A Series of ~ o ~ o g r a p h s Editor: JACK CAZES Cherry Hill, New Jersey
1. Dynamics of Chromatography, J. Calvin Giddings 2. Gas Chromatographic Analysis of Drugs and Pesticides, enj jam in J. Gudzinowicz 3. Principles of Adsorption Chromatography: The Separation of on ionic Organic Compounds, Lloyd R. Snyder 4. MulticomponentChromatography:Theory of lnte~erence,Friedrich He/fferich and Gerhard K/ein 5. ~uantitative Analysis by Gas Chromatography,Josef ~ o v a k 6. High-speed Liquid Chromatography, Peter M. Ra~csanyiand €/isabeth ndamentals of Integrated GC-MS (in threeparts), en jam in J. Gudzinowicz, Mic~ae/J. ~udzinowicz,and Horace F. arti in 8. Liquid Chromatography of Polymers and Related Materials,Jack Cazes 9. GLC and HPLC Determination of Therapeutic Agents (in three parts), Part 1 edited by Kiyoshi Tsuji and Walter Morozowich, Parts 2 and 3 e d i t e ~ by ~ i y o s hTsuji i 10. Biological/Biomedical Applications of Liquid Chromatography, edifed by Gerald L. Hawk 11. Chromatography in Petroleum Analysis, edi~edby Klaus H. A/tgelt and T. H. Gouw 12. Biological/BiomedicalApplications of Liquid Chromatography I I , edifed by Gerald L. Hawk 13. Liquid Chromatography of Polymers and RelatedMaterials I I , edited by Jack Cazes and ~ a v i eDelamare r 14. Introduction to AnalyticalGasChromatography:History,Principles, and Practice, John A. Perry 15. Applications of Glass Capillary Gas Chromatography, edited by Walter G. Jennings 16. Steroid Analysis by HPLC: Recent Appli~tions,edited by 17. Thin-Layer Chro~atography:Techniques and Applica and Joseph Sherma 18. ~iological/~iomedical Applications of Liquid Chromatography I l l , edifed by Gerald L. Hawk 19. Liquid Chromatography of Polymers and Related ~ a t e r i a ~Ill, s edifed by Jack Cazes iological/~iomedical A~plicationsof Liquid Chromatography, edited by Gerald L. Hawk atographicSeparation and E~ractionwithFoamed Plastics and rs, G. J. Moody and J. D. R. Thomas cal Pyrolysis: AComprehensjve Guide, Wi/liam J. in Chromatography Detectors,edi~edby ~ h o m a sM. Vick~ey 24. High-Pe~ormanceLiquid Chromatography in Forensic Chemistry, edited d John D. Wi~wer,Jr. quid ~hromatographyof Polymers, edi~edby ~ o s e~f ~ f f c ~ 26. HPLC Analysis of Biological Compounds: A Laboratory Guide, Hancock and James T. Sparrow
27. 28.
33.
36. 37. 38. 40 41
49. 50.
51.
52. 53.
56
*
60. Modern Chromatographic Analysis of Vitamins: Second Edition, edifed by Andre P. De Leeffheer,Wi//yE. lam be^, and Hans J. Ne/is 61, Preparative and Production Scale Chromatography,edifed by G. Ganefsos and P. E. Barker 62. Diode Array Detection in HPLC, edifed by Ludwig Huber and Stephan A. George 63. Handbook of Affinity Chromatography, edifed by Toni K/ine 64. Capillary ElectrophoresisTechnology, edited by N o r b e ~ o A. Guzman 65. Lipid Chromatographic Analysis,edifed by ~akayukjShibamoto 66. Thin-Layer Chromatography: Techniques and Applications, Third Edition, Revised and Expanded, Bernard Fried andJoseph Sherma 67. Liquid Chromatography for the Analyst, Raymond P. -W.Scoff 68. Centrifugal Partition Chromatography, edifed by A/ain P. Foucau/t 69. Handbook of Size Exclusion Chromatography,edited by Chi-San Wu 70. Techniques and Practice of Chromatography, Raymond P. W. Scoff 71.Handbook of Thin-LayerChromatography: Second Edition,Revisedand Expanded, edited by Joseph Sherma and BernardFried 72. Liquid Chromatography of Oligomers, Consfanfin V. Ug/ea 73.ChromatographicDetectors:Design,Function, and Operation, Raymond P. W. scoff 74.ChromatographicAnalysisofPharmaceuticals: Second Edition,Revised and Expanded, edjted by JohnA. Adamovics 75. Supercritical Fluid Chromatography with Packed Columns:Techniques and Applications,edifed by Klaus Anfon andC/aire Berger 76. Introduction to Analytical Gas Chromatography: Second Edition, Revised and Expanded, Raymond P. W. Scoff 77. Chromatographic Analysis of Environmentaland Food Toxicants, edifed by Takayuki Shibamoto 78. Handbook of HPLC, edifed by Hena Katz, Roy Eksfeen, Peter Schoenmakers, and Nei/~ j / / e r 79. Liquid Chromatography-Mass Spectrometry: Second Edition, Revised and Expanded, W. ~.A. Niessen 80. Capillary Electrophoresis of Proteins, Tim Wehr, Roberto ~Odrjguez-Djaz, and ~ i n g d eZhu 81. Thin-Layer Chromatography: Fourth Edition, Revised and Expanded, ~ e r n a r dFried and Joseph Sherma 82. Counfercurrent Chromafography, edited by Jean-Michel Menet and Didier Thiebaut 83. Micellar Liquid Chromatography, A/ain B e ~ h o dand Celia ~arcja-A/varezCoque
Modern Chromatographic Analysis of Vitamins,Third Edition, Revised and Expanded, edited by Andre P. De Leenheer, ~ i / /E.ylam be^, and Jan F. Van Bocxlaer
M A R C E L
MARCEL DEKKER, INC. D E K K E R
NEWYORK BASEL
This book is printed on acid-free paper.
Marcel Dekker, Inc. 270 Madison Avenue, New York, NY10016 tel: 2 12~696-9000;fax: 2 12-685-4540 Marcel Dekker AC Hutgasse 4, Postfach 8 12, CH-4001 Basel, Switzerland tel: 4 1-61-26I -8482; fax: 4 1-61-26 1-8896 llttp://www.dekker.conl The publisher offers discountson this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing atthe headqLlarters address above,
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, micro~lming,and recording, or by any information storage and retrieval system, without permission in writing from the pLlblisher. Current printing (last digit): 1 0 9 8 7 6 5 4 3 2 1
The ~ I software~was written ~ exclusive^ R for this~ volume by~ Jose' R a n ~ ~Torresn L a p a ~ i~~J, ~ i v e r sofi ~~ a ~ p l o nSpain. a,
When studying micelle-catalyzedpeptide and oligonucleotide formation in late 1974 and 1975, I had no idea that this research would eventually lead to a newbranch of separation science. In the course of these synthetic studies, we hadto separate a variety of products producedin the merization reactions. We subjected the solution containing our products to traditional liquid chromatography on columns up to two meters in length. We noticedthat the elution behavior of the standards we used was completely different when wechromatographed the standards alone as compared to the spiked reaction solution. It soon becameclear that the micellar "catalyst" in the reaction solution was altering the separation. Soon afterwards we found that we could actually use some forms of chromatography to measure the binding constants of various molecules to micelles. In retrospect, my decision to try using micellar solutions as mobile phases in LCwas a logical progression of this work, but at that time, the deliberate use of micelles in analytical chemistry was unheard of and was met with a good deal of skepticism. The first paper on the deliberate use of micelles as a mobile phase for LC was published in 1977. This was quickly followed by successful applications in TLC and HPLC and a theoretical treatment based on the micellar pseudophase. Micellar liquid chromatography (MLC) was about to become a novel part of the great HPLC boom that occurred from the late 1970s through the 1980s. In early July 1980, a session devotedto the analyticalapplications of micelles was given as part of the International Symposium on Solution Behavior of S u r f ~ c t ~ (in n t ~Potsdam, NY). It was a huge success, o V
with standing-room-only crowds. The 1981 Gordon Conference on Catalysis in Micellar and Macromolecular Systems (Wol%oro, NH) also devoted a session to micelles in chemical analysis. The co-organizer of that session was Willie L. inze, who first demonstrated the use of micelles in enhancing spectroscopic analysis. Also, by that time we were using cyclodextrins as beneficial pseudo-phases in separations. At about this same time, the useof separation-based techniques to measure association constants to micelles, cyclodextrins, etc.was also popularized. A largenumber o scientists began investigation^ involving micelles in separations. y the mid-1980~~ the journal ~ ~ a Z ~ t i c a 1 a separate review category for micellar liquid C ~ e ~ i ~had t restablished y chromatography. It is interesting to note that our work on MLC led directly to the use of cyclodextrins in chromatography (and the "chiral separatio~srevolution") as well as to their use in capillary electrop~oresis, The authors of this book are among the leading European scientists in the area of micellar separations research. 0thhavemade significant contributions to the field. Early onI was fortunate to meet AlainBerthod because of his work in MLC. As a result, wehave successfully collaborated on many projects for over one and a half decades. Looking back, two things are most g r a t i ~ i n gabout our development of micelle-based separations. The first is that micellar methods are widely used and have solved many problems for scientists and technicians. Second, I have hadthe opportunity to meet andinteract with many wonderful people all over the world as a result of our joint interest in micellar or pseudo-phase~basedseparations. This is a particularly opportune time for this book. The field is sufficient~ymature that one is able to examine it with a proper perspective and to see how it gave rise and spread into other areas of separation science and analytical chemistry. ~ i c e l l a r ~ i q u i d C ~ r ois~a a t o ~ r a ~ thorough scholarly and practical presentation of all areas of this special separation technique. It is likely that it will be the de~nitivereference volume in this area of research and technology for years to come. Daniel W. Armstrong Curator Profissor Head of the ~nalyticalDivision Department of chemist^ University of ~ i s s o u r ~ - R o ~ l a
Our main concern in starting this almost three-year work was to fill a gap: there was no dedicated book exploring thoroughly the theme of Micellar Liquid Chro~atography(MLC). Several reviews appeare~in different journals. Chapters dealt with MLC in different analytical c h e ~ i s t r ybooks. ever, it was necessary to have a handbook exclusively dedicat the MLC technique. This is the goal of this book. We tried to collect as much infor~ationas poss adding the chemical knowledge needed to develop it. exhaustively the literature for MLC articles up to early to cite every reference with its full title so that the reader has a good idea of the covered topic. We wrote for the specialist, but wetried to explain the d i f ~ c u ~ t points starting from the basics. We shared the work equally. One of us, being interested in the physicochemistry of the micellar media, the role ofthe stationary phase and the efficiency problem, prepared the chapters on these topics. The other, being anexpert in micellar partitioning and modeling and optimization of retention, wrote the corresponding chapters. We corrected each other’s work to obtain some h o m o ~ e n e i ~ . We think that the text will be usefulfor people wanting to start MLC analyses and also for people working in related fields such as the separation methods using micelles and capillary electro~horesis or electrochro~atography.Students specializing in analytical chemist^ will find useful information as well.
From left to right, Alain Berthod, Celia Garcia Alvarez-Coque and Daniel W. Armstrong at HPLC’99 in Granada (Spain), June 1999.
We would like to thank first Jack Cazes and Russell Dekker, who invited us to prepare this book and published it after waiting patiently throughout the writing process. During the progress of the book, we realized that the exposed algorithms, neededto optimize new appIications, were difficult to use. We then asked Professor Jose Rarncin Torres-Lapasici to develop software to facilitate the mathematical treatment. This gave birthto MICHROM, a software program included on a CD-ROM with this book. We cannot thank him enough for this great enhancement of our work. We thank Christelle Garon-Boucher and Laurent Veyre for scanning the literature for crnc and micellar partition coefficient values that allowed usto prepare the important lists of such data included as Appendices. We thank also Joe P.Foley for reviewing our chapter on modeling. We conclude this preface thanking the inventor of the MLC technique, Daniel W. Armstrong, for his continuous support for more than twelve years. Lyon (France) and Valencia (Spain), October 1999 Alain ~ e r ~ and ~ oCelia d GarciaA~varez-Co~ue
Foreword Preface S y ~ b oand l ~ Abbrevia~ions
1.
Presentation of the Book
How theAuthorsBecameInvolvedinMicellarLiquid Chromatography 11. The Technique 111. The Book References
V
vii xiii
1
I.
2.
PhysicochemicalProperties of MicellarMedia 1. 11.
Introduction Surfactant Molecules 111. Surfactant Solutions IV. SolubilizationinMicellarPhasesandMicroemulsions V. Conclusions References
3.
HistoricalDevelopment of MicellarLiquid chromatography I. Ion Pair Chromatography 11. 111.
The Birth of MicellarLiquidChromatography ModernMicellarLiquidChromatography References
9
57
I. Introduction 11. S t a t i o n a ~Phase Characterization 111, Surfactant ~dsorption IV.StationaryPhasesandSelectivity V.Equilibrationand Care oftheColumn References
5.
115 I. Introd~lction 11. The Three-phase Model 111. HybridMicellarMobilePhases IV.EvaluationofPartitionCoefficients V. The SolubilityLimitTheory VI.EffectofpHonRetention VII. Effect of Ionic Strength on Retention VIII. Thermodyna~icProperties Based on the Three-phase Model References
I. Introduct~on 11. ChromatographicProcessandEfficiency 111. ChromatographicProcessandMicellarPhases TV. RemediationofReducedEfficiency V. Conclusion References
173
03 1. Introduction 11. Elution Order Reversals 111. The Choice of Surfactant as a Means to Control Selectivity IV.SolubilityLimitTheoryandSelectivity V.EffectofOrganicModifiersontheElutionStrengthandSelectivity VI.RelationshipBetweenElutionStrengthandSelectivity VII.InfluenceofTemperatureonSelectivity References
24 1 I. Introduction 11. RetentionBehaviorinPureMicellarEluents 111. RetentionBehavior in HybridMicellarEluents at ConstantpH IV.PhysicochemicalMeaning of theEmpiricalModels V.SimultaneousEffect of pH,MicelleandOrganicModifier VI. Neural Networks VII. Factors Affecting the Prediction Capability of the Models VIII. Optimization of the Resolution IX. Easy Modeling and Optimization in MLC with the MIC References
Y
293
9. ~uantitationof I. Introduction 11. RetentionBehaviorofHomologousSeries 111. Quantitative Stru~ture-retentionRelationships IV. Quantitative Retentiol1"activity Relationships References
10. Anal~icalUse of Micellar Liquid Chromatography
343
I. Introduction 11. MicellarChromatography of ProteinsandEnzymeActivity 111. Controlof Pha~aceuticalPreparations IV. Miscellaneous Applications References
apid Analysis of Untreated Physiological Fluids I. Introduction 11. The DirectInjectionApproach 111. Use of MicellarMobilePhases IV. ~ackgroundSignaloftheMatrixFluid Application V. Development VI.Screeningof Illegal Drugs in Sport VII.PreviousSeparationoftheDrugs References
429
12. Enhanced Detection in Micellar Medium I. Introduction 11. Enhanced Fluorescence Detection 111. Sensitized Terbium Fluorescence IV.Micelle-stabilizedRo0m"temperaturePhosphorescence V. Inductively-coupled Plasma-mass Spectrometry VI. Amperometric Detection References
463
13. E x ~ a n d i nthe ~ Micellar Liquid Chro~atographyField I. Introduction 11. UseofOriginalMobilePhases 111. UseofOriginalStationaryPhases IV.MicellarPhasesandIons V.MicellarSeparationsWithout aClassic Column Conclusion VI. References
Appendix I:
M I C H ~ O MSoftware
Appendix 11:
Critical Micelle Concen~ationsof Selected S
501 ~
a 503c
~
~
xij
Appendix 111: Surfactant Association Coefficients
527
Appendix IV: How to Prepare a Ternary Phase Diagram
593
~
597
n
~
~
x
first constant of the Knox equation (flow anisotropy) concentration of free solute in bulk water concentration of solute associated to the micelle second constant of the Knoxequation (longitudinal diffusion) asymmetry factor measured at WO.1 H third constant of the Knox equation (mass transfer) concentration of an added salt to a micellar solution longitudinal diffusion contribution to plate height eddy diffusion contribution to plate height correlation factor for PAH calculated as [(number of secondary double bonds) + (numberofprimaryand carbon atoms) - 0.5 for a nonaromatic ring] mass transfer contribution to plate height alkyl chain lengthof a linear alcohol alkyl chain length ofthe tail of a surfactant molecule stationary phase masstransfer Contribution to plate height stagnant mobile phase mass transfer contribution to plate height solute diffusion coefficient in the mobile phase solute diffusion coefficient in the stationary phase stationary phase particle diameter Fisher coefficient
ICCSO k
AM AS
ieS
MD
Kws
L
fraction of mobile phase inaccessible to the solute fraction of stationary phase unavailable to the solute free energy of cavity formation Cibbs free energy of reaction height at the peak maximum standard enthalpy of transfer from mobile phase to stationary phase 50% inhibitory growth concentration of phenols in the culture of ~ ~ t r a hp y y~ r~ ~ ao ~ ~ ~ ~ retention factor (previously called capacity factor and noted k') retention factor at zero micelle concelltration relative variation in the concentration of solute in bulk water upon addition of a modifier solute-micelle equilibrium constant partition constant between stationary phase and water multiplied by the phase ratio CPC constant solute size depending calculated or predicted retention factor experimental retention factor constant of the ion-exchange equilibriu~at the micellesolution interface constant of the ion-exchange equilibrium at the stationary phase-solution interface relative variation in the concentration of solute in micelle upon addition of a modifier autoprotolyse constant of water (= 10-14at 20°C) relative variation in the concentration of solute in the stationary phase upon addition of a modifier retention factor with a mobile phase of 100% water micellar partition coefficient or solute-micelle distribution constant solute-stationary phase equilibrium constant average distance of a molecule from the wall of a FFF channel
ratio of the ma~imali~ed ien~h-to-breadth ofthe rectan enclosing the molecu~es protonation constant methionine o surfactantformingmicelles(totalconcentration surfactant minus cmc) r molecules perAOT surfactant l ~ o ~ e c u l e micelle a~gregation number(number of surfactant molecules in one micelle) number of carbon atoms in an homologue compound 1-octano~-waterpartit~ollcoefficient coefficient betweens t a t i o n a phase ~ and micelles partition coefficient betweenwaterandmicellesper surfactant molecule pa~itioncoefficient between water ands t a t i o n a ~phase -log of the acid dissociation constantK, amount of adsorbed surfactant quantitation of hydrophobici~index constant of perfect gases resolution factor collcentration of a solute in a micellar medium (=[ [AMI) elution strength parameter in hydro-organic mobile phases elution strength parameter in hybrid micellar systems entropy variation standard entropy of transfer from mobile phase to stationary phase solvent-related properties of solutes absolute temperature delay time (time before the gradient reachesthe top ofthe column) retention timeof a neutral solute moving withthe electroosmotic flow in CE gradient retention time
retention time of amicelle in MEKC dead time time at the peak maximum mobile phase linear velocity total volume of eluent needed to elute a given solute from the column pore volume in GPC stationary phase column dead volume solute retention volume (= V,) volume of the active surface of the stationary phase width of a FFF channel tryptophan width of the peak at 10% of peak height tyrosine
chromatographic selectivity (k2/kl,2 is the most retained solute) methylene selectivity selectivity of a carbonyl group binding selectivity to micelles separation factor or selectivitybetween solute i and solute it- 1 stationary-phase pa~itioningselectivity complexation co"nstant of a metallic ionwith a H2L cornplexing agent complexation constant of a metallic ionwith a HI, complexing agent obstruction factor in stagnant mobile phase obstruction factor in porous or granular material degree of counterion binding to micelles mole percentage of solute in the micelles conce~tratio~ of modifier
organicvolumepercentage in a micellarsolutionor microemulsion phase ratio (=Vs/Vo) thickness of the stationary phase useful layer total concentration of surfactant partial specific volumeof the monomers of surfactant in the micelle P for the R, substituent log , standard deviation linkedto peak width chromatographic variance reduced mobile phase velocity(= ud, /D,) stagnant mobile phase fraction S
ACN AES AOT AY BET BSA CE cmc CMX CTM CTAB C1,TAB CTAC DDTC DF DMA
acetonitrile atomic emission spectrometry sodium diethyl hexyl sulfosuccinate (also called Aerosol OT@) alanyl-tyrosine Brunauer, Emmett and Teller method (using gas adsorption to measure the surface area of a porous material) bovine serum albumin capillary electrophoresis critical micelle concentration (mol/L) cefmenoxime hemihydrochloride cefotiam dihydroch~oride c e ~ ~ ~ i m e t h y ~ a m morohexadecyl~~ethylammonium niu~ bromide tetradecyltri~ethylammoniumbromide hexadecyltrimethylammonium chloride diet~yldithiocarbamate aspa~l-phenylalanine dimethylarsenic acid
ii
NED
dodecyltrimethylammonium bromide dithiocarba~ate ethyl a~lthraqLlinone e~ectrochromatography ~vaporativelight scattering detector ~ a ionization ~ e detector phenyla1any~-phenylalanine field flow fractionation glycyl-leucyl-~rosine gel permationchro~atography(also called sizee ~ c ~ u s i o n chromatography) hydrogen bond acceptor h y d ~ o ~ ebond n donor height equivalent to a theoretical plate high-perfo~anceliquid chro~atography S,8-dideaza-isopt~royl-~-glutamyl-~-glutamic acid inductively coupled plasma massspectromet~ Internationa~Olympic Committee internal surface r~versed-phase limit of detection linear solvation energy relations~ip leucy~-~rosine leucyl-t~ptophan methionylaspa~l-hu~an growth hormone micellar elec~rokinetic chromato~rap~y maleic acid icellar liquid c~romatography ~onomethylarsonicacid micelle-stabilized room t e ~ p e r a t u r e p h o s p h o r i ~ e t ~ N-ace~~-~-cysteine N-( 1-naphtllyl)et~yienediaminedihydroch~oride nuclear magnetic resonance octadecyl-bonded silica 0-ptha~aldehyde
O/W PAH
RPLC SDS
SFC THPA TLC W/O
oil in wateremulsion or microemulsion (aqueous continuous phase) polycyclic aromatic hydrocarbon qua~titative retention-activi~ relationship quantitative structure-activi~ relatio~ship quantitative structure-retention relationship Reversed-Phase Liquid Chromatography sodium dodecyl sulfate (sodium lauryl sulfate) supercritical fluid chromatography tetrahep~lammoniumbromide thin layer chromatography water in oil emulsion or microemulsion (organic continuous phase)
erthod investigated the p micell~rsolution microemulsio~sbyusing electroch~ liquid c ~ o m ~ t o g r ~ p h y strong on micellar liquid chro erthod decided to study the ~ o d i ~ c a t i o no sf the station chromatogr~phiccolu
odmetmost
o f theresearchersinvolvedin
Upon returning to Europe, Berthod began work on the reduced efficiency problem in MLC[121. Garcia-Alvarez-Coque began workalso, on theinprovement ofderivatization reactions of organic compounds by the useofmicellarmedia 1131. In the early nineties, Dr. Laserna, another inefor~er'sgroup, involved both Berthed and Garcia-NvarezCoque in the analysis of drugs used illegally in sport [14, 151. At this tine, on Garcia-Alvarez-Coque also began to work onMLC. She wrote a review the solute-micelle and solute-stationary phase interactions [161. Afterwards, ~arcia-Alvarez-Coquebecame very interested in modeling the retention behavior ofsolutes 117, 181. Togetherwith Dr. Torres Lapasio, they prepared the software program~~C~~~~ for the development of MLC applications, that is included withthis book.
Micellar liquidchromato~raphyis an alternative to conventional reversedphase liquid chromatography ( P L C ) with aqueous-organic mobile phases. It joins the advmtages of rnicellar mediawith the separation capability of LC. Themobilephases are aqueoussolutionsofa surfactant at a concentration above the critical micellar concentration (cmc), that is, in a medium where micelles exist. The variety of possible interactions between solutes,micellesand stationary phasegivesa large versatility to this t e c ~ q u eand makes it appropriate for a wide range of solute analyses. ~ i ~ rofehydrop~lic s and hydrophobic compounds c m be separated in one run. This adap~bilityis perhaps the most important characteristic of MLC. Another advantage of the use of micellar solutions as mobile phases is the solubilization of nonpolar molecules. The necessary low amount of organic solvent used in micellar phases is very positive. It reduces the toxicity, ~ ~ a b i l i t y , e n ~ i r o impact ~ e n t aand l cost ofthese phases. Micellarmobilephases have posed,however,someserious problems, that have slowed the development and widespread useof in MLC than that First, the chromatographic efficiency is often much lower
1-
observed with similar colurnns in conventional RPLC. Secondly, the eluent strength of micellarsolutionscanbeextremelyweak. Fortunately, the addition of an organic modifier, such as an alcohol, can greatly remediate both of these problems. The position (retention) and shape (efficiency) of the chromatographic peaks depends on the nature and concentration of the surfactant and modifier, but it can also depend on the mobile phase pH, temperature and ionic strength. All these parameters can be modeled and optimized to provide an adequate separation of a mixture of solutes. The main use of the technique is in the analysis of physiological fluids. Micellar mobile phases are able to maintain proteins in solution. They allowthe direct injectionof biological samples inMLC systems. The protocol of the analysis is dramatically simplified.
. The pioneering workof Armstrong andthe early research performed in focused mainly on the study of the retention mechanisms and principles. ~lthoughsomeinteresting applications were also reportedfrom the beginning, most of the applied work has been done in recent years. The technique has reachedmaturity,andneedsareference organi~ingthe information given by the more than three hundredpapers, publishe~ date. to
It can be thoughtthat working on MLC just is mixing the su~actant with water, and flushmgthe solution throughthe column. As with any other technique,workingwithoutaminimumknowledgewillonly produce sapp pointing or misleading results. The use of micellar solutions, without special care, can easily damage the colurnn and even the chromatographic system. One of the purposes of this book is to expose the procedure to be followed andthe reasons for the problems that can develop. The knowledge of the physicochemical properties of the micellar phases is required in order to use them properly. This is outlined in the following chapter. The use of surfactants in chromatography was first implemented in ion-pair chromatography. This lead to the e m p l o ~ e nof t
in layer c ~ o ~ a t o g r development is reviewed in
to a property ofsurfactants is their ability at stationa~ phasesisdiscussedin several related topics. el e ~ p l a ~ i the n g beh~vi solutes inside the r this model was slightly modi~edand micelle-re~elle~ solutes [ZO] and hybri
re related to efficienc to remedy these problemsare descri~edin ~ h a p t e 6. r True ~icellarphases ( s u ~ a c t+~water) t can be consi~eredmobile phas~swithlowelution str forganic ~ o d i ~ e rmainly s, alco~olsare
to run the s o ~ w ~ r e . c o e ~ c i escale ~ t may not the bebest tool
chapters. ~~ectrophotometric detection has been utilized in most of the also interesting in ons, but other detection modes are Chapter13providesotheruses for organizedmedia, such as ~icroemulsionsor supercriticalfluids,inchromatographic separation. icellar electrokinetic c~omatographyhas attracted more a~entionthan other separatio iques using micellar phases. The connectionof this S also discussed. t e c ~ i ~ with ue odium dodecyl sulfate is the most commonly used s u r f a c t ~ in t ever, there are several thousand varieties of surfactants and can be used as well. ~ppendixI1 tabulates the physicochemical properties selected collection micellar of of partitioning coef~cientsgathered LC literature are also given in ppendix es 111. Fin of diagrams ofphase systems few re1 way to make one that cannot be found inthe literature.
3. 4.
reant, Micellar ~ ~ e cont s rotribu~lamine~mulsionand on ionic Surfactant Solution, Anal. Chem., 53: 1579 (1981). ges, ~ b t e n t i o nd'une ~mulsionde Fine et Mesure de la ~ a i l l ede ses Particules, J: Chim. Phys., 79: 373 (1982). Nome, partition in^ ~ e h a ~ i of o r~olutes bile Phasesin LC, Anal. Chem., 53: 1662 , strong, Application ofPseudophase L i ~ u i d Chromato~raphy:~ i ~ hSelective l y ~ o b i l e ases for present and ~ u t u r e Separa~ons,Am.Lab., 13: 14 (1
5.
6. 7. 8.
9. 10.
11.
12. 13.
14. 15.
A. Berthod, I. Girard and C. Gonnet, MLC. Adsorption Isotherms of two Ionic Surfactants on Jive Stationa~Phases, Anal. Chem., 58: 1356 (1986). A, Berthod, I. Girard and C. Gonnet, Additive Effects on Su~actant Adsorption and Ionic Solute Retentionin MLC, Anal. Chem.,58: 1362 (1986). D.W. Armstrong, T.Ward and A. Berthod, Micellar Effects on Molecular Diffusion: Theore~cal and Chromatographic Considerations,Anal. Chem., 58: 579 (1986). A. Berthod, I. Girard and C. Gonnet, S t a t i o n a ~Phase in U C : Surfactant Adsorp~onand Interaction with Ionic Solutes, ACS Symp.§er. ,342: 130 (1987), in Ordered Media in ~hemical Separations, edited byW.L. Hinze andD.W. Armstrong, American Chemical Society, Washington,D. C. A. Berthod, K.P.Li, T. Yuand J.D. Winefordner, A Simple, Versatile, Low Volume Fluorescence Detectorfor HPLC, Anal. Chem.,59: 1484 (1987). G.Ramis Ramos, M C . Garcia Alvarez-Coque, A.M.O'Reilly, I.M. Khasawneh and J.D. Winefordner, Paper-Substrate RoomTemperaturePhosphorimetry of Po~aromatic ~ydrocarbons Enhanced by Surface-ActiveAgents, Anal. Chem., 60: 4 16 ( l 988). G. Itarms Iiamos, M.C. Garcia Alvarez-Coque, A. Berthod andJ.D. Winefordner, ~luorescencein Microemulsions andReversed micelle^. A Review and New Results, Anal. Chim. Acta, 208: l (1988). A. Berthod and A. Roussel, The Role of the §tationar~Phase in MLC, J: Chromatogr., 449: 349 (1988). J.S. Esteve Romero, E.F. Sirno Alfonso, M.C. Garcia AlvarezCoqueand G. Ramis Rmos, Micellar enhance^ Spectrophotometric Determination of Organic Species, Trends Anal. Chem., 14: 29 (1995). erthod, J.M. Asensio andJ.J.Lasema, MLC for ~ a p i ~ Screening ofIllega1Drugs in Sport, J: Liq. Chromatogr~,12: 262 1 (1989). J. Sabater ~ o n t e s ~ oM. s , C. Garcia Alvarez-Coque, G. Ramis Ramos and J.J. Lasema,On the UseofMLC for Detection ofDrug Misuse in Sport, Quim. Anal. ,11: 163 (1992).
16.
17.
18.
19.
20.
21.
M. J.Medina Hemhdez and M C . Garcia Alvarez-Coque, Solute~ o b i l Phase e and Solute-Stationary Phase ~nterac~ons in Mz;C, Analyst, 117: 831 (1992). J.R. Torres Lapasio, R.M. Villanueva Camailas, J. M.C. GarciaAlv Mallols, M.J.Medina Hemimdez and Modeling of the Retention Behavior of Solutes in ~C with Organic Modl~ers,J ~hromatogr. ,639: 87 (1993). J.R.Torres Lapasio, R.M. Villanueva Camaiias, J. Mallols, M. J.Medina Hemhdez and M.C,Garcia AV Interpretive Strategyfor ~ptimization of Surfactant and Alcohol on cent ration in MLC,J. Chromato~r.A, 677: 239 (1994). M.F. Borgerding, F.H. Quina, W.L. Hinze, J. B o w e ~ a s t e rand H.M. McNair, Investigation of the Retention ~ e c h a n i s min on ionic MLC using an A l ~ l ~ e n ~ e n e ~ o m o Series, l o g o uAnal. s Chem., 60: 2520 (1988). P.Jandera and J. Fischer, Chromatographic Behavior in RPLC with Micellar and SubmicellarMobile Phases, JChromat~gr.A , 728: 279 (1996). M C Garcia Alvarez-Coque, J.R. Torres Lapasio and J.J. Baeza, Description of the Partitioning Behavior of Solutes and Data ~reatmentin Mz;C withModi~ers,Anal. Chim. Acta, 324: I63 (1996).
en the surfactant concentration is above a s p ~ c i value: ~ c the c~itical e l k concentratzo~(cmc), the surfactant solution ~ecomesa ~ i c e l l a r chromato~~aphy ( mic~lles, as mobile phases. systems is, thus, required t capabilities of this chromato~raphict e c ~ i q u e . e basic and s ~ p l i ~ e d b a c ~ ~ r o s y s t e ~ sisprovided. The rameters aEectin the cmc
was deli be rat^^ chosen to refer to review articles or books rather than to the manyori~inal ist of micellar
r~actant is a contractionof su y of this class of surfa~eor interfacial t~nsion. ents and a ~ ~ h i ~ h ifrom l e sthe
"philo," whch meanrespectively,bothandloving. molecules love bothpolar and nonpolarmedia. 11.I. ~
e
The amp~phile
~~ e sr c ~r ~ ~ i ~ ~
The ~ p h i p h i l i ccharacter of s u r f a c t ~molecules t is due to the association of twoparts with very differing polarities inside the same molecule[2]. One part is highly nonpolar, hydrophobic or lipophdic, usually an alkyl chain. Another part of the s u ~ a cmolecule ~ t is polar or hydrophilic. It canbe a nonionic chainwith polar groups, such as ether, alcohol or amine groups, or an ionic(anionic or cationic) group. Figure 2.1 shows the schematic representation of a s u r f a c t ~ tmolecule.Some s u r f a c ~ t shave two nonpolar tails or two polar heads,as illustrate^ in the figure. The natureof the sur~actantpolar head is used to classify the molecules.
Schematic representation of ionic surfactants (left) and nonionic surfactants (right). A-c~assicalsurfactants (e.g., SDS,CTAJ3, Brij); B-two-tailed surfacta~ts(e.g., AOT); C-two polar head surfactants (e.g., betaine, Pluronics, Tween 80).
According to the electrical charge of their polar head, the three main classes of surfactants are: anionic surfactants, cationic surfactants, and nonionic andor multisurfactants. A fourth class isaddedfortheamphoteric firnctional surfactants [33. a) Anionic ~ ~ r f u c t a n t s
Inaqueoussolution, the anionic surfactant dissociatesgiving an anion carrying themphiphilic properties and an inactive cation (e.g., Na" or h i o n i c surfactants are the mostcommonlyused active principlesin industrial and household detergent preparations. Soaps, sulfonated compounds, alkylsulfates and alkylphosphates are the four main anionic sufiactant fmilies. Table 2.1 Correspondence Between theHistorical and Normalized Names of Fatty Acids Whose Salts Make theSoaps
number
Nstorical name
IUPAC name
6 8 10
caproic acid caprylic acid capric acid
hexanoic acid octanoic acid decanoic acid
12 14 16 16:1
lauric acid myristic acid palmitic (cetic) acid palmltolelc acid
dodecanoic acid tetradecanoic acid hexadecanoic acid 9-hexadecenoic acid
18 18:1 18:2 18:3
stearic acid oleic acid linoleic acid linolenic acid
20 20:4 22 22: 1 24
arachidic acid arachidonic acid behepc acid qrucrc acid llgnocerlc acid
octadecanoic .acid. 9-octadecenoic acid, 9,12-octadecadienoic acid 9,12,15-octadecatrienoic acid elcosanoic acid 5,8,11,14~e~cqsatetranoic acid docosanoic ap;ld 13 -docosenolcacid tetracosanolc acid
saponification of natural oils and fats produces glycerol an fresh ashes is, probably, the is the reason whyfatty acids a ient n ~ e s dfrom i ~ e r ~ ~ rical names along with the whose sodi~mor potassium salts are the constituents of soaps. oaps are most oftena mi~tureof several s d u r n onified natural oil. roduces a soap whose compositionis oleate, 17% sodiumlinoleate, 6%
y connected to a carbon atom of the
,and the metal cation, The al~ylsulfonates r than 200 m. The ~ Z ~ Z ~ e ~ z e ~ e
d
cleans^^ a~ents.
LL s ~ ~ o s ~ c c have i ~ atwo ~ ehy~ophobic ~ chains, and
are salts of
~ a ~ l s u l f o s u c c i n aesters t e with the formula:
e ~ OT~(sodium ~ o~ ~2y ~~ esulfosuccinate) x y l is a sur [4]. form readily reverse micelles, in organic nonpolar solvents ~ulfosuccinatesabsorb slig~tly light at short w a v e l e n ~ s . the ester group is sensitive to h ~ ~ o l y s iso s ,they can only be used near in neutral solutions.
~
A Z ~ Z s ~ ~ The ~ ~salts e s . of sulhric acid ester form the alkyl sulfate surfactant family. They are produced by sulfation with sulfur trioxide of linear alcohols:
The al~lsulfuricacid formed is a strong acid, easilyneutralized by s o d i ~ or potassium hydroxide. The alkylsulfate aqueous solutionsare neutral and sis occurs in veryacidic solutions. O,Na is the most c o ~ o member n S physicochemical properties have been e~tensively investigated [3,4].
AZ
~
~
~ hisfamily ~ includes s ~
alkylesters of phosp
d
The alkylphosphates are low in aqueous solutions, even at extreme
~ the ~salts ~of thee mono ~i-
.
1
~
TH
t Anionic ~ ~ u ~ ar c t a n t sThe . hydrophilic part of many industrial
surfactant molecules associates an anionic group with a nonionic one [3]. These surfactants have the general behavior of anionic surfactants. Two examples ofsuch composite surfactants are the alkylethers~~ates:
and the carbo~methyletho~ylates~
h aqueous solutions, cationicsudactants are ionized ina cation canyng the amphip~licproperties,and an inactive anion, such as Cl- or Br-. The cationic group is most often a quaternary a ~ o ~ group u m [5]. The general formula is:
R’, Ry’and R,”groups can be identical e hydrophobic chain. The . The cetyl or hexadecyl trimethyl onium bromide C ~ ~ H 3 ~ - N ( c HBr3)~, is the cationicsurfactant most studied. Alkyl t r i ~ e t h y l a m ~salts o ~ iand ~~ d i a l ~ l d i m e t h y l a m m o salts n i ~ ~are non UV-absorbing sudactants, very stable in aqueous solutions.
A variety of cationic sudactants are derived from pyridine and imidazole, such as the a l ~ l ~ y r i d i n salts i ~ mandthe a l ~ l i~lda~olidi~ salts. These salts are very stable in aqueous solutions, but absorb whichmakesthem difficult to beusedin MLC. Also, surfactants associating a nonionic polyosyethylene chain with a cationic terminal group have been designed [5].
Nonionic surfactants are not salts, obviously, they do notgiveions in solution [6]. The hydrophilic part of their molecule containspolar groups
such as ether, alcohol, carbonyl or amino groups. Nonionic surfa~tants are stable in aqueous solution andare a little sensitive to the ionic stre are c o ~ e r c i a l l y water hardness. A variety ofnonionicsuxfactants available: 90% or more of these surfactants are obtained by polycondensation of ethyleneoxide. The alkyl ethuxylates:
can be referredto asC,E,, with m,the carbon number of the alkyl chain and n, the ethylene oxide number of the hydrophilic polyoxye~ylenechain. ese s u ~ a c ~are t snon absorbing molecules, usefir1 in MLC. e ethylene oxide pol~erizationstep in their synthesis?the alkyletho are mixtures of molecules with the same hydrophobic alkyl chain and a hydrop~licchain, having a different number of ethylene oxide units [6]. For example, the polyethylene 20 stearylether, marketed as rij@78, was found to contain only 30% of C18E20. It also contains 25% of Cl8E,1, 25% of C18E1g7 10% of a mixture of C18E22 and longer hydrophilicchains and 10% of a mixture of molecules with lessthan 19 ethylene oxideunits. It is possible to adjust both the hydrophobic chain length and the ethylene oxide condensation in the nonionic surfactant S properties of the final molecule can be tuned well to the needs. a lineofnonionic chemicalcompanieshavedeveloped suxfactants. A s an example, Table 2.2 lists the characteristics s u x f a c ~ t sfrom the IC1ChemicalCompany [6]. The physicochemical properties of solubili~ation, emulsification, wetting, cleansing and detergency, foamingor antifoaming, are very different amongthe members of the Brij farnily. The hydro~~lic-lipophilic balance value (HL as the ratio of the molecular weight the of hy~ophilicgroup to the molecular weight of the nonionic surfactant time 20, is used to sort the suxfactants in a 0-20 scale, from the more hydrophobic (low HLB value) to the more h y d r o ~ ~ l i Two c . surfactants with similar HLB values may have differing behavior, due to the very differing molecular weights. Furthermore, some batch to batch variation of the homologue distribution may occur for the same suxfactant. It must be pointedout that the formula given in Table 2.2 correspond to the average homologue molecule. The actual dist~butionin
ethylene oxideunits around the indicated value mayvary widely. It canbe noted that the CAS (ChemicalAbstract Service) number givento the CmE, chemicals is often relatedto the alkyl chain lengthonly.
Alkyl EthoxylateNonionic Surfactants of the BrijO Seriesa
30 35 52
C12E4 C12E23
10.2 17.2 5.6
52744-68-0 9002-92-0 9004-95-9
56 58 72
C16E10
13.2 16.0 5 .2
9004-95-9 9004-95-9 9005-00-9
12.7 15.5 5.2
9005-00-9 9005-00-9 9004-98-2
12.7 15.5
9004-98-2 9004-98-2
18.9 15.7
9005-00-9 9005-00-9
76 78 92
c1sE2
c16E20 C18E2 c18E10
c1sE20
C18:IE2
97 99 700 72 1
C18E100
C18E21
C,E, is the alkylethoxylate surfac~ntwith m ethylene oxideunits and n carbonatoms in the alkyl chain. c18 and cl,;,correspond tothe stearyl and oleyl hydrophobic groups, respectively.
a
The c o ~ o Z y ~are e r another ~ family of nonabsorbing nonionic sufiactants, in which the hydrophilic chain an is ethylene oxide polymer, and the hydrophobi~chain is a propylene oxide polymer.The two polymers can be connected in one or several points. For example, the PluronicO series dotte Chemicals Corporation are block copolymers withthe general formula:
Structure and Commercial Denomination of the Block Copolymer Nonionic Surfactants ofthe PluronicOSeries'
10%
20%
40% 30%
50% 80% 70% L35
L3 1
I900
1100
L42
L43
L44
L6 1
L62
L63
L64
2000
2200
l500
1700 2500
4800
2000
F6
8700
2900
L72
4100
2800
P85
L8 1
4500
2500
F77 6800
F87
7500
L101
4 700
3600
L121 4500
122
P123
P104 5400
0 14000
4600
103
F3
F108
105
16200
6500
F127
13300
b is the number of oxypropylene units (hydrophobic) the in surfactant molecule. The room temperature appearance ofthe surfactant is:L = liquid, P = paste, F = flakes. The numbers in italics under the codes are the surfactant average molecular weights.
a
y changing the oxyethylene (a) and oxypropylene (b) unit numbers, it is possible to build luronic homologues with a wide variety of properties. Table 2.3 lists the~olecularweights of the various embers of the Pl~ronic series. The m~ufacturercode includesa letter: L for liquid (low molecular weight), P for paste and F for flakes (solid with high molecular weight), and a two or three-digit number. The first or the first two digits correspondto approximately one fifth of the oxypropylene units in the molecule, and the last digit indicates the oxyethylene percentage in the molecule. For example, Pluronic F68 is a solid whose meanb value is about 6 .5= 30, and whose mole~ulecontains 80% of polyoxyethylene. The approximate mass of the lipophile pol~oxypropylenecentral block is1740. This corresponds to 20%
of the molecular weight ofF68, which is estimatedto be 1740/0.20= 8700, with 7000 (80%) as the mass ofthe two hydrophilic polyoxyethyleneblocks. The a value is close to 80. ther nonionic surfactant molecules includethe et~oxyzatedaZ@Z p~enoZs,which strongly absorb W light, and the e t ~ o ~ ~ afatty t e dacids, fatty esters andal~ano~amides, which are slightly absorbing nonionic su~actants[6]. The amine oxides, such as the alkyl dimethyl amineoxides:
are no~onic su~actants in basic and neutral solutions. They are protonated in acidic solutions and, thus, represent the transition to cationic su~actants.
Amphoteric su~actantsare ionic sudactants containing positive and negative charges on the same molecule [7]. They can be true amphoteric ions, such as the betaines:
with n r 1, that have a cationic nature in strongly acidic media and an amphoteric structure in neutral and basic solutions. These su~actantshave a mo~ileproton with a pKA valuearound 4. In s~~obetaines, the carboxylic a sulfonic group, -SO3- (pKA
-
The aminocarboxyZi~acids:
are ampholytes: they are cationic surfactants in strongly aci&c media, and anionic s u d a c t ~ t sin basic media, and possess an isoelectric point at an inte~ediatepH value, with an inner salt amphoteric structure. Natural
L-amino-acids are the building blocksof proteins and made a science theme by themselves. Other amphoteric surfactants include the betaines derived from imidazolines,which absorb W light and the lecithins or phosphati&cholines, whch are naturally occurring phospholipid§[7]. surfactants associating a nonionic moiety with a charged one are considered ionic surfactants.
~urfactantmolecules are water soluble and produce solutions with unique properties, due to the amphiphilic nature of the solubilized molecules. The two mainproperties of surfactant solutions are the a ~ s o ~ ~ t at i oany n interface andthe micelleformation. These are a consequence of polar and nonpolar interactions withthe sufiactant molecules.
a)
~ o ~ e c u1nteractlons ~ar
The interactions between moleculescan be sorted following an increasing polarity order as: (i) van der Waals forces, (ii) hydrogenbonding interactions, and (iii)electrostatic forces,
van der Waals Forces. These forces cover a variety of short distance range interactions, whose intensity decreases with the sixth power of the distance between two molecules. They include the London or dispersion forces between nonpolar molecules, the Debye forces between the p e ~ a n e n t dipole of a molecule and the induced dipole of another molecule, and the Keesom forces or dipole-dipole interactions between two polar molecules. The order of magnitude of van der Waals forces is 2 kJ/mol for polar and moderately polar molecules, and less than 1 kJ/mol for nonpolar molecules
[SI
*
e notion o f ~ y ~ ~ oint~~action ~ ~ o ~ was i c we1 r solute is dissolved in solute tends to locally restrict the motion of water molecules. water molecu~esis associate^ with the is entropy increase is responsible for the surface aq~eoussolution [g]. a c t i v i ~and micelle f o ~ a t i o of n s~rfactantmolecules.
ilic character of surfactant molecules e~plains interface. Two phases of different pol larity difference attracts the surfactant molecules because this can ze the entropychangeby putt in^ their polar r phase and their nonpolar part in the less polar phase. surfactant molecule arrangementat the liqui~-liqui The adso~tionof surfacta~tmolecules inter~acialtension. The decrease ofthe water-ai prope~y.The addition of a sur ers emulsion ~ o ~ a t i possible on liquid i~terfacialtension. etting and detergency are two imp0
properties of s u ~ a csolutions ~ t dueto surEactant adsorption at S e decrease of surface tension bysu~actantadsorp In ~ a t ~ r - ainterface ir depends on the s u r f a c ~ nconcentration. t s u ~ a c tadsorbs ~t on the s ~ t i phase o ~ c h~ a n ~ i nits~ nature and p o l a ~ ~ .
re Surfactant adsorption at interfaces. Left, liquid-liquid interface, the hydrophobic tails points towardthe less polar liquidphase. Right, air-water interface.
The second most important property of sudactant solutions is the micelle fo~ation.
Figure 2.3 shows that the physicochemical properties of a surfactant solution, such as the surface tension, undergo an abrupt change over a narrow concentration range. This concentration is called the critical micelle c o n c e n t ~ ~ t ~(cmc) o n of the surfactant. Above this concentration, micelles are present. Figure 2.4 shows a typical micelle for an ionic su~actant[101. The su~actantmolecules orientate in such a way that their polar heads face the water phase and their nonpolar tails aggregate in the waterless micelle core.
P5
Figure 2.3 Physicochemical changes related to micelle formation at the critical micelle concentration(cmc).Density andrefractive indexare two parametersalso affected by micellization (not shown).
~"--"-"""""
"
-I_""."
"
e
-"
X to 40 n
Figure 2.4 Representation ofan ionicmicelleaccordingto Hartley [lo]. The GouyChapman layer corresponds to the diffuse layer under the electric fielddue to the micelle. The Stern layer co~espondsto the ionic groups.
b) A Very S
~~
~o
~~
~e
e~
Using a closed association model, micelle formationcan be representedby:
in which N is the uggregution number or number of s u ~ a c tmolecules, ~t S, in the micelle aggregate. The equilibrium constant corresponding to Eq. 2.1 is:
G
~ t r o d u c the ~ gsurfactant concentration, C, and the percentage ofs u r f a c ~ t molecule forming micelles,"Prn,the following is obtained: K =
Figure 2.5 shows the valueof for bfferent sodium concentrations. At a critical c entration, the cmc, sharply. The cmc can be calculated as:
C . -
Micelle andmonomerproportionrelated to thesurfactant (SDS) concentration. Left ordinate: Amountof surfactant molecules in micelles (Q in Eq. 2.3).Right ordinate: monomer surfactant concentration.
micelle, andthe cmc is 8.2 the theoretical value molecule§ addedab
mc value). The micelle concentration was then only a tenfold increasein the overallsu~actantconcentration
verysimple, the closed association modelgives surfactant micellization.
an approximation to
The free energy of micellef o ~ a t i o n , C", is expressed by:
is the gas constant and T is the absolute temperature. Using the chemical potential, yo,the law of mass action is:
which in the subscripts stand for s u r f a c t ~ tmonomer an surfac~ntin micelles, respectively. From Eq. 2.4, it can be derived:
qs, I!.5 and 2.7 and notingthat the agg~egationnumber, most cases higher than 20 gives:
Finally, the entropy change,
O,
due to the micelle formation is:
S" = d A ~ / d T= R In cmc +
(2.9)
Increases in the cmc with temperature are often observed experimentally, which means that the AS" term is most oftenpositive. icellar solutions are sometimes called ordered media [121. The chemical order in a micellar solution seemsto be greater than in a classical solution. Equation 2.9 shows that the micellization ofsurfactant molecules obeys the second principle oft h e r m o d y n ~ c s .It seems that the surfactant h y d r o ~ b o nc h s have a much higher freedom of motion inside the micelle core than in the water bulk [131. The micelle structure minimizes the molecule energy. The large entropy increase of water molecules associated with the removal of nonpolar surfactant tails from the aqueous solution (hydrophobic eEect) is the main micelle driving force. Electrostatic forces tend to separate the polar heads that bear the same charge. The whole micelle is an e~uili~rium between these forces. This equilibrium is very sensitive to any chemical additive orparameter that can act on any of the forces, such as: salts, polar or nonpolar solutes,temperature and/or pressure.
The micellestructure is dynamic. The whole micelle aggregate has a finite mean lifetime.Fu~hermore,a given micelle existingat a given time, rapidly exchanges some surfactant monomers with the neighboring solution [141, The dynamic character of micellesis characterized by two relaxation times, as illustrated by Figure 2.6 [l51. The firstrelaxation time, r ~ g i n fiom g lO-* S to lom4 S , corresponds to a rapid monomer exchange between the micellar aggregate and the bulls solution. It can be considered as the mean lifetime of an individual surfactant moleculein the micelle [l S]. The second relaxation time, in the range of l 0-3S to -1 S, measures the lifetime of the aggregate itself. A decreasein sizeisfirst observed,followedby the vanishing of the aggregate, while other s u ~ a cmolecules ~ t are aggregating to form a new micelle (Figure 2.6).
FS
Figure 2.6 Dynamic exchange inmicelles. Slow process (msto S): micelle formation from monomers tomicelle through intermediateaggregates. Rapid exchange (in the ps range): exit and entrance of individualsurfactant molecules.
The micelle core is nonpolar. It contains somewater molecules [161, but its overall polarity is much lowerthan thatof water. The micelle core can be, thus, compared to a nonaqueous phase. However,due to the dynamic micelle ~ u c ~ a t i o nits ,is not exactly a phase. The micellar phase is often o - ~ it~ isunot ~ possible e , to separate it fiom the called a ~ ~ e ~ ~ because aqueous phase. When the aqueous phase is removedbydrying or evaporation, the crystalline form of the surfactant is obtained. This form is not identical to the micellar pseudo-phase, which is a hydrated form of sudactant molecules.
In a micellar solution,it is convenient to view the volume occupied by the surfactant molecules associated inmicelles as a phase. From a
practical point of view, the volume percentage ofthe micellar phase,qm,is Vm
-
= (C cmc) V
(2.10)
where v is the sur~actant~oZar v o Z ~ in ~ L/mol, e or volume occupied by one mole of s u r f a c ~in t the a ueous solution. For example, in a 0.1 ,v = 0.246 L/mol), 2.26% ofthe volu solution(cmc = 8.2 qm= 0.0226 (Eq. 2.10). The aque micellarpseudo-pce occupies the remaining 97.74% of the liquid volume, correspo qm.The q p ,value is usedto obtain the aqueous and micellar pseudo-p~ase volumes, 'va and Vm,respectively:
and
is the actual volume ofthe cellar solution. evaluate the solute and Vmvolumes are us ses, the solute micellar p m coe~cient, also call micellar ~ i ~ ~co~stant, ~ i and ~ ~ the ~ solute o local n concentr micellar or aqueous phase.
e micelle size depends onthe aggregation number. cture of micelles should be kept in mind. There is size due to the endless micelle creationldest size distribution curve for the surfactant solution is illustrated in Figure 2.7. ~ZceZZesare, in fact, an a~imateddist true single monomers;4 . 5 ~0-3 1 M (56%) are ihmers, trimers nts of higher aggregates (Fi~ure2.7). The average micelle aggregation number at 20°C and 0.1 surfactant concentration is
ED1
62, although there is, a c ~ a l l ya, Gaussian distribution centered onthe value
F i ~ ~2.7 re The distribution curve of theaggregatescontained ina 0.1 M SDS solution at 20°C. The SDS micelle aggregationnumber varies from 50 to 80 with an average value of 62.
The micelle shape dependson thesurfactant concentration. Spherical micelles(Figure 2.4) are found at a sudactant concentration near the cmc. The average diameter of SDS micelles is about 2 nm or -2OA in concentrations near the cmc. As the sudactant concentration increases, the micelle size andshape can change. Oblong, rod-lke or cylindrical micelles may be formed. The length and breadth of the rods or cylinders are in the low nm range (few tens of A). At high surfactant concentration, a viscous middle phase, a hexagonal structure andor a neat phase with a liquid crystal structure are observed with some surfactants [171. Figure 2.8 shows the molecular a~angementin some possiblestructures thatmay be found[181.
THOD an
It is notpossible to useviscoussolutions as mobile phases in liquid . Only relatively diluted micellar solutions with spherical chromato~raphy micelles will beuseful.
Possible physicochemical structures observed whenthesurfactant concentration in water increases. The lamellar and hexagonal structures have liquid crystal properties (from[1S]).
As s h o w in Figure 2.3, the particular concentration at which numerous properties of the surfactant solution change, is the cmc of the s u d a ~ ~ t . This value is very importantto h o w . The micelle is formedby an equilibrium betweenthe hydrophobic eRect, due to the lipophilic part of the sudactant molecule, and the polar forces of the polar part of the molecule.
This equilibrium is sensitive to numerous parameters, and any change in the surfactant or e ~ p e ~ m e n tconditions, al willaEect the cmcvalue as described below.
a) Su~actantS~uctureand the cmc The first parameter acting on the cmc value is the s u ~ a c t a n t s t ~ c t A u~e. longer hy~ophobicchain. will renderthe surfactant more hydrophobic and promote micelle f o ~ a t i o n .For the homologous surfactant series, the cmc can be relatedto the alkyl chain lengthby: lncmc=a-bn,
(2.13)
in whch a and b are constants, and is the number of carbon atoms in the alkyl chain. Table 2.4 lists the regression parameters for Eq. 2.13, fitted with the crnc data obtained from members of the three main classes of s u r f a c ~ t s .Table 2.5 gives the cmcs of the three homologous series. The polar head nature affects the cmc. The nonionic surfactants have much lower cmc valuesthan ionic surfactants with similarlipophilictail. The cmc value also decreasesrapidlywith the numberof carbon atoms in the nonpolar alkylchain. It can be seenthat the cmc of the two ionic s u r ~ a c ~ t homologues decreasesto a half when the nonpolar alkyl chain increases by one m ~ y l e n unit. e With one more methyleneunit, the cmc of the n o n i o ~ c surfactant homologues is decreased bya factor hgher than 3.Therefore, the cmc ofa non-ionic surfactant is two or three orders of maptude lower than the cmc ofthe ionic surfactant with a similar alkyltail. Regression Parameters of the In (cmc) vs. n, Curvesa
Homologous series
a
aa
ba
r2a
3. lb 4.3 4.8
0.65b 0.70 1.21
0.996 0.998 0.999
a = intercept, b = slope, r = regression coefficient. 4OoC,other values are for 25".
LV Experimentai Values of cmc(n")for Homologous Series of Surfactants
18
atomsCarbon
C,SO,
Na+
r-
8 16
10 14
12
140
33
8.6 0.23 0.582.2
300
69
16
3.7
0.9
0.3
C, corresponds to an alkyl chain withn carbon atoms; CnEmis the alkylethoxylate surfactant of the anionic alkyl with m ethylene oxide units andn carbon atoms in the alkyl chain. The cmc sulfate surfactants are given at 40°C [19]. Thecmcof the cationic [20] and nonionic surfac~nts[21] are given at 25°C. a Values estimated using the regression parameters of Eq. 13 given in Table 2.4 due topoor water solubility.
Values of cmc( m w for Different n-Dodecyl Surfactants at 25°C [22]
20
crnc
8.2 1.3"
crne
15
4 3 0Polar
10
12b
-E3
head
crnc
0.058
0.080
At 54°C Potassium tridecanoate or C,,-COOK c Dodecylsultaine Dodecylbetaine.
a
l .2
1.5
Table 2.6 compares the cmc values of a n u ~ b e r surfactants. It shows that the number of charges on the polar he p a r ~ e t emore r import than the electrical sign (positiveor counterion is e~ually~ p oby ~ its n~mber t o~charges. especially, nonionic su~actants have lower cmc v c o r r e s p o n ~ gcharged su~actants. b) ~ ~ eofc~ te ~ ~ e r a t u r e
e forces involved with the polar heads of ionic surfa ar or hydrogenbonding fo forces very diEerent from rature acts very di~erently with n o n i o ~ csurfactants. types of forces, therefore, the effect of temperature on s h o u l ~be studied separately from the effect on nonionic
Igure 2.9 shows the variation in. solubility with t
~ o zwas ~ t described by
ionic solided
surfactant i
later found that
,the micelle structureis an e~uilibriu lar heads andthe a~ractionforces b
chains. The tempera~remay affect differently thesetwo comp~ing forces. "h effect of temperature on the cmc values is often complex. At tempera~reshigher than the KraB point, the cmcof most ionicsudactants increases slightly.
0
Temperature dependanceof the SDS solubility. The lSraflEt point of SDS
is close to 15°C. The Kraffl boundary plateauis around 18°C.
u cmc mvalue can esist. This is the case of S cmc is at 29°C and increasesat hgher temperatures, as illustrat~~ by Figure 2.10. A 20% increase in the cmcof SDS is noted between 30°C and 60°C. The cmc of d o d e c y l ~ chloride, o ~ ~ a at 25°C (Table 2.6)and 1.7 X10m3 examples givean idea of the magnitudeof the effect of temperature on the cmc of ionic s u ~ a c t a n ~ .
A~
~~
.7 %am Point ("C) of Some Surfactants 1131.
JSraffk point
8
15
30
45
Surfactant
Sodium dodecanoate
Sodium tetradecanoate
Potassium tetradecanoate
Calcium tetradecanoate
36
53
8
>loo"
"."."....*.....~..."."*"~~."*~."..~."..~"*..~".*.*..~.~*."..~~~.*"*.~~.....".".."..~~"~~......."..~
IG-affk noint
The high Kram point of calcium surfactant salts is responsible for the precipitation of calcium soaps in hard waters .
35
~
O
n
1
T~mp~ratu (oC) r~ 2.10 Temperature dependance of some surfactant cmcs. Thick line and left ordinate: SDS. Thin lines and right ordinate: polyoxyethylene octylphenol with two ethylene oside units (E2), five units (ES)and nine units @g).
~ ~ ~ - ~z~ ~r ~f z~ c c These ~ ~ surfactants ~ ~ s . donot present the
wever, a nonionic micellar solution becomes turb ases when the temperature israise , Ths is the c Z o ~ ~ i n ~ polyoxyethylene chain,the polar art of most on-ionic s u r f a c ~ t s is , progressively dehydrated as the temperature raises. Losing water molecules, the polyoxyethylene chain becomes less polar and, at a particular temperature, a turbidity, the clouding, appears. n t nonionic surfactant solution. is called the c Z o ~ ~ ~ oofithe S e nonionic rnicellar solution onionic surfacta~t,and an containing the major part of the su~actant.
T
Cloud Point in "C
of SornePluronic@Nonionic
Surfactants."
?h ~ u r f a c ~Code nt
L3 l L6 1
6
Ethylene oxide unitsb
Solution Solution
4
37 24 20 1 1
29
5.4 8
12 77 2
22 40 46 50 74
82 5
1
?h
17 16 11 10 0 2 6 6 90
Surfactants are described in Table 2.3. In Phonic@ copolymers, the number of ethylene oxide units is referred as 2a, e.g. L31 contains 8 ethylene oxide units.
The cloudpoint temperature dependson the concentrati~nof surfactant. Itincreases si~ificantlywith the polyoxyethylene chain length, but scarcely depends on the polar chain length. As an example, the cloud points of some nonionic Pluronic surfactants (see Table 2.3) are listedin Table 2.8. The nonionic micelle size and aggregation number increases dramatically with temperature. It can be stated that the aggrega~onnumber becomes idlnite at the cloudxng temperature. As illustrated by Figure 2.10 (polyoxyethyleneoctylphenol surfactants with 2 (E2), 5 (E5) and 9 (E9) oxyethylene units), the cmc of a minimumvalue for a specific nonionic su~actantsmaypresent tempera~re. Often the increaseof temperature produces a continuous increase of cmc (E2 on Figure 2.10). h all cases, the cmc variations with temperature are on the order of 1% per "Cor less.
c) E ~ e cotf Added E ~ e c t r o ~ t e s The cmc of ionic surfactants is always depressed bythe presence o f a d ~ ~ d electrolytes. Shinoda [24] proposed the semi-em~iricalequation: / kT -A6/z In (cmc +
(2.14)
in which is the carbon number of the surfactant alkyl chain, free energy difference per methylene group between the monome~cand micellar state, 6 is the degree of counterion bindingto the micelle, z is the valency of the counterion, C,, is the added electrolyteconcentration, and A and B are constants. This equation was proposed in a similar form by other authors [25, 261, q.2.14 shows that the cmcofionic surfactants decreases e~ponentiallywith the concentration of any added electrolyte. It has been shown that the nature of the ion with the same charge as the ionic surfactant is not critical, i.e., sodium chloride, sodium sulfate or sodium phosphate similarly decrease the crnc of sodium dodecyl sul potassium or cesium chloride would change the counteri which wouldproduce a different slo e in the In (cmc) versus
NaI, and to 2.1 X lo-, M (a 75% decrease at 40°C), when 0.03 M CsClis added [22]. The cmc of nonionic surfactants is also decreased by added salts
[l l]. The effectiveness ofsalts in alteringthe cmc of nonionic s u r f a c ~ t s follows, approximately, the lyotropic series [111, whch is for anions:
and for cations:
The magni~deof the effect is, however, much lowerfor nonionic than for ionic s u r f a c ~ t s .For example,the cmc of C,4>E,,is 1.l X lom4 M at 25°C. The addition of 0.03 M NaCl decreases the cmc by5%, that is,only to 1'05 . It startsto s i g ~ f i c ~ t ldecrease y when the added salt on cent ration is above 0.1 M. Thus, to obtain a 55% decrease in the crnc, it is necessa to add as much as 0.86 M NaCl[22] (cmc = 5.5 XlO-' M). 0, or NaSCN, or 0.43 M Na2S04 or CaNO, are added, becomes 5.5 M, 8.5 XlO-, M, 2.3XlO-'Mand '7.6 The depression ofthe cmc by added salts is the result of a decrease in the repulsive forces between the ionic headgroups of the surfactant molecules. The micellization becomeseasier, since the hydrophobic effect on the non-polar chains is little modified or slightly enhanced (salting out effect) by the added electrolytes[28]. Also, the aggregation number of ionic micellesenlarges dramatically whichmaychange the micellarphase viscosity. The effect of salts on the cmc of a surfactant should be kept in mind when usingMLC. It is often necessaryto buffer the micellar phases using electrolytes.
d) ~
~of S eo ~ ~ c e n~~ s
~n~~ress~re
The addition ofan organic modifier is common in MLC, however, this can change the cmcvalue ofthe s u r f a c t ~used. t It is also important to consider the effect of pressure on the micellar state of surfactants, since pressure is a driving parameter in liquidc~omatography.
S OF M I C E L MEDIA ~~
Table 2.9 Effect of Short Chain Alcohols and of Pressure on the cmc of SDS and Decyl Trimethyl A~moniumBromide (C,,TAB) at 25°C [43]"
cmc
Alcohol V h
M
~~~
Mole fraction
%
Me~anol
0.45 0.57 0.14 3.6 0.88 7.7 1.9 4.013.2 0.122 20.0 5.9
0.9
%
(M)
0.0025 16.2 23.9
0.008
0 079
0.063 0.059 0.068
Ethanol 7.0 13.4
l .2 2.3 3.1 0.010.101 4.6 22.4
1
18 27
14.8
0.045 0.064
0.006
0.056 0.055 0.060
n-Propanol 3.2
14.3
0.43 7.0 0 -94 1.5 17.2 2.3
l
5.7 0.03 11.2 0.0013 0.030 9.2
0.018 0.039 0.0038 0.0006
Pressure
l 0.019
cmc ( ~ )
kg/cm2
p.s.i.
SDS
500 l000
7000 14000
0.0090 0.0094
ClOT 0.067 0,067
"The crnc of SDS and C,,TAl3 in pure water are0.0082 M and 0.068 M,respectively, at25°C and atmospheric pressure.
~uzvents.Short chain alcohols decreasethe cmc of ionicsurfactants at low concentration, but a cmc increase is also oRen observed for methanol and ethanol at larger concentrations (Table 2.9). Longer alcohols decrease the cmc, as illustrated by Figure 2.11. It has been observedthat the longer the alkyl chainof the linearalcohol, the higher the eEect on the cmc of potassium dodecanoate 129,431. Dioxane, the solventwith the lowest polarity and still fully miscible with water, decreases the solution average dielectric constant. It was generally foundto increase somewhat the cmc of surfactants. For example, the cmc of S 50%) in a 10% w/w and 25% w/w dioxane solution, respectively, at 2 122). Urea is a compound known for m o ~ ~ i the ng liquid water structure and for increasing the solution dielectricconstant [l l]. slightlyincreases th c ofionic surfac~nts. S is unaffectedby 2 rea andbecomes 9.0 urea solution at 25°C 1221. The cmc of dodec romide is 1.5X 10-2M in pure water, it becomes 2. (+200%), in 2 M and 4 M urea solutions, respectively [22]. ~ r e s s ~ rTable e . 2.9 also lists the eEect of pressure on the cmc of . Extreme pressures (2000 kg/cm2 or 28,500 p.s.i.) produce only small changes ( 4 0 % ) in the cmcs. It can be considered that the pressure used inclassical liquid chromatography,in the 40-200 kgjcm2range, will not change the cmc ofthe surfactant used. e) ~eterm~natlon of cmc V ~ ~ u e ~ Figure 2.3 shows a variety of physicochemical changes of the surfactant solution occurring when micelles form. The measurement of any one of these properties can be usedto determine accurately the cmc ofa s u r f a c ~ t . ~ ~ ens ion ~ ~ e ~ s ~ cr e ~The ee nmost t , commonly used cmc determination method is the surface tension measurement [22]. Figure 2.2 shows that the surfactant molecules orientate at the solution air-water i n t e ~ a ~ e . This surfactantadso~tiondecreases the surfacetension. The magnitude of the tension decrease depends onthe free monomer surfactant concentration. As shownby Figure 2.5,the free surfactant monomer concentration reaches a plateau for surfactant concentrations above the cmc. As the surfactant concentrationisincreased in the aqueous solution, the surface tension
1 EEect of normal alcohols on the CMC of potassium dodecanoate (25OC). Micelles does not form in concentrated alcohol solution(3M propanol 23% v/v; 3M butanol = 27.5% v/v). Milky emulsion formation occurs with very low amounts of long chain alcohols (from [29]). =I
(0.0727~Nlrn), the surface te to few dyneslcm (l0” Nlm) when the crnc is e~trapolationof the decreasingpart of the surface tension log ( s u r f a c t ~concentration) t plot (below the c (~bovethe crnc) gives the crnc v
conductivity can be used for cmc dete~ination[22]. Conductivity measurement is an accurate way to determine ionic surfactant cmcs in aqueous (e.g. 50% vlv methanol solutions. It should be used in organic rich solutions solution) to check if micellization does occur with ionicsurfactants.
~ ~ e c t ~ o ~ h o The t o ~absorption e t ~ . spectrum of numerous dyes changes when they solubilize in the micelle core. Other dyes are essentially waterinsoluble, they dissolve significantly in the presence of few micelles.The absorption spectrum of dye-containing surfactant solutions is recorded for increasing surfactant concentrations. M e n micellization occurs, an abrupt spectrum change is easily detectable. Sometimes a color change can be determinedvisually.uinaldineblue(pinacyanolchloride), fluorescein, rhodamine 6 6 and bromophenol blueare examples of dyes whose spectrum changes upon micelle solubilization. OrangeOT (1-~-tolylazo-2-naphthol), scarlet red (Sudan IV) or methyl yellow 2 ~ - d i m e t h y l ~ n o a z o b e n ~ eare ne) water-insoluble dyes usedfor cmc determination 1221.
Other ~ e t h o ~ Turbidimetry, ~. light scattering and refractive index measurements are spectrophotometric methods used to obtain surfactant cmcs. Viscosity, difhsion coefficient measurements, flow injection analysis and electrochemical methods (pote~tiometryand polarography and even capillary electrophoresis) canalso be used. The accuracy of the quoted methods is variable. However, all of them can rapidlygive an essentialinformation: a breakinthe studied f ~ ~ a t i o An continuous . change property is a reasonable proof for micelles of the studied property should bea warning signal: is there any micelle in this medium?
. A big interest in micellar solutions that is these mediaare able to solubilize, at the same time,polar, ionic solutes and nonpolarsolutes. The location of the solubilized compounds in the micelle or surroundings depends on the polarity of the solutes. On the other hand, when the amount of nonpolar solventincorporatedinamicellar phase becomeshigh,emulsionand
microemulsion structures can be formed. Classical emulsions cannot be used as mobile phases in liquid c~omatography,but microemulsions can. The microemulsion physicoche~calstructure and properties are exposed briefly below.
IK 1. SoZ~~iZz~ation in Mi~eZZa~ Me~za Small mounts of nonpolar compounds can dissolve in the nonpolar core of the micelle, ionic compounds are located in bulk water, and polar compounds can partition between the polar layer (Stern layer in ionic micelles,Figure 2.4) of the micelle and bulkwater. The polar layer of the micelles f o ~ e d by nonionic surfactants is larger than the Stern layer of theionic micelles (Figure 2.12). In nonpolar solvents, water andpolar solutes are located in the core of reverse micelles, and nonpolar solutes are dlssolved in the nonpolar solvent (Figure 2.8).
i
ic
.l2 Localization of solutes in the micelles. The apolar solutes are located in the micelle core. The polar solutes are locatedin the ionic palisade(Stern) layer. Alcohols may form mixed micelles.
ften, the solubilized CO ,the cmc can be in~reased e ~ o ~of nonpolar t s solutes can produce swo les that is the first step towards emulsion fo r alcohols tend to solubilize in micellesas illu oup of the alcohol is located in ated in the nonpolar micelle decrease the electrical repulsionbetween the first step t o ~ a r d microemulsion s fo~ation. h s term means that the of medium or long chainalcohol. There is no clear limit. S that there is 1 “alcohol molecule every a mixed micelle, micelles, it is not possible to speak of mixed micelles. the number of alcohol molecules should be comparab lecules. A mixedmicelleshould contain several alcohol molecules per micelle.
en a solute is addedto a micellar solution, distributes it i t s e ~ b e ~ ethe en aqueous solventand the micelles. To quantitate this &strib r ~ a r t i ~ o n c ~ e ~orc ~i ei nc te ~distrib~tion ~ar constant, as the ratio of the solute concentration in the micelles to the solute concentration in the nomicellar phase. It is expressed by [30]:
(2.15)
is the aggregation number, [ ]the solute ~o~centration in the cellar ~ e d i u mp, is the mole percentage ofthe solute in the micelles, (1 p)is the solute fraction located in the aqueous pseudo-phase, and qrnthe volume percentageofthe micellar pseudo-phase(Eq. 2.10). The aggregation number is not always known, therefore, the solute partition c o e ~ c i e nper t lecule, Pm ,is often used in
(2.16)
The solute pa~itioncoefficient and its location are linked. 2.13 sl~owsthe plotof p values vs. sudactant concentration i ~ c r e a s i nfrom ~ 100 to 200,000are ules of a solute 100-fold moresoluble in the micelles than in water (Kw = 100) are located inthe micellar phase of eason is that the micellar phase volume is only 2.7% of thetotal volume. Otherwise, 42% and morethan 99% of the rnoleand 200,000, respectively, are cules of a solute with ICw values of 2 located in the micellar phase of the 0.1 SDS solution. Ths shows that a solute with a ICw 200,000 can be considered fully locate = I :
Localization of solutes in the micellar phase as a function of thesolute partition coefficient. The indicated valueis the solute partition coefficient, Km. It should be divided by 62 (SDS aggregation number) toobtain the distribution constant per SDS molecule, pm*
LV
S micellar phase. Solutes with Km values between 1000 and 207000 are located in both phases. Solutes with lower located mainly in the aqueous pseudo-phase ofthe 0.12 S concentrationis commonly used in the mi . If the analyzed solute interacts with the 1000, number of SDS that its Km constant is higher than is about 62 molecules per micelle for 0.12 that a value Km = 1000 corresponds to molecule of only]Pw= 16.
c) S o l u ~ i l i ~ a toifoSalts ~ a~~~~ Salts and ionic compounds have a hgh affinity for the aqueous pseudophase. Their distribution constant is, thus, close to zero. This means that they are completely excluded from the micelles. The consequence that is the concentration of salts in aqueous solvent may be hgher than the bulk concentration. For example, for1 M SDS (288 gL), the volume percentage of the micellar pseudo-phase isqm = 0.246 (Eq. 2.10).Therefore, one liter olution contains246 mL of micelles and754 m L of water. If salt, say sobumnitrate, is bssolved in this micellar solution, the ed in the aqueous pseudo-phase7where its concentration can higher. The actual ion concentration willbe actually be 0 .O133 3% overall concent~ation. l/( l cp,)higher th
-
Ion location is even more i m p o ~ when t considering the p micellar solution. The proton concentration in the aqueous pseudo-phase will be: (2.17)
as:
the concentration ofprotons introduced inthe micellar media. ation can be writtenfor the hydroxide ions. The autoprotolysis or ionic product of water in the micellar m d u r n can be written (2.18)
The pK;, valueof a micellar solution will be expressed by [3 11:
(2.19) at 20°C. The J value was ~troduced totake into account ion-exchange phenomena or electrical interactions with the organic interphasethat may act on the ion location[3l]. To illustrate the use of Eqs. 2.17-2.19, let's suppose that 0.0 l HGl is added into one liter of the 1 M SDS solution previously mentio It should be pointed out that a l M SDS solution is a highly conce micellar solution.The aqueous pseudo-phase pH is 1.87 andnot 2.0, of a classical solution; the pK, value in 1 M SDS is 14.25, slightly higher than the value inpure water (14.00). The pH shift is only 0.13 unit (6.5%) 9 value of 24.6%. The pH shift is smaller than 0.05 in micellar with a , values P,lowerthan lo%, such as those usedas mobile phases media with ( in MLC. This pH change can, consequently, be ignored. However, when concentrated emulsionsor microemulsions are used, the pH shifts must be taken into account.
Micellar solutionsare two component mixtures: water and surfactant . ~~n a small amountof a third component is added, itcan be solubilized inthe a third nonpolar component, called rnicellar media. M e n a large amount of "oil," is added to a micellar solution, the solubility limit is exceeded and another system, biphasic innature, foms an emulsion,
The emulsion s t ~ c t u r is e obtained by adding energy to the mixture of water, surfactant and oil. The energy is necessary to increase the interface area. ~emulsions ~ are ) fomed Droplets tend to form. il in water ( oil droplets are dispers in a water continuous phase. Water in 01 emulsions have a continuous oil phase with water droplets insid 2.14). The droplet diameter range from0.02 pm for very fine emulsionto 50 pm and more forcoarse emulsions. The emulsion structure is notstable. reaming or sedime~tation ith time, the droplets fuse by coalescence.
oc~ursand, even^ ly, the oil phaseseparates from the aqueous phase 1321. ins and t~rbidityof such systems proscribe their use as mobile es
il in water
Physicochemical structures of the organized liquid systems. Micellar solutions, emulsions and microemulsions. Open areas: aqueous phase; dashed areas: apolar phase, redrawn from Ref.[3 l].
A microemulsion is an emulsion with droplets so small (c40 se visible light (Fi~ure2.14). A ~croemulsionlooks
be so low that the interfacial area tends to increase spontaneously E34.1. ome microemulsionsare thermo
longer (months to years) compared to the a classical emulsion [33,34].
e 131'
S
y ~ i ~o~three ~ r eco~ponents, A, e~uilateraltriangle volume, mass or mole, is selected. The relative ~ o u n t of s ressed in erc cent age of the selected parameter,such as:
% + C % = 100% rcentages are independent, the third one c gure 2.1S shows how a given mixturecan the crossing of the res
of the triangle. solvent, andthe top
(2.20)
The ternary phase diagramrepresentation. Point M represents the liquid mass, composition with59% ofA, 9% of €3and 32% of C (total 100%). The percentage can be volume or mole percentages.
ost ~ c r ~ m u l s i oare n s made withfour components. In this case, be used, unless a pseudo-component is defined, such as a given ratio of surfactant toalcoholic cosudactant. This active mixture is considered as the third component andis placed at the G apex. Figure 2.16 shows the phase diagram of the ternary system w a t e r ~ e p t a n e / s o dbis i ~ (2e~ylhexyl)sulfosuccinate (Aerosol OT or AOT) [35]. AOT is an anionic surfa~tantableto form ~/~ microemulsionswithout the need of a ~osurfactant.Figure 2.17 shows the phase &agram of the pseudo-ternary + n-butanol) [3l]. C syste~ water~eptane/(GTA~ s u d a ~ that t needs to be associated with a cosurfactant toform microemulsions. The ratio C T A ~ ~ u t a nwas o l constant (1/1 w/w) for all compositions represented inthe phase diagram. The hatched areas corres-
pond to emulsions or polyphasic mixtures. The open areas show the microemulsion domains and the dotted areas correspond to liquid crystal domains.
Mass phase diagramofthe ternaryliquidsystemmade of sodium diethylhexyl sulfosuccinate (AOT), heptane and water (25°C). This system presents a vast L2 (see Figure2.14) area. The L1area is lenticular.The dotted areais a viscous, birefringent and milky area, probably a suspensionof liquid crystalin L2 phase, redrawn from[36].
,111 Mass phase diagram of the quaternary liquid system made of CTAl3huhol (U1 in mass) as thesurface active mixture(surfactant f cosurfactant), heptane and water (NaC1O.lM). This system presents acontinuous change from the L1 structure to the L2 structure (see Figure 2.14). Thedotted areacorresponds toviscous milky compositions, redrawn from [3l].
~ i c r o ~ ~ u l sdomain. ion
onpolar solutes are locate
4.21
*
Evolution of the microemulsion realm of existence according to oil variation (from left to right: dodecane, benzene or toluene) and alcohol cosurfactant variation (from top to bottom: hexanol, pentanol, butanol, to the same 2/13 SDS/alcohol molar ratio) redrawn from [37].
L
solution ofsurfactant molecules inwater makes the micellar e media are easy to prepare, but their p~ysico-che~cal rties should be known in order to be able to use them correctly in . The two important points to remember are: (i) surfactants adsorb at e, and (ii) sur~actantassociations are dynamic; they f o m and or less rapidly. The first point partly explains the u n c o ~ o n solute retention and selectivity obtained in MLC due to modifications ofthe s ~ t i o n phase a ~ by s u ~ a c tadso~tion. ~t The second point may explain the slow mass transfer often observed in LC and may be responsible for the reduced c ~ o m a t o ~ r a p h i c e ~ ~ i e n c y .
1. 2. 3.
chick and F.M. Fowkes (series eds),
,Marcel Dekker, NewYork.
Surfactant Science
.
Gelbart, A.Ben-Sauland D. Roux (eds), Micelles) runes, ~icroemulsions,and ~onolayers,Springer Verlag,
rfactants, in ~ l l m a n'nS~ncyclope~ia of ~ n ~ u s t r i a ~
chemist^, VCH, W e i ~ e i m ,Germany, Vol. A25, pp 747-817 4.
5. 6. 7.
8. 9. 10.
11. 12. 13. 14.
15.
(1994). . Linfield (ed.), Anionic Surfactants, S u ~ a c Science ~ t Series, Marcel Dekker, New York, Vol. 7 (1976). E. J u n g e (ed.), ~ ~ Cationic Surfactants, S u ~ a c Science ~ t Series, Marcel Dekker, New York, Vol. 4 (1970). M.J. Schick (ed.), on ionic Sur~acta~ts: Physical Chemistry, ~ ~ ~ Science a cSeries, ~ Marcel t Dekker,NewYork, Vol. 23 7). . Lornax (ed.), Amphoteric Su~actants,S u ~ a c Science ~ t enes, Marcel Dekker, New York,Vol. 59 (1996). .A t h s and J.A. Beran, General chemist^, 2nd ed., ~ c i e n t i ~ c erican Books, New York( l 992). C.T d o r d , The ~ydrophobicEflect, 2nd ed., Wiley, New York (1980). G.S. Hartley, Ion Aggregation in Solut~onsof Salts with Long ~ a r a Chains, ~ n ~ o l l o i dZur., 88: 22 (1 939). G. C.Kresheck, in Water, a Comprehen~iveTreatise, F. Franks Plenum Press, New York, Vol. 4,pp. 95-167 (1975). Hinze and D.W. strong, Ordered Media in C h e ~ i c a l s i ACS, ~ W a s ~ ~ DC, o nVol. tions, ACS S ~ ~ oSeries, . Shaw, ~ntroduction to Colloid and Surface Chemistry, uttenvorths, London (1980). ,~ i n a rand~ulticomponent y So~utio~ ofAm~hi~hilic Co~pounds)Solubilization andthe or mat ion) S t ~ c t u r eand of Liquid Crystalline Solutions) Chem. ~ h e o r e ~ cSignl~cance al Rev., 68: 1 (1968). o and W. ~ ~ J. Lang, C.Tondre, R. Zana, R. Bauer, H. H Ubricht, Chemical ~ e l ~ a t i Studies on o f ~ i c e l l a r ~ ~ u i ~ iJ b:r i u m ) Phys. Chem., 79:276 (1975).
16. 17. 18. 19. 20
*
21. 22. 23.
24. 25. 26. 27. 28. 29. 30.
L. Laaksonen and J.B. Rosenholm, Molecular ~ynamics Simulations of the ~ater/OctanoateInte~acein the ?resence of Micelle, Chern. ?hys. Letters 216:429 (1993). J.F. goo^ and T. Walker, Micellization in Aqueous Solutions, in Colloid Science, D.H. Everett(ed.), The C h e ~ c a lSociety, London, ‘Vol. 3, Ch. 5, pp. 230-252 (1979). and P. M ~ e r ~ eine Mice~lization, , Solubilization and K.L. Micr ions, K.L. Mittal(ed.),Plenum Press, NewYork, ‘Vol. 1 (1979). Zana, Ionization of CationicMicelles: Effect of the ete er gent ucture, J Colloid Interface Sci., 18:330 (1980). .J. Rosen, Sur~actantsand Interfaci~l ~henomena, New York( l 978). D. Atwood and A T . Florence, Su~actantSystems, hap pm Hall, London (1983). P. Muke~eeand K.J. Mysels, cal Micelle Concentrations of AqueousSu~actantSystems,N S-NBS 3 6 , W a s ~ n ~ o n ( 1 9 ~ 0 ) . F. KraB and H.Wiglow, Ueber des Verhalten der fettsauren Alkalien und des Selfen in Gegenwart von Wasser, III Die ~ e ~ e n als ~ ~ s t a l l o i d e , W D Seven i e als Colloide, Chem. 3erichte, 28: 2566 (1895). K. Shinoda, in Colloidal S~rfactants,K. Shinoda, B. T ~ ~ u s h i and T. Isemura (eds.), AcademicPress, New York ( l 963). N. Funasaki, ~~ermodynamics of Micel~izationof Su~actantsin ?resence or Absence of Salts, J Colloid ~nterfaceSci., 61: 384 (1978). N. Nishiludo, inMixed SurfactantsSystems, (eds.), Sudactant Science Series, Vol. 46, York, p. 23 (1993). M.L. Corrin and W.D. Harkins, The Effectsof Salts on the cmcof Colloida~Electrolytes, J Am. Chem. Soc., 69: 683 (1947). Becher, Non Ionic Surface Active Compounds. Y: ~ ~ e ofc t ectrolytes, J :Col~oidScience, I 7: 325 (1962). K.~ ~ i n o dand a T. Nakagawa, in Colloidal Surfactan S ~ o d a B. , Tamamushi and T. Isemura (eds.), Academic New York (1963). D.W. Armstrong, T.J. Ward and A. Berthod, M i c e l l ~ r E ~ eonc t ~ o l e c u l a r ~ ~ ~ ~ s i heo ore n :tical and C~romato~raphic Considerations,Anal. Chem., 58: 579 (1986).
31. 32. 33
*
34. 35. 36.
Com~entson the ~ e ~ n i t i oofnMicroemulsions, Surfaces, 4: 201 (1982). an (eds.), Micel~arSolutions and lag, New York(1990). . Laughlin, The Aqueous Phase ~ehaviorof Surf~ctants, ~cademicPress, London (1994 . orthault, L2 Microemulsions as a t o ~ r a ~ hAnal. y , hem., 62: 1402
317.
38. 39.
. Tadros,
Su~factant~ in A~rochemicals,
40.
41. 42.
43.
Colloid ~nterfaceSci., 57: 1 (1995).
S:
A Review, A h .
process~g,surfactant ~ctionsby m a ~ n d . soap was done by heating fresh ashes mixed w odd to c o ~ p a rdoing e l a ~ with ~ chromato~r d ~
LC can be considered and others develo
y co~lombicforces o f a cation o~lombicforces, the two ions as§ociat f the solvent is low and/or because th ro~hobiceffect and the electrostatic interaction both r f o ~ a t i o n .The ion-pairing agent or counter-ion 7
58
ERTHOD and ~ A R ~ ~ A -
is most often a strong electrolyte with an alkyl chain of various length. uatemary onium alkyl halides are used as pairing agents for anions. Sodium alkyl sulfates or sulfonatesare anionic salts used as pairing agents for cations. The polarity of the ion-pair is much lower than the polarity of eachindividual ion. Ion-pairing is used to transfer ionsfroma polar aqueous phase to a nonpolar organic one [2]. In its simplest form, the partition equilibriumcan be represented by
where C’ and A- are any organic cation and anion, respectively. The ionpair (C’A‘) behaveslike a polarorganicmolecule and willtherefore preferentiallydissolveinapolarorganicphase such as an alcohol, a chlorinated solventor solvent-mixtures suchas ether and alcoholor ketone and alkane.
In 1973, ion-pair formation associated with phase transfer was used by Eksborg and Schill [3] to develop normal-phase ion-pair chromatography. A polar bare silica stationary phase was impregnated withan aqueous ionic solution. The eluent, a mixture of hexane, chloroform and pentanol, could [4] and h o x separate various benzoic and salicylic acid derivatives. Karger [S] used the normal phase mode with impregnated stationary phases to separate catecholamines by ion-pairing. Soon, h o x thought to use the hydrophobic eRect to f o m ion-pairs in aqueous phases. He developed the first use of reversed-phase ion-pair chro~atographyin 1976 [6]. He termed it “soap chromatography.” He added ionic surfactants to the polar hydro-organic mobile phase. They adsorb on the alkyl-bonded stationary phase. They also associate with the hydrophobic and ionic analytes. Two mechanisms can explain the analyte retention: (i)A classical ion-exchange mechanism, the ionic solute interacts with the counter-ion covered stationary phase. (ii) An ion-pairing mechanism, the hydrophobic ion-pairpartitionswiththealkyl-bonded stationary phase. The true
mechanism may depend on the nature of the ions. It is accepted that a mechanism intermediate between ion-exchange and hydrophobic p~~oning is likely. The dynamic ion-exchange involves differential adsorption of both the analyte ion and the surfactant counter-ion as described by Melander and Horvath [7].
The nature and concentration of the pairing ion of and the organic modifier, the ionic strength,the pH and the character of thestationary phase are the main factors acting on the retention behavior ofthe solute-ion. these factors are interdependent.
a) Ion- air A ~ s o r ~ ~on i othe n S t ~ t ~ o n se a~ The pairing ion adsorbs on the stationary phase. With~ydro-organicmobile [S]. Figure 3.1 phases, Freundlich-t~eadsorption isotherms were obtained shows the adsorption isotherms of sodium octyl sulfonate, CH3(CH2)~~ Na' (top) and sodium dodecyl sulfonate,C H 3 ( C H 2 )Na' ~ ~(bo~om) ~ ~ ~ on Lichrosorb W8 sorbent with different acetonitrile (ACN)-water mobile phases. The adsorbed amount of pairing ion increases exponentially the with carbon number in its alkyl-chain. It decreases rapidly with the organic modifier content (Figure3.2). The ionic solute retentionfactors are directly proportional to the amount of pairing-ion adsorbed the on stationary phase. The ion-pair adsorption phenomenon renders the equilibration of the s~tionaryphase critical. Workmg with none~uilibratedstationary phases will produce nonreproducible results. The time needed to eq~ilibratea column can be surprisingly long. For example, let's estimate the mobile needed to equilibrate a 25 cm column containing 2 g of 8 (3 05m2/g) witha water/ACN 80120 v/v solution containing 0.005 M sodium dodecylsulfonate. Figure3. l (bottom) shows that the ionpair adsorption amount is 0.8 pmol/m2. In the 25-cm column, the surface area is 700 m2,then 560 pmoles of sodium dodecylsulfonateare needed to equilibrate the column. This amount is contained in 112 mL, of mobile at phase. If the working flowrate is 1mL/min, the 25-cm column will need least l 12 minor approximately two hours to be correctly e~uilibrated.
Adsorption isotherms on a C8 bonded silicastationaryphase. Column 25 cm x 4.6 mm id. Lichrosorb RP8, 350 m2/g, 10 pm particle diameter. Top: sodium oc~lsulfonatein acetonitrile (ACN)-water mobilephases. Bottom: sodium dodecylsulfonate in ACN-water mobile phases, 20°C, data from Ref. [8].
e ion-pair adso~tionon the s t a t i o n a ~ phase precludes the use of organic modi~ergradients that would completely modify the s ~ t i o n a ~ phase adsorbed layer (Fig~re3.2).
F i ~ 3.2 ~ r ~ Rapid surfactant desorption with the increase of mobile phase organic modifier content, same experimental conditions as Figure 3.1.
b) S ~ ~ uStructure te
In ion-pair chro~atography,Tomlinson and &ley[g] have estimatedthe ion retention factors using hnctional group values, T, related to the constants. This shows that the hydrophobic part of the ionic solutes is res onsible for the retention as well as its ionic part. The later part may be d ionic strength dependent.
The acidity constant, pKA, is used to measure the ionization degree of a ,is 99% or more in the A’ solute at a given pH value. An acidic solute, ionic formfor pH values such as pH ~ P K $A- 2
(3‘2)
Similarly, a basic solute, B7is 99% or more in the protonated ionic form, BH’, for the pH values: PH
(3-3)
If both conditions are not fulfilled, the ionizable solute is retained partly in its molecular form andpartly through the dynamic ion-exchange process. The retention factor of ionizable c o m p o ~ d sshows a sigmoidal dependance on the mobile phase pH. The lower retention of the sigmoid corresponds to the molecular formpartitio~ng.The lugher retention values correspond to the dynamic ion-exchangeof the ionized form of the solute. A11 variations are possible. For example,the retention ofthe solute ~-c~orobenzoic acid is not sensitive to a pH change on the practical 1-8 range with a q u a t e m a ~onium pairing ion andACN/water 50150 v/v mobile phases [g]. Its PI(, value is 2.9 in pure water. The retention factor of the solute 4-aminobenzoic acid plotted versus the mobile phase p H is a sigmoidal curve with a maximum retention variation in the p a p p r o ~ ~ a t epH l y 4 [g]. The PICA value of 4-aminobenzoicacid is 4.9 in pure water, but it can be closeto 4 in A~N/water50/50 v/v solutions.
It is established in ion-pair c~omatographythat saltsalways reduce the ionpair retention factors by shielding ion-ion interactions. The plot of the retention factor, k, of a ionic solute versus the inverse of the mobile phase ionic strength7I-’, is linear with a positive slope [g], Often, an increase in ionic strength decreases the retention factor and reducesthe efficiency ofthe system. From a practical point of view,the mobile phase salt concentration (ion-pairing agent, solute and buffer) should be maintained belowthe 0.02 M limit [101.
e) ~ ~ g a n i c
~ o d ~ iand e r on~cent a t ration u~e
~ e t h a n o lacetonitrile , or isopropanol are the organic modifiers commonly usedinmixtures with water. Higher organic modifier concentrations produce drastic reduction of ionicsolute retention factors. “his is dueto the increased elution strength of organic rich mobile phase and also to the decrease ofthe pairing ion adsorption on the stationary phase (Figure 3.2). As illustrated by Figure 3.1, the ion-pair adsorbed amount depends onthe nature and concentration of the organic modifier and on the alkyl chain length ofthe ion-pairing agentitself.
fl ~ a t u r eo ~ tI~o ne- ~ aAgent ~r Changing the ion-pair agent is the most readily available way for molfying the ionic solute retention factors. This is achieved by selectinga ion-pair of the same class but with a different alkyl chain length. If the alkyl chain length increases, the ion-pairing agent adsorbed amount increases (Figure 3.1) and the ionic solute retention factors increase. Also the hydrophobic character of the ion-pair increases. Practically, alkyl chain length ranging from 6 methylene units (hexyl) to 16 units (hexadecyl) are available in chemical catalogues.
An increase in temperature is almost always associated with a decrease of the retention factor f lo]. This is again a result of the ion-pairing agent adsorption whose isotherm changes with temperature. This is a problem in the dailyuse of ion-pairing chromatography. Variations ofambient temperature result in a poor reproducibility ofthe retention factor values. Table 3.1 summarizes the parameters acting on the solute retention in ion-pairing chromatography. The big interest of the technique is its ability to analyze and to separate ionic solutes together with molecular solutes using LC columnandhydro-organicmobile phases with low ~ o u n t of s the ion-pairing agent. A satisfying long term reproducibili~of the results may be difficultto obtain, however.
Parameters Acting on the Ionic Solute Retention in Ion-pairing Chromatography
aramete~
Effect on retention
ty alkyl bonded bonded
ion-pairing alkyl chain length ion-p~ringconcentration organic modifier concentration organic modifier nature PH solute on ionic strengthI temperature
exponential increase linear increase exponential decrease change selectivity depends pKA decrease withI-' decrease
An increase of the listed parameter produces the indicated effect on the ionic solute retention factor.
The con~ntrationof surfactant in the solutions usedas mobile phases in ionpair c~omatographyshould notpass the cmc value. It was h o r n that the solute retention was different when micellespresent were in the mobile phase [S-lo]. Figure 3.3 illustrates the dramatic retention change observed when the surfactant cmc is passed [l l]. The retention factor ofbenzyl t r i m e t h ~ l ~ o n i ubromide m (BTAB),a cationic solute, is plotted versus the mount of §D§adsorbed on the stationary phase (a cyanopropyl bonded TAB retention factor is directly proportional to the §
adsorbed mount. concentration rea retention factor d
that first developed . chemical separatio~s
tant solutions above invented people by studying
the cmc for the use
physicochemical properties.
Retention factor of the cationic solute benzyltrimethylammoniumbromide
(BTAB)versus the adsorbed amount of SDS on a CN bonded phase. Column 10 cm x 4.6 mm i.d., CPS Hypersil 5 pm. Inset shows log k vs. the SDS concentration in the aqueous mobile phase. Reprinted from Ref. 11 with permission of the American Chemical Society.
a) A Pioneering
or^
In the early sixties Herries and Richards [123 studied the differences in reactivity of proteinmoleculesbeforeand after dena~ration. They consideredmicelles as possiblemodelssince surfactants werewellestablished denaturating agents. Their studies led them to be involved in kinetic studies of reactions where the reactants partition differently between an aqueous and micellar phase. For their calculationstheyneeded the partition coefficient of the various reactant molecules between the aqueous and micellar phase. The gel permeation chromatography (GPC) technique was used for this purpose. Their idea was: GPC separates the big molecules from the small ones. A. micelle is a molecular association big enoughto be excluded from the GPC stationary phase gel. The monomersurfactant molecules are small and retained insidethegelpores.Thereactantmolecules that partition between the micellarandaqueousphasewillhave a retention volume i n t e ~ e ~ abetween te the micelle retention volume (close to the exclusion volume) and the permeation volume. Then, the reactant retention Volume can be related to its micellar partition coefficient.
*
In an article exposing their results[121, they werethe first to use the arti in and Synge theoretical derivation to establish the equation relating the solute retentionto its micellar partition coefficient. Although their Sephadex G-25 column presented a strong adsorption for some solutes, they were able parameter versus the SDS to obtain straight lines plotting the Vi/(V~-V*) surfactant concentration, thesubscripts i, e ando refer to the internal (pore) and the external volume of the GPC gel,thesoluteelutionvolume of the lines (exclusion)volume,respectively.Theslopeandintercept allowed themto determine accuratelythe micellar partition coefficient of the seven compounds studied [121. They were not interested in the separation go firther in this direction. ability of the technique and did not
7
b) The ~ e ~ i b e r aUse t e of ~ i c e ~Phases ~ a r for Se~arat~on In the late seventies, aniel Wayne Armstrong was working as a graduate student in the Janos Fendler’s group at Texas A&M University (College Station, Texas). His workwasrelated to synthesisandchemicaland biochemical reactions in micellar media.To measure the micellar partition coefkient of transfer ribonucleic acids (tRNA), he usedthe GPC method [121. He got thepartition describedbyHenries,Bishopand&chards coefficients ofthe tRNA solutes he was lookingfor. He readily noticedthe separation capability of the method and immediately announced this fact in his veryfirst article on the topic:The relativelow cost, the ease of operation and shor~essof timefor a completerun merit, we believe, the development of this technique to i t s ~ lpotential l [131.
The first MLC chromatogram.Mobile phase:0.2 M SDS,column: 30 cm
x 4 mm i.d. MicropackMCH-10,lO pm C18 particles, flow rate varying between 2 mL/min
and 3 mL/min as indicated by the dashed line. Solutes: a varietyof phenols, PAHs and other aromatic compounds. Reprinted from Ref. 16.
Armstrontook an
separations with another more affordable technique: thin layer chromatography (TLC). He successfblly used micellar normal and reversed phases with TLC plates to separat investigated anionic micellarsoluti reversed micelles with sodium dioctylsul reversed micelles in cyclohexane [141. broadening of the initial spots on the micellar devel thought to use cyclodextrins the same way as micelles a s e Chr ” to designate all the term: ~ ~ P s ~ u d o - ~ hLiquid chromatographic techniques, GPC, that C, use a secondary chemical e~ullibrium,withmicelles or cyclodextrins, in the separation process [151.
1980, ~rmstrongmoved to eorgetown Univer apply hs ideas onthe capabilities of mic C article presented the separation of ph LC c h r o m a t o ~ ris~ shownin #7)was only 80 efficiency of the 0-cresol peak (Peak column (h.e.t.p = 375 pm or 38 particle hameters). This low efficiencywas probably not pointedout because a flow gradient was used [161. ith Famk Nome, an ex-memberendler group that was s p e n d ~ gsome time as a post-doc with A co~erstonearticle establish~gthe theory o proposed the three-phase model (see Chapter 5) and established it with experiments and asupport~gtheory based onthe following equation:
where 4 is the phase ratio, +=V,N,, the ratio of V,, the stationary p volume and Vm,the mobile phase volume insidethe c retention factor; v is the surfactant molar volume; concentration in the mic form ([M] =total surfactant concentration cmc), and the P, and constants are the dimensionless solute pa S and water coefficients between the ,respectively. s~tionaryphase and wate determine the solute micellar partition coefficient affinity [1'71, b) The ~
'
a S ~r r e o ~af ~ the MLC ~ e c h ~ i ~ u e
Interested by the use of micellar mobile phases, several research teams started to workin the LC field. LindaCline-Love, at Seton es range, NJ)was studying the spectroscopic ~ r o p e ~ i of strong proposed to use theseproperties to the detection of fluorescent and phosphorescent compounds [lS]. Love further developed such use of the micellar phases [191. group verified confirmed Armstrong's theory [201. John University of Florida (~ainesville,FL), was the first to i nze7 eE1ciency problem in another Fendler former studentthat took an aca Salem, NC), showed eory (Eq. 3.4) and could be used as well as ionic surfactants E221 'rkbrigh~, at the Universi~of M ~ c h e s t e r ( ~ n g l ~used d)7 separate salts of dithiocarbamates with cationic s u ~ a c t ~[23], ts
. III.1.
1982-1986, a GrowingInterest
er 1982, more and more research groups located worldwide started to now a professor investigate the capabilities of MLC. In the US, Armstron~y at Texas Tech University (Lubbock, TX), was still active in the field [24251. Armstron~[26-271 and the otherAmerican scientists E281 wrote reviews to increase the di&sion of the technique. Researchersin I d y [29], China [30-311, Spain [32], France [ l 1, 331 or Japan [34-351 increasedthe knowledgeof the foundationsofMLCand its application range. The Liquid Chromato~raphywas progressively deno~nation:~~eudo-phase replaced by: Micellar Liquid C h r o ~ a t o ~ r a [ ~1h1y2 1 331.
b) ~~e 1
~ New 8 York ~ A~S~eeting
strong andHinzeproposed
to the AmericanChemicalSociety to organize a symposiumon the useof Ordered Media in Chemical ~eparation~ at its 19lStMeeting held in New York inApril 1986. A high percentage ofthe people working in the MLC field attendedor were invited, With the texts of the conferences, the two symposium organizers edited a book contain~ga great partof the MLC knowledge at this time [36). It is ~ p o to point ~ t out that Armstrong hmself gave a talk on cyclodextrin formation. He was happy that people used and developed C techque he invented,but he thoughtthat the technique had oms. The use ofCDs for chemical separations was already his main researchinterest. He was so successhl applying theCD chemistry to the difficult task of chiral separations that so far7he never went backto LC limitations were clearly seen during this s y m p o s i when ~ the capa~ilitiesof the micellar phasesassociated with the then newcapillary e1ectrophoresis techque were exposed. Electropherogr~sshowing efficiencies in the 100,000 platesh range were displayed [37].
III.2.
198 7 - ~ r e s e ~Mt ,~ t ~MLC re
After the 1986 symposibecame clear that LC wouldnot supplant any separation methodser, the interest of the scientific c o ~ u n i ~ was still signific~tas LC publications appearing S seen as an interesting t e c ~ i ~ u that e per year (Figure 3.5 deserved studies intwo main directions: (i) the ~ n d ~ e n tway: a l eluent strength, selectivityand e~lciencyan (ii) the applied way: hydrophobici~studies and original separations.
40
Figure 3. Number of articlesentirelydedicated to IdLC sorted by publicationyear from 1977 to 1998.
?
~ ~~ t r e~n ~ te~ , ~ ~ e ~ etc t i v i ~ a nThe ~ ~very ~ i clow i eefficiency ncy. obtained with purely aqueous micellar phase was greatly improved by the 3% vlv propanol addition r e c o ~ e n d e dby weak eluent strength of the aqueous micellar a ~ ~ i t i oofnalcohols. These studies were initiated by rsey’s student [38-391. Hinze showed that the retenti C models(logkproportional compoundsdidnotobeytheestablished to homologuenumber)[40]. He propadirect transfer ~ e c h ~ i s m between the micelles and the stationary phase [4 l]. C were thoroughlyinvesti~atedby Love [44] and others [45]. e reasons for the low efficiency observed in were investi~atedby and ~ ~ t i ~ i z a t i o n ~undamental . studies increased the ~ o w l e d ~ofethe micellar retention mechanisms It was then possible to propose theoretical models that allowed chromatographers to predict and C separations [48-501.
~ i ~ ~~ eo~ a~r ~i t ci o ~ S~early s~. as 1985, Cline-Loveshowed that the surfactant molecules contained in the micellar mobile phase were able to bind to proteins and to reduce greatly their a ~ s o ~ t i on o nthe silica based stationary phases. The directi of biologicalsamplessuch as s e w ithoutproteinpreciitationinsidethe or urinesamples can bedone i colu~ [511. This property initiated a great interest for number of works in thisarea [32, 52-5 51. b i c i ~an^ Qu~ntitative~ ~ u c t ~ r e ~ e t e ~ partition coe~lcientof a sol~tecan be are l data for the nshi [56-591. research field was investigated by numerous groups ~ e ~ a r a t i o n s o ~ r o n s a n ~ r ~ o r ~ aThe n i caqueous ~ ~ e cphase i e s . of the micellar mobile phase c m dissolve polar solutes
3. of
4. 5.
Organic Co~pounds, J:Chro~atogr., 83: 99
.L.Karger, High Perfor~anceIon Pair Partition The Separationof Biogenic Amines and their matogr. Sci. 12: 52 1 (1974). and, ~eparationof Tricyclic Psychosedative eed Ion Pair Partition and ~ i ~ u i d - S o l i d
6. 7.
Pair Chro~ato-
9.
10. 11.
graphy, Adsorption is other^^ of Two Ionic Surfuctants on Five
12.
13.
~olutesbetween Mic Gel ~iltrationand Eflect on the ~ineticsof Some ~ i ~ o l e c u l a r ~eactions,JPhys. Chem. 68: 1842 (1964). . Fender, ~ ~ e r e n t ipartition al in^ of t ~ A between s Micellar and AqueousPhases: a ~ o n v e n i ~Gel nt ~ i l t r a t i oMethod ~ for Separation of t ~ A sBiochi~. , Bio~hys. Acta, 478: 75(1977). .Terrill, Thin Layer Chromato~ra~hic and ~ucleosides Separation of Pesticides, ~ecachlorobi~henyl with Micellar Solutions, Anal. Chem., 51:2160 (1979).
15. 16.
17. 18.
19.
20. 21. 22
I)
23. 24. 25.
26. 27. 28.
D .W. A ~ s t r o n g , Pseudophase Liquid Chromatography: Applications to TLC, J: Liq. Chromatogr., 3: 895 (1980). D.W. h s t r o n g and S.J.Henry, Use of an Aqueous Mobile P ~ s e for Separation of Phenolsand P M s via HPLC, J: Liq. Chro~atogr.,3: 457 (1980). o Solutes r D.W. A ~ s t r o n gand F, Norne, Partitioning ~ e h ~ i of luted with ~ i c e l l a r ~ o Phases b i l e in Liquid Chromatography, Anal. Chem., 53: 1462 (1981). D.W. A ~ s t r o n gW.L. , Hinze, K.H. Bui andH.N. Sin& ~ n ~ a n c e d Fluorescence and Room ~emperaturePhosphorescence ~etection in Pseudophase Liquid Chromatography,Anal. Letters, 14: 1459 (1981). . Weinberger, P. Yamchuk and L.J. Cline-Love, ~ i q u i d ChromatographicPhosphorescence ~ e t e c ~ owith n ~icellar Chromatography and Post-column ~ e t e c t i o n ~ o d eAnal. s , Chem., 54: 1552 (1982). P. Yamchuck, R. Weinberger, R.F. Hirsch and L.J. Cline-Love, Selectivity in Liquid Chromatographywith ~ i c ~ l l a r ~ o b i l e Phases, Anal. Chern.,54: 2233 (1982). J.G. eEtchegaray and J.S. Landy, ~ ~ c i e n c y Enha cellar Liquid Chromatogra~hy,Anal. Chem., 55: 924 (1983). M.F. Borgerding andW.L.Hime, Characteri~ationa n d ~ v a l ~ a ~ o n of the Useof on ionic ~olyoxyethylene(23) dodecanol @rij 35) Micellar~obilePhases in RPLC, Anal. Chem., 51: 2183 (1985). G.F. Kirkbrightand F.G.P. Mullins, Separation of ~ i t h i o carbamates by HPLC using a ~ i c e l l a r ~ o bPhase, i l e Analyst,
G.Y. Stine, Selectivi~ in Pseudophase Liquid 1. Chem., SS: 2317 (1983). G.Y. Stine, Selectivi~in valuation and ce~le-Solute~nteructlons,J: Am. Chem. SOC.,
, A~plications of Pseudophase Liquid C~romatography:Highly Selective~obilePhasesfor Present and Future Separa~ons,American Lab., 13: 14 (1981). D.W. Armstrong, ~ i c e l ~ eins Separations: A Practical and heo ore tical Review, Sep. Pur$ Methods, 14: 2 13 (1985). L.J. Cline-Love, J.G. Dorsey and J.G. Habarta, The ~ z c e l l e Analytical Chemistry ~nterface,Anal. Chem., 56: 1132A (1984).
29.
.Pelizzetti,The Useof a in the HPLC ~ e ~ a ~ a toifoHn y ~ ~ o x y b e n ~ e n e ~ e r i v a ~ v ~ s , A
30. 31. 32.
33. 34.
i, Cation-~xchan~e C ~ ~ o ~ a t o ~ofr aC ~a th~yc h o l a ~ i n ~ s cella^ ~ o b i l Phases, e JPhar~acobio-~yn., 7: 5 ( l 984). ads, ~ i c e l l e ~ x c l u s i o n C h r o ~ a t o ~ r aof~ h~yn o r ~ a n i c
36. 37.
39. 40.
41.
he^., 62: 130 (1990).
43. 44.
~Role ers C : ~ e l e c t i vin i ~~ Chem., 64: 1894 (1992 F.P,Tomasella anclL.J. based onthe ~ h r e e - ~ h a s e
~~~
45. 46. 47. 48 49. 50.
51. 52. 53
*
54.
56.
~ o l v e n t ~ t r e nand ~th and~icelles, Anal. dynamic ~ r o ~ e r t iin es ~ ~ u i l i b r iAnal. u ~ Chem., ~odel,
arina, ~etermination ~ o l u t Association e Constants of ~ o B ~ ne e and ~ a ~ h t h a l e n e
7
57. 58.
59.
60. 61. 62. 63.
IA-ALV
Correlation with the Octanol- Water System, Anal. Chim. Acta, 212: 171 (1988). di andE.D. Breyer, ~uantitationofHydrophobici~with al. Chem., 61: 1040 (1989). , Sanchez and F. lez, MA. R o d r i ~ e z - ~ e l ~ a d o.J. ~arcia-Nontelon~o,Solute-Micelle Association Constants and t s H y d r ~ p h o b i cfor i ~ PAHs by Correlation of Pot, ~ o e ~ c i e nwith MLC, Chromatographia,34: 627 (1992). I. Ueda andH. Fujiwara, Micelle Water N ~ ~ u rK.a Hayash, , rtition Properties of~henolsete ermined b y ~ CProposal . for a Versatile Measureof Hydrophobyci~,Chem. Pharm.Bull.,43:
.P.Nullins, HPLC of Organic and~norganicAnions: Use of cellar ~ o b i l Phases, e ACS Symp. Ser., 342: 115 (1987). terpretation of ~etentionBehaviors of rans sit ions in MLC using an Ion-exchange ~ o d e l Anal. , Chem., 64: 589 (1992). iura, MLC Separation of ~anadiump )Chelate with 2-(2~hia~o~la~o)-S-dimethylaminophenol, Fresenius J Anal. Chem., 344: 294 (1922). S. Terabe and Z.Deyl (eds.), Micelles as Separation ~ e d i in a Chromatography and~lectrophoresis,J Chromatogr.A, 780: 1 361 (1997).
-
The two main properties ofsurfactant molecules are micelle formation. and adsorption at interfaces. In Micellar Liquid Chro~atography( micelle formation property is linkedto the mobile phase. Micelles play the role of the organic modifierRPLC. in Nonpolar solutespartition themselves between the micelle apolar core and the apolar bonded stationary phase. 5. The surfactant adsorption This partitioning willbe the subject of Chapter property is linked to the stationary phase. A significant numberof s u r f a c ~ t moleculesmay adsorb on the stationary phase surface changing its properties. The study of such adsorption andits associated problems is the main subject ofthis chapter.
Although new materials are rapidly emergingfor c~omatographicstationary phases, silica is still the principal medium used as the material for liquid chromatography (LC)pachng [1-31.
II.l.
P ~ y s i c o c ~ e ~ iPro~erties cal of Silica
The three physicochemical parametersfor silica relevantto chromatography are the particle size, thesurface area and the porosity. The other important parameters are the particle shape,the carbon percentage, the ligand density and bonding structure and the surface silanol concentration [l]. 7
0
0000 - 0 O F
W"
mmm+
vi"
2 vivi-0 G 0 t7
o\W M M
000
0
00000
o!o\.o! ? Y Y Y Y
000 M M M M"
22s v i n vin W"
TH
E
f”i
00
U
0000 vivlvivi
“mm
000
M “
lrivivi
f”if”if”i
600 W“lri” vi”
v??
mm-
-m*
00
e?=? 000
0
a) Particle Size ilica packings are always obtained withparticle a distribution, i. e., a stated -pm particle diameter isan average diameter. The polydispersity of a silica powder is related to the width of the particle size distribution. It can be given. by the ratio d90/dl0,particle diameters of which 90% and 10% o f t particle are smaller than or equal to, respectively. The ratio dgO/dlo shou be as close to unity as possible (monodispersesilica). Particle size is always associated to particle shape which can be spherical or irregular. Table 4-1 lists the particle size and shape of some commonly usedc o ~ e r c i asilical based LC packings. b) Surfcrce Area
The a m o ~ h o u ssilica used for LC packings is highly porous. The specific surface area of a silica sample is equal to the sum of its internal (pores) and external areas 141. The specific surface area, A, of a porous silica sample is expressed in m2/g. iiis due to internal area at more than 99% [l]. It is easily measured by the BET method that uses gas a d s o ~ t i o n . i s o t h e ~ s . Surface area and porosity are linked. c) Pore Size andPore Volume
Pores are holes, cavities and/or channels communicating with the silica surface. Pores can be regarded as cylinders with a diameter, Pd, usually Their total volume is the pore measured in m, s o m e t ~ e isn A (=O.1 m). volume, Vp,measuredincm3/g or mL/g of material. The pore size distribution is measured by the mercury intrusion method [1, 41. A, good chromatographic silica packing should havenarrow a pore size distribution. In such case, pore volume, pore diameter and specific surface area can be linked by the a p p r o ~ ~ a t i relation: ve
t?- = 4
0 0 0 ~ ~ ~ ~
with the units m2/g, cm3/gand nrn for t?-, Vp and Pd, respectively. Eq. 4.1 can be used to estimate the pore volume of a given silica packing when the surface area andpore diameter are given. The typical pore size of classical LC packing ranges between8 and 20 m (80-200 B\)(Table 4.1). wide pore packings with30 to 100 nrn (300-1000 /S)pore size are mar~eted
Apolar surfaces are obtained with alkyl groups. A variety of can be bonded on silica surfaces producing chromatographic p a c h g s with a widerange of polarity and properties.
A bond^ silica packing is characterized by its carbon load, %C, its surface coverage in ~ ~ o l / m and 2 , the bonding type, monomeric or p o l ~ e r i c end-capped , or not [l, 3, 41. It was shown that the maxi mu^ bonding coverage density was 4.5 pmol/m2 for monomeric dimethyl-alkylsilyl groups. A lower coverage density means that some silanol groups remain a~ainable to the solutes. They are called “residual silanols.” They are responsible for the tailing observed onthe peaks of basic compounds and sensitivity of the packing. Table 4.1 lists the physicochemical properties of a selection ofcomercially available HPLC packings.
. classical 4.6 m i.d. LC column contains about l g of packing material per l 0 cm of column length. The packingsufiace area is much higher, inthe hundredth ofm2range, than the tubing surface area (a few cm2). ~urfactant adsorption on the porous packing should be consideredsince it may change the s t a t i o n a ~phase surface properties.
Several methods can be used to estimate the amount of surfactant molecules adsorbed on the s t a t i o n a ~phase surface in aLC column. The titration, the brea~hroughand the stripping methods will bedescribed because they can produce reliable results with common chromatographic equipment. The results obtained with agivenmethodcanbeused to checkand to corroborate the results of another method. Accuracy ofthe method dep~nds on the exact knowledge of the column dead volume which should be the first p a r ~ e t eto r establish.
a) Dead ~ o ~ u ~ e
~ e a s u r ~ ~ e ~ t s
The ~ e t e ~ i n a t i oofn the void or dead volume of a column is too often neglected by chromatographers. The dead volume,Vo, is the mobile phase volume inside the column. It corresponds to the inter-particle volume, Vi,
plus the porous or intra particle volume, V,. It is also the mobile phase volume needed to elute a nonretained solute. Vo=toF
(4.3)
where 1;, is the retention time of a nonretained solute and is the F flow rate. A n on retained sol LC with polar mobilephases can be pure acetone or methanol or asorbingioniccompound such as sodium nitrate or uracil [7]. The dead volume obtained should be compared to the column internal volume,VC: VC dc2= n;/4
L
(4.4)
where dcis the column i n t e ~ a l d i ~ eand t e rL, the column l e n ~ h .For the popular 4.6 mm i d . tubing, the column internal volume is 0.166 mL per column cm (-1.7 mL for 10 cm). The ratio V ~ is cthe total porosity of the column. It is a dimensionless number that should be somewhere between0.7 and 1. If the V ~ ratio c is found to be higher than unity, it means the dead volume is c is found to over~stimated:the solute used was retained. If the V ~ ratio belowerthan 0.7, thedeadvolumeis too small. The test solute was probably excluded from the internal pore volume ofstationa~ the phase. A practical rule of thumb for the 4.6 mm isd.column only is E71: V. (rnL) =: 0.1 L (cm)
(4.5)
A l0 cm 4.6 m i.d. column should have a dead volume closeto 1 mL. For adsorption isotherm determination, theaccurate knowledge of the dead volume is required. In mostcases, the dead volumetracer method produces appropriate results. When a cross-check the of dead volume value is necessa~,the ~ e ~~ se it can ~~ beo used. ~ First, the columnis carefilly flushed and filled with methanol which has a density of 0.79 14 g/cm3 at 120°C. It is disco~ectedfiom thepump and its two sides are closed to avoid evaporation. The column is weighted. Itsmass is WveOH.Next, the column is similarly flushed and filled with chloroform, d=l.483 g/cm3at 2O"C, and weighted. The second mass, WcHc13, higher than the first one, allows to obtain the dead volume with a high accuracy using:
HAS
7
The 1.446factor is only valid for methanol and chloroform at 20°C. Any pair of solvents with a significant density difference and compatibili~with the stationary phase can be used instead of methanol and chloroform. The 1.446 factor is trivially the reverse ofthe solvent density difference.
It is often necessary to determine the surfactant concentration in a given solution.Ionic surfactants canbedeterminedby the two-phaseliquid method. Ionic dyesare soluble in aqueousphases. They form ion-pairs with surfactant molecules bearing an opposite charge. The ion-pairs are apolar and canbe extracted intoan apolar organic phase. Chloroform is commonly used. The typical protocol for titration of a cationic surfactant, e.g., c e t y l t r i m e t h y l ~ o nbromide i~ (CTAB), is as follows. A known volume of the cationic surfactant solution to be titrated is introduced ina test-tube. It is diluted withwater. A l0 mL volume of chloroform is added along with few drops of a cationic indicator dye and a few drops of an anionic dye. Sunset yellow, mordant red 72, disulfine blue W 150 or alizarins are examples of anionic dyes. Cationic dyescan be dimidium bromide, methyl red, neutral red or chrornoxan green. The color of many indicators is pH dependent, and it may be necessary to add a buffer (or an acid) in the aqueous phase. Few cationic surfactant moleculesand the anionicdye associate forming ion-pairs that are extracted into the chloroform phase coloring it. A known SDS solution, say, exactly 0. drop. The tube isshaken. An emulsion forms eeds a minute to separate because it is stabilized by the cationic surfactant. andthecationic surfactant formion-pairs that chloroform phase. As the titration progresses, the emulsionseparates faster and faster because there is less and less cationic surfactant to stabilize it. When the endpoint is reached, the SDS molecules anionic dye-cationicsurfactant ion-pair and forman pair extracted by the organic phase. This produces the phases. The equivalent volume ofthe S S solution allows to calculate the cationicsurfactant concentration in the titrated volume. The protocol for
an anionic surfactant is similar. A CTA standard solution is used for the titrating agent. Nonionic surfactant cannot be titrated by this method. They are oRencompoundswithverylowvolatility. Their concentration can be d e t e ~ i n e dby weighting the r e m a ~ n gsolid matter aRer complete solvent evaporation. c) The ~
~ e u ~ t eth h ~hodo u ~ ~
The accuracy of the b r e ~ ~ o u method gh depends on the ability to detect the surfactant inthemobilephase. A refractive index ( )detector is most oftenrequiredsinceusable surfactants shouldnot absorb An evaporative light ring (ELS) detector can also be used su st surfac~antsare solids at room temperature.
A t h e ~ o s t a t e dchromatographicsystemshouldbe used. The ,packed with the studied s t a t i ~ ~ a phase, ry is first e~uilibratedby pass in^ the aq~eousphase, pure water or buffer or water and organic modifier, without anysurfactant molecule. The eluent is then switchedto a mixture made ofthe same mobile phasecon~ininga h o r n concentration, C, of surfactant. A surfactant concentration front thus moves down the c o l u ~ . If the surfactant is orbed by the packing material, the ,will exceed the dead or void volume, b r e ~ h r o u g hvolume of the front ~ ~ This ) ~ ~ o u n oft surfactant adsorbed will then be ( ~ B - [8]. shouldbeexpressedpreferablyinpmol/m2 for convenient compa~isonbetween phases. It should be pointedout that the brea~hroughvolume may notbe easy to determine. Figure 4.1 shows the idealb r e ~ h r o u g hshape and some curves that maybeobtaineddependingon thecolumn efficiency. A progressive surfactant adsorption may be due to pore size polydispersity of the stationa~phase: the larger pores are filled before the smaller pores. Also, successive surfactant layers may form. These c i r c u m s ~ c e smay in the detector response. In case of produce slow baseline changes or steps corrected. Theadsorbed ~ o u n of t steps, the VB volumeshouldbe s u ~ a ccorresponds ~ t to the area shown in Figure4.1. the signal is propo~ionalto the surfactant concentration.The recorder trace can be usedto correct the VBvolume. The responseof most ELS detectors
is not propo~ionalto the surfactant concentration. A calibration must be p e r f o ~ e d€or each case or the titration control shouldbe use.
Possible detector signal obtained during breakthrough experiments. An efficient column gives a sharp of rise the detector signal. A less efficient column gives a smooth detector rise. The shaded area shows the detector changes obtained in a real study [lo] and corr~spondsto the adsorbed surfactantamount.
or accurate results, it is advisa~letocontrol
the sur~actant a d s o ~ t i o n p e r € o ~ i nag titration. The effluentleaving the c o l u ~is correspond in^ to the collected untilthe detector response reaches plateau a absorbance value of the initial surfactant solution. concentration inthe collected effluentis d e t e ~ i n e dby titrati su~actantmass is adsorbed onthe stationa~phase in the column. The mass obtainedby titr~tionshould corroborate the one obtained using the brea~hroughvolume.
RTHO
Anotherway to determinetheamountof surfactant adsorbedonthe stationary phase of a column is to strip it off with a strong eluent and to titrate the collected emuent. This methodassumes that there is no irreversible surfactant adsorption. m s is not always the case. Irreversible adsorptions were mentioned by h o x for anionic surfactants [9] and by inze [101 for a nonionic surfactant. However, in most cases, adsorbed surfactant can be fully strippedoff the stationary phase by pure methanol, a mixtureof methanol and propanol orpure propanol. The protocol iss t r ~ ~ M~e r complete o ~ ~c o l~m . ~ ~ i l i b r a t i o n two with thesurfactant solution at concentration G, the column is flushed by or three void volumes of aqueous phase without surfactant. This effluent is not collected.This step washesout the water solublesurfactant contained in the void volume without desorbing the surfactant of the stationary phase. Next, the column is rinsed with at least 10 column void volumes of pure methanol (or propanol). The effluent, containing the desorbed sudactant, is collected. ethanol is evaporated and the dry surfactant is re-dissolved in a measured volumeof water. The amount ofsurfactant is determined by titration.
III.2. A ~ s o r ~ tIisoo~t ~ e rwith ~ s Aqueous ~ ~ a s e s a) Anionic ~urfactant ~ e ~ h studied o d the adsorption ofSDS on five different HypersilB stationary phases [l 11. Five 10 cm x 4.6 m i d . LC columns were specially selected for the study. All experiments were done in,a thermostatic bath at 30°C. The amount of SDS adsorbed on the phases was determined by crosschecking the resultsobtainedwith(i) the brea~hrough method,(ii) controlling the concentrationof the effluentsolution,and(iii)by the within an 8% stripping method. The 3 resultswereinagreement experimental error. Figure 4.2 presents the S S adsorption on bare , bonded phase,SAS Hyp ~ ~ e r s i l CPS B , ~ ~ e r s i l aBcyanopropyl a methyl bonded phase,MOS HypersilB, an octyl bonded phase and ersilo, anoctadecyl bondedphase. The physicochemical prope~iesof these phasesare listed inTable 4. l. A very differing behavior was obtained
HA
with nomicellar mobile phases and micellar mobiles phases. Figure 4.2 can be compared with the adsorption isotherms obtained in ion-pair chromato~raphy(Figure 3.1 and Chapter 3) 191.
0
0.002
0.004
0.0043
0.008
0.0
SDS ~ n c ~ n t r ~ t (mol/L) ion
SDS adsorption isotherms at 30°C on five different Hypersil@ A) nonmicellar mobile phases; B) micellar mobile phases, data from [1l].
phases.
S
the low concentration a1 point of view, the bile phase can be very large. The l.9 ~ m o ~ m 2
S
indicated byan arrow. Thethree nonpol ersil@,show a Lang~uir-likeadsorpti a concave shape or L type in the Giles’ classi~cation[121. phase shows a convex shapeor S type cooperative adsorpti increases in the order: bare silica < exactly the decreasing order of the
ypersil@passes the one on mainly responsiblefor
e [121:the amount of adsorbed suda
linear for a ionary and mobile phase, respectively, should be isotherm. Figure 4.3 shows the Freundlich plots of the five a d s o ~ t i o nisotherms. For the 4 bonded phases, there are clearly tvvo
di~erentslopes sepa by the micellization area (logcmc = -2. l). increase of adsorbed ismuchlowerwithmicellar solutions than pre-~icellarsolution. There is still a S adsorption increase. S phase. These results were possible to speak of saturationonlywith 'S workon an Alt hase [131. A diEerent result o noted a sharp adsorption m a ~ ~ for u man just before the micellization conc 153. The substrate was a mono-c~stallinesilicon wafer that was by nitric acid to produce an optically flat silica plate that was d end-capped. The bondingdensity was 2.8 chlorodimethyloctadecy pmol/m2, similar to the ersil8 (2.9 pmol/m2).This flat substrate is diE1cult to compare withporous silica used for LC pachngs. '
l
.3 S t a ~ o n phase a ~ versus mobile phase SDS concentr~tionsin log-log coordinate showing a clear break for the cmc value (cmc= 0.0082 M,log cmc = -2.1).
6) Cationic S~rfactant
The adso~tionof CTAB was similarly studied on the five H ~ e r s i l ~ s t a t i o n a ~phases listed in Table 4.1 [111, The results are presented in xres 4.4 and 4.5
GTAB concentration (mol/L)
o
0.b2 ab4
0 . h 0.08 o:r 0.i2 0.i4 CTAB concentration (mot&)
o.is
0 . i ~ 012
Figur CTAE3 adsorption isotherms at 30°C on five different Hypemil@phases. A) nonmicellar mobile phases; B) micellar mobile phases.
~ o n - M i c ~Mobile ~ l a ~ Phases. Figure 4.4A shows the CTAB adsorption for sub-micellar solutions. The conclusions are exactly similar to those obtained for SDS adsorption. The b r e ~ h r o u g hvolumes were much higher. The 2.1 ~mol/m2 CTAB adsorption withthe 0.0003 M solution corresponds to 0.00022 moles of CTAB. This amount was contained in 2.2 liters of mobile phase. The experiment duration was two days at l mL/min. The three apolar phases show L-type concave adsorption isotherms. The two polar phases show S-type convex isotherms [121. The notable dif3erence is the CTAB adsorption on bare silica. This adsorption is a proof that ~y~rophobic interaction is not the onlydriving force for CTA adsorptionon the poroussilica surface. S i l ~ o l - q u a t e ~armno a~ interactions are likely. They induce a cooperative adsorption of more C molecules by hydrophobic interaction.
log Cm ([CTAB] in mol/L)
.5 Stationaryphase versus mobilephase C T A B concentrationsinlog-log coordinate showing a clear breakfor the cmcvalue (cmc = 0,0009 M, log cmc = -3.05).
.5
~ i c e l l a~r o ~ iPhases. l e Figure 4.4 shows the 30°C adsorption i s o ~ e ~ s concentration range [1l]. The five i s o t h e ~ are lateau for concen~ationhigher than the cmc [121. Figure 4.5 shows the corresponding logC, versus log C, ~reundlichplots The curves are clearly made by two lines. The slo e chan e at the cmc concent~ation(log cmc= -3 -05)is apparent an a d s o ~ t i o nsaturation is observed for C cmc. For the three other phases, a adsorbed ~ o u nist noted (Figure 4. what was observed with the S S surfactant (Figur
c) ~ o n ~ o n i c
~urfact~~~
inze studied the adsorption of nonionic s u r f a c ~ ton s a Cl (Resolve G 18, Table 4. l) [161. H-type isotherms similar to the ones obtained with ionic surfactants (Figures 4.2 and 4.4) were e s ~ b ~ i s h efor d two polyosyethylene dodecyl ether surfa~tants (BrijB22 and adsorption increased beyond the surfactantcmc.The 22 ~ o u nincreased t from l .4 pm0 )to 2 pmo~m2at 1700 cmc (0.16 unt was almost increase of the BrijB 35 adsorb cont~uous.It was about 0.3 p and 0.9 pmo~m2 at 850cmc (0. G, versus log C, Freundlich plot ofthe BrijB 22 bilinear curve s i ~ i l ato r the plots found in Figure m Freundlich plot of the BrijB 35 break at the cmcconcentration. ater soluble surfactant, itmay have a low affinity for the apolar bmded solve C l 8 s t a t i o n aphase ~ [16].
a) ~hysicoche~ical ~tructure of the Ionic ~urfactantLayer The s t ~ c ~ofr e
ed on porous silica 5 pm, 6 m pore re studied first [171 so stu~ied[181. with high power proton
FT
es used to study the sur orbed molecules. linked to polarity di~erencesfound with s u r f a c t ~covered t phases.
~urfactantadsorption on monomeric C 18 bonded silica phases.A) the ionic B) the cationic sulfate groups protrude outside the bonded layer making it negatively charged; 8 layer; C) Sil~ophilicinteractions are another way ammonium groups are buried insideC 1the of adsorption; D) the cyano to explain the lower ionic charge observed in case CTAl3 strongly bindto thecharged surfactant head either positive or negative. A, B and D from studies [17, 181. C from chromatographic studies [19].
n a~yl-bondedphases, the W results seem to indicate that the surfactant is located insidethe trimethyl onium polar head of the CTAB thickness ofthe bonded layer (Figure 4.6~4)when the sulfate polar group of de as illustrated by Figure 4.613 [1’71. This would covered alkyl-phases are more polar with a negative charge than the TAB covered phases. These results are coherent with o d that theSDS adsorption could occur Berthod’s work[191. ~ e ~ thought ophobic adsorption as seenby (Figure 4.6). However, ways to describe the adsorptio B: (i) h y ~ o p h o ~and ic (ii) silanop~lic adsorp~on (Figure 4.6C) [191. The silanophilic adsorption is the interaction between the Lewis acid silanol groups and the basic cationic onium groups.It is the onlyway to explain the CTAB adsorption on the bare silica (Figure 4.4). The studies showed that there was a strong interaction the surfactantcharged heads andthe cyano groups bonded on the rb phase. The charged heads ofthe adsorbed surfactant molecules are ‘‘tied up” by the cyano ~ n c t i o n a l of i ~the bonded phase (Figure 4.6 [181. The negative SDS charge is almost canceled and so is the positive charge. The stationary phase has just a slightly higherpolarity after surfactant adsorption. The surfactant-c~anogroups’ strong inte~actionwas difficult to disrupt: part of CTAB or SDS was irreversibly adsorbed on ~ i c r o s o r b[1S]. Such irreversible adsorption was not observed with the ypersilB phases [l 1, 191.
b)
ha^^^ of the
Silica Surface
It was shown that the adsorption of s u r f a c t ~molecules t onthe silica surface changes its polarity. ~bviously,it also changes its s t ~ c t u r esufiace , area, pore volume and pore size. It was shown that, with S S and BrijB 35, the surface arealost beyond the cmc was closeto 0.6m2p adsorbed surfactant ole [161. For example, 0.5 pmol/m2 of BrijB 35 were olve C 18 whenthe aqueous phase concentration was 0.02 this phase was 180m2/g The total amount ofadsorbed is 90 pmoles. The surface area decline is close to face area of the stationary phasecovered bythe is only 126 m2/g, a 30% loss. The surface area loss was only 0.3 m2 per adsorbed pmole of BrijB 22.
The surface area decrease may be to distortion due of the pore shape or to pore filling with surfactant. The pore structure of the phase was studied by nitrogen porosimetry [161. It was found that, for 2 and SrijO 35and for anionic SDS, the generalporeshape no of 18 phase was retained. The pore volume was decreased by surfactant adsorption. These observations suggested that the s u ~ a c ~ t molecules were coating the interior walls of the pores without completely filling them[161. The thick surfactant film formed onthe top of the organic bonded layer is responsible for apart of the loss of efficiencyobserved in L C with purely aqueousphases. 111.4. Eflect o ~ ~ o b i ~ eA~~itives P ~ on ~ Ss e~ ~ a c At ~~ s~o t~ ~ ~ i o n a,)Buffers and Salts
Ionic compounds are frequently added to micellar mobile phasesfor p 2 described the significant changesthat ionic strength adjustments. Chapter anysalt additionproducesonthephysicochemical s t r u c ~ r eofionic surfactant micelles (mainly: decrease of cmc and increase the aggregation of number). The slope changes in the log C, versus log C, Freundlich plotfor adsorption data obtained with CTAB and SDS mobile phases with 0.1 NaCl added salts corresponded to the surfactant cmcs in such solutions (0.0002 M and 0.0014 M, respectively, comparedto 0.0008 M and 0.0082 M in pure water) [20]. The amount of adsorbed surfactant was found to depend on the stationary phase [1 1,16, 191. A “salting-out” effect was observed with the apolar alkyl phases. Inmost cases, it increased the amount of adsorbedsurfactant by loweringthe electrostatic repulsions and enhancing the hydrophobic interactions [rZO]. Table 4.2 lists the effect of NaCl addition on the adsorption of various surfactants on different L C phases. The addition of NaCl increases the surfactant adsorption by 15 to 40%. The exception,5pmol/m2of SDS on HypersilO SAS C1 phase dropping to 4.4 ~mol/m2 when 0.1 M NaCl is added (Table 4.2), was that are enhanced by NaC1 on explained by silanol ion-exchange phenomena hase [rZO]. The Si-0- ionized groups have no affinity for the anionic *
~ u ~ a c t+a ~ t NaCl
~dsor~ed tat. Phase ~191 POI 1191
5.2
-id-
4.6
Davisil C18 -id-id-id-
3.2 .8
.o
5.0
~191 POI 1191 c201 1.5
7
33
P11
3.4 -id-
[21] 3.6
150
-id-id-
3.1
1’70
-id-id-id-
P11 C211 c211 [H]
1.8 2.6 [21] 130
[21] c211
Table 4.2 presents the results obtainedwith W absorbing pyridinium chloridesurfactants [2 l]. The authors studied the “salting out” S stationary phase. ef5ect that increased the adsorbed ~ o u n on t an They also studied the desorption rate of the four ethoxy pyridinium chloride surfactants by passing pure NaCl solutions onthe surfactant-coveredphase. exponential decrease ofthe surfactant loadmg was obtained. The half time, tl,2,to desorb 50% of the surfactant, was between 1.5 hours and 170 hours (a week) [2l], NaCl dramatically decreasedthe sudactant desorption rate. This is a logical consequence of the “salting out” effect.
The first studies of surfactant adsorption isotherms were donewith hydroses in ion-pairing chromatography.The CTA rsil8 was found to be 1.2 pmol/m2 with 5 thanol-water mobile phase containing 0.0 1 is valueis 30% lower than the value obtain water (1.7 pmol/m2, Figure 4.4). h o x obtained even lower values on a HypersilB SAS phase (0.2 pmol/m2 with a 30-70% v-v prop~ol-water e was the first to establish phase containing 0.055 surfactant adsorption isotherms 191. The adsorbed amou S was as high as 8 pmol/m2 with a 20-80% watermobile phase containing 0.07 sodium salts and 0.0 Saturation of the stationary phase by AB was observed o C 18phase when 60-40% v-v methanol-water (with0. l a CTAB concentr It was established that organic modifiers decreasedthe ~ o u noft adsorbed surfactant by competingfor adsorption sites and/or bydecreasi~g the hydrophobic interactions [24]. For ionic surfactants, the ability of an organic modifier for s u r f a c ~desorption t was related to its hydrophobici~. Figure 4.7 shows that the amount ofadsorbed surfactant decreases linearly with the organic modifier concentration. The slope of the lines is related to the surfactant desorption s t r e n ~ h . Pentanol strongly desorbs ionic surfactants, methanol is less efficient [24]. hysicoche~calmethods such as contact meas~rementsand ourier t r a n s f o ~ e dinfrared (FT
le, surface tension spectroscopy were
f ~ a of~additive ~ n
ole fraction of additive
Effect of organic additives on the plateau of concentration of adsorbed surfactant. Stationary phase: Hypersil ODs; mobile phase concentration,SDS =0.2 M,CTAB = 0.1 M, data from 1241,
used to investigate the structure of the alcohol-silica bonded layer [25]. It was found that the alkylchains of propanol and longern-alcohols inte~enetratethe C18 bonded alkyl chains to form a single monolayer similar to the one formed by the alkyl chains of the surfactant molecules (Figure 4.6). The hydroxyl group is oriented toward the aqueous phase. Competitionbetweenalcohol and surfactant molecules for a d s o ~ t i o n explains the decreasedamount of adsorbed surfactant with increasing concentration of alcohol inthe mobile phase (Figure 4.7). The organic additives of the micellarmobile phase affect the su~actant-adsorbedlayer. This changes the chromatographicselectivity and efficiency obtainedfor a set ofanalytes with the same column. and efficiency are studied in otherparts of this book.
It was shown that surfactants adsorb at any interface, especially the stationary phase surface. The question is: once a column is well ~uilibrated with a micellar phase, are we working with a given bonded stationary phase or with a surfactant stationary phase? In other words, does the surfactant coating of the stationary phases render themall similar?
To answer to the introduction question, astudy of the retention behaviorof a set of solutes of various polarities was done with different stationary phases and identical micellarmobile phases [26]. SDS and C phases were used withfivedifferent i18 phases. The retention of the strong's three phase polar, apolar as well as ionicsolutesobeyed model(eq. 3.4, Chapter 3). Table 4.3 lists the solute-micelle partition coefficient and the PWs solute-stationary phase affinity coe~lcientfor different solutes on four bonded phases. values are similar within experimental errors (see Chapter 5).Th alues are similar for the and C 18 bondedphases. hese two phases do behave similarly in classical LC. The more polar 1 and CN bonded phases do show signific~tly ower a&ty coefficient The surfactant layer is there: cationic solutes are retainedby the four bondedphasesandhyd anic mobile phases. eirhighaffinity for the stationary phaseswithobile phases isdue
a~itionCoef~cients, hase ~~~, Pws,for Four
Solute
SAS C1
tationary phase ODs C18
MOS C8
38 37 40 ctylbenzen~sulfonate 160 e n z y l t r i ~ e t ~ y l a ~1700 on.
300 2000
0 35 1900 Cetylpyridi~umchloride 3000
360 23 3200 4700
35 1400
~tylbenzenesulfon~te 9.4 e n ~ y l t r i ~ e t ~ y l420 ~on.
190 190 Ca~eine 3.6 1.8 enzoic acid 870 550 760 Cetylpy~di~um chloride 850 610 1000
Average value
38 220 1900
36 l10 500
200 1500
390 2800 2900
190 34 2000 4300
300 33 2500 3700
33 1500
10 51
. .
2.3
2.0 90
63
m
.=
910
Data from Ref.28. Toluene = apolar solute, cafEeine= polar solute, sodium octylben~enesulfonate and benzoic acid= anionic solutes,b e n ~ l t r i m e ~ y l a m m o n i ~ = cationic solutes. H~ersil stationa~ bro~ideand cetylp~ridi~ium c~loride phase data are listed in Table 4. l.
enol) ere more retained on
the
b) ~ ~ t e ~ f aTension ci~l
was s h o that ~ the surfactant adsorption decreasedthe interfacial tension between the stationary and mobile phases. In the absence of micelles (ionpair chromatography), the nonionic solute retentionfactor was agected by t h s interfacial tension change accordingto eq. 4.7 [321:
where k,is the retention factor in the absence ofsur~actant,C is the mobile phase surfactant concentration (Kcmc), is the Gibbs free energyof adsorption of the surfactant at idkite dilution and4; is a constant relatedto the change in interfacial tension caused by the added surfactant. A. negative 4; value produces a decrease of the solute retention factor with an increase of surfactant concentration [33].
c) ~ a ~~ i cde l ~ Gradient ar Capabili~es The surfactant adsorption isotherms showan almost horizontalplateau for sur~actantconcentration above the cmc (Figures 4.3 and 4.5). This property was used to perform rapid analyses with a micelle concentration gradient a micellar solution a change in total surfactant concentration serves hange the concentration ofthe micelles. If the surfact~t-a~orbed layer depends only on the free s u r f a c ~concentration, t thenthe stationa~ phase is not altered bychangesintheconcentrationof the micelles. erefore it is possible with micellar concentration gradientsto speed the ion of strongly retained solutes without affecting the stationary phase t [34]. This allows a step gra~ientback to the initial conditions ~ t h o u ree~uili~ration time [l31. This rapid gradient capability is not universal in horizontal plateau in thesurfactant isotherm adsorption ficant increases of s u r f a c t ~ adsorption t above the cmc were observed especially with nonionic surfactant [l61 and/or with polar statio^^ phases [.191. In such cases, some time (=several column volumes) will be necessary to re-equilibrate the column after a micellar gradient analysis.
7 *
It is always recommended to use the same column with the same type of surfactant. A column should be dedxated to the anionic su~actants,a second oneto the cationic sudactants, etc. The reproducibility of the results in MLC dependsonthecolumnequilibration. The adsorbed layer of surfactant should be donecorrectly. It was shownthat the time to reach the equilibrium between the stationary phase and the mobile phase could be very long in ion-pair chromatography with sub-micellar mobile phases. Two days at 1 mLlmin were necessaryto equilibrate a 15 cm x 4.6 mm i.d. column of HypersilB ODs with a mobile phase containing 0.0003 These low surfactant concentration solutions do notconta they are not used inMLC. With a micellar phase,the equilibration time is reduced. It is possible to use the rapid gradient capability just mentioned above. Typically,a mobile phase containinga hgh surfactant concentration (10 to 100 cmc)can be usedto quicklysaturate the column withsurfactant. Then 5 to 10 column volumesare used to rinse the column withthe mobile phase containingthe desired amount ofsurfactant.
The recommendationto use the same column withthe same surfactant was done because, in some instances, it may be impossible to fillyremove the su~actant-adsorbedlayer. It was not possible to eliminate a small part of SDS adsorbed on a 1980 HypersilB ODs phase even when using a pure methanol mobile phase [S]. However, a complete SDS desorption from a 1988Hypersil8 ODs phase was possible with pure a methanol mobile phase [24]. Part of the nonionic Brij 35 surfactant could not be removed from a Resolve C 18 phase after a 24-hour elution of water-acetonitrile 70-30% v-v [161. Conversely, thereare many examples where thesurfactant layer can be completely stripped off the stationary phase [1 1, 19, 2 1,241. It was suggested that partial irreversible surfactant adsorption was dueto a tight insertion of the s u r f a c ~ alkyl t chains in thealkyl moieties of the bonded A column could be used with ~erent layer of densely grafted phases [35].
S (anionic and catio c) afker a complete methanol [M].
"
ost s u r f a c ~ t are s very soluble In pure m phase is r e c o ~ e n d e dfor s u r f a c ~ t methanol-propanol75-25%v-v may be n e c e s s a ~in rare occasions. 20 to 50 column volumes is the minimum ~uantityof methanol to elute and to remove the adsorbed surfactant. h experime ensure a complete su~actantdesorption. The a decreases the solute mass~transferbetween th mobilephase. This producesband-broadeni erved [l6, 24, 353. The e x p e r ~ e n t atest ~ can be test m i ~ u r e is prepared to measure the ef~ciencyo LC newcolumnwith a classical hydro-organic mobile p e~ample,a 50 pmol/L benzophenone + biphenyl solution was stationary phase and a~ethanol-water75-2 ntion p a r ~ e t e r sof the test solutes (retenti es) are measured with the hydro-organic mobile phase and the any surfactant exposition. These p a r ~ e t e r sare simil~rly column has been exposedto s u r f a c ~ t srinsed , with pure quilibrated with the reference hydro-or~anicmobile phase [24]. If the retention factors and the peak ef~cienciesof the test solutes re cove^ ofthe initial s~tionary similar to their original values, a complete phase surface can be assumed. small decrease of the col^ e ~ c i e n c y be due to column aging.
rsolutionsused C containmore than 90% vlv water. e aqueous solutions are to dissolveminute ~ o u n t sof silica ecially at elevated temperature~30'~) andlor values hi~herthan 6. 1ca dissolution is deadly for thecolumn. It uldbereduced to a S problem is solved by saturating the aqueous mobile phase is done, without causing any band broadening, bylaci in^ a s a ~ r a t i n g p r e c o l just u ~ after the pump and b re the injection valve. This precolumn can be a short (5 cm x 4.6 m i. column packed with a
are silica phase to reduce the pressure buil precolumn also acts as a filter, it should be checkedregularly.
A micellar solution should stay not in contact with the bon based s~tionary-phaseswhen the chromatogra~hicsystem is not first reason is that the c~omatographi dware does not like S such as ionic surfactant solutions. osionmay occur in systems conta~ingionic surfactant solutions. that are weighted to prepare the micellar soluti an produce crystals’ forma~onaround the pump rystals may obstruct the system producing c may also scratch the p u p piston. The sec rich micellar solution may produce silica Si-C bond ruptures that can ruin the column. A very essential rule of care is: a m i ~ e l ~ phase ar s h o u never ~ ~ stay motionless in a c h r o ~ a t o ~ r a p h l ~ system. cleaning procedure will beindicated. It may be every evening andto re-equilibrate it everymo that a micellar phase does not harma c ~ o m a t o g r a p ~syc is not static. This means that it is possible to keep a micellar phase overnig if the pump is not turned off. If the operator knows he willgo on WO with the same micellar phase and he wants to leave, hec m value (often 0.1 mL/min) to reduce the pressure he shouldplace the detector outletTe the mobile phase r e s e ~ o i rto recycle the micellar phase. ensures that the system will not run dry (unless a leak occu also guarantees the good equilibrium of the adsorbe column. Last, itsaves the stationary phase since the recycled mobile phase is saturated in dissolve silica and will not dissolve moreof it. LC system will not be used for the weekend or d according to the following three step pro First, the micellar phase is replaced by 100% pure water. pump is flushed rapidly and the system, valve, column, detector tubing, is rinsed with 10to 20 column volumes of pure water. higher flow rate can be used if the pressu does not increase too much. The pure water viscosity is only 1
Next, a 100% methanolmobilephase is used to remove the adsorbed surfactant in the chromtographic system. A IowJEow rate (0.5 mLlmin or less) should be used at the beginning because the methanol-water mixture has a hgh viscosity whichis signi~cantly increased by the desorbed surfactant. Once the pressure decreases, the flow rate may be resumed. At least 10 column volumes pure of methanol should be passed throughthe column. Then, the hydro-organic mobile phase used to test the column efficiency can be loaded in the system (5- 10 column volumes). The test itself will be done when the system willrestarted. be The power can be turned off. The column contains a classical hydro-organic phase without ionsthat will not damageit. The complete procedure takes about half an hour. It should not be n e g s t d . The 100% water step should not be bypassed. A brutal change fiom a buEerd micellar mobile phaseto a 100% methanol phasem y produce salt c r y ~ l i ~ ~ The o n column . and even thep ~ p system ~ g could be damaged. The last step, with a hydro-organic mobile phase, can be bypassed. But it is good practiceto check for column performances once every week. the micellar phase during the If thetwo rules, a low flow rate with night and the30 min cleaning procedurefor the weekend,are respected, the life~imeof the columns used inMLC are comparable and evensuperior to LC.These rules were followed witha chromatographic performances were system using a 15-cm ODS column. The CO ma~tainedfor morethantwoyearsofintensive use.
2.
3.
erthod, Silica: Backbone ater rial of Liquid Chromatography Column Packing, J Chromatogr., 549: 1 (199 1). Unger, F.E. and Regnier R.E. matography Packings,J Chromatogr., 544: 1 (1991). A. Berthod, ~ o r m a l and Reversed Phases in Liquid Chromatography, Encyclopedia of Analytical Science, Academic ress, New York,pp. 2567 (1995).
4.
K.K. Unger, Porous Silica, J. Chromatogr. Library, Elsevier, ~ s t e r dVol. ~ ,16 (1979).
5.
J.C. Giddings, Dynamics of Chromatography,Marcel York (1965).
6.
Fhenomenex, ~ P L C ~ a t e r iSorbent al Characte~stics, Commercial Catalogue (1996).
7.
L.R. Snyder, J.J. Kirklandand J.L. Glajch, Prac~calHPLC eth hod ~evelopment,2nd Ed. VViley Interscience,NewYork (1997)).
8.
J.H. . b o x and G.R. Laird, Soap Chromatography)a New HPLC Techniq~e for Separation of Ionizable ~aterials) J Chromato~r. , 122: 17 (1976).
9.
Knox and R.A. Hartwick, Mechanism of Ion-pair Liquid omatography ofAmines, Neutrals, Zwitterions andAcids using Anionic ~etaerons,J Chromatogr., 204: 3 (198 1).
10.
and Evaluation erding andW.L. Hinze, ~haracte~zation ofNonionic~~cellar~obile Phases in .RP-HPLC, Anal. Chem., 57: 2183 (1985).
11.
A.Berthod, I. Girard and C. Gonnet, MLC: Adsorption Isotherms of Two Ionic Surfactants on Five Sta~onary Phases, Anal. Chem.,
58: 1356 (1986). 12.
C .H. Giles, Anionic Surfactants, E.H. Lucassen-Re~ders (Ed.), Surfactant ScienceSeries, Marcel Dekker, New York, Vol. 1 1, 4 (1981).
13.
orsey, M.G. Khaledi, J.S. Landyand J.L. Lin, Gradient E l u t i o n ~ CJ, Chromatogr., 316: 183 (1984).
14.
M.E. Montgomeryand M.J. VVirth, Eflect of SDS onthe c in a C18 ~rientational B e h ~ i o rof a ~ y ~ r o p h o b iProbe Monolayer Bonded to Silica, Anal. Chem., 64: 2566 (1992).
15.
.A.Fiaseckiand M.J. Wirth, Spectroscopic Probing of the Interfacial ~oughnessof SDSAdsorbed to a Hydrocarbon Surface, Langmuir, 10: 19 13(1994).
17.
Interface Sei., 165: 497 (199~). 18.
ooper, Solid-State onCyanopropyl o n ~ e d ~ h a sIrnplieations es: for MLC, JI Colloid Interf~ceSei., 179: 341 (1996).
19.
,A d ~ i ~ v e on ~ ~urfaetant ~ects
20. 1362 (1986). 21.
surf act ant^ on ODS ~ilica:Role of ~l~ctrostatics, J. C o l l o i ~ Interface Sei., 155: 124 (1993). 22.
.A. Wall, ~ y d r o p h o b i c C h ~ o ~ a t o ~ rwith aphy , S1 Dyna~icallyCoated Sta~onaryPhases, J.~ h r u ~ a t o g r1.74: (1979).
23.
i e r s e ~ a ,~ e v e r s e d ~ h a s e ~ ~ with L CC ~ t r i ~ i Containing de ~ l u e nJ.~Lig. , C ~ r o ~ a t o g4:~ . , 1961 (1981).
24.
ussel, he R o ~ eof the Station~ry~ h a s ein C. A ~ s o r ~ t i oand n ~ ~ c i e n cJ y., Chrornatogr.,449: 349 &h, Spectroseopic Study of the aC18 Surface by L o n ~ - ~ h na~n ~lcohols, Anal. Chern., 66: 680 (1994).
27.
29.
30.
1.
32.
kda, ~ i ~ u l t a n e o u s ~ e ~ a r a tofi o Ionic n and on ionic C o ~ ~ o u n dussin^ ~eversed-~hase ~ieell~r Chro~ato~ra~~y Anal. ~ e i ~ n ~9:e59 s , (2993).
eming, T h e r ~ o d y n a ~ i c Ion- air LC, Anal. Chem., 54: 225 1 e~ersed-~hase
,Interfacial Tension
33.
1.C h e ~,55: . 425 (1
radient Ca~abilities inze, Investi~ationof the C, J: c h r o ~ a t o ~ r556: . , 263
35. (1991).
. icellar Liquid Chromatography(MLC), the mobile phase consists of an ~queoussolution of surfactant at a concentrationabove the critical mi concentration (cmc), in contact an with alkyl-bonded stationary phase. versed-PhaseLiquidChromatography LC) mode, with micelle S a mobile phase modifier. It is also o necessary to add a small concentration of an organic solvent to improve the efficiency and br control the elution strength and selectivity of the mobile phase. technique has also been named as pseudo-phase liquid chromatography. This term was coined to describe separations where s i ~ f i c a n t p a ~ i t i o n i n g of a solute occurs to discrete aggregates dissolved in.the mobile phase, rather than to bulk solvent. of anionic, cationic, nonionic and icellar mobile phases zwitterionic surfactants are usedinconjunction with stationary phases(including CS, C18 and cyano). su~actant(usually
solute-statio~a~ phase hydr ar ith an anionic surfactant: (a) n o n ~ o ~ soiut
icellar systems share the basic components of an at is, a nonpolar stati phaseand a polar aqueousmobile phase. owever, con~entional with a ~ u e o u s - o r g ~mobile c phases homogeneous, whereas micellar solutionsare microscopicallyheterogeneous, being co~posedof two distinct media: the ~ p h i p h i l i cmicellar aggregates e surrounding bulk aqueous or aqueoussurfactant monomers in a concentration approximately equal to the cmc. Inmicellar solutions, the solutes are preferential~ysolubiliz~ into or onto the micellar assembly, a process which is d ~ ~Thecsolutes . localized in micelles experiencemicroenviro~ent a that is d r ~ a t i c a l l ydifferent from that of bulk solvent interns of polarity d fluidity. This is reflected by micelle-inducedp e ~ ~ r b a t i o nofsphysicochemical prope~iesof the solutes, i~cludingchanges in solubility, acidity, photo~hysical prope~ies and reaction rates. L C is muchgreater than that of conventional
solvents, because of the number of possible ~teractionswith both mobile and stationary phases (Fig. 5.1). The solutes in the mobile phase can interact electrostaticallywith the charged outer-layer of ionic micelles, and hydrophobically with their lipophilic interior. steric factor can also be i m p o ~ ~The t . modificationof the stationary phase by a d s o ~ t i o nof surfactant mono~ers,which creates a"micelle-li surface, gives riseto similar ~teractionswith the solutes. The co~bination of these interactions cannot be duplicated by anytraditional pure or mixed ile micellar solutions will never totally replace traditional aqueous-organic eluents, they offer several interesting alternatives to separation work. The basic mechanism ofseparation in L C is fairly well understood and there is a reasonable theoretical foundation on whichto build. a fascinating example of the use of a secondary chemical equili liquid chro~atography.The primary equilibrium is the pa~itioningof the solute between bulk mobile phase and stationary phase, and the secondary equilibrium is the pa~itioning tomicelles. This secondary equilibrium is affected bya great variety of p a r ~ e t e r s :type and concentrations of surfactant and additives such as salts or org c modifiers (for instance, . The current knowledgeon C interactions isexposed
1
in this chapter. The 0 plications of such interactions on the elution strength and selectivi~of micellar eluentsare considered in
mstrong and Nome [l]proposed a three-phase (stationary phase, bulk aqueoussolvent and micellarpseudo-phase)model, to explain the chromatographic behavior in an W L C system of a solute eluted with a mobile phase containing a surfactant above the cmc. Such a treatment has allowed a theoretical description of MLC, and greater understanding and utilization of this chromatographic technique. ~ccordingto the three-phase model, the retention of a solute is controlled bythree competing reversibleequilibria, namely partitioning (or b ~ d i n g )from bulk aqueous solvent to micelles, p a ~ i t i o ~ nfrom g bulk aqueous solvent to the alkyl-bonded stationary phase, and direct transfer from micelles to the stationary phase. The first equilibrium takes place inside the mobile phase, and the latter equilibrium can be neglected in most si~ations,but it is significant for highly nonpolar (water-insoluble) solutes, which have a great affinity for both stationary phase and micelles. The partition in^ equilibria are described by three coef5cients: Pm (between water and micelles), Pws (between water and stationary phase), and,,P (between micelles andstationary phase). It is the first partition c Pw ,that imparts uniqueness to MLC. The coefficients Pwsand opposing effects on the retention of solutes: as Pwsincreases the retention due to increased increases, whereasas P,, increases the retention is reduced partition in^ into micelles. The retentionbehavior of solutes willdependon the type of interactions with the micelles and with the surfactant-modified stationary phase. Nonpolar solutes should only be affected by hydrophobic interactions (Fig. 5. la). For these solutes, different proportions of nonpolar, dipoledipoleand proton donor-acceptor interactions between solutes and
surfactants are expected. But for solutes that are charged, two distinct situations can be considered:
(i) the sign ofthe charges on the solute and surfactant arethe same (Fig. 5.lb), or (ii) the sign ofthe charges onthe solute and su~actantare opposite (Fig. 5. IC).
The first situation is encountered when an anionic solute is eluted with an anionic surfactant, or a cationic solute is eluted with a cationic surfactant [e.g.,dissociated phenol with the anionic surfactant sodium dodecyl sulfate (SDS), and protonated benzylaminewith the cationic ),on a C 18 column] surfactantdodecyl t r i m e t h y l ~ o n i u mbromide (D [2]. Electrostatic repulsionfrom the micelleshoul affect the retention as the solute will still reside in the bulk solvent phase, and therefore, will move downthe column. In contrast, repulsion fromthe su~actant-modified the solute may elute stationaryphase shouldcause a decreased retention, and in the dead volume. However, it may be retained by hydrophobic ~teraction with the stationary phase, although this effectwill be reducedby the electrostatic repulsion. Because due to different hydrophobic interactions, dissociated phenol and 2-naphtholare well separated with S The second situation appears when a solute is c~omatographedwith an oppositelycharged surfactant? where electrostatic attraction occurs between both species. ~lectrostaticattraction between solute and micelle will complement any hydrophobic interaction?and thus, it can be expected that the solute will remain inthe mobile phase for a longer period of time, decreasing the retention. However, electrostatic and hy~ophobic intera~ions with the stationary phase are often sufficiently large to offset micellar attraction and thus thee retention will increase. That is why ~ssociated phenol and 2-naphtholare retained to a greater extent with SDS on a Cl8 column [2].
on ionizable solutes should only e~perience hydrophobic interactions. It has been shown, however,that any molecule with moment can profoundly be affected byelectrostatic effects [3].
nction of the micellarpseudo-phasein has beencompared to that of the organic modifierin traditional S for most solutes an increase in the concentration of surfactant in the mobilephase results in a decreased retention. This is in contrast to reversed-phase ion-interaction chromatography, where the surfactant concentration is belowthe cmc (i.e., no micelles exist), and the addition of an ionic surfactant will retention of compounds that interact electros~ticallywith it. ,the elution strength increaseswith micelle concentration only ifthe solute interacts with micelles.
strong and Stine [4] proposed a classification of compounds into three groups, according to their elution behavior with a micellar mobile phase: (i) compounds bindingto micelles (ii) nonbinding cornpounds (iii) antibinding compounds
c om pounds that associate or bind to micellesshowdecreased retention when the concentration of micelles inthe mobile phase is increased. or compounds that do not associate with micelles, the retention may remain unalteredby the micellecontents (nonb~ding), may increase with c creasing micelle concentration (antibinding). Fig. .2 shows plots of the n factor (lk)vs. micelle con~ntration([ nding, nonbindingand antibinding compo lines of positive, zero and negative slope, respectively. The behavior most freque~tlyobserved is binding to micelles, while the antibinding behavior is quite ~ n c o ~ o n .
Typical plotsof MLC retention vs. mobile phase micelle concentrationfor a: (a)
bind in^ solute, (b)non bind in^ solute, (c) anti bind in^ solute.
ith anionic micelles are n e ~ a t i ~ e l y
lsion is an i m ~ o factor ~ ~ in t so ~ositi~ely charged corn ~~~
The a n t i b ~ d behavior ~g cannot be observed with stationary phases that adsorb an appreciable mount of surfactant (i.e.,C8 or C 18 bonded phases). For these phases, when the column acquires the same charge as the micelles, and no hydrophobicinteraction occurs, similarly charged solutes tend to elute in the dead volume of the column. In contrast, when using stationary phases which do not adsorb large quantities of surfactant (i.e.,C 1 or preferably cyano-bonded phases), one can observe increased retention when eluting with higher a concentration of micellesthe in mobile phase. Antibinding results from a compound being strongly excluded or repelled from the micelle. In fact, the solute is not only excluded from the micelle, but also from the double-layer aroundthe micelle. Thesolute is thus forced onto the stationary phase. A. negativewater-to-micelle pa~ition 0) has been assigned to these compounds, whichhas in principle no physical meaning. However, as compounds that bind to micelles each have a characteristic coefficient, compounds that are excluded from micelles may have characteristic a negative parameter. This parameter may be usefbl, sinceinsome cases it canbe correlated to the degreeof electrostatic repulsion between solute and micelles. There are a variety of methodsto measure partition coefficients of solutes to micelles. Much less attention is givento solutes that do not seem to interact with micelles or, at least, do not follow models that require a specific type or amount of interaction. The repulsion of a solute from a micelle is sometimes implied (for instance, in the micellar idibition of a reaction), but rarely treated in a rigorous quantitative manner. Part of the reason for this is the lack of experimental techniquesthat can measurezero or negative interactions. In addition, experimental phenomena that can be explained bythe binding ofsolutes to micelles are often thought to be more relevant and interesting. Using MLC one can detect and measure positive (i.e. binding), negative (i.e., antibinding) or zero (i.e. nonbinding) interactions between solutes and micelles. ~ u ~ h e ~ oone r ecan , monitor a solute's change in behavior (from binding to nonbinding to antibinding or vice versa),when a small change the in e n v i r o ~ e nand/or t micelles occurs.
3
11.3. ~ e s c r ~ t i o ofnthe ~etentlonBehavior In MLC, the retention factor is relatedto the concentration of monomers of surfactant in the form of micelles (i.e.,total concentration of surfactant minus cmc), through very simpleequations. Three theoretical approaches, basedon the three-phasemodel,havebeenproposed to formulate the retention of binding solutesfor liquid chromatographic systemsat various micellar concentrations:the p a ~ i t i o model ~ g of Armstrong and Nome [l], and the equilibrium approaches of ~ n y a n and a ~Cline-Love [5], and Foley [6]. Jandera andFischer 171 extended the mathematical treatment to antibinding solutes. The equationsderived are similar and allow the evaluation of the strength of solute-micelle and solute-stationa~phase interactions. Also, another pseudo-phasecan be substituted for the micelle ~ (such as a cyclodextrin or crown ether), without appreciably c ~ a n g i nthe theoretical equations.
a) ~ q ~ a t i of nArmstrong andNome Armstrong and Nome [l]extended to micellar media the classic model of ~ a ~and i Synge, n andHerries et al. [S]. The model describes the s i ~ a t i o n that a solute experiences inan ideal system formedof a numberof identical plates, where pa~itioningequilibria take place. The transitions that can occur in the three e n v i r o ~ e n t sthat exist in a micellar chromatographic system (i.e . ,the aqueous phase,the micelles andthe stationary phase) were considered (Fig. 5.1). The mass fraction of solute ineach e n v i r o ~ e nand t ,,P and ,,P . The PM,coeficient each theoreticalplate was obtained firom was not included in the model, since can it be obtained by combination of the two former (PM, = P,, /P,,). Based onthe expression givingthe maximally occupiedtheoretical plate, the following equation was obtained:
to the
-Love [5] rewrote the pseudo-phase retentio~ e~uation y used retentionfactors and bin ing constants, rather gelutionvolumes and rtition coefficients. a1 e~uilibriawere consi
ssociation of the solute i hase ind ding sites,
ssociation or binding of the solute in bulkwater with o~omerof surfactant in the rnicelle:
(iii) direct transfer of the solute from the micelle to the s t a t i o ~ phase: a~ S
+
cted inthis approach. in the expression givi
is quantity can be included in the pa~ition is referredto the asso~iationof the solute with a surfactant monomer in the micelle, and should be multiplied bythe aggregation numberto find the constant referredto the whole ~ c e l l e .
e as§umptions made in the equilibrium model proposed by not veryhfferent from thoseof the previous authors [1,5]. The associ~tion between the solute in bulk water and the micelles (eq. 5.4) is considered a secondary equilibrium, which affects the retention of the solute in the absence of micelles, given byk,.
This equationis again similar tothe equation prop (eq. 5.l). The retention factor of fiee so in eq. 5.1 and K , [S] in eq. 5.6, whereas corresponds to the same constant in eq.5.6, and to the product U in eq, 5.1 when the volume of aqueous phase is taken as total volume of mobile phase. Also, the basic pseudo-phase equation of retention (eq. 5. l) can be al~ebraicallyrearranged in at least 12 different related forms [g]. Although obtaining one form fr another is mathematically trivial, there are certain stati§tical and other antages in using di~erentvariations.
d) ~ a r ~ ~~o oe ~n c i e n t ~
The three models given by eqs. 5.1, 5.6 and 5.7 lead to similar linear equations lk vs. [ 1,a d can be rewritten as:
or
a .3 Solute-micelle, Km, and solute-stationaryphase, KM,binding constants in micellar mobilephases.
Figure 5.3 illustrates the three-phase model with indication ofthe partition constants Km and KAs. The values of Km and K A S give i n f o ~ a t i o nabout the solute affinityto micelles andto the stationary phase, respectively. One consequenceof the partitioning treatmentis that the measured values of K ICMsforacompoundusing the samemicellarmobile phase, but stationa~phases, will result different in values of KAs and KMs ,but identical values of K,,. By the same token,the measured values of a compound on a single stationary phase,but with two micellar mobile phases, will result in differentvalues of K, and identical valuesof ICAs . The constant C I s should indicatethe difEerent degreeof of surfactant. This was not however observed in a study p e r € o ~ e dwith severalsolutesand CTAB as surfactant [lo]. The highvalues of ICAs obtained with cyano-silica, and the unexpectedly low values G 18-silicaY with clearly showedthat the adsorption ofsurfactant was not the only factor to take into account. The subjacent stationary phase (the bonded moiety) probably still plays a role in the interaction of the solute between bulk water and stationary phase. In spite ofthe sufiactant coverage, the polar nature of the bondedstationary phase is maintained.
e) Jandera and Fischer Approach Equations 5.8 and 5.9 describe the retentionofsolutes that can form associates, or inclusioncomplexes,bothwithmicelles and s u r f a c ~ t adsorbed on the stationary phase. Tks is the case for neutral compounds and compounds with an opposite charge to the sufiactant. However, as indicated above, compounds having the same charge as the sufiactant will be excluded from the micelles and repelled the by modified stationa~phase, unless other interactions exist that neutralize the electrostatic repulsion. For solutesrepelled from themicelles, the retention increaseswith the concentration ofsurfactant, which supposesan apparent negative valuefor K, . Since this constant is the ratio of the eq~ilibriumconcentrations of solute between bulk water and micelles, it should be positive.
era and Fischer1’71pointed out to the severe inconsistency of the retention stated above, to describe the behavior of tib bind in^ pos~latedthat asthe repulsion forces hinder the molecules of a solute to come into close contact with the molecules of s u ~ a c ~a tpart , of thes~tionaryphase with adsorbeds u ~ a c tbecomes ~t inaccessible. This a~ m e ~that s only a action of the real volume ofs ~ t i o n pha the phase ratio in the column. The relative re~uctionin acc the amount of adsorbed s u ~ a c t ~ t
(5.10)
imilarly, a fraction of the volume of mobile phase could notp a ~ i c i ~ ain. te the interacti~nswith the solute: (5.l l) mo
where fs and fmare the fractions of stationary and mobilephases inaccessible to the solute, respectively, which depend on both solute and s u ~ a c t ~ t . ~ c c o r ~ton this ~ model, four equations were f o ~ u l a t e d correspond in^ to diflerent si~ations.The first coincides witheq. 5.8, which is valid for solutes that are not excluded from both s ~ t i o n phase a ~ and or solutes repelled by the adsorbed monomers of s u r f a c t ~ on t the s~tionaryphase, but associated with the micelles:
(5.12)
whereas for solutes repelled by both the stationary phase and micelles:
similar equation can e obtained for solutesnonexc d by the micelles = 0 in eq. to antibin~ngsolutes. It is not likely, however, the stationary phase bythe adsorbe ciated withthe surfactant in the mobile phase, or vice versa,owing to the mherently similarnature of the interaction forces between the molecules inthe two phases.
us
inally, all the equations describing the retention in f o ~ u l a t e das:
where the coefficients c. and cl will have a different S n according to the nature of the interactions in the stationary phase and micelles.
~uations5.1 5.6 and ,7 predict decreased retention concentration, if the solute pa~itionsto the micelle. ount of surfac~nt7 over a wide range ofs u r f a c t ~ concentration, t l show that below the cmc th will be little change in retentions unless re are ion interaction effects. the other hand7at very h concentration, the solute can elute near the dead increases in concentration will not lower the retention. taken near this region might deviate from theory and should *
for a he e~uationsof retention have been verified experimentally er of solutes (nonpolar7polar, neutral an
1
(anionic, cationic, nonionic and zwitterionic), and column materials (CS, C 18 and cyano). They are also valid for mobile phases containing constant a amountof organic modifier.However, it should be noted that certain assumptions have been made to formulate the equations, regarding the micellar and chromatographic system. It was assumed that:
(i) the cmc and aggregation number of micelles are not affected by the small mounts of solute being chro~atographed, (ii) binding of the surfactant to the stationary phase does not appreciably affect the retention timeof the solute or, if the retention is affected, this binding reaches a level of saturation at or before the cmc of the surfactant is reached, (iii) an increasing concentrationof micelles does not alter their aggregation numberor geometry, (iv) the stoichiometry of the solute-micelle 'lcomplex" is 1:l.
In cases where these assumptions are not valid, deviations from the expected behavior can be found. In fact, the three latter assumptions are only reasonable at a limited range of low surfactant concentration. For instance, adsorption of SDS does not occur on naked silica until 0. l M concentration is reached. Therefore,the model shouldf d with silica as the stationary phase. In fact, it was observed that the change on the silica sudace due to SDS adsorption caused a variation in the KASvalues of several nonionicand ionic solutes, so that the plots of lk vs. SDS micellar [IO]. concentration were no longer linear At high concentration, some surfactants such as CTAB, undergo "sphere to rod" transitions. Also, upward curving plots will occur with 12 stoichiometry or greater. Apparently, owing to these reasons, nonlinearity was observed in the lk vs. total SDS concentration plots for some amino acids and peptides (Fig. 5.4) [l l]. Two different lines were obtained for
IT1
YST
1
these com ounds, depending onthe examined range (0.02-0.08
0.
0.2
.SDS concentration alanyl-tyrosine (AY), han (W), eluted at pH sing two different ranges of surfactant concentration: Solid lines, 0.02-0.08 M SDS; dashed lines, 0.1-0.2 M SDS.Reprinted from Ref. 11 with permission of Elsevier.
Finally, deviations fromthe model are observed for highly and low retained solutes. In the former case, irreversible adsorption of the solute on the stationary phase may occur, whereas in the latter, the elution of the solute willalways take place in the dead volume. The high errors obtained for hydrophobic compounds mayarise, at least partially, from using large
concentrations of surfactantwhch are needed to elute these CO ase column, in a reasonabletime.
.
a) ~ n ~ ~ n c eof~ ti e nti ton ~
t ~ e n ~ t ~
ne of the most serious problems of pure micellar eluents is their we elution s t r e n ~ h often , lower than the elution stre of methanol-water obile phases. Shorter retention times can be achieved by using higher s~rfactantconcentrations andshorter chain-len~hbonded stationa~phases. The elution strength can also be increased by the a ~ ~ i t i oofnan o solvent. The term hybrid was proposedfor t e ~ eluents a ~ of water-o solvent-micelles, and will be used here[121. of l - p r o p ~ oto l mi cell^ mobile addition of small per ey et al. [U],to enhance the first r e c o ~ e n ~ e d chromatographic ef~ciencyand decrease the a s ~ eof t ~ e then, several organic solvents have been stu
micellar medium,it increases to ca. study was made onthe effect of different organic additives (i.e., diols, alkanes, alkylnitriles, and di~olara rotic s o ~ v e ~ t s such as dimethyl sulfoxide and dioxane) upon the elution st
hexadecyltrimeth~la~onium chloride ( phases, for two neutral test solutes, benzene and 2-eth~lanthra~u~oney retained on aC 18 stationary phase [171. The test solutes were chosen due to the fact thatbenzene is relatively water soluble and 2-ethylanthra~u~one is virtually waterinsoluble. Thus, they represent two extremesin hydrophobicity. The results indicated that the presence of alcohols, diols, alkylnitriles?and dipolar aprotic solvents, in the micellar mobile phases, resulted ina diminution ofthe retention factors for the two test solutes. contrasty the presenceof the alkane additives (i.e.,pentane, hexane, cyclohexane) did notgreatly alter the retention. 0.25
0.20
I/ k 0.15
0.10
0.05 0.07
0.08
0.09
0.10
Ef5ect of the hydrophobici~of alcohols obilephases containing 3% l-propanol
1.
0.11
0.12
0.13
the retention of arniloride, eluted
an
LV
The use of alcohols as additives in MLC can result in dramatic reductions in the elution times of solutes, compared to those observed with pure aqueous micellar mobilephases. The retention factors decrease as the carbon number (hydrophobicity) ofthe alcohol is increased(Fig. 5.5),and the reduction is more pronounced for micellar solutions containinggreater mounts of the additive (Fig. 5.6). The organic modifier effects uponthe retention howeverare attenuated as the surfactant concentration is increased. n the other hand, the eEects are greater for more hydrophobic solutes. Thus, thereductioninretention factors observeduponthe addition of organic modifiers to micellar mobile phases depends, not only upon the identityandconcentrationof the organicmodifier, but also uponthe concentrationof surfactant andthe hydropho~icity of the test solute. mounts oforganicmodifier can greatly affect solute LC. This is particularly importantwhenattempting to separate very hydrophobic components. 0.3
EfXect ofthe concentration ofalcohol on the retention of: (a)naphthalene, and (b) 2,3-benzofluorene, inSDS mobile phases modified with l-butanol. Reprinted from Ref. 20 with permission of Elsevier.
MlC
"le elution strength of the alcohol is usually greater than that of the surfactant. However, for positively charged solutes such as catecholamines in acidic medium, anionicsu~actantscan originatean important reduction in the retention due to the hgh aE1nity of the solutestowards the negatively charged micelles [2 l ~. b) ~ i s ~ ~ ~ cof~ art r niteionn in^ ~~ ~ u i ~ i b r i ~
The relative effectiveness of an organic modifier,in. LC appears to be directly relatedto its own ability to partition and bind to the micellar pseudophase. That is, the bettertheorganicmodifier binds to the micellar assembly, the greater its ability to alter the retention ofsolutes. Eq. 5.9 is also valid for hybrid micellar eluents (Fig. 5.5). For these eluents, solute binding c o n s t ~ t to s micelles and the partitioning intothe stationary phase both decreaseas a result of the addition of the modifier, especially hghly for hydrophobic solutes. However, the Km/KMratio increases and, therefore, the elution power of the mobile phase isgreater. The addition of alcohols to micellar mobile phasescauses changes in certain micellar properties, suchas the aggregation number andthe cmc are of thes u r f a c ~.tThe observed reductions in retention in hybrid systems however too large to berationalized in termsof these changes. The reductions should be explained by the modification micro-enviro~ent of the of the micelles and the stationary phase.
MLC retention may be due to the The organic additive effects upon. modification of the micelle produced by the introduction of molecules of modifier in its palissade, and to the change in thenature of the surfactantmodified stationary phase. The addition of organic modifieralso increases the aE1nity of the solute forthe bulk aqueous solvent in the mobile phase, which becomes more nonpolar. This significantly alters the equilibrium of the solute away from the micelle and stationary phase towards the bulk aqueous phase which becomes more nonpolar.
The concentration of organic solvent added to the micellar mobile phase should not be very high, since this might reduce the role of micelles and
bring the system closer to an ~ueous-organicsystem. Large concentrations of modifier can totally disrupt the micelle structure, as the h~drophobic effect, the maindriving force for micelle formation, is reduced. The maximum allowable concentration depends on the type of modifier and surfactant, and is usually not known. As a rule of thumb, the voluie percentage of organic solventshouldbe kept below15-20%. separations reported in the literature were carried out with hybrid mobile phases containing too large amounts of modifier, the role and integrity of micelles were thus not entirely clear 122-241. en hybrid micellar eluents were first used, they were severely criticized. It has been argued that when organic mo fiers are used, this type of chromatography loses some of its appeal 1251. owever, most reported procedures in MLC, since 1985, utilized these eluents. The experience acquired in our laboratories on the development ofanalytical procedures for the deter~nationof different drugs (e.g., P-blockers, diuretics, narcotics, steroids, stimulants and sulfonamides)has shown that, in most instances, the retention ofsolutes with pure micellar eluents is excessive, which forces one to add a modifierto achieve adequate retention times. It may seem logicalto assume that the separation mechanism with aqueous surfactant-organic solvent mobile phase is similar to that with conventional aqueous-organic solvent, rather to pure aqueous surfactant mobile phase, since the effect of adding an organic solvent is to sacrifice the role and effectiveness of micelles. However, as long as the integrity of micelles ismaintained, addition of modi~er a to a micellar mobile phase will not create an aqueous-organic system, even though in hybrid systems the interactions between solutes and micelles are reduced, and the s t a t i o n a ~ phase is more similar to that of a conventional aqueous-org~icsystem. In effect, organic solvents modifythe structure and CO ase, as they solvate the hydrocar~onaceousbonded phase. [26] studied the adsorption isothermsof different organic phases and showedthat over 90% of the C 18 phase was covered with l -propanol at a concentration of 3% (w/v). "herefore, with a micellar mobilephase that includes a small concentrationof organic solvent, the s t a t i o n a ~phase resembles a solvated phase in an aqueous-organic system, rather than in a pure aqueous micellar mobilephase.
The c~omatographic ehavior of alkyl homologousseries is useful for the investigation of retention. h aqueous-organic systems, th of a homologous series is related to the number of carbon atoms, for each solute inthe series, through the following relations hi^ [L271:
(5.15)
where n, is the number of carbon atoms in the homolog., a nonspeci~cselectivity of a me~ylenegroup, and p is the contri retention from the hnctional group c o ~ o to n the series. increase in retention due to the addition of a methylene group as a measure of hydrophobic interaction ina given In contrast to this behavior, severalauthors have notedthat a linear relationship between k and n, seems to exist in LC, when either a pure aqueous micellar or a phase used is [28, 291. hasobserved rij@ been 35 as micelle-for 8 s t a t i o n a ~phases, and n-alkylben surfactants with 2-al~ylant~aquinones and ~-alkylphenones ashomologous series. been suggestedthat a different retention mechanismex series inan aqueous surfactant-organic solvent mobile an aqueous-organicsolventsystem,and that ure hybrid micellar mobile phasesare similar. d from the linear k vs. n, relationship, i of solutes of the homologous series will be e isocratic mode, com~aredto that of conventiona . This behavioris hrther c o ~ e n t ine ~
Fritz [30] studied t e t r ~ h e p t y l a ~ o ~bromide um
effect of some surfactants [i.e.., S, ~octylsulfosuccinate,
rij@ 30), polyoxyethylene(~0)sorbitan ethylene-polyox~ropylenecopolymer 1,2-decanedioll, d inconventional nitrile-water and methanol-water mobile phases with C18 columns. er the workingconditions,the f o ~ a t i o nofmicelleswasunlikely. S even in pure water because it has a wever, the presenceof su~actants as additives in aqueous-organic mixtures greatly improved the separation of several groups of compounds: alkylben~enes, polycyclic aromatic alkylphenols, and some otheraromatic compounds. ns obtained without thes u r f a c ~ t sshorter , retention times and sharper peaks were obtained. The retention times of late-eluting S were reduced by a larger percentage than the retention of earlier S effect is similar to that obtained with gradient elution but in h s case isocratic elutionwith an aqueous-organiceluent,containing an rfactant, was used. raphy can be considered a bridge between C, because surfactants were used at concentrations above their cmc in aqueous solution, but without micelle e observed additive effect couldarise in two ways. First, the surfactant molecules might be adsorbed on the s u ~ a c eof the stationary e, with the long alkyl orpolyox~ro~ylene chains interacting withthe chains and the hydrophilic head groups sticking out. This would give more hydrophilic surface and thus reduce the retention times of solutes. econd, hydro~hobicinteractions might be established betweenthe solute molecules and the longalkylorpolyoxypylenechainsof the discrete surfactant molecules,in the mobile phase. drogenbonding rolewhenbothsoluteand surfactant containpotentialhyd @ 30 o ation centers,as in al~ylphenolseparations with Brij Inthe exper~entalconditionsofthese separations, no adsorption of molecules could occur on the surface of thestationary phase, the second mechanism wasthus probably correct. Increasingconcentrationsof s u ~ a c in ~ the t aqueous-organic solvent eluents resulted in progressively lower retention factors. stingly, eq. 5.9 was followed for these mobilephases, however, instead concentration of micelles, the concentration oftotal sur
used. The treatment assumeda 1:1 association between solute ands u ~ a c ~ t , and consequently, the linear range was limited to rather low concentrations of surfactant. On the other hand, a linear relationship was found between log k of alkylbenzenes and a~ylphenols,and the carbon number of the series,for all surfactants studied. Ths relationship suggests that the retention mechanism ofthe systems agreed withthat found in conventional aqueous-organicLC,rather than in MLC.
JK 1. A ~ v a n t a o~fethe ~ MLC Approach There is a significant amount of experimental support for the three-phase model in MLC. First, plots of retentiondata fit the theoretical e~pressions (see Section 11.3), for a variety of solutes whch bind to micelles. ~econd, the experimental values ofKm and KAsare independent of the s ~ t i o n a ~ phaseandmicellarsystem,respectively. F u ~ e ~ o rmicellar e, binding constants evaluated byMLC are in good agreement with those obtained by otherindependentmethods (e.g., spectroscopic, potentiometric, kinetics, solubility, etc.) [S ,11,3l]. As an example, Table 5.1 shows the c~culated ,,P and PM, and related binding constants, for several partition coefEicients phenols, quinols and catechols, obtained from chromatographc elution with SDS, compared to previous data. The agreementis satisfacto~for phenols where spectrophotometric methods were used. Slight differences, however, are observed between the chromatographic and kinetic data for quinols and cons~ts catechols. Appenbx I11 lists the partition coefficients and binding that we found preparingthis book. According to eq. 5.9, a linearized plot of I l k vs. [M] yields a slope
KAhn/KAs (units M” or L/g) and a dimensionless intercept1KM. Thus, the
binding constantKm (M-’ or L/g) can be calculatedas the ratio of the slope to the intercept, and KA, coincides with the reciprocal of the intercept. eme em be ring that Km = U ( Pm 1 ), thecalculation of the
-
Partition CoefXicients (Pwsand Pw) and Binding Constants (Km) for Several Compounds Eluted with Sodium Dodecyl Sulfate Mobile Phases on a C 18 Column [31] compared to non-chromatographic methods.
Compound Phenol or Benzene- l-ol 4-~ethylbenzenel-ol 4"~thylbenzene-1-01 4-~-Propylbenzene-1-01 4 - ~ - ~ u ~ l b e n z e1-01 ne3,5-~i~ethylbenzen~-l-ol 2,4,5-Trimethylben~ene-l-ol
28.6 75.8 174.2 389.0 69.4 179.2
2,3,5,6-Tetr~ethylbenzene-l-01
(460) 6.3 17.6
Catechol or Benzene- 1,2-diol 4-~ethylbenzene-l,2-diol 4-~yanobenzene-l,2-diol 3-~-Propylbenzene-1,2-diol 4-~-Bu~lbenzene-l,2-diol Hydroquinone or Benzene-1,$-diol
2-~ethylbenzene-l,4-diol 2-Chlorobenzene-l,4-diol 2-Phenylbenzene-l ,&diol 2 - ~ - B ~ ~ l b e n z e1n,&diol e2,3,5-Trimethylbenzene- 1,4 -diol
a
9.7
5.6 75.2 156.2
1.7 3.0 4.7 32.4 50.2 9.4
40.2 95.0 245.0 450.0 995.0 229.6 513.6 (1 150) 27.4 72.4 28. l 225.0 43 1.0 14.5 21.1 34.0 202.0 244.4 53.8
10.2 24.4 63.5 116.7 258.4 59.5 133.3 (300) 6.9 18.6 7.1 58.3 111.8 3.5 5.2 8.6 52.2 63.3 13.7
8"
9.9" 24" 65" 140" 280" 64" 125" 200" 4b lob 5b 37b
75b 3b 6b
7b 85b 65b 12b
Spectrophoto~etricmethods; Kinetic methods, in parenthesis, extrapolated values.
ensionless water-micelle partitioncoe~lcientis trivial if the correct unit views have been published on the evaluation of these C [6, 321. ~ n f o ~ n a t e l the y , d i ~ e ~ s i oproblem n was
Km Binding Constants in Llrnol for Diverse Neutral Solutes in SDS and CTAB Micellar Systems.
Co~poun~ E21
PI
References
E101
E111 1291 E311
P 3 1 1341 E351 1361 E371
mobile phases 5325 Anthracene Benzene 25.8 18.9 23.520.319.217.1 cohol Benzyl 11.8 8.8 10.4 Na~hthalene 23 242 217 5 245 353 290 Nitrobenzene 22. l 23.2 23.1 9.4 Phenol 10.2 9.6 9.5 Pyrene 8065 Toluene 52.9 50.0 63.3 54.2 59.3 52.4 56.5
5440
Compound
25.9
10.5
76.1 44.8
References ~71
Benzene 47.2 40.2 27.2 ~enzonitrile 19.4 17.5 ohol Benzyl 12.8 104 Chlorobenzene 548 54.7 ~itrobenzene 36.9 27.0 38.5 Phenol 79.5 Toluene
[l01
35.9
l291 ~361 ~331
l381
23.7 17.3
13.5 157
71.4
83.3 204 142 129 142
Although bindingcon~ants(Krn) are usually reported,it should be noted that the figure depends on the units of the partial specific volume of monomers of surfactant in the micelle (Llmol or Llg). It may be more convenient, therefore,to indicate water-micelle partitioncoe~lcients(
A nonexhaustive literature compilation of these coefficients, as measured by LC, is given in AppendixIII. The binding constants are listed addressing e unit problem. In some cases, the partition coefficients were calculated in the referencedarticles. It can be observed from the binding constants given that chemically bonded reversed-phase columns were generally employed, with octadecylsilica and octylsilica stationary phases mostly used. Sometimes cyano, C l and silica columns werealso used. Binding constants SDS and C T D micellar systems. Nonionic have been calculated mainly for and zwitterionicsurfactants have also been studied, although only to a small extent. Most work was done using pure and unbuffered micellar mobile phases at room temperature. Whena modifier was introduced inthe mobile phase, it consisted of a short or medium chain alcohol, such as methanol, ethanol, propanol or butanol, or a salt such as sodium chloride at low concentration, When pH was fixed, a phosphate buffer wasusually used. ome solute-micelle binding constants for aromatic compounds with CTAB, obtained by differentauthors under identical experimental conditions, are included in Table 5.2. Only the chromatographic column could change (octadecylsilica and octylsilica). Good agreement exists when different Km values are available, especially in thecase of S even for veryhydrophobiccompoundsfor whch the error inthese dete~inationsis usually high. The evaluation of solute-micelle binding constants has important implications outside thefield ofchromatography, suchas micellar catalysis, tertiary oil recovery, and enzyme and membrane modeling. powerhi technique for thedete~inationof these constants, in comparison to classical methods. The use of MLC inthis field has several advantages:
(i)theevaluation can bemadefor allthe compounds experiencing a c~omatographicretention inthe system that varies when the concentration of surfactant in the mobile the compounds phase is changed. %S is quite common and do not need to experience a change in their spectroscopic characteristics in micellar media, as when spectroscopic methods are used,
(ii) simultaneous determination of the constants of several solutes is possible, as the retention timesof a mixture of the solutes can be measured from a single injection. (iii) solute concentrationneed not be know, and (iv) impurities, present ifsample, the in are chromatograp~callyseparated and will not interfere.
Moreover,bindingconstants can beobtainedfromindependent sources to predict retention factors. This is usehl to facilitate systematic optimization. h example of this type of prediction has been given for perylene which showeda high retention and the interceptof the Ilk vs. [ plot was essentially zero with aG 18 colum and SDS in the mobile phase [S]. The retention factor at a given micelle concentration and a reported value ofthe binding constant were taken to estimate the value of the intercept in eq. 5 -9, and tohrther predict the retention at varyingmicelle concentration. The agreement between experimental and predicted values was excellent.
It? 2. S o ~ ~onsi~erations e on
art it ion in^ ~ a ~ c ~ ~ a t i o n
a) Critical ~ i c e l l eConcen~ration For any aqueous surfactant solution, there is a relatively small range of concentrations below which virtually all surfactant is present as monomers, and above which virtually all additional surfactant is present in micellar form. This micelli~ationphenomenon causes significantchangesin the bulk physical propertiesofthe solution. The cmc an is important parameter which deserves careful attentioninorder to minimize uncertainties in measurement.
The cmc values for the most widely used surfactants in MLC are wellknow(e.g., 8,1~lO-~MforSDS, 1,3~lO-~MforGT~,and 1.0 rij@ 35) [39]. However, in a hybrid micellar mobile phase where an organic solvent is present, the cmcvalueschange and are often not
referenced. Therefore, they should be evaluatedprior to the calculation of binding constants.
ifferent modifiers show different behavior. A pro~ressivedecrease in the cmc has been observedin S~~ solutions, with respect to the absence of modifier, for all alcohols, methanol excepted [40]. The reduction in the cmc is larger when the length of the alcohol chain is increased, and occurs even for small amountsof alcohol. At moderateconcentrations of propanol, butanol and pentanol,the concentration of micelles almost coincides withthe total concentration of surfactant. Similarly to methanol, the cmc increases with the concentration of organic solvent for acetonitrile and t e t r ~ ~ d r o ~ r ~
l401.
I
7
-iD
-5.0
4.0 "3.0
Loll [w-351
-2.0
Log k vs. log
(0)aniliniumchloridein(a) only, and (Q) sodium benzoate in (b) only. Reprinted from Ref. 41 with permission of Elsevier.
e determination of cmc values is made by the measurement of some property of the solution, as a h c t i o n of surfactant concentration. the extrapolated segments o cmc is taken as the point of intersection of easured propertyat concentrations ofsurfactant below and above the cmc. ne of the properties that has been used to measure cmc values is the retention in a micellar mobile phase of a neutral solute strongly with micelles[4 l]. Figure 5.7 shows plots oflog k surfactant concentrationforacetophenone, a n i l i ~ u mchl nitrobenzene, phenol and sodium benzoate, and the three su and BrijB 35. Thesolutions also contained 3% l -propanol to e the efficiency of the chromatographicpeaks. The d r ~ a t i change c in retention as micellization occured is quite obvious.The values of cmc for the three surfactants were calculated from these plotsand the results were compared to conductimetric measurements. A relatively good agreementof the chro~atographicand literature values(chromatographicmethod: 4 . 7 lo3, ~ 3 . 6 10” ~ and 3.43 10-nductimetricmethod: 7.3 x l Om3, 1 . 3 ~ 1 0and - ~ L O X ~ OM, - ~for S and BrijB 35, respectively) was evidence that it is truly solubilization of the solutes by the micelles that causes elution, at relatively high concentration ofsurfactant. b) ~ o l u of ~ ~et a t i o ~ aPhase ry
or convenience, the partition coefficients ,,P and ,,P are U considering the entire volume of the silica support particles. in mind, however,that only the surface bonded layer isa the pa~itioningequilibria. The calculation of ,P from eq. 5.1 requires ,, which cannot be determined easily. Usually, the difference knowledge of V between the empty column volume and the packed column void volume is taken as Vs ,which gives an overestimation ofthis volume since it includes the entire volume occupied by the silica solidsupport particles, rather than just the true stationary phase. Although theuse of such a value will result in accurate values for ,,P ,the values obtained for,,P and expected to be significantlyin error, withunderestimated partition coefficients. An approach that completely excludes any volume associated with the base silica material should thus be used.
1
orgerdinget al. E291 developed an accurate procedure for d ~ e the c~~ o mg a t o g r a pphase ~ c ratio.In thw procedure, measurementof the total weight of the paclung material in the colurnn, and ofthe cumulative pore volumes of unbonded silica support and bonded C 18 packing materials, before andafter exposure to the m i c e l l e - f o ~ gsurfactant, is made. In this way, any volume associated with the base silica material is completely excluded, sincethe stationary phase volumeis assumed to be that portion of the silica pore volume filled upon bonding the odadecylsilane (or other alkyl ligand) chainsand, in the presence of micelles, by sorption of the surfactant to the C 18 chains.
c) ~ r r o r in s the Slopeand~nterceptof the ~ q ~ a t i of o n~ e t e n ~ o n The evaluation of and Km is usually made from linear regression of the experimental data of retention factors and micelle concentration, accordmg to eq. 5.9 (remember KAs= 4,Kws). It is evident that K,, will be affected by a larger error than KAs. Considering that: K,,, = NB
(5.16)
where A and B are the slope and interceptof the straight-line, the error will be givenby:
AKm = (B AA + A AB)/
(5.17)
being the errors in the slope and intercept, respectively. It should be noted that the error in K,, is not only affected by the errors in the slope and intercept, but also by their values. The intercept ofthe straight-line decreases(KM increases) withthe hy~rophobicityof the compound. High unce~aintiesin KM and even higher in K,, ,will result for compounds showing very low intercepts (close to zero).These are compoundswith an importanthydrophobic character,
IT1
experiencing a high retention in the MLC system (e.g.,anthracene pyrene). The high unce~intiesin the retention factors of hy compounds often yield negativeintercepts, that is, negative values for and K&. It has been shown that the application of weights to the linear fitting (eq. 5.9) of (l k , [M] )experimental data, or the use of nonlinear regression (eq. 5.8) improves greatly the accuracy of the binding cons [42]. With these data treatments, achievement of negative interce frequent. The statistical treatment of lk vs. [ plots indicates that the error a compound having either in the slope tends to increase when working very small or very large binding constants. There are, therefore, ce limits to the applicability of the procedure with a given stationary ph~se. Consider, for example, the retention of naphthalene with a C 1 phase column [l]. Because of naphthalene's great affinity for the s t a t i o n a ~ phase andlow affmity for water, an appreciable concentration of micelles in the mobile phase was needed to achieve elution in time. In fact, it was difficult to obtain a su~lcientamount enough surfactant concentration to calculate an accurate @usstatiomry phase. In constant for n a p h ~ e n eusing , did not interact as strongly with a cyano stationary sufficient data could be collected to calculate the bindi the slope of the lk vs. [ ] plot was large and the intercept close to zero, thus, small changes in the slope ofthe line resulted in relatively large changes in the intercept. n the other hand, the interaction of hydroquinone with a ionary phase was very low. This lack of interaction was even more pronounced with the cyano column [l]. Consequently, the retention times of hydroquinone with a cyano c o l m were not sufficiently di~erentor consistent inthe presence or absence of micellesin the mobilephase. particular case, it was difficult to collect adequate data for proper tre Limitations such as those described for naphthalene (with the column) and hydroqu~one(with the cyano column) would be moreserious if one were limited to a singlestationary phase. Fortunately, several ~fferent polarity bonded stationary phases are available. As a result, a very wide variety of compounds can be chromatographed using micellar mobile phases
and treated accordingly.It has been proposedthat a stationary phase should be changed for a more polar oneto decrease the errors in the evaluation of S for hydrophobiccompounds(the order of increasing polarityisthefollowing: C 18 C8 C 1cyano silica). This should decrease the partition coefficients betweenthe aqueous phase andstationary phase, and increase the intercept of the straight-line.It was however checked that the reduction in the errors was minor when a C8 column was used instead of aC 18 column[321. The advantage of usinga more polar colurnn the compounds out ofthe c o l u or ~ reducing is still the possibility of eluting their retentiontimes. If the solubility of a hydrophobic compoundthe inaqueous phaseis increased by the addition of an organic modifier, such as an alcohol, the dis~lac~ment of the equilibria towards the aqueous phase will enable the calculation of the binding constants. Ineffect, the additionofalcohol AS and KM ,and the associated errors. Figure 5.6 shows the increase in the intercept of the straight-line of Ilk:vs. [M] for hydrophobic solutes, when l-butanol at different concentrations is added to an aqueous micellar solution. The Km binding constants of several benzene derivatives andP eluted with hybrid §I)§ micellar mobile phases, are given in Table 5.3 for di~erentalcohols at varying concentration. The decrease in KbMdue to the addition, of alcohol to the mobile phase was higher for the most hydrophobic co~pounds.It was also more important with increasing length (decreasing polarity) of the alcohol chain, Finally, Km decreased withthe percentage of alcohol in the mobile phase. It is interesting to note that it was possible to calculate the constants for all compounds in the presenceof butanol. This is in contrast to the results obtained when no additive was utilized the in mobile phase. an alcohol usually decreasedKm ,regardless of the romatic compounds, this has been checkedfor §D§ wever, in the presence of cationicCTAB micelles, the binding constants were higher than in the presence of anionicSI)§. Th~s can be explained by the favorable electrostatic interactions between the positively charged CTAB head groups and the unlocated charge of the
aromatic ring of the solutes. In fact, the difference in Km values with ,relative to SDS, was greater for naphthalene derivatives.
. The approximately zero intercepts obtained for hydrophobic compounds, such as the bgher molecular weight homologues, merelyreflect their large aEmity for the micelles ands u r f a c ~ t ~ o a t stationary ed phase, c~mparedto bulk aqueous component in the mobile phase. Solubility data inhcate that these compounds are virtually insoluble inwater. This ~ n d i n gimplies that as the amount of such hydrophobic solutes in the aqueous phase is almost negligible, they can only be transported between the micelles in the mobile phase and the surfactant-modi~edstationary phase by a direct transfer process, and no partitioning will take place between aqueous phase and micelles, or between aqueousphase and stationary phase. Therefore, only a single equilibrium willexist, which is described by P,, (Fig. 5,8) [29]. Although mentioned in theories of pseudo-phase c~omato~raphy, the possibility of a direct transfer of an insoluble or a sparingly water-soluble solute, from the micellein the mobile phase to the su~actant-coated stationary phase, was largely ignored. For this situation, a m o d i ~ form e ~ of quation of Armstrong and Nome (eq. 5.l) was derived [29], which successfullyaccounts for the dependence between and k [M], observed inthe elution of such hydrophobic solutes:
The drawback of the solubility limit theory is the dif~cultyin defining an exact limit for this behavior. Also, if the intercept in eq. 5.9 is to zero, the binding constants Km and very small, although not equal PS, values can be be calculated, but the errors will be very high. Only evaluated for very hydrophobic solutes. It was checked that this pa~ition
coe~cient,determined at any given micellar mobile phase composition fiom for othermobilephase eq. 5.18, is useful to predictsoluteretention concentrations with little error, over the entire alkylbenzene homologous series range [29].
Representationofthe direct transfer process for distributionofa solute between the micellar pseudo-phase and the surfactant-modi~edC 18 stationary phase. Reprinted from Ref 29 with permission of the American Chemical Society.
It can be observed, inFig. 5.6, that at low alcohol concentrationthe retention m e c ~ s m is a directtransfer between the micelles andthe modi~ed stationa~phase, andat the highest alcoholconcentratior~s,perhaps owing to an e~ancementin the solubility, the dissolved portion of solutes in bulk solvent interacts with both micelles and stationa~phase.
. VL 1. ~ ~ a nin~Acid-Base e s ~ e ~ ~ viniM o ir c e ~ ~ ~ r M e d i ~ The retention of weak organicacids and bases is affectedby the pH of the micellar mobile phase. Solute-micelle binding constants of the dissociated or cationic micellesare and undissociated forms of a compound with anionic
different.Smallchangesinp can thus significant1 alter the chromatographic retention, particularlywhen the mobile phase is close to the value of the roto on at ion constant (log KH ). Therefore, fo c o m ~ o u n ~the s, must be specified retention data are values in micella a is c ~ c i a for l a o etention mechanisms des and in better understanding optimi~ationstrategies in The influenceof micelles on the acid-base properties of a n ~ ~can e s be significant andis reflected in changes in logI C H . A good example is the acid-base behavior of amino acids and peptides, which present the f o l l o w ~ g equilibria [44, 451:
espite large structural diEerences among amino acids and peptides, their log KH values in bulk aqueous solution are very similar. In micellar media, the eEect of chemical s t ~ c ~on r elog K, is much more ~ r o n o u n c e ~ of solute-mi (Table 5.4),which is indicative of the profound influence interactionsontheacid-base behavior of thesecompounds [44J apparent protonation constant in micellar media is expressed as:
are thebindingconstants where Km andKm undissociated species, respectively.
of the dissociatedand
As shown inTable 5.4, both carboxylateand amine groups of amino
acids andpeptides are weakeracidsin §D§micelles. The contributo~ factors to the positive log K, shiftin an SDS system are, first, the electrostatic repulsion between the carboxylate group and S
1 Tab! protonation Constants of Several Amino Acids and Peptides Measured Potentio~etrica~~y in Aqueous and Micellar Solutions[44].
omp pound
Ionization Step
Glycine Aspartic acid
Glycyl-~lyc~e Benzoic acid Lysine
Lysylphenylal~e
1 2 1 2 3 1 2 l 1 2 3 l 2 l 2 3 1 2 1 2
0.1 M
0.1 M
log Kma l o g k b Alog Kc log kbAlog Kc 9.74 9.72 -0.03 2.34 2.57 +0.23 9.80 9.76 -0.04 3.45 3.68 +0.23 2.04 2.39 t-0.35 8.08 8.22 +O.14 2.89 3.34 +0.45 4.15 4.76 +O .47 3.18 -0.97 10.56 10.55 -0.0 1 9.32 +0.46 9.78 +l S9 3.88 2.29 +o.02 8.71 -0.54 9.23 9.25 0.00 + l .86 2.21 2.21 4.07 10.45 10.26 -0.19 +0.68 8.03 7.35 +2.18 4 '94 2.76 -1 -03 8.18 +OS7 9.2 1 9.68 -0.21 2.11 +2.27 4.59 2.32 d d 8 -06 4.73
I&,,,,=Protonation constants in nonrnicellar solutioncontaining 0.1 M NaCI; &,= Protonation constants in micellar solution containing 0.1 M NaCl; c Shift of protonation constant, The solubility wastoo low tobe measured potent~ometrically.
a
second, the hydrophobicand electrostatic attraction of the alkyl and protonated amine groups to the SDS micelles; andthird, the localization of protonated weakacid, 'HA-, in an e n v i r o ~ e nwith t lowerdielectricconstant. In addition, the molecular size(andor aqueous solubility) of the compounds also seem to play a role in in~uencingbinding constants, and consequently, e log KH the mag~tudeof the log KH shift. As expected,the m a ~ ~ ofdthe shift is larger for compounds experiencing greater interactions with S micelles (2.e.,log KH = 1.6-2.3 for amino acids andpeptideswith hydrophobic and basic groups, as compared with log KH = 0.2-0.5 for the more polar compounds). In cationic CTAB solution,the shft sign of logKHis opposite of that in SDS solutions [44].The compounds are stronger acids in GT media than in aqueous solutions. This is probably due to the strong electrostatic attraction of the negative chargeon the conjugate bases, to the cationic head groups of CTAB micelles. Incontrast to SDS, the shift in log KH1 is larger than in log Km . The large log KHshift of benzoic acid in CTAB solution.ascompared to the shifts observed for amino acids, and to the same compoundin SDS solution, can be attributed to the strong electrostatic attraction and hydrophobicinteractionsof the conjugate base of benzoic acid, and perhaps more importantly, to the lack ofelectrostatic repulsion whch is evident for the other solutes. Apparent protonation constants were evaluatedfor several solutes in SDS micellar solutions, from MLC retention factors at varying concentration of micelles and 1 -propanol [46). It was observedthat themodifier decreased micelle the values of log ICKapp, whereas these wereincreasedwith concentration. The shifts inlog ICKapp upon variation ofmodifier concentration are causedbyboth the modification of the t h e ~ o d ~ a m i c constants in the aqueous-organic solvent, andthe displacement ofacid-base equilibria due to the modification of the interaction of the solute with the micelles, inthe presence of modifier.
major constraint of silica bonded phases is the limi range of between o yretainedstrong acids and bases, whose protonation constants are either outsideor close to the boundariesof this range. In suchcases, ~troductionof electrostatic to theeluent of a small amountof interactionsthroughtheaddition surfactant, with a charge opposite that of the solute, provides ade~uate retention. Thus, the use of micella DS mobile phases shifts the apparent protonation constants of solutes to milder p conditions, and benefits the observation of the maximal limiting retenti within the operable limits of silica-based columns. Also, the wider distribution of protonation constants of compoundsinmicellarmediagivesa greater selectivity to the chromatographic separations. .S and 7.5, This istroublesomeintheseparation
of the micellar mobile phase is usually buffered using the ric acid-base systems. Potassium ion cannot be used with ,as ~otassium dodecyl sulfate precipitates from aqueous solutions due to its high Kram point (see Chapter 2). columnshouldbeequilibrated themobilephaseuntilbeforeand after the c o l u is ~ onepeakisobservedi a t o g r ~ of s weakacidsand bases because prototropic equilibriaare much faster than the solute-micel~e or solut~-stationa~ phase d ~ a m i c s . *
Cyano-bondedand Cl8 columns interact verydifferentlywith surfactant monomers, resulting in a different elution behavior of organic acids and bases,as a function of micelle concentration and pH in the mobile yano paclungs do not appear to adsorb S S monomers and generally have the capability of retaining various species. ionic C 18 c o l u ~ s are modified by adsorption SDS of monomers, withthe negative headgroups in contact with the mobile phase, such that the surface is charged and will repulse anionic species. Some examples of retention behavior for weak organic acids and bases are shown below.
7
9 Change in chromatographic retention with pH lurnn, at various SDS molar concentrations: (0)0.02, 0.10. Reprinted from Ref. 47 with permission of the American Chemical Society.
a) Cyano C o ~ u ~ n s
For somesolutes, the pH of micellar solutions is ~ofn d ~ e n timpo~ance al in determining the sign of the slopes in the l/k vs. [M]plots (positive, negative or zero). Forbenzoic acid at pH 4.5, the retention factors decrease withincreasing SDS concentration, because ofhydrophobic interaction of the neutral acidic specieswith the SDS micelles, but at pH > 4.5, the retention factors increase owing to electrostatic repulsion between the negative charges of the basic species and micelles (Fig. 5.9) [47]. Hence, completely opposite behavioris observed depending onpH. In the intermediate range of pHvalues, there is an isoeluting point where k is completely independent ofSDS con cent ratio^. This is the p
whch the two species, acid and base in equilibrium witheach other, have the same retention. This is analogousto the isosbestic point in spectroscopy and would be expected to give the protonation constant in the micellar medium (for benzoic acid, log KHis 4.2 in water, and 4.7 in SDS micellar solution, as obtained from the isoeluting point). It has been c o ~ e n t e dbefore, in Section 11.2, that if solute retention is constant with increasing micelle that the solute-micelle binding concentration,h s does not necessarily mean constant is zero. Theisoelutingpoint (50% ionization point) clearly emo on st rates that thelack of change ink, with micelle concentration in the mobile phase, is the result of the balancing or opposing hydrophobic and electrostatic effects caused by the change inthe form of the species. n the other hand, inspection of the retention factors for weak bases using cyano columns andSDS eluents showsthat the largest k values occur in more basic solution, where the neutral free-base form is present, and the smallest in acidic solution where the protonated, positively charged form exists, which has favorable electrostatic attraction to the negatively charged micelles (Fig. 5.10) E471
10 Change in chromatographic r various SDS molar concentrations: Reprinted fromRef. 47 with permissionof the American Chemical Society
b) C18 C o l u ~ n s The effect of pH on the retention factors of solutes, eluted with an anionic surfactant, is very similar on C l 8 columns to that obtained using cyano columns, when hydrophobic interactions dominate. However, less hydroCl8 phobicandnegativelycharged solutes willeluteveryquicklyon columns, because of repulsion from both micelles and negatively charged modified stationary phase, Fig. 5.l 1 shows plots of k vs. pH with the typical
1s C"
6
3
4
=
3
4
5
0
a
Q
pw
7
d
n
7
Modification ofthe retentionwith pH in a C 18 column for: (V) benzocaine?(U) ) furosemide, (A) tyrosine, and (A) sulfanilamide?for 0.05M SDS-8% (v/v)l-propanol (a,b), and0.15M SDS without alcohol (c,d) mobile phases. Reprinted from Ref. 46 with permission of Elsevier.
.11
behavior found when weak acids are eluted fromC 18 columns. of the retention factors with pH is shown for mobile phases of conta~ing8% (v/v) l-propanol and without&us modfier. Inspection of the curves revealsthat the largest k values are achieved in acidic solutions, where the neutral or cationic forms of the solutesare present, and the smallest, in more basic solutionwheretheanionic or neutralforms dominate. The dependence ofk on pH at a constant~ n c e ~ t r a t i oofnsurfactant and modifier is si~moidal,which resemblesthat of conventional acid-base titration curves. The observed increase in retention at lower pH could beascribed to the fact that the interaction of solutes with the surface of the su~actant-modi~ed stationary phase is stronger than with micelles.
Change in chromatographicretention with pH for aniline in a C 18 column with 0.05 h4SDS mobile phases. Reprinted from Ref. 47 with permission o f the American Chemical Society
~ ~ o ~ tofi anionic o n surfactant monomers onthe surface of a s t a t i o n a ~phase also causes protonated organic bases to be retained a longer period of time than the neutral free-base forms, because of electrostatic onsequently, the elution behavior of protonated bases will mimic n C l 8 columns (Fig. 5.12), in contrast to the behavior observed on cyano columns. Figures 5.10 and 5.12 aremirror images of each other.
No isoeluting point exists with G 18 s ~ t i o n phases, a~ but it can be where the change in kper unit change in average log KHvalues obtained from ,with a C18 column, were in good agreement withthose obtained fromthe isoeluting point using a cyano column [471.
aC1, can also influence c~omatographicretention. ation predicts a decrease in activity coe~lcientswith ength, with a c o n c o m i ~ increase t in solute solubility (salting-in efliect). In this context, the terms "salting-in" and "salting-out" apply to the solute becoming more or less soluble in bulk aqueous phase, respectively, and do not refer to the behavior in the micelle. solubility ofionic organic species, as a ~ n c t i o nof salt concentration, depends onboth their ionic andcarbocyclic portion, it will bethe result of a co~binationof electrostatic and hydrophobic eEects. The behavior of antibinding and binding solutes should be consideredseparately.
If antibindingis mainlyan electros~ticphenomenon, one wouldexpect to see de~nite salt effects on this chromato~ra~hic behavior. In effect, m o ~ ~ c ~ t i o ~ of ionic strength might be su~lcient to change completely an antibinding to a bindingtype behavior (i. e. ,salting-out into the micelle).
salt, bromophenol blue isan antibinding compound, whereas the in presence of as little as 0.02 NaC1, it appears to be nonbinding. At slightly higher salt concentration, the compound binds strongly to SDS micelles [48]. In fact, binding of bromophenol blue to SDS micelles increases substantially more than might be expected from the linear increasethe inconcentration of NaCl. The chromatographic behavior of bromophenol blue plotted as k vs. pH,withandwithoutadded salt, is shownin Fig. 5.13 [47]. Large going differences ink occur at high pH, and the two curves tendto co~verge towards low pH values, indicating that salt effects (electrostatic effects) should be larger for anionic speciess dand l e r for neutral species. Also, the elution behavior vs. micelle concentration at high pH with salt reverses, compared to that without added salt. The bromophenol blue anion is less retained using 0.05 M SDS compared to 0.08 M S S, but has the opposite behavior whenNaC1 is added, indicating that the anionic species in the presence of salt behave more like the neutral species (i.e., the sign of the slope ofthe lk vs. [M] plot changes from negativeto positive). For most antibinding solutes,the slope of l k vs. [M] becomes less negative with increasing ionic strength, althoughallnot compounds showthis trend,indicating that more than simpleelectrostatic eEe considered. For example, the interaction of naphthol green micelles appears to be largely unaffected bythe addition ofsalt. Conversely, the slope oflk vs. [M] for thiocyanate ion becomes even more negative with increasing NaCl concentration [48], Therefore,itseems that for the transition from antibindingto nonbinding andfirther for binding to occur, the solute ion must have sufficient hydrophobic character to associate with the nonpolar portion of the micelle, once electrostatic repulsions have been mini~zed.
In lyophobiccolloidal systems, theadditionofsufficient salt between colloidal generally brings about flocculation. h analogy flocculation andthe nonbinding-binding transition has been drawn [48]. The counterion layer around the micelle is narrowed in a solution containing higher concentration of ions, which facilitates appro~imation the of the solute to the micellar assembly. Hydrophobic interactions between the solute and the nonpolar core of the micelle can then be established.
c
.l3
Salt effects on the retention factor of bromophenol bl various pH values. Micellar mobile phase: (0)0.05 M SDS, 0.05 M SDS-0.10 M NaCI, and (A) 0.08 M SDS-0.10 M NaCI. Reprinted from Ref. 47 with permission of the American Chemical Society.
The behaviorof thiocyanate is difficult to explain in terms of electrostatic criteria. However,itmaybe possible to rationalize by considering the negligible hydrophobic character of this ion. added, the predominate factor may bethe ions increasedaffinity for the bulk solution (salting-inef-fect), resulting ina lesser association with micelles and
1
decreased retention with increasing salt concentration. In contrast, many organic ions tendto be salted-out to the micelle.
nly a few studies have been made on the effect of ionic strength on the ehavior of binding solutes [36, 43, 491. The main effect of NaCl is to decrease electrostatic interactions. The salting-out effect corresponds to a reduction ofelectrostatic repulsions byNaC1, but the addition of NaCl also decreases electrostatic a~ractions, In fact, no uniform behavior seems to exist with ionicsolutes. The presence ofsalt has ~ifferenteffects on solutebinding constants, depending on the nature of the c o ~ ~ o u nFor ~. benzene derivatives, in general, the binding constants increased in the presence of NaCl. In contrast, for naphthalene derivatives, these only increased forl-nap~thol, while for naphthaleneand 2-naphthol, they decreased 13(51.
.
The effectof temperature on the retentionofcompoundsprovides t h e r m o d ~ ~data i c thatdescribe the chromato al. [131 proposed the use ofhigh ~mperatures off e~uationto micellar mobilephases:
S" are the standard enthalpy and e solute from the mobile phaseto the s ~ t i o n phas a~
for transfer of solutes from bulk water to the stationary phase )anisole, (A) l-naphthalenemethanol, (0)toluene, and (B) naphthalene. Mobile phase: 0.10 M SDS. Reprinted from Ref. 50 with permission of the American Chemical Society.
ff plots for transfer of solutes from bulk water to micelles for: (+) isole, (0)toluene, (A) 1 naphthalenem met ha no^, and (B) naphthaiene. Mobile phase:0.10 M SDS.Reprinted from Ref 50 with permissionof the American Chemical Society.
consider, however,the multiple equilibria involved inMLC, as described by the three-phase model. Tomasella et al. [34,50] proposed a more rigorous approach that accounts for thetwomainequilibria,andprovides the t h e ~ o d ~ ~associated i c s with eachtransfer. The t r ~ s f e of r solutes from bulk aqueousphase to the stationary phase is described by:
(5.22)
If the standard enthalpy, A As ,andthe standard entropy, Aso~s ,are f temperature over the temperature range of interest, a plot of [M])] vs. 1/T willbelinear(Fig. 5.14). AHoAsdoes not change with increasing concentration ofsurfactant, since mice effect on this transfer. The general trend is a more negative sing hydrophobicity of the compounds, as observed for a group of (Table 5.5) [S l]. This means that the transfer fiom bulk aqueous phase to stationary phase becomes enthalpically favored for more hy~ophobiccompounds. The less negative valuesof AH",, found inMLC, with respectto conventional LC,indicate that the effectof temperature on k is less pronounced with micellar eluents. In addition, the change in enthalpy for the transfer of solutes from bulk aqueous phase to the micellar pseudo-phase can be calculated by the following e~uation: (5.23)
vs. 1Rgives astrai~ht-linewhose slope and intercept will d Asom(Fig. 5.15). Again, AHom becomesusually greater with increasing compound hydrophobicity. However, a change in
tempera~redisplays a minor change in Km ,when compared with the changes that arecaused by the ad&tion of a few percentof organic m o ~ i ~ e r to the mobile phase. Change in Enthalpy (Kcallmol) for Transferof Solutes from Bulk Aqueous Phase to the Stationary Phase and fromBulk Aqueous Phase toMicelles [S l].
~aphthalene
-2.65
-4.49 -6.20
Acenaphthylene Fluorene
-3.08
-4.84
Anthracene
-5.40
-8.19
P h e n ~ t ~ e n e -6.58
-8.76
9-~ethylanthracene -6.12 luoranthene
-9.05 -8.58
Pyrene
-6.88
-9.29
Chrysene
-10.5
-13.3
-9.68 ~enzor~lanthracene
-12.2 -13.9
en~o[~]~uorant~ene enzo[~]pyrene enzo[e]pyrene Perylene
1.8 -10.6
-1
-4.43
-7,05
-1 1.4
-14.9 -12.8 -14.1
-1 1.7
1. 2.
3.
4. 5.
trong andF.Norne, Partiti ar Mobile Phases inLC, Yarmchuk, R. Wein~erger~ Selectivi~in LC with Micellar Mo rial. Chem.,54: 2233 (1982). . Vera, Study of the arina, 0.Jimenez, on Selectivi~of a of Benzene and ~aphthalene ~erivativesin M C , Microchem. J: ,53: 2 15 (1996). .W.A ~ s t r o and n ~ G.Y. Stine, ~ e l e c t i vin i ~Pseudo-phase LC, Anal. em., 55: 2317 (1983). l l a r on LC ~ ~ n ~ a nand a rL.J. t Cline-Love, ~ o d e l f o r ~ i c eEfects ~etention Factors and for ~et~rminationof ~icelle-Solute ~ ~ u i l i b r Constants, i~m Anal. Chem,, 56: 1557 (1984). Foley, Critical Compilation of Solute-~icelleBinding Constants elated Parameters~om~C Measurements,Anal. Chim. Acta, 231: 237 (1990). Jandera and J.Fischer, Chromato~ra~hic Behavior in W L C with A, 279 cellar and Submicellar Mobile Phases, J Chromato~r. 728: ./
~
6.
7.
.G. Herries, W. Bishop and F. Solutes between Micellar and AqueousPhases: ~easurementby Gel ~iltrationand Efect on the Kineticsof some Bimolecular~eactlons, J: Phys. Chem., 68: 1842 (1964). 9. .Armstrong,~icellesin Separations:A Practical and ~ h e o r e ~ c a l Review, Sep. Pur$ eth hods 10. rthod, I.Cirard and C, ten~onStudy of Solutes of Various Polarities, Anal. Chem., 58: 1359 (1986). 11. .Strasters and M.C. Khaledi, Comparatlve Study of the ~eterminationof ~ o l u ~ e - ~ i c e linl edin^ Constants by MLC and ~ i c e l l aElectro~netic r CapillaryChromato~ra~hy, Anal. Chim.Acta, 246: 131 (1991). aledi, J.K. Strasters, A. ~imult~neous ~nhancement of ~eparati~n Selectivi~ and So~vent 8.
13. 14.
15.
16. 17. 18.
19. 20,
21.
22
I
23
Strength in W L C using Micelles in ~ydro-OrganicSolvents, Anal. Chem., 62: 130 (1990). .T.DeEchegarayand J.S . Landy, ~ ~ c i e n c y ~nhancementin MLC, Anal. Chem., 55: 924 (1983). Torres Cartas, M.C.arcia Alvarez-Coque and R. C a m ~ a s~etermination , of Anabolic Steroids in Pharmaceuticals by LC with a Microemulsion of Sodium DodecylSu~ateand Pentanol as Mobi~ePhase, Anal. Chim. Acta, 302: 163 (1995). .Garcia Alvarez-Coque, E.F. Simo Alfonso,G. Ramis steveRornero, ~igh-PerformanceMLC Determination of mides in Pharmaceuticals after Azodye Precol~mn Derivatization,J. Pharm. Biomed. Anal. 13: 237 (1995). ssel, The Roleof the Stationary Phasein ~ C , J. Chromatogr., 449:349 (1988). Jr., W.L. Hinze and F. romatogr., 12: 1367 ( offman and E. Bonvini, HPLC Separation of Ino~itol ersemploying a Reversed-Phase Column and Micellar ~ o b i l Phase, e J. Chromatogr., 529: 65 (1990). .N. Kayali, S. Rubio Barroso and L.Polo Diez, ~ e t e r m i ~ a of ~on PAHs in Particulate Air by I " , J. Liq. Chromatogr., 17: 3623 (1994). .L.Marina, A ~ o d e Describing l the tion of Alcohols to the Mobile Phasein MLC?J: Chromatogr. A, 719: 15 (1996). .Villanueva Camaiias,J. edina Hernandezand M.C.Garcia Alvarez-Coque, Modeling of ntion Behavior of Solutes in h.17;C with Organic M o d ~ ~ e r s ~ J. C~romatogr.,639: 87 (1993). irkbright and F.G.P. Mullins, ~eparationof ~ i t h i o c a r ~ HPLC usinga Micellar Mobile Phase, Analyst, 109: 493 ( .Wu and J.H.Aiken, ~ i c e l l a HPLC r ~eparation Bilirubin Species with Direct SampleInjection, J. Liq. Chro~atogr., 18: 1219 (1995).
24.
Separation and Phase-~olute Interaction by MLC, J Liq. Chromatogr., 18: 2397 (1995). 25. sch and L.J. Cline-Love, Eflects iency o f ~ CJ. ,Chromatogr., 283: 47 (1984). 26. cott and C.F. Sirnpson, Solute-So~ventInteractions on the Surface of ~eversedPhases.Interactive Characteristicsof some ShortChain Aliphatic Moderators h a v i n ~ rent ~unctionalGroups, raday Symp. Chem.Soc., 15: 69 (1980). 27. . Jandera, W L C of Homo~ogousSeries. A General Method for re diction of ~ e t e n t ~ oJ n ,Chro~atogr.,314: 13 (1984). d J. Ngeh-Ng~ain~i, ~etention ~ehavior of
29. ir, Inves~gationof the ~etention Mechanism in on ionic MLC an Al~lbenzeneHomologous Series, Anal. Chem., 60: 2520 30. 31.
32.
. Li
and S. Fritz, ~ o v eAdditives l for the Separation of Organic by HPLC, JC~romatogr.A, 728: 235 (1996). and E.Pelizzetti, The Useof a Micel~arMobile Phase in Separation of Hydroxyben~ene~erivatives,Anal. Chim.
istribution Coe~cients
33.
cel~e-~oluteAssociation Constants of some ~enzene and S by Micel~arHPLC, Chro~atographia, 28:
35.
~ctanol- at er Coe~cientswith Hydrophobici~for Po~cyclic Aromatic-~vdrocarbonsby MLC, Chromato~raphia, 34: 627 (1992).
LL
36.
.L. Marina, In~uenceofAlcoho1 OrganicMod~ers uponthe Association Constants andRetention Mechanism for Aromatic Compounds in MLC, J Liq. Chromato~r.& Rel. Technol.
.Khaledi, J. S. Landy andJ.L. Lin, ~radient-Elution kiZCqJ Chromatogr., 316: 183 (1984). an, Solute Retention in MLC,JI 38.
37.
39.
orsey, TheMicelle A ~ a ~ t i c a l
40.
41. 42.
43
*
44.
45.
46.
47.
HybridkiZC, Chromatographia,48: 655 (1998). J.S. Landy and J.G. Dorsey, Characterization o f ~ i c e l l a rMobile Phases for W L C , Anal. Chim. Acta, 178: 179 (1985). C. Garcia Alvarez-Coque, J.R. Torres Lapasio and J.J scription of thePartitioning ~ehaviorof Solut Treatment in kiZC with M o d ~ e r sAnal. , Chim. Acta, 324: 163 (1 996). .A. Garcia, S. Vera and M.L. Marina, ~eterminationof MicelleSolute Association Constants of someBenzeneand ~aphthalene ~erivativesby Micellar HPLC with ~ u t a n o land Sodium-Chloride Additives to Mobile Phase, Chromatographia,32: 148 (1991). ASH.Rodgers, Micellar-Mediated S h ~ t sof .G. Khalediand Ionization Constants of Amino Acids andPeptides, Anal. Chim. Acta, 239: 121 (1990). .G. Khaledi, In~uenceof pH on Retention and Consequences of Micellar-~nducedS ~ ~ oft s ‘zationConstants,Anal. Chem., 66: 327 (1994). .Torres Lapasio, J.J. Baeza Baeza and M.C. Garcia Alvarez-Coque, ~escriptionof the ~etentionBehavior in M C as a Function of pH, S~rfactantand Modl~erConcentration,J Chromatogr. A, 769: 155 (1 997). M. Arunyanart and L.J. Cline-Love, In~uence of Micelles on Partitioning Equilibria of Ionizable Species in LC: pH and Ionic Strength Effects, Anal. Chem., 57: 2837 (1985).
0– strong and .U.Stine, valuation and Perturbation of Micelle-So~uteInteractions, J: Am. C h ~ mSoc., . 105: 6220 (1983). erthod, I. Girard and C. Gonnet, Additlve Ejcfeects on Surfactant A~sorptionand Ionic Solute Retentionin MLC, Anal. Chem., 58: 1362 (1986). .P.Tornasella and L.J. Cline-Love, Thermodynamic Properties in MLC based on the Three-Phase ~ ~ u ~ l l bModel, r i u ~Anal. hem., 62: 13 15(1990). 51. Rodriguez Delgado, M.J. Shchez, Gonzalez and F.Garcia elongo, Role of Temperuture in the h ~ i o orf PAHs in MLC. ~hermodyna~ic Aspects, J: C h r o ~ a t o ~Science, r. 33: 647 (1 995).
. The first historical use of micellar phases in chromatography was with the Gel Permeation Chromatography technique (GPC) (see Chapter 3). The low efficiencyproblem was not notedin the early use of micellar phases because GPC is not a technique that produces sharp peaks. However, Arrnstrong observed very early that the TLC micellar spots were broader than nonmicellar ones[1].Broad peaks,i.e., poor efficiency, were obtained when micellar mobile phaseswere used insteadof classical hydro-organic mobile phases in cIassicalLC [2, 31. Dorsey attributed the reduced miceIIar efficiency to a poor wetting of the apolar stationary phase by the aqueous surfactant phase [ 3 ] . He proposed the addition of 3% l-propanolto remediate the problem [3,4]. Yarmchuck et al. thought that a slow mass transfer between the micelles and the stationary phase was responsible for reduced efficiency [5]. Armstrong [6], Berthod [7] and Hinze [8] concluded that poormass transfer within the stationary phase itself was mainly responsible for the observed low efficiency. Two main approaches were proposed to enhance the efficiency in MLC: (i)-the addition of low amounts (3% v h ) of l-propanol [3] to decrease the amount of adsorbed surfactant and to increase the stationary phase wetting, (ii)-to raise the temperature [4, 5) to reduce the liquid viscosity and to increase the chemical reaction rates. In this chapter, thepossible causes ofthe reduced efficiencyin MLC will be surveyed anda rernediation path will be proposed and discussed. The chromato~ra~hic process is rapidly exposed pointing out the thermodynamics (retention time) and the kinetics (peak efficiency). The differences between a classical hydro-organic reversed mobile phase and a micellar phase are recalled. The kinetics of the chrornato~raphicprocess can be modeled using the Knox equation that relates the reduced plate height to the 17
l
reduced mobile phase velocity. Knox plots with classical hydro-organic mobile phases and micellar mobile phases are compared and discussed. ~ u r f a ~ t a n t a d s o ~ont ithe o n s t a t i o n a ~phase and solute exchange between micelles and the bulk are the two main differences between classical and micellar phases. Thesetwo mechanisms are likely responsiblefor the major part of the efficiency loss. The effect of the adhtion of various organic solvents onthe micelle physicochemical structure and surfactant adsorption is exposed.The efficiency enhancementsobtainedwithtem~eraturechanges and/or organic modifier additionsare described.
. The t h e ~ o d ~ ~ofi the c sc~omatographicprocess is responsible for the solute r~tention.The retention volume,VR,is related to the solute affinity constant for the stationary phase, K, by
in whichV, and V, refer to the mobile and stationary phase volume inside the column. The retention factor, k, is expressed by
where Q, (=V, No), is the phase volume ratio. The peak retention volume and factor are due to chemical equilib~~a which are temperature dependent. The classical Gibbs free energy equati&n can be applied:
Combining eqs. 6.2 and 6.3 yields: Ln k = -AHo/RT + ASo/R + Ln 4
C64
The plots of Ln k versus UT,the Van't Hoff plots, will be linear if ASo are constant over the temperature range studied. The slope of the Van't Hoff plot gives the enthalpy change associated with the solute transfer from the mobile to the stationary phase. The intercept is related to the entropy change. Nonlinearity of these plots would indicatechanges in the chromato~aphicmechanism with temperature.
The kinetics of the chromatographic process is linked to efficiency. The number of theoretical plates, N,that measures column efficiencies, is best estimated using the Foley-Dorsey equation when a departure from the Gaussian symmetrical peak shape is noted [9]:
BM + 1.25 in which W,.1, is the peak width at 10% peak height, B/A is the peak asymmetry factor measured by the ratio of the back (B) to front (A)half portion of the W,.,, peak width referring to its t, retention time. This equation takes into account the peak asymmetry. In MLC, the efficiency decreases are always associated with peak tailings, i.e., asymmetry. All plate counts given were obtained using the Foley-Dorsey equation.
According to Snyder and Kirkland [IO], the con~ibutionsto band broadening in a column can be represented in H, the column plate height, by the equation:
in whichthe C values are constant plate height coefficients related to eddy diffusion (e), mobile phase mass transfer (m), longitudinal diffusion (d), sta~nantmobile phase mass transfer (sm) and s t a t i o n a ~phase mass transfer ( S ) , D, is the thickness of the stationary phase layer, D, the solute iff fusion coef~cientin the mobile phase, D, the solute diffusion coefficient in the stationary phase layer and U is the mobile phase velocity. TheKnox equation [111 is used in LC to d e t e ~ i n ethe contributions to the final solute band width. It can be expressed as:
where A', B' and C' are the constants of the Knox equation, h is the reduced plate height calculated as h=H/d,, where H is the column plate height (H=LIN, L beingthe column lengthand N the number of theoretical plates), d, is the stationary phase particle diameter, U is the reduced mobile phase velocity, i.e., U = ud,/Dm,with the average mobile phase velocity, U, The A', € 3 ' and C' terms are related to flow anisotropy, molecular lon~itu~inal diffusion andmass transfer processes, respectively. The theoretical support for the Knox equation was derived by Horvath [12). The A' term cannot be expressed simply. The theoretical treatment links A' to structural parameters of the column packing,porosity, pore volume, pore diameter and tortuosity [121. A' is related to the flow pattern and the general band spreading due to "eddy" diffusion [13). The B' term (longitu~ina~ molecular diffusion) was written as [131:
where ym and y, are obstruction factors for diffusion through granular and/or porous material. The C’ term of the Knox equation represents the mass transfer contributionto solute band ~roadening.It was written as [131
where the y terms are obstruction factors with the subscripts defined for eq. 6.6, 9 is the stagnant mobile phase fraction and q is a geometrical factor depending on porosity [12-141. and broadening and/or unusual retention are also known to occur when specific interactions take place between particular solutes and the stationary phase. The most common cause of such band broadening is a stationary phase overload. It can bealso an exclusion phenomenon due to the small size of the stationary phase pores and the large size of the solute. Acid-base interactions between the surface silanols and basic compounds, especially the amine compounds:,dramatically broaden the corres~onding peaks. Peak deformations, either frontings or tailings, are observed.
I13.~ x t e r ~n ~ ~~
roade n den in^
The plate number, i.e., the chromato~raphicefficiency, is a parameter difficult to estimate correctly [151. The use of methods assuming Gaussian peaks may produce grossly overestimated peals efficiency. The a s y ~ m e t ~ based method [g] (eq. 6.5) was designated as the manual method giving plate numbers closest to the exact plate numbers obtained with the moment method [15, 161. The second central reducedmoment of a peak too often overlooked, that corresponds to its variance [171. It is known, but is a combination of the column variance plus the overall peakvariance, a2:, all the other extra-column variances
(6.10)
17
where the subscripts refer to the column, connecting tubing, injector, detector and other variance contributions (e.g., electronic time constants, frits, unions).
.l Efficiencyexperimentalmeasurement. 1- observedefficiency for a 10,000 plate column, the extra-column bandbroadening is heavy in decreasing the column efficiency band broadening and efficiency for low k solutes. 2-observed efficiency with extra-column enhancement combined withk reduction, amaximum N value seems toexist, 3-true column efficiency when enhancements are used in MLC. Reprinted from [25] with permission of Elsevier.
It is possible and desirable to reduce all extra-column variances. However, external band broadening cannot be fully eliminated. The ~ e a s u r e d and a variance is always the sum of the actual column variance, 02c01,
constant term. If the column efficiency is constant.fora variety of solutes, increases linearly with time the square root of the column variance, acOl, or elution volume. The other variance contributions are rather constant with time. As the retention time increases, the external band broadening contribution to the global variance becomes less significant. The measured efficiencies of highly retained peaks appear higher than the efficiencies of lowk peaks. Figure 6.1 (curve 1) shows that a plateau of constant efficiency, the true column efficiency, is rapidly reached. for improvi~gthe efficiency in LC reduce the retention factor at the same time that they decrease the parameter. Then, the e x ~ e r i ~ e ~ t a ~ observed efficiency may show a maximum value [121 for a given ls value as illustrated by Figure 6.1 (curve 2). Itmeans that the ef~ciency enhancementcanbe annihilated by the extra-column effects. This is specially true when nonoptimized chromatographic systems are used.
c ~ ~ ~ ~ ,
.
The three phase model was exposed in the previous chapter. besides the solute exchange between the stationary and the mo the micelles introduce a secondary chemical equilibrium. Each equilibrium, micelle-aqueous phase, stationary phase-aqueous phase and stationary phase-micelles, has its thermodynamic constant, PWM, PSM,respectively, linked together and to the micellar concentrati the phase ratio, 4, and the surfactant molar volume, v, by the ~rmstrong equation [181.
(6.1 1)
The PSMconstant, not contained in eq. 6.l 1, is the ratio of the two other constants [181:
(6.12)
A.1~0,each e~uilibriumis not reached instantane~uslyan kinetic constant. All three kinetics constants intervene in the global observed efficiency and are studied individ~ally therea~er.
In the early studies, the micellar reduced efficiency was attri [S]. di~inishedrate constant for solute exit from the ~icellar a~gregates If the solute stays inside the micelles that move withthe mobile phase for a long t i m ~ it, interacts with less stationary phase. It “sees” a s h o ~ e r column. Assuming the solute entrance in the micelle is diffusion controlled and e~uivalentfor all solutes, the exit rate constant is inversely ~ r o p o ~ i o nto a lthe PWM constant. l’he solutes with the highest affinit~for the micellar phase shauld be the least r~tained.Actually, they are the most retained solutes which shows that the solute affinity for the micelles (Pm) are strongly correlated to the affinity for the surfactant covered stationary These combined effects with the h ~ ~ r o p h osolutes ~ i c show that they are the ones with the poorer efficiency [5]. The reduced solute exit rate from the micelle produces a decreased solute diffusion coeffi~iellt in the mobile phase, D,. It was shown that the m iff fusion ~oefficientwas ~ecreasedby ,,P constant [6]. l’he overall solute micelle iIlclusion related to the diffusion coefficient, D,, in a micellar solution depends on the micelle diffusion coefficient, DM,and the solute iff fusion coefficie~t i n the aqueous phase, D,,
D m = D, +”-----D , l+
l
1 -l”
(6.13)
is related to the solute a f ~ n i t yc,onstant, PwM,~ e a s u r ~ byd
The product [Mlv is the volume percent of micellar phase and l-[M]v is the volume percent of the aqueous phase 161.
/
Decreases of the benzene diffusion coefficient due to its inclusion in micelles of different surfactants (Data from [6]). Reprinted from 1251with permission of Elsevier.
Figure 6.2 illustrates the decrease of the benzene diffusion coefficient in various ~ i c e l l a rsolutions. Table 6. l lists the diffusio~ coefficients of various solutes in different micellar media. Depending on the solute affinity for the micellar phase, the D, diffusion coefficient can be decreased by a factor varying from 2 to 10 (Figure 6.2). Such a D, decrease may highly increase the B‘and C’ terms of the Knox equation as
shown by eqs. 6.8 and 6.9, respectively. The corresponding reduced velocity, U, is also increased by the D, decrease which should reduce the observed negative effect at low flow rates and magnifL it at high flow rates. An i m ~ o ~ a surfactant nt concentration does not seem to be favorable for ef~cie~c~. .l
DiffbsionCoefficients in VariousMicellar Phases at 24°C
Solute
P,
P
D, (x l o6 cm%)
Brij 22, 5% w/v; 0.08 M Benzyl alcohol Benzaldehyde Benzene Micelle
16.5 21.6 72.2
1600 2100 7000
5. l 4.7 2.9 0.87
400 640 1800
6.3 5.5 4.4 0.94
4800 15,000 43,000 120,000
6.8 3.4 1.7 0.94 0.57
-
Brij 35,5% w/v; 0.042 M Benzyl alcohol Benzaldehyde Benzene Micelle
10 16 45
-
~~~,1.4% w/v; 0.05 M Benzene Toluene ~thylbenzene Prop~lbenzene Micelle
77.4 242 694 1940 "
P is the partition coefficient referring to the micellar phase (=P,,,,xN,,,). Brij 22 = C12E10, mw = 610, v = 0.9542mL/g or 0.582L/mol, Naggr, = 97, cmc= 0.09 mM. Brij 35 = C12E23, mw = 1200, v= 1.064 Limo1 or 0.867 mL/g,NBRPr,= 40, cmc = 0.1 mM. SDS,mw = 288, v= 0.246 Umol or 0.854 mL/g, N8ar, = 62, cmc = 8.2 mM. Data from [7],
III 2. cellar Phase- ~ t a t i o ~Phase ar~ ~ x c h a ~ ~ e s Chapter 4 extensively described the surfactant adsorption occurring when micellar phases are used. This adsorption takes place with any surfactant,
LE
ionic (see Figures 4.2 and 4.4) as well as nonionic (Figure 4.1). surfactant adsorbed layer changes the solute-stationary phase and micellestationary phase interaction and kinetics. The amount of adsorbed material depends on the surfactant and the stationary phase used. It is about 4.5 pmol/m2(-1.3 mg/m2 or -140 mg/(g of C 18 phase) or 14%) for the SDS surfactant on a C18 phase. Similar measurements gave 4.3 pmol/m2 (-2 mg/m2or -.2 10 mg/(g of C 18 phase) or 2 1%) for the CTAB surfactant on the same C 18 phase (see Chapter 4). These amounts of adsorbed SDS or CTAB are close to two surfactant molecules per bonded moiety for the C 18 phases. Corresponding results were obtained for nonionic surfactants [8, 191. The adsorbed surfactant molecules fill up part of the silica pore volume and in doing so, reduce the stationary phase surface area [19,201. The surfactant adsorbed layer increases the thickness of the statio~ary phase organic layer, 0, Assuming a regular layer, the previously indicated adsorbed mass of SDS on the C 18 phasecorresponds to 1.06 mm3/m2or an average thickness of about 1.06 nm. The same calculation for the CTAB layer on the C 18 phase gives a thickness of about 1.6 nm. These estimated of the surfactantmolecule lengths. The breadths roughly correspond to --M @,parameterintervenes quadratically in the last term of eq. 6.6 that means the surfactant layer may dramatically reduce the chromato~raphic efficiency.
As shown by eqs. 6.7, 6.8 and 6.9, the A',B' and C' terms of the Knox equation are not directly related to the kinetic constants of the three equilibria occurring with micellar mobile phases. The slow exchange of the solute between the micellar and aqueous phases increases the three terns of the h o x equation. It increases the flow anisotropy (A'term). It reduces the molecular diffusion. increasing the B' term (eq. 6.8) and the C' term (eq. 6.9). The reduction of the stationary phase-micellar phase soluteexchanges due to the adsorbed layer acts mainly on the C' term.
Table 6.2 lists the A',B' and C' values experimentally obtained by several workers [19-22]. To be meaningful, the Knox plots should be obtained on the same column, with the same solute and different mobile
phases. In column #l, with the benzene solute, the efficiency loss is due to A'increases (+300%) and B' increases (+700%) compared to the values Parameters of the Knox Equation
Optimal Flow
Column Phase Mobile Solute
A'
k
mUmin
Benzene
25°C 45°C
MeOH-water 70/30 v/v SDS 0.05M (1.4% w/v) ACN-water 30/70 1.2 Brij 35 0.042M(5% w/v) 0.2 SDSO.O5M(l.4%w/v) SDS 0.05M(1.4%w/v) 0.7
~t~y~benzene MeOH-water 70130 v/v SDS 0.05M (1.4% w/v)
1 1
2 2,
1.1 3.1
CO.1
0.2 20 0.2, 16.4 0.026 6 0.03 5.6 0.16 11.4 0.05
1.1
3.2 0.65 1.0 3.4 2.7
15.6 17.3 26. l 17.8 16.3
0.2 0.3
1.1 3.1
3.8 23
3.5 45
0.1
5.4
0.26 5.8
0.2 0.3
3 3 1 1
Cu~~arin 0.01M SDS (0.3% w/v) 2 O.OlM+2%C30H SDS 0.018 8.5 4.0 0.2 2 SDS 0.01M+1% G50H 0.007 8.8 3.0 0.32
0.15 0.5
-
-
Column #l: 10 cm, 4.6 mm id., Radial-Pack ODs, 10 pm (Waters), datafrom [21], Column #2: 10 cm, 4.6 mm id., Apex ODs, 5 pm (Jones), data from [21]. Column #3: 10 cm, 4.6mm i.d., C14 bonded phase, 5pm, 3.5 pmol/m2,data from [22]. Room temperature, unless otherwise indicated.
obtained with the methanol-water mobile phase. The C' term is little ith the ethylbenzene solute, the A' and B' values are similar to the benzene values but the C'term isalso increased by 350%. In column #2, the efficiency reduction compared to the acetonitrile (ACN)-water phase is due to moderate increases of both the A' (+54%)and C'(+16%) terms; the decrease of the B' term is not significant considering the raw data set E2l]. In column #3, the hydro-organic reference values were not given. Assuming the column #l values can be used, the efficiency loss is ain due to the A'and B' term increases (+300% and +80%, respectively). hen the temperature was raised from 25°C to 45"C,this decreased the A' and C'terms by -21% and -70%, respectively. The effect of alcohols on
nox parameters is seen by a decrease of the A' terms by -26% and
-45% and a dramatic decrease of the C' parameters by -93% and -97%
is shown when only 2% v/v propanol and 1% v/v pentanol, respectively, were added [22]. The use of the b o x plots to study the causes of micellar reduced efficiencies leads to the following conclusions. The micellar phase flow anisotropy seems to be much higher than the flow anisotropy obtained with a hydro-organic phase of Comparable viscosity (increased A' term). This is only partly due to the micellar viscosity. The main reason of such differences in flow patterns is the partial clogging of the stationary pores by adsorbed surfactant molecules [19, 221. A temperature raise decreases the mobile phase viscosity and the amount of adsorbed surfactant [22]. Both effects decrease the flow anisotropy and the A' term. It will be exposed thereafter that alcohol additions to a micellar phase dramatically reduce the amount of adsorbed surfactant. ' term is linked to solute diffusion coefficient which is nclusion (eq. 6.8 and Figure 6.2). The increase of the ) , change andalso to thek increase (Table in micellar phases is due to the I 6.2). Considering the C' term equation (eq. 6.9), it should be expected that micellar phases decrease C' because k values are higher and coefficients are lower. This is not observed experimentally. It means t the surfactant adsorption causes dramatic decreases of the D, parameter. The solute diffusion coefficient in the surfactant covered stationary phase is very dif~cult[6,7,20-2 l]. It slows down the solute mass-transfer. This effect becomes dominant for lipophilic solutes that have a high affinity for the stationary phase (ethylbenzene, Table 6.2). The peaks correspo to lipophilic solutes becomevery broad with micellar mobile phases. a temperature raise and alcohol addition decrease the amount ofads surfactant [19, 231.Both actions reduce the C' term and improve the observed efficiency.
Clearly, the Knox plot study points out that surfactant adsorption on the stationary phase is responsible for the bulk of efficiency loss observed using micellar phases. A slow solute exchange between the micelle apolar core and the aqueous phase is another possible explanation for efficiency loss.
. e serious waste of efficiency in MLC was immediately noticed. It was a major drawback hindering the development the of .technique. To be useful for the practicing chromatographer, the efficiency observed in least approach that of conventionalHPLC. As early as 1983, suspected a slow mass transfer coming from a poor wetting ofstationary the phase by the aqueous micellarphase. They proposedthe addition of3% vlv 1 -propanol to the micellar phase and heating the chromatographic system to 40°C [3]. The dramaticenhancement of the MLC efficiencytriggered numerous studiesof the effect of organic additives temperature and in
a) Polar Solutes withan Amino Group kers were recently separated byMLC on a classical C 18 phase with micellar mobile phase [24]. These experimental resultsare typical be used to illustrate the efficiency problem in MLC. Most pblockers are amino-alcohol derivatives. They are difficult to quantify by classical LC because their aminogroup is positively charged. Interactions with the negative residual silanols of the stationary phase produce some peak tailing. The problem is easily solved by the addition of a small amount of triethylamine to the hydro-organic mobile phasethat quenches the silanols. The use of decjlcated columns for basic compounds with very few residual MLC can also be usedto separate silanol groups is another elegant solution. and quantify these compounds. Table 6.3 lists the experimental chromatographic parameters obtainedfor4p-blockers [24]. alcoholswerechecked as possibleefficiencyimprovers.Figure 6.3 illustrates these results.
As described extensively in Chapters7 and 8, any alcohol additionto a micellar phase increases the elution strength. The full lines of Fig. 6.3 correspond to the peak efficiency; the dotted lines showthe corresponding retention factors. A significant efficiency improvement (from 5 to 15 times increase) is obtained with the addition of 15% vlv propanol to the SDS
Alcohol content (% v&)
tire 6.3 Combined changes produced by alcohol addition on efficiency (full lines) and retention factor (dotted lines) oftwo P-blockers: Propranolol (top) and Carteolol (bottom) (Data from [24]).
Retention Factor and Associated Efficiencyofthe P-blockers’ Peaks Eluted with Alcohol Containing Micellar Phases
~ e b ~ t o l o l~ ~ ~ e o l o leli~rolol k N k N k N
~ o b i l Pe ~ a ~ ek SDS 0.1 M 100 47 +p~opanol5% 940 16 1200 10 +prop~nolIO% +propanol 15% 8 +butanol 5%300 10 +pentanoE 2% 7.5 SDS 0.I5 M +propanol 5% 10.5 +propanol 15% 5.5
rmolol N
430 39 11 1550 7 1550 5 6.5 5 300
70 96 196 190 1200 46 1260510 24 105014 1500 29 10.5 1150 2100 19 1900 13 300 550 28 130 9.5 240 31 90 500
700 1400
800 400 16 1100 7 1450
7.5 3.5
1100 29 1850 12
Data &om [24].
utanol and pentanol were also used addition of 5% vlv butanol to the 0. fficiency upto four times (celiproplol), it can decrease it 1). A 2% vlv pentanol addition has a similar eEect on n and efficiency than the addition of 5% vlv butanol (Figure 6.3 e huge decrease in the P-blocker ret2ntion prod~cedby the 2% vlv pentanol (and 5% vlv butanol) addition is due to a [24]. The alcohol seemsto reduce significantlythe mount of a S adsorbed ions decrease, the s t a t i o n a ~phase. positive P-blockers are less retained. *
b) Apolar ~ o l u ~ e ~
en~ene,toluene, 2-et~ylant~ra~uinone (EA )and ace top hen on^ were used as test solutes of low or medium polarity [8, 20, 231. The linear alc from methanol to hexanol were studied as organic additives with S [25]. Table 6.4 lists the Pw and o ince the phase ratio was not given) soluteswith diEerent micellar phases. Clearly,the alcohol additions produce a p~rallelvariation of the two constants P. It means that these additions
change the solute interactions with the micelles and the stationary phase in a co~parablem a ~ e [24, r 261. Effect of Alcohols onthe Solute Mlnity for the Micellar and StationaryPhases
rg 35 + 15% ethanol Acetophenone
5rij 3
+ 15% ethanol 14
Toluene
CTAB
CTAB + 5% methanol
39
40
55 56 31 22 16 29
49 23 11 8 6 20.4 15.4
240 200 128 114 390 240
140 124 86 55 190 125
The OPws value are only suggestive sincedifferent Cl8 columns wereused. Data from [25].
Figure 6.4 shows the efficiency e~ancementfor benzene and EA obtained with 5% v/v alcohol additions to a 0.285 M S The stationary phase isa classical C 18 phase(4= 5p r almost proportional to the a1 log Pm value of the alcohol, a par solution, the benzene and carbon number [27& h pure efficiencies were respectively1400 and 49 plates (HE "he m a x i ~ u mefficiencies, 3400 and l 100 plates (5.5 d, and18 dp),
respectively, were obtained with 5% vlv pentanol. The benzene efficiency a wasmore than doubledandtheipophilicsolutewasmultipliedby factor 22(Figure 6.4). "he EA ention factor wasreducedby82%, from 45 in pure SDS phase to 8vlvpentanol[25]. The maximum ency increasefor benzene was obtained for butanol and pentanol. The efficiency seems identical with the addition ofpentanol and hexanol ( ~ i ~ u 6.4), r e but the hexanol efficiency value was obtained with a 0.5 rnLlmin flow rate (instead of I mLlmin) due to the high viscosity of the -1hexanol phase. With this lower flow rate, hexanol producesan inferior efficiency comparedto the 5% vlv pentanol addition. Pentanol seemsto be the best alcoholfor MLC efficiency enhancement withSDS solutions.
6.4 Efficiency enhancement obtained with 5% v/v alcohol addition to a 0.285 M
SDS phase. Column 10 cm, 4.6 mm id., 5 pm C18 Astec (Whippany,NJ),flow rate 1 mL/min
(0.5 mL/min for hexanol),23°C. Reference efficiency: benzene 1400 plates (left Y-axis), EAQ 49 plates (right Y-axis) in pure SDS micellar phase. Reprinted from E251 with permission of Elsevier.
The main effect of alcohol addition to a micellar phase is the increase of elution strength as reviewed in Chapters 7 and 8 and modeled by the ~ I softwareC(see Appendix ~ Iand ~ attached ~CD-ROM).~ The decrease of the Pvalues observed forthe apolar compounds (Table 6.4) can be generalized to any solute. These effects are related to thermodynamics. The kinetics of the chromatographicprocess is also affected by the addition of alcohols to the micellar phases. a) Efect o ~ A ~ c o h oon~ ~s i c e ~o~utions ~ ~ a ~
As briefly exposed in Chapter 2, the addition of mediumchain alcohols to micellar solutions change the micelle structure. It was demonstrated that there are two relaxation times in the kinetics of micelle formationdestruction: 1) A fast relaxation time, in the low nanosecond range, co~espondingto the exchange rate of a surfactant molecule with the micelle. 2) A slower relaxation time, in the microsecond range, was linked to the average lifetime of a micelle [28]. Methanol and ethanol at low e on cent rations do not partition signi~cantlywith the SDS micelles. Both alcohols do not affect notably the two micelle relaxation times. The C3-C7 alcohols were found to increase the kinetics of the exchange rate of an SDS surfactant or alco~olmolecule and to decrease the two relaxation times [29]. The formati0~-destructionof the SDS micelles is highly accelerated by these alcohols. The alcohol molecules incorporate in the micelles with their polar hydroxyl groups in the Stern layer and their alkyl chain in the micelle cores. This moves the ionic sulfate groups further away [30]. The charge density of the micellar interface decreases which mayfacilitate the inte~icellarmigration of solutes. Similar results were obtained with the cationic CTAB surfactant [3 l]. Low concentrations of C3-C6 alcohols increased the intermicellar migration of a dye solute by several orders of magnitude [321. For MLC, it is n e c e s s a ~to summarize the vast amount of work done on the effects of alcohols on the ionic micelle structure distin~uishingWO points: (i) low amounts(5 5% vlv) of methanol andlor ethanol have ~inimum effect, (ii) the ratio of the alkyl chain length of the alcohol over that ofthe surfactant, CnOH/Cnsurf, can be used as a reference parameter. If the ratio Cn~HICnsurf is lower than 113, e.g., propanol and butanolwith SDS micelles,
or propanol, butanol and pentanol with CT micelles, the alcohol increases the kinetics of both thefo~a~ion-destruction of the micelle andthe exchange is hgher than 113, the concentration of rate molecule-micelle. If Cn~HICns~f the alcoholshouldbe compared to the surfactant ~oncentra~ion.Low amounts of alcohol increase the kinetics of the micellar exchanges which is LCefficiency. § i ~ ~ c aalcohol n t ~ n ~ n t r a t i o dsorganize ns the micelle structure which is unfavorable for MLC efficiency. At a constant surfactant concentration^ the maximum beneficial alcohol concentration decreases as the C ~ H I C nratio s ~ increases. ~ Alcohol Effect on Surfactant Adsorption
obile Dhase
+ 3% vlv
+ 2% vlv
amount Adsorbed
propanol pentanol
+ 5% vlv methanol + 3% vlv propanol + 2% vlv pentanol
4.6 4.4 3 .O 3.1 4.5 4.4 3.2 2.2
Column ODs Hypersil, C l 8 , 5 pm. Data from [7].
b) Efect of A ~ c u ~ oon l s ~tationaryPhases
Alcohols andsur~actantmolecules competefor ~ o r p t i o non the stationary phase. Alcohol addition reduces the mount of adsorbed surfactant [19,20, able 6.5 lists the amounts of adsorbed sudactant when micellar mobile phases withorwithoutalcoholwasused[7]. The desorption is dependent on the alcohol chain length [7, 191. deso~tionis linked to a of reduction that should decrease the 6.6). The kinetics of the surfactant adsorption-desorption process is enhanced by alcohol addition[S, 7,2 l].§pectroscopic studies have shown a monomeric C 18 silica bonded layer that short c h n alcohols (Cl-C3) without changing its organization. ositely, long chainalcohols (C7-C 10)
inte~enetratewith the l 8 chainsC331 These alcohols decrease tension and viscosity of the C 18 bondedlayer in minute~ o u n t [s3 3 ] . I)
c) ~~c~ Alcohol Is Best and How ~
~ to Add? c h
The 1983 work ofthe orsey research group established the use of 3% v/v 1 -propanolin micellar phasesto reduce the MLC effkiency problem [3]. is nowc o n f i ~ e that d the adhtion of alcoholto micellar phases: (i) increases the rate of the solute mass-transfer betweenthe micelles and the aqueous phase by increasing the solute micelle exit rate constant, (ii) increases the solute mass transfer kinetics betweenthe stationary phase and the aqueous phase by decreasing the stationary phase viscosity and the amount of adsorbed surfactant. The problem of alcohol additionto micellar phases is that kinetics e~ancementscannot be dissociated from t h e ~ o d ~ ~ i c s changes. The efficienciesincreaseand the retentiontimes decrease. A hybrid alcohol-micelle mobile phase has necessarily a higher solvent strength than a purely aqueousphase [34]. It was shownthat alcohols were changing the micelles andthe stationary phase in a comparable manner [26] as noted parallel variations in Table 6.4. As shown by Figure6.4 the maximum efficiencyincrease for benzene was obtained for butanol and pentanol that have C n ~ ~ / C n ~ ~ f of p a113 r~eter and 0.42, respectively. 5% v/vhexanolseems to be too muchwitha C%H/Cn,f parameter of 1/12, the benzene efficiencydecreases; the reason may be that the es are disorganized.Pentanolseems to be the alcohol wit best ions.
A limiting parameter for the choice of alcohol is viscosity. higher viscosity hybrid micellarphases obtained with alcohol addition ahave than the corresponding micellar phase without alcohol. ~iscosityincreases atically with the alcohol chain length. The viscosit of the 0.285 M phase was 1.65 CPproducing a 95 kg/cm2 (9.5 M pressure at 1 mL/min with a 10 cm column. The visco S phase was 2.32 CP producing a 127 kg/ ressure p5]. The viscosity of the 5% hex precluded its use at l mL/min. 0.5 mL/min flow rate was likely produced overestimatedesciencies for the hexanol
1
Efficiency enhancement versus the pentanol to SDS concentration ratio. Same experimental conditions and reference efficiencies as in Figure6.4. Reprinted from E251with permission of Elsevier.
S 5% v/v the right amount for alcohol addition? Considering that large ~ o ~of talcohols s with a C n ~ ~ / ratio C n ~higher ~ than 1/3 can disorganize the micelle structure [35], the alcohol to surfactant ratio is igure 6.5 shows the efficiency e~ancementobtained for benzene plotted versus the pentanol over SDS con cent ratio^ ratio. Two sets of experiments were done: (i) constant SDS concentration (0,285 and increasing pentanol additions, and (ii)4% v/v pentanol and increasmg S concentrations 11251. The efficiencies obtainedare clearly linkedto the p e n ~ o l / ~ratio. D ~ The efficiency e~ancementreachesa plateau for conce~trationratios higher than 6. Similar effic e~ancementswere (C18) column and obtainedwith the cationic CTAC surfactant, an CN) as the organic additive [20]. A plateau at a three fold
efficiency enhancement was obtained for ACNICTAC concentration ratio higher than 12. Although these results were obtained for only two ionic surfactants and Cl8 columns, it seems possible to extend them to other micellar mobile phases and bonded silica stationary phases. The alcohol to surfacta~tratio acts on both solute exchange rates: micelle-aqueous phase exchanges and stationary phase-mobile phase exchanges. It is directly related to the number of alcohol molecules per micelle. Also, this ratio dictates the total amount of surfactant surface coverage and fluidity of the composite organic layer of the stationary phase [7,25]. In conclusion, the best amount of alcohol to add depends on the surfactant concentration used. It means again that efficiency enhancements cannot be dissociated from micellar solvent strength.
IK 3, ~ t ~~ eo urt efsor ~ f ~ c i e ~nhancements n~y Alcohol addition is the most widely used solution to improve efficiencyin MLC. Knowing that the inferior efficiency is due to a slow solute exchange rate between micelles andthe bulk aqueous solution and/or the surfacta~t covered stationary phase, two other routes are possible to enhance these exchanges: (i) to raise the temperature and(ii) to decrease the flow rate.
a) ~
~ of ~ e ~ c~ ~ ~ e ~ a ~ u ~ e
Figure 6.6 shows the effect of temperature on the experimental efficiency of anthracene [5] and acetophenone [22] eluted with pure micellar solutions. The efficiency of acetophenone on Ca18 column was multiplied by 12 when 25°C to 73°C [22].Athreefold the tempera~rewas raisedfrom enhancem~ntwas obtained for anthracene 15). The retention factors were decreased by temperature. The magnitude of the k decrease is far less i m ~ o ~ a than n t what was observed withalcoholadditions[34]. The k decrease with temperature are predicted by the Van? Hoff equation (eq, 6.4). The equilibrium enthalpies and entropies of the retention process were measured in MLC for various surfactants [36]. These studies showed that the micellar retention process is essentially governed entropic by effects [36]. Enthalpy-entropy compensation studies are useful in the investigation of separation mechanisms. They were used in the study of the retention
1
25 403530
45 5550
T e m ~ r ~ t u r(QC) e
€30
65
70
75
6.6 Combined changes producedby temperature raiseson efficiency (full lines) and [53 retention factor (dottedlines) of Anthracene (top) and Acetophenone (bottom) (Data from and [22]).
mechanisms by C 18 phases [37] andthe study of the mechanism of chiral separatioris with cyclodextrin bonded phases in GC [38]. The enthalpy change, AH,, of the Pm values of 6 solutes (the more lipophilic one was naphthalene) were determined in pure SDS micellar phase and 3% and 10% vlv propanol modified SRS phases [39]. The plot of the Ln k value of each solute versus the co~espondingAH,, value should produce straight lines with similar slopes if the separation mechanisms are similar [37-391. Such studies showed that the alcohol addition altered the micellar retention mechanism. Propanol altered the stationary hase as well as the micellar phase [39]. The solubility of silica in hydro-organic phases is enhanced by the rise in temperature. The rate ofthe silica solubilization depends on the pH and temperature of the mobile phase and also on the bonding quality of the s t a t i o n a ~phase [40]. Silica solubility may become a serious problem when working at tem~eraturehigher than 50°C [5,22,39]. Column life is shortened. b) Flow Rate
Considering the Knox plots, an obvious way to improve efficiency is to work at the optimum flow rate. Table 6.2 lists the o p t i ~ u mflow rates for different mobile phases. The strength of the pure micellar phases is much lower than that of hydro-organic phases [34]. 0.3 mL/min or lower flow rates will produce rebutting retention times. Table 6.2 shows that both alcohol addition and, especially, a rise in tempera~reincrease the optimum flow rates. It was shown that both these remediation methods also increase the solvent strength, decreasing the retention times. If time is not a priority, it should never be forgotten that a decrease of the flow rate can easily further enhance the micellar efficiency and consequently the chromatographic resolution.
c) Classical E f ~ c i e n c ~ E n ~ a n c e ~ e n t s Column overload and silanol interaction may broadenthe chromatographic bands with or without micelles in the mobile phase. In case of a peak deformation due to a stationary phase overload, the decrease of the injected
ount will enhance the peak shape. In the case of peak tailing due to silanol-amine interactions, it was shown that the peak shapes greatly improved whenthe mobile phase pH was decreased to below 5.5 [25]. The observed efficiency increase was attributed to protonation of the silanol groups. T ~ e ~ y l a m i is n eclassically added to reversed-phase mobilephases to bind the silanol groups. This reduces the tailing of basic com ounds which is not due to the micellar phase. The peak of propranolol in a ( n o ~ o su~actant) ~c mobile phase was greatly improved and its a s ~ e t r y by a small addition of triethylamine to the micellar mobilephase gheraddition of t r i e t h y l ~ n eproduced a decrease of the propranolol retentionfactor with a slight fiurther amelioration of peak shape and efficiency. This effect is probably due to an organic modifier effect similar to the,effect describedfor alcohols.
To s u ~ a r i z ethe , efficiency loss observedin MLC using purely aqueous micellar mobile phases is due to three causes: (i) a slow solutetransfer from the aqueous to the micellar phase, (ii) a slowtransfer from the stationary phase to the aqueous phase and (iii) a change of the flow pattern in the column, a change due to surfactant adsorption that changes the porosity and surface of the stationary phase. efficiency can be enhanced: (i) by reducing the mobile phase flow rate to work closerto the optimum of the Knox plot, (ii)by increas~g the temperature whch decreases the viscosities, increases the rate constants and decreasesthe amount of adsorbed surfactant, (iii) especiallyby adding an organic modifier such as an alcohol whose alkyl chain has a lengthsuch be as the ratio C n ~ ~ / close C n to ~ 113, ~ ~ The amount of added alcohol should increased if the surfactant concentration is increased to keep constant the alcohol to surfactant ratio. inetics and t h e ~ o d ~ a m i care s intimatelylinked. If flow rate reduction is excluded, all the other efficiency reme~iationmethods produce thinner peaks elutingearlier. Then, the resolutionfactor may or may not be improved.
~
1.
D.W. Armstrongand R.Q. Terrill, Thin Layer Chromatographic Separation of Pes~cides,Decachlorobiphenyl and~ucleosideswith Micellar Solutions, Anal. Chem. 51:2 160 (1979).
2.
D.W. Armstrong, W. Hinze, K.H. Bui and H.N. Singh, ~nhanced ~luoresce~ce andRoom T~mperatureLiquid ~hosphorescence Detection in ~seudophaseLC, Anal. Lett., 14: 1659 (198 1).
3.
J.G. Dorsey, M.T. DeEchegaray and J.S. Landy, ~ ~ c i e n c y Enhancements in MLC,Anal. Chem., 55: 924 (2983).
4.
orsey, MLC, Adv. Chromatogr., 27: 167 (2987).
5.
P. Yarmchuk, R. Weinberger, R.T. Hirschand L.J. ClineLove, Eflects of ~estrictedMass Transfer on the E ~ c i e n c yof MLC, J: Chromatogr., 283: 47 (1984).
6.
D.W. Armstrong, T.J. Ward and A. Berthod, Micellar Eflects on Molecular D~fusion: Theoretical and Chromatographic Considerations,Anal. Chem., 58: 579 (1986).
7.
A. Berthed and A. Roussel, The Roleof the Stationary Phase i n ~ C , J: Chromatogr., 449: 349 (1988).
8.
M.F. Borgerding andW.L. Hinze, Characterizationand valuation of the Use of on ionic Surfactant Micellar Mobile Phases in W HPLC, Anal. Chem., 57: 2183 (1985).
9.
J.P. Foley and J.G. orsey, Equations for Calculation of Chromatographic Figures of Merit for Ideal and Skewed Peaks, Anal. Chem., 55: 730 (1983).
10. L.R. Snyderand J.J. Kirkland, Introduction to Modern L i ~ u i d ~hromatogr~phy, Wiley-~ntersci~nce, New York (1 979). 11. J.H. b o x : and J.N. Done, The Performance of Packrngs in High Speed LC, II-ZIPM the Eflectof Particle Size, J: Chromatogr. Sei., 10: 606 (1972). 12. C. Horvath andH. J. Lin, Band Spreadingin LC, J Chromatogr.149: 43 (1978).
13. J.C. Giddings, ~ynamicsof Chromatograp York (1965). 14. J.C. Giddings, UnrJied SeparationScience, York (1991).
15. A. Berthod, On the Use of the Knox quat ti on, 11-The ~ ~ c i e n c y Measurement Problem, J:Liy. Chromatogr., 12: 1187 (1989). 16.
idlingrneyer and F.Y 56: 1583A (1984).
rren, Plate Countin LC,Anal. Chem., Schetter and J.C.
17. 18.
.W. A ~ § t r o n gand F. Norne, Partitioning Behavior of Solutes Eluted with Micellar Mobile Phases in LC, Anal. Chem., 53: 1662 (1981).
19.
orgerding, W.L. lin, Investiga~onof StationaryPhase Mod~~cation by the Mobile ase Surfactant in MLC, Anal. Chem., 61:1353 (1989).
20
I)
orgerding, R.L.~illiarn§,W.L. Hinze and F. Perspective in MLC,J:Liq. Chromatogr,, 12: 1367 . Borgerding and W.L
21.
d E ~ c i e n c yin MLC,
e, Investigation of the omatogr. A, 556: 263
(1991).
22
*
.Lavine and S. Hendayana, Band Broadening in MLC, J: Liq.
romatogr. & Rel. Technol., 19: 101 (1996).
23.
erthod, I. Girard and C.Gonnet, Additive ~ ~ e cont sSurfactant Adsorp~onand Ionic Solute Retention in MLC, Anal. Chem., 58: 1362 1986 .
24.
arcia Alvarez-Co~ue and LC ~rocedure for the Evaluationof lockers ~illanueva Cmanas, in Pharmaceuticalsusing ~ y b r l d ~ i c e l l aMobile r Phases, J: C~romatogr. A. 765: 22 1 (1997).
25.
erthod, Causes andRemedia~onof ~ e d u ~ e d J Chromatogr.A, 780: 191 (1997).
~ ~inc ~i e n cC y
,
26.
aledi, ~ydrophobic Selectivi~ in ~ i c e l l a and r ~ydro-organic "LC, Anal. Chem., 60: 876 (1988).
27.
and T.h d o , ~eparationof S ~ u l l Chromatography Columns and ~ i c e l l a Solutions, r J: Chro~atogr.Sci.,27: 653 ( l 989).
28. E.A.G. ~ i a n s s o nand S.N. Relaxation Spectromet~,E. (1975). 29.
30.
h&, ~ y n a m i Properties c o f ~ i c e l l a Solu~ons. r ain Alcoho~sandPo~mers on~icellar Stabili~, J: ColloidInterface Sei., 113: 484 (1986). .Almgren and S. Swamp, The Size of S ~ S ~ i c e l lwith e s various Additives, a Fluorescence ~uenchingStudy, in Surfactant in .L i n ~ a n (Eds.), York (1984).
31.
manesco andF.Nome, The Effect of Cosolve~t Formation on the of Aqueous Solutions, in Surfactant in Solutio L i n h a n (Eds.), Plenum Press, New York(1984).
32.
alliaris, J. Lmg, J. Slwm and R.Zana, Intermicellar~i~ration of Reactants: Effect ofAddi~onsof Alcohols, Oils, and ~ l ~ c t r o ~ t e s , JPhys. Chem., 91: 1475 (1987).
33
J. W i d , Spect~oscopic Study of the ontgomeryand ~ o l e c u l a r3usis of ~ e t ~ o nf ga C18 ~urfaceby ~ong-chain Alcohols, Anal. Chem., 66: 680 (1994).
*
34.
ledi, J.K. Stasters, A.H. Si~ultaneous ~nhance~ent of ~eparation Selectivi~ and So~vent Strength in "LC using ~ i c e l l e sin ~ydro-organicSolvents, Anal. Chem., 62: 130 (1990).
35. J. L a g and R.Zana, Effect ofAlcohols and Oils on the ~ i n e ~ofc s ~ i c e l ~For~ation-3reakdown e in Aqueous Solutions of Ionic Surfuctants, J: Phys. Chem., 90: 5258 (1986).
&er, ryhythe elations ship betweenthe Logarithm of k’ and ~omologue~ u ~ bineMLC r is not Linear?, Anal. Chem., 63: 180 (1991). .~
~ s t r oMultiple n ~ , Enantioselective ~etention Mechan~sms on ~erivatized Cyclode~trin Gas Chromatographic ~ t a ~ o n a r y ~ h Anal. a s e s ,Chem., 64: 873 (1992). 39.
o~asella,J. Fett and L.J. Cline Love, Effects of Organic er and Temperatureon MLC, Anal. Chem., 63: 474 (I 99 1).
40.
erhod, ~ilica,Backbone Material of LC C o l u ~ n Pachngs, J: ~hromatogr. ,549: 1 (199I).
41.
line-Love and J.J. Fett, ~ptimizationof ~ e l e c t i v in i ~MLC Procedures for the ~eterminationof Drugs in Urine by Direct ~ ~ j e c t i oJn:, harm. Biomed. Anal., 9: 323 (199 I).
I,1.
~ i ~ e l l Liquid ar Chromato~raphyvs. Conventional ~ e v e r s e d - ~ h aLs ieq ~ i d~hromato~raphy
Micellar LiquidChromatography (MLC) canbeused for separations commonly accomplished by ion exchange, normal phase andlor ReversedPhase Liquid Chromatography (WLC), by only selecting an appropriate combination of surfactanttypeandbondedpacking.From the early publications, micelles were indeed thought to replace organic modifiers in W L C and, in fact, many attempts have been tomade de~onstratethe unique advantages of MLC over aqueous-organicWLC. Despite the capabilities of MLC, the technique has not however received enough attention in solving the "real world" problems in analytical laboratories, for two apparent reasons. First, W L C with aqueous-organic eluents is a powerful and popular technique that would be difficult very for any other chromatographic method to replace. A reasonable argument is needed for an alternative technique. Second,the two techniques (MLC and conventional RPLC) have been often compared in an area in which MLC has clear disadvantages,that is, for the separation of uncharged and hydrophobic compounds. MLC is a poorchoice in this respect fromthe three impo~ant aspects of efficiency, elution strength and selectivity.
As shown in the previous chapter, column efficiency in MLC is very ofteninferior to thatachievedwithaqueous-organicsystems,and deteriorates as the hydrophobici~of solutes increases. Also, pure micellar eluents are generallyweak:longanalysis times are thus observed for hydrophobic solutes. Separation selectivity for these compounds is small,
as compared to aqueous-organic mobile phasesat similar elution st due to the similar e n v i r o ~ e n t of s mobile and stationary phases in The problem with the efficiency, elution strength and selectivity, can be improved bythe addition of a small concen~ationof an organic modifierto the micellar mobile phase; however, even this system may not beas suitable as aqueous-organic systems for the separation of hydrophobic compounds. n the other hand, itis impossible for any homogeneous aqueousmixed organic mobile phaseto achieve the range of interactions LC.Owing to the amp~philicnature of surfactants, solutes can associate with micelles through a combination of electrostatic, hydrophobic and steric interactions. For this reason, micellar eluentsare compatible with ionicandwater-insolublecompounds. The mainstrength of MLC lies precisely in the capability of performing and controllingthe separation of mixtures of cationic, anionic, and uncharged polar and nonpolar solutes, with isocratic elution. In this respect, micellar mobile phases oEer a clear advantage in terns ofchromatographicselectivityoverconventional aqueous-organic eluents.
for ~anipulating It appears that MLC can offerseveralmechanisms retention times and, as a consequence, controllingthe selectivity. such as the type and concentration of surfactant and organic solvent, pH, ionic strength and temperature,can be used for this purpose. ith thewidevariety of surfactants available, including ionic cationic), nonionic and zwitterionic, many uniqueseparations and applications can bedeveloped. ause organicmodifiersdonot the equilibria equally for all sol they can also change the sele a major the mobile phase. Moreover, with electrostatic inte ions playing role in the selectivity, it is important to specify the of the mobile phase. For instance, the measured pH of a 0.1 sodmm dodecyl sulfat l~onium bromide solution is 6.1 and that of 0.5 M cetyltrim values can obviously be changed by appropriate buEers, so that some solutes can be present in ionizedform.
ST
Some examples will be given this in chapter to show the interest and consequences of manipulating all these variables. The importance of pH in controlling the selectivity in the separation of mixtures of solutes experiencing protonation equilibria is noteworthy. However, it will not be discussed in this chapter, butin Chapter 8, where some new knowledge on the combination of surfact~t-mediatedequilibriawith protonation equilibria is introduced.
Interesting selectivities were reported from the early publications of MLC, mostly in the form of elution reversals. Peak crossing is ofien considered proof of enhanced selectivity. However, although elution order reversals are indicative of a unique separation mechanism in MLC, peak crossing in a complex mixture does not necessarily mean an improvementin the overall separation.
ure 7.1 Effectof sodium ,benzene (U),nitrobenzene( with permissionof the American Chemical Society
Chromatogramsof: (1) nitrobenzene, (2) toluene and (3)2-naphtholYeluted with 0.02 M SDS (A), and 0.20M SDS (B). Reprinted from Ref. 1 with permission of the American Chemical Society.
en the logarithm ofthe retention factor (log k) is plottedagainst the l o g a r i ~of surfactant concentration, as in Fig. 7.1, one important feature of MLC becomes apparent. The linear plots are not parallel, but
7
intersect one another. Convergence, divergence and reversal the peaks of can occur as a result of an increase in micelle concentration. Thus, not only the retention factor but also the selectivity coefficient, a (defined as the ratio between the retention factors of two solutes, k,/kb,where a is larger than l), are changing, In Fig. 7.1, the plots for nitrobenzene, toluene and 2naphthol cross. The consequences of these crossings are i~lustratedin Fig. 7.2, where the elution order is nitrobenzene, toluene, 2-naphthol, with 0.02 M SDS, while it is reversed with. 0.20 M SDS. The separation can therefore be controlled by the simple measureof changingthe concentration of surfactant.
5) describes the retention in MLC The Arrnstrong equation (Chapter as [2]:
where Q, = V, /V, is the phase ratio (V,is the volume ofthe active surface o f the stationary phase and V, the column dead volume), U the partial specific volume of the monomers of surfactant, ,,P and PwMthe partition coefficients between water and stationary phase, and water and micelles, respectively (remember Km = U (PWM- 1) is the solute-micelle binding constant), and [M] the micelle concentration. In Table 7.1, several solutes are listed in their elution order by the use of 0.02 M SDS. At this low micelle concentration, the system resembles ,,P controls the retention, as seen by the increased retention RPLC and ,,P values. As the concentration of surfactant is times with increasing increased, the importance of K,, increases, due to the larger n u ~ b e of r micelles present in the mobile phase. The larger the value of K,, ,the greater the effect of the increasing Concentration, as seen by the steeper slopes in the log-log plots for the more retained compounds (Fig. 7.1). The effect of ,,P on retention is independentof SDS micelle concentration,since the amount of surfactant adsorbed on the stationary phase remains essentially constant after equilibration, once the concentration is above the critical micellar concentration (cmc).
.l at er- station^ Phase and Water-Micelle Equilibrium Constantsfor Sodium Dodecyl Sulfate MicellarChromato~aphicSystem [l]
SDS
Benzene
5.O 8.7 9.4 13.8
Nitrobenzene
22.4
367
90
Toluene
41.3
875
215
2-Naphthol
59.5
1600
393
Phenol ~-Nitrophenol
nitroan aniline
160
39
314
77
342
84
314
77
P,, and P, are dimensionless constants, KAM is in Llmol because the molar volume value for SDS used was U = 0.246 L/mol.
2-Naphthol has larger KM value than the other solutes and, consequently, the steepest slope. A s the concen~ationof SDS increases, the retention ofthis compound decreases faster than the retention of nitrobenzene or toluene, so that elution reversals occur. Since KM for toluene is also higher than for nitrobenzene, complete reversal of elution order takes place for 2-naphthol, toluene and nitrobenzene, going from 0.02 M SDS to 0.20 M SDS. While KMfor 2-naphthol is much higher than for the early eluting solutes such as phenol, the difference in the values of P,, is so large that for these two compounds to very high SDS concentrations would be needed show a change in elution order.
. IIL I . Char~eof the ~ ~ r f a c t a ~ t A mode of selectivity can be employed via the selection of ionic surfactant (e.g., anionic or cationic), since polar solutes can interact electrostatically,
as well as hydrophobically with these surfactants. As previously stated, there is considerable adsorption of surfactant onto a nonpolar s~tionary phase, thereby providing electrostatic interactions of ionizable solutes with the stationary phase, as well as with the chargedmicelles. Withan appropriate surfactant, mixtures of polarandnonpolarsolutes can be resolved adequately. It is often possible to perform the separation of a greater number of compounds when working with mobile phases of low surfactant concentration, dueto the higher retentions. However, the analysis duration is longer.
Figure 7.3 Chromatograms of: (1) phenol,and (2) benzene, eluted with 0.05 M SDS (A), and 0.05 M DTAE3 B). Reprinted from Ref. 1 with permissionof the American Chemical Society.
TW
One interesting example of the selection of the surfactant charge was provided by the comparison of two surfactants, one anionic (SDS), the other cationic (dodecyl ~imethylammoniumbromide, DTAB), in the separation of a mixture of aromatic compounds. These surfactants form comparable micelles, differingessentially only in the nature of the polar end group. Both contain a linear C-l2 chain as their primary hydrop~obicportion and have similar cmc of 8 . 1 ~ 1 0M ' ~and 1 . 6 ~ 1 0M, - ~respectively [l]. Figures 7.3 and 7.4 show that dissociated phenol and benzene are not well resolved with DTAB, but are completely resolved with SDS. In contrast, p-nitroaniline andp-nitrophenol are not separated with SDS, but are well resolved with micellar phases. Also, benzylamine (not shown), whichis unretained TAB, interacts strongly with SDS and is retained (k =21 with 0.10 M
SDS).
c
L
l
~ h r o m a t o g r of: ~ s(1)p-nitroaniline, and (2)p-nitrophenol, elutedwith 0.05 M §D§ (A), and 0.05 M DTAB (B). Reprintedfrom Ref. 1 with permission of the American Chemical Society.
Y
111.2. 1 n t ~ ~ ~ c ~with i o nt s~ Set ~ t i o n a ~se The ability of micelles to solubilize and to interact selectively with solute molecules is believed to be the basis of separation in surfactant molecules readily adsorb on bonded stationary phases, such as C 18 or C8,and the modifications produced can have profound implications regard to retention and sele olute-stationaryphase interactions S some ofthe reported differences LC are thus very import selectivity, in between with a~ueous-organicmobi are some mea in due ation of the s t a t i o n a ~ adsorbed surfactant .
Figure 7. Structures of SDS- andCTAB-modifiedC 18.Reprintedfrom Ref. 3 with permission of theAmerican Chemical Society.
The role of the surfactant-modified stationary phase in an separation process was investigated by employing a set of six vanillin compounds (vanillin, 0-vanillin, isovanillin, ethylvanillin, vanillic acid and as retention probes [3]. ifferencesin selectivity between cou and micellar mobile phases ere found to be due t nature of the surfactant bonded-phase association. For hydrophobic alkyl tail is associated with the bonded alkyl pha head group projected away from the surface (Fig. '7.5) would lead therefore to the formation of ahydrophilic 1 phase. This simplifiedmodelwas also suggeste Armstrong et al. [ ~to] C. According to these explain poor stationary phasemass transfer in authors, a hydrophobic solute has to traverse the hydrophilicbo~ndarylayer '
formed by sulfate head groups, ions and associated water,in order to gain access to or leave the s ~ t i o n phase. a~ The b o u n d a ~layer can obstruct mass t r ~ s f e between r bulk solvent and bondedphase. adsorbed on C 18, the nitrogen head group is probably oriented closerto the silica sudace due to hydrophobic interaction between 7.5). the N-methylgroups of thes u d a c ~and t the bonded phase (Fig. en headgroup is thus probably i ~ c o ~ o r apartially t e ~ or totally into the bonded phase, giving rise to a modified bulk phase that is significantly denser. ~nco~oration of CTAB would ensurethat much ofthe hydrophobic character of the original bulk phase is retained.
F i ~ ~7. re Chromatograms of a mixture of vanillin compounds, separated on a C 18 column with micellar mobilephases of different nature at pH 3. From top to bottom: methanolwater 20230 (v/v), 0.02 M SDS and 0.02 M CTAR From Ref. 3.
Figure 7.6 shows the separation ofthe mixture ofvanillins on a C I 8 column with the following mobile phases: methanol-water 20:80 (v/v), 0.02 M SDS and 0.02 M CTAB. Several things are apparent from an examination of the chromato~rams.First, the test mixture is completely the a~u~ou§-organ.ic separated by the SDS micellar mobile phase, but not by mobile phases. Second, the efficiency of the chromato~raphic process 1s poor for a11 three mobile phases, and this implies that t variability in. the resolution of the test mixture, among the three mobile
7.7 Chromatograms of a mixture of vanillin compounds, separated on a C 18 column with micellar mobile phases, at pH 3 and varying SDS concentration. Molar concentration of SDS from top to bottom: 0.02, 0.03,0.04, and 0.05. From Ref. 3.
10
Separation of two binding compounds (lined peak, naphthalene; dotted peak, ~-nitrophenol),and two antibinding compounds (solid peak, 2-naphthol-4-sulfonic acid; open peak, sodium naphthalenesulfonate) with a cyano column and SDS ihicellar mobile phases. Molar concentration of SDS: (A)0.025, (B) 0.15, (C) 0.25, and (D) 0.40. From Ref. 5.
phases, is due to dif5erences in selectivity.Third, the retention time of four of the six vanillin compoundsis greater with 0.02 M CTAB, whch suggests that the interaction between each of these four compounds and the stationary phase is greater with C T ~ - m o d i ~ eCd18than with SDS-modi~edC 18. This would explain the superior resolutionachievedwith S vanillin compounds, which are also hydrophilic and probably undergo some type of selective hydrogen-bonding interaction withthe bonded layer. Although SDS adsorption enhances the selectivity of the statio^^ phase toward the vanillin compounds,SDS micelle-solute int contribute to the selectivity of this separation. For example, interact more strongly with vanillin than with isovanillin,as evidenced bythe greater ISm binding constant for vanillin, and this interaction is responsible, at least in part, for the baselineresolution of these two compounds. Nevertheless, the successfbl separation of the vanillin compounds withthe 0.02 M SDS mobile phase is primarilydue to solute-stationary phase interactions, which isalso the reason whythe separation of the vanillin test mixture is more favorable at lower SDS concentrations (see Fig. 7.7). III.3. Separation of ~ i x t ~ rofe gin s din^ and A n t i ~ i n d i n ~
~o~pounds Some separations may involve mixtures containing binding, nonbinding and antibinding compounds. By choosingthe correct micellar mobile phase, it is possible to separate anddistinguishbetween these three classes of compounds. Figure 7.8 illustrates the separation of a mixture containing binding (i.e.,naphthalene and p-nitrophenol), and antibinding (i.e.,2naphthol-6-sulfonic acidand sodium naphthalenesulfonate) compounds.In each successive chromatogram (A through D), the concentration of S the mobile phase was increased. m e n the SDS contents ofthe mobile increased, the retention of naphthalene and p-nitrophenol decreased, whle at longer 2-naphthol-6-sulfonic acidand sodium naphthalenesulfonate eluted times. Note that optimum resolutionwas obtained eitherat high or low concentrations (i.e.,chromatograms A and D,respectively), and that peak inversion occurred.
nd
The solubility limit theory explains the direct transfer of h i ~ l hydrophobic y solutes from micelles to the stationary phase [6]. The retention factor for these compounds can be expressed as:
where ,,P is the pa~ition coef~cient between micelles and s t a t i o n a ~ phase ,,P /PWM and ,,P >>l). (from eq. 7.1 with,,P = The implications on selectivity ofthe direct transfer furnishes a new evidence of the solubility limit theory. The retention mechanism of several hydrophobic compounds (i.e., benzene derivatives, polycyclic aromatic hydrocarbons (PAHs), and dihydropyridines) was studied in SDS and CTA micellar systems, by comparing e~perimentalselectivity coef~cientswith those theoretically calculated assuming a direct transfer mechanism [?’,g]. A mathematical expression was derived by using the three-pa~ition equilibria theory, whichexplains the tendency of selectivity coefficiel~ts to the ratio of,,P coefficients of the solutes, when the concentration of surfactant increases. Expressing the equation that relates the retention with the concentration of micelles as a function of,,P and ,,P :
the selectivity coefficient for a given pair of compounds (i.e., a and b) will be defined by:
'"1
Variation in the experimental (0)and theoretical selectivity ( as a function of micelle concentration, for three pairs of solutes: pyrene-acenaphthene (top), pyrene-toluene (middle), and pyrene-benzamide (bottom), in SDS-5% l-propanol (a-c) and CTAB-5% l-butanol (d-f) mobile phases. From Ref. 8.
As the P, coefficients for hydrophobic compounds are high, it is possible l. Also when [M) is increased, v[M] can be to consider that (1 -1&M) + U P w M ) = v[M]. It finally results with: large enough to make (v[M] F=
r"jMS,b
Figure 7.9 shows the variation of the theoretical (eq. 7.5) and experimentalselectivity coefficients (calculated from the retention factors), as a fbnction of micelle concentration, in SDS-5% 1-propanol and CTAB5% l-butanol mobile phases,for three pairs of solutes: pyrene-acenaphthene which are both very hydrophobic and for which a direct transfer mechanism can be assumedfor any surfactant concentration, pyrene-toluene in which only for pyrene can a direct transfer mechanism be assumed for all surfactant concentrations, and pyrene-benzamide in which benzamide does not experience a direct transfer, except at very highsurfactant concentration. Whenboth solutes experience direct transfer, the e~perimental and theoretical selectivity coefficients are very similar (Fig. 7.9 a, d). It is possible to predict the selectivity coefficient from PM,pa~itioncoefficients. In contrast, when one of the two solutes does not experience a direct transfer mechanism, the theoretical and experimental selectivity are different but this difference decreases under the conditions in which the direct transfer is favored (Fig. 7.9 b, c, e, Q. Therefore, for certain solutes, the selectivity coefficient tends toward a value that does not depend on the concentration of surfactant in solution. This tendency is due to a change in the retention mechanismfrom a threepa~itionequilibria mechanism to a direct transfer of the solutes, from the micelles to the stationary phase. This change is favored when: (i) the ~ydrophobiccharacter of the solute increases, (ii) CTAB instead of SDS is used as surfa~tant,and (iii) the polarity of the aqueous mobile phase is increased, With regard to the surfactant nature, it should be said that the partition coefficients of aromatic compounds are generally greater with than with SDS. This is due to electrostatic interaction between the positively charged CTAB micelles and the unlocated charge of the aromatic rings. When the polarity of the mobile phase is decreased by adding an
alcohol, the affinity of a hydrophobic solute for the aqueous bulk phase increases, subsequently increasing the difference betweenthe theoretical and experimental coefficients.This difference is minimal when using 1 -propanol in the mobile phase, which hasa higher polarity than l-butanol.
. icellar eluents composed of only surfactantare generally weakand suffer from poor efficiency. Althoughthe elution strength can be increased adequately incertain instances by increasing the micelle on cent ration, the chromatographic efficiency usually deteriorates. Addition of an organic solvent to the micellar eluent may give an adequate elution strength, but can also improve the chromatographicefficiency and lead to selectivity enhancements. All this will have a favorable efEect on both resolution and analysis time. The use of an organic modifierishowever not always appropriate. Selectivity enhancements might not lead to an improvement in resolution ifthe retention falls below the optimum k range, as a result of an increase in elution strength. In other instances, the addition of an organic solvent to micellar eluents may have a beneficial effect on retention, but the efficiency may remainlow. The rate of change in retention of different solutes varies with their charge and hydrophobicity, as well as with the nature of surfactant and organic modifierin the mobile phase. The partition coefficients of hydrophobic solutes decrease morethan those of hydrophilic solutes, with an increasing concentration of alcohol. Hence, the selectivity is modified. A s an example ofthe influence of the modifier on the separation, Fig. 7.10 shows the chromatogr~sof a mixtureof PAHs eluted with mobilephases containing a fixed concentrationof micelles (0.15 M SDS), in the absence of modifierandwithdifferentmodifiers.Asseen, the separation is impossible inthe absence of modifier, as the most hydrophobic as very wide overlappingpeaks, not well difEerentiatedfrom the baseline. better Amobile phase of 0.15 M SDS-15% 2-propanolgivesmuch separation with shorter analysis time.
1
1 G,O15
!
l
i i ‘c’
6
o.o’sl i butanol O.OO0
l . . . . . . . . . . . . . .
0
1
0
2
0
3
...-. .... 0 u o s
.*
J
~
Retention skne (min.)
~ r ~ ~ , ElutionofamixtureoftenPAHswithamicell~mobilephaseofO.15M lO SDS: (a) without modifier,(b) with 15%methanol, (c) with 15% 2-propano1, and (d) with 7% 1butanol. PAHs: (1) naphthalene, (2) acenaphthylene, ‘(3) fluorene,(4)anthracene, (6) 9-methylanthracene,(8) pyrene, (9) chrysene, (12) benzo[a]pyrene, (14) peylene, and (15) dibe~~ac]anthra~ene, Reprinted from Ref. 9 with permission of Elsevier.
Figure 7.1 1 illustrates the chromatographic selectivities of M SDS, 2-propanol, acetonitrileand tetrahydro~ran,in the presence of 0.02 for a mixture of seven amino acids and peptides. The volume fraction of organic solvents were adjusted so that the total analysis times of the three mobile phases were approximately the same. The elution order and selectivity of all solutes were similarfor the three modifiers, except for the different elution order of peaks land 3 for 2-propanol, as compared to those for acetonitrile and tetrahydrofuran, and poor resolution of peaks 5 and 6 for 2-propanol and acetonitrile.The chromato~ramsof amino acids and peptides for different~oncen~ations of 2-propanol and 1-butanol, at a fixed micelle concentration (0.08 M SDS) in the hybrid systems, are illustrated in
~
Figure 7.12. The strengthso f both hybridmobile phases werealso adjusted so that the retentionof the lastpeak remained the same.It can be o ~ s e ~ e d that, while all peaks were well separated for 2-propanol, there existed strong overlaps and coelution of peaks 3 , 4 and 5 (Fig. 7.12b), and 1-2and 3-4 (Fig. 7.12d), for l-butanol.
1
Figure 7.11 Chromatograms of a mixture of amino acidsand peptides eluted with 0.02 M SDS mobile phases containing: (a) 3% 2-propanol, (b) 12.5% acetonitrile, and (c) 3% tetrahydrofuran. Compounds:(1) tyrosine, (2) methionine, (3) alanyl-tyrosine, (4) t~ptophan, (5) asp~yl-phenylalanine,(6) leucyl-tyrosine, and(7) glycyl-leucyl-tyrosine. Reprinted from Ref. 10 with permissionof Elsevier.
ure 7.12 Chromatograms of a mixture of amino acids and peptides eluted with 0.08 h4 SDS and: (a) 8% 2-propanol, (b) 1.9% l-butanol, (c) 14% 2-propapl01, and (d) 3.6% 1butanol. See Figure 7. l 1 for compounds 1-7. Other co~poundsare: (8) leucyl-triptophan, and (9) phenylal~yl-phenylalanine.From Ref. 10.
An extensive investigation was made byKhaledi et al. to better understand the effect of adding organic solvents to micelles, for controlling the retention and selectivity in MLC, and how it compares with aqueous-organic RPLC systems. The retention behavior in MLC can be quite different from that in aqueous-organic RPLC despite the fact that, for both systems, hydrophobic interactions are the main driving force for retention. Ineffect, addition of an organic solvent to micellar eluents does not create an aqueous-organic system [1 1, 1rZl. The retention characteristics of solutes with a ternary mi~elles-water-organicsolvent eluent are similar to those in a pure aqueous
23
micellar eluent. This means that in the hybrid mobile phases, it is micelles that influence the role of the organic cosolvent in the mobile phase.
In RPLC with binary aqueous-organic mobile phases, the relationship between the retention factor (i.e., log k) and the volume fraction of organic modifier (e.g., methanol), c p,is often a quadratic equation E13]: log k = log kW + Aq + Bcp2
(7.6)
where the coefficient A is expected to be negative and B positive, and log kW is the logarithm of the retention factor of a solute in pure water, a measure of the interaction of the solute with a given stationary phase. Coefficients A and B are directly related to the elution strength of the organic modifier. Over a limited range of cp values, the relationship between the retention factor and cp can be reduced to:
where the slope of the line, S, is the solvent or elution strength parameter, which is generally proportional to the retention and molecular weight of solutes. The linearity of eq. 7.7 deteriorates in the low (less than 20% v/v) and highconcentrations (more than 80% v/v) of organic solvents. For two solutes a and b, where kb >k, ,Sbis often larger than S,. A s a result, the selectivity between the two solutes will decrease with an increase in organic modifier ~on~entration. Khaledi et al. [l l]suggested that a similar relations hi^ was valid in MLC with hybrid eluents of micelles-organic modifier:
an
Yo vfv Plots of log k' vs. cp for diverse amino acidsand peptides eluted with SDS mobile phasescontaining: (a) 2-propanol, (b)acetonitrile, and(c) t e t r ~ y ~ o ~ rCompounds: an. tryptophan (W), glycyl"leucyl-~osine(GLY), leucyl-tyrosine (LY), alanyl-tyrosine(AY), and methionine (M). Reprinted from Ref. 10 with permission of Elsevier.
where Shy, is the solvent strength parameter in hybrid micellarsystems and log k, is the retention in pure aqueous micellar eluent (i.e., without organic modifier). Figure 7.13 shows plots of eq. 7.8, for several amino acids and small peptides, asa h c t i o n ofvolume fractionof three organic modifiers: 2-prop~ol, acetonitrile and t e t r ~ y d r o ~ r aover n , a 3-1 5% v/v range. Although the linearity achieved is not good enough,log~ithmic the relationship between log k cp and is usefulto compare the elution strength for each modifier. The solvent strength parameters, S and S,, represent the sensitivi~ of solute retention with volume fraction of organic modifier in aqueousorganic and hybrid systems, respectively. Otherwise, the relationship between the slope and the intercept of eqs. 7.7 and 7.8 has a signi~cant effect on chromato~aphicselectivity. The selectivitybetween thosesolutes, whose slopes and interceptsare directly related to one another, will decrease with an increase in organic solvent concentration. In contrast, for cases where there is no direct relationship between the slope and the intercept, the selectivity may increase with organic solvent concentration [l 0,141. In fact, in WLC with methanol-water mobile phases, linear correlations have been reported for the slope vs. intercept of eq. 7.7, for a large group of compounds [131 For hybrid eluents in MLC, plots of slope vs. intercept of eq. 7.8 are given in Fig. 7.14. As shown, unlike conventional aqueousorganic eluents in the presence of micelles, no correlation was observed between s h y , and log ko for 2-propanol, acetonitrile and tetrahydrofuranmodified micellar eluents. One can then anticipate, therefore, a different selectivity behavior for these organic solvents in the presence of micelles.
C, retention and selectivity are governed thee by the compet~g equilibria: p~itioningfiom bulk solvent to micelles to the and stationary phase, or direct transfer from micelles to the stationary phase. Equation 7.1 can be r e ~ i ~ in e logarithmic n form as:
@P,,) and log (1 + Km [ if the relationships between bothlog ( are linear, the following can be expressed:
2
ac
S
ma6
&Y
LY
0.00
OF
LW
0.ar.
FF
0 25
3
AY
05
4
4.5
LY
W
Relationship between S, and log k',in eq. 7.8 for diverse amino acidsand peptides, eluted with SDS mobile phases containing: (a) 2-propanol, (b) acetonitrile, and (c) tetr~ydrofuran.Compounds: alanyl-tyrosine (AY), tyrosine (U), methionine(M),leucyltyrosine (LY),asp~l-phenylalanine(DF),glycyl-leucyl-tyrosine (GLY), tryptophan (W), 1eucy~-triptophan(LW), pheny~alanyl-phenylalanine (FF). Reprinted fromRef. 10 with p e ~ ~ s s i of o n Elsevier.
(7.1 1)
Pws0 and Km0 are the partition coefficients for pure aqueous micellar eluents [151. The parameters S, and S, represent the sensitivity of variations in solute pa~itioningfrom bulksolvent into the stationary phase and into the micelles, respectively, with changes in cp. From equations '7.l and 7.9-7.1 1, it is derived that s h y b is dependent upon S, and S,, according to: (7.12) The negative sign in equation 7.12 clearly reflects the competing equilibria (into stationaryphase and micelles). nature ofthe two pa~itioning In the absence of micelles, S, == 0 and Shyb = S,, which represents the solvent strength parameter in conventional aqueous-organic RPLC. This equation S values) in hybrid micellar systems also shows that the elution strength (i.e., will be generally smaller than in aqueous-organic eluents.
VI.2. ank king o s E l u t i ~ nStrength of Organic M o ~ ~ e r s A. widely accepted technique for characterizing solvents, in liquid chromatography, is the Snyder's selectivity triangle [161. This technique classifies organic solvents on the basis oftheir relative ability to engage in proton acce~ting,proton donating and strong dipolar interactions. ~e~ the resulting values are plotted on three axes in the form of a triangle, solvents having similar functionalities tend to fall within the same area of the triangul~plot(e.g., 2-propanol and 1-butanol belong to group 11of Snyder's triangle, tetrahydrofuran to group 111, and acetonitrile to group VIb). In principle, the solvents grouped in the same area of the triangle should have similar chromato~aphicselectivity, while solvents from other groups should exhibit different selectivity, for a given separation. This theory has been
widely accepted and hasoften formed the rationale for solvent selection for optimizing a given RPLC separation. The presence of micelles in the mobile phases of RPLC has great influence on the chromato~aphicselectivity of organic solvents. As a result, the classification established bySnyder seems to be no longervalid in with hybrid mobile phases. Thus, according to Snyder's solventclassi~cation, the chromato~aphicselectivity for 2-propanol and l-butanol in aqueousorganic mobile phases should be the same at equal elution stren~hs, ~ecause they belong to the same selectivity group. The S values of some amino acids and peptides, for 2-propanol-water and -butanol-water, 1 are given in Table 7.2 [l 01, The compounds are ranked according to the S values in2propanol-water mobile phases. As expected, the S values in l -butanol-water are larger than those for 2-propanol-water, and the ranks of S values of different solutes for both alcohols are the same. In other words, the selectivity of solutes in 2-propanol-water and 1 -butanol-water mobile phases will be similar at equal elution strengths. able Solvent Strength Parameters for l-Propanol-Water and l-Butanol-Water Mobile Phases, Without Surfactant and With 0.02 M SDS [lo]
Compound
2-PropanolWater
1-Butanol- 2-Propanol1 -ButanolWater 0.02 M SDS 0.02 M SDS
Aspartylphenylalanine
14.1
35.5
8.77
25 .'7
T~ptopha~
16.3
36.3
11.8
34.0
17.5
37.1
8.77
22.1
~lycyl-leucyltyrosine
18.4
44.6
7.49
21.1
Leucyl~ptophan
18.7
50.2
7.07
23.3
7.32 Phenylalanyl53.2 p~e~ylalanine
20.4
28.3
The Shy,values for the same test solutes in the hybrid systems of 2-propanol-SDS and l-butanol-SDS, at a micelle concentration of 0.02 s are also shown in Table 7.2. Since, in MLC, organic ~ o d i f i e r associate with micelles and, on the other hand, compete with micelles to interact with the solutes,the rank of S, values of some solutes for 2-propanol is different from that for 1-butanol. For example, s h y , for phenylalanyl-phenylalanine with 2-propanol is the second lowest and with 1-butanol is the second highest, or for leucyl-~osine for 2-propanol isthe second highest, while for l-butanol is the second lowest. Also, in a stu aromatic compoun~s,in methanol- at er and hybrid [141, anthracene showed S = 12.6 (the largest), benzyl alcohol was ranked ninth with S = 5.OS,and phenol hadthe smallest value of 3.13. As a result, the retention of anthracene in conventional aqueous-organic sensitive to variations in the concentration of organic solvent than benzyl alcohol or phenol, In the presence of CTA micelles, however, the co~espondingShy,values and the ranking were quite different and, in fact, opposite to that of conventional aqueous-organiceluents. Thus, s h y , =0.47 for anthracene (ranked tenth, one of the smallest), 1.53 for benzyl alcohol (the largest), and l.31 for phenol (ranked fourth). A comparison of the values of S and s h y b shows that the elution strength decreased due to the inclusion of micelles in the aqueous-organic media (i.e,, s h y , S). The smaller variations in S,,, mean that solute size is less important compared to the solvating effect of the alcohol modifier. The m a ~ i t u d of e the reduction in elution strength depends upon the degree of interactions of solutes and organic solvents with micelles. ~ i c e l l e s control the solvation ability of organic solvents and, as a result, their chromato~raphicselectivity. Consequently, one can expect that the ranks and the ~agnitudesof Sflyb for the alcohols also change with micelle concentration.
In Table 7.3, the s h y , values for some PAHs of environmental concern, with a wide range of hydrophobici~,are listed for methanol, 2propanol and l-butanol as modifiers, with SDS and CTAB micelles. As shown, the s h y , values can be ranked as:
7.3 Absolute Values of the Slopes of In k vs. Percentage of Organic Modifier, for Several PAHs Eluted with Hybrid Eluents of 0.15M SDSCalculated by Linear Regression, Alcohol and 0.02 M CTAB-Alcohol [17].
§I)§ 0.15 M
Compound
2-Propanol Methanol
Dibenz[ac]anthracene
CTAB 0.02 M
l-Butanol
Methanol
2-Propanol
l-Butanol
2.7
2.7
8.3
1.9
2.5
8.8
e n z o ~ g ~ ~ p e r y l e n e2.7
2.9
7.9
1.3
2.2
8.0
Fluoranthene
2.4
3.1
7.9
1.3
l .9
9.1
Chrysene
2.3
3.1
8. l
l .7
2.1
7.2
B e n z o ~ ~ ~ a n t h r a c e n2.3 ~
2.9
8.2
1.4
1.9
7.4
2.1
2.8
7.6
1.4
2.6
6.4
Perylene
2. l
2.9
8.1
1.6
2.7
6.7
Benzo~~~pyrene
2.0
3 .O
8.7
1.S
2.2
6.7
Pyrene
2.0
3.2
8.0
1.4
2.5
7.9
l .9
3 .O
7.3
1.4
2.6
8.7
~enzo~~~fluoranthene l .9
2.9
7.9
2.0
2.4
6.9
~~enanthrene
1.8
3 .O
8.3
1.2
3.2
9.2
Anthracene
l .8
3.2
8.O
1.2
3.1
9.6
Naphthalene
1.8
3 '3
7.3
1.4
4.1
6.9
Acenaphthylene
1.7
3.2
7.5
lS
4.2
6.3
Fluorene
1.6
2.9
8.0
1.4
3 *2
10.2
enzo~e~pyrene
which issimilarto conventional aqueous-organic systems, as 1-butanol is the strongest solvent and methanol the weakest. The larger S,, for 1-butanol (35 times) corroborates that &us solvent interacts more strongly with micelles and can compete better for the interaction with solutes. Because each alcohol modifierinteracts differently with micelles,the selectivity changes from one another, Finally, the different impact of micelles of CTAB on the ranking of solutes, and the importance of solute size and hy~rophobicityshould be noted.
U. 3.
Si~altaneoas Enhance~ent of Elution Stren~thand Selectivi~with ~ y ~ r ~ d ~ i c Elaents ellar
A s commentedabove,inconventional
WLC, a systematic decrease in selectivity occurs usually as a result of an increase in volume action of organic modifier (i.e., elution strength). In contrast, in the presence of micelles, the selectivity may increase, decrease or remain m changed with elution strength. The elution strength may thus beenhancedwithout sacrificing the selectivity. It should bereminded that improving the resolution and reducingthe separation duration are the two important goals in many optimizationstrategies. Simultaneous selectivity enhancement with elution strength can lead to a better separation in a shorter time period. The simultaneous enhancementin elution strength and selectivity, often observed in MLC, can be attributed to the existence of the three competing equilibria. Examplesare given bythe separation of a mixture of amino acids and small peptides, and a mixture of substituted benzenes, eluted with hybrid eluentsSDS of and 2-propanol(Figs. 7.15and 7.16) [151. Figure 7.15 shows the influence of the concentration of 2-propanol on selectivity, for several pairs of compounds at a constant micelle concentration. Selectivityvariations occur systematically andmonoto~cally for different peaks, as a result of an increase in 2-propanol phase content. Figure 7.16 illustrates the changes in selectivity €or the same pairs of compounds dueto the variations in micelle concentration, at a constant 2propanol composition. Interestingly, micelle concentration had an opposite effect on selectivity as compared to 2-propanol. For those pairs of peaks whoseselectivitywerereducedwithincreasing 2-propanol content, an enhancement in selectivity was observed as a result of increasing micelle
35
2.@
QiLY/oF 0.S 0
I
Me.
, '
/
I
d
Variation in selecvolume fraction of 2-propanol at 0.02 M SDS, for amino acids and peptides (top), and substituted benzenes (bottom). Compounds: benzene (B), benzyl alcohol (B Alc.), naphthalene(Napht.), benzonitrile (B. Nit.), be~aldehyde(B. Ald.), nitrobenzene(Nitro B). See other symbolsin Fig. 7.14. Reprinted fromRef. 15 with permissionofthe American Chemical Society.
N
0.s .J 0.ED
V~iationin selectivity celle concentrationat 12% 2propanol, for amino acidsand peptides (top) and substituted benzenes (bottom). See symbols in Figs. 7.14 and 7.15. Reprintedfrom Ref. 15with permission of the American Chemical Society.
0.
0.
concentration and vice versa. These observations suggest that although the elution strength increases withthe concentration ofboth micelle and organic solvent, the effect of both on selectivity could be quite different, even opposite. Micelles and 2-propanol compete to interact with solutes, and as a result, theyinfluencethe role of one another in con~olling the retention and selectivity. eq. 7.X:
MLC can be derived fiom An expression describing the selectivity in
(7.13)
where am is a function of the binding selectivity to micelles: (7.14)
and ,,a
is the stationa~-phase pa~itioning selectivity: (7.15)
and have opposite effects on chromatographic selectivity. The values of Km and P, usually decrease with an increase in volume fraction of modifier. The degree of decrease of these parameters, however, is not equal for different solutes. This can lead to changes in selectivity. To study this behavior, the partition coefficients of a compounds (Le., amino acids, peptides and benzene derivatives) measured at different concentrations of 2-propanol. For those compounds for which the selectivity increased as a result of increasi~g2propanol, the partition coefficients ofthe less retained compound inthe pair decreased to a higher degree, as compared to the more retained comp~und. an increase For other pairs of solutes for which the selectivity decreased with in 2-propanol, the reduction inthe coefficientsofthe compound showing the
higher retention in the pair was larger than for the compound with lower igure 7.17 illustrates variations in Km and Pwsfor three solutes, with an increase in percentag~of 2-propanol. As shown, the rank of the solutes is t ~ p t o ~ h (W)> a n alanyl-tyrosine (AY)>leucyl). Leuc~l-triptophan/t~ptophan(LW/W) and leucyltriptophan/alanyl-tyrosine(LWIAY) belongto the first group of solute pairs (for which the selectivity increases as a result of an increase in organic solvent concentration)9and W/AY belongs to the second group of solute airs (for which the selectivity decreases as a result of an increase in organic solvent concentration). ,,a and am for the three pairs of solutes, with The variations in increasing 2-propanol, are shown in Figure 7.18 (top and middle). The selectivi~is con~olledby the co~petitionb e ~ ~these e n two p a r ~ e t e r(eq. s 7.13). Figure 7.18 (bottom) illustrates the variation in chromatographic lectivi~ with volume fraction of organic modifier.It is shown the in figure that there is a systematic change in selectivity with organic solvent concentration, which may increase or decrease.
Another i ~ p o ~ afactor n t is the effect of micelle concentration on am, and therefore, on c~romatographicselectivity. For the first group of pairs of compounds, like LW/W, both,,a and aAMincrease with
01, but the rate of increase of the former is greater (Fig. 7.1 8). micelleconcentratiol~increases at fixed 2-propanol, the rate of increase in am becomes larger than the rate of increase,,in a ,and the selectivity decreases. For the second group ofsolutes, like W/AY9both aws and am decrease with an increase in 2-propanol, but the degree of decrease ,,a is more than that of am. A s a result of an increase in micelle in concentration at fixed percent 2-propanol, the degree of decrease in aAM es larger than that of aWs9 and the selectivi~increases.
,when
In a~ueous-organicsystems, it is common practice to first adjust the ,and then optimize the selectivity at a constant elution id systems, where an increase in elution strength can often enhancement in selectivity, a separate optimization of d selectivity is inefficient. This is the main subjecttreated ere it is shown how the si~ultaneous opti~ization of cing elution strength and selectivity can be performed.
7-
2.4
2.2
2
1.
1.
ure '7.1'7 Linear regression of log {1 + KAM[M]} and log {+Pws}vs. volume fraction of 2-propanol for: leucyl-~iptophan(LW), tryptophan (W), and alanyl-tyrosine (AY). Reprinted from Ref. 15 with permission of the American Chemical Society.
c " " " f - - - - " - - L " -
" " " " " "
l
a
t
ure 7.18 Variation in ,,a ,a A M , and linear regression ofa,,/a,, vs. volume fraction of 2-propanol for: l e u c y l - t r i p t o p h ~ t ~ p t o p(LMJIW) h~ and t~ptopha~alanyl-tyrosine ( ~ / A ~The ) , concentrationof SDS in the mobile phases was 0.02 M. Reprinted from Ref.15 with permissionof the American Chemical Society.
IVIT
7
The effect on selectivity of increasing the temperature of the chromatographic column is usually ignored, perhaps because the improvements in selectivity are not very substantial. However, with highly hy~rop~obic solutes,the in~uenceof temperature may be impo~antsince the rate of transfer of solute between bulk aqueous phase and s t a t i o n a ~phase, or between micellar pseudo-phase and stationary phase, will probably be slow. An increased temperature will favor the transfer processes. Figure 7.19 shows the chromato~ramof a test mixture of nine PAHs. At 30"C,the mixture was unresolved and all peaks showed bad symmetry and efficiency, especially for the most hydrophobicsolutes (peaks 4-1 5). At 40"C, the separation was still poor. At 5O"C, the improveme~t was evident, the efficiency of the separation of peaks 8-15 increased considerably. The best situation occurred at 60"C, where the ~ i ~was~ r totally resolved and the analysis time diminished in comparison with the initial situation at 30°C. The enhanced silica dissolution is a problem that should not be neglected when working at elevated temperature with micellar phases.
e
Ti
in)
igure 7.19 Chromatograms of a mixture of PAHs, using 0.15 M SDS as mobile phase. Temperature: (A)30°C, (B) 40°C, (C) 50°C and(D) 60°C. The compounds aresthesame as in Fig. 7.10. Reprinted from Ref. 17 with permission of Elsevier.
1.
P.Yarrnchuk, R. Weinberger, R.F. Hirsch and L.J. Cline-Love, Selectivi~in LC with Micellar Mobile Phases,Anal. Chem,, 54: 2233 (1982).
2.
D.W. Arrnstron~and F. Nome, Partitioni~gBehavior of Solutes luted with Micellar Mobile Phases in LC, Anal. Chem., 53: 1662 (1981).
3.
B.K.Lavine, S. Henda~anaandJ. Tretreault, Selectivi~in Micellar W L C : C18 and C8 Alkyl Bonded Phases,Anal. Chem., 66: 3458 (1 994).
4.
D.W. Arrnstrong, T.J. Ward and A. Berthod, Micel~ar~ ~ e cont s Molecular usio ion: ~ ~ e o r e t i c a and l Chro~atographi~ Co~siderations,Anal. Chem., 58: 579 (1986).
5.
D.W. Amstrong and G.Y. Stine, Selectivity in PseudophaseLC, Anal. Chem., 55: 23 17 (1983).
6.
M.F. Borgerding, F.H. Quina, W.L. Hinze, J. Bowermaster and H.M. McNair, Investigation of the lietention Mech~nism in on ionic MLC usingan Alkylbenzene ~omologousSeries, Anal. Chem., 60: 2520 (1988).
7.
J.M. Saz and M.L. Marina, lietention Mechanism and rmplications for Selectivity for a Group of ~ihydropyridinesin Ionic MLC,
J. chroma to^. A, 687: 1 (1994). 8.
M.A. Garcia and M.L. Marina, Influence of Alcohol Organic Modl~ers upon the Association Constants and lietention Mechanis~ for Aromatic Compounds in MLC, J. Liq Chromatogr. & lid. Technol., 19: 1757 (1996).
9.
M.A. Rodriguez Delgado, M.J. Sanchez,V.Gonzalez and F. Garcia Montelongo, Influence ofAlcoholic Modl~erson the Selectivi~of the Separation of a Group of Po~cyclicAromatic ~ydrocarbons by MLC,Anal. Chim. Acta, 298: 423 (1994).
10.
.G. Khaledi, Selectivi~of Organic Solvents in MLC of Amino Acids and Peptides, J. Chromatogr., 631: 125 (1993).
11.
aledi, E. Peuler an J. Ngeh-N~ainbi,Retention~ e h ~ i o r o ~ ~ o m o l o ~Series o u s in RPLC using Micellar, Hydro-~rganic and Hybrid Mobile Phases, Anal. Chem., 59: 2738 (1987).
12.
i, Hydrophobic Selectivi~in Micellar and ~ y ~ r o C, Anal. Chem., 60: 876 (1988).
13.
P.J. Schoenmakers, ~ p t i ~ i z a t i oofnChromatographicSelectivi~, Journal of C ~ r o ~ a t o ~Library, a ~ h y Vol. 35, Elsevier, Amsterda~, 1986, Ch. 2.
14.
M.G. Khaledi, J.K. Strasters, A.H. Rodgers and E.D. Simultaneo~s ~nhancement of Separation Selectivi~ and Solvent Stren~thin RPLC ~ i nMicelles g in ~ y d r o - ~ r g a nSolvents, ic Anal. Chem., 62: 130 (1990).
15.
.Kord and M.G. Khaledi, Controlling Solvent Stren~thand ~ e l e c t i vin i ~MLC Role of Organic Modl~ers and~icelles, Anal. Chem., 64: 1894 (1992).
16.
V.J. Barnick, S ~ a t e ~ i for e s SolventSelection.ALiterature Review, Trends Anal. Chem. 16: 293 (1997).
17.
.A.Rodriguez Delgado, M.J.Shchez, V.Gonziilez andF. Garcia Monte~ongo,Role of Temperature in the 3 e h ~ i o rof PAHs in MLC. ThermodynamicAspects, J. C h r o ~ a t oScience, ~. 33: 647 (1 995).
. Chapter 5 showed howthe presence of micelles in a ~eversed-PhaseLiquid Chromatographic ( ~ L Csystem ) can provide a great variety of interactions (Fig. 5.1). In the micellar mobile phases, the solutes can remainoutside the micelle associated with the polar head of the surfactant, can form a part o f the outer palisade layer, or can penetrate into the micelle core. Also, the monomers of ionic surfactants can be adsorbed on alkyl-bonded stationary phases, mainly through hydrophobic interaction between the tail of the surfactallt and the alkyl chains of the stationary phase. In this case, the charged headofthe surfactant will remain in contact with the polar solution. Solutes can experience hydrophobic interactions with either nonpolar tails ofthe adsorbed surfactant and/or nonpolar bonded moieties of the stat~onary phase, and polar interactions with the ionic head of the adsorbed surfactant and free silanol groups on the stationary phase. Nonpolar solutes will only be affected by hydrophobic interactions with both micelles and stationary phase, but chargedsolutes will give rise to two distinct additional situations, according to the sign of their charge, which can bethe same or opposite to the sign o f the head of the surfactant and can, therefore, be repelled or attracted by the surfactal~t. Most reported procedures for the determination of compounds in Micellar Liquid C~romatography(MLC) make use of micellar mobile phases containing an organic modifier, usually a short-chain alcohol or acetonitrile. These modi~ersincrease the elution strength, which is pa~icularlyimportant for the most hydrophobic solutes, and often improve the shape ofthe chromato~rap~ic peaks. The most hydrophilic alcohols do not penetrate the micelles, but butanol and pentanol caninserted be into the
micelle with their hydroxyl group orientated towards the Stern layer, and their hydro~arbol~ chain remaining inside the nonpolar micelle core. The their hydrocarbon chain remaining inside the nonpolar micelle core. The inco~orationof the alcoholin the micelle can resultin additional interactions with the solutes. On the other hand, the modifiers solvate the bonded stationary phase and reducethe amount of surfactant adsorbed, the effect beinglarger with increasing concentrationand hydrophobicity of the alcohol. The rigidity of the surfactant-al~l-bondedligand structure may also be affected.
A decrease in retention times is usually observed when either the micelle concentration or the concentration of organic solvent is increased. owever, different components in a mixture respond in different ways to changes in concentrationof surfactant and/or organic modifier, resulting in changes in resolution. Selection of pH in the mobile phase is also often extremely important for the resolution of complex mixtures, owing to the side acid-base reactionsof many solutes. Other variables to be considered are temperature and ionic strength. The chromatographer is concerned with the achievement of the optimum mobile phase that permits the separation of the compounds in a mixture, in the minimum time. This task may be reallydifficult when two or more variables are involved in the opti~zationprocess. The optimization strategy utilized may be sequential or interpretive. In a sequential strategy, the retention of the solutes is not known CT priori, and each set of mobile phasesisdesignedby taking into account the retentionobservedwith previous eluents.In contrast, in an interpretive strategy, the experiments are designed before the opti~zationprocess and used to fit a model that will permit the prediction ofthe retention of each solute. This strategy may be much more efficient and reliable. A sequential strategy will be inadequate when several local (or secondary) maxima exist (as occurs in chromatography), and may not givethe best m a ~ m u mthat , is, the optimum. The necessity for an adequateexperimentaldesignbecomes especially important when dealing with forms of liquid chromatography suitable for the simultaneous analysis of ionic and nonionic compounds, such LC, whereseveral variables should be controlled (i.e.,type and concentration of surfactant and organic modifier, pH, temperature and ionic stren~h).The method development strategy must provide the chromatographer with an answer to whchvariables (factors) should be used, and how
to setup initial experiments to search the appropriate variable space in an efficient way. The separation process in a micellar chromatographic system requires a structured approach inthe development of practical applications. o Ideally, the resolution of complex m i ~ ~ rshould es be made and in a short time, with minimal consumption of reagents. The reader should not be repulsed by the somewhat complex equations developed in this chapter. They are incorporated in the M ~ C software ~ ~ supplied ~ M with the book. Appendix I describes the computer assisted way to model the retention behavior of a given mixture in MLC. With a few guided experiments, the software will be able to m0 the retention of an actual mixture components with surprising accuracy.
I. 2. ~ r e ~ i ~ t a~etention ble ~e~avior As commented above, in MLC the chromatographer issupplied with several tools to refine a separation and optimize the resolution for a given multicomponent mixture. However, this advantage canonly be exploited to the fullest when all variables are taken into account sim~ltaneous~y. using the proton, surfactant?or organic modifier concentration, or by v a ~ i n g one after optimizing the other, the best separation can easily be missed.The use of an inte~retive optimization strategy, which needs specific illfo~ation on the retention of the individual components in a mixture, may require a relatively low number of experiments to derive an acceptable separation. The accurate predictionof the retention of each compoulld aingiven mixture is moreimportant in MLC than in conventional WLC, due to the negative effect of the broadening and distortion of the chromatographic peaks for some solutes and eluent com~ositions,with micellar mobile phases. ~ortunately,as shown in this chapter, solute retention in accurately predicted. A regular change in the retention behavior is observed with the variation of the concentrations of surfactant and organic modi~er, and pH. The simultaneous optimization of the resolution and analysis time s, is thus possible, on the basis of a limited number of e ~ p e r i ~ e n teven though these experiments were relatively far apart in the variable space. As seen below, modeling ofthe retention behavior in MLC has also allowed a
b e ~ eunderstanding r of the processes occu~inginside the chromato system.
ace
113. o election o s t ~ e~ariable
The boundaries ofthe variables included in the mode~ing set by the operator, on the basis of previous experience. T are imposed by the practical limitations of the chromato~aphicsystem: the lower surfactant concentration must be well above the critical micellar conce~tration(cmc), and strong enough to cause elution of all components. The upper surfactant concen~ationis determine^ by a combination of the so~ubilityof the s u r f a c t ~ tthe , acceptable viscosity of the resulting mobile phase (i.e., maximum pressure drop over the column), and the degradation of the efficiency at higher concentrations. The concentration of organic modifier must consider the retention times, and is limitedto a maximumto ensure the integrity of micell S. Finally, the pH range for a silica reversedphase support, between 2.5 and 7.5, should be considered. en the concentration of organic m o d i ~ ebecomes r too high, the characteristicsof the micellar pseudo- ase change: a microemulsion can be created or the micelles can com tely isa appear. The maximum er reported for modeling purposes, with sodium concentrations of surfactant, havebeen 15% propanol, 6% butanol, dodecyl sulfate (S 3% pentanol, and 20% acetonitrile (v/v concentrations).
. I11 I ,
~ i c e l l eConcentrationas ~ n i ~ ~x ~ ~ e r ie~ e n t a l ~ a r i a b l e
The three-phase theory[l], which postulates the following e~uilibriabetween a solute, A, the micelle, ,and the stationary phase, S:
predicts the retention factors, k,in micellar media at a given pH,according
to a very simple equation: which relates the retention with the concen~ationof monomers of surfactant 1. In this equation, AM is the binding constant le, and KASthe partition coefficient between stationary phase and water multiplied by the phase ratio (see Chapter 5). This equation can be rewritten as: 1
-= c* + k
Cl [M]
It has been extensively commented in Chapter 5 that eq. 8.3 has been verified, experimentally, for a large number of solutes (ionic, and polar and nonpolar neutral), different types of surfactants (anionic, cationic and nonionic), and diverse column materials (mainly C8, C 18 and cyano). It is also valid for mobile phases containing an organic modifier. Deviations from the model are, however, observed for very highly and poorly retained solutes.
The separation of mixtures of compounds showing a wide range polarities of can be advantageously made by using a gradient of ionic surfactant. In MLC, gradient elution is favored because at moderate concentrations of ionic surfactant, the composition of the stationary phase remains constant during the gradient. Therefore, the only reequilibration process necessary, before the next gradient run, is flushing the chromato~aphicsystem withthe initial mobile phase. Prediction of the retention in gradient conditions in MLC has been. made based on the gradient elution theory developed by
Snyder [2], assuming the linear model given byeq. 8.3 and a linear change in micelle concentration [3]:
V being the volume of the mobile phase delivered. The equation finally derived was:
where tgis the gradient retention time, tothe dead time, VJt0 the volume flow rate, klthe retention factor ofthe solute at the initial mobile phase, and t, the delay time (time before the gradient actually reaches the top of the column). This model has shown excellent results for a variety of solutes and SDS gradients in the 0.10-0.50 M range.
11 3. ~ i ~ ~ l t a nElffect e u ~usf p H and ~ i ~ eCuncentratiun ~ l e eth hod development in MLC can be largely potentiated by extending the m o ~ e ~ i nofgthe retention along the pH scale (Fig. 8.1). ~ r u n y a n aand ~ Cline-Love [43,and Rodgers et al. [5,6], studied the effect of pH on the MLC retention of weak acids and bases. These authors combined the equations that describe the acid-base equilibrium for a monoprotic system in water:
A + H * HA
se .l Solute-micelle andsolute-stationary phase interactions in micellar mobile phases
of surfactant and organic modifier, at varying pH.
and the retention factor in a micellar chromato~raphicsystem:
where h is proton concentration, K, is the protonation constant, [AS], [AM] and [A] refer to the basic species, and [HAS], [HAM] and [HA] to the acidic
ewriting this expression, an equation similar to the reciprocal of eq. 8.3 is obtained: ‘AS
+
‘ M S KH
h
1 + KHh
k =
l +
Km
KwKHh [W l + KHh -t
-
C S
1
+
‘k[W
(8.10)
where KHAsand AM are apparent constants with respect to proton concent~ation.Equation 8.9 may also be rewritten as follows:
(8.1 l )
or
1 + KFh
(8.12)
where kA and kHAare the retention factors of the basic and acidic species, respectively. Equations 8.1 1 and 8.12 indicate that the retention varies with
following a sigmoidal behavior between the retention of the acidic species and the basic species. The effect of the concentration of surfactant on the apparent protonation constant is also shown in eq. 8.1 1. This eq~ationmay easily be extended to solutes exhibiting several protonation e~uilibriain the pH working range of reversed-phase colu protic system, the retention as a function of micelle concentration will be given by:
2 are the consecutive protonation constants9and KH2AS and KH2m the partition constants of the diprotonated species. Figure 8.2 shows simulated retention of ~ i ~ e r i o ncompounds ic in MLC, based upon eq. 8.13, with nonionic and anionic micelles [6]. The retention behavior with nonionic micelles (Fig. 8.2, top) is similar to conventional WLC, with the retention passing through a minimum. For anionic micelles (Fig. 8.2, bottom), electrostaticrepulsion betweenthe solute and surfactant will result in lowretention at elevated pH. For both nonionic and anionic surfactants, the retention decreases with increasing micelle concentration. Itis important to note that a large degree of error in predicting the retention is to be expected as a result of small errors in p m e a s u r e ~ ~ n tin s )the region where the solute is being protonated. For monoprotic systems, the apparent protonation constant can be determined by measurin~the retention factor at several pH values and ~ ~ i n g the e~perimentaldata to eq. 8.12, via nonlinear regression. A similar equation can be used for diprotic systems. Figure 8.2 shows micellarinduced shifts of the apparent protonation constants, with increasing anionic micelles. The logarithm of the first protonation constant(log K, =p t ~ p t o p ~ aphenylalanine n, and lysine in water is approximately 2.4. For the are between 3.60 and 4.56. Therefore, a micellar mobile phase, log KHM
greater range of protonation constants exists in the micellar mobile phase. This results in larger changes in selectivity with pH in MLC due to the ~ifferentialshift in protonation constants b e ~ e e nthe solutes, which n increases the i ~ p o ~ a n of c epH in the o ~ t i ~ i z a t i oprocess.
Predicted retention of a zwitterion (pK, and pK,, are acid-base dissociation constants in water and micellar solution, respectively, and [S] is surfactant concentration), as a function of pH and micelle concentration with nonionic micelles (top): solute-micelle association constants for cation, zwitterion and anion are 3, 0.1 and 3, respectively, cmc = 2.4~10-~ M; and with anionic mice~~es (bottom): solute-miceilesoc cia ti on constantsare 10,000, 10 and 0.1, respectively, crnc = 8 . 3 ~ 1 0 M.~ The derivatives of the sigmoidal curves at each micelle concentrationare also shown to give the apparentdissociation constants. Reprinted from Ref. 6 with permission of the American Chemical Society.
. IIL 1. A ~ o ~ e l Convent~onal ~ o r RPLC A s a result of the increasing use of modifiers in MLC, the need for an
adequate description of the retention in hybrid micellar mobile phases appeared. Khaledi et al. [?,81 were the first which intended to model the retention of solutes in these systems. They assumed that the linear p, relationship between log k and the volume fraction of organic modifier, c followed in conventional RPLC over a small range of values of c p,was also valid in hybrid MLC at constant micelle concentration:
According to these authors, log k,in eq. 8.14 was the logarithm of the retention factor at a given micelle concentration in the absence of modifier. However, linear relationships are only actually obtained with methanol as modifier, probably owing to its weak elution strength. Figure 8.3 shows log k vs. cp plots for several aminochromes [S]. With the data plotted, excluded the point for U,= 0, the value of kowas calculated from the intercept of the fitted straight-line according to eq. 8.14. It was found that the difference between the experimental and calculated kowas larger for an increasing alkyl chain-length of alcohols. Curiously, a linear relationship existed between this difference and the number of carbon atoms in the alcohol. IIL 2. It~~ative Re~ression Strategy In order to model the retention in a hybrid micellar mobile system, Strasters et al. [g] proposed a procedure that used the retention data of only five mobile phases: four measurements at the corners of the selected two-dimensional variable space, defined by the concentrations of surfactant and modifier, and the fifth in the center (design VI in Fig. 8.4). In this method, the rectangular variable space is divided into four triangular subspaces defined by three of the five measurements: two neighbor corner
Logkvs. cp plots for~inochromes:( l ) noradrenochrome, (2) adrenochrome, (3) dop~inechrome,and (4)isopropylnora~enochrome.The mobile phase contained0.05 M SDS. Reprinted from Ref. 9 with permission of Elsevier,
W
V
Some experimental designs used in hybrid MLC. The abscises represent the concentration of surfactant, and the ordinates the concentration of modifier.
4
triangles. 1
L
points and the central point (Fig. 8.5). For this reason, it will be called here method of triangles. The authors assumed that the retention of solutes was linearly related to the mobile phase variables within a selected portionofthe space, and fitted a separate logarithmic linear function: l o g k = c , + c , [M]+c,cp
(8.15)
in each triangular subspace, The method was applied to the Optimization of the separation of a set of fifteen phenols eluted with hexade~~ltrimethylammonium bromide (CTAB) and 2-propanol at pH ' 7 ,and thirteen amino acids and small peptides eluted with SDS and 2-propanol at pH 2.5. Torres Lapasitj et al. [101 later used, instead of the logarithmic function, a hyperbolic dependence. The method of triangles can be considered as i n t e ~ e d i a t between e a true sequential method and an interpretive method. The retention in other mobile phases is calculated by inte~olation linear inside each triangle where the coordinates belong. Fu~her,a chromato~amis simulate^ and compared with experimental data to verify the quality of the prediction. When experimental and predicted chromatograms coincide, a onf firm at ion of the assumed linearity is obtained. However, when strong deviations ofthe linear model are observed, additional data-points should be included to refine the prediction, and this is usually performedwithout leaving the region of the variable space initially chosen, by a further sub~ivisionof the response surface into smaller triangles (see Fig. 8.5). This, obviously, can result in an undesirable large number of experiments, and can cause the elimination of significant maxima inthe first steps of the optimization process. If the optimum is found near one of the experimental points of the design, the prediction will be reliable. On the contrary, if it is at the center of one triangle subspace, serious errors may result. Certainly, the success or failure of the method of triangles depends on the correctness of the linearity assumption of the retention model. The use of the logarithm of a function instead of the function to make a linear interpolation is a common practice, when the range of variation of the function must be reduced. The division of the variable space into four triangles reduces even more the variations of the function (i.e.,a function
that does not fit in the whole variable space may be acceptable over a smaller range).owever,ithasbeendemonstrated that the use of a hyperbolic firnction makes less necessary the addition of new mobile phases to refine the predictions, dueto the better fittingaccuracy [IO].
The scheme of interpolation followed in the iterative regression strategy explained above is not very simple from a practical point of view. The method of triangles requires at least four different equations, onefor each established subspace. The use of a single equation to describe the retention behaviorofasolute,inthewholevariable space, seems to be more convenient to predict the retentionof a solute in any mobile phase, with a minim^ effort. Table 8.1 shows some ofthe models (equations) that have been considered, where thelo gar it^ (eqs. 8.15-8.22), or the reciprocal of the retention factor (eqs. 8.23-8.30) are related to micelle concentration and volume fraction of organic modifier. To obtain an adequate equation, a detailed examination of the retention behavior is required. Figure 8.6 shows plots of the reciprocal of the retention factor of noradrenalinevs. §D§concentration for constant 1propanol concentration, andvs. l -propanol concentrationfor constant § concentration [g]. Good agreement between experimental and calculated data isobserved.Theretention factor of noradrenalinedecreases at §D§concentration studied increasing 1 -propanol volume fraction, for each (Fig. 8.6b). However, this effectisattenuated as theconcentrationof surfactant increases, that is, the elution strength ofl-propanol decreases at increasing concen~ationof surfactant. The same behavioris observed with the surfactant (Fig. 8.6a) (i.e., its elution strength decreases at increasing concentration of modifier).Thus?noradrenaline showed a similar variation of the retentionfactors with respectto both surfactant and modifier. After observing these results,Torres Lapasio et al. [S] decided to make a more extensive study of the capability of the empirical equations given in Table 8.1, to describe the retention behavior at any surfactant and modifier concentration.The errors in the prediction the of retention for a set
+ + + +
o"o"O"0"
+ + + +
.6 Retention behaviorfor noradrenaline:(a)Ilk vs. total concentration of surfactant plot for: (1) 0, (2) 0.015, (3) 0.05, (4) 0.085, and (5) 0.10 (vlv) l-propanol; (b)l/k vs cp plot for: ( l ) 0.035, (2) 0.052, (3)0.092, (4) 0.133, and (5) 0.150 M SDS.Solid lines represent theoretical curves obtained from eq. 8.26; circlescorrespond to experimental kvalues. Reprinted from Ref. 9 with permission of Elsevier.
an
of five catecholamines were evaluated, using several models and more than one hundred experimental designs, some of them represented inFig. 8.4. The results given by the logarithmic equations are poorer. These functions systematically give larger k values[101, Equation 8 2 3 also gives a bad prediction of the retention when usedto describe the whole variable owever, this simplified equationyelds acceptable results in a small region of the variable space,as when applied in the method of triangles. The smllest errors were obtained witheqs. 8.26-8.30 [g, 1l]. It is thus evident that a term including both concentrations of s u r f a c t ~ tand modifier is needed to model the retention. The most simple equation givingacceptable results is eq. 8.26. It has been checkedthat the prediction capability of this equation is similar to the other more complex (eq.8.27-8.30), for polar or moderately nonpolar compounds, such as amino acids [121, sulfonamides [131, P-blockers [141, and diuretics [151 (C 18 column and mobile phases of S S and l-propanol (C8or C 18 or 1 -pentanol), and some benzene and naphthalene derivatives or l -butanol) columns and mobile phases ofSDS or CTAB, and 1 -propanol [g, 16). Thus, for these compounds, the plot of the reciprocal of k vs. concentration of modifier is linearat a fixed surfactant concentration. It is im~ortantto remark that the accuracy of the model has been checked with data obtained by severalauthors. Further, Garcia et al. [161 found that changes inthe concentration of surfactant and alcohol, inthe mobile phase, influencethe relative errors obtained with eq. 8.26, which were lower for CTAB with respect to S and for 1 -propanol with respectto 1-butanol. This simple model failed in taking into account someinteractions ofthe solutes,which are more porta ant in veryh~drophobicsystems and when solute-micelle ~teractions are diminished. It should be consideredthat the amount of su~actanton the column desorbed by 1 -butanol is greater than by l-propanol. Also,1butanol can compete to a greater extentthan 1 -propanol with the micelle, in the interaction withthe solutes. Eq. 8.28 (with a 'p2 term) provided a better description of the retention for highly hydrophobic polycyclic aromatic h~drocarbons(PAHs) than eq. 8.26 (Fig. 8.7). For these compounds, the plot of l & vs. cp was nonlinear. Similar results were obtained for several steroids eluted with S §-acetonitrile eluents [171.
elative error (94)
l
2
3
4
5
6
7
Ey.28
8
Eq.26
9 10 l1 12 13 14 15 16 17 18 19 20 21 22 23
~
o
~
~
~
d
.7 Relativeerrors(%)obtained in SDS-l-butan01 mobilephases,for the retention factor prediction of: ( l ) benzene, (2) benzylic alcohol, (3) benzarnide, (4) toluene, (5) benzonitrile, (6) nitrobenzene, (7) phenol, (8) 2-phenylethanol, (9) chlorobenzene, (10)phenylacetonitrile, (1 1) 3,5-dimethylphenol,(12)naphthalene, (13)l-naphthol, (14) 2-naphthol, (1 5) l -naphthyla~ine, (16) pyrene, (17) phenanthrene, (20) fluoranthene,(21)acenaphthylene,(22) (18)2,3-benzofluoreneY(19)fluorene, acenaphthene, and (23) anthracene. Adaptedfrom Ref. 16.
Table 8.1 shows that the addition of a [”jq” term (eq. 8.29) improves the accuracy of the predictions for some experi~entaldesigns. ~lthoughe~perimentaldesigns of four, five and six points, such as designs I, VI and XIV in Fig. 8.4, are enough to achieve the fitting parameters of eqs. 8.26,8.28 and 8.29, respectively, an additional ~easurementshould be used at least to check the accuracy of the fittings. Calculated k values according to eq.8.26 and design VI (Fig. 8.4) are plotted in Fig. 8.8, against experimental values, for: (i) five catecholamines (noradrenaline, adrenaline, adrenalone, dopamine, and isoprenaline) and thirteen mobile phases (0.035-0.150 SDS and 0-15% 1-propanol), (ii) fifteen phenols (4-benza~idepheno1,~ - h y d r o ~ y - b e n ~ l alcohol, 4-hydroxyphenemethyl alcohol, 4 - h y d r o ~ y b e n ~cyanide, l 4-hydrox~acetophenone,4-h~drox~ben~ldehyde, phenol, 4-~uorophenol,4-
ropiophenone9 4-methylphenol, 4-nitrophenol, 4-hydroxybenzophenone, 4-isopropylphenol, 4-hydroxydiphenylmethane, 4-tert.butylphenol) and five mobile phases (0.04-0.12 M CTAB and 0-10% 2propanol), (iii)thirteen aminoacids and small peptides (arginine, methionine9 t~ptophan,tyrosine, alanyl-tyrosine, ar~inyl-phenylalanine7aspartylphenylalanine, leucyl-t~ptophan, leucyl-tyrosine, lysyl-p~enylalanine7 phenylalanyl-phenylal~ine,glycyl-leucyl-~rosine,and glycyl-phenylalanylleucine) and five mobile phases (0.1-0.4 M SDS and 045% 2-propanol), and (iv) six aromatic compounds (anisole, benzene, naphthalene, 1naphthalenemethanol,phenol and toluene) and fifteen mobile phases (0.060.14 M SDS and l 0% 1-propanol) 193. The equations for the fitted straightlines (linear least squares) were kcalc= 0.22 + 0.98 kexp(r = 0.99 cate~~olamines, kcalc= 0.05 + 1.OOkexp(r = 0.998) for the phenols, kcalc= -0.4 l + 1.05 kexp(r = 0.9996) for the amino acids and peptides, and kcalc= 0.24 + 0.99 kexp(r = 0.996) for the diverse aromatic compounds. The proximity of the slope to unity and the low intercept revealed the absence of systematic errors.
calc
ure experimental vs. kvalues according eq.to 8.2 Fig. 8.4) for: (a) five catecholamines and thirteen mobile phases, (b) fifteen phenols and five mobile phases, (c) thirteen amino acids and peptides and five mobile phases, and (d) six aromatic compoundsand fifteen mobile phases. Reprinted from Ref. 9 with permission of Elsevier.
The prediction of retention factors for mobile phases showing the smallest elution strength, such as pure micellar solutions with a concentration of surfactant close to the cmc, is often very poor, especially for nonpolar solutes. It must be consideredthat for these mobile phases, an extrapolation is p e r f o ~ e din a region of strong change in k, where small variations in the concentrations ofsu~actantand modifierafEect highly the retention. Finally, the accuracy of the model for each component in a mixture can be variable, and the degree of tolerable variability for any component in a given separation is dependent upon how imp0 particular component isto the overall quality of the separation. III.4. ~redictiQn of the ~etentiQn and S e ~ e c t i v i ~ e isolines in Fig. 8.9 represent mobile phase compositions where the same retention is from expected the retention data mobile of phases of -10% 2-propanol, 0.08 propanol. It is obviousthat when both micelle and modifierconcentrations are increased, the respective eEects are combinedand an even shorter retention is observed. In t h s way, the retention decreases from a retention factor of 63.8 in the lower left corner of the diagram to 10.2 in the upper right corner. Although the above trend is observed for all components in a mixture, it must be stressed that the amount of reduction will not be the same, thus resulting in changes in the selectivity coefficient, a, which is defined as the ratio of the retention factors of two components, where numerator and denominator are selected such that the resulting value is larger than 1. Fig. 8.10 (top) shows the effect of varying surfactant and alcohol concentration the selectivity on phenol of and 4-hydroxybe~aldehyde:the selectivity is scarcely aEected by 2-propanol, due to a similar change in retention of both compounds. increasing the surfactant concentration, an increase in selectivity is observed due to the fact that 4-hydro~ybenzaldehyde shows retention than phenol. The final separation in 0.12 superior to the ones observed at other mobile phase CO reverse is observed when the retention behavior of 4-hydro~~ropiophenone
Retention factor of 4-nitrophenol as a function of surfactant andmodifier concentration. The lines connect points defined by equal retention factors.
Selectivity, a, of: phenol and4-hydroxybe~aldehyde(top), and 4-hydroxypropiophenone and 4-nitrophenol (bottom), as a function of surfactant and alcohol concentration. The lines connect points of identical selectivity.
and 4-nitrophenol is examined (Fig. 8.10, bottom): areduction in selectivity occurs when the surfactant concentration is increased, and the two components coelute in 0.12 M CTAB. In contrast, when the alcohol concentration is increased from 0 to 10% (v/v), the selectivity increases, since the reduction in retention of 4-hydroxypropiophenone is stronger than the change in retention of 4-nitrophenol. A much stronger change in retention of a compound in a mixture than the su~ounding Components, with varying conditions,causes ~umerous cases of coelution and peak-crossings in i n t e ~ e d i a t emobile phase compositions. Therefore, a satisfactory separation can only be found through a systematic search of the variable space.
The equations that describe the retention on hybridmicellar mobile phases were first derived on a pure empirical basis. A further concern was to find an interpretation of these equations, based on physico-chemical properties. This permitted the improvement of the descriptive models, and the evaluatio~ of the parameters of interaction between the three environments involved in MLC: stationary phase, bulk water, and micelles, according to equilibria and 8.2 [181. The coefficients in eq. 8.26 were related to several ~arameters of retention. From the reciprocal of this equation: (8.3 1)
an expression similar to the reciprocal of eq. 8.3 can be obtained:
being:
This equation can be rewritten as: k =
Kxs
(8.33)
where K,' and K,' are apparent constants with respect to the concentration of modifier. It was observed however that the convergence of the iterative process inthe nonlinear regression was more rapid andstable when the experimental data were fitted to eq.8.3 1 instead of eq. 8.32. In eq. 8.32, the constants K A D and KMDmeasure the relative variation in the concentration of solute in bulk water and micelle, respectively, in the presence of modifier taking the pure micellar solution (without modifier) as the reference: (8.34)
(8.35)
where [A] and [AM] are the concentrations of free solute in bulk water and solute associated to the micelle in a pure micellar solution, and A[A] and A[AM] are the changes in the concentrations produced by the modifier. On the other hand, eq. 8.28 can be rearranged as follows:
k =
l + KADq+ K q 2
This equation introduces a new c~nstant,K, that implies a qu with c p,whichmay sup hyperbolic variation in KTAsand excessive dependence of the retentio~with c p,and produce high err an extrapolation is made in a region of large concentrations of Therefore, an alternative model was proposed for highly hydrophobi~ solutes,which considers the additional change in the c o n c e n ~ a t ~ o n ~ ~ s o l u t associated tothe s t a t i o n a ~phase, produced by the presence of ~ o d i ~ e r [l8]:
This equation is an extension of eq. 8.32, and has been checked to significant i ~ p r o ~ e m e nint the prediction of the retention of highly hydrophobic solutes, such as pyrene. The constants K,, and K,, in eq. 8.37 account for the displacement of the water-~icelle equilibri~m,whereas KsD and Kmdescribe the modification ofthe water-stationa~ phase equilibrium(see Fig. 8. l), These changes are due tothe diminution in thepolarity of water, and the m o ~ i ~ c a t i oofn the i~teractionsof the solute with micelles and s t a t i o n a ~ phase, when a modifier is added.
. V: l,
Iterative ~ e ~ r e s s i oStrategy n
~ l t h o u g htwo variables usually suffice for samples of moderate complexity to obtain a satisfactory separation, inclusion of a third variable will often further improve the quality of the separation, with respect to both resolution and analysis time. In the first methods developed in MLC for the analysis of compounds showing an acid-base behavior, the pH was usually previously chosen and only the concentrations of surfactant and m o d i ~ e rwere optimized. The best pH for the separation was selected after examining the retention in a reduced number of mobile phases, at two or three pH values. However, for an adequate method development, this variable should be simult~eously optimized with the concentrations of surfactant and modifier, especially because the protonation constants of solutes suffer shifts, depending on the composition of the mobile phase. This is caused by the different partitioning of the acidic and basic species of solutes in the micellar pseudophase, due to electrostatic interactions. Strasters et al. [191 were again concerned with this problem, and considered the simultaneous optimization of pH, and concentration^ of s~rfactantand modifier, for the separation of a mixture of several amino acids and peptides. For this purpose, the authors usedth trian~lesadapted to three dimensions, and linear functions of surfactant and modifier. The procedure beganwith a design of located in a three-dimensional space, which was divided into (Fig. 8.1 1). Therefore, 24 different equations of retention should be fitted, The retention in other mobile phases was calculated by linear i ~ t ~ r ~ o l a t i o n inside eachtetrahedron. The range of pH values examined by the prediction of the retention was intentionally reduced to prevent deviations from linearity in the retention behavior of and peptides. This was, obviously, a great limitation of th
Figure 8.11 Initial experiments inathree-parameteriterative regression optimization, The solid dots represent points located on the visible outside of the cube, the open (3,4,7 dots and 10) are me~urementson the invisible sides of the cube, and the open square (8) is located in the centreof the cube. Oneof the subspaces, the tetrahedron (2, 5,8,9) is indicated by the dashed lines.
K.2. Global ~ o ~ e l A mathematical model was further developed [20], which extended the description made with eq. 8.32 to take intoaccount the influence of pH on retention. The effect of the modifier on the description of the retention can be considered by substitution, in eq. 8.1 l, of eqs. 8.34 and 8.35, and other similar changes for the protonated species. The following was obtained:
(8.38)
where:
may be rewritten as:
1 +K?h
(8.39)
is the apparent protonation constant of the solute, that depends where KHMq on the concentration of both surfactant and modifier and the on association capabili~ of both acid-base species the with micelle. The modifier decreases ,Mv, whereas this constant increases slightly with the concentr~tionof surfactant.
Equation 8.38 contains nine constants (KAs,Km, KHm,ICHAD, K H M D and ICH), and describes the change in the retention factors of acid-base solutes at any pH, and any concentration of surfactant and m o d i ~ e rin the mobile phase. However, the model does not consider the modification of K, in the water-modifier bulk solvent, due to the change in the concentration of modifier. A more complete model would require the
introduction ofa new constant. It was checked, however,that the inclusion of this constant doesnot improve significantly the description of the retention, sincethe large number of parameters the of model providesa high flexibility to the fitting, absorbing the deviationsproduced by this simplification. Partial fittings of the data can be madeto obtain initial values that facilitate the rapid and reliable convergence of eq. 8.38 towards the correct solution, and avoid local minima.Thus, the retention factors in four mobile phases atpH 7 couldbe used to calculate the four constants of the nonprotonated species (eq.8.32). The parameters of the protonated species could be obtained, similarly, at a sufficiently acid medium, whereas the estimation of the protonation constant will require additional mobile phases at inte~ediatepH values. Usingtheseinitial parameters, the global nonlinear fitting of the completeset of experimental data can be made more easily. The determination of the parameters of the model givenby eq. 8.38 requires experimentaldata from at least nine mobilephases, but extra data should be usedto improve the reliability of the predictions. The model was successfully applied to the prediction of the retention behavior of seven solutes (benzocaine, bumetanide, ethacrynic acid, furosemide, su~fanil~de, tyrosine and xipamide) insidethe total pH range of a C18 column, and for the concentration ranges of SDS and l-propanol, 0.05-0.15 (v/v), respectively, witherrors lower than 6% [201.
?? 3. Sepa~ationof Amino Acids The importance of p e r f o ~ i n ga simultaneous optimization of the three variables, pH, and concentrationsof micelles and modifier, isillustrated by the separation of a mixture of amino acids and small peptides (Fig. 8.12) [191. Apparently, a good separation is given at pH 2.5for 0.1 M S 4% (vh) 2-propanol. The resolution is more than adequate, but the analysis time is relatively long (ca. 40 min). The same applies to the optimum 3.5, where similar resolution and andysis time are observed and 1% 2-propanol. However, when the full variable S ace is taken into consideration, an even betterseparation is obtained. At p
0.24 M SDS and 2% 2-propanol, the optimum situation is obtained. Not only is the resolution improved (in fact, an unnecessary improvement) but, more i ~ p o ~ a nthe t , analysis time is drastically reduced to 20 min. Hence, two variables are sufficient to obtain a satisfactory separation for this mixture, but three variables give a better chromatogramwith respect to both separation and analysis time. ovvever, the overall analysis time will be a function of pH only if the degree of dissociation of the latter components in the chro~atograms change in the examined pH range.As a consequence,the pH can be usedto fine tune the selectivity, without in~uencing the elution strengthof the mobile phase.
I
m
i
Three intersectionsin the original variable space at different pH values, forSDS and 2-propanol mobile phases used for the separation ofa mixture of amino acids and small peptides. In each intersection, isoresponse lines show thebehavior ofthe resolution criterion. The predicted optimum at each pH (indicated by the dot)is displayed onthe right. The optimum at pH 3. I is the predicted global optimum ofthe three-parameter optimization. Reprinted from Ref. 19 with permission of Elsevier.
. In recent years, neural networks have been widely used for solving different chemical problems, including the field of liquid chromato~aphy,such as the modeling and prediction of the retention using structural descriptors, chromato~aphicpeak classification and deconvolution of overlapped peaks. In MLC, Xie et al. [2 l] applied multi-layer feed-forward neural networks, trained with an error back-propagation algorithm, to model the retention behavior as a function of pH, concentrations of surfactant and organic modifier, and temperature. Other contributions came later [22, 231. Accurate retention predictions were always achieved. The soft models defined by the weights of the neural networks are capable of accommodatingall types of relationships, being especially useful when the dependence of the retention behavior with the mobile phase variables is unknown. However, neural networks learn the relationships from the data themselves, and hence, more experimentalpoints are needed with respect to hard-modeling methods. The use of neural networks is, therefore, only recommended for those cases where adequate theoretical or empirical models do not exist, such as retention modeling in MLC with four variables (e.g., pH, surfactant, modifier, and temperature).
. VI1l. ~ i n e a r i ~ a t i o n o f tohqua e ti on^ of ~ e t e n t i o n In the literature, linear regression has often been applied tofit the experimental data to the model that describes the dependence ofthe retention factor with the concentration of surfactant (eq. 8.3). However, the linearization process introduces a perturbation in the fitting. The least-squares method performs an implicit weighting of the experimental points, that is proportional to the sensitivity of the signal with respect to the variable being fitted. When an equation is linearized, the fitting data are modified, and consequently, the implicit weighting is perturbed [24]. A weighting strategy should be employed to diminish the systematic errors
produced with the linear transformation, and obtain output data similar to those given by nonlinear regression. The weights are given by:
(8.40)
where f and F are the nonlinear andlinearized equations, respectively, and P is a parameter of the model 1243. Substitution in eq. 8.40 of eqs. 8.238.30, for n = 2, results in:
F = . -1 k4
(8.41)
Fitting errors, in average 25% larger, havebeen obtained with the unweighted linearized model, compared to the weighted fitting [181. The constant KM in eq. 8.3 is related to the retention in a mobile phase at the cmc. Therefore, its calculation requires an extrapolation in a region of large variation in k. When unweightedlinear fitting is performed, the errors can lead to negative intercepts in the Ilk vs. [M]plots. The application of weights corrects these errors, at least partially.
VII.2. Dead Time Measureme~t The reliable calculation of retention factorsrequires the accurate evaluation of the dead time of the chromatographic system. This is not an easy task when a micellar mobile phase is used. Due to changes in the surfactant adsorbed layer, the shape, height and sign of the p e ~ r b a t i o n appearing s at the head of the chromatogramsare unpredictable, especially when the nature of the injected solution is very differentfrom that ofthe mobile phase. In the MLC literature, the dead time is usually measured by injection of water, aqueous solutions of NaN03, NaI and KI, or organic solvents such as
methanol and acetonitrile. The criteria applied to locate the dead time are either the measurement of the position of the maximum of the first peak in the chromatograms, or the measurement ofthe time from the injection to the first deviation of the baseline. The first criterion is very simple, but the variability in the shape and positionof the first peak with mobilephases of diverse composition yields, frequently, approximate values. In contrast, the start of the first main peak is fairly reproducible [ z ] .
In general, it is questionable that the values given in the literature were real dead times, although the measured times were probably close to them. It is implicitly assumedthat the injected compound used to determine the dead time doesnot interact with the stationary phase. This is obviously notthe case for methanol or acetonitrile that disrupt the adsorbed surfactant layer. It has been observed howeverthat the time measured with the diEerent salt solutions mentioned above is variable. More frequently, only a characteristic time in the head of the chromatograms, or reference time, will be determined. This time should be at least reproducible. In fact, even the dead time measured by injection of water is not very reproducible, especially when the data are taken at the maximum of the first perturbation of the baseline. The measurement is better when performed by injection of micellar solutionsof the analytes [125]. The experimental work can provide a great number of replicates which can be usedfor this purpose. Figure 8.13 shows that whenwaterisinjected, two well differentiated regions appear at the beginning of the chromatogram. The first region is rather irregular and unpredictable, and the second region keeps its shape and position whenthe wavelength changes. Onthe other hand,the a solution with asimilar composition as the signals obtained by injection of mobile phase (blank solution) keep the position of the maximum. In this case, only the height of the signal is modified, and the second perturbation is scarcely observed. The noise increase observed with the wavelength, when water is injected,is probably associated with the state of the detector lamp and of the material on the cell windows. The first perturbation observed when water is injected has probably a refractometric origin, andthe second oneis an absorptiometricsignal. The irregular shape of the first peak obtained with water is due to the large difference in composition between the injected solutionand the mobile phase,
Figure 8.13 Influence of wavelength in the measurement of the referencetime obtained by injection of: (a) micellar solution, and (b)water. A 0.05 M SDS mobile phasewithout modifier was used, Wavelengths are given in nm. Reprinted from Ref. 25,
Influence of SDS concentration in the injectedsolution ontheshape of thesignal at the beginningofthe chromatogram,when amobilephase of 0.05 M SDS without modifier is used. Reprinted from Ref. 25. a
time, S
whichoriginates erratic fluctuations in the refractive indexwhenboth solutions are mixed. In contrast, when the blank solution is injected, the homogenization of the mixture is more simple. Figure 8.14 shows another set of injections, wherethe concentration of SDS in the injected sample is varied from a value below the concentration of the mobile phase up to a concentration above it. It may be observedthat when the compositions of bothinjectedsolutionandmobilephase are matched, the second peak decreases and even disappears, and krther, a positive signal is achieved. Ths behavior confirmsthe absorptiometric nature of this peak. The referencetimedoesnotchange appreciably with the composition of the eluent in hybrid mobile phases. Therefore, the same value can be usedto predict the retention fkctorsof solutes eluted in a given chromatographic column, with mobile phases containing variable amounts of surfactant and alcohol [18,251. Furthermore, an excellent predictionof retention timescan be obtained usingapproximate values of the reference time, sincethe retention factor is only used as an intermediate variable in the prediction of the position of the chromatographic peaks. The evaluation of physicochemical retention coefficients requires, however, the use of an accurate value of the reference time. Otherwise, ill-conditionedfittings will give negative coefficients.
VH.3. Critical Micel~a~ C~ncentrati~n Another important factor that should be considered, when ph~sicochemic retention coefficientsare calculated fromchromatographic data in MLC,is the correct subtractionof the cmc from the total concentration of surfactant. The cmc of SDS solutions decreasesat low concentrationsof alcohols used as modifiers, except for methanol, and increases for acetonitrile and tetrahydro~ran. m e n the cmc is not subtracted from. the total on cent ration of surfactant, the errors in the calculation of the retention coefficients can be larger than 100%. However, as occurs with the dead time, for many solutes there is no significant difference between the use of total or micellar concentrationof surfactant in the prediction of the retention behavior (retentionfactors) [1 1,181.
an
ractions that the solutes experience in
a cellar
i~nific~t
v a rthat i~~les
of resolutionof complex mi~turesof ionic and high accuracy in the prediction of retention its the reliable and relatively rapid optimization of the mobilephase for the separation ofamixtureof retive method and a reduced number of mobile iable, four or fivefor two variables, andnine for three variables). The optimization of the resolution of a mixture of compounds comprises several steps,as the following:
(i) ~ c ~ ~ v e mofe the n t equations of retentionfor each solute. earch ofthe o p t ~ u m mobile phase withthe aid of maps of global resolutionfor the m i ~ ~ofr esolutes. (iii) Simulation of the c h r o m a t o ~ for r ~ the optimum mobile phase. (iv) Search of a new maximum, when the selected optimum is not satisfactory.
The four criteria of resolution given below based on the normalized product, r, of different properties,Xi,i+l,associated to pairs of consecutiv~ peaks, have been applied inMLC [g, l1,261:
VI0
r =
n-l
77
./Y ,,it 1
i=l
(8.42)
The individual properties studied were the selectivities, separation factors, valley-to-peak ratios and overlapped fractions. The two first criteria only consider the position of the chromatogra~hicpeaks, and the latter two their position and shape. The three latter functions may vary from 0 to l . The combined function of resolution, r, is maximized to obtain the optimum mobile phase, and its proximity to unity indicates the quality of the separation. Since the product of all observed resolutions is used, coelution will effectively cause the criterion to dropto zero. Extremely long chromatograms with a number of unnecessarily large resolution values will also be represented by low criterion values.
VIII l. Strute~ies Stric~ly Bused Solute on ~ e t e n t i o n The most simple criteria used inchro~atographico p t i ~ i z a t i oare ~ based on properties that only depend on the retention of solutes [Z l], such as the modified selectivity:
(8.43)
and the separation factor:
(8.44)
where ai,i+l is the selectivity, and ti, ti+l, and ki, ki+l,are the retention times and retention factors of consecutive peaks.
The positional criteria can lead to reliable resolution optima using the retention data from a few mobile phases. However, as the shape and width of the chro~atographicpeaks are not considered, for largely overlappe~asymmetric peaks, an unacceptable optimum can be obtained. If this is the case, only an increase in plate count will provide the desired separation, using for instance two identical columns in series [8]. ~lternatively,a response surface related to a different criterion must be examined.
VI112. ~trategiest ~Take ~ into t Account the Position a~~ Peak ~ ~ a ~ e The major drawback of practical separations applying MLC is still the low c~omatographicefficiency, caused by the resistance to mass transfer in the processes involving micelles and a sur~actant-modi~ed stationa~ phase as osed in Chapter 6. This is especially impo~antsince the increase in ~ i c e l l e ~ o n ~ e n t rcauses a t i o na decrease in plate count, resulting in a varying efficiency over the variable space. It is thus ~ o ~ h w h itol eexamine the inclusion of the expected peak shape in the expression of the chromatographic quality. The infor~ationgiven bythe positional-s~apecriteria is interesting, not only when the chromatographic peaks are asymmetric or have low efficiencies, but also with sy~metricpeaks which are very close to each other. Poorly defined optima obtained with a positional criterion often become clearer when the shape of the peaks is considered. Two positional-shape criteria have been developed [1l],the valleyto-peak ratio: (8.45)
and the overlapped fractions:
oi=l--
W;
(8.46)
In these equations (see Fig. 8.15), h, is the height of the valley betweentwo adjacent peaks, h2 the interpolated height between the maxima of two adjacent peaks measured at the abscissa of the valley, W; is the total area of a givenpeak, and wfithe area of this peak overlapped by other peaks. The overlapped fractions extends the global function of resolution. to all individual peaks in the chromatogram, n- 1should thus be substituted for n in eq. 8.42.
~hromatographicpeak properties. Reprintedfrom Ref.ll with permissionof Elsevier.
an
VU13. P ~ e ~ i ~oft Peak i o ~Shape The models used to predict peak shape, based on Gaussian distributions, have the advantage of using very intuitive parameters, related to properties which can directly be measured onthe chromatograms (position and height of the maxima, and width of the peaks). The equation ~escribinga pure Gaussian peak is:
(8.47)
where t is the time, H@and t, are the height and time at the peak maximum, the standard deviation. However, in MLC, skewed peaks with low and CT, efficiencies are often obtained, and in such cases, the assumption of a Gaussian model mayyield large errors. A convenient modification of the Gaussian model is the substitution of the standard deviation in a pure Gaussian peak, by a polynomial function (polynomial modified Gaussian model, PMG), which varieswith the distance to the maximum of the peak [27]:
(8.48) L
With this approach, accurate descriptions of peaks showing large asymmetries can be obtained, including those showing deformation either to the right or to the left. Also, eq. 8.48 with a linear or a parabolic function can be applied to refine peak parameters, such as the efficiency and asymmetry factor, estimated by direct measurementof the chro~atographic signal. The method of Powell can be used to fit theexperimental data to the nonlinear functions [28]. The coefficient soin eq. 8.48 coincides with the standard deviation of a symmetric Gaussian peak describing the central region of the peak, and
1
remains almost constant when new terms are added to the polynomial i ~ disto~ion. function, whereas s1and s2are coefficients that ~ u a n t peak The coefficient sl only depends on the asymmetry factor;a large value of this coefficient indicates a strong asymmetry, and a negative value, a left bias; s2and upper terms in polynomiaof higher degree correct small deviations to better fit the shape of the peaks. However, the use of a large number of S coefficients lessens the practical application of the model.
A linear standard deviation function in eq. 8.48 approximates satisfactorily the real shape of the peaks in a chromatogram, but the accuracy ofthe fitting can be easily improved by increasingthe degree ofthe polynomial. The model with a linear hnction permitted, however, the development of a simple method to simulate chromatograms andto optimize the resolution of mixtures of compounds, using criteria that consider not only the position of the chromatographic peaks, butalso their shape 1271. With the linear function, the number of chromatographic peak parameters (i.e., position, height, efficiency, and asymmetry factor)coincides withthe number of coefficients in the h(t) function (Ho,t,, soand S , ) . The coefficients soand s1may be easily obtained from the asymmetry factor and efficiency. The steps to follow in the calculation of these coefficients, for the simulation of a peak in a chromatogram, are the following:
(i) The retention time of a peak in a given mobile phase is predicted from the equations that describe the retention. (ii) The efficiency, N, and asymmetry factor, B/A,are estimated by interpolation after fitting to a plane the values of these parameters for experimental mobile phases close to the predicted mobile phase. (iii) The width of the peak at 10% of peak height is calculated from:
?To., = A +
B
=
41.7 ti
N
(1.25 + B / A )
(8.49)
28
and the individual values of A and B are obtained from BIA:
A =
B-
1
6 . 1
+
WO.1 l +-
(iv)
BIA
1 BIA
and
(8.50)
(8.51)
The height of the normalized peak can be determined by assuming a triangular profile: H = EIWo.l,being E a normalization constant. If a higher precision is desired, other functions can be used. In the simulation of a real c h r o m a t o ~ rthe ~ , normalization step can be obviated.
(v) Finally, the coefficients so and si of the standard deviation of the skewed peak are calculated from eq. 8.48, for t = tR-A, and t = tR + B, making h(t) = 0.1 H. , After solving the system of two equations, the following is obtained:
(8.52)
so =
B( 1 " S 1 /m)
J%iG
a
0.466B( 1 - 2 . 1 4 6 ~ ~ )
(8.53)
83
VIII 4. ~ ~ n t o Maps u r of ~ e s o l u ~ i o n Different regions ofthe variable space will often associated be with different critical peak-pairs. The resolution of a multicomponent mixture thus requires an analysis involving all components in the whole variable space. allow the evaluatio~ Inspection of the contour maps of global resolution will of the robustness of the optimum. Figure 8.16 shows the contour maps for the separation of a set of fifteen phenols (the same cited previously in this chapter), with mobile phases of CTAB and 2-propanol, where an efficiency N = 2500 was considered for all solutes. For the positional criterion (separation factor), the optimum was found for a mobile phase of 0.12 M CTAB-10% 2-propanol (Fig. 8.16a, upper corner of the variable space), whereas for the valley-to-peak criterion it was 0.102 M CTAB-10% 2propanol (Fig. 8.16b), and for the overlapped fractions, 0.107 M CTAB10% 2-propanol (Fig. 8.16~). The simulated chromatograms for 10% 2-propanol and three different concentrations of surfactant are giveninFig.8.17. The disagreement amongthe positional and positional-shape criteria is due to the retention behavior of peaks 13-15. As the concentration of surfactant decreased from 0.12 to 0.10 M (Fig. 8.17a and 8.17b), the valley-to-peak ratio improved, and the overlapped fractions decreased to a lesser extent. The positional resolution however became worse. A further reduction in the concentration of CTAB decreased both the positional and positional-shape resolution (see peaks 9-10 and 13-15). The combined use of the diverse optimization criteria may give complementary information for selecting the optimum mobile phase. The positional criterion gives a rough appro~imationof the region where the peaks will be separated, but it does not evaluate accurately the quality of the separation. The valley-to-peak ratio indicates the region where the peaks will be apparent. Finally,the overlapped fractions show the region where the peaks will be better quantified, because a larger surface of each peak will be exposed. As the application of the positional-shape criteria requires a good prediction of the position and shape of the chromatographic peaks, they are in principle more susceptible to errors. However, these criteria are always preferable, even when asymmetric peaks are assumed to be symmetric. The most satisfactoryoptima should be observed in the contour maps of the three
.l6 Contour mapsofglobal resolution forthe mixtureof fifteen phenols eluted with CTAB-2-propanol mobile phases, according to different criteria:(a) separation factor,(b)valleyfrom Ref.11 with pe~issionof~lsevier, to-peak ratio, and (c) overlapped fractions. Reprinted
II
1
F i ~ u 8.1 r ~7 C h r o m a t o g r ~of s a mixture of phenols in mobile phases containing 10% 2propanol and diverse CTAB molar concentrations: (a) 0.12, (b) 0.10 and (c) 0.08. Peaks: (1) 4-be~am~depheno~, (2) 4-hydroxy~nzy~ alcohol, (3) 4-hydroxyphenemethyl alcohol, (4) 4-hydroxybenzyl cyanide, (5)4-hydroxyacetophenone, (6) 4-hydroxybenza~dehyde, (7) phenol, ( 8 ) 4-fluorophenol, (9) 4-hydroxypropiophenone, (10) 4-methylphenol, (1l) 4-nitrophenol, (12) 4-hydroxybenzophenone, (13) 4-isopropylphenol,(14) 4-hydroxydiphenylmethane, and (15) 4-tert.-butylphenol. Reprinted from Ref. 11 with permission of Elsevier. criteria; the global resolution for the valley-to-peak and overlapped fractions should both be high. Other additional factors should beconsidered in the selection of a mobile phase. Thus, an optimum found in a region of large variation in the
global resolution function will not be adequate in practice, as the errorsin the prediction of the retention and in the preparation of the mobile phases may lead to results differentfrom those expected. For complex response surfaces showing several maxima and minima, additional experimental mobile phases should be prepared in the region where the optimum appears. In other cases, the optimum will correspond to a high experiment duration, and mobile phases giving shorter retention times will be more acceptable.
Using the above strategy, the profile of a chromatogram can bepredicted on the basis of a limited number ofexperiments, even thoughthese experiments were relatively far apart in the variable space. An example of this prediction capability is the separation of a mixture oftwo diuretics (chlo~halidoneand a~iloride)and five steroids (boldenone, testosterone, methyltestosterone, medroxyprogesterone acetate and dydrogesterone) with mobile phases of SDS and acetonitrile. The six experimental mobile phases used for the prediction were 0.075 M SDS, 0.20 M SDS, 0.075 M SDS-lO% acetonitrile, 0.14 M SDS-10% acetonitrile, 0.075 M SDS-20% acetonitrile, and 0.20 h4 SDS-ZO% acetonitrile [171. The position of the maximum of the peaks was calculated using eq. 8.37, and the efficienciesand a s y m m e t ~ factors were interpolated from thevalues obtained with thethree experimental mobile phases closer to thepredicted mobile phase fromthose available. The overlapped fractions were used to optimize the resolution. Fig. 8.18 shows the chromatograms of the mixture of diuretics and steroids, eluted with mobile phases of 0.14 M SDS (Fig. 8.1 sa) and 0.18 M SDS1’7% acetonitrile (Fig. 8.18b). As observed, predicted and experimental chromatograms, located in regions far away fromeach other in the variable space, are in good agreement. Another example of the good performance of the models shown in this chapter is given in Fig. 8.19, which shows the predicted and experimental chromatogram for a mixture of four diuretics, three P-blockers and a vasodilator, eluted with a mobile phase of 0.15 M SDS and 7% 1propanol at pH 3 [14].
7 2
0
time, min
time, min Predicted (a,c) and experimental (b,d) chromatograms for the separationof a mixture of diuretics and steroids, eluted with mobile phases of 0.14 h4 SDS (a,b) and 0.18M17% acetonitrile(c,d). The prediction of the position ofthe peaks was made with eq. 8.37, and the shape of each peak with eqs. 8.49-8.53. The area of the peaks was not normalized. (2) ~ i ~ o r i d e(3) , boldenone, (4)testosterone, (5) Compounds: (1) ch~o~halidone, methyltestosterone,(6)medroxyprogesteroneacetate,and (7) dydrogesterone. Reprinted from Ref. 27 with permissionof the American Chemical Society.
HT
AT
3
.l9 Predicted (a) and experimental (b) chromatograms of amixture of compounds eluted with a 0.15 M SDS"7% l-propanol mobile phase, buffered at pH 3.0. Compounds: hydrochlorothiazide (HT),chlo~halidon~ (CL), bendro~umethiazide(BE), atenolol (AT), amiloride (AM), hydralazine (HY), metoprolol (ME), and oxprenolol (OX). Reprinted fromRef of the Royal Society of Chemistry. 14 with permission
The use of the optimization procedure described in this chapter can be assisted by t h e ~ I C software. ~ R ~ ~Torres Lapasio was interested bythe general t r e a ~ e nof t chromatographic data using personal computers [g- 11, wrote a suite of programs which he called ~ I C (from ~ R icellar Chromatography). T h e ~ C ~ R ~ ~ s o f tincludes w a r e the modeling of the retention behavior and shape properties ofchromatographc peaks, and the optimization ofthe resolution ofmixtures. The first release o f M I C ~ R ~ ~ developed was to take stages of the analytical process [29]. It allows the determi times, smoothing of chromatograms7 measurement of peak parameters, modeling of skewed peaks and deconvolution of overlappedpeaks. It also includes several tools for the experimental design, optimi~ationof mobile phase composition (concentrationsof surfactant and modifierat afixed p management of sets of data, optimization and regression analysis, and simulation of chromatograms. The user can interact with the software at several levels, from a semiautomatic to a fully manual mode. In the manual mode,the progressive of predicted chromatogramscan changes with mobile phase composition the be graphically observed.utines for the graphical representation of chromatograms, resolution surfaces and contour maps are implemented to better displaythe available information. All thise n v i r o ~ e nallows t a fast, reliable and easy resolution of chromatographic problemsof diverse n a ~ r e and complexity. ~ I C ~ R ~been ~ hused a sinour laboratories since 1994. Several analytical procedures,commentedin Chapters 10 to 12, were developed withthe aid of this software. This book is accompanied with abeta version of the second release o ~ ~ I C whch ~ Rcontains ~ ~ only ? two modules dedxated tothe modeling of the retention and further optimization of mobile phase composition. to ~ C ~ (release R ~ 2.0)Mincludes important improvements with respect the first release. Problems involving three factors (concentrations of modifier, andpr-r)are considered. Advances in modeling
~
optimization are implemented, and the graphicalintedace is more flexible. will reveal thegreat power ofMLC in the prediction The use o f ~ C H R O M of chromatographic data. It is strongly recommended to read Appendix I before using thesoftware.
1.
W. Amstrong and F. Nome, Partitioning Behavior of Solutes uted with ~ i c e l l a r ~ o bPhases i l e in LC, Anal. Chem., 53: 1662 (198 1).
2. L.R. Snyder, in C. Horvath (editor), HPLC.AdvancesandPerspectives, Academic Press, New York, Vol. 1, Ch. 4, 1980. 3. L.S. Madamba-Tan, J.K. Strasters and M.G. Khaledi, Gradient ~ ~ u t i in o nMIC. I. Micelle Concentration Gradient, J Chromatogr. A, 683: 321 (1994). 4.
. Arunyanartand L.J. Cline-Love, I n ~ ~ e n cofe ~ i c e l l e on s Partitioning ~quilibriaof Ionizable Species in LC: pHand Ionic St~engthEflects, Anal. Chem., 57: 2837 (1985).
5. A H . ~ o d g e r s , J.K. Strasters and M.G. Khaledi, ~imultaneous MLC, O~timization of pH and ~ i c e l l e Concentra~on in J: Chromatogr., 636: 203 (1993). 6.
. Rodgers and M.G. Khaledi, In~uenceof pH on Retention and ectivity in MLC: C o n s e ~ u e ~ ~of e ~s i c e l l a r - I n d ~ c e d S h o~f t s Ionization Constants,Anal. Chem., 66: 327 (I 994).
7.
J.K. Strasters, A.H. Rodgersand E ltaneous ~nhancementof Separation Selectivi~a ~ t r e n g in t ~@LC using Micelles in Hy~ru-Or~anic Solvents, Anal. Chem., 62: 130 (1990).
8.
. Strasters, E.D.Breyer, A.H. Rodgersand M.G. Khaledi, Simultaneous ~ptimizationof Variables i n ~ u e n c i nSelectivity ~ and ~ ~ u t i Stren~th on in MLC, J.Chromato~r.,51l :l 7 (1990).
. Khaledi,
1
9. J.R. Torres Lapasio, R.M. Villanueva Camailas, J.M. Sanchis Mallols, M.J. M e ~ n a H e r n ~ dand e z M.C. Garcia Alvarez-Coque,~odeling of the ~etentionBehavior o f Solutes in MI;C with OrganicM o d ~ e r s , J: Chromatogr., 639: 87 (1993). 10. J .R. Torres Lapasio, M.J. MedinaHernrindez, R.M. Villanueva Caxnailasand M C . Garcia Alvarez-Coque, Description of the ~etentionBehavior of Solutes in MLC with Organic ~ o d l ~ e r s : Comparison of two ~ e t h o d sChromatographia, , 40: 279 (1995). 11. J.R. Torres Lapasio, R.M. Villanueva Camailas, J.M. Sanchis Mallols, M. J. Medina Hernindez and M.C. Garcia Alvarez-Coque, ~nterpretive st rate^ for ~ptimizationof~urfactantand AlcoholConcentra~onin MLC, J: Chromato~r. A,677: 239 (1994). 12. M J. Medina Hernrindez, M. Catala Icardo and M.C. Garcia AlvarezCoque, Correlation between ~ y d r o p h o b i c iof~ Amino Acids and ~etentionData in W L C ~ i t h ~ i cEluents, e l ~ a rChromato~rap~ia, 41 : 455 (1995). 13. ELF. Simo Alfonso,G, Ramis Raxnos, M.C. Garcia Alvarez-Coque and J.S.Esteve Romero,Determination of Su~onamidesin ~ u m a nUrine by Azo Dye Precolumn Derivatization andMLC, J: Chro~atogr.B, 670: 183 (1995). ado Martinez, M.C. Garcia Alvarez-Coque andR.M. V i l l ~ u e v a Camail~s,~erformance o~Micellar ~obilP e ~ a s in ~ sRPLC for the Analysis of ~harmaceuticals containing ~ B l o c ~ e rand s ot~er Antihypertensive Drugs,Analyst, 121: 1677 (1995). 15.
E.Bonet Dorningo, J, Torres Lapasio, M.J. Medina Hernhndez and ,C. Garcia Alvarez-Coque, Chromatographic ~onitoring of iuretics in Urine~ a ~ p lus e s a Sodium DodecylSu~ate-Propanol ~ i c e l l a r ~ l u eAnal, n t ) Chim. Act@,287: 120 1(1994).
16. M.A. Garcia, 0,Jimhez and M.L. Marina, Comparisonof t ~ e ~ o d e l s scribing the~ e t e n t i ~ innMLC with ~ ~ b~ l ur e ni tfso~r a Group of nzene ~ ~ r i v a t i v eand s P o ~ y c ~ ~ Aromatic lic ~ydrocarbons, J: C~romatogr. A,675: 1 (1994). Villanueva Camailas and M C Garcia Alvare ~ ~ t i ~ i z ~o ftStero~ds i o n with a ~ i c e l ~ a r ~ o b i Phase of ~ o ~ D i ~ ud ~~ cate y ~ Containing A~etonitrile,Anal. Chim. Acta, 333: 31 (1996).
-
2
18.
arcia Alvarez-Coque, J.R. Tor Baeza, Description of the Partitioning~ e h ~ i ofo Solutes r and Data Treatment inMLC withMod~ers) Anal. Chim. Acta, 324: 163 (1996).
19.
trasters, S.T. Kim and M.G. ale&, Multiparam~ter tions inMLC using the Iterati gression ~ p t i ~ i z a ~ o n ,586: 22 1 (1991). Strategy)J. Chro~atogr.
M.C. Garcia Alvarez20. J.R. Torres Lapasio, J.J. Baezaezaand ~ as a ~ u n c Coque, Description of the ~etenti e h ~ i ino MLC of pH) Su~actant and M o d ~ e rConcentration)J: Chromat~gr. A, 769: 155 (1997). 21.
ie, J. J.Baeza Baeza, J. orres Lapasio, M.C. Garcia Mvarezand G. Rarnis Rarnos, ling and re diction of Retention in HPLC by using Neural N e ~ o r ~ ~ ) C h r o ~ a ~41o:g435 ~ a(1995). p~ia,
22.
Garcia and M.L. arina, Neural N e ~ o r ~ ntion Modeling in MLC with H y b r i ~ ~ l u e n t s , J. Liq. Chro~utogr.& Rel. Technol., 20: 73 1 (1997).
23.
.Jhenez, I. Benito andM.L. arina, Ne~ral ~ e ~ oasra Tool ~ s for Modeling the Ret~ntion ~ehavior of Dihydropyridines in MLC, Anal. Chim. Acta, 353: 367 (1997).
24.
aeza and G. Ram S, eduction ofthe R ~ l u ~ Stan~ard ve S ~itting of linearize^ ~ ~ u a t i o by ns ~eviationin the Least using S e n s i ~ i vWeights, i~ Anal. Chim. Acta, 31 6: 173 (1995).
25. J.R. Torres Lapasio, J. J. Baeza Baeza and Coque, On the Measure~entof Dead Ti Chromatogr. & Rel. Technol., 19: 1205 (1996).
Separation of Ionic and Nonionic 26. T. Okada, Simultaneous Co~poundsusing Reversed-PhaseAdLC, Anal. Science, 9: 5 9 (1993). 27 J.R. Torres Lapasio, J.J. aezaBaeza and Coque; A Model f o r the D cription) S i ~ u l a of Skewe~ Chromatographic Peaks,Anal. Chem., 69: 3822 (1997). *
28. 29.
0, Opti~ization:Theory and Applications, Wiley, New
R. Torres Lapasio, M.C. Garcia Alvarez-Coque an aeza, Global Treatmentof Chro~atographicData with Anal. Chim. Acta, 348: 187 (1997).
aeza
~
~
I, l. ~ o ~ a rofi~tc ~ t a ~n o ~ - ~ aarti t e rit ion ~ o e ~ i c i eas n ta ~ e a s u r eof ~ ~ ~ r o p ~ o ~ i c i ~ ydrophobicity is commonlyunderstood as ameasureof the relative tendencyofasolute to preferanonaqueousrather than an aqueous environment, or as a measure of the tendency of two (or more) solute molecules to aggregate in aqueous solution. The hydrophobicity of solutes is arelativepropertyanddependsmostlyontheenvironment.en comparingthebehaviorofvarioussolutesinthesame e n v i r o ~ e n t ,a quantitative scalecan be establishedto demonstrate the abilities of indwidual solutes to participate in hydrophobic interactions. The partition coefficient in the biphasic solvent system l-octanolP ,was proposed as ameasureofhydrophobicityoforganic water, , [l]. Since that time it has become compounds in the early 1960s by Hansch the standard for measuring hydrophobicity. Large compilations exist of log , P data. The log , P hydrophobicityscale has the advantageof its universality, continuity and additive nature. On the basis of this additive rule, in orderto estimate the partition coeflicient values for new compounds, Hansch and Leo [2] derived constants for diEerent hnctional groups. However, despite numerouseEorts made by many researchers using P is still problematic. a variety of techniques, the measurement of , conventional shake-flask method has several disadvantages such as being tedious and time-consuming. Also, the solute must be inherently pure and Plogin excess ofabout 6 are available in reasonable quantities. Values of, very difficultto measure, dueto the very low concentration of solute the in aqueous phase. 3
2 any attempts have been made to determine P,, by other means different to the classical shake-flask method, which include counte chro~atography[3], and ~eversed-PhaseLiquid Chromatography with C8, C 18, l "octanolcoated C 18 andphenyl stationa es [4]. The P,, establis~entof a correlationbetweenretention data inandlog assumes that theextentofchromatographicretentionreflects the hydrophobicityof a solute. This approach is known as quantitative ~[S]. C~omatographic ~ ~ )techniques structure-retention relationships( oEer a number of advantages over the static method. A great amount of relativelypreciseandreproducible data can be readilined,and the a very small dete~inationsare rapidandeasy to beautomated. amount of sample is required, a wide dynamic range may exist and the impurities present inthe sample can be s i ~ u l ~ e o u sseparated. ly All conditions can be kept constant in a chromatograp~cprocess. Solute structure becomes thus the single most independent variable in the system. ever, each pa~itioningchromatographic system yields an individual scale of hydrophobi~i~. The question arises whether difEerent chromatograp~chydrophobicity parameters should be used for predictive purposes or whether a chromato~raphicsystem should be developed that mimics the log P,, hydrophobicity scale. In eithercase, the chromato~raphic measures of hydrophobicity should be defined and reproducible. MicellarLiquid C ~ o ~ a t o g r a p h(MLC) y has shown to be an LC techmque to measurethe hydrophobicity of solutes. Diverse contributions of several authors in the fields of the relationships of carbon atoms between the retention with micellar eluents and the number in homologous series, several descriptors of hydrophobici~,such as log P,,, and the biological activity of compounds, are reviewed in this chapter.
I. 2. Solute ~nteractions in pa cellar ~ ~ u e n t s icelles are dynamic structures wherearapid exchange of surfactant ~ o n o m e r stakes place with bulk solution, with other micelles, and with surfactant molecules adsorbed on any solid surface. Compounds solubilized by these systems participate in similar equilibria; they exist in dynamic equilib~umwith bulk solvent,micellarpseudo-phase,andany suIface
TITAT1
LC,the solutes partition fi-om bulk solvent into micelles into the stationary phase. the s~ilaritiesand Khalediet al. [6, 71 wereconcernedwith differences in retention behavior between the mode of LC which employs micellareluentsand that withaqueou ic solvents.These t e c ~ q u e s share the basic components of an system, that is, nonpolar a stationary phase and a polar aqueous mobile phase. The hydrophobicity of solutes should play an important role in governing the retention in both systems, which is easilycontrolledbyadjustingsolute-mobilephase interactions. However, the differences in interaction mechanismcan cause significant differencesin retention behavior. A~ueous-org~ solvents c are homogenous, while micellar mediaare ic with microscopically no~omogeneous. Micelles are ~ p ~ p h i laggregates anisotropic microenviro~ents that provide both hydrophobic and electrostatic sites of interaction with solutes. Three sites of solubilization can beidentifiedinthemicelles:thecore(hydrophobic), the surface (hydrop~lic)and the palisade layer (the region between the head group an the core). Solutes are solubilized in the micellar phase depending on the nature of solutes and micelles. Hydrophobic neutral solutes enter into the core of the micelle, relatively polar solutes are inserted in the palisade between surfactant molecules, and highly polar solutescan remain outside the micelle, adsorbed onits surface through electrostatic interactions. soluteswillbe attracted to the surfactant ionic groups inmicelles, or repelled, dependmg on their charge. The interactions of solutes with micelles can occur through three different mechanisms: pseudo-phase extraction (partitioning), solute-surfactant coassemby (comicellization), and surface adsorption. These interactions create a unique situation in which solutes occupying variouslocations idon micellemayexperiencedifferent microenviro~entpolarities in a given micellar mobilephase. The characteristics of a~yl-bondedstationary phases in micellarand aqueous-orga~cmobilephases are anothernoteworthy difference. The extraction of organic solvent by the grafted alkyl phase depends on the composition of aqueous-organic mobile phase, which has a profoundeEectonretention.micellareluents,however,alkyl-bonded phases may be modified with an appro~imatelyconstant concentration of
monomers of surliactant which is approximately equalto the critical micellar concentration (cmc). As a result, the structure of the stationary phase are independent of mobile phase composition. In other words, solutes experience a stationary phase with unchangedcharacte~stics at different compositions of the micellar mobile phase. On the other hand, the adsorption surfactant of monomers onthe stationary phase reduces the silanophilic interactions and increases the hydrophobicity of thestationary phase.
11.1. ~orre~ations between ~etentionFactors and ~ a r ~ o n The chromatographic behavior of alkyl homologous series is usefid for the investigation of bothmobile and stationary phasecontributions to the retention mechanism in RPLC, and for the calibration of retention insuch systems. The regular linear increase of retention due to the addition of a methylene group is recognizedas a measure of hydrophobic interaction in a LC system. Also, the existence of such a linear relationshp makes the retention studyofahomologousseries particularly attractive for comparative purposes. The chromatography of these compounds is quite interesting and provides valuable ~ f o ~ a t i about o n the differences between micellar and aqueous-organic mobilephases. a) Log k vs. nc or k vs. nc ~ ~ r r ~ l a t ~ o n s
With aqueous-organic mobile phases,the logarithm ofthe retention factor, k, is linearly related to the number of carbon atoms or repeat units in the normal chain of the homologous series,n, ,in the following form:
although some deviations from linearity have been noted [ 8 ] . The slope, as log a(CHz), is a measure of methylene or hydrophobic selectivity, defined
~ ~
IC1
7
the ratio of the retention factors of two solutes differingfrom one another by a methylene group:
while the ~tercept,log p, reflects the specific interactions between the ~ n c t i ogroup ~ l of the molecules, and mobile and stationary phases. The presence of micelles in the aqueous mobile phase a profound eEect on the chromatographic characteristics of di et al.[6, 71and ~orgerding etal. [g] reported that, in thes separations, it is usually k (and not log k) which is linearly related to the number ofcarbons:
where a and b are fitting coefficients without any physical m e ~ i n g .The plot of logk vs. % for these systems hasa clear curvature, which indicates )c h g e s with the number of carbons of the homologous c o m p o ~ d s .A.quadratic equation would providea better co~elation:
The relationship between log aqueous-organic [S], however been e ~ ~ l a i n ethrough d a ion qu coefficient of the second degreeterm is small enoughto neglect linear correlation is thus observed. A.pparently, this is not the micellar and hybrid mobilephases (Fig. 9. l).
An estimation of how well the data points fit a straight-line is often made by e~aminationof the correlation coefficient (r or r2);this statistic is however easily ~ s ~ t e ~ r ebecause t e d nonlinear data in character may give
a relatively high r2 value. h order to confirm how the regression explains the variability of the data, the Fisher coeflicient (F) can be used. Large F values ensure good agreement between data and the linear models.
"E 1.8
1 I
1
1.6
1.s 1.4
1.3
Linear and quadratic fits for k and log k vs. number of carbon atoms for nalkylbenzenes. Symbols:experimental points (a), seconddegree polynomial regressionpoints (x), linear regression points (--). Mobile phases:0.2 M SDS (a, b),and 2-propanol-water3555 (v/v)(c, d). Reprinted from Ref. 6 with permission of the American Chemical Society.
A linear behavior between k andn, was found for n-alkylbe~enes eluted with pure rnicellar eluents of the anionic sodium dodecyl sulfate ), cationic c e ~ l t r i m e ~ y l ~ o n ibromide u m (CTAB), and nonionic ~o~y~oxyethylene(l0 or 23)ldodecanol (Brij8 22 or Brij8 39, and with hybrid eluentsconta~ingSDS or CTAB and 2-propanol. The homologues of n-alkylphenones eluted with pure and hybrid mobile phases of CTAB showed the same behavior,but linear relationships were observed between SDS mobile phases, the same log k andn, for these compounds eluted with type of correlation found in aqueous-organic systems. It seems that the
23
correlation between k and n, dependsonthehomologousseriesand surfactant type. Table 9.1 MethyleneSelectivitybetweenSuccessive Members of Homologous Series, Eluted with Micellar MobilePhases in a C8Column [7]
Alkylbenzenes
0.06 M 0.10 M 0.20 M 0.07 M SDS SDS SDS CTN3
n-Hexylbenzenea
1.29
1.14
0.15 M
0.18 M
1.11
1.11
1.11
1.12
n-Pentylbenzene
1.20 1.13
1.16
n-Butylbenzene
1.27
1.20
1.13
1.16
1.13
n-Propylbenzene
1.39
1.31
1.24
1.21
1.18
1.18
Ethylbenzene
1.47
1.36
1.26
1.26
1.19
1.20
1.53 Tolueneb 1.38
1.55
1.76
Akylphenones n-Hexaphenone
1.40
1.37
1.36
1.27
1.23
1.23
n-Valerophenone
1.46
1.41
1.39
1.31
1.27
1.26
n-Butyrophenone
1.56
1.49
1.43
1.41
1.32
1.31
n-Propiophenone
1.59
1.50
1.41
1.62
1.45
1.42
It shouldbenoted that the nonlogarit~icbehavior(eq. 9.3) of alkylbenzenes and alkylphenones did not change upon addition ofup to 20% of 2-propanol ['7J This was interpreted as follows: The addition of 2propanol reduces solute partitioning into micelles. The stationary phase resemblesmore an aqueous-organic(water-propanol)systemsince the 4). adsorbed sufiactant layer is reduced by propanol additions (see Chapter
owever, that are still the solute-micelle interactions which aplay major role in controlling the retention behavior in propanol-surfactant systems.
Methylene Selectivity between SuccessiveMembers of Homologous Series, Eluted with Aqueous-Organic MobilePhases in a C8 Column [7]"
50% 60% 70%
90%
Allsylbenzenes MeOH ACN MeOH
a
c
35%
50%
2-PflH 2-PrOH
n-Pentylbenzeneb
1.56
1.34
1.77
1.16
1.55
1.30
n43utylbenzene
1.53
1.31
1.75
1.19
1.58
1.29
n-Propylbenzene
1.55
1.30
1.72
1.13
1.66
1.31
Ethylben~ene
1.48
1.27
1.61
1.12
1.61
1.28
Toluenec
1.47
1.25
1.62
1.13
1.67
1.31
n-Hexaphenone
1.54
1.31
1.73
1.13
1.68
1.33
n-Valerophenone
1.51
1.27
1.69
1.13
1.67
1.31
n-~utyrophenone
1.50
1.26
1.59
1.08
1.65
1.32
n-Propiophenone
1.56
1.22
1.56
1.12
1.68
1.30
ACN = acetonitrile; MeOH = methanol; PrOH = propanol; v/vconcentrations are given. )= kpentylbenzene kbutylbenzene. ktoluene /kbenzene*
The results commented above show that,in LC,usually methylene selectivitydecreases as the carbon numberincreases(Table 9.1). The is, for instance, methylene selectivity between n-pentyl- n-bu~lbenzenes and The observed variationsare smaller than between ethylbenzene and toluene. quitesmallandincertaincasestheir statistical significancemightbe questionable. However, the decrease in @(CH2)values with an increase in the molecular size is observed in almost all cases. For aqueous-organic
mobile phases, variations a(CH,) in measured fordiffierent successive pairs a regular trend(Table are also observed, but these are rather random without 9.2). The variations in selectivity along an homologous series impliesthat a larger number of compoundsare eluted per unit time with micellar mobile phases, as compared to the traditional aqueous-organic mobile phases. It can be observed in Fig. 9.2, that between ca. 30 and 120 min, three pe are eluted with a '75:25 methanol-water mobile phase,as compared to six or more with an aqueous 6% Brij8 35 micellar mobile phase, although the ability to resolve solutes is less for the micellar phase. The observed selectivity drffierence is notattributable to a diflierence in solvent strength, as the alkylbenzenes begin to elute earlier with the methanol-water mobile phase. 3
I
0
0.0
.0
17.0
M.0
Slt.0
u.0
U.O
MlNUTES
102.0
119.0
136.0
iU.0
170.0
Figure 9.2 Chromatograms showing the separation ofn-alkylbenzene homologues eluted with: (A) 6% Brij-35, and (F3)methanol-water 75:25 (v/v).Elution order in A: propylbenzene, ca. 50 min, butylbenzene, amylbenzene, phenylheptane, phenyloctane, phenylnonane, and phenylundecane. Reprinted from Ref. 9 with permission of the American Chemical Society.
b) ~ ~ eof cthet ~unctiona~ Group on ~ e ~ e c t i v i ~
Since a(CH2)is the ratio of the retention factors oftwo compounds differing only in a methylene group, it should be independent of the series type for a given mobile and stationary phase system, such is the case for the aqueousorganic mobile phases. In contrast, as shown in Table 9. l, for micelles the a(CH2 )values are dependent on the type of series, as the methylene selectivity for allsylphe~onesare consistently greater than for alkylbenzenes E71 *
The relative retentionof homologous compounds in both allsylbenzene and alkylphenone series can be defined as:
where toluene and acetophenone are used as the parent compounds. Therefore, the selectivity ratio of p~enones/benzenesis:
or
where a(C0) is the selectivity of carbonyl group. For aqueous-organic solvents, a(CH2) of two homologous pairs remains nearly constant at a different carbon number and theratio in eq. 9.8 is unity. In rnicellar eluents, the changes ina(CH2)with carbon number forakylbenzenes are farsmaller than those for alkylphenones. Therefore, the difference ina(C both types of homologues increases with carbon number.The is diEerentto unity, sincethe carbonyl selectivity between acetophenone and toluene is different fromthat of more hydrophobic compounds.
The same authors that found the existence of a nonlinear relationship MLC, intended to givea betweenk and n, ofhomologousseriesin rationalized explanation of this peculiar behavior. The different reported approaches are given below. a) ~ i ~ ~ o e n v i ~ ~oonl a~~ e i tni e~s
The curvature in the log k vs. rq: plots, observed with micellar and hybrid mobile phases, was first attributed to the different locations (with different microenvironme~tpolarities) in the micelle, for dfferent members of a homologous series [6, 7 ' 1. Methylene selectivitydecreases as the difference between mobile and stationary phase polarities is reduced. For a given mobile phase composition, the larger and more hydrophobic homologous compounds are located in a less polar environment of micelles, it is then conceivable to assume that these compounds experiencea smaller change in their ~ c r o e n v i r o ~ epolarity nt upon being transferred from the micellar pseudo-phase to the bondedalkyl stationary phase. The a(C between n-pentylbenzene and n-butylbenzeneis smaller than it is between ethylbenzene and toluene (Table 9. l), because the former pair is located in a more nonpolar environmentthan the latter. For micelles, the typical selectivities for alkylphenones and alkylbenzenes are between 1.l and 1.6 for SDS concentrations inthe 0.06-0.5 M range, which is similar to the values observed for organic-ri aqueous-organic systems. The overallsmaller a(CH2 ) in indication of how closely the environment of a methylene group in the micellar mobilephase resembles that of the alkyl-bonded stationary phase.
icellesandgraftedalkyl-bondedphaseshave a similarmolecular organi~ation,with chain molecules and a large su~ace/volumeratio, and their prope~iesvary with depth from the surface. A constant amountof free su~actantadsorbedonthealkyl-bonded stationary phasemakes the e n v i r o ~ e n t in s the two systems even moresimilar. The appearance of the a methylene group stationary phase is "micelle-like." It is not surprising that doesnotfindmuchdifferencebetween its e n v i r o ~ e n tin surfactant aggregates and in the alkyl-bonded phase. This situation is similar to that found in organic-rich conventional RPLG where thee n r i c ~ e nof t an alkylbonded phase by organic solvent makes the stationary phase e n v i r o ~ e n t more similarto that of the mobile phase. The effiect of a polarhnctional group in a homologous series onthe ) canbe also examinedfrom the perspective of its eoverall polarity ofthenonionicmolecules [7]. The carbonyl group of alkylphenones is located in a more polar medium of of micelles than alkylbenzenes. As a result, the methylenegroups alkylphenones would see a different mobile phase e n v i r o ~ e nand t undergo a larger change in polarity as transferred from micellar eluent to stationary phase. A situation that does not exist inaqueous-orga~csolvents. b) Free Energy of Cavity ~ o r ~ a ~ i o ~
The linear relationship between k and n, observed with micellar mobile phasesmay also resultfromenergeticsunique tothepseudo-~hase e n v i r o ~ e n.t In conventional C withaqueous-organicmobilephases, the linear dependence between log k and n, has been a~ributedto the direct propo~ionalitybetween log k and the free energy of retention, which is in turn a linear combination of the free energy increments associated with the constituent parts of the molecule [8], The predominant contribution to the free energy derives fiom the cavity formation within thephase mobile solvent structure that is required to accommodate the solute molecule transferred from the stationa~phase:
logk = -
IiT
+- logc
This equation impliesa constant contributionto the solute, between mobile and stationary phases, with the chain length of the homologue:
(9.10)
In other words, since the fiee energy ofcavity formation, of the cavity surface area, it is proportional to the regular increase in molecular volume upon addition of each homologous structural unit. In contrast, for MLC, it has been postulated that the free energy of cavity formation would not contribute the to overall free energy of retention, because the cavity created when the solute partitions from the micellar pseudo-phase to bulk water would be lostwhen the solute partitions from bulk waterto the surfactant-modi~edstationary phase, p r o d u c ~ gno netfiee energy change [9]. The retention of solutes in MLC can be described by the following equation:
(9.1 1) where Km is the solute-micelle binding constant,4 is the phase rat the partition coefficient between stationary phase and water, and concentration ofsurfactant in the form of micelles[lo]. As was derived in Chapter 7 for a pair ofcompo~ds,the chromatographic selectivity is simply the difference between micelle andstationary phase selectivities:
Equation 9.12 indicates that the net free energy oftransfer of a methylene group from mobile phaseto stationary phase is the diflerence between the
fiee energy of transfer fiom bulk solvent (e.g. ,water) to s~tionary phase, and from bulk solventto micelle [7]. By analogyto the RPLC theory, the individual partitioning processes within the MLC mechanism would be expected to exhibit a logarithmic relationship for any partitioning process involvingwater as a phase 16, 91: log K,, (Pws)= a r4: + b
(9.13)
The slope ofthis equation would be a measure ofthe fiee energy of transfer of a methylene group from bulk solventto micelle (for logK,, vs. n,),or from bulk solvent to stationary phase (for log Pws vs. n, ). The intercept represents again the interaction betweenthe residue of the homologue with micelles or the stationary phase. However,the linear relationship cannot be extended to thehigherhomologuesbecausenegativeintercepts are consistently encountered in theIn( vs. [M] regressionanalysis. c) ~ o ~ u b Limit i ~ Theory i ~
The solubility limit model (see Chapter 5) appears to provide the best prediction of the retention for the entire homologous series [g, 1l]. The behavior displayed by the homologous compounds can be ~ualitatively two independent equilibria between bulk water explained by focusing on the and stationary phase, and micelle andstationary phase. At low homologue number, the solutes are the most water soluble, so the retention is best described as being due to its distribution between the surfactant-modified stationary phase and thenomicellar portion of the mobile phase, which is largely water. This equilibrium is typified by ordinary WLC, and one expects a significant free energyof transfer for the methylene group. The slope of the log k vs. curve reaches its largest value (Fig. 9.3). As the homologues become larger, they are less water soluble.In the limitof water insolubility, the homologues partition directly between the chemically similar micelles and hemimicellar modified stationary phase, the partition coefficient approaches unity and log k becomes independentof n, . The log k vs. n, curve flattens. For this process, AGcmust be nearlyzero. Because AGGis a hnction of homologue size, plots of log vs. k m, are nonlinear.
fiee energy of transfer fiom bulk solvent (e.g. ,water) to s~tionary phase, and from bulk solventto micelle [7]. By analogyto the RPLC theory, the individual partitioning processes within the MLC mechanism would be expected to exhibit a logarithmic relationship for any partitioning process involvingwater as a phase 16, 91: log K,, (Pws)= a r4: + b
(9.13)
The slope ofthis equation would be a measure ofthe fiee energy of transfer of a methylene group from bulk solventto micelle (for logK,, vs. n,),or from bulk solvent to stationary phase (for log Pws vs. n, ). The intercept represents again the interaction betweenthe residue of the homologue with micelles or the stationary phase. However,the linear relationship cannot be extended to thehigherhomologuesbecausenegativeintercepts are consistently encountered in theIn( vs. [M] regressionanalysis. c) ~ o ~ u b Limit i ~ Theory i ~
The solubility limit model (see Chapter 5) appears to provide the best prediction of the retention for the entire homologous series [g, 1l]. The behavior displayed by the homologous compounds can be ~ualitatively two independent equilibria between bulk water explained by focusing on the and stationary phase, and micelle andstationary phase. At low homologue number, the solutes are the most water soluble, so the retention is best described as being due to its distribution between the surfactant-modified stationary phase and thenomicellar portion of the mobile phase, which is largely water. This equilibrium is typified by ordinary WLC, and one expects a significant free energyof transfer for the methylene group. The slope of the log k vs. curve reaches its largest value (Fig. 9.3). As the homologues become larger, they are less water soluble.In the limitof water insolubility, the homologues partition directly between the chemically similar micelles and hemimicellar modified stationary phase, the partition coefficient approaches unity and log k becomes independentof n, . The log k vs. n, curve flattens. For this process, AGcmust be nearlyzero. Because AGGis a hnction of homologue size, plots of log vs. k m, are nonlinear.
C igure 9.3
Idealized plot of eq. 9.17.
From the definitionof k in MLC: 'S
k = 'W
+
(9.14) 'm
or the reverse:
-+""1
__
w '
nm
'S
'S
(9.15)
n, ,nw andn, being the molesof solute inthe stationary phase, bulk water of the partition coe~lcients andmicelles,respectively.Bysubstitution between mobile and stationary phases, Pws and PMs,in eq. 9.15 :
(9.16)
Eq. 9.16 rewritten in terns of free energies yieldsthe following:
1 =
(9.17)
kf
where the superscripts "fg" and "c" correspond to ~nctionalgroup and methylene, respectively.It isevident fromeq. 9.17 thata linear relations~p between log k and r4: is in principle not expected, nora linear relations hi^ between k and r4: m
The point on the curve in Fig. 9.3, where the break occurs, is the of n, that makes the WS part equal to the MS part in eq. 9.17, that is:
(9.18)
nc =
c
8
of inputs are nee ed: a set of q ~ a n t i ~ t i v e l y c o ~ p a r a ~ l e ciently large n ~ b e ofr solutes (dependent variable) and ~ a r i o u~uantities ~ assumed to reflect s t ~ c ~ rfeatures al of the solutes S
(independentvariables). Throughthe use of diverse chemometric tec~iques, the retention p a r ~ e t e r are s characterized in terms of various combinations of solute descriptors or in t e r n of systematic knowledgeextracted ( l e a ~ e d ) can provide otherwise inaccessible from these descriptors. Goo i~ormationofthe solutes and atographic system involved: (i) the prediction of the retention for a new solute,(ii) the mostinformative structural descriptors, (iii) the molecular mechanismof separation operating a given chromatographic system, (iv) complex physicochemi~al lained in this chapter havebeen properties. The basic concepts on QS extracted mainly from reviewswritten by Kaliszan [S, 121 and Khaledi [4], and should be consultedto extend the knowledge on this field.
a)
The Logari~h~ic Correla~lon
LC with aqueous-organic mobile phases depends ma' hydrophobic interactio . Ingeneral,some correlation between retention data andlog is observed for a given ser observedrelationships ort the assumption that in rather than adsorption processes are decisive for retention. LC, the relationship between retention and log expressed inthe logarithmic formas: log klog =a
Pow+ b
(9.19)
which is aspecial case of the Collanderequation that predicts linear relations~psbetween the logarithm ofthe partition coefficients measured in two different partitioning systems, provided that solute-solvent ~teractions are similar inthe two systems [l ] : log P, = a log P,+ b
(9.20)
an
nedisadvantageofthe LC system is that it offers distinct h y ~ o p h o b i cmi ~ ~ ~ e Forn example, ~ . at a fixed ~ n ~ n ~ a ~ o n of a given solvent, soluteX may appear more hydrophobicthan solute Y, whereas the reverse seemsto be true at another concentration.The presence of organic modifier in mobile phase makes the interactions d e t e ~ n ~ g chromatographic retentionextremelycomplex.High on cent rations of organic modifier weaken the hydrophobic effect, and consequently, result in under or overestimation of hydrophobicity. Therefore, an initial ~uestion that must be addressed in these studiesis both the choice organic of modifier and its concentration.
t has been generally recognized that thebestmeasure of hy~rophobicityis the retention factor with a mobilephase of 100% water, kW. The advantages of using kWare that it is independent of any specific organic modifier effects, it reflects polar-nonpolarpa~itioningin a manner similar to s h ~ e - f l a ~measurements, k andis dependent on solute'sstructure and polar ~nctionalities. owever, kWis difficult to obtain exper~entally for most compounds because of prohibitively long retention times, and of a plotof log kvs. usually it must be obtained by the extrapolated intercept volume percent organic solvent. Mhle these plots are linear for a narrow range of organic solvent, it is well known that deviations from linearity occurs if a wide range is e x ~ i n e d Even . more troublingis that for a given s o l u t e - c o l u pair, ~ e~rapolationof log k vs. volume percentorganic solvent gives significantly different intercepts for different organic solvents. b) C o n ~ e n e orf~~~o ~ ~ t e s
The correlations betwee on factors and log ,P are limited to congeneric compounds. different relations~ps probably will be observedfor different sets of compounds, becauseof the complex nature of the factors that contribute to retention, which isdrastically different from the transfer of solutes betweentwo isotropic aqueous andorganic solvents. The term congenericis not just limited to homolo~ous compoun~s of a given ifferentsolutes fiom variousclasses(hydrogenbonddonors, hydrogen bond acceptorsor nonhydrogen bond donors)can be classified as a congeneric group if they show similar partitioning behavior in solvent systems with dissimilar phase polarities. Accurate definition of the term congeneric is however difficult andc ise d y open to Merenti n t e ~ r ~ t i o ~ .
Several researchers have attempted to find a single equation that relates log k inWLC to log Powfor a wide range of compounds that differ in size, shape, functional groups and type of interactions. The primary approach has been to adjust the chromatographic conditions (i.e., composition of mobile and stationary phases) such that the transfer of solutes from the aqueous media to the nonpolar bonded phases of closely resembles the partitio~ngprocess in an octanol-water system. Hydrophobicity is as much a phobia againstthe aqueous e n v i r o ~ e n t and the chemistry involved the in as a “philia” toward nonpolar species, thus, contact of a solute with thestationary phase cannot be neglected. In order to apply the Collander equationto the relationship betweenthe retention and log Pow, a similar mechanism should exist between the solutes and the organic phases, in addition to a similar molecular interaction. stationary phases used octanol is an isotropic liquid while the alkyl-bonded in WLC are anisotropic and heterogeneous media, whose propertiesvary with depth from the interface. Also, the structure and composition of the alkyl-bonded phase depend upon the type and concentration of organic cosolvent in the aqueous mobile phase. Furthermore, the residual silanol on the silica sudace caninfluencetheretention of certain solutes. The accessibility of silanol groups (which are reported to be solvated with water andor hydrogen bonding organic cosolvent) depends the on composition of mobile phase.
(ODs) stationaryphases have been For years, octadecyl-bonded silica comrnonly employed for hydrophobicity studies. However, the retention data ODS columns -even though theyare basically the obtained using individual same type of material-are hardly comparable. The stationary phases vary in termsof endcapping, surface area of the bonded phase,carbon content and concentration of residual silanol groups. Another disadvantage o reversed-phase materials is their instability at pH >8. With these lim in mind, researchers have attemptedto introduce reversed-phase materials for the construction of a universal, continuous chromatograp~chydro.g., specially deactivatedODs-based phases and polymerer knowledge of thenature of thepartitio~ngprocesses involved may make possible the design of systems, in whichthe chromaits relative tographic retentionof a certainsolutedependssolelyon hydrophobicity.
a)
Cor~elationsbetween ~etentionFactors ana' log P,,
Some reports have appeared in the literature showing LC as a promising t e c ~ i q u for e thequantitation of hydro bicity [3]. It appears that retention in analysis. a obtained with micellar eluents LC can be used lcelles are uniquemedia whch o flexibility in controlling specific interactions alongwithhydrophobic forces, by the careful selection of surfactant (chain length and headgroup) and solvent additive. Several studies have been published on the correlations between the LC and log P,, . Cago et al. [131 and Conzalez et al. [14.1 reported good correlations between logk and log ,for 1l monosubstituted benzenes, and groups of12 to 16 polycyclic aromatic hydrocarbons ( Fig. 9.4), respectively, eluted with micellar eluents of S rij@ 35 from C 18 columns. The same was observe Lavine et al. [l51 -propanol. r 2 l aromatic compounds with a mobile phase of and 1 correlations In contrast, KhalediandBreyer [161 repexcellent for k (instead of log k )vs. log ,P for anioni S) and cationic (tetradecyltrimethyla~oniumbromide, C ,TA )surfactants on C8 and phenylbonded stationary phases, with c o m p o u n ~ having variousfunctionalgroups. The logkvs. log ,P plots consistently showed acurvature. The two sets of 16 and 35 compounds used bythese authors with C8 and phenyl columns, respectively, covered more than 4-orders ofm a ~ i t u d in e hydrophobicityand represented a relatively broad range of molecular interactions, molecular shapes and sizes. In a sense, the sets consisted of several noncongeneric compounds.
,for micellar and Figure 9.5 illustrates plots of kand logk vs. log , aqueous-organic mobile phases and a set of 16 aromatic compounds. As shown in Fig. 9.5b, there is a clear curvature in the plot of log k vs. log , P for the micellar eluent (a quadratic fit provides a better regression than a linear fit). Also, the r data in Table 9.3 indcate that for the micellar mobile phases, k vs. log,P fits are better than log k vs. log ,P fits. Therefore, the same phenomenon appears in the log k vs. n, and log k vs. 1 which can be explained by the solubility it theory proposed b or compounds with low log values, the solutes are relatively
rij
8
Figure 9. P l o t s of log k vs. log P, for several PAHs eluted with (surfactant molar concentration is given from top to bottom): (a) SDS (0.06, 0.08, 0.10, 0.13, 0.15 M); (b) CTAE3 (0.02, 0.04, 0.06, 0.08 M); (c) Brij-35 (0.02, 0.04, 0.06, 0.08, 0.10 M).PAHs: naphthalene, acenaphthylene, fluorene, anthracene, phenanthrene, 9-methylanthracene, ~uoranthene, pyrene, chrysene, benzo[a] a n t h r a c e nbee,n z o [ b ] fluoranthene, benzo[a]pyrene, benzo[e]pyrene, perylene, dibenz[ac]anthracene, dibenz [ahlanthracene, and benzo[ghi] perylene. Reprinted from Ref. 14 with permission of Vieweg Publishing.
Comparison of the Correlation Coefficients of the Fittings k vs. log P, and log k vs. log Pow for Micellar and ~queo~s-Organic Eluents
r
,P
Eluent
log k vs. log, , P
n = 16 [l61 0.04 M CTAB
0.970
0.925
0.04 M CTAB + 3% 2-propanol
0.980
0.933
0.08 M CTAB + 3% 2-propanol
0.965
0.910
0.12 M CTAB + 3% 2-propanol
0.959
0.913
0.04 M SDS + 3% 2-propanol
0.964
0.925
0.08 M SDS + 3% 2-propanol
0.969
0.930
0.12 M SDS + 3% 2-propanol
0.966
0.878
Met~anol-water70:30 (vh)
0.873
0.901
Met~anol-water40:60 (vh)
0.821
0.982
n = 11 E131 0.10 M SDS
0.988
0.969
0,016 M CTAE3
0.997
0.946
0.05 M CTAB
0.984
0.984
0.10 M CTAB
0.987
0.989
0,016 M Brij.35
0.953
0.05 M Brij-35
0.972
0.980
Me~anol- at er
0.985
0.970
0.986
Figure 9.5 Plots showing the relationship between k andlog k vs. log P, for 0.04 M C,4TA13-3%2-propanol (a, b),andmethanol-water 40:60 (vh) (c, d). Compounds: (1) benzylamine, (2) benzyl alcohol,(3) acetanilide, (4) phenol, (5) benzaldehyde, (6) benzonitrile, (7) acetophenone, (8) nitrobenzene, (9) benzoicacid, (10) anisole, (11) benzene, (12) propiophenone, (13) butyrophenone, (14) chlorobenzene, (1 5) naphthalene, and (16) anthracene. Reprinted from Ref. 16 with permissionof the American Chemical Society.
water soluble, so water-stationary phase p a ~ i t i o ~ plays n g its largest role. Highly hydrophobic solutes become insoluble in water, and the micellestationary phase equilibriumthen becomespredominant. In this e~uilibrium, the two phases are chemically similar and the partition coefficient approaches unity, becoming independent of hydrophobicity. The log k-log ,P curve flattens.
This explains why log k-log, P correlations improve whenthe most hobic compounds are elimin from the curve. Thus, for a 0.04 -3% 2-propanol eluentona tationary phase, elimination of ited setof 16 aromatic c o m p o ~ d s more hydrophobic compounds from 4.45 to 2.19) resulted in a (whch reduced the upper 1 coeE1cient for the log kvs. lo the same mobile phase on a p log ,P provided better correlations for two sets of 28 and 23 compounds with differentupper ~ydrophobicitylimits of 4.88 and 2.69 [161. Thecationicmicellareluentsof CI4 gavebetterlinearcorrelations ,than the anionic surfactant SDS [161. This might indicate nicmicellarsystemshaveasimilar hydrophilic/lipophilic balance to octanol-water, andlor the polar interactions betweenthe solutes and the cationic head group better resemble those in octanol-water. For the chromatographic correlations, oneshould also note that the cationic su~actantsadsorbed on the stationary phase can better shield the residual silanol groups on the silica surface, than S S. Consequently, the reduction (or e l ~ t i o n of ) the silanophlic ~ t e r a ~ i o would ns increase the correlations LC retention and log, P . thenvise, the addition of alcohol to the micellar mobile phase seems to improve log k-log, P [1'7, 181 and kvs. log ,P [l61 linearcorrelations. The alcohol apparently provides an e n ~ r o ~ ethat n t is more closely related to that of octanol-water than pure aqueous micellarsystems. These results show that the type of general relationship depends not only uponthe set of compounds but also on thecharacteristics of mobile and stationary phases. This was confirmed in a study of the chroma tog rap hi^ 'or of a group of 11 benzene derivatives and 12 PAHs, in S micellar mobile phases modified with methanol, l-propan butanol at different percentages[191. The authors demonstrated effectively that the hydrophobici~range of compounds isan i m p o ~factor t in the klog , P or log k-log , P correlations. For the whole set of 23 aromatic ,P from 0.64 to 5.03), k always correlated better compounds studed (log, with log ,P than log k, irrespective of the nature of the surfactant present in the mobile phase, andthe nature and percentageof the alcohol usedas a owever, for a group of 15 benzene and naphthalene derivatives
17
(log ,P from 0.64 to 3-37)),and for 10 monosubstituted benzenes (log,P fiom 0.64 to 2.84), k correlated better with log , P than log k only when was used, whereas withSDS log k-logP,, correlations were similar to or even better than k-log ,P Thus, depending onthe situation, either k or log k might betterfit the data and one must thereforetry both and use the one with the best correlation.
Figure 9.6 Calculated vs. experimental log P,, values for 0.05 M CTAB3% l-propanol mobile phase for a group of benzene derivatives and PAHs. Reprinted from Ref. 19 with permission of Vieweg Publishing.
The values of log ,P can be calculated from experimental k data, using the equation of the fitted straight-line. Figure 9.6 shows calculated P plotted vs. experimental values for a mobile phase of 0.05 M CTAB log , modified with 3% 1-propanol. The average of the relative error obtained between the calculated and experimental values 9.6%. was Finally, it should
be notedthat deviations from the established correlations can be expectedfor caseswhereother factors (e.g., structural properties,otherspecific molecular interactions besides hydrophobicity) significantly contributeto the partitioning process.
b) ~ y d ~ o p ~ o bofi Amino c i ~ Acids Isoindolesof0-phthalaldehyde (0PA)~-acetyl-L-cysteine (NAC) are comonly used for detection of amino acids (RICH(COOH)~H~) in A particular hydrophobicity scale was established with the retention data of thesederivativesinmobilephasesof §D§ at pH 3, using the glycine derivative as a reference [20]. Linear relationships were obtained between the ratios of log k of each amino acid derivative to log k of the glycine derivative (whch was calledquantitation ofhydrophobic it^ index, QH), and log ,P for the R1 substituents (nRl). The interactions of the amino acid derivatives with a modifiedC 18 SDS are mainly of columnandmicellarmobilephasesoftheanionic hydrophobic and electrostaticnature. The basic structure of the amino acid isoindoles is thesame.Consequently, in theabsenceof electrostatic interactions, the hydrophobic character of the R1 substituent should be responsible of the retention. The nitrogen atom in the isoindole heterocycle is nonprotonated at pH 3, and thus, electrostatic interactionscan only exist with ionizable groups giving a positive charge to R, in the molecule. This occurs with arginine and histidine, which have amino groups that do not react with OPA. Figure 9.7 shows log QHvs. nRl plots for hybrid eluentsof §DS and methanol, l-propanol and l-pentanol. For all these eluents, the correlation For the derivatives of alanine, betweenlog QHand nRl wasgood. asparagine, aspartic acid, cysteine, glutamic acid, glutamine, serine, and threonine, the QH index was less than unity andnR1was negativeor slightly positive. For these compounds, the order of hydrophobicity was the same according to the QH index andthe nRIvalue. Valine, leucine and isoleucine H > 1, also showed this behavior. On the other hand, for some compounds showing QH> l, a diflerentiated behavior was observed.This was the case of the derivatives of arginineand histidine, which showed a
H index due to electrostatic attraction of the additional protonated amino group in the R, substituent with the anionic surfactant. Another separate group was formed by tyrosine, tryptophan and p h e n y ~ a l ~ n e , because of the presence ofan aromatic ring inthe R, substituent.
l .0
OS
8
_ .
0.0
-0. S
-a.e 22
"1
b
I
R1 't.Q
X
0.S
U
m
ao
-
1.0
i
4
Figure 9.7 Plots of log QH logindex vs. P,, of R, substituent for amino acid OPA-NAC derivatives, eluted with 0.05 M mobile SDS micellar phase at pH 3 containing: (a) 5% methanol, (b) 3% l-propanol, and (c) 1% 1pentanol. Amino acids: alanine, asparagine,asparticacid, glutamic acid, glutamine, isoleucine, leucine, phenylalanine, serine, threonine, tryptophan, tyrosine, and valine. Reprinted from Ref. 20 with permission of Vievveg Publishing.
c) An Apparent Linearity LC, the retention is influenced by two competing equilibria ofsolute interactions with micelles in the mobile phase (controlled by Km) and their pa~itioninginto the stationary phase (controlled by Pws)(see eq. 9.11). Both of the solute. The partitioning processes depend on the hydrophobicity reciprocal of the intercept in eq. 9.1 1is the retention factor at zero micelle on cent ration, h, which is a parameter similarto kW,and as shown below, is useful in hydrophobicity measurements. Let's now suppose that log k, and log Km are correlated with the hydropho~icityof solutes: log k, = A,
+ AI log P,,
(9.2 l)
log Kp&f= B, + B, log P,,
(9.22)
then eq. 9.1 1becomes:
and in logarithmicform: log k = A,
+ AI log P,, -log( 1 + [M] 1
O@O
+
1
Bilogpow)
(9.24)
From eqs. 9.23 and 9 -24,a nonlinear relationship should be expected , , P for a constant between the retention of a compound (k or log k) and log, micellar concentration. Two extreme situations can be considered:
1
(i) For solutes with low hydrophobicity (low logP,, )or for very lowmicellarconcentrationsin the mobile phase, the term ]is negligible and eq. 9.23 becomes:
k k0 = 1o(Ao + Ai log Pow)
(9.25)
which gives a nonlinear relationship between k and log esides, from eq. 9.24:
(9.26)
which translates into a linear relationship between log k and
,P or for high (ii) Forhighly hydrophobic solutes (highlog ) micellar concentrations inthe mobile phase,eqs. 9.23 and 9.24 result in: (9.27)
from where: log k = A, -B,
+ (A, B,)log P,,
log [M]
(9.28)
Equation9.2'7provides an apparent linearrelationshipbetween k and log P,,, and eq. 9.28 describes a linear relationship between log k and log P o w *
Accordingly, when logk is plotted vs. log P,, ,a break should occur when Km [M] =: l. The value of log ,P for this point is:
3
(9.29) A s can be deduced from eq.9.29, the linearity rangeof the plots increases when the micellar concentration in the mobile phase decreases. There is therefore an apparent parallelism with the fact that a decrease in organic modifier of aqueous-organic mobile phases, in conventional WLC, results k vs. log ,P . in better correlations of log
lo t
S
Log k vs. logP,, (a) andk vs. logP, (b) relationships predictedby eqs. 9.23 and 9.24 (solid lines), and experimental values (symbols) for a series of monosubstituted benzenes: acetanilide, acetophenone, benzaldehyde, benzene, benzonitrile, benzyl alcohol, benzylamine, bromobenzene, butyrophenone, hexaphenone, methyl benzoate, methyl phenyl ether, nitrobenzene, propiophenone, toluene, and valerophenone. Molar concentrations of SDS in mobile phase:( 1 , ~0, ) (2,*) 0.016,(3,O)0.05, ( 4 , ~0.1, ) and (5) 0.15. Reprinted from Ref. 21 with permissionof Elsevier.
Figure 9.8 depicts themodeled logk vs. log,P (left) andk vs. log, P (right) relationships for several neutral compounds, at various micellar concentrations, togetherwithexperimentalpoints.Ingeneral,good agreement between predicted and experimental values was obtained. The
relationships between log k and log ,P are nonlinear (curves 2-5), as predicted byeq, 9.24,although two different regions ineach curve with an apparent linearity are observed at low and highlog ,P values. A decrease in micellar concentration expands the range of linearity. In contrast, as predicted by eq. 9.23, k-log Powrelationships are always nonlinear. Statistical Analysisof Linear Regressions Between Chromatographic Parameters andlog P,, for Several Setsof Compounds [21]."
Stationary 1% p*, Series Phase/Su~actant Variable r n range I
C18/0.l M SDS
I1
CW0.1 M SDS
Dependent
F 0.93 1 0.954 0.989 0.760 0.986 0.7 18 0.977 0.961 0.982 0.913 0.856 0.922
45 71 308 10 250 7 227 133 277 55 28 57
0.987 0.991 0.992 0.868 0.979 0.866
293 430 497 25 182 25
C18/0.1 M I11
CTAB
1.10-2.99 log10
k k log k;,
k, log K M KM
.."~."."".~~~~~~________._________._". .""**..
Series
IV
V
VI
Stationary log pow Dependent phase/Su~actant range n rVariable G1810.05 M r i j 8 35 1.10-2.99 10 0.959 0.989 0.993 0.833 0.974 0.817 G810.12 M 1.10-4.45 l 1 0.924
91 372 608 18 146 16 58
0.966 0.991 0.863 0.987 0.889 0.95 1
139 500 29 354 34 95
0.984 0.991 0.864 0.982 0.917
301 534 29 248 48
C1810.12 M
SDSb
l . 10-4.45
l1
F
Series I: anthracene, benzene, biphenyl, l-bromonaphthalene, l-methylnaphthalene, naphthalene, pyrene, toluene and p-xylene. Series l": acetanilide, acetophenone, benzaldehyde, benzene, benzonitrile, benzyl alcohol, benzylamine, bromobenzene, butyrophenone, hexaphenone, methyl benzoate, methyl phenyl ether, nitrobenzene, propiophenone, toluene and valerophenone. Series V-W: acetophenone, anthracene, benzaldehyde, benzene, benzonitrile, benzyl alcohol, butyrophenone, chlorophenone, naphthalene, nitrobenzene and propiophenone. Mobile phase with 3% 2-propanol.
a
d)
Log ko vs. log P,, and log Km vs. log P,, as a Means to Probe
Solute ~nteractionsin the ~ ~ r o ~ a t o g r a pSystem hic Log h vs. log P,, and log Km vs. log P,, relationships have also been examined 116, 2 1, 221. Tbe results suggest that Km does not reflect the hydrop~obicityof compounds as well as k, (Table 9.4). Figure 9.8 also reveals the superior capability of log h data (curve 1 in Fig. 9.8a) with solutes. Note also the respect to log kor k to predict the hydrophobicity of P plot. greater sensitivity (slope) of the log k,-log , Lavine etal. [153 made an interesting study, using a set of2 1 aromatic compounds as probes, that demonstrates the interactions of solutc;?swith mobile and stationa~phases in MLC. In Fig. 9.9, log P,, is plottedagainst log k for methanol-water and S S mobile phases. It can be seen that the data set is divided into twogroups. The first group of compounds (groupA) is mainly composedof monosubsti~tedbenzenes, whereas the second group )consists entirely of phenols. Compound 3 (2,4-dinitrophenol), which does not belongto any group, existed mainly inthe anionic 3.5
3.0
I4
2.5
n,
M 2.0
0
U
I .5
f .O
0.5
Plots of log Pawvs. log k for a series of 21 aromatic compounds. A C18 column was used with (a) methanol-water 50:50 (v/v), and (b) 0.05 M SDS with 2% 1propanol. Compounds: (1) benzyl alcohol,(2) benzaldehyde, (3) 2,4-dinitrophenol, (4) bemonitrile, (5) acetophenone, (6) nitrobenzene, (7) p-nitroanisole,(8) methylbenzoate, (9) anisole, (10) benzene, (11) toluene, (12) chlorobenzene, (13) bromobenzene, (14) ethylbenzene, (1 5) resorcinol, (16) catechol, (17) phenol, (18) p-nitrophenol, (1 9) o-chlorophenol, (20) obromophenoland (2 1)2,4-dichlorophenol. Reprinted from Ref. 15with permission ofElsevier.
3
form at the pH of the mobile phase (pH 6.3) and therefore, eluted off the column in the dead volume. The only difference betweenthe two data sets (methanol-water and SDS mobile phases) is that catechol lies in group A instead of group B, but there is no difference in the elution order of the compounds. The slopes of the lines drawn through the sets of points are approximately equal for eachset.
P and log Km vs. log , P were made for Next, plots of log vs. log, the aromatic compounds, to better understand the reasons for the similar results in the aqueous-organic and micellar eluents. A nice straight-line could be drawn through allthe data points in the log Km plot (Fig. 9. lo), but for log ,the samedichotomy as in Fig. 9.9 wasobtained. It was the concluded that the differences in the chromatographic behavior between phenols andthe other aromatic compounds tested, observed the in log kvs. plot could not be attributed to a mobile phase effect, but in all likelihood, to the presence of unhindered silanol groups the onbonded phase surface, as occurs in conventional RPLC. Similar experiences (not shown) were made with CTAB as s ~ r f a c t ~For t . CTAB, group A was on the left ,,P and the three and groupI3 was on the right in the plot of logk vs. log, plots (log kvs. log ,P ,log ko vs. log ,P ,and log K,, vs. log, ,P )were similar. Evidently, in this case, the mobile phase effect was important. The interaction ofthe solutes with aCTD-coated C 18 phasealso diEered from S-coated C 18 phase probably due to a change in the sufiace charge ofthe stationary phase. e)
~~~
ofIonic C o m ~ o ~ n d s
any biologically active compoundsare ionic at physiological p the degree of ionization is the same for structurally related compounds,the difference in retention is dueto the difference in hydrophobici~.However, ,P for ionic compounds with diEerent degree of ionization, linear log k-log relationships will not be obtained. In this case, the following model has proved to give good correlations [23-251: log k = a log ,P
+ b6 + c
(9.30)
2. t.
1
.IO Plots of log P,, vs. logarithm association constants between: water and micelles (top), and water and stationaryphase (bottom), using SDS mobile phases with 2% 1propanol. The compounds are the same as in Figure 9.9. Reprinted from Ref. 15 with permission of Elsevier.
an
where a and b are fitting coefficients, and 6 is the molar fraction of the charged form ofthe compound, which can becalculated as: (9.3 1) ICH being the
protonation constant. It was found, with a group ofcatecholmines, that the value ofaremained constant W . This indicated th the contribution ofhydrophobicity to the retenti ot vary when the p of the mobile phase changed[23].
53
~ d v a n t a ~ofthe e s MLC Appr~achto ~ s t i ~ athe te ffydrophobici~
LC provides several advantages timating the hydrophobici~of compounds, instead ofconventionalwith aqueous-organic mixtures. In an MLC system, silanophilic adsorption of surfactant monomers (in equilibriumwith micelles) on the alkyl-bonded stationary phase is produced. The stationary phase becomes more hydrophobic and the concentration of residual silanol groupson the silica surface (especially for cationic ,the residual activity of alkylsurfactants) is reduced. Inconventional bonded stationary phases caninfluenretentionof certain solutes, As a consequence, kocould giving inadequate prediction of hydrop be better than kW for these predxtions. r, for charged solutes eluted with an ionic surfactant, electrostatic interactions will occur. In this case, hydrophobicity predictions using k, will fail. Adequate selection of the nature of surfactant andp of the mobilephase could eliminate this problem. *
.
n the other hand, the e n v i r o ~ e n tof the surfactant-modified stationary phase is independent of rnicelle concentrationthe in mobile phase (for most surfactants and stationary phases), and similar to that of pure aqueous eluent systems. As a result, the alkyl-bonded station have both invariable mphiphilic and anisotropic properties. aqueous-organic LC, the composition and s t ~ c t ~ofr ethe a~yl-bonded phase change with the concentration of organic modifier in mobile phase.
The intercalation of organic solvent in the al~l-bondedphase plays an role in in~uencingretention and selectivity. ina ally, the retention behavior of compounds (nonpolar, polar or ionic)chromatographeanionic,cationicandnonionic surfactants has .9. l l). This permits the correct evaluation of been accurately mode h,while largeerrors in measuringkWvalues are ~ e ~ u eThe ~ tonly . problem with k, is that it cannot bemeasured for very hydrophobic compounds owing the near zero interceptfor the l& vs. ,although this probl improvcd by choosing a shorter chai stationa~phase, or ic m o ~ i ~toe rincrease the elutionstr
a)
Prediction of the ~ e t e ~ t i oofnPAHs based on Pro~erties
more complex approach in is the description of the retention ehavior of solutes by a multipar~eter e~uation that includes several escriptorsrelated to diEerent s t ~ c t ~ r properties. al ~hromatographic rctention data must besome ~ n c t i o nofthechemical structure of haseandmobilephase,allofthemmutuallyinte ever, there is no general, strict, canonical equation relatingthe retention ese variables. Even if the stationary and mobile phases in a given chromatographicsystemremain c o n s t ~ t ,still a precise ~uantitative ~escriptionof the retention of a series of solutes appears proble~atic, culties increase with the diversity of the solutes considere problem is also not trivial for homolo~ues. 7
studies is to generate a multitude of t retention data. The m i n i ~ u m ~ u a t i o n y i e lthe d ~ gc retention data, in satisfacto~agree with the observed values, is selecte o b s e ~ i n gall the statistical rules.numberof descriptors th individual solute is v i ~ a l l yunlimited. If tens or of descriptors are used, then most lkely, several e~uationswith similar
predictive abilities but consisting of different sets of variables can be derived. Linear regression models showing the dependence between retention and molecular structure make the selection of the optimum conditions for the separation and identification of unknownpeakson chromatograms of multicomponent mixtures possible. However, studies that are not inte~retableinphysicalterms are notvery i n f o ~ a t i ~regarding e the me~hanismof retention. Rodr~guez-~elgado et al. used statistical and principal component analysis to establish some general equations that relate the retention, in ,to severalmolecular descriptors of 1’7 S eluted with mobile S, CTAB and BrijB 35withoutalcohol [22], and with ropanol or l-butanol [26]. PAHs are widely distributed pollutants in the e n v i r o ~ e n tandshowmutagenic and carcinogenic properties. Therefore, great eEorts have been madeto develop methods for their quantitation and measurement of hydrophobicity. Nonpolar species possessing polarizable electrons, such as P W s , have beenfound to reside near the polar head groups rather than deep within the core of micelles. owever, a precise location is impossible to establish, especially dueto the fact that many of these solutes have dimensionswhich are comparable to those of micelles. The results obtained from the QSRR study inhcated thatthe shape, size and hydrophobicity PofM s are the dominant molecular characteristics for determining their retention in MLC. The behavior of unsubstituted S was well established usingthe model: log k = a CF + b L/B + c
(9.32)
where CF is a correlation factor calculated for each P double bonds)+ (number ofprimary and secondarycarbon atoms) -0.5 for a nonaromatic ring^, and L/B is the ratio of the ~ a x i m a l i ~ elength-tod breadth of the rectangle enclosing the molecules. The parameter CF is directlyrelated to the size of the moleculeandnondirectly to its hydrophobicity,whereas L/B refers to the shape of the molecule. The descriptors retention of unsubstituted P W s increased with both geometric
CF and LB. The regressioncoefficient of the fittedline for methylsubstituted P M s washoweververylow,whichwasexplained by the n o n p l ~ t of y the molecules due the to methyl group. For these compounds, two new descriptors (selected among 13 starting descriptors subjected to factor analysis) wereadded to themodel.The fraction ofnonpolar § ~ unsaturated surface area over total surface area ~ U § ~gives information about the electron delocalization along the aromatic rings. The ~ the Nsecond ) descriptor. The retention dipolemoment ( ~ F ~ was increasedwhenthe . " § U N ratio decreasedandwhen increased, although the relevance of this descriptor was lower and only acquired importancein describing the electronic properties of isomers. The relevance of these latter descriptors is in agreement with the polarity can play an important role in the retention behavior of F
)
The coefficients of the independent variables in eq. 9.32 change with the nature of surfactant and alcohol modifier and their concentrations, through first- and second-degree polynomia. These equations may the allow evaluation oftheretentionofasolute for anygiven chromato~raphic condition.
b) Linear Solvation EnergyRe~ations~ips Linear solvation energy relationships &§ER) have also been appliedfor the evaluation of the retention behavior in WLC. In LSER, solvent-related properties of solutes,SF (e.g., log k, log P,, or log aqueous solubility),are described in terms of solvatochromic parameters in the following general forrrl:
SP = SPo
+
m V + sn* + b p + aa
100
(9.33)
where SP, is the regression constant,V is the molar volume of solutes related to the solubility behavior, R* is a measure of solutes' ability to engage in dipolarity/pol~~abili~ interactions withthe solvent, p the solutes' basicity, and a the solutes' acidity. The coefficients m,S, b and a are related to the chemical nature of the solventsystems. The l o g a r i t ~ i crelationships
an represent the free energy transfer of of solutes from one phase (e.g., mobile phase or aqueous phase)to another phase(e.g., stationary phaseor 1-0ctano1 phase). The termmV/100representsthedisfavorableendoergiccavity formation process of separating the solvent moleculesto provide a suitably sized enclosure for the solute; V/100 instead of V is used to adjust the tude de of the cavity term within the same range as the other independent variables in eq. 9.33. The termSE* measures thefavorable exoergic eEects of solute-solvent dipole-dipole and dipole-induced dipole dielectric interactions, and bp and aa measure the exoergic eEects of hydrogen bondinginvolvingthesolvent as ahydrogenbonddonoracid and the (HE3~)base, or the as an solute as hydrogenbondacceptor base and the solute as an HE3 acid, respectively.
L VangandKhaledi[27]applied to the evaluation of the (k and Km) of un retentionbehavior ed substi~ted arornatic compounds of diverse hydrophobi diphenyl c o l u ~ sand , pure and hybrid mobile phases of with 2-propanol andltan01 as modifiers. Zou et al. [28] made a similar study with ws, and found high correlations betweenthe retention and solvatochrornic p a r ~ e t e r of s solutes. e regression coefficients ofLS as well as log k in conventi eluents, and the solvatochromic parameters for a group of 16 ar compounds are shown in Table 9.5. One common feature amongall 1s in the table is that the cavity termis ,the correlationwasbetterusing 1 d better with the solvatochromic p for the relationship between k compounds. The aa termwaspositive for solutes bind in^ to cationic rnicelles, while negative for anionic micelles. This suggested that rovides a more basic e n v i r o ~ e nto t micelles are less dipolar (more negativeS values) for solutes than
LSER Regression in MLC (eq. 9.33) for 16 Aromatic Compounds [27].a
§P log K,,
Eluent
§P,
m
S
b
a
r
SDS +
0.34
2.81 (0.18)
-0.29 (0.14)
-1.36 (0.13)
-0.35 (0.11)
0.958
C,,TAB + 3% 2-propanol
0.97
2.67 (0.14)
-0.88 (0.10)
-1.44 (0.10)
0.32 (0.08)
0.970
0.08 M SDS
1.27
1.52 (0.14)
-0.92 (0.11)
-0.78 (0.10)
-0.92 (0.08)
0.970
-5.52
121.9 (9.0)
-40.4 (6.9)
-59.9 (6.4)
-8.6b (5.4)
0.948
3% 2-propanol
log K,, log k
k
+ 3% 2-propanol 0.08 M SDS + 3% 2-propanol
log k
0.08 M C,,TAB + 3% 2-propanol
1.02
1.01 (0.09)
-0.23b (0.13)
-0.76 (0.06)
-0.02b (0.01)
0.943
k
0.08 M C,,TAB + 3% 2-propanol
0.53
53.9 (2.7)
-8.22 (2.04)
-32.5 (1.9)
1.3fib (1.58)
0.975
~e~anol- at er
-0.33
3.22 (0.08)
-0.32 (0.06)
-1.73 (0.06)
-0.23 (0.05)
0.994
-0.16
l .84 (0.11)
-0.3 l (0.09)
-1.43 (0.08)
-0.22 (0.07)
0.976
log k
40:60 (v/v) log k
Propanol-water 40:60 ( d v )
Uncertainties at the 95% confidence level are given. Values are not statistica~~y significant at the 95% confidence level.
The biological activity of many organic compounds, the bio-accumulation of organicpollutmts and soil sorptionof e n v i r o ~ e n t acontaminants, l have all been. attributed to thehydro aracter ofmolecules. m imp0 structure-activi~relationship studies play porary drug design, toxicology and e n v i r o ~ e n t amonitoring. l ,the ~iologicalactivity is viewed as a s u ~ a t i o nof the different
interactions that a compound undergoes both duringthe transport through biological membranes andin the reaction with the sitesaction of (receptor). These ~teractionsare assumed to be governed by the chemical structure of the compound. A change in structure can result in a change in biological response. The partitioning of solute molecules into lipid bilayers and biological membranes is the basis for drug and metabolite uptake, passive transport across membranes and bioaccumulation. For the prediction of the net eEects of complex p h a ~ a c o ~ i n e t iand c p h a ~ a c o d ~processes, ~ c the i n f o ~ a t i o nextracted from diversifieddata may be more usefulthan based on individual, one-dimensional hydrophobicity scales. To date, Hansch's hydrophobicity parameter, log ,,P is the most widely used reference to characterize the hydrophobic contributionto the free energychange in the CT are used to biological response, while the &pole moment and Hammett's describe electronic contributions;T a g s steric parameter, molar refiactivity and van derWaals volume are often used for size andsteric contributions. Another approach in QSAR has been the application of chromatographic techniques, mostly P L C . The use of chromatograp~cparameters gives rise to a new field: quantitative retention-activity relationships(
The l-octanol-water system is challenged as the only valid hydrophobicity scale. It may not bethe best model for studying the partitioning process in LC model better this biomembranes. Does not the retention in partitioning? It has been arguedthat the chemical bonded stationary phase resembles more the hydrocarbon chains of membranes, but the bonding density of almost all commercial columns may be too low to provide a suitable model.
In contrast, micelles have long been recognized as simple chemical models for biomembranes [29]. Indeed, structurally, micelles are more similar to biomembrane~ than l -0ctanol or RPLC stationary phases. Several researchers have demonstrated that the solubilization (or partitioning of solutes) into micelles closely resembles that of lipid bilayers. Both micelles
and biomembranes have amphiphilic properties and are anisotropic media that provide both hydrophobic and electrostatic sites interaction. of The use of micelles to model biomembranes in however received little attention. Perhaps the difficult measuring micelle-waterpartition coefficientsby conventional methods have been the major obstacles in conducting an extensive investigationof the use of micellar systems. The following contributions showthat a combination of the unique characteristics of micelles and the capabilities of physicochemical studies should be quiteusefix1 in QRAR research. b) ~stimutionof Phenol Toxicity and Anesthetic Action
Breyer etal. [3 01 reported interesting results on capacity the of predictionfor t~ The biological MLC of the toxicity of a group of 26p - s u b s t i ~ phenols. as log lIIGC50 (log VC), where activity of these compounds was measured IGC50 is the50% inhibitory growth concentration of phenolsthe in culture enu A graphical representation of the effect of of ~ e ~ r u h y ~pyriformis. mobile phase composition onthe quality of the predictions is given in Fig. 9.1 1. Micellar systems gave better prediction of toxicity than conventional RPLC systemswithaqueous-organicmobile phases. Addition of 10% 2-propanol to the micellar aqueous phase resulted in a better correlation, Also, a diphenyl column gave a slightly better correlation toxicity with than the more a C 18 column, since the former could better differentiate between hydrophobic phenols. The polarizability and configuration of the stationary phase, and the possibility ofm”interactions between solutes and s~tiona~ phase may have considerable effect on the retentionof more hydrophobic phenols. The predictive ability ofthe MLC QRAR model is compared in Fig. 9.12 with a three-variableQSAR model. The predicted value of log 1/C was calculated from the fitted log 1/C vs. k andlog 1/C vs. (log P,,, acid dissociation constant pK, and resonance parameter R) plots, by using the leave-one-outtechnique(thecompoundpredicted was left, out in the derivation ofthe model). As observed, the QSAR model had difficulties in predicting the toxicity of highly lipophilic phenols, as indicated by the curvature in this region. The results show that a single MLC retention
t
Observed log "IC
Plots of predicted vs. observed log 1/C for the three-variable QSAR model (with log P,, ,p& and R as descriptors, see text for meaning) (top, r = 0.928), and the one-variableh4LC QRAR model (bottom, r = 0.936). Reprinted from Ref. 30 withpermission ofthe American Chemical Society.
k Correlations betweenlog 1/C and k (or log k ) , for a seriesof26 p-substituted phenols eluted with: 0.04 M CTAB (top, r = 0.917), 0.04 M CTAB-lO% 2propanol (middle,r = 0.946), and methanol-water 40:60 (v/v), (bottom, r = 0.812). A diphenyl column was used. Reprintedfrom Ref. 30 with permissionof the American Chemical Society.
ure 9.1
ITY
p a r ~ e t e ris capable of describing the bioactivity of phenols, while three R needed to achieve a structural descriptors conventionally used in~ S A are similar correlation. f course, the addition of other structural p a r ~ e t e r s to k would firrther improvethe correlation with log116. Since the toxicity of phenols was measured at a physiological p 7.4,the good correlation between log 1/C and k obtained at
surprising. It is possiblethat because the micellar system close1 biological system, the same form (molecular, ionized or partially ionized) of the compound responsible for particular a biological responsealso exists in the chromato raphic column. In situations like this, SAR since measuring MLC retention ent molecular descriptors needed to bu Good relationshipsbetween the retentionin MLC and some biological activities of local anesthetics (bupivacaine, lidocaine, mepivacaine,prilocaine,procaineand tetracaine), such as anestheticpotency, concentration of compound that produces an effect similar to a reference concentration of cocaine, duration of the action, toxicity and time t iminate half the drug present in the body, havealso been reported [24]. omeanestheticactionsof barbiturates also correlatedwellwith the retention: ~ n i m u meffectivehypnoticdosein rabbits, molar drug concentration necessaryto reduce cell division, and molar drug conc~tration required to reduce 50% the inhibition of oxygen respiration onthe brain of a rat in vitro [25].
c)
~orre~ation betweenthe ~etentionOfDiuretics and their Site of Action within the Kidney
A finalexampleofthe capability of MLC to characterize bioactive substancesisthereportof mhdez et al. [3 l], related to the action of diuretics in the nephro variety of compounds with different chemical structures have beendescribed to act as diuretics. compoundsenhancerenalexcretionofwaterandelectrolytesthrough interference of the mechanisms of ionic transport all along the nephron, which is constituted of a glomerule and a long tubule where a filtration process takes place (Fig. 9.13). Each individual segment in the nephron
(proximal tubuleyloop of Henle, distal convolutedtubule and collecting duct) has a different function, and different diuretics possess specific sites of action. Even moreythe action of a diuretic produces a characteristic profile regarding the excretion of water and electrolytes, and such profile may suggest preciselyits site of action within the nephron. To define these sites of action in vivo and invitro studies are usually pe~ormedin humans and laboratory animals. According to their action,the compounds are classified as hgh,intermediate and lowefficacy diuretics.
. l 3 Sites of action along the nephron and retention factors for some diuretics eluted with a0.03 M SDS mobile phase at pH 7.
The therapeutical action of diuretics has been attributed to their hydrophobic character, amongother properties. As commented, the therapeutical and physiological classifications of diuretics are correlated. The site of action for high efficacy diuretics is the loop of Henley for intermediate efficacy diuretics is the distal tubule and for low eEf1cacy diuretics is the proximal and distal tubules and the collecting duct. It has been shown that LC with SDS mobile phases and C 18 columns offers a scale for hydrophobicity, which leadsto a further correlation between the retention andthe sites of action of diuretics within the nephron. The site of action of those dmretics showing the lower retentionin MLC is the proximal
Y
tubule (acetazolamide) and the loop of enle (loop diuretics: burnetanide, furosemide, ethacrynicacid), for those withan inte~ediateretention is the distal tubule (thiazides: benzothiazide, chlorthalidone, bendro~umet~azide, d~ydrochlorothiazide, xipmide),and for those witha long retention isthe collecting duct (potassium sparing diuretics: amiloride, triamterene, spironolactone). However, hydrochlorothiazide, an intermediate diuretic, showed a retention similarto high efficacy diuretics (acting in the loop of Henle). T h s behavior may be in accordance with the fact that the nephron. Thus, it competes hydrochlorothiazide acts in other sites within with uric acid for the secretion of organic acids at the level of the proximal transporter system. The chromatographicretention of diuretics decreased as the concentration of S S inmobile phase increased,whichmade r times closer to each Among the micellar mobile phases, 0.03 gave the best correlations between retention factors and site of action. It is interesting to note that the correlations hold in spite of the largely diverse chemical structures of diuretics. The order of retention attained with other RPLC systems was checked to be completely mobile phases in conventional different fromthat of MLC, and a correlation with regard to physiological properties could not bedrawn. *
LC should be developed further as a means of characterizing bioactive substances. The chromatographic system is dynamic in nature, which suggeststhe possibility of using MLC to mimic physiological systems such as the nephron, where both hydrophobic and electrostatic interactions, as well as kinetic phenomena can be important.
1. C.Hansch, in Drug Design, E.J. Ariens, ed., Vol. 1, Academic New York, 1971. 2. C.Hansch and A. Leo, ~ u ~ s t i t uConstants ~nt for Correla~on Ana~sis in C ~ e ~ ~ sand t r Biology, y Wiley-Interscience, NewYork, 1979.
atogr. A, 656: 4 17 (1993).
,New York, 1987.
Lo~arithmof k and ~ o m o l o ~ uumber e in MLC is not ine ear¶ Anal. Chem., 63: 1808 (1991). aliszan, ~uantitative Structure-~etention ~ e l a t i o ~ s hAnal. ~s, Chem.) 64: 619A (1992). lvare~-Builla, J. El~uero and J.C. of Octanol-Water arti it ion Coe~cients with city M e a s u r e ~ ~ nobtained ts by MLC, Anal. Chem., 59:
Octanol-Water Coe~cientswith ~ydrophobicityfor ~ o ~ c y c l i c i a627 , (1992). Aro~atic-~ydrocarbons byMLC, C h r o ~ a t ~ ~ r a p h34: Lavine, A.J. m i t e and J.H. an, Solute ~etentionin matogr., 542: 29 (1991). .Breyer, ~uantitationof ~ydrophobicitywith MLC¶Anal. Chem., 61: 1040 (1989).
astor and S. Vera, A Statistical Study the Correlation between k or log k and log P,,, for a Group of Benzene and Naphthalene Derivatives in MLC using a C l 8 Column, Chromatographia, 40: 185 (1995). 18. .C. Garcia Alvarez-Coque an Villanueva re for Evaluation the lockers in ~har~aceuticalsusing ~ y b r i d Micellar Mobile Phases, J
17.
of
19.
20.
21. 22.
23.
24.
25.
26.
Study of the k or log k-log P,, Correlation for a Group of Benzene Derivatives and Po~cyclic Aromatic ~ydrocarbonsin AdLC with a C8 Column, J: Chromato~r. A, 687: 233 1994). Catala Icardo andM.C.Garcia Alvarezn ~ydrophobicityof Amino Acids and Retention Datain RPLC withMicellar Eluents, Chromatographia,41: 455 (1995). edina Hernandez and S. Sagrado, C~romatographic Quantl~cationof ~ydrophobicityusing ~ i c e l l a rMobile Phases, J Chromatogr. A, 718: 273 (1995). elgado, M.J. Shchez, V. Gonzalez Garcia elations between Retention Data lycyclic ocarbons in MLC and SeveralMolecular Descriptors, 345: 748 (1993). .Villanueva Camailas,S. Sagrado and iindez, Quantitative Retention-Structure andRetentionActivity Relationship Studiesof Ionic and Nonionic Catecholamines by MLC, Chromatographia~46: 605 (1997). s Escuder Gilabert, S. Sagrado, R.M. Villanueva C ~ a i l a and edina Hernandez,Quantitative Retention-Structure and Reten ctivity Rela~ionshipStudies of Local Anesthetics by MLC, Anal. Chem, 70: 28 (1998). M. Cuenca Benito, S. Sagrado, R.M. Villanueva Carnailas and edina Hernandez, Quantitative Retention-Structure andReten~onActivity Rela~onshipsof Barbiturates by nnZC, J Chromatogr. A, 814: 121 (1998). .A. do, M.J. Sanchez, V. Gonzalez and F.Garcia ofRetentionfor Substituted and ~nsubstituted ont Polycyclic Aromatic ~ydrocarbonsin nnZC in thePresence of Organic Modl~ers,J Chromatogr. A, 697: 7 1(1995).
3
27. S. Yang and .G, fialedi, Linear Solvation EnergyRelationsh~sin ~C and ~ i c e l l a r~ l e c t r o ~ n e t i cC a ~ i l l a ~C h r o m a t o ~ r a ~ h y , J: C h r o ~ a t o ~ r . 692: A , 30 1 (1995). .F. Zou, Y.K. Zhang and P.C. h , Eflect of~olecularStruc~ureon the Solute-~icelleand Solute-~tationaryPhase ind din^ Constants in “ L C , Anal. Chim. Acta, 310: 46 1 (1995). Fender, Interactions and Kinetics i n ~ e m b r a n e ~ i m e Systems, tic U. Rev. Phys. Chem., 35: 137 (1984). .Breyer, J.K. Strasters and M.G. Khaledi, ~uantitativeRetention ~ i o l o ~ i c a l - A c t i v i ~ R e l aStudy ~ i o nb ~y h~~L CAnal. , Chem., 63: 828 (1991). ina Hernhndez, E. Bonet Dorningo, G. h i s Ramos and cia Alvarez-Coque, On the ~ e t e ~ ~ o n o ~ ~ in i uMLC retics and their Site of Action within the ~ e ~ h r oAnal. n , Lett., 26: 1881 (1993).
The first applications of MicellarLiquid Chromatography ( analytical chemistrywere published in 1984. The reports showed the interest of using micellar mobile phases in the analysis of plasma [l], proteins [2] and pesticides [3]. However, the research on MLC has been largely dedicated to the retention mechanism inside the chromatographic system and the relationslxps between the retention behavior and hydrophobicity of solutes. Indeed, micellarmobile phases are quite interesting from a mechanistic standpoint, but some practical analytical utility had to bedemonstratedbeforetheycould be considered as an alternative to the more traditional aqueous-organic mobilephases. In the nineties, the number of analytical procedures in MLC proposedbyseveral authors increased,especially in the field of the and phamaceutical detemination of drugs inphysiologicalfluids preparations. The analysis of othertypes of samples wasless frequent. This chapter shows the experimental procedures to befollowedinhandling micellar mobile phases for the analysis of several types of samples, with the exception of physiological fluids, which will be considered in the next chapter, owing to the particular characteristics and interest of these samples.
A practical requirement of any micellarmobile phase is its compatibilitywith the sample matrix. Micellareluents can be used to dissolve samplesor extract analytes. Sudactants can cosolubilize nonpolar and polar compounds, derivatization reagents andproducts. They can also induce favorable shifts in the equilibrium constants and spectral properties, inhibit undesirable reactions,stabilize reaction intermediates, and expedite reactions by meansof micellar catalysis [4,5]. Other potential advantages
ofmicellareluentsinclude a muchlower cost than traditional liquid chromatographic grade solvents, adjustable solvent strength andlower i~itabilityand toxicity. u~risingly7almost all the applied work in LC to date appears to haveinvolvedonlyionicmicobile phases composedof either anionic sodium dodecyl sulfate or cationic he~adecyltrimethyldodecyltrimethyl), and a~onium bromide or uncharged nonionic onium of (Dbromide ants, such as NeodolB 9 1-6 and polyoxyethylene(~3)dodecanol 35) are exceptions. It is believed that whenmicellarmobileph chromatographic columns deteriorate very easily. adequate experimental me tho do lo^ (see Chapter 4), no of thechromatographicperformance willoccur. Hundreds of injections can be madewithout modi~cationof the retention ofthe compounds or pressure buildup in the chromatographic system. In our laboratories, we have used the same columnsfor a year or even longer(at least 600 injections)7without any apparent deterioration. The chromatograp~csystem should comprise a p r e c o l m located between the pump and the injector, packed with silica gel, in order to saturate the mobile phase with silicic acid, thus increasing the lifetime ofthe analytical column. Caution must be taken to insure that the steady-state conhtion required for reproducible chromatography is reached. Before the sample is injected into the chromatograph, the system must beequilibrated with the micellarmobile phase. Special care is required to workwith micellar mobile phases such as those given in Chapter 4 (Section TV). Finally, it should be reminded that for micellar solutions of ionic rfactants, the temperature of the column should always be above the affi point (e.g.,15°C for SDS), in order to avoid clogging and possibly ruining the column. Nonionic surfactants, instead, have a cloud point temperature at which phaseseparation occurs (see Chapter 2). In this case, l chromatographic work should be conducted below this temperature. owever, this depends onthe on cent ration of surfactant and is usually very high (e.g.,approximately 100°C for aqueous 1-6% Brij-35).
.
a) Use of
on ionic Surfactant in the Sepa~a~on of rotei ins
haseLiquidChromatography ( LC) is an impo~anttool in rotein chemistry. E x ~ i n a t i o of n sorptionisotherms alcoholic uffers desorb not didproteins near physiological systemsy containing a poly(ethoxy alcohol) surfactant did not desorb 2, but they did at pH 7 with concentrations of s u ~ a c t apparently well above the critical micellar concentration (cmc) [2]. It has been proposedthat a necessary condition for the desorption of a protein from a surface is that the surface tension ofthe solvent falls between that of the protein andthe surface [6]. This condition is fulfilled for many proteins with ctmt solutions near conditionsof physiological and ionic strength. erefore, it was expected that separations of protei ould be achieved in these conditions. It was effectively found that a mobile phase of Neodol8 9 1-6, a nonionic surfactant, separated proteins on areversed-phasecolumn at [2]. Neodol8 91-6 is a blend of C9, C 10 and an average of 6 moles of ethylene oxideper m wing to the several chain-len~hsand degrees of surfactant has a cmc range rather than a fixed value. proteins were injected into a C8 column, their retention volumes fell into three broad groups, depending onthe isocratic concentration of Neodol9 1-6 ate required for elution. Table 10.1 shows these groups (low, i ~ t e ~ e ~and high retained proteins), together with the molecular weights, isoelectric points, and average hydrophobicities of the proteins. The chromatographic behavior was rationalizedin terms of favorable van der Waals attraction of the proteins by the nonionic micelles and the reduced surface tension provided by the aqueous micellar mobile wcver, littlecorrelation of the retention withthe molecular weight was observed, although the most strongly retained proteins were of low
3
molecular weight. The accessibility of these proteinsto a larger pe ofthepacking's surface area mayaccount for the observedretention. Sudace charge also has minimal effect, as indicated by the occurrence of basic proteins in all three groups. The average hy~rophobicityhas been correlated with protein properties, such as solubility, aggregation phenomena, and thermal stability, but no direct or inverse correlation was observed inthis case. 10.1 Properties ofProteins of Low, Intermediate and HighRetention in SurfactantReversed Phase System @H ' 7 ) [2].
Catalase Carbonic anhydrase
58
5.7
1.04 32
7.3
Ferritin
800
5.0
Apoferritin
24
Bovine serum albumin
65
4.8
~yroglobulin
335
4.6
C h ~ o t9.2 ~sinogen bonuclease
23
Lysozyrne Cytochrome c
14
11.0
0.97
13
10.0
1.11
~-~actoglobuli~
18
5.2
1.23
1.05 1.12 1.05 9.4
LYTICAL USE OF MICEL
7
I
Figure 10.1 Gradient micellar chromatographyof protein mixture:1.ovalbumin, 2. bovine serum albumin, 3. thyroglobulin, 4. chymotrypsinogen, 5. (3-lactoglobulin, G. lysozyme. 0.05 M phosphate at pH7. Reprinted Column: Supelcosil LC-8; mobile phase: Neodol91-G in from Ref.2 with permissionof the American Chemical Society.
Unlike c ~ o ~ t o g r a p of h y most low molecular weight solutes, where increases in micelle concentration produce reductions in retention, small increases in protein decreases of surfactantconcentration caused exponential retention. Therefore, to obtain adequate separations of mixtures of selected proteins, shallow gradients were employed.Fig. 10.1 illustrates the potential of micellar mobilephases for separating components witha wide range of properties, as is comonly the case for biological isolates.For some purposes, isocratic elution gave sufficientseparation, although caution was taken to assure that no protein remained sorbed on the support. hexample of separation of proteins is given in Fig. 102 .
Chromatogram of a beaf heart myoglobin preparation. First peak is myoglobin, second peak is apomyoglobin (tentative), later peaks are iron porphorins (tentative). Mobile phase: NeodolG3 9 1-6 in 0.05 M phosphate at pH 7. Reprinted from Ref. 2 with permission of the American Chemical Society.
b) ~ e t e r ~ i n a t i o n o ~ ~ e c o ~ bui n ~a Growth nat n ~ o r ~ o n e A method was described for the ];C dete~inationofrecombinant methionylaspa~yl-humangrowth hormone ( -HGH) in.E~cherichiacoli (E. coli)f e ~ e n ~ t i broth o n [7], which utiliz obile phases containing the anionic sudactant SDS and l-propanol, under micellar conditions. A C4 column was used at 60°C for the separation. The methodology is directly analysis of samples solubilize^ via sul~tolysisin. the ,and offers superior resolutionin comparison with chromatography in the absence of the surfactant.
The analysis of heterologous proteins in recombinant hosts, such as E. coli, presents many challenges to the analytical biochemist. The cells must be lysed and the inclusion bodies solubilizedprior to ~uantification. Cell lysis and protein solubilization can accom~lished be chemically through S. A s theproteinsininclusion bodies can exist as a distri~utionof forms, such as covalent and noncovalent polymers, it is crucial to convert the target protein into a single molecular entityprior to analysis. This canbe achieved by unfoldingthe proteins and disrupting the inter- and intr~oleculardisulfide bonds via reductionor sulfitolysis. The complexity of the matrix adds to the difficulty inthe determination of the recombinant protein,as both the whole cell and the inclusion bodiescan also contain nucleicacids, salts, lipids, and other host molecules, in addition to proteinaceous material. In fact, the employment of ionic surfactants, such as S traditionally been avoided in chromatograp~c the analysis of proteins owing to the generation of a strongly denaturing e n v i r o ~ e n t .However, for the dete~ination of recombinant proteins in fermentation broth, the characteristics of these surfactants are useh1 for the enhan~ementof the solubility of unfolded denatured proteins, the elimination of irreversible a d s o ~ t i o non the stationary phase and the facilitation of uni selectivity. Inthe procedure reported for the dete~natioo n sulfitolysis solubilization reagent was prepared with SDS, Tris, anhydrous sodium sulfite, anhydrous potassium tetrathonate and disodium ethylenedi~~etetraacetate &hydrate at pH 8.5-8.7. The sulfitolysis at room process was complete within 6-8 hours under these conditions temperature. Next, the reaction was quenched andthe solubilized protein stabilized by adjusting thepH of the solutionto 5.8-6.2 with maleicacid. The optimum mobile phasepH for the determination appeared to benear 6.4. The recovery,evaluatedbyspik volume of fermentation broth sample with a nonsulfitolyz standard solution at four levels ranging from 100 to 800 pg/ml, mounted in average 103.5%. This indicated complete extraction of the sulfitolyzed protein from thebroth.
3
Folylpolyglutamate hydrolase specifically catalyzes the hydrol~iccleavage of peptide bonds involving the y-carboxyl group of glutqmic acid. folates in food occur mainly as polyglut~ateswhich are poorly absorbed, much of the interest in these enzymes has centered on their role in the bioavailability of dietary folates. However, folates may also play a role intra~llularlyin the control of one-carbon metabolism. ~tracellularfolates are almost entirely polyglutamyl derivativesand changes in poly~~utamate chain-len~hsare known to have dramatic effects onthe kinetics of a number of folate-requiring enzymes. Early microbiological assays of folylpolyglu~atehydrolase were sensitive, but also lengthy, hard to reproduce, subject to interference and lacked of specifity. The proposal ofa more simple andsensitive method for the ~ u ~ t i t a t i ~etermination ve of the activity of this enzyme in crude tissue extracts was thus attractive. The newprocedure was based on separation of folate analogue mono- andpolygluta~ateson a C 18 column, S in water as the mobile phase under isocratic conditions [S]. ~ t e ~ substances e ~ g in tissue extracts were removed by gelfiltration on centri~gally-elutedminicolumns of Sephadex6-25, prior to incubation of p o l y g l u ~ a t esubstrate with tissueextract hydrolase. Reactions were t e ~ n a t e by d denaturationofthe enzyme inSDS, which su~se~uently served as the rnicellar solvent systemfor the chromatographic separation. i g l u ~ aare ~ usually s the substratesof choicefor the qu~titation of hydrolase activity since, unldse longer chain-length cogeners,they have only a single y-glutamyl peptide bond. Thus, the product of hydrolysis c ~ o subse~uently t become a substrate. Also, the analysisis not complicated bythe endopeptidase activity associated with somehydrolases. The procedure was performed with 5,8-dideaza-isopteroy1-y-g1u~y1-Lglutamic acid ( 1 ~ ~ - 6 1 because u) of its relative ease of availability. Fig. 10.3A shows the complete separation of the peak at the solvent front. Progressive hydrolysis of the substrate is shown in Figs. 10.3B-D. The baseline resolution ofthe substrate &om the Q, permits quantitative assessment of hydrolysis rate. The assay was linear over a period of at least 2 h. Moreover, the amount ofproduct formed was directlyproportional to the amount of added
extract.The results indicated that the method was suitable for the dete~inationof folylpoly~lut~ate hydrolase activity in cell-free extracts from' diEerent mouse tissues, as well as extracts from a wide variety of sources. otvcnt
rant L
I
Figure 10.3
Hydrolysis at
37°C of WHQ-Glu by crude
extractsfrommousekidney. Hydrolysis time:0 min (A), 15 min (B), 60 min (C), and 120 min (D). Reactions were stopped by introduction of S D S , b e f obr e i n g ~hromatographedisocratically with 0.2 M SDS. Reprinted from Ref. 8 with permissionof Elsevier.
0
1.a
l
n1'
l
I
Themicellarsolventsystemusedin this assay has several advantages, in addition to rapid separation: S S is an effective-denaturing agent and,as such, canbe usedto stop reactio precipitation withtrichloroaceticacid or heating.Since also has a unique solubilizing power, can it be used fordxect injection of concentrated protein solutions into the W L C system, without time- ons sum in^ steps to remove protein precipitates or extraction of folate analogues. Therefore, the assay is simple,rapid,inexpensive,andapplicable to crudehydrolase pr~parations.
L
icellarmobile phases canreplace,inmanyinstances,conventional aqueous-organic mobile phases in the of control p ~ ~ a c e u t i cprepar~tions al withgood results. A beliefexists that the analytical procedures using micellar eluents are inferior, owing to the frequent low e~cienciesof the chromatographic peaks. This is not always the case, as was demonstrated inacomparativestudyof the p e r f o ~ ~ of c e LC with mic~llarand aqueous-organicmobilephases,intheanalysis of pha~aceuticals [9). c o n ~ i ~ P-blockers ng
LC analysis of pha~aceuticalpreparations, the samples are usually treated with a cellar solution. The drugs are easily e ~ r a c t e din this medium, which produces an important reduction in the time employed in the p~eparationof the sample. The solutions ofthe pha~aceuticalscan chroma tog rap^ withoutanyothertreatment than beinjectedintothe filtration. ome features of analytical procedures developed for explained below, including samplepreparation, derivatization of the drugs and o~timizationof thechromatograp~icseparation.
alytical procedures havebeen reported for the rmaceuticals presented as tablets, pills, capsules, gel sprays, suspensions and oily injectable doses. Although the drugs are easily solubilized in a micellar solution,the excipientsare frequently not soluble in this medium. It may also be convenient to treat, first, the pha~aceuticals with a small ~ o u noft ethanol andto add, afterwards, a micellar solution to assure the extraction of the drug. Inany case, filtration solutions is required previously to its injectionin the c~omatograph. operation (e.g.,through can however be performeddirectly on the a u t o s ~ p l e vials r d for the ~ i ~ e r e n t Teflon 0.45 pm membranes). Ther e c o ~ e n d e treatments types of pharmaceutical samples are the following: Ta~letsand Pills. Five to ten tablets or pills should be weighed, powdered and homogenized in a mortar, a portion taken, weighed and dissolved in 0.05-0.1 using an ultrasonic bath. Watercanbe use but the last dilution should preferably be made with the solution used as mobile phase, in order to reduce the noise at the b e g i ~ ~ofgthe c h r o m a t o g r ~when the drug solution is injected. similar procedure should be followed with the contents of capsules, whose weight can be determined by the di rence between the weight of the filled and empty capsules. The capsul hould be carefully cleaned to obtain an accurate weight of the capsule contents. pharmaceuticals, the capsules can be dissolved in the micella without opening.In this case, a morecomplex chro obtained due to the presence of the capsule materials pigments). adequate amount should be weighed and dissolved in 0.05-0.1
~ r o p s~, ~ p o u land es and mixed with0.05-0.
. An aliquot of the solution shoul
solution.
uspensions should bem e c h ~ c a l l ystirred an ltrasonic bath in alternate periods, before talung the S liquot of the homogenized suspension should be dissolve
SDS, using an ultrasonic bath. Clear micellar solutions are easily obtained. Oily ~ ~ j e c t a ~Doses. Ze Most often the individual dose is contained in a breakable vial. The oily contents of two or three vials can be mixed, an aliquot is taken and a comparable volume of 0.1 M SDS is added. A homogeneous emulsion is formed by alternating 5 min periods of mechanical stirring and ultrasonic bath. The emulsion should be diluted with 0.1 M SDS to obtain a clear solution foranalysis.
The modi~cationof the polarity of some drugs through the f o ~ a t i o nof derivatives may be convenient to increase their retention in the c~omatographiccolumn, or enhancethe selectivity ofthe separation in the analysis of mixtures of drugs in complex matrices. The absorption bands of the analytes are shifted to longer wavelengths upon chromogenic derivatization andthis can facilitatetheir detection. Examples that make use of the advantages of precolumn derivatization have been reported in MLC for the dete~inationof s u l f o n ~ i d e and s amino acids. The derivati~ation reactions were readily performed in a micellar medium of SDS, leading to rapid and simple procedures to control the compounds in pha~aceutical preparations. U)
~ e t e r m i ~ u tof i o~~u ~ o ~ a m i d e s
iazoti~ationandcouplingwiththe ~ra~on-Marshallreagent (W(lnaphthy1)ethylenediamine dihydrochloride, NED), combined with spectrophotometricmeasurement, is perhapsoneof the most popular procedures for the d e t e ~ ~ a t i of o narylmines. In the analytical protocoles, theamine group is diazotizedwithsodiumnitrite, the excessnitrite eliminated withs u l f h c acid, and the diazonium ion produced coupled with NED to form an azo dye (Fig. 10.4). The azo dyes show, inacid medium, an absorption maximum close to 550 nm and a high molar absorptivity (40,000-50,000 mol-’ 1 cm”). The maximum wavelength shifts to 490 nm at pH>4 and shows lowerabsorptivity.
N=N+
I
Diazotization of sulphonamides naphthy1)ethylen~diaminedihydrochloride.
and
coupling with
.N-(l-
It has been demonstrated that these spectrophotometric analytical procedures are largelyimprovedwhenperformedin an medium, where the coupling reactions of the diazotized a are positively catalyzed (the reactionsare almost instantaneous), and the protonation constants of the azo dyes are shifted to higher pH values 1. In nomicellar solution, diazonium ionsare formed at pH <:1, whereas > 2 is necessary for coupling at a reasonablespeed. A ~ ~ modi~cationof pH is usually made in order to measure the absorbance of the azo dyesintheirprotonated forms. In an S S micellarmedium, the combined effects of micellarcatalysis ofthe coup reaction and the earlier azo dyes make these pH changes u~ecessary, d measurement steps can be carried out in a 0. sults from the addition of the coupling reagent to the sed to diazotize the arylamines. The pharmaceutical industry commercializes a great variety of formulations that contain sulfonamides, which are used as antibacterial agents in medicine and veterinarypractice. When some sulfonamides were directly chromatographed with0.1 a M SDS mobile phaseat pH 7, the drugs eluted with the void volume. AtpH 3, broad peaks with retentionfactors, k, in the 3-4.5 range were observed. In these conditions, the analysis of formulations containing a sulfonamide and other drugs was not possible. The formation ofthe azo dyes of sulfonamides increased the retention of the
h
rugs, and p e m i ~ e dthe use of a micellar eluent of and l - p e n t ~ oof l . . ion strength. This im~rovedthe selectivity of the det azo dyes formedi ~ e d i a t e l and y werestable during S even.when exposed to lightandoxygen,except sulfacet~ide. The radationof this azo dyewasobservedby the diminution of the chromatogra~hicpeak at 4-5 min, and the appearance of a secon a s ~ o ~retention er time. If the solution of the sulfacet~ideaz injected before 2 h from its f o ~ a t i o nonly y one peak was observed.
b) ~ e t e ~ ~ i n a tofAmino ion Acids ofamino acids withspectrophotometric mation of derivativesy becaus recolumn derivatizationis usually S thederivatizationre It reacts with prima^ ~ i n groups o .5 androom tempera~reto f o m ig. 10.5). The derivatives show hly ~uorescent,with excitation w a v e l e ~ ~ath340 nm and emission at 445 n m . more extensively used than other thols for the d mercaptoethanol isoindoles are unstable. stability of isoindoles is of improved when ~-acetyl-L-cystei~e is instead used in0 acidsare found in diverse~ h a ~ a c e u t i c a l s , matological preparations, andappetite s t ~ u l ~ t s
protonationoftheircarboxylate e derivatization wa equimolar n reagent with
group.
imumresolution mixin by
molar ratio 2 10). The concentration of isoindoles are less stable at increasing reproducibility was good ifthe injection into the chromatograph was made one ~ n u t after e mixing the reagents.
2
+
+ OP
C
2
Derivatization of amino acids withOPA and thiol to form l-alkylthio-2-al~yl substituted isoindoles.
III.3. ~ ~ t i ~ i ~o a~ t ~i ooP~~ a is e~~ o e~ ~ o s i t i o ~ Several variables should be considered in the development of an procedure: the nature of surfactant and modifier, their concentration a surfactant solution is used as mobile phase, the retention of be adequately controlled through the addition ofa small ~ o u n t of alcohol, and through variation ofpcoho1 usually also improves the efficiency ofthe chromatographicp retention and efficiencyare temperature and io procedures are performed at room temper given by the con~binationof the su~actant ied as a separate variable.
The report commented above for the determination of sulfonamides in pharmaceuticals [l l] is a usefbl example of the development of an analytical procedure, where a sequential optimization is made. Next, the develop~entof a procedure forthe analysis of mixtures of~ - b l o c ~ eand rs diuretics [141 will show the usefulness of the inte~retive opti~zation strategy shownin Chapter 8, whch was assisted by the s o ~ a r e ~ (see Appendix I).
The first procedurewasdevelopedforthedetermination of several sulfonamides (sodium sulfacetamide, sulfadiazine, su~fagua~dine, sulfamethizole, sulfamethoxazole, and sulfanilamide), in different pharmaceutical preparatio~(tablets, pills, capsules, drops, and suspensions), after azo dye f o ~ a t i o n .The retention of the azo dyes was excessive (analysis times longer than 40 min) with pure micellar SDS mobile phases. The retention times were still high (> 25 min) after addition of l-propanol as modifier. Therefore, an alcohol giving a higher elution strength,such as l-pentanol, was preferred. In an anionic micellar solutionof S S, the primary amine group of s u l f o n ~ d eazo dyes is protonated in very weak acid media, whereas the protonation of the aryl-alkyl secondary amine group, in para position with respect to the azo bridge (Fig. 10.4), takes place usually in the 3.5-4.5 pH range? oneto two pH units higherthan in a nonmicellar solution[101. Thus, S, the single and double charged cationic forms of the pH 9 and pH <4, respectively. The higher retention obtained at decreasing pH indicated that the solute interacted more strongly with the anionic modified stationary phase than with the anionic micelles in the eluent. The stronger interaction also decreased the efficiency of the c~omatographicpeaks. Thus, pH 7 was selected for the mobile phase. SDS-l-pentanolmobilephasewasused, an enahybrid increasingconcentrationof SDS gavelowerretention,but also lower efficiency. A mobile phase containing a low concentration of SDS was thus n the other hand, an increasin~amount of l-penta ,together with increased efficiency. Finally, a 0.05 2.4% l-pentanol mobile phase was selected. With this mobile phase, the s u l f o n ~ d azo e dyes were eluted in lessthan 15 min, followingthe elution
C
~
order (k in parentheses): sodium sulfacetamide(4.9, sulf~ethizole(5 .0), sulfaguanidine (9.l), sulfamethoxazole (10.5), sulfadiazine (1l.8), and sulfanilamide (12.1). For the analysis ofthe most retained drugs, another eluent of higher elution strength was also used. Table 10.2 shows the composition, recoveries and reproducibilities obtainedintheanalyses of severalpharmaceuticalscontaining the sulfonamides. The analyses were performed by derivatizing five aliquots of the dissolvedpharmaceuticals.Therecoverieswithrespect tothe compositions given bythe manufac~rerswere usually closeto 100%. The the detection wavelength. accompanying compounds did not give anyatpeak These resultsare given as an example ofthe good performanceof anMLC procedure appliedto the control of pharmaceuticals. b) Interpretive ~ p t i m i z ~ t i oStrategy n
Several pharmaceutical formulations that contain a P-blocker, together with one or two diuretics, are commercialized. Some of these preparations also include the vasodilator hydralazine. The combination of the three types of drugs hasan additive eEect inthe treatment of hypertension, producing the d i ~ n u t i o nof arterial pressure through diverse mechanisms. The simultaneousadmmistrationofthese drugs is recommended,when an appropriate controlofhypertensionisnotpossiblewitheach of them separately. There are only a few reports on chromatographic procedures, where comb~ationsof a P-blocker and a diuretic are analyzed using the same aqueous-org~icmobile phase for both drugs. One example is a procedure recommended in the USP-XX~~ Pharmacopeia [151 for combinations of metoprolol and hydroc~orothia~idein tablets, where both sample preparation and chromatographic conditions are different for each drug. Sample preparation is especially tedous for metoprolol, since it involves dissolution in HGl with sonication and heating, filtration and triplicate extractionwithchloroform,eliminationofthe organic solventand redissolution. The possibility of using the same micellar mobile phase for the analysis of pharmaceuticals, where the P-blockers atenolol, metoprolol and oxprenolol, the diuretics amiloride,bendro~umethiazide,chlorthalidone and
2 m
!
fl
09
? fl
N
fl
2
4
M. (\i
2
t-4
N
6)
c'! M
2
M
!Y
r ; 0
hydrochlorothiazide, and the vasodilatorhydralazine, are associated, was studied [14). It was considered that a mobile phase allowing the complete separation of the eight drugs, in an appropriate analysis time, should be usekl for the analysis of pha~aceutical any containing two or more of these drugs. The retention of the drugs was checked to be accurately described by the equation:
ln(=c,+c,[M]+c,cp+c,[M]cp
(10.1)
cp the concentrationof modifier where [M] is the micelle concentration, and (v/v). The optimization of the resolutionwas p e r f o ~ e dusing the normalized product of the overlapped fractions: n
r = i=1
(10.2)
being:
oi = 1 -wyw;
(10.3)
where wi is the total area of a given peak, and wti the area of the peak overlapped by other peaks. Because of the excessive retention of metoprolol and oxprenolol in mobile phases containing low concentrationsSof 1 -propanol = (k100 and 178 for metoprolol and oxprenolol, respecti S mobile phase without modifier), theoptimization study was ,made within the concentration ranges 0.10-0.15 M SDS, and 5 4 5 % (vi;) 1 -propanol.
The contour map of global resolution depicted in Fig. 10.6 was d r a using ~ the chromatographic data (retention factors, ef~cienciesand factors) of the drugs obtained in five mobile phases (0. l
can be observed that the resolution is m a x and ~ scarcely ~ modified over a wide region of the variable space, between 5 and 12% l-propanol. In this f the eight drugs appeared well resolved. A mobile phase -7% l-propanol was selected for the analysis of the ince it permitted the dete~inationof the P-blockers, azine, withinthe same run, with analysis times below20 min. The chromatogram ofthe mixture of drugs for this mobile phase has already been shown in Chapter 8 (Fig. 8.19). For the analysis of other combinations of these or other drugs, the simple o~timizationmethod described will permitthe rapid selection of a suitable mobile phase. I
1
n
Contour map of global resolution forthe separation of a mixture of three pblockers (atenolol, metoprolol and oxprenolol), four diuretics (amiloride, bendro~umethiazide, chlo~halidoneand hydrochlorothiazide), and hydralazine, eluted with mobileofphases SDS and propanol at pH3. Reprinted from Ref. 13 with permission ofthe Royal Chemical Society.
LI
Y
icellar mobile phases are often claimed to be a real alt based on their solvent properties and different selectivity. LC literature, only a fewreports compare the performane analysis of samples when both types of mobile phases are U reports deals with the determination of nine P-blockers (atenolol, carteolol, celiprolol, labetalol, metoprolol,nadolol,oxprenolol, propranolol and timolol) in tablets, capsules and ophthalmic solutions, and represents an example wherethe MLC technique is really competitive [9]. lockers are used in the treatment of neurological, neuropsyc~~tric and cardiovascular disorders. The compounds are isoprenaline derivativesand contain an akanolamine side-chain terminating in a secondary aminogroup. Their wide range of hydrophobici~makes the C determination of mixtureswithaqueous-organicmobile phases owever, interesting results were obtained e in of the separation of P-blockers showedthat a mobile phase of S- 15% 1 -propanol at pH 3, gave a relatively large solvent strength, which permi~edthe elution of all P-blockers in lessthan 15 min. The addition of l-propanol and an acid medium greatly increased the efficiency of the chromatograp~cpeaks. Aqueous-organic mobile phasesof diverse com~ositionhave been recommended for the determination of different p-blockers rmaceuticals. The chromatographic procedures proposed in the U I1Pharmacopeia [151,British Pharmacopeia [161, and Analytical phases con~ining 3560% g Substances[171, employ aqueous mobile cetonitrile or methanol, buffered at pH 3-4 with phosphate. It was checked that 40:60 methanol-0.05 M phosphate at pH 3 elutedmost P-blockers in adequate times. Atenolol requi lower strength, such as 22:78 methanol-0.0 l its chromatogra~hicpeak overlapped with the noise at the head of the chromatogramwith the previouseluent. The retention of celiprolol, labetalol, metoprolol, oxprenolol, and propranolol, with the weaker mobile phase was however too high [9].
3
M l-4
c? d
52
W.
M M
0
N
0
N N N
=?
\9
v?
0 0. r4
0 N
S
d
0
VI,
P4
rc-
2
7"
0 0
rc-
m-4
m-4
0
N.
3-4
8z
!
7-4
M.
W
0
c?
0
N
M. r-4
22
N
v!
c?
! -
0 0
cn
. I . (
N
0
U
f!
U
3
m
0 0
U
r-4
2
52
00
0
d
.-
P-
r?
0
d-
22
2
0
M.
cn
N
M. l-4
0 v)
M l-4
9 v)
U
0
Ei d
Q)
r l
m:
M M
0
d
v;
0
f?
n
The chromatographic parameters (retention factors, efficiencies and ry factors) obtainedwith the selectedmicellar and methanolwater mobile phases are shown inTable 10.3. The elution order is the same in both cases, except for celiprolol, metoprololand timolol, which elutedat micellareluents. the otherhand, the peaksobtained opanol eluents had b r characteristics than the peaks in ic eluents, with a tenfoldincrease in. efficiency for some aks in the aqueous~rganicmobile phases were often very a s ~ e t r i c a l The . dete~inationof the whole setof drugs was possible with a unique micellar mobile phase, with retentiontimes of less than 15 min, whereas two different aqueous-organic mobilephases were neededto make the same analyses. ives the results of the analysis of pha~aceuticals containing several p-blockers, using micellar and methanol-water mobile phases. In both cases, the phar~aceuticalswere dissolved in the solution used as mobile phase. The analyses were made by qui~tuplicate,and an average value of the measurements was taken to calculate the amount of drug. The recoveries were usually in the 96-103% range for the micellar mobile phase, and inthe 89-103% range for the aqueous-organic eluent.The low recovery obtained for some p-blockers withthe aqueous-organicmobile phase was probably due to an incomplete extraction of the drug, when the S dissolved. The 107% recovery achieved with ial10 was checkedbyrepeated analyses of the
Table 10.5 shows the experimental conditions (stationary phases, mobile phase compositionsand detection wavelen~hs)and figures of merit (linear ranges and limitsof detection) of severalchromatographic procedures, that employ micellar mobile phasesfor the analysis of diverse procedures havealready been commented on this in chapter. aspects still deservebereviewed, as the use of ecial mobile phases containing two surfactants or microemulsions of S
ccc
m
0 0
E1
m Q)
cd
c,
d
0 0
G
m-4
4
2
2
n
d Q)
E! d m 52
i '
2 v, ? l
2
!
ctfn:
Q)
c,
cd
0
c, m m
? l
0 9 1 " . ?l
..
2 M.
2
? l
S ? l
5
5
? l
v) F-(
? l ? l
v)
2
d
2 Q
r-i
N
cu 0
N N
M N
I
..m
I
I
M l-4
S
13'
CB
N
.n
n
CB
eous quantitation of the activ ingredients c o ~ o n l y lets, acet~inophen,pseud ephedrine and chlorph cult if not impossible using conventional rent fclnctional groups and polarities. f phenol, and can be retained on a re only by using very weak aqueous-organicmobile phas a s e c o n d a ~amine derivative of benzyl alcohol is al chromatographic behavior similar to aceta~nophen. In contrast, heniramine is a nonpolar tertiary ~ i n requiring e very stron mobile phases or ion-pairing to elute in a reasonable amount of time. inophen is present at much lxgher levelsth or chlorpheniramine, and has a large molar absorptivity. The d i l e ~ in a developing a chromatographic procedure for the three compounds is to separate acetaminophen and pseudoephedrin~adequately, and still elute chlorpheniramine in a reasonable time. Theuseof an isocratic mixed micellar mobile phase succeeded inthis simultaneous dete~ination[201. column choice in parallels column choice in C. Indeveloping the ical method for the three drugs, C 18, C8 and cyano columns were all tried with anionic, cationic and nonionic surfactants, with varying degree of success, Chlorphenir~ine not beelutedfrom C18 o columns, unless the s u r f a c t ~ t his system was undesirable since ntration was greater than 0.2, the viscosity of the mobilephasecausedexcessive backpressure, and acetaminophenandpseudoephedrinewerenot suitably retainedon the lower concentrations of surfactants, it was not possible to taminophen and pseudoephedrine adequately, and c ~ o ~ h e n i r ~was i n retained e too long. therwise, when the cyano column was used with a mobile phase of Brij-35, pseud~phedrineeluted in the dead is system was however usefcll for the separation of aceta~inophenand c h l o r p h e ~ r ~ n Instead, e. when mobilephases of SD were a~emptedon the cyano column, acet~inophenand pse~doephedrine were adequately separated, but chlorpheniramine was retained too long, resulting in a broad rounded peak.
obtainedon the cyan o surfactants could yiel
this concentration,
r a better control of the separatio~. The ef~ciencyof the
Chromatogram of a pharmaceutical preparation: (A)acetaminophen, (B) pseudoephedrine, (C) chlorpheniramine, ( Column: cyano at 65OC; mobilephase: 18 g Brij@ 35 and 3.46 g §D§dissolved in 1 liter of water containing 4.5% of npropanol. Reprinted from Ref. 20 with permission of Elsevier.
37 b)
Use of a ~ i ~ r o e ~ ~ l sofiSDS o n and 1-~entanol as ~obile~
~
a
~
habolic steroids stimulate protein synthesis and produce a considerable weight increase. These compounds are used in specific dietarytherapies in l ~ case , pathological processes characterized abynegative ~ ~ o g e n b asuch some debilitating chronic diseases, during radiotherapy and after major surgery or trauma. The high hy~ophobicityof steroids makes them strongly of a C l 8 stationary phase.The associated withthe n o ~ o ~ alkyl ~ e chains d retentionof a group ofsteroidsusually a ~ i n i s t e r e d(dy~ogesterone, medroxyprogesterone, medroxy-progesterone acetate, nandrolone, nandrolone decanoate, progesterone, stanozolol, testosterone enanthate, and a pure micellar mobile testosterone propionate) was thus extremely high with S. Therefore, the elution strength hadto be increased through the addition of l-pentanolto the mobile phase. However,for low l -pentan01 contents, the retention factors were still excessively high. Only the on of '7% l-pentanol to 0.1 M SDS, whch created a microemulsion, gave adequate retentions (k < 14) and acceptable efficiencies [28]. In a water7% l-pentanol, the minim p e n t ~ o l - S ~ ternary S system, with concentration of SDS required to obtain a microemulsion was 0.065 Below this value, a emulsion was obtained. Almost any other compound present in the formulations eluted at the beginning of the chromatograms with a mobile phase of0. l M SDS-7% l-pentanol of high elutions t r e n ~ h . The proposed method is therefore strongly group-selectivefor the steroids, and interesting separations inside thegroup are also possible. any formulations containing steroids are injectable solutions. The used in these formulations isan important interference inthe analytical procedures reported in the literature, even inpr those ~harmacopeias[15, 161, that utilize spectrophotometry or aqueous-organic mobile phases. These procedures require previous separation ofthesteroids,whichfrequentlyleads to low and variable recoveries. With theMLC procedure, the samplepreparation step is simple, since the oily mediaare easily dissolved in the SDS solution and injected directly inthe chromatograph.
e
. IK 1. Pesticides Micellarchromatographyshowspromise for the separation of labile ~ ~ ~ b a mspecies a t e with metabolites of toxicological significance, which are commonlyemployed as pesticides, and hence,maybe present in industrial effluents. Conventional methodsfor the determination of dithiocarbamates are susceptibleto interferences and require careful sample cleanup procedures prior to derivatization, to obtain accurate and reproducible results.
3
4
5
A simple MLC proce~urewas described for the simultaneous determinationof five dithiocarbamates (sodium ~-methyldithiocarbamate, sodium~~-dimethyldithiocarbamate, a m m o n i ut emt r a m e t h y l e n e dithiocarbamate, sodium ethyldithiocarbamate and disodium ethylenebis~thiocarb~ate) [3],which employed Cl8 or cyano columns and micellar solutionsof ~ T A ~ - m e ~ a n o l buffered at pH 6.8 with 0.02 M phosphate. Pond-water samples were used to assess the potential of the analysis of e n v i r o ~ e nsamples. ~l For the analyses, pond water was filtered and passed through a C 18bondedcolumn to remove aromatic compounds. Resorcinol was added as an internal standard. A typical chromatogram is shown in Fig. 10.8,
Figure 10.8 Chromatogram of dithiocarbamate (DTC) salts (10 yg/ml) in spiked pond water: 1. Resorcinol (internal standard), 2. N-methyl-DTC, 3. N,N-dimethyl-DTC, 4. ammonium tetramethylene-DTC,5.sodiumN,N- diethyl-DTC.Column: Cl*; mobile phase: M CTAB-70% methanol in phosphate buEer at pH 6.8. Reprinted from Ref. 3 with permission of the Royal Chemical Society.
e ~ractic~l ~otential of nonionicwas
~ e ~ o n s t r aby t e the ~ use of
concentration o
ration betweeninjections. Th cedureproved to providegoo navarietyof tobacco S resultsin the routinedetermination o with average recoveries of95% and a detection limit of120 pg/g.
Chromatogram of a tobacco sampleat a 80 pg/m1 levelof maleic hydrazide. Column: Hypersil ODs; mobile phase: CTA€3 in phosphate buffer at pH 7. Reprinted from Ref. 26 with permission of Elsevier.
Adequatebufferstrength W important for peak shape sensitivity. ~nitially,a buffer of 0.0 1 phosphate was used. This buffer gave a reasonable peak shape for a small volume ofstandard compound, but distorted peaks, and hence, poor limits of detection were observed for tobacco samples. A possible source for this peak distortion could be the presenceof a highconcentrationof NaCl (e.g.,appro~imately5 produced from the acid-base neutralization, which caused slower so d i ~ s i o in n the injected solution than in bulk mobile phase. This was fbrther confirmed by obtaininga similarly distorted peak froma standard solution in NaCl. The peakshapeof in tobacco samples sphate). This also astrongerbuffer (0.04 M ssible causes of peak distortion, one of which may ralized tobacco solution may not be neutral.
With greater regulation of over-the-counter products by the Food andDrug A ~ i ~ s t r a t i o an ,facile analytical method for the determ~ationof sun screen agents incosmetics was desirable. Determinations basedon LC have been reported, but due to the complex nature of cosmetics, the preparation of the sample may requirean extraction step. A simple chromatographic procedure for the determination ofthe leading W rcial cosmetic products: 2-ethylhexyl-p-dime~ylaminoethylhexyl-~-methoxyci~~ate (EMC) and 2-hy&oxy= 4-methoxybenzophenone (oxyben~one)was described. The procedure takes advantage of the easy solubilizationof the samples in micellar solution, and employs a micellarmobile phase of S S-10% 2-propanol withgood resolution [23].The use ofexternal standards m~imizessample preparation even further. The analysis of parabens, utilized as antimicrobial agents in cosmetic preparations, was also demonstrated. The method was applied to a variety of cornmercialpreparations with a sun protection factor ranging from 4 to 15 spf.
~, and Acidity ~ ~ e cint HPLC s 1. P.Yarmchuk, S e l e c ~ v i EfJiciency, using Micellar Mobile Phases: Application to the Analysis of Drugs in Blood Plasma, Diss. Abstr. Int. B, 44: 2 15I ( l 984). arford and B.J. Sliwinski, MLC of Proteins, Anal. Chem., 56: 1554 (1984). irkbright andF.G. P, Muliins, Separation of Dithiocarba~ates by HPLC using a ~ i c e l l a r ~ o Phase, b i l e Analyst, 109: 493 (1984). ittal (editor)~Micelli~ation, So~ubiliza~on a~ Microe~ulsions, Vol. 2, Plenum, New York, 1976. 5. J.S.Esteve Romero, E.F. Simo Alfonso, M C Garcia ~ v a r e z - ~ o q u e and C. Ramis Ramos, MicellarEnhancedSpectrophotometric Determination of Organic Species, Trends Anal. Chem., 14: 29 (1 995).
6. C.J. Van Oss, D.R.Absolorn and A.W. Neurnann, Repulsive van der Waals Forces.II.Mechanism of Hydrophobic Chromatography~Sep. Sci. Technol.? 14: 305 (1979). s Human Growth 7. M.A. Strege andA L . Lagu, A n a ~ s i ofRecombinant Hormone in Escherichia Coli Fermentation ~ r o t h b y ~ i c eHPLC, llar J:Chromatogr. A, 705: 155 (1995). . Stratton, J.B. Hynes, D.G. Priest, M.T. 8. G.L. Asleson, Micellar HPLC Determinationof Folylpolyglutamate Hydrolase Activity, J. Chromatogr., 357: 183 (1986). 9. I. RapadoMartinez, M.C. Garcia Alvarez-Coque and Cmaiias, LCProcedure for theEvaluation Pharmaceuticals using Hybrid Micellar ~ o b i l e Phases, J: Chromatogr. A, 765: 221 (1997). Rornero, M.C. Garcia Alvarez-Coque and is 10. J.S. Esteve hrnos, For~ationRates andProtonation Constants of Azo Dyes in 38: 1285 a SodiumDodecyl S u ~ a t eMicellar Solu~ion, ~alanta? (1991). 11. M C . Garcia Alvarez-Coque, E.F. Sirno Alfonso,G. Rarnis J.S. Esteve Rornero,Micellar HPLC ~ e t e r ~ i n a t i ofnS ~ ~ o n ~ m i ~ e s in Pharmaceuticals aper Azo DyePrecolumn Deriva~zation, J:Pharm. Biomed. Anal., 13: 237 (1995). 12. M.C. Garcia Alvarez-Coque, M.J . MedinaHemimdez, Villanueva Carnaiias and C. Mongay Fembdez, Studies o Formation and Stability of Isoindoles derived from Amino Acids, 0-Phthalaldehyde and N-Acetyl-L-cysteine, Anal. Biochem., 180: 172 (1989). 13. M. Catala Icardo, M.J. Medina Hembdez and M. C.Garcia AlvarezCoque, Determination of Amino Acids by Micellar HPLCand Precolumn Derivati~ationwith 0-Phthalaldehyde and N-Acetyl-Lcysteine, J. Liq. Chroma r., 18: 2827 (1995). rcia Alvarez-Coque and R. 14. I. Rapado Martinez, M C Cmaiias, Performance of ~ i c e l l a r ~ o bPhases i l e in RPLC for the Analysis of Pha~maceuticals containing,&Blockersandother Anti~ypertensiveDrugs, Analyst, 121:1677 (1 996). XXII Revision, U.S. P ~ a ~ a c o p e i a l 15. The United States Pharmacopeia, Convention, Rockville, M 16. British ~harmacopeia,Stationary Office, London, 1988.
17. Analytical Pro~lesof Drug Substances, 1980-1990. orgerding andW.L. Hime, Characterizationa n d ~ v a l u a ~ of on 18. of on ionic Po~oxyethylene~23~dodecanol Micellar Mobile in W L C ,Anal. Chem., 57: 2183 (1985). C. Wu and B.Ghen, Analysis of Berberin 19. ~ i z o m and e Chinese Patent Medicines Xuebao, 21: 458 (1986). , ~imultaneous Analysis of Acetaminophen, 20. T A . PseudoneHydrochlorideandChlorpheniramineMaleate ina Cold Tablet using an Isocratic,MixedMicellar HPLCMobile Phase, matogr., 410: 206 (1987). U, U. Wangand , Lu, Determination of 21. Compound Aspirin Tablets by MLC, ~aowu- enx xi-~azhi, 9: 162 (1989). r, HPLC of ~tabilizedSolutions of~henothiazine 22 T.Jira and G. rmazie,46: 849 (1991). Drugs by ML . Tomasella, P. Zutingand L.J. Cline-Love, Determination of 23. screen Agents in Cosmetic Products by MLC, J. Chromato~r., *
24. lvarez-Coque, valuation of Diuretics in Pharmaceuticals by HPLC witha 0.05M~odium Dodecyl ~ u ~ a t e - 3 ~ Propanol Mobile Phase, Analyst, l 17: 843 (1992). 25.
~endro~umethiazide,Chlorthalidone, Spironolactone and . l l :7 1 1 Triamterene in Pharmaceuticals, J. Pharm. B i o ~ e dAnal., (1993). Determina~onof Maleic ~ydrazidein Tobacco by MLC, 26. togr., 595: 346 (1992). 27. containing Catecholamines by MLC with Spectrophotometric ~etection, Analyst, 120: 1767 (1995). 28. S. Torres Cartas, .C. Garcia Alvarez-Coque and C ~ ~ aDetermination s , ofAnabolic Steroids in Pharmaceuticals by
LC with a Microemulsion of odium ~ o d e c y l ~and u~ate Acta, 302: 163 (19 agrado and 29. M.J. Chromato~raphic~eterminationof Ca~einein Pharmaceu~cal 43: ~ormulationsusing ~ i c e l l aMobile r Phases, Chromatogra~h~a, 149 (1996). 30. ~~ectrophotometric ~etection, J:Liq. Chrom. C% Rel. Tec~nol.,19: 1957 (1996). 31. Gonzalez and A. rcia, E. of etr ram iuram ~ l s u ~ (Thi de at er by HPLC: Micellar versus Conventional "LC, C~romato32.
que, On the Retention of ~ i u r e t i c in s within the ~ e p h r o nAnal. , Lett., 26: (1993).
The analysis of drugs and their metabolites in physiological fluids is an important issue in pharmaceutical research, clinical chemistry, doping and food quality control and toxicology. Normally, in clinical chemistry, druga is monitored when any of the following conditionsexist: (i) the drug has a narrow therapeutic range, (ii) thereis a danger oftoxicity, (iii) thereis a lack The therapeutic of therapeutic effect, or (iv)there is aforensicneed. efficacy of many drugs is closely related to their concentration in blood and tissue, which depends on the dosage, route and frequency a~inistration. of ~ e r a p e u t i drug c monitoring may be necessary to adjust the dose of adrug to the patient's needs. Therefore, appropriate a n a l ~ i c a assays l that are simple and reliableare desired. Thesedeterminations have beengreatlyenhanced through the ( developmentofHigh-PerformanceLiquidChromatography tec~ology.However, the assay of drugs in physiological fluids presents many problems. ~requently,the drugs are inverylowconcentration, strongly boundto proteins and in a complex matrix where inte~erencefrom numerous endogeneous co~poundsis expected. For this reason, the dire~t ersed-PhaseLiquid ~hromatography(RPLC) withaqueousor bile phases, such as mixturesofwaterwithmethanol acetonitrile is usually not feasible. The high-molecular-massproteinsfound in these samples are particularly troublesome. When conventional RPLC silica supports are used, proteins tend to denature and precipitate the in injection valve orat the column head, thus producing obstruction of the interparticulate space and clogging of the system,. Ths leads to a rapid degradation ofc ~ o ~ t o g r a p ~ c 7
e and an increase in back pressure. fluids are those that contain a large fraction of protein, mainly blood, plasma and serum. ~erebrospinaland interstitial fluids, as well as urine, are in general more compatible with liquid chromatograp~csystems, due to their low protein content. The harmhl proteinaceous material must be removed from sample prior to injection inan s o ~ t i o nto the stationary p s t r e a ~sample ~ e prep~ationfor physiological fluids. is to reduce this step to the precipitation of proteins b acetic acid or sodium hydroxide, owever, very often, the drugs must also be e ~ r a c t e dfro preconcentrated beforeassay. Typically, the developed methods include S S such as proteinprecipitation,liquid-liquidor solid-phase extraction posable cartridges, and evaporation, prior to injection into the c procedures are time-consuming, require he and introduce additional sources of error, since complete recovery of the analytes and dilution. the use of robotics can allow complex sample preparation to be carried out with high precision minimal labor cost, the e~uipmentcost and development timeof such methods are only justifiable for cases where high sample throughput over an extended time period is expected. An additional problem is the use and disposal ofthe toxic solvents and chemicals usedin theextraction processes. Toxic solvents possess danger not only to the individual working with them but also to the e n v i r o ~ e n t . The complexity of these la~or-intensive methods have preventedtheir application for routine clinicaluse. revious samplepreparation steps consti~tein many applications the main share of the total bioanalytical method. For ~aboratorieswith a large number of routine samples for chemical analysis, this means that a lot of time must be devotedto sample work-upprocedures. This is largely the reasonwhymuchefforthasbeen put into the development of liquid chromatographic systems that can tole physiological fluids. Anyof the existing sample e n r i c ~ e n t a n d o radeproteinizat
injection one-step methods, and should preferably include the use of a suitable internalstandard to correct for possible errors in the overall procedure and incomplete recovery (even a simple deproteinization step may cause analyte losses due to drug protein binding).
everal authors haveintentionallyavoidedthesample preparation step entirely, by making a direct ection of the physiological samplein the chromatograp~csystem [l I. nimal sample handling reducesthe cost and time of the analyses, i~creasessample throughput, and decreases error sources owing to minimized risks of losses and chemical changes in the reducednumberof steps. This improves the limitsof S), and the accuracy and precision of the determinations. Anotheradvantageofferedisthelowsampledemand of onlyafew microliters. For blood analysis, for instance, ~ n g e ~ u n c samples ~ r e are adequate for analysis. This is an impo~antconsiderationin pediatric applications. There are, however, also occasions where direct injectio~ ~rinciples are unsuitable. en the analytes are chemically or biochemically transformed in the iological fluid, it may be necessaryto rapidly transfer a1 e n v i r o ~ e n tbyaliquid-liquid or li~uid-solid ,if onlya single specimenhas to be analyzed, it can be directly injected into a suitable chromatographic system; the arises when there is a series of samples with unstable compounds. cases, it may still be possible to stabilize the analytes by the addition of antioxidants andlor enzyme i ~ i b i t o r.s irect injection isalso limited by the fact that there isno e l ~ n a t i o n of interferences of both exogeneous and endogeneous compounds, and there is no preconcentration of the drug. Thus, frequently, large volumes of untreated ~hysiologicalfluids need to be injected,in. order to determin~ drugs
3
whose therapeutic ranges lie the in ng/mL range. Injectionof large volumes creates thefkrther problem of blockage of the analytical column and buildup of strongly retained endogeneous compounds. Also,because there is no a large signal sampleclean-up, the physiologicalmatrixwillproduce eclipsing the peaks of early eluting compounds. ~ptimizationshould be achievedby judicious selectionofthedetectionschemeanddetector wavelength.
The past decade has witnessed a proliferation of methods and media for direct injection of physiological fluids into€PLC systems. The numerous s ~ d i e sandapplicationsoftheprincipleofdirectinjection clearly demonstrate that the technique now has maturedto a stage where it can be used routinely in bioanalytical laboratories,for the automatedanalysis of a large number of samples[11. LCmedia are notadequate to tolerate direct injection of small analytes in protein-containing matrices. "here fore, special column packings have been designed to minimize the deleterious ef5ects of the adsorption ofproteins.Mostsystemsutilizeaprecolumn for the reception ofthe physiological fluids,trace enrichment and preliminary cleanup, before the analyte fractionis transferred to the analytical column. The p r e c o l u ~is changed regularlyto keep the system backpressure at normal levels. The most common solid phase forthe precolumn is silanized silica, but many other types have been used. The stability index used in these to a certain total studies is takenas the number of injections (corresponding volume of physiological fluid)that reduces the c ~ o m a t o g r a p ~efficiency c or, alternatively, increases the pressure by a given percentage. Systems which can handle the injection of25 mL total physiological sample volume, without the need to exchange the precolumn, have been reported. An impo~antfeature of direct injection is that because the same precolumn can be used for all assays in a series of specimens, the risk of nonreproducibili~of off-line procedures where a new precolumn is used for each sample, is eliminated.Great care should be taken in order to keep the endogeneous proteins solubilized during chromatography. Certain eluent
restrictions are associated with the use of precolumns to ensure that the (e.g., the concentration of the organic matrix components remain in solution modifier should be keptas low as possible, preferably below10%). The large differences in molecular magnitudes between drugs and macromoleculeshavebeenexploited.Oneinteresting approach is the pretreatment of small pore silica with plasma [2] and/or solutions con~ning bovine serum albumin ( SA) [3]. The pores are so small that the proteins do not penetrate into the support, and the macromoleculesare only adsorbed onto the external surface of the particles, leaving the largest part of the surface area unaltered. Another conceptis the use of a solid phase prepared by attaching a hydrophobic tripeptide to glycerylpropyl-derivatizedsmall pore silica [4].This packing, named internalsurface reversed-phase (IS exposes internal and external surfaces of the support: the former permits hydrophobic partitioning and latter the is a hydrophilic phase, nonadso~ti~e toward protein. The precolumn servesthe dual fhction of acting as a guard column, as well as effecting a preseparation of the analytes from the physiological fluid matrix. It can be used in two different ways: in the foreflush or backflush mode. Figure l 1.1 shows two common valve configurations [S]. In thefirst configuration (A), the extraction column is between the injector and the. column switching valve and is always operated in the foreflush mode. Solvent delivery to the extraction column is performed by pump 1 which pushes the matrix proteins to the waste. During drug transfer from the extraction column to the analytical column, pump 1 delivers the wash solvent to both columns connected in series. Next the proteins are washed off the extraction column and pump 2 separates the analytes inthe analytical column. In thesecond co~guration(B), the extraction columnis set across the switching valve. The initial positionoperates similarly to configuration A,but when the valveis actuated, solvent deliveryto both columns in series is performed by pump 2. As in co~lgurationA,the extraction column is forward flushedfor elution of the drug. Another con~gurationcan be derived fromB. Again theextraction column is set across the switching valve. At this valve, the line from the analytical pump (pump 2) and the line to the analytical column have their positionsswitched. Similar to configuration B, when the valve is
Alternative valve configurations: loading pump (pump l), analytical pump (pump 2), extraction column (column l),analyticalcolumn (column 2), extraction solvent (M), extraction column wash solvent (W), analytical mobile phase Reversing the Pump2and Column 2 connections will reverse the flow path in Column 1 during the drug transfer step, allowing solute enrichment at the head of Column 1. Reprinted from Ref. 5 with permission of the American Chemical Society.
m).
actuated, solvent deliveryto b columns is provided by pump the flow path along the extr n column is reversed thereby the e ~ t r a ~ i column on to e component ofinterest. analytes are not transfer analytical column passln~a n the contrary, since they are still retained at e backwashed. In t h s way, the analytes are enrlc on the top of the precolumn. The system can alsoconsist of two altem precolumns, which provides an efficient and time- savin^ operation o Thesampleisinjectedandloaded a purge phase, the a n a l ~ e sare ba uring the purge phase a new sample is is elutedon the analytical column, the n ckflushe~onto the anal~icalcolumn from the second rec column ith theset e c ~ i ~ u ethe s , lifetime of theanalytical column is so high ons in the low p1 range can be perfo st longer by using small injecti rge total volume of mobile phase mative ofdirectinjection can befound in the use LC columns and eluents capable of sol~bilizin~ the protei ical matrix, such as triethyla~oniumac ,or ase~plained next,solutions of a b o v e ~ critical e micellar concentration (cm choice for direct ~jection.
separation of the p r o t e ~ spresent in the ved to be themost promis~gand useful, in S possible to the chromatographic colu
injectionisused.Thecompatibilitywithconventional p a c ~ n g sis particularly attractive. icelles tend to bind proteins competitively[73, thereby releasing protein-bound drugs. Therefore, theseare free to partition intothe stationary phase, whereas the proteins rather than precipitating on the column, are solubili~edand swept away harmlessly, eluting with or shortly after the solvent front. Anotheradvantageis that surfactants are nontoxic, n o ~ a ~ a band l e relatively inexpensive, in comparison to aqueous-organic solvents. The use of surfactantsin h e c t injection isalso much less complex than column-switchingprocedures requireadditional inst~mentation (precolumns,switchingvalvesandpumps),and accurate and precise timing of valve switching for a successful separation. *
LC,untreated physiological fluids directly injected into
the possess no difficulty, allowing hundreds of repetitive serial injections with no increase in system pressure, no noticeable cloggingofthe in retention factors, or system injection portoranalytical column, no changes c o n t ~ i n a t i o nevident. However, it should be reminded that the analytical column should be protected with a guard column to saturate the micellar mobilephasewith silica. This guard columnshouldbecontrolled and replaced after repetitive injections the of p~ysiolo~ical samples. It should be u ~ e c e s s to a ~comment that samples must be filtered previously to injection to eliminate any particulate matter. Also, it is convenient to inject daily a probe solute, in order to check for possiblechangesinretention. The pressure ofthe system shouldalso be monitoredto evaluate the precipitation of proteinaceous material into the column.
has been applied extensively and seemsto be rather useful and valid fordifferentsolidphasesandmicellareluents. The anionic surfactant ~is the~most~commonly ) used for direct injection sodium dodecylsulfate ( of physiological fluids, but the nonionic surfactant polyoxy dodecyl ether (Brijm 35) has also been employed successful s ~ ~ a c t a ncause t s protein to precipitate and cannot be used.
A~emptshave been madeto use other surfactants, such as sodium decyl sulfate and sodium pentadecylsulfate [8]. However, it was o b s e ~ e d that none of these are able to quantitatively elute BSA when methanol is presentintheeluents. In fact, sodium pentadecyl sulfate has limited solubility in eluents containing less than 30% methanol (v/v concentrations oforganicsolvents are given).One n o n d e n a ~ ~ nanionic g surfac~t, sodium desoxycholate was found to elute BSA from model serum injections. A zwi~erionicsurfactant known as C 12 APS (3-(dimet~yldodecylonium propanesulfonate) hasalso been successfil in drug analysis [ g ] .
It is necessary to avoid conditionsthat are u~avorablefor micelle formation or for protein solubilization. Grohs etal. [8] studied the range of s u ~ a c ~ t and organic solventcompositionswhichwillpermitdirectinjectionof physiological samples onto C 18 chromatographic columns, and indicated that the eluents typically used in LC may only represent a subset of the surfactant-containing eluents that ill permit direct injection. The a u ~ o r s sing a model serum of0.05 M BSA and a mobile phase 15% methanol (the cmc of SDS was not reached), that good results and no column clogging occured after a hundred i~jections. However, some li~tationson the use of organic solvents as additives were found. Theselimitatio~are not based on considerations of micelle stability, but rather upon the need to favor surfactant-protein, as well as s u r f a c ~ t packing interactions. SDS interactions with both proteinand C 18-bonded phase appears to be too limited the in presence of relatively high amounts of methanol, to allow direct injection of physiologicalfluids. It is even more p r o b l e ~ a t with i ~ other alcohol modifiers. Several authors have reported problems ofpressure increases and irreproducible retention times, when low concentrations of surfactants are employed (below 0.05 M). Sentell et al. [101 used a micellar mobile phase of 0.02 M SDS-10% l-propanol in phosphate buflier at pH 3.5, for the determination of the diuretic bumetanide. After several injections of spiked serum, the operating pressure increased noticeablythe and retention time of the drug decreased. Using a similar mobile phase, Ameer and [1l] also observed that serial injections of plasma containing bumetanide
brought about shortening retention times of the analyte. This was e~plained as changes in thestationary phase characteristics produced bythe complex e appreciably i plasma matrix. The dete~inationo u m e t ~ d was of . The chromatogr increasing by the concentration nd, probably due to a better solubili~ati r and Kelly [121 attempted direct inj alidone and~rosemidewith a mobil S and 5% l-propanol, but arapidincreaseincolumn backpressure occurred, as well as loss of analyte sensitivity. ne reason to lirnit the amount of organic solvent in eluents ~esig~ed for direct injection might be the need to ma^^^ an adequate coating of surfactant onthereversed-phasepacking,andtherebyprevent 4). Protein materialcan only beeEe adsorption (see Chapter from the samples and the column packing material, if a S surfactant concentration is used in the mobile phase. For this reason, the concentration of the s u r f a c ~should t be maintained well above the cmc, the organic solvent content relatively low. The role and integ micelles in the mobile phaseis not clearat a low concentrationof (e.g.,10% propanol). d high concentration of alcohol e type of surfactant and modifier, their concentrations, the andp of the mobile phase have been identifiedas the key parametersthat can be a drug and the compone~ts varied to obtain the required resolution between of a physiological sample.
In most cases, the drug bound to proteins is displaced by the surfact released andlor monomers for ~ a r t i t i o ~ g CO injection s tdirect ~ t i owith n aphase. ~ ever incomplete, and doubled peaks can appear ascribed to the p r o t e ~ ” nd and unbound drug (this depends on the nature of the drug, protein binding and mobile phase composition). been given forthe ~uantitationof cephalosporins 13,141 and m e ~ o t r e ~ a t e ~
E151-
en Haginaka etal. [131 applied the direct injectionapproach to ation of ceph~osporins,using aG 18 column andan nds were not always observed as single peaks, the eluent. Fig. 11.2 sh noxime hemihy~rochloride ed with a0.08 M SDS authors claimed that, as the protein bindingof these drugs is a p p r o ~ ~ l y 80% and 7-8%, respectively, the two peaks of CMX might be dueto drug bound (the peak at a shorter retention time) and not bound (the peakat a longer retention time)to serum proteins. Acidi~cationof the serum sample !
Chromatograms of cefmenoxime hemihydrochlo~de (CMX) andcefotiam dihydrochloride (CTM) in: (A) distilled water, (B) serum sample, and (C) acidified serum sample. Mobile phase: 0.08 M SDS-8% 2-propanol in 0.047 M phosphate bufTer at pH 3.3. The figuresover the peaksrepresent the retentiontimes. Reprinted from Ref 13 with permission of theAmerican Chemical Society.
-CO
(pH 1.2)changed the two peaks ofCMX into a single peak, as shown inthe figure. Also, after ultrafiltration of the serum samples, the peak at longer retention time was only observed in all eluents used. the assumption On that the two peaks of CMX were due to bound and unbound drug, the protein binding estimated as the peak area ratio of the bound to total CMX, was 79%. %S suggested that the protein binding ofdrug a might bed e t e ~ i n e d For other drugs, or for cephalosporins and methotrexate in other conditions, the recoveries from physiological samples were often close to l OO%, which indicated that the protein-bound drug was completely displaced by the micellar phase. These examples also show that, for a strongly bound drug, the alteration of drug-protein binding by changing the conditions, such ,is required for the recovery and quantitation of the total drug.
.
A hitation of the direct injection method with micellar mobile phases is the high background signal of matrix the at the beginningof the chromatogram, whch may completely overlap the peaks of the analytes, resulting in a useless zone(Fig. l 1.3) [161, The profile of the background depends on the composition of the mobile phase andis due to the presenceof proteins and endogeneous compounds. The protein-surfactant complexes, excluded from the pores of the stationary phase support, appear as a broad band at the a solventfrontand an unidentifiedendogeneouscompoundproduces p r o ~ n e n tpeak that standsoutamongothersmaller peaks. Both may representoccasionally serious intederence, but can bereduced atan increased detection wavelength.The background signalof the matrix is also LC,even with previous separation steps [171. observed in conventional It may be convenient to decrease the retention of the protein band and the prom~entpeak, as much as possible, for the dete~inationof many compounds. Only if the elution of the drug is produced at the end of the
protein band, will the determina~ionbe possible. Drugs eluting at shorter times will needextraction procedures.
Figure 11.3 Chromatogramsofthe background of urine m a t r ~ e l u t e different d ~ ~ mobile phases [pH was 4.5 in all cases except for (b) where it was 6.91: (a) and (b) 0.05 M SDS, (c)0.15 M SDS, (d) 0.03 M SDS-2% l-propanol, (e) 0.05 M SDS-4% l-propanol, (Q 0.07 M SDS-2% 1-propanol. Reprinted from Ref. 16 with permission of the Sociedad Espafiola de Quimica Analitica and Elsevier.
Fett [183 studied the effectand concentration of and Brij8 35)ontheretentioprominentpeak,in the c ~ o ~ t oofurine g r ~ matrix. These authors obse the retentionof t h s peakwasminimizedandremainedconstant -range i 5.5-7.5, ith SDS mobilephases, for all surfactant concentrations I5.5, an impo~antincreasein the retentionof the endogeneous compound was observed for surfactants. both The shape of the k vs. pH curve of the endogeneous compound is sigmoidal, which indicates that it is protonated in acid solution, with a logarithm of the protonation constant, log ICH:= 4.5-5.0 (Fig. 1 1.4) [161. The retention is longer at a decreasing concentration of surfactant (Fig. 11.5). micellar mobile phases, the pH should be in the ge of 6-7 in order to a larger concentration m i ~ i z the e retention of the matrix, or at a lower should be used. However, the elution of the drugs should also red. The retention of many acidic compounds increases at lower p whereas at larger S S concentration, the retention is decreased and the atographic peaks deterioratesappreciably. efficiency of the c alternative of reducing the retention of thematrix is the addition of analcohol to solutions having moderatecon~entrationof surfact fact, most procedures developed for the dete~inationof drugs by as Surfactant together with an alcohol as modifier, due to the excessive retention with pure micellar eluents. A study was made on the variation of the background of physiological matrix (urine) when mobile [161. The retention of the phases with severalmodifierswereused prominentpeakinthematrixdecreased, as usual, withincreasing l "pentanol, followingthe elution con~entrationof methanol, 1 -propanol and strength of these alcohols. A smaller effect was observed withthe protein band. All these studies showed that the retention of the endogeneous c o m p o ~ giving d the prominent peak leaves a detection window for the drugs a and is reproducible. The mean retention time of the prominent peak with mobilephaseof 0.04 -4% l-propanol at p 4.5, for smples taken from eleven healthy females, was mea red to be 8.3kO.3 min (relative standard deviation, RSD = 4%). On the other hand, the
the position of this peak for 12-16 urine samples of two volunteers, taken long 36 h, was 3 4 % .
Influence of pH and of SDS on the retention of the prominent peak in the chromatogram of urine matrix. Mobile phases: ( l ) 0.05 M SDS,(2) 0.1 M SDS, (3) 0.15 M SDS. Reprinted from Ref. 16 with permission of the Sociedad Espaiiola de Quimica Analitica and Elsevier.
30 r
25
20 l5
l0
5 0 0
0 00 .. 004
0.0
Figure 11.5 Change in the retention factor of the prominent peak in urine matrix, as a (A) SDS mobile phases at pH: function of mobile phase surfactant concentrationat various pH: )and 7.5 (m); and (B) Brij-35 mobile phases at pH: 3.0 ( Reprinted from Ref. 18 with permission of Elsevier.
The optimization of mobile phase composition should take into account not onlythe retention of the compound to be analyzed,but also the retention of thematrix. The determinationof anticancer6-thopurine drugs and their metabolites in untreated serum is a usehl example (Fig.1l.6) [191, ith a mobile phase of 0.04 M SDS in 0.01 phosphate buffer at p the blank serum produceda background responsethat had completely eluted after 6 min,withtheexception of a peak at 8 min. The peaksof 6thiogu~idine riboside and 6-thioguanine were well resolved, but unfo~unately, thethreeearlier-elutedcompounds(6-mercaptopurine riboside, 6-thioxanth~eand 6-mercaptopurine) were overlapped by the matrix peaks. A lower pH (2.0) permitted complete se~arationof 6erc cap top urine from the serum background signal. Under these conditions, 6-thioguanine elutedtoo late (at around 40 min).
3
d
(A) Chromatograms of serum blank and (B) and (C) serum spiked with 1 pg/mL of 6-thiopurine compounds. 18 C column and mobile phase of 0.04 M SDS in 0.0 1 M phosphatebuffer atpH2.3 for (A) and (B) and pH2.0 for(C).Compounds: (l)6mercaptopurine riboside, (2) 6-thioxanthine, (3) 6-mercaptopurine, (4) 6-thioguanidine riboside and (5) 6-thioguanine. Reprinted from Ref. 19 with permission of Elsevier.
. i? I . The State of the Art It has been demonstrated that MLC is a reliable chromatographic method, which can greatly improve boththe speed and simplicity of the analysis for compounds of interest in physiological fluids. The versatility of this technique encompasses the wide range of drug classes normally monitored, such as analgesics, anticancer drugs, antidepressants, antiepileptics, bactericides, P-blockers, bronchodilators, catecholamines and diuretics, among others. Some of the reasons that explain the few applications reported up-to-date for this technique may be that most of the present extractio~reconstitutionmethods are well established, andlor the method development in MLC is unfamiliar. The ability to inject physio~ogi~al samples directly into an MLC system will no doubt in the future be exploited for clinical applications. The experimental characteristics of some reported chromato~aphic procedures, that employ micellar mobile phases for the determination of drugs in physiological fluids, are given inTable 11.1. In these procedures, ITV-visible detection was used, Other procedures using other detection modes will be co~mentedon in Chapter 12. Spiked samples were usually employed for the development of the procedures. However, an appropriate study of the possibilities of the technique should include the use of metabolites and urinary excretion controls. The physiological samples should be collected at several time intervals post-dose, frozen immediately and stored in the dark, until being analyzed. The stability of the drugs in the samples should also be controlled. Comparison of the samples with a blank serum is always necessary.
Y:2. ~ p t i ~ i z a tProtocol io~ Mobile phase parameters should be properly chosen to improve the selectivity of the separation between a drug and the endogeneous com~onents in the physiological fluids. Optimal control of mobile phase composition, including pH and type and concentration of surfactant andmodifier, permits the development of MLC procedures for the determination of virtually any drug in these samples.
U
m
0
Table 4 1.1
~co~tinu~~)
Compounds
Samples; station^ Phases; Mobile Phase Compositions; Detection ~ a v e l e n g ~Checked s; Linear Ranges; Limits of Detection Ref.
S u l f o n ~ d e ss~u l f a c e ~ d e ,sulfadiazine, sulfa- Hmm urine and cow milk; ~ y ~ o p ~endcapped lic ODs at 40'C; 0.07 M merazine, sulfathiazale, sulfamethazine,sulfmeth- SDS-6% 1-propmol. o x ~ ~ ~ i snu lef a, c ~ o r a p ~ ~ i nsulfmonoe, methoxine, s u l f a b e ~ ~ d~e , a ~ e ~sulfao ~ ~ e , quinoxaline and sulfisomidine
30
Sulfonamides: sulfadiazine (I), sulfagu~dine(11), Urine; Spherisorb ODs-2; 0.05 M SDS-2.4% I-pentanul, azo dye presulfmethizole (ID), sulfamethoxazole (IV) and column derivatizationwith nitrite and N-( 1-naphthyl)e~ylenedi~ne di~ulfa~iazole (V) h y ~ o c ~ o r i d550 e ; m;LODs (pg/mL), 0.1 (I, IV),0.2 (11, V) and 0.3 (III).
31
Alkaloid: nicotine (I) and its ~ e ~ b o l icotinine te (11) Urine; Econosphere CN-bonded silica; 0.2 M SDS-3% 2-propanol at pH 4.6 and 40°C; linearity, up to 3 pg/mL; LUDs (pg/mL), 0.2 (I) and 0.1 (II).
32
~ t i ~ ~ drug: a 5-lipoxygenase t o ~ inhibitor Urine; CN-bonded silica; 0.025 M SDS-3% 1-propanol in 0.01 M Zileuton and its N-dehydroxylated metabolite phosphate buEer at pH 3; 262 nm; linearity, 0.25-5 pg/mL; LOD, 0.1 Pg/d*
33
Steroids:hydroxycorticosterone (I), corticosterone
Urine>Spheri-5 W-18; 0.05 M SDS-9% 1-butanol; 245 m; linearity, 0.1-10 pg/mL; LODs (ng/mL), 50 (I, 11) and 100 (111, IV, V, VI).
34
S e m ; c o l ~ - s ~ t c extraction ~ g , column: ODs and 0.01 M SDS, analytical column: Adsorbosphere ODs and methanol-w~ter65:35 (vh); 242 m; linearity, 30-30~0nglmL.
35
(a),~ o r ~ ~ e r(111), o n etestosterone (IV), medroxyprogesterone acetate (V) and progesterone (W) Diazepam
W1
% F M
M
.r( e,
8
t:
p! Frc
W M
C B M
HY
of the micellar mobile phase isan impo~antfactor fo analysis of ionizable drugs using nonpolar column s t a t i o n a ~phases 11.7 shows the c h r o m a t o ~ r of ~ sacetylsalicylic acid (log 3 .Oand 6.5, eluted from aC 18 column with a0.0 20). It can be seenthat at the higher p evident. The anionic solute probably eluted ver, when pH was reduced to 3 S, the ne 3.5 min, appreciably distin rug eluted at appro~imately components.
(A) and (C) Chromatograms of a serum sample containing 25 pg/mL of acetylsalicylic acid, eluted with a 0.08 M SIX mobile phase at pH 3.0 and 6.5, respectively. (B) Chromatogram of blank serum using the same mobile phase as in (A). Reprinted from Ref. 20 with permission of the American Chemical Society.
procedure forthedeterminationofdiureticsofdiEerent therapeutical character: high(bumetanide,ethacrynic acid, ~rosemide), intermediate (bendro~umethiazide,chlo~alidone,hydrochloro~azide, x i p ~ d e and ) low (acetazolamide,miloride, spironolactone, triamterene) efficacy diuretics, and the uricosuric agent probenecid, in urine samples, illustrates a method development implying the control of pH,surfactant and modifier [23]. The greatest analytical problems in the detection of these compounds are basically their wide variety of chemical structures, ~ c t i o n a l groups andprotonation constants. This implies the use of several experimental conditionsfor their analysis with conventional aqueous-organic mobile phases and laborious liquid-liquidor solid-liquid extraction prior to chromatographic separation. In contrast, the s m e micellareluent can produce a satisfactory separation after direct injection. Some diuretics (ethacrynic acid, bumetaide, furosemide, probenecid andxipamide)show an acid-basebehaviorin the pH range of 3-7. Significant changes in retention were thus observed as the pH of the anionic mobile phase was modified, with longer retentions in acidic media where the compounds were protonated. In contrast, their retentions were very low at pH >6 where the anionic form of the compounds dominated. The other diuretics (acetazolamide, amiloride, ben~o~umethiazide, chlo~halidone,hydrochlorothiazide, triamterene and spironolactone) did not experience changes in retention withpH. The use of SDS micellar mobile phases withouta modifier, at diEerent pH values inthe range of 3-7 and in an adequate separation of the the presence ofan alcohol at pH 7, did not give mixtureofdmretics.Also,bumetanide,ethacrynicacid,furosemide, probenecid and xipamide were overlapped by the broad band of urine near the solvent front, at pH 6-7. Maximum separation of the peaks of these diuretics was observedat pH 4.5. On the other hand, the retention factor of the most retained diuretic, SDS (0.15 M), wastoo spironolactone, at a relatively large concentration of high (retention factor, k = 29). horganic mochfier was required for lower ethanol scarcely variedthe retention of the diuretics and urine matrix In contrast, l-pentanol excessively reduced the retention and some diuretics were overlapped by the background of urine. l-Propanol was found to be mostappropriate because of its intermediate behavior. However,
11-
t
7
(nata)
Chromatograms of: (a) urine matrix, (b) urine matrix spiked with a mixture of 1 pg/mL of each diuretic and (c) an aqueous solution the of diuretics, all eluted with a 0.042 M SDS-4% l-propanol mobile phase in phosphate buffer at pH 4.5. Compounds: (l) furosemide, (2) chlorthalidone, (3) ethacrynic acid, (4) bendro~umethi~ide,(5) probenecid, (6)bumetanide, (7) amiloride, (8) xipamide, (9) spironolactoneand(10) triamterene. Reprinted from Ref. 23 with permission of Elsevier.
acetazolamideandhydrochlorothiazideappeared overlapped by the broad band of urine.
at the solvent front,
urther optimization ofmobilephasecompos 1 concentration) was made using the method dev . [40], assisted by the ~~C~~~~ so imumresolutioncorresponded to shows the experimental chromatogram for this mobile le splked witha mixture ofthe diuretics, together with a similar c h r o m a t o g r ~obtained with an aqueous solution. The retention times for both chromatograms were similar. The peaks of amiloride and ~ p ~ i done the , one hand andt r i ~ t e r e n and e spironolactone onthe appeared m u ~ a l l yoverlappedinthe chromatogram§.The pe bendro~umet~azide was overlappedby the prominent peak of urine. pportunities to fixrtherreduce matrix inte~erencesthrough chemometric techniques and diode array detection present challenges for fixture studies.
example of the capability of in theseparation of m given bythe screening of u l f o n ~ i d e sin human cow milk with an ~-2~propanol mobile phase and (Fig. 11.9) [30]. Using a 0.07 propanol eluent at pH 3.0, isocratic separation of the sulfonamides was achieved within 15 min. S
A s c o ~ e nabove, t ~ the use of a surfactant different from
anecdotalindirectinjection MLC procedures.For this reason,the comparisonofthe dete~ination of a drug such as ~ e o ~ h y l l i n e (l,3-dimethylxanthine), using two different surfact~ts,is interesting. et al. [ g ] r e p o ~ e da procedure for the direct d e t e ~ ~ a t i oofnthis human serum by using a pphase zwi~erionic theof sur containing 3% 1propanoland W detection at273 m.Later, artinez;et al. [29] loped a procedurefor theophylline, caffeine (1,3,7-trimethylxanthine) eobromine (3,7-dimethylxanthine), in urine, with a
column, a 0.075 M" S% l-propanol eluent and UV detection at thesame wavelength. The two surfactants gave similar results, since theophylline eluted in bothcases in 5 min, sufficiently apart from the broad bandof the matrix, the chromatographic peak wasalmost symmetric and the LO near 0.5 pg/mL. With the SDS mobile phase, the peaks of caffeine and its meta~oliteswere well separated. Certainly, the good characteristics and availability of SDS will make it difficult to be replaced in the future by other surfactants in liquid chromato~aphicdrug analysis.
n Separation of sulfonamides in: human urine (top) and cow milk (bottom), Mobile phase was 0.07 h4 SDS-6% l-propanol in 0.02 h4 phosphate buffer at pH 3.0. Compounds: (1) sulfacetamide, (2) sulfadiazine, (3) sulfamer~ine,(4) sulfathiazole, (5) sulfamethazine, (6) sulfamethoxypyridazine? (7) sulfachloropyridazine? ( 8 ) sulfamonomethoxine?(9) sulfabenzamide, (10) sulfadimethoxine, (1 1) sulfa~uinoxalineand (12) sulfisomidine. Reprinted from Ref. 30 with permission of Elsevier.
icellar mobile phases containing modifiers permit the adequate of mixtures of solutes of dif%erenth y ~ o p h o b i c iwith ~ , isocratic owever, for samples containing compounds with a wide range of retention, gradient elution can be used to hasten the elution of strongly retained compounds.This will reducethe analysis time and improvegreatly the sensitivit of detection. In fact, rapid gradient capability is another LC,where the stationary phase must not be reequilib~ated l mobile phase composition after each run. However, the rocedures using gradient elution are few. One example is of the diuretic bumetanide inserum, using a mobile phase pan01 with phosphate buffer at pH 3.5 [IO], where the ration of 0.1 M was held until the peak of b u m e t ~ d e was completely eluted (7 min) and then, the concentration of increased to0.2 M over 5 min to ensure that any adsorb componentswerewashedfrom the column. Another application is the determination of thiazide diuretics and furosemide, wherean SDS gradient was used to accelerate the elution of furosemide [24]. ixed surfactant mobile phases canbe useful to improve the required in difficult separations [181. Although hydrochloro thiazide was well separated from the urine matrix with a Brij-35 mobile phase, no resolution could beachievedbetween this diuretic and its hydrolysis product 5-chloro-2,4-disulf~oylaniline, which is formed in aqueous solution upon standing at room temperature. In contrast, baseline separation was obtained between the two compounds with SDS mobile phases. However, SDS eluents did not provide separation of the diuretic inent peak in urine. The optimized mobile phase contained (below its cmc), in addition to 0.02 M Brij-35, The small was added to obtain resolution betweenthe diuretic and its hydrolysis product, without compromising the separation of the drug from the urine background.
W detection is the most available mode in liquid chromatography, its sensitivity is however insufficient for some trace amount determinations. Therefore, other detectionmodeshavebeenutilized to enhance this
capability. The use of precolumn derivatizationto improve the sensitivity and selectivity in MLC d e t e ~ t i o niss commentednext. are described in Chapter 12.
UV monitoring. A ost amino acids cannot be detected by direct wasdevelopedforthedeterminationof a mixture of these compounds(proline,glutamine,threonineandtyrosine)in urine, by Cu(I1) complexes and detection at 235 m,using formation of the DS-8% l-propanol mobile phase at pH 5.5 [25]. therwise, the high background at the beginning of the chromatogram ofa physiological sample,that appears when the detection is performed at 230-280 m, is eliminated inthe visible region. To exploit this fact, the azo dyes of several sulfonamides (sulfamethizole, sul sulf~ethoxazole,sulfadiazine and sulfathiazole) with the Br reagent EN-( 1-naphthyl)ethylenedi~inedihydrochloride,N in micellar solution before injection in an MLC system. The derivatization reaction is rapid, can be automated and yields high molarabso tivities at large wavelengths. The elution was performed with 0.05 M S pentanol and the detection at 488 nm [3 l]. The chromatogram of urine matrix treated withthe derivatization reagents gaveonly one peak at 13.5 min,whichcorresponded to an unknownendogeneouscompound that probablyformed an azo dye. This peakdidnot interfere with the dete~inationof the sulfonamideazo dyes, which elutedat shorter retention times. The describedmethodgathered the advantages of precolumn derivatization and chromato~raphywithmicellarmobile phases. The derivatization of the sulfonamides increased the retention by reducingthe polarity of thedrugsandimproved the signal-to-noise ratio and the resolution ofthe c~omatograms.This produced the adequate separation and detection of thes u l f o n ~ i d e sin urine.
an MLC procedure with precolumn Another example of derivatization is the determination of thiazide diuretics, after hydrolysis and formation of the NE azo dyes [24]. The bazide diuretics originated only two difTerent arylamines by hydrolysis, depending on the existence of a -Cl two peaks were or -CF3substituent in the thiazide nucleus. Therefore, only ( observed in the chromatograms for mixtures of these compounds
11.10). Underivatized diuretics could not be detected,especiallythose with a -C1 substituent, due to overlapping with the band of the proteins and the prominent peak of an endogeneous compound,in the physiological sample. 3
E
4
l
Chromatograms of: (a) spiked urinecontain in^: (1) hydroch~orothi~ide, (2) hydroflumethi~ide,(3)furosemide and(4) endogeneous compounds; and (b) urine blank. The samples were subjected to hydrolysis and derivatization with the ~ra~on-Marshall reagent. Reprinted from Ref. 24 with permission ofElsevier.
S
been used to analyze physiological samples for assay of drugs usedin sport. Theuseofperformance enhancing drugs by sportsmen and sportswomen is recognized today as one ofthe key problems in sport, toxicology and sport medicine [4 l]. ~ompetitors are ~equently pressured to beat marks andsucceed ich, unfortunately, leads to drug misuse to enhance their performance. ing practice in modern sport has through the years. The extent of drug use is u ~ o a l ~ o u,g h cators show that it is wider than just a few isolated cases (e.g. Tour e France1998). In addition to ethical considerations, several of the classes of drugs routinelyabusedcan cause adverse eEects in the organis including changes inbloodviscosity, cardiac ~ n c t i o nan resistance. These changes may be deadly in combination with inducedphysiological erations, such as those related with h electrolyte balance, flu alance andmetabolic rate. heuseof perfo~ance-enhancing drugs constitutes a ~ a ~ e d ctice in most official athletic competitions. The experien controls performed by severalinternationalorganizat~ons . ive pharmacological ofbanned drugs, whichisupdatedcontinuous1 categories of dopingdrugs are ~nsideredby the lympic ~ o ~ i (IOC): ~ e stimulants, e narcotic anabolic steroids, P-blockersand diuretics [4 l]. analytical methods to determine unequivocally these comp metabolites in physiological fluids has been one of the inclusion of these banned classes. Thus, although it wa olic §teroid§ were being mis~sedin sport before 196 an the class until1975,when suitable methods of
siological fluid of choice for doping control bec aking the samples, relative simplicity of its c
comparedwithotherbodyfluids,andaccumulatedexperience. The complexity of the dope problem has forced the development of a varietyof advanced analytical methods, in order to detect andcharacterize any kind of banned drugs. These include gas chromatography, liquid chromatography, hyphenated gas chromatography-mass spectrometry and liquid chromatography-massspectrometry,fluorescencepolarizationand i~unoassay methods. Most methods require some degree of sample conditioning prior to detection, starting with the extraction of the drug from the urine matrix. For co~poundsexcreted as conjugates in urine, hydrolysistheofmetabolites is required. Cas chromatographywithconventionaldetectors or in combination with mass spectrometry requires additional sample derivatization, to increase the volatility of the drugs. Even compounds excreted free in urine need liquid-liquid or solid-liquidextraction before this chromatographic analysis. everal aspects of the official me~odologyshould be considered in dopingcontrol.Screeningproceduresneed to bebasedonlyonsemi~uantitativeapproaches, since the purpose is detection. Suspect samples should be submitted for Eurther confirmatory analysis by a technique based on a different analytical principle, before making an such important decision as to whether somebody has used a drug before competition. Increasing sample throughput is becoming extremely important in the sport-oriented laboratories. More than 200 drugs and metabolites must be tested for a variety of doping agents in sporting events. Laboratories are receiving more and more samples and are expected to keep turn-around timesto a minimum. In.this context, methods needing little or no sample preparation, resulting in a reduction of analysis time,are of great interest. Among the different testing methods approved for illegal drugs in the Olympics, only a few liquid chromatographc procedures are included. The tedious sample preparation and lengthy analysis times preclude the general use ofthe existing conventionalHPLC procedures for rapid screening. Two Spanish groups headed by Laserna andGarcia Alvarez-Coque studied the possibilityof MLC inthe screening ofillegaldrugs [26,42,43]. Two ~ p o ~ aadvantages nt of this technique are the direct injectionof urine, and of different the possibility of determining, from a single injection, compounds of chemical structures. In MLC, the rapid elution of proteinaceous material urine opens a large window for the detection of banned drugs.
11-
many drugs are excreted as conjugates in urine andthe advantages of direct sample injection methodsare of decreased relevance. Also, unfortunately, the availability of real urine samples from doped athletes is limited and not always can a real sampleanalyses be made inthe development procedures. Further evaluation of the methods is needed in order to implement it as a complementary methodologyfor doping control.
. YII.1. ~ x t r a c t ~ oProcedures n uantitation at low concentration levels of some drugs by MLC with direct injectionmaynotbe feasible for severalreasons: (i) the drug peak is overlapped bythe protein band and peaks of endogeneous compounds in the physiological matrix, (ii) thedrug peak is overlapped by the peaks of other h g s consumed bythe individual or (iii) the LOD is insufficient. The approach improves column lifetime and provides a mechanism for rele strongly protein-bound drugs. However, it implies moderate efficiencydue to poor mass transfer and, with large injection volumes,the inefficiency is amplified and the high amountsof proteins in the sample can deteriorate the chromatographic column. As with any singular dimension approach with a small injection volume and the lack ofan enrichment step, the sensitivity of the determination is inadequatefor many applications. Some authors have considered the MLC approach still interesting for the dete~inationof drugs in physiological fluids, when a previous separation step is required. It is expected that the combined selectivity of to avoid extraction procedures and MLC will provide wide resolution power p r e s ~ b l e r e ~ c t i oThus, n s . the antipyrine metabolites4-~noantipyrine, 4-methy~aminoantipyrine and 4 - f o ~ y l ~ n o a n t i p y r i nwere e in plasma samples and eluted in less than 5 min with a 0 pentanol mobilephase and C 18 column. Extraction with methylene chloride [27], or through disposable C l 8 bonded porous silica cartridges IrZS], with methanol as an eluent, were necessaryto remove the metabolites from the
n the procedures, the organic extraction solvents wereevaporate to dryness and reconstituted with mobile phase before injection into the chromatographic system. Furosemide was used as an internal standard.
C separation, chlo~halidonewas extracted intodiethylether-2-propm plasma, usingxipamide asan internal re extracted from urine through alumina chloric acid [44] and clen~uterolin urine was eluted from a cation-exchange sorbent withthe same solvent used as he detennination ofsteroids is dif~cultdue totheir high hydrophobicityandverylowconcentrat ~ h y s i o l o g i csamples ~ (in the lowng/mLlevel). A mobile phase of and l-pentanol pennitted the separation and adequate elution of a m i ~ u r eof medr methyltestosterone, medroxy~rogesterone acetate, st r ne propionate and testosterone monitoring was not enoughfor its quantitation in ~ r i nsamples, ~ when direct injection made. Attempts were also unsuccess~lto reach an adequate by using sensitized terbium fluorescence detection[47]. Thus, for theseanalyses, a previousse~aration is imperative.
an e n r i c ~ e n tstep precludessimpleone-dimensional c separations withdirectinjection from providing the for many detenninations. doptingamultidimensional rove the sensitivity by all ing large injection volumes yet pennits trace e n r i c ~ e n tand peak compression sing. This a~proachdemonstrates a practical use of the singular d~ensionalchromatographic process.
481. The procedures operate with aconsuma~le
extraction column that can last for several hundred injections. chromatographic ensionprovides the sample extraction and clean-up withamicellar S ile phase. Theseconddimension, coupled on-line to the first, utilizes conventio reversed-phase media and organic solvents for the analytical separation. the prote~ceousmaterial removed in the extraction step, any reversed-phase packingmaterial is compatible with the system. Thethree column-switch~g con~gurations shown in Fig. 11.l have been used. ulticomponent analysis is facile undera set ofwelldefined conditions: (i) the compounds separate from excludedmaterial, (ii) elute off the extraction columninasinglebandand (iii) are separable on the analytical column, (iv) the mechanism ofseparation differs (i of the columns and (v)separation and detectionare adequate eluentisselected to provide the proper retention characteristics, but generally requiresonlymodest modification for differing applications. ~ r r e g a r ~ eof s s thetype of bondedphase, the excluded componentsfiom the extraction column should occupy the first minutes of the c h r o m a t o g r ~with the micellar eluents used. Also, the analytical phase should bemore retentive for the drug component than the extraction phase with the eluent used during the purge phase (see Fig. 11.1). Because of trace e n r i c ~ e n t , the purity of all reagents, primarily SDS, is especially critical. Posluszny et al. [5], in one ofthe first procedures, cut thedrug onto the analytical column with micellar mobile phaseas it eluted off the tail of the exclusion front. This style of recovery hadthree significant disadvantages that markedlyimpacted the selectivity: (i) endogeneous substances were eluted onto the analytical column, (ii) the micellar mobile phase was drastically diEerent from the analytical mobile phase, resulting in a large baseline shift since as much as 4 mI, might be cut onto the analytical column and (iii)the relatively weak eluting power of the micellar mobilephase resulted in the retentionand build-up of hydrophobic substances on the analytical column. These disadvantages wereovercomebyproviding for a more selective elution of the extraction column [48]. This was achieved in the following stepwise m a ~ e r(i) : the physiological sarnple was injected under micellar conditions (the micellar solvent and column were selected to ensure
s i ~ i ~ c aretention nt of the drug), (ii) after clearance of the proteinaceous material, the drug was eluted witha nonmicellar recovery mobile phase (the extraction and analytical columnswereplaced in series and the drug recovered onto the analytical column), (iii) the valve was switched to the analytical mobile phase andthe separation was performed onthe analytical column and (iv) simultaneously with step (iii), the extraction column was washed and reequilibrated with micellar mobile phase.
WASH
I
trfcyc tics
Chromatogram ofthe extractioncolumn for tricyclic antidepressant separation in plasma. Extraction column: 4 cm x4.6 mm id. polys~rene-divinylbenzene,10 pm spherical polymer. Extraction eluent: 0.06 M SDS-8% (v/v)acetonitrile-0.2%triethylamine in 0.02 M phosphate buffer at pH 7.2. Recovery eluent: 0.08 M SDS-56% acetonitrile-12% methanol in 0.09 M phosphate buffer at pH 3.0. Wash eluent: 0.06 M SDS-90% acetonitrile. Reprinted from Ref. 48 with permission ofElsevier.
This protocol resulted in the following ordering of mobile phases flowing through the extraction column: (i) extraction solution (the micellar solvent that washes endogeneous material off the column while permitting analyte retention), (ii) recovery solution (the solventthat elutes the analyte off the extraction column onto the analytical column) and (iii) wash solution (a strong solvent employedto clean-up the extraction column). The recovery
11-
solvent isusually an aqueous dilutionof the analytical mobile phase. This measure generally provides for solvent focusing on the head of the analytical column. Theend result is a dramatic sharpening ofthe peaks of interest both on the extraction column, as well as on the analytical column. sensitivity and selectivity are improved.
A column-switching separation of an injection of 500 pL of blood plasma con~ining270 ng/mL, of a mixture of the tricyclic a n t i d e ~ r e s s ~ t drugs, doxepin, desipramine,no~riptyline,imipramine and amitrip~line,is shown in Figs. 11.11 and 11.12 1481. Figure 11.11 illustrates the selective recovery of the tricyclics from a polyrneric extraction column. The five components are separated as a single peak from the solv recovery mobilephase. The recovery is accomplished by a extraction solvent pH was adjusted to 10 with triethylamine, while the recovery solventwas formulated withoutthe base to ensure that nothmg of 0 NOR
I
DE
IMI
2 Multidimensionalchromatogram of tricylic antidepressants in blood plasma. Extraction column and extraction, recovery and wash eluents as in Fig. 11.11. Analytical column: Spheri-5, RP-18,22 cm x 4.6 mm i.d. Analytical eluent: 0.08 M SDS-56% (v/v) acetonitrile-32% methanol-O.OS% triethylamine in 0.09 M phosphate buffer at pH 3.0. Compounds: doxepin (DOX), desipramine (DES), nortriptyline(NOR), imipramine (M)and amitriptyline (AMI). Reprinted from Ref. 48 with permission of Elsevier.
extractionsolventpassed onto the silica-based analytical column. Triethylamine was again added to the analytical mobile phase to minimize peaktailing. ther applications ofc o l ~ - s ~ t c h i are n g given inTable 1 1.1.In DS was added to the analytical mobile phase to m ~ i m i z e artifacts from the switching processor to act as an ion-pair reagent for the analyte. At the high levels of organic modifiers present in the solutions, micelles ifstill presentare severely altered compared to more aqueous media. Also, a specified amount of organic modifier was usually added to the micellarextractionmobilephase to increasetheelutingpowerofthe otherwise relatively weak micellar solution. ~aturationof the s t a t i o n a ~ of s u r f a c ~ t phase inthe extraction column was maintained by the addition to the wash solution. the extraction and analytical column chromatography should independently. Afterwards, the multidimensional system will be integrated with a column-switching valve. ample throughput can be increased by reducing the time necessary for washing andre-e~uilibratingthe e~tractioncolumn and malung use of faster analytical chromatography. Also, one sample can be loaded on the precolumn while another is being chromatographed on the analytical column.
esterlund, Direct Injection of Plasma into Column LC Systems, romatographia, 24: 155 (1987). shida, I. Morita, T. Masujima and H. Imai, Direct Enrichment ~~ptop~ and a nits ~etabolitesin Plasma onto a Precolumn . 30: 3827 (1982). lowed by P L C Analysis, Chem. hida, I. Morita, G. Tamai, T. Tsuru, N. Takai and .Imai, Some Cha~acteristicsof a P~otein-CoatedODS C o l ~ m n and its Use for theDetermination of Drugs by the Direct Injection A n a ~ s i sof Plasma Samples, Chromatographia, 19: 466 (1984).
4. 5.
6.
7.
9.
erton, ~nternalSurfac Chem., 57: 1757 (1 berger, Determina~onofDrug Su~stances vect ~njection ~ultidimensional LC with a New Chromatographfor Pharmaco~neticDrug ect ~njectionof Body Fluids, J. Chrom~togr.,222: Karneyama and T. Takagi, ~ ~ e of cSalt t r on the Binding of Sodium Dodecyl S u ~ a t e of the Protein ~ o l y ~ e p t i derived de from and on the Vrscosity Behavior Bovine Serum Albumin in the Presence of the Surfactant, Bio~him. ~iophys.Acta, 1161: 79 (1993).
Direct
u e ~ ~ermouche, u cand ~ e of ~heophyllinein Human serum
10.
11.
12.
ce~lar Mo~ile Phase. Comparison say ~ e c h n i ~ uAnalyst, e, 118: 151 1 (1993). orsey, ~ r ~ c ~Approach cal to Direct a1Fluids w i t h ~ L C : D e t e ~ m i n a ~ o n ~ a t o ~ r a ~4:h 35 y , (1989). ur~in~arne, Determination of Bumetanlde in man Pl~smaand Urine by HPLC with Fluores~enceD e t e ~ ~ o n , Anal. Lett., 21: 1589 (1988). termination of Chlor in e .kLLC,A n a ~ s t1, 13: 8) . ~ a k a g a w aDirect , Serum ehavior and Recovery of Cephalosporlns,Anal. Chem., 59: 2732 (1987). .Yasuda, Direct SerumInjection i insand Assay of Hydrophilic '
*
13.
~eterminationof Methotrexate in ~ntreatedBody F l ~ i d sby ~ L C , Anal. Chem., 61: 946 (1989).
C.Garcia Alvarezes in MLC, Quim. in andW. Worsley, Simple andSelec~ve er~inationof ~ydrochlorothiazi~ein
tion of Selectivi~in MLC Procedures for the Determination of Drugs in Urine by Direct Injection, J harm. Biomed. Anal., 9: 323 (199 1). 19. Fraga, E. Blanco Gonzalez and A. S of some Anticancer 6-Thiopurine Compou with Micelles as the ~ o ~ i Phase, l e Anal. Chim. Acta, 212: 181 (1988). e Luccia, M. A ~ n y a n aand ~ L.J. Cline-Love, Direct Serum Injection with MLC for Therapeutic Drug~onitoring,Anal. Chem., 57: 1564 (1985).
~ i c e l l a Eluent, r Anal. Chim. Acta, 287: 20 1 (1994). . Carda Broch, M.C. Garcia Alvarez-Coque, E.F. Simo Alfonso and J.S.Esteve ~ o m e r ~ , De~ermination ~LC ofDiuretics by Diazotiza~on and Coupling with theBratton-~arshall~eagent) Anal. Chim. Acta, 353: 215 (1997). J.Issaq, P)irectirnjec~onProcedure n Unextracted Urine with ~icellar ~ydro-Organic~ o b i l ePhases containing Copper Ions, J Liy. Chromatogr.)12: 1085 (1989). Bonet is 26. Maldonado, J.J. Laserna, E. and n of Banned Drugs in Sport by MLC, Anal. Chim. Acta, 259: 203 (1992).
LY
27.
28.
29.
30.
31.
32.
33.
34. 35.
36.
37.
38.
etero, and l J.J. Laserna, Determination ofAntipyrine n Plasma by MLC, J: Liq. Chromatogr., 16: olites 2767 (1 993). I. Carretero, J.M. Vadillo and J.J. Laserna, Determination of Antipyrine Metabolites in Human Plasmaby Solid-Phase ~xtraction Analyst, 1.20: 1729 (1995). artinez, S. Sagrado and M.J. Medina Hernandez, A Rapid Procedure for theDetermination of Ca~eine,Theophylline and Theo~rominein Urine by MLC and Direct SampleInjection, Anal. Chim. Acta, 304: 195 (1995). . Khaledi, MLC ~eparationof ~ u ~ o n a m i d in es on-Column Injection, les using Direct J: Chromatogr. A, 692: 3 1 1(1995). E.F. Simo Alfonso,G. Ramis Ramos,M C . Garcia Alvarez-Coque and e sHuman Urine teve Romero,Determination of ~ u ~ o n a m i din Dye Precolumn Derivati~ationand MLC, J: ~hromatogr.B, 670: 183 (1995). J. Reynolds and S.J. Albazi, ~imultaneousDetermination of ~icotine and Cotinine in Untreated Human Urine by MLC, J: Liq. omatogr., 18: 537 (1995). . Thomas and S.J. Albazi, Simultaneous Determination of the ~-Lipoxy~enaseInhibitor ~ileuton and its ~-Dehydroxylated Metabolite in Untreatted Rat Urine by M C , J: Liq. Chromatogr. & Rel. Technol., 19: 977 (1996). Z.L. Chen andS.F.W a g , Determination of Steroids in Human Urine b MLC, Anal. Lett., 30: 2315 (1997). .J.Koenigbauer and M.A. Curtis, Use ofMice~lar Mo~ile Phases andMicro-Bore Column Switching for the Assay of Drugs in Fluids, J: Chromatogr., 71: 277 (1988). . Chen, Direct Injection A n a ~ s i sof HexamethyleneBisacetamide in Biological Fluids by an HPLC C~lumn- witching Techniquewith a Micellar Mobile Phase, C~romatographia,34: 63 .Li, L.R. Chen andY. ofMicellar Mobile Phasesand an r Direct-Injection Determination C Column-Switching Cortisol, J. Liq. Chromatogr., 16: 2583 (1993). . Li, S.X. Jiangand L.R. Chen, Direct-Injection Analysis of ~itomvcin-Cin Biolonical Fluids bv Multidimension
PLC with a Micellar Mobile Phase, J. Liy. Chromatogr. t& Rel. ~echnol.,19: 1255 (1996). 39 . ~este~lund Separation , of P~enolic Compounds and correspond in^ ~lucuronidesby Couple~-Column 46: 72 (1997). MLC, Chromato~raphia, 40.
st rate^ for Optimizationof Surfactant andAlcoho1Concentrationin
~
CJ. C~romatogr. , A, 677:239 (1994).
41.
en andIC. Tittel (Eds.), ~ e d i c i n eInte~ational , lympic C o ~ i t t e e , 1, 1988. 42. .Asensio and J. J. Laserna, MLC for ~ a~creening ~ i in Sport, J.Liq. Chromatogr.,12: 262 1 (1989). 43 8
44
*
45.
46.
47
I
48.
os, Determination of Catecholamines as A m i n o c ~ r o ~be s y ~ C ~hermalLens Spectrophotometric Detection,Chromatogra~hia, 38: 365 (1994). iosca,J.J. Baeza Baeza and of ~lenbuterolin Urine by A Derivati~ationand MLC, C~romatographia, 44 .Torres Gartas, .C.Garcia Alvarez-Coque C ~ ~ aDetermination s , of Anabolic Steroids in Pharmaceuticals by as LC witha~icroemulsionof Sodium Dodecyl Suljate and Pentanol ~ o b i l e ~ h aAnal. s e , Chim. Acta, 302: 163 (1995). ~a, of arrington and R.V o n ~ a n d ~ sDetermination Steroids in Urine by MLC with Detection by Sensitized ~ e ~ b i u m Fluorescence, Anal. Chem., 6.5:2346 (1993). J.V. Posluszny and R.~ e i n ~ e r ~Optimization er, ofMultidi~ensiona1 HPLC for the Determi~tionof Drugs in Plasma by Direct ~njection, ~ i c e l l a Cleanup r and Photodio~eArray Detection, J.Chromatogr., 507: 267 (1990).
~
The impressive ability of micelles to manipulate the ~ i c r o e n v i r o ~ e n t experienced by solubili~edmolecules may greatly alter their properties. In previous chapters, aqueous solutions of micelles have been s h o w to produceunique, effective separationswhenused as mobilephasesin liquid chromatography. ut another aspect that is not adequately considered is the eEects on detection. In recent years, there has been a rapid growth in the number of publications that report the use of surfactant monomers or micelles to improve the analytical performance of various spectroscopic (U spectrophotometry, fluorimetry, phosphorimetry, chemiluminescence and atomic spectroscopy), and electrochemical (especially ~perometry) s been recognized methods [l]. The unique properties of s u r f a c ~ t have as being very helpful to overcome many problems associated withthe use oforganicsolventsinthesemethods, ~ u r f a c t a n t - ~ o d i ~procedures ed yield sensitivity and/or selectivity improvements in determinations commonly performed in homogeneous solution, whereascertain analytical methods (such as room-temperature phosphorescence in solution) can be exclusively conducted in organized media. maxima co tensity -visible absorption the adhtion of surfactants. 1ar Liquid C ~ o m a ~ ~ a p h y visible dete~ionwere presented in Chapters 10 and 11. analysis of p h a r m a ~ u t i cprepara~ons ~ and p h y s i o l o ~ c ~ ghit was obviousthatseveralcompounds expe~enced
an
.._*_~..._.~...~..*.___I________._______~~~"~~.~~~-~"~.~**-"~.~~*..*~..*--~.~.~~..."~.---.*"~.-~.*""...*~.~~..~*~.~"..~ Samples; Detection Mode; StationaryPhases; Mobile Phase Compositions; Checked Linear Ranges; Limits oE Detection Ref
6 o ~ ~ u n ~ s
Codeine (I), morphine (II), propranolol (ID), S e m and urine; fluorescence (A, = 215 nm and Am = 300 m); pBondapak 618 and Supelcosil LC-CN; 0.02-0.05M SDS-10% I-propanol; quinidine (N)and quinine (V) lin&ty ( p g / ~ ) ,0.5-2.0 (I), 0.4-1.2 (II), 0.04-1.2 (III), 0.2-1.0 (N) and V). 0.2-1 .O (V);LODs (pg/mL), 0.3 (I, II), 0.01 (IU)and 0.03 (N, ~de, Spheri-5RP-18;f l u o r e s after ~ ~ i s ~ ~ o c yd ~ e ~~~ ta et ~ a t i0.1 o nM ; SDSAlprenolof , ~ e ~ d r o ~ u m e t h i a ~furosemide, nadolol, p h e n y l e p ~ epropranolol , and ~ ~ t e r e n7% e 1-pentanol. p-Blockers: acebutolol (I), atenolol (11), celiprolol Urine; fluorescence [Aexc = 230 nm and A,, = 440 m (I, a, N), 300 m (E, (III), labetalol (IV),metoprolol (V), nadolol (VI) and V, VI) and 340 nm (VII)]; Spherisorb ODs-2; 0.1 M SDS-15% I-propanol-1% ~ e t h ~ l ~ e - 0M . 0phosphate 2 buRer at pH 3; linearity propranolol (W) ( p g / ~ ) ,0.14.2 (I), 0.05-2.1 (a),0.54.2 (III), 0.1-0.9 (N),0.05-0.5 (V), 0.05-2.2 (VI) and 0.004-0.1 (W);LUDs (nghnL), 30 (I), 19 (11), 200 (El), 20 (IV), 16 (V), 8 (VI) and 3 (VII). Acyclovir S e m and plasma; fluorescence (A,, = 285 m and A, = 370 m);Separon SGX 618; 0.05 M SDS-0.05 M phosphate at pH 2.05; LOD, 0.08 pg/mL,, Plant growth regulators: indol 3-yl acetic acid (I), Plant extracts; fluorescence (Aexc= 281 nm and A,, = 340 m);Lichrospher 2-( 1-naphthyl) acetic acid (If), indol 3-yl propionic alkylnitrile;0.010 M SDS with pH gradientelution; linearity (pg/mL), 0.01acid (III), 2 ~ 2 - n a p h ~ yacetic l ) acid (IV), indol 3-yl 4 (I), 0.01-8 (11, 111), 0.02-8 (N), 0.01-20 (V), 0.02-20 (VI) and 0.05-20 butyric acid (V), 2-(l-naphth~l)acetamide (Vr) and (W);LODs (pg/g), 0.3 (I, a, V), 0.8 (II), 1.0 (IV) and 1.1 (VI, W). indol3-yl acetic acid ethyl ester (VII) ~~i~ S e m ; fluorescence(A, = 370 nm and A, = 504 m);Gapcell Pak Ra; pH1; 0.0 10 M SDS-20% acetonitrile ~ p r e c ohle~t i c d i ~ e r e n t i a ~ mode o~ formation of the aluminium 8-quinolinol chelate); LOD, 1 nglmL. ..._.*_l_..*.**l_****.
"
f
"
.
-
-
.
.
"
~
"
.
*
~
~
.
*
~
.
.
.
_
_
_
_
_
_
_
_
_
_
1
_
_
1
_
_
_
_
1
.
.
.
~
~
.
.
~
~
.
"
~
~
.
.
~
~
~
.
~
~
.
~
~
~
.
~
~
~
.
"
~
*
.
~
~
~
.
~
~
.
.
.
~
.
"
~
~
.
~
~
~
~
"
"
~
~
.
~
~
f;
7
10
11 ~
"
~
~
*
.
~
~
.
~
.
~
.
.
~
~
.
"
~
*
.
*
~
~
*
c?
0
enhanced absorption in the micellarmedium, addressed in mostanalytical reports.
t h s question was not
In this chapter, the features of other detection methods that have beenutilized in LC:conventional and sensitized fluorimetry, roomtemperature phosphorimetry? inductively coupledplasma h ~ h e n a t e dwith mass spectrometry, andamperometry, are examined. Table 12.l gives details of somereported procedures. Most o appeared during the 9Os, when the developmentof applications in increased. The study and use of new detection systems can result in enhanced flexibility and efficiency for the separation analyst.
.
The heterogeneouschemical microenviro~ent providedbymicelles influences excited-state equilibria by imposing additional constraints and pathways on the molecule's behavior. Fluorescence parameters such as excitation and emission wavelengths, quantum yields, fluorescence lifetimes and relaxation processes for the excited states may change. The di~erencesobserved result frequently in increased fluorescence intensities andlor reduced interference from impurities. From an analytical point of view, these phenomena can very be usefbl. Micellar-enh~ced fluorescence is a method that seems very promising and interest in it is steadilyincreasingwitha view to developingmoresensitiveand convenient methods for the determination of molecules and metal ions. LC procedures havetaken advantage of this method. The observedenhancementinemissionintensityinmicellar systems must derive from an increase in either solute molar absorptivity at the exciting wavelength andlor quantum yield, compared with that in bulk solvent alone. The rate constants for deactivation of the excited states by radiationless processes are also significantly reduced when the
LL
solutes are in the presence of micelles. The reason is that the organic compounds are effectively compartmentalized when they partition to or bind the micellar system and the emi~ingexcited state is thus protected from quenching. The bound solutes are probably much more restricte (less mobile) insuch an e n v i r o ~ e n tcompared , with the situation in bul solvent. In addition, the micropolarity is reduced and the microviscosity increased [1, 2 l]. of the solution is also an important parameter that will influence the luminescence characteristics of organic species that exhibit acid-base properties. In manyinstances, the chemical and physical properties of electronically excited molecules differ markedly from those of the ground-state molecules,becauseof the difEerent electro~ic distribution. Therefore, most of the excited molecules show protonation constants (log KH) whichdiffer greatly from those measuredin their fferences inlog KH of more than 6 units have been of compounds. A s for the ground state, acid-base equilibria in the excited state are drastically altered by the surfact aggregates, which can result in ahrther increase in sensitivity. The gain insensitivityis usually measured by the micellar enhancement factor, defined as the ratio between the fluorescence intensity in micellar solution and inhomogeneoussolvent, at the same fluorophor concentration. The observedintensities are usually many times greater than in the corresponding homogeneous media. Thus, the fluorescenceofacyclovir (a drug withaconsiderable activity viruses of the herpes group) was increased by a factor of five emission maxima was shifted by about 15 nm to the short-wavelen~h region,by addition of sodiumdodecyl sulfate (SDS) to the aqueous Inincrease in the relative intensity of solution [g]. slight fluorescence with concentration was already visible in the W the critical micellar concentration (cmc) concentration regi (Fig. 12.l). This was causedby the influence of n the protonation constant of acyclovir. The constant increased the f o ~ a t i o nof the protonated compound which is responsible ofthe fluorescence signal. In contrast, the addition of organic solve solutions may reduce the fluorescence e~ancements,
acyclovir as a function of concentration ddition of 5% 2-propanol, in a solution of 0.05 M phosphate at pH2. Reprinted from Ref.9.
eluents are compared in Table 12.2. mobile phase e ~ a n c e dthe fluorescence signal in a range from 1.Sfor fluoranthene to 10 for pyrene. Comparison of LODs (ng/mL or ppb) for somePAHs obtained in MLC and Conventional W L C with Fluorimetric Detection
ompoun~
Eluent Eluent Micellar
Acenaphthylenea
100
270
~thraceneb
0.2
0.2
0.5
2.0
270
480
0.2
0.5
0.2
0.7
luorantheneb
2.5
3.8
Naphthaleneb
0.3
1.2
1.7
17.4
0.25
2.6
~iphenylb
yrenea Pyreneb a
Micellar eluent: 0.035 MSDS;aqueous-organic eluent: methanol-water 30:70 (v/v) [13]. Micellar eluent: 0.024 M SDS;aqueous-organic eluent: methanol-water 40:60 (v/v) [2].
urther e~ancementsin detection were obtained by the use of laser-induced fluorimet~with one of the visible lines of an argon ion laser (488 m) for e~citation[7]. The relevant properties of this source ofradiation are its highintensity and excellent spatial resolution. A procedure was developed to determineseveralbanned drugs in sports, such as ace~zolami~e,amphetamine, atenolol, chlorthalidone and ~ r o s e m i ~ after e , separationwith a 0.1 M §D§mobile phase. A s the fluorescentdrugsdidnot absorb at488 m,the hghly fluorescent
~uoresceineisothiocyanate derivatives were formed before injection in the chromatographic column.
The development of selective and sensitive analytical methodologies for the analysis of minute quantities of drugs, in physiological fluids, have attracted considerable interest in analytical toxicologyand therapeutic drug monitoring. As shown in Chapter l 1,MLC provides a solution to direct injection of physiologicalsamples by solubili~ingthe protein components,via sudactant coating of the analytical column to avoid clogging. In addition, the surfactant monomers appear to displace the drug bound to the protein, releasing it for partitioning to the s t a t i o n a ~ phase. The possibility of direct sample introduction greatly simpli~esthe treatments and improvesthe accuracy of the procedures. Ascommentedabove,mostof the reported chromatographic IJV detection. This is not procedures usingmicellareluentsutilized optimal for many drugs and samples. Fluorescence detection of compoundswhichyield measurable fluorescence emissionmaybe preferable, owing to the higher sensitivity and selectivity when compared to absorption methods. h certain cases, the obtainable LO s i ~ n i ~ c ~lower, t l y more than adequate for therapeutic drug monitor in^ of concentration ranges normally encountered in serum andurine. The fluorescence background signal of the physiological matrix at the solvent front, due to unretained proteins, is similar to that observed with W detection (see Chapter 11). Again this is the limiting factor in S , although in some cases, the signal seems to be substantially reduced. The background response level can be varied by changing the excitation wavelength. It has been found that serum background can be completely eliminatedby using a 470 nm cuttoff filter [6].
The d ~ t e ~ i n a t i oof nnonfluorescent analytes through reaction with suitable fluorescent reagents extends the number of compounds that can
CTIO
NC
be detected using micellar-e~ancedfluorescence. h example is given c~omatographic-fluorimetric by a highly sensitive and selective procedure for aluminum inhumanserum, after formation of the 8-quinolinol complex [ll]. The interest in determining this metal is its relation withthe pathology of Alzheimerdisease and dialysis dementia. The effect of micellar aggregates on the luminescence of organic compounds can be easily explained, whereas the introduction of a metal ionin these microheterogeneoussystems significantly increases the difficulty to rationalize the observed effects. In fact, the mechanisms which are operating in micellar-e~ance~fluorescence of metal complexes have not been completely elucidated. However, neutral complexes appear to be protected against deactivation pathways by the surrounding micellar environment [l].
I I
10 0 R~~ention time I min
0
Figure 12.2 Chromatogram of 8-quinolinolato-me~lcomplexes. (a) Absorbance at 370 504 nm. Thesample contained 4x10" M Fe3+, nm, (b) fluorescence at h,, = 370 nm, h, Ni2+,Cu2+, Zn2+, Co2+, Pb2+4and ~ 1 0M - ~A13+.Reprinted from Ref. 11 with permission of Chemical Letters. I=
ong the common metal ions, only aluminum and cobalt gave W' eaks when complexed -acetonitrile mobile phases. to each other withs~ectrophotometricdetectio inum was only possible S well as several other surfactants to the al~minum complex solution, increased the fluorescence intensity. The not require deproteini~ationprior to analysis. The most commonly used t e c ~ i ~ ufor e aluminuminhumanserumis graphite-fu~ace atomic a b s o ~ t i o nspectrophotomet~,which is often limited dueto serum matrix inte~erence.
*
abolicsteroids are generallyused as therapeutic agents inclinical practice, but are also widelyabused as ~ ~ r f o r m a n cenhancing e drugs. The undesirable side ef5ects ascribed to this practice have le regulation or total prohibition most in countries. methodologies suitable for the d e t e r m ~ a t i oof~ anabolic steroids in urine should meet twocriteria: S for free steroid should be below nglmL.
1-10
imultaneous determination different of steroids and their metabolites should be possible. Exceptfor a numberofaromatic estrogens, the majorityof steroids are no~uorescent. any steroids react with concentrated sulfuric and phosphoric acid to form fluorescent derivatives, but the lack of selectivity and the f o ~ a t i o nof a large number of uncharacterized products limit the application of this method in drug testing.
ause of these consider
~ r ~ ~ stof ~e r
~I o ~ s
~
t
~
e luminescence gu a n t h ~ i d ions, e such as europium, have shown to be substantially increa§e~ when highly absorbing sen§itizer is U ~ertain co~pounds h carbonyl groups can to l ~ t h ~ i dions e
S
The testosterone derivat common a §t~ctural car~onylgroupin the -ring, that also occurs in nonas proge§teroned cortisone (Fig. 12.3). ct as a goo^ acceptor, the following condit ~
~of testosterone t ~ e
(1)
heexcitedenergylevelof the donor should be slightly above 2 ~ , 0 0 0cm” to minimize the energy di~erence with the excited singlet state of terbium.
(ii) eous
enviro~ent shoul~be ~ ~ to prevent e ~uorescenceby water. the s t r o n ~ g u e nof c ~terbium g
d
(iii)Thedonorandtheacceptorshouldbeless apart.
than
-100
The first excited singlet level of a typical semi~uinonoidsteroid is at 41,666 cm", corresponding to strong absorption at 240 m (Fig. 12.4). The important feature for l a n t h ~ d esensitization is the a triplet level at 26,041 which produces steroid existence of phos~horescencethat can beobserved at 7 Atroom temperature, this triplet energy levelcan be used to efficiently pump the 5d3 level of terbium ion7 which occurs at 26,000 cm". M e r relaxation to 5d4,this undergoes a radiative transition to the 7f ground level, resulting in the characteristic terbium ion fluorescence. E~lcientenergy transfer to lanthanide ion is made possible by the longlifetimeof the excited donor. Fluorescence e~ancementswere also observed in trivalent europium, samarium and dysprosium fluorescence, but the effect was strongest with terbium, due to the close energy match between the donortriplet level and the terbium , excitedsinglet state. Among the differ en^ steroids s ~ d i e d bol~sterone and testosterone acetate produced the greatest enhance~ents (approximately ~ 1 8 0with ) ~ LODs in the 0.5 ng/mL range.
43
Steroid donor 1
36
27 18
9 0
Energy level (Jablonski) diagram of typical semiquinonoid steroid donor and lanthanide ion acceptors. Reprinted from Ref. 12 with permission of the American Chemical Society.
LL
Fluorescence was found to be quenched in physiological fluids such as urine. This problem was effectively overcome by putting together the terbium acceptor and the steroid donor in SDS micelles. A modest (50%) increase in terbium ion fluorescence was found at the cmc of whichcorresponds to the typical increaseinlanthanidequantum when passingfromwater to an organicmedium. This suggests that terbium was solubilized in the micelle, probably by a c c o ~ o d a t i o nin the palisade layer. The ion experiences shelding from the aqueous mediu and is broughtintoenforcedproximity tothedonor also micelliz There should be however not only an effect of compa~mentalization,the surfactant is also effective in separating excited steroid molecules from each other, reducing triplet-triplet~ ~ i l a t i o n . The use of a micellar mobile phase simplifies the procedure by p e ~ i t t i n gthe direct injection of urine into the column, without sample preparation. The micellar eluent contained 0.01 M Tb(N and 20% acetonitrile. A typical chromatogram showing the separation of testosterone, methyltestosterone, bolasterone, progesterone and testosterone acetate is shown in Fig. 12.5.
igure 1 Steroid separation by m C : (A) Aqueous solution with 20 ng of each steroid and (B) 200 yL of urine sample spiked with 300 ng/mL of testosterone (a) and methyltestosterone (b) and 100 ng/mL of bolasterone (c), progesterone (d) and testosterone acetate (e). Reprinted from Ref. 12 with permission of the American Chemical Society.
.
this nonradiative decay of the triplet has been to avoid collisions by freezin a rigid frozen matrix (low-t rel~ctanceof anal~ical c~emists cool sarn les to c ~ o g e ~ c t e ~ p e r a ~ r e s
calledmicelle-stabilizedroom
order to observe
tem~er~t~re
amount phosphorescence. of exception is found with those ounds having internalheavyatoms, such as bromonaphthalene. out the heavy atom, there is usually insu~lcientspin-orbit coupling is necessaryforiystemcrossingfrom the excitedsinglet state S , the triplet state is not populated ~hosphorescencecannot occur.
It? 2.
~ ~ v ~ n t ~ ~ e s ~ o ~lC~h ir~co e~ a~~t~~ a~ in ari ~ h y~ Phos~horescence~ t ~ ~ i l i z a t i o n
icellar solutions permit the observation of phosphorescence by using a conventional spectrofluorimeter and micro flow-cells. this inst~mental simpli~cation was used to design a oration of a chromatographic column also proved to be quite S for thestabilizationofphosphorescence.Conventional static measurements of M ~ - ~ T arePlabor intensive, subject to variabili~ fiom incompleteandinconsistentremovableoxygen fkom solutionand require extensive sample puri~cation.For good precision, it is necessary to illuminate the sample inside the apparatus for a fixed time (e.g.,15 min), until the signal stabilizes. and resolving power of LC facilitate the for three major reasons: (i) (ii)
(iii)
h insituon-column puri~cationallows the detectionof pure solute, unencumbered by quenchers. The puri~cationprocessrendersitunnecessary to degas the individualsamples;oxygen is notretainedon the column and elutes on the solvent front undetected. The precise flow characteristics ofthesolventdelivery system enable each solute to pass through the detector flow-cell under consistent conditions, producing a solute illuminationtime whch isconstantfrom sample to sample.
As a result, relative standard deviations of the phosphorescence signal <2% have been routinely observed for replicate measurements in
LC. Sample preparatio~measurementtime was also reduced from 45 min to <5 min 1131. l reports using MS-RT detection employed SDS as surfactant [2, 131. The requisite micellar reagent can be introduced into the system either as the mobile phase or as a postcolum reagent. The s u r f a c t ~ t concentration can therefore afliect both the c~omatographyand the detector response. This stresses the i m p o ~ c of e c a r e ~ l l y s e p a r a tthe ~g c~omatographicand spectroscopic variables before comparing detection sensitivities. The presence of thalliumionused as heavy atom in the mobile phase didnotnot sig~ficantly alter the elution behavior of compounds, comparedto thatin a pure SDS mobile phase [2]. P detection is generally less sensitive than fluorescence. compounds, the percentage of fluorescencequenchedby thallium is >go%, but not all of that quenching results in intersystem crossing. Also, the micelle itself, although eEective,is not totally protective of the solute against radiationless pathway esults in l ~ i n e s c e n c e intensity losses. The sensitivity of is still s u ~ r i s i n gconsi~eringthat phosp~orescencelifetimes are several orders of magnitude longer than fluorescence lifetimes,thereby p e ~ ~ i fewer n g excitatio~emissioncycles per unit time. In the first report on detection in MLC, w r i ~ by ~ nA ~ s t r o n get al. 121, the authors expressed their concern on the relative weds phosphorescence intensity of several S, compared to their fluorescence. Careh1 elimination of oxygen was made but a systematic weakness of the system, such as a small air le&, was presumed. In another ,lineardynamic ranges covering three orders of m a g n i ~ d eand as low as 5 ng/L (5 ppb) were obtained for several P M s [131. The principal a d v ~ t a g eof M S - ~ T Pdetection is the improved selectivity via detection ofsignals in the red regionof the s p e c t ~ mwhich , is inherentlylesscrowded than the lowerwavelengthregion. This is accomplished by red shifting the signal by intersyste~crossing to the triplet state, which allowsa hgher wavelength cutoflifilter to be used.
12-
7
IK 3. ~haracterizatio~ of the ~ h o s ~ h o r e s c e ~si^^^ ce ~onventional phosphor~eterscan scan the spectrum of the s ~ p l to e allow identification of the emission, based on energy/intensity profiles. This is nothowever possible withmost chromatograp~cfluorescence detectors that use cutoff filters topass a broad bandof radiation. Additionally, in phosphorimeters7 some sort of temporal d i s c ~ ~ a t i o n is normally used to select an observation time window when the shortlivedfluorescenceand scatter has decayed to zero, but the long-lived phosphorescence is still present. W e n using a c~omatographicflow-cell withnotimedelay available on the detector, itis necessary to devise of determining the nature of the observed signal. anothermeans scribing a measured emission to fluorescence, phosphorescence, scattered light or a combination, requirescarehl verification. Three approaches can be used to discriminate betweenthe various types of signals. All relyon the factthat the excited tripletstate of moleculesisdifficult to populate, especially for those with large fluorescence quantum yieldsandonceachieved,it is very easy to deactivate by using radiationlessprocesses. First, elimination ofthe heavy atom would virtually eliminate phosp~orescencefor most species. signalis \still observed,it can be ascribed to fluorescence or sc (except for molecules with a heavy atom substituent). Second, osygenation of the mobile phase will introduce a quencher of the triplet state phosphorescence of the compound. If a signal exists in the presence as being ~uorescenceor ofoxygen, it can also besafelydesignated scatter. These two methods of signal discr~inationare illu~ratedin 12.6for a mixtureof2-naphthol,biphenyland phe~~ene. in Fig. 12.6A wasobtained using a thallium free, S mobile phase. The signals observedare due to fluorescencepassedby a 41 8 nrn lower cutoff emission filter in the detector. Upon substitution of 30% of the sodium counterions in the S micelle with thallium, a similar c~omatogramis observed (Fig. 12. These signals also arise fiom fluorescence and are d ~ i n i s h e din intensity ig. 12.6A7because thallium is effectively depopulating the excited singlet state from whch fluorescence occurs. Thus, Fig. 12.6A
describes a sample in which virtually no triplet states are populated and describes a sample in which the triplet state is appreciably populated but is radiationlessly quenched by oxygen.
c
Chromatograms of0.8 pg 2-naphthol(a), 16 pg biphenyl (b) and 0.4 pg phenanthrene (c), eluted from a cyano column with mobilephases of: (A) oxygenated 0.1 M SDS, (B) oxygenated 0.1 M SDS-TI@) dodecyl sulfate (70:30)and (C) oxygen-free version of (B). Full scale rangeexpansions areindicated on each chromatogram.Reprinted from Ref. 13 with permissionof theAmerican Chemical Society.
Fig. 12.6G, obtained with an oxygen-free micellar mobile phase with thallium ion, represents phosphorescence emission from
the
S
to the third appro
l and phenanthrene,respe ’ ing ~uorescence(Fig. 12. henyl could be increased er which would discrimin
lspeciation is b e c o ~ n gmore and more im ental toxicity andbiological impo~nce dependon their oxidation states and diEerent ch ccepted todaythat the mostreliable approache tandem t e c ~ i ~ u esuch s , as hybri analytical methods involvi chromatography/atomic e ission spectrometry y/inductively coupled plasma-m low level detection capabili~of especially attractive as an element-speci~c chrom c~o~ato~raphy. ~ n ~ o ~ u n a t e l y , mobile c o ~ oph n lvents, which may be detrimental to the decrease in sensitivi~can result due to er plasma ins~bility,i ~ o ~ a t i oofnmolecu the conventional organic solvents is icellar mobilephases have been prop
of alkyltin and arsenic compounds [14- 16, 251. The characteristics of these ~ a l y t i c a lprocedures are comented on next. The p r e l i ~ n a ~ optimization of the chromatographic separations was carried out with an detector, becauseof its ease of operation and accessibility ( ment time is often limited).
'y:2.
~peciutionof Alkyltin C o ~ p o ~ n ~ ~
rganotin compounds are widely used as biocides, catalysts and polymer stabilizers. Very sensitive (subpart per billion detection) analytical methods are required to assess theireffect on the e n v i r o ~ e n t . In addition,speciation i n f o ~ a t i o nis necessary,since the toxicity of organotin compounds is strongly dependent on the number and nature of the organic substituents. The feasibility of using MLC to separate organotin compounds [ with a C 18 reversed-phase column was investigated three types surfactants of r~ethyla~onium bromide, S was found to resolve these charged or nonionic micelle mobile phase resulted ina lack of interaction withtheorganotin cations. Therefore,thesemobilephasescaused the compoundseither to come off with the voidvolume or to become irre~ersiblyadsorbed on. the stationary phase. Salt depositionoccurred at theconstricted area of the torch injector when the S S solution was introduced in the ICPregular torch positioned horizontally. This produced a sig~~cant d i ~ n u t i o nof the analytical signal with time. After appro xi mat el^ one hour, the injectorwascompletelyblocked.Several torch modi~cations were a~empted,including a 2-m-i.d. injector with a reduced taper and an injector withthe taper regionmovedclose to the injectorinlet. A s t r a i g h t - ~ ~injector e with a 1 . 5 - m i.d. was also tried. Salt deposition still occurred. Finally, the deposition was eliminated by using a Leeman torch with a removable injector. The cloggingproblemwas firther reduced by s w i t c ~ n gthe column outlet to a 1% (vh) nitric acid solution between c~omatographic
IUM
runs. Nitric acid solution also rinsed the nebulizer, preventing it from clogging. With this interface, nodeposition occurred over an 8 h continuous run. Addition of oxygen to the nebulizer gas was unnecessaryas is usually the case with ~ethanol- at er or acetonitrile-water mobile phases. This reduced i n s ~ u ~ e ~complexity tal and prolonged the life-time ofthe sampling cone.
.7 Chromatogram of a mixture of: (A) monomethyltin trichloride, (B) dimethyltin dichloride, (C) trimethyltin chloride, (D) triethyltin bromide and (E) tripropyltin chloride. The concentration of SDS was programmed from 0.02 M to 0. l M for 2 min and constant at 0.1 M. Reprinted from Ref. 14 with permission of Applied Spectroscopy.
The concentration of SDS in the mobile phase was also kept to a mini mu^ (e.g., 0.1 M) to prevent clogging ofthe torchand sampling orifice on the ICP-MS detector. A good separation of organotin
ds with short alkyl chains ration, but butyltin compounds S 12.7 shows the c h r o m a t o g r ~of a m i ~ r of e alky splitting was observed for monomethyltin trichlori equilibriumbetweendifferentmono b-mglLlevel L . Compared to an detection provided better sensitivity.
enic has found wide use in pesticides, herbicides, wood desi~cants. Arsenic exposure can thus occur in many including occupations (e.g., coal mines and smelters), air and fo toxicity of arsenic varies widely with different chemical ),monomethylarsonicacid ( )anddimethylars are the arsenic species studied most in the literature. ~ n o r ~ a narsenics ic are more toxic than organoarsenicals, while trivalent arsenic corn ounds A are are more toxic than their pentavalent c o u n t e ~ ~ ~ s .A the two main metabolitesknown to be toxic. epossibilityof perfo~in~ uch as physiological fluids, in S studied. In aqueous solvents, the arsenic species dissociate to give ons. Inthisinstance, the cationic c e t y l t r i m e ~ y l ~ o n i ubromide m )became the surfactant of choice since it solutes into the micellesdue to electrostatic in themobilephase to avoid waskeptbelow 0.05 ositionof salt at the 'zer, spray c h ~ b e and r torch injector tip. wever, below 0.1 M luted at very long times (retention very broad erefore, 1-propanol was added to factor > 15) with the mobile phase to decrease on and improve the peak shape. ignal-to-noiseratiosimproved d r ~ a t i c a l l yfrom 5% to 10% (v/v) 101, due to improvedchromatographywhile having negligible dation on plasma stability or analyte sensitivi~.
Figure 12.8 shows a chromatogram of a ~ i x ~ofr e and ~ s ( V ) Total . arsenic found was 0.52 pg/mL which was in a ~ e e m e n t with the c e ~ i ~ value e d of 0.48 pg/mL. The last peak however was very represented probably more than one arsenic species.
.
Electrochemical detectors are very popular inliquid chromatography. Electron transfer processes offer highly sensitive and selective methods for detection of solutes in flowing streams. Various techniques have been devised for these measurements, with the most popular being based on the thousands
30
0
.
n U
Arsenic speciationin urine. Peaks1 and 2 are different forms of chlorine.
DMA is dimethyl~senicacid. Column:Hamilton PRP-l. Mobile phase:0.05 M CTAB-lO% l-propanol at pH 10.2. Reprinted from Ref. 16.
application of a fixed potential to a glassy carbon solid electrode. In spite of the wideuseof this detectiontechnique,onevery serious problem exists. As amperometric detection is based onan interfacial rate process, it is inherently sensitive to surface c o n t ~ i n a t i o n . ~ e r c udrops, r y which periodicallygivesrenewed surfaces, are diflcultto use in flowing streams. Solid electrodes, which are better suited for these systems, are easily fouled. any surface studies have been performed to better understand thesurface reactions that occur andprevent the fouling process [26]. Coating of solid electrodes with polymeric materials has beenreported to avoid the access to the electrode surface of large adsorbents, such as proteins. This approach is not applicable, however, tosmall analyte molecules whch adsorb during or afierthe electron transfer process. Furthermore, the electroactive species must difise through the coated layer to reach the electrode surface and this may increase peak variance and have adverseeffectson chromatographic resolution. While many otherexternal treatments work well, they are time consuming as the cellmustbedismantledandreassembledbetween trea~ents. Kirchhoff et al. [27] investigated the factors influencing electroanalytical measurements in aqueous sufiactant mediaandfound that anionic SDS, cationic CTAB and nonionic Triton X400 possess a wide potential window with low background currents within which electrochemical measurements can beconducted. W i l e surfactants adsorb onto electrode surface, electron transfer can still occur between solute and electrodesurface. The use of mobile phases contain in^ a surfactant to prevent the effects of adsorptive fouling ofglassy carbon electrodes has been reported [28]. Aromatic compounds were utilized as probe solutes, the oxidation of which proceeds through a radical cation mechanism. Two hypotheses wereconsidered by which s u ~ a c ~ might ts serve to reduce the adsorptive fouling of solid electrodes. First, micelles might solubilize the radical cations generated in the oxidation reactions and carry them past the electrode surface before adsorption can occur. h this sense, micelles of anionic s u ~ a c t ~would t s be best, as there will be both hydrophobi~ and electrostatic attraction of the radical cation to the micelle structure. Second, a cationic surfactant would adsorb onto the electrode surface and
electrostaticallyrepeltheradicalcations,thuspreventing the electrochemically generated products f?om depositing on the electrode su~ace. Figure 12.9 shows three series of sequential injections of p-nitrophenolin an aqueous-organicmobile phase and m phases of cetyltrimethyl~oniumchloride (CTAC) and S in response was virtually identical for the mobile phase with no s u r f a c t ~ t and for the SDS micellar mobile phase, but the response for CTAC is approximately constant. Possibly, enough of the negatively charged S was adsorbed onto the electrode surface and attracted the radical cations. Meanwhile, the carrier stream containing the cationic surfactant prevented adsorptive fouling and allowed accurate, reproducible measurements of many repetitive samples before electrode cleaning would be necessary. Indeed,noloss of electrochemicalresponse was observed after 55 sequential injections. As the preventive mechanism appeared to be a lack of adsorption of the surfactant onto the electrode surface and not any interaction with the micelles, sub-cmc concentrations of surfactant in the aqueous-organicmobilephaseswill probably show the same electrode stabilization.
c
Figure 12.9 Normalized current vs. injection number forp-nitrophenoland various mobilephases:O.lOMCTAC(~),O.lOMS~S()and acetonitrile-0.025M phosphate 7624 );pH was always 5.1. Applied potential: 1.l V vs. Ag/AgCl. Reprinted from Ref. 28.
The adsorbed surfactant can change the double-layer s t ~ c t u r ~ , the rate of electron transfer and the apparent half-wave potential, electroactive species. Thus, it is clear that electrochemical conditions d e ~ e l o p ~for d an a ~ u ~ o u s - o r g ~ separation ic are probably notdirectly transfera~le to micellar mobile phases, New hydrodyna~ic v o l t ~ o g r a m swillbenecessary to d e t e ~ i n ethe o p t i operating ~ ~ S cyclic v o l t ~ ~ o g r a mof s dopaminein ~icellarsolution. In micellarsolution, dopamine is oxidized at loweroperating potentials whichshouldbe a benefit for selectivity.Also the oxidizing current and the rate of molecular difision towards the electrode are lower, the latter produced by the higher local viscosity.
Cyclic voltammogramsofdopamine obtained with 0.05 MHCl (solid line) and 0.1 M SDS (dotted line) vs. SCE. Reprinted from Ref. 18.
a polyester sheet, the solution from the jet im e ~icroelectrodeand spread radially overthe
awbacks of electrochemical elution, which is intere of time and solventsavi the background curr e compositio~,so anyc the baseline. The gra~ent-i table value by a d j u s t ~ gthe c o n t r i b ~ t i ~ factors. ric detection, the application of a potential to the transient or charging current, which ckground current is composed of tvvo compo~entresults from electrochemical processes ode surface. For glassy c a r ~ o nelectrodes, these d from the oxidation of~nctionalgroups on the electro n. of oxide layer(§) whenever ox~genevolution cies and oxide layer, second component is uritiesin the solution ,the m a g n i ~ d eof the residual
is dependent on the surface condition of the solid electrode and the rate of impurity and solvent oxidation. The variation of mobilephasecomposition during gradient elution can alter the extent of the electrolysis of the mobile phase and impurities, thus resultinginachangeof the residual current which contributes to the baseline shift. Also, changes in the double layer of the electrode induces a charging current flow (at least transiently). h general, changesin the properties of the mobile phase, such as pH, viscosity, electroactive impurities, ionic strength andsolvent type, along other parameters such as temperature, flow-rate, cell resistance, potential and electrode sensitivity, influence the baseline shift [17,129).
O ! a
1 Specificconductance vs. micellar SDS concentration (top) and percentage of methanol in amethanol-water mixture (0.05 M NaClO,) (bottom). Reprinted from Ref. 1’7 with permission of the American Chemical Society.
orsey et al. [17, 301 demonstrated advantages associate a micellar concentration gradient in MLC. Conventional with gradient elution requires a long column reequilibration time before the next injection. However, in MLC, only the mixer and injector must be ~ u s h e dprior to the next injection. This is possible because any change in total surfactant concentration changes only the micelle concentration. The concentration of free surfactant remains appro~imatelyc o n s t ~ and t sinceonly fiee surfactant interacts with the stationary phase, m i n ~ a l columnequilibrationis necessary. This fact suggested thatmice~lar mobile phases might offer advantages for gradient elution with electrochemical detection. Due to the approximately constant concentration of free surfactant, it is likely that the double-layer s t ~ c ~ r e and electrode surface conditionswill also remain virtually unchanged during the course of a micelle gradient, reducing anypossible contribution to the baseline shift from these factors. *
Constant bulksolventcompositionduringamicelle gradient makes control of p a r ~ e t e r ssuch as pH, conductance and even mobile phase impurities, easier compared to aqueous-organic mixtures. constituent in the solvent, aqueous buffers can be used even in the presence of a small but fixed percentage of organic modifier. The conductance change of a micelle gradient is shown in Fig. 12.11. Since the conductance of micellar solutions is directly proportional to the concentration of ionic sudactant, this change can be minimized by adding more supporting electrolyte to the solution of lower aqueous-organic micelle on cent ration. This cannot beachievedin gradients as the conductivi~o f methanol-water mixtures passes through a minimum duringthe gradient. gradient concentration from 0.01 (wider than necessary for many pr was checked to give good results. Figure 12.12 shows the chromatogram of a mixture of eight phenols. This figure is a demonstration o f the slight dconditions. Compatibility gradient-inducedbaseline shift under CO of gradient elution techniques with is determined both by chromatographic and electrochemical conditions.
Micelle gradient chromatogram of phenols (ca.400 ng of each) with detection at 1.20 V vs. Ag/AgCl.Solvent A: 0.05 M SDS-3% l-propanol at pH 2.5 (phosphate buffer). Solvent R:0.1 12 M SDS3% l-propanol at pH 2.5. NaC10, was added to balance conductivity with solvent R. Gradient program: Solvent A to E3 in 15 min. Compounds: (1) hydroquinone, (2) resorcinol, (3) catechol, (4) phenol, (5)p-nitrophenol, (6) o-nitrophenol, (7)~-chlorophenoland (8) p-bromophenol. Reprinted from Ref. 17 with permission of the American Chemical Society.
auro and E. Pelezetti, Surfactants in Analytica Organized Amphiphilic ~ e d i
1. 2.
3.
~nhanced ~luorescence and Room ~ e ~ p e r a t u r eLiquid Phosphorescence Detectionin ~seudophase LC, Anal. Lett., 14: 1659 (1981). .H. F a t e ~ i , Separation and clic Aromatic ~ydrocarbons b y ~ L C u s i n ~ ~ i c e l l a Phase r ~ o band~luorescence ile Detector, JLiq. Chromatogr., 18: 2569 (1995).
4.
5.
6.
7. 8.
9.
Injection with ~~C for ~ h e r a p e u ~Drug c ~onitoring,Anal. Chem., 57: 1564 (1985). ~ ~ n andy L.J.~Cline a Love, ~ Determination of Drugs in treated Body Fluids by MLC with Fluorescence Detection, JChromatogr., 342: 293 (1985). aldonado and J.J. Laserna, Laser ~nduced e Detection of 3anned Drugs in Sport by al., 11: 107 (1992). . Villanueva C ~ a ~ and a s Alvarez-Coque, MLC: A ort thy ~echnique for the Determinution of antagonists in Urine Sa~ples,Anal. Chem., 71:3 19 (1 999). Ilkova, OPl lood ndSe with Fluorometri~~ e t e c ~ o Jn:,Liq. Chroma 3)-
10.
arcia ~ a n c ~ eA. z, Plant ~ r o w t hRegulators Detected by Derivati~e Fluoromet~, JChromatogr. A, 723: 227 (1996).
11.
12.
13.
14.
15.
16.
17,
18.
19.
20
*
21. 22.
Sato,H. Yoshimura, bi, S. Hatakeyama, E. shino and T. Yotsuy Determination of A1 man Serum by Kinetic D~erentiation ode Reversed-Phase C with ~luorometricDete on, Chem. Lett., 3: 203 (1996). . h i n , K. Harrington andvon W a n d r u s ~ aDetermination , of Steroids in Urine b y ~ L C with ~etectionby sensitize^ ~ e r b i u ~ ~luorescence,Anal. Chem., 65: 2346 (1993). R. Weinberger, P. Yarmchuk and L.J. Cline-Love, LC Phosphorescence Detection withM C and Postcolu~nReaction ~ o d e sAnal. , Chem., 54: 1552 (1982). . Heitkem~er,J. Greed and 3. Caruso, ~ C Pas-a ~ ~ tion of Alkyltin Compounds, Appl.
kochi, The Speciation of O~ganotin Compounds in Seawater ~ C / ~ C P - ~ S , ~ u n s 44: e ~56 1 K a g a ~ (1995). J.S. Wang, J. ey and J.A. Caruso, Arsenic J. Chromato~r. A, n by MLC with Detection, rsey, ~ y d r o - ~ r g a n iana' c ~ic~llar ctrochemical~etection, Anal. Chem.,
57: 2190 (1985). . Hu and €?L.Zhu, Direct Urine Injec~onwith A " . ometricDetection for Dopamine ~onitoring,J: Liq. Chromato~r.~ 14: 2755 (1991). azzeo, I.S. Km11 and etermination of ~ u m a Plasma n using Photolysis with ~lectrochemical~etectionin LC, J:Pharm. ~ i o m e dAnal., . 11: 999 ( l 993). eng, T. Li, H.M. Li and E.K. ang, Direct Injection of Udne and Determination OfAcetaminophen by MLC with a allJet Cell/Carbon~ibre~icroelectrode, Anal. Chim.Acta, 298: 4 15 ar Spectposcopy:eth hods lmm, ~ o ~ e c u lLuminescence and Applications, Wiley,ChichesitedKingdom (1985). J.L. Ward, G.L. Walden and J meforher, A Review of Recent UsesO~Phosphorimetry for OrganicAnalysis, ~alanta,28: 201 (1981).
23.
24.
25.
26. 27.
28
*
29. 30.
T.V o - D i ~ ,Room-~emperature Phos~horimetryfor Chemical Analysis, edited by P.J. Elving and J.D. Winefordner, York (1984). L.J. Cline Love,M. Skrilec and J.G. Habarta,Analysis byMicellein Solu~on, Anal. Stabilized Room~emperature ~hosphorescence Chem., 52: 754 (1980). A. SanzMedel, M.R. Femandez de la Campa, Hevia Temprano,B. Aizpun Femandez andV.M. Liu, Surfactantbased OrderedMediain AnalyticalAtomic Spectrometry~~alanta, 40: 1759 (1993). R.E. Shoup in HPLC: Advances and Perspectives, Vol. 4, edited by CS.Horvath, Academic Press, New York, 1986. J.R. Kirchhoff, E. Deutsch and W.R. Heineman, Factors In~uencing ~lectroanalytical MeasurementsinAqueous Su~actant Media, Anal. Lett., 22: 1323 (1989). J.F. Clos and J.G. Dorsey, ~nhanced~ t a b i lof~lectrochemical i~ ~etectionwith Surfactant Containing ~ o ~ iPhases l e in LC and low-Injection Analysis, Anal. Lett., 23: 2327 (1990). and L.R. Faulkner, ~lectrochemical ~ e t h o d s , 1s and Applic~tio~s, Wiley, New York, 1980. .G. Khaledi, J.S. Landy and J.L. Lin, ~ r a d i e n t ~ l u t i o MLC, n J: Chromato~r.,31 6: 183 (1984).
. uses micellarmobilephaseswith classical LC columns. This chapter expands the field to includesomemobile phases that can be considered close to micellar phases, such as normal and reverse microemulsions, bilesalt solutions, and surfactant solutions in supercritical fluids. Also, this chapter rapidly surveys the use of micellar mobile phases with LC stationary phases such as sizeexclusion or gel permeation polymer phases. Allied t e c ~ i q u e susing micellar phases such as ionexchange chromatography and capillary electrophoresis are also briefly presented. ~urfactantenhancedchemical separations are obtained through coacervation, liquid membranes, ultrafiltration, foams andor other interactions with phospholipids, proteins and biomolecules. these All topics were deliberately excluded. Theyare well exposed inthe literature [1-31.
. Chapter 2 discussed ~icroemulsionstructure. 'These organized media are stable and transparent. They are possible candidates for mobile phases in chromatography. ile salt solutionsare another kind of special micelles with chiral properties that can be used in MLCas well. Supercritical fluids ( were also used as su~actantsolvents to perform micellar SF of MLC.
a)
~ h y s i c o c h e ~ iStructure ca~
The effect of the addition of short chain alcohols onthe chromatographic selectivity and peak efficiency was extensively exposed in previous chapters. The addition of such alcohols to a micellar solution forms mixed micelles. This is the first step toward the achievement of microemulsions with ionic surfactants. The oil in water microemulsion(L1 structure, see Chapter 2) has acontinuousaqueous phase containingoilswollen micelles or an alcohol-surfactantinterphase layer. The micro~ropletsof oil stabilized by medium istransparent and stable, however it has a dynamicstructure. Then, it is interesting to see if L1 microemulsion mobilephases could be useful in
we 13.1 Mass phase diagramofthe system water-heptane-S~S-pentanol(ratio % w/w). The open areas show the clear microemulsion compositions. The circle focuseson MLC useful compositions. The corr~spondinginset figure shows the numbered microemulsion whose compositions are listed in Table13.l.Reprinted from Ref. 4.
6)
The System
Figure 13.1 shows a microemulsion systemthat was extensively studied in LC [4,5]. Appendix IV at the end of the book describes the experimental protocol that allows to obtain the mass phase diagram of a microemulsion system not found in the literature. The phase diagram of the heptane-waterpentano~/SDSsystem (Figure 13.1) was established using this protocol. Only compositions located in the water-rich corner of the diagram were tested. The cp parameter, in v/vpercentage, corresponds to the nonaqueous microemulsion content (oil +alcohol + surfactant). The high viscosityofthe o/w microemulsion systems with less than about 85% w/w water ('p>--20) precluded their use in LC.
Mobile Phases Microemulsion Compositions Corresponding to Fig. 13.l Microemulsion
#
% w/w
SDS
Pentanol %w/w
Heptane % w/w
Water % w/w
cp %v/v
l
2.88 2.73 2.75 3.52 1.44
5.76 5.5 1 5.5 1 7.00 2.87
0 0.3 l 0.59 l .44 0.34
91.36 9 1.45 91.15 88.04 95.35
8.64 9.2 l 9.60 13.37 5.09
2 3 4 5
cp is the organic volume fraction, 1-9is the aqueous volume Eraction. Data from [4].
c)
~ i ~ r o e ~ u l s for i o nthe s Rapid Screening of ~llegalDrugs in Sports
Oil in water microemulsions were first used as mobile phases in MLC to solve a practical problem: the screening of drugs illegally used in sports 14.1. Eleven drugs were separated on a 25 ~ 0 . 4 cm 6 C 18 5 pm column (Spheri 5, Brownlee Labs) with the five microemulsion mobile phases whose compositions are listed in Table 13.1 andrepresented by a dot in Fig. 13.1.
6)
The System
Figure 13.1 shows a microemulsion systemthat was extensively studied in LC [4,5]. Appendix IV at the end of the book describes the experimental protocol that allows to obtain the mass phase diagram of a microemulsion system not found in the literature. The phase diagram of the heptane-waterpentano~/SDSsystem (Figure 13.1) was established using this protocol. Only compositions located in the water-rich corner of the diagram were tested. The cp parameter, in v/vpercentage, corresponds to the nonaqueous microemulsion content (oil +alcohol + surfactant). The high viscosityofthe o/w microemulsion systems with less than about 85% w/w water ('p>--20) precluded their use in LC.
Mobile Phases Microemulsion Compositions Corresponding to Fig. 13.l Microemulsion
#
% w/w
SDS
Pentanol %w/w
Heptane % w/w
Water % w/w
cp %v/v
l
2.88 2.73 2.75 3.52 1.44
5.76 5.5 1 5.5 1 7.00 2.87
0 0.3 l 0.59 l .44 0.34
91.36 9 1.45 91.15 88.04 95.35
8.64 9.2 l 9.60 13.37 5.09
2 3 4 5
cp is the organic volume fraction, 1-9is the aqueous volume Eraction. Data from [4].
c)
~ i ~ r o e ~ u l s for i o nthe s Rapid Screening of ~llegalDrugs in Sports
Oil in water microemulsions were first used as mobile phases in MLC to solve a practical problem: the screening of drugs illegally used in sports 14.1. Eleven drugs were separated on a 25 ~ 0 . 4 cm 6 C 18 5 pm column (Spheri 5, Brownlee Labs) with the five microemulsion mobile phases whose compositions are listed in Table 13.1 andrepresented by a dot in Fig. 13.1.
~ r o e ~ ~ l § i #3, o n §#4 and
an early e l ~ t i peak n ~ whose width was related to the ~ i c r o e ~ ~ l s i o n
~e~tan
r-
Direct injectionof a urine sample after 50% dilution with the mi~roemulsion mobile phase. Column:25 cm x 4.6 mm Ld., 5 pm C18 Spheri5, flow rate 0.5 rnllmin, 10 pL injection,detection UV@254 nm. Left chromato~ams;Microernu~sion#l (Table 13.1) pressure drop 184 kg/cm2; right chromatograms:Microe~ulsion#2, pressure 160 kg/cm2. Reprinted from Ref. 4.
Two points were noticed.The first point concerned selectivity. drug retention could be modeled usingthe very simplerelationship: log k =a
cp + b.
(13.1)
The classical Ilk; vs. cp relationship was poorly linear. All the k vs. log cp lines were crossing together close to the coordinate range: log cp "1.18 k 0.04 and k = 0.5 k 0.05 [4]. The second point concerned peak efficiency. dramatic 3-time decrease of plate count was noticed between microemulsion # 1,#2 and #3 when the k factors were reduced by10 to 30%. To find the reasons for these observations,a complete study was done with the alkylbenzene homologousseries. d) ~ehavlorofthe ~ o ~ o l o g oAlkylbenzene us Series
ixteen different water-rich microemulsion compositions from Fig. 13.1 sted as mobile phases on a x15 0.46 cm Spherisorb 5pm C 18 column ,France). The alkylbenzenehomologous series, from toluene to decylbenzene, was used as the test solute [5]. SeZec~vz~. A linear relationship between the retention factors and the alkyl chain carbon number, nc, was obtained with all 16 microemulsion mobile phases:
k=an,+p
(13.2)
The regression coefficients were higher than 0.984, Such linear relationships were previously obtained with micellar mobile as phases detailed in Chapters 7' and 8. The slopes, a, and intercepts, p, of these lines were found to be related to the organic volume fraction, cp,of the microemulsion:
(13.3)
The e, f, g and h constants were determined so that the general empirical expression for the alkylbenzene retention factor with such microem~lsion could be expressedas:
"
"
.
I . " . . " -
" " " "."..v~"
organic volume fr~ctjonphi ("hv/v)
4
13.3 Retention factor of four alkylbenzenesversus the organic volume fraction, cp, of the microemulsion mobilephase. Inset: The same data plotted versus log cp seem linear.
Figure 13.3 shows the k vs. cp curves. The inset shows the k vs. log cp curves for the Composition with logcp ranging between 0.170 and 1.l:! (5% cp 13%). These curves seem linear. A regression analysis done on these points only returned the slopes and intercepts of the straight lines with
regression coeE1cients inthe 0.98 1-0.995 range. Furthermore, all the lines crossed together around the coordinate range log cp = 1.15 (cp = 14%) and k 5. Of course this result is an artifact with no chemical meaning. It shows that one should be careful of such possible mathematical artifacts when analyzingany set of experimentaldata with powerfulmodem so~ware. The onlysignificant result is that organic rich microemulsions havestrong a solvent power. The retention factors of all members of the alkylbenzene family become similar with a nilselectivity. ~ ~ c i e ~ It has c y been . shownthat anincrease of the microemulsion pentanol
content can increase the efficiency [S]. Conversely, a drastic decrease of efficiency has been observed upon increasing the heptane content in the microemulsion [S]. The toluene efficiency dropped fi-om 7000 plates with Microemulsion ##1 (without heptane) to 3300 plates with Microemulsion #3 (heptane 0.59% w/w). The decylbenzene peak efficiency was divided by 8 with the same addition of heptane(1150 plates and 140plates, respectively) [5]. It was speculated that the significant decrease in the kinetics of the solute exchange betweenthe stationary phase and the microemulsion droplets was linked to physicochemical structural changes of the system [5]. The large solubility powerofo/wmicroemulsion systems, up to 4 &L of decylbenzene in a 90% water rich system, produced a unique selectivity. ~ n f o ~ u n a t ethe l y low-efkiency problem hinders the use of such systems.
11.2. ~
a in Oil ~ ~ i~c r or e ~ ~ l s i o n
The use of water in oil microemulsions(L2 type) was also termed “normal phase MLC .” Dorsey was the first to use such reversed micellar mobile phases with two polar silica stationary phases, an unbonded andNH2bonded phase [6]. His goal was to suppress the retention and selectivity variations caused by the water content of apolar solvents. It was shown that the retention of phenol, naphthol and dinitrotoluene was not sensitive to the water content (range 0. l-1% v/v) of an AOT-hexane mobilephase. In this concentration range, water molecules are bound tightly to the ions (A sulfonatepolar heads and sodium counterions)with practically no free water molecules left to adsorb on the polar stationary phase.
TH
of water (2to 40% w/w) are added to hexane or W/AOT, becomes the ewaterlAOTmolarratio, ion factors of test solutes were dep heptane content. Fig. 13.4 shows variations were not monotonous.It was demonstratedthat the h dration of the polarunbsilicasurfacechanged signi~cantlyin the O< range. The surfactant and water adsorbed on the si1 ~ f o ~ i n ag composite layerso thxk that it changed the c o l u pe~eability and dead volume[7]. The combinedeffect ofstationary phase hydration and adsorbed layer was responsible for the observed retention factor variations (Figure 13.4). The M,,,,, ratio is directly relatedto the water droplet size so the retentionfactor of a test solute can be relatedto the s t ~ c t u r eof the L2 ~croemulsion. ~ ~ c i e ~The c y peak . efficiencies obtained with the L2 ~ c r ~ m u l s i o n mobile phases for various solutes were cornpared to the ones obtained with classical he~ane-2-prop~ol phases. A 2- to 3-fold lower plate count was o b ~ i n e dwith the L2 microemulsionmobilephases compared to the traditional mobile phases[6]. In another work, the efficiency obtained with T-water micr~mulsionwas in the 200plate range, 10 times alues obtained for the same solutes and the same column with tra~itional hept~e-propanol phases [73. nce again, the exchange of thesolutebetweenthe stationary phaseandreversedmicellesis slow. The efficiency problem in the case of ~ c r o e ~ u l s i omobile n phases is so serious that the practical use of these mobile phases in chemicalanalysis is questionable [S]. orphysicochemicalresearch such as~icroemulsion structure investigation^ MLC can help.
a)
~ h y ~ i ~ ~ c h e~tructure ~ical
The supercriti~alstate of pure compounds exists at elevated temperature and pressure. Above a compound-speci~ccritical tempera~reand critical pressure, the liquid state and gas state &sappear, replaced by a unique supercritical state. Supercritical fluidsassociate some propertiesof gas (low
Retention factor of polar solutes with L2 W O microemulsion mobile phases. Column: 15 cm x 4 mm id., 10 pm XWP 250 Grace bare silica, flow rate 0.5 mL/min. Reprinted from Ref. 7 with permission of the American Chemical Society.
viscosity, high compressibility) to some properties of liquids (density, solvent power). 'Ihs last property makes themuseful in separation and extraction. The solvent power of a supercritical fluid depends on its density, i.e. it is adjustable in changingtemperature or pressure. Carbon dioxide isthe most commonlyused supercritical fluid. Its critical parameters are 3 1.O"C (304. l K) and 7.38 MPa (73.8 bars, 1060p.s.i.). Propane can also be used (T,= 97°C or 370K; P, = 4.33 MPa or 43.3 bars or 620 p.s.i,), aswell as nitrous oxide, ethane, methanolor even water. The d i ~ s i o ncoefficient of solutes dissolved insuch phases is inte~ediatebetween the corres~ondi~g gas and liquidvalues. Supercriticalphases, CO2or propane, but even water, are rather apolar phases. The idea was to adjust their solvent power using surfactants inthereversemicelle fom. Only supercritical ethane and propane were able to solubilize the AOT surfactant and some water in reversed micelles [g]. Is it possible to use such phases in chromatography?
b) ice^^^^ SFC There are several appealing factors for the use of micellar supercritical phases in chromatography. The peak efficiencies obtained in SFC are higher than in LC because the solutediffusioncoeE1cients are higherin supercritical fluids than in liquids. In SFC, masstransfers are enhanced by the combination of high d i ~ s i o ncoefficients and low viscosities. This could compensate for the low efficiency induced by micelles. The polar aqueous core of the reversemicelles should allow the separation of hydrophilic or even ionicsoluteswith supercritical fluids. These polar compounds are difficult to analyze in SFC [lo]. ~ n f o ~ u n a t e lthe y , use of reverse micelles in supercritical propane was somewhatdisappointing. Phenol, naphthol and resorcinolwere separated using a 0.05 M AOT-0.25 M water reverse rnicellar phase in supercritical propane. However the ef3ciencies obtained were average, in the 5000 plate range (h = 10 dJ for a 250x 1m 5pm column [g]. Neither universal detector northe mass spectrometer could be useddue to molecules. The adjustment of the AOTlwater ratio required a ing arrangement. SFC as a separation technique is used less icellar SFC would need more work to fully understand the physical and chemical processes involved in the separation of polar and apolar solutes. At the moment, this path seems to be set aside.
lic
11.4. M i ~ e ~ l Bile a r Salt Mubi~e~ ~ a s e s a) Bile Salt ~escriptionand Properties
ile salts are biosurfactants naturally formed by cholesteroldegradation in the liver. Bile acids are stored in the gall-bladder. After derivation with glycine or tauric acid, the sodium salts of the derivatives are essential inthe digestion process to emulsifyfood triglycerides. "he pancreatic lipase enzyme can only split emulsified fatty globules. The three main bile acids are cholic acid, lithocholic acid and deoxycholic acid.
R1 H H
name H
lithocholic
Figure 13.5 shows that the steroid skeletonof the molecule has a polar side and a lipophlic apolar side. Micelle formation ispossible with asi~nificant solubilization powerfor apolar molecules. The bile salt cmc depends onthe pH and ionicstrength. It is approximately5-10 m M for cholic acid and 3-5 mM for the deoxycholic acid[1l]. The great and appealing property of the bile salt molecules is their chiral structure with 3 or 4 asymmetric centers. Investigating micellar bilesalt mobile phase in LC, it was hoped that their micelles would beable to distinguish enantiomers. b) ~ i c e l l aBile r Salts ~ o b i l Phases e
inze was the first to investigate the capabilities of micellar bile salt mobile phases [11, 121. He found that a significant amount (-5% v/v) of a long chain n-alcohol (pentanol, hexanolor heptanol) was useful to minimize the bile salt adsorption on the C18 stationary phase. A wide range of solutes could beseparated by thesephases, PA.Hs, quinones, steroids, indoles, polar and lipophilic vitamins. These phases were also able to resolve optically big enantiomers such as binaphthyl derivatives [121. Such compounds are
loyed in or~anic sy~thesis schemes for enantioselective synthesis. The s were thorou~hly e~aluated for thee~antioresolution ves with different anionic, catio~ic,~ o n i o ~ i c id derivatives 1113-151. ~ e l e c t i vvalues i~ a hthyl derivatives, However, the ~ o b i l ~ s i ~ c ite contai~ed7 0 vlv~ acetonitrile ers hases were not able to se~aratee ~ a ~ t i o ~with
The cholic acid molecule and its aggregation pathway.
ermeation Chromatography(GPC)is also called Size Exclusion Chromatography ( EC). The principle is simple:big moleculesor “objects” are sorted by size sing some kind of stationary phase acting like a sieve. e exact ~echanismis notactually a sieving process. The stationary phase contains pores o f a given diameter. “he molecules bigger than the pore size are not retained, they are “size excluded.” The smaller molecules visit the pores and need moretime to pass through the column. The peak retention volume on the chromatogram corresponds to the molecule size, the bigger molecules eluting first, the smaller ones last. The molecule sorting size depends on the pore distribution o f the polymer used as a stationary phase in the GPC column. a) S e ~ a r a t i oof~S ~ a l~l o l e c u ~by e sGPC
h 1964, Herries thought o f using micelles as the big “objects” not retained or excluded by aGPC polymer phase. Small molecules are retained by the pore of the stationary phase. They are less retained when solubilized inside an excluded micelle. The idea was to measure solute affinity for micelles through solute retention times. n s was the first time a micellar phase was usedinchromatography [l61. This part ofMLC history was already exposed in Chapter 3. Terabe and Okada developed a slightly different approach to model the small molecule and ion retention in micellar 181. The equation is:
1+ V(PW -1)[
v, KD + Pm -
(13.5)
vl
in which the subscripts i, R, o and S for the volumes, V,correspond to the internal pore, retention, void and packing polymer volume, respectively. The ,VVS and I) for the distribution coefficients respectively the solute micelle-aqueous phase coefficie (adsorption) of the solute for the polymer stationary phaseand GP
coefficient very close to unity for small solutes [171 (K, = l for ions [l S ] ) . [M] is the concentration ofmicelles. The plots ofthe left memberof eq. 13.5 versus the micelle on cent ration were linear. Their slopes and intercepts allowed to obtain the P and K parameters for polar alcohols and phenols.
6) Ion ~e~aratiun Okada separated ions using ion-exchanges between ionic micelles and the aqueous phase. The principle is similar: the micell~sare excluded from the pore volume of the polymer stationary phase; they travel faster than a small ion that visits all pores. When the ion is attached to a micelle, it also travels faster. Ion retention volumes allow to estimate ion micellar affinity. The surfactant-covered polymer stationary phase was also responsible for part e P of the ion retention [l 8, 191. Equation 13.5 was used to d e t e ~ i n the coefficients of several inorganic anions with cationic micelles [181 and several metal-cations with anionic micelles [191. Carboxylic anions were separated by mixed micelles made of SDS and Brij@ 35 (C 12E23) [20]. Several factors affect the selectivity in micellar CPC of inorganic anions [21]. The ionic strength and the amount of adsorbed surfactants were especially important. The pH and the micelle charge density also affected the ion separation. These factors indicate that ion-exchange play a significant role in the retention obtained in micellar CPC [2l]. This will be exposed further in Part IV thereafter.
a)
~ e s ~ r ~ tand i o Advanta~es n
The first uses of micellar phases by A ~ s t r o n were g done in CPC and thin layer chromatography (TLC). This was described in Chapter 3. TLC was a useful tool for the determination of solute partition data in micellar systems. The micellar pa~itioncoefficient, PwM,the solute-stationary phase interaction coefficient, Pws, andthe micellar binding constant, K,, could be obtained from the solute-Rf parameters with an equation very similarto eq. 13.5 [22]. A number of solutes were separated by micellar TLC such as phenols and dyes [22], indicators, caffeine, biphenyl, naphthol and benzamide [23], PAHs andamino acids [24], vitamins [25] or fluorescein derivatives [26]. The low cost, low toxicity, peculiar selectivity and ease of
operation of micellar TLC renders the technique very useful in teaching laboratories. A number of practical demonstrations can be performed at practically no risk for the students [23]. b) ~ r ~ ~ a c ~ s
Surfactant adsorption on the thin layer support cannot be avoided. It isthe cause of a micellar gradient concentration between the solvent front and the mobile phase reservoir. Adsorption upto saturation of the sorbent depletes the surfactant concentration in the mobile phase. A double solvent front was observed, the upper one was a dilute non micellar surfactant solution, the lower second front corresponded to the micellar front [22-25, 271. The problem is that this phenomenon introduces potential error in the identification of the exact Rf parameters of the solutes, inducing accuracy concerns of the K coefficients [28]. This adso~tion-inducedmicellar concentration gradient was used to separate the polar solutes in the nonmicellar region (between the two solvent fronts) from the hydropho~ic solutes separated by the slow moving micellar phase [22,27]. The typical elution times in micellar TLC are in the hour range.Such long development times are required due to the relatively high viscosityof the micellar solutions compared to organic solvents. The micellar phase viscosity was also responsible for the increase in spot size (low efficiency) plates because the [23] Small amounts of sample should be depositedtheon solubilizing power of micellar phases is lower than that of organic solvents. The last original stationary phase used with micellar phases is the wall of capillary open tubes. The forces driving the micellar phase are not mechanical (pressure) but electrical (electricfield). This is part of the world of capillary electrophoresis. It deserves special consideration found at the end of this chapter.
. When ionic solutes must be separated using MLC, electrostatic interactions will occur between the analytes and the ionic surfactants. Ion-pairing andlor ion-interaction can occur simultaneously with the ionic surfactant covered
stationary phase and with the mobile phase containing charged micelles. 1on-pairing complements the micelle partition mechanism described for nonionicsolutes.Theionretentionis the result of this multifarious mechanism.
ionswillbe separated by cationi 1> ullins obtained the elution order c e t y l t r i m e t ~ y l ~ o n i uchloride m herisorb column [29]. This is the usual eluotropic order obtaine~ basic anion exchangers. They showed that the ion retention creaseofboththeionic stre and the surfa~tant S of lk vs. p*'2and lk vs. were linear. The first linear relationship is typical of ion-exchange mechanisms, the plot means that the ani~n-micelleinteraction obeys the A ~ s t r o ~ ~ model for molecule-mi~ellepartition. ith a shorterchaincationic su~actant, ~exadecyl tr' chloride (HTAC), and a polar stationary phase, A s ~ i p a c ed a reversed selecti >I-. The ~ o ~ t adsorbed on the polar stationary phase is in the 0.5 pmole/m2 range, one order ofm a ~ i t u d elower than thesurfactant coverage ofapolar C 18 or C8 er 4). The ion-exchange mechanism is less impo~ant Ionic strength changes "transportation by micelles produced drastic selectivity variation micelles interacted with the GPC gel. the ionic strength wa -exchange elution orde both the micelle size and the s u r f a c ~ t a d s o ~ t i oThe n. ~ r e n ~increase^ h the second one favoredthe first effect decreased the micelle-gel interaction, i o n - e ~ c h ~ g e m e c h[18,301, ~ s m This shows that the anion elution order depends on the mobile phase composition and thenature of the stationary phase. ~0~~
L
a k ~ in g account three possible exchange equilibria, (i) s t a t i o n a ~ph ulk solution, (ii) charged micelles-bul~solution and (iii) com etitio ion-e~changesites b e ~ e e the n analyte ded to buffer the ionic stre equation [3 1:
(13.6)
The s~bscripts refer ie to the ion-exchange equili ionary phase and the solution-~icelleinterface, respecti~ely. entrations are the counter on concentration in the phase, aq, includin~added salts, and the one on the s t a t i o n a ~phas is the micellar counterion dissociation constant. @ is thec o l m pha llar concentrationandk is the anion retention factor. the classical ained from ion-exchange equilibria resembles of pa~ition coe~cient
(13.~)
a~ition coef~cient [3l]:
As soon as a moderate concentration of salt (C = 0.05 M) is added to the ,,P values can be regarded as constants when micellar phase,the Pw and the micellar concentration changes.Table 13.32 lists the ICieM, KieSand P, values o b ~ i n e dfor some ions[30,3 11. The K values depend onthe micellar ~ n c e n t r a t i obut ~ not onthe added salt concentration. The Pw constant is independent of the micellar concentrationbut decreases if the ionic strength increases. Partition VersusIon-Exchange to Model Anion MicellarInteraction.
I-
0.lM
1.77
l .37
35
0.lM
4.32
2.77
60
o. 1M
615 30.1
12.1 34 1 226
0.2M 0.3M
.
__
0.lM 0.2M 0.3M
2.16
l .74
38 24 16
0.1M 0.2M 0.3M
6.17
3S 3
100 57 43
" _ _ _ _ ~ ~ _ _ _ _ ~ ~ ~ ~ "
"-~-____
Micellar phase:CTAC 0.08M, Data from [30].
a) ~ e t e n ~ i o n ~ e h a v i o r
It is difficult to separate cations of the same charge by LC. Ligands are added to the mobile phase to form metallic complexes. This enhances the . Since complexationcan produce charged ion-exch~geselectivity in MLC or ~ o n ~ hspecies a r ~that ~ interact with all micelles, the inte~retationof the
retention data in MLC of cations is difficult [30]. Complexescan be involved in ion-exchangeequilibria and micellar partitioning simultaneously. Anionic micellar phases will interact with cationic species but cationic micellar phases can be usedas well to separate neutral or negatively charged metal complexes. Sodium diethyldithiocarbamate (DDTC)wasused to separate transition metal cation with a CTAB micellar phase andC 18 a column[32]. e limits of detection obtained with atomic absorption spectroscopy were in the tens of picograms injected. Since a highconcentration of I-propanol (45% v/v) was added to the 0.03 M CTAB mobile phase, the presence of ellesmaybe discussed. Simpleion-pairingbetwee TC metal species may explain the observed selectiv also used as a ligandfor transition metal cations with a phase and aG 18 column [33]. *
kada showed that it was possible to simultaneously separate nonionic compounds andcations in a complexed form[34,35]. The separation of phenols was optimized on a C 18 column with a §D§micellar mobile phase. Next, ~-hydroxyisobu~ric acid was addedto the S as a ligand for rare earth cations. This mobile phase wa at the samet h e , a mixture of 10 phenols and 15cations a 136 .
If the cations are separated directly with an anionic micellar mobile phase, the equations derived for the LC anion separation (eqs. 13.6, 13.7 and 13.8) can be used. Most often a complexing agent, a ligand to the micellar solutionforming the following equilibriawith
2t-
+ H2L
+
with the constant pm
(13.9)
L + H'
with the constant PI,
(13.10)
MHL'
and
Simultaneous separation of para substituted phenols (Bottom: detection U ~ ~ 2 nm) 8 0and cations (Top: postcolumn derivatization+ visible 540 nm). Column:15 cm x 4.6 mm i.d., 5 pm Inedsil OS-2, mobile phase: 0.1 M SDS + 0.08 M hydroxyisobu~ric acid, pH 4.05, 1 mllmin. Adapted from [35].
Sodium ions bufferingthe ionic strengthandor added as counterions of p bufTer salts compete withthe analyte cations. Ion-exchange with micelles:
2 ~ a ++ , M2+aq 2Na',,
+
2+m
with the constant
andasimilarion e x c h ~ with ~ e the st phase ( c o n s t ~ t da derived the equation for the place. ons side ring these exchanges apparent cation micellar partition coefficient 130, 331:
X
(13.12)
and the corresponding equationfor the apparent Pwscoefficient changingthe sub- or superscripts m by S 2331. In these equations, K2 is the second hssociation constant of the ligand molecule,H2L(tartaric acid, for example 1301). Experimentalmeasurementsshowedsome discrepancy with the predicted results. This was due to the shift of the dissociation equilibria of molecule in the presence of micelles. nce the micellar ApKa shift complex in^ mobile was taken in account, the cation retention with micellar phases was correctly predicted [30, 33-351.
mple ions can be separated by micellar mobilephases. Alth e ion retention can be interpreted on the basis of ion physicoche~calmeaning of such models ais little of ions withprobably nonmicellar phases contain of o r g ~ solvents, i ~ the omission of solvation size, shape or ionization state or local heteroge
4
the present models. However, these models have proved their usefixlness for the pr~ictionand optimization of ion MLC separations. Utilization of new original surfactants, fixnctionalized micelles or micellar catalysis will give rise to fixrther possibilities in the MLC of inorganic ionsstill modeled with simple ion-exchangeinteractions [301.
The column is the core of the LC separation technique. To expand the field LC, a rapid description of twoseparation techniques that do notuse a colurn, but canuse a micellar phase is presented. The twotechniques are field flow fractionation (FFF) and capillary electrophoresis (CE).
K 1.
~ l c e l l a Field r Flow ~r~ctionatlon
FFF was introduced by Giddings in 1966 to separate “objects” such as proteins, particles, latexes, cells,polymers,powdersandother macromolecules. FFF uses an external field applied ona long ribbonldcechannel. The injected “objects” are driven toward the lower wall of the channel. Because of the parabolic flow profile in the channel, the objects forced agai~stthe wall travel more slowly andare retained relative to the solvent or objects that interact less or do not interact with the field. The range in od that molecular weight is lo4to l 0l6 1361. A ~ s t r o n gand ~ e ~ h thought micelles and/or microemulsion droplets were “objects” falling in the FFF molecular weightrange. So, a micelle ora ~croemulsiondroplet could be retained in a FFF channel. Since small molecules can partition with the micelle or microemulsio~droplet, they could also be retained in the FFF channel by secondary chemical equilibrium as illustrated by Figure 13‘7. T h s would extendthe applicabilityof the FFF technique to s d l molecules. These two authors studied the theoretical feasibility of the use ofa micellar mobile phase in a FFF channel [3’7].
THE MLC FI
7
b) ~ e ~ a r a ~ i ~~ ~~l e c oubyl~~e ~i~c~e lal FFF al ~l
The theoretical study showed that micelles or microemulsi~ns droplets should be retained in an FFF channel. Injected solutes should partition between the aqueous nonretained and the organic retained pseudo-phases. The solutes are more or less retained according to their partition coefficient, . The solute retention time is:
Figure 13.7 Principle of micellar FFF. Field sensitive micelles move more slowly thanthe carrier. A: sample injection;B:the polar solute (darkcircles) moves withthe carrier, the apolar solute (hexagons) moves with the micelles. Reprinted from Ref. 38 with permission of the American Chemical Society.
(13.13)
the channel width and L is the average in which to is the dead time, is distance of the solute from the wall. This distance depends on the solute partition coefficient, but also on the field strength and on the micellar concentration inthe carrier [37]. igure 13.8 shows the first separation of small molecules by Ascorbic acid was separated from toluene through a sec equilibrium with field-retained microemulsiomdroplets. i. e., the exchange betweenthe aqueous phase and the swollen micelles is low, efficiency is low andbroad peaks are obtained (Figure 13.8). There are so many powerhl techmques for small moleculeseparation that micellar FFF was not used for this purpose. Its interest could be inthe physicochemical study of the micellar or microemulsion structure. For example, inthe case of the Figure 13.8 experiment, the separation allowed the estimation of the 1O-I6g)and average mass of the mobile phase microemulsion droplets ( 1 . 4 ~ consequently, its radius (35 nm) [381. These values can be obtained by S such as smallangleneutron sca~eringor highresolution hemeth icellar FFF can be an easy alternative in such studies.
The capillary electrophoresis (CE)technique can be used with micellar phases. The first use of such phases with CE was presented in 1984 by Terabe, who called the technique: micellar electrokinetic chromatography EKC) [39] Its success was tremendous becauseit opened the use of C to noncharged molecules and species, CE became an essential separation to ‘ally in the fieldofbiology. Today ten applications are published MLC. Numerous books and review articles in for only one in describe the CE technique includingthe use of micellar phases [40-43]. A simplified survey is presented here. I)
ur The first micellar FFF separation.Rotationspeed 1400 rpm,field 337 G, flow rate 0.22 mL/min. The system peak the is detector response with flow changes. Reprinted from Ref. 38 with permissionof the American Chemical Society.
a) ~ ~ i n c i p l e
CE exists in astrong electric field(4 to 70 kWm) appliedin an open-tubular silica capillary (id. 20 to 100 pm, length 20 to 100 cm)filledby an electrolyte. Two smallbeakersreceive the capillary ends and the two platinum electrodesof a high voltage unitthat generates the electric field. ~lectroo~motic Flow. Due to surface charges on the capillary wall, an electroosmotic flowis generated bythe electric field. This flow depends on the strength of the electric field, thetemperature, the pH7the ionicstrength, the solvent (viscosity and dielectric constant) and the nature of the charges on the capillary wall. Without surfactant and with anionic surfactants in the rnobilephase, the electroosmoticflowis oriented toward the negative electrode (cathode). The electroosmotic flow does not have the parabolic profile of any laminar flow in an open tube, rather its profile is flat. A narrow band injected at the beginning of the capillary travels with the electroosmotic flow without broadening due to the parabolic flow profile. ~lectrophoretic~ o ~ i l Any i ~ charged . species moves in an electric field toward the electrode of the opposite charge. The electrophoretic mobilityof the charge is related to its speed in the electric field. The electrophoretic mobility depends on the pH that can change the ionization state of the species, and onthe ionic strength and viscosity of the electrolyte medium. The electrophoretic mobilityof an ion can be in thesame direction or in an opposite direction as the electroosmoticflowdependingon its charge. Neutral solutes are not sensitive to the electric field, they move with the electroosmotic flow. Micelles of ionic surfactant are charged so they have an electrophoretic mobility. Neutral solutes can partition withthe micelles. The solutes will be separated according to their micellar affinity. Figure 13.9 illustrates the MEKC principle. b) ~hromatographic~ a r a m e t e r ~
~ e t e n t i o ~ .Micelles have a retention time, Lc7depending on their charge and size. A neutral solute that does not interact with the micelles, moves with the electroosmotic flow. Its retention time ist,,. All the other neutral solutes have a retention time, tR,between thesetwo values that delineate the time window,LC-t,,, of the technique. The retention factor, k, is expressed by [39]:
: : : .. . ..: . ...'...
2
Figure *I 3.9 Principle of MEKC. Top: anionic micelles, the electroosmotic flow has the anode (+)to cathode (-)direction. The anionic micelles are attracted by the cathode (+). Bottom: cationic micelles. Everything is reversed. Solute 1 (open bold hegagons) has a lower affinity constant for the micellar phase than Solute 2 (filled hexagons). eof = electroosmotic flow, mc = electrophoreticmicelle motion. Adapted from [46].
(13.14)
k is propo~ional to the solute micelle affinity coefficient, surfac~ntconcentration, C, L421: v (C, -cmc) =
C, the retention factor is not propo~ionalto the ret esoluteaffinity for themicelle is veryhigh (larg tb is not high but close to the retention factor is very high, the retention time, micelle retention time?L,. ~ ~ c i e ~ cE~ciencies y. in the hundredofthousandsplatelmrange are . T h s is due to the square nature of the electrophoretic obile phase flowing does not induce any band broadening as it does in all other chromatographic techniques. Longitudinal difision is the only p h y s i c o c h e ~ ccause ~ of band broadening. In this si~ation,the fision coefficient ofthe bulky micelles that was a problem in in A solute located apter 6) turns into a big advantage MEKC. a micelle has the low diffwsion coefficient of the micelle itself while it is retained. Consequently the observedpeak of this solute is very sharp. owever, if the plate numbers obtained with range) is impressive when comparedto those o b t a ~ e d CE that plate/m range), itis not as high as what can. be achieved with classic can approach the million plate/m. The resistance to mass transfer that is intro~ucedby solute pa~itioningbetween the bulkbuEer and the micelles was greatly responsible for the low MLC efficiency. It is again respo~ible M E ~ and C classic CE s~parationsC44, for the efficiency diEerence between 451.
~esoZ~tion.The resolution equation between olutes 1 and
2 is [40,42]:
In eq. 13.6, the three first terms are the classic Snyder resolution equation . A high e ~ c i e n c yand a high selectivityfactor are favorable to efil resolution. There is, however, an optimum rangeof k due to the limitedelutionwindow,around1 k < 5, withinwhich ma~imum resolution can be achieved. Outside this optimum, i.e., for poor or strong [46]. The last term of eq. retainedsolutes,resodecreasesrapidly G. It t&esinaccountthe t,,/t,, ratio that 13.16 isspecific to corresponds to the time windows of the technique. means t,, >> t,, so the ratio teo/tmcisverysmall. S is a favorable condition for a high resolution. Usually anionicsurfa er time windows than cationic micelles. The separation capabilities of KC is hgher with anionic micelles than those with cationic micelles [47].
c) Pract~caZpera at ions
A primary reasonfor the exploding interest in MEKC, both in thea c a d e ~ c and industrialarea, is its combination of high efficiency,versatility, speed, ease of use and adjustable selectivity. ~ p t i ~ i ~ a t i oActing n. on the instrumental parameters is the first way to optimize a sep~ation.The applied voltage,the capillary dimensions andthe injection mode should be optimized. The temperature should not vary and current should be kept low (inthe tens p heating by theJouleeffect.This is done by adjusting the electrolyte concentration and composition. Analysis reproducibility can be e ~ ~ cbye d carefil cleaning ofthe capillary betweenruns.
EKC can also be Optimized by controllingthe migration behavior of the solute. This is done through manipulationof the,,P solute binding cons~ntsand change of the elution windows. The type of su~actant,its
an
concentration,and the additionofmodifiers such as organic solvents, cyclodextrins, urea or even another surf'actant (mixed micelles) to the mobile phase are someways to control the solutemigrationbehavior [48]. Introduction of secondary chemical equilibria in the mobilephase, the primary equilibrium being partitioning with the micelle, can also greatly change the selectivity in MEKC. pH changes, ion pairing or complexation equilibria are examples of such possible secondary equilibria allowing to enhance a separation [41, 42, 46, 481. ost of these changes will affect some physicochemical parameters of the electrolyte, viscosity, conductivity or dielectric constant that will modify the current (Joule heating) or the electroosmotic flow direction and/or intensity. The instrumental parameters should consequently be adjusted. ~ ~ ~ Z i c ~ Figure ~ i o ~13.10 s . showstheelectropherograms of a complex
mixtureofpharmaceuticalprinciple with analgesic, antipyretic and antitussive properties obtained with different micellar phases. It shows the effiect of different micellar phases on the selectivity and resolution ofthe depharmaceuticalseparation, MEKC hasbeenusedin fferent separations in practicallyall chemistry related fields such asthe biologicalfield,arnino acids, peptidesandnucleic acid constituents;thebiomedicalandpersonal care field, pha~aceuticals, cosmetics,metabolites,vitaminsandclinical applications; the chemical engineeringfield,organic acids andbases,hydrophobicand structural isomers, inorganic and organometallic compounds; the agrochemical and foodfield,foodcontent, sugars, glucosides, fatty acids andlipids; the ,PCBs, water analysis and e n v i r o ~ e n field, ~ l prioritypollutants, P research withseparation of chiral rnolecul In conclusion, both the great resolving power and the ease of use of C explain the great interest that this technique receives. The limited windows is a disadvantage, however, the problem is greatly improved of mixed micelles, polymeric phases andorganic modifier t recent bench top mass spectrometers with electrospray MEKC withoutquick con.. introduction are able to workdirectly tamination of the ion-source. The MEKC coupling will further broaden the scope of the technique.
6
6+7 4+5
2
2
Figure 13.10 Effect of the surfactant nature on the electropherograms of a complex antitussive preparation containing 14 solutes (analgesic, antitussive, antipyretic principles, conservatives and excipients). Buffer: 0.02 M phosphate-borate, pH 9. Capillary 50 pm, 65 cm fused silica. Detection UV 210 nm. Adapted from [47].
an
It is perhaps comme ilfaut to finish this book with MEKC, the analytical technique that is most successful with micellar phases. the Atmoment, the is electrochromatography latest separation technique still under development columns (EC). EC is a microchromatography technique with filled capillary in which the pump is replaced by an electric field. The electroosmotic flow is solely responsible for mobile phase motion. The flat flow profile passes are retainedthrough the million plateh in efficiency.Neutralsolutes interaction withthe stationary phase. Thereis no theoretical objection to the use of micellar mobile phases withEC. ~nfortunately,it can be predicted than surfactant will adsorb onthe stationary phase andthat the solute mass transfer between the bulk and micellar phase will be slow. The ef~ciency obtained with micellar EC will be likely lower than that obtained with EC andhydro-organicmobilephases.However, the adsorption of ionic surfactants will completely changethe stationary phase surface potential, consequently it will change the electroosmotic flow and may produce origina and interesting results.
1.
2.
3. 4.
W.L. Hinze and D.W. Armstron~(Eds.), Ordered ~ e d i ain Chemical Separations, ACS Symp. Ser.342: ACS Pub., Washin~on (1987). E. Pramauro and E. Pelizzetti, surf act ant^ in Ana~ticalChemistry, edia, Applications of O r g a n i ~ e ~ A m p h ~ h i l i c ~ Comprehensive Anal. Chem. Ser.2LKX1, Elsevier, Amsterdam (I 996). S. Terabe and Z.Deyl (Eds.), ~ i c e l l eas s Separation~ e d i a , Issue, J. Chro~atogr. A, 780: 1-360 (1997). A. Berthod, J.J. Laserna and E. Carretero, Oil in at er ~icroemulsionsas ~ o b i l Phases e for ~ a p i Screening d of 1llegal Drugs in Sport,J. Liq. Chromatogr., 15: 3 1 15 (1992).
o, Oil in
5.
at er ~icroemulsionsas
. Chem., 64: 2267
(1992).
rthault, at er in Oil Anal. Chem. 62: 1402
7.
?
(1990).
9.
10.
11. 12.
13.
14.
15.
16.
,~icroemulsionsin Separation Sciences,J Ch~omatogr. A, 780: 93 (1997). . Smith, Reverse ~ i c e l l e Supercritical ~ l u i dChromatography, Anal. Chem., 59: 1977 (1987). et, La C h r o ~ a t o ~ r a ~en hie Est-elle Possible??Analusis, ~
New
,Z.S. Fu and W.L. Hinze, ~ i c e l l a rBile Salt for the LC Separationof Routine Compounds and ~ o b i l Phases e Optical, ~ e o ~ e t r i cand a l Structural Isomers,J Chromato~r.Sci. 28: 292 (1990). ,T. Takeuchiand H. aguchi, Retention ~ e c h a n i s mof omeric Se~arationbywith ~icellar~ile-~alt~obile ~ h a s e sChro~atographia,3 ,T. Takeuchi and H. tiomeric Sepa~ation of 1,1'-Bina~~thyl-2,2'-diyl ate by ~icrocolumn LC with ~ i c e l l a Bile r Acid~erivativesas ~ o b i l Phase e Ad~i~ve, HRC-J of ~ i g Resolution h Chromato~r.,15: 275 (1992). U, T.Takeuchi and aguchi, Retention Behaviour of lnaphthyl Compoun~si tiomeric Separation by ~ i c r o column LC with ~ i c e l l a rBile-salt ~ o b i l e ~ ~ a Chromatoses, ~raphia,33: 58 (1992). chards, The partition in^ of Solutes between ~ i c e l l a rand Aqueous Phases: ~easurementby Gel ~iltrationand Eflect on the Kinetics of Some Bimolecular ~eactions,J Phys. Chem. 68: 1842 (1964).
17.
18. 19. 20. 21.
22.
23. 24.
25.
26.
27. 28.
29.
30.
Tanaka, K. C)tsuka and T. h d o , Separation of Small ith SizeExclusionChromatography Columns and Micellar ~olutions,J Chromatogr. Sci., 27: 653 (1989). da, ~ i c e l l e ~ x c l uChromatography sion of InorganicAnions, em., 60: 1511 (1988). elle ~xclusionChromatograp~yof ~ e ~ - m e t a l nal. Chem., 60: 12116 (1988). on-exclusion Chromatography with ~ i ~ ~ei cde l l a r es, Anal. Chim. Acta, 230: 9 (1990). ctors Aflecting Selectivi~in Micelle Exclusion hy of Inorganic Anions, J Chromatogr., 538: 34 1 . Amstrong and R. Terrill, TLC Separation of Pesticides, ~ecachlorobiphenyland cleosides with Micellar Solutions,Anal. Chem., 51: 2160 (1979). .A ~ s t r o n gK.H, , Bui and R. Barry, Use of Pseudo-Phase em. Ed., 61: 457 (1984). in ~eachingLaboratories, J . Amstrong and NI. y, Use of ~ i c e l l e sin theTLC ration ofpolynuclear c Co~pounds and Amino Acids, Anal. Lett., 12: 1285 (1979). ei and Y. Cheng, Separation and Detection of tami ins B1 B6 (Pyridoxine)a n d ~ i c o ~ i n a mini ~ e iamine), B2 (~ibo~avine), Compound Vitamin B ~abletswith Use of Mlcellar TLC, Yaowu~ e n x i - ~ a ~10: h i 37 , 1 (1990). Parshina and S.S.Lopukhova, Use ykov, E.G. Sumina, E.V. lar Mobile Phases for Separating Fluorescein~erivatives b Means of TLC, JAnal. em., 50: 684 (1995). ,A m s t r o ~ g and K.H. i, Use ofMicellar ~ o b i l e ~ ~ina s e s RP-TLC, J Liq. ~hromatogr.,5: 1043 (1982). . Armstrong and G.U. S h e , k valuation of Partition Coefficients to ~ i c e l l e and s Cyclodextrins via Planar Chromato-g~aphy, J Am. Chem. Soc., 105: 2962 (1983). ,GP. NIull~sand G.F. Kirkbright, Determination of Inorganic Anions by Hplc Usinga Micell~r Mobile Phase,Analyst, 109: 12 17 (1984). kada, Micellar Chromatography of ~norganicCompounds, J Chromatogr.A, 780: 343 (1997).
31.
32.
33
*
34. 35. 36. 37, 38.
39.
40, 41.
42.
43
(I
.Shirnizu, Retention Mechanism ofAnions in MLC: Interpretation ontheBasis of an Ion-Exchange Model, J: Chromatogr. A, 706: 37 (1996). .S.Andres, S. 'Vera andM.L. Marina, ~eterminationo f ~ i ~ I ) , Co(II) and Cu(IiJ as ~iethyldithiocarbamatecomplexes by HPLC using a CTAB Mobile Phase, J: Chromatogr. A, 685: 27 1 (1994). T. Okada, Interpretation of ~ e t e n t i o n B e h ~ i oorfs TransitionMetal Cations inMLC using an Ion-ExchangeModel, Anal. Chem.,
ultaneousSeparation of Ionic and on ionic g RP-MLC, Anal. Sciences, 9: 59 (1993). taneous Cation and Reversed-Phase Chromatography, J: Chromatogr., 607: 135 (1992). J.C. Giddings, Field Flow Fractionation,Anal. Chem., 53: 1l7OA od and D.W. Amstrong, Theoretical Study on the Use of Secondary Equilibriafor the Separation of SmallSolutes by FFE Anal. Chem., 59: 2410 (1987). ,D.W. Amstrong,M.N. Meyers and J.C. Giddings, Use for the Separation of SmallSolutes byFW, ry Equilibria Anal. Chem., 60: 2138 (1988). S. Terabe, K. Otsuka, K. Ichiwa,A. Tsuchiya and T. Ando, Electro~neticSeparationswith Micellar Solutions and OpenTubular Capillaries,Anal. Chem., 56: 11l (1984). P .G. hghetti (Ed)., Capillary Electrophoresis in Analytical Biotechnology CRC Press, Boca Raton, FL (1996). .J. Sepaniak, A.C. Powell, D.F. Swaile and R.O. Cole, Funda~entalsofMEKC, in CapillaryElectrophoresis, P and J.C. Colburn (Eds.), Ch. 6, pp. 159-189 Acadern New York, NY (1992). K. Otsuka and S. Terabe, MEKC, in Capilla~ ~~ectrophoresis ~ u i d e b o oK. ~ ,Altria (Ed.),Methods in Molecular Biology, 5 2 , Ch. 12, pp. 125-155 Hurnana Press Inc. Totowa, NJ (1996). N.A. Guzrnan (Ed.), Capillary Electrophoresis ~ e c h n o l oChro~, matographic ScienceSeries, 64: Marcel Dekker Inc, NewYork, (1994). M
The enclosed CD-ROM contains everything needed to run MICHROM. This software is able to take the results obtained with a set of compounds and several Compositions of hydro-alcoholic micellar phases, and calculate the affinity constants to predict the results for compositions of mobile phase (surfactant concentration, modifier concentration and pH). ICHROM includes the first module, MICHROMl ,used to rocess the data for the user selected model andto generate a series that canbeusedby ICHROM2. The second module can produce contour plots, 3D surfaces, chromatograms for a given composition and even dynamic simulations. The required first step is to read the User’s Manual found on the CD-ROM. Acrobat Reader@,from Adobe Systems Incorporated, can be installed from the CD. It should be used to visualize andlor print the User’s Manual PDF file. It is recommendedto print the 40 pages of the user’s manual titled “Getting Started with MICHROM.” The figures of user’s manual are in color. They can be seen on a color screen using Acrobat Reader@ andcan be printed as such with a color printer.
It is possible to try MICHROM directly from the CD-ROM using one of the .CRS files loaded on the CD-ROM (fases~l.CRS, mic 1.CRS or sulfa1 l .CRS). In order to process a personal set of data, it is necessary to install M I C ~ R O M land 2 onto your hard disk. The installation is straightforward by just copying the MICHROM files into a dedicated directory.
The chromatographer is always wondering where heshe orcan getthe cmc of a particular surfactant. Since MLC makes exclusive use of surfactant solution above the critical concentration, we thought the reader would appreciate using our collection of cmcvalues.When the physical parameters needed in MLC were known, they were included. This table was prepared by and is intended for chromatographers. ~hysico-chel~ists would complain that the methods used to obtain the cmcs were not listed. Furthermore,most of the valuescamefrompreviouscompilations, especially the 1993 compilation of Van Os, Haak and Rupert (l) and the NSF (now NIST) compilation of 1971 by Mukerjee and Mysels(2). These references contain the original referencesfor the methods used. The table issorted first byglobal surfactant formula,thenby increasing temperature and finally by additive. Whenseveral different values for the cmc of a particular surfactant were found, further references were examined (see Additional Reference list) and the most often cited valuewasselected. This wouldbelisted in Ref. 2 as a “questionable criterion.” l)the global surfactant formula, The 13 columns give successively: 2) its molecular weight, 3) its charge with S (both the anionandthecationhave zwitterionic, and of course, + and - for cati
4) the t~m~erature, 5) the
in molelL, 6) and 7) the medium with nature and co~centration,respectively, 8) the molar volume in Llmole, 9) the micel~ea ~ ~ r e ~ a tnumber, i o n 10) the Krafft temperature for ionic surfactant of a 10 g/L solution in "C, 1 1) the counter ion binding, whose micelle has 62 molecules means that 65% of the 62 molecules have a sodium ion associated, the micelle bears about 22 ne~ative char~es, 12) the reference and 13) the surfactant chemical name.
a,
gP
U
a8eq:
rlr If
0-
N N I
d-
N
W
I
N N 0
d-
W
O
00
m
J W
v10
W N
0
-
I
Nd- C I
N
d-
W
_ _ I _ _
506
r-I’.
t v?
-e
1: N
I
d
M
Y
x, x,
l
ERTHOD and G A R ~ ~ A - A ~ V A R
t
doo
O M -4N
N v3
d
507
"
U t C
v
3 .
~
4fg
d o
5
$4
w
0
"
W
2
I
i j
t
3
13
a8.1eq:
m o O O O O O O m m U N N N N N P 4 N P 4 N d . U
N
t
E. O
c?e c
o e
U
m c N
I
2 m
B
N U
0000
m m 0 m m m
and ~ A ~ ~ ~ A - A L V A
Z 3
n -4
N
2
i
2
L
11
l 2222
#I+ + + + + + m
m
0
+ W
m m
ov,v,v,c m m m m V :
v, W
9.
gs;,
c,
LL
8 P.
c
\o
N
N
3 7
E
v)
0
9
0 -
3
t
W
+
T-
d.
W
v3 W M
N
tn
N c-
0 0
3
M
%
N
22
N 0 CV
-
+
+
0
M
o -
5
d-
z N
E3
m
3
N -
0 (-.l
M
z Sd-
I
sz! o
E M
I
N
v)
-
l
P-
M
W P-
n
W
I
n v)
2 z v)
2 3
v
L
I
PS
D
3
nd
r I
t r3
I +
Z En O
0 0
I
I / +
+
r l
i
.l
a8.q
W
N 0
UD *
0 0
%
et
0
v7
cu
0
N
N
0
‘ c -
> U
0
t l
ovtv,v, \ D c u N N
CRITICAL ~ I C E ~ L E
CO~CE~TRATI~~~
U
n
U
519
C U
-
m N 0 0
z v7
m
XI
9
U
vs v7
“c
r*.(
v,
0
z;
v,
M
I _ _
F
~
0
v,
0
F
0
0
8
N
v,
N
z 2
z
W
m Q
c-4
22
0 0
m
I?
r-
oc
5
h 4
-
E
a9mqc
cr;t
L>
x)
x)
If:
c-
P-
2
A
C C
V
n
~
F-
3
2
k
A C
E
-
n
3
?
3
ICELLE C O N C E N T ~ T I O N ~
A
~
A
W
0
52
0
13
u n
3
x
ec
I, -
3
1.
2. 3. 4. 5. 6.
7.
9.
10. 11.
12.
.Van Os, J.R. Haak and L.A.M. Rupert, Physico-Chemical Properties of Selected Anionic, Cationic and on ionic Su~actants Elsevier, Amsterdam (1993). P. Mukerjee and K.J. Myseis, Critical Micelle Concentration of ~ q u e o uSurfactant s S y s t e ~ sN, S ~ S - N € 3 36, S U.S. Department of erce, Washington (l 97 1). urrelandR.S.Scheschter, Microem~lsionsandRelated Science Series,Vol. 30, M. Dekker,New York S y s t e ~ sSurfactant , (l 988). G.C.Kresheck,in F. Franks(ed.), at er, A Comprehensive Treatise, Vol. 4, Plenum Press,New York (19’75)Ch. 2. A.Berthod, I. Girardand C. Gonnet, in W. Hinze and D.W. Ar~strong(eds.), Use of ~ r d e r e dMedia in Che~icalAnalysis, ACS Syrnp. Ser. Vol. 342, ACS, Washington (1987). S. Puvvada and D. €31ankschteinyin K.L. Mittal and D.O. Shah (Eds.), Su~factantsin Solution, Vol. 1l, Plenum Press, New York (1991) pp. 95-111. P.M.MacDonald, J.R. Rydall,S.C. Kuebler and F. Winnik, SynthesisandCharacterizationof a ~ o ~ o l o ~ Series o u s of ~witterionic Su~actants based on ~hosphocholine,L a n ~ ~ u i7:r , 2602 (199 l). J. Alnrhar, Y.Chevalier, €3.Gallot, P. Le Perchec,X. Auray andC. Petipas, Solution Propertiesofzwitterionic ~urfactants,L a ~ ~ ~ u i r , IO: 3435 (1994). s ~~rfactant A. Cipiciani and S. Prirnieri, The ~ ~ e coft~witterionic Systems upon~ ~ o m a t~ucleophilic ic Substitutions, 3: Chem. Soc. Perkin Trans. 2, 1365 (1990). N.Nishido,in K. Oginiand M. Abe(ed ,Mixed ~urfactant S y s t e ~ sSurfactant , Science Series, Vol. 46, Dekker, New York
. Yamauchi,Y.Yoshida, T. Moriya, K. To inoshita, ~rchebacterialL ~ i dModels:Form Vesicles from Single sopr re no id Chain-Amph~hiles,~ i o c h i ~ . iop~ys.Acta, 1193: 41 (1994). Zana and H. Levy, Mic~llarProperties of ~ ~ y ~ r ~ x y a 2 ~ 1 tri~ethylammonium~ r o m i ~ in e sAq~eousSolutions, 3: C o l l o i ~ ~nterfaceSei., 1’70: 128 (1995).
13.
R. Zana, Aqueous Surfactant-Alcohol Systems: A Review, Adv. Colloid Interface Sci., 57: l (1995).
J, Cross, Anionic Surfactants: Analytical ~ h e ~ i s t rSurfactant y, Science Series, Vol. 73, M.Dekker, New York (1998). H.W, Stache, Anionic Surfactants, Organic Chemistry,Surfactant Science Series, Vol. 56, M. Dekker, New York (1996). D.W. R u ~ i n and ~ h P.M. ~ o ~ l a nCationic d, Sur~actants,Surfactant Science .Dekker, New York (1991). .L.Laughlin, The A~ueousPhase ~ e h a ~ i ofrSurfactants, ~ c a ~ e r n i c Press, London (1994). E.G. Lornax, Am~hoteric Su~actants, Surfactant Science Series, Vol. 59, M.Dekker, New York (1996). .J.Schick, ~Onio~icSurfactants: Physical Che~istry,~urfactantScience Series, Vol. 23, M.Dekker, New York (1987). N.M. Van Os, on ionic Su~actants: Organic Chemistry, Sur~actant Science Series, Vol. 72, M. Dekker, New York (1997).
Exposed in several chapters is the fact that a strong point of Micellar Liquid Chromatography (MLC) isits capability to give the micellarpartition coefficient of the analyzed solute. The Armstrong-Nome equation was presented in Chapters 3 and 5, Equations 3.4and 5.1. The variations ofthis equation were also exposed (Eqs. 5.8 and 5.9). Chapter 5 discusses the unit problem that exists in the literature listing micellar partition coefficients. The Armstrong-Nome equationgives the ,,P solute-micelle affinity coefficient. It isa dimensionless coefficient but valid for one surfactant molecule. The true solute micellar partition coefficient ,is,P x N with N the micelle aggregation number. Equations 5.8 and 5.9 giveICAM, an equivalent coefficient whose dimension depends on the concentration unit used. All equations relate linearly Ilk, the retention factor, with[M], the micellar concentration (= surfactant concentration -cmc) in mol/L or in g/L. The relations between the two coefficients are: ISAM
= slope/intercept = U (P,,
or P,
= K,,/u
+1
- 1)
(111.1)
(111.2)
The possible units for U,the surfactant molar volume,are g/L,or mol/L. To add to the confusion, the molar volumeof CTAB is 0.999 g/mL. This means 27
that the value of the CTAB ,,P coefficient is numerically the same as the corresponding K A M value expressed in glmL. The U value of SDS is 0.854 mL/g producing K,, values numerically 15% lower than the corresponding P,, values. This 15% difference is within the error margin obtained in such coefficient determination (accuracy -20%). The solute affinity for the surfactant covered s t a t i o n a ~phase is ,,P or K A S . These two coefficients are iden is rarely known, the dil~ensionlessproduct able was prepared after reviewing original slopes and intercepts of the Ilk versus [M] lines were given, the K,, value was calculated according to Eq. 111.1 and the P,, value was taken as ,,P and the other K A M values were calculated using Eqs. llintercept. Then 111.l and 111.2. A modern data processor was used producing 12 digits in all calculations. Due to the seven orders of magnitude range of the P or K values, it was not possible to round these numbers correctly. Only the first two digits are significant, e.g. the listed P,, value of 956 in CTAB for dinitrophenol should be read as 960 &l90 or log P,, = 3.0 f 0.06. The dinitrophenol micellar partition coefficient, P,can be estimated as 86500 since the CTAB aggregation number is 90 (Appendix 11) or log P = 4.9 f 0.06. The aggregation number changes if any modifier is added to the micellar phase (see Chapter 2 and Appendix 11). In a lower extent, the molar volumes are also affected by the modifiers. However, the U values in aqueous micellar phase were used throughout the table for calculations. The table is first sorted by global solute formula, then by alphabetic order of the isomers and finally by surfactant in the mobile phase. The 1 l columns give successively: 1)the global solute formula, 2) its name, 3) the stationary phase used, 4) the surfactant used in the micellar phase, 5) the additive in the micellar phase, 6) the temperature, 7 ) the dimensionless P, value, 8) the K,, value in Llmol, 9) the K,, ,value in mL/g, 10) the dimensionless K,, value or the (bK,, value and 1 l) the reference. The reference titles were omitted to save space and because they are already listed in Chapter 5 and elsewhere in the book.
m p .
N N ’ c ”
E E E
1 2 %
I
l
l
v l v l v l
000000
0 0 0
I
-
compound
,C6H5CIO
C6H5CIO C6H5CIO C6H5CIO C6H5CIO
name
stationary phase
iurfactant
chlorobenzene chlorobenzene
Spherisorb C8 - 5pm Spherisorb C8 - 5pm
SDS SDS
chlorobenzene chlorobenzene chlorobenzene chlorobenzene chl~robenzene chlorobenzene 2-chlorophenol 2-chtorophenol 3-ch~orophenof 3-chlorophenol ~-chlorophenol o-chlorophenol ~-chlorophenol
SDS SDS SDS SDS SDS SDS SDS SDS SDS
nitrobenzene nitrobenzene iitrobenzene iitrobenzene iitrobenzene
Altex ODS Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Rainin Microsorb C 18 3pm Bakerbond C 18 Nucleosil C18 5pm Bakerbond C 18 Nucleosil C18 5pm Bakerbond C 18 Rainin Microsorb C18 3pm Rainin Microsorb C 18 3pm pBondapak C 18 Yovapack C18 4pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm
iitrobenzene iitrobenzene
Spherisorb C8 - 5pm Spherisorb C8 - 5pm
2-chlorobenzene- 1.4-diol
SDS SDS CTAB SDS SDS Brij-35
CTAB CTAB CTAB CTAB CTAB CTAB
additive % vfv
roc
5% n-BuOH 25 25 10% n-BuOH 3% 2-PrOH 3% 25 10% BuOH 25 5% BuOH 25 NaCl0,tM 25 25 2% 1-PrOH __ 40 3% 2-PrOH 40 40
40 3% 2-PrOH 40 35 2% 1-PrOH __ 25 24 __ 25 25 3% n-PrOH 25 5% n-PrOH 25 25 10% n-PrOH 3% n-BuOH 25 5% n-BuOH 25
pWM
rc,MorK2
KAM
Or K t
(bpws or
ref:
tLfmol)
fmW
KAS
116 72
28 18
99 61
44 16
8 8
68 279 188 221 334 294 26
57 238 160 187 285 250 21 15
227 I69 40 77 130 147
I1
8.1
3 -
23 34 417 76 30 52 150 119 102 84
14
4 5 4
34 60 151 120 103 85
16 68 46 54 82 72 6.2 4.2 9.0 6.5 10 120 22 8.6 62 55 43 37 30
172 18 3.7 38 59 41 35 24
2 45
29 16
81 43
22 13
38 27 41 489 90
31
7
9 9 9 6 4
6 6 I0 7
8 8 8 8 8
8
E 2
m
E;"
4 M
1
compound
name
stationary phase
surfactani
additive
-
r oc
WM
(L/mol)
% YfV
Nucleosil C18 5pm Rainin Microsorb CIS 3 p m Supelcosil LC-1 5 pm Micropak C 18 1Oprn Micropak CN lOpm Supelcosil LC-1 5 pm Rainin Microsorb C 18 3 p m Supelcosil LC-18 Supelcosii LC-18 Radial-PAK C 18 - 1Opm Radial-PAK C 18 - 1 Opm Novapack C18 4pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5prn
SDS CTAB DTAB SDS SDS SDS SDS Brij 30 Brij 30 Brij-35 Brij-35 Brij-35 CTAB CTAB CTAB CTAB
Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm
CTAB CTAB CTAB
AItex C8 Ultr~phere Altex Octyl Ultrasphere Novapack C 18 4pm Novapack C 18 4vm
CTAB CTAB CTAB CTAB
I
re$
M Or
3% 2-PrOH
40 35 25 22 22 25 2% 1-PrOH 25 60% ACN 60% ACN
I5 468 2 33 34 78 62 40 40 54
3.5 170 61 8.1 8.2 19 15 19 19 13
12 467 200 28 28 66 52 51 51 45
10 233 120 14 1.5 14 7.7
85
6 I2 13 13 12 6 14 14
I5 15
15% EtOH
3% n-PrOH 5% n-PrOH 10% n-PrOH 3% n-BuOH 5% n-BuOH 10% n-BuOH 3% 2-PrOH
5% BuOH
5
~
7
25 25 25 25 25
34 131 131 4 26 104
35 47 47 46 37
29 130 130 125 103
36 55 49 47 34
25 25 25
102 64 49
37 23 17
101 63 48
31 21 12
8
38 31 25 25
26 67 113 75
9.1 24 41 27
25 66 I12
51 52 43 24
11 16 7 9
74
8
8 8 8
8 8
I
name
stationary phase
surfactan -w
benzene benzene benzene benzene benzene benzene benzene benzene benzene benzene benzene benzene
Rainin ~ i c r o s o r bC 18 3pm Supelcosil LC- 18 Supelcosil LC-18 Supelcosil LC-I 5 pm Hypersil ODS C 18 Hypersil ODS C 18 Supelcosil LC-1 5 pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm
benzene benzene benzene
Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm
SDS SDS SDS
benzene benzene benzene benzene benzene benzene benzene 3enzene
Altex ODS pBondapack 6-18 (Waters) Rainin Microsorb 5pm octyl Rainin Microsorb 5pm octyl Rainin Microsorb 3pm - C 18 Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4ym
SDS SDS SDS SDS SDS SDS SDS SDS
CTAB DOSS DOSS DTAB MB-14 SB-12 SDS SDS SDS SDS SDS SDS
additive % vlv
_ I
r
W W S 01
_.
re$
KAS
35
100
25
69
25 25 €O% MeOH 25 3% n-PrOH 25 5% n-PrOH 25 10% 25 n-PrOH 1% n-BuOH 25 5% n-BuOH 25 10% 25 n-BuOH 3% 2-PrOH 38 25 31 3% PrOH 31 31 25 10% BuOH 25 5% BuOH 25
78 97 72 79 65 70
36 15 15 21 63 34 19 24 17 19 16 17
99 35 35 68 212 101 66 82 61 66 55 59
27 36 34 23 37 23 25 25 19
6 I4 I4 12 I7 I7 I2 8 8 8 8 8
68 55 45
16 13 11
57 46 38
19 18 9
8 8 8
24 103 70 66 86 79 78 66
5.6 25 17 16 21 19 19 16
19 87 59 56 73 67 66 56
62 27 56 43 82 38 16 21
I1
39% ACN 39% ACN
48
I8 I6 16 I6 7
9
9
13
d
oi
v)
2213 2
d
v)
&
S ~ ~ F A C T A NAFFINITY T COEFFICIENTS
M P-
r3
N
2 2 2 2 2 2 2 2 d000000
v)
2 m
2
535
1
1
:
000000
u u u
E
2
m
K jK
37
538
E
I
l
I
El
I
1
and ~ A R ~ ~ A m A L V A R E ~ = ~ O ~ U E
001
E 2
m
00
I
I
I
I
I
E
=L m
P
0
P
-
-
0
N
N
3:
$9 c
29
0 ' E !
0
0
4E
E
1
1
Z88
1
E E E
2 222 1
8
l00
E
SU~FACTANTAFFINITY COEFFICIENTS
3:
$3 2
S v)
3:
9 cp
v)
S
compound
name
station^ phase
iurfactanl
additive % v/v
r oc
re$ wxem+wA
;7H8 :7H8 ;7H8 ;7H8 3H8
toluene toluene toluene toluene toluene
Spher~sorbC8 - 5pm Spherisorb C8 - Spm Spherisorb C8 - 5pm Spherisorb C8 - Spm Spherisorb C8 - 5pm
SDS SDS SDS SDS SDS
:7H8 :7H8 7H8
toluene toluene toluene
Spherisorb C8 - 5p.m Spherisorb C8 - 5pm Spherisorb C8 - Spm
SDS SDS SDS
:7H8 :?H8
toluene toluene
pBondapack C-I 8 (Waters) Hypersil S h ~ d o n- 5pm
:7H8
toluene
.7H8
25 25 25 25 25
310 I84 22 223 179
76 45 55 55 44
264 157 190 190 152
144 69 82 90 55
8 8 8 8
25 25 25
181 159 68
44 39 17
153 I35 57
57 52 15
8 8 8
SDS SDS
25 mb
200
2 3 ~
49 56
I70 196
70 1
18 24
CPS Hypersil Shandon 5pm
SDS
amb
170
42
144
23
24
toluene
SAS Hypersil Shandon 5pm
SDS
amb
59
204
19
24
.7H8
toluene
MOS Hypersil Shandon 5pm
SDS
mb
290
71
247
150
24
7H8
toluene
ODS Wypersil Shandon 5pm
SDS
mb
2
59
204
140
24
7H8 7H8 7H8 7H8 7H8
toluene toluene toluene toluene toluene
SAS Hypersil Shandon 5pm ODS Hypersil Shandon- 5pn Rainin Nicrosorb 5pm octyl Rainin Nicrosorb 5pm octyl Rainin Microsorb 3pm ODS
SDS SDS SDS SDS SDS
25 25 31 31 31 -
23 21 22 216 574
58 53 55 53 141
20 1 I84 191 184 489
20 170 210 160 650
25 25 I6 16
10% MeOH 3% n-PrOH 5% n-PrOH 10% n-PrOH 1% n-BuOH 5% n-BuOH 10% n-BuOH
3% ProW
8
f6
name
anisole anisole anisole anisole anisole anisole anisole anisole anisole anisole anisole anisole benzyi alcohol benzyl alcohol benzyl alcohol benzyl alcohol benzyl alcohol benzyl alcohol
s t a t i ~ phase n~
Brownlee RP-18 Spheri-10 Brownlee RP-18 Spheri-I0 Brownlee RP-18 Spheri-10 Brownlee RP- 18 Spheri-10 Brownlee RP- I8 Spheri-I0 Brownlee RP-18 Spheri-I0 Brownlee RP- 18 Spheri-10 Brownlee RP-18 Spheri-10 Brownlee RP- 18 Spheri-10 Brownlee RP- I8 Spheri-1 0 Rainin Microsorb C 18 3p.m Waters Radial-PAK C 18 1Opm Novapack C 18 4pm Gaskro Kogyo - Inertsil ODS C18 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5wm
:urfactanl SDS SDS SDS CTAE CTAB SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS Brij-35 Brij-35 Brij-35 Brij-35
CTAB GTAB
additive % vfv
r "C
W w s 0'
Whl
re$
KAS
ihosph. buf. 20 24 3% 2-PrOH 38 35 3% 2-PrOH 38 35 25 35 45 3% PrOH 25 3% PrOH 35 3% PrOH 45 10% PrOH 25 10% PrOH 35 10% PrOH 45 2% I-PrOH 25 15% EtOH 25 25 25 25 3% n-PrOH -
5 42 95 34 160 27 54 119 112 105 55 53 49 51 48 45 99 12 110 19 9.5
I1 10 24 12 39 6.3 13 29 27 26 13 13 12 12 12 If 24 296 222 19.4 9.5
49 40
17.3 14.2
37 35 85 33 135 22 45 101 95 89 47 44 41 42 40 38 83
2.6 2.5 23 62 50 57 20
36
3 23 I0 I1 6
I1 19 19 19 I9 I9 I9 I9 I9 19 19 6 I5 15 7
7,7 16.2 7.6
6,O 4,o 5.3 5.0
20
47.5 38.9
150.5 10.1
8
8
w
~ ~ ~COEFFICIE~T~ F A C AFFI~ITY TA~T
w
E E E
A A A
w
E
A
m
$ m
547
P-
F
" .
L
P-
E
" .
t:
N
N
N
C
hi
m
d
v E M W
3:
9 9
ii cf4 S M
S M
E E E
S . % %
v r m m
3:
2 S
m
I
I
/ EE
C
3:
3
Q $ ! m
5
#
N
BERTHOD and GARC~A-ALVAREZ-CO~~E
# # o o # # o o # w o o w o v M C U m m N N m m N h m m N r
-
m
P - ( v P - 0
P
0000 "
I
I
8
8
8
8 (VnJC\INC\IT\IC\I
O O O " , O O O O
8
l
cod (U-
E
rf-
I
m
55
N
c
E
A
" .
L-
c
L-
5 A
BERT~O and ~ ~A~C~A=ALVARE~=CO
3 0 0 0
2zzz
P
S
:E'5
0
m
Ec n
Q1
m m Ec n
Id
7
n
S:
G.
c)
c)
.-0
name
naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphthalene naphtha~ene naphthalene naphthalene naphthalene
stationary phase
Spherisorb C8 - 5pm Spherisorb C8 - 5pm Altex ODS pBondapack C-18 (Waters) Novapack C 18 4pm Bakerbond C 18 Novapack C 18 4pm Novapack C18 4pm Novapack C18 4pm Brownlee RP- 18 Spheri-10 Brownlee RP- 18 Spheri- 10 Brownlee RP- 18 Spheri-I0 Brownlee RP-18 Spheri-10 Brownlee RP- 18 Spheri-10 Brownlee RP- 18 Spheri- i 0 Brownlee RP-18 Spheri-I0 Brownlee RP-I 8 Spheri-I0 Brownlee RP- 18 Spheri- 10 Brownlee RP- 18 Spheri-10 Novapak C18 - 4pm Nucleosil C 18 5pm -Novapack C 18 4pm
jurfactanf SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS SDS
additive % vlv
roc
$%n-BuOH
25
10% BuOH 5% BuOH NaCl 0,lM
3% PrOH 3% PrOH 3% PrOH 10% PrOH 10% PrOH 10% PrOH 3% 2-PrOH 5% MeOH
'AM
Or
I(L
(Llmol)
5% n-BuOH 25
10% n-BuOH 3% 2-PrOH
PWM
25 38 25 25 40 25 25 25 35 25 35 45 25 35 45 25 35 45 25 40 25
~
KAM
Or
395 99 87
97 24 21
tmLk> 336 83 74
238 956 1216 235 245
58 23 5 299 58 60 70 I18 50 353 362 248 51 51 47 42 41 40 246 25 79
202 816 1038 200 208 243 410 174 1225 1257 86 1 178 177 163 145 142 139 854 89 275
286 481 204 1436 1473 1009 210 208 192 171
324
re$
K2
142 48 20
8
909 320 833
I1 18 7 4 9 9 9 19 19 19 19 19 19 19 19 19 19 30 5
70 130 230 100
730 138
8 8
2
W 0 m M N N
2 0
2
.-.. PP-
0
m W
moo M M
0 0 0
0 0
I
l
l
w w w
u u o
" M
0 0
1
E E E
2 2 %
l
l
m m m I
000000
v u u
E E E
I
l
l
A l l m m m 0 0 0 0 Q 3
0 0 0
I
v?
“5
VI
N
W
m -
N
2
vslnvs N N N
vsvsvs
I3131 0 0 0
f 22%
vs
i5 31
0
W
C
0Ov
63 1
I
S s s
‘i; $22”
" "
00000000
I
I
I
I
elnlnln
x
2
Q M
$
l
Ef
8 8 8
118
8
82
-
ZL 988 60 I
ZZEI 61 I Z9 I PLZZ 58
I8€ EP 65 828 00 I LL
6PSC OZ C P9 C LLtZ
$6 E CE
96 I 6PP
00 I SOZ
Lt
zz
8t ZLZ PI E 6ZP LLP 8SPI
OOE I
OLP
s1 I 62 1 LP PP
009 8PP €9 I €S I
62 62 62 62
LEE
€8
L8Z
62
992
62
62 62 62 62 62 OE 8 8 8
92s 26 L 08 C
69s
I6 PZ 1. 8€1 OZP
OZE
8L
E6
€2
90s 09s 80L G
SZ
HOrd-u %f: HO"8-U %O I
SZ HOn8-u Yo!
SZ
SZ
C6 OEZ
ZZ 9s
-
O1HZl.Z OlHZl-3 01HZ13 OlHZl-=I
WL3 WL3
OlHZtS
sas
WL3 SE-C!38
-
HOrd-Z SZ %OZ HOJd-2 Y O S1 SZ Hold-Z SZ %O I SZ HOrd-Z %S SZ HOW-I %S SZ HOW-I %i H0*8- I SZ %O I SZ HOaM %OZ SZ HOaliV %S I SZ HOWJ %O I SZ HOW4 %S SZ
SZ
SZ
LZS
621
cL -
far -
HOn8-u Yo0I i 0 w - u %S i 0 n a - u Yo€
8HZ 13 8HZ 13 8HZ 13
sas sas sas
8HZ 13 8HZ 13 8HZ 13 8HZ 13 8HZ 13 8HZ 13
sas sus sas sas sas sas
8HZ13 8HZ13 8HZ 13 8HZ 13
sas sas sas sas
8HZ13
sas
8HZ 13
sas
Ieicspns
punodUIo~
1
1
1
NT ~FFINITYCOEFFlCl
1
u u u u
w w w w
,
1
1
1
"d
0000Do
0 0 0
I
I
I
I
I
0000000000
""00"
I
I
I
l
l
1
00000000
o o u u
I
W
W
W
W
E W
5
S
W
S
G 5 -
e4
e
U U U U
2222
~ ~ ~COEFFICIE~T~ F A C AFF~~ITY TA~T
W
ij 83
e
5
S
W W
e
I :
71
om pound
station^ phase
name
additive Yo VIV
14H10 14H10 14H10
phenanthrene phenanthrene phenanthrene
Novapack C 18 4pm Nova~ackC 18 4pm Novapack C 18 4pm
CTAB CTAB CTAB
14HI 0 14HlO 14H10 14H10 14H10 14H10
phenanthrene phenanthrene phenanthrene phen~threne phenanthrene phenanthrene
nucleosd C 18 Spin Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm Spherisorb C8 - 5pm
SDS SDS SDS SDS SDS SDS
14H10 14H10 14H10 14H10 14H10 14Hl~ 14H10
phenanthrene phenanshrene phenanthrene phenan~hrene phen anthreiie phenan~hr~ne phenanthrene
Bakerbond C 18 Novapak C 18 - 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C18 4pm Novapack C 18 4pm
SDS SDS SDS SDS SDS SDS SDS
14H10 14HlO 14H10 14H10
phenan threne phenanthrene phenanthrene phenanthrene
Novapack C 18 4pm Novapack C 18 4pm ovap pack C 1 8 4pm Novapack C 18 4pm
SDS SDS SDS SDS
I
r "C
re$
(Llmol)
5% 1-BuOH 40
7% I-BuOH 10% 1-BuOH 3% 2-PrOH 3% n-PrOH 5% n-PrOH 3% n-BuOH 5% n-BuOH 10% n-BuOH
WM
1193
434 407 326
1190 1116 894
188 258 658 708 307 91
46 63 162 174 75 22
161 220 56 1 604 26 1 77
300 3700 488 3 79 246 222 31
1041 12845 1491 13I3 855 77 1 106
62 63 356 262
214 217 1236 909
40 40 40 25 25 25 25 25
40 25 5% MeOH 25 10% MeOH 25 15% MeOH 25 20% MeOH 25 25 10% 1-BuOH 7% I-BuOH 25 5% 1-BuOH 25 5% 2-PrOH 25 10% 25 2-PrOH __
1220 15~42 11983 1540 1003 905 125
256
7 066
29 29 29
306 204 429 323 147. 24
16000
5
8 8 8 8 8 4 30 29 29 29 29 29
29 29 29 29
b - m
E E E E E
2
d-
s.2.2.2.2 .
d-d-d-d
U
2 a
>
s
$ g &
28-03
r:-i
name
station^ phase
;urfactant
9-methylanthracene
Novapack C 18 4pm
SDS
9-meth ylanthracene
Novapack C 18 4pm
SDS
9-methylanthracene
Novapack C 18 4pm
SDS
9-methyl-anthracene 9-methyl-anthracen~ dansyl-glycine sodium octylbenzenesulfonate
Novapak C 18 - 4pm Novapak C 18 - 4pm Nucleosii C 18 5pm Hypersil - 5pm
Brij-35 SDS SDS SDS
14H21Na03S
sodium octyibenzenesul fonate
sodium octylbenzenesul fonate 14H21Na03S 14H21Na03S sodium octylbenzenesulfonate
sodium octylbenzenesulfonate sodium octylbenzenesulfonate
additive Yo v/v
r oc
10% 2-PrOH 15% 2-PrOH 20% 2-PrOH
25
1448
356
1234
29
25
929
22 8
792
29
25
767
188
654
29
123 36993 38
160
130 9100 9.2 39
31591 32 136
275 50000 11 0.2
30 30 5 24
W W S
PWM
0'
re$
KAS
25 25 __ 3% 2-PrOH 40 3mb
110
CPS Hypersil
- 5pm
SDS
unb
110
27
93
I0
24
SAS Hypersil
- 5pm
SDS
unb
160
39
136
9.4
24
MOS Hypersil
- 5pm
SDS
imb
300
74
255
35
24
ODS HypersiI
- 5pm
SDS
imb
220
54
187
33
24
SAS Hypersil - 5pm ODS Hypersil - 5pm SAS Hypersil - 5pm SAS Hypersil - 5pm ODS Hypersil - 5pm ODS Hypersil - 5pm
SDS SDS SDS SDS SDS SDS
155 221 651
38 54 160 34 100 38
132 187 555 118 347 -132
11 36 83
25 25 25 25 25 25
25 25 NaCl -0.IN 25 5% MeOH 25 NaCl -0.1M 25 5% MeOH 25
139 408 155
8,O 48 22
SURFACTANT AFFINITY COEFFICIENTS
l
1
1
G G G
575
8 a
A
>
6)
3 S
G
E
% tf
2 U
%
8
A
z"
l
0 0 0 0 0 1
57
E d
E
€ w
t
rl
d
rl
w
I
2 d t
oo
3
-$
_t .
d
oo
-
benzo(a)anthracene benzo(a)anthracene benzo(a)anthracene benzo(a)an thracene benzo(a)anthracene benzo(a)anthracene benzo(a)anthracene benzo(a)anthracene
INovapack C 18 4pm Hypersil C 18 5pm Novapack CIS 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm
Brij-35 Brij-35 CTAB CTAB CTAB CTAB CTAB CTAB
benzo(a)anthracene benzo( a)anthracene benzo(a)anthracene
Novapack CIS 4pm Novapack C I8 4pm Novapack C 18 4pm
CTAB CTAB CTAB
)enzo(a)anthracene ~enzo(a)anthracene ~enzo(a)anthracene 3enzo(a)anthracene )enzo(a)a~thracene >enzo(a)anthracene
Novapack C I8 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C 18 4pm Novapack C I8 4pm
SDS SDS SDS SDS SDS SDS
Novapack CIS 4pm xnzo(a)anthracene )enzo(a ) ~ ~ r a c e n e Novapack C 18 4pm )enzo(a)~thracene /Novapack C 18 4ym
SDS SDS SDS
40 40 40 40 40 40
198 311 28847 12364 7144 12364 9616 5770
2 10 330 10500 4500 2600 4500 3500 2100
178 280 288 17 12350 7 136 12350 9606 5763
40 40 40
4946 4~97 3160
1800 1600 1150
4940 4391 3 156
25 25 25 25 25 25
73578 9313 9053 7$05 7568 227
18100 229 I 2227 1920 1862 56
62835 7946 7724 6659 6457 193
25 25
703 917 6515
173 225 1602
599 781 5558
25
5% MeOH 10% MeOH 15% MeOH 3% 2-PrOH 5% 2-PrOH 10% 2-PrOH 5% I-BuOH 7% 1-BuOH 10% I -BuOH 5% MeOH 10% MeOH 15% IvleOH 20% MeOH 10% 1-BuOW 7% 1-BuOH 5% 1-BuOH 5% 2-PrOH
25
430 322
30 28 29 29 29 29 29 29 29 29 29
127000
30 29 29 29 29 29 29 29
29
name
station^ phase
;urfactant
9-methylanthracene
Novapack C 18 4pm
SDS
9-meth ylanthracene
Novapack C 18 4pm
SDS
9-methylanthracene
Novapack C 18 4pm
SDS
9-methyl-anthracene 9-methyl-anthracen~ dansyl-glycine sodium octylbenzenesulfonate
Novapak C 18 - 4pm Novapak C 18 - 4pm Nucleosii C 18 5pm Hypersil - 5pm
Brij-35 SDS SDS SDS
14H21Na03S
sodium octyibenzenesul fonate
sodium octylbenzenesul fonate 14H21Na03S 14H21Na03S sodium octylbenzenesulfonate
sodium octylbenzenesulfonate sodium octylbenzenesulfonate
additive Yo v/v
r oc
10% 2-PrOH 15% 2-PrOH 20% 2-PrOH
25
1448
356
1234
29
25
929
22 8
792
29
25
767
188
654
29
123 36993 38
160
130 9100 9.2 39
31591 32 136
275 50000 11 0.2
30 30 5 24
W W S
PWM
0'
re$
KAS
25 25 __ 3% 2-PrOH 40 3mb
110
CPS Hypersil
- 5pm
SDS
unb
110
27
93
I0
24
SAS Hypersil
- 5pm
SDS
unb
160
39
136
9.4
24
MOS Hypersil
- 5pm
SDS
imb
300
74
255
35
24
ODS HypersiI
- 5pm
SDS
imb
220
54
187
33
24
SAS Hypersil - 5pm ODS Hypersil - 5pm SAS Hypersil - 5pm SAS Hypersil - 5pm ODS Hypersil - 5pm ODS Hypersil - 5pm
SDS SDS SDS SDS SDS SDS
155 221 651
38 54 160 34 100 38
132 187 555 118 347 -132
11 36 83
25 25 25 25 25 25
25 25 NaCl -0.IN 25 5% MeOH 25 NaCl -0.1M 25 5% MeOH 25
139 408 155
8,O 48 22
SURFACTANT AFFINITY COEFFICIENTS
l
1
1
G G G
575
8 a
A
>
6)
3 S
G
E
% tf
2 U
%
8
A
z"
l
0 0 0 0 0 1
FACTANT AFFINITY COEFFICIENTS
e,
f;
E
9 , n
a
W
E
D
0 *
0
3
d
3=
v s v s v s v s v s v s N N N N N N
2
8 . (
S m
W
~
v s v s v s v s N N N N
" " 4
w w w w
W
W
v v v v v v v v v v
W
W
W
d
"+,.-id
v
v
W
vs N
W
vs N
v
I
I
a,
2x
P,
n W
W E
13
U " " "
2222
a,
a,
E
N
a
n
a
v
K R W
A
a,
W
13
1
I
3:3:
00000000
G G G G
U
r
r
c
" -
lgl""
7
n r
o c
-
N ' ; f
N
Surfactants : SDS= sodium dodecylsulfate; CTAB = cetyl t r i m e t h y l ~ o n i u mbromide; E3rij@ 35 = polyoxyethylene 23 dodecyl ether or C12E23; Brij830 = polyoxy ethylene 4 dodecyl ether or C12E4; DOSS = dioctylsulfosuccinate de sodium or Aerosol OT; MB-14 = myristyl (C 14) betaine; SB-I 2 = dodecyl dimethyl (3-sulfopropyl) ~ ~ o n hydroxyde; ~ u m THPA = tetraheptyl bromide (a nonmicelle forming surfactant); Tween@ 60 = polyoxyethylene 20 sorbitan monostearate. For the numbered dihydropyridine esters, see Ref 3 1.
8
Lu S
e
591
1.
F.G.P. Mullins and G.F. Kirkbright, TheAnalyst, 109: 12 17(1983).
2.
S. Terabe, H, Tanaka, K. Otsuka and T. Ando, J. chroma to^. Sci., 27: 653 (1989).
3.
E. Bonnet-Domingo, M.J. Medina-Hernandez, G. Ramis-Ramos t , 843 (1992). and M C . Garcia-Alvarez-Coque, The A ~ f f l y s117: A.S. Kord, J.K. Strasters and M.G. Khaledi, Anal. Chim.Acta, 246: 131 (1991).
4.
5.
L.S. Masamba-Tan, J.K. Strasters and M.C. Khaledi, J. Chromatogr.A, 683: 32 1 (1994).
6.
B.K. Lavine, A.J. White and J.H. Han, J. Chromatogr., 542: 29 (1991).
7.
28: M.L. Marina, S. Vera and A.R. Rodriguez, Chromatogra~hiff, 379 (1989).
8.
M.A. Garcia and M.L. Marina, J. Liq. Chromatogr. &: Relat. Technol., 19: 1757 (1996).
9.
M.A. Garcia, S. Vera and M.L.Marina, Chromatographia,32: 148 (1991).
10.
E. Pramauro and E. Pelizzetti, Anal. Chim. Acta, 154: 153 (1983).
11.
M.G. Khaledi and E.D. Breyer, Anal. Chem., 61: 1040 (1989).
12.
P. Yarmchuck, R. Weinberger, R.F. Hirsch and L.J. Cline Love, Anal. Chem., 54: 2233 (1 982).
13.
D.W. Armstrong and F. Nome, Anal. Chem., 53: 1662 (1981).
14.
X. Li and J.S. Fritz, J. Chromatogr.A, 728: 235 (1996).
15.
M.F. Borgerding and W.L. Hinze, Anal. Chem., 57: 2183 (1985).
16.
M.G. Khaledi, E.Peuler and J. Ngeh-Ng~ainbi,Anal. Chem., 59: 273 8 (1987). S A . Zibas and L.J. Cline Love, Anal. Chim,Acta, 299: 17 (1994). M. Arunyana~and L.J. Cline Love, Anal. Chem., 56: 1557 (1984). F.P. Tomasella, J. Fett and L.J. Cline Love, Anal. Chem., 63: 474 (1991).
17. 18. 19.
20.
7’.
21
F. Garcia-Sanchez, A.Navas-Diazand C ~ r o m a t o ~A,r .723: 227 (1996).
*
22. 23 24
*
m
25. 26. 27.
Okada, J. Chromatogr., 607: 135 (1992).
A. Garcia Pareja, J.
G M . Janini and S.A. Atari, Anal. Chem., 55: roch, Ph.13. 13issertatio~( ~ a l e n ~ iSpain) a s (l 999).
A. Berthod, I. Girard and C. Gonnet, ana^. Chem. 58:1359 (1986). A. Berthod, I. Girard and C. Gonnet, Anal. Chem. 58: 1362 (1986). B.K. Lavine, S. Hendayana and J. Tetreault, A~~Z.Chem. 66: 345 (1994) .Sanchez-Mallols, R.M. ~ i l i a n u e v a - ~ a ~ a and n a s C. Rarnos, Chromato~~a~hia, 38: 365 (1994).
28.
D. Lopez-Lopez, S. Rubio-Barrosoand Chromatogr., 18: 2397 (1995).
29.
M A ,Rodriguez-Delgado,M.J. Sanchez, V.Gonzalezand F. Garcia
L. Polo-Diez, J. Liy.
Monteiongo, Chromatographia,38: 342 ( l 994).
30.
V. Gonzalez, M.A. Rodriguez~Delgado, M.J. Sanchez andF. Garcia Monte~ol~go, ~ h r o ~ ~ t o g r a p h34: i a 627 , (1992).
3 1,
687: 1 (1994). J.M. Saz and M.L. Marina, J. Chromato~r. A,
icellar solutions simply are binary solutions of surfactant in water. wever, very often a ternary component -an organic modifi physicoche~~stry of a ternary system is commonly studi dia~ram. Numerous diagrams are pu r, it islikely that the particular phase dia ill not befound. It is notdif~cultto prepare a ternary phase diagram in less than a day’s work. However, the procedure is rarely taught to students in analytical chemistry. Therefore it is presented here. Any compositioll of a system with three components esented in two dimensions if a relation is made b oportions. The relation is:
can be volume%,mole % or, the easiest to measure, mass shows how any givencomposition can be represented in an e ~ u i l a t e rtriangle. ~l To prepare the ternary dia am in mass, four mixtures o will be considered. They are 8 0 ~ 0 + 20% C,60% B + 40% C,
60% C and 20%
+ 80 % C.
In each mixture, small amounts of A are
added to the I3 + C mixture. Let us consider an actual example: we were unable to find in the literature the water/heptane/sodium dioctylsulfosuccinate (AOT) ternary diagram. To obtain the phase diagram presented by Figure 2.16, we prepared a mixture of 80% heptane and 20% AOT. Actually 0.8 g (1.17 mL)of heptane was introduced in a 30-mL test tube closed by a rubber cap to avoid evaporation. 0.2 g of AOT (a wax) was added and fully dissolved in heptane. This composition corresponds to 80% heptane + 20% AOT near the lower right corner of Figure 2.16. Next, the test tube and its holder are put on a balance that is zeroed. The cap is opened and a drop of water is added. The balance indicated 0.064 g. Shaking the tube made the water drop disappear and forma clear solution. The composit~onis calculated as followed: mass of heptane = 0.8 g, mass of AOT = 0.2 g, mass of water = 0.064 g, total mass = 1.064 g. The mass percentages are :heptane = 75.19%, AOT = 18.80% and water = 6%. This composition is located on a line starting from the 80-20 heptane-AOT initial composition and going to the water apex at the lower left of the triangle. By visual observation, the homogeneous water solution is referred to as a L2 system (Chapter 2). A second drop of water was added to the mixture, the balance read 0.108 g. The heptane and AOT masses were unchanged. The total mass was 1.108 g, the water mass was 0.108 g or 9.75% wlw. After shaking, the second drop solubilized in the heptane phase. We were still in the L2 area. More drops of water were added andthe test tube was weighted andshaken in a similar way. It is observed that the water dissolution takes more and more time and shaking energy. When the balance read 0.592 g, a cloudy heptane mixture was obtained. After more shaking, the cloudy solution remained. The limit of the L2 zone were was reached. The composition was: total mass = l .592 g, heptane = 50.25%; AOT = 12.56% and water = 37.19%. Continuing the water addition in the test tube, a viscous cloudy solution was obtained for a water mass of 0.728 g. When placing the test tube between a polarizer and crossed analyzer glasses, some lightcould be seen. The viscous cloudy liquid was a heterogeneous system containing a liquid crystal suspended in an L2 solution (dotted area in Figure 2.16). The viscous mixture became fluid again for a water mass of 1SO6 g (water = 60% w/w). Letting the test tube stand for 2 min showeda phase separation. Water was further added up to fill the test tube. The final water mass was 25 g giving a biphasic system containing 96.15% of water, 3.08% of heptane and 0.77% of AOT.
The same procedure was followed starting with a test tube containing 0.6 g (0.88 mL) of heptane and 0.4 g of AOT (line 60-40% w/w). Water was added dropwise and Table I V J lists the observed phenomena. Once the 40-60 and 20-80lines were diluted the same way, the respective L2, L 1 and liquidcrystal areas became almost drafied. The same procedure was followed starting with water + AOT mixtures diluted by heptane. Then, special mixtures in the water rich corner were prepared to exactly delineate the L1 and liquid crystal areas.
Observations ~ i I u t ~ the n g 60%-40% Heptane-~OTMixture with Water. Water Total Water mass g mass g % 0 0.050 0.100 0.150 0.200 0.300 0.550 0.900 1.223 1.634 2.337 3.359 4.276 18.00
1.000 1.OS0 1.100 1.150 1.200 1.300 1.550 1.900 2.223 2.634 3.337 4.359 5.276 19.00
0 4.76 9.09 13.04 16.67 23.08 35.48 47.37 55.0 1 62.03 70.03 77.06 8 1.OS 94.74
Heptane AOT % YO 60 57. I4 54.55 52.17 50.00 46.15 38.71 3 1.S8 26.99 22.78 17.98 13.76 11.37 3.16
40 38.10 36.36 34.78 33.33 30.77 25.81 21.05 18.00 15.19 1 1.99 9.1 8 7.58 2.10
Observation initial 1g mixture clear and fluid L2 phase L2 phase L2 phase L2 phase L2 phase L2 phase L2 phase
mixture of L2 phase and liquid crystal
true liquid crystal cloudy biphasic system clear and fluid L1 phase cloudy fluid phase lightly cloudy fluid phase
When more than three components are needed, two dimensional representation is still possible if two relationsh~psare imposedon the system. For example, Figure 2.18 shows phase diagrams of water, oil, SDS and alcohol. The constant molar ratio 2 SDS molecules for 13 alcohol molecules was imposed onto the system.
197 Acetophenone 20,65-66 Adsorptioll 99, 105 adsorbed layer effect of organic 10l -103 additives 99- 100 effect of salts 90-98 isotherms 13,47 1-472 Aerosol OT Affinity of solutes for micellar phases 162 525-576 table of value Aggreg~tionnumber 29 Alcohols, see organic mo~ifiers effect on micelles 192- 194 and efficiency l 87-19 1 and surfactant deso~tion 194 Al~ylbenzene 299-300 in microemulsions 469-47 1 A~kylsulfates 11-13 ~lkylsulfonates 11-13 Al~yltins 1-452 45 Amino acids 154,22l, 222,270 deri~ationby OPA 357 determination 356 hydrophobici~ 3 18-3 19 Amperomet~ 453-454
Amphi~hiles 9 Amphoteric surfactant 19 Anabolic steroids 440-44 1 Anesthetic action (QSRR) 336 Anionic surfactants 11 210-21 1 versus cationic Anthracene 197 Antibinding compounds 120-122,214-216 sait effect on 161-164 ~ntidepressants 423 A r ~ s t r o n g - ~ o mequation e 69 123,207 Arsenides 452-453 Background signal 399-403 Band broadening 176-178 189,366-369 P-Blockers in sports 418-419 Bile salts 475-476 Binding compounds 120-122 salt effect on 164 selectivity 214-216 Binding constants 140-142 525-576 table of value Bioactivity 333-334 Biomembranes 334-335 ?
~reakthroughmethod 88-89 Brij@
16
Capillary electrophoresis 488-495 Carbon number in homologous series 296-298 Cation separation 482-484 Cationic surfactants 14 versus anionic 2 10-212 Cavity formation 304-305 Cholic acid 476 Chromato~aphicprocess 179 Cloud point 36-37 cmc 3 1-44 measure of 4 1-42 in modeling 275-276 and partitioning 144-145 table of values 503-522 Column C18 160 159care 108-109,344-345 cyano 157-158 packings 78-93 preparation 107 switching 420-424 Concentration of surfactant and gradient 106,245,457 and retention 69, 123,207 2 14,215,247-250 and selectivity 2 14-2l 5 Conductan~e 458 Congenerity 310-31 l Contour maps 283-284,364 Control of pharmaceutical preparations 352-3 78 25 1 Conventional WLC Critical micelle concentration see cmc 1-520 50
Cyano columns Cyclic voltametry
157-158 456
Dead time 272-275 Dead volume measurement 273 Derivatization 354-3 57,4 14-4 17 diazotation 355 orthophthalaldehyde 357 sulfonamides 415 Desorption of s~rfactantby alcohols 194 Detection 429-460 Diffusion coefficients 180-182 Direct injection 388-398 advantages 3 89 drawbacks 390 mobile phase needed 396 surfactant to use 394-395 technique 39 1-393
15 1-152 216-217 Displacement of bound drugs 396-397 Diuretics 287-288 optim~zationof direct analysis 409-4 12 separation 359-363 site of action in kidney 338 in sports 417-418 retention 337 Drugs displacement 396-397 overlapping peaks 398-399 Direct solute transfer
-202173 Efficiency enhancement by alcohols 187-194
99
INDEX
by reduced flow rate 198 by temperature 196- 197 in microemulsions 47 1-472 El~trochemicaldetection 453-460 and gradient 457-458 Electrolytes and cmc 37 Electroosmotic flow 490 Electroosmotic mobility 490 Elution strength 132-136,203-240 effect of alcohols 228-23 1 measurement of 223-227 and selectivity 222-228 Empirical models 264-265 Enhanced detection 429-460 Enzyme activity 350-352 Exchange of solutes micelle-bulk 180-181 stationary phase-micelle 183 Extra column variances 179 Extraction of drugs 419-420 Fatty acids 11 Field-flo~-~actionation486-488 Fluorescence 434-444 Foley-Dorsey equation 176 Gel permeation chromatograp~y 66-67,477-478 Glassy carbon electrode 453-457 Global retention model 267-269 Gradient capabilities 106 and electrochemical detection 457-460 elution 245-246 Height equivalent to a theoretical plate
176
Historical development 57-78 Homologous series 296-308 Hybrid micellar mobile phases 132-137,231-236 retention behaviorin25 1-263 Hydrophobicity 293-342 amino acids 318-319 congenerity 310-311 the MLC approach 328-329 and P,, coefficients 320-323 and QSRR 309-3 10
Illegal drugs 4 17-419,467-468 Inductively coupledp l ~ m ~ 9 - 4 5 3 Intercept (equation) 146-147 Interfacial tension 106 Ion exchange 483-485 Ion pair chromatography 57-64 Ion separation by GPC 478 inorganic 480-485 Ionic strength 161-164 Isotherm of adsorption 90-96 anionic surfactant 90-9 1 cationic surfactant 94 nonionic surfactant 96 sub-micellar phases 92,95 Iterative regression strategy 25 1-255 with pH and organic modifier 266-267 Jablonski diagram
442
Kinetics of mass transfer 175 Knox equation 176 efficiency studies 183- 186 Krafft point 33-35
Linear solvation energy relationship 1-33333 Lyotropic ionic series
38
Meaning of empirical models 263-265 Measurement of cmc 4 1-42 elution strength 223-227 surfactant adsorption 85-90 Metal complexes, fluorescence 43 8-439 Methods column care 108-109 column preparation 106 Methylene selectivity 298-300 Micellar clean-up 420-424 Micellar electrokinetic 488-495 chromatography 2 1-44 Micellar phases for drug analysis 399-403 gradient of concentration 106,245,457-460 with two surfact~ts 376-3 77 Micelles conductance 458 dynamic of 26-27 effect of alcohols 134-136 formation 2 1-25 shape 29 size 28 solubilization site 43-44 MICHROM software 243,289-290,501-502 user’s manual or, the CD-ROM Microelectrodes 457 Microemulsions 47-49,378 alkylbenzenes 469-47 l
drugs in sports 468-469 oil in water 466-467 water in oil 47 1-472 Microenvironment polarity 303 Modeling of retention 24 1-291 Molecular interactions 20 ulticomponent analysis 42 1
Neural networks 27 1 Nonbinding compounds 120- 122 Nonionic surfactants 14-18 Nonmicellar phases 137- 139 Octanol-water partition coefficients 293-294 congenerity 310-311 log k vs. log P , , plots 325-326 and micellar constants 312-318 309 and QSRR Optimization strategy 24 1-243 for drug analysis 403-4 10 359 interpretive 276-288 of resolution 358 sequential Organic modifiers effect on cmc 39-40 effect on efficiency188- 192 effect on selectivity 2 19-222 eRect on surfiactant adsorption 101-103, 193-194 elution strength 132-136, 228-23 1 high concentration 13 7-139 si~ultaneouseffect with
on
PH 266-270 table of cmcs 503-525 ~rthophthalaldehydederivatizat 357 ~ v e r ~ a ~ fractions ped 278-279
Packings, see Silica PA effect of alcohols 230 enthalpy of transfer 166 fluorescence 43 7 QS 329-330 selectivity 220,238 Partition coefficients 44-45, 3 26- 127 evaluation 139-140 measurement 141-142 Partitioning 115-1’72 effect of alcohols 135 effect of cmc 144145 e~uilibrium 135 model of retention 123- 130 Peak shape 280-28 1 Peptides 154 Pesticides 379
PH
effect on retention 152-155, 246-250 in micellar media 46-47 simultaneo~seffect with organic modifier 266-270 P ~ a ~ a c e u t i cpreparations al 352-378 Phase diagram 49-52,466 preparation of 593-595 Phosp~orescence 444-449 Physiolo~icalfluids 387-425 fluorescence detection 43 8
Pluronic~ 17 Polarity, rnicroenvironrnent ~ 0 3 Polyaromatic hydrocarbons, see FAH Polynomial modified ~ a u s s i a n f~nction 280 Pore size Pore volume Precolurnn derjvatjzation 354-357,414-415 Prediction of factors affecting 27 1-273 peak shape retention QSRX and PAHs selectivity Pressure effect on cmc Protein purification 345-349 Pseudo-phase 28-29
~uantitativeretention-act~vity relationship 333-339 bioactivi~ 333-334 ~uantitativestructure~retention relationsh~p 308-333 ionic compounds 326-32
Rare earth cations Recombinant growth hormone Reduced efficiency remediation Resolution optimization of contour283-284 map Retention behavior
276-2
244-245
444-449
203-240
equations 123-130 147-148 errors in 80-82 global model 267-269 homologous series 296-308 prediction of 243,261-262 strategy of optimization 277-279 Retention factors and carbon number 296-298 micellar concentration 123, 20’7,214,247-250 octanol-water coefficients 3 12-318,321-326 450-453 Reversal of elution order 205-207 Room temperature phosphorescence 145-146
1 particle 80-8 size porosity shape 81 surface area 80-82 Slope of retention ~quations 146- 147 Soaps I2 Solubility limit theory 15 1-152 and ~ydrophobici~ 3 06-3 08 andselectivity216-219 Solubilizatio~ 45,294-295 Solventeffectoncmc39-41 Speciation Sports, illegal drugs in 4 17-419,467-468 Stationary phase 79-1 13 volume Steroids 287 Salts and cmc 87-33 Strategy of optimization 277-279 Sample preparation 53,4 3 19-424 interpretive 359 Selectivity sequential 358 and affinity constants method Stripping 90 207-208 Su~fonamides 354-355 1 alkylbenzene 299-30 in milk and urine 4 12-4 13 and carbon number 297-299 13 Sulfosuccinates effect of s t a t i o n a ~phase screen agents Sun 372 Supercritical fluid 473-475 103-105 effect of surfactant 208-21 1 Surface silanols 84 ~ n c t i o l ~groups al 3 02-3 03 Surfactant and elution strength 222-228 adsorbed 97-99 layer and organic modifiers adsorption20,65-66,90-96 11 classification 2 19-222 predictio~of 242-243 501-521 1, 3 crnc and solubility limit 2 16-2 19107-108 desorption molecule Shift of dissociation 10 constants 153-154 84 Silanol groups Temperature effect on Silica 78-83 33 cmc 80-82, 84 bonding
603
196-197 efficiency selectivity 237-23 8 Terbium, fluorescence 440-443 Testosterone 44 1 Thallium 444,448 174Thermodynamics Thin layer chromatography 68,478-479 S 18- l20 Three phasemodel t h e ~ o d y n ~ iof c s164-167 Tobacco 380-381 Toxicity of phenols by QSRR 336-337 Transfer direct 151-152
enthalpy Transition metals Triangular subspace Triethylamine
166- 167 484 253-254 199
Valley-to-peak ratio 278 Van? Hoff plots 164165 Variable253-256 space Volume of stationary phase 145- 146
Zwitterionic compounds retention 249-2
50
