and more accurate results.

Effect on the Solvent Peak Width

The initial band width of the solvent corresponds to the du­ ration of the splitless period (the time solvent vapor is intro­ duced into the column). If solvent effects interfere, i.e. if the column temperature is clearly below the solvent boiling point, the solvent peak is usually broader. Its width is then determined by the evapo­ ration time of the solvent in the column inlet ("solvent trap­ ping" process) and the broadening which results from retar­ dation by "phase soaking" in the coated column. Sample transfer broadens the solvent peak only if its duration ex­ ceeds the width determined by solvent effects. The latter contribution can be determined by on-column injection.

Delayed Elution after Solvent Effects

If solvent effects interfere, chromatography of the solutes is delayed until solvent evaporation is complete; the early peaks are further retarded by phase soaking [4]. This means that peaks are eluted later than with split injection, which im­ proves their separation from the solvent peak. For this reason, the somewhat broad solvent peaks usually obscure

2.3. Purging the Injector

265

hardly more early peaks than do the narrower ones obtained by split injection (see also Section 7.4.1). 2.3.2. Purge Flow Rate Required Tailing Solvent Peak

Purge of the vaporizing chamber at the end of the splitless period requires a flow rate of 10-20 ml/min. The minimum purge flow rate can be determined by the shape of the solvent peak. The purge flow rate is reduced until the solvent peak becomes broader or shows in­ creased tailing, indicating that removal of residual sample vapor from the vaporizing chamber is too slow. The statement cannot be reversed: tailing solvent peaks do not necessarily indicate poor purging ofthe vaporizing cham­ ber. Accumulation of involatile sample by-products or sta­ tionary phase in the column inlet and other imperfections of the column [5]. as well as dead volumes in the detector, have the same effect.

Degrading By-Products

High purge flow rates are useful if samples contain a large amount of involatile by-products which are degraded in the injector. They reduce the proportion of the volatile degrada­ tion products entering the column and keep "ghost" peaks small.

2.4. Septum Purge

The septum purge was introduced by Kurt Grab in 1972 [61. As the name implies, it serves to keep septum bleed away from the vaporizing chamber and the column. Removal of septum bleed is particularly important if high injector tem­ peratures are used and injector heads are well heated.

Removal of Solvent Vapor

The septum purge was, however, introduced for another purpose which is specific to splitless injection: even if the vaporizing chamber is not overloaded with sample vapor (see below), some vapor diffuses backwards out of the liner into the septum area. Purging the vaporizing chamber at the end of the splitless period does not effectively remove these materials. If there is no septum purge, such vapor slowly diffuses back into the stream of carrier gas directed towards the column entrance. This results in broadened and tail­ ing solvent peaks.

Purge Flow Rate

Suitable septum purge flow rates range between 1 and 10 ml/min. As they are rather uncritical, they are usually adjusted just once for an intermediate carrier gas inlet pres­ sure. The needle valve can be replaced by a restrictor. New instruments often regulate a fixed septum purge of ca. 3-5 rnt/rnin. The design of the septum purge will be subject of Section E4.5.

2.5. "Ghost" Peaks from

Despite the use of a purge flow, the septum can generate "ghost" peaks, which are commonly eluted above 200°C from standard apolar columns. The corresponding com­ pounds are carried through the septum purge area and in­ troduced into the vaporizing chamber by the syringe

Septum Material

266

0 2. How to Perform Split/ess Injection

needle. Either they stick to the outer needle wall (oily mate­ rial) or they are contained in the small septum particles which are cut, like a Christmas cookie, from the septum by the nee­ dle (stuck in the orifice ofthe needle) and are then blown out of the needle by the sample when the needle is fully inserted. The size of these "ghost" peaks is poorly reproducible and often seems to depend on the sample solvent; particularly large peaks are experienced when benzene, toluene, and alcohols are used. Confirmation of the Source

Because there are other potential sources of "ghost" peaks, e.g. the sample, the solvent, and impurities in the car­ rier gas or in the gas supply lines (Section 07.3.6), tests must be performed to determine whether or not the observed "ghost" peaks arise from the septum. The test is based on comparison of split and splitless transfer of the "ghost" peak material. A normal splitless injection is repeated introduc­ ing the syringe needle before the split exit is closed. a) The split flow rate is increased to, e.a., 50 mllmin; b) the sample is withdrawn from the syringe needle into the barrel of the syringe and the empty needle is intro­ duced into the injector; c) the split exit is closed; d) the sample is injected and the syringe is withdrawn (fol­ lowed by the splitless period). Insertion before closure results in most of the material re­

sponsible for the "ghost" peaks being removed through the

split outlet. If the "ghost" peaks arise from material intro­

duced from the septum, they will now be reduced in size

by, approximately, the split ratio, i.e. a factor of at least

ten.

Standard Procedure

For manual injection, it is recommended that the split exit is

always closed after insertion of the syringe needle. During

the time required for closure the syringe needle is pre-heated,

as required for a "hot needle" injection.

For automated injection with a fast autosampJer this pro­

cedure is not possible. For autosamplers capable of perform­

ing hot needle injection it would be a desirable feature.

2.6. Septum Purge During the Splitless

Below, arguments for and against closing the septum purge

during the splitless period are discussed. The subject is of

interest, of course, only if there is a choice. If the septum

purge and the split line end in the same closing device, they

must be closed simultaneously. When, on the other hand, a

flow/back pressure regulated system (Figure 05) is used, the

exit cannot be closed.

Period

2.6.1. Arguments in Favor of Closing

Possible loss of sample material through the septum purge

is a reason for closure. losses through the septum purge

can occur only if the chamber is overfilled (Figure D7). This

occurs rather frequently, however. Sample vapor can be

2.6. Septum Purge During the Splitless Period

267

driven from the liner by the pressure wave from sample evaporation or by diffusion during the time-consuming trans­ fer into the column. Vapors rinsed Carrier gas supply

::::::::::::::::;;;;;;;;

_~,\¥% through septum

purge

Figure 07 Losses of sample material through the open septum purge when the vaporizing chamber is overfilled.

Discrimination against Volatite Material

Losses through the septum purge are larger for the volatile components, because high-boiling components are partly retained on surfaces in the vaporizing chamber and diffuse less rapidly. Hinshaw [7] found that discrimination against the volatile components increased with increasing septum purge flow rate. Results with the flow/backpressure regula­ tor system, involving a high gas flow rate into the septum purge and by-pass line, were correspondingly poor: for 2 ul, injections into a 2 mm i.d. liner, discrimination against n-Ca compared with n-C20 reached a factor of two. In this experi­ ment, the vaporizing chamber contained some glass wool, which increased discrimination because it reduced losses of the high-boiling components during overflow (Section D4.1.4).

Pressure Increase

The last argument in favor of closing the septum purge ap­ plies if the carrier gas is pressure-regulated. Expansion of the sample vapor in an essentially closed system causes the pressure to increase. This compresses the vapor, enhanc­ ing the capacity of the vaporizing chamber to retain sample material, and increases the flow rate into the column, i.e. the rate of sample transfer. The effect is stronger the smaller are the accessible volumes around the vaporizing chamber into which the gas volume can expand (Section E2.2).

2.6.2. Sample Material Entering the Carrier Gas Supply Line

Closure of the septum purge exit during sample transfer pre­ cludes losses through this exit. The problem of the overfilled vaporizing chamber is not solved, however, because the sam­ ple vapor now penetrates the carrier gas supply line, which contains a considerable volume of compressible gas.

Loss of High Boiling Com­ pounds

Upon backflow the vapor rapidly leaves the heated part of the gas supply line and recondense on a cool wall. When

268

D 2. How to Perform Splitless Injection

the flow direction changes back to normal, the most volatile solutes re-evaporate and probably still reach the column during the splitless period. Somewhat higher-boiling mate­ rial returns to the vaporizing chamber only after the end of the splltless period and is mostly lost through the split out­ let. Still higher-boiling compounds do not return at all and contaminate the gas supply line. Occasionally they become visible again in later chromatograms as "memory effects". Such phenomena can prove puzzling and are described in some detail to save the analyst a negative experience. Test for "Memory Effects" from the Gas Supply Line

When he observes "memory effects" and can rule out the syringe as the source, the analyst is likely to proceed as fol­ lows: he repeats the run with all the manipulations involved, including closure ofthe valves, but without injecting a sam­ ple (instrument blank). Typically, he will no longer find "memory effects" and conclude (erroneously) that the "memory effects" did not arise from the gas chromatograph. He then injects the solvent used for his sample, with the result that the" memory effects" are:back. Most probably he infers that the solvent has become contaminated with the sample and pours it to waste. He will be very surprised to discover that solvent from a new bottle is no better. Did the impurities originate during production of the solvent? Even a completely different solvent is, however, likely to pro­ duce the same peaks. At this point the search becomes des­ perate, because neither the instrument nor the solvent seems to be the cause. It can come as another surprise to discover that a small reduction in the volume injected strongly re­ duces or even eliminates the "memory effects".

Source of the "Memory Effects"

This riddle is answered as follows. The sample volume injected overfilled the liner. The material causing the "memory effects" is of sufficiently low volatility to remain on the cool wall of the carrier gas line, which is the reason why the "instrument blank" did not reveal its presence. Another injection of an excessive amount of sample or sol­ vent caused vapor to condense in the cool zone and re­ dissolve the deposited material. Then the carrier gas pushed the resulting solution towards the heated part of the gas line, where the solvent and some of the dissolved mate­ rial evaporated and reached the injector. As only a limited amount of the deposited material was moved sufficiently close to the heated zone, the "memory effects" can be ob­ served many times.

2.6.3. Reasons to Leave the Septum Purge Open

In the event of an injection overfilling the liner, a septum purge flow rate of several mLJmin prevents contamina­ tion of the carrier gas supply line. Sample vapor is flushed away before this can happen. Immediate loss is probably preferable to partial loss in the gas supply line and subse­ quent "memory effects".

2.6. Septum Purge During the Split/ess Period Vapor Entering Split Line

269

An active septum purge helps to limit the pressure increase during injection, which can be advantageous: if the split out­ let line has a considerable internal volume, pressure increase causes sample vapor to be pushed into it. As the pressure decreases again, some vapor returns, but the high-boiling solutes are retained in this usually contaminated line.

3. Sample Volumes Suitable for Splitless Injection Description of the maximum sample volumes suitable for splitless injection is split into two parts: this section deals with the classical concept, Section D4 with the injection of large volumes using special techniques. Expanding Solvent Vapor

In classical splitless injection, performed with an empty liner, the maximum sample volume is determined by the require­ ment that it must be possible to store the vapor until it is transferred into the column. This is a more severe problem than generally accepted. De Veauxand Sze/ewski[81 confirmed by a statistically designed experiment that the size of the vaporizing chamber and the sample volume are among the most critical factors.

No Simple Guidelines

It is, unfortunately, impossible to give simple guidelines, because maximum sample volumes depend on the volume and design of the vaporizing chamber, the sample solvent used, the inlet pressure, and other factors.

3.1. Calculated Volumes of Solvent Vapor

The volume of vapor formed inside the injector is largely determined by the sample solvent. Table D1 lists calculated Table D1 Volume of undiluted vapor generated by 1 !1L of solvent, calculated for an injector at 250°C and a carrier gas inlet pressure of 30 kPa.

Solvent Pentane Heptane Toluene Diethyl ether Dichloromethane Chloroform Methanol Water

Vapor volume [!1Ll

300 210 260 300 450 320 750 1500

270

0 3. Sample Volumes Suitable for Splitless Injection values for 1 III of solvent, not considering dilution by the

carrier gas. Vapor volumes depend on the conditions: the

higher the injector temperature, the more the vapor expands,

whereas an increased inlet pressure compresses it.

Chromatographic software sometimes includes a calcu­

lator enabling the determination of vapor volumes for a range

of solvents and conditions.

1 III of the most important solvents produces 250-450

III vapor. Solvents of low molecular weight, such as metha­

nol and water, expand to far larger vapor clouds. That pro­

duced by 1 ttl water overfills most of the splitless injectors

currently available, even if dilution by carrier gas is neglected.

Number of Molecules per Volume of Liquid

The volume of vapor depends on the number of molecules

contained in 1 III of liquid, which in turn is given by the mol­

ecular weight of the solvent (the smaller the molecular

weight, the larger is the number) and its density. Heptane

produces less vapor than pentane because of the larger

molecular weight; the vapor volume of dichloromethane is

large because of its high density.

3.2. Determination of Injector Capacity

Calculated vapor volumes are of limited usefulness because

we do not know the extent to which clouds of sample vapor

are diluted by carrier gas. This information can be obtained

by performing the following experiments.

3.2.1. Determination

Splitless injections of a test sample are performed twice, with

widely opened and closed septum purge exits, and the sam­

ple volume is increased until peak areas differ. With the exit

open, overflowing solute material is lost and peak ar­

eas are reduced. This test is not possible with flow/back pres­

sure regulation because the septum purge exit cannot be

closed.

The experiment can be overcritical: slight overloading of the

injector does not result in losses when the septum purge is

closed.

from Peak Sizes

Conditions

Because solvent recondensation in the column inlet ac­

celerates the sample transfer, the injector capacity is found

to be increased. To avoid this effect, the column tempera­

ture should not be more than 10° below the solvent boiling

point. Reconcentration by cold trapping then requires that

the test component be eluted at least 50° above the col­

umn temperature during injection. It should, furthermore,

be sufficiently volatile to return from the carrier gas supply

line without substantial retention. For instance, a solution of

1 III n-tetradecane in 20 ml of the solvent of interest is suit­

able (FID).

3.2.2. Detection of Solvent in the Septum Purge

A more rapid test [9] is based on determining whether or

not solvent is flushed through the septum purge exit, assum­

ing that similar proportions of solute material would over­

flow. The solvent is detected by means of a flame.

3.2. Determination of Injector Capacity

271

The experiment presupposes an instrument which enables the septum purge to be kept open while the split exit is closed. The septum purge flow rate is adjusted to ca. 10 mllmin. The exit is equipped with a short piece of 0.32 mm i.d. fused silica capillary (Figure 08). Large dead volumes should be avoided, because they delay the arrival of vapor at the flame. Solvent recondensation within the outlet also provides an incorrect (delayed) picture. Needle valve Flame Fused silica capillary

Septum Septum purge

outlet

Gas supply

Vaporizing chamber Expanding

vapors

Figure DB Determination of injector overflow by detection of solvent vapor in the septum purge effluent (a flame serving as de­

tector).

Solvent Detection by Means ofa Flame

Solvent vapor is detected by lighting the septum purge ef­ fluent. Hydrogen as carrier gas burns as a colorless flame, visible merely as a glow at the tip of the fused silica capil­ lary. The flame turns yellow and larger on combustion of organic material. It is extremely bright with benzene or tolu­ ene; dichloromethane and methanol are more difficult to see. To detect water, the addition of ca. 20 % of an organic sol­ vent, such as propanol or acetone, it is recommended. If helium is the carrier gas, a small flame (e.g. that of a small candle) is positioned just below the exit of the fused silica capillary to ignite the effluent at the moment it contains sol­ vent vapor.

Column

The carrier gas inlet pressure should be appropriate for the column in use. Normally discharge into the column is slow and has little influence on the results. If, however, the vapor recondenses in the column inlet, transfer is accelerated and larger sample volumes can be injected before injector over­ flow is observed.

Procedure

The split exit is closed and solvent is injected. If the vapor overfills the liner the flame turns yellow shortly after the plunger is depressed. The first thrust of vapor results from the pressure pulse initiated by solvent evaporation. The flame rapidly gets smaller again and then persists for a few tens of seconds as vapor diffuses backwards from the va­

272

0 3. Sample Volumes Suitable for SplitJess Injection porizing chamber into the purge gas stream. Finally, the split exit is opened to flush the vapor from the vaporizing cham­ ber. When injections are performed by the "bot needle" tech­ nique, the flame turns slightly yellow during introduction of the needle, because the small amount of solvent coating the internal needle wall evaporates and the first vapor is eluted into the septum purge zone.

Parameters of Interest

The experiment can be used to test the following conditions. 1 The maximum volume of sample which does not cre­ ate significant overflow. 2 This maximum sample volume is determined for a vari­ ety of solvents. 3 When a narrow bore liner is used, vapor overflow be­ comes massive. When a short (5 ern) needle is used, losses are more 4 pronounced than with a long (7 cm) needle (unless a fast autosampler is used). 5 Increasing the inlet pressure compresses the vapor cloud and also accelerates transfer into the column, re­ ducing overflow. 6 Increasing the injector temperature causes expansion of the vapor and enhances the overflow. 7 Solvent recondensation in the column inlet can sub­ stantially reduce overflow. 8 Expansion of the vapor is usually not reproducible: the size of the flame varies when identical injections are re­ peated.

3.2.3. Measurement of Losses through the Septum Purge

By use ofthe system shown in Figure D9, part ofthe septum purge effluent was fed into an FID [10] to quantitate losses resulting from injector overflow. Seplum purge line Carrier gas

~

10FID

VaporiZing chamber

Capillary column

Figure D9 Quantitative determination of injector overflow by feeding part of the septum purge effluent into an FID. (From ref. [10].1

3.3. Results

Injection volumes were increased in steps of 1 Ill, starting with 1 III (needle volume; manual hot needle injection) and proceeding up to 4 III (reading "3Ill" on the barrel of the

3.3. Results

273

syringe). losses of solvent vapor were calculated as propor­ tions of the last microliter added. For instance, the areal amount observed on injection of 2 III was subtracted from that of 3 III to give the loss resulting from injection of the third microliter.

3.3. .,. Pressure Wave versus Diffusion

There are two mechanisms of escape of sample material from the liner which should be distinguished because the final conclusions are different.

Pressure Pulse

Expansion during solvent evaporation can drive vapor out of the injector and is registered at the septum purge exit as a sharp signal, as shown in Figure 010A for 1-3 III injections of hexane into a 80 x 2 mm i.d, vaporizing chamber (internal volume 250 Ill). 3

~I

B

A 2 IJI

3

~I

1 IJI

w·­

Figure D10 Signals recorded at the septum purge exit after injecting 1, 2, and 3 III volumes of hexane; 71 mm syringe needle. A, sharp peaks resulting from vapor expelled from an 80 x 2 mm i.d. liner. B, broad peaks from vapor diffusing backwards from a 80 x 4 mm i.d. liner. (From ref. [10].)

Back Diffusion

The second type of loss occurs as a result of back diffu­ sion out of the vaporizing chamber. It is slower and the resulting signals are correspondingly broad. Their tail reflects the decreasing amount of vapor leaving the chamber as more is transferred into the column. Figure 0108 shows peaks obtained from 1-3 III injections of hexane into an 80 x 4 mm i.d, (1 ml) liner. Forthe 1 III injection, backflow resulted solely from back diffusion. For larger injections the two types of loss overlap, resulting in a sharp peak at the beginning fol­ lowed by a broad tail.

3.3.2. Volume of the Vaporizing Chamber

The lower curves in Figure 011 show losses observed when an empty 80 x 4 mm i.d. (1 ml) liner was used. Upon injec­ tion of 1 III of hexane (needle volume), 8 % was lost, solely as a result of back diffusion. 2 III resulted in loss of 22 % of

274

D 3. Sample Volumes Suitable for Splitless Injection the extra material injected, most of it again by back diffu­ sion. Under the conditions used (50 kPa, injector at 250°C), 2 III of hexane produces 420 III of vapor. The overall loss of 15 % indicates dilution with carrier gas by a factor ex­ ceeding two. For a 4 III injection, producing ca. 850 III vapor, more than half of the last microliter was lost. 100

250 III liner

7 em needle

~

...J . ::L ~

c;; c

75

s :g

1 rnt, liner

E 50

.g

7 em needle

lJl

lJl

.3 25

3 III 4 III Sample volume

Figure 011 Losses of sample material by back flow from the vaporizing chamber: dependence on the size of the vaporizing chamber and the length of the syringe needle. Column temperature, 80°C; hexane. (From ref. [10].)

With the 80 x 2 mm i.d. (250 Ill) liner, losses from a 1 III injection of hexane increased to 14 %. This amount is rather small bearing in mind that the volume of vapor corresponded almost to that of the injector. It tells us that in this narrow bore tube dilution of the vapor with carrier gas is weak (re­ duced turbulence during expansion of the vapor). When an extra microliter of hexane was injected, almost none of this material was retained in the injector. It is useless to inject more than a needle volume (1 Ill) into an 80 x 2 mm i.d. liner.

3.3.3. length of the Syringe Needle

When the sample is nebulized (as in the experiment de­ scribed here), the length of the syringe needle determines the location of the center of sample evaporation within the vaporizing chamber. With a fast autosampler, the band of liquid travels until stopped, i.e. the center of evaporation is determined by the position of the packing or obstacle.

Principle

Vapor expands primarily backwards, because it cannot dis­ place the carrier gas in the bottom of the chamber. Hence to make best use of the volume available the sample should

evaporate near the bottom of the chamber. Results

With the 80 x 2 mm i.d. (250 Ill) liner and a 51 mm needle, losses upon injection of 1 III of hexane reached 35 %, com­ pared with 14 % for the 71 mm needle. With the 80 x 4 mm

3.3. Results

275

i.d. (1 mL) liner, losses increased from 8 to 21 % for the first microliter and from 22 to 35 % for the second (Figure 011).

Reduced Standard Deviation

Snell et al. [11] confirmed the importance of using long sy­ ringe needles with the finding that standard deviations de­ creased by a factor of ca. four. They found that 34.5 % less 2,6-di-tert.-butyl-p-cresol and 45 % less cyclohexanone reached the column when a 51 rnrn, rather than a 71 mm, needle was used. Hence the high standard deviation was the result of poorly reproducible losses. 1.9 ~L volumes (includ­ ing the needle) of hexane solutions were injected into an 80 x 2 mm l.d, liner.

3.3.4. Inlet Pressure

Figure D12 shows the effect of inlet pressure and column temperature on the losses resulting from use of an 80 x 4 mm i.d. (1 mL) liner and a 71 mm syringe needle. A higher inlet pressure compresses the size of the vapor cloud and increases the column flow rate, i.e. the transfer rate. Both result in an increase in injector capacity. Losses from the second microliter of 2 ~L hexane injections decreased from 52 % at 30 kPa (column flow rate, 2 mLlmin) to 8 % at 75 kPa (5 mLlmin). 100

------------------------.---------------­

~

...J :::L

:: 75 co c:

30 kPa, 80°C 50 kPa, 80°C

o

<=

~

E 50

g

~

75 kPa, 80°C

25 50 kPa, 25°C

3IJl

4IJl

Sample volume

Figure D12

Losses resulting from backflow from the vaporizing cham­

ber: dependence on gas inlet pressure and column tempera­

ture (recondensation in the column inlet). (From ref. [10).)

3.3.5. Solvent Reconden­ sation

Recondensation of the solvent in the column inlet strongly increases the rate of transfer into the column (see also effects on the split ratio, C8.3.3, and on splitless transfer, 05.5). It requires a column temperature at least 20° below the sol­ vent boiling point. When hexane is injected at 25°C (instead of the 80 °C chosen for the data shown above). losses from the second microliter were reduced from 22 to 7 %, and those from the fourth microliter from nearly 60 % to 15 %.

3.3.6. Volume of Vapor from Solvent

The volume of vapor per microliter depends on the solvent, as shown in Table 01. Methanol, for instance, produces more

276

0 3. Sample Volumes Suitable for Splitless Injection than three times as much vapor as hexane (smaller molecu­ lar weight and higher density). As shown in Figure D13, losses through the septum purge exit are correspondingly higher. When the volume of methanol injected into the 1 mL liner was increased from 1 to 2 ~L, only ca. 0.2 ~L of the extra material remained in the injector, i.e. 80 % was lost. Even from the first microliter nearly 30 % was lost. 1 Jll metha­ nol behaves like 3 Jll hexane. 100

-------------------------------------~-----x

~

-' :L ~

Methanol

75

Iii c: .9

Hexane

~

~ 50

g ljl o -'

25

1~

2~

3~

4~

Sample volume

Figure D13 Losses through the septum purge exit for hexane and meth­ anol. 80 x 4 mm i.d. liner; 71 mm syringe needle; 50 kPa inlet pressure; 80°C column temperature. The volume of vapor produced from methanol is slightly more than three times that from hexane. (From ref. [10].)

Mixing with Carrier Gas

Results for methanol with the 1 mL liner were worse than those for hexane and the 250 ~L liner, despite the four times larger liner capacity for a vapor volume only three times larger. This is another indication that dilution with carrier gas rapidly increases as the liner is widened and means that en­ larging the diameter ofthe liner is not very effective means of increasing capacity. Elongation is more effective.

Solvent Effects on Response Factors?

The large effect of the nature of the sample solvent on injec­ tor overflow has led to the suggestion of "solvent effects on response factors" [12). Puzzling results were reported which seemed to show that absolute and relative peak areas of polynuclear aromatic hydrocarbons (PAHs) de­ pended on the solvent used. With acetonitrile or metha­ nol as solvent peak areas were between 26 and 97 % of those obtained with toluene; with isooctane they were up to 56 % larger. A closer look at the conditions revealed that "3 ~L" volumes were injected into a liner with an internal volume of less than 400 ~L and that the syringe needle reached the center of the liner only, i.e. even the small liner volume available was ex­ ploited only partially. As 3 ~L methanol generate more than 2 mL vapor, the injector was hopelessly overloaded. For

3.3. Results

277

solvents forming smaller vapor clouds overloading was less drastic. This is the reason why the losses (and, ultimately, the peak areas) depended on the solvent used (13).

3.3.7. Liners with a Constriction at the Top?

Liners with a constriction at the top ("goose neck") have been advertized for splitless injection with the claim that these would prevent backflow from the vaporizing chamber. The bore of the constriction was ca. 0.8 mm, which fit the 22 gauge syringe needles of autosamplers and headspace syringes. Restriction, 2mm long

A

Restriction, 8mm long

c

B

Figure D14 Experimental liners with and without constrictions at the top.

Experimental Test

To check the effectiveness of such a constriction, the three liners of 3 mm l.d. shown in Figure D14 were compared. Losses are shown in Figure D15 for hexane and the condi­ tions used above (50 kPa, 80°C, 71 mm needle). Curve A: straight 565 ul, liner.

Curve B: liner B with a short (2 mm) constriction.

Curve C: liner C with a long (8 mm) constriction.

100

z ...J

75

n;

A C

.g

B

::L ~

<::

'0

al 50

E

g Ul

Ul

s

25

3IJl

4IJl

Sample volume

Figure D15 Losses of sample vapor (hexane) resulting from the use of the 3 mm i.d. liners shown in Figure D14. .

278

D 3. Sample Volumes Suitable for Splitless Injection Differences were significant, but not spectacular. 1 For the needle volume (1 Ill) losses with the long con­ striction amounted to only half ofthose of the open tube, but were small anyway. 2 For the 2 III injection losses from the second microliter were reduced from 70 % to 45-50 %. 3 For larger injections differences were no longer signifi­ cant.

Reduced Loss from Diffusion

Constrictions cannot prevent backflow of vapor resulting from the overpressure generated by excessive sample volumes, because the resistance against gas/vapor flow is low, but they can substantially reduce losses resulting from diffusion, be­ cause the latter depend primarily on the cross section of the escape route. They did, indeed, reduce losses as long as small volumes were injected, but the improvement was no longer significant for injections exceeding 2 Ill, because 2 III hexane produces ca. 420 III of vapor which, after dilu­ tion with carrier gas, is near the limit to overfill the liner of ca. 500 III internal volume (reduced by the constriction).

Pre-Evacuated Injectors?

The limited effectiveness of the restriction was also criticized by Hammar [141. who suggested evacuation of the vaporiz­ ing chamber before splitless injection to avoid dilution with carrier gas. Hinshaw [7] reported some puzzling results, but since he compared a packed straight liner with an empty one with constrictions at both ends, the results were determined by the packing rather than by the constrictions.

Alternative: Closure of the Septum Purge

loss resulting from diffusion into the septum purge was clas­ sically prevented by closing the septum purge outlet during the splitless period. In fact, jf the vapor is not removed it is likely to return into the vaporizing chamber. This most sim­ ple measure is, unfortunately, not possible with most of the instruments currently in use. Hence liners with a constric­ tion at the top are useful for systems which do not enable closure of the septum purge; they are unnec­ essary for others.

3.3.8. Valve to prevent Backflow

Kaufmann [15] suggested preventing backflow from the va­ porizing chamber by installing a valve above the liner. It closes upon pressure increase in the chamber during expan­ sion of the sample vapor and opens again when pressure returns to normal.

Jade Valve

On a Hewlett-Packard 5890 instrument he modified the re­ movable injector head by adding a Jade airlock just above the liner. The Jade valve was actually conceived as a replace­ ment ofthe septum and consists of two balls which precisely fit into a seat, but can be moved, e.g. by the syringe needle during injection. The carrier gas entered above the valve, but below the septum purge zone. With a slight overpres­

3.3. Results

279

sure it made its way through the additional valve into the chamber. The valve closed upon development of overpres­ sure in the vaporizing chamber. Fast Injection

Closure of the valve presupposes that the syringe is with­ drawn rapidly because the needle keeps the valve open and might prevent it from closing at exactly the most critical moment. Because sample evaporation takes at least a few hundred milliseconds, this can be achieved by use of an autosampler.

Test Experiment

Kaufmann used a 2 mm i.d. liner and tested the device by injecting 5 III volumes of an acetone solution. 1.25 mL vapor was generated, five times exceeding the internal volume of the chamber. Peaks were massively larger than forthe stand­ ard injector without the valve, although not all by the same extent. There was still strong discrimination against vola­ tile and higher-boiling components.

Displacement of the Carrier Gas

Closure of the vaporizing chamber at the top requires a fun­ damental change of concept. because the carrier gas can no longer be displaced backwards into the gas supply or septum purge line to provide room to the expanding vapor. If the center of sample evaporation is in the lower part of the liner, the carrier gas in the upper part of the chamber is trapped and the vapor is driven into the elastic volumes of the split outlet, from where only volatile solutes will return to the column. Perhaps the center of evaporation should be po­ sitioned near the top, just below the valve, filling the cham­ ber from the top towards the bottom.

3.4. Pressure Increase during Splitless Injection

Hewlett-Packard introduced electronic pressure control and its use to increase the carrier gas inlet pressure during the splitless period I" pressure pulse", Wylie et et., [16]). The cloud of sample vapor is compressed, which increases the liner capacity.

Required Increase

To have a noticeable effect the pressure increase must be substantial. When the absolute pressure is doubled, the volume of vapor is halved, i.e. twice the volume of sample can be injected into a vaporizing chamber of given size. If the inlet pressure was initially 100 kPa (200 kPa absolute pres­ sure), it must be increased to 300 kPa (400 kPa absolute). Hence a large increase in injector capacity can be achieved only for applications involving low inlet pressures, e.g. when hydrogen is used as carrier gas. The enhanced injector capacity is, mostly, merely a positive side effect of the accelerated transfer into the column which also results from the pressure increase. This technique will be discussed in Section 05.4.

3.4.1. Auto-Regulation1

Injection into a hypothetical closed chamber, with a volume corresponding to the vapor generated by the sample, dou­

280

o

3. Sample Volumes Suitable for Splitless Injection bles the content and, hence, the absolute pressure. If the in­

let pressure is 100 kPa before injection, it increases to 300

kPa. This self-regulates the capacity of the chamber - theo­

retically the chamber has infinite capacity: when more is

injected, pressure is higher.

The pressure increase also accelerates transfer into the col­

umn and, again, as more is injected, the transfer becomes

more effective. At the end of transfer the pressure automati­

cally returns to normal. This seems an attractive concept,

but today's injectors are not built this way.

Small Volumes around the Vaporizing Chamber

Back pressure regulation immediately discharges a vol­

ume corresponding to that of the vapor formed by the sam­

ple. Hence, there is no gain.

Pressure-regulated systems are essentially closed during

sample evaporation, because the exits of the split and the

septum line are closed and the pressure regulator is tight

against backflow; the column is the only outlet. Some pres­

sure increase is, indeed, observed, but the effect is weak,

often merely 5-10 kPa, because the volumes around the va­

porizing chamber accessible to expanding gas and vapor are

several times larger than the volume of the vaporizing cham­

ber. If all the volumes indicated in Figure D16 are added,

they typically total 10-15 mL.

Carrier gas======ij-:I

i = = = =

1

regulation, supply line

~:gt~~ti~~~ethe

"\=====

["

Filter, regulation of split flow rate

Space between the liner and the injector body

Figure D16 The accessible volumes around the vaporizing chamber are ca. ten times larger than the volume of the chamber itself.

A self-regulating pressure pulse resulting from sample evapo­ ration would provide a simple and efficient tool for splitless injection. The volumes around the vaporizing chamber would, however, have to be clearly smaller than the vaporizing cham­ ber itself. This is difficult to achieve.

3.5. Slow Injection?

We originally suggested adjusting the rate of injection to the rate of transfer into the column, with the aim of immediately transferring the vapor generated into the col­ umn and thus increasing the volume that can be injected [171. At a carrier gas flow rate of, e.a., 3 mLlmin, 50 IlLls gas

3.5. Slow Injection?

287

and vapor are transferred; if half the mixture consisted of sample vapor, ca. 0.1 ~lIs of liquid could be injected. At Varian, Yang et al. [181 seized upon this idea and proposed an injector with a very small vaporizing chamber (150 ~L). They recommended avoiding overloading of the liner by slow introduction of the sample. Needle Problems

This concept convinces only in theory, because during slow injection, the sample evaporates inside the syringe nee­ dleand leaves the high-boiling solutes largely on the inner wall. In fact, slow injection resulted in the worst discrimina­ tion against high-boiling compounds (Part A).

3.6. Conclusions

In splitless injection, the vapor of the injected sample must be stored inside the vaporizing chamber until it is swept into the column. The storage capacity depends on numerous fac­ tors, which hinders provision of simple guidelines. For in­ stance, high inlet pressures or strong recondensation in the column inlet enable injection of much larger volumes, and there are large differences between the volumes of vapor formed by different solvents. It would be rather simple, how­ ever, to provide the user recommendations by using soft­ ware which takes into account physical laws and some ap­ proximations.

Monitoring by Flame Test

The testing procedure with the flame at the outlet of the purge exit facilitates the rapid determination of whether or not the liner is overloaded. Experience shows that an injec­ tion volume producing a small yellow flame can be toler­ ated as losses are smaller when the septum purge is closed (provided it can be closed).

Closure of Septum Purge or Goose Neck Liner

To avoid losses by diffusion backwards from the vaporiz­ ing chamber, either the septum purge should be closed dur­ ing the splitless period or a goose neck liner, with a con­ striction at the top, should be used.

Recommendations on Sample Volumes

An 80 x 4 mm i.d. liner enables splitless injection of up to: 2 ~L for samples in most of the solvents commonly used ("1 ~L" read on the barrel of the syringe, if the needle content is transferred), 1.5 ~L ("0.5 ul,") for samples in chlorinated solvents (e.g. dichloromethane), and 1 ~L (the needle content) for methanol and ethanol. At inlet pressures above 150 kPa, these values can be doubled. An 80 x 2 mm l.d, liner enables splitless injection of up to 1 ~L ("0 ~L") of solutions in solvents such as hexane or ethyl acetate. Solvents producing larger volumes of vapor can only be injected at inlet pressures above 100-200 kPa. Splitless injection of aqueous solutions is problematic, because of the extremely large volume of vapor generated (among other reasons).

282

D 4. Injection of Large Volumes

4. Injection of Large Volumes For most trace analysis, injection of 1-3 JlL is inconven­ ient, because it complicates sample preparation: extraction must occur with less solvent than is desirable, reconcentra­ tion is required before injection, and extracts as small as 10 III must be handled to enable the analysis of at least a quar­ ter of the material available. There are possibilities of large volume injection by splitless injection, but the methods most commonly used are the on­ column/retention technique or PTV solvent splitting.

4.1. Overflow Technique

Visual experiments suggested that injector overflow might be acceptable if sample evaporation occurs from surfaces of a packing material and the sample does not contain highly volatile components.

4. 1. 1. Evaporation from Cool Surfaces

When a sample is nebulized and an excessive volume of vapor causes overflow from the vaporizing chamber, solute material is lost with the solvent. Evaporation from a pack­ ing, however, is fundamentally different: the solvent evapo­ rates first. The solutes are retained by the solvent (solvent trapping) and the low temperature (cold trapping); they are vaporized only after solvent evaporation is complete. Hence, it should be possible to vent solvent vapor without losing solute material. Vapor overflow

cil"

Carrier gas

~

';I" ~':

~

• II

a

II

b

II

c

'~."

{,11) d

Figure D17 Steps of sample evaporation in splitless injection of large volumes by the overflow technique. The surface of a pack­ ing material is shown as a small plate. (From ref. [19J.I

4.1. Overflow Technique

283

Only Solvent Vapor Escapes

The process is summarized in Figure D17. a The sample liquid is deposited on to surfaces, which presupposes that these are cooled to the boiling point of the solvent (packing of low thermal mass). b The solvent evaporates. The consumption of energy keeps the zone at the solvent boiling point until al­ most all the solvent has evaporated. c As the solute material is retained at the cooled site of evaporation, solvent vapor can be allowed to expand outside the vaporizing chamber and to escape through a widely opened septum purge exit. d After completion of solvent evaporation, the injector temperature is restored and the solutes evaporate. The carrier gas again flows into the vaporizing chamber and transfers the solute material into the column.

Temperature Measurement

Temperatures at the site of evaporation were measured by means of a fine thermocouple [20). During the first 1-2 s af­ ter introduction of 200 ~L hexane at 250 °e, the temperature read was 80 °e (Figure D18), i.e. corresponded to the boil­ ing point of hexane at the 50 kPa used. 2 s after the injection, 102 °e was measured. It took more than 20 s to restore the temperature. The same amount of 1-propanol initially cooled the evapo­ ration site to 106 "C (standard boiling point, 97 "C). After 2 s, the temperature was 110 "C and 20 s later it had reached 220 "C only. It increased more slowly because of the larger amount of energy extracted from the environment.

at the Evaporation Site

1:1

50 ~L propanol, 330 'C

~ ~

I-

240

.......-:~_ _- - - 200 ~L hexane, 250 'C

I

I

200

200 ~L propanol, 250 'C

160 J 120

1

~ L~,~~~--.~---,o

10

20

30

40 Time after injection

50

[sl

Figure 018

Temperatures measured in the center of the evaporation zone

after injection of the solvents and volumes indicated; injec­

tor temperatures, 250 and 330 °e. (From ref. [20).)

Fast Solvent Evaporation

These measurements showed that solvent evaporation is rapid: for a 200 ~L injection, the temperature remained low for a few seconds only. Warming the site of evaporation back to the regulated temperature, on the other hand, easily re­ quires more than half a minute. The duration of the splitless transfer period must be prolonged correspondingly.

284

D 4. Injection of Large Volumes

4.1.2. Injection Rate

Injection must occur faster than the solvent evaporates, to ensure that the site of evaporation remains cooled to the (pressure-corrected) solvent boiling point. If it is too slow, the solvent evaporates concurrently with its introduction, heat is supplied too rapidly to result in cooling to the boiling point, and the volatile components are not retained optimally. 100 III of a hexane solution containing n-alkanes was injected into a vaporizing chamber at 330°C [201. Introduction in 1 or 5 s produced almost identical results, but that in 10 s led to increased losses of the components up to ca. n-C25. This suggests that 100 III hexane evaporated within about

5 s. 4.1.3. Keeping the Liquid in Place

The packed bed in the vaporizing chamber must keep the injected sample liquid in place against rather strong forces resulting from violent evaporation. If, for instance, 200 III solvent evaporate in 3 s, vapor is formed at a rate of ca. 1000 rnt/rnln and escapes at a velocity of 5-20 m/s.

Mechanism of Expulsion

Visual experiments with perylene solutions showed the

course of events if retention of the liquid in the pack­ ing was too weak or the sample volume injected too large. Expanding vapor pushed liquid from the main evaporation site into the upper region of the packing. As this brought liquid into a hot zone, evaporation was accelerated. This gen­ erated even more vapor and further increased the forces. Vapor is formed primarily along the liner wall, the heat source. It must flow pastthe liquid sitting like a drop in the center of the packing (Figure D19). There are also channels (streams of vapor) through the liquid. material Packing

i~ij~1

Liquid driven upwards by vapor

\

Heat supply -- supporting / evaporation

"iii

3

Cii c:

1~~mll

1

Overpressure driving vapor upwards

::J

)

(

Split outlet

Column

Figure D19 The sample liquid in the packed liner must be kept in place to resist the strong flow of vapor driving. it upwards. (From ref. [19].)

4.1. Overflow Technique

285

Retention by Wetting Forces

The liquid is retained in compartments surrounded by pack­ ing material. The wetting forces keeping it in place. If wet­ ting forces per unit surface area are assumed to be similar for different packing materials, forces increase with increas­ ing surface area, i.e. the strongest forces are obtained from particulate packings which have small inter-parti­ cle compartments. If compartments are too small, however, the release of vapor is hindered.

Wool

A dense packing with silylated glass wool (6 cm x 4 mm) safely retained 50-100 ul, hexane. Retention of other solvents seemed more difficult. At 240°C it retained at most 40 ul, propanol; at 330 °C not even 20 ul.,

Tenax

On 20/35 or 35/60 mesh Tenax TA, retention of 200 ul, 1­ propanol was no problem, not even at 350°C. This suggests that particulate packings must be used if large volumes are to be injected. In a 5 mm Ld.liner (Carlo Erba/CE Instruments, Series 8000 GC) packed with a 6 cm plug of Tenax, 2000 ut, hexane or 1000 ul, propanol were safely retained. More prob­ lems are observed when water is injected, because it does not wet the Tenax surface [2'1]. Tenax is not suitable for dichloromethane solutions, because it dissolves in this sol­ vent. Particulate packings must be kept in place by dense plugs (ca. 5 mm long) of deactivated wool on each end, since the explosion-like evaporation tends to blast a cavity into the packing and to carry packing material into the column or septum purge line.

Channels

As the syringe needle penetrates the Tenax, channels are formed through the packing which are not refilled spontane­ ously afterwards and might facilitate expulsion of liquid. This and the accumulation of septum particles at the posi­ tion of the fully inserted needle tip suggest that the packed bed should be replaced regularly.

Maximum Sample Volume

The maximum volume of liquid safely retained by a packed bed of given size must be determined experimentally by injection of increasing volumes; peak areas of high-boiling components (for which loss by co-evaporation with the sol­ vent is ruled out) increase in proportion to the sample vol­ ume until overloading occurs. A sufficient safety margin must be included since expulsion from the packed bed is not re­ producible.

4. '.4. Retention of

The forces retaining the solute material at the site of evapo­ ration are generated by low temperature and solvation with sample solvent. The process is not, however, ideally organ­ ized. As solvent evaporation advances, an increasing pro­ portion of the solute material is deposited on to "dry" surfaces. This solute material is in danger of being carried

Volatile Components

286

D 4. Injection of Large Volumes away by vapor from still active evaporation, especially if the "dry" zone is already heated.

Losses

As shown in Figure D20, losses of volatile solutes depend on the sample solvent. The results refer to 50 III injections into a 4 mm i.d, liner packed with Tenax, kept at 330°C. Pen­ tane provided the best results. The low boiling point caused the evaporation site to be cooled to a low tempera­ ture. 1-Propanol, with the highest boiling point (98°C), was worst. The results also show that the losses depend on fac­ tors other than the boiling point: with methanol (b.p, 65°C) they were higher than with hexane (b.p. 69°C) because of poor solvation (retention).

l m400

~

~ 300 200

~/ penlane-~ / / Hexane

/

Methano

t-Propanol

Ethyl acetate

100 ­

o .L,

r-----"

12

-,-----,---~-._.,--...-"'~

16

20

,-----,-----:--------,--------,----'1

24

28

32

n-Alkanes(e-number) Figure D20 Dependence of solute losses on the sample solvent. 50 ilL injections at an injector temperature of 330 °e. (From ref. [20].) Increase with Increasing Sample Volume

losses of volatile solute material increase with sample vol­ ume. On increasing it from 50to 250 Ill, the loss of the n-C13 alkane increased from ca. 30 to 55 % and that of n-C17 from hardly 10 to 30 %. 500 III injections resulted in a substantial further increase of the losses (hexane solution injected into a 4 mm i.d. liner packed with 35/60 mesh Tenax, 330°C [20]).

4. 1.5. Desorption of Solute Material

The packing of the liner also influences transfer of the solute material into the column. Splitless transfer no longer means displacement ofthe gas phase of an empty chamber into the column, but a kind of chromatography in a liner behaving like a packed precolumn. If, for instance, the solute material is spread over a 40 x 4 mm plug of packing material at the bottom of the bed, the solute material furthest from the column must pass through ca. 500 III of packing. If the carrier gas flow rate into the colum n is, e.g., 2 ml./min, 500 III of packing is flushed 5-6 times a minute, enabling a component with a partition coefficient, k, of 5-6 to reach the column within a 1 min transfer period. The duration of the splitless period must be further prolonged by the time required to return the temperature of the evapo­ ration site to that of the injector. This is easily 30 s.

4. 1. Overflow Technique 4. 1.6. Instrumental Requirements

287

Splitless injection by the overflow technique can be per­ formed with any split/splitless injector, provided the split outlet and the septum purge can be regulated independently (Figure D21). The vaporizing chamber must, furthermore, have an internal diameter of 4-5 mm to facilitate flow of the vapor past the liquid sitting in the packed bed. Syringe

Carrier gas supply ---tE===;;:

.SeptumCUr.gf:

Glass wool

:';t~'~~~i?

Vapor overflow Expanding solvent vapors

Sample liquid

)(

Split exit

Column

Figure D21

Injector for splitless injection with overflow. (From ref. [19J.)

Pneumatics for Extremely Large Injections

Conventional carrier gas supply and septum purge are suit­ able for injection volumes of 100 III at least. For substan­ tially larger volumes, a modification ofthe type described in [22] was used. During solvent evaporation, a switching valve interrupted the carrier gas supply and widely opened the septum purge exit. This ruled out flow of vapor into the carrier gas supply, enabled wide opening of the septum purge without loss of carrier gas, and minimized pressure (hence also temperature) during solvent evaporation.

4.1.7. Syringe Needles

The length of the syringe needle determines the point in the packed bed where the sample liquid is released. Since the vapor expands primarily upwards, most of the packing material should be situated above the site of initial evaporation. Needles 71 or 80 mm long were used, depos­ iting the sample ca. 25 mm above the bottom of the liner. 50, 100,250,500, and 1000 III syringes equipped with such nee­ dles are not standard, but available on request, e.g., from Hamilton. Needles with side port holes are preferable be­ cause they reduce the amount of septum particles introduced into the packing.

4.1.B. Flow Rate through the Septum Purge

The septum purge exit must be sufficiently widely open to enable ready escape of the vapor expanding backwards from

288

0 4. Injection of Large Volumes

the vaporizing chamber. If the rate of expansion exceeds that of discharge, the pressure rises and vapor penetrates the carrier gas supply system. This becomes a severe prob­ lem if vapor reaches the manometer and pressure regulator. When a septum purge flow rate of 120 rnt./rnin was set, prob­ lems were encountered only when 1 mL volumes were in­ jected too quickly. 4.1.9. Column Tempera­ ture During Injection

Column temperatures well below the pressure-corrected solvent boiling point promote strong solvent recondensa­ tion in the column inlet and, as a consequence, severe flood­ ing. If, for instance, 50 IJ-L of solvent recondense, the liquid penetrates the column by approximately 6 m. As in conventional splitless injection, transfer into the col­ umn is slow and the initial bands of the solutes are corre­ spondingly broad. Reconcentration is required, by means of either cold trapping (Section 07.3) or solvent trapping (Sec­ tion 07.4). With large volumes, solvent trapping is delicate, because some recondensation is needed, but it should not be so excessive that it results in severe flooding. As a rule of thumb, the oven temperature should be ca. 20°

Guideline

above the standard solvent boiling point if inlet pres­ sures up to 100 kPa are applied, and 30° above at inlet pres­ sures up to 200 kPa.

4.1. 10. Examples

Figure D22 shows AFIO (NPO) chromatograms of some tri­ azine herbicides in 1-propanol, obtained from a 12 m x 0.32 mm i.d. column coated with Superox 0.6 (a high mo­ lecular-weight polyethylene glycol) by use of a 4 mm i.d.liner with Tenax TA at 280°C. The 50 IJ-L volume was injected in

50 IJI Injection 200 IJI Injection 2 IJI Injection

L

120 Qq+--Progr. 5 Q/min-I 150 QC 230 QC

Figure D22 Triazine standards in 1-propanol: 2 and 50 IJ-L injections of the same solution; the solution for the 200 IJ-L injection was diluted fourfold. Prop. propazine; Terb. terbutylazine; Atra. atrazine; Prom. prometryne; Sima. simazine; Dese. desethylatrazine; Desi. desisopropylatrazine. (From ref. [19].)

4.7. Overflow Technique

289

less than 1 s, 200 ~L in 2 s. The septum purge flow rate was

set to 120 ml/min, but reduced to 10 ml/min 1 min after

injection. The splitless period lasted 4 min. The 2 ~L and 50

~L injections of the same solution demonstrate the gain in

sensitivity by a factor of 25. For the 200 ~L injection, the so­

lution was diluted fourfold.

Mineral Oil on Jute

Forthe determination of mineral batching oil on jute bags

used for packing foods such as rice, coffee, or cocoa beans,

20-50 ~L of raw extracts of the fabric were injected on to a 3

m x 0.25 mm i.d. capillary column [231. The hydrocarbons of

interest ranged from n-C16 to n-C32'

4.2. Precolumn Solvent

The main obstacle to overcome in large volume injection is

the large volume of solvent vapor. Overflow solves the

problem by rapid discharge through a septum purge line of

low resistance.

Splitting

Column Flow Rate Too Low

Discharge through the column is difficult because column

flow rates are too low. Even when extremely high (e.g. 10

ml/min) and assuming no dilution with carrier gas, the vapor

of merely ca. 50 ~L hexane or 25 ~L dichloromethane can be

discharged per minute. If the sample were injected at a rate

adjusted to the discharge of the vapor, at most some 0.4-0.8

~L of liquid could be injected per second. This is inconven­

ient and creates problems relating to evaporation inside the

needle. Hence discharge must be accelerated by other

means.

Early Vapor Exit

In 1994, Suzuki et al. [241 suggested the discharge of the sol­

vent vapor through an early vapor exit, i.e. an outlet posi­

tioned between an uncoated precolumn and the sepa­

ration column (Figure D23). This enabled injection of more

than 200 ~L of a solution in dichloromethane.

The early vapor exit was introduced in 1988 for on-line LC­

GC by on-column techniques [25,261. A precolumn, either

coated ("retaining precolumn") or uncoated ("retention

gap") is mounted upstream of a press-fit V-connector which

-.l)"lical Col.

Cold trap Col.

Y-cOl'lllle<:tor

Solvent. divenion col. Oven

Figure 023 Splitless injection with an early solvent vapor exit ("solvent diverting column n ) . (From Suzuki ef al. [24].)

290

0 4. Injection of Large Volumes feeds to the separation column and to the solvent vapor exit. The exit is open during solvent evaporation and almost closed during analysis (leaving a small purge flow).

Instrument Configuration

The injector of a Hewlett-Packard instrument was equipped with a 4 mm l.d, liner. The 3 m x 0.53 mm i.d. precolumn was deactivated, but uncoated. Between the press-fit V-piece and the electromagnetic valve at the vapor exit was a 2 m x 0.53 mm i.d. "solvent diversion column".

Conditions

The injector was at 250°C and operated with a splitless trans­ fer period of 1 min. The column head pressure (helium) was 100 kPa with the vapor exit closed and decreased to 48 kPa when it was open (flow regulation). The gas flow rates through the split outlet and the septum purge were 60 mLJ min and 2 mLJmin, respectively. When the vapor exit was open, the flow rate through the split outlet collapsed to less than 0.01 mLJmin. The samples were injected manually at a rate of ca. 2.5 IlLJmin. This resulted in concurrent evapo­ ration, i.e. discharge of the vapor at the same rate as it was formed. The vapor exit was closed 5 s after the injection was complete. The split ratio in the V-piece was such that less than 1 ~L dichloromethane entered the column (flame pho­ tometric detection!).

Retention of the Solutes

Retention of the solutes, alkylated phosphates, was opti­ mized by variation of conditions such as inlet pressure, injection rate, and oven temperature. The system proved robust for the solutes of interest, as conditions could be var­ ied within a wide range without loss of components.

Pesticide Analysis

In a further paper [271, the authors described the application of this method to the analysis of pesticides in water. Splitless injection was preferred to on-column injection to avoid col­ umn contamination with non-evaporating material. In addition to the system described in the first paper, a 3 cm x 0.25 mm l.d. "regulation column" was used between the V-piece of the vapor exit and the solvent diversion column (2 m x 0.53 mm i.d.). It was coated, whereas the solvent diver­ sion column was uncoated but deactivated. The geometry of the vapor outlet was explained by a required difference between the resistances determining the ratio of the flow rates in the V-piece. 100 ~L hexane solutions were injected manually at 2.5 /lLJs and an oven temperature of 40°C. Se­ lection of conditions enabling full retention of the pesticides in the uncoated precolumn is discussed.

Open Questions

From later experience the following questions arise: Vaporization of the sample liquid in the hot cham­ 1 ber is a critical step: it tends to be irregular and violent. Suzuki et al. did not mention how they dealt with this aspect. According to Boderius et al. [281, formation of

4.2. Precolumn Solvent Splitting

2

3

291

droplets must be prevented by direct introduction ofthe liquid into a packing material. Why was the precolumn uncoated? As solvent recon­ densation was not involved, a retaining precolumn would have retained the solutes better. The design ofthe vapor outlet (columns for regulation and solvent diversion) seems more complicated than necessary. It might be because of the peculiarity of Hewlett-Packard instruments delivering the carrier gas through a flow regulator. Alternative systems involve pressure regulation.

4.3. Evaluation

Large volume injection is having a hard time being ac­ cepted, although there seems to be no debate about the util­ ity and advantages of such a tool. 250 I!L PTV solvent split injection was described at the end of the seventies, 100 I!L on-column injection in the early eighties, but both found lit­ tle response. Both have their limitations, but also a broad range of applications with robust performance. Is there a need for large volume splitless injection (or direct injection, see Section D8.1.3) beside these well elaborated PTV and on­ column techniques? Could it even be that a confusingly broad choice of options makes potential users hesitate?

Standardized Method

Standardization of methods has the severe drawback that it hinders the introduction of new technique, as already ex­ perienced in the eighties when capillary GC replaced packed column GC. As nearly all existing methods require a split! splitless injector, few instruments are equipped with on-col­ umn or PTV injectors. This in turn is an obligation for new method development to rely on split!splitless injection again.

4.3.1. Overflow Tech­ nique

Large volume splitless injection with overflow was investi­ gated in the hope of finding a technique involving an injec­ tor available on every GC instrument and which is simple to perform. It did not find its way into routine analysis, maybe because performance is inferior to the alternatives.

4.3.2. Solvent Splitting

The system described as "splitless injection of large volume" by Suzuki et al. is similar to what we later termed "vapor­ izer/precolumn solvent split system" [29]. The latter was conceived as an on-column technique with a vaporizing chamber acting as a filter to retain non-evaporating by-prod­ ucts. These are, of course, different views of the same con­ cepts.

Single, Generally Applicable Technique

This might, in fact, be the best choice for a universal large volume injection technique. The separation of the solvent from the solutes (i.e. the retention of the solutes while the solvent escapes) should be performed in a capillary precol­ umn, because solvent trapping is the most powerful process available. A vaporizing chamber is required for contaminated

292

0 4. Injection of Large Volumes

samples (e.g. extracts for pesticide analysis). A substantial amount of work has been invested into this field and the proc­ ess has been described in much detail, but the literature clas­ sifies these techniques as on-column injection or on-line LC-GC transfer and this is also why no further discussion is added here.

5. Sample Transfer into the Column This section deals with the conditions required for virtually complete transfer of the sample vapor from the injector into the column, i.e. with the duration of the splitless period and suitable column flow rates. Nearly Complete Transfer Required

For quantitative analysis it is essential that most (e.g. ~95 %) of the sample material is transferred from the vaporizing chamber into the column. The loss of sensitivity is of less concern than the accuracy and precision of quantitative re­ sults. With incomplete transfer, different sample components are lost unequally, resulting in a distorted picture of the com­ position of the sample analyzed (discrimination). Poor re­ producibility of the losses increases standard deviations, but also introduces systematic errors if, e.g., the deviation for a sample differs from that for the calibration mixture.

5.1. Spreading in the Vaporizing Chamber

An important source of deviations is irreproducible spread­ ing in the vaporizing chamber. In one instance, concentrated vapor expands upwards like a piston in a cylinder, in another there is turbulence and the vapor rises in the vaporizing cham­ ber like a pillar of smoke, or the liquid is repelled a long way from the bottom. If merely 30 % of a component enters the column during one injection, it might well be 20 % or 45 % during the next, caus­ ing the results to vary by 33 %. If transfer is nearly com­ plete. such irregularities no longer have a significant effect.

5. 1. 1. Observations with the Iodine Experiment

As described in Section 84.2.1, injection of a concentrated solution of iodine in acetone into a heated glass tube ena­ bles observation of the spreading of the solute vapor. Hot needle injection, nebulizing the solution, produced a fog in a fairly reproducible manner, whereas injection through a cool needle, hence with band formation, initiated widely differ­ ent processes.

Empty Liner

In an empty liner at 240°C, vaporization was accompanied by violent motion. The iodine vapor was' driven upwards by

5.1. Spreading in the Vaporizing Chamber

293

the solvent vapor. The movement looked different for each injection. Sometimes vapor expanded backwards as a plug, removing the gas like a piston and ending as a con­ centrated cloud sitting in the lower part of the tube. Another time, a trail of vapor moved through the carrier gas further up. A third time, vapor spiraled up the liner like smoke rings. As a consequence of this the distribution of the iodine vapor within the vaporizing chamber varied from one injection to the next. Packed Liner

Injection through a cool needle (band formation) into a small amount of wool at the bottom of the tube totally changed the distribution of the iodine vapor. The solvent evaporated first and a concentrated cloud of iodine vapor was formed in the close vicinity of the evaporation site (mechanism ex­ ploited in the overflow technique).

Losses Depending on Spreading

Movement through the carrier gas causes greater dilution than plug-like expansion, and if sample transfer into the col­ umn is incomplete, more material is lost (Figure D24). In addition, when the volatile material is enriched at the front of the expanding vapor, more volatile than high-boiling ma­ terial is lost, with the result that not only the total amount of sample material entering the column varies, but also its com­ position. This explains why complete transfer of the sam­ ple material from the injector into the column is of paramount importance in quantitative analysis. _ _c::::D _ _c::::D

I

I

Vapor expands Irregularly

Plug-like expansion

r ransferred to column

Incomplete transfer

r ransferred to column

Complete transfer

Figure D24 The expansion of the vapor is often not reproducible. If sam­ ple transfer is incomplete (i.e. material in the upper part of ~the liner is not transferred), losses depend on the extent of ,/ the expansion, and. hence. are poorly reproducible.

5.2. The Transfer Process

The transfer of the sample vapor is more demanding than is often recognized. Ideally, the carrier gas should push the vapor cloud into the column like a piston. A single volume corresponding to that of the vapor cloud, at most equal to the injector volume, would have to be displaced. At a col­

294

0 5. Sample Transfer into the Column umn flow rate of 3 ml/min and with a 1 mL liner transfer would take 20 s. This model is approached at best if carrier gas flow rates are high.

Transfer versus Dilution

The carrier gas is a poor piston because sample vapor mixes with it by diffusion. The size of the vapor cloud steadily increases during sample transfer. At low gas flow rates, this growth can even become predominant, as the vapor cloud appears to grow more rapidly backwards up the injec­ tor than it is driven down into the column (Figure D25). Under such conditions, transfer remains unsatisfactory even after extremely long splitless periods.

,

Carrier gas -----,-iI-I

it it

Gas flow driving vapor cloud towards column Sample vapor diffusing upwards

::\":". :.\~~~.:.:

."::".:.

J

Sample vapor transferred into the column

)(

•

Split outlet

Column

Figure D25 Sample transfer in splitless injection: displacement towards the column and growth of the vapor cloud upwards.

Gas Velocity in the Vaporiz­ ing Chamber

During the splitless period, the gas flow rate through the vaporizing chamber is equal to the column flow rate. The gas velocity in a 4 mm i.d. vaporizing chamber is 256 times lower than in a 0.25 mm l.d. column. If the average velocity in the column is 50 crn/s, that of the compressed gas in the column inlet might be 35 cm/s and that in the vaporizing chamber is merely 1.3 mm/s. Diffusion speeds at the el­ evated injector temperatures are similar, confirming that the growth of the vapor cloud is by no means negligible.

5.3. Flow Rate and Duration of the Splitless Period

The duration of the splitless period required for essentially complete sample transfer must be determined experimen­ tally, e.g. by measuring peak areas of a test component in­ jected with transfer periods of different duration (30). Unless stated otherwise, the experiments described below involved 2 ~L ("1 ~L") hot needle injections of a n-tridecane solution in hexane into an empty 5 mm i.d. liner (internal volume, 1.6 mL; gas chromatograph model Top. CE Instru­

5.3. Flow Rate and Duration of the Splitless Period

295

ments/ThermoQuest) by use of a 71 mm syringe needle. The 105 mm liner had a constriction at the bottom, reducing its useful length to 80 mm (adjusted to the length of the syringe needle, Figure E3). After the splitless period, a split (injector purge) flow rate of 100 rnt/rnin was used to remove efficiently sample material not yet transferred into the column. The column temperature during transfer was 100 "C, which ruled out solvent recondensation.

5.3.1. CaTTie, Gas Flow Rates

Figure D26 shows experimental results for carrier gas flow rates varied between 0.5 and 4 mllmin (hydrogen). For elu­ tion of the component, the flow rate was re-set to 2 rnt/rnln to ensure constant FID response. At a flow rate of 4 mlJmin, transfer was almost com­ plete within 35 s. At 2 mlJmin, the required duration of the splitless pe­ riod was 90 S, and even then the peak area did not fully reach that obtained at the higher flow rate. At 1 mlJmin, only ca. 94 % of the sample was trans­ ferred by use of a splitless period of 3 min. At 0.5 mlJmin, the transfer efficiency reached 75 % af­ ter 2 min and 82 % after 4 min.

'i

90

'2

:J 80

e

~

:e.!!.

82% --_o-0-.5-m-u"'-m7in"

70

60

m ca 50 ~

=

a.

40~

5 mmLd.liner hydrogen n-Iridecane in hexane

30 20 10

60

120

180

240

Duration of splltless period [s]

Figure D26 Sample transfer (peak area of a test solutel plotted against the duration of the splitless period for different carrier gas flow rates. (From Maja Fritschi, 1999.1

Increased Dilution at Low Flow Rates

The sample vapor did not fill the space available in the liner (1.6 rnl.): 2 J.1L hexane translates into ca. 500 J.1L vapor; the septum purge was permanently open. At a carrier gas flow rate of 4 mlJmin, the direct transfer (carrier gas acting as a piston) of a vapor cloud of less than 1.6 mL should have taken less than 24 s. In reality, it was 35 s, indicating correspond­ ing mixing during transfer. .

296

D 5. Sample Transfer into the Column The transfer time might be expected to be doubled when the

carrier gas flow rate is halved, but in reality the required du­

ration of the splitless period increased to ca. 90 s. Upon fur­

ther reduction of the flow rate, the transfer remained incom­

plete even after far longer times. The vapor cloud expands

and renders the already slow transfer even more diffi­

cult.

Minimum Carrier Gas Flow Rate

There is a minimum carrier gas flow rate below which growth

of the vapor cloud becomes predominant. For a 5 mm i.d.

liner it is ca. 2 mllmin.

Because the test solute was essentially unretained in the

vaporizing chamber (clean liner and sample, non-adsorptive

component), the above results represent the most favorable

situation. For retained high-boiling or adsorptive solutes,

higher flow rates and longer splitless periods are required.

5.3.2. Liner Bore

Figure D27 shows the influence of liner diameter on sam­

ple transfer. Results obtained from a 3 mm i.d. liner (565

J,lL internal volume) are compared withthose from the 5 mm

i.d. liner. With 2 mllmin carrier gas, transfer was complete within ca. 30 s. It was faster than for the 5 mm i.d. liner with 4 mLJmin. At 0.5 mllmin, fairly complete transfer was achieved after 1 min. This is as fast as with a flow rate of ca. 3 mLJ min when using the 5 mm i.d. liner. Ui'

:g

90

2mUmin

" !

eIi!! 70

:E .!. 60 as as

e 50

i

~

40

30

0.5mUmin

• /0....·-·····

::J 80

[}--------------D

-.- .-. ·-··-0:5--··--·

2 mUmIn

mUmin

j;// I!.

:'

- - 3 mm i.d. liner ••..••.. 5 mm Ld.liner

20 10

60

120

180

240

Duration of splitless period [s]

Figure D27 Sample transfer (peak area of a test solute) plotted against the duration of the splitless period: comparison of 3 and 5 mm i.d. liners. (From Maja Fritschi, 1999.)

Loss in Capacity

Transfer with narrow bore liners is more efficient because a 2.8 times faster carrier gas works against the same diffu­ sion. On the other hand, the internal volume is also 2.8 times

5.3. Flow Rate and Duration of the Splitless Period

297

smaller and, hence, the capacity for intermediate storage of vapor is small.

5.3.3. Diffusion Speeds

If the growth of the vapor cloud upwards against the carrier gas flow is largely determined by diffusion, it should depend on the type of carrier gas and the molecular weight of the solute.

Type of Carrier Gas

The above experiments were performed with hydrogen as carrier gas. Oiffusivity in hydrogen and helium is similar, whereas that in nitrogen is ca. four times lower. This sug­ gests that spreading in nitrogen should be noticeably slower (as observed in the separation process). The viscosities of these gases are also different: helium and nitrogen are about twice as viscous than hydrogen. Hence inlet pressures for a given fldw rate are different and, because diffusion is slower at higher pressure, this should result in slower spreading when helium or nitrogen are used. Figure 028 compares transfer obtained at equal gas flow rates (2 mt/mln). There is indeed somewhat less spreading in nitrogen than in hydrogen or helium, whereas no signifi­ cant difference between hydrogen and helium is apparent. Because, however, nitrogen requires substantially lower col­ umn flow rates (for the same reason), there remains no rel­ evant difference between the three carrier gases. lIDO

~ .:

80

i 60

40

5 mm Ld. liner 2mUmin 250 'C injec10r n-tridecane in hexane

20

30 60 90 120 Duration 01 splltless period [51

Figure D28 Sample transfer (peak area) plotted against the duration of the splitless period for three different carrier gases. (From Maja Fritschi, 1999.)

Molecular Mass of Solutes

As diffusion speeds increase rapidly with decreasing molecu­ lar mass, small molecular weight solutes should be signifi­ cantly more difficult to transfer than high molecular-weight material. Figure 029 shows that methane spreads clearly faster than the n-alkanes C13 or C40, but considering the large difference in molecular mass the effect is small. All this suggests that turbulence/convection might be more impor­ tant than diffusion.

298

D 5. Sample Transfer into the Column

5 rnrn i.d, liner hydrogen 2rnUrnin

4

2 30

60

90

120

150

180

Duration of splitless period [s]

Figure D29 Sample transfer plotted against the duration of the splitless period for three solutes of different molecular mass. (From Maja Fritschi, 1999.1

5.4. Accelerated Transfer by Pressure Increase

A carrier gas flow rate of 2 mL/min is often not acceptable for GC separation, and the vacuum pumps of many mass spectrometers do not take carrier gas flow rates in excess of 0.5-1 mLlmin. Solutes which are adsorbed or retained in the vaporizing chamber can, on the other hand, be satisfactorily transferred only at gas flow rates above 2 mLlmin. Small-bore liners substantially improve results obtained from use of low gas flow rates, but their capacity to house sample vapor is often insufficient. An old idea enabling escape from this dilemma is decoupling of the flow rate during injection from that during analysis and optimizing them individually. This has been used for test­ ing purposes since the seventies, but only became practical for routine analysis with programmable electronic gas regulation (Wylie et al. [16]).

5.4.1. Principles

During the splitless period, the inlet pressure is increased above that suitable for chromatography and it is reduced again when sample transfer is complete (Figure 030). This "pressure pulse" causes the flow rate into the column to reach maybe 10 mLlmin, which greatly improves the trans­ fer characteristics and thus eliminates one of the bottle necks of splitless injection.

Absence of Chromatography

Duri ng sample transfer chromatographic migration must be stopped anyway to reconcentrate the solutes of inter­ est in the column inlet. This is achieved by either cold trap­ ping or solvent trapping. During this period, excessive car­ rier gas flow rates do not disturb chromatography. Problems occur only, if pressure is not decreased early enough at the end of solvent trapping.

5.4. Accelerated Transfer by Pressure Increase

,njecti°r.l~-+Pressure increase

299

_ _-..

Splitless

period

End 01transler'l split exit opend, pressure de­ I creased

'---''-,---' L_---'

L-

Pressure suitin anal sis

Time

Figure D30

Improving sample transfer by increasing the flow rate dur­

ing the splitless period: pressure profile and chromatogram.

5.4.2. Advantages

Splitless transfer with flow increase has a.nurnber of impor­

tant advantages:

1 Appropriate splitless transfer becomes possible in con­

junction with analysis at a low column flow rate, i.e. with narrow bore columns. 2 Transfer of somewhat retained sample components is improved. This is important for adsorptive and high­ boiling compounds. 3 The reduced residence time in the injector reduces the thermal stress on labile solutes. 4 The vapor cloud is compressed. Larger sample vol­ umes can be injected or narrower bore liners employed, further contributing to improved transfer.

Reduced Degradation

Wylie et al. [311 reported a large reduction of the degrada­ tion of endrin to endrin aldehyde and endrin ketone as well as of DDT to DOE and DOD. At constant pressure, a highly active liner resulted in 86 % degradation of endrin. With an inlet pressure of 552 kPa during transfer, these losses were reduced to 20 %. This pressure was maintained for 0.3 min and then returned to 83 kPa at 683 kPa/min, i.e. in ca.

40s. Improvements for Organo­ phosphorus Pesticides

Wylie and Uchiyma [321 compared peak areas obtained by splitless injection with those from on-column injection. With splitless injection at a constant column flow rate of 5 mU min (30 m x 0.53 mm i.d. column) and a 1 min splitless period, recoveries of the organophosphorus pesticides acephate, omethoate, and methamidophos reached only 57, 63, and 71 %, respectively. With a pressure increase to 22.5 psig for 70 s, creating a flow rate into the column of 50 mll min, 86 % of the acephate reached the column; at 70 psig (298 mllmin) it was even 97 %. They also showed that the

300

0 5. Sample Transfer into the Column

matrix-induced response enhancement (Section D6.2)

could be reduced.

When a 30 m x 0.25 mm i.d. column was used for GC-MS,

recoveries were substantially lower as a result of the smaller

flow rate. With constant inlet pressure and a flow rate of

1.5 mL/min, recoveries of acephate, omethoate, and methamidophos were 13, 26, and 40 %, respectively. At 50 psig, maintained for 2 min, increasing the flow rate to 8.8 rnt/min, recoveries increased to 28, 49 and 66 %. When injected as a spiked extract from green beans, they further increased to 125, 96, and 91 %, respectively, owing to the temporary deactivation of the vaporizing chamber by the sample matrix.

Hydroxylated PCBs

Vincenti et al. [33) substantially improved the analysis of

hydroxylated PCBstandards. Tetrachlorobiphenyls were the

internal standards; a dihydroxy tetrachlorobiphenyl was the

most important solute. Injection at a constant pressure of

5 psig (helium, 25 m x 0.21 mm l.d. column, 0.66 mLlmin)

resulted in a peak area 1.5 times larger than that of the inter­

nal standards. With a constant pressure of 10 psi, the rela­

tive peak area increased to 5.7.

Constant flow during the temperature-programmed

analysis (applying a corresponding pressure program) in­

creased the relative areas to 4.9 and 8.6, respectively, sug­

gesting that this was one of the abnormal examples of an

increase in the gas velocity in a temperature program reduc­

ing adsorptivity.

With a pressure increase to 30 psig during the 1 min

splitless transfer, the relative area was further increased to

18.6. The result from on-column injection (relative area 34.6,

i.e. 23 times larger than initially!) indicated, however, that

still little more than 50 % of the solute had reached the col­

umn.

The example also teaches a lesson on possible errors in

splitless injection: with the initially applied conditions over

97 % of the solute was lost - and this was not even a particu­

larly difficult component.

5.4.3. Extent of Pressure Increase

The volume of the vapor cloud is inversely proportional to the absolute pressure, i.e. twice the volume of sample can be injected if the absolute inlet pressure is doubled (in­ creasing the sample volume by less than that is hardly worth the effort). Substantial gain is possible when inlet pressures are low. At an initial inlet pressure of 30 kPa (130 kPa absolute pressure), doubling is easy, because a transfer pressure of 160 kPa (260 kPa absolute) is sufficient. Increase by a factor of four (420 kPa) is possible. If, on the other hand, the inlet pressure is, e.g., 150 kPa to begin with, doubling means a transfer pressure of 400 kPa.

Sample Volume

5.4. Accelerated Transfer by Pressure Increase

301

This can also be seen from another angle: at 30 kPa inlet pressure the sample capacity of the liner is low and an in­ crease important whereas capacity is almost double to be­ gin with when the inlet pressure is 150 kPa. Sample Transfer

Increasing the inlet pressure accelerates sample transfer more efficiently than it enhances injector capacity, because the column flow rate increases in proportion to the inlet pressure (overpressure). With a short, rather wide-bore col­ umn and hydrogen as carrier gas, implying inlet pressures of ca. 20-40 kPa, a pressure increase to 100 kPa enhances the transfer rate by a factor of 2.5-5. Effects are again less spec­ tacular if inlet pressures are high to begin with (e.g. if helium is used as carrier gas).

Increase in Practice

According to literature the pressure was usually increased to the limit of the system, over 500 kPa. For instance, Wylie et al. [161 used a 25 m x 0.32 mm l.d. column and he­ lium as carrier gas. At the inlet pressure for analysis, 83 kPa, a column flow rate of 2.8 mLjmin was obtained. During in­ jection, they increased pressure to 552 kPa, resulting in a flow rate of 56.4 mLjmin. Wylie and Uchiyma reported a maximum pressure pulse de­ termined by the safety system of the mass spectrometer: at a flow rate of 8.8 mt/mln, the source pressure was near the point at which their instrument shut down. This maxi­ mum depends, of course, on the mass spectrometer used. Klick [341 reported that high standard deviations obtained with splitless injection could not be improved by increasing the pressure to 220 kPa (30 s). She assumed that septum leaks were the cause. Septum leakage will certainly be more common at the high pressures, but it should not affect quan­ titative results when the vaporizing chamber is not overfilled.

Achievable Flow Rates

Flow rates through the column, defined in terms of volumes at ambient pressure, increase more than in proportion to the inlet pressure (overpressure) because ofthe compressibility of the gas. The most drastic increase is again achieved when chromatography requires low inlet pressures, as shown by the examples in Table D2. With a 15 m x 0.32 mm i.d. column and hydrogen as carrier gas, normally used at 25-30 kPa, a pressure in­ crease to 100 kPa produces a flow rate of 15 mLjmin (calculated for a column temperature of 60 "C), At 500 kPa, a fantastic flow rate of 170 mLjmin is achieved. For a 60 m x 0.25 mm i.d. column with helium, a flow rate of merely 10 mt/rnln requires an inlet pres­ sure of 610 kPa.

Flow Rates in the Injector

The increase of the flow rate in the injector is smaller than that at the column outlet: the higher the inlet pres­ sure required, the more the gas is compressed and the lower the real flow rate.

302

D 5. Sample Transfer into the Column Table D2

Carrier gas Ilow rates obtained at different inlet pressures

lor two opposed situations, namely a short column used with

hydrogen as carrier gas and a long column used with he­

lium. Flow rates at ambient pressure and at the inlet pres­

sure.

Inlet pressure [kPa]

25 100 150 200 300 400 500 610

Flow rates [mLJmin] 60 m x 0.25 mm i.d., 15 m x 0.32 mm i.d., He H2 Outlet Injector Outlet Injector

2.7 14 25 39 73 116 169

2.2 7.2 10 13 18 23 28

1.1 1.6 3.0 4.9 7.1 10

0.44 0.53 0.75 0.98 1.2 1.4

For the 15 m x 0.32 mm i.d. column used with hydro­ gen, a flow rate of 10 mLJmin can still be achieved at a readily accessible inlet pressure of 150 kPa. For the 60 m x 0.25 mm i.d. column, an inlet pressure of more than 400 kPa is needed just to reach 1 mLJmin. The high inlet pressure cancels most of the effect seen at the column outlet. The cloud of sample vapor is com­ pressed correspondingly. Hence, a smaller volume must be transferred and the gain in transfer efficiency remains effec­ tive. If, however, the higher inlet pressure is used for inject­ ing a larger volume of sample, at the limit again filling the vaporizing chamber, the reduced flow rate at the high pres­ sure must be considered. Furthermore, when considering retention of solutes on surfaces in the injector, merely the real (i.e. small) flow rate is effective. .

5.4.4. Duration of the Pressure Pulse

The duration ofthe pressure pulse is determined by two con­ siderations. Pressure should be kept high up to the end of the sample transfer, including transfer of the most diffi­ cult (adsorbed or retained) solutes. Prolongation helps improve quantitative results. Excessively long periods of high flow rate can cause peak broadening, because conditions are far removed from the optimum for the separation process. . High carrier gas flow rates have no effect on chromatogra­ phy as long as the solute material is cold trapped or solvent trapped in the column inlet. Afterwards, however, they cause band broadening.

5.4. Accelerated Transfer by Pressure Increase

303

If the solutes of interest are cold trapped, i.e. the column temperature during transfer is so low that they are not yet chromatographed, the moment the inlet pressure is reduced is not critical. Normal flow rates must be resumed before the separation process is started, i.e. some 60° below the elu­ tion temperature. Critical Early Eluted Peaks

Reduction of the pressure is more critical if the first solutes of interest are eluted at or near the column temperature dur­ ing transfer. For such applications, solvent trapping is the reconcentration technique of choice. Chromatography starts at the end of solvent evaporation and pressure should be returned to normal exactly then. Because accurate timing is difficult, early eluted components often produce broad peaks with strongly reduced (and maybe poorly reproducible) re­ tention times (see Wylie et al. [16)). If such effects disturb the analysis, the duration of the pressure pulse and the extent of the increase must be reduced. It should be kept in mind that at a flow rates of, e.g., 10 mLl min, transfer of unretained solutes is complete in ca. 10 s, and that the transfer time at 50 rnt/rnln is very short.

Reducing the Pressure

The inlet pressure must be reduced such that backflow from the column is not possible. Either the pressure is reduced by a sufficiently slow program, or a rapid pressure drop is avoided by slow discharge of the overpressure. The lat­ ter is most easily achieved by relieving the pressure regula­ tor while the split exit is still closed. As the extra volume of gas in the injection system is discharged into the column, backflow from the column is prevented.

5.4.5. Accentuated

Pressure increase during transfer easily increases the dew point of the solvent by several tens of degrees and can

Solvent Recondensation

result in strong recondensation of solvent in the column in­ let when no recondensation is encountered otherwise. This can be an advantage for a solvent which does not enable solvent trapping at normal pressures (e.g. pentane) or where a higher oven temperature can be used, accelerating the cooling step (e.g. dichloromethane). Band Broadening in Space

In other situations, such recondensation results in broaden­ ing or distortion of peaks, particularly when the pressure increase is also used for injection of larger volumes. Partial recondensation of 5 III of solvent easily causes severe flood­ ing and band broadening in space. Peak broadening can be avoided by use of either an increased column temperature during injection (reducing recondensation) or an uncoated precolumn (reconcentrating initial bands).

Column Deterioration

Wylie and Uchiyma observed rapid deterioration of chroma­ tography when they injected pesticide samples with a strong pressure pulse. 10-20 cm of the column inlet had to be re-

UNIVElUUDAD DEANTIOQUlr. BIBLIOTBCA CENTRAL

304

D 5. Sample Transfer into the Column

moved to restore performance. The liner was packed with ca. 7 mg deactivated glass wool. It was argued that this might not have been sufficient to retain non-evaporating material inside the injector. The faster transfer leaves less time for deposition of aerosol-forming non-evaporat­ ing by-products on surfaces in the injector. Because addi­ tional glass wool would have probably worsened the loss of labile solutes, the use of an uncoated precolumn was sug­ gested. Deterioration can also have been caused by solvent recon­ densation. Acetone solutions were injected at a column tem­ perature of 40°C. Without pressure increase, acetone (b.p, 56°C) did not recondense, whereas there must have been substantial recondensation at 50 psig. Recondensed solvent transports the nonvolatile material towards the front end of the flooded zone and accumulates it there, causing the ef­ fect of the latter to be stronger.

5.4.6. Recommendations

Pressure increase during splitless injection should be ap­ plied whenever possible. Since there are seldom draw­ backs, it is not even worth checking whether or not there is a gain from it. If solutes are eluted at temperatures substantially above that during injection (reconcentration by cold trapping), the in­ crease can be large and for a duration exceeding the mini­ mum, e.g. 30 s. The upper limit is determined by the pres­ sure ofthe gas supply in the laboratory or the pressure regu­ lators of the instrument and is usually in the range of 350­ 500 kPa. Such high pressures are needed only if the inlet pressure required by the column is high to begin with (he­ lium, long, narrow bore columns), however. It is hardly mean­ ingful to exceed a column flow rate of 15-20 mllmin. If solutes are eluted at or near the injection temperature, there is less peak broadening when the pressure increase is re­ stricted to a factor of 2-3. This also facilitates optimization of the duration of the pulse. Increasing the flow rate is most effective with short col­ umns and hydrogen as carrier gas - yet another argu­ ment in favor of keeping columns short and preferring hy­ drogen.

5.5. Accelerated Transfer by Solvent Reconden­ sation

Recondensation of the sample solvent in the oven-ther­ mostatted column inlet accelerates sample transfer. The mechanism was described for split injection, where more sample entered the column than expected from the pre-set split ratio (Section C8.3.3).

Mechanism of Acceleration

At the beginning of the splitless period, concentrated sol­ vent vapor enters the column. Recondensation on the col­ umn wall reduces the vapor volume by a factor of 100-500, i.e. removes most of the incoming gas/vapor mixture. The resulting vacuum is immediately filled by further sample

5.5. Accelerated Transfer by Solvent Recondensation

305

vapor from the injector - which might again recondense. In this manner, the flow of vapor into the column inlet is easily many times greater than normally. Only when the bulk of the concentrated vapor has been transferred does the flow rate drop back to normal, because the vapor is now too di­ lute to cause a substantial pressure drop upon reconden­ sation. Acceleration only for Evapo­ rated Solutes

Recondensation of solute vapor eliminates such small vol­ umes from the gas phase that it has no measurable effect. Solute vapor is, however, pulled into the column with the recondensing solvent. In this way, recondensation of the solvent also accelerates the transfer of the sample com­ ponents, provided they are in the gas phase. Recondensa­ tion does not accelerate transfer of components which slowly evaporate from surfaces; these anyway difficult solutes ar­ rive at the column entrance at a moment when suction into the column has ended.

5.5.1. Efficiency of the Recondensation Effect

The power of the recondensation effect depends on many factors and is, as a consequence, difficultto predict. It de­ pends on the dilution of the sample vapor with carrier gas, i.e. on the geometry of the vaporizing chamber, the length of the syringe needle, and on mixing with carrier gas during sample evaporation.

Column Temperature Below the Dew Point

To achieve recondensation, the column temperature dur­ ing transfer must be well below the dew point of the solvent vapor/gas mixture. When the concentration of vapor is 100 %, the dew point corresponds to the solvent boiling point (to be corrected for the inlet pressure); the lower the vapor concentration, the lower is the dew point. Recondensation extracts vapor from the gas phase until the remaining vapor/gas mixture has a dew point correspond­ ing to the actual column temperature. In other words, the concentration of vapor remaining in the gas phase corre­ sponds to the vapor pressure of the recondensed liquid. At a given column temperature, introduction of concentrated vapor will cause relatively strong recondensation, resulting in a strong vacuum and an increase in the rate of transfer, whereas a more dilute vapor might not recondense. When the column temperature is reduced, the concentration of vapor remaining in the gas phase is lower, i.e. more solvent recondenses and recondensation also occurs for more di­ lute vapor.

5.5.2. Experimental Results

In Figure D31, the effect of solvent recondensation is shown for a test sample of n-C22in hexane (b.p.. 69°C) and an 80 x 4 mm i.d. liner. During the splitless period, the column was at 25°C, i.e. nearly 50° below the pressure-corrected boiling point. The carrier gas flow rate of ca. 1.5 mLlmin was below the minimum enabling reasonable transfer without solvent

306

0 5. Sample Transfer into the Column recondensation. The transfer was, nevertheless, complete within ca. 40 s. In the absence of recondensation (column at 100 °C), hardly 80 % ofthe n-C22 entered the column in 100 s. peak area ___- - - - -... 25°C

column temperature

10

30

60 100 splitless period [51

Figure D31 Accelerated sample transfer as a result of solvent recon­ densation in the column inlet. (From ref. [301.) Acceleration at the Begin­ ning

Transfer was accelerated particularly during the first part of the splitless period: during the first seconds three times more material entered the column than at 100 °C. After 10 s, the difference still amounted to a factor of two, reflecting reduced recondensation owing to more dilute vapor. The rate of transfer fell rapidly after ca. 25 s, indicating that the con­ centration of solvent vapor in the gas phase had dropped below that corresponding to the vapor pressure of hexane at 25 °C. Solvent recondensation enabled rapid sample transfer at a carrier gas flow rate otherwise too low for splitless in­ jection. The effectiveness of the recondensation effect can­ not be predicted, however, and the results shown should be taken as an example and not for deriving general rules.

5.6. Tests on Complete­ ness of Sample Transfer

There are two situations when a test on the completeness of the sample transfer is needed.

Optimization of Conditions

Basically rules could be elaborated on how to achieve com­ p/ete sample transfer for a given sample, injector design, needle length, and type of autosampler. None is available today, because they would be exceedingly complex. It is sim­

5.6. Tests on Completeness of Sample Transfer

307

pier to start with reasonable conditions and then to check

the transfer efficiency, correcting the conditions when

needed.

Dubious Results

A check on the efficiency of sample transfer is advisable if

high standard deviations are observed, response factors

deviate from expectations, or if absolute or relative peak ar­

eas depend on whether the column temperature during in­

jection is above or considerably below the solvent boiling

point. High standard deviations can, however, also have other

sources: irreproducible transfer from the syringe needle or

matrix effects when samples contain large amounts of in­

volatile material.

5.6.1. Rapid Check via

A rapid check of sample transfer during a sequence of analy­

ses requires just one extra run: the injection of an already

analyzed sample repeated under conditions accentuating

sample transfer.

During sample transfer the carrier gas flow rate (inlet pres­

sure) is, e.g., doubled and the duration of the splitless pe­

riod extended to several minutes. If possible the column tem­

perature during injection is reduced to at least 25° below the

solvent boiling point to promote recondensation. At the

end of sample transfer the carrier gas inlet pressure is re­

duced to the previous level (otherwise detector sensitivity

might be changed). This must occur slowly enough to avoid

flow of compressed gas backwards out of the column.

Accentuated T,ans'e, Conditions

Interpretation of Results

If peak areas obtained from the test run exceed those previ­

ously obtained, conditions must be improved. Experiments

are required to determine whether prolongation of the

splitless period is sufficient or further changes are neces­

sary.

Areas obtained under accentuated transfer conditions do not,

however, necessarily represent the full amount of solute

material to be analyzed, because some types of loss can­

not be eliminated:

1 2

3

5.6.2. Check via On­ Column Injection

losses in the dead volume below the column entrance

or in the split outlet line; losses through the septum purge or in the carrier gas supply line (although increased inlet pressure reduces them); and adsorption or degradation in the vaporizing chamber.

A mora comprehensive check is obtained via on-column injection. It must be performed on the same instrument, because comparison of absolute peak areas presupposes the same detector. Comparison of the absolute peak areas obtained in splitless and on-column injection requires some caution. The sample volume injected on-column is accurately known (the volume measured), whereas in splitless injection the needle content

308

D 5. Sample Transfer into the Column may have been added. For volatile solutes, the amount in­

troduced can then be derived from the volume of solvent

injected, but higher-boiling solutes may partly have remained

inside the needle.

Many modern gas chromatographs are equipped with an on­

column and a vaporizing injector. If not, a makeshift on­

column injector can be constructed from a press-fit T-piece

as described in 1351. If performed only occasionally, the sam­

ple can also be deposited directly into the column inlet with­

out use of an injector.

On-Column Injection into a Detached Column Inlet

For injection directly into the column inlet, a syringe with a

needle fitting the column bore is required. If no on-col­

umn syringe with a 32 gauge (0.23 mm o.d.) steel needle or

a fused silica needle of 0.23 or 0.17 mm o.d. is available, a

0.53 mm i.d. precolumn at least ca. 5 cm long is attached to the column inlet, accommodating the ordinary 265 gauge needles of syringes used for manual injection into vaporiz­ ing injectors. The capillary inlet is dismantled from the (hot) injector and left for a time sufficient for backflow of carrier gas from the column inlet to come to a halt. For a long column, this can take a few minutes, for a short (e.g. 15 rn) column merely some 10-20 s. The syringe needle is then introduced and the sample slowly injected. Introduction of the needle is facili­ tated by means of a press-fit connector (loosely) attached to the column entrance, acting as a funnel guiding the nee­ dle. The syringe needle must be withdrawn with care, be­ cause the sample liquid tends to adhere to its tip (there is no carrier gas driving it onwards). Rapid withdrawal risks some liquid being pulled out of the column. The column inlet is held vertically (check the movement by visual control), Before re-installation of the column in the injector, a small carrier gas inlet pressure is applied to ensure that the sample liquid and vapor are pushed further into the column during installation (an excessively high inlet pressure results in a stream of hot gas which feels like needles penetrating the finger when the column is mounted).

5.7. Fast GC/Narrow Bore Columns

Numerous studies have been performed on the acceleration of GC analysis. Fast GC is easy when separation efficiency can be sacrificed (simply use columns only, e.g., 3 m long), but more difficult when efficiency must be maintained. Con­ clusions are that columns should be shortened and their in­ ternal diameter reduced to ca. 0.1 mm.

Low Carrier Gas Flow Rates

According to the above rules on the required VOlume of the vaporizing chamber and the minimum carrier gas flow rates necessary to achieve satisfactory transfer, there seems to be no possibility of performing splitless injection: columns of 0.1 mm i.d. are commonly used with carrier gas flow rates of 0.3-0.5 mLJmin.

5.7. Fast GC/Narrow Bore Columns

309

High Inlet Pressures

Van Ysacker et al. [361 came to a more optimistic conclusion. They used columns of (1) 5 m x 0.067 mm i.d., (2) 10 m x 0.10 mm i.d., or (3) 10 m x 0.05 mm i.d., which required inlet pres­ sures (helium) of 600 kPa (columns 1 and 2) or 1800 kPa (col­ umn 3). At these high pressures, the volume of vapor is massively reduced and, hence, smaller bore liners can be used. At 600 kPa (700 kPa absolute pressure), the volume is almost one-fifth that at 50 kPa (150 kPa absolute). Instead of the 4 mm i.d, liner, a 1.8 mm bore is sufficient.

Narrow Bore Liners

For the experiment shown in Figure D32 all possibilities of optimization were exploited. Column 1 was used with an in­ let pressure of 650 kPa, resulting in a flow rate of 0.7 ml)min and a high gas velocity at the column outlet (3.4 rn/s), As a hexane solution was injected at a column temperature of 50 °C, recondensation in the column inlet strongly accelerated sample transfer (at 650 kPa, the boiling point of hexane is 139°C). Even a 0.8 mm i.d. liner resulted in an area for the test solute (n-nonane) suggesting full transfer. Transfer was complete in some 20 s. The 0.8 mm i.d, liner was found to be sufficiently large for a 1 ul, injection of a hexane solution. No data are provided for the transfer under conditions with­ out solvent recondensation and solvents producing larger volumes of vapor.

0.8 mm Ld.

III

~

-" III Q)

a.

o

20

40

60

80

100

120

Splitless period [sl

Figure D32 Peak area of a test solute obtained by splitless injection into a 0.067 mm i.d. column at an inlet pressure of 650 kPa. with liners of different internal diameter and splitless periods of different duration. (From Van VBaeker [36].) Retention Gap

The chromatograms showed severe band broadening in space since the recondensed solvent flooded some three times further into the column than normally and the flooded zone was anyway large in proportion to the short column. The early-eluted compounds are reconcentrated by solvent trapping and form sharp peaks. Starting some 30° above the column temperature during injection severe peak distortion is observed whereas the high-boiling compounds form sharp peaks again, because they are retained in the column neck and do not reach the flooded zone (Section 07.5).

310

D 5. Sample Transfer into the Column The problem is solved by using an uncoated precolumn. A 1.5 x 0.32 mm i.d. precolumn was connected to the separa­ tion column by means of a press-fit connector additionally tightened with ferrules on both sides, similar to the VU un­ ions from Restek. Maybe a 50 cm x 0.25 mm i.d. precol­ umn would have been more appropriate.

Drug Analysis

In 1988, Kokanovich et al. [371 suggested the use of 0.1 mm i.d. capillary columns for rapid analysis of drugs. At a carrier gas velocity of 140 cm/s, the flow rate was about 0.7 ml/ min. They used a 2 mm i.d. liner, in some instances contain­ ing "Pureeol" inserts of ca. 1 mm i.d. packed with a small plug of glass wool. The column entrance was positioned 1 cm below the needle exit. The injection volume was 1 ul.: the splitless period 54 s. Relative standard deviations of a few percent were obtained for a number of drugs.

5.8. Splitless Injection for SPME

So far the sample has been assumed to be a dilute solution in a solvent. There is, however, an important exception to be mentioned here: splitless injection in solid phase micro ex­ traction (SPME). The sample is extracted from a liquid or a gas on to a fiber coated with stationary phase. Injection in­ volves thermal desorption from this stationary phase. Hence SPME is solvent-free injection.

Volume of Vapor

In trace analysis requiring splitless injection, the fiber might be loaded with 100 ng of solutes of a molecular mass typi­ cally ranging between 50 and 200 Daltons. If all this material were evaporated at once, it would form ca. 20 nl vapor, i.e. not even the smallest imaginable liner is too small. Further­ more, this vapor does not need to be stored before transfer. At a column flow rate of 1 ml/min, this cloud is calculated to be transferred in ca. 1 ms. Hence a narrow bore liner is suit­ able, preventing band broadening during slow transfer.

Narrow Bore Liner

Many applications have been performed on Varian instru­ ments equipped with septum-equipped programmable injectors (SPls). These are well suited to the purpose: the liner is of small bore (0.8 mm) and can be directly connected to the column entrance by means of a press-fit seal. A 2 cm section of an 0.8 mm i.d. liner has a volume of ca. 10 ul, (further reduced by the inserted fiber), which is transferred into the column in a fraction of a second. For other injectors a thick wall liner of similar bore is rec­ ommended. If no suitable glass tube is available, a narrow bore glass tube can be inserted into a 4 mm i.d. liner, pro­ vided it sits rather firmly, such that there is no significant flow of carrier gas between the two tubes. 1 mm bore may be preferable to 0.8 mm, because sometimes septum par­ ticles accumulate until the liner is almost plugged (which depends on the septum - hard, highly thermostable septa are not suitable).

5.8. Splitless Injection for SPME

311

Band Broadening by Slow Desorption

This does not mean that SPME necessarily produces sharp initial bands. If the geometry of the vaporizer is appropriate, the width of the initial band is primarily determined by the desorption time from the fiber. Langenfeld et al. [381 deter­ mined the profile of the initial band by direct coupling of the injector to a mass spectrometer. They found that desorption of naphthalene (b.p. 218°C) at 200 °C took ca. 12 s, whereas more than 1 min was required for benzoanthracene (but only 15 s at 300°C).

Retention Power of the Fiber

The long desorption time for benzoanthracene at 200°C is less surprising than that it has been measured at all: the coat­ ing of the fibers is 7 or 100 urn thick. In the liner, the section housing the fiber can be regarded as an open tubular col­ umn with the stationary phase on an inserted rod instead of on the outer wall. A capillary column coated with a 7-100 urn film would never be considered for analysis of benzoanthra­ cene. The retention power of the fiber is extremely high and requires a correspondingly high desorption temperature.

Reconcentration of Broad Initial Bands

If desorption results in excessively broad bands for chroma­ tography, a reconcentration technique must be applied. As solvent trapping is ruled out (lack of solvent), cold trapping is the only option. If the solutes are eluted at temperatures above ca. 60°C (maybe by use of a long, thick film column), cooling of the oven to 35-40 °C during solute transfer might be sufficient. For more volatile solutes, cryogenic focus­ ing is required. Either the whole GC oven is cooled or a cold trap, which merely cools a short section of the column inlet, is used. It can be helpful to perform desorption at a reduced carrier gas flow rate. Inlet pressure is reduced by maybe a factor of four before the fiber is introduced and brought back to normal ca. 20 s after. Vapor released from the fiber is mixed with less gas and driven less far into the column. This method works if broadening is determined by kinetics, but has no effect when partitioning is unfavorable.

Comparison with Injection of Liquid Samples

Snowand Okeyo [391 compared contributions to band broad­ ening for splitless injection by SPME with that for liquid sam­ ples, interpreting the difference as the result of solvent ef­ fects. Because pentane was injected at 40°C, reconcentra­ tion was weak and phase soaking alone helped to recon­ centrate the first eluted components.

5.9. Conclusions

We have previously claimed that the volume of a vaporizing chamber suitable for splitless injection should not exceed 1-1.5 mL, otherwise transfer at the carrier gas flow rates commonly used in capillary GC would not be feasible - as­ suming flow programming is not involved. This statement can now be justified.

312

D 5. Sample Transfer into the Column

5.9.1. Diameter of the Vaporizing Chamber

The critical factor for splitless sample transfer is the gas ve­ locity in the vaporizing chamber. Experiments indicate that efficient transfer from a 4 mm i.d. liner is possible at flow rates down to 2.5 mLJmin, which corresponds to a gas ve­ locity of 3.3 mm/s. Similar sample transfer characteristics can be expected with liners of other internal diameters, if the linear gas velocity is unchanged.

Maximum Diameter

When the internal diameter of the liner is increased from 4 to 6 mm, the volume of the vaporizing chamber is enlarged from 1 to 2.25 mL, which enables injection of approximately twice the volume of sample. At the same time, however, the carrier gas velocity is halved, and satisfactory transfer now requires a carrier gas flow rate of at least 5 mLJmin (3 mLJ min if a strong recondensation effect can be achieved). This substantially exceeds the optimum flow rate for 0.32 mm i.d, columns and, hence, presupposes flow programming.

Minimum Diameter

If low carrier gas flow rates are used, splitless transfer at constant pressure is possible only if a liner of narrower bore is used. Reduction to 2.8 mm doubles the gas velocity and enables satisfactory sample transfer at a flow rate of ca. 1 mllmin. This also halves the maximum sample volume, i.e. injection is now restricted to samples/solvents generat­ ing modest volumes of vapor - barely more than 1 ~L.

High Inlet Pressures

These estimates the were based on the assumption of inlet pressures up to around 150 kPa. They no longer apply when high inlet pressures are used for narrow bore columns or pressure is increased to render transfer more efficient. 1 1.2 mm i.d. liners can be used when pressure is in­ creased to 500 kPa, because the vapor volume is com­ pressed. 2 Injection with a 6 mm i.d. liner becomes possible jf a higher gas flow rate is used. If pressure is increased to 500 kPa, the internal volume is sufficient to house 11.3 mL vapor (calculated in terms of normal pressure), which corresponds to more than 20 flL hexane when mixed 1:1 with carrier gas.

5.9.2. Duration of the

Many manuals recommend spfitless periods of only 30 s duration. At the carrier gas flow rates commonly used many such splitless injections are effectively split injections with a substantial and irreproducible split ratio and probably poor linearity also.

Splitle•• Period

Rather Excessively Long

There are no severe drawbacks when splitless periods are longer than necessary (except maybe a broadened solvent peak). Hence there is no reason for minimization. Additional time may be useful for delayed transfer of solutes which are somewhat retained in the vaporizing chamber, e.g. on a layer of involatile by-products, or adsorbed on, e.g., a pack­ ing of glass wool. '

6.1. List of Problems Discussed in Other Parts

313

As mentioned above, it is difficult to provide firm rules. The transfer should be checked for the sample of interest whenever there are doubts about its completeness. Transfer without Solvent Recondensation

When a 4 mm i.d, liner is used, the minimum carrier gas flow rate required for virtually complete sample transfer without solvent recondensation is ca. 2 mUmin. Atthis flow rate, a safe duration of the splitless period is 90 s. At 4 mU min, the recommended splitless period is 50 s. These conditions should also be used if only weak solvent recon­ densation is involved, e.g. using dichloromethane or carbon disulfide as solvent at a column temperature of 30 "C.

Transfer with Solvent Recondensation

If the column temperature is at least 30° below the solvent boiling point, recondensation accelerates sample transfer to such an extent that a flow rate of 1.5-2 mUmin and a splitless period of 40 s provide satisfactory transfer. At a flow rate of 4 mLJmin, a splitless period of 30 s is sufficient.

6. Problems with Quantitative Analysis Splitless injection cannot be regarded as a highly accu­ rate sample introduction technique. Knowledge about the most critical parameters helps to improve results. 6.1. List of Problems Discussed in Other Parts

Below the known sources of problems in quantitative analy­ sis with splitless injection are listed, together with the symp­ toms to be expected. If the quality of the results is unsatis­ factory, the reader can go through the list to check which aspect could be improved, but sometimes also just to con­ firm that his problem is inherent to the technique and diffi­ cult to solve. Some of the problems have been described in other parts and will just be mentioned. Others require more complete presentation.

6.1.1. Selective Evapora­ tion from the Syringe Needle

Partial evaporation in the syringe needle can cause problems, particularly if the injector head is poorly heated. It is an im­ portant cause of inaccurate absolute peak areas (volume of sample injected) and of discrimination against high­ boiling components when samples cover a wide volatility range. The extent of the discrimination can be tested by re­ injection of the needle content (Section A5.2). Evaporation in the needle is avoided by using a fast autosampler or a very short syringe needle combined with a cool injector head, or a solvent with a boiling point sufficiently high that evapo­ ration inside the needle is suppressed.

314

D 6. Problems with Quantitative Analysis

6.1.2. Poor Sample Evaporation

Section B described problems related to sample evapora­ tion inside the vaporizing chamber. When a fast autosampler is used the liner must be packed or contain obstacles stop­ ping the liquid just above the column entrance. Otherwise sample material is lost at the bottom of the chamber, which again tends to discriminate against the high-boiling sol­ utes. Sample liquid jumping around in the vaporizing cham­ ber renders the process poorly reproducible. Thermospray hot needle injection largely eliminates these problems and enables working with an empty liner. Solute evaporation from layers or droplets of non-evaporating sam­ ple by-products is a problem discussed below under the head­ ing of matrix effects.

6.1.3. Injector Overload­ ing

Injection of excessive volumes of sample, use of vaporizing chambers of insufficient size, or use of syringe needles which are too short cause expansion of vapor backwards out of the liner. If the septum purge is left open (or cannot be closed, as with classical flow/backpressure regulation), overflowing solute vapor is removed through this exit. Volatile compo­ nents are lost preferentially. Otherwise vapor penetrates the carrier gas supply line and eventually recondenses in the cooler regions. This causes the opposite discrimination ef­ fect: the volatile compounds are more likely to return. Com­ ponents of intermediate to high-boiling point can lead to "memory effects" for weeks (Section D2.6.2). Tests can be performed by injecting smaller volumes of sample or by in­ creasing the inlet pressure during the splitless period.

6.1.4. Incomplete Trans­ fer of Sample Vapor

If the sample vapor is incompletely transferred from the va­ porizing chamber into the column, results tend to be inac­ curate and poorly reproducible.

High Standard Deviation as the Tip of the Iceberg

A high standard deviation is certainly undesirable as such, but often it is just the tip of the iceberg. If the sample is trans­ ferred only partially, the proportion of material entering the column depends on a number of poorly controlled factors and often differs for different samples. If, for instance, this proportion is different for the calibration mixture and the sample, systematic errors occur. These are as difficult to recognize as the ice beneath the water surface.

High Standard Deviations as a Warning

Standard deviations are sometimes regarded as inherent in any physical process and that there is no need to question their origin. This obscures the fact that there is a real proc­ ess behind them which should be identified ifthe results are to be improved. If 100 % of the sample were transferred, the results would not deviate from the true values (apart from errors arising from the syringe needle problem; injector overflow is as­ sumed to be avoided). This also means that standard devia­ tions would be low, because an accurate value cannot vary, but only the deviation therefrom.

6.1. List of Problems Discussed in Other Parts

315

If a relative standard deviation of, e.g., 30 % is observed, it must be concluded that sample transfer was far from com­ plete. Losses in the syringe needle cause less variation. In­ jection resulting in the highest peak areas at best achieves 100 % sample transfer. To obtain a relative standard devia­ tion of 30 %, other injections must have introduced less than 50 % of the sample. A high relative standard deviation should be considered as an alarm signal indicative of poor sample transfer. Obviously, the opposite statement is not necessarily true: a low varia­ tion does not prove complete sample transfer because un­ der certain conditions even a reduced proportion of the sam­ ple might be introduced reproducibly. Inaccurate Absolute and Relative Peak Areas

Incomplete sample transfer into the column not only reduces absolute peak areas to a poorly reproducible extent, but it also causes discrimination effects. because losses may be different for different compounds, as shown by the fol­ lowing example from the pharmaceutical industry. A dichloromethane solution containing a drug and an inter­ nal standard was analyzed by GC-MS. The ratio of the peak areas of the sample component and the internal standard differed by a factor of two, depending upon whether the sam­ ple was injected at a column temperature of ca. 50 DC, which resulted in cold trapping alone, or at ambient temperature, to exploit additional solvent effects. Obviously the observed deviation was caused by accelerated sample transfer as a result of the solvent recondensation (results obtained at the low column temperature were, indeed, closer to expecta­ tions). A 0.20 mm i.d, column was used with helium, resulting in a column flow rate far too low for satisfactory splitless trans­ fer without the help of solvent recondensation. Conditions were typical of those used for GC-MS with mass spectrom­

eters equipped with small vacuum pumps. An Experimental Optimiza­ tion

Kaufmann [40] undertook optimization of a variety of condi­ tions by statistical techniques (fractional factorial design and central composite design). The factors of interest in­ cluded the design of the liner, the sample volume, the dura­ tion of the splitless period, and the sample solvent. The test solutes were silylated compounds such as lactate, glycerol, phosphate, succinic acid, malic acid, glucose, and trehalose, dissolved either in ethyl acetate or in 1:1 BSTFA/pyridine. Injections were performed with an HP 6890 autosampler. A 30 m x 0.25 mm i.d. column was used at 12.5 psig inlet pres­ sure (helium) and an initial temperature of 70 DC. The liners tested had a constriction at the bottom (goose neck), two with a plug of wool just above the constriction, the other two with a short cyclo section (Restek). 2 mm i.d. liners performed better than those of 4 mm i.d. provided only 1 J.1L volumes were injected and column tem­ peratures were low (recondensation). In fact, transfer was

316

0 6. Problems with Quantitative Analysis faster and, hence, more complete. Quantitative performance

was improved at the expense of sensitivity.

The quality of the results obtained with glass wool in the

liner depended less on sample volume and column tempera­

ture than those without, obviously because deposition on to

the packing created conditions as used for the overflow tech­

nique. Wool could be tolerated because the silylation rea­

gent prevented degradation of labile solutes.

The primary message of the paper was the interdepend­

ence of all the factors which renders optimization com­

plicated and results in every recommendation having a re­

stricted range of application only.

6. 1.5. Adsorption and Retention in the Vaporiz­ ing Chamber

Adsorption or retention on surfaces in the vaporizing cham­

ber delay transfer to the column, because the material may

be released from the adsorptive or retentive site only when

the split outlet is re-opened and thus largely be lost (Section

86.1.2). On-column injection provides a quick test, but pre­

supposes a fairly clean sample. Higher injector temperatures,

nebulizing injection into an empty liner, and increase of the

flow rate (inlet pressure) during the splitless period are other

suggestions.

6.1.6. Degradation of Labile Solutes

Splitless injection exposes the solutes to high thermal stress

owing to the long residence time in the hot vaporizing cham­

ber. Labile components may decompose, rearrange, or po­

lymerize. In split injection, cooling during solvent evapora­

tion and rapid discharge through the injector provide milder

conditions. Comparison with results obtained by on-column

injection most rapidly indicates whether there are related

problems. Lower injector temperatures, nebulizing injection

into an empty liner, and increase of the flow rate during the

splitless period are the suggestions to try.

6.2. Enhancing Matrix Effects

Interference of a matrix effect means that the sample ma­

trix influences the analytical result, the matrix being the

solvent. the components, or the by-products. In the most

important instance, a given amount (concentration) of sol­

ute produces peaks of different area depending on whether

it was injected as a standard solution or in a real sample.

6.2.1. Definition

Here enhancing and reducing matrix effects are distin­

guished.

With an enhancing matrix effect, the sample produces a larger signal for a given amount of a solute than the solution of standards. Hence the response is increased by the presence ofthe sample by-products ("matrix-in­ duced chromatographic response enhancement", the term introduced by Erney et al. [41]). The reducing matrix effect causes the solute signal in the sample to be reduced compared with that from the solution of standards. .

6.2. Enhancing Matrix Effects

317

Matrix effects are defined in terms of the resulting phenom­ ena, not the mechanisms involved. They can result from an effect of adsorptivity or chemical decomposition in the in­ jector or in the column, as well as from other processes in­ cluding elution from the syringe needle, movement of the sample liquid during evaporation, and transfer into the col­ umn.

Systematic Errors

Matrix effects introduce systematic errors when calibration of response or correction factors is performed by use of "clean" mixtures of standards. Results are excessively high when enhancing matrix effects interfere and too low in the event of reducing effects. Because systematic errors are detected only upon specific testing, the characteristics of these effects should be kept in mind to make sure that ana­ lytical procedures include appropriate testing or are designed to be immune against matrix effects.

6.2.2. Description of the Effect

Erney et al. demonstrated that the response to some pesti­ cides is higher when they are injected in a sample extract than in a solution of standards in pure solvent. This is the result of a temporary deactivation of surfaces (in the injector or the column) by other sample components and enables explanation of recoveries exceeding 100 %, as rather fre­ quently observed.

The Dynamics of the Enhanc­ ing Effect

The most fundamental work was performed some years pre­ viously. Miillerand Stan [42] described an experiment which presented the nature of the enhancing matrix effect as a rather complete picture. They studied the effect of changes of catalytic activity in the injector on the decomposition of selected labile insecticides. The performance of the system (injector and column) was monitored for 100 consecutive injections, starting with 15 injections of a mixture of stand­ ards in pure solvent, followed by 85 injections of a spiked spinach extract.

Improvement during Re­ peated Analysis of Standards

The first injection of the standards into a clean liner (of 260 I!L internal volume, containing ca. 10 mg freshly deactivated glass wool) produced peaks for the rather stable internal standard (trl-allate) and the two organophosphates dimeth­ oate and azinphos-ethyl (Figure D33). The two carbamates, methiocarb and ethiofencarb, were completely decomposed, as evidenced by the presence of the resulting phenols. In the chromatograms obtained from subsequent analyses the peak of azinphos-ethyl increased in size, reaching a plateau after a few injections. The other peak areas did not change.

Improvement and Decline with Food Extracts

The injection ofthe first spinach extract strongly increased the responses of dimethoate and azinphos-ethyl, whereas that of tri-allate remained constant (presumably transfer has been complete to begin with). Continuing injections further

318

D 6. Problems with Quantitative Analysis

Standards

Spinach extract

AZinphos-ethyl Dimethoate

.---./ j'--./~ Methiocarb

11 13 15 I'S

n1i1fii18 30 k Jil

~

JO 10 Jo ifO ,00

Number of injection Figure D33 Dynamics of an enhancing matrix effect. Peak areas of three insecticides relative to tri-allate as internal standard. First 15 injections. solution in pure solvent; injections 16 to 100. resi­ due-free spinach extract spiked at the same concentration. (Miill., and Stan [42].)

increased the response of the two organophosphates; this reached a maximum after ca. 15 injections. at which it re­ mained for further 30 analyses. Theresponse then started decreasing with a simultaneous increase of variance. Peaks of intact carbamates were still observed after ca. 30 injec­ tions. but remained far too small. The work included results obtained analogously for on-col­ umn and PTV splitless injection. On-column injection pro­ duced fairly constant signals for the carbamates, confirming that the loss observed with splitless injection occurred in­ side the injector. There was. however. a significant increase of relative peak area for azinphos-ethyl (factor of two), sug­ gesting that activity in the column interfered. For dimethoate a slow increase of the relative peak area was followed by a rather sharp decrease after ca. 40 injections, probably reflect­ ing increasing contamination of the column inlet.

Temporary Deactivation

These results demonstrate the strong effect sample compo­ nents other than those to be analyzed can have on the on­ going analysis. The influence of contaminants is not always negative: in this instance it reduced the activity of the sur­ faces in the injector, i.e. rendered the surfaces more inert than careful silylation applied immediately before start­ ing the experiment. After more than 30 injections of the spin­ ach extract. the amount of matrix material accumulated to such an extent that it became detrimental, probably because its retentive power increased the residence time in the injec­ tor.

6.2.3. Effect on Quantita­ The clarity of the experiment is striking and reminds us old tive Analysis experience: the first injection in the morning often produces an un­ reliable result; the analyst injecting one sample in the morning and a few samples in the afternoon usually obtains results of

6.2. Enhancing Matrix Effects

319

quality inferior to those ofthe colleague who works with

a more regular rhythm;

"priming" of packed columns, i.e. repeated injections

of a contaminated sample before starting analysis, im­

proves performance.

Method Validation

The results also demonstrate a severe problem often ne­ glected in the validation of methods: chromatographic performance is unstable. Specification of the precision and accuracy of a chromatographic method by statistical evaluation of results is unsatisfactory because it does not properly quantitate the changes in the system. Changes oc­ cur over longer periods (aging column, liners becoming con­ taminated), but also from one run to another (such as after an injection with deactivating material).

Effect on Pesticide Analysis

Erneyand Poole [431 compared the response from constant amounts of 25 pesticides in pure solvent and in a residue­ free milk matrix. 18 of 19 organophosphorus compounds and 6 out of 7 organochlorine pesticides produced larger peaks when injected in the milk matrix. The response enhance­ ment ranged between 6 and 92 0/0. The experiments were performed in two laboratories and, as shown in Table D3, results deviated substantially. The Varian 3600 instrument was fitted with a splitless flash injec­ tion liner thermostatted at 250 "C, the HP 5890 instrument with the same liner at 220 "C. There is no indication of the sample volume nor of whether injection was performed by means of an autosampler. Columns were different, maybe also carrier gas flow rates. Table D3 Response of pesticides, injected in a milk extract. reletive to injection in a solution in pure solvent. (From Erney end Poole [43].)

Compound Methamidophos Acephate Omethoate Dimethoate Diazinon Ethion Parathion methyl

Varian 3600

HP 5890

1.51 1.83 1.78 1.44 1.27 1.36 0.90

1.28 1.36 1.23 1.11 1.06 1.09

Such differences are hardly of importance, however, because the values obtained are not thought to be typical for the two systems. The results show that the increase in response re­ sulting from the matrix is different for different components (probably larger where adsorption was more severe) and for

320

D 6. Problems with Quantitative Analysis

different systems. Not shown, but discussed above: the re­ sults are hardly more than a picture taken at a given mo­ ment and would have turned out different had the experi­ ment been repeated. Temporary Deactivation

The response enhancement was explained by the protec­ tive effect of matrix material resulting in increased trans­ fer of solute material from the injector into the column, i.e. temporarily suppressed adsorption or chemical activity in the injector. Hajslova et al. [441 provided a wealth of data on matrix ef­ fects observed during the analysis of oranges, cabbage, and wheat for a broad range of pesticides after clean-up by gel permeation chromatography (GPC). Although most matrix effects were positive (chlorothalonil being the worst with a doubled peak area), some negative ones were also observed, particularly after the analysis of a long series of samples. Polarity correlated with the effect, which confirms that ad­ sorption was the source of the problem. Bernal et al. [45] described similar phenomena for the analysis of grape must and suggested a method for optimized cleanup.

6.2.4. Proposed Solu­ tions

If the response of the components to be analyzed is calibrated by injection of standards in solvent, systematic errors often reach 30-70 %, sometimes a factor of two, results being too high.

More Efficient Cleanup?

Additional cleanup of the sample should eliminate the ma­ trix and render the sample more similar to the calibration solution. If the matrix effect is different for every compound to be analyzed and changes even during the day, it is best to eliminate it. This approach is usually of limited practicality since it means additional sample preparation steps which themselves introduce errors. Multi-methods are, furthermore, often de­ signed for a broad spectrum of components with widely dif­ ferent characteristics, inevitably resulting in broad overlap with by-products. It seems, furthermore, to solve the problem from the wrong end. The source of the error is severe loss during the injection of the standards; the analytical result for the sample is closer to the correct value. Following the philoso­ phy that high losses are seldom reproducible, improvement should be achieved by reduction of the losses during cali­ bration.

Deliberate Contamination?

Erney et al. [41] made attempts to achieve more or less dura­ ble deactivation of the injector. Because sample by-products seemed more effective than regular deactivation procedures, they injected a milk extract without clean-up or added some edible oil to an extract. This follows the idea of "prim­ ing" the system, with the hope that the active sites could be

6.2. Enhancing Matrix Effects

321

saturated for some time. The experiments failed since ma­ trix effects still enhanced the areas of omethoate and dimeth­ oate by a factor of 1.4 to 1.9 and activity remained unstable. Matrix-Matching Standard Solutions

Many authors have suggested preparing standard solu­ tions in extracts containing the sample matrix, i.e. by addition of the standards to an equivalent sample, prefer­ ably one free from the solutes of interest (a version of the method of standard addition). This has the advantage of bringing the results of the calibration closer to the proper values - rather than worsening the results of the sample to­ wards those of the calibration without matrix. Basically each sample should be analyzed with and without the standards added (classical method of standard addition) if matrices are to be truly matching. In reality it is probably sufficient to prepare the calibration solution with a similar matrix, still leaving open whetherthe analysis of, e.g., spin­ ach requires calibration with spinach or just with a vegeta­ ble. As long as the deactivating material is unknown, "equiva­ lent" or "similar" remain ill-defined. Often It might primarily be humidity, which in turn is influenced by components act­ ing as detergents. Erney et al. [46] investigated the dependence of the area en­ hancement effect on the concentration of the matrix material for a milk extract. As shown in Figure 034, there was a range of matrix concentrations in which the enhance­ ment reacted positively to more matrix material. It then re­ mained fairly constant and finally dropped again when the concentration was doubled and tripled beyond the maximum shown. Hindered evaporation from the matrix material was assumed to be the cause. Such results differ for each type of sample and also depend on the sample preparation proce­ dure. 01.8

~

c::: 1.7

I~-----------------­

Aceph~

Q)

~ 1.6

o

/

g.1.5 Q)

t-i

c::: 1.4 1.3

-..

Methamidophos

1.2

Dimethoate Ethoprop

1.1 /Ronnel

1

o

2

4

6

8

10

12

14

16

18

20

Amount of extract from milk Figure D34 Response ratios (response with matrix divided by that with­ out) for the pesticides indicated using solutions containing an extract prepared from 0-20 g of milk fat. (From Erney et al. [46].)

322

0 6. Problems with Quantitative Analysis

Recommended Procedure

The following procedure was recommended. 1 Identify the pesticides which are subject to matrix ef­ fects for a given type of sample and prepare a corre­ sponding data base. 2 Plot response against solute concentration for matrix­ matched solutions (calibration curve). 3 Prepare standard solutions by addition to samples free from these components. For all types of product of in­ terest, samples found to be free from the components are kept in stock for preparing such standard solutions.

Single Component Additive Simulating Matrix

Erney and Poole [431 searched for a single compound which could be added to the standard solution and the sample and would result in equal response enhancement. It should mask the active sites at least as well as did the sample matrix and be eluted from the column clearly before the solutes of inter­ est (non-evaporating material would accumulate and finally cause the performance to deteriorate as a result of exces­ sive chromatographic retentive power). The compounds tested, 1,2,3-tris(2-Gyanoethoxy) propane, N,N,N',N'-tetrakis(2-hydroxypropane)ethylenediamine, glyc­ erol, polyethyleneglycol 200, formic acid or formamide, were added at concentrations of 1-5 %. The matrix material in the milk extract was given as 0.2-0.4 % (including, e.g., water?). None of the compounds tested improved the response as well as the milk extract, i.e. the peaks of the standards (orga­ nophosphorus pesticides) remained up to 28 % smaller than in the extract.

Improved Sample Transfer

The term "matrix-induced chromatographic response en­ hancement" is, to some extent, an euphemistic expres­ sion for splitless injection with poor sample transfer. As discussed in previous sections, transfer should be virtu­ ally complete even in the event of some retention (adsorp­ tion) in the vaporizing chamber. In fact, the enhancing ma­ trix effects can be reduced by improving the transfer condi­ tions. As shown by Vincenti et al. [331, increasing the ca r­ rier gas flow rate (pressure pulse during splitless trans­ fer) improves the results. More complete transfer dur­ ing calibration with matrix-free solutions brings the peak areas closer to what they should be and, hence, reduces the enhancing matrix effects. Often enhancing matrix effects were large when fast autosamplers were used depositing the sample mate­ rial on to a packing. Nebulizing injection into an empty liner might perform better. Effects do not seem to be restricted to injection with a fast autosampler, however. Increasing the injector temperature tends to reduce interactions with surfaces and can also improve the re­ sults of calibration with clean solutions (see Erney et al. [46]).

6.2. Enhancing Matrix Effects

323

Finally it should not be forgotten that the effects result primarily from insufficient deactivation of surfaces, primarily of packing materials. Improvement of this is the primary task.

6.3. Reducing Matrix Effects

The presence of high-boiling or involatile material in the sample can cause reducing matrix effects, i.e. peaks for a given amount of solute from the matrix-containing sample are smaller than those from the mixture of standards in sol­ vent [47]. Occurrence of reducing matrix effects might be restricted to nebulizing injection into empty liners.

6.3.1. Contaminant. Simulated with DC-200

Figure D35 shows the influence of 5 % involatile material in a test sample on the peak areas of the C'O-C34 n-alkanes (so­ lution in hexane), using the relatively low molecular weight dimethylpolysiloxane DC-200 to simulate the contamination. Injections were performed manually by the hot needle tech­ nique (nebulization) into an empty 80 x 4 mm i.d, liner at 250 DC. retonve peak area

1.0

0.9 0.8

clean calibration mixture

:~

0.7 0.6

0.5

~

0.4

1 2

~:

0.3 0.2 0.1

l>

10

14

18

22

26

30

~

QJ

.Q

E

~

c:

c=

~ U

QJ

c­

-

QJ

Ci. E 10 c

'"

34

n-olkane

Figure D35 Reducing matrix effects in splitless injection. Peak areas obtained from a solution of ...alkanes in hexane ("clean cali­ bration mixture") are compared with those from injections no. " 2, 4, 5, and '0 of a test sample containing 5 'Yo DC­ 200. (From ref. [47].)

324

D 6. Problems with Quantitative Analysis

"Clean" Sample

Firstly, a clean (DC-200-free) solution in hexane was in­ jected into a freshly cleaned liner, using conditions ensuring virtually complete transfer of the solute material (carrier gas flow rate, 3 mLJmin, H2; splitless period, 40 s; injection at 30 °C column temperature, providing a strong recondensation effect). Peak areas decreased with increasing molecular weight of the alkanes, as is typical of losses in the syringe needle (1.2 III injections into an old injector with a fairly strong temperature drop towards the septum). The relative standard deviations of the absolute peak areas were below 5%.

Contaminated Sample

A solution in which 5 % of the solvent was replaced by DC-200 was then injected 10 times. Even the first injection into the clean liner resulted in a massive reduction in peak areas: the n-C10 peak was reduced by 15 %, even though the boiling point of this solute is below the injector temperature used (250°C). losses increased to ca. 40 % for later eluted alkanes. Re-injection of the needle content showed that these losses were not a result of increased retention inside the sy­ ringe needle, but of reduced efficiency of transfer from the injector to the column.

"Dirty" Sample compared with "Dirty" Liner

During the 10 injections of the DC-200-containing sample, a total of ca. 0.5 mg of DC-200 accumulated in the vaporizing chamber. The peak areas obtained from injections 1 to 10 did not, nevertheless, differ significantly (with the exception of that of n-C34' see below). Sample transfer was, there­ fore, affected almost exclusively by the DC-200 present in the sample. This was confirmed by the finding that the peak areas obtained from subsequent injections of the clean solution were almost identical with those at the beginning of the experiment, with the still clean injector.

Retention on Wall of the Liner

The only peak area to fall steadily with each injection of the "dirty" sample was that of the test component with the high­ est boiling point, n-C34. Obviously this component was increasingly retained by the ever thickening layer of DC-200

Table D4 Effect of the concentration of DC-200 in the sample on the efficiency of sample transfer; peak areas from the test samples containing DC-200 divided by those from the clean mixture. (From ref. [47].)

DC-200 [%)

10

14

0.01 0.1 1 5 20

1.0 1.0 0.87 0.86 0.88

1.02 1.0 0.81 0.69 0.78

Relative peak areas, n-alkanes 18 26 22 1.01 0.96 0.80 0.57 0.62

0.84 0.80 0.56 0.61 0.50

0.86 0.70 0.53 0.56 0.52

30

34

0.86 0.74 0.55 0.63 0.32

0.82 0.68 0.54 0.44 0.12

6.3. Reducing Matrix Effects

325

(a polymer actually used as a stationary phase), and increas­ ing amounts of it no longer reached the column during the splitless period. Effect of DC-200 Concentra­ tion

The experiment was repeated with test samples containing lower concentrations of DC-200.Average peak areas obtained from the DC-200-containing samples were divided by those from the DC-200-free test solution, such that a value of unity indicates no effect of DC-200 on the peak areas. As shown in Table D4, a DC-200 concentration as low as 100 ppm reduced the peak areas of ncC22 to n-C34 by 14­ 18 %. Such concentrations of high-boiling or involatile ma­ terial are present even in relatively clean samples. Losses increased strongly when DC-200 concentrations reached 0.1 and 1 % and amounted to about 50 % for alkanes above n­ C20·

6.3.2. Triglyceride. in the Sample Matrix

Similar experiments were performed with a test sample con­ taining triglycerides, the most common matrix material in biological and food samples. The test components were stig­ masterol, a plant sterol with a strongly adsorptive hydroxyl group, and n-triacontane (n-C30), frequently used as inter­ nal standard for sterol analysis in fats and oils. The test solu­ tions were either in hexane or in hexane containing 20 % olive oil (identical solute concentrations). The injector tem­ perature was again 250 cC.

Absolute Peak Areas

Figure D36 shows the peak areas obtained for n-C30 during a sequence of injections. First the clean test mixture was in­ 100

~-x,

,,

,,

peak area

,,

,

75

"

calibration "

,, mixture ,, ,, ,,

,,

50

,, ,, ,

25

,

,,

sample

x'

.x,

-,

,

I

i

5

10

15

x

number of iniection

Figure D36 Peak areas obtained from 100 ppm n-C 3 0 in either hexane or hexane containing 20 % olive oil during a sequence of 16 injections, starting with a freshly cleaned liner. (From ref. [47J.)

326

0 6. Problems with Quantitative Analysis

jected twice (simulating calibration of an analytical proce­ dure), using conditions ensuring virtually complete transfer into the column. For the first injection of the mixture con­ taminated with oil the peak area of n-C30 dropped by a factor of ca. five. Subsequent runs resulted in transfer of merely 10-15 % (with a high standard deviation). Quantitation of n-C30 by use of absolute peak areas calibrated by injec­ tions 1 and 2 would have produced results which were too low by a factor of 5-10. Effect of Oil on Liner Wall

The clean solution was re-analyzed as injections 13, 14, and 16. In contrast with the DC-200 experiment, transfer was massively reduced, indicating hindrance of sample trans­ fer by the oil on the liner wall. The proportion of the n­ C30 transferred was, however, still twice as high as for the sample containing oil. Hence calibration of absolute peaks areas by use of the clean mixture would still have resulted in a systematic error of a factor of two.

Relative Peak Areas

The second experiment simulated analysis with n-C30 as in­ ternal standard for the determination of stigmasterol in ol­ ive oil. Figure 037 shows the area ratios for a mixture con­ taining equal concentrations of the two components (but with different FID response and stigmasterol only ca. 85 % pure) for the above sequence of injections. 4

011 sample

C Q>

<;

5

10

15

number 01 injectIOn

Figure D37 Area ratios for equal amounts of n-C30 and stigmasterol in hexane and in a hexane solution containing 20 % olive oil; same sequence of injections as in Figure D36. (From ref. [47].)

Changing from the clean solution to the sample containing oil, the area ratio changed by a factor close to two. This means that the absolute peak area of stigmasterol was reduced more strongly than that of n-C30' presumably be­ cause of stronger retention in the layer of partly decomposed

6.3. Reducing Matrix Effects

327

and polymerized olive oil. Subsequent injections ofthe clean mixture (13, 14, and 16) resulted in area ratios which ap­ proached those from the oily sample. Hence, area ratios were affected less strongly by the matrix than abso­ lute areas. Increased Standard Devia­ tion

The oil in the sample not only caused the area ratio to change, but also the standard deviation to increase sharply. This dem­ onstrates once more that processes affected by deviations are usually poorly reproducible; it is the deviation from the correct result (complete transfer) which is poorly reproduc­ ible.

Selection of the Internal Standard

According to an often-heard rule, the internal standard should be selected to be eluted close to the solute of interest to furnish stable area ratios also in situations when proc­ esses are imperfect. The results in Figure 037 show that this is not generally true. Retention in the injector is often differ­ ent from that in the column. The polarity of the nearly black oil layer in the liner was higher than that of the stationary phase (SE-52) in the column and stigmasterol was more strongly retained than n-C30. The best internal standard is not only similar in molecular weight, but also similar in the parts of the molecular structure which determine ad­ sorption or retention in the injector.

6.3.3. Interpretation of the Experimental Results

The model considering the liner as a matrix-coated chroma­ tographic column seems to be deficient, because the above phenomena cannot be fully explained by partitioning be­ tween the gas phase and a retaining liner wall. For instance, retention of decane in a film of OC-200 at 250°C is low.

Droplets Retaining Solute Material

It should be remembered that all experiments were per­ formed by thermospray injection into an empty liner. Sam­ ples containing high-boiling or involatile material can­ not evaporate completely. After nebulization of the liquid and evaporation of the solvent the matrix material forms droplets which retain or adsorb solutes and transport them. Most droplets are transferred to the liner wall and "glue" the dissolved solute material to it (Figure D38) Evaporation from there is hindered. Other droplets might be "shot" to the bot­ tom of the injector, particularly when they are relatively large, as is assumed for samples containing high loads of non­ evaporating material (e.g. 20 % edible oil).

"Shots" by the Column Entrance

Doubling the duration of the splitless period (and other means of promoting transfer) hardly reduced the losses. This indicates that retention of the solute material in the matrix layer was strong or that the material was lost at the bottom of the injector. The observation that glass wool between the needle exit and the column entrance reduced the matrix ef­ fect (see below) suggests that losses from "shots" to the bottom of the vaporizing chamber are important.

328

0 6. Problems with Quantitative Analysis

~

II

clecn muture

dirly sample

.~ dirt layer

glass :', :

..

msert

,'"

Figure D38 A possible explanation of the redu'cing matrix effect in splitless injection with nebulization. Clean samples (left) evaporate in the gas phase and the vapor has little contact with the liner. Samples loaded with non-evaporating mate­ rial (right), however, "glue" the higher-boiling solute mate­ rial to the liner wall or pull it to the bottom of the vaporizing chamber. (From ref. [47].)

6.3.4. Effects on Quanti­ tative Analysis

The above results on the reducing matrix effect can be sum­ marized by four points. 1 Splitless injection of samples containing a high-boiling or involatile matrix can result in incomplete transfer of the solutes. Even components of rather low boiling point are affected. 2 The effect is observed even for the first injection of a contaminated sample into a still clean liner. The problem cannot, therefore, be solved by more frequent cleaning of the liner. 3 Sometimes only the contaminants in the sample are important; other times, especially if high-boiling solutes are involved, deposits of matrix material on the liner wall have an effect also. 4 Incomplete sample transfer resulting from the presence of high-boiling or involatile sample by-products causes increased standard deviations. The reducing matrix effect was investigated for splitless in­ jection, but it probably also affects results obtained from split injection.

External Standard Method

Quantitation based on external standards starts with (re­ peated) injection of the components of interest in pure sol­ vent. The samples are then injected under identical condi­ tions and the peak areas are compared: The above results

6.3. Reducing Matrix Effects

329

indicate that this can result in large systematic errors. Er­ rors can only be assumed to be small if the solutes are vola­ tile (more volatile than, e.q., n-decane) or the sample con­ tains less than ca. 50 ppm of poorly evaporating by-prod­ ucts. Internal Standard Method

Shatkayand Flavian [48,49] showed for packed column GC that even the method of internal standard{s) can lead to er­ rors. Matrix effects are one reason for this. Analyses using an internal standard are sensitive to unstable discrimi­ nation. The ratios of the areas of the internal standard and the components of interest must be equal for the calibration mixture and the sample. If splitless transfer is incomplete, its efficiency may vary for different solutes and depend on the concentration of non-evaporating sample by-products. Both matrix effects cause such unstable discrimination.

Example 1: Dioctyl Sebacate, n-C30, and Cholesterol

Dioctyl sebacate, n-C30, and cholesterol were analyzed in a mixture containing 5 % triglycerides in hexane (spiked, simu­ lating an extract from foodstuffs). First, relative standard deviations resulting from injection of a mixture of stand­ ards in hexane and a small amount of acetone were checked in order to see whether "the method works". The following results were obtained: Compound Dioctyl sebacate n-C30 Cholesterol

RSD(%) 4

6 6

The analyst concludes that he has chosen appropriate con­ ditions, and maybe that he has a "good" splitless injector. He proceeds to analyze the samples containing 5 % oil. Re­ producibility was also checked: Compound Dioctyl sebacate n- C30 Cholesterol

RSD(%)

20 30 30

He finds these far higher standard deviations irritating. In the absence of a convincing explanation he remembers some statistics and concludes that he needs a larger number of injections to obtain a sufficiently reliable result. Here he is wrong: the high relative standard deviations represent the tip of an iceberg and he should take it as a warning of poor sample transfer. 100 % transfer cannot vary by ±30 % be­ cause there is no transfer above 100 %. The mixture with oil containing the same concentration of solutes as the calibration mixture showed that peak areas obtained from the sample were, in fact, smaller than those from the calibration mixture. Peak areas (integrator counts/ 1000) were:

330

0 6. Problems with Quantitative Analysis

Dioctyl sebacate

n-C30 Cholesterol

Hexane 200 350 320

Hexane+oil 140 180 100

Assuming 100 % transfer of the solutes in the calibration mixture, only 70 % ofthe sebacate, 51 % ofthe n-C30, and 31 % of the cholesterol from the oil-containing sample reached the column. Quantitation by the external standard method would have led to results 30 % too low for sebacate and a factor of 3.2 too low for cholesterol. Quantitation of cholesterol with n­ C30 as internal standard and calibration of the response factor with the clean mixture would have led to a result 40 % too low. Example 2: Analysis of PNAs

Polynuclear aromatic compounds (PNAs) are usually ana­ lyzed by splitless injection because the concentrations of in­ volatile by-products tend to be excessive for on-column injection. Ranging from fluorene to coronene (or even fur­ ther), the PNAs comprise a broad range in volatility. With benzofluorene as internal standard (eluted in the center of the chromatogram) and a correction factor determined with a clean solution of standards, benzopyrene concentrations in cigarette smoke were systematically too low by a factor of 2-3.5. Apparently discrimination in the sample was more severe than in the calibration mixture. Effects of unstable discrimination can be overcome by using series of internal standards. As it was difficult to find a sufficient number of PNAs absent from the samples, n­ alkanes were preferred and response factors determined for the n-alkane nearest to the PNA. When discrimination be­ tween two neighboring internal standards was severe, the response factors were extrapolated, e.g. by using the aver­ age peak area of the two n-alkanes for a PNA eluted half way between them. There is no doubt that such sophisticated methods reduce systematic errors, but they still do not ensure accurate re­ sults: as shown for the mixture of stigmasterol and n-C30, discrimination does not parallel elution in the chromatogram. Discrimination would, for instance, obviously not change if a column of different selectivity were used.

6.3.5. Minimizing the

Reducing matrix effects can be diminished by optimizing the conditions during sample transfer. A high carrier gas flow rate is helpful, because it generally improves transfer, and the splitless period can be prolonged to improve the trans­ fer of material retained by the sample matrix. Experiments, however, indicate that improvements are modest. If, as a re­ sult of matrix effects, only 20 % of a solute was transferred,

Matrix Effect Optimized Conditions

6.3. Reducing Matrix Effects

331

doubling the splitless period increased this proportion to hardly more than 25 %. When samples are free from involatile by-products, injec­ tor temperatures can be fairly low. Increasing the injector temperature has, however, proven to be an efficient means of overcoming reducing matrix effects. The optimum injec­ tor temperature for real samples is near the limit above which degradation becomes a problem, because it weakens the retentive power of the non-evaporating material. It does, on the other hand, also accelerate the degradation of sample by-products in the injector, giving rise to "ghost" peaks.

Calibration with Matrix­ Containing Solutions

Systematic errors can be avoided when calibration of re­ sponse factors is performed with solutions containing the matrix of the sample. This is, however, a makeshift approach because it means that sample transfer during calibration is deteriorated to the same extent as during analysis of the sam­ ples. On moving further from properly functioning sample transfer, repeatability suffers and the danger of systematic error grows. Synthetic mixtures of the solutes of interest and the matrix material of the sample can sometimes be prepared, e.g. if the matrix consists essentially of triglycerides. More often, however, the composition and concentration of the matrix material are unknown and the method of standard addi­ tion must be used.

Classical Method of Stand­ ard Addition

The classical method of standard addition requires two ana­ lytical runs for each sample, one with addition of a known amount or concentration of the solute(s) of interest, the other without. To avoid the use of small differences between peak areas (suffering from the random errors of both determina­ tions), the amount of solute material added should corre­ spond to at least twice the content of the sample.

Analog of the External Standard Method

The second procedure resembles the method of external standards, because a series of samples is analyzed with a single calibration. Peak areas per amount or concentration of solute material are again calibrated by the method of stand­ ard addition, but the calibration for the first sample is also used for all subsequent samples. This time-saving method is appropriate if all samples contain a similar matrix at a similar concentration. It must be checked whether the increasing amount of matrix material accumulated in the injector causes drifting peak ar­ eas or area ratios (discrimination). On the other hand, fre­ quent cleaning of the liner may not be appropriate be­ cause changes in adsorptivity and enhancing matrix effects are most accentuated for a freshly cleaned injector.

Analog of the Internal Standard Method

In general, quantitation on the basis of internal standards or surrogates is clearly preferable to externalstandard proce­

332

D 6. Problems with Quantitative Analysis dures. Calibration of the response factors must, however, be performed by adding known amounts of the solutes of

interest and of the internal standard(s) to the sample. If a series of samples is analyzed, it is again important that the sample matrices are similar and that progressive injec­ tor contamination does not cause drifting results. 6.3.6. Glass Wool in the

Liner?

In 1984, experiments were performed to determine whether glass wool would reduce the matrix effect [501. C'O-C30 fatty acid methyl esters in hexane and in hexane containing 0.1­ 10 % triglycerides were injected manually (1.5 ut., hot nee­ dle) into a light plug of silanized glass wool ca. 15 mm long positioned such that the inserted syringe needle entered it by about 5 mm (Carlo Erba, mod. 4160; 250 "C},

Reduced Standard Devia­ tions

For the clean test sample, relative standard deviations of absolute peak areas were almost equal, irrespective of the presence of glass wool (Table 05). On addition of triglycer­ ides, they increased 3- to 5-fold if no wool was used. Glass wool almost completely suppressed this increase for solutes up to a certain volatility limit, and this limit became lower the higher was the triglyceride concentration. With the injec­ tor at 250 ec, the C26 and C22 esters were at this limit for 1 and 10 % triglycerides, respectively. For higher-boiling es­ ters, the relative standard deviations increased at least as strongly as those measured in the absence of glass wool.

Efficiency of Sample Transfer

The rule that high standard deviations in splitless injection should be regarded as the tip of an iceberg, poor solute trans­ fer being the bulk, also applies to these results: they were directly related to losses. The more completely a solute was transferred into the column, the smaller was the room for random errors.

Table 06 shows losses of solute material as a result of sam­

ple transfer being hindered by the matrix effect. They were calculated by comparison with the solution free from oil. If no glass wool was used, the losses were high: with a triglyc-

Table D5 Relative standard deviations [%] with (+GW) and without (-GW) glass wool in the liner. Fatty acid methyl eaters C 10-C30 in hexane or hexane containing 0.1-10 % oil. (From ref. [50].)

O%oil +GW

Ester

-GW

10 14 18 22 26 30

1.3 2.1 2.2 2.0 2.1 3.3

1.2 2.1 2.4 2.6 1.9 2.3

0.1 % oil -GW +GW 1.7 3.7 6.2 3.0 3.7 3.7

1.4 1.9 2.5 1.8 2.1 2.1

-GW

1 % oil +GW

3.2 10 5.0 6.5 5.0 6.9

1.4 2.8 2.9 1.8 3.4 7.1

10 % oil -GW +GW 8.3 12 9.4 15 18 43

1.3 1.0 2.8 7.7 24 37

..

6.3. Reducing Matrix Effects

333

eride concentration as low as 0.1 %, 10-20 % of the solute material was lost (including the rather volatile C14 ester). With wool in the liner, these losses were practically elimi­ nated. The relative standard deviations given in Table 05 indicated a limit of solute volatility beyond which the repro­ ducibility rapidly became worse. This same limit is also ob­ served for the losses. Table D6 Proportion ["10] of sample material not transferred into the column as a result of the presence of 0.1·10 "10 oil, as ob­ served with and without glass wool (GW) in the liner; same experiment as Table D5. (From ref. [50].)

Ester 10 14 18 22 26 30

0.1 % -GW +GW

3 9 14 16 21 21

0 2 2 0 2 5

1% -GW +GW 6 16 30 30 29 32

0 5· 7 5 10 27

10 % -GW +GW 20 32 37 37 42 48

0 4

3 10 44 58

Filter Effect

Evaporation from the surface of glass wool distinguishes between evaporating and non-evaporating materials more sharply than vaporization after nebulization in an open tube. Such filtration reduces contamination of the column inlet, but may also start in a volatility range encompassing that of the solutes of interest. With an injector at 250 DC, op­ timum quantitative analysis was obtained for methyl esters up to C22-C30, depending on the concentration of the edible oil.

Concluding Remark

This study remained incomplete insofar as neither the mechanisms resulting in the reducing matrix effect, nor their elimination by use of wool have been properly understood. In the practice of routine analysis, the use of wool positioned at the tip of the inserted needle improved results and reduced column contamination in some instances, but not in others. Wool cannot generally be recommended because of its adsorptivity and chemical activity.

334

0 7. Reconcentration of Initial Bands

7. Reconcentration of Initial Bands One of the fundamental laws of chromatography states that the initial solute bands (the bands at the beginning of the chromatographic process) must be sharp and contribute negligibly to the final peak width. More specifically this means that 1 the sample must enter the column during a short period of time; and 2 the column inlet over which the sample material is spread must be short compared with the length of the solute band eluted from the column. Only Split Injection Produces Sharp Bands

Unless split flow rates are small, split jnjection creates initial bands which are sharp in both senses, i.e. sample vapor en­ ters the column during maybe a second and the amount is so small that it forms a short plug in the column inlet. All other injection techniques violate one or both of the laws mentioned above. In splitless injection, the sample material enters the column over a period which can be as long as several minutes, and if solvent recondenses in the column inlet. flowing sam­ ple liquid may spread the solute material over a column length of up to several meters. Splitless injection should, therefore, produce broad and distorted peaks. This was the reason why nobody seriously experimented with splitless injection for a long time. There are, however, methods for reconcentrating these bands, i.e. sharpening them before the separation process is started. Injection techniques such as splitless injection are, therefore, reliant on tricks, and the user must know about these if he is to achieve good chromatography.

7.1. Distinction between the Two Band Broadening Effects

Two band broadening effects are distinguished, because their characteristics are fundamentally different, as are the meth­ ods available for their reconcentration [51].

7.1.1. ... in Space

If solute bands are broadened "in space", the common char­ acteristic of the bands of all the solutes is spreading over the same length of the column inlet. Often the spread­ ing results from a flow of the liquid sample in the column inlet, which is no different for volatile and high-boiling, or polar and apolar compounds. Since the effect on peak dis­ tortion also depends on column diameter, the broadening is not termed "in length", but "in space". The distribution of the sample material within the "flooded zone" is usually uneven, resulting in distorted, often even

7.1. Distinction between the Two Band Broadening Effects

335

split peaks. A description of the spreading process during on-column injection was given in ref. [52]. In splitless injec­ tion a similar process occurs if the sample solvent recon­ denses in the column inlet. Characteristics

Band broadening in space is recognized by the following characteristics. There is no broadening effect for peaks eluted at col­ umn temperatures up to ca. 50° above that during injec­ tion. Then the broadening effect gradually increases and reaches its full extent ca. 100° above injection. 2 In an isothermal run (more than 50° above the injection temperature), peaks are broadened in proportion to their retention time (Figure D39). 3 In temperature-programmed runs (more than 50° above the injection temperature) the distortion of all peaks is similar (Figure D40). 4 All solutes are affected similarly, irrespective of chemi­ cal characteristics. A contaminated column inlet can also cause peak distortion or splitting, but affects different compounds differently. 5 The peak distortion pattern usually changes if conditions are altered and can also be poorly reproducible from one run to another.

7.1.2.••• in Time

If solute bands are broadened "in time", the common char­ acteristic for all solutes is equal broadening in terms of gas chromatographic retention time (or millimeters on the chart paper). For all compounds, the last material enters the column with the same delay and migrates behind that entering first by the same distance in time. C20

Band Broadening in Space C21

Band Broadening in Time

wl_

C21 C20

i_A_i!

C22 C23

I C20

C21

No Band Broadening C22

C24 C23

Figure D39 Comparison of the two band broadening effects for isothermal runs. Left: band broadening in space; peak broadening growing in proportion to the retention time. C2 0-C2 4 n-alkanes in acetone; 1.5 IiL on-column injection at 30°C; after 30 s, ballistic heating to 180°C. 10 m x 0.30 mm i.d. glass capillary column coated with a 0.07 lim film of OV·1 (methyl silicone). Upper right: band broadening in time in splittess injection; constant broadening of all peaks, determined by the duration of the sample transfer. Isothermal run at 180°C. Bottom: no peak broadening; split injection of a more concentrated solution at 180 °C.

336

D 7. Reconcentration of Initial Bands

Band Broadening In Space

C21 C20

C24

I

C22

C23

Band Broadening in Time

1I C23

C22 C21 C20

Figure D40 Comparison of the two band broadening effects in temperature-programmed runs, using the same column and test mixture as in Figure D39. • Band broadening in space is visible only at column temperatures more than about 50 °C above that set during injection and remains constant throughout the rest of the tem­ perature program. 1.5 ilL on-column injection at 30 °c, temperature program from 120 °c. • Band broadening in time results in peak broadening which is reduced and finally elimi­ nated by temperature programming (cold trapping). Splitless injection at 130 °c, tem­ perature program started after elution of the solvent peak. The most important source of band broadening in time is slow sample introduction into the column. It occurs in splitless injection, in split injection with low split flow rates, and as a result of slow transfer from traps, packed pracol­ urnns, etc. Characteristics

Band broadening in time is recognized by the following char­ acteristics. In isothermal runs, all peaks are broadened approxi­ mately equally as measured in seconds or millimeters on the chart paper (Figure 039); strictly speaking, the contribution to the final peak widths is the same. 2 Peak broadening is severe at the column temperature during injection and diminishes with increasing tempera­ ture (the exact opposite of band broadening in space). 3 No peak broadening is observed for compounds eluted at temperatures more than 50-80 0 above injection (Fig­ ure 040). This characteristic is used for the reconcentra­ tion ("cold trapping"). 4 Peak distortion is reproducible and reflects sample trans­ fer. A typical example is shown in the middle chromato­ gram of Figure 039.

Different Initial Band Lengths

In contrast with band broadening in space, bands broadened in time are not spread over equal lengths of the column in­ let. Volatile solutes form long initial bands, because the most advanced material migrates relatively far into the column before the last material enters, whereas high-boiling solutes are retained in a short section of the column inlet by cold trapping (shown schematically in Figure D41).

7.1. Distinction between the Two Band Broadening Effects

337

Band Broadening in Time Solutes spread equally in respect of GC retention time High boiling solute

J

Volatile solute

7/777777777777777777777777777777777777777. Column inlet Reconcentration by: • solvent effects . cold trapping

Band Broadening in Space Solutes spread equally in respect of column length Independent of volatility

7/777777777777777777777777777777777777777 Reconcentration by retention gap technique

Figure D41 Comparison of the two types of band broadening in terms of the lengths of the initial bands.

No Effect of Changes in Retentive Power

Band broadening in time is not influenced by changes in re­ tentive power (film thickness). In particular, a precolumn free from stationary phase (retention gap), used for reconcen­ trating bands broadened in space, does not change band widths in terms of time. It is true that the solute bands are spread over a longer section of an uncoated column inlet because the most advanced material travels further during a given period ohime. In terms of time, however, the advanced material is still separated to the same extent from the rear material, and it is this difference in time, which determines peak broadening.

7.2. Band Broadening in Time

The shape of a peak resulting from splitless injection with­ out proper reconcentration is described in order to facilitate its recognition. This should prevent a column being con­ sidered poor when the problem is insufficient reconcentra­ tion. Furthermore, peaks obtained from non-reconcentrated bands provide information about the sample transfer.

7.2. 1. Shape of the Band

A simple experiment enables the observation of initial band shapes. 300-500 III of diluted methane (e.g. fuel gas diluted in air) is injected splitless. The stationary phase is irrelevant as methane is virtually unretained. The peak primarily re­ flects the shape of the initial band, possibly somewhat "rounded off" by diffusion processes in the column.

Leading Edge

If the sample is introduced close to the column entrance (as appropriate for splitless injection), the upward slope of the initial band is steep (Figure D42), because concentrated sample vapor enters the column immediately. If the syringe needle is too short, however, a plug of carrier gas will sepa­

338

0 7. Reconcentration of Initial Bands rate the vapor from the column entrance, the vapor moves slowly towards the column, and the front of the vapor cloud will have been diluted before its arrival, i.e. the pen rises relatively slowly and with some delay. Concentrated vapor

enters the column

Vapor inside the injector is Increasingly diluted Split exit opened

I Injection

\

Band shape when the split exit is not _ opene~

------------~---

Purged material

Figure D42 Initial band shape reflecting sample introduction into the column.

Falling Edge

At the beginning of sample transfer concentrated vapor en­ ters the column, as shown by the maximum height of the first section of the signal. Efficient sample transfer is charac­ terized by a rather rapid downslope, indicating rapid empty­ ing of the vaporizing chamber (left-hand peak in Figure D43). If sample transfer is slow, however, a fairly constant amount (concentration) of sample vapor enters the column over an extended period of time, resulting in a peak with a slow downslope (right-hand curve in Figure D43). In this instance, a significant amount of solute is frequently lost through the split exit because of premature opening of the latter. High transfer efficiency

Low transfer efficiency Slow upslope: short syringe needle

~'-~ wuAZiiW Material flushed through split exit

Figure D43

Shapes of initial bands. Left: high carrier gas flow rate. long

syringe needle. Right: low carrier gas flow rate. short sy­

ringe needle.

High Gas flow Rate: No Reconcentration Required

Increased carrier gas flow rates lead to substantially improved peak shape. Figure D44 shows peaks obtained from 500 ul, splitless injections of methane diluted with air. At a carrier

7.2. Band Broadening in Time

339

gas flow rate of 0.8 mUmin, transfer remained incomplete even after 3 min. At 1.5 mUmin, it took more than 75 sand the initial band was correspondingly broad. At 8 mUmin, the shape of the methane is acceptable under most con­ ditions. Splitless or direct injections into 0.53 mm i.d, col­ umns are, in fact, usually performed without reconcentra­ tion effects. 4.4mlmlin 2.7mlmlin 1.5mlmlin 0.8mlmlln

12s ,

,

I

,) »1-1---.------r----r+

1234(minl

I

2

3

Figure D44 Methane peaks obtained by splitless injection at different carrier gas flow rates. 20 m x 0.25 mm i.d. separation col­ umn; splitless period. 3 min; 80 x 4 mm i.d.liner; 30-150 kPa inlet pressure (hydrogen).

Duration of the Splitless Period

The pen of the recorder drops back to the baseline when the split exit is opened (after a delay equal to the retention time of the column) and the sample material remaining in the in­ jector is purged through the split outlet. The removed tail of the band represents lost sample material and enables rapid determination of whether the duration of the splitless period was sufficient.

7.3. Reconcentration by Cold Trapping

There are two techniques for focusing bands broadened in time: cold trapping and solvent trapping (solvent effects). Sometimes the terms "thermal focusing" or "stationary phase focusing" are used synonymously with "cold trap­ ping", but we retain the older term [17]. "Cold trapping" might suggest an actively cooled trap, but here this is not so, as will be shown below.

7.3. 1. Principle

Reconcentration of bands broadened in time requires the migration of the solute material first entering the column to be slowed or (ideally) stopped to give the material enter­ ing last a chance to catch up. At the end of the transfer, or shortly after, the solute material must be released in such a manner that all the material starts the separation process as a sharp band.

Temporary Increase in Retentive Power

Arresting the migration ofthe solute material is achieved by means of a temporary increase in gas chromatographic re­ tention, either by reducing the temperature (cold trapping effect) or temporarily increasing the thickness of the film of

340

0 7. Reconcentration of Initial Bands stationary phase employing the sample solvent as liquid phase (solvent effects). The extra retentive power is re­ moved when the column is heated orthe solvent evaporated.

Reduced Column Tempera­ ture

The cold trapping effect exploits one ofthe characteristics of band broadening in time described above, namely that peak broadening is reduced and eventually disappears when the column temperature is increased during a run. This is be­ cause the migration speed ofthe solutes of interest is low if the column temperature during transfer is well below their temperature of elution. For practical work this means that the column temperature during splitless injection must be reduced substantially below the elution temperature of the solutes of interest; it is increased again for elution ofthese components at the end of the splitless period. The solvent and other volatile sample constituents are, of course, not retained in this way and, therefore, not reconcentrated.

7.3.2. Reconcentrating Power

The reconcentration obtained by cold trapping is determined by the ratio of the migration speeds of the solutes of interest at the temperatures during injection and elu­ tion. If, for instance, the advanced material of a solute mi­ grates ten times more slowly (less far) during the sample transfer period than at the elution temperature, the band is reconcentrated by a factor of ten. This raises the question of the temperature difference required to change the velocity of gas chromatographic movement by a given factor.

The "15-Degree Rule"

The dependence of migration speed on column temperature may be roughly estimated by the" 15 degree rule", which was derived from a study of the effect of film thickness on the elution temperatures of selected solutes in temperature­ programmed runs [53,54). Changing film thickness by a fac­ tor of two caused the elution temperature of a component to increase or decrease by ca. 15°. As retention times at a given column temperature are proportional to film thickness, this also means that the migration speed of a solute is dou­ bled or halved if the column temperature is changed by 15°. A more thorough treatment is given by Bartle [55).

Estimated Reconcentration Factors

Applied to the cold trapping effect, the 15° rule predicts that bands are reconcentrated by a factor of 2 for each 15° reduction in column temperature below the elution tem­ perature. Table D71ists such calculated reconcentration fac­ tors.

7.3.3. Reconcentration Required

The width of the initial band must be reduced to such an extent that it contributes negligibly to the final peak width. Thus the reconcentration factor required depends on the width of the initial band and on the chromatographic broadening of the final peak.

.... 7.3. Reconcentration by Cold Trapping

341

Table D7

Dependence of reconcentration by cold trapping on tempera­

ture differences between injection and elution of the solute

of interest.

Temperature difference [0]

Reconcentration factor

15

2 4 8

30 45

60

16

75

90

32 64

105

128

Example

As an example, an initial band width of 60 s and a chro­ matographic peak width of 4 s are assumed. The final peak width is determined from the width of the initial band and the chromatographic band broadening by a square relation­ ship (Sternberg [56]). An initial band of 2 s broadens a peak 4 s wide to 4.5 s, or by little more than 10 'Yo. If this is consid­ ered acceptable, the initial band must be reconcentrated by a factor of 30, which is achieved by cold trapping in­ volving a temperature difference of 75°.

Temperature-Programmed Run

The temperature difference required for efficient cold trap­ ping can also be deduced experimentally from a tempera­ ture-programmed run starting at the injection temperature. Figure 045 compares chromatograms of diesel oil diluted in pentane. The lower chromatogram was obtained by split injection, i.e. shows peak widths (separation efficiencies) determined solely by the column. For the upper chromato­ gram, a more dilute solution was injected splitless at 50°C, which precluded reconcentration by solvent effects. At low elution temperatures, reconcentration by cold trap­ ping was weak and peaks correspondingly broad. Peaks be­ came sharper as the temperature was increased, and were no longer significantly broadened when it was ca. 60° above that during injection. A temperature difference of 60° was sufficient, primarily because peaks were broader than the 4 s assumed in the example calculated above.

Conclusion

A temperature difference between injection and elu­ tion of 80-90° is the minimum if peak broadening is to be avoided. For less critical applications, a temperature differ­ ence of 60° is sufficient.

7.3.4. Practice of Cold Trapping

Cold trapping complicates analysis involving splitless injection, because the column must be cooled for each injec­ tion, and fully isothermal chromatography is impossible.

342

0 7. Reconcentration of Initial Bands

C 14

SPLITL.ESS C 17

C II C 20

SPLIT

175

0

150'

125·

100 0

75

0

50

0

Figure D45 Cold trapping effect during a temperature-programmed run. Bottom: split injection of diesel oil dissolved in pentane on to a 13 m II 0.30 mm i.d. glass capillary column coated with an 0.5 11m film of SE-52. Temperature program of 5 o/min from 50 to 180°C. Top: splitless injection under conditions ruling out solvent effects. Early eluted peaks are severely broadened (and correspondingly smaller). n-C , 4• eluted ca. 60° above the injec­ tion temperature. produced the first peak of virtually perfect shape.

Fast Procedure for Manual Injection

As an example, the component of interest is assumed to be eluted at 200°C (isothermally or during a temperature-pro­ gram). The above rule suggests a maximum injection tem­ perature of 120°C. The GC oven can be reconditioned to this lower temperature. As this is time-consuming, for manual injection the oven door can opened instead, which cools the column to below 120°C within seconds (Figure D46). It is not relevant whether the oven walls remain hotter - touch­ ing the column with the fingers gives a better idea of the temperature than the thermometer of the instrument. Injec­ tion is performed as soon as the column has cooled suffi­ ciently. Cooling is continued, as the current of ambient air insulates the column from the hot oven walls. When sample transfer is complete and the split exit re-opened, the oven door is closed and the column temperature increased to elute the solute. Whether this is achieved by rapid temperature programming or by ballistic heating depends on the reten­ tion time reproducibility required (see below).

7.3. Reconcentration by Cold Trapping

343

Solute of

interest

l!!

I~ven door ,opened

~ Gl

a. E

.l!!

c:

E

:>

o o

Previous run

\

80°C

t

Oven door closed, heating Time

~

Figure D46 Changes in column temperature required to achieve cold trap­ ping: rapid method for manual injection.

Possible Side Effect: Solvent Recondensation

If the column temperature drops considerably below the boiling point of the solvent, solvent can recondense. Recondensed sample flows into the column and spreads the dissolved solutes, causing band broadening in space. Resulting peak broadening is usually weak, but ifthe solvent does not wet the surface of the column inlet, flooded zones can be long and peak broadening severe. Recondensed solvent can, furthermore, dissolve stationary phase material in the column inlet, carry it further into the column, and deposit it at the point where the last portion of the solvent evaporates ("phase stripping"). Such problems seldom occur, because the usual solvents do not recondense when the column is warm. Severe band broadening in space has, however, been experienced with aqueous samples or samples in water-containing solvent mixtures. Then cooling ofthe column to temperatures lower than necessary must be avoided.

Reproduction of Absolute Retention Times

If injection is performed manually and a cold trapping effect is involved, reproduction of absolute retention times is of­ ten insufficient for reliable automated peak recognition. Automated injection more accurately reproduces the chro­ matographic cycles, including the duration of the cooling period. The overall retention time of a component is made up of the splitless period, the time required by the heating step, and the retention time involved in the separation process. If injection is performed manually, deviations of absolute re­ tention times result primarily from the second contribution, because the amount of heat left in the oven walls varies from one run to' another, and this affects the amount of heat re­ quired to increase the oven temperature.

UNlVERSlOAD DE ANTIOQUlJ" BIBLI01'BCA CENTRAL

344

D 7. Reconcentration of Initial Bands The rumor that rapid ("ballistic") heating of the oven could damage capillary columns has survived from the early days of capillary GC. We have never found experimental evidence of this, neither in the literature nor in our laboratory. If repro­ duction of absolute retention times is important, however, ballistic heating should be avoided for another reason: the time required for a given temperature increase depends on the amount of heat consumed, i.e. on the residual heat in the GC oven and, thus, on the duration of the preceding cooling period.

7.3.5. Problems with Disturbed Baselines

Cooling and heating of the oven, as required for cold trap­ ping, often produces unstable baselines and "ghost" peaks (peaks not belonging to the sample injected). To facilitate the recognition of the sources, the mechanisms involved and the resulting phenomena are listed below.

Changing Carrier Gas Flow Rate

Standing currents (baselines) of GC detectors usually depend on the flow rate of the carrier gas. This flow rate decreases when the column is heated. If hydrogen is the carrier gas, less fuel gas is fed into the detector; if other carrier gases are used, less inert gas dilutes the detector gases or otherwise affects the standing current. This may just as easily increase or reduce the baseline. Detectable volatile impurities in the carrier gas (e.g. meth­ ane for the FID, or oxygen and humidity for the ECD) hardly cause disturbance as long as the input to the detector is con­ stant, i.e. chromatography is isothermal: the baseline is flat, although higher than it might otherwise be. In temperature­ programmed runs, however, the amounts of impurities reaching the detector decrease with increasing column temperature by the same extent as the flow rate of the gas through the column decreases. This causes the baseline of the FID to fall.

Contaminated Column Outlet

Contaminated column outlets or detector components (e.g. the flame jet of the FID) may also cause the baseline to fall as temperature increases. After a run reaching a high tem­ perature, the column outlet or the detector region may be­ come contaminated with column bleed or high-boiling sam­ ple material. Most outlet systems have a cold spot, which first retains and later slowly releases such contaminants. As the detector block is essentially an isothermal system, the amount of such material released into the detector cell per unit time depends on the carrier gas flow rate, i.e. the base­ line tends to drop during temperature-programmed runs (at least for the FID). This problem can normally be reduced by increasing the temperature of the detector block, preventing contami­ nation at the cold spot. This, however, also causes the col­ umn bleed to become more obvious, as it is discharged im­ mediately into the detector.

7.3. Reconcentration by Cold Trapping

345

Overheated Column Outlet

When the temperature ofthe detector block is increased, the fall in the baseline during a temperature program may also become more pronounced, overheating of the stationary phase in the column outlet being the source of the bleed. Attachment of a short piece of uncoated (but deacti­ vated) capillary to the column outlet would solve this problem. Simple patience usually also serves the purpose, because such problems tend to disappear by themselves. The above effects can be simulated by reducing the carrier gas inlet pressure at any isothermal column temperature. This should be by a factor of about two, because a long range temperature program easily halves the mass flow rate of the carrier gas.

Bleed from Ferrules

Increase of the baseline during a temperature program of­ ten results from column bleed, but bleed from the ferrule of the column attachment to the detector has a similar effect, because the temperature of the ferrule varies with the oven, i.e. it increases when the oven is heated. The bleed leaving the ferrule towards the detector sooner or later reaches the stream of make-up or fuel gas directed towards the detector (Figure D47). The delay between the emission of the bleed and the rise in the baseline depends on the con­ struction of the detector block. Bleed from the ferrule can be distinguished from column bleed by the reaction time of the baseline on rapid oven cooling by opening the door: it falls almost instantly if the bleed is from the column but only slowly if the ferrule is the source (slow cooling of the fitting and slow diffusion into the detector gas, as nicely shown by Oehme [57]). Detector

i===: ""',.---- Supply of air

~=== ~ Supply of fuel and

make-up gas

Column outlet

Figure D47 Schematic diagram of the detector block. A ferrule bleeding in the column attachment releases material into the gases directed towards the detector. Because the ferrule tempera­ ture varies with the column temperature, such bleed may produce drifting baselines similar to tho_ arising from col­ umn bleed.

346

D 7. Reconcentration of Initial Bands

Delayed Release of Solvent

Many chromatograms are disturbed by a "hump" eluted af­ ter the solvent peak, similar to that in Figure D48. It can arise from solvent slowly released from the surface of the capillary column. The solvent is retained in small pores, analogous to retention by molecular sieves [51. Temperatures considerably above the solvent boiling point are required to release it. Such pores may, for instance, be generated by leaching during column preparation. In isothermal runs, the baseline after the solvent peak slowly returns to the original level. Rapid heating accelerates the release of the vapor, forming a "hump". The amount of solvent retained, i.e. the size of the "hump", depends on the method used to prepare the column, the solvent, and the temperature, but hardly on the amount of solvent injected or the injection technique applied. Usually "humps" are par­ ticularly large for solvents of small molecular weight, such as acetonitrile and methanol; with ECO, water (humidity in the sample) may form large "humps". There is no possibility of "deactivating" the column in this respect, nor does the effect disappear during prolonged use of the column. The "hump" can be reduced by using an­ other solvent, by slowing the temperature increase (ren­ dering it broader and correspondingly lower), or by increas­ ing the column temperature during injection.

"Hump" of solvent released after a delay

1-------1 Rapid temperature increase

Figure D48

Typical pictures of solvent peaks distorted as a result of sol­

vent retained in pores in the column surface. Rapid tempera­

ture increase accelerates the release and leads to the forma­

tion of a ·'hump".

7.3.6. "Ghost" Peaks

"Ghost" peaks arise from materials which were not injected with the sample. They can have many origins, and it is often difficult to trace these. The following paragraphs should help the reader if "ghost" peaks cause him trouble. "Ghost" peaks are not only encountered after splitless injec­ tion involving cold trapping, but with temperature pro­ gramming following any type of injection, or even with­ out injection. They are discussed in this context because they are particularly troublesome in trace analysis.

7.3. Reconcentration by Cold Trapping

347

Sharp Initial Bands

Well shaped "ghost" peaks presuppose that the correspond­ ing compounds start chromatography as sharp initial bands. Substances originating from inside the column (e.g. bleed from the stationary phase) do not form normally shaped peaks and those released from the column outlet do not form peaks at all. Hence "ghost" peaks represent material intro­ duced into the column by the carrier gas or released from the column inlet (e.g. by degradation of components deposited there by previous injections).

Reconcentrated Bands

In isothermal runs, materials carried continuously into the column by the carrier gas are eluted at a constant rate (after some equilibration, they leave the column at the same rate as they enter) and just increase the baseline slightly. Most probably such contaminants will not even be detected (except by MS). If the column is cooled, however, e.g. for splitless injec­ tion with cold trapping, all except the most volatile materials introduced by the carrier gas are retained (cold trapped) in the column inlet and form a narrow initial band. When the column is heated, such contaminants are chromatographed together with the solutes of interest. "Ghost" peaks are often somewhat broadened or show some tailing, firstly because reconcentration over a long cooling period may be incomplete (the initial band has a width of, perhaps, more than 30 min) and, secondly, because the cold trapping effect becomes weaker as the column temperature is increased, i.e. the material introduced during programming is no longer co-eluted with the bulk accumulated previously.

Contaminated Carrier Gas

The material forming the "ghost" peaks can originate from the carrier gas itself. If so, the "ghost" peaks grow as the gas cylinder empties, because the constant vapor pressure represents an increasing proportion of the decreasing car­ rier gas pressure and, hence, an increasing concentration. It is probably caused by incidental contamination of a cyl­ inder and has little to do with gas purity. In fact, a compo­ nent present at a level of 0.01 ppm can cause a large "ghost" peak, but would correspond to merely 1 % of the total impu­ rity of a 6.0 grade gas

Gas Supply

More often "ghost" peaks are from the carrier gas supply line to the injector, e.g. as a result of contamination by pre­ vious overloading of the vaporizing chamber (Section D2.6.2). Pressure regulators of the GC instrument or the gas cylin­ der are other sources (especially if equipped with plastic membranes). Many laboratories have centralized the gas supply or ac­ cept long supply lines from gas cylinders located outside the laboratory. All too often, such supply systems release con­ taminants for several years. It is difficult to clean them. Some achieve success by rinsing them every few days by means

348

D 7. Reconcentration of Initial Bands

of a high gas flow rate through a large leak. With hydrogen,

some care is required as the gas may ignite itself.

Degraded Contaminants from the Injector

In the injector, high-boiling or involatile material can

slowly degrade into more volatile products (e.g. triglycer­

ides into free acids) which are constantly rinsed into the col­

umn. Considering the amount of contaminants often accu­

mulated, there is sufficient material for a large number of

"ghost" peaks!

Often the release of such degradation products is not con·

stant. "Ghost" peaks can be small when running an "in­

strument blank" (by cooling and heating the column as dur­

ing analysis, closure and opening of the split exit, insertion

of the syringe needle, but not injecting anything). The sizes

of the "ghost" peaks usually increase when certain solvents

are injected, creating the (wrong) impression that the peaks

originate from the solvent. Rather typically there is a further

massive increase on introduction of a real sample. Perhaps

the sample by-products "soak" the lacquer-like layer and fa­

cilitate the release of volatile material, or components such

as humidity initiate degradation of involatile material.

This behavior is also responsible for "memory" effects

(appearance of solutes of interest although they were not

injected). "Memory" effects can be strongly enhanced when

a real sample is injected and, hence, it is dangerous to sub­

tract the size of a "memory" peak observed on injection of

solvent from that observed when a sample is injected.

Septum Material

When the syringe needle is introduced into the injector,

septum material is often transferred into the injector, per­

haps on the outer needle wall, maybe also as small particles

in the tip of the needle. Such effects were described in Sec­

tion D2.5, which concluded with the recommendation that

the needle be inserted before the split valve is closed.

Tests to Determine the Source of Contaminants

The following tests help discovery of the source of the

"ghost" peak material.

1 Is the "ghost" peak related to injection? Run a chro­

matogram without an injection. If the" ghost" peaks are present, they represent material from the carrier gas cylinder or the supply system. 2 If there is continuous input of contaminants from the instrument or the gas supply system, the size of the "ghost" peaks is proportional to the duration of the trapping period. Hence the sizes ofthe "ghost" peaks obtained after short and long cooling periods should be compared. The statement cannot be reversed: if there is no proportionality, the contaminants may still be from the instrument. 3 If the source of the contaminants is within the in­ strument, e.g. the gas supply line, warming enlarges the "ghost" peaks as it increases volatility. The easiest means of increasing the temperature within the GC in­

7.3. Reconcentration by Cold Trapping

4

5

6

7

8

9

349

strument is to heat the oven. In a first run, the oven is cooled immediately after elution of the "ghost" peaks. In the following run, the temperature is programmed to the limit of the column and left there for maybe another 30 min, warming the whole instrument. The material released during this heating passes through the column without forming "ghost" peaks, but the instrument will remain warm for a while during subsequent cooling, such that an increased amount of contaminants should be accumulated at the column entrance. The maximum effect is achieved by rapid cooling of the column, i.e. by opening the oven door. Components of interest inside or outside the instru­ ment are warmed, e.g. with a heat gun or a hair dryer, while the GC column is cooled in order to trap the re­ leased material. It must be kept in mind that a source of contaminants may contaminate the whole system down­ stream and that the source of "ghost" peaks thus de­ tected may be secondary, i.e. a site accumulating mate­ rial released further back. For this reason the test works best in or near the instrument. Carrier gas is supplied directly to the regulators of the instrument from an extra gas cylinder. It may even be of interest to connect the gas cylinder directly to the in­ jector, using the regulator of the cylinder. This test is indicative of sources in the supply line. If the "ghost" peaks appear only when solvent is intro­ duced, the search should be directed towards the gas supply line nearthe injector being contaminated by pre­ vious overloading of the injector. A slight reduction of the volume injected typically reduces the size of the "ghost" peaks far more than proportionally, confirming that the peaks are not from a solvent contaminant (Sec­ tion 2.6.2). If the peaks also appear when the syringe needle is in­ troduced without injecting, septum material is a likely source. The size of such"ghost" peaks tends to be poorly reproducible. Sometimes the peaks only appear when the outer needle wall was wetted with solvent or sam­ ple solution shortly before injection, perhaps dissolv­ ing septum material. The split flow rate before and after a splitless injec­ tion has no effect on the size of the" ghost" peaks if the carrier gas is saturated with the material or if the carrier gas is contaminated to begin with. If, however, the con­ taminants are released into the gas at a constant rate, increasing the gas flow rate results in dilution and, hence, reduction in transfer into the column. If the size of the"ghost" peaks is reduced by increasing the split flow rate before injection, the source is likely to be in the instrument or the gas supply. If, however, in­ creasing the split flow rate after the splitless injection reduces their size, the contaminants probably originate

350

D 7. Reconcentration of Initial Bands from the sample or the injector. The release of trapped

material or degradation products from a layer of contaminants in the injector tends to be fairly constant in terms of amount per time (provided the same sample is injected). If a higher gas flow passes through the in­ jector, the released material is more diluted and a smaller proportion of it enters the column. 10 If material from the injector is the source ofthe "ghost" peaks, increasing the injector temperature is likely to enhance the effect. The test can also be reversed: if the injector is not heated, the "ghost" peaks should disap­ pear.

Charcoal Traps

If it is suspected that the source of the contaminants is out­ side the instrument, installation of a small charcoal trap should remove the "ghost" peaks. Such a filter is useful for diagnosis as well as therapy. A short piece of 1/8 inch tubing can be filled with charcoal between plugs of glass wool. The charcoal must be firmly retained to prevent small particles from being blown into the GC system. Itcharcoal is not read­ ily available in the laboratory, it can be taken from the filters of (new) cigarettes. Charcoal has enormous capacity: 100 mg is sufficient to trap the amounts of contaminants introduced over a period of several months at least. The effect on the "ghost" peaks is not always immediate and complete, because material from the main source might have contaminated the entire gas system. For efficient clean­ ing within the gas chromatograph, the GC oven is heated as high as possible with a high gas flow rate purging the sys­ tem through the split exit, possibly using air instead of car­ rier gas.

7.3.7. Application of Cold Trapping

Cold trapping is the method of choice for the reconcentra­ tion of initial bands broadened in time if solutes are eluted above a column temperature of ca. 100 cC. The need to cool and heat the column merely for injec­ tion may be inconvenient, but if compounds are injected splitless and eluted at elevated temperatures, there is usu­ ally no realistic alternative. If some sensitivity can be sacri­ ficed, split injection at a minimum split ratio should be considered, because this enables fully isothermal runs. Cold trapping fails for solutes eluted below ca. 80 cC, because a sufficient difference between the column tempera­ tures during injection and elution can be achieved only by cooling below ambient temperature. It is not needed, how­ ever, because such splitless injection can be performed us­ ing solvent effects. Cooling to or below ambient tempera­ ture would cause solvent to condense anyway, i.e. solvent effects cannot be avoided.

7.4. Reconcentration by Solvent Effects

Bands of volatile solute broadened in time are reconcentrated by "solvent effects". We prefer the plural to the previously used "the solvent effect", because at least two mechanisms

7.4. Reconcentration by Solvent Effects

351

are involved, namely solvent trapping and phase soaking.

Detailed descriptions are given in [58-61] and summarized

in [62].

Solvent effects enable splitless injection at the column

temperature of the analysis or at the beginning of a tem­

perature program. This is of particular importance for sol­

utes eluted at low oven temperatures. Solvent effects pre­

suppose that the sample solvent and the column tempera­

ture are adjusted to each other.

The classical, simplified explanation concentrates on sol­

vent trapping. It is adequate unless distorted peaks ("par­

tially trapped" components) disturb the analysis, as is fre­

quently observed for solutes eluted before the solvent, but

rather seldom for those eluted afterwards. If solute peaks

are distorted, improvement requires a rather detailed under­

standing of both solvent effects to find more appropriate

solvents and stationary phases.

7.4. 1. Recondensation of Solvent

Reconcentration by use of solvent effects presupposes recondensation of solvent in the column inlet. The conden­ sate formed just below the nut of the column attachment (Figure D49) is driven further into the column by the carrier gas until a mechanically more or less stable layer is formed. Solvent-saturated carrier gas flows into the column, but the solutes are trapped in the solvent film. Vapor introduced Carrier from injector gas

•

Column attachment Layer of recondensed solvent and trapped solutes

_t

.

_

I

.­

I

I

Vapor and .,:", carrier ~as

flowing into

column

':."

1

2

3

~~ 4

.

,"~.i

-;'"

'.'..!..

5

Figure D49 The solvent trapping mechanism. 1 Solvent vapor from the injector recondenses in the col­ umn inlet below the attachment to the injector. The sol­ utes are trapped in the solvent layer. The carrier gas drives the liquid deeper into the column. 2 End of sample transfer. 3 The solvent evaporates from the rear of the layer. The liquid ahead traps the solutes. 4 Shortly before the end of solvent evaporation. 5 With the last solvent, the volatile solutes are released and chromatography is started.

352

0 7. Reconcentration of Initial Bands When the transfer from the injector is complete, the solvent film evaporates from the rear towards the front be­ cause the carrier gas is saturated with solvent vapor at the rear and unable to pick up more solvent further ahead. The condensed solvent downstream of the site of evaporation traps the volatile solutes and prevents them from passing into the column together with the solvent vapor.

Temporary Stationary Phase

For every microliter of recondensed solvent, the layer spreads ca. 20-30 cm into the column. It is 10-20 J.lm thick and has a correspondingly high retentive power. It acts as a tempo­ rary stationary phase and halts the migration of the solutes. At the end of solvent trapping, the solvent barrier disap­ pears by evaporation and releases the solute material. Con­ densed solvent can be present in the column inlet for be­ tween 30 s and several minutes, depending on the volatility ofthe solvent atthe column temperature. the carrier gas flow rate, and the volume of sample injected.

An Example

The essential features of solvent trapping are shown in Fig­ ure D50, obtained from a complex mixture of isomeric alkanes with n-nonane as the major peak. A Split injection under conditions ensuring sharp initial bands in the absence of a reconcentration effect. B Slow split injection, with the plunger of the syringe de­ pressed at such a rate that the sample was introduced over a period of 30 s; the initial bands were broadened accordingly.

0:1 rerenuon

lime, min

length and movemenl of

c

B

A

o

3:

3'i 20

1-235 -I

0'8"5

232 464 1-232 --I

no wet

no wet

zone

lone

mm

wet zone

Figure D50 "Nonane fraction", a mixture of isomeric alkanes dissolved in hexane; 20 m x 0.34 mm i.d. gla88 capillary column coated with Ucon LB 550 (a polyethylene-polypropylene glycol). A Normal split injection; B slow split injection; C splitless injection with a splitless period of 30 s. The distance the condensed solvent penetrated the column inlet beyond its attachment to the injector is s,hown below the chromatogram. (From ref. [17J.)

7.4. Reconcentration by Solvent Effects C

353

Splitless injection of a tenfold diluted solution with sam­ ple transfer lasting 30 s. Although this produced initial bands as broad as those in B, ten times more solvent entered the column and largely recondensed in the col­ umn inlet (23 °C), giving rise to reconcentration by sol­ vent effects.

Spreading of the Condensed Sample Solvent

Spreading of the recondensed sample (i.e. the "flooded zone") is shown below chromatogram C. From the point of recondensation just below the column attachment to the in­ jector, the liquid rapidly flooded ca. 50 cm of the column. As the supply of solvent vapor from the injector was exhausted, the layer evaporated from its rear, while liquid continued to flow further into the column. The last portion of liquid evaporated from a point ca. 70 cm downstream of the column attachment, 1.5 min after introduction of the sample. The solutes were released at that point and moment.

Retarded Chromatography

Thus the actual chromatographic process started 1.5 min after the injection upon release of the solutes by the solvent. With the combination of solvent (hexane) and stationary phase (Ucon LB 550, a polyethylene-polypropylene glycol copoly­ mer), only solvent trapping was effective and all the peaks were eluted with a delay equal to the solvent evapo­ ration time (see time scale below the chromatograms). The broad solvent peak, typical of splitless injection with solvent effects, might create the impression that many early solute peaks were obscured. Comparison of chromatograms A and C shows that this is not true: the early peaks are separated from the solvent at least as well as after split injection (the major peak is a solvent impurity and, therefore, larger in C).This is because the broad solvent peak "contains" the 1.5 min extra-retention time of all solutes, i.e. the solvent evaporation period during which no chromato­ graphyoccurred.

Incomplete Solvent Trapping

There are limits to solvent trapping. Extremely volatile com­ ponents evaporate from the solvent film at a significant rate. In particular, when the solvent poorly solvates the solutes (such as a polar solute in an apolar solvent), retention by the solvent layer may not be sufficiently strong and solute ma­ terial escapes prematurely. This produces distorted peaks (chair or stool shapes) owing to partial solvent trapping. The second solvent effect, phase soaking, which results from swelling the stationary phase film with solvent, can reconcentrate these bands.

Phase Stripping

Phase stripping by flooding liquid can be a problem if the stationary phase of a column is not fully immobilized ("bonded"), e.g. cyano silicones, or depolymerized during use, as occurs with Carbowaxes.

354

D 7. Reconcentration of Initial Bands

The recondensed solvent is driven into the column by the carrier gas and dissolves stationary phase material. The lat­ ter is carried towards the point where the last portion of the solvent evaporates, which is 10-60 cm from the attachment of the column to the injector, depending on the amount of recondensed solvent. There the dissolved mate­ rial is piled up and causes general peak broadening and dis­ tortion. Phase stripping is seldom observed with the commonly used, well immobilized stationary phases and is eliminated by us­ ing an uncoated precolumn at least as long as the flooded zone (e.g. 1 rn),

7.4.2. Requirement. for Solvent Effect.

The amount of solvent recondensed in the column inlet must be sufficient to leave some liquid up to the end of the splitless period, such that the solute material transferred last (maybe adsorbed material or material retarded other­ wise) is still trapped in condensed solvent. Towards the end of the splitless period, more dilute solvent vapor is transferred into the column,and at a given point less vapor enters than evaporates. Now a sufficient amount of liquid must be there from the first part of the transfer pe­ riod to maintain solvent trapping. This is one reason why the column temperature during the transfer must be sub­ stantially below the solvent boiling point.

Temperature below Dew Point

Recondensation of solvent vapor in the column inlet occurs if the carrier gas is oversaturated with solvent vapor at the given oven temperature, i.e. if the oven temperature is be­ low the dew point of the gas/vapor mixture. This dew point depends on the column temperature, the dilution of the solvent vapor with carrier gas, and the inlet pressure. The more dilute the vapor, the further is the dew point below the boiling point, but it increases with increasing column inlet pressure. Hence, efficient solvent trapping is obtained up to a higher column temperature when the vapor is con­ centrated and the inlet pressure is high. As a rough rule of thumb, boiling and dew points increase by ca. 20-25° when the inlet pressure is increased from 0 to 100 kPa, and by a further ca. 15° when pressure is increased from 100 to 200 kPa. On the other hand, the dew point decreases to a similar extent when the vapor concentration is halved. This description is simplified because just one of the two solvent effects, solvent trapping, is considered. Experimen­ tal experience shows that weak solvent recondensation is sufficient in one instance (when phase soaking is efficient), whereas stronger recondensation is needed under other con­ ditions.

Experimental Determination

Such considerations indicate that for successful solvent ef­ fects the column temperature must be below a certain limit and that this will be below the solvent boiling point (except

7.4. Reconcentration by Solvent Effects

355

inlet pressures are extremely high). There are no stringent rules on a minimum column temperature. Because there is no formula for calculating the upper tem­ perature limit, it must be determined experimentally for a given injector geometry, sample volume, and solvent. Start­ ing with a column temperature maybe 10° below the stand­ ard solvent boiling point, which hardly ever produces no­ ticeable solvent effects, the column temperature is re­ duced stepwise in increments of, e.g., 5° until the solute peaks are sharp and the separation of the mixture is as effi­ cient as with split injection. If solvent trapping is the only efficient solvent effect, the change from broad to sharp peaks tends to occur within a range of a few degrees. It tends to be more gradual if phase soaking interferes. In Figure D51, such an experiment was performed with a rapidly eluted, complex mixture, the "nonane fraction".

,

;

.'

2 ~l P.ENT ANE .

l

Irl

1 ~l HEXANE'

I~

2 ~J HEXANE

L

I f­ 2 ~l HEPTANE

l

Figure D51 Experimental determination of the maximum column tem­ perature providing reconcentration by solvent effects. 13 m x 0.32 mm i.d. glass capillary column coated with 0.15 Ilm SE-54, isothermally at 45°C; isomeric C9 -e 10 alkanes. No reconcentration with pentane (b.p. 36°C).

With heptane (b.p. 98°C), the solvent effects were strong

and the peaks as sharp as with split injection.

Only weak solvent effects with hexane (b.p. 68°C) when

a 1 III (needle) volume was injected.

Injecting 2 III of the hexane solution, the higher con­

centration of vapor in the carrier gas rendered recon­

centration complete.

Under critical conditions, a small change in the vapor con­ centration (or column temperature) tips the scale.

Guidelines

For an injector geometry fulfilling the requirements discussed in previous sections and an intermediate inlet pressure it is usually found that the column temperature must be at least 15 to 25 °C below the standard boiling point of the sample solvent. Not all of the commonly used solvents are suitable for split­ less injection with solvent effects. With a column tempera­

356

D 7. Reconcentration of Initial Bands ture of ca. 25°C, i.e. with the oven door open, the most vola­ tile solvents providing solvent effects are dichloromethane, acetone, diisopropyl ether, and carbon disulfide (used for liquid desorption from charcoal traps). Diethyl ether and pentane require cooling of the column below ambient tem­ perature. Hexane can be injected at temperatures up to ca. 40°C, cyclohexane and ethyl acetate up to 50°C, i.e. enable automated analysis with the oven door kept closed. The requirements of solvent effects must be consid­ ered at an early stage of method design, for instance when selecting the solvent used for extraction or sample clean-up.

7.4.3. Effects on Reten­ tion Times

As solvent effects arise from slowing or trapping of the sol­ ute material during the transfer period, they increase re­ tention times.

Effect of Solvent Trapping

Solvent trapping causes all solutes to be retained in the layer of condensed solvent for an equal amount of time, namely that of solvent evaporation. In an isothermal run, the re­ tention times of all peaks are correspondingly increased (in Figure D50, the delay was 1.5 min). There are no further ef­ fects on the chromatogram, i.e. the distances between the peaks and the peak widths are the same as for split injection. In temperature-programmed runs, the extra-retention time is reduced with increasing temperature and becomes negli­ gible ca. 60-80° above the injection temperature. Solutes eluted at higher temperatures do not migrate at the injection temperature (they are cold trapped). At column temperatures near that used for injection, repro­ ducibility of absolute retention times can be a problem. It is determined primarily by the reproducibility of solvent trapping, which in turn is sensitive to factors influencing the solvent evaporation time. If evaporation times are long, an increase in the sample volume by 0.1 III can cause 10 s more extra-retention time.

Effect of Phase Soaking

Extra-retention times are more complex if phase soaking is involved. Phase soaking results from co-chromatography of solutes with the strongly overloaded solvent peak through part ofthe separation column and slows the chromatographic migration differently for different components. Delay is strong for a component which co-migrates through a large part of the column and weaker for a later eluted substance which was separated from the solvent earlier. The resulting extra retention time may amount to more than a minute for an early eluted peak, decreases for solutes eluted later, and becomes negligible towards the end of an isothermal run. In temperature-programmed runs, effects are restricted to a temperature range comprising hardly 20°. These characteristics of phase soaking cause the early part of the chromatogram to be focused, i.e. peaks are sharper than with split injection, but also closer toqether, The sharp­

7.4. Reconcentration by Solvent Effects

357

ness of the peaks might create the impression of increased separation efficiency. Efficiency is, however, slightly lower because distances between peaks are reduced more than are the peak widths. Theoretical plate numbers must not be determined from chromatograms which are influenced by solvent effects as the absolute retention times are prolonged without the occurrence of chromatographic band broadening. For early eluted peaks, plate numbers of several millions are easily obtained, even if the column is of modest efficiency.

7.5. Band Broadening in Space

Band broadening in space in splitless injection was first de­ scribed in 1981 [631; a more detailed description was given in 1985 [641. It had not been discussed before because it sel­ dom produces spectacular effects, such as distortion or split­ ting of peak. Peak broadening is readily measurable, how­ ever; it seemed accepted that "certain" analyses involving splitless injection resulted in somewhat reduced separa­ tion efficiency. This need not be accepted.

Side Effect of Solvent Trapping

In splitless injection, band broadening in space is a conse­ quence of flow of recondensed sample in the column inlet. Often solvent recondensation is wanted to achieve reconcentration of solute bands broadened in time by sol­ vent effects. Sometimes it is an uninvited side effect when cooling for cold trapping reduces the column temperature below the dew point of the vapor/gas mixture.

Spreading in the Liquid Phase

As the layer of recondensed solvent rapidly becomes too thick to be mechanically stable, the carrier gas drives liquid further into the column. When glass capillary columns are used, the rather dramatic flooding process can be ob­ served by eye as moving rings, representing waves driven by the "wind" of the gas. The front of the liquid layer ad­ vances (although with decreasing speed) until the solvent has evaporated.

Visible at Increased Column Temperature

The evaporating solvent deposits the solutes on the capil­ lary wall. The higher-boiling solutes remain there and are, hence, spread over the whole flooded zone. The volatile com­ ponents evaporate as soon as released by the solvent enve­ lope and start chromatography from the short zone in the capillary inlet where the last portion of solvent is vaporized. Hence solvent trapping reconcentrates volatile solutes not only in time, but also in column length (space), which is why band broadening in space causes peak broadening only for components eluted some 50 DC above the column tem­ perature during injection.

7.5.1. Shape of the Initial Band

As a first approximation, all higher-boiling solutes have ini­ tial bands of the same length in terms of centimeters of col­ umn inlet: This is so indeed for on-column injection: all

358

0 7. Reconcentration of Initial Bands

the solute material is spread throughout the flooded zone. Assuming homogeneous distribution (which is not quite the case), the band has a rectangular shape: the concentration profile is vertical at the ends of the flooded zone and flat in between. Temperature Gradient at Column Attachment

In splitless injection, the situation is more complicated be­ cause some solute material is retained in the column in­

let section which is thermostatted inside the injector (Figure D52). It does not reach the flooded zone and is not spread, therefore.

~c:rrier

..:'.,•. -,::,.:. :..,::.. '.:"~':.> -.:.::",:;

::e 01 vaporizing chamber

~:'.J::':' ~'.: ';"~":J.~':: '.

':" ':;:';.~ ::~~~~" '.:'.:.:

Sample vapor entering column

~~. ~njeClor temperature '

.•.. :

:i

GC oven

Retained solute material

:.

~

Temperalure gradienl Cool column attachmentscrew Steep temperature drop

Column

Recondensed solvent spreading into column

Figure D52

The section of the column inlet located in the region of the

column attachment is at an intermediate temperature. It re­

tains some of the solute material, i.e. prevents it from reach­

ing the flooded zone. but no solvent recondenses there.

The column entrance is positioned in the truly thermostat­ ted part of the injector. The inlet then passes through a tem­ perature gradient, from the less intensely heated bottom part of the injector to the column attachment with the fitting and the screw protruding into the oven. Near the bottom end of the column attachment the temperature drops, in a steep profile, to that of the oven. Retention of Solute Material

The solute material is pre-separated in this gradient. As the injector temperature usually exceeds the maximum oven temperature during the analysis, all the solute material of interest rapidly passes through the first part of the column neck. The high-boiling material should then be retained in the somewhat cooler region 1-2 cm below the column en­ trance. That of intermediate volatility should pass slightly further to find a point near the column attachment which halts its progress. Only the solvent and the volatile solutes should advance into the oven-thermostatted column.

7.5. Band Broadening in Space Visual Observation

359

Experiments only partly confirm these expectations. In a glass capillary column, the initial band of the moderately high­ boiling perylene can be observed visually owing to its strong fluorescence. 2 III of a solution in dichloromethane were in­ jected splitless at an injector temperature of 270°C. The GC oven door was open to promote solvent recondensation, but also to mark the front end of the flooded zone. At the end of solvent evaporation (about 2 min after injection), the column was dismantled and the distribution ofthe perylene observed in a dark room under UV light (366 nm). Most of the fluorescence was observed ca. 1.5 cm below the column entrance (Figure D5~, the column inlet in the injector being some 45 mm long). Its intensity decreased to­ wards the point where the column left the attachment, but ca. 10 % of the injected perylene had entered the oven-ther­ mostatted column. There, it was redissolved by the recon­ densed solvent and spread throughout a flooded zone ca. 40 cm in length. From the low volatility of perylene it should have been expected that all of this material would remain in the column neck. :! :

splitless injection syringe

hot injector

warm

lone cold

oven

Figure D53 Visually determined distribution of perylene in the column inlet after splitless injection with an open oven door. (From [64].)

Incomplete Retention in the Column Neck

Chromatographic experiments confirmed that part of the solute material always enters the flooded zone, although the proportion decreases as the volatility of the solute decreases; it corresponded to ca. 50 % for n-decane and decreased to 15 % for n-e30 (Figure D54). Obviously the column inlet inside the injector was too short for complete partitioning. 1.5 III of an acetone solution of C'O-C36 n-alkanes was in­ jected at ambient column temperature and eluted from a 5 m x 0.30 rnm i.d. glass capillary column coated with OV-1

360

D 7. Reconcentration of Initial Bands

10

cold on-column

12

18 24

22

16

\4

20

splitless

I 3OO'C

Figure D54 In on-column injection, peak deformation due to band broadening in space is determined by the distribution of the sample material in the flooded zone. In splitless injection under condi­ tions causing solvent recondensation it is further complicated, because during the flooding process part of the solute material remains in the column neck inside the heated injector. (From ref. [63].) (100 % methylpolysiloxane). The shortness of the column and the restricted wettability of QV-1 by acetone accentu­ ated peak distortion.

Shape of Initial Band

In Figure D55, the shape of the initial band is shown in the form of a chromatogram as it would be recorded by a detec­ tor positioned right after the flooded zone. The material brought to the front of the flooded zone is eluted first. As the flooded zone is maybe 20-35 cm long and the band retained at the bottom of the injector only 5-10 rnrn, the peak eluted last is extremely sharp compared with the rest. In reality, the initial band shape loses much of its clarity during the chro­ matographic process. It is clearly visible only if the col­ umn is short or the sample solvent does not wet the sta­ tionary phase surface and the flooded zone is correspond­ ingly long.

Defocusing Effect of the Cool Injector Base

It should be added here that similar peak distortion can re­ sult from a cool column neck inside the injector. The tem­ perature of the base of the injector tends to go up and down with that of the oven, usually clearly exceeding the latter. It can, however, be a cold spot if a rapid temperature pro­ gram is applied. Initial bands of higher-boiling solutes are enlarged owing to slow release from thlssite of high reten­ tive power. The effect is accentuated if the corresponding

7.5. Band Broadening in Space

Chart

361

• Retained in bottom part of injector

Spread in flooded inlet inside the oven

Figure D55 Peak (initial band) after splitless injection with solvent recondensation before chromatographic diffusion processes "washed away" its initial clarity; solute eluted at least 50° above the column temperature during injection (no recon­ centration by solvent trapping). The distribution of solute material within the flooded zone, i.e. the shape of the broad part of the peak. is poorly reproducible.

capillary section is contaminated and thus has higher gas

chromatographic retentive power. Many types of sample by­

product tend, in fact, to accumulate there.

The resulting effects vary between tailing peaks, distorted

peaks, and peaks of insufficient area, the latter being the

most insidious effect as quantitative results can turn out in­

correct without any visible warning from the chromatogram.

7.5.2. Extent of Peak Distortion

Lengths of flooded zones per volume of liquid were deter­

mined by visual experiments, primarily by use of on-column

injection [65,661. The results can be summarized by the fol­

lowing rules.

Length of Flooded Zone

In a 0.32 mm i.d, coated column inlet, the sample liquid

flows 15-25 cm per microliter. If only part of the sol­

vent is recondensed, the flooded zone is correspond­

ingly shorter. Hence, 2 ~L of sample (including the nee­

dle volume) creates a flooded zone up to some 40 cm

long.

The length of the flooded zone is inversely propor­

tional to the column diameter, because a wider bore

capillary has a larger surface to form a sample film. In

0.53 mm l.d. inlets, the flooded zones are typically 10-15 crn/ul, of recondensed solvent, whereas they reach 50­ 80 crn/ul, in 0.1 mm i.d, columns. If the surface of the column inlet (the stationary phase or deactivated uncoated precolumn) is not wetted, the sample is unable to form a film. It leaves some droplets here and there and moves up to several meters per microliter. In extreme situations, some sample liquid can flow directly into the detector.

362

D 7. Reconcentration of Initial Bands

The maximum tolerable length of the flooded zone is deter­ mined by the length of the solute bands resulting from the chromatographic process. It should not contribute noticeably to the terminal band length, the length of the band when leaving the column. Terminal Band Length

As deduced by Saxton [67]. the terminal band length is iden­ tical for all components running through a column under the same conditions; peak widths do not, in fact, vary because terminal band lengths vary, but because these bands leave the column at different speeds. If chromatography is assumed to provide an efficiency cor­ responding to a height equivalent to a theoretical plate (HETP) equal to 1.5 times the column diameter (a column used not far from its optimum), the terminal band lengths are as given in Table DB. Table D8 Terminal band lengths [cm] calculated for different column lengths and internal diameters under the assumption that the efficiency in terms of HETP corresponds to 1.5 times the column diameter.

Column Terminal band legth [cm] Length 0.10 mm 0.25 mm 0.32 mm 0.53 mm 7m

15 m 25 m 50 m Broadening Effect

13 19 24 35

20 30 39 55

23 34 44 62

30 44

56

80

The calculation of peak broadening from the size of the ini­ tial band is complicated by the non-Gaussian shape of the band (see below), but a rough comparison of the initial and the terminal band lengths is sufficient to reveal that band broadening in space in splitless injection with solvent recondensation is by no means negligible. For practical work, initial band lengths (flooded zones) cor­ responding to a quarter of the terminal band lengths produce negligible peak broadening, whereas those cor­ responding to half of it cause measurable, but not immedi­ ately obvious broadening. This means that with a 15 m x 0.32 mm i.d. column, producing terminal bands ca. 35 cm long, initial bands ca. 15 cm long can be tolerated without substantial loss of column performance. These 15 cm corre­ spond to the flooded zone generated by ca. 1 ilL of recon­ densed liquid.

Sample Wetting the Station­ Figure D56 shows chromatograms obtained from fatty acid ary Phase Surface methyl esters in hexane, analyzed on a 12 m x 0.31 mm i.d. glass capillary column coated with a polydimethylsiloxane. Chromatogram C was obtained by split injection and serves

7.5. Band Broadening in Space

A)

SPLITLESS WITHOUT SOLVENT

RECONPENSATION

363

n I' ,

!lO~~=====~~~~=========:;:.~~:o.J ~

210°C E3)

C)

7°/min

SPLITLESS WITH SOLVENT RECONDENSATION

SPLIT

16 15

14

12

10

18

8

6

----''---~~~--L--__A___--------~

~ 1-------­ ---------1 7°/min

210·C

25°C 40°C

Figure D56 Band broadening in space in splitle.. injection with a sample solvent wetting the stationary phase surface: Ce-C,s fatty acid methyl esters in hexane. C, 1.5 III split injection (30:1); no broadening by initial bands. B, 2 III splitless injection; column temperature during sample transfer, 25°C, t.e. with strong solvent recondensation. Peaks are broadened by 25 to 35 %. A, as B, but sample transfer at 60°C, Le, without recondensation. Attenuation is twice that used for B. (From ref. [64].)

for comparison. Chromatograms B and A were obtained from splitless injections (2IlLI under conditions either resulting in solvent recondensation (column temperature during injec­ tion, 25 °CI or excluding the latter (column temperature, 60°CI.

Without solvent recondensation (chromatogram AI, the first peaks were broadened in time. There was no recon­ centration by solvent effects and cold trapping was still weak. The broadening effect was modest, because of the relatively high carrier gas flow rate (5 rnt/rnln, hvdroqenl. Peaks from ester C10onwards, reconcentrated by cold trapping, had the

364

D 7. Reconcentration of Initial Bands same widths at half height as those in chromatogram C. In fact, there was no flow of recondensed sample and, thus, no band broadening in space. Peak broadening in chromatogram B is not obvious, but compared to chromatograms A or C it amounts to 25-35 %. It is more apparent from the peak heights: the same amounts were injected as in A, but attenuation was halved. Broaden­ ing reduced the separation efficiency, in terms of separation number tTZ), by 25-35 % and in numbers of theoretical plates by a factor of nearly two, i.e. to an efficiency expected from a column half as long. . Effects of band broadening in space are weaker when longer separation columns are used, because the initial band con­ tributes less to a longer terminal band. If a 25 m capillary column had been used the peak broadening would have been hardly significant (although still measurable).

Non-Wetting Samples

The chromatograms in Figure 057 were obtained with the above column and the conditions of chromatogram B, but using acetone and methanol instead of hexane. A)

ACETONE

6 IB

16

15

14

12 10

7'/min

B

--------1

25'C

40'C

Figure D57

Peak distortion arising from band broadening in space. Ac­

etone wets polydimethylsiloxane stationary phases poorly.

methanol not at all. Sample. column, and conditions as in

Figure D56B. (From ref. [64].)

Polydimethylsiloxane stationary phases are only partially wetted by acetone, which results in a flooded zone about twice as long and correspondingly more severe peak distor­ tion. The distortion pattern shown in Figure 048 is now ap­ parent as pre-peaks or shoulders in the up-slope (except for the C6 ester). The upper halves of the peaks are as sharp as

7.6. Uncoated Precolumns - Retention Gap Techniques

365

those obtained by split injection since they represent the narrow part of the initial band. Polysiloxanes containing at least 5 % phenyl are wetted by acetone. Methanol does not wet apolar stationary phases; 1 ~L flows several meters into the column and peak distortion is correspondingly severe. The height of the bands eluted be­ fore the sharp signals indicate that the amount of solute material deposited in the flooded zone decreased towards the front of the flooded zone. This is probably because of extraction of the esters from the flooding methanol into the stationary phase; the extraction seems to be more rapid for the higher molecular weight esters. Methanol wets polysi­ loxanes with at least 50 % phenyl substitution and Carbowax­ type phases.

7.5.3. Avoidance of Peak Distortion

When a wetting solvent recondenses in the column inlet (sol­ vent effects or side effect of cooling during cold trapping), peak broadening tends to be considerable if the column is short and of narrow bore, but can be neglected if the column is at least some 30 m long. Measures must always be taken when the sample does not wet the column inlet.

Cold Trapping without Solvent Recondensation

Band broadening in space is avoided when solute reconcen­ tration can be achieved by cold trapping with a column tem­ perature sufficiently high to prevent solvent recondensation. This precludes analysis at low temperatures. If, for instance, a sample is dissolved in hexane, injection must be performed at a column temperature of at least 50°C. Since satisfactory reconcentration by cold trapping requires a temperature in­ crease of 60-80°, the first perfectly shaped peaks are obtained at ca. 120°C.

Non-Wetting Solutions

Injection of aqueous samples creates several problems in capillary GC. Those relevant to initial bands are: no stationary phase is wetted by recondensed wa­ ter; condensed water attacks siloxanes, rendering the col­ umn adsorptive and chemically active; no precolumns wetted by water [68,691 and resisting chemical attack by water are available [70,711. Hence solvent effects are not applicable. Samples in non­ wetting solvents, such as water, can be analyzed by splitless injection only when cold trapping without solvent recon­ densation is applied. Water vapor does not attack column surfaces. As the column temperature during injection must be approx. 90°C at least, the minimum elution tempera­ ture generating sharp peaks is ca. 160 °e. Thick film columns or columns coated with polar stationary phases can be used to increase elution temperatures.

7.6. Uncoated Precol­ umns - Retention Gap Techniques

Instead of avoiding the flooding process, bands broadened in spaeecan be reconcentrated by the retention gap tech­ nique. In on-column injection, this enables the introduction

366

0 7. Reconcentration of Initial Bands of up to many hundreds of microliters. It presupposes un­

coated precolumns which, in these extreme cases, are usu­

ally 10 m long.

Uncoated precolumns are also used for a completely differ­

ent reason, namely the reduction of the effects of non-evapo­

rating sample by-products in the column inlet.

7.6.1. Reconcentration of Bands Broadened in Space

If the flowing sample liquid spreads the solute material in an

uncoated precolumn, the solute bands are reconcentrated

at the beginning of the coated column. Migration of

solute material through an inlet of low retentive power (re­

tention gap) occurs more rapidly and at a temperature well

below that of elution (see Figure D58). Initial bands are com­

pressed on encountering the region of high retentive power

in the coated column. The mechanism of reconcentration are

discussed in depth under "Retention Gap Techniques" in the

context of on-column injection in ref. [731.

'c:"

iii

Fast migration

al "0 o o

u::

A

B

... Slow migration

I I

c

o

Sharp solute band

Figure D58 Flooding sample liquid and reconcentration of the solute bands by the retention gap technique in a temperature-pro­ grammed run. A The sample liquid flows into the column. a The solvent evaporates. At this temperature, higher­ boiling components remain spread throughout the flooded zone. C The temperature has been increased. The low retention power enables the solute material to migrate through the uncoated inlet. It is stopped at the entrance of the separation column. D Solutes wait there until the temperature is further in­ creased to enable chromatography through the separa­ tion column. (Adapted from ref. [74].)

7.6. Uncoated Precolumns - Retention Gap Techniques

367

High Efficiency

Reconcentration of solute bands by the retention gap tech­ nique easily exceeds e fector of 100 [751; hence an initial band of, e.g., 1 m (rather long for splitless injection) is short­ ened to a residual initial band in the separation column of negligible 1 cm. The technique is, therefore, far more effec­ tive than is needed for splitless injection.

Required Length of Precol­ umn

The uncoated precolumn must be et leest es long es the flooded zone, but may be longer without disturbing chro­ matography (except adsorptivity is a problem). 1 ~L of wetting liquid floods 20-30 cm of an 0.32 mm i.d. uncoated precolumn or 25-40 cm of a 0.25 mm i.d, precol­ umn (65). which is slightly more than for a coated column. Considering that not all of the injected solvent recondenses, uncoated precolumns of 50 cm x 0.32 mm i.d. or 60 cm x 0.25 mm i.d. should serve the purpose (injection volume, 2-2.5 ~L).

Deactivation of Uncoated Precolumns

For compounds of low adsorptivity, rew fused silice is suit­ able, but mostly deactivated precolumns are preferred. De­ activation is usually achieved by silylation. For reasons of wettebility, the reagent must introduce phenyldimethylsilyl groups ("phesil" surface) [651. Trimethylsilylated precolumns are not wetted by solvents of high surface tension, such as benzene, toluene, dichloromethane, ethyl acetate, or acetone. If the precolumn is not wetted, the recondensing sample liq­ uid is not retained and floods into the separation column almost as if there were no uncoated precolumn. "Phesil" precolumns are reasonably well wetted even by methanol.

Chemical Stability of the Deactivation

A problem of silylated precolumns is their instebility to­ werds eggressive semple meteriel, such as condensed water (humid extracts, e.g. in ethyl acetate). The deactiva­ tion is hydrolyzed and the silyl group possibly lost with the attacking molecule (water vapor seems to be harmless). Sta­ bility can be improved by use of polymeric silylation rea­ gents, bonding the reagent several times and, hence, pre­ venting its loss upon breakage of a bond (which might be re­ formed at higher temperatures). Precolumns deactivated with a very thin layer (0.5-1 nm) of OV-1701 were a step in this direction [76,771. Deactivation procedures with hydrosil­ oxanes seem to be of similar efficiency [78,791. Numerous kinds of deactivated precolumn are available com­ mercially, usually with little information on the type of pro­ cedure applied. There are also "polar" and "apolar" precol­ umns. Their distinction is important when long uncoated precolumns are used, but not for those of up to 2 m com­ monly used in splitless injection. Because the characteristics of these precolumns cannot be reduced to a scale between "good" and "poor", it might be useful to compare various products for a specific problem.

368

0 7. Reconcentration of Initial Bands

7.6.2. Uncoated Precol­ umn as Waste Bin

Uncoated precolumns are also used as disposable inlets re­ taining non-volatile sample by-products. When "dirty" sam­ ples are analyzed, the column inlet often becomes contami­ nated. Most of the involatile material reaches the column as an aerosol. i.e. as small particles suspended in the gas phase (like smoke). Other, scarcely volatile compounds, e.g. fat or waxes, enter the column as vapor, but do not get beyond the inlet owing to insufficient volatility.

Effects of Contaminants

The effects of involatile matrix materials contaminating the column inlet were discussed in detail in ref. [801. 1 The most frequently observed effect is peak broaden­ ing and tailing because the contaminants exert reten­ tive power, releasing the solutes slowly and after a de­ lay. A 20 cm column inlet contains roughly 10 ~g sta­ tionary phase. When non-evaporating material corre­ sponding to a fraction of this amount is deposited there it modifies the chromatographic properties. It usually accumulates to a kind of droplet and locally builds up high retentive power. 2 Other material primarily increases adsorptivity. Hy­ droxides, as a drastic example, perhaps co-extracted from basic solutions, obviously adsorb acidic sample components. Polar material tends to have adsorptive properties. 3 Some contaminants support the degradation of sam­ ple components. For instance, basic material saponi­ fies esters. Degradation is particularly severe for com­ ponents chromatographed at high temperatures. 4 Contaminants can degrade the stationary phase in the column inlet. As degradation and elution of the poly­ mer fragments is accelerated by increasing the column temperature, this causes a drifting baseline which is easily misinterpreted as general column bleed.

Example

Figure D59 shows test chromatograms from a column de­ graded by ten splitless injections of a crude hexane extract of lemon peel. Chromatogram B shows general peak broad­ ening (reduction of separation efficiency from TZ 32 to 18). Many analysts are used to curing deteriorated columns by "conditioning", which simply means strong (brutal) heat­ ing. For capillary columns this often has a clearly negative effect - in our case it seems to have killed the patient (C). Removal of the contaminated inlet completely restored the quality (D). Peak broadening in chromatogram B resulted from accumulated contaminants. "Conditioning" probably ruptured the stationary phase to form droplets. In glass cap­ illary columns, such droplets were usually visible by eye.

Reduced Effect of Contami­ nants

Uncoated precolumns reduce several of the problems aris­ ing from contaminated column inlets [821.

7.6. Uncoated Precolumns - Retention Gap Techniques

369

A larger amount of retaining sample material can

2

be deposited in the column inlet because the "retention gap" can be filled before the retentive power reaches that of the coated column. Degradation of labile solutes is reduced because solutes leave the contaminated column inlet at a far lower temperature.

A 121=29.8 E12

Ill' aI

•

10

122=33.4 Ell

D

EIO A

P

j

j

...

Sr

.1'1

.w

'--J

I

120°

40°

122= 19.4

TZ 1=17.5

po'

B aI

A I

~

~l..JV

J,./~ ~

~

c

121° aI

D 122=29.4

I I

120.5°

I

12,=32.1

t.-.

t\

~

l

1

Figure D59

Standardized column test (temperature program 40 to 120 OCt from a 15 m x 0.30 mm i.d.

glass capillary column coated with non-immobilized SE-52.

A, Almost new column. B, After 10 splitless injections of a crude extract of lemon. C, After

"conditioning" 15 h at 250 °c. D, After removal of 1.2 m column inlet.

Components: D, 2,3-butanediol; 01, 1-octanol; ai, n-nonanal; P, 2,8-dimethylphenol; A, 2,6­

dimethylaniline; S, 2-ethylhexanoic acid; E,•.".", methyl esters; am, dicyclohexylamine; 10,

11, n-alkanes. (From ref. [81].)

370

D 7. Reconcentration of Initial Bands Decomposition of the stationary phase by aggres­ sive material is ruled out because there is none. Hence drifting baselines can no longer be caused by degraded stationary phase from the inlet (but may still result from degraded sample by-products). On the other hand. adsorptivity caused by matrix material is not reduced by an uncoated precolumn. 3

Disposable Column Inlet

The use of uncoated precolumns is convenient when a con­ taminated inlet must be removed frequently. Because of lac­ quering. contaminants can seldom be rinsed out with solvent, compelling the removal of the contaminated piece. Precolumns can be replaced. leaving the column untouched. If a chromatogram shows signs of inlet contamination, re­ moval of a 5 cm section positioned inside the injector is often sufficient. If this does not help, it must be assumed that the contaminants have penetrated the oven-thermostat­ ted column. where their fate depends on the conditions ap­ plied.

Length of the Contaminated Zone

If no solvent recondensed. the involatile material remained at the spot where the aerosol droplets were deposited. i.e. within a section less than 10 cm long. If solvent recondensed. contaminants were transported further into the column (Figure 060). Recondensed solvent dissolves the material from the column wall and carries it deeper into the column. spreading it throughout the flooded zone. After the first injection. the material is likely to be more or less evenly distributed. During the subsequent injection it is redissolved and carried forward. After many injections the contaminants end up being accumulated at the front end of the flooded zone. possibly within a section hardly 1 cm long.

Contaminants in the column neck

Flooded zone coated with contaminants

Material flushed forwards with every flooding process

Figure D60

Deposition of conteminants in the column inlet if the sol­

vent recondenses.

7.6. Uncoated Precolumns - Retention Gap Techniques

371

Removal of Front of Flooded Zone

If the solvent recondensed during at least some of the injec­

tions, the removed inlet section must long enough to include

the main deposit at the front end of the flooded lone. As up

to, e.g., 35 cm of an uncoated 0.32 mm i.d. capillary are

flooded per microliter of liquid, 30-60 cm ofthe precolumn

must be cut away to restore performance safely.

Longer Precolumns

In extreme circumstances, a contaminated column inlet must

be replaced daily (which still takes far less effort than further

clean-up of all the samples). Use of precolumns ca. 1.5-3

m long is recommended, because this enables removal of

several pieces until a new connection to the separation col­

umn must be installed. Since the precolumn contributes neg­

ligibly to the retention power of the system, its length has

little influence on retention times and shortening has virtu­

ally no effect.

Particles Driven Far into the Column

Sometimes particles are driven far beyond the flooded lone.

Occasionally they probably pass through the entire column.

They appear to move like dust. Small sodium sulfate crys­

tals, suspended after drying of a sample solution, have been

observed up to 15 m from the entrance of a transparent glass

capillary. At several sites they formed whitish deposits with

a sharp cut-off at the front, as if they were piled up against

an obstruction. The stationary phase is, apparently, not al­

ways sufficiently sticky to retain particles. After the first are

halted, however, others remain attached to them. In an ex­

treme case (50 ul, on-column injections), a column was al­

most blocked after some 100 injections. Sodium sulfate could

be rinsed from the column with water.

The use of an uncoated precolumn does not help this

problem, because the wall is not sticky.

Sodium sulfate particles have sorprisingly little effect on col­

umn performance; silica gel tends to be more adsorptive.

Presumably the surfaces of the particles are readily deacti­

vated, e.g. by fragments of stationary phase (column bleed).

Triglycerides Passing through the Precolumn

Another unresolved problem concerns materials with criti­

cal intermediate volatility, such as fat or wax esters (from

plant extracts): they can pass through the uncoated pre­

column, but not through the separation column.

As a rule of thumb, components pass through an uncoated

precolumn at a temperature 100-140° lower than through a

standard separation column. This means that, e.g., triglycer­

ides and waxes move slowly through the uncoated pre­

column at temperatures of 200-220 °e. They are stopped

at the beginning of the coated column and accumulate there

unless baked out at high temperature. Solute peaks will start

to tail and become broad. The analyst removes pieces ofthe

uncoated precolumn, but is disappointed, because this does

not eliminate the disturbing material: the first 20-40 cm of

the separation column must be cut off.

372

D 7. Reconcentration of Initial Bands To solve this problem, either oven temperatures are kept below ca. 180°C to keep the triglycerides inside the pre­ column, or increased to ca. 350°C to discharge the trig­ Iycerides through the separation column. Since triglycerides polymerize rather rapidly, they must be removed after every few runs. This presupposes, of course, a column of low re­ tention power and sufficient thermostability.

Precolumn with Thin Film of Stationary Phase

For the analysis of pesticides in edible oil by injector-in­ ternal headspace analysis (Section D8.3), precolumns coated with a thin film of stationary phase were used to retain the triglycerides up to the upper end ofthe temperature program (250 DC), but rapidly release them at ca. 320 DC through a purge exit positioned between the precolumn and the sepa­ ration column.

An Example

An application involving such problems, the analysis of "GX­ 071" in animal feed, was described by Arrendale et al. [831. Quantitation based on an internal standard resulted in a rela­ tive standard deviation of 0.8 % when using on-column in­ jection and of 2.8 % with splitless injection. A large amount of later eluted by-products (presumably fat or wax esters) created problems unless the column was heated to 320°C for at least 30 min after each run. Peaks became broadened and were split, without significant difference whether injection occurred splitless or on-column. Obviously even in splitless injection nearly all ofthe high-boiling mate­ rial was transferred into the column.

7.6.3. Press-Fit Connec­ tions

Various methods of coupling the precolumn to the separa­ tion column have been discussed [841. The press-fit connec­ tions introduced in 1986 by Rohwer et al. [851 have almost completely replaced alternatives such as butt connectors. They are practically free from dead volume and highly in­ ert (at least after brief conditioning).

Polyimide Seal

Press-fit connections are sealed by contact with the narrow ring of the polyimide coating on the tip of the capillaries to be joined (Figure 061). In fact, the coating consists of a material similar to Vespel, which is widely used for manu­ facturing thermostable ferrules. The fused silica must be bro­ ken squarely by means of a single, clean score (using, e.g., a Seal between the poly imide coating and the connector Fused silica \

Polyimide

/ ~----+------J

Tip of fused silica capillary

Press-fit connector

Figure D61 Fused silica capillaries are tightened against the press-fit connector by their own polyimide coating.

-7.6. Uncoated Precolumns - Retention Gap Techniques

373

piece of a silicon wafer or broken quartz), avoiding dam­ age of the polyimide. Breakage of Fused Silica

Cutting/breaking of fused silica (and glass) capillaries was studied by Roeraade [86). He found it difficult to obtain con­ sistent results. Fast cooling of the fused silica during manu­ facture results in high compressive stress, which is released upon breakage of the tubing and often results in cracks and jagged surfaces. Cracks can tear apart the polyimide and thus affect the seal. The best results were obtained by scoring the tube with a silicon wafer fragment and subsequent pulling with slight bending.

Forming the Seal

The fused silica butt should be wetted with a small amount of liquid (saliva being among the most successful) and gen­ tly pushed into the conical seat of the connector. High pres­ sure can result in excessive lateral force and cracks in the fused silica, resulting in small leaks. Much depends on the polyimide coating on the fused silica. Some high temperature fused silica tubing did not sit firmly in the connector, nor tighten or adhere to the connec­ tor after heating. This polyimide was, apparently, excessively hard and had no residual chemical reactivity to enable bond­ ing to the connector [87). It seems that the polyimide has since been improved.

Test for Tightness

Press-fit connections must be tested for tightness. After installation, the carrier gas inlet pressure is increased to a high level and the connection tested for leakage by means of a droplet of soap solution or a leak detector. In the event of a leak, stronger pressing rarely helps. The connection must be dismantled and the butt newly cut. Occasionally the conical seat in the connector is not concentric, necessitating replace­ ment of the connector. It is advisable to check tightness again after several heating cycles. If connections then prove to be still tight, they usually stay so for a long time.

Re-Using Connectors

Once press-fit connections have been heated above ca. 200°C, the polyimide sticks firmly to the connector. When an attempt is made to dismantle the connection, the capil­ lary usually breaks, rather than becoming detached. Connec­ tors with a piece of fused silica in the conical seat can be re­ used after heating to 500-550 °c for a few hours. This degrades the polyimide and the pieces of fused silica be­ come loose (88). The same can be achieved in a yellow flame, but the connector is easily overheated (deformed).

Thermostebilitv

Press-fit connections are thermostable up to ca. 350 °C; at 360°C, occasional leakage has been observed, but mostly disappeared again by itself. A plausible assumption is that leakage occurs as a result of degradation of the polyimide, but press-fit connections with aluminum-coated fused silica were not significantly more thermostable.

374

D 7. Reconcentration of Initial Bands

Additional Tightening

Press-fit connections are sufficiently tight for most applica­ tions, but when used with a mass spectrometer, additional air is often observed. Such leaks are too small to be discov­ ered by use of leak detectors or soap solutions and it is an open question whether they are relevant. Vecchi and Walther [89] proposed improving the tightness of press-fit connections with glue. The idea is convincing, but our experiments were rather disappointing. lightening of assembled connections by addition of a drop of polyimide resin (Figure D62) was temporarily successful, but after some heating, the connections usually started leaking [87]. Different thermal expansion of fused silica and glass might have pulled the capillary out of the press-fit seat. Experiments in which the resin (1:20 dilution in pyridine or methylpyrro­ Iidone) was applied to the region of the press-fit seat also failed. Drop of glue applied onto assembled connection

Glue applied before breaking capillary

Connector Press-fit seat

Figure D62 Improving the tightness of press-fit connections by use of glue? Drops of glue at the entrance of the connector seldom resulted in lasting tightness. Application of epoxy glue at the seat of the press-fit seal providad hardly better rasults.

The best results were obtained by using the two-component epoxy glue Epo-Tek H77-S, as proposed by Bemgard and Ostman [90]; a similar glue (Epo-Tek 353ND) was used by Clark and Jones [91]. It was applied as a thin layer after the fused silica had been scored, but before it was broken. Ex­ cess glue was removed with tissue. After installation, the columns were heated to 150°C for 15 min (with flow of car­ rier gas) to harden the glue. These seals were tight for an extended period of time, but the upper temperature limit was below 300°C. Several suppliers have introduced press-fit connectors with ferrules for additional tightening (e.g. VU unions from Restek). They are hybrids between butt connectors and press­ fits and, compared with butt connectors, have the advan­ tage of avoiding dead volume and contact with polymers in the region of the capillary tips.

7.7. Examples of the Use of Reconcentration Effects

Some practical aspects of the techniques used for reconcen­ trating initial bands are more easily discussed by use of real­ istic examples. It is unlikely that the reader will actually have to analyze the samples discussed, but the principles should be transferable.

7.7. Examples of the Use of Reconcentration Effects

375

7.7.1. Dioctyl Phthalate

Dioctyl phthalate is one of the most important plasticizers in polymers. It is eluted from standard apolar columns at ca. 220°C and is, therefore, a typical candidate for the use of cold trapping: cooling by at least 80° during splitless trans­ fer is no problem.

Manual Injection

If the sample is injected manually, the column is cooled in seconds by opening the oven door. The door is left open during the splitless transfer period provided band broad­ ening in space does not cause relevant peak broadening (the liquid wets the column inlet and the column is rather long), if the sample solvent does not recondense (volatile solvent), or if an uncoated inlet is used. Otherwise the oven should be closed before injection and thermostatted at, e.g., ca. 140°C (less than 20° below the solvent boiling point). After transfer is complete, the oven can be heated ballistically to the elution temperature.

Autosampler

If injection is performed by means of an autosampler or re­ tention times must be highly reproducible, the instrument is therrnostatted at 140°C and the temperature increased by means of a rapid program.

7.7.2. Traces of Tetra­

Tetrachloroethylene, primarily used for cleaning, is analyzed in waste water. The detection limit required is 100 ppb, which is easily achieved, even with FID, by reconcentrating extrac­ tion and splitless injection.

chloroethylene

Volatility of Extraction Solvent

The extraction solvent must be selected not only on the ba­ sis of extraction efficiency, but also the requirements of split­ less injection. With an apolar column of standard film thick­ ness, the elution temperature is ca. 35-40 °C, which tells the analyst that application of cold trapping is impossible and solvent effects are needed. Pentane (b.p, 36°C) does not recondense, neither at this temperature nor after cooling the column to ambient temperature. Hexane (b.p, 68 DC) is pref­ erable, as it boils about 30° above the analysis temperature and solvent effects provide reconcentration even in fully iso­ thermal analysis. Band broadening in space is no problem because hexane wets all stationary phases and peak broad­ ening would occur only at temperatures more than ca. 50° above that used for injection. If the column is longer or has a somewhat thicker film, the analysis temperature might be ca. 60°C. If the chromatogra­ pher still uses hexane as extracting solvent, he must cool the oven to a temperature below 45°C for the sample trans­ fer period to achieve solvent recondensation. He probably prefers a higher-boiling solvent to enable injection at the elu­ tion temperature. Cyclohexane (b.p. 80°C) is critical, but isooctane or n-heptane (b.p. 100°C) are well suited.

7.7.3. Extraction of Water with Pentane

"Micro-pentane extraction" [921. using 0.6 to 0.8 mL pen­ tane to extract 1 I water, is rapid and sensitive for the analy­

376

D 7. Reconcentration of Initial Bands

sis of a broad spectrum of organic compounds in surface or drinking waters. Pentane is chosen as solvent because it is readily available in high purity and avoids a broad solvent peak obscuring important gasoline components. It does not, however, recondense at ambient temperature. To achieve sol­ vent effects, the column temperature must be reduced below ca. 15 °e. The column can be cooled by means of a subambient ther­ mocontrol system. For manual injection it is quicker and cheaper to immerse the column into a bath of cool water which is stored in a refrigerator between the analyses. On-Column Injection

Whenever solvents are highly volatile, on-column injection is more convenient, because the maximum column tempera­ ture corresponds to the boiling point of the solvent at the inlet pressure (as a rule of thumb, standard boiling point + 1°/10 kPa inlet pressure) and solvent trapping is always in­ volved. This was actually the solution: on-column injection enabled injection at a column temperature of 35-40 °C. Fur­ thermore volumes up to 250 III are introduced, which greatly increases sensitivity and simplifies the analysis.

7.7.4. Semivolafiles in Cigarette Smoke

The semivolatile components of cigarette smoke (part of the tar) can be recovered from the glass fiber filters of smoking machines by extraction with diethyl ether. Ether (b.p. 36°C) does not, however, recondense in a column inlet kept at ambient temperature (open door). From this point of view diisopropyl ether (b,p, 56°C) or methyl tert-butyl ether (MTBE, b.p. 46°C) are preferable.

On-column injection is not advisable because the high load

of non-evaporating by-products would rapidly contaminate

the column inlet.

Co-Solvent

If the choice remains with diethyl ether, a higher-boiling co­ solvent, which recondenses at ambient temperature, can be added. Hexane seems ideal as far as its volatility and early elution from columns of intermediate to high polarity are concerned, but it solvates poorly and only weakly retains the more polar of the volatile constituents to be analyzed. Such solutes are partially solvent trapped and eluted as distorted peaks ("chair" or "stool" shapes). Dichloromethane, added 1:1 to the sample in ether, solved this problem (primarily by exploitation of phase soaking). It was either added to the vial or drawn into the syringe needle before the sample (per­ formed like a solvent flush injection). A higher-boiling co­ solvent is needed when autosampler injection requires a higher oven temperature (closed door).

Methanol as Extraction Solvent

Instead of diethyl ether, methanol is often used to extract the smoke components. Boiling at 65°C, methanol easily re­ condenses at column temperatures up to about 45 °C. There are, however, two problems to be considered. If the column is coated with a stationary phase of low polarity (less than

7.7. Examples of the Use of Reconcentration Effects

377

50 % phenyl in the siloxanel, recondensed methanol floods far into the column owing to poor wetting. This causes se­ vere band broadening in space, hence distortion of peaks eluted considerably above the temperature of injection. The problem can be solved either by using a wettable uncoated precolumn (e.g. "phesil" surface) or a column coated with a more polar stationary phase. The other problem concerns partial solvent trapping, i.e. insufficient retentive power of the polar methanol for vola­ tile solutes of low polarity. With polar columns of the Car­ bowax type, peak distortion is reduced owing to partial recon­ centration by phase soaking.

7.7.5. Solvent Residues in Pharmaceutical Prepa­ rations

Pharmaceutical tablets are analyzed for traces of solvents. It is most convenient to dissolve them in a late eluted solvent which does not interfere with the solvents to be analyzed (e.g. dimethyl sulfoxide, dimethylformamide, or ethylene glycol monoethyl ether). In capillary GC, solutes eluted before the solvent tend to be incompletely solvent trapped. They are too volatile to be retained by the solvent layer and tend to be eluted as se­ verely deformed peaks. There is no generally applicable means of avoiding this problem, although long columns with very thick (e.g. 5 urn) films of stationary phase reduce the peak distortion [93,94). Split injection is usually used to re­ duce the amount of solvent entering and recondensing in the column; this is, of course, paid for by reduced sensitivity.

7.7.6. Headspace Analy­ sis

Headspace analysis commonly involves split injection, so that acceptably narrow initial bands are obtained. Sensitiv­ ity is, however, often lacking, and the analyst looks for ways of introducing the sample in splitless mode. There are in­ deed some possibilities of reconcentrating initial bands, pro­ vided there are no extremely volatile components to be analyzed.

Cold Trapping

Reconcentration by cold trapping depends on the possibility of increasing the column temperature between injection and elution. For headspace analysis, these possibilities are usu­ ally limited, but they might be sufficient, especially if some peak broadening can be tolerated. Columns should have maximum retentive power, i.e. should be long, coated with a stationary phase similar in polarity to the solutes, and with the thickest film available. The oven temperature during in­ jection should be as low as possible. With manual injection, the oven door can be opened, which reduces the tempera­ ture by as much as 10°C below those achievable by auto­ mated operation with a closed door. Reconcentration is improved by cooling the column (or a part of it) below ambient temperature. Cooling increases the retentive power down to the temperature at which the sta­ tionary phase solidifies (at still lower temperatures the re­ tentive power first drops drastically before increasing again).

378

0 7. Reconcentration of Initial Bands

Many immobilized stationary phases of low polarity retain their liquid properties down to -20 °C (or even lower, if only short-term cooling is applied). Solvent Effects

Kurt Grob [95J showed that solvent effects can be used to reconcentrate broad initial bands in headspace analysis, pro­ vided the solvent peak does not obscure important solute peaks. Injection is performed in two steps using two syringes. First, pure solvent (e.g. dichloromethane) is injected. When the bulk of the solvent has entered the column, form­ ing the desired layer of liquid, the headspace sample is in­ jected. liming is important: if the headspace sample is in­ jected too soon it displaces the solvent vapor from the col­ umn entrance and prevents a sufficient amount of solvent from reaching the column before the sample itself enters; if it is injected too late the solvent evaporates before transfer of the headspace sample is complete.

7.7.7. Solvent Effect. at Elevated Column Tem­ peratures

For analyses performed at elevated temperature, cooling of the column during the splitless period can be avoided if the sample is injected with a solvent boiling ca. 30-40 °C above the analysis temperature, i.e. by using high-boiling solvents which recondense and create solvent effects. It is, for instance, possible to perform isothermal splitless analyses at 200 °C by using a solvent boiling at ca. 230-240 °C (e.g. n-tridecane).

Advantages

The use of high-boiling solvents has some important advan­ tages. It saves time, because the column temperature can be kept constantly high. 2 Solvent recondensation accelerates transfer of the sample from the injector into the column. 3 High-boiling solvents prevent evaporation of the sample inside the syringe needle and related prob­ lems. 4 In the vaporizing chamber, solutions must be depos­ ited on to surfaces (band formation). Evaporation is slow and smooth, which renders processes in the injec­ tor more reproducible. 5 High-boiling solvents produce small volumes of vapor, i.e. more can be injected and transfer into the column is faster.

Sample Preparation: Solvent Exchange

A major practical limitation ofthis technique is the purity of

high-boiling solvents.

Sample preparation with high-boiling solvents is difficult,

because, e.g., reconcentration by evaporation is impossible.

The sample must be worked up with a volatile solvent.

Then the solvent is exchanged by addition of the high-boil­

ing solvent and evaporation of the solvent used for sample

preparation. If the high-boiling solvent is admixed as a co­

solvent without removal of the volatile solvent, some of the

above advantages are lost.

7.7. Examples of the Use of Reconcentration Effects Experimental Result

379

BrOtell et al. [961 experimented with double injections, in­ jecting hexane solutions after introducing a high-boiling co­ solvent, such as pentadecane (b.p., 271°C), and determined the conditions providing optimum column efficiency for DDD and DDT ti.e. minimum band broadening arising from injec­ tion). 5 ul, of co-solvent were rapidly injected together with 5 ul, (!) of sample (introduced in 3-4 s). Fully isothermal runs were performed at 210°C.

8. Related Injection Methods 8.1. Direct Injection

Direct injection is distinguished from splitless injection by

the use of an injector with no split outlet, allowing neither

split injection nor rinsing of the vaporizing chamber after

splitless sample transfer. The connection from the vaporiz­

ing chamber to the column is "direct".

Direct injection is a vaporizing technique, i.e. the sample is

introduced into a hot chamber and transferred to the col­

umn as a vapor. Injection directly into the column is not called

"direct", but "on-column".

Complete Sample Transfer

The main advantage is that the permanent and exclusive

discharge from the liner into the column ensures complete

transfer of vaporized solute material into the column. loss

as a result of premature purging of the injector is ruled out.

If the solvent peak has a reasonable shape, not only the sol­

vent, but also the vaporized solutes are well transferred. In a

temperature-programmed run, adsorbed or retained mate­

rial can reach the column only after several minutes, yet still

be eluted with the bulk as a peak. This does not, of course,

rule out that more strongly adsorbed or retained materials

remain in the injector.

Use for GC with 0.53 mm i.d. Columns

Direct injection became popular with the 0.53 mm i.d, col­

umns, particularly in North America. There were several rea­

sons for this.

1 Sample splitting is less important because "megabore"

columns are usually also "megafilm" columns with a capacity approaching 1 Jlg. 2 0.53 mm i.d. columns are used at high flow rates, which helps overcome the major problem of the technique. 3 Direct injection is often applied with instrumentation designed for packed column GC, because packed col­ umn injectors can easily be adapted to capillary columns as long as no split outlet must be installed.

B. 1. 1. Injector Design

Direct injection must obey the basic rules of splitless in­ jection and the injector must fulfil the corresponding re­

380

D 8. Related Injection Methods quirements. The internal volume of the vaporizing chamber must be sufficiently large to accommodate the cloud of di­ luted sample vapor. A thorough treatise on this subject was published by Silvis [971.Even with column flow rates of, e.g., 10 mLJmin, sample evaporation produces vapor at a far higher rate than can be discharged into the column: if 2 ul, of sample evaporate in 0.5 s, vapor is generated at 60-120 niLJ min.

High Flow Rate during Transfer

Because the injector is not rinsed at the end of the sample transfer, narrow solvent peaks can be obtained only when transfer of the sample vapor is efficient, i.e. if there is little mixing with carrier gas in the vaporizing chamber (high gas velocity). This calls for high column flow rates or a tem­ porary increase of the flow rate during transfer. The solvent peak becomes acceptable at carrier gas flow rates of ca. 10 mLJmin, in a 4 mm i.d. vaporizing chamber producing a gas velocity of 13 mrn/s.

Column Connections

The bottom end of the vaporizing chamber and the connec­ tion with the column must be designed such that there is no dead volume below the column entrance from which vapor might enter the column during extended periods of time (causing broadening and tailing of, primarily, the solvent peak). This rules out arrangements like that of the conven­ tional split/splitless injector (shown as A in Figure D63).

'0El'

_ _'Elc'septum _ _ Carrier

purge

gas

Press-fit seal Split

Q~outlet

Splitlsss injector

Glass liner with seal 10 column

Melal liner directly coupled to column

Vaporizer with on-column injection

A

B

C

o

Figure D63 Injector designs in relation to direct injection.

Butt Connection

The capillary column can be butt-connected to the bottom of the liner. If a packed column injector is used, this is, in fact, an inexpensive means of adapting it for use with capil­ lary columns: a glass tube with the outer diameter of a packed column is fitted into the injector by means of the normal fit­ ting. The base of the glass tube is drawn to a narrow bore to enable connection with the column. The connection is, how­ ever, fragile - tightening requires well controlled manipula­ tion.

8.1. Direct Injection

381

Restek introduced the deactivation of metal parts by the "Silcosteel" procedure. This enables the construction of steel liners as shown in C of Figure D63. The column is connected to the liner by means of a ferrule and screw fitting to a threaded liner. This could replace fragile glass liners. Shrinkable PTFE TUbing

In the early days of capillary GC, the connection between the liner and the column was often prepared with shrinkable PTFE tubing, which is used for the insulation of electric wires and can be obtained from corresponding sources. Thermosta­ bility is, however, poor: above 180°C, PTFE absorbs solute material and above 230°C its mechanical stability becomes critical.

Press-Fit Connections

Fused silica columns are most frequently connected to the liner by means of a press-fit seal (8 in Figure D63). Demands on tightness are not high since the seal against ambient pres­ sure is made by the ferrule of the column attachment. The press-fit seal should, nevertheless, be checked by setting pressure on the carrier gas supply and loosening the col­ umn attachment. A strong leak causes solvent to pass the connection point and return slowly, deforming the solvent peak, as shown by Mehran (98]. After heating above 200°C, the polyimide of the fused silica tends to stick to the seat of the connection so firmly that it breaks if attempts are made to remove it. This means that the liner must be replaced whenever the column is replaced. The capillary butt remaining in the seat can be removed af­ ter heating the liner to 500°C (decomposition of the polyim­ ide).

B.1.2. On-Column Injec­ tion?

In 1979, Kern and Brander (99] described a direct injector with an 0.8 mm i.d. liner directly connected to the column, being a kind of on-column injector with a heated tube elon­ gating the column inlet. Later, kits to convert packed column injectors into capillary column direct injectors were advertized to enable on-column injection. The 0.53 mm i.d, column or a precolumn of the same bore is positioned nearthe top of the injector (D in Figure D63). The constriction in the liner forms the conical seat for the press-fit connection and serves to guide standard 265 gauge syringe needles into the col­ umn (Jennings and Mehran (100], Hinshaw (101l). Depend­ ing on the length of the syringe needle, the point of injection is somewhere in the center of the injector.

Excessively High Tempera­ ture

Such techniques do not fulfil the two most basic requirements of on-column injection. Although the heater of the injector can be switched off, the injector temperature is usually far above the boiling point of the sample solvent (the injector is heated from the oven and from the heating block of the de­ tector; its insulation prevents dissipation of heat). The high temperatures have two consequences.

382

D 8. Related Injection Methods

2

They are likely to cause sample evaporation inside the syringe needle, with the consequences of selec­ tive losses of high-boiling material and uncontrolled in­ jection volumes. If the point of injection is inside the injector, part of the sample liquid flows into the oven-thermostatted column, but some of it remains on the capillary wall inside the injector. Vaporization of high-boiling solutes requires an injector temperature not far below the maximum used for analysis (or a PTV injector to increase the tempera­ ture after the injection); this brings us back to "hot on­ column" injection.

Low Oven Temperature During Injection

Owing to the lack of room in the column inlet for the vapor

generated by fast sample evaporation, conventional on-col­

umn injection requires that the oven temperature during in­

jection be below the solvent boiling point. This limits the

amount of vapor formed to that which can be discharged by

the carrier gas. With hot on-column injection, the large vol­

ume of vapor generated inside the injector expands by over­

pressure. Only recondensation into a cool oven-thermostat­

ted column inlet might pull it onwards at a rate sufficient to

prevent backflow out of the column entrance. Hence oven

temperatures during injection must be clearly below the sol­

vent boiling point.

Conclusions

In capillary GC, on-column injection provides the most accu­

rate results and avoids loss of solute material by degrada­

tion. These advantages are lost in "warm" or "hot"

on-column injection.

The technique might produce better results for adsorptive

components because the column inlet is usually more in­

ert than the liner.

Involatile sample by-products are a problem as in on­

column injection, since they are partly carried into the oven­

thermostatted column.

8. f .3. Injection of Large Volumes

In 1990, Watanabe and Hashimoto (1021 reported injection

of up to 100 III of liquid samples by a combination of di­

rect injection and retention gap technique. The sample

was introduced slowly (ca. 5 IlLJS) into an injector at 250­

300 cC. The vapor was discharged concurrently by means of

a high carrier gas flow rate and solvent recondensation in

the column inlet.

Injector

A Hewlett-Packard 5890 instrument was used with a packed

column injector containing a 2 mm i.d. liner. The tempera­

ture profile of the injector had to be improved to elimi­

nate tailing peaks for the solvent and the high-boiling sol­

utes. Insulation of the injector head was improved and an

additional heater added to the column attachment zone.

8.1. Direct Injection

383

Carrier Gas Supply

The carrier gas supply was modified to enable an increase of the column flow rate during solvent evaporation. The gas supply line was split and a pressure regulator mounted parallel to the mass flow regulator. The flow regu­ lator delivered the high gas flow rate (10-20 mLJmin) for the solvent evaporation period. After completion of the injection process, this supply line was shut and the pressure regula­ tor delivered the column flow for analysis. The separation efficiency of the system was reduced by ca. 10 %, possibly because of the large precolumn.

Uncoated Precolumn

An uncoated precolumn of 17-20 m x 0.3 mm i.d. was used. It had the capacity to retain 100-120 ul, liquid [1031. The hexane solution introduced at 5 JlLJs produced vapor at a rate of almost 1 mLJs. At a carrier gas flow rate of 10 mLJmin (0.167 rnt/s), ca. 15 % ofthe solvent was discharged through the column concurrently with the introduction, while 85 % spread as a racondensed liquid in the uncoated precol­ umn.

Comment

The technique described by Watanabe and Hashimoto is a precursor of large volume splitless injection with a solvent vapor exit as described in 1994 by Suzuki et al. (Section D4.2) and of the vaporizer/precolumn solvent splitting system de­ scribed in 1996 [291, where a hot vaporizing chamber was added to an on-column system.

Valve against Backflow

Section D3.3.8 summarized the proposal of Kaufmann in which a valve (e.g. a Jade airlock) was introduced above the liner to close the vaporizing chamber during the period when solvent evaporation causes overpressure. In another paper [104). Kaufmann proposed the use ofthis valve for large vol­ ume injection. His system brought together elements from Watanabe and Hashimoto (see above) as well as from Suzuki. He used a 4 mm i.d. direct liner (Uniliner, Restek) with the Jade valve above it. The liner was packed with fused silica wool to slow down evaporation (7). The sample was transferred to a 15 m x 0.53 mm i.d. uncoated precolumn leading to an early vapor exit and the separation column. 50 III of a solution in ac­ etone was injected at 50°C oven temperature, hence recon­ densing part of the solvent.

B. 1.4. Evaluation of Direct Injection

For general evaluation, direct injection must be compared with splitless injaction, because it is a non-splitting method suitable for the analysis of dilute samples, including samples loaded with involatile by-products.

Shape of the Solvent Peak

In direct injection, the shape of the solvent peak is often a problem when the detector is sensitive to the solvent. Two solutions are viable: the use of high carrier gas flow rates (at least 10 mt/min) or of an increased gas flow rate during trans­ fer (implying flow or pressure programming).

384

0 8. Related Injection Methods

In splitless injection, a perfect solvent peak can be achieved even under conditions of poor sample transfer. While this is certainly of advantage to the visual appearance of the chro­ matogram, it hides imperfections affecting quantitative analy­ sis. Quantitative Analysis

Shagena and Hinshaw [1051 found an improvement in both detectability and reproducibility of results when comparing wide-bore columns and direct injection with packed column GC. Mallet and Mallet [1061 performed a similar comparison for the analysis of organophosphorus pesticides and con­ cluded that although the separation efficiency improved, the reproducibility of the results deteriorated somewhat.

Comparison with Splitless Injection

No experimental data comparing direct and splitless injec­

tion are available, apparently because ofthe widespread (but

unjustified) belief that the application of direct injection is

restricted to 0.53 mm i.d columns and that of splitless injec­

tion to those of narrower bore.

1 Quantitation by direct injection is expected to be more

reliable because it rules out losses from premature purging of the injector (see above). 2 Direct injection is simpler because there is one fewer variable to select, i.e. the duration of the splitless pe­ riod. 3 Owing to the direct connection of the vaporizing cham­ ber to the column, there is no possibility of shooting the sample past the column entrance or of vapor being driven into the split outlet by the pressure wave during evaporation. This also means, however, that the sample must pass through possible deposits of septum particles and other dust located above the column en­ trance (the same as in splitless injection when wool is placed above the column or the column entrance is po­ sitioned below the orifice of a "goose neck" liner).

"Ghost" Peaks

The permanent and complete transfer of material from the injector into the column can also be a drawback: involatile sample by-products on the liner wall often slowly de­ grade to more volatile products and with direct injec­ tion, all this material is transferred into the column. In iso­ thermal chromatography it merely elevates and disturbs the baseline, but in temperature-programmed runs it leads to the formation of" ghost" peaks (after reconcentration by cold trapping). In splitless injection, these materials are split (ex­ cept during the splitless period); if necessary, the split flow rate can even be increased to remove more of the degrada­ tion products.

Column Attachment

Particularly after the introduction of programmable carrier gas supply (high flow rate for transfer), direct injection seems highly competitive with splitless inj~ction. It is, never­

-8.1. Direct Injection

385

theless, rather seldom used. This certainly has to do with the conservative attitude of analysts in routine laboratories, but also with the difficulty of mounting the column. None of the systems marketed so far can compete with the simplicity of column installation into a classical splitless injector.

8.2. Solid Injection

"Solid injection" means introduction of a sample free from solvent. The term is also used when the solvent-free resi­ due of a sample is actually a liquid. Numerous solid injection techniques have been conceived, but most have disappeared again. In the early nineteen sev­ enties, sample extracts were loaded into glass capillary tubes 2 cm long. Some 30 could be placed in a manifold (Carlo Erba, Milan) in which the solvent was evaporated un­ der vacuum. The capillaries were then loaded into a revolv­ ing device which dropped them into a heated liner. After a number of analyses the accumulated tubes had to be re­ moved from the liner. Another system involved metallic cap­ sules which were filled with sample and closed. In the injec­ tor, they were opened pierced by a thorn.

Damage by Solvent?

At that time, there was a widespread belief that the solvent should be removed from the sample before injection. The main reason was the fear that the solvent could damage the column. Today we know that this is not true (for the most commonly used solvents at least), not even for large volume injection. Detectors, also, are not affected by the amounts of solvent introduced in splitless injection. On the other hand, problems related to large volumes of vapor (required size of the vaporizing chamber, slow discharge through the col­ umn) do, indeed, arise as a result of the solvent, but this has not been used as an argument for solid injection.

8.2.1. Moving Needle Injection

In the early days of capillary GC, the strongest competitor with splitless injection was the moving needle injection in­ troduced by van der Berg and Cox in 1972 [1071. It found particularly wide application in France. The semple was deposited on to the tip of a glass nee­ dle at ambient temperature, held vertically in a tube mounted above the heated injector chamber (Figure D64). The sol­ vent was evaporated in a stream of carrier gas leaving the tube through the top. The needle was then lowered into the vaporizing chamber (by means of a magnet), where the sam­ ple components were vaporized and transferred into the col­ umn under splitless conditions.

Evaluation

Deposition of the sample liquid on to the cool needle rules out problems related to evaporation inside the syringe nee­ dle. Volatile solute material is lost by co-evaporation with the solvent, restricting the application to components of in­ termediate to high-boiling point (starting, for instance, with methyl palmitate). The vaporizing chamber can be small, enabling efficient transfer into the column.

386

D 8. Related Injection Methods Resistance for controlled release of gas and solvent vapors Magnet to move the needle; up position for loading

Syringe depositing extract onto the needle tip

Gas supply Vaporizing chamber Heating block

Column

Figure D64

Moving n_dle injector.

Albaiges et al. [1081 compared splitless and moving needle injection for a test mixture comprising the C16-C36 n-alkanes. Splitless injection provided better results in terms of discrimi­ nation, but the reproducibility for the higher alkanes was slightly better for the moving needle technique. Automation

Automation of moving needle injection is difficult. This might be one of the reasons why the technique has largely disap­ peared. Analytical Services (Poitiers, France) introduced a system called "Injecteur Automatic en Phase Solide" (APS) which automates a moving needle injection of different de­ sign. The sample is applied to a cool needle from which the solvent is evaporated. This needle is introduced into a nar­ row vaporizer for thermal desorption. After withdrawal, the needle is heated to pyrolyze and remove the remaining involatile material. Heating of the needle can also be used for pyrolysis analysis.

8.2.2. Direct Sample

More recently, solid injection was revived for the direct in­ troduction of a piece of sample for thermal desorption of the components of interest. Instead of the needle the APS sys­ tem mentioned above can be equipped with a tool designed to hold fibers, such as hair (e.g. for drug analysis), plant or polymer fibers, or wire. A third tool is a vial for powders or small objects (e.g. a piece of cork to check for off-flavor).

Introduction

8.2. Solid Injection

387

Advantages

The following advantages were listed: reduced sample manipulation, avoidance of solvents, higher concentration (no dilution during an extraction), retention of involatile material in the sampling system, better transfer into the column (small vaporizing cham­ ber). The main drawbacks are probably related to the difficulty of thermal desorption from a complex matrix and the small size of the sample being representative of highly homoge­ neous samples only.

Pesticide Analysis

Jingand Amirav[109] described a device marketed by Varian as the "ChromatoProbe", which fits the Varian 1078 or SPI injectors. They used 1.6 mm o.d.{1.2 mm i.d. sample vials 10-15 mm long in a vial holder system replacing the septum cap of the injector. Sampling involved the following steps: 1 1-5 ul, of sample extract or homogenate in solvent were transferred to the vial and the vial was introduced into the holder. 2 The vial was transferred into the injector at 80-90 °C for solvent evaporation, with the column at 70°C for cold trapping of the solutes. At a split flow rate of 50 mLJ min, solvent evaporation lasted 1 min. 3 The injector was heated at 300 a/min to 250°C and held there for 30 s, the split outlet being closed. During this splitless period, the components of interest (pesticides) were "thermally extracted" while the matrix material remained in the vial. 4 The vial was removed from the injector and disposed of. While the injector was open, a protective carrier gas flow prevented air entering. As an example, 150 g tomatoes were blended with 300 ml acetone, of which 1.6 III was placed in the sample vial. Diazinon, parathion, methyl parathion, methyl trithion, and ethion were analyzed with the pulsed flame photometric detector (PFPD), with a detection limit of a few ppb. A carrier gas flow rate of about 5 rnt/rnln was required, because re­ covery of the high-boiling components was low at 1 mLJmin.

Drug Analysis

For the analysis of drugs in hair, a single 1 cm piece of hair was placed in the vial with 10 III methanol [110]. The vial was introduced into the injector at 120°C, then heated to 250 °C. This enabled the detection of cocaine and 6-mono­ acetyl morphine. Other analyses included drugs in urine (the urine sample was placed directly in the vial),

Thermal Desorption

Quantitative thermal desorption is a major problem in direct sample introduction. 1 The sample cannot be heated to high temperatures be­ cause thermal degradation of the matrix produces "forests of peaks" in the chromatogram.

388

0 8. Related Injection Methods

2

The solute material may be enclosed in solids, particu­ larly after drying at high temperatures. 3 Deep and narrow bore sample vials do not enable effi­ cient extraction of solute vapor, because there is no gas flow sweeping the sample out; solutes must leave the vial largely by diffusion. If the analysis must merely determine the presence of a given component, and this is often the only requirement, incom­ plete thermal desorption is, of course, of secondary concern.

8.3. Injector-Internal Headspace Analysis

The method to be described here could also be termed "di­ rect sample introduction with thermal desorption". The origi­ nator, Morchio [111), did not give it a name. We prefer "In­ jector-Internal Headspace Analysis" since this was, at least initially, the main field of application. It differs from the above technique by restricting the sample matrix to a high-boil­ ing liquid, the most important being edible oil or fat. Seen from this angle it could also be termed "thin film desorption". The technique was first described in 1982, but remained an Italian secret because most ofthe workwas published in the Italian language.

Steps of the Process

2

3 4

The sample is injected by syringe and transferred to the liner wall (Figure D65). It forms a film from which the volatile components evaporate and are transferred into the column (in split or splitless mode). Most of the non-evaporating material (oil) slowly flows along the wall to the bottom of the liner. The oil accumulates in a bag. Syringe Septum

Needle with side port hoi

1 Oil transferred to liner wall 2 Volatiles evaporate

3 Oil flows to bottom

4 Oil accumulated from previous injections

Figure D65 Liner for injector-internal headspace analysis and the four steps of sampling involved. (From ref. [112].)

8.3. Injector-Internal Headspace Analysis

389

High Temperature

Injector-internal heads pace analysis occurs at a higher tem­

perature than is usual for headspace analysis. The injector

temperature is primarily limited by the volatility of the ma­

trix, because the matrix should not evaporate and enter the

column. For edible oils and fats, the maximum is ca. 200­

220°C.

Supported by intense stripping of the solutes from the thin

film, evaporation is complete for solutes of up to rather high­

boiling points, such as pesticides. This greatly extends the

range of application and simplifies quantitation, because

evaporation and transfer are mostly complete, rather

than in equilibrium with the matrix.

Liner with Bag

Injector liners with the specially designed bottom are avail­

able for ThermoQuest instruments.

Packing of a standard straight liner with wool also stopped

the oil above the column entrance and was simpler to use,

but after a few injections, peaks of pesticides were broad­

ened and reduced in area. The oil accumulated and built up

increasing retentive power. With the open tubular liner, most

of the 'oil slowly flows to the bottom, making way for the

next sample.

Column Entrance

The column entrance should be positioned above the ori­

fice of the indentation (Figure D65) such that vapor or deg­

radation products from the oil diffusing out of the bag are

carried away into the split outlet and cannot reach the col­

umn.

Injection

1-3 u], of oil or fat, undiluted or mixed 1:1 with acetone, were

injected by use of a syringe with a side port hole needle.

The needle was inserted 4 em, depositing the sample in the

upper part of the chamber. A splitless period prolonged to 4

min improved the results, because it enabled complete trans­

fer also for components which were somewhat retained by

the oil.

Dilution with solvent helps withdrawal of viscous or solid

samples into the syringe. Visual experiments showed, how­

ever, that the volatile solvent can hinder transfer to the

liner wall. For this reason samples should not be diluted by

more than a factor of about four.

Cleaning of the Liner

Cleaning of the liner with dichloromethane is simple as long

as the oil is not polymerized and should, therefore, be per­

formed at the end of every day.

Oil polymerizes in the presence of air, for instance when the

carrier gas is switched off overnight. The resulting yellow or

brownish polymer is best transesterified in warm methoxidel

methanol solution, as this is less aggressive than aqueous

alkali and leaves behind surfaces which are less adsorptive.

The liner should then be rinsed with 1-5 % hydrochloric acid

(to remove highly adsorptive vicinal silanols) and dried in

the hot injector with the septum cap removed (1-3 min).

r !

390

0 8. Related Injection Methods

Contamination of the Column

After some 20 injections, peaks usually started to become broad as a result of high-boiling material accumulating in the column. The disturbing material probably consists of trig­ Iycerides and minor components, e.g. mono- and diglycer­ ides, free sterols, and squalene.

Uncoated Precolumns?

Uncoated precolumns do not help when oven temperatures exceed about 200°C. Because of their low retentive power, even triglycerides slowly move through them and ac­ cumulate in the inlet of the separation column; replacement of the precolumn is no longer effective. The problem can be solved by heating to 350°C, discharging the material through the whole column. Such cleaning must occur before the un­ saturated compounds polymerize.

Cleaning through Purge Exit

Because the removal of high-boiling material must occur rather frequently and the separation column is often not suit­ able for baking out triglycerides, the disturbing material was discharged through a purge outlet (Figure D66) consisting of an 0.32 mm i.d. fused silica capillary placed behind a 2 m x 0.25 mm i.d, precolumn coated with an 0.08 11m film of a methylsilicone. On the one hand, this thin film enabled discharge of the triglycerides at a modest temperature. On the other, retentive power was sufficiently high to prevent high-boiling material from reaching the separation column up to oven temperatures of ca. 250°C. Every evening, the purge exit was opened and the oven tem­ perature increased to 330 °C for 5-10 min. The high gas flow rate through the short precolumn results in rapid discharge. The exit was closed either by means of a press-fit cap, pre­ pared from a press-fit connector divided into two and flame­ sealed in the center, or a soft septum.

Injector I

Purgeexit with press-fitcap

Detector

¥­

v\

Thin film Separation column

precolumn

Figure D66 Precolumn and purge outlet for the discharge of matrix material of intermediate volatility, such as edible oil. (From ref. [112].) .

-8.3. Injector-Internal Headspace Analysis Applications

397

The method was first used for the determination of solvent residues (hexane) in edible oils. Mariani and Fedelithen ap­ plied it for the analysis of antioxidants, such as BHA, BHT. and Jonox 100, in edible oil [1131. later for chlorinated sol­ vents [114]. Morchio, de Andreis, and Verga [115] described the determination of organophosphorus pesticides in edible oils (or fats) by injection of the oil. Using flame photo­ metric detection (FPD),the detection limit was ca. 1 ppb. Droz [116] applied the technique to the analysis of chlorinated pesticides in edible oils and fatty foods (such as extracts from fish), as well as for detection of fqod irradiation by analysis of the olefins cleaved from triglycerides. We have used it for the analysis of flavor components in extracts from fatty foods, in particular for the gamma- and delta-deca­ and dodecalactones.

9. General Evaluation of Splitless Injection It is difficult to answer a question as general as "how accu­ rate are results obtained by splitless injection" or rather "how accurate can they be under thoroughly optimized conditions". Many papers, primarily from the nineteen eighties, deal with this question, but they actually report data on precision (re­ producibility) rather than accuracy. More data are avail­ able from method validation, but now splitless injection is merely a contributor to a more complex system of uncer­ tainty. 9.1. Data on Precision from the Literature n-Alkanes

Generally Valid Conclusions?

In 1982, Watanabe et al. [117] found coefficients of variation for split/ess injection on to capillary columns to be far higher than for packed column GC. For the area ratio of the C12 and C28 n-alkanes in hexane and an injector temperature of 250°C it was nearly 40 %. When the injector temperature was in­ creased to 350°C, it decreased to 2 % and even for the n-C12/ C44 area ratio it was only 6 %. Such high coefficients of variation at 250°C injector tem­ perature should be understood as a sign that something went fundamentally wrong. Discrimination against high-boiling compounds was, in fact, extremely high: when the injector temperature was 250 °C. 80 % of the ,..C28 was lost; at 350°C, however, only 20 % of the n-C28 and less than 40 % of the n-C44 were missing. Some results are summarized in Figure 067. The question arises as to the extent to which such results can be used as a basis of generalized statements. According to our experience, they are poor for clean test samples.

392

0 9. General Evaluation of Splitless Injection

..g• .

1.20

u.. 1.00

..

;0 0

.... ..li ~

0.80

U!

U!

ll!

0.60

~ ...

.~..

0.40

(C)

0.20

CARBON NlIMBER

Figure D67 Discrimination curves for C12-C44 ...alkanes with different injection techniques. (From Watanabe et al. [117].) A. Capillary GC. on-column injection. B. Capillary GC. split­ less injection. injector at 350°C. C. Same. but injector at 250°C. D. Packed column GC. on-column injection. E. Packed column GC. flash evaporation injection. injector at 350°C. F. Same. but injector at 250°C.

At an injector temperature of 250 °C, coefficients of variation for C28 are usually 3-6 % and discrimination, primarily as a result of losses inside the syringe needle, below 15 %. The results suggest that some (then still unknown) elemen­ tary rules were neglected. Neither the syringe needle han­ dling technique is described nor is information provided on injector geometry, the temperature of the injector head, or the length of the syringe needle. The carrier gas flow rate (hardly 1.3 mLJmin) was low. On the other hand, most of today's injectors are similar, and in routine laboratories the know-how on optimizing conditions is probably inferior. Hence such performance could easily be obtained even to­ day. Maybe the data should be understood as a warning about the possible extent of errors. Polyaromatics

Springeret al. [118] did not obtain better results, even though they merely used standards in a pure solvent. The mix­ ture contained polynuclear aromatics (PNAs) and a few n­ alkanes serving as internal standards. A very high injector temperature (380°C) was found to provide the best results. The coefficient of variation for peak areas of pyrene (4-ring system eluted at ca. 170-200 °C) normalized with respect to n-decane was, nevertheless, as high as 15 % (only 1 % if n­ eicosane was the internal standard). The carrier gas flow rate at the column temperature during sample transfer was

-9.1. Data on Precision from the Literature

393

hardly 1 mllmin and the splitless period lasted for only 30 s. The coefficient of variation for absolute areas of a compo­ nent like pyrene should not exceed 5 %. It should be lower if an internal standard is used, irrespective of whether it is n­ decane or n-eicosane. In 1985, McMahon [1191 reported results of a collaborative study on the quantitative results obtained by split, splitless, and on-column injection using a sample of polyaromatic hydrocarbons (naphthalene to chrysene). Results were based on an internal standard. After excluding outliers, the relative standard deviation of the results from 23 laboratories was ca. 6 %. On-column injection produced clearly better results.

Comparison with Split Injection

Schomburg et al. [1201 tested splitless injection with a mix­ ture of C18-C36 n-alkanes in different solvents and came to the conclusion that although the performance of splitless injection was better than that of split injection, it de­ pended strongly on solvent volatility. Most of the deviations resulted from syringe needle problems.

Phenols

Kalman [1211 reported results on the determination of free phenols. Although he did not use conditions which could be considered optimized, standard deviations were mostly 1-2 % when an internal standard was used and about three times higher with use of an external standard (25 ng/flL solu­ tions).

Chlorinated Benzenes and Biphenyls

In 1983, Onuska et al. [1221 evaluated splitless and on-col­ umn injection for the analysis of chlorinated benzenes and biphenyls. They concluded that quantitative results obtained by splitless injection were not satisfactory, particularly so if mixtures contained components with a wide range of vola­ tility. "The splitless injector acts like a non-linear splitting device and delivers unpredictable and irreproducible quantities of individual components on to a WCOT column. " This is an unusually clear statement in a scientific journal. Manual injection resulted, in fact, in data with relative stand­ ard deviations of up to more than 30 % and which changed from one day to the next. With an autosampler, re­ producibilities were within a few percent. The experiments were performed on two types of instrument, both equipped with liners which were much too small.

Drugs

Plotczyk [1231 reported coefficients of variation of 5-12 % for absolute peak areas of drugs when the injector was kept at 200-250 °C. With an internal standard, relative stand­ ard deviations were ca. 5 %. The test mixture was synthetic, hence free from involatile by-products.

Pesticides

Stan and Goebel [1241 tested the repeatability of splitless and on-column injection for mixtures of pesticide standards,

394

D 9. General Evaluation of Split/ess Injection

using the internal standard method. Relative standard de­ viations for chlorinated pesticides varied between 0.2 and 5.5 % (mean 1.7 %), those for the organophosphorus pesti­

cides between 1.1 and 11.8 % (mean 3.4 %).

Response factors obtained with splitless injection were

checked with those from on-column injection. Compared with

the internal standard, up to two thirds of the later eluted

material was lost, whereas the peaks of some other compo­

nents were up to 60 % too large (owing to discrimination

against the internal standard).

9. t, t, Limited Utility of Literature Data

9. '.2. Message to a Lawyer

The utility of such data for a generalized evaluation of split­ less injection must be questioned for the following reasons. For the determination of accuracy, data on precision (re­ producibility) is of limited value. Systematic errors of­ ten far exceed random errors (standard deviations). Deviations arising from discrimination or incomplete sample transfer often change from one set of conditions (e.g. for the calibration mixture) to another (e.g. for the analysis of the sample). Data on precision can, there­ fore, only be used to estimate the minimum probable error - and this is hardly what we are interested in. 2 Hardly any of the reports adequately specifies the ex­ perimental details. It is, therefore, impossible to con­ clude whether high standard deviations were a result of poorly optimized conditions or of the splitless injection technique per se. Unfortunately most of today's stand­ ardized methods are no better at specifying injection conditions. Hence analysts cannot be expected to pro­ duce better results. 3 The reproducibility tests reported were performed with clean, synthetic mixtures. Real samples containing a considerable concentration of high-boiling or involatile by-products tend to be characterized by higher devia­ tions (drifting matrix effects). Most results from analytical laboratories eventually end up on the desks of engineers or lawyers who have to make decisions based on our results, but know nothing about analytical chemistry (except that our results are expen­ sive and not very reliable). If they asked for some general indications on accuracy and reliability, we might give them the following answers. 1 Despite the much higher cost, trace components can­ not be measured as reliably as temperature in the office. 2 It is probably more difficult to determine the range of uncertainty than the concentrations or amounts them­ selves, because standard deviations and linearity tests provide minimum rather than maximum deviations. 3 If sampling and sample preparation do not signifi­ cantly contribute to deviations, relative standard de­

9.1. Data on Precision from the Literature

4

5

6

7

S

9

9.2. Comparison with Alternative Techniques

9.2. 1. On-Column Injec­ tion

Better Performance

395

viations easily reach 10 %. Then statistics tell us, e.g., that there is 65 % probability that the true value is between 9 and 11. To obtain 95 % certainty, the range must be widened by a factor of two. Statistically there is, however, still 5 % probability that the true value is below S or above 12 - and in reality, the probability is ratherhigher than 5 %. Standard deviations do not include systematic errors. Systematic deviations are frequent, often rather sub­ stantial, and difficult to quantitate. If, e.g., extraction and derlvatizatlorr are involved, rela­ tive standard deviations are closer to 20 % and the true value within 40 % of the result given. Hence the value is likely to be somewhere between 6 and 14. Systematic deviations (such as matrix effects) can cause a deviation of 60 %. Hence, if the analyst had to guarantee the result by staking his personal sal­ ary~ he would haveto indicate a value between maybe 4 and 16 and still make sure that he has enough food in the freezer to survive in case he lost his salary. This staggering uncertainty has nothing to do with poorly performed work, such as miscalculation, inac­ curate weighing, or spilling sample liquid on the lab bench. The major part of the deviations has to do with, e.g., uncontrolled processes during injection. The accuracy of the results could often be substan­ tially enhanced. This would presuppose perfecting techniques (such as spHtless injection) and methods (e.g. by introduction of verification procedures), but presently nobody seems to be willing to fund such an investigation. Working hard on the accuracy of their results, many' analysts overestimate their achievement and deliver results like 10.2 when, in reality, it would be more appropriate to state" a value between Sand 12", l.e. 10±2.

Splitless injection cannot be classified as accurate, reliable. and well developed, but is applied still far more widely than alternative techniques. This vote from the users either flat­ ters splitless injection or confirms that chromatographers are conservative. An absolute judgment is rather useless: splitless injection must evaluated in comparison with the alternatives. In real­ ity it has to compete mainly with cold on-column and PTV injection. On-column injection is the method of choice whenever its application is possible. With no other technique are the re­ sults more reliable; there is no discrimination or other devia­

396

0 9. General Evaluation of Split/ess Injection

tion affecting sample introduction apart from adsorptive ef­ fects, and the danger of systematic errors is small. Its applica­ tion is simple, as there are few working rules to be followed. Larger Sample Volumes

On-column injection is also more flexible with regard to sam­ ple volume: from 0.2 to 500 III can be injected without loss of volatile components. In fact, the limitation of the sam­ ple volume in splitless injection to a few microliters is a draw­ back which is much more severe than commonly recognized, because most analysts are used to it and methods are de­ signed accordingly.

Only for rather Clean Sam­ ples

Considering these overwhelming advantages, it was believed that splitless injection would soon be replaced. There remain, however, important applications where on-column injection cannot compete. On-column injection is primarily hampered by the fact that involatile sample material is deposited in the oven-thermostatted column inlet and can severely dis­ turb chromatography. Even if an uncoated precolumn is used, peaks are broadened and tail as soon as the accumulated amount of involatile material exceeds ca. 10 Ilg [125]. i.e. after a few injections of samples containing 0.1 % of in­ volatile material.

9.2.2. Splitless Injection for Analysis of "Dirty" Samples

The major field of application of splitless injection is trace analysis of "dirty" samples, e.g. many biological materials, foods, and environmental samples. Splitless injection toler­ ates far larger amounts of involatile material than on-col­ umn injection, because most of it is deposited inside the in­ jector. Usually many injections of solutions containing as much as 10 % non-evaporating material can be per­ formed before the liner needs cleaning or replacement and a contaminated section ofthe column inlet must be removed. Colored and even rather viscous samples can be injected. Although often chromatograms with at least reasonably shaped peaks are obtained, the analyst should not forget that involatile material can seriously affect quantitation (matrix effects). Fortunately, for many typical applications involving samples loaded with by-products, such as pesticide residue analysis or the determination of veterinary drugs in meat or urine, accuracy is not of primary importance. Sampling (e.g. selec­ tion of the apple being analyzed for pesticides) usually intro­ duces the predominant uncertainty.

Minimizing Sample Clean-Up

The often-heard statement that only clean samples should be analyzed is unrealistic. For many samples, clean-up suffi­ cient for on-column injection requires excessive effort. Clean­ up is, moreover, a permanent source of uncertainty, e.g. from varying recoveries and blanks. It is, therefore, certainly worth­ while devoting some effort to the analysis of contaminated samples (such as checking for matrix effects) to save the far larger amount of work needed for sample clean-up.

-9.2. Comparison with Alternative Techniques

397

9.2.3. PTV Splitless Injection

Programmed Temperature Vaporizing (PTV) injection in the

splitless or solvent split mode is another serious competitor

against classical splitless injection. In terms of accuracy and

simplicity it cannot compete with on-column injection, i.e. it

is not the first choice for the analysis of clean samples, but

has important advantages which make it a tough competi­

tor against classical splitless injection in the field of

the matrix-loaded samples.

Vaporization Process

Because the sample is introduced into a cool injector, PTV

injection rules out problems arising from evaporation of the

sample in the syringe needle. Sample evaporation occurs

from a surface (liner wall or packing material), which keeps

the process under control and ensures that only vapor

reaches the column, ruling out contamination ofthe column

inlet with involatile material. So far it resembles splitless in­

jection under conditions of band formation.

For the analysis of high-boiling, adsorptive, and labile com­

ponents, PTV injection does not always produce superior

results. This reflects the difficulty of vaporizing these solutes

from active, possibly contaminated surfaces. Nebulization in

a hot chamber and evaporation in the gas phase. as com­

monlyachieved in classical splitless injection by the hot nee­

dle technique, is unsurpassed in this respect.

Matrix Effects

Enhancing matrix effects may be strong also in PTV injec­

tion (they have the same source as in classical splittles in­

jection). Reducing matrix effects, however. were found to be

strongly reduced [1261. probably because of the well con­

trolled deposition of the sample liquid in the vaporizing cham­

ber.

Sample Volume

PTV splitless injection is possible for sample volumes up

to 20 JlL without loss of volatile components. which is an

important advantage in trace analysis. It is difficult to under­

stand why this has not developed into an important com­

petitive advantage. In solvent split mode, the sample vol­

ume can be further increased by a factor of ten. One factor

limiting these techniques is again the lack of inert packing

materials for the liner.

9.2.4. Outlook

In the nineteen eighties it seemed obvious that a technique

as unreliable and difficult to control as splitless injection

would soon be replaced by better technology. This did not

happen and is why efforts to improve classical splitless

injection should be continued.

Connected with the improvement of splitless injection, there

should also be standardization on a single principle (neb­

ulization or band formation, with all the consequences in­

volved), such that methods can be perfected also in this prob­

ably most sensitive step of the analysis.

398

0 References

References D

2 3 4 5 6

7 8 9 10

11

12

13 14 15

16

17 18 19

20

Kurt Grob and Gertrud Grob, "Splitless Injection on Capillary Columns, Part I, The Basic Technique; Steroid Analysis as an Example", J. Chromatogr. Sci. 7 (1969) 584, and "Part II, Conditions and Limits, Practical Realization", J. Chromatogr. Sci. 7 (1969) 587. Kurt Grob and Konrad Grob, "Isothermal Analysis on Capillary Columns without Stream Splitting; the Role ofthe Solvent", J. Chromatogr. 94 (1974) 53. K. Grob, "Solvent Effects in Capillary GC", J. Chromatogr. 279 (1983) 225. K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg (1987) 270 and 302. K.Grob and A Artho, "Distorted Solvent Peaks in Capillary GC: A Symptom of Porous Column Surfaces", J. High Resol. Chromatogr. 13 (1990) 803. Kurt Grob and G. Grob, "Techniques of Capillary GC. Possibilities of the Full Utilization of High-Performance Columns. Part I: Direct Sample Injection", Chromatographia 5 (1972)

3. J. V. Hinshaw, "The Effects of Inlet Liner Configuration and Septum Purge Flow Rate on Discrimination in Splitless Injection", J. High Resol. Chromatogr. 16 (1993) 247. R.D. De Veaux and M. Szelewski, "Optimizing Automatic Splitless Injection Parameters for GC Environmental Analysis", J. Chromatogr. Sci. 27 (1989) 513. K. Grob, M. Biedermann, and Z. Li, "Checking the Capacity of a Splitless Injector - a Simple Test", J. Chromatogr. 448 (1988) 387. K. Grob and M. Biedermann, "Injector Chambers for Classical GC Splitless Injection Must be Large!", J. High Resol. Chromatogr. 12 (1989) 89. R.P. Snell, J. W. Danielson, and G.S. Oxborrow, "Parameters Affecting the Quantitative Performance of Cold On-Column and Splitless Injection Systems Used in Capillary GC", J. Chromatogr. Sci. 25 (1987) 225. H.B. Lee, R. Szawiola, and R.S.Y. Chau, "Solvent Effects on Response Factors for Polynu­ clear Aromatic Hydrocarbons Determined by Capillary GC Using Splitless Injection", J. Assoc. Off. Anal. Chern. 70 (1987) 929. K. Grob, "Solvent Effects on Response Factors - Problems in Performing Splitless Injec­ tion", Letter to the Editor, J. Assoc. Off. Ana1. Chern. 71 (1988) 76A. C.-G. Hammar, "New Injector Design for Splitless Capillary Column GC", J. Chromatogr. 249 (1982) 167. A Kaufmann, "Prevention of Vapor Overflow in Splitless Injection by a Novel Injector Design", J. High Resol. Chromatogr. 21 (1998) 258. Ph.L. Wylie R.J. Phillips, K.J. Klein, M.O. Thompson, and B.W. Hermann, "Improving Splitless Injection with Electronic Pressure Programming", J. High Resol. Chromatogr. 14 (1991) 649. K. Groband K. Grob, "Splitless Injection and the Solvent Effect", J. High Resol. Chromatogr. Chromatogr. Commun. 1 (1978) 57. F.J. Yang, AG. Brown, /II, and S.P. Cram, "Splitless Sampling for Capillary-Column GC", J. Chromatogr. 158 (1978) 91. K. Grob, S. Brem, and D. Frohlich, "Splitless Injection of up to Hundreds of Microliters of Liquid Samples in Capillary GC: Part 1, Concept", J. High Resol. Chromatogr. 15 (1992) 659. K. Grob and S. Brem, "Splitless Injection of up to Hundreds of Microliters of Liquid Sam­ ples in Capillary GC: Part 2, Experimental Results", J. High Resol. Chrornatoqr, 15 (1992) 715.

D References 21 22

23 24

25

26

27

28

29

30 31

32 33 34

35 36

37

38 39

40

41

399

K. Grob and D. Frohlich, "Splltless Injection of Large Volumes of Aqueous Samples - A Basic Feasibility Study", J. High Resol. Chromatogr. 16 (1993) 224. K. Grob and D. Frohlich, "Splitless Injection of Large Volumes: Improved Carrier Gas Regulation System", J. High Resol. Chromatogr. 15 (1992) 812. K. Grob and Ch. Siegrist, "Determination of Mineral Oil on Jute Bags by 20-50 JlI Splitless Injection onto a 3 m Capillary Column", J. High Resol. Chromatogr. 17 (1994) 674. T. Suzuki, K. Yaguchi, K. Ohnishi, and T. Yamagishi, uGC Detection of Tris(2-chloroethyl) and Tris(2-butoxyethyl)phosphate in Groundwater by Large Volume Injection", J. AOAC Int. 77 (1994) 1647. Th. Noy, E. Weiss, T. Herps, H. Van Cruchten, and J. Rijks, "On-Line Combination of LC and Capillary GC. Preconcentration and Analysis of Organic Compounds in Aqueous Sam­ pies", J. High Resol. Chromatogr. Chromatogr. Commun. 11 (1988) 181 K. Grob and E. Muller, "Co-Solvent Effects for Preventing Broadening or Loss of Early Eluted Peaks when Using Concurrent Solvent Evaporation in Capillary GC. Part 1: Con­ cept of the Technique", J. High Resol. Chromatogr. Chromatogr. Commun. 11 (1988) 388. T. Suzuki, K. Yaguchi, K. Ohnishi, and T. Yamagishi, "Determination of Pesticides in Water by Capillary GC with Splitless Injection of Large Sample Volumes", J. Chromatogr. A 662 (1994) 139. U. Boderius, K. Grob, and M. Biedermann, "High Temperature Vaporizing Chambers for Large Volume GC Injections and On-Line LC-GC", J. High Resol. Chromatogr. 18 (1995) 573 . K. Grob and M. Biedermann, "Vaporising Systems for Large Volume Injection or On-Line Transfer into GC: Classification, Critical Remarks, and Suggestions (Review)", J. Chroma­ togr. 750 (1996) 11. K. Grob and A. Romann, "Sample Transfer in Splitless Injections in Capillary GC", J. Chromatogr. 214 (1981) 118. Ph.L. Wylie R.J. Phillips, K.J. Klein, M.a. Thompson, and B.W Hermann, "Using Elec­ tronic Pressure Programming to Reduce the Decomposition of Labile Compounds Dur­ ing Splitless Injection", J. High Resol. Chromatogr. 15 (1992) 763. P.L. Wylie and K. Uchiyama, "Improved GC Analysis of Organophosphorus Pesticides with Pulsed Splitless Injection", J. Assoc. Off. Anal. Chem. 79 (1996) 571. M. Vincenti, C. Minero, and M. Sega, "Optimized Splitless Injection of Hydroxylated PCBs by Pressure Pulse Programming", J. High Resol. Chromatogr. 18 (1995) 490. S. Klick, "Evaluation of Different Injection Techniques in the GC Determination of Ther­ molabile Trace Impurities in a Drug Substance", J. Chromatogr. A 689 (1995) 69. K. Grob, A. Artho, Ch. Frauenfelder, and I. Roth, "Charcoal Open Tubular Traps for the Analysis of Air and Headspace Samples", J. High Resol. Chromatogr. 13 (1990) 257. P. Van Ysacker, H.M. Snijders, H.-G. Janssen, and C.A. Cramers, "The Use of Non-Split­ ting Injection Techniques for Trace Analysis in Narrow Bore Capillary GC", J. High Resol. Chromatogr. 21 (1998) 491. J.D. Kokenovlch, T.F. Simonick, and V.W Watts, "High Speed Analysis of Underivatized Drugs by GC/MS", in: P. Sandra (Ed.}, Proc. 9th Int. Symp. on Capillary Chromatography, Monterey 1988, HGthig, Heidelberg (1988) 550. J.J. Langenfeld, S.B. Hawthorne, and D.J. Miller, "Optimizing Split/Splitless Injection Port Parameters for SPME", J. Chromatogr. 740 (1996) 139. N.H. Snow and P. Okeyo, "Initial Band Width Resulting from Splitless and SPME-GC In­ jections", Proc. 18th Int. Symp. Capillary Chromatography, Riva del Garda, 1996, P. Sandra and G. Devos (Eds.), HGthig, Heidelberg, 1996, 709. A. Kaufmann, "Maximum Transfer Conditions for Splitless Injection", J. High Resol. Chromatogr. 20 (1997) 193. D.R. Erney, A.M. Gillespie, D.M. Gilvydis, and C.F. Poole, "Explanation of the Matrix-In­ duced Chromatographic Response Enhancement of Organophosphorus Pesticides dur­

400

42

43

44

45

46 47 48 49

50 51 52 53

54 55 56

57 58 59

60

61

62 63 64

0 References ing Open Tubular Column GC with Splitless or Hot On-Column Injection and FPD", J. Chromatogr. 638 (1993) 57. H.-M. MOiler and H.-J. Stan, "Pesticide Residue Analysis in Food with CGC - Study of the Long-Term Stability by the Use of Different Injection Techniques", J. High Resol. Chromatogr. 13 (1990) 697. O.R. Erney and C.F. Poole, "A Study of Single Compound Additives to Minimize the Ma­ trix Induced Chromatographic Response Enhancement Observed in the GC of Pesticide Residues", J. High Resol. Chromatogr. 16 (1993) 501. J. Hajslova, K. Holsdove, V. Kocourek, J. Poustks, M Godule, P. Cuhrs, and M. Kempny, "Matrix-induced effects: a critical point in the GC analysis of pesticide residues", J. Chromatogr. 800 (1998) 283. J.L. Bernal, MJ. del Nozal, J.J. Jimenez, and J.M. Rivera, "Matrix Effect in the Determi­ nation of Acaricides and Fungicides in Must by GC with ECD and NPD", J. Chromatogr. 778 (1997) 111. DB. Erneyand T.M Pawlowski, "Matrix-Induced Peak Enhancement of Pesticides in GC: Is There a Solution?", J. High Resol. Chromatogr. 20 (1997) 375. K. Grab, and M Bossart, "Effect of Dirt on Ouantitative Analysis by Capillary GC with Splitless Injection", J. Chromatogr. 294 (1984) 65. A. Shatkay and S. Flavian, "Unrecognized Systematic Errors in Quantitative Analysis by GLC", Anal. Chern. 49 (1977) 2222. A. Shatkay, "Effect of Concentration on the Internal Standards Method in GLC", Anal. Chern. 50 (1978) 1423. K. Grob and H.P. Neukam, "Glass Wool in the Injector Insert for Quantitative Analysis in Splitless Injection", Chromatographia 18 (1984) 517. K. Grab, "Band Broadening Effects and Reconcentration Techniques Related to Injection on to Capillary GC Columns", Anal. Proc. (London) 19 (1982) 293. K. Grab, "On-Column Injection in Capillary GC", Huthiq, Heidelberg (1986,1991) 110. K. Grab and K. Grob, "Are we Using the Full Range of Film Thickness in Capillary GLC?", Chromatographia 10 (1977) 250. K. Grab, G. Grob, and Kurt Grab, "Comprehensive, Standardized Quality test for Glass Capillary Columns", J. Chromatogr. 156 (1978) 1. K.O. Bartle, in "GC, A Practical Approach", P.J. Baugh (Ed.), IRL Press, Oxford (1993) 10. J.C. Sternberg, "Extra Column Contributions to Chromatographic Band Broadening", in: J.C. Giddings and R.A. Keller (Eds), Advances in Chromatography, Vol. 2, Marcel Dekker, New York (1966) 205. M Oehme, "Hochaufl6sende Gaschromatographie", Huthlq, Heidelberg (1986). K. Grab, "Partial Solvent Trapping in Capillary GC. Description of a Solvent Effect", J. Chromatogr. 251 (1982) 235. K. Grob, "Solvent Trapping in Capillary GC. Two Step Chromatography", J. Chromatogr. 253 (1982) 17. V. Pretorius, C.S.G. Philips, and W Bertsch, "Solute Focusing in GLC, Using the Solvent Effect: A General Description", J. High Resol. Chromatogr. Chromatogr. Commun. 6 (1983) 232. K. Grab and B. Schilling, "Observation of a Peak under the Action of "Phase Soaking", a GC Solvent Effect. during Passage through a Capillary Column", J. Chromatogr. 259 (1983) 37. K. Grob, "On-Column Injection in Capillary GC", Huthiq (1987, 1991) 245. K. Grab, "Peak Broadening or Splitting Caused by Solvent Flooding after Splitless or On­ Column Injection in Capillary GC", J. Chromatogr. 219 (1981) 13. K. Grob, "Band Broadening in Space in Splitless Injection", J. Chromatogr. 324 (1985) 251.

D References

401

65 K. Grob, H.P. Neukam, and M.-L. Riekkols, "Length of the Flooded Zone in the Column Inlet and Evaluation of Different Retention Gaps for Capillary GC", J. High Resol. Chromatogr. Chromatogr. Commun. 7 (1984) 319. 66 K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg, (1987,1991) 118. 67 WL. Saxton, "Chromatographic Terminal Band Lengths", J. High Resol. Chromatogr. Chromatogr. Commun. 7 (1984) 117. 68 K. Grob and Z. Li, "Introduction of Water and Water-Containing Solvent Mixtures in Cap­ illary GC. I. Failure to Produce Water-Wettable Precolumns (Retention Gaps)", J. Chromatogr. 473 (1989) 381. 69 K. Grob and Z. Li, "Introduction of Water and Water-Containing Solvent Mixtures in Cap­ illary GC. II. Wettability of Precolumns by Mixtures of Organic Solvents and Water; Reten­ tion Gap Techniques", J. Chromatogr. 473 (1989) 391. 70 K. Grob and Z. Li, "Introduction of Water and Water-Containing Solvent Mixtures in Cap­ illary GC. III. Water-Resistant Deactivation of Uncoated Precolumns7", J. Chromatogr. 473 (1989) 401. 71 A. Pouwelse, D. de Jong, and J.H.M. van den Berg, "Use of Polar Coated Retention Gap for the Introduction of Large Volumes of Polar Solvents in On-Column Capillary GC", J. High Resol. Chromatogr. Chromatogr. Commun. 11 (1988) 607. 72 K. Grob and A. Artho, "Carbowax-Deactivated GC Precolumns Capable of Resisting Wa­ ter?", J. High Resol. Chromatogr. 14 (1991) 212. 73 K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg, (1987, 1991) 368. 74 K. Grob, "Sample Introduction in Capillary GC", Kemia-Kemi 11 (1984) 732. 75 K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg, (1987, 1991) 403. 76 R.F. Arrendale, J. T. Stewart, and R.M. Martin, "Effects of the Pre-Column in Automated On-Column Injection Capillary GC", J. Chromatogr. 518 (1990) 307. 77 G. R. van der Hoff, P. van Zoonen, and K. Grob, "Deactivation of Precolumns for Capillary GC by Thin Layer of OV-1701", J. High Resol. Chromatogr. 17 (1994) 37. 78 C.L. Wooley, K.E. Markides, M.L. Lee, and K. D. Bartle, "Deactivation of Small Diameter Fused Silica Capillary Columns with Organosilicon Hydrides", J. High Resol. Chromatogr. Chromatogr. Commun. 9 (1986) 506. 79 M. Hetem, G. Rutten, B. Vermeer, J. Rijks, L. van de Ven, J. de Haan, and C.A. Cramers, "Deactivation with Polymethylhydrosiloxane. A Comparative Study with Capillary GC and Solid State 29Si Nuclear Magnetic Resonance Spectroscopy", J. Chromatogr. 477 (1989) 3. 80 K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg (1987,1991) 159. 81 K. Grob, "A Current, Easily Reparable Form of Damage to GC Capillaries", J. High Resol. Chromatogr. Chromatogr. Commun. 1 (1978) 307. 82 K. Grob and B. Schilling, "Uncoated Capillary Column Inlets (Retention Gaps) in GC", J. Chromatogr. 391 (1987) 3. 83 R.F. Arrendale, J. T. Stewart, ME Mispagel, and B. Vitavirasuk, Comparison of Cold On­ Column and Splitless Injection Using an Automatic Liquid Sampler: Application to the Determination of GX-071 in Animal Ration", J. High Resol. Chromatogr. 12 (1989) 749. 84 K. Grob, "On-Column Injection in Capillary GC", Hlithig, Heidelberg (1987, 1991) 441. 85 E.R. Rohwer, V. Pretorius, and V. Pretorius, "Simple Press-Fit Connections for Flexible Fused Silica Tubing in GLC", J. High Resol. Chromatogr. Chromatogr. Commun. 9 (1986) 295. 86 J. Roeraade, "Cutting of Glass and Fused Silica Capillaries", J. High Resol. Chromatogr. Chromatogr. Commun. 6 (1983) 140. 87 K. Grob, M. Biedermann, K. Bernath, H.-P. Neukam, and M. Galli, "Call for Fused Silica Tubing Furnishing light Press-Fit Connections", J. High Resol. Chromatogr. 15 (1992) 613.

402 88

D References

C. Wesen and H. Mu, "Re-Use of Press-Fit Connectors and Splitters for GC Capillary Col­ umns", J. High Resol. Chromatogr. 15 (1992) 136. 89 M. Vecchi and W. Walther, "Simple and Versatile Method for Connecting Fused Silica and Glass Capillaries", J. High Resol. Chromatogr. Chromatogr. Commun. 11 (1988) 337. 90 A. Bemgard and C. Ostman, "High Temperature and High Pressure Stable Gluing of Press­ Fit Connectors for Fused Silica and Metal Capillary Tubing", J. High Resol. Chromatogr. 15 (1992) 131. 91 J. Clark and B.A. Jones, "Fused Quartz Couplings for Capillary Columns and Restrictors in SFC", J. High Resol. Chromatogr. 15 (1992) 341. 92 K. Grab, K. Grab, and G. Grab, "Organic Substances in Potable Water and in Its Precur­ sor. III, The Closed Loop Stripping Procedure Compared with the Rapid Liquid Extrac­ tion", J. Chromatogr. 106 (1975) 299. 93 K. Grab, "Broadening of Peaks Eluted Before the Solvent in Capillary GC. Part 1: The Role of Solvent Trapping", Chromatographia 17 (1983) 357. 94 K. Grab and B. Schilling, "Broadening of Peaks Eluted Before the Solvent in Capillary GC. Part 2: The Role of Phase Soaking", Chromatographia 17 (1983) 361. 95 Kurt Grob, "The Glass Capillary Column in GC.A Tool and a Technique", Chromatographia 8 (1975) 423. 96 H. Br6tell, N.-D. Ahnfelt, H. Ehrson, and S.Eksborg, "ECD GC with Splitless Injection on Isothermally Operated Wide Bore Glass Capillary Columns", J. Chromatogr. 176 (1979) 19. 97 P.H. Silvis, "Optimizing Injection into 0.53 mm i.d. Capillary Columns", LC-GC Int. 2 Nr. 9 (1989) 19. 98 M.F. Mehran, "Large Diameter Open Tubular Columns in GC Analysis", J. High Reso!. Chromatogr. Chromatogr. Commun. 9 (1986) 272. 99 H. Kern and B. Brander, "Precision of an Automated All-Glass Capillary GC System with an ECDfor Trace Analysis of Estrogens", J. High Resol. Chromatogr. Chromatogr. Commun. 2 (1979) 312. 100 W Jennings and M.F. Mehran, "Sample Introduction in GC",J. Chromatogr. Sci. 24 (1986) 34. 101 J.v. Hinshaw, Jr; "Modern Inlets for Capillary GC", J. Chromatogr. Sci. 25 (1987) 49. 102 C. Watanabe and K. Hashimoto, "Direct Injection of Large Sample Volumes into Capillary Columns with Packed Column Injector", J. High Resol. Chromatogr. 13 (1990) 610. 103 K. Grab and B. Schilling, "The Length of the Zone Flooded by the Injection of Large Volumes onto Retention Gaps in Capillary GC", J. High Resol. Chromatogr. Chromatogr. Commun. 7 (1984) 531. 104 A. Kaufmann, "Large Volume, Low Discrimination GC Injection into a Modified Splitless Injector", Chromatographia 46 (1997) 275. 105 E.C. Shagena and J. V. Hinshaw, "Quantitation on Wide-Bore, Open-Tubular GC Columns", in: P. Sandra (Ed.), 9th Int. Symp. Capillary Chromatography, Monterey 88, Hlithig, Heidelberg (1988) 182. 106 C. Mallet and V.N. Mallet, "Conversion of a Conventional Packed-Column GC to Accom­ modate Megabore Columns. 1. Evaluation of the System for Organophosphorus Pesti­ cides", J. Chromatogr. 481 (1989) 27. 107 P.M.J. van der Berg and Th. Cox, Chromatographia 5 (1972) 301. 108 J.M. Bayona, X. Aparicio, and J. Albaiges, "A Comparison of Vaporizing Injectors for Trace Analysis in Capillary GC", J. High Resol. Chromatogr. Chromatogr. Commun. 9 (1986) 59. 109 H. Jing and A. Amirav, "Pesticide Analysis with the Pulsed-Flame Photometer Detector and a Direct Sample Introduction Device", Anal. Chem. 69 (1997) 1426. 110 S.B. Wainhaus, N. Tzanani, S. Dagan, M.L. Miller, and A. Amirav, "Fast Analysis of Drugs in a Single Hair", J. Am. Soc. Mass Spectrom. 9 (1998),1311.

D References

403

111 G. Morchio, "Rapida determinazione GC dell'esano residuo negli oli di sansa grezzi, di sansa e di oliva rettificati", Riv. Ital. Sostanze Grasse 59 (1982) 335. 112 K. Grob, M. Biedermann, and A. M. Giuffre, "Determination of Organophosphorus Insec­ ticides in Edible Oils and Fats by Splitless Injection of the Oil into GC (Injector-Internal Headspace Analysis)", Z. Lebensm. Unters. Forsch. 198 (1994) 325. 113 C. Mariani and E. Fedeli, "Determinazione gascromatografica di BHA, BHT e Jonox 100", Riv. Ital. Sostanze Grasse 60 (1983) 667. 114 C. Mariani, S. Venturini, and E. Fedeli, "Sulla presenza di prodotti alogenati volatili negli oli vergini di oliva", Riv. Ital. Sostanze Grasse 67 (1990) 239. 115 G. Morchio, R. De Andreis, and G.R. Verga, "Indagine sui contenuto di composti fosforganici presenti negli oli vegetali e in particulare nell'olio di oliva", Riv.ltal. Sostanze Grasse 69 (1992) 147. 116 Ch. Droz, Official Food Control Authority of the Canton of St. Gallen, Switzerland, unpub­ lished results, 1998. 117 Ch. Watanabe, H. Tomita, K. Seto, Y. Maseda, and K. Hashimoto, "Accuracy and Repro­ ducibility in Splitless, and Packed and Open Tubular CoolOn-Column Injections", J. High Resol. Chromatogr. Chromatogr. Commun. 5 (1982) 630. 118 D.L. Springer, D.W Phelps, and RE Schirmer, "Obtaining Ouantitative Data for PNA's with Capillary Column GC, Using Split and Splitless Injection", J. High Resol. Chromatogr. Chromatogr. Commun. 4 (1981) 638. 119 D.H. McMahon, "A Collaborative Study to Evaluate Ouantitation Utilizing Different Injec­ tion Modes for Capillary GC", J. Chromatogr. Sci. 23 (1985) 137. 120 G. Schomburg, H. Husmann, and R. Rittmann, "Direct' (On-Column) Sampling into Glass Capillary Columns. Comparative Investigations on Split, Splitless and On-Column Sam­ pling", J. Chromatogr. 204 (1981) 85. 121 D. Kalman, "Optimized Injection for Determination of Free Phenols by GC Using Fused Silica Columns", J. High Resol. Chromatogr. Chromatogr. Commun. 6 (1983) 564. 122 F.I. Onuske, R.J. Kominar, and K. Terry, "An Evaluation of Splitless and On-Column Injec­ tion Techniques for the Determination of Priority Micropollutants", J. Chromatogr. Sci. 21 (1983) 512. 123 L.L. Plotczyk, "Application of Fused-Silica Capillary GC to the Analysis of Underivatized Drugs", J. Chromatogr. 240 (1982) 349. 124 H.-J. Stan and H. Goebel, "Evaluation of Automated Splitless and Manual On-Column Injection Techniques Using Capillary GC for Pesticide Residue Analysis", J. Chromatogr. 314 (1984) 413. 125 K. Grob, "Effect of Dirt Injected On-Column in Capillary GC; Analysis of the Sterol Frac­ tion of Oils as an Example", J. Chromatogr. 287 (1984) 1. 126 K. Grob, T. Liiubli, and B. Brechbiihler, "Splitless Injection - Development and State of the Art, Including a Comparison of Matrix ("Dirt") Effects in Conventional and PTVSplitless Injection", J. High Resol. Chromatogr. Chromatogr. Commun. 11 (1988) 462.

E Injector Design

405

E Injector Design

This section summarizes important aspects of the design of the vaporizing chamber and the surrounding system. It ad­ dresses: 1 users who want to know more about the background of injector design; 2 analysts thinking of buying a new instrument. When in­ struments are compared, the vaporizing injector should certainly be one of the first components to be evalu­ ated; and 3 people developing methods and instrumentation. The discussion is restricted to subjects of general interest. For manufacturer-specific aspects the reader is referred to the manuals. Same Injector for Split and Splitless Injection

Although the requirements are quite different, split and splitless injection have always been performed with the same injector. This is worthy of discussion, because many instruments are used either for split or for splitless injection almost ex­ clusively. Maybe specific injectors could be improved. On the other hand, there is a fairly continuous transition between the techniques: split injection at low split flow rates ap­ proaches splitless injection with regard to deposition of the sample and intermediate storage of sample vapor. As split injection is older, it was splitless injection that called for the modifications and compromises. It did this with lim­ ited success. Even today (2000), injectors with vaporizing chambers which are too small (2 mm i.d. liners) are mar­ keted, because the requirement that the sample vapor must be stored before it is transferred into the column has been ignored for far too long.

406

E 1. Vaporizing Chamber

1. Vaporizing Chamber The vaporizing chamber serves several purposes: 1 vaporization of the sample liquid, which can include stopping and holding of sample liquid injected as a band (Part B); 2 housing of the vapor up to its discharge into the col­ umn or through the split outlet; 3 mixing of the sample vapor with the carrier gas such that it is homogeneously distributed across the cham­ ber when it arrives at the split point; and 4 retention of non-evaporating sample by-products to pro­ tect the column from contamination.

1.1. Classical Teaching

Presently (year 2000) marketed injectors basically follow the concept laid down by Kurt Grab in 1978 [1]. The split injector of that time was modified to accomodate splitless injection according to the belief that a splitless injector per­ forms for split injection at least reasonably, but not neces­ sarily vice versa.

Housing the Sample Vapor

In split injection, the sample vapor is introduced into a gas stream (primarily split flow). Basically an unlimited amount can be injected and there are no limits to the capacity of the injector. When the rate of vapor formation substantially exceeds that of the discharge, however, the vapor must be stored

Gas is driven backwards into t supply line

4~====lt= Split outlet Column entrance

Gas cannot escape and prevents vapor reaching the bottom

Figure E1 Classical concept of the filling of the vapqrizing chamber by sample vapor.

-1.1. Classical Teaching

407

until it can be moved onwards in split or splitless mode. This is critical because the 1-2 III of liquid commonly injected produces a volume of vapor approaching (sometimes exceed­ ing) the volume available in modern vaporizing chambers. 1. 1. 1. Longitudinal Axis

At the beginning of sample evaporation, the chamber is filled with carrier gas. The sample vapor must displace the gas to liberate the room it needs for its expansion.

Filling from the Bottom

According to classical teaching, the vaporizing chamber is filled from the bottom towards the top (Figure E1). The center of sample evaporation is positioned slightly above the column entrance. The expanding vapor dis­ places the carrier gas backwards into a large volume in the gas supply system, compressing the gas. A cushion of gas prevents vapor from reaching the bottom of the chamber, which presupposes that the vapor does not expand into the split outlet line.

Center of Evaporation

From routine analysis with manual hot needle splitless in­ jection (thermospray) it was known that involatile sample material was transferred to the liner wall forming a ring ex­ tending from ca. 5 mm above (behind) the needle exit to ca. 15 mm below. This indicated that the center of evaporation was located 5-10 mm below the needle exit. Today we know that this is correct provided the sample liquid is nebulized at the needle exit. In fact, for solutions in volatile solvents thermospray was normal before the fast auto­ sampler was introduced.

Length of Syringe Needle

This concept determined the longitudinal axis of the injec­ tor. The syringe needle should release the sample 15-20 mm above the column entrance such that the vapor remains above the column entrance even when it expands somewhat further down. If the maximum convenient length of the sy­ ringe needle is 71 mm (3 inch), the column entrance should be positioned 85-90 mm below the septum (Figure E2).

Position of the Column Entrance

Recommendations on the column position were based on

two factors.

1 The column entrance should be clearly above the bot­

tom of the vaporizing chamber because septum parti­ cles and other materials are accumulated there (empty liner) and should be out of the way of the sample va­ pors. 2 For splitless injection and split injection at low split flow rates, the volume below the column entrance should be small since sample material driven there will hardly return to the column entrance: there is no gas flow to bring it upwards to the column entrance. It was concluded that the column should enter the vaporiz­ ing chamber by about 5 mm.

408

E 1. Vaporizing Chamber

Septum purge

Carrier gas supply ====:r~

liner, e.g. 80 x 4 mm i.d.

5mm :m#~==

Split outlet

Column

Figure E2 Geometry of the splitless injector commercialized by Carlo Erba in 1976.

Septum Purge Zone

The length of the injector head, including the septum and the septum purge zone, should be minimal because any space they occupy is at the expense of the useful length of the va­ porizing chamber. If the injector head is 10 mm long, the vaporizing chamber can be ca. 80 mm long.

Advantage of Longer Cham­ bers

Split injectors designed before the introduction of splitless injection were usually longer to provide more distance for vaporization and homogenization of the vapor. As men­ tioned several times in this book, however, there are no data substantiating the view that such elongation noticeably im­ proves the results.

1. 1.2. Internal Diameter for Split/ess Injection

The optimum internal diameter of the vaporizing chamber depends on whether or not vapor must be stored. If 2 IJ.L of sample solution is evaporated in 0.5 s, vapor volumes be­ tween some 0.2 mL (e.g. hexane, 150 kPa inlet pressure) and 1.5 mL (methanol, low pressure) are generated at a rate of 24-180 mLJmin. Intermediate storage is required if split flow rates are substantially lower.

Intermediate Storage of Vapor

It would be desirable to have the capacity to house the vapor from at least 2 III of sample (including the volume from the needle when performing thermospray injection). As shown in Section 03, with a 4 mm i.d. chamber of 1 mL internal volume, losses from 2 IJ.L injections of a solution in hexane were significant, but probably tolerable. For a 2 IJ.L injection of dichloromethane or methanol, the chamber is usually far too small, wich is an unsatisfactory situation.

-1.1. Classical Teaching

409

Maximum Internal Diameter

The arguments against wider liners were: 1 The gas velocity in the liner becomes too low for sat­ isfactory transfer into the column. 2 The wider the liner, the more the vapor expands like a smoke trail instead of a plug. This increases dilution with carrier gas (enlarging the cloud). The extra volume is poorly used. For this reason, elongation increases stor­ age capacity more effectively than does widening. Introduction of flow or pressure programming, which enable increase of the flow rate for the transfer and pressure to com­ press the vapor cloud, improved somewhat on this.

Liners of Narrower Bore

2 mm i.d. liners are preferable when carrier gas flow rates are low, such as when narrow bore columns are used. For splitless injection, the internal volume is sufficient only if inlet pressures are high: at 200 kPa, a vapor cloud at am­ bient pressure comprising a volume of 0.75 ml can be stored. This is the lower limit for 2 III of hexane and still not enough for 1 III methanol.

1. 1.3. Internal Diameter for Split Injection

If the split flow rate is higher than the rate of sample evapo­ ration (around 50 rnt/mtn). storage of vapor is no longer necessary and other arguments become important.

Deviation of the Split Ratio

As shown in Section C8.4.1, a large vaporizing chamber is preferable, because it minimizes the pressure wave and the deviation of the true split ratio from that pre-set. Recon­ densation of the sample matrix (solvent) in the column inlet has, furthermore, less effect, because of stronger dilution of the vapor with carrier gas. These effects favor the use of 4 mm i.d, liners.

Homogeneity of the Vapor

On the other hand, distribution of the vapor across a na~ row bore liner (2 mm) is faster, which should improve the reproducibility of the splitting process. Use of a wider bore (4 mm) liner with obstacles could be an alternative.

Sharp Initial Bands

In fast GC, particularly when performed isothermally at or near the temperature during the injection (no cold trapping), special attention is required to obtain initial bands of suffi­ cient sharpness. Van Lieshout et al. (2) studied the optimum liner diameter for methane as test component. With a 5 mm i.d. liner and a split flow rate of 100 mt/rnln, the peak width at half height was a totally unacceptable 6 s, With a 1.25 mm i.d. liner, it was reduced to 0.55 s. Reduction to 0.75 mm did not further sharpen the peak, indicating that the ini­ tial band no longer contributed noticeably.

Undiluted Samples

Samples of intermediate to high-boiling point (undiluted or in high-boiling solvents) evaporate slowly and there is, there­ fore, no large vapor cloud to be stored nor a pressure wave to keep low. Transfer of the liquid to the wall of an empty liner is, furthermore, safer when the diameter is small.

410

E 1. Vaporizing Chamber

1. 1.4. Conclusions

The outer diameter of the injector liner is given by the size of the cavity in the injector body, but the internal diameter can be varied by selecting liners of different wall thickness. Lin­ ers of 2 and 4 mm internal diameter should be made available, and methods should specify which liner to prefer.

Liners with a Narrow Bore Bottom Section

The length of the vaporizing chamber can be varied by using liners with a restriction at the bottom for splitless injection and split injection at low split ratios. This restriction (Figure E3) reduces the effective length of the chamber to the needs of the latter injection method, whereas open tubes can be used for split injection at higher flow rates. The liner at the left is designed such that the long svrlnqe needle reaches to within 15-20 mm of the column. That at the right is suitable for injection involving split flow rates exceeding ca. 50 mLJ min.

r E E

:g

E E o

'"

1

Liner for injection at low split flow rates and splitless injection

Liner for injection at high split flow rates

Figure E3 Liners suitable for injectors longer than 80 mm. The restric­ tion at the bottom shortens the effective length of the va­ porizing chamber to the needs of splitless injection and split injection at low split flow rate (dimensions of the CE Instru­ ment.nbermoQuest instrument).

1.1.5. Column Installa­ tion

As shown in Part B, the position of the column entrance in the vaporizing chamber can severely affect the results. Because there is no stop pin aiding correct positioning of the column entrance, the analyst must insert it to the proper height by monitoring the length of the inlet introduced from the oven.

Correct Height

In a first step, the appropriate height ofthe column entrance in the injector must be determined. Figure E4 shows the two most common positions. The classical installation, intended for thermospray injection, introduces the inlet into the vaporizing cham­ ber by about 5 mm (see above). If the liner is packed or contains obstacles, septum material and other particles accumulate above the col­

1.1. Classical Teaching

411

umn entrance and remain between the sample and the column. Even then insertion into the chamber by a few millimeters is appropriate to avoid contact of the sam­ ple vapor entering the column with the metallic surfaces of the injector body. With a goose neck liner, there is the dilemma of whether to keep the column entrance above the accu­ mulated garbage, i.e. inserting it ca. 5 mm above the constriction, or to make sure no sample vapor is formed below the column entrance, which requires location of the column entrance within the constriction, as shown in the figure.

Classical style; "Goose neck" liner,

column entrance column below

above the "garbage" bottom 01the liner

Piece 01 wire or column

Figure E4

Column positions in the vaporizing chamber and measure­

ment of the length of the column inlet to be introduced"

Length of Inserted Inlet

The length of the column inlet to be inserted into the injector must be determined for each type of instrument, also taking into account the fitting of the column attachment used. Usu­ ally the instrument manual gives some instructions, al­ though not always with an indication of the position reached when the column inlet is introduced by the given length.

Measurement

The correct length of the inlet section to be introduced can be measured, preferably after cooling ofthe injector. The septum cap, the septum purge device, and the liner are re­ moved and a glass rod or a pencil introduced to stop a piece of fused silica capillary or wire introduced from the bottom. The ferrule and the screw of the column attachment must be installed and tightened such that they are in the same posi­ tion as during column installation. The point at which the capillary or wire enters the attachment is then marked. The length of the introduced piece is measured, from which the length of column inlet to be introduced is calculated.

Marking the Column Inlet

To install the column correctly, the length of the inserted in­ let is marked, i.e. the point which should finally be located

412

E 1. Vaporizing Chamber adjacent to the nut of the column fitting. This mark can be made with a felt-tip, but the white fluid for typewriter correc­ tions (such as "lipp-Ex") is more easily visible.

ASTM

Column installation is described by an ASTM Standard Practice [31 which includes subjects such as carrier gas choice, deactivation of liners, choice of ferrule, setting the flow rate, column conditioning, column testing, and use of precolumns. It does not comment on the height ofthe column entrance in the chamber, however.

1.2. Newer Develop­ ments

As shown above, the classical concept of injector design left little room for variation: length and internal diameter were restricted within narrow limits. For splitless injection it com­ pelled us to accept a compromise between storage ca­ pacity and transfer performance which never really sat­ isfied. In the meantime, the concept received two new in­ puts, both of which influenced the premises and would have enabled fundamental changes.

1.2.1. Pressure and Flow Programming

Electronic flow and pressure regulation introduced the pos­ sibility of increasing the inlet pressure/column flow rate dur­ ing the splitJess period. Pressure increase compresses the vapor cloud and, hence, increases the injector capacity (although not spectacularly). The increased flow rate into the column as a result of the pressure increase enables es­ cape from the deadlock that a reasonable injector capacity (internal liner diameter) could not be combined with a low column flow rate, as, e.g., required by mass spectrometers with weak vacuum pumps. With pressure/flow programming, the gas velocity in the vaporizing chamber can easily be increased to a level suffi­ cient for effective sample transfer with a liner widened to 5-6 mm i.d. Widening from 4 to 5 mm i.d, enlarges the in­ ternal volume by 56 %. As pressure increase further enhances the capacity, a gain of at least a factor of two is achieved.

1.2.2. Fast Autosampler

In 1985, Hewlett-Packard (Agilent) introduced the fast autosampler with the intention of suppressing sample evapo­ ration inside the needle. It went unnoticed for a long time that this also had a substantial effect on sample evaporation inside the injector.

Transport of Liquid

Resulting band formation fundamentally changes the above logic of injector design. It is no longer true that sample evapo­ ration occurs ca. 10 mm below the needle exit. As shown by the videos, solutions in volatile solvents can be shotthrough hot tubes for distances longer than 20 cm without significant evaporation and without risk of the liquid touching the liner wall. Hence it is no longer true that the length of the vaporizing chamber is restricted by the length of the syringe needle. '

1.2. Newer Developments

413

As shown in Part B, injection with band formation requires means of stopping the sample liquid above the column entrance, the choice being a plug of packing or an obstacle trapping the liquid. The position of this stopper determines the center of evaporation. It can be located far from the nee­ dle exit, which, basically, would have enabled the construc­ tion of a far longer chamber and its use with a short syringe needle as early as 1985 (Figure E5). Injection with thermospray

Center of evaporation ';;i:i! near needle exit

IIJ

Injection with band formation

Sample liquid moving as a band to the obstacle, determining the evaporation site

Packing or obstacle

Figure E5

Injection with band formation enables the construction of a

long vaporizing chamber that can be used with a short sy­

ringe needle.

Neglected Opportunity

Elongation ofthe vaporizing chamber increases the capac­ ity for temporary storage of the vapor more efficiently than widening: there is less mixing with carrier gas and the gas velocity during the splitless period is higher. This would have been particularly important for the injectors from Hewlett­ Packard because their vaporizing chamber is fully open at the top (carrier gas flow redirected into the septum purge line).

1.3. Room for Improve­ ment?

Every few years, manufacturers present new GC instruments. A closer look reveals, however, that many of the vaporizing injectors are still the same as those produced 15-20 years ago - as if there were no room for improvement. Color, outside architecture, and user interfaces change, but not those parts which ultimately determine the quality of the re­ sults. A vaporizing chamber designed according to today's knowl­ edge would look different from what we currently see. Be­ low some options are discussed, starting from the require­ ments of splitless injection since the corresponding injec­ tors are also suitable for split injection.

414

E 1. Vaporizing Chamber

1.3. 1. Preference for Thermospray or Band Formation?

An update of the injector design requires a decision to be taken on whether to go for thermospray or band formation (Figure E6). It will probably be optimized for one of the two options, and not give a choice between the two. Both op­ tions offer important advantages.

Injection into hot vaporizing chamber

,;

'\

Cool needle Fast autosampler or short needle

Hot needle Manual injection and analogous autosamplers

Band ofJ/ liquid '" Mechanical I spray upon , deformed Band of liquid needle tip must be stopped ?

Thermospray

J/ Light packing

....... Trapping in obstacles

•+

Fine droplets slowed in gas

J/ in Vaporization gas phase

'" Transfer of particles to liner wall

Figure E6 The two strategies for injection into a hot vaporizing injec­ tor. (From ref. [4].1

1.3.2. Optimized Thermospray

Thermospray provides more gentle evaporation. It is less

affected by increasing injector contamination during a se­

ries of analyses, but aerosol formation can cause problems.

Optimization of the injector might involve the following ele­

ments:

Thermospray presupposes some evaporation inside

1 the needle. Negative side effects of this (excessively large sample volume, discrimination) can be minimized by reducing the volume of the spraying device (e.g. the needle). A narrow bore orifice barely 1 cm long is suffi­ cient if the heat supply is adequate. This orifice can be part of the syringe or of the injector. 2 The vaporizing chamber should be ca. 150 mm long. As the needle (sprayer) should be shorter, it can no longer be introduced from the end opposite to the column in­ let, but from the same side as the column, filling up the chamber from the column entrance towards the gas supply. No such injector has yet been constructed, although several

rather detailed projects have been designed by some instru­

ment manufacturers. One of these designs resembles that

shown at the left in Figure E7.

1.3.3. Optimized Injec­ tion with Band Formation

The alternative approach involving band formation and

evaporation from a packing or a trap between obstacles could

be optimized in the following manner.

Evaporation in the needle is suppressed. Hewlett­

1 Packard/Agilent did this by combining fast injection with a cool injector head, which excludes the possibility of manual injection. Alternatively the length ofthe syringe

1.3. Room for Improvement

415

Syringe

Gas supply

Gas

supply Band of liquid shot to bottom

Injector liner Expanding nebulized sample

Heating block

Heating block

Packing stopping the sample liquid

Syringe

II

II

Thermospray

Band formation

Column

Split outlet

needle is reduced to maybe 15 mrn, such that evaporaFigure E7

Possibilities of designing vaporizing injectors with larger internal volumes based on the prin­

ciples of thermospray or band formation. The chamber can be as long as 20 cm. (From ref.

[4].1

2

tion in the needle is avoided even for manual injection and autosamplers working at normal speed. The injec­ tor head is again kept at a low temperature (Figure E7). The bottom of the liner is equipped with an obstacle stopping the liquid, a trap or a volume of well deacti­ vated wool which is sufficient to pick up the amount of sample liquid injected, which could be as much as 10

I!L. 3

4

The injector cavity houses liners 15-20 cm long and with internal diameters up to 5 mm. Different wall thick­ nesses result in internal diameters of 1.5, 2.5, 4, or 5 mm. Primarily the vapor of solvent and some highly volatile components expand towards the top of the chamber. The more difficult, high-boiling solutes remain in the region of the center of evaporation, hence close to the column entrance. This enables heating of the base of the injector only; the top must be warmed just to the temperature which ensures that the sample liquid is re­ pelled from the liner wall.

416

E 2. Surroundings of the Vaporizing Chamber

2. Surroundings of the Vaporizing Chamber 2.1. Seal between Liner and Injector Body?

There have been long disputes about whether or not injec­ tor liners should be sealed against the injector body. The two options are shown in Figure Ea. 1 In the injector on the left, the liner fits tightly into the metal body, but there is no seal. The carrier gas supply and the split outlet are far apart, strongly favoring flow through the vaporizing chamber over that around it (in­ struments from Carlo Erba/Fisons up to ca. 1990). 2 In the design on the right (probably all injectors mar­ keted today), there is a seal between the liner and the injector body. To eliminate dead volumes, the carrier gas supply is positioned just above this seal, the split outlet just below. -""=== Septum purge Carrier--,,=as~-===u outlet supply Split outlet Seal

Split outlet Column

Liner tightly fitting cavity

Column

Seal between liner and injector body

Figure E8 Injector designs without and with ferrule between the liner and the injector body.

Arguments Against a Seal

The arguments against the use of a seal are: a seal complicates replacement ofthe liner (which should be easy because frequent exchange may be necessary); there is a risk of the fitting releasing material into the incoming carrier gas, generating "ghost" peaks; If the seal is not tight, the carrier gas is likely to take the wrong route: it flows from the supply directly into the split outlet. Such leaks easily remain unnoticed.

Arguments in Favor of a Seal

The situation is different if the liner is densely packed with a long bed of packing material. A worst case scenario: at a split flow rate of 50 mLJmin, the pressure drop over a 50 x

2.1. Seal between Liner and Injector Body

417

4 mm bed consisting of a material ofthe type used in packed column GC is ofthe orderof3 kPa and increases to ca. 10 kPa for a split flow rate of 150 mljmin.

Flow Around Vaporizing Chamber

Such a pressure drop across a packing in the liner can cause most of the carrier gas to flow through the gap between the injector body and the liner (Figure E9). If the regular flow rate through the vaporizing chamber drops to almost zero, a side stream enters it from the bottom and proceeds upwards into the column entrance, i.e. largely around the sample. The split ratio is severely affected. If an injector is to be suitable for the use with a densely packed liner, the latter must be sealed against the injector body. Hardly anybody uses such packings, however.

.....

Carrier gas supply .:::....-----'IE==::;_

Gas flow around

vaporizing chamber

Densely packed vaporizing chamber hindering gas flow

Gas entering the column from bottom

•• Split flow

11~==.......-1

Column

Figure E9 Densely packed liner with no seal to the injector body. The carrier gas flows (partly) around the vaporizing chamber and, in extreme circumstances, enters the column (mostly) from the bottom.

Rinsing the Space Between Liner and Injector Body

If there is no seal, the carrier gas is supplied at the top of the chamber to provide an easy route into the liner and maxi­ mum resistance against flow around it. For the same rea­ son, the split flow leaves through the bottom. The space between the liner and injector body is left without controlled rinsing. It is accessible to vapor, the solvent vapor being of primary concern, from the top and bottom, i.e. vapor overfilling the chamber (top) and leaving towards the split outlet.

Distorted Solvent Peak

Vapor slowly leaving this space upwards into the car­ rier gas supply reaches the column after a delay and gen­ erates a distorted solvent peak. Return downwards into the bottom of the injector has no adverse effect, because the carrier gas flow purges it into the split outlet before it can diffuse upwards to reach the column (column inlet entering the vaporizing chamber).

418

E 2. Surroundings of the Vaporizing Chamber

The typical pattern of solvent peak distortion sometimes observed for injectors without a seal is shown in Figure E10. The mechanism might be that the vapor enters the space between the liner and the injector body from the bottom. Some vapor diffuses upwards and enters the carrier gas stream at the top of the liner. The gas carries it into the va­ porizing chamber and to the column. Because diffusion to­ wards the top takes some time, this vapor forms a separate peak (instead of just a tailing solvent peak).

Injection

Solvent entering the column with delay

~ Figure E10 Distortion of the solvent peak as a result of vapor returning from the space betw_n injector liner and metal body. The solvent eluted after a delay can form a separate peak (as shown), a peak fused to the main solvent peak (broadening of the solvent peak), or a flat, very broad tail.

Material Used for the Seal

The seal must be manufactured from of a material which fulfils several requirements. 1 It must not bleed, because the material released from the upper region contaminates the carrier gas flowing into the column and produces "ghost peaks". 2 It must be thermostable (without bleed) up to the in­ jector temperatures commonly used, i.e. 350°C. 3 It must not allow diffusion of solvent from the split outlet into the carrier gas supply through the ferrule. 4 The ferrule must not stick to surfaces inside the in­ jector, because it would then be difficult to remove, it. Sticking can be avoided by housing the ferrule in a metal case, exposing only those parts of the ferrule which need to form a seal.

Graphite

The material best fulfilling these requirements is graphite. It is thermostable without bleed, does not stick to surfaces, is inexpensive and can be used many times when handled with sufficient care. It is preferable to embed the graphite in a metal case to prevent breakage and deformation during tightening.

Diffusion of Solvent

Graphite forms an excellent seal against gas flow, but is a poor barrier against diffusion. When the solvent leaves

2.1. Seal between Liner and Injector Body

419

through the split outlet, some vapor penetrates the graphite. This can pass through the ferrule into the carrier gas on the other side, which brings it back to the column. A minute amount of solvent passing through the ferrule may cause tailing of the solvent peak such that the analysis of early eluted components is disturbed (depending on the detector). Release from the graphite may last more than 10 min after passage of the solvent vapor through the split outlet. Diffusion through graphite can be reduced by application of high pressure, i.e. securing the seal tightly. It does not com­ pletely eliminate it, however. Tailing Solvent Peak

Figure E11 shows solvent (hexane) peaks obtained by split injection (10 ml.lmin split flow rate) with loose and strongly compressed graphite ferrules. At intermediate attenuation, the FID showed tailing last­ ing for at least 5 min (top chromatogram). 2 For diagnostic purposes, the split flow rate was increased to 100 ml.lmin 30 s after injection and reduced again 5 s later (center). This eliminated the vapor from open vol­ umes. The fact that the solvent residues (tailing) returned shortly afterwards is proof of a deposit slowly releasing hexane vapor, such as an adsorbing material. 3 The ferrule was then compressed, firmly securing the screw above it. This substantially improved the shape of the solvent peak (bottom), but did not fully eliminate the problem: for several minutes the baseline remained higher than before injection. The solvent would be more disturbing with a more sensitive detector, such as MS.

Loose ferrule 10 mUmin split flow rale

"'lOUlIhout Solvent vapor

diluted in 10 times

t~8 :::::~"Q

the

ferrule centaminates

incoming carrier gas

~

Loose ferrule fOmUmin after 30 S, 100 mUm;n for 5 s then baell:to 10 mUmin

Purge at 100 mUmin

Compressed ferrule 10 mUminthroughout

Time (min)

Figure E11

Solvent (hexane) peaks with a liner sealed against the injec­

tor body by means of a graphite ferrUle. "1 Ill" injection;

250°C injector temperature; 90 °C column temperature.

420

E 2. Surroundings of the Vaporizing Chamber

Viton at Moderate Tempera­

tures

Modified PTFE materials, such as Viton or Kalrez, or polyim­ ide and polyimide/graphite (Vespel) are virtually tight against diffusion, but they are of limited thermostability (250-280 °Cl, tend to bleed, and also to stick to surfaces. Because there seems to be no generally suitable material, Viton is recommended for work at low temperatures, par­ ticularly when using GC-MS for the analysis of low molecu­ lar weight components and the ions of the solvent interfere. Graphite must be used for high injector temperatures. These are usually used for the analysis of high molecular weight material, when tailing solvent is less disturbing. Maybe the use of deactivated steel liners will increase in the future, which enables sealing with metal ferrules.

2.2. Accessible Volumes around the Vaporizing Chamber

We like to think of the injector as a closed chamber waiting to be filled with sample vapor, similar to the loop in HPLC injection valves. This is wrong: 1 the chamber is already full of carrier gas and there is no "waste" outlet to release the carrier gas as in HPLC injectors; 2 the vaporizing chamber is connected to the gas supply and two outlets with large volumes, and there is no valve to prevent the vapor from leaving the cham­ ber and entering these systems. Instead of expanding in the liner, the vapor might well expand into the split outlet line.

Pressure Increase

Injection into a vaporizing chamber creates a pressure in­ crease, because sample vapor is added to a chamber full of gas. If the vapor expands to a volume corresponding to half of that of a hypothetically closed injector, absolute pres­ sure increases by 50 %. If, for instance, the inlet pressure is 100 kPa (200 kPa absolute pressure), pressure should increase to 200 kPa (300 kPa absolute). The pressure increase observed on the manometer or pres­ sure sensor is much less because the chamber is not closed. In a system with pressure regulation and needle valve, the needle of the manometer jumps up by hardly 10 kPa. The expanding vapor drives gas and possibly vapor into the accessible volumes around the vaporizing cham­ ber. If the total accessible volume is 10 times larger than that of the vaporizing chamber (usually it is even more than that), pressure increases 10 times less, i.e. by 5 % or the 10 kPa observed in reality. This means that most of the extra volume added by the sample flows out of the vaporizing chamber. It is important to ensure that only carrier gas is leaving.

Boundaries of the Vaporizing Chamber

Upstream of the vaporizing chamber, the system is open up to the gas regulation device. Mechanical pressure regula­ tors close when the pressure at the outlet exceeds that regu­ lated; flow regulators continue feeding gas and prevent flow

2.2. Accessible Volumes around the Vaporizing Chamber

421

backwards. Electronic devices (proportional valves) do the same as soon as the pressure sensor has noticed the pres­ sure increase. Downstream, flow of gas and vapor is stopped at the closure of the split outlet. Direction of Expansion

If the resistance to flow is neglected, the flow from the va­ porizing chamber is directed primarily into the largest of the accessible volumes since pressure is leveled out over the whole accessible gas volume. Instrument design must con­ sider this and direct the expansion of the sample vapors by appropriate volumes in the gas supply and the split outlet. The center of sample evaporation must be positioned ac­ cordingly.

Vapor Cloud in the Center?

It seems convincing that the two volumes upstream and downstream of the vaporizing chamber (gas supply plus septum purge, split outlet) should be equal and that the sam­ ple should evaporate in the center of the chamber. Then the vapor expands equally in both directions. This might be so if the resistances against gas flow were really identical or negligible, but this is not realistic: the vapor would first ex­ pand in the direction of lower resistance, maybe leave the liner on that side, and shortly afterwards flow backwards in the other direction. In the end, the bulk of the vapor would be situated in the liner, but higher-boiling components would be lost on surfaces outside the chamber.

Classical Teaching

The classical concept defines the direction of expansion as backwards towards the gas supply system, and the center of evaporation as near the column entrance (short dis­ tance for high-boiling components). The carrier gas supply line was of a large bore and included a mechanical manom­ eter of large internal volume with a rather large T-piece (Fig­ ure E12).

Pressure regulator

Vapor expanding upwards

Center of sample evaporation

4...!====~1:-~P~i1 ;>:~t~~t

Column

Figure E12 Classical concept: expansion of the sample vapor towards the large volumes in the gas supply system.

422

E 2. Surroundings of the Vaporizing Chamber

Small Volume in Split Outlet

No Filters

When filling the chamber from the bottom (column entrance) upwards, flow into the split outlet must be avoided. The vol­ umes in the region of the column attachment and the split outlet to the closing valve must be small. In fact, the split outlet of the Carlo Erba instruments comprised 20 em x 0.5 mm i.d. tubing with an internal volume of ca. 40 ul., The volumes in the connections to the needle valve and a possible device for automatic closure were small. Nowadays volumes in the split outlet tend to be much

larger. 1

2

3

The outlet passes through the volume between the liner and the injector body up to the ferrule and the outlet line (several hundred microliter). The line tends to be long because closure occurs in a module positioned as far as possible from the hot oven (typically somewhere at the rear of the instrument), It contains a filter, a tube packed with charcoal, easily of 2 mL volume, to prevent contamination of the electric valves by material injected in split mode.

Solvent Recondensation in the Outlet

Another aspect is indirectly related to the pressure increase and may have a similar effect: if solvent vapor is adsorbed in the charcoal filter or recondensed on the walls of a cool split line, it sucks further vapor into the split outlet (as dis­ cussed for solvent recondensation in the column inlet). The cool parts must be far enough from the injector that they cannot be reached by the vapor.

Conclusions

1

2

3

2.2. f. Reversed Split Flow?

So far, no alternative to the classical concept has been proposed. The accessible volume in the split out­ let must be small to enable a well defined expansion of the sample vapor from a site close to the column inlet. Electronic regulators in the split outlet must be protected from sample material, and the filter is, therefore, an es­ sential part of the system. To keep the accessible vol­ ume small, the closing valve must be positioned near the vaporizing chamber, upstream of the filter. Deposition of the sample liquid on to a packing (injec­ tion with band formation) renders problems related to vapor expansion less critical. Higher-boiling solutes stay on the packing until solvent evaporation is complete and the flow of expanding vapor comes to a standstill (ef­ fect also used for large volume splitless injection by over­ flow). Volatile solutes expand and flow with the solvent vapor, however.

Kaufmann [5) proposed preventing flow of sample vapor into the split outlet by use of a gas flow backwards from the split outlet into the vaporizing chamber. Using a 2 mm i.d. liner with a constriction at the bottom, he observed that even an extremely small flow rate (0.25 mLJmin) eliminated the broad tail of the solvent peak.

2.2. Accessible Volumes around the Vaporizing Chamber

423

Purge Inwards

He installed a flow regulator delivering a small flow of car­ rier gas into the line between the injector and the solenoid valve of the Hewlett-Packard system. At a flow rate of 0.25 ml/min, the tail of the solvent peak was substantially reduced. Higher flow rates broadened the solvent peak because this purge flow counteracts the gas flow rate transferring the sam­ ple vapor from the vaporizing chamber into the column. The purge flow rate must, in fact, be small compared with the column flow rate. After the split outlet is opened, the purge flow is vented through the split line.

Stopping a Pulse?

A purge flow of 0.25 ml/min cannot prevent a pressure pulse driving vapor into the compressible volumes of the split out­ let (the vapor is formed at a rate 160-1000 times higher). It can only return lost material into the vaporizing chamber. Ifthe accessible volume in the split outlet was 2.5 mL, it takes a long time to purge it at 0.25 ml/min. It remains question­ able whether higher-boiling and somewhat adsorptive com­ ponents will also return to the column.

2.3. Septum

The septum has been one of the most widely discussed sub­ jects almost since the invention of GC, septum bleed being the predominant problem in early times (e.g. [6].) Numerous inventions have been made to replace the septum by a better means of closing the injector, but hardly any has been commercially successful. This prompted pro­ ducers to improve classical septa with regard to bleed, thermostability, puncturability, particle formation, and adhe­ sion to metal surfaces. An excellent summary is given by Restel< [7]. The thermostability of septa was discussed in Sec­ tion A8.3.

2.3.1. Required Tightness

The septum seals the injector chamber against ambient at­ mosphere, preventing escape of carrier gas and diffusion of air into the injector (possibly against a stream of escaping gas). How tight does it need to be?

Loss of Sample Vapor?

Some analysts fear that a leak through the septum causes sample material to be lost together with the escaping gas. They might overlook that such loss can occur only if the vaporizing chamber is overfilled, i.e. under conditions which are anyway unacceptable. With the Hewlett-Packard system, during splitless transfer the split flow rate is directed through the septum purge line and the vapor does not even reach the septum leak, because it has already been removed through the septum purge. This also removes air diffusing inwards through a leak.

Gas Regulation System

The effect on the analysis of a leak in the septum primarily depends on the gas regulation system. A leak can affect split injection, but not splitless injection. The mechanically pressure-regulated system is ro­ bust: it delivers an extra flow of carrier gas correspond-

UNIVERSIDAD DE ANTIOQUlA BffiUOTECA CENTRAL

424

E 2. Surroundings of the Vaporizing Chamber

ing to that leaving through the leak; the column and split flow rates remain the same, hence also the split ratio. This is no longer true for electronic pressure regula­ tion. As the proportional valve in the outlet regulating the split flow is guided by a flow sensor in the gas sup­ ply line (Section E4.3.2), the split flow rate will be too low by the extent of the leak flow through the septum. Excessively large peaks are obtained. Flow/backpressure-regulated systems adjust too Iowa split flow rate whether based on mechanical or electronic devices: the flow rate through the split outlet is lower by the amount leaking through the septum. Peaks turn out to be excessively large.

2.3.2. Septum Bleed

Packed column gas chromatographs and corresponding in­ struments adapted for capillary GC were highly sensitive to septum bleed because they had no septum purge installed.

Septum Purge

Material evaporating from the lower face of the septum dif­ fused into the gas stream directed to the vaporizing cham­ ber and the column. With an active septum purge (Section E4.5), this problem is eliminated: the small purge gas flow constantly removes the bleed material.

Septum Particles

There remains, however, the problem of septum particles cut or ruptured from the septum and carried into the va­ porizing chamber by the syringe needle. Material bleed­ ing from these particles is not removed through the septum purge. Their effect on the chromatograms depends on nu­ merous factors, including the selectivity ofthe detector used. In split injection, the material is split in the same way as the sample components while almost the full amount enters the column during splitless injection (unless the needle is intro­ duced before the split outlet is closed, as recommended in Section D2.5).

Influence of Column Position

Packings or obstacles in the liner positioned above the column entrance stop the septum particles. Bleed material evaporates into the gas stream directed to the column. With an empty liner and a column entering it by some 5 mm (classical concept, Figure E4), the particles drop past the column entrance and bleed material will hardly show up in the chromatogram. In split injection, the gas flow immediately carries the material out of the injec­ tor. In splitless injection, the vapor is unlikely to be swept upwards to the column entrance and is also discharged at the end of the splitJess period. Septum particles occasionally cling to the liner wall above the column entrance. Evaporated material will then reach the column, causing septum bleed to show up occasionally.

-2.3. Septum

425

Poor Reproducibility

Septum bleed originating from particles dropping into the vaporizing chamber typically produces "ghost peaks" of widely varying size. There might be 10 injections without a fresh septum particle being introduced and the "ghost peaks" are absent, whereas the next produces a forest of "ghost peaks". Hence a rapid test involving a single injection of sol­ vent easily leads to a false conclusion.

Testing

The fingerprint of the septum material should be known to recognize it when a chromatogram is severely disturbed by it. Because even a large number of solvent injections might not produce the peaks, a test with a small piece of septum is more effective. The piece is cut from a septum of the type used and inserted into the vaporizing chamber. It should be handled with tweezers to avoid a "fingerprint" chromato­ gram in the proper sense -large amounts of squalene, cho­ lesterol, branched wax esters and fatty acids are transferred by a single touch [81. A liner must contain some wool or an obstacle to support the piece of septum. The piece is dropped into the chamber, the chamber closed and the gas switched on again. After some 10 s with an elevated split flow, purging material from the surfaces, the split outlet is closed for splitless transfer with the column temperature low enough to achieve cold trapping. The split outlet is then opened widely and the oven temperature programmed. The analysis can be repeated with a longer splitless period (maybe 10 min) to compensate for the fact that thermal desorption of the bleed material might already be fairly complete.

Bleed Material

As shown by chromatograms in the Restek Guide, the com­ position of septum bleed can vary widely. It may contain low molecular weight siloxane monomers and oligomers, plasti­ cizers, such as phthalate esters, mineral oil hydrocarbons used on tools for cutting the septa, or material transferred from the fingers during handling and mounting.

2.3.3. Effect of Particles

After some time, septum particles no longer release bleed, but can still affect chromatographic performance. Because they are manufactured by the same material as the most widely used stationary phases, their retentive power is sub­ stantial and they tend to hinder vaporization of high-boiling compounds, resulting in discrimination against the late­ eluted components. The video showed how the sample liquid tends to be sucked into these particles of low thermal mass, which accentuates the problem. The sample is "filtered" through retaining septum particles provided the particles are located above the column entrance, i.e. when the liner contains a packing or an obstacle, or when the column in mounted low in a goose-neck liner. Classical injector design avoided the problem by positioning the col­ umn entrance ca. 5 mm above the bottom of an empty liner.

on Sample Evaporation

426

E 2. Surroundings of the Vaporizing Chamber

2.3.4. Recommendations

The first recommendation is certainly that the most suitable septum should be selected from those offered commercially. High-temperature septa are not necessarily the best. When the injector head is at a modest temperature only, a softer septum will last longer, particularly when used for thick needles, such as those used for headspace analysis. Particle formation is also reduced.

Septum Conditioning?

In the early times of GC, before septum purge became stand­ ard, many GC ovens contained a small beaker containing several septa which were baked out with every GC run over weeks or even months. This indeed removed the volatile material, but also oxidized the septa and rendered them harder, reducing their performance with regard to punctur­ ability and fragmentation. With septum purge and the sub­ stantial improvement of the septa, conditioning is no longer considered appropriate.

Puncturing at Same Point

A needle guide in the septum cap should help the syringe needle to pierce the septum at the same spot every time such that just a single channel is cut. Autosamplers facilitate this by a reproducible movement of the needle. Manual injec­ tion, however, typically introduces the needle at many sites and in different directions, cutting many channels which fi­ nally cause this region of the septum to fragment into small pieces. Sooner or later these particles fall out and are trans­ ferred to the vaporizing chamber. For this reason, septa last longer when used with autosamplers.

Tightening the Septum Cap

With regard to securing of the septum cap, there are two opposing arguments. On the one hand, greater force results in higher pres­ sure closing the core, which keeps the septum tight for more injections - as long as the needle always passes through the same channel. On the other, it increases the probability of pieces being cut out and a hole being cored which can no longer be closed by pressure. This is particularly likely to occur when the needle cuts the septum at many points. Practice seems to suggest the following two rules. 1 Particularly for manual injection, septa should be com­ pressed as little as possible in order to provide an easy passage for the syringe needle. The septum cap is tightened and then loosened again until the leak detector registers escaping carrier gas. The cap is then slightly re-tightened. 2 Septa must be tightened when hot. They expand strongly upon temperature increase and build up high pressure when the cap is secured at ambient tempera­ ture. This is also one of the reasons why injectors should be permanently heated: shrinking upon cooling may cause a leak.

2.3. Septum

427

Syringe Needle

With regard to the lifetime of a septum, needles with a side port hole are preferable to those of the beveled or cone style. They should not be used for the injection of samples in volatile solvents, however, because ofthe uncontrolled move­ ment of the sample in the vaporizing chamber. Hence they are useful only for the injection of high-boiling samples (trans­ ferred to the liner wall) and headspace analysis. The thicker needles used for the latter tend to be particularly harmful to the septum. Particularly when of the beveled style, the tip of the nee­ dle must be checked frequently, because a small defor­ mation rapidly ruptures the septum. Fingers feel deforma­ tions before they are visible by eye. Small hooks or edges are easily removed by drawing the needle through two pieces of fine sandpaper which are pressed together.

2.3.5. Merlin MicroBeal

Pre-cut septa sealed by the inlet pressure have been avail­ able for some time, but they could not be used at high tem­ perature. A more recent development, the Merlin Microseal, might ,be more successful, provided the septum cap does not exceed moderate temperatures. It consists of a pre-cut septum and is actuated by a small spring which keeps the needle entrance closed. The Merlin Microseal is supposed to last for many thousands of injections and should not release septum particles.

2.4. Heating of the Injector

Today's instruments show the injector temperature with the precision of a single degree. This pretends accuracy which is neither real nor useful. A difference of 100 is hardly of importance, considering that during sample evaporation the vaporizing zone is often cooled by many tens of degrees, at exactly the time the injector temperature is really important.

Heating up

Strong deviations can occur when the injector is freshly heated, because the temperature sensor is situated near the heating cartridge and reaches the set point more rapidly than the vaporizing chamber. The injector head is heated particu­ larly slowly. As this severely influences the evaporation in­ side the needle, the first analysis often produces odd results (e.g. greater discrimination). This is another reason for not switching off the injector heating overnight or at week­ ends.

Temperature Distribution

Only a small section of the vaporizing chamber, situated just below the center, is really thermostatted at the set point. The injector head, including the septum, is often more than 1000 cooler. The temperature of the part reaching into the oven commonly depends on the oven temperature. The temperature profile through the injector of the Hewlett­ Packard 5980 instrument is shown in Figure E13. The ex­ treme temperature drop towards the septum cap is a pre­ requisite to prevent evaporation inside the needle when in­

428

E 2. Surroundings of the Vaporizing Chamber jecting with the fast autosampler. A flat profile for an injector for thermospray injection, was shown in Figure A21.

-

InjlctlGII PlIIt So1Pollll Tlmpllllla,.

ollllplum

35O·C

~,

.........:O;;;;;;;;~~

10

20

:

I I I

30

Sy'TIp itI,"

40 50

&0 10

&0

".IGI

90

35·C

a.ln

tllliclioo PlIIt 50

100

150

200

250

300

350

Tlmpllo1... In Gil SIrllm (·CI

Figure E13 Temperature profile through the injector of the Hewlett­ Packard 5980 instrument. (From Klee [9].1

2.4. 1. Injector Head

In the early days of GC, a steep temperature drop towards the septum was a prerequisite for the use of elevated injec­ tor temperatures as septa were insufficiently thermostable. Today this is no longer a valid argument, firstly, because ther­ mostable septa are available (they withstand ca. 300 °CI and, secondly, because the septum purge prevents septum bleed from reaching the column.

Evaporation in the Needle

As discussed in Section A8.2.3, the temperature of the injec­ tor head is relevant to sample evaporation inside the syringe needle. A cool injector top helps to keep the needle tempera­ ture low and prevent evaporation inside. If evaporation in the needle cannot be avoided or is ex­ ploited for thermospray injection, the injector should be well heated up to the septum cap.

Condensation of High Boilers

In splitless injection, sample vapor expands backwards up

to the top of the liner. If the liner wall is substantially below

the regulated temperature, high-boiling components are

likely to recondense and might no longer reach the column

during the splitless period. Related losses are poorly repro­

ducible because the expansion of the sample vapor tends to

be different for every injection.

Measuring the Temperature Profile

When methods are transferred to other instruments,

setting the same nominal injector temperature does not en­

sure identical conditions, because the temperature gradients

towards the top and bottom are often different.

The user should have an idea of the temperature distribu­

tion within his injector. Measurement with a thermocouple

2.4. Heating of the Injector

429

is, in fact, easy. The thermocouple is introduced through the septum cap from which the septum and the narrow bore septum purge device have been removed; temperatures are measured at different depths in the injector. 2.4.2. Base of the Injec­ tor

Heating of the base of the injector, including the column at­ tachment, is primarily responsible for the temperature of the column inlet. Because the first few centimeters of the col­ umn are often contaminated with retaining material enter­ ing as aerosol, the temperature should be at least as high as that of the oven.

Dependence on Oven Temperature

The temperature of the base of the injector depends on that of the oven and can drop below 100°C after prolonged cool­ ing of the latter. There is no basic objection to this, provided the vaporizing chamber is not cooled and the temperature increases again at such a rate when the oven is heated that it never lags behind that of the oven. The latter is critical for fast programs, for instance after injection at a low oven temperature to obtain a cold trapping effect.

Minimized Heating of the Oven

Large changes of temperature also cause the ferrule of the column attachment to expand and contract, affecting the reliability of the seal (see below). This is in favor of a long heating block which keeps temperature high for the base of the injector also. On the other hand, heating of the oven by the injector must be minimized as this hinders chromatogra­ phy at low temperatures. This undesirable heat transfer also depends on the exposed surfaces and on the insulation in the oven roof.

3. Autosamplers Autosamplers are increasingly complex instruments combin­ ing many functions. Here only capabilities concerning the basic injection process are considered.

3. 1. Injection Speed

Since the introduction of the fast autosampler, the market is split into two types of samplers performing fundamentally different injection.

Fast Autosampler

The fast autosampler, patented by Hewlett-Packard (Agilent) in 1985, performs the whole injection in less than 500 rns (definition in the patent), sufficiently rapidly to prevent evapo­ ration inside the needle if the injector head is kept rather cool. Advantages and drawbacks have been extensively dis­ cussed in previous chapters.

430

E 3. Autosamplers

Non-Fast Autosamplers

All other autosamplers inject more slowly - speed relating to the process from inserting the needle to its withdrawal, not necessarily the depression of the plunger (see below). Hence "slow autosampler" would not be an appropriate term. They essentially imitate manual injection, with the ad­ vantage that manual and automatic injection result in the same evaporation process.

Needle Preheating

The more sophisticated non-fast autosamplers are program­ mable to perform hot needle injection, i.e. thermospray. The sample is withdrawn into the barrel of the syringe and the needle preheated in the injector during an adjustable period of time before the plunger is depressed. The function of leav­ ing the needle in the injector for an additional period of time after injection is less important.

Withdrawal of Solvent

Some autosamplers also enable withdrawal of solvent be­ fore or after the sample. When withdrawn before, it serves to perform solvent flush injection. Picking up ca. 0.2 III of solvent after the sample serves to clean the needle be­ fore it is introduced into the injector. It flushes the layer of sample liquid from the needle wall into the barrel. This prevents formation of a small pre-peak as discussed in Sec­ tion A5.3.4.

3.1.1. Injection Rate

Many autosamplers have a programmable injection rate, i.e. speed of depression of the plunger. This function has a vari­ ety of applications apart from split and splitless injection, but for vaporizing injection, fast depression of the plunger is always preferred. The fast autosampler injects in rapid mode to prevent evaporation inside the needle, but, as shown by the vid­ eos, is still slower than normal manual injection. For the other autosamplers, the plunger should be de­ pressed as rapidly as possible to enable the best per­ formance of hot needle injection. Slow depression of the plunger results in strong discrimination against the high-boiling analytes (Section A5.3.2).

Slowing in Last Phase

The drawback of depression by means of a stepper motor (the way programmable injectors are constructed) is the de­ creasing speed when the plunger approaches the bottom of the syringe. The movement must be slowed to avoid exces­ sive wear when the plunger hits the bottom. This slowing is undesirable but not totally avoidable.

3.1.2. Adjustable Depth of the Needle

Some autosamplers enable adjustment of the depth of pen­ etration by the syringe needle: the needle can be inserted only partly. This is useful in at least two instances: Split injection at elevated split flow rates profits from a longer distance between the needle exit and the col­ umn entrance, particularly when the liner is empty: there

3.1. Injection Speed

431

is more room for homogenization of the vapor. At split flow rates exceeding ca. 50 mLJmin, it is sufficient to insert the needle 3 cm. When the needle is inserted merely a minimum distance (ca. 15 rnrn), sample evaporation inside the needle can be avoided when the septum cap is not very hot and the solvent not too volatile. This results in the same performance (band formation) as with the fast auto­ sampler. In this mode of use, heating of the needle be­ fore and after depression of the plunger is avoided.

4. The Gas Regulation Systems The two fundamentally different concepts of gas regu­ lationfor split injection were described in Section C1.3: the pressure regulator/flow restriction and the flow/backpressure regulation system. Electronic systems differ somewhat from the early concepts using mechanical devices - approaching each other. Systems with mechanical regulators are also described be­ cause there are still many instruments of this type around and need not be replaced as long as there are no basic im­ provements. 4.1. Mechanical Pressure Regulation/Flow Restric­ tion

The pressure regulator/flow restriction system, shown in Fig­ ure E14, is the oldest concept. The gas supply is adjusted by a mechanical pressure regulator followed by a manom­ eter. The flow rates through the outlets of the septum purge and split line are regulated by means offlow restrictors, such as needle valves. Downstream ofthese valves there are usu­ ally devices for automated closure of the exits for splitless Manometer

~

y.

Needle valve

~ Septum purge

Il!IIIj!!....-

Pressure

regulator

Vaporizing

chamber

Needle

~

Flow meter

11

valve

cOlumn~L---y.--+ Split outlet

Figure E14

The classical pneumatic regulation system according to the

pressure regulationlflow restriction concept involving me­

chanical components.

432

E 4. The Gas Regulation Systems injection. Some instruments direct the two lines to the same closing valve, which reduces costs, but renders adjustment of the flow rates less convenient.

4. 1. 1. Pressure Regula­

tors

Basic Design

Here some general information on mechanical pressure regu­ lators is collected because these are used on many older GCs and in almost all gas supply systems, such as on the gas cylinders and in the gas distribution system. Mechanical pressure regulators contain a membrane which is in contact with ambient pressure (air) on one side and the gas to be regulated (e.g. the carrier gas) on the other (Fig­ ure E15). A spring compensates for the overpressure on the side ofthe regulated gas and can be loaded through a screw with a knob. The membrane actuates the valve. If pressure is too low (the sum of ambient pressure plus the force of the spring exceeds the pressure of the regulated gas), the valve allows gas to pass and to increase the pressure. It closes when the membrane approaches the equilibrium, i.e. the required pressure is reached. Knob to regulate pressure on the membrane

Figure E15

Schematic design of a mechanical pressure regulator.

Dependence of Pressure on Flow Rate

One of the weaknesses of mechanical pressure regulators is that the output pressure depends on the flow through the device. The valve should stop delivering gas as soon as the set point is reached - closure should occur at equilibrium. This is impossible because some force (overpressure) is required to obtain a seal. Pressure regulators are not designed for complete closure of the valve, but for the adjustment of a restrictor. The lower the gas flow rate, the tighter the valve should close, which requires increased overpressure. For this reason, the regu­ lated pressure tends to increase as the flow rate is reduced, with the effect that retention times become de­ pendent on the split flow rate. This is a problem primarily at flow rates below about 10 mL/min.

4.1. Mechanical Pressure Regulation/Flow Restriction

433

Effect on Retention Times

In split injection, the flow dependence of the regulated pres­ sure can affect retention time reproducibility when split flow rates are varied or when the split flow rate is strongly, but not reproducibly reduced some time after the injection is completed (gas saver). In splitless injection, closure of the exits of the split and the septum purge line can increase pressure and the column flow rate during the sample transfer period. This corresponds to a weak pressure pulse. It is probably not of significant usefulness, but has no adverse effect either.

Temperature Dependence

Pressure released by mechanical regulators depends on tem­ perature because the tension of the membrane and the pressure of the spring depend on it. This is one reason why retention times tend to drift at the beginning of longer series of analyses: pressure regulators for the carrier gas are usually installed as far removed from the oven as possible, but warming from a hot oven cannot be avoided entirely.

Precision, On/Off Valves

Some, users maintain the inlet pressure also during longer stand-by periods, such as weekends. They argue that re-ad­ justment is inaccurate and responsible for instability of the retention times. They often pay for this in the consump­ tion (waste?) of large amounts of carrier gas. Some instruments are equipped with on/off valves between the pressure regulator and the manometer. This enables switching off of the carrier gas without touching the pres­ sure regulators.

Air Tightness

Ambient pressure, i.e. air, is on one side ofthe relatively large membrane, carrier gas passes on the other side. The mem­ brane must not permit diffusion of air, because otherwise the carrier gas will be contaminated with oxygen and hu­ midity.

Flexibility of the Membrane

For accurate regulation, the membrane should be highly flex­ ible. So far, the best membranes have been made from rub­ ber or plastic. Polymers are open to diffusion of air and humidity, although to a widely varying extent, and one should keep in mind that diffusion is also possible against a pressure drop. It makes no sense to buy highly pure carrier gases when pres­ sure regulators contaminate them. Oxygen is a severe prob­ lem for polyglycol (Carbowax) type stationary phases, but also causes lacquering (excessive crosslinking) of silicones at temperatures above ca. 300°C. Humidity does not seem to be a problem for the column, but it does affect the per­ formance of ECDs.

Metal Membranes

Diffusion-free membranes are made from metal, mostly con­ structed as bellows for flexibility. As an inexpensive trick, plastic membranes can be covered with thin metal

434

E 4. The Gas Regulation Systems foil. The regulator is dismantled and aluminum foil (e.g. from a chocolate bar, the chocolate rewarding the effort of saving money) is placed on the membrane. Thin steel foil is actually preferable because of the lower risk of its being torn during re-assembly ofthe regulator [101.

Regulators in the Gas Distribution

Regulators on gas cylinders and in gas distribution systems are usually more of a problem than those in the instruments. Some have extremely large membranes and enable passage of a correspondingly large amount of oxygen and humidity. It is recommended that regulators with steel membranes are used for these installations, also because precision is of minor importance there.

4.1.2. Manometer.

Accuracy of pressure regulation primarily depends on the manometer, because the regulator is adjusted to reach the set point shown by the latter. Mechanical manometers usu­ ally contain a hollow spring that tends to stretch (straighten) with increasing internal pressure. By means of a reversing lever, the position of the end of the spring turns the needle of the manometer.

Problematic Accuracy

Mechanical manometers are sensitive to overpressure and sudden increases of pressure, because the spring is easily deformed. In fact, when manometers are checked after some years of use, some easily show deviations of 20-30 kPa.These are often the main cause of variations in the retention times obtained from different instruments. The pressure indicated depends, furthermore, on temperature, because it depends on the tension of a spring.

Check Column Head Pres­ sure in Injector

Inaccurate column head pressure is often the main reason for deviations in retention times. If such deviations cause problems, inlet pressures should be checked regularly, e.g., by use of a manometer equipped with a syringe needle. The needle is introduced through the septum, as shown in Fig-

Carrier gas

Figure E16 Measurement of the carrier g8S pressure inside the injector by means of a manometer equipped with 8 syringe needle.

4. 1. Mechanical Pressure Regulation/Flow Restriction

435

ure E16. It should be long enough to enable penetration through possible packings, because this enables compari­ son of the pressure above and below such a resistance. The most reliable pressure reading is obtained in the "quiet" corner just below the septum, because a high gas velocity at the needle exit can pull gas from the manometer, causing it to indicate too Iowa pressure. 4.2. Mechanical Flow/ Backpressure Regulation

In the late seventies, Hewlett-Packard introduced the flow/ backpressure regulation system in response to the dense packings (occasionally) used in the vaporizing chamber at that time. The mechanical system, still incorporated in the HP 5890, is shown in Figure E17. It includes a flow regula­ tor in the carrier gas supply line, a needle valve in the septum purge outlet, as well as a charcoal filter, manometer, and backpressure regulator in the split outlet.

-b-o---.,r-I---- ~

,Needle valve

Flow regulator

Septum purge

Flow

sensor

Manometer Filter Column

Split outlet Backpressure regulator

Figure E17 The mechanical flow/backpressure regulation system of the HP 5890 system for the split-only inlet.

Pressure Drop over Dense Packing

Backpressure regulation in the split outlet was introduced to enable reliable control of the column head pressure even when there was a substantial pressure drop in the injector liner (dense packing). It was a further development of the system proposed by German and Horning [11], who found that retention times depended on the split flow rate, as a result of varying pressure drops over the packing. As mentioned above, pressure drops become noticeable only under conditions combining a high split flow rate with a nar­ row bore liner packed with a long and dense plug of pack­ ing. A 1 cm bed of commonly used column packing material creates less than 10 kPa pressure drop even when the split flow rate is as high as 300 mllmin. Dense plugs of glass wool are clearly more permeable and do not produce noticeable pressure drops under any conditions. It should, furthermore, be remembered that high split flow rates cause variations of the column head pressure for other reasons also. Hence, backpressure regulation was introduced to solve a problem that no longer exists.

436

E 4. The Gas Regulation Systems

Splitless Injection

As the regulated flow could not be stopped during the splitless period, the flow passing through the vaporizing chamber was re-directed. As shown in Figure E18, an elec­ tric three-way valve was installed in the split outlet. During the splitless period, the main gas stream passed into the septum purge line and around the injector directly into the split outlet. To purge the injector (normal position), the sole­ noid valve was switched to feed the gas flow through the vaporizing chamber, closing the re-direction route. In con­ trast to the pressure regulator/flow restriction design, no on/ off valve was required at the exits of the septum purge and split line (closing the septum purge exit was useless since there was anyway a high gas flow rate passing the top of the vaporizing chamber).

I

Needle valve

4-o--n,....----T"""-.y--.. Flow regulator

Septum purge

Flow sensor Re-direction of the split flow during the splitless period Manometer

Split outlet

Figure E18 The flowlbackpressure regulation system of the HP 5890 instrument as used for splitleaa injection.

4.2. 1. Comparison of the Two Systems

Both the pressure regulator/flow restriction and the flow/ backpressure regulation system have advantages. Beyond those mentioned above, the following points are important. The flow/backpressure regulator system basically ena­ bles adjustment of the split flow rate without flow measurement at the split exit. 2 The maximum split flow rate is limited by the capac­ ity of the flow regulator, usually 500 mt/m!n. 3 Split flow rates and, hence, split ratios controlled by the flow/backpressure regulator system are more strongly affected by the pressure wave resulting from sample evaporation than those regulated by pressure/flow re­ striction systems (Section C8.3.1). 4 In the event of severe leakage or if the column is dis­ mantled without the carrier gas supply being switched off, the loss of carrier gas into the GC oven is limited by the flow regulator of the flow/backpressure regulator system, whereas it can be extremely high in the other. If hydrogen is the carrier gas, however, safety considera­ tions still call for a sensor to check hydrogen concentra­ tions in the oven atmosphere.

-4.2. Mechanical Flow/Backpressure Regulation

5

4.3. Electronic Regula­ tion Systems

437

The flow/backpressure regulator system is a drawback for splitless injection since the high gas flow rate pass­ ing over the head of the liner into the septum purge line accentuates the loss of sample material backwards out the vaporizing chamber.

In the early nineteen nineties, fully electronic regulation sys­ tems were introduced in response to the need of automa­

tion and regulation by means of data systems. Data systems enable downloading of complete meth­

ods, ruling out erroneous adjustment of parameters.

Quality assurance and GLP require complete documen­

tation.

Electronic devices tend to be more accurate than me­

chanical regulators.

Because mechanical regulators could not be directly trans­ lated into electronic devices, the transition brought about some modifications. Proportional Valves

Mechanical pressure and flow regulators were replaced by proportional valves which feed a gas flow according to a set point which can be a pressure or a flow rate. This enhanced the flexibility of the system. A flow and a pressure sensor were built into the system which enabled regulation by ei­ ther principle and measurement of the other for calculating column and split flow rate.

4.3.1. Flow/Backpre88ure Regulation

A diagram of the pneumatic system of the Hewlett-Packard/ Agilent 6890 instrument is shown in Figure E19. Proportional valve 1

~ ...i.l.ter .....

F?

P?

Flow sensor

Pressure sensor

Pressure R .. I estnction regu ator /

~~-+-SePtumpurge ~ Onloff valve

Fixed septum purge regulator

t--......-.... Split outlet

Figure E19

Electronic carrier gas control of the HP 6890 instrument.

Split Mode

The philosophy ofthe system for split injection corresponds to that with the mechanical regulators. Proportional valve 1 feeds gas according to the set point for the flow rate controlled by the flow sensor down­ stream in the supply line. The combination of the valve and the sensor acts like the previous flow regulator. Proportional valve 2 in 'the split line controls the split flow rate. It is regulated by the pressure sensor and a

438

E 4. The Gas Regulation Systems pressure set point from the software for the column head pressure. It acts, hence, like the previous backpressure regulator. The split flow rate is equal to the total flow rate regulated by the flow sensor minus the fixed septum purge flow rate (3 mt/rnln) and the calculated flow rate through the column.

Pressure Sensor Above Liner Packing

In contrast with the earlier design, the pressure sensor is positioned in the septum purge line. On the one hand, this protects it from sample material passing through the split outlet. On the other, a reduced column inlet pressure result­ ing from a pressure drop over a dense packing in the vaporizing chamber can no longer be prevented, hence the original justification of flow/backpressure regulation is aban­ doned.

Splitless Mode

Because of the flexibility of the components, splitless trans­ fer can now be performed by the principle of pressure regu­ lation/flow restriction. During the splitless period, proportional valve 1 feeds gas according to a set point which can be a constant pressure or a pressure program, controlled by the pres­ sure sensor in the septum purge line. The flow through the split outlet is stopped by an on/off valve. After the splitless period the system switches back to backpressure regulation mode, valve 1 guided by flow, valve 2 by pressure. In its essence, the pneumatics active during splitless injec­ tion made the transition back to the classical design.

Flow Adjustment by Means of Pressure Regulation

There is frequently confusion about electronic regulation of column flow rates in capillary GC. Direct flow regulation is possible for packed columns, but not for capillary GC with split/splitless injection because of the far higher flow rates through the split and the septum purge exits. Column flow rates are adjusted by means of the inlet pressure; the soft­ ware calculates the required inlet pressures for given oven temperatures, adjusts it is flow performance is pro­ grammed, and defines the pressure set points accordingly.

4.3.2. Pressure Regula­ tion/Flow Restriction

The electronic and data-system-controlled version of the pneumatics for the pressure regulation/flow restriction sys­ tem is shown for the TRACE GC from CE Instruments. The components used are basically the same as those found in the HP 6890 instruments (Figure E20).

Regulation system

The proportional valve 1 in the supply line feeds gas accord­ ing to the pressure set-point and the pressure determined by the pressure sensor. The desired split ratio is entered into the data system, where the split flow rate is calculated from the column flow rate and the fixed septum purge flow rate (5 mljmin for helium). The split flow rate is controlled by

4.3. Electronic Regulation Systems Proportional valve 1

F?

P?

Flow sensor

Pressure sensor

Pressure A I 'c!' I esm Ion regu ator /

~~sePtumpurge

~ . . . iller •...

439

~

On/off valve

Fixed septum purge regulator

1-..... -...

Split ounet

Figure E20 Electronically controlled pneumatic system of the TRACE GC from CE Instruments.

proportional valve 2 in the split outlet which is guided by the

flow sensor in the supply line.

For splitless injection, the system incorporates two on/off

valves, but works as in split mode.

Reaction on the Pressure Wave

Electronic regulation systems react to the pressure wave ini­

tiated by sample evaporation in a way other than the two

mechanical systems described above. Immediately after in­

jection, the pressure sensor observes a value exceeding the

set point and closes the proportional valve. The flow sensor

observes a reduction and tends to open the proportional valve

in the split outlet. As this would, in fact, increase the split

flow rate (the weakness of backpressure regulators), the re­

action of the proportional valve is dampened to such an

extent that it does not respond to a wave.

4.4. Charcoal Filters in the Split Outlet

Many instruments are equipped with a charcoal filter in the

split outlet line. There are good reasons for this, but also

some problems.

4.4.1. Advantages

Some analysts are concerned about the toxicity of the solute

material leaving the split outlet - in split injection, 90-99 % of

the sample is vented through this outlet. Handling of toxic

material certainly requires proper attention, but samples with

toxic components are usually highly diluted and injected by

a non-splitting method. Much more material is spilled, fur­

thermore, e.g. during cleaning and loading of the syringe.

Unless the toxic substances are highly volatile, they will

hardly ever leave a split line. They undergo a kind of capil­

lary GC in the cool outlet tube coated with a thick layer

of previously deposited sample material. In fact, such tubes

can even become plugged.

Retention of Toxic Sub­ stances

Protection of Regulators

Regulation of the split flow rate by means of electronic ele­

ments requires a filter preventing contamination of the

devices. Needle valves were easy to clean (and might have

required cleaning every year to ensure good performance),

but this is no longer possible with modern systems.

440

E 4. The Gas Regulation Systems

Stability of the Split Ratio

The last (and oldest) argument concerns the stability of the split ratio in systems regulated by means of a restrictor (such as a needle valve). The viscosity of the solvent vapor differs from that of the carrier gas, and when passing through the restrlctor, it alters the flow rate. A strong adsorbent, such as charcoal, is used in order to retain solvent vapors for a short period of time at least, such that the change in the split flow rate occurs only after the whole sample has passed the split point.

Viscosity of Sample Vapor

The least viscous vapor, that from pentane, is nearly three times Jess viscous than helium and 20 % less viscous than hydrogen. The most viscous vapor (dichloromethane, chlo­ roform, acetone) is slightly more than half as viscous as he­ lium and some 15 % more viscous than hydrogen [121. Hence, vapor viscosity differs rather little from that of hydro­ gen, and even compared with helium, differences are not spectacular when it is considered that the vapor is diluted several fold. Backpressure regulation is not affectedby viscosity changes, nor are modern electronic systems with proportional valves.

4.4.2. Drawbacks

At least when the charcoal is fresh and absorbs the vapor, it sucks the sample vapors into the filter, i.e. into the split out­ let, and increases the split ratio. In most instruments, the charcoal is not exchanged for years (many users not even know of its presence) and has lost most of its effect (includ­ ing the negative one). As filters have an internal volume of ca. 2 mL, they are a severe problem for splitless injection: as described above, pressure increase during sample evaporation pushes vapor into the split outlet. Closure ofthe line should occur upstream of the filter.

4.4.3. Suitable Size

Present filters are large, probably to have a high capacity to retain solvent. Assuming that charcoal retains an amount of solvent similar to its own weight (several hundred milli­ grams), after a few hundred injections it is saturated and the solvent is slowly released again. Release will occur at low concentration, such that the filter still has a dampening effect and prevents solvent recondensation in the cool regu­ lating devices. The capacity for higher-boiling material is high and ensures that the vapor pressure in the passing gas re­ mains low enough to prevent condensation in the critical parts. In trace analysis it takes years to deposit milligrams of mate­ rial. Injection of undiluted samples in split mode can, on the other hand, consume the capacity of the filter in weeks. If the injector is primarily used for splitless injection, char­ coal filters should be small (e.g. 300 ~L). When also used for split injection of concentrated samples, they must be regu­ larly replaced even when their internal VOlume is ca. 2 mL.

4.5. Septum Purge

441

4.5. Septum Purge

The septum purge was introduced by Kurt Grob in 1972 (13). As the term implies, it serves to keep septum bleed away from the vaporizing chamber and the column. Removal of septum bleed is particularly important if high injector tem­ peratures are used.

Removal of Solvent Vapors

The septum purge was, however, introduced for another purpose specific to splitless injection: even if the vaporizing chamber is not significantly overloaded, some vapor diffuses backwards from the liner into the septum area (Section D3.3.1I. Purging the vaporizing chamber at the end of the splitless period does not remove this vapor. If there is no septum purge, such vapor slowly diffuses back into the stream of carrier gas directed towards the column entrance and causes broadened and tailing solvent peaks. The septum purge may be constructed as shown in Figure E21. The carrier gas entering the injector is split into a main stream directed to the column and a small purge stream flow­ ing towards the septum and through a separate exit. Septum bleed and solvent vapors diffusing towards vaporizing chamber

Septum

Carrier gas in

===--~;;r::N

Side stream of carrier gas stopping diffusion

Injector liner to column and split exit

Figure E21

Injector head with septum purge. A small proportion of the

carrier gas flows towards the septum and through a sepa­

rate exit.

Design

The key component is a narrow channel through which carrier gas is diverted from the vaporizing chamber towards the septum purge outlet. The relatively high velocity of the gas in this channel rules out diffusion from the septum re­ gion towards the vaporizing chamber. The septum purge device must be removable to enable re­ placement of the injector liner and must be tight fit in the injector body so that it does not create new dead volume. A ferrule between the device and the injector body should be avoided because this might again cause bleed and "ghost" peaks. As there is no significant pressure drop, it is sufficient if the device fits closely into the injector body.

Gas Velocity

The linear velocity of the gas in the narrow channel di­ rected towards the septum must be at least equal to the ve­

442

f 4. The Gas Regulation Systems locity of the vapors tending to diffuse from the septum zone towards the column. This velocity is in the order of millimeters per second. The minimum purge ge8 flow rete required is smaller the narrower is the channel; the channel must be just large enough to enable passage of the syringe needle. For a 0.8 mm i.d. channel, the minimum flow rate is calculated to be ca. 0.1 mLlmin (which is difficult to measure).

Purge Flow Rate

The flow rate can exceed this minimum substantially. Dur­ ing splitless injection, high gas velocities can cause turbu­ lence in the liner and suck out sample vapor. Suitable septum purge flow rates range from 0.5 to 10 mLlmin. As the septum purge flow rate is rather uncritical, it is usu­ ally adjusted just once, by application of an intermediate carrier gas inlet pressure. The needle valve can be replaced by a (cheaper) restrictor consisting of 0.5 mm i.d. steel tub­ ing crimped to provide the required flow rate. New instruments often regulate a fixed 8eptum purge of 3-5 mL/min. This is achieved by means of a pressure regula­ tor releasing the gas at a constant and' low pressure (below the inlet pressures likely to be used), followed by a restrictor. Hence, the pressure drop overthe restrictor is kept independ­ ent ofthe carrier gas inlet pressure by means ofthe pressure regulator.

E References

443

References E K. Grab and K. Grob, "Splitless Injection and the Solvent Effect", J. High Reso!. Chromatogr. Chromatogr. Commun. 1 (1978) 57. 2 M. van Lieshout, M. van Deursen, R. Derks, H.-G. Janssen, and G.A. Crsmers, "The Influ­ ence of Liner Dimensions on Injection Band Broadening in Split Injections in Fast Capil­ lary GC", J. High Reso!. Chromatogr. 22 (1999) 116. 3 American Society for Testing and Materials (ASTM), "Standard Practice, Installing Fused Silica Open Tubular Capillary Columns in Gas Chromatographs", ASTM E 1510-93, 1993. 4 K. Grob and M. Biedermann, "Visual Experiments on Sample Evaporation: Conclusions on Split and Splitless GC Injection and Injector Design", Anal. Chern. (2001) 5 A. Kaufmann, "Maximum Transfer Conditions for Splitless Injection", J. High Resol. Chromatogr. 20 (1997) 193. 6 D.M. Ottenstein and P.H. Silvis, "GC Septa: A Comparison of Bleed Characteristics", J. Chromatogr. Sci. 17 (1979) 389. 7 Anonymous, "A Guide to Minimizing Septa Problems", Restek Corporation 1998. 8 M. Biedermann and K. Grob, "GC Ghost Peaks due to Fingerprints", J. High Resol. Chromatogr. 14 (1991) 558 9 M.S. Klee, "GC Inlets - An Introduction", Hewlett-Packard Co., Avondale (1991) 42. 10 K. Grob, "Diffusion-free Pressure Regulators in Capillary GC", J. High Resol. Chromatogr. Chromatogr. Commun., 1 (1978) 173. 11 A.L. German and E.G. Horning, J. Chromatogr. Sci. 11 (1973) 76. 12 K. Grob, "On-Column Injection in Capillary GC" (1991) 529. 13 Kurt Grab and G. Grob, "Techniques of Capillary GC. Possibilities of the Full Utilization of High-Performance Columns. Part I: Direct Sample Injection", Chromatographia 5 (1972)

3.

Appendix 1

445

Appendix 1 Selection of the Injection Technique

Classical Split Injection

Use for: relatively concentrated solutions: 20 ng/Ill to 1 % (FID) per component; analysis of undiluted samples; headspace analysis; and/or fast, fully isothermal analysis. Simple to handle; flexible with regard to sample concentra­ tion, solvent, and column temperature during injection; op­ timum reproducibility of absolute retention times; demand­ ing for analyses requiring high accuracy; high risk of sys­ tematic errors.

PTV Split Injection

Eliminates problems resulting from evaporation inside the needle: enables injection of small volumes «1 Ill) and avoids discrimination against high-boiling material as a result of se­ lective elution from the needle; samples always evaporate from surfaces (liner wall or packing), i.e. reproducible proc­ ess, but sensitive to surface effects.

Splitless Injection

Use for: dilute samples: 0.5 to 50 ng/Ill per component, 10 pg/ III with vapor overflow (FID); and/or "dirty" samples, especially if highly accurate results are not the first priority. Produces relatively accurate results for volatile solutes; prob­ lems with quantitation of high-boiling solutes (matrix ef­ fects!); requires reconcentration of the initial bands by cold trapping or solvent effects, which often necessitates cooling of the column during injection.

PTV Splitless Injection

For differences compared with classical splitless injection, see PTV split injection.

PTV Solvent Split Injection

For splitless injection of components of intermediate to high boiling point; most of the solvent is vented (all if performed by backflushing). Used for large volume injection (up to 500 Ill) and if the solvent or a derivatization reagent disturbs the detector.

446

Appendix 1

On-Column Injection

Dilute samples: 300 to 0.01 ng/Ill per component (FlO); best method for producing highly accurate results and analyzing labile compounds; not suitable for highly contaminated sam­ ples (samples containing more than ca. 0.1 % of involatile by-products); injection requires cooling of the column be­ low the solvent boiling point (except for high oven tempera­ ture on-column injection).

Appendix 2

Selection of Conditions for Classical Split and Splitless Injection Evaporation Inside Syringe Needle

If possible, prevent sample evaporation inside the syringe needle. Use high-boiling solvents, low injector temperature (including cool septum cap), short needle, fast injection. If evaporation inside the needle cannot be prevented, the amount of sample material transferred from the needle should be maximized by the use of a high injector tempera­ ture (including a hot injector head) and hot needle injection.

Sample Volume

If evaporation inside the syringe needle cannot be avoided, use 5 or 10 III syringes, although the minimum volume in­ jected then corresponds to the needle volume, i.e. 0.5-1 ul., Injection of "0.5" to "1 Ill" (volumes read on the barrel of the syringe, i.e. in addition to the needle volume) improves transfer of high-boiling components from the syringe nee­ dle. Splitless injection without vapor overflow: maximum sample volume ca. 2 ul., Split injection: small sample vol­ umes (ca. "0.5" Ill) tend to produce better quantitative re­ sults.

Length of Syringe Needle

long needles, releasing the sample near the column entrance, for split/ess injection and split injection with low split flow rates (exception, fast autosampler injection), as well as for headspace analysis. Short needles for split injection with high split flow rates and for samples in high boiling matrices or matrices which evaporate with difficulty.

Injector Temperature

Minimum injector temperature if sample evaporation inside the syringe needle can be avoided. Otherwise maximum in­ jection temperature not degrading solute material. Splitless injection: high injector temperature improves sample trans­ fer and reduces matrix effects.

Appendix 2

447

Width of Liner

The sample vapor must not overfill the injector liner (except for the vapor overflow technique). Splitless injection: liners of 3-5 mm i.d, Split injection: split ratios tend to be more accurate with 3-5 mm i.d. liners; if split flow rates are low, 2 mm i.d. liners provide sharper initial bands.

Packed Liners; Liners with Obstacles

Rule to start with: 1 packed liner or liner with built-in obstacle for fast autosampler injection; 2 packed or narrow bore empty liner for samples in high boiling matrices; 3 empty liner for injection with thermospray. Packing the vaporizing chamber with, e.g., silanized glass or quartz wool, or use of liners with built-in obstacles prevents unevaporated sample liquid from flying past the column entrance after injection with band formation. Such techniques might also reduce entry into the column of involatile by-prod­ ucts after injection with thermospray. The packing material might promote decomposition of labile solutes, cause "ghost" peaks, and retain (adsorb) high-boiling components.

Carrier Gas Flow Rate

Splitless injection: high carrier gas flow rates improve trans­ fer of solutes from the injector into the column; below 2 mLl min sample transfer usually becomes unsatisfactory (below 1.5 mLlmin if solvent recondensation accelerates transfer). If

possible, increase pressure during transfer.

Split injection: high carrier gas flow rates to obtain maxi­

mum sensitivity, but low flow rate for high split ratios. At

high carrier gas flow rates, the best separation efficiencies

are obtained with 0.32 mm i.d. columns.

Duration of the Splitless Period

At a carrier gas flow rate of 4 mLlmin and without solvent recondensation, a splitless period of 40 s is usually sufficient; at 2 mt/rnln, the duration should be extended to 90 s.

Column Temperature During Injection

Split injection: whenever possible avoid column tempera­ tures more than 20° below the solvent boiling point to pre­ vent recondensation in the column inlet (sucking more sam­ ple material into the column than expected from the pre-set split ratio). Splitless injection: reconcentration of bands broadened in time requires either reduction of the column temperature to at least 60-90° below the elution temperature of the solutes of interest (cold trapping) or a column temperature at least 20-25° below the solvent boiling point (solvent effects).

448

Appendix 3

Appendix 3 Glossary of the Most Important Terms Used in the Text Air plug injection: syringe handling technique for injection of liquids; use of an air plug be­ tween the sample plug and the plunger. Autosampler. device for automatic injection. Band broadening in space: spreading of the initial solute bands in a manner which causes their length to be equal in terms of column length; caused, e.g., by flow of sample liquid in the column inlet. Band broadening in time: spreading of the initial solute bands such that they have the same width in terms of gas chromatographic retention time at the temperature of injection; caused, e.g., by slow transfer of the solute material from the injector into the column during splitless injection. Band formation: injection through a cool syringe needle such that the liquid leaves as a band (jet) or row of aligned droplets; typical of fast autosampler injection. Butt connector. "low" or "zero dead volume" unions for connecting capillaries by means of a compression fitting and ferrule. Capacity of the column: maximum amount of a component which can be injected without causing overloading of the column (peak asymmetry). Cold split/splitless injection: synonymous with programmed temperature vaporizing (PTV) injection. Cold trapping: technique used for reconcentrating solute bands broadened in time: the sol­ utes are introduced into the column at an oven temperature substantially below their elution temperature. Concurrent solvent evaporation: evaporation of the sample solvent during introduction of the sample into the column inlet. Classical vaporizing injection: injection into a permanently hot split/splitless injector. Cool needle injection: the sample liquid is withdrawn into the barrel of the syringe, the needle is introduced into the injector, and the plunger is depressed without delay. Dead time: see Gas hold-up time. Direct injection: non-splitting vaporizing injection into a liner connected directly to the col­ umn entrance (there is no split outlet). Discrimination: discrimination against a solute occurs if the proportion of the solute entering the column is smaller than that of other solutes used as the basis of comparison.

Appendix 3

449

Dynamic headspace analysis: analysis of the volatile components of a sample by stripping with a flow of gas or by continuous removal of the headspace gas above the sample; components are usually enriched on a trap. Elution temperature: column temperature at which a component is eluted from the column and enters the detector. Enhancing matrix effect presence of matrix material in the sample causes peak areas for a given amount of solute to be larger for the sample than a solution of standard in solvent (e.g. calibration mixture); results from adsorption in the injector or column. External standard method:. determination of a concentration or amount represented by a peak by comparison of the peak area or height with that obtained by analysis of a mixture of standards containing a known amount or concentration of the solute of interest. Fifteen degree rule: rule of thumb that the chromatographic migration speed of a solute changes by a factor of two for each 15° step in column temperature. Filled needle injection: the sample is not withdrawn from the needle into the barrel of the syringe before injection. Flash evaporation: vaporizing injection under conditions resulting in virtually instantaneous sample evaporation in the gas phase; term coined before sample evaporation in a hot injector was investigated. Flooded zone: column inlet section over which the sample is spread by flow of the liquid; occurs in on-column injection or in splitless injection with solvent recondensation. Fluctuating split ratio: changing ratio of the split and the column flow rates during the split­ ting process. Full trapping: complete retention (solvent trapping) of a solute within the flooded zone during solvent evaporation, resulting in sharp initial bands. Gas hold-up time: time required for the carrier gas to pass through the column (also called "dead time"). Ghost peaks: peaks in a chromatogram not arising from material injected with the sample. Headspace analysis: indirect analysis of solutes in liquid or solid samples via the gas phase, i.e. the gas above the sample. Hot split/splitless injection: injection into a conventional, permanently hot injector. Hot needle injection: the sample liquid is withdrawn from the needle into the barrel of the syringe; the inserted needle is pre-heated in the injector for 3-5 s before rapid depression of the plunger. Injection temperature: column temperature during sample introduction. Initial band length: length of the column inlet over which sample material is spread (corre­ sponds to the length of the flooded zone in on-column injection or in splitless injection with solvent recondensation). Initial band width: width, in terms of time, of the solute band in the column inlet. Insert also injector insert or liner; tube housing the vaporizing chamber. Internal standard method: determination of concentrations or amounts of solutes by com­ parison of peak areas or heights with those obtained from a known amount or concentra­ tion of a substance added to the sample injected.

450

Appendix 3

Linear splitting: the split ratio is equal for all the components of a sample; the composition of the sample entering the column is identical with that inside the injector. Liner. glass or metal tube in the injector housing the vaporizing chamber. Matrix effect influence of the sample matrix (solvent, involatile sample by-products) on quan­ titative results. Matrix-induced chromatographic response enhancement see "enhancing matrix effect". Memory effect appearance of peaks arising from samples injected previously. Moving needle injection: solvent-free splitless injection by deposition of liquid samples on the tip of a needle at ambient temperature, evaporation of the solvent by a stream of carrier gas, and introduction of the needle into a vaporizing chamber to transfer the sol­ utes to the column. Needle rinse injection: determination of solute material remaining inside the syringe needle by sucking solvent into the syringe and injecting the needle washings. On-column injection: direct introduction of the sample liquid into the oven-thermostatted col­ umn inlet without a vaporization step in a separately heated chamber. Overflow technique: discharge of solvent vapor by expansion of the vapor (in contrast with vapor carried away by a gas stream). Partial solvent trapping: incomplete retention of a solute within the sample layer in the col­ umn inlet during solvent evaporation, causing peak deformation ofthe "chair" or "stool" type. Peak splitting: although there are numerous mechanisms resulting in splitting of peaks, "peak splitting" is often used synonymously with "band broadening in space". Phase soaking: solvent effect taking place in the coated column beyond the flooded inlet for solutes which are co-chromatographed with the solvent or another strongly overloading component; solute migration is slowed owing to swelling of the stationary phase. Press-fit connection: connection of capillaries by pressing a fused silica capillary into a ta­ pered seat of a (glass) connector, T-piece, widened glass capillary butt, or another ta­ pered seat. Pre-set split ratio: split ratio set by adjustment of the gas flow rates through the split outlet and into the column. Pressure wave: increase of pressure in the injector caused by rapid evaporation of the sample liquid. Programmed temperature vaporizing (PTV) injection: injection into a cool injector; the solute material is transferred into the column in split or splitless mode after rapid heating of the injector. Recondensation effect increase of the flow rate into the column inlet or the split outlet by recondensation of a substantial proportion of the injected sample material (mostly of the solvent), affecting the true split ratio and accelerating sample transfer into the column in splitless injection. Reducing matrix effect presence of matrix material in the solution injected reduces the peak area for a given amount of solute compared with a solution of standard in solvent (e.g. calibration solution); probably results from transfer to the liner wall in droplets of invola­ tile material.

Appendix 3

451

Retention gap: column inlet of low retentive power compared with the separation column, usually consisting of a deactivated but uncoated pre-column; used for reconcentration of solute bands broadened in space or for reduction of the effects of involatile sample by­ products. Sample transfer. transfer of the sample material from the vaporizing chamber into the col­ umn. Sandwich injection: the plug of sample liquid in the syringe is placed between plugs of sol­ vent, each separated from the sample plug by an air bubble. Septum purge: purge line leaving a vaporizing injector between the septum and the entrance of the carrier gas into the vaporizing chamber; a small flow of carrier gas through this exit removes septum bleed, or sample vapor which entered the septum area from the injector. Solid injection: injection of liquid samples involving removal of the solvent (and volatile sol­ utes) before transfer of the solute material of interest to the column (e.g. moving needle or solvent split PTV injection). Solvent effects: effects of the sample solvent or another strongly overloading component on the chromatography of the solutes (solvent trapping, phase soaking, band broadening in space). Solvent evaporation temperature: column temperature during evaporation of the sample sol­ vent within the flooded zone. Solvent evaporation time: time required for the sample solvent to evaporate from the flooded zone. Solvent flush injection: the syringe needle is filled with solvent, followed by an air bubble, before the sample liquid is picked up. Solvent split injection: injection technique involving use of a programmed temperature va­ porizing (PTV) injector. The sample solvent is evaporated and vented through the split exit at low injector temperature, followed by heating of the injector and splitless transfer of the solute material into the column. Solvent trapping: solvent effect describing the behavior of solutes during solvent evaporation within the flooded zone.

SPI: Septum-equipped temperature-programmable injector; PTV-like injector from Varian In­ struments. Split flow rate: carrier gas flow rate leaving the exit at the base of a vaporizing injector. To­ getherwith the column flow rate it determines the split ratio, but is also used to purge the vaporizing chamber after splitless injection. Split injection: injection under conditions causing a (usually relatively small) proportion ofthe sample vapor to enter the column and the main stream to leave through the split outlet. Splitless injection: injection technique resulting in almost complete transfer of the sample material into the column, involving a vaporizing injector also suitable for split injection; the split line is closed during sample transfer. Splitless period: period of time during which the split exit is closed for transferring the sample material into the column.

452

Appendix 3

Split line: outlet line at the base of the vaporizing chamber of a split/splitless injector, used for discharging sample material in split injection or purging the vaporizing chamber after splitless sample introduction. Split point zone in the vaporizing chamber around the column entrance where the sample vapor is divided into the portion entering the column and the portion purged through the split line. Split ratio: usually understood as pre-set split ratio: ratio of the carrier gas flow rates passing by the column and entering it before the injection is performed (adjusted split ratio). Static headspace analysis: analysis of volatile components in the gas phase above a sample after a single equilibration step. True split ratio: effective split ratio, calculated from the proportion of the solute material enter­ ing the column. Standard addition: method for quantitating solute concentrations or amounts by adding to the sample a known concentration or amount of the solutes of interest. The added mate­ rial corresponds to the difference between the peak areas or heights obtained from the analysis ofthe sample with and withoutthe added standard material. The concentrations or amounts are calculated therefrom. Terminal band length: column length over which the solute material is spread when reaching the end of the column. Thermospray injection: introduction of a liquid sample into the injector through a hot syringe needle causing the sample liquid to be sprayed at the needle exit; partially evaporated solvent acts as a propellant. Vapor overflow injection: solvent vapor leaves the vaporizing chamber by expansion as a result of its vapor pressure and because the volume of vapor exceeds that of the vaporiz­ ing chamber. Vaporizing chamber: volume in the liner of a vaporizing injector where sample evaporation is (or supposed to) take place. Wet needle injection: method for introducing very small volumes of liquid samples into va­ porizing injectors. The sample is withdrawn from the needle into the barrel ofthe syringe and the needle introduced into the injector without depressing the plunger (injection of the 30-80 nL of sample liquid coating the needle wall).

Subject Index

453

Subject Index (referring to the book only)

A Absorption in septum particles 95,425

Accuracy of sample volume 2, 57

Acoustic flow meters 159

Addition of standards, use of 10 ~L syringes

12 '

Adsorption

effect of contaminants 368

in syringe needle 45

in vaporizing chamber 129, 130, 132,

135,222,316

on glass wool 236

Adsorption suppressors 46

Aerosol formation 88, 115, 229

Air plug, solvent flush injection 32 33

Air, effect on column, solutes, ECO' 39

Air plug injection 23,37,448

Alcoholic beverages 248

Amino acids 174

Aqueous samples 365

Autosamplers 429, 448

fast autosamplers 4, 62, 232, 412, 429

B Back diffusion from vaporizing chamber 273

Backpressure regulator 152,435

Band broadening

avoidance 339,350,365

basic description 334,335,337,357

distinction of effects 334

in space 303,377,448

in time 334,448

Band formation 62,72,83,88, 116,414,448

Band widths in space and time 164

Baseline problems 344, 368

Batching oil, jute bags 289

Benzene in gasoline 212

Beveled needle tip 10

Bleed from ferrule of column attachment 345

Boiling points of some components 109

Brominated alkanes 128

Buffer volume in split outlet 201,420

Butt connector 448

c Calibration, problems of 209,224,237,316,

323

Capacities of standard capillary columns

150,448

Carbamate insecticides 126

Carbofrit 103

Carrier gas control

CE Instruments TRACE 439

HP 6890 437

Carrier gas flow rates

at different inlet pressures 302

dependence on oven temperature 156

split injection 183, 189

splitless transfer 295, 298

Carrier gas overnight? 142

Charcoal filters in the split outlet 201,422

439 '

Charcoal traps to avoid ghost peaks 350

Charge separation during thermos pray 114

Chemical stability of deactivation 367

Chlor~nated benzenes and biphenyls 393

Chlorinated pesticides in edible oils 391

Chloroalkanes 128

Chlorohydrin 127

Cholesterol 136.329

Christmas tree effect 172

ChromatoProbe 387

Cleaning

injector liners 143

454

Subject Index

syringes 12

Closure of the septum purge 278

Cognac 222

Cold trapping 336, 339, 375, 448

Column Flow Rate

at different inlet pressures 302

dependence on oven temperature 156

split injection 183, 189

splitless transfer 295, 298

Column installation 410

Column packing material for packed liners

103

Column temperature during injection 447

split injection 198, 204

splitless injection 341,351,375

Column-labile compounds 123

Comparison of injection techniques 126,251,

291,395

Concentrated samples 185

Concentration of vapor in the carrier gas 204

Concentration per component suitable

for split injection 163,445

for splitless injection 445

Concurrent solvent evaporation 448

Conditioning of columns 368

Conical style needles 10,84

Contaminants

acting as retaining stationary phase 132

evaporation from 19, 112,323

retention in injector 114

retention in precolumn 368

Contaminated carrier gas 347

Contamination

of column inlet 114, 229

of column outlet 344

Conversion of gas velocity into flow rates

160

Cool (cold) needle injection 23, 28, 448

Cooling of the injector by sample evapora­

tion 74

Cup or "Jennings" liner 96, 244

Cycloliner 98

D DC-200 simulating contaminants 323

DDT 136

Deactivation

by sample material 131,141,222,316

by silytation 138, 140

by stationary phase 140

of glass and quartz wool 133, 140

of liners 133, 138

of uncoated precolumns 367

Dead time 448

Deformation

of needle tip 84

of syringe plunger 7

Degradation of solutes

countermeasures against 125

in injector 128, 135,316

in injector or column? 122

in precolumn 368

in syringe needle 4, 31

mechanisms of 124

of endrin 299

testing of injector activity 134

Degraded contaminants from injector 265,

348

Deposition on surfaces 73, 121

Depression of the plunger 174

Detector block 345

Determination of

column flow rate 157

enzyme activity 174

injector activity 134

injector capacity 270

injector overflow 272

losses in the needle 24

split ratio 156

Deviation from the pre-set split ratio 202

Dew point ofthe gas/vapor mixture 108, 111,

200,354

Diameters of syringe needles 10

Diffusion in the injector 297,312

Diffusion speeds, non-linear splitting 214

Dilution in the injector 109, 118, 178,274,

408

Dimethoate 45

Dioctyl phthalate 375

Dioctyl sebacate 329

Direct injection 379,448

Direct sample introduction 386

Discrimination effects 425,448

split injection 154,213

splitless injection 267,315,392

syringe needle 3,26,41,42,47,51

Distortion of solvent peak 263,265,346,417,

419

Divinylcyclobutane 126

Drifting baseline 368

Duration of

pressure pulse for splitless transfer 302

solute evaporation 206

Subject Index solvent evaporation 75, 77, 105,283

splitless period 264,294,298,312,339,

447

Dynamic headspace analysis 176, 449

E Early vapor exit 289

ECDs, effect of air 39

Effect of injecting air 38

Ejection from syringe needle 18

Electronic flow and pressure regulation 279,

298,412,424

reaction to pressure wave 439

Electronic flow meter 159

Elution temperature 449

End face ofthe column 216

Endrin 134

Enhancing matrix effect 238,300,316,449

Evaluation of GC instruments 55 .

Evaporation from surfaces (packings) 96,

116,222,231,414

Evaporation in gas phase 87,107,226,414

Evaporation inside syringe needle 15, 446

Expansion of the sample vapor 195,269,421

External standard method 210, 328, 331,

449

F Fast autosampler 232,412,429

avoiding evaporation in syringe needle

4,62

resulting in band formation 62,83

Fast GC 172, 308, 409

examples 174

Fat as sample impurity 371

Fatty acid methyl esters 208, 245

Fatty acid silyl ester 136

Ferrule between liner and injector body 416

Fifteen degree rule 340, 449

Filled needle injection 22, 26, 449

Filters in the split outlet 201,422,439

Flash evaporation 87,226,449

Flavor components 238

Flooded zone 334, 353, 449

Flooding process 351,357

Flow meters

electronic 159

with floating particle 161

soap bubble meter 158

Flow/backpressure regulation 424

455

split injection 152, 197

splitless injection 263

regulation systems 435, 437

Fluctuating split ratio 207,220,247,449

Full solvent trapping 350, 449

Fused silica, cutting of 372

Fused silica wool, see "Glass wool

G Gaseous samples 251

Gas hold-up time 449

Gas regulation systems 151, 196,263,431,

437

Gas syringes for headspace analysis 10

Ghost peaks 384,418,449

from degraded sample by-products 237

from liner packing 140, 237

from septum material 265, 425

search for 344, 346,

Glass bead liner 98

Glass capillary columns 138

Glass frits 102

Glass wool 101,118,136,140

filtering out nebulized matrix 115

improving evaporation 112

split injection 235, 246, 249, 250

splitless injection 284,332

Goose neck liner 93, 95, 411

Graphite ferrules 418,419

H Headspace analysis 10,39, 176,251,377,

449

Heat

for solvent evaporation 71, 73

from liner wall 75

Heat capacity

of glass 74

of hydrogen 74

Heating ofthe injector 50,53,414,427

overnight and at weekends 142

Herbicide analysis 242

High-boiling sample solvents 36, 59, 239,

242,378

High-boiling samples 239

Historic background

split Injection 152

splitless injection 257

Homogenization of vapor across the liner

227,243

456

Subject Index

Hot needle injection 23,29,227,231,449

Hump eluted after the solvent peak 346

Hydroxylated PCBs 300

I

Incomplete sample evaporation 70,93,217,

200,236

Influence of

initial band width on final peak width

170

liner diameter on sample transfer 296

vapor viscosity on discrimination 230

Initial band 150

effect on peak width 169

length 449

shape, splitless injection 334, 338, 360

width 151,164,449

Injection of large volumes

splitless overflow 282

direct injection 382

Injection point 182

Injection speed 46

autosamplers 430

discrimination 28

fast analysis 173

splitless injection 280

Injector design 149,405,407

for splitless overflow 287

direct injection 379

Injector temperature 47,106,121,228,250,

322,446

Injector-internal headspace analysis 388

Injector-labile substances 122

Insert: also injector insert or liner 449

Internal diameter of vaporizing chamber 447

for split injection 181, 182,202,206,409

for splitless injection 273,296,309,312,

408

Internal standard method 209,329,331,449

Interpretations of "sample volume" 20

Inverted cup liner 97

Iodine experiment 117,292

Isokinetic splitting 215

J Jade valve 278

L

Laminar liner 99, 244

Leaching of silica 138

Leak through septum

after cooling of injector 56

dependence on needle style 10

effect depending on gas regulation 423

Length of

contaminated column inlet 370

flooded zone 361

initial bands 337

injector liner (vaporizing chamber) 408,

413

syringe needle 60, 446

split injection 167,183,190,202,

228,234,250

splitless injection 274,407,414

uncoated precolumn 367

Linear gas velocity 160,215,312

Linearity of splitting 213,248,450

Liner diameter, see "Internal diameter of

vaporizing chambers"

Liner types 450

baffles 96, 246

constriction at bottom 93,95,216

constriction at top ("goose neck") 277

cup 96,244,245

cycloliner 98

for injectors longer than 80 mm 410

glass bead 98

glass frit 102

laminar 99,244

Losses through septum purge 266

M

Manometers 434

Matrix effect 113,114,210,229,237,450

elution from the syringe needle 20

enhancing 238,300,316,449

in PTV injection 397

reducing 239,323,450

Matrix-induced chromatographic response

enhancement 450

Matrix-matching standard solutions 321

Matrix material as contaminants 114,368

Maximum peak height 167

Maximum sensitivity, split injection 176

Maximum tolerable initial band widths 169,

173

Measurement of

evaporation time 80

gas pressure inside injector 434

temperature drop 79

see also "Determination of.. ."

Subject Index Mechanical

flow/backpressure regulation 263,435

pressure regulation/flow restriction 262,

431

spray effect 84

Mechanisms of solute degradation 124,125

Memory effects 11,268,348,450

Merlin microseal 427

Message from standard deviations 193

Message to a lawyer 394

Metal liner 99

Metal surface at base of injector 93, 128

Methane peaks 168,337,339

Method of standard addition 177,209,331,

452

Micro-pentane extraction 375

Minimum flow rate, splitless injection 296

Minimum split flow rate 170

Mixing with carrier gas 74,227,244,276

Moving needle injection 385,450

Mustard oils 31, 127

N Narrow bore columns 172, 189,308

Nebulization of sample liquid 72, 87, 90

limits to 85

Needles, see "Syringe needles"

Needle

rinse injection 24, 450

dwell time 62

valve 262

Non-linearity of the response 130

Non-wetting samples 364, 365

o Observation of

initial band shapes 337

sample evaporation 81,352,359

On-column injection 360,376,381,395,446,

450

into detached column inlet 308

One-microliter syringes 9, 57, 188

Optimized split flow rate 177

Organochlorine pesticides 319

Organophosphorus pesticides 299,319,391

Overflow technique 282,450

Overheated column outlet 345

Overlapping chromatograms 175

Oxidized sample 39

Oxygenated dibenzothiophenes 127

457

p Packed liners 233

column packings 103, 140

see also "Glass wool"

Partial solvent trapping 61,353,376,377,

450

Partial vapor pressure 108

Particles attracted to liner wall 229

Particles driven into column 371

Peak broadening and distortion 334,361,368

effect of splitting on 169

for peaks eluted before solvent 242

solute degradation 123

splitting 450

Perylene 81,359

Pesticide analysis 299,319,372,387,391

Phase soaking 450

extra-retention times 356

Phase stripping 343, 353

Phenols 134

Phesil surfaces 139

Plugged syringe needles 14

Plunger guides 7

Plunger-in-barrel syringes 6

Plunger-in-needle syringes 9,57, 188

Plungers 6, 183

Pneumatic system

HP 6890 instrument 437

CE Instruments TRACE 439

for large injection by splitless overflow

287

Polyimide seal, press-fit 372

Polynuclear aromatic compounds 235,330

~2 ' Poor reproducibility 4,204, 219

of absolute peak areas 200

Position of column entrance 167,182,407,

410

Pre-evacuated injectors? 278

Pre-heating of carrier gas 186

Pre-peaks, hot needle injection 31

Pre-separation of solutes in the injector 221

Pre-set split ratio 194,450

Precolumn solvent splitting 289

Precolumns

uncoated 365,367

for contaminated samples 368

with thin film of stationary phase 372,

390

Press-fit connection 372,450

additional tightening 374

458

Subject Index

testing for tightness 373

Pressure and flow programming 412

Pressure increase, volumes around chamber

420

Pressure increase during splitless injection 279,298,412 Pressure pulse 279,298,412 Pressure regulator/flow restriction system 151,197,431,438

Pressure regulators 432

Pressure wave, split injection 195,205,220,

273,450

Prevention of aerosol formation 116

Priming ofthe injector and column 141,231,

320

Programmed temperature vaporizing (PTV)

injection 253,396, 445,450

Proportional valves 437

Purge exit 390

Purging injector after a splitless injection

263,265 Q

Quantitation on basis of pre-set split ratio 209

Quantitative results 394

obtained by split injection 153,251

obtained by splitless injection 313,391

R

Random deviations 193

Raw fused silica wool 136

Re-using press-fit connectors 373

Reconcentration of initial bands 259,334

by cold trapping 171,340

by solvent effects 171, 350

by bands broadened in space 366

Recondensation in the column inlet 450

split injection 198,220,249

splitless injection 261,275,351

Reducing matrix effect 239,323,450

Removable needles 11, 47

Removal of the contaminated inlet 368

Reproduction of absolute retention times

343,356 Reproducibility of

absolute peak areas 200

quantitative results 4,29,204,219,329,

332

Repulsion of liquid from hot surface 91

Resistance in split line 186

Retention gap technique 337,365,382,451

Retention power in the injector 132,222,316

Retention power of a surface 119

Retention times 433

influence of solvent effects 356

Reversed split flow 422

5

Sample clean-up 396

Sample evaporation 174

effects if incomplete 70,93,217, 2QO,

236

in the needle 2, 15,59

in the injector 63,69,93,200,217

Sample matrix, importance of 211,224

see also "Matrix effects"

Sample transfer into column 292,451

accelerated by pressure increase 298

accelerated by solvent recondensation

304

as aerosol 88, 115, 229

Sample volume 36,41,42,228,446

adjustment to liner volume 179

for gaseous samples 179

for splitless injection 269

Sandwich injection 23,37,46,451

Seal between liner and injector body? 416

Selection of the internal standard 327

Semivolatiles in cigarette smoke 376

Separation of liquid from needle tip 187,240

Septum 348, 423, 426

bleed 141,349,424

particles 94, 95, 424

thermostability of 55

tightening of 56,426

Septum purge 265,424,441,451

flow rates 265, 442

Shrinkable PTFE tubing 381

Silylating the syringe? 46

Silylation

of glass and quartz wool 133, 140

of liners 133, 138

of uncoated precolumns 367

reagents 12

Slow carrier gas: nitrogen 189

Slow injection? 23,28,280

Soap bubble flow meters 158

Solid injection 385,451

Solid phase micro extraction (SPME) 310

Subject Index Solute concentration in the injector 110

Solute evaporation 106

Solvent

damaging the column? 385

diverting column 289

evaporation 43,71,81,282,269

flush injection 23,32,451

peak, distortion of 263,265,346,417,

419

residues in pharmaceuticals 377

split injection 451

Solvent effects 261,350,375,451

at elevated column temperatures 378

on response factors? 276

Solvent recondensation

accentuated by pressure increase 303

split injection 198,220,249

splitless injection 261,275,351

split outlet 200

Solvent trapping 350,449,451

effect on retention times 356

mechanism 351

Sources of heat for solvent evaporation 73

Speed of evaporation, n-alkanes 120

Speed of sample liquid 72

SPI 451

Split

injection 149,171,451

outlet line 149, 200, 452

point 149,452

Split flow rate 150, 168,451

adjustment of 161

maximum 186

Split ratio 155, 177, 185,452

changes during injection 221

commonly applied 163

problems concerning 192

Splitless injection 35,257,445,451

Splitless period 451

Standard addition 177, 209, 331, 452

Static headspace analysis 176, 452

Stationary phase focusing 339

Stigmasterol in olive oil 326

Stop flow split injection 230

Syringe cleaners 13

Syringe needle handling 22

Syringe needles 9

attachment 47

for autosamplers 10

preheating 430

tip 10

with sideport hole 10, 183,240,389

459

Systematic errors 192, 193, 223, 225, 229,

317,329

T Temperature drop during injection 79

Temperature gradient

column attachment zone 358

towards septum 48, 52

Temperature profile through the injector 51,

428

Tenax TA, vapor overflow 285

Terminal band length 362, 452

Testing

completeness of sample transfer 306

inertness of injector 134

elution from syringe needle 24, 450

Tetrachloroethylene 211,212,375

Thermal capacity of syringe needle 85

Thermal conductivities of gases 76

Thermal focusing 339

Thermospray injection 84,87,107,121,227,

414,452

Thickness of needle wall 86

Thioglycollic acid 212

lightening septum cap 56,426

lightness of syringe plunger 7

lime for sample evaporation 71

Toxic substances, retention in split outlet 439

Transparent injector 82

Triazine herbicides 288

Triglycerides 44

air as carrier gas, 39

in the sample matrix 325, 329

passing through precolumn 371

True split ratio 194,452

u Uncoated pre-columns 365

as waste bin 368

Undiluted samples 185, 217

v

Valve to prevent backflow 278, 383

Vapor

concentration in the injector 179

overflow injection 282, 452

pressure of solutes 107

Variation of the split ratio 220

Video taping 81

460

Subject Index

Viscosity of sample vapor 201, 440 Viscous sample liquids 186 Visual observation of sample evaporation 81 Viton 420 Volatility of the solvent 44, 204 Volume around the vaporizing chamber 280, 420 of solvent vapor, calculated 269 of vaporizing chamber 181,273,311, 410,414

w Wax esters 371 Wet needle injection 23, 188,452 Wettability of liners 139 stationary phases 364 Williams distillate 249