~WI LEY-VCH
Konrad Grab
Split and Splitless Injection for Quantitative Gas Chromatography Concepts, Processes, . Pract...
54 downloads
1009 Views
20MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
~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.