Preparation and evaluation of fused silica capillary columns for gas chromatography and micellar electrokinetic chromatography

Citation for published version (APA): Wu, Q. (1992). Preparation and evaluation of fused silica capillary columns for gas chromatography and micellar electrokinetic chromatography. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR371918

DOI: 10.6100/IR371918

Document status and date: Published: 01/01/1992

Document Version: Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.tue.nl/taverne

Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.

Download date: 10. Oct. 2021 PREPARATION AND EVALUATION OF FUSED SILICA CAPILLARY COLUMNS FOR GAS CHROMATOGRAPHY AND MICELLAR ELECTROKINETIC CHROMATOGRAPHY

QUANJIWU

. " .· ~- :" . ." PREPARATION AND EVALUATION OF FUSED SILICA

CAPILLARY COLUMNS FOR GAS CHROMATOGRAPHY

AND MICELLAR ELECTROKINETIC CHROMATOGRAPHY PREPARATION AND EVALUATION OF FUSED SILICA

CAPILLARY COLUMNS FOR GAS CBROMATOGRAPHY

AND l\fiCELLAR ELECTROKINETIC CBROMATOGRAPHY

PROEFSClll:FT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, Prof,dr. J.H. van Lint, voor een commissie aangewezen door het College van Dekanen in het openbaar te verdedigen op dinsdag 14 april 1992 om 16.00 uur

door

QtrANJI wt1

Geboren te Liuzhou (P.R. China)

Druk: Boek en Offsetdrukkerij Le1ru, Helmond, 04920·37797 Dit proefschrift is goedgekeurd door de promotoren: prof.dr.ir. C.A. cramers en prof.dr. P.J.F. Sandra

copromotor: dr.ir. J.A. Rijks To Rongrong and Fanda To our parents CONTENTS 1

CHAPTER 1 4 GENERAL INTROOUCTION 4

1.1 Introduction 4 1.2 Scope of This Thesis 6 1.3 References 9

CHAPTER 2 • 10 DEACTIVATION OF FUSED SILICA CAPILLARY COLUMNS BY POLY-HYOROSILOXANES WITH DIFFERENT FUNCTIONAL SUBSTITUENTS • • • . • • • • • • • 10

2.1 Introduction 10 2.2 Theory 14 2.3 Experimental 17 2.4 Results and Discussion 21 2.5 Conclusions • • 34 2.6 References 37

CHAPTER 3 • 39 A SOLID STATE 29si NMR STUDY OF THE DEACTIVATION OF SILICA SURFACE USING NON-POROUS SILICA PARTICLES AS MODEL SUBSTRATE • • • • • • • • • • • 39

3.1 Introduction ...... 39 3.2 Theory . . . . 42 3.3 Experimental ...... 46 3.4 Results and Discussion . . . . 50 3.5 conclusions . . . . 57 3.6 References . . . . 58

CHAPTER 4 • 60 PREPARATION OF NON-POLAR AND MEDIUM POLAR FUSED SILICA CAPILLARY COLUMNS FOR GC • • • • • • • • • 60

4.l Introduction 60 4.2 Experimental 68 4.3 Results and Discussion 72

-1- 4.3.l Evaluation of Deactivated Capillaries 72 4.3.2 Evaluation of Columns coated with Non-polar Stationary Phases . • • 73 4.3.3 Evaluation of Columns Coated with Medium Polar Stationary Phases 79 4.4 conclusions 82 4.5 References 84

CHAPTER 5 • • • • 90 APPLICATIONS IN GAS CHROMATOGRAPHY 90

5.l Introduction 90 5.2 Experimental 93 5.3 Results and Discussions • 95 5.3.l High Speed Analysis on Narrow-Bore Capillaries • • • • • • • • • • • • • 95 5.3.2 High Temperature Columns •••••• 99 5.3.3 Analysis of Drugs, Pollutants, Amino Acids, and Underivatized Steroids 105 5.4 Conclusions • 110 5.5 References 111

CHAPTER 6 • 114 MICELLAR ELECTROKINETIC CAPILLARY CHROMATOGRAPHY ---THE INFLUENCE OF COLUMN MODIFICATION ON ELECTROOSMOTIC FLOW • • • • • • • • . . • . • . • 114

6.1 Introduction 114 6.2 Theory 119 6.2.l Electrical Double Layer 119 6.2.2 t-Potential and Electroosmotic Flow (EOF) • • • • • • • • 120 6.2.3 Factors Influencing EOF l24 6.3 Experimental ••••• 127 6.4 Results and Discussion 130 6.4.l Repeatability of EOF Data and Peak Area • • • 130 6.4.2 Influence of Column surface modif ication on EOF • • • • 13l

-2- 6.4.3 Discussion on the Influence of surface Treatment and Addition of SOS on EOF 133 6.4.4 Stability of the Coatings • • •••• 138 6.5 Conclusions • • ••••••• 140 6.6 References •••••••••• 141 CHAPTER 7 • 145

SEPARATION OF HERBICIDES BY MICELLAR ELECTROKINETIC CAPILLARY CHROMATOGRAPHY 145

7.1 Introduction 145 7.2 Theory 147 7.3 Exper ilnental 153 7.4 Results and Discussion 156 7.4.1 Influence of surfactants 156 7.4.2 Influence of Methanol . 162 7.5 Conclusions • 168 7.6 References 169

SUMMARY •• 172 SAMENVATTING 177 CURRICULUM VITAE 182 ACKNOWLEDGEMENT • 184 PUBLICATIONS 185

-3- CBAPTER 1

GENERAL INTRODUCTION

1.1 INTRODUCTIOR

Capillary columns known for their high efficiency, separation speed, inertness and thermal stability, nowadays dominate the field of separation science. Capillary chromatography (including GC, LC and SFC) and capillary electrophoresis have been developed into powerful tools for scientists and analysts specialists werking in widely different areas. Applications include process control, environmental control, chemical, medicine and food industry. Also biological and biochemical research such as the analysis of trace amounts of drugs and metabolites in various body fluids, the separation of proteins and synthetic oligonucleotides and DNA sequencing fragment separation etc. would not advance as fast without open tubular column technology.

The capillary column is the heart of a capillary chromatography system. Different materials such as stainless steel, nickel, teflon, glass and fused silica can be used to make capillary columns with inner diameters between ca. 10 µm and 1 mm. Several manipulations are involved in the preparation of capillary columns, some of them are still not fully understood due to lack of suitable physical or chemical techniques for direct characterization of the inner surface of the capillaries.

-4- In GC, generally, a glass or fused silica capillary column needs a preliminary hydrothermal treatment in order to obtain a controlled density of silanol groups on the column wall (l- 5]. The deactivation of the capillary column completes the column pretreatment and gives the column its necessary inertness, thermostability and wettability for stationary phases of different polarity. After coating of the deactivated capillaries with the desired stationary phase, the coated stationary phase layer is usually subjected to immobilization.

The introduction of flexible fused silica columns by Dandeneau and Zerenner [ 6] in 1979 brought a breakthrough in the application of capillary columns. The main advantages of fused silica capillaries are their mechanica! strength and the increased inertness in comparison with glass columns. The inner surface of the column can be easily modif ied using a wide range of different modification methods (7-12] to be compatible with the desired stationary phases.

Recently, capillary electrochromatography (EC), a separation method in which the electroosmotic flow (EOF) is the driving force for the mobile phase grows rapidly. This technique, which is based on the basic principles of chromatography, uses similar equipment as electrophoresis and can achieve very high efficiencies in a relatively short time [13).

In EC, specially in micellar electrokinetic chromatography (MEKC), a predictable and reproducible EOF through the capillary column is important. The capillary column is used

-5- with or without pretreatment in EC techniques. The column modification, however, can both provide reproducible EOF values and eliminate the interactions between the solutes and the column wall. This is particular important for the separations of proteins and biomolecules. Both dynamic and statie modification of the capillary inner wall can be obtained by different methods.

Numerous parameters are involved in the modification and coating procedures for fused silica capillaries, all of them control the quality and performance of the final columns. In general, maxima! interaction of the solutes with the stationary phase and minimal interaction with the residual active sites on the column inner wall surface is desired. The repeatability of the preparation and the stability of the modification layer and stationary phase are very important factors to obtain reproducible analytica! results.

1.2 SCOPE OF THIS THESIS

As pointed out in the general introduction fused silica is the material of choice for capillary separations. For their application it is essential that the column properties are determined by the bulk properties of the stationary phases and not by the residual active sites of the column wall. constant surface properties are also essential to obtain controllable and reproducible electroosmotic flow in EC.

Therefore deactivation of the column wall and compatibility

-6- of the modified surface with the stationary phases is essential for reproducible analytica! results.

This thesis describes some recent developments in methods of deactivation and crosslinking of the stationary phases for fused silica capillary columns. The experimental study of the deactivation process is supported by solid state 29si NMR measurements on model substrates like fumed silica. Columns prepared following optimized procedures are evaluated by applications both in capillary GC and in EC.

In Chapter 2 the deactivation of fused silica capillary columns by polyhydrosiloxanes with different functional substituents is described. A series of vinyl modified polyhydrosiloxane deactivation reagents containing different functional groups is synthesized for this purpose. The optimization of the deactivation reaction is investigated and evaluated by GC with a dual column system. Also the incorporation of different functional groups into the reagents and their effect on the column performance is investigated.

Due to the small surface area of a capillary column, a model substrate has to be used to characterize the deactivated surfaces by solid state NMR. A study of the deactivation process of silica surface using non-porous silica particles as the model substrate by 29si NMR is described in Chapter 3. The effect of the silylation temperatures and reaction times on the deactivation process of silica surfaces is investigated. The effect of the functional groups incorporated

-7- in the reagents on the deactivation process is studied.

In Chapter 4 the preparation of non-polar and medium polar capillary columns for GC is described, using the optimized deactivation procedures outlined in the previous chapters. crosslinking and/or chemical bonding of the stationary phases by thermal treatment and radical initiation is investigated. The effect of hydrogen and of vinyl groups, which exist in the deactivation layer resulting from the deactivation reaction, in the crosslinking process is studied.

The applicability of the prepared non-polar and medium polar fused silica capillary columns in GC is illustrated in Chapter 5. Narrow-bore columns are used to demonstrate high speed separations for selected samples, whereas normal bore columns are used for the analysis of underi vatized steroids, drugs and priority pollutants. The performance of the capillaries with respect to inertness, thermostability and efficiency is evaluated.

In Chapter 6, a systematic investigation is carried out on the electroosmotic flow (EOF) in micellar electrokinetic capillary chromatography (MECC). Fused silica capillaries are treated with different reagents to investigate their effect on the magnitude and reproducibility of the EOF. Experiments with and without sodium dodecylsulphate (SDS) addition to the buffer are carried out in order to elucidate the effect of this surfactant on the EOF. A model is evaluated to explain the mechanism of the influence of column wall modifications on the

-a- electroosmotic flow.

MECC is applied for the separation of a number of chlorophenoxy acid herbicides (Chapter 7). Sodium dodecyl sulphate (SDS), Brij 35, cetyltrimethylammonium bromide (CTAB) and methanol are added in various concentrations to the buffering eluents to investigate their effects on the separation of the herbicides. Besides that, the differences in the electroosmotic flow (EOF), due to the different composition of the various eluents, are correlated with the viscosities of the mobile phases.

1.3 REFERENCES 1. K. Grob, G. Grob and K. Grob, Jr., Chromatographia, 10 (1977) 181. 2. K. Grob, J. High Resolut. chromatogr. & Chromatogr. Commun., 3 (1980) 493. 3. M. L. Lee and B.W. Wright, J. Chromatogr., 184 (1980) 235. 4. M.W. Ogden and H.M. McNair, J. High Resolut. Chromatogr. & Chromatogr. Commun., 8 (1985) 326. 5. B. Xu and N.P.E. Vermeulen, J. Chromatogr., 445 (1988) 1. 6. R.D. Dandeneau and E.H. Zerenner, J. High Resolut. Chromatogr. & Chromatogr. Commun., 2 (1979) 351. 7. K.K. Unger, Porous Silica, Elsevier, Amsterdam, 1978. 8. K. Grob, G. Grob and K. Grob, Jr., J. High Resolut. Chromatogr. & Chromatogr. commun., 2 (1979) 31. 9. K.K. Unger, G. Schier and u. Beisel, Chromatographia, 6 (1973) 456. 10. R.C.M. de Nijs, J.J. Franken, R.P.M. Dooper, J.A. Rijks, H.J.J.M. de Ruwe and F.L. Schulting, J. Chromatogr., 167 (1978) 231. . ll. C.L. Woolley, R.C. Kong, B.E. Richter and M.L. Lee, J. High Resolut. Chromatogr. & Chromatogr. Commun., 7 (1984) 329. 12. C.L. Woolley, K.E. Markides and M.L. Lee, J. Chromatogr., 367 (1986) 23. 13. W.D. Preffer and E.S. Yeung, Anal. Chem., 62 (l.990) 2178.

-9- CHAPTER 2

DEACTIVATION OF FUSED SILICA CAPILLARY COLUMNS BY POLYHYDROSILOXANES WITH DIFFERENT FUNCTIONAL SUBSTITUENTS

A series of vinyl modified polyhydrosiloxane deactivation reagents containing different functional groups is synthesized. The optimization of the deactivation reaction is investigated and evaluated by GC with a dual column system. Due to the high activity of silylhydride groups in the polyhydrosiloxanes, relatively low temperatures and short reaction times can be used for the silylation. The incorporation of different functional groups into the reagents is emphasized. Some vinyl groups are intentionally left to facilitate crosslinking and/or chemical bonding of the stationary phases. The contribution of the functional groups of the deactivation layer to the polarity of the prepared column is discussed.

2.1 INTRODUCTION

It is generally accepted that inertness is one of the most important parameters of capillary columns in gas chromatography. This aspects is more significant in high-speed GC and trace analysis where columns with reduced inner diameters are used. To prepare highly efficient, inert, heat and solvent resistant columns, they have to be subjected to pretreatments before coating [ 1-3] • The inner surf ace of glass and fused silica is, without adequate treatments, not satisfactorily wettable, at least not by stationary phases having high surface tensions. The films coated on such surfaces are usually irregular and unstable, and tend to form droplets which results in deteriorated column performance.

The weakly acidic silanol groups on the fused silica column wall interact with and adsorb compounds with high peripheral

-10- electron densities. The strained siloxane bridges on the column surface can act as proton accepters in hydrogen bonding interaetions, and cause adsorption of proton donating components.

on a glass surface, in addition to the problems mentioned above, alkali and alkaline earth metal ions cause even more severe problems of adsorptive effects and catalytic activity. The latter problem may also cause the stationary phases to decompose at higher temperatures resulting in increased bleeding of the stationary phases.

The main purpose of deactivation is to create a stable layer which should shield all the active sites of the column surface, resulting in perfectly neutra! columns. The other goal of the deactivation is to obtain good wettability for the stationary phases to be coated. This can be achieved by applying similar materials as the stationary phases in the deactivation procedure. Tremendous amounts of research have been done in the development of column technology and many methods and techniques have been proposed and evaluated by different authors. Aue et al. (4) introduced and ethers successfully developed [S-9) the method of Carbowax 20M deactivation. However, some investigators have shown (9, 10) that Carbowax 20M deactivation layers are not stable for extended use above 250°C. Grob and Grob [11] claimed that polar polyethylene

glycol deactivated surf aces are 11 poorly wettable by apolar

phases 11 •

-11- currently, many deactivation techniques are based on silylation reactions. Silylation reactions are flexible and the regents are widely available. lt has been stated in literature (12-18] that only the free surface hydroxyl groups are involved in silylation reactions, whereas hydrogen-bonded hydroxyl groups show virtually no reaction with the silylation reagent.

..:L 'si-0-H / ....i... 'si-O-H . I \/ >3.1 A 2.4-2.8 A 0 " \ I \ ...... T -Si-0-H T /Si-0-H /

Free •U rfaee Hydrogen·bonded .urface hydroxyl 11roup• hydroxyl groupa Trimethylchlorosilane (TMCS) and hexamethyldisilazane (HMDS) were frequently used to silylate the column surface at lower temperature (<250°C) in the early years of column technology (19-21]. Welsch et al. (22] were the first to attempt high temperature silylation {HTS) of glass capillary columns and showed the benefits of high reaction temperatures in the deactivation process. Their most favourable surface silylation was obtained with pure HMDS at 300°C for 20 h. Based on Welsch's work, different HTS methods ware suggested using various reagents (23-42]. Schomburg et al. (43] described a deactivation procedure known as the PSD {polysiloxane degradation) in which a non-extractable layer of polysiloxane phase is produced on the column surface (44-47]. OH-terminated polysiloxanes were also applied to deactivate capillaries [4S,

49].

-12- Quite high temperatures (normally between 370-450°C} and long reaction times (4-20 hours) are required with above mentioned methods in order to achieve a neutra! and inert surface prior to coating. This might damage the polyimide outer coating of the capillaries resulting in fragile columns.

Recently, methyl-, phenyl- ànd cyanopropyl-polysiloxanes containing silylhydride groups are applied to deactivate fused silica capillary columns at moderate temperatures (50-55]. The deactivation reaction involves dehydrocondensation and yields neutra! and highly inert capillaries.

It is essential that the coating and im.mobilization of the stationary phase after deactivation should not deteriorate the inertness of the deactivated surface. Deterioration is observed when common free radical initiation techniques with peroxides and azo compounds are used for the immobilization. The peroxide residues can be bonded to the polymerie stationary phase and result in an increased adsorptivity and decreased thermal stability (56-58]. A second popular deactivation reagent is azo-tert-butane (ATB). ATB only slightly contributes to the polarity of the immobilized phases (33). The performance of ATB, however has been found to deteriorate with time [59, 60]. Moreover, columns treated with aged ATB showed high activity levels, solute retention shifts, high bleed and erratic immobilization efficiencies [59, 60).

Based upon these observations and considerations, an ideal process consists of low temperature deactivation and

-13- immobilization of the stationary phases without the use of any additional reagents. Polyhydrosiloxanes can be an attractive choice due to their relatively low reaction temperatures and short deactivation times.

In this chapter, the applicability of polyhydrosiloxane deactivation reagents is investigated. A series of vinyl modified polyhydrosiloxane based deactivation reagents containing different functional groups were synthesized f or this purpose (61). Vinyl groups were intentionally incorporated into the polymer chain for facilitating later immobilization of the stationary phases. The reaction conditions for the deactivation reaction are optimized. All GC experiments were performed on a dual column system. Gas chromatographic retention indices were measured for critical solutes of different polarity in the test mixture. The peak shapes of the recorded chromatograms were compared in order to evaluate the degree of deactivation obtained with these reagents. The influence of surface modification on the polarity of the stationary coating is also discussed. The effect of hydrogen and vinyl groups in the crosslinking process of the stationary phases will be discussed in Chapters

3 and 4.

2.2 'l'HEORY

Fairly high temperatures are required for deactivation with octamethylcyclotetrasiloxane and polydimethylsiloxanes, due to the acidity of the silanols on the fused silica column wall

-14- [50]. on the contrary, with polyhydrosiloxanes, covalent bonds are formed from the dehydrocondensation between hydride silicon bonds and silanols on the column wall [50, 62, 63]:

' 1 1 1 -si-OH + H-Si-Me ----> -si-o-si-Me + H2 t (2 .1) 1 1 1 1

This reaction is favoured by acidic surface conditions and can reach 100% conversion under medium temperature conditions. Basic column surfaces cause deprotonation of silanols thereby hindering the above mentioned reaction (64]. Hydrolysis of Si-H bonds according to equation (2. 2), is essential for intra- and intermolecular cross-linking of PMHS.

1 1 Me-si-H + H20 ----> Me-si-OH + H2 t (2.2) 1 1 The physisorbed water on the column surface induces initial hydrolysis at the PMHS-water interface. The hydrolysis product can react further with Si-H as shown in the equation (2.3).

1 1 1 1 Me-Si-OH + H-Si-Me ----> Me-si-o-si-Me + H2 t (2.3} 1 1 1 1

The covalent bonds between the polymer molecules are formed by eq 2. 3. The reactions shown in eq 2. 1 and 2. 3 are irreversible at temperatures above 200°C. Therefore, under the optimum conditions for deactivation, a thin and densely crosslinked and unextractable layer was formed on the column wall.

The hydrolysis products can react also according to:

-15- 1 1 Me-si-OH + HO-Si-Me ----> Me-Si-O-Si-Me + H20 (2.4) 1 1 1 1

This reaction is reversible in the presence of water formed in the reaction. Therefore, the deactivation process will be deteriorated by the presence of an excess of water on the column wall [65].

Substitut ion of other functional groups instead of methyl groups will decrease the reactivity of the Si-H bonds because of both the steric hindrance effect and the electron withdrawing effect of the other functional groups [51, 52). Additionally, the irregular distribution of the silanols and siloxane bridges on the column surface should be noticed. The increased molar ratio of other substituted functional groups make it more difficult to obtain a complete deactivation compared with PMHS.

(2.5)

The critical surface tension of a column deactivated with the reagents shown above depends on the ratio of m, n, and p in formula 2.5. In order to minimize the effect of polar sites in the deactivation layèr on the retention of the solute in coated columns, an optimized amount of the different functional groups in the deactivation reagents should be used.

-16- 2.3 EXPERIMENT.AL 2.3.l Materials and Reaqents

1,3,5,7-tetramethylcyclotetrasiloxane (D4H), 1,3,5,7-tetra­ vinylmethylcyclotetrasiloxane (D4Vi), bis-(cyanopropyl)di­ chlorosilane, diphenyldimethoxysilane, polymethylhydrosiloxane (PS 122, 85 es, fluid) were all purchased fro:m Petrarch systems (Bristol, PA, USA). These reagents were used without further purification.

Fused silica capillary columns with inner diameters of 0.22 mm and 0.32 mm were obtained fro:m Chrompack International BV

(Middelburg, NL), the 50 µ.m columns from Scientific Glass Engineering PTY LTD. (Ringwood, Australia). All solvents were PA grade (Merck AG, Oarmstadt, FRG), deionized water was used throughout this study (Millipore Syste:m, Millipore Corp., Bedford, USA).

2.3.2 Synthesis of Polyvinylmethylhydrosiloxane (PVMHS)

D4Vi (2 .1 g) and o4H (13. 2 g) were mixed together (molar ratio, 10:90) with 3 :ml dichloromethane. A few drops of concentrated sulfuric acid were added to catalyze the polymerization. The mixture was stirred at room temperature for 48 h. Then 5 drops of hexamethyldisilazane were added to the mixture and stirring was continued for another 16 h to enhance the endcapping. This endcaps the ter:minated hydroxyl groups of the linear polymers. The mixture was then dissolved in 10 ml of CH2cl2 and washed 8 times with an equal volume of

-17- water. The solvent was removed by distillation. Finally, the fluid was dried at 60°C under vacuum for 6 hours.

2.3.3 synthesis of Polydiphenylvinylmethylhydrosiloxane (POPMHS)

Diphenyldimethoxysilane (4.88 g) and D4Vi (0.172 g) was mixed

in 2 ml CH2c12 and stirred after addition of 5 drops of

concentrated sulfuric acid at room temperature. Then D4H (1.08

g) dissolved in 2 ml CH 2Cl2 was dropped into the mixture (molar ratio, 50:5:45) during 30 minutes in an Argon atmosphere. The mixture was stirred for 60 h. For the endcapping, the same procedure as used for the synthesis of PVMHS was applied.

2.3.4 synthesis of Polybis-(oyanopropyl)vinylmethyl­ hydrosiloxane (PBCPVMHS)

The monomer, bis-(cyanopropyl)dimethoxysilane, was achieved according to the following conversion reaction: Cl OCH3 1 1 R-Si-R + R-Si-R + (2.6) 1 1 Cl OCH3

The dichlorosilane was added in drops to the stirred excess trimethylorthoformate to reach full conversion. Thereafter the polymerization was performed according to the procedure for the synthesis of PDPVMHS.

-18- 2.3.5 Pretreatment of the :rused Silica Capillaries

The capillary columns were hydrothermally treated by filling the column with a 10% (w/w) solution of HCl to 90% of their volume and then heated to 150°C for 1 h. Next, the column was rinsed with a 2% HCl solution and with deionized water until neutral. Finally the columns were washed with 5 ml of methanol and dried at 250°C for l h under a flow of Helium.

2.3.6 Deactivation and ~esting Procedure

Each column was dynamically coated with a 5% or 10% (v/v) solution of the deactivation reagents in pentane or dichloromethane. The speed at which the coating solution was pumped through the column was kept at about 25 to 30 cm/min. After flushing the columns with Helium for 30 min and flame sealing, the columns were heated to different temperatures during different reaction times. After deactivation the capillary columns were carefully rinsed with 10 ml of dichloromethane and dried at 250°C for l h under a flow of helium. The deactivated fused silica capillaries were tested by GC with a dual column system consisting of a coated precolumn and the column to be tested. A test mixture (cf. Table 2 .1) which contains components with different adsorptivity was used for evaluation of the deactivated c.olumns ( 65) • The test method is essentially as described by Schomburg et al. [10]. It allows the examination of peak shapes and peak areas for a quantitative comparison of the column activity for the components in the test mixture. The column polarity is

-19- measured by means of Kováts retention indices for the deactivated column as well as for the coated capillary. Zero­ dead volume connectors (Unimetrics Corp., CA, USA or Gerstel

GmbH, Mülheim a/d Ruhr, FRG) were used to connect the deactivated column to a high quality coated column, (home

made, PMHS deactivated OV-1 precolumn, L = 10 m, i.d. = 0.25 mm, film thickness df = 0.25 µm). An average linear velocity of 30-40 cm/sec (Helium) was used.

Tabla 2.1 Composition of the test mixture. Peak No. is the elution order on the precolumn. Peak Components conc. Representation for No. (mg/l) 1 decane 119.0 --- 2 1-octanol 112.4 silanol, hydrogen bond 3 2,6-dimethylphenol 118.3 basic sites 4 undecane 116.6 --- 5 2,6-dimethylaniline 114.7 acid sites, silanol 6 dodecane 114.4 --- 7 decylamine 116.4 acid sites, silanol, exposed siloxane bridge 8 tridecane 107.8 --- 9 nicotine 121.0 acid sites, silanol 10 tetradecane 128.4 ---

The amount of each component injected onto the column system is about 1 ng. The measurements are performed on a Packard 430 gas chromatograph (Packard, Delft, NL) equipped with a split injector and an FID. All experiments were carried out

isothermally at ioo 0 c.

-20- 2.4 RESULTS AND DISCUSSION

2.4.l Deactivation with Pol1Jllethylhydrosiloxane (PMRS)

The deactivation conditions used for the deactivation of the fused silica capillaries with PMHS are given in Table 2.2.

Table 2.2 survey of the deactivation conditions of fused silica capillaries. Column Reaction conditions No. ID (µm) L (m) T (OC) Time (h} Reagent 1 220 5 ------untreated 2 220 10 240 4 PMHS 3 220 10 280 4 PMHS 4 220 10 300 4 PMHS 5 1) 220 10 ------6 2) 220 10 280 4 PMHS 7 320 4 240 4 PVMHS 8 320 4 280 4 PVMHS 9 320 4 300 4 PVMHS 10 320 4 320 2 PVMHS 11 320 4 350 0.5 PVMHS 12 3) 320 4 ------l) After reconditioning of column No. 3 at 300°C for 30 h with Helium flushing. 2) Flushed with water saturated Helium for l hour prior to deactivation. 3) After rinsing and reconditioning column No. 10 at 300°C for 30 h with Helium flushing.

-21- l

a 1 b 4

2 4

6 3 6

7 s 8 10

0 2 4 6 B 10 min

1

c d

!~ s 4

6 5

6 7 8 9 8 10 9 10

0 2 4 6 8 0 2 4 6 B 10 min

Fiqure 2.1 Representative chromatograms of the test mixture. (a) precolum.n; (b) precolumn plus untreated column No. 1; (c) precolumn plus PMHS deactivated column No. 2; (d) precolumn plus PMHS deactivated column No. 3. For test conditions see text and peak identification, see Table 2.1.

-22- The normalized peak areas (NA), corrected for the weight difference in the test mixture but not for FID response factors and the corresponding Kováts retention indices measured at 100 °c for both the precolumn and the evaluated capillaries are given in Table 2.3.

Table 2.3 Normalized peak areas (NA) and Kováts retention indices (RI) at 100°c of the test components of the PMHS deactivated columns. Col. Test components No. 1-octanol 2,6-DMP 12, 6·-DMA Decylamine Nicotine NA RI NA RI NA RI NA RI NA RI O* 86 1051.3 96 1081. 0 101 1136.5 86 1233.3 92 1309.1 1 11 1087.3 70 1093.8 61 1159. 0 0 -- 0 -- 2 73 1052.5 94 1093.7 99 1138.l 11 1251. 5 53 1313.8 3 86 1052.5 95 1081. 4 103 1136.l 81 1234.4 90 1310.8 4 85 1052.4 94 1081. 6 98 1137.0 75 1235.5 85 1311.1 5 86 1052.6 95 1081. 7 102 1136.8 80 1235.1 89 1311. 4 6 47 1053.0 80 1082.2 87 1137.9 21 1244.3 64 1312.6 * The precolumn.

Considering the representative chromatograms for the precolumn and the differently deactivated columns given in Fig. 2.1 and the data presented in Table 2.3, it is obvious that excellent deactivation is obtained after heating the column to 280°c for 4 hours (#3) or at a higher temperature during a shorter time {#4). The deactivated column #3 was heated at 300°C for 30 h and rinsed with pentane and dichloromethane respectively. No significant changes in the normalized peak areas and the retention indices of the components were observed. This high thermal stability and resistance to solvent rinsing was explained by Woolley et al [55]. They assume that a very thin

-23- (several monolayers) but densely crosslinked film was formed by dehydrocondensation (eqns. 2.1-2.3) and that the surface is covered by anchored Si-Me groups with methyl protruding upwards from the surface. This is in agreement with the observation that the Koväts retention indices of a well deactivated column are in acceptable agreement to those of the precolumn. As shown in equation (2.2), physically adsorbed water on the surface plays an important role in the formation of a densely cross-linked network. An excess of water, however, will deteriorate the deactivation layer due to the presence of unreacted Si-OH groups (cf. eq 2.4). This is illustrated by the strong adsorption of 1-octanol and aminodecane on column No. 6 (cf. Table 2.2 and 2.3).

2.4.2 Deactivation with Polyvinylmethylhydrosiloxane (PVMHS)

The deactivation conditions (e.g. temperatures and times) for a number of short fused silica columns (L = 4 m) with PVMHS are summarized in Table 2.2. Representative chromatograms of the PVMHS deactivated columns (at 320°C during 2 hours) before and after solvent rinsing are given in Fig. 2. 2. The corresponding normalized peak areas and retention indices with this copolymer are presented in Table

2.4.

For reasons of comparison, the deactivated capillaries were tested before and after solvent rinsing. Comparing the columns

#3 (Table 2. 3) and #9 (Table 2. 4) , it can be seen that a

-24- slightly higher temperature is needed for complete deactivation with PVMHS as compared to the PMHS deactivation. This can be explained by steric effects so that a higher deactivation temperature is required to form a densely cross­ linked layer and to cover the residual active sites in this case.

:5 5

b 6 a 6

8 8 7 7 9 9 10 1C

' ...... -....__ _,..,, _____ --- L-·-~ .....__ ... ..,, __~ ---- 0 2 4 6 8 min 0 2 4 6 8 min

Fiqure 2.2 Representative chromatograms of the test mixture on the precolumn plus PVMHS deactivated column No. 10 (a} before solvent rinsing; (b} after solvent rinsing. For test conditions see text and peak identification, see Table 2.1.

Comparing the Kováts indices given in Table 2.4 for one of the most optimal columns (e.g. #9) and the precolumn an increase of the retention indices can be observed. This indicates that the incorporation of vinyl groups into the cross-linked deactivation layer can be considered successful. Further

-25- studies with solid state 29si NMR on model substrates, like cab-O-Sil [66] were required to confirm these results and will be described in Chapter 3.

'l'able 2.4 Normalized peak areas (NA) and Kováts retention indices (RI) at 100°c of the test components of the PVMHS deactivated columns.

Col. Test components No. 1-octanol 2,6-DMP 2,6-DMA Oecylamine Nicotine NA RI NA RI NA RI NA RI !NA RI 0 86 1051.3 96 1081. 0 101 1136.5 86 1233.3 92 1309.1 Before rinsing wi:th solvent 7 80 1054.6 - -- - - ~~ •• ""0 --- ~1319.7 8 85 1054.6 92 1086.5 99 1142.4 27 1241.8 1314.5 9 86 1054.8 93 ~1238,4 88 1315.5 10 87 1055.8 94 1238.8 88 1316.5 11 86 1055.7 95 1238.0 90 1316.4 After rinsing with solvent

7 80 1054.6 90 1086.l 99 1142.l 0 1--- 36 1319.0 8 86 1054.2 90 1086.6 100 1141. 5 30 1241.1 87 1313.6 9 86 1055.2 91 1085.6 100 1142.6 75 1238.l 89 1314.7 10 86 1055.7 93 1087.1 100 1143.5 78 1238.1 89 1315.5 11 85 1055.1 92 1087.4 100 1145.0 83 1238.8 90 1316.7 12 85 1053.0 96 1090.1 100 1147.2 80 1240.5 87 1318.5

2.4.3 Deactivation with PDPVMHS

The deactivation conditions (e.g. reaction times and tempera­ tures) for the deactivation with PDPVMHS are summarized in

Table 2.5.

A survey of the corresponding normalized peak areas and Kováts retention indices with POPVMHS before and after solvent

-26- rinsing is given in Table 2.6. Corresponding representative chromatograms of the test mixture (cf. Table 2.1) are shown in Fig. 2.3.

Table 2.s Survey of the silylation conditions of fused silica capillaries, all columns are 0.32 mm i.d. * 4 m.

Column No. Reaction conditions T (OC) Time (h) Reagent 13 260 4 PDPVMHS 14 280 4 PDPVMHS 15 300 4 PDPVMHS 16 320 2 PDPVMHS 17 350 2 PDPVMHS 18 1) ------19 260 4 PBCPVMHS 20 280 4 PBCPVMHS 21 300 2 PBCPVMHS 22 320 1 PBCPVMHS 23 350 0.5 PBCPVMHS 24 2) ------1) After rinsing and reconditioning column No. 17 at 350°C for 40 h with Helium flushing. 2) After rinsing and reconditioning column No. 23 at 280°C for 1 h with Helium flushing.

-27- 1 1 4 4 a b 2 1!3 '"'3 5 6 5 6

a a

7 9 10 10 _...._ 9 LJ. l [_ - '; 0 2 4 6 a 2 4 6 l min 0 a min 14 1,..3

5 c d

6 3 56

7 a 7 a 9 10 9 10

l~ ..._ -- ...... 0 2 4 6 a min 0 2 4 6 a min

Figure 2.3 Representati ve chromatograms of the test mixture on the precolumn plus PDPVMHS deactivated column No. 14 (a) before solvent rinsing; (b) after solvent rinsing; and precolumn plus PDPVMHS deactivated column No. 17 (c) before solvent rinsing; (d) after solvent rinsing. For test conditions see text and peak identification, see Table 2.1.

-28- Table 2.6 Normalized peak areas (NA) and Kováts retention indices (RI) at 100°c of the test components of the pDPVMHS deactivated columns.

Col. Test components No. 1-0ctanol 2,6-DMP 2,6-DMA Decylamine Nicotine 1 NA RI NA RI NA RI NA RI I 0 86 1051.3 96 1081. 0 101 1136.5 86 1233.3 92 1309.l

Before ~insing ~ith solvent 13 82 1056.7 88 1088.6 100 1147~ 1252.1 71 1344.5 14 83 1056.2 89 1092.4 99 1151 1243.5 81 1328.5 15 83 1058.6 93 1099.0 100 1157.5 53 1243.3 85 1337.6 16 85 1065.6 91 1113.7 100 1178.6 59 1254.0 86 1350.2 17 84 1067.3 91 1118.2 99 1183.8 77 1245.9 88 1351. 5 ~ After dn1ing with solvent 1057. 2 87 1090. 3 100 1146. 2 33 1249.5 76 1335.6 1055.8 901089.6~ 1147.444 1241.7 70 1317.5 ~055. 7 ~1091.8 99 1150.5158 1240.6 86 1325.l 057.0 921090.81001150.761 1242.7 86 1322. 5 17 84 1058.1 91 1093.7 100 1152.2 82 1240.5 90 1323.7 18 84 1057.9 92 1092.8 100 1152.0 59 1239.4 87 1322.7

Similar steric effects and shifts of Kováts retention indices were observed with PDPVMHS as with PVMHS deactivation.

Obviously, a deactivation layer which incorporates polar functional groups will show a shift of Kováts indices to a higher value. It can be seen that there are larger retention

index shifts of the test components in the mixture for the freshly deactivated columns compared to the rinsed columns

(#16 and #17). From that it can be concluded that some unreacted or decomposed short chain materials are physically

attached or just left on the surface before solvent rinsing.

The residual contribution to the retention behaviour of the

-29- test components after solvent rinsing, even if the deactivation layer is very thin, indicates the incorporation of the characteristic functional groups in the deactivation layer.

After heating the column. No. 17 at 350°C for 40 hours, no significant changes in the column activity were noted, which indicates the good thermal stability of this deactivation layer. In previous studies [23-25, 66], the loss of the phenyl groups from the silica surface at temperature above 250°C was observed when 1,3-diphenyltetramethyldisilazane {DPTMDS) was used as the deactivation reagent [43, 67]. The two phenyl groups are attached to different silicon atoms, bath together with two methyl groups. Blum and Eglinton [68] pointed out that asymmetrical substitution leads to a dipole moment at the respective silicon atoms, which makes it more susceptible for nucleophilic attacks by silanol groups or traces of water. Improved thermal stability of the polymer will be obtained by symmetrical substitution. This is in agreement with the re sults shown above. Moreover, the deactivation layer is solvent resistant, neither the normalized peak areas nor the retention indices were changed by rinsing the column with 10 ml each of n-pentane, acetone and dichloromethane.

2.4.4 Deactivation with PBCPVMHS

Deactivation conditions {e.g. reaction temperatures and times) for this reagent are presented in Table 2.5.

-30- A survey of the corresponding normalized peak areas and Kováts

retention indices is presented in Table 2. 7. Corresponding representative chromatograms of the test mixture are shown in Fig. 2. 4.

12 1 4 4

5 6 3 a b 2

~5 6

8

8 9 9 10 10

7

t L _,_,_...... ,.__1u1 ___J _ L" J 0 2 4 6 8 min 0 2 4 6 8 min

Figure 2.4 Representati ve chromatograms of the test :mixture on the precolwnn plus PBCPVMHS deactivated column No. 21 {a) before solvent rinsing; (b) after solvent rinsing. For test conditions see text and peak identification, see Table 2.1.

-31- Table 2.7 Normalized peak areas (NA) and Kováts retention indices (RI) at 100°c of the test components of the PBCPVMHS deactivated columns.

Col. Test components No. 1-0ctanol 2,6-DMP 2,6-DMA Decylamine !Nicotine NA RI NA RI NA RI INA RI INA RI 0 86 1051.3 96 1081. 0 101 1136.5 86 1233.3 92 1309.1 Before rinsing with solvent

19 78 1060.5 89 1118. 6 99 1164.9 19 1236.5 93 1344.5 20 83 1063.1 90 1130.2 99 1173.2 25 1236.8 92 1328.5 21 83 1064.9 89 1139.6 100 1181.1 34 1237.5 96 1337.6 22 85 1065.9 91 1143.2 100 1182.9 45 1237.5 95 1350.2 23 83 1066.5 93 1143.4 100 1183.2 66 1237.8 97 1351. 5 After rinsing with solvent 19 81 1058.5 90 1107.0 99 1156.2 0 --- 87 1335.6 20 83 1061.9 90 1126.1 98 1169.5 0 --- 89 1317.5 21 84 1062.0 90 1125.5 99 1169.8 0 --- 92 1325.1 22 85 1063.3 93 1134.0 98 1174.5 0 --- 90 1322.5 23 84 1063.9 91 1134.5 100 1174.5 0 --- 91 1323.7 24 83 1065.6 91 1138.3 99 1178.4 21 1238.6 92 1322.7

As is shown in Table 2.7 the normalized peak areas and the retention indices of the polar components in the mixture for

PBCPVMHS deactivated columns increase with silylation temperature. Most probably this again can be explained from the steric hindrance effect. In the case of higher temperatures, the steric effect is decreased and a densely cross-linked network is formed at the column wall.

From Fig. 2.4, it can be seen that the decylamine peak totally vanished after solvent rinsing. This demonstrates that some

-32- residual silanols still existed on the column surface after the physically attached polymer layer was washed away. Consequently this caused the adsorption of aminodecane, one of the most critical components in the test mixture, After shortly reconditioning the column (No. 23, cf. Table 2.5 and 2. 7) by heating, the decylamine peak elutes again, however not quantitatively. Reheating the rinsed column resulted in a further reaction between the silylhydride and the residual silanols on the column wall. Apparently, higher temperatures and/or longer reaction times are required to complete the deactivation of fused silica capillaries with PBCPVMHS.

The wettability of the capillary columns by polar stationary phases depends on the critical surface tension (CST) of the column wall. The (CST) can be improved by increasing the content of polar functional groups in the deactivation layer. On the other hand, the polarity of the deactivation layer should not be increased too much. A high content of polar functional groups in the deactivation layer, which enhances the wettability by the stationary phases, may result in an increase of the overall column polarity expressed in the Kováts retention indices. This effect is illustrated in a chromatogram of the Grob test mixture given in Fig. 2.5 for a column which has been deactivated with a 50% cyanopropyl molar ratio of PBCPVMHS. The column was coated with OV-1701 (df • 0.3 µm), which contains only 7% cyanopropyl groups in the polymer. The inverted elution order of 2,3-butanediol and n-decane compared to the chromatograms in ref. [69) illustrates that the deactivation layer significantly

-33- contr ibutes to polarity of the column. Matching of the polarities of the deactivated column wall and the stationary phases should be carefully optimized in order to achieve not only a high column efficiency but also a defined column polarity.

4 3 s

6 7 s 9

LJ11ll•• ______• ._...... _~.____...... _ ___

0 10 20 min

Figure 2.s Representative chromatogram of Grob mixture #2 on a column (320 µm I.C. * 4 m) coated with OV-1701, df= 0.3 µm (see text for the details). GC condition: He, 25 cm/sec., oven temperature from 40°C to 200°c at 4°C/min; peak identification: 1 = decane, 2 = butanediol, 3 = octanol + dodecane, 4 = 2,6-dimethylphenol, 5 = 2,6-dimethylaniline, 6, s, 9 =methyl caprate, undecanoate and laurate respectively, 7 = dicyclohexylamine.

2.5 CORCLUSZOHS

The synthesized vinyl modified polymethylhydrosiloxanes with different functional groups incorporated in the polymer chain are very useful for silylation of silanol groups on the column wall of fused silica capillaries. Excellent deactivation of the fused silica surface was obtained with PMHS and PVMHS.

-34- Relatively high temperature and/or long reaction times are needed for diphenyl and bis-(cyanopropyl) substituted reagents compared to the methyl substituted deactivation reagent. This is due to steric hindrance effects and an electron withdrawing effect of the functional groups in the polymer chain. Complete silylation of the silanol groups on the column wall appears most difficult for the bis-{cyanopropyl) substituted polymers.

Physisorbed water on the inner surface plays an important role to achieve a thin and densely crosslinked layer on the column wall. Too high a water content, however, will deteriorate the deactivation layer due to the presence of unreacted Si-OH groups caused by the hydrolysis of Si-H groups (see eq. 2.2).

Compared to methyl deactivated columns an increase of the Kováts retention indices of polar test compounds indicates a successful incorporation of functional groups in the deactivation layer. If different functional groups are present in the deactivation layer, the individual contribution of each of these groups to the solute retention cannot be distinguished. A careful optimization of the content of functional groups in the deactivation layer and the stationary phase is required in order to achieve efficient columns with a reproducible polarity.

Confirmation of these conclusion with solid state 29si NMR on a model substrate, Cab-0-Sil, and a more detailed interpretation of the deactivation process will be presented in Chapter 3.

-35- 2.6 REFERENCES l. M.L. Lee and B.W. Wright, J. Chromatogr., 184 (1980) 235. 2. L.S. Ettre, Anal. Chem., 57 (1985) l419A. 3. B. Xu and N.P.E. Vermeulen, J. Chromatogr., 445 (1988) 1. 4. W.A. Aue, C.R. Hastings and s. Kapila, J. Chromatogr., 77 (1973) 299. 5. W.A. Aue and D.R. Younker, J. Chromatogr., 88 (1974) 7. 6. D.A. Cronin, J. Chromatogr., 97 (1974) 263. 7. O.A. Cronin, J. Chromatogr., 101 (1974) 271. a. J.J. Franken, R.c.M. de Nijs and F.L. Schulting, J. Chromatogr., 144 (1977) 253. 9. R.C.M. de Nijs, J.J. Franken, R.P.M. Dooper, J.A. Rijks, H.J.J.M. de Ruwe and F.L. Schulting, J. Chromatogr., 167 (1978) 231. 10. G. Schomburg, H. Husmann and F. Weeke, Chromatographia, 10 (1977) 580. 11. K. Grob and G. Grob, J. Chromatogr., 125 (1976) 471. 12. V.Y. Davydov, C.T. Zhuravlev and A.V. Kiselev, Trans. Faraday Soc., 60 (1964} 2254. 13. C.G. Armstead and J.A. Hockey, Trans. Faraday Soc., 63 (1967) 2549. 14. E.K. Lippincott and R. Behoeder, J. Chem. Phys., 23 (1955) 1099. 15. G.C. Pimental and A.L. McClellan, The Hydrogen Bond, Freeman, San Francisco, CA, 1960. 16. M.L. Hair and w. Hertl, J, Phys. Chem., 73 (1969) 2372. 17. G.E. Berendsen and L. de Galan, J. Liq. Chromatogr., 1 (1978) 403. 18. M. Mauss and H. Engelhardt, J. Chromatogr., 371 (1986) 235. 19. M. Novotny and K. Tesarik, Chromatographia, 1 (1968) 332. 20. M. Novotny, L. Blomberg and K.D. Bartle, J. Chromatogr. Sci, 8 (1970) 390. 21. G.A. F .M. Rutten and J .A. Luyten, J. Chromatogr., 74 (1972) 177. 22. Th. Welsch, W. Engewald and Ch. Klaucke, Chromatographia, 10 (1977) 22. 23. K. Greb and G. Grob, J. High Resolut. Chromatogr. Chromatogr. Conunun., 1 (1978) 527. 24. K. Grob, G. Greb and K. Grob, Jr., J. High Resolut. Chromatogr. Chromatogr. Commun., 2 (1979) 31. 25. K. Grob, G. Grob and K. Grob, Jr., J. High Resolut. Chromatogr. Chromatogr. Commun., 2 (1979) 677. 2 6. K. Grob and G. Grob, J. High Resolut. Chromatogr. Chromatogr. Commun., 3 (1980) 197. 2 7 • K. Grob, J. High Reso l ut. Chroma togr. Chroma togr. Commun., 3 (1980) 493. 28. M. Godefroot, M. van Roelenbosch, M. Verstappe, P. Sandra and M. Verzele, J. High Resolut. Chromatogr. Chromatogr. Commun., 3 (1980) 377. 29. L. Blomberg, K. Markides, J. Buijten and T. Wännman, J. High Resolut. Chromatogr. Chromatogr. Commun., 3 (1980) 527.

-36- 30. L. Blomberg, J. Buijten, K. Markides and T. Wänmnan, J. Chromatogr., 239 (1982) 51. 31. S.R. Lipsky and W.J. McMurray, J. Chromatogr., 239 (1982) 61. 32. T. Welsch, R. Müller, w. Engewald and G. Werner, J. Chromatogr., 241 {1982) 41. 33. B.W. Wright, P.A. Peaden, M.L. Lee and J.J. Stark, J. Chromatogr., 248 (1982) 17. 34. J. Buijten, L. Blomberg, K. Markides and T. Wännman, J, Chromatogr., 237 (1982) 465. 35. J. Buijten, L. Blomberg, K. Markides and T. Wännman, Chromatographia, 16 (1982) 183. 36. K. Markides, L. Blomberg, J. BuijtenandT. Wänrunan, J. Chromatogr., 254 (1983) 53. 37. K. Markides, L. Blomberg, J. Buijten and T. Wännman, J. Chromatogr., 267 (1983) 29. 38. K. Markides, L. Blomberg, s. Hoffmann, J. Buijten and T. Wännman, J. Chromatogr., 302 {1984) 319. 39. M.L. Lee, R.c. Kong, C.L. Woolley and J.s. Bradshaw, J. Chromatogr. Sci., 22 (1984) 136. 40. J. Buijten, L. Blomberg, s. Hoffmann, K. Markides and T. Wännman, J. Chromatogr., 289 (1984) 143. 41. W. Blum, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 120. 4 2. T. Reiher, J. High Resolut. Chromatogr. Chromatogr. Commun., 10 (1987) 158. 4 3. G. Schomburg, H. Husmann and H. Borwi tzky, Chromatographia, 12 (1979) 651. 44. G. Schomburg, H. Husmann and H. Behlau, Chromatographia, 13 (1980) 321. 45. G. Schomburg, H. Husmann, w.s. Ruthe and M. Herraiz, Chromatographia, 15 (1982) 599. 46. M.A. Moseley and E.D. Pellizzari, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 404. 47. M.A. Moseley and E.D. Pellizzari, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 472. 48. W.Blum, W.J. Richter and G. Eglinton, J. High Resolut. Chromatogr. Chromatogr. Commun., ll (1988) 148. 49. F. David, P. Sandra and G. Diricks, J. High Resolut. Chromatogr. Chromatogr. Commun., ll (1988) 256. 50. C.L. Woolley, R.C. Kong, B.E. Richter and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 329. 51. K.E. Markides, B.J. Tarbet, C.L. Woolley, C.M. Schregenberger, J.S. Bradshaw, K.O. Bartle and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 378. 52. K.E. Markides, B.J. Tarbet, C.M. Schregenberger, J.S. Bradshaw and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 741. 53. C.L. Woolley, K.E. Markides and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Conunun., 9 (1986) 506. 54. C.L. Woolley, K.E. Markides and M.L. Lee, J. Chromatogr., 367 (1986) 9. 55. C.L. Woolley, K.E. Markides and M.L. Lee, J. Chromatogr., 367 (1986) 23. 56. E.M. Barral II, R. Hawkins, A.A. Fukushima and J.F. Johnson, J. Polym. Sci., Polym. Symp., 71 {1984) 189.

-37- 57. A. Aerts, J. Rijks. A. Bemgard and L. Blomberg, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 49. 58. L. Blomberg, J. Microcol. Sep., 2 (1990) 62. 59. B.A. Jones, T.J. Shaw and E.S. Francis, Pittsburgh Conference, 1989, Abstract No. 1319. 60. T.J. Shaw and B.A. Jones, llth International Symp. on Capillary Chromatogr., May 14-17, 1990, Monterey, CA, USA. 61. Q. Wu, c. Cramers and J. Rijks, lOth International Symp. on capillary Chromatogr., May 22-25, 1989, Riva del Garda, Italy. 62. M.G. Voronkov and N.V. Kalugin, J. Appl. Chem. USSR, 32 (1959) 1612. 63. I.A. Shikhiev, R.Yu. Gasanova and B.M. Guseinzade, J. Gen. Chem. USSR, 38 (1968) 188. 64. E. Wiberg and E. Amerger, "Hydrides of the Elements of Maingroups I-IV", Elsevier, New York, 1971, Chap. 7. 65. M. Hetem, G. Rutten, J. Rijks, L. van de Ven, J. de Haan and c. Cramers, J. Chromatogr., 477 (1989) 3. 66. G. Rutten, J. de Haan, L. van de Ven, H. van Cruchten and J. Rijks, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 664. 67. T. Welsch and H. Frank, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 709. 68. W. Blum and G. Eglinton, J. High Resolut. Chromatogr. Chromatogr. Commun., 12 (1989) 290. 69. K. Greb and G. Grob, J. High Resolut. Chromatogr. Chromatogr. Commun., 6 (1983) 153.

-38- CHAP'l'ER 3

A SOLID STATE 29s1 NMR STUDY OF THE DEACTIVATION OF SILICASURFACESUSING NON-POROUSSILICAPARTICLES AS THE MODEL SUBSTRATE

The effect of the silylation temperatures and reaction times on the deactivation process of silica sur.taces is investigated with solid state 29si NMR spectroscopy. These model experiments using fumed silica as a model substrate for fused silica can reveal the structure and the chemical properties of the deactivation layer on the surface of the particles. Interpretation of the results yields valuable information on the chemical details of the deactivation reaction. The advantages and disadvantages of solid state 29si NMR as a characterization method f or the surface properties are discussed.

3.1 INTRODUCTION

Tremendous improvements in the deactivation of capillary surfaces have been achieved by high temperature silylation and, more recently, by low temperature dehydrocondensation [ 1- 6]. The development of column technology clearly benefited from the increasing understanding of the reaction mechanisms involved in the deactivation process.

The characterization of deactivated surfaces has proved useful for the description of the modif ied surface and is performed mostly by chromatography. Other techniques such as cross­ polarization magie angle spinning {CP-MAS) NMR and lH NMR spectroscopy, normal and total reflectance IR spectroscopy, GC-MS, scanning electron microscopy, secondary-ion mass spectrometry, electron spectroscopy for chemical analysis and capillary rise measurements have also been used to fulfil this purpose. The surface area of the inner column wall, however,

-39- is far too small to fit the requirement of most of these characterization techniques. The combination of solid state 29si NMR spectroscopy and experiments with a fumed silica material like Cab-o-sil as a model for the column wall may give the researchers · some representative pictures of what occurred inside the column during the deactivation process. In fact, it has been proved to be a powerful tool for the characterization of such species [7-14].

In the framework of a long term systematic study on column technology, Chapter 2 presented the results of a study on the deactivation of fused silica capillary columns with polyvinyl­ methylhydrosiloxane (PVMHS) and polydiphenylvinylmethylhydro­ siloxane (PDPVMHS). Excellent deactivations were obtained at a temperature of 300°C for PVMHS and 350°C for PDPVMHS, respectively. According to Lee and coworkers [3-6] this type of reagents serves two purposes in the process of deactivation. The first one is to get rid of the active sites of the column wal! by the reaction between the silylhydride and surface silanols present at the column wall. The second important characteristic of the deactivation layer is that it increases the critical surface tension of the column inner surface. Due to this increase in the surface tension of the column wall wettability is obtained for phenyl containing stationary phases without loss of the deactivation of the column wall.

It has been stated in literature that with PMHS deactivation, all silylhydride groups react with silanol groups to form a

-40- layer of a densely crosslinked polymer with its methyl groups protruding upwards from the surface [6]. It was found, however, by 29si NMR studies [ 14] that the reaction of silylhydride groups with the surface silanols is not complete due to the intrinsically low reactivity of the silylhydride groups and steric hindrance. In previous studies [11, 15-19], the loss of phenyl groups from the silica surface was observed at temperatures above 250°C when 1,3-diphenyltetramethyldisilazane (DPTMDS) was used as the deactivat ion reagent [ 11, 18, 19] • The two phenyl groups are attached to different silicon atoms, both together with two methyl groups. Blum and Eglinton [20) pointed out that asymmetrical substitution leads to a dipole moment at the respective silicon atoms, which makes it easier accessible for nucleophilic attacks by silanol groups or traces of water. An improved stability of the polymer can be obtained with the symmetrically substituted reagent. This was the third reason to select PDPVMHS as the deactivation reagent in this study.

In this chapter, a parallel investigation of the deactivation with PVMHS and PDPVMHS was carried out by solid state 29si NMR spectroscopy. A fumed vitreous silica, Cab-o-Sil MS, was used as the model substrate for detailed studies on the nature and the chemical structure of the deactivation layer and for confirmation of the intended chemical reactions at the column wall. Estimation of the relative amounts of the functional groups as determined by 29si cross-polarization magie angle spinning (CP-MAS) NMR combined with elemental analysis can elucidate the reaction mechanisms of the deactivation

-41- reactions.

The effect of the vinyl groups, which were incorporated in the synthesized copolymer on the deactivation process are described. The advantages and disadvantages of 29si NMR as a characterization method of the surface properties are also discussed.

3.2 THBORY

The deactivationprocess which occurs on the surface of silica particles can be studied by 29si CP-MAS NMR. The specific chemical structure of the functional groups are determined by NMR signals of deactivated silica samples. This gives information on the properties and relative amounts of the siliceous groups incorporated in the deactivation layer. Typical capillary columns with inner diameters between 50- 530 µm have surface areas far too small to allow the application of most spectroscopie analysis techniques for surface characterization. Therefore, model materials such as glass sheets (21, 22], silica gels (7-10] or fumed silica (11- 14, 18] have to be used for model studies on surface deactivation and modification utilizing these techniques for surface characterization. A non-poreus fumed silica, Cab-O-Sil MS, was chosen because of its compatibility with 29si CP-MAS NMR (high surface area of about 200 m2 /g) and its close similarity to the fused silica column wall (no pores, hence the reaction is not limited by diffusion processes).

-42- Below a brief overview of the principles of 29si CP-MAS NMR is given.

If a nucleus is placed in an external magnetic field it will interact with this external field. Moreover, it will interact with the magnetic field generated by spinning neighbouring nuclei. The spin states are determined by the spin quantum number. The corresponding energy levels are governed by Planck's law:

( 3 .1)

where E: the energy level, h: the Planck constant, vo: Larmor frequency.

v0 is related to Bo (magnetic strength of the external field) according to:

y *Bo \10 (3. 2) 2 * 'lt where -y: the gyromagnetic ratio, a physical property of the nucleus.

The sensitivity of NMR measurements is determined by the natural abundance of a nucleus and its 'Y value. The shielding of the nuclei by surrounding electrons causes the resonance frequency to shift. The extend, to which the resonance frequency shifts depends on the chemica! structure of the molecules and provides qualitative information on the structure. NMR line in NMR spectra of stationary, solid samples are

-43- broadened by a number of factors. Two of the more important interactions are chemical shift anisotropy (C.S.A.) and dipolar interactions between nuclei. C.S.A. arises because the shielding of nuclei in a sample by the bonding electrons depend on the direction of the sample {e.g. the main axes of a crystal) with respect to the external magnetic field. In fact, the shielding should be described by a tensor, 3 . In liquids, rapid molecular tumbling reduces this to the trace of the tensor: ae = (axx + ayy + Uzz)/3. The line broadening, resulting from C.S.A. can be described a Hamiltonian of the general form:

Hc.s.A. = Y * h * Ï * ä * B (3.3) ....where I spin quantum number, = ....a shielding tensor, B external magnetic field.

In a sample consisting of small crystallites (powder) this can be written as:

(3. 4) where c1 : a constant, r: the distance between the nuclei, 8: the angle between the magnetic field direction and the internuclear axis.

When the sample is spun around an angle with the magnetic field for which (3 * cos28 - 1) = 0 1 the C.S.A. broadening is greatly reduced. This technique is called Magie Angle Spinning (MAS) and 8 = 54°44' is called the magie angle. Moreover, antifacts ("spinning side bands") remain in the spectrum of

-44- the sample. They disappear gradually when the spinning speed is enhanced.

Dipolar interactions between nuclei arise by coupling through space. The appropriate Hamiltonian for these interactions is:

2 2 - '::t - HD = y * h * I * D * I (3.5) where

::$ D : the dipolar coupling tensor. In contrast with C.S.A., the dipolar interactions in Hertz does not depend on the external field strength. Equation 3.5 can be simplified to:

3 * cos2 6 - 1 r3 (3. 6) where C2: a constant.

The broadening effect of dipolar interactions will only disappear when the MAS spinning frequency is larger than the dipolar interaction in frequency units. In practice, this means that small interactions (like e.g. in 29si-o-1H over larger distances can be removed by MAS. In 13c solid state NMR, however, one bas to cope with very large dipolar interactions between 13c and lH nuclei (40 kHz for stationary, solid samples). These interactions are removed by dipolar decoupling (irradiation of the protons, while exciting and observing e.g. 13c NMR signals).

Unfortunately, however, the sensitivity of 29si MAS-NMR is quite low because of the low natural abundance of 29si (4.7%).

-45- As the sensitivity is proportional to the square root of the measuring time, a significant gain in sensitivity can only be obtained at the expense of very long measuring time. On the other hand, measuring time also depends on the relaxation time

T1, the time necessary for a nuclear spin to return from energy level E2 to E1 to restore equilibrium. Only when equilibrium is restored can excitation occur again.

cross-Polarization (CP) techniques can shorten the relaxation time. In this technique a large magnetization of e.g. protons polarize the 29si nuclei and this is possible only between closely spaced atoms. By adjusting the frequencies of the proton and the 29si to meet the Hartmann-Hahn condition, a link between the proton and the 29si magnetization is formed (24). This results effectively in a reduction in the relaxation time of 29si nuclei. Only 29si nuclei near a spatially fixed proton which are obviously present solely at the surf ace of the silica particles contribute to the NMR signal in 29si CP-MAS NMR. This makes 29si CP-MAS NMR a surface analysis technique suited for silica materials.

3 • 3 EXPERJ:MEHTAL

3.3.1 Materials

1,3,5,7-tetramethylcyclotetrasiloxane (D4H), 1,3,5,7-tetra­ vinylmethylcyclotetrasiloxane (D4Vi),diphenyldimethoxysilane and polymethylhydrosiloxane (PMHS, 85 es, fluid) were all purchased from Petrarch Systems (Bristol, PA, USA). These

-46- reagents were used without further purification.

PVMHS was synthesized from o4H and o4vi (molar ratio= 90:10) according to the procedure developed in our laboratory [25, 26). Similarly, PDPVMHS was synthesized from diphenyldi­ methoxysilane, o4vi and D4H in a molar ratio of 50:7:43.

The Cab-O-Sil M5 (Cabot Corp., Tuscola, Ill., USA) was a gift from Heybroek & Co. Handelmij. NV (Amsterdam, NL). The specific area of the sample is 202 m:i /g determined by the saET method after the heat and rehydration treatment. This value will be used in this chapter. All solvents used were P.A. grade from E. Merck (Darmstadt, FRG). The deionized water was obtained from a Milli-Q System (Millipore, Bedford, MA, USA).

3.3.2 Silylation of Cab-O-Sil and Solid state 29si NMR Measurements

The Cab-o-sil was heated to 720-750°C for 48 h to remove "water inclusions" [ 27). After cooling, the Cab-O-Sil was

rehydrated by refluxing it with pure water for 5 hours, then the water was removed by a rotary evaporator. Afterwards, the rehydrated sample was dried in a vacuum oven at 110°c overnight and stored over P205 in a vacuum desiccator for several weeks.

This dried Cab-O-Sil was coated with the deactivation

reagents. Approximately l g of Cab-O-Sil was added to solutions of 10% of PVMHS and PDPVMHS in n-pentane respectively and mixed by sonification. Aftar slowly

-47- evaporating the solvent at room temperature with a rotary evaporator, a polymer film with a film thickness of approximately 5 nm was deposited on the silica surface.

The coated Cab-O-Sil samples were then dried in an oven at

100°C at reduced pressure for 1 h. A quartz ampoule which was filled with about O. 3 g of coated Cab-O-Sil sample then evacuated was repeatedly filled with Helium up to atmospheric pressure. After sealing the tube to a volume of about 8 ml, the ampoule was wrapped in an aluminum foil and heated to the desired silylation temperature for 16 hours. After silylation the ampoule was opened and the content was rinsed twice with dichloromethane and subsequently dried overnight at 110°C under reduced pressure.

Solid. state 29si NMR spectra of the samples listed in Table

3.1, were obtained with a Bruker CXP300 spectrometer at 59.63

MHz, in a similar way as discussed extensively by Hetero et al.

[ 14].

The carbon contents of the silylated Cab-O-Sil samples were

determined with a Perkin Elmer 240 Element Analyzer (Perkin

Elmer, Norwalk, CT, USA). The IR spectra of the synthesized

copolymers were measured with a Perkin Elmer 237 Grating

Infrared Spectrophotometer.

-48- Tabla 3.1 Survey of the silylated Cab-O-Sil samples.

sample Silylation conditions No. Temperature ( oc) Reagents 1 240 PVMHS 2 280 PVMHS 3 320 PVMHS 4 360 PVMHS 5 280 PDPVMHS 6 300 s 7 320 PDPVMHS 8 360 PDPVMHS

(A}

(B)

-30 -32 -34 -36 -38 -32 -36 -40 -44 -48 ppm ppm

:riqure 3.1 High resolution liquid 29si NMR spectra of the synthesized reagents, pulse interval time 10 s, acquisition time 0.1 s. (a) PVMHS, molar ratio= 90:10 (Si-H:Si-Vi), NS (numbers of free induced decays) = 666; (b) POPVMHS: molar ratio = 50:7:43 (Ph:H:Vi), NS= 5300.

-49- 3.4 RESULTS ABD DISCUSSION 3.4.1 Characterization of the Synthesized Reagents.

A representative high resolution liquid 29si NMR spectrum of the synthesized PVMHS and PDPVMHS deactivation agents is given in Figure 3.1. The relative amounts of the functional groups determined by the intensity ratios of the 29si NMR signals are comparable with the molar ratios as given in the experimental section. The structure and functionality notation codes of the most relevant chemical shifts for the liquid reagent as well as the silylated Cab-O-Sil samples are presented in Figure 3. 2. Obviously all the required functional groups are present in the synthesized deactivation reagents. code: T, a,

CH, CH, 0 CH•CH,H H OH CH, OH O- -o-l-cH-CH -l-~o-~-o-l1-o-~-o-L-o-~-o-L-o-:-~-È: 1 0- 0- 0 CH CH CH CH 0 - 0- 0- l • ·ri r' r' r' r' r' r.' r. t ppm: r·20 -20 ~I-46 -~ .3: .3: .5; -66 -101 ·110

Figure 3.2 Structures and functionality notation codes of the most relevant siliceous surface moieties and their corresponding chemical shifts.

The IR spectra showed in Figure 3. 3 also confirmed the presence of all the functional groups in the reagents.

-50- 100

î C•C

1.1111 Si-H 3000 2000 1600 Si-C113 f 1000 WÁVENUHBER ( ctf I ) Si-0-Si

\00 ·-~ITT·, ! !J 1·1 ~ • " ,

, 1· _,.1, I' 1, ' ·1 • • . " . : -~~ l· ·' .. • • . ". j t ·il ~ J i i Si-Ph ~ ~- ;, ; Si-H

• ! ; .•• 1 • . ·jl·L; d1 11. · " . ; 0 1.ILE~: l tl. /iOOO 3000 2000 1609 tiooo WAVENlJMBER cci-r ) Si-0-Si

Fiqure 3.3 Infrared spectra of the synthesized copolymers. Upper: PVMHS, lower: PDPVMHS.

3.4.2 Evaluation of PVMKS Silylated Cab-o-sil Samples with Solid State 29si NKR. Spectroscopy

Solid state 29si CP-MAS NMR spectra of the PVMHS silylated Cab-O-Sil samples (Table 3.1: #1-4) are given in Fig. 3.4. The corresponding relative ratios of the 29si CP-MAS NMR signals of bonded siliceous moieties of these samples are presented in Table 3.2.

-51- Table 3.2 Relative ratios of the 29si NMR signals of bonded siliceous moieties of the PVMHS silylated Cab-O-Sil samples. Sample Relative ratio of the 29si NMR siqnals No. D2ED2 D2H D'2H D2Vi T2 T3

1 4 61 29 2 -- 4 2 4 49 41 -- -- 6 3 5 39 45 -- -- 11 4 7 31 43 -- 2 17

Figure 3.4 clearly illustrates that different results were obtained with different reaction conditions. rt is evident that excellent deactivation was achieved by this reagent (note the disappearance of the geminal and vicinal silanols which usually give signals at -90 and -101 ppm respectively). The main silylation products, having chemical shifts in the -35 ppm region show that methylhydrosiloxanes remained after the silylation. A flexible polymer chain is formed on the surface as is indicated by the appearance of narrow peaks at chemical shifts around -35 ppm and -36 ppm in the spectra (Figure 3.4). The presence of intra and intermolecular crosslinking in the surf ace polymer layer of the Cab-O-Sil samples is indicated by the signal of methyltrisiloxysilane (T3) (see eq. 2.1 in Chapter 2). The intensity of the signal increased with

increasing temperatures. The T3 groups were formed mainly by the reaction between silylhydrides and surface silanols. The number of T3 groups increases at higher temperature due to the degradation of the polymer chains.

The 29si NMR signal of D2ED2 , which is a non-desirable crosslinking product of the silylation, indicates the reaction of silicon hydride and vinyl groups incorporated in the

. -52- polymer. The presence of vinyl qroups is not evident after the silylation reaction. This is due to the relatively low amount of vinyl groups in the polymer (even lower after the silylation reaction) and the lack of sensitivity of the 29si NMR technique.

4-'

3 ----"'"\,.__

2

-20 -40 -60 -80 -100 -120 ppm

Figure 3.4 Solid state 29si NMR spectra of PVMHS silylated Cab-O-Sil samples (Table 3.1: #1-4). Q4 is the signal from the silica surface.

The amount of vinyl groups in the copolymer prior to the deactivation is 5% on a molar basis. Because of the higher density of silanol qroups on the cab-O-Sil particles as compared to fused silica capillary columns, the silylation

-53- reaction may not be exactly the same as that on the capillary column surface. Higher surface concentration of silanol groups may cause more pronounced crosslinking and chemical bonding of the deactivation agent, which can also result in a higher conversion of the vinyl groups. It can be seen from Table 3.2 that the amount of D2ED2 increases with increasing temperature, this is in agreement with the above explanation.

The carbon content of the PVMHS silylated Cab-O-Sil samples is 10.8%. This matches very well with the expected carbon content of the samples and shows a complete crosslinking and chemical bonding of the polymer layer to the particle surface.

3.4.3 Evaluation of PDPVMRS Silylated Cab-o-sil samples with Solid state 29si mm. Spectroscopy

Figure 3.5 presents the solid state 29si CP-MAS NMR spectra of the PDPVMHS silylated Cab-O-Sil samples (Table 3.1: #5-8). The corresponding relative ratios of the 29si NMR signals of the bonded siliceous moieties of these samples are given in Table 3.3.

Ta:ble 3.3 Relativa ratios of the 29si NMR signals of bonded siliceous moieties of the PDPVMHS silylated Cab-O-Sil sam les. Sample No. D ED D H + D' H D T 5 7 26 40 10 17 6 6 17 39 9 18 7 7 18 39 36 8 11 14 39 36

-54- -20 ~60 -100 ppm

Figure 3.5 Solid state 29si CP-MAS NMR spectra of PDPVMHS silylated Cab­ o-sil samples. (Table 3.1: #5-8).

Analogous to the situation in the PVMHS experiments the intra and intermolecular crosslinking in th.e surface polymer layer of the Cab-O-Sil samples is observed. The amount of T3 groups increases with increasing temperature again indicating the formation of a more rigid and densely crosslinked layer at the Cab-O-Sil surface.

Again, o2Eo2 (-20 ppm) was produced by the reaction of silicon hydride and vinyl qroups incorporated in the polymer as a non- desirable crosslinking product. Referring to a recently

-55- published study on silylation reactions with PMHS reagent (14], it can be estimated from the 29si CP-MAS NMR spectra of the silylated Cab-O-Sil samples, that a relatively thin layer was formed (l-2 nm) instead of 5 nm calculated from the amount of the reaqent. It should be emphasized, however, that the measurements with CP excitation are semiquantitative.

Experimentally, an averaqe carbon content of 17. 6% was obtained by element analysis of the rinsed silylated Cab-O-Sil samples. Compared to the expected carbon content (about 30%) in the coated Cab-O-Sil, a considerable amount of the deactivation reaqent is washed away by solvent flushing of the deactivated Cab-0-Sil samples after silylation. Most probably, further crosslinking of the polymer is obstructed by the large ster ic hindrance effect of diphenyl groups. The thermal degradation of the polymers should also be taken account in this respect (26). Although, a significant part of the reagent is lost by solvent rinsing, owing to inferior crosslinking, the final result is a rigid, thin and stable network consisting of short polymer chains bonded to the silica surface. This is indicated by the appearance of broadened peaks in the spectra (Figure 3.5).

Onder the silylation conditions applied in this study, the 29si CP-MAS NMR resonances showing thermal decomposition of diphenyldisiloxysilane groups were not observed. Bath phenyl groups remained chemically bonded to the siloxysilane copolymer at silylation temperatures up to at least 360°C. Loss of phenyl groups would have been accompanied by

-56- transforxnation into other moieties, which have not been observed in the spectra. With 1,3-diphenyltetramethyl­ disilazane, which has been aften used as a silylation reagent, phenyl losses were observed already at temperatures above 250°C [11]. This leads to a decreased wettability of the capillaries for phenyl containing stationary phases.

3.S CONCLUSIONS

PVMHS and PDPVMHS are efficient deactivation reagents for fused silica capillary columns. The main reaction between silylhydride groups and the surface silanols produces a densely crosslinked and chemically bonded layer which covers the particle surface and blocks the interaction with active sites on the surface. 29si NMR provides valuable information on the structures and properties of organo-siliceous groups at a non-porous silica surface after silylation. The reaction between Si-H and si-Vi groups by thermal treatment is confirxned by solid state 29si NMR experiments. The quantity of the reaction product, D2ED2 increases with increasing temperature. The existence of vinyl groups on the silylated Cab-O-Sil samples, however, is not evident, due to the low amount of vinyl groups in the copolymer and the lack of sensitivity of the 29si NMR technique. Semi-quantitative results of the thickness of the polymer layer can be obtained from the 29si NMR spectra and elemental analysis. Thick layers (> 3 nm) give narrow and sharp peaks while thin films (< 2.5 nm) show broadened signals. From this

-57- point of view, the thickness of the modification layer of PDPVMHS silylated Cab-O-Sil samples can be estimated to be about 1-2 nm. The spectra obtained show that the deactivation layer is stable up to at least 360°C. No loss of phenyl groups from the modification layer was observed. The thermal stability of the silylation layer is enhanced using the reagent with two phenyl groups connected to one silicon atom. This is expected because of the electron donating properties of the geminal phenyl groups and their chemical stability.

3.6 REFERENCES 1. C.L. Woolley, R.C. Kong, B.E. Richter and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 329. 2. K.E. Markides, B.J. Tarbet, C.L. Woolley, C.M. Schregenberger, J.S. Bradshaw, K.D. Bartle and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. co:mmun., 8 (1985) 378. 3. K.E. Markides, B.J. Tarbet, C.M. Schregenberger, J.S. Bradshaw and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. commun., 8 (1985) 741. 4. C.L. Woolley, K.E. Markides and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Co:mmun., 9 (1986) 506. 5. C.L. Woolley, K.E. Markides and M.L. Lee, J, Chromatogr., 367 (1986) 9. 6. C.L. Woolley, K.E. Markides and M.L. Lee, J. Chromatogr., 367 (1986) 23. 7. G.E. Maciel and o.w. Sindorf, J. Amer. Chem. Soc., 102 (1980) 7606. 8. G.E. Maciel and o.w. Sindorf, J. Chromatogr., 205 (1981) 438. 9. O.W. Sindorf and G.E. Maciel, J. Phys. Chem., 86 (1982) 5208. 10. o.w. Sindorf and G.E. Maciel, J. Amer. Chem. Soc., 105 (1983) 1487. 11. G. Rutten, A. van de Ven, J. de Haan, L. van de Ven, and J. Rijks, J. High Resolut. Chromatogr. & Chromatogr. Commun., 7 (1984) 607. 12. G. Rutten, J. de Haan, L. van de Ven, A. van de Ven, H. v. Cruchten, and J. Rijks, J. Resolut. Chromatogr. & Chromatogr. Commun., 8 (1985) 664. 13. L. van de Ven, G. Rutten, J. Rijks and J. de Haan, J. Resolut. Chromatogr. & Chromatogr. Commun., 9 (1986) 741. 14. M. Hetem, G. Rutten, B. Vermeer, J. Rijks, L. van de

-!!>S- ven, J. de Haan, and c. Cramers, J. Chromatogr., 477 {1989) 3. 15. K. Grob, G. Grob, and K. Grob, Jr., J. High Resolut. Chromatogr. & Chromatogr. Commun., 2 {1979) 31. 16. K. Grob, G. Grob, and K. Grob, Jr., J. High Resolut. Chromatogr. & Chromatogr. Commun., 2 (1979) 677. 17. K. Grob and G. Grob, J. High Resolut. Chromatogr. & Chromatogr. commun., 3 (1980) 197. 18. T. welsch and H. Frank, J. High Resolut. Chromatogr. & Chromatogr. Commun., 8 (1985) 709. 19. T. Reiher, J. High Resolut. Chromatogr. & Chromatogr. commun., 10 (1987) 158. 20. w. Blum and G. Eglinton, J. High Resolut. Chromatogr. & Chromatogr. commun., 12 (1989) 290. 21. M.L. Lee, D.L. Vassilaros, L.V. Phillips, O.M. Hercules H. Azumaya, J. W. Jorgenson, M. P.. Maskar inec and M. Novotny, Anal. Letters, 12 (1979) 191. 22. B.W. Wright, M.L. Lee, s.w. Graham, L.V. Phillips and O.M. Hercules, J. Chromatogr., 199 (1980) 355. 23. c.s. Yannoni, Acc. Chem. Res., 15 (1982) 201. 24. S.R. Hartmann and E.L. Hahn, Phys. Rev., 128 (1962) 2042. 25. Q. wu, C.A. cramers and J.A. Rijks, in P. Sandra (Ed.), Proc. of the lOth International Symposium on Capillary Chromatography, Riva del Garda, Italy, May 1989, Hüthig, 1989, p. 188. 26. Q. wu, M. Hetero, C.A. Cramers and J.A. Rijks, J. High Resolut. Chromatogr., 13 (1990) 811. 27. J.A.G. Taylor and J.A. Hockey, J. Phys. Chem., 70 (1966) 2169. 28. N. Grassie and I.G. Mcfarlane, Europ. Polymer Journal, 14 (1978} 875.

-59- CHAPTER 4

PREPARATION OF NON-POLAR AND :MEDIUM POLAR FUSED SILICA CAPILLARY COLUMNS FOR GC

Non-polar and medium polar capillary columns are prepared based on the deactivation desaribed in previous ahapters. Crosslinking and/or chemical bonding of the stationary phases by thermal treatment and the azo-tert-butane (ATB) initiation is investigated. The effect of hydrogen and of vinyl groups, whiah exist in the deactivation layer resulting :trom the deactivation reaation, on the crosslinking process are discussed. Evaluation of the prepared columns is carried out on the aspects o:t efficiency, inertness, thermal stability and resistance to commonly used solvents. Narrow bore capillary columns down to 23 µm i.d. with a non-polar stationary phase were successfully prepared by a similar procedure.

4.1 INTRODUCTION

Although efficient sample introduction systems, sensitive detectors and other auxiliary devices are essential components in modern high resolution gas chromatographic systems, the column remains the haart of the analytical instrument. The rapid growth in the application of capillary columns in chromatography promotes the development of column technology [1-3].

Tremendous efforts have been invested in the pretreatment of the inner surface of glass and fused silica capillaries in order to improve the quality of the resulting columns. The aim of the pretreatments is to eliminate and/or to shield the active sites on the inner wall of the columns, while simultaneously improving the wettability of the column wall by the stationary phases. This has been extensively discussed

-60- in Chapters 2 and 3.

For obtaining high separation efficiencies in any type of capillary column it is essential that the stationary phase is deposited as a smooth, thin homogeneous film. This film must also maintain its integrity without forming droplets when the column temperature is varied. The wettability of the column is related both to the nature of the stationary phase and the column surface to be coated. A number of studies (4-13] on the wettability of column surfaces has been reported. capillary rise method (1, 11) was used to determine the critica! surface

tension (CST, 70 ). From the Zisman plots [14, 15], in which the eosine of the advanced contact angle (9) is plotted

against the surface tension of the test liquid (11) 1 can be obtained (see Fig. 4.1 and 4.2). These graphs can be used to characterize the column surface.

Fiqure 4.1. Diagram f or the contact angle

At equilibrium, the relation of three interfacial tensions can be described by Young's equation:

COS 8 = Ysg - Ysl ( 4 .1) Y1g where îsg• îsl and îlg are the surface tension between solid

-61- and gas (vapour of the wetting liquid) , solid and liquid, liquid and its vapour respectively.

The significance of the critical surface tension ('Yc) is that, in general, liquids with n :S 'Yc will completely wet the surface while liquids with 'Yl > îc tend to form droplets.

0.80

o.so

0.40

0.20

0.00 ,______._ ___,______,..__ _ _.______. 10 20 30 40 50 60 v1 (N/m)

Fiqure 4.2. Zisman plot of a silylated fused silica capillary column, data from ref. ( 11].

Wettability of the column can be enhanced by roughening the surface and by chemica! modification of the surface. Unfortunately, the first method is not applicable to fuse silica columns as has been shown by Lee et al. (1, 16].

Replacement of surface silanols with silyl ether groups (silylation} or other inert reagents containing functional

-62- groups similar to those in the stationary phase is the most successful way for both deactivation and enhancement of the wettability of the surface (see Chapter 2).

Coating of the capillaries can be accomplished by either dynamic or statie methods. The dynamic method described by Dijkstra and De Goey [17] is a simple coating technique but often results in non-uniform films [l-3, 18]. Many solutions have been proposed to improve this method, such as controlling the coating speed and temperature [19-22], controlling the solvent evaporation rate (23] and the volatility o.f the solvent [18], concentration of the coating solution [24J, and using the "mercury plug" method [25, 26]. Different approaches were suggested to calculate the film thickness {df) in dynami­ cally coated capillaries (19, 23, 27-35). The following equation was found to be the most useful (33], except for low viscosity solutions.

(4.2) where r: the column radius (m), u: the coating velocity (m/sec), c: the concentration of the solution (% V/V), ~= the viscosity of the solution (kgm-lsec-1), ~= the surface tension of the solution (N/m).

In contrast to dynamic coating, in statie coating the column is filled with a solution of the stationary phase in a volatile solvent. After sealing one end of the column the solvent is evaporated and the stationary phase deposited as a film on the column wall. This technique comm.only provides

-63- more efficient columns [36-55]. The film thickness of the stationary phase can be easily calculated from the equation:

(4.3)

where P is the phase ratio e.g. ~ = Vm/V6 , Vm and V6 are the volumes of the mobile and stationary phase respectively. 1/~ is equal to the concentration (% V/V) of the coating solution. An important advantage of statie method is that the p is known, therefore, the df can be accurately determined. The density of the stationary phase must be known for this calculation. For a number of frequently used stationary phases density values are tabulated in ref. [40]. In practica, it is important that the coating solution is dust free and degassed to eliminate bumping during the solvent evaporation step. Furthermore it is important that no air or vapour bubbles exist in the column, especially at the sealed end [36, 40, 43- 55].

The main drawback of the statie method is its time consuming nature. Higher coating temperatures [56] and more volatile solvents [57 / 58] were applied to accelerate the coating procedure [59, 60). A high temperature-high pressure coating method {61) and the free-release statie coating procedure (62- 65] were also proposed to speed up the coating procedure. These methods are especially suitable f or coating narrow bore capillary columns as the typical statie technique with dichloromethane or n-pentane as the solvent is extremely time consuming for this type of columns.

-64- A precipitation coating method was introduced by Dluzneski and Jorgenson ( 66 J to coat s µ;m. and 10 µ.m columns for open-tubular LC. This method involves filling of a heated capillary column with a hot stationary phase solution and then cooling the column rapidly so that the stationary phase precipitates from the solution onto the column wall.

Immobilization of the stationary phase by in situ crosslinking and/or chemical bondinq on the column surface after coating is one of the significant advances in column technology. The physical and chemical stability of the stationary phase and its ability to resist the strain caused by mechanica! vibration, high temperatures and the injection of large volume of polar solvents is enhanced (16, 67-69].

Early efforts on im:mobilization of the stationary phases involved ionic reaction mechanism in the crosslinking process, which results in Si-o-si linkages [70-81). High column activities and low column efficiencies were obtained by these approaches due to the large number of residual silanols and alkoxy qroups. In addition, sample molecules generally become less soluble in the stationary phase as a result of the high degree of cros slinking. A more recent development ( 83] is crosslinking by using OH- or methoxy terminated silicones with different functional groups [84-98]. Excellent therm.al stability was obtained by this method owing to the stability of the Si-O-Si bonds formed by the condensation between the surface silanols and the terminal silanol groups in the stationary phase (16, 99].

-65- currently, free radical initiation, first introduced by Grob [100] is the most frequently used method to immobilize stationary phases. Peroxides [67, 100-116] such as dicumyl peroxide (DCP), and azo compounds (e.g. azo-tert-butane, ATB)

[16, 59, 103, 115-124] are normally applied for this purpose.

Radiation by high energy sources [108, 125-131] has also been used as the means of initiation of stationary phase immobilization. However, the necessity to have special facilities severely hinders the application of this technique. Also ozone has been used for the crosslinking of various stationary phases [70., 106, 109, 132], even at room temperature [70]. It has been found, however, that the

presence of tolyl groups in the stationary phases is a prerequisite for this method. In the absence of tolyl groups the immobilization of the stationary phases was not possible.

An excellent review on the immobilization of stationary phases

has recently been published [133].

The main drawback of immobilization initiated by peroxides is

that peroxide residues and by-products can be incorporated into the phases (133-135]. This may result in an increased

adsorptive activity, decreased thermal stability and solute

retention indices shifts (133]. The oxidation of tolyl groups

in the stationary phases has also been observed when high

percentages of peroxide were used to cure the column [115,

116].

'' Immobilization of silanol-terminated stationary phases may

require fairly high temperatures or the addition of tri- or

tetra- functional crosslinking agents to improve the

-66- immobilization yield. It has been stated that additional silanol groups must be present throughout the polymer chain in order to immobilize the stationary phases via thermally induced condensation [136). In some case, a compromise had to be made between the degree of immobilization and maximum selectivity of hydroxyl terminated cyanopropylsilicones (137].

Azo-tert-butane (ATB) introduced by Wright et al. (103) contributes only slightly to the polarity of the immobilized stationary phases because its by-product is an alkane. Unfortunately, however, a small portion of the radical generator is inevitably incorporated in the phase matrix [ 133 J. Additionally, the performance of ATB has. been found to deteriorate with time (138, 139]. The columns treated with aged ATB showed high activity levels, high bleed rates, erratic immobilization eff iciencies and solute retention shifts. The regeneration or clean-up of ATB is required to maintain or increase its activity for the immobilization process [133, 139]. Another type of immobilization was developed by Schomburg and coworkers (70, 140, 141). In the crosslinking reactions, stationary phases containing 1-alkene groups were hydrosilylated by polymethylhydrosiloxane in the presence of a chloroplatinic .acid catalyst, as shown in the following equation:

j H2PtCl6 j 1 -ri-H + H2C=CH-ri------> -ri-CH2-CH2-ri- (4.4)

In this chapter, the preparation of non-polar and medium po lar

-67- capillary columns was investigated for columns deactivated with the deactivation reagents described in Chapter 2. Vinyl modified deactivation reagents were synthesized in order to facilitate the inunobilization of the stationary phases. In contrast to the platinum catalyzed method, the immobilization was performed by thermal treatment to simplify the preparation procedure and to avoid possible undesired products originating from the catalyst. The radical ini tiation method was also used to compare the performance of vinyl and non-vinyl containing stationary phases and to elucidate the effect of hydrogen and vinyl groups in the crosslinking process. Polysiloxanes with varying methyl and phenyl contents were coated on the deactivated fused silica capillary columns. The performance of the prepared columns was evaluated with respect to inertness, efficiency, thermal stability and crosslinking efficiency.

4.2 EXPERIMENTAL 4.2.1 Materials and Apparatus

The materials, solvents and columns are described in the experimental section of Chapter 2. A Packard Becker GC (Model 430, Packard Becker, Delft, NL) and a Varian GC (Model 3400, Varian Instruments, Walnut Creek, CA, USA) were used to evaluate the deactivated and coated capillaries. Evaluation of the narrow bore capillaries was carried out on a carlo-Erba GC (Model 2900, Carlo-Erba Instruments, Milan, Italy). All GC experiments were performed with hot split injection and flame ionization detector (FID).

-68- 4.2.2 synthesis of the Deactivation Rea9ents

The synthesis of PVMHS and PDPVMHS is described in Chapter 2. In order to obtain deactivation layer with different phenyl 9rOUp contents, a fixed volume of PDPVMHS copolymer was mixed with various volumes of PMHS [142, 143] (see Table 4.1).

'l'a:ble 4 .1 Survey of the deactivation reagents and the stationary phases used in this study.

Stationary phases* Deactivation reagents Ratio of Ph:Vi:H silane OV-73 (5.5% diPh) 6:1:93 OV-7 (20% Ph-Me) 21:3:76 OV-61 (33% diPh) 35:5:60 ov-11 (35% Ph-Me) 135:5:60 OV-17 (50% Ph-Me) 50:7:43 * The remaining substituant of the polymer chain is dimethyl.

4.2.3. Column Preparation

Fused silica capillary columns were pretreated and deactivated under different deactivation conditions as described in

Chapter 2. The deactivated columns were statically coated with an appropriate concentration of the stationary phase solution. Then, i:mmobilization of the stationary phases was achieved either by thermal treatment or radical initiation with ATB.

The following procedure was used for the thermal method: the coated column was heated from 40°C to 250°C overnight (> 16 h) under a slow flow of helium (5-10 cm/sec). The column was then

-69- rinsed with 10 ml dichloromethane and conditioned by progranuning the temperature from 40°C to 280°C at 4°C/min under a normal flow of helium (25-30 cm/sec) and kept at 280°C for 10 hours. For ATB cured columns, a procedure similar to that described in ref. [103] was used for the immobilization of the stationary phases. The regeneration of ATB was carried out according to the procedure described in ref. [133]. Aged ATB was used for the immobilization of one column to show its effect on the quality of the crosslinked column.

4.2.4 Evaluation of the Prepared capillaries

The inertness of the columns was tested both aftar deactivation and aftar coating using the test mixture listed in Table 2.1. The split ratio was adjusted so that approximately 1 ng of each component was injected onto the column. For the deactivated capillaries, a dual column system as described by Schomburg et al. [144] was used, which consists of a high quality home made PMHS deactivated ov-1 precolumn (10 m * 250 µm i.d., film thickness df = 0.25 µm) coupled in series to the column to be tested. Isothermal chromatograms of the test mixture were recorded for the precolumn alone and the precolumn connected to the deactivated column. The test chromatograms were evaluated with respect to the peak shape, peak retention ( expressed as kováts retention indices) and the normalized peak area of the critical components. Additionally

for the coated capillaries, Grob 1 s test mixture [122, 145] was used for the sake of comparison.

-10- Column efficiency is characterized by calculating the coating efficiency (C.E.). C.E. is defined as the ratio of the theoretica! plate height to the experimental plate height, Hexpi under optimal conditions [16, 36, 142):

C.E. = ( · Htbeor ) (4. 5) H•xr; min

The minimum theoretica! plate height, Htheor1 is represented by the simplified Golay-Giddings plate height [ 146, 14 7] equation:

1 H r * k 2 + 6 * k + 1)2 (4.6) tbeor =( {J3'(l+k)} )*(ll where r is the column radius and k the capacity factor. This simplification can only be used for thin film columns under conditions of low pressure drop. The Kováts retention index [148, 149] of a solute is calculated by using the following equation:

1 1 ( log t log t ) I(x) = 100 * Z + 100 * z,x - r,; l (4.7} ( / ( log tr,(z+l) - log tr,x ) where I (x) is the retention index of component x, t'r,z1 t'r,z+l are the adjusted retention times of component x and of the alkanes just before and after the peak of component x, respectively.

Thermostability of the deactivated and coated columns was determined by platting the baseline drift against temperature. The crosslinking efficiency, CLE, was obtained by determining the capacity factor of tetradecane before and after rinsing the crosslinked columns with dichloromethane. The crosslinking efficiency is given by:

-71- (4.8}

where ki and k2 are the capacity factors before and after rinsing the column with the solvent respectively.

4.3 RESULTS AND DISCUSSIONS

4.3.1 Evaluation of the Deactivated capillaries Representative chromatograms of the precolumn and the precolumn connected to one of the deactivated columns are shown in Fig. 4.3 (for more detailed discussion see Chapter 2). These chromatograms show no significant differences in peak shape, which indicates the excellent deactivation obtained by the reagents and the method used in this study.

1 ·z:34

s b ' a 4 23 ~

1 5 7 1 6

8 7 9 8 10 9 1 j ( \ i l - '- min 0 5 10 b 16 Figure 4.3 Representative. chromatograms of the test mixture. (a) precolumn; (b) precolumn plus a deactivated column; temperature 100°C isothermal, peak identification: l=decane, 2=octanol, 3=DMP, 4=undecane, S=DMA, 6=dodecane, 7=decylamine, S=tridecane, 9=nicotine, lO=tetradecane.

-72- The thermostability of these deactivation layers is illustrated in Fiq. 4.4. It can be seen that the deactivated columns have a qood thermal stability due to the thin and densely crosslinked polymer layer formed on the column wall. In addition, the inertness of the deactivated columns was tested before and after the bleeding test. In these experiments no significant changes in peak shape, normalized peak areas and retention indices were observed (see Chapter

2) •

10

8

< .s. El s I! 1 c '6"" " 4 ii5" 2

2

0 100 200 300 400 500

Temperature (°C) Fiqure 4.4 Bleeding test of the deactivated columns, 1-PMHS, 2-PVMHS, 3- PDPVMHS.

4.3.2 Evaluation of the Columns coate4 with Non-polar Stationary Phases

A number of columns with different non-polar stationary phases has been prepared as described in the experimental section. A survey of the experimental conditions and the characteristic properties of these columns are outlined in Table 4.2.

-73- Table 4.2 survey of the experimental conditions and column characteristics.

Col. Stat. Reagent I.D. df curing C.E. CLE No. Phase (µm) (µm) Method a b 1 OV-1 c PMHS 50 0.23 therm. 93 93 97 2 OV-1 PMHS 250 0.25 therm. 89 47 10 3 OV-1 PMHS ATB 89 89 27 4 OV-1 c PMHS • 25 therm • 83 82 95 5 ov-1 c PMHS 220 0.58 ATB d 95 91 96 6 OV-101 PMHS 250 0.25 therm. 88 70 15 7 SE-54 PMHS 250 0.25 therm. 90 90 94 8 SE-54 PMHS 250 0.25 ATB 86 86 95 9 ov-1 c PVMHS 250 0.25 therm. 93 93 96 10 ov-1 PVMHS 250 0.25 therm. 86 71 21 11 ov-1 PVMHS 250 0.25 ATB 85 89 83 12 SE-30 PVMHS 250 0.25 94 78 24 13 SE-54 PVMHS 250 0.25 99 99 95 14 OV-101 PVMHS 250 0.25 85 85 85 15 OV-73 PVMHS 250 0.14 88 87 17 16 OV-73 PVMHS 250 0.14 88 88 17 OV-1 c PVMHS 23 0.12 therm. 89 89 95 C.E. = coating efficiency, CLE = crosslinking efficiency. a) coating efficiency before crosslinking treatment. b} coating efficiency aftar crosslinking treatment. c} vinyl modified. d) aged ATB reagent.

Some typical representative chromatograms of the coated columns are shown in Figure 4.5. It can be seen from the data in Tabla 4.2 and Figure 4.5 that highly efficient and very inert columns can be prepared by the method described in this study. In addition, high crosslinking efficiencies were obtained for the columns coated with vinyl modified stationary phases and crosslinked by thermal treatment without the

-74- addition of a catalyst. The column efficiency and the column activity were not significantly affected by this crosslinking procedure. This indicates that the reaction shown in equation ( 4. 4) reaches a high conversion without the addi tion of a catalyst.

p 5 b

234

.5 a

6 7 8

7 JO 9

a.~ ~ l r .1-LLL

L1 .__ 0 10 20 min o !O min

Fiqure 4.5 Representative chromatograllls of the test mixture on the coated columns, (a) column No. 4, FID, f.s.d. 2•10-12 A; (b) column No. 7, FID, f.s.d. 16•10-12 A (cf. Table 4.2); temperature 100°c isothermal, peak identification see Fiqure 4.3.

In the case of PMHS deactivated columns, the hydrogen groups left on the surface after deactivation play an important rele in chemically binding the stationary phase to the column wall. This role however, fainted for non-vinyl containing stationary phases. Even if ATB was used as a radical initiator these columns show very poer crosslinking efficiencies (cf. column No. 2, 3 and 6 in Table 4.2). For PVMHS deactivated columns, both hydrogen and vinyl groups play a role in the crosslinking

-75- and chemica! bonding process. Thermal treatment of the non­ vinyl containing phases did not yield a sufficiently high crosslinking efficiency (cf. column No. 10, 12 and 15). By using ATB or dicwnyl peroxide as the initiator, the cross­ linking efficiency was improved. In this case the main function of the vinyl groups in the deactivation layer is to facilitate crosslinking and chemical bonding of the stationary phases.

The advantage of immobilization of the stationary phases by thermal treatment is that the introduction of any kind of reagents needed to initiate the process is avoided. The introduction of additional reagents is a potential source of additional activity and thermal instability of the resulting columns. This is illustrated in the chromatograms shown in Figure 4.6, in which the column No. 5 was evaluated with the test mixture before crosslinking and after crosslinking with aged ATB. The activity of the column is increased, resulting in the disappearance of the aminodecane peak. Moreover, the bleeding of the column is also increased at elevated temperatures.

Also narrow-bore capillary columns which are suitable for fast GC analysis and for SFC, can be prepared using the procedures described earlier in this chapter (cf. Figure 4. Sa). Inertness of the low inner diameter columns is very important because of the limited sample capacity of the column.

-76- . a b

1 J

4

23 5

5

6

7 8 8 9

1•..,._,_._.I ...... ___.__. __ ___,[__[_ 1 1 L ~::::::::::::;:::=:::::::::==::;=:::::'.::=-- 0 20 40 60 0 20 40 60 min l!'igure 4.6 Representative chromatograms of the test mixture on column No. 5, (a) before crosslinking, and (b) after crosslinking with aged ATB (cf. Table 4.2); temperature 100°c isothermal, peak identification see Figure 4.3.

Another example is presented in Figure 4.7. This figure shows the separation of the test mixture on a 23 µm i.d. column. The injected amount was approximately 7 picogram per component. As can be seen in the figure, the column has a very high degree of inertness illustrating the potential of the methods described here for the preparation of narrow-bore columns. The thermostability of the coated columns is shown in Figure

4. S. From this figure it can be concluded that thermally stable columns can be obtainedwith low bleeding levels in the temperature ranges of interest.

-77- 4 3 5 6 2 7

0 5 min Fiqure 4.7 Representative chromatogram of the test mixture on a narrow bore capillary column (23µm i.d. * 1 m), column.temperature 100°C isothermal, about 7 pg of each component onto the column, FID, f.s.d. l*lo-12 A, peak identification see Figure 4.3.

9

8

7 1 6 2 <,!?, 3 !" 5 c: 'ti"' 4 iii" 3

2

0 - 100 200 300 400 500

Temperature (°C) Figure 4.8 Bleeding test for the coated columns, 1 = SE-54, 2 = ov-1, 3 = OV-17, 4 m OV-61.

-78- 4.3.3 Evaluation of the Columns coated with Medium Polar stationary Phases

A number of medium polar columns with different phenyl contente (Table 4.1) in the stationary phases was prepared. Corresponding experimental conditions and the performances of these columns are outlined in Table 4.3.

'l'able 4.3 Survey of the experimental conditions and column character. Col. Stat. Reagent df Curing C.E. CLE No. Phase (µ.m) Method a b 18 OV-73 PDPVMHS 0.11 therm. 86 76 16

19 OV-73 PDPVMHS 0.11 ATB 85 85 87 20 OV-73 PDPVMHS c 0.11 ATB 86 85 86 21 OV-7 PDPVMHS 0.18 therm. 78 61 13 22 OV-7 PDPVMHS 0.18 ATB 77 79 83 1 23 OV-61 PDPVMHS 0.11 therm. 77 70 9 24 OV-61 PDPVMHS 0.11 ATB 77 80 79

1 25 OV-11 PDPVMHS 0.19 therm. 81 68 9 26 OV-11 PDPVMHS 0.19 ATB 80 80 78 27 OV-17 PDPVMHS 0.26 therm. 83 48 9 28 OV-17 PDPVMHS 0.26 ATB 83 82 70 C.E. = coating efficiency, CLE = crosslinking efficiency. a} coating efficiency before crosslinking treatment. b) coating efficiency after crosslinking treatment. c} with Ph:Vi:H ratio= 21:3:76.

Representative chromatograms of some of these columns are presented in Figures 4.9-4.11. In agreement with the results reported in Chapter 2, it fellows from Fig. 4.9, that the effect of the retention behaviour of the deactivation layer on the overall performance of the final column cannot be

neglected. The relativa positions of 2,6-dimethylaniline (OMA) and n-dodecane and nicotine compared to n-tridecane have

-79- initiator, such as ATB or OCP, is needed to obtain a reasonable crosslinking efficiency.It can be seen that the crosslinking efficiency decreased with an increasing phenyl content in the stationary phases. This is most likely due to the ster ie effect of the phenyl groups which tend to hinder the crosslinking process [103, 151].

14

Fiqure 4.10 Representative chromatograms of different test mixtures on column No. 26, (a) the test mixture of Table 2.1, and (b) the Grob test mixture (cf. reL [145]), C10 = decane, C11 = undecane, 01 = octanol, al = nonanal, EH = 2-ethylhexanoic acid, P = 2,6-0MP, A = 2,6-0MA, E10' E11' E12 = methyl esters of decanoic acid, undecanoic acid, and dodecanoic acid respectively.

Oue to the electron donating properties of the geminal phenyl

groups and its chemical stability the diphenyl substituted polysiloxanes showed an increased thermal stability as was

expected. This is illustrated in Chapter 5.

-81- 1 4 6 2 8

10

3 5

9 L L o 6 12 min Figure 4.11 Representative chromatogram of the test mixture on column No. 28, temperature 90°C isothermal, peak identification see Figure 4.3.

4.4 CONCLUSIONS

Using the methods described in this chapter highly efficient and inert fused silica capillary columns with non-polar and medium polar stationary phases were obtained. C.E. > 80% are routinely obtained during this study. In addition these columns have a high thermal stability and solvent resistance. The highest thermostability was achieved by crosslinking diphenyl substituted polysiloxane phases.

For vinyl modified non-polar stationary phases, immobilization can be achieved by thermal treatment. This gives minimal

-82- effects on column characteristics, such as the activity and the retention behaviour. For non-vinyl containing stationary phases, the immobilization of the PMHS deactivated columns is less satisfactory both for thermal treatment and ATB initiation. For these stationary phases, immobilization of the PVMHS and POPVMHS deactivated columns is obtainable by the ATB initiation method.

The crosslinking efficiency of the phenyl containing phases decreases with an increasing phenyl content of the stationary phase. This is caused by the steric hindrance effect of the phenyl groups in the stationary phase.

Non-polar narrow bore capillary columns can be prepared using the same method. Again highly inert and efficient columns were obtained.

Aged ATB was observed to yield deteriorated performance of the im:mobilized columns. Careful regeneration of the ATB can avo id this problem.

Optimization of the deactivation procedure is required to obtain proper wettability of column surface by the stationary phases. The contribution of the deactivation layer to the retention mechanism of the coated column should be minimal to avo id the dual solute retention mechanism. Otherwise the selectivity of the stationary phase may be altered.

-83- 4.5 REFERENCES

1. M.L. Lee and B.W. Wright, J. Chromatoqr., 184 (1980) 235. 2. L.S. Ettre, Anal. Chem., 57 (1985) 1419A. 3. B. Xu and N.P.E. Vermeulen, J. Chromatogr., 445 (1988) 1. 4. F. Farré-Ruis, J. Henniker and G. Guiochon, Nature (London), 196 (1962) 63. 5. M. Gassiot-Matus, J.O. Pascual-Calveras and A. Serra­ Maciá, Chromatographia, 5 (1972) 328. 6. K.D. Bartle and M. Novotny, J. Chromatogr., 94(1974)35. 7. G. Alexander and G.A.F.M. Rutten, J. Chromatogr., 99 (1974) 81. 8. J.L. Marshall and D.A. Parker, J. Chromatogr., 122 (1976) 425. Sa. J.C. Marshall and M.W. Sanderson, Chromatoqraphia, 12 (1979) 782. 9. F. Riedo, M. czencz, o. Liardon and E.sz. Kováts, Helv. Chim. Acta, 61 (1978) 1912. 10. G. Korosi and E.sz. Kováts, Colloid Surf., 2 (1981) 315. 11. K.O. Bartle, B.W. Wright and M.L. Lee, Chromatographia, 14 (1981) 387. 12. J.A. Hubball, P.R. Dimauro, S.R. Smith and E.F. Barry, J. Chromatogr., 302 (1984) 341. 13. M.W. Ogden and H.M. McNair, J. Chromatogr., 354 {1986) 7. 14. W.A. Zisman, Ind. Eng. Chem., 55 (1963) 18; Advan. Chem. Ser., 43 (1964) 1. 15. J.J. Bikerman, Physical Surfaces, Academie Press, New York, 1970, p.73. 16. M.L. Lee, F.J. Yang and K.O. Bartle, Open Tubular Column Gas Chromatography: Theory and Practioe, Wiley, New York, 1984, p.62. 17. G. Dijkstra and J. De Goey, in O.H. Desty (Ed.) Gas Chromatography 1958, Academie Press, New York, 1958, p.56. 18. O.A. Parker and J.L. Marshall, Chromatoqraphia, 11 (1978) 526. 19. J. Roeraade, Chromatographia, 8 (1975) 511. 20. R.L. Levy, O.A. Murray, H.O. Gesser and F.W. Hougen, Anal. Chem., 40 (1968) 459. 21. J.P.J. van Dalen, Chromatographia, 5 (1972) 354. 2 2. M.L. McConnel 1 and M. Novotny, J. Chromatogr. , 112 (1975) 559. 23. L. Blomberg, Chromatographia, 8 (1975} 324. 24. L. Blomberg, J. Chromatogr., 138 (1977) 7. 25. G. Schomburg, H. Husmann and F. Weeke, J. Chromatogr., 99 (1974) 63. 26. G. Schomburg and H. Husmann, Chromatographia, 8 (1975) 517. 27. R. Kaiser, Gas Phase Chromatography, Vol. II, Butterworths, London, 1963, p. 45. 28. M. Novotny, L. Blomberg and K.O. Bartle, J. Chromatogr. Sci., 8 (1970) 390. 29. M. Novotny, K.O. Bartle and L. Blomberg, J. Chromatogr., 45 (1969) 469.

-84- 30. K. Tesarik and M. Neèasová, J. Chromatogr., 65 (1972) 39. 31. G. Guiochon, J. Chromatogr. sci, 9 (1971) 512. 32. M. Necasová and K. Tesarik, J, Chromatogr., 79 (1973) 15. 33. K.O. Bartle, Anal. Chem., 45 {1973) 1831. 34. M. Novotny and K.O. Bartle, J. Chromatogr., 93 (1974) 405. 35. K. Grob and G. Grob, Chromatographia, 4 (1971) 422. 36. J. Bouche and M. verzele, J. Gas Chromatogr., 6 (1968) 501. 37. G.A.F.M. Rutten and J.A. Luyten, J. Chromatogr., 74 (1972) 177. 38. G. Alexander, G. Garz6 and G. Pályi, J, Chromatogr., 91 (1974) 25. 39. K. Grob, J. High Resolut. Chromatogr. Chromatogr. Commun., 1 (1978) 93. 40. G.A.F.M. Rutten and J.A. Rijks, J. High Resolut. Chromatogr. Chromatogr. Commun., l (1978) 279. 41. M. Verzele and P. Sandra, J. High Resolut. Chromatogr. Chromatogr. commun., 2 (1979) 303. 42. B.L. Goodwin, J. Chromatogr., 172 (1979) 31. 43. P. Sandra and M. Verzele, Chromatographia, 11 (1978) 102. 44. M.A. Cueman and R.B. Hurley, Jr., J. High Resolut. Chromatogr. Chromatogr. commun., l {1978) 92. 45. M. Giabbai, M. Shoulto and W. Bertsch, J. High Resolut. Chromatogr. Chromatogr. Commun., l (1978) 277. 46. J.C. Thompson and N.G. Schautz, J, High Resolut. Chromatogr. Chromatogr. Commun., 3 {1980) 91. 47. G. Anders, D. Rodewald and Th. Welsch, J. High Resolut. Chromatogr. Chromatogr. Commun., 3 (1980) 298. 48. C.H. Lochmüller and J.D. Fisk, J. High Resolut. Chromatogr. Chromatogr. Commun., 4 (1981) 232. 49. T.L. Peters, T.J. Nestrick and L.L. Lamparski, J. High Resolut. Chromatogr. Chromatogr. Commun., 4 (1981) 588. 50. H. Husmann and G. Schomburg, J. High Resolut. Chromatogr. Chromatogr. commun., 3 (1980} 655. 51. F. Janssen and T. Kalidin, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 107. 52. F. Janssen and T. Kalidin, J. Chromatogr., 235 (1982) 323. 53. K. Grob and G. Grob, J. High Resolut. Chromatogr. Chromatogr. commun., 5 (1982) 119. 54. A.G. Calder, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 324. 55. K.R. Lin, L. Ghaoui and A. Zlatkis, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 571. 56. R.C. Kong and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 6 (1983) 319. 57. R.C. Kong and M.L. Lee, Chromatographia, 17 (1983) 451. 58. T. Wännman, L. Blomberg and s. Schmidt, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 32. 59. R.C. Kong, S.M. Fields, W.P. Jackson and M.L. Lee, J. Chromatogr., 289 (1984) 105. 60. S.M. Fields, R.C. Kong, M.L. Lee and P.A. Peaden, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 423.

-85- 61. F.I. onuska, J. Chromatogr., 289 (1984) 207. 62. B. Xu and N.P.E. Vermeulen, Chromatographia, 18 (1984) 520. 63. B. xu and N.P.E. Vermeulen, Chromatographia, 18 (1984) 642. 64. B. xu and N.P.E. Vermeulen, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 181. 65. K. Janak, v. Kahle, K. Tesarik and M. Horka, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 843. 66. P.R. Dluzneski and J.W. Jorgenson, in P. Sandra (Ed.), Proc. of 8th International Symposium on Capillary Chromatography, Riva del Garda, Italy, May 1987, Huethig Verlag, Heidelberg, 1987. 67 • K. Grob and G. Grob, J. High Resolut. Chromatogr. Chromatogr. commun., 5 (1982) 349. 68. B.W. Wright, P.A. Peaden and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 413. 69. L. Blomberg, J. Buijten, K. Markides and T. Wännman, J. Chromatogr., 279 (1983) 9. 70. L. Blomberg and K. Markides, J. High Resolut. Chromatogr. Chromatogr. Commun., 8 (1985) 632. 71. R.G. Sinclair, E.R. Hinnenkamp, K.A. Boni, O.A. Berry, W.H. Schuller and R.V. Lawrence, J. Chromatogr. sci., 9 (1971) 126. 72. K. Grob, Helv. Chim. Acta, 51 (1968) 729. 73. c. Madani, E.M. Chambaz, M. Rigaud, J. Durand and P. Chebroux, J. Chromatogr., 126 (1976) 161. 74. M. Rigaud, P. Chebroux, J. Durand, J. Maclouf and c. Madani, Tetrahedron Lett., 44 (1976) 3935. 75. c. Madani, E.M. Chambaz, M. Rigaud, J. Durand, P. Chebroux and J .c. Breton, Chromatographia, 10 (1977) 466. 76. c. Madani and E.M. Chambaz, Chromatographia, 11 (1978) 725. 77. c. Madani and E.M. Chambaz, J. Amer. Oil. Chem. soc., 58 (1981) 63. 78. L. Blomberg, J. Buijten, J. Gawdick and T. Wännman, Chromatographia, 11 (1978) 521. 79. L. Blomberg and T. Wännman, J. Chromatogr., 168 {1979) 81. 80. L. Blomberg and T. Wännman, J. Chromatogr., 186 (1979) 159. 81. L. Blomberg, K. Markides and T. Wännman, J. Chromatogr., 203 {l.981) 217. 82. L. Blomberg, J. Buijten, K. Markides and T. Wännman, J. Chromatogr., 208 (1981) 231. 83. M. Verzele, F. David, M. van Roelenbosch, G. Diricks and P. Sandra, J. Chromatogr., 279 {1983) 99. 84. E. Geeraert and P. Sandra, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 431. 85. F. David, P. Sandra and G. Diricks, J. High Resolut. Chromatogr. Chromatogr. Commun., 11 (1988) 256. 86. S.R. Lipsky and W.J. McMurray, J. Chromatogr., 279 (1983) 59. 87. S.R. Lipsky and W.J. McMurray, J. Chromatogr., 289 (1984) 129. 88. S.R. Lipsky and W.J. McMurray, J. Chromatogr. Lib., 32 (1985) 257.

-86- 89. S.R. Lipsky and M.L. Duffy, J. High Resolut. Chromatogr. Chromatogr. Conunun., 9 (1986) 376. 90. K. Grob and G. Grob, J. Chromatogr., 347 (1986) 351. 91. w. Blum, J. High Resolut. Chromatogr. Chromatogr. co:m:mun., 8 (1985) 718. 92. w. Blum, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 350. 93. W. Blum and L. Damasceno, J. High Resolut. Chromatogr. Chromatogr. Conunun., 10 (1987) 472. 94. w. Blum, w.J. Richteer and E. Eglinton, J. High Resolut. Chromatogr. Chromatogr. Co:mmun., 11 (1988} 148. 95. W. Blum and E. Eglinton, J. High Resolut. Chromatogr., 12 (1989) 290. 96. A. Bemgard, L. Blo:mberg, M. Lymann, s. Claude and R. Tabacchi, J, High Resolut. Chromatogr. Chromatogr. Commun., 10 (1987) 302. 97. P. Schmid and M.D. Milller, J. High Resolut. Chromatogr. Chromatogr. Commun., 10 (1987) 548. 98. A. Bemgard, L. Blo:mberg, M. Lymann, s. claude and R. Tabacchi, J. High Resolut. Chromatogr. Chromatogr. conunun., 11 (1988) 881. 99. K.A. Andrianov, Metalorganic Polymers, Interscience Publishers, New York, 1965, p.48. 100. K. Grob, G. Grob and K. Grob, Jr., J. Chromatogr., 211 (1981) 243. 101. L. Blomberg, J. Buijten, K. Markides and T. Wännman, J. Chromatogr., 239 (1982) 51. 102. s.R. Lipsky and W.J. McMurray, J. Chromatogr., 239 (1982) 61. 103. B.W. Wright, P.A. Peaden, M.L. Lee and T.J. Stark, J. Chromatogr., 248 (1982) 17. 104. L. Blomberg, K. Markides, J. Buijten and T. Wännman, J. High Resolut. Chromatogr. Chromatogr. Commun., 4 (1981) 578. 105. J. Buijten, L. Blomberg, K. Markides and T. Wännman, Chromatographia, 16 (1982) 183. 106. J. Buijten, L. Blomberg, s. Hoffmann, K. Markides and T. Wännman, J. Chromatogr., 289 (1984) 143. 107. K. Markides, L. Blo:mberg, J. Buijten and T. Wännman, J. Chromatogr., 254 (1983) 53. 108. K. Markides, L. Blomberg, J. Buijten and T. Wännman, J. Chromatogr., 267 (1983) 29. 109. K. Markides, L. Blomberg, s. Hoffmann, J. Buijten and T. Wännman, J. Chromatogr., 302 (1984) 319. 110. P. Sandra, G. Redant, E. Schacht and M. Verzele, J. High Resolution Chromatogr. and Chromatogr. commun., 4 (1981) 411. 111. K. Grob and G. Grob, J. Chromatogr., 213 (1981) 211. 112. K. Grob and G. Grob, J. High Resolut. Chromatogr. Chromatogr. commun., 4 (1981) 491. 113. K. Grob and G. Grob, J. High Resolut. Chromatogr. Chromatogr. commun., 5 (1982) 13. 114. P.A. Peaden, B.W. Wright and M.L. Lee, Chromatographia, 15 (1982) 335. 115. B.E. Richter, J.C. Kuei, J.J. Shelton, L.W. castle, J.S. Bradshaw and M.L. Lee, J. Chromatogr., 279 (1983) 21. 116. B.E. Richter, J.C. Kuei, L.W. Castle, B.A. Jones, J.S. Bradshaw and M.L. Lee, Chromatographia, 17 (1983) 570.

-87- 117. B.E. Richter, J.C. Kuei, N.L. Park, S.J. Crowley, J.S. Bradshaw and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Conunun., 6 {1983) 371. 118. J.C. Kuei, J.J. Shelton, L.W. Castle, R.C. Kong, B.E. Richter, J.S. Bradshaw and M.L. Lee, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 13. 119. M.L. Lee, J.C. Kuei, N.W. Adams, B.J. Tarbet, M. Nishioka, B.A. Jones and J.S. Bradshaw, J. Chromatogr., 302 (1984) 303. 120. J.C. Kuei, B.J. Tarbet, W.P. Jackson, J.S. Bradshaw, K.E. Markides and M.L. Lee, Chromatographia, 20 (1985) 25. 121. E.D Lee, K.E. Markides, R. Bolick and M.L. Lee, Anal. Chem., 58 (1986) 740. 122. K. Grob, Making and Manipulating Capillary Columns for Gas Chromatography, Hilthig, Heidelberg, 1986. 123. R.F. Arrendale and R.M. Martin, J. High Resolut. Chromatogr. Chromatogr. Commun., 11 (1988) 157. 124. I. Benecke and G. Schomburg, J. High Resolut. Chromatogr. Chromatogr. Conunun., 8 (1985) 191. 125. G. Schomburg, H. Husmann, w.s. Ruthe and M. Herraiz, Chromatographia, 15 (1982) 599. 126. w. Bertsch, v. Pretorius, M. Pearce, J.C. Thompson and N.G. Schnautz, J. High Resolut. Chromatogr. Chromatogr. Commun., 5 (1982) 432. 127. J.A. Hubball, P. Dimauro, E.F. Barry and G.E. Chabot, J. High Resolut. Chromatogr. Chromatogr. Commun., 6 (1983) 241. 128. Gy. Vigh and O. Etler, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 620. 129. o. Etler and Gy. Vigh, J. High Resolut. Chromatogr. Chromatogr. Commun., 7 (1984) 700. 130. V. Tatar, M. Popl, M. Matucha and M. Pesek, J. Chromatogr., 328 (1985) 337. 131. K. Lakszner and L. szepesy, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 441. 132. J, Buijten, L. Blomberg, s. Hoffmann, K. Markides and T. Wänrunan, J. Chromatogr., 283 {1984) 341. 133. L. Blomberg, J. Microcol. Sep., 2 (1990) 62. 134. E.M. Barrall II, R. Hawkins, A.A. Fukushima and J.F. Johnson, J. Polym. Sci., Polym. Symp., 71 (1984) 189. 135. A. Aerts, J. Rijks, A. Bemgard and L. Blomberg, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 49. 136. G. Schomburg, in P. Sandra (Ed.). Proc. of the 8th International Symp. on Capillary Chromatography, Riva del Garda, Italy, May 1987, Hilthig, Heidelberg, 1987, P.143. 137. F. David, P.Sandra and G. Diricks, in P. Sandra (Ed.). Proc. of the 8th International Symp. on Capillary Chromatography, Riva del Garda, Italy, May 1987, Hüthig, Heidelberg, 1987, P.264. 138. B.A. Jones, T.J. Shaw and E.S. Francis, Pittsburgh Conference, 1989, Abstract No. 1319. 139. T.J. Shaw and B.A. Jones, in P. Sandra (Ed.), Proc. of llth International Symposium on capillary Chromatography, May 1990, Monterey, CA, USA, Hilthig, 1990. 140. w. Röder, F.-J. Ruffing, G. Schomburg, and W.H. Pirkle,

-88- J. High. Resolut. Chromatogr. Chromatogr. co:mmun., 10 (1987) 665. 141. F.-J. Ruffing, J.A. Lux, W. Röder and G. Schomburg, Chromatographia, 26 (1988) 19. 142. Q. wu, C.A. cramere and J.A. Rijke, in P. Sandra (Ed.), Proc. of lOth International Symposium on capillary Chromatography, Riva del Garda, Italy, May 1989, Hüthig, 1989, p. 188. 143. Q. Wu, M. Hetem, C.A. cramere and J.A. Rijke, J. High Resolut. Chromatogr., 13 (1990) 811. 144. G. Shomburg, H. Husmann and F. Weeke, Chromatographia, 10 (1977) 580. 145. K. Grob, G. Grob and K. Greb, Jr., J. Chromatogr., 156 {1978) 1. 146. M. Novotny and A. Zlatkis, Chromatogr. Rev., 14 {1971) 1. 147. M.J.E. Golay, in D.H. Desty (Ed.), Gas Chromatography 1958, Academie Press, New York, 1958, p. 36. 148. J.C. Giddings, Anal. Chem., 36 (1964) 741. 149. E.sz. Kováts, Helv. Chim. Acta, 41 (1958) 1915. 150 E.sz. Kováts, in J.C. Giddings and R.A. Keller (Eds.), Adv. Chromatography, Dekker, New York, N.Y., 1 ( 1965) p. 229. 151. W. Noll, Chemistry and Technology of Silicones, Academie Press, New York, 1968. ·

-89- CBAP'l'ER S

APPLICATIONS IN GAS CHROMATOGRAPHY

The applicability of the prepared non-polar and medium polar fused silica capillary columns was illustrated by different examples. High speed GC separations were performed on narrow­ bore capillary columns with inner diameters down to 23 µm. Columns with a high thermal stability were prepared for the analysis of less volatile and high molecular weight compounds. The separation of some underivatized steroids, drugs, priority pollutants and methyl esters of amino acids is also shown. The performance of the capillaries with respect to inertness, thermostability and efficiency in these applications is evaluated.

S.1 INTRODUC'l'IO:N

The knowledge of column technology has grown dramatically [1, 2). Nowadays, high resolution GC is one of the most widely employed separation techniques in analytica! organic chemistry.

capillary columns allow the complete separation and specific analysis of many complex samples which were difficult to analyze using packed columns. capillary GC has been applied to a wide variety of analytica! problems because of its repeatability, reliability in quantitative and qualitative information, and shortened analysis time for a given plate number.

The main application areas of high resolution gas

chromatography include chemical, petrochemical (3-5] t pharmaceutical industries, life science (6-10), food (11) and consumer products, and environmental analysis [12-15). The

-90- demands of'these applications with regard to analysis time, trace analysis and the ability to separate less volatile compounds promoted column technology. Narrow bore capillary columns for high speed analysis, high temperature columns for high boiling point components and very well deactivated columns for trace analysis were the results of these demands.

Theory [16-20] shows that the reduction of the inner diameter of the capillary column is one of the most practical and convenient ways to reduce the analysis time. High plate numbers per unit of length ( inversely proportional to the column diameter, see equation 5.1, [17, 18]), or per unit of time (inversely proportional to the column diameter squared,

see equation 5. 3, [ 20 J) were obtained with small diameter columns. For thin film columns the contribution of the stationary phase for the plate height can be neglected and follows:

2 (5 .1)

where 2 112 F(k) = ( 1 + 6 * k + 11 * k ) (5.2) 2 3 * { 1 + k )

and d0 is the column inner diameter, k is the capacity factor of the component.

2 dN) 48 * Dif ( 1 + k ) (5.3) ( dt U,O.Pt = --d-~--- *

where DG 0 is the diffusion coefficient of the solute in the mobile phase at the column outlet pressure. In case of large values of P, P being the inlet to outlet

-91- column pressure ratio, usually encountered in narrow-bore

2 columns, (dN/dt)u,opt is no longer proportional to 1/d 0 but to 1/dc (19].

Apart from high pressure drops other problems, such as sample capacity and speed of sample introduction and detector compatibility may be encountered in the use of narrow bore capillary columns. Promising results, however, were obtained by different authors [16-29].

The advantage of high temperature columns is that less volatile or high bolling compounds can be analyzed on this type of columns. High molecular weight samples such as the oligomeric series of polyethylene glycol, polywax and triton X-100 and simulated distillation products which up to now are normally analyzed by supercritical fluid chromatography (SFC) are now also amenable to high temperature GC (30-32]. Compared with the results obtained by SFC, high temperature GC offers superior speed of analysis, increased column efficiency and improved resolution of the component bands. (30-44].

On the basis of the knowledge of column technology described in the previous Chapters, different fused silica capillary columns were prepared. Their applicability to the selected samples is illustrated in this Chapter. Examples of high speed GC separations, high temperature GC analysis and the separation of methyl esters of amino acids are given. The performance of the capillaries with respect to inertness, thermostability and efficiency is evaluated.

-92- 5. 2 EXPERJ:HENTAL s.2.1 Apparatus

The high temperature separations were carried out on a Varian 3400 gas chromatography (Varian, Walnut creek, CA, USA), or on a Fractovap 4160 gas chromatography (Carlo Erba, Milan, Italy) equipped with a split injector and a flame ionization detector (FID). The hot split injection technique was applied throughout the experiments. The chromatograms were recorded with an SP 4000 integration system (Spectra Physics, Santa Clara, CA, USA), or with a strip chart recorder (Kipp & Zonen, Delft, The Netherlands).

The experiments with narrow bore capillary columns were carried out on a Fractovap 2900 GC (Carlo Erba, Milan, Italy), equipped with a hot split injector and a FID. According to the specifications provided by the manufacturer, the time constant of the electrometer amplifier was 50 ms for the 1 V integrator output and about 150 ms for the 10 mV recorder output. A Tescom high pressure regulator (Model 44-1100, Tescom Ine., Minnesota, USA) proceeded by a dust filter was used to control the helium flow rate. The split injector was modified to allow a high split ratio for delivery of a small amount of sample onto the column.

The other experiments were performed on a Packard 430 gas chromatograph (Packard, Delft, The Netherlands) or on the Varian 3400 GC equipped with a split injector and a FID.

-93- s.2.2 Materials and Chemicals

All the reagents used for the deactivation of the fused silica capillary columns were detailed in Chapter 2. The empty columns used in this study were purchased from Siemens (Siemens, Germany) and can be used isothermally at 400°C. The procedures for the preparation of different inner diameter capillaries were summarized in Chapter 4 of this thesis. All the solvents were PA grade and purchased from E. Merck AG (Darmstadt, Germany).

Below follows a short description of the samples used. The chemica! structures of the polymer additives, a gift from DSM (Geleen, The Netherlands), are given in Table 5.1. Pólyvinylmethylhydrosiloxane (PVMHS) was dissolved in pentane (10%, v/v). The fatty acid methyl esters (FAME) mixture (No. 4-7080) and the volatile organic compounds mixture (VOC mix #3, No. 4- 8779) were obtained from Supelco (Bellefonte, PA, USA). The standards of the sedative hypnotic barbiturates were a gift from Professor Zlatkis (Houston, Texas, USA). The methyl or ethyl esters of some amino acids and fatty acids were obtained by derivatization from the corresponding acids according to the procedure described in ref. [45]. The polyethylene glycol sample (PEG 400) was purchased from E. Merck AG (Darmstadt, Germany).

-94- Ta:ble 5.1 Chemical structures of the polymer additives

Trade name Chemica! name or •tr11ct11re MW

Oleamlde 0 271 H, )·<(CHt 7( OK NH1

Cyasorb 531 326 Ó!-Ooc,H"

H ·0.c-(C!l,.1-o0 H Tlnuvln 770 () v Q 480

C(CH,)1

lrgafH 168 P·(OO·C(CH,>~. 647

ID•

lrganox 1076 .,,.,.,,olc•,.c•Q-o• 531 . ..

Tlnuvln 144 ,~.... 685 C,H, COO.Q·CH,).

0 0 Il Il OSTOP H,,C.j>-CClt,ctl,SC~,11,,, 683

1178 lrganox 1010 M·l.

5.3 RESULTS AND DISCUSSIONS 5.3.1 High Speed Analysis on Narrow Bore Capillaries

For narrow bore capillary columns (de< 100 µm), high inlet pressures are normally needed for obtaining the optimum velocity of the carrier gas. Special injection techniques (such as cold trap injection) are required for very fast analysis [19, 28, 29) to avoid an excessive contribution of the injection to the overall band width. Under optima! conditions, hot split injection can be used for narrow bore capillary columns with inner diameters around 10 µm. High

-95- column efficiencies were obtained using the hot split injection technique (cf. Chapter 4).

..."'

0 2 4 min

Figure s.1 Representative GC chromatogram of volatile organic compounds (VOC mix #3, Supelco) on a narrow bare column, 2 m * 23 µm, df = 0.12 µm OV-1, T0 from 60°C to 200°C at 20°C/min, inlet pressure, Pi = 15 Bar; peaks in elution order: 1,1-dicploro­ propylene, 1,2-dichloroethane, trichloroethylene, 1,2-di­ chloropropane, cis-1,3-dichloropropylene, trans-1,3-dichloro­ propylene, 1,1,2-trichloro-ethane, 1,3-dichloropropane, 1,2- dibromoethane, 1,1,1,2-tetra-chloroethane, 1,1,2,2-tetra­ chloroethane, 1,2,3-trichloropropane, 1,2-dibromo-3-chloro­ propane, hexachlorobutadiene.

Figure 5.1 presente the chromatogram of a mixture of volatile organic compounds which was separated on a 23 µm column. On this column 14 components can be separated within 3.5 min, while it takes 35 min on a 60 m * 0.75 mm i.d. column. Moreover, the GC oven could not reach the desired temperature programming rate of 20°C/min. Otherwise, the analysis time would have been even shorter than 3 min. A representative chromatogram of the separation of a fatty acids methyl ester mixture on the same column is depicted in Figure 5.2. The total analysis time is less than 7 min. The same mixture with compounds ranging from 2-hydroxydecanoic acid to eicosanoic acid and having a wide bolling point range required a separation time of more than 30 min on a 30 m * 0.25 mm i.d. column.

-96- 1 2 ",,;" ::J ..... ; ... 4 n 5

3 ""0) "1() i:"·f. " ...... •• M

0 7 min Fiqure 5.2 Typical GC chromatogram of FAME mixture on the same column as in Figure 5.1, T0 at 200°C isothermal, peaks identification: 1-methyl dodecanoate, 2-methyl tetradecanoate, 3-methyl 3- hydroxytetradecanoate, 4-methyl cis-9-hexadecanoate, 5-methyl trans-9-octadecanoate, 6-methyl nonadecanoate, 7-methyl eicosanoate. For ether conditions see Fig. 5.1.

3 Il 13 19 4 10 22 9

1. 24 2 'l l l 0 10 min Fiqure 5.3 Representative GC chromatogram of methyl esters mixture of 24 benzoic and related acids, column, 3 m * 50 µm, df = 0.25 µm ov-1, Tç from 100°c to 290°C at 20°C/min, Pi = s Bar, peaks identification: 1) benzoic acid, 2) p-OMe benzoic acid, 3) cinnamic acid, 4), 9), 10) o-, m-, p-OH phenyl acetic acid, respectively, 5), 7), 8) o-, m-, p-OH benzoic acid respectively, 6)mandelic acid, 11) 3-0Me-4-0H-phenyl acetic acid, 12) 3-0Me-4-0H-benzoic acid, 13) m-OH-cinnamic acid, 17) 2, 5-diOH-phenyl acetic acid, 19) 3-0Me-4-0H-cinnamic acid, 22) sinapic acid, 24) 3,4-diOH-mandelic acid.

-97- Figure 5. 3 represents another example of the high speed separation of a methyl ester mixture containing 24 benzoic and related acids. Except the isomers of m- and p-hydroxy-benzoic acids, all other isomers were baseline separated within 8 min. The problem again was the heating rate of the oven, which delayed the elution of the later eluting components.

Typical chromatograms of the separation of a super gasoline and a natural gas condensate are shown in the Figures 5.4 and 5. 5, respectively. It can be seen that these narrow bore capillary columns are quite suitable for this kind of volatile and complex hydrocarbon samples due to their relatively small phase ratio {J.

0 7 min

Fiqure 5.4 Typical GC chromatogram of super gasoline on the same column as in Figure 5 .1, Pi = 15 Bar, Te from 40°C to 220°c at 20°C/min, no peak identification was performed.

-98- O 5 JO 15 min P'igure s.s Typical GC chromatogram of a natural gas condensate. Column, 8 m * 50 µm, df = 0.25 µm OV-1, T0 from 40°C to 100°C at 8°C/min, then to 220°c at 15°C/min, no peak identification was performed. s.3.2 Hiqh ~emperature columns

Fdr the elution of less volatile or high boiling compounds, the GC oven has to be programmed to very high temperatures. This dose not cause problems when glass capillary columns are used. The outside coating of fused silica capillary columns, however, can be damaged at temperatures above 370°C. Generally, the capillary column becomes brittle after prolonged use at high temperatures. Specially manufactured polyimide coatings [39], or aluminum clad capillary columns po, 31] have to be used in high temperature applications. Polymer additives (antioxidants, uv stabilizers), which are used in food packing materials or in pipes for water distribution systems can represent a source of contamination through the migration of these substances into the food or

-99- water. Some of them are known to be toxic agents (46, 47]. Normally these additives are analyzed by SFC or HPLC [48, 49]. Figure 5.6 shows a chromatogram of a polymer additives mixture on an OV-61 column. The last peak, Irganox 1010, having a molecular weight of 1178 daltons (cf. Table 5.1), eluted from the column at 420°C. Similar separations were achieved by Blum and Damasceno [34] using an OH-terminated diphenyl phase. Representative chromatograms of the separation of a mixture of phenyl substituted cyclosiloxanes and of a triglyceride sample on the same column as used above are shown in Figures 5.7 and 5.8, respectively. In the first example, octa- and decaphenyl substituted cyclosiloxane (with a molecular weight of 792 and 990 daltons, respectively) was easily separated and quantitative results could be obtained. Similarly, quanti­ tative analysis of triglyceride samples can be performed on a routine basis with this type of columns.

2 3 4 s 7 1 Oleamide 2 cyasorb 3 Tinuvin 770 4 Irqafos 168 S Irqanox 1076 6 Tinuvin 144 7 DSTDP ' S Irqanox 1010 j 8

0 10 20 m.l.n

Fiqure 5.6 Representative GC chromatogram of polymer additives, column, 5 m * 0.25 mm i.d., df = 0.07 µm OV-61, Te from 150°C to 300°C at 20°C/min, then to 420°C at 10°C/min, (cf. Table 5.1).

-100- 2

Fig. 5.7 Fig. 5.8

:So

O 5 J 0 min 0 10 20 min i'igure 5.7 Representative GC chromatogram of a phenyl cyclosiloxane mixture on the same column as in Figure 5.6, Te from 180°C to 420°C at 15°C/min, peaks identification: 1 = octaphenylcyclo­ tetrasiloxane, 2 = decaphenylcyclotetrasiloxane. Figure S.B Representative GC chromatogram of a triglycerides sample on the same column as in Figure 5.6, Te from 40°C to 240°C at 20°C/min, then to 380°C at 10°C/min. Tso represents a triglyceride with a total carbon number of 50.

Figure 5. 9 presents the GC chromatogram of a polyethylene glycol sample (PEG 400) with an average molecular weight of 400 daltons on the same column as used above. PEG samples can also be analyzed by SFC, but the solubility of glycols in pure carbon dioxide is limited. The addition of polar modifiers such as methanol to the mobile phase is needed in order to improve the separation of PEG samples. Some other exa:mples are shown in Figures 5. 10 and 5 .11, respectively. Figure 5. 10 is the chromatogram of a plant

extract containing hydrocarbons up to c50 • The separation of the same sample with SFC takes about 40 min (19).

-101- S 6 Fig 5.9 Fig. 5.10 i

7 c,_2 c 4 " I 46

8

0 1 0 20 minO 10 20 min

Figure 5.9 Representative GC chromatogram of the polyethylene glycol (PEG 400) sample on the same column as in Figure 5.6, Te from 100°c to 370°C at 15°C/min, the numbers above the peak indicate the repeat unit in the polymer chain. Fiqure s.10 Representative Ge chromatogram of a plant extraction product on the same column as in Figure 5.6, Te from 150°C to 350°C at 20°C/min, hot split injection.

0 40 min Fiqure s.11 Representative GC chromatogram of a paraffin sample on a 15 m * 0.25 mm i.d. column, df = 0.18 µm ov-1, Te from so 0 c to 380°C at 8°C/min, hot split injection.

-102- Another example of the high separation power that can be obtained in GC is shown in Figure s.11. This figure shows the separation of a paraffin sample on a 15 m * 0.25 mm i.d. column.

Capillary GC can also be applied to analyze the deactivation reagents used for the preparation of GC columns. Both the cyclic and the linear polysiloxane molecules show a high degree of dispersion with oligomers starting from the tetramer up to some 27 siloxane units. Figure 5.12 shows the separation of Polyvinylmethylhydrosiloxane (PVMHS) copolymer on an ov-1 column. Solid state 29si NMR spectroscopy studies of the deactivation process with this reagent using fumed silica particles as the model substrate are described in Chapter 3. Figure 5. 13 represents a chromatogram of the extraction product of the PVMHS silylated Cab-O-Sil samples. Comparing Figures 5.12 and 5.13 it can be concluded that the short chain siloxanes can be extracted from the surface of Cab-O-Sil particles after silylation. The crosslinked and longer chains are covalently bonded to the surface and can not be extracted by dichloromethane. The solid state 29si NMR results of this silylated Cab-O-Sil sample indicated that a crosslinked thin layer is formed on the surface of the Cab-O-Sil particles. (cf. Chapter 3).

-103- 0 10 20 30 min Fiqure s.12 Representative GC chromatogram of polyvinylmethylhydrosiloxane on the same colwnn as in Figure 5.11, Te from 80°C to 380°C at 10°C/min, hot split injection.

0 12 24 min Figure 5.13 Representative GC chromatogram of the extraction product of the silylated Cab-O-Sil sample on the same column as in Figure 5.11, Te from 40°C to 200°C at 20°C/min, then to 340°C at 10°C/min, hot split injection.

-104- S.3.3 Analysis of Drugs, Pollutants, Amino Aci4s, an4 On4erivatized steroids.

High resolution GC plays an essential role in the analysis of pharmaceutical products and drugs. It is used in quality control, analysis of new products and in monitoring metabolites in biological fluids [50]. An example of the analysis of sedative hypnotic barbiturates mixture is shown in Figure 5.14. With a short, narrow bore column, this mixture could be analyzed in our laboratory within 40 sec [29]. The symmetrical shape of all components indicates that the column was very well deactivated and quantitative analysis can be achieved.

34567 8 9 10

2

0 8 16 min Figure S.14 Representative GC chromatogram of sedative hypnotic barbiturates mixture. GC conditions: column, 20 m * o.25 mm i.d., df = 0.25 µm ov-1, Te from 120°C to 300°C at 8°C/min; peaks in elution order: barbital, aprobarbital, amobarbital, pentobarbital, secobarbital, hexobarital, mephobarbital, phenobarbital, butabarbital, butabital.

Phenols are widely regarded as important priority pollutants and are known to be difficult to analyze. They exhibit high affinities to the activa sites on the column surface. It is even more difficult to analyze halogen and nitro-substituted phenols. Total or partial loss of the components and badly

-105- tailing peaks can often be observed. Figure 5.15 presents the GC chromatogram of a chlorophenols mixture analyzed on a 40 m * 0.25 mm i.d. ov-1 column. The 40 m column consisted of two 20 m columns connected by a glass press-fit connector (Unimetrics Corp., CA, USA). Apparently the separation efficiencies and peak shapes were not affected by the connection.

0 10 20 30 min Figure s.1s Representative chromatogram of a chlorophenols mixture, GC conditions: 40 m * 0.25 mm i.d. column, ov-1, df = 0.25 µm, Te from 80°C to 200°C at 4°C/min, peaks in elution order: 1- chloro-, dichloro-, trichloro-, tetrachloro-, pentachloro­ substituted phenols.

The separation of a chlorobenzene mixture is given in Figure 5.16. Again it shows perfect inertness to these components and almost baseline resolution for all critica! pairs. The high separation power of the capillary columns makes them suitable for the analysis of ester mixtures of fatty acids and

amino acids. Figure 5.17 shows a chromatogram of the c7-c9 fatty acids derivatized with 2-chloroethyl chloroformate, the

-106- derivatization reaction is not complete therefore the sample still contains free fatty acids. !t is surprising to see that even the free fatty acids elute from the column with quite good shapes. Normally, leading peaks were observed for free fatty acids on non-polar columns.

1 -2

Fig. 5. 1 6 Fig. 5. 1 7

4 3 - - 6 M 5 ,.. 4

•l 3 s 1 z "..,. 6 in •l 1 -··. " ~ ~-) ))

"·.;,; M ". L~ - ~:i,.; ~- 0 s 10 :n1n 0 a min Fiqure s.u Representative chromatogram of chlorobenzene mixture, column, 20 m * 0.25 mm i.d., df = 0.25 µm ov-1, Te from so 0 c to 200°c at 4°C/min, peaks in elution order: l=chloro-, 2=dichloro-, 3=trichloro-, 4=tetrachloro-, 5=pentachloro-, 6=hexachloro­ substituted benzene. Fiqure 5.17 Representative chromatogram of chloroethylation reaction product of c 7-c9 free fatty acids, column, 19 m * 0.25 mm i.d., df = 0.25 µm ov-1, Te from 70°C to 200°C at 10°C/min, peaks identification: 1-3 and 4-6, c7-c9 free fatty acids and corresponding chlqroethyl esters respectively. \

The chromatogram of another methyl ester mixture of fatty acids on an OV-17 column is presented in Figure 5. 18. The isomers containing more than two double honds are separated from the saturated and one double bond isomers.

-107- 6

2 3

4 5 7 9 12 1011 1314 J 0 8 16 min Fiqure 5.18 Representative chromatogram of the methyl esters of fatty acids, column, 20 m * 0.32 mm i.d., df = 0.26 µm OV-17, Te from l00°C to 2so 0 c at 10°C/min, peaks in elution order: C12:01 C14:01 C15:01 C16:0 + C16:l• C11:01 C1s:o + C1s:11 C1s:21 C1s:J1 C19:01 C20:01 C21:01 C22:0• C23:Q1 C24:0· 3 56 Fig. 5.19 Fig. 5.20 1 7 l l

2 7 4

s 10 1 2 4 9

9 6

J 8 5 lO

,\_ \._ - ! ..._..______j\ t ' - O 10 O 10 min Figure 5.19 Representative chromatogram of the separation of the ethyl esters of 10 amino acids, column: 4 m * 0.32 mm i.d., df = 0.26 µm OV-17, Te from 80°C to 270°C at 20°C/min, peaks in order: alanine, leucine, serine, proline, cysteine, hydroxyproline, aspartic acid, hysine, arginine. Figure 5.20 Representative chromatogram of the separation of underivatized steroids, column, 20 m * 0.25 mm i.d., df = 0.25 µm ov-1, Te at 260°C isothermal, peaks in elution order: C2~· C241 ethylestrenol, c26 , androsterone, methandriol, dianabol, pregnanediol, c28 , estriol, C30·

-108- The ethyl esters of 10 amino acids were chromatographed on an OV-17 column (see Figure 5.19). Steroids are often derivatized prior to analysis in order to increase their volatility. Figure 5.20 shows, however, the separation of an underivatized steroids mixture. This separation was performed isothermally at 260°C. It can be seen from Figure 5.20 that the column has a high efficiency and inertness. The same column also shows little interaction with oxygen containing compounds at low temperatures (see Figure 5.21).

l 2

3

9 4 7 10 14 6 12 5 8

11 13

0 10 min

Figure s.21 Typical GC chromatogram of a test mixture on the same column as in Figure 5.20, Te from 40°C to 160°C at 4°C/min, peaks in elution order: 1-butanol, toluene, c8 , chlorobenzene, 3- heptanone, c9 , p-chlorotoluene, 1-heptanol, trimethylbenzene, o-dichlorobenzene, 2-nonanone, c11, 1-nonanol, benzylaceton.

The chromatograms shown in Figure 5. 22 are from hexane extracts of oranges and of opium poppy straw respectively. No

-109- attempt was made to identify the components present in these two samples. The components in the poppy straw extract are most likely steroids and alkaloids.

i 1 a b 1 l 1 1 i

i 1 1 1

_l 1ll .11 L 11 1 1 t •. l. ! " 0 10 20 min 0 5 min

Fiqure s.22 Typical GC chromatograms of the hexane extraction products of (a) oranges, column, 20 m * 0.25 mm i.d., df = 0.58 µm OV-1, Te from 40°C to 250°C at 10°C/min, splitless injection, 1µ1; (b) poppy straw, column, 5 m * 0.25 mm i.d., df = 0.14 µm ov- 73, Te from l80°C to 300°C at l0°C/min, split injection.

5.4 CONCLUSIONS

The narrow-bore capillary columns prepared according to the methods described in Chapter 4 are very well suited for high speed analysis of different samples. Besides the inertness, a high separation efficiency was obtained with the small bore columns and the time required to obtain a given separation was reduced compared with the analysis time required using normal bore columns.

The normal bore columns prepared in this study have a high

-110- thermal stability and inertness. Less volatile and high boiling compounds, such as polymer additives, glycerides and polyethylene glycol 400 can be analyzed using these columns. Compared with the results obtained by SFC, shorter analysis time, higher column efficiencies for the separation of this type of compounds can be obtained by high temperature columns.

The successful separation of a number of selected samples, such as sedative hypnotic barbiturates, underivatized steroids and chlorophenols demonstrates the inertness of the columns prepared here. They are also suitable for the analysis of ester mixtures of free fatty acids, amine acids and different extracts.

5.5 REJ!'EREHCES

1. M.L. Lee and B. W. Wright, J. Chromatogr., 184 (1980) 235. 2. B. Xu and N.P.E. Vermeulen, J. Chromatogr., 445 (1988) l. 3. H. Boer, P. van Arkelp and W.J. Boersma, Chromatographia, 17 (1980) 500. 4. E.G. Boeren, J. Chromatogr., 349 (1985} 377. 5. J.P. Durand, Y. Boscher, N. Petroff and M. Berthelin, J. Chromatogr., 395 (1987) 229. 6. v.w. Watts and T.F. Simonick, J. Anal. Toxical., 10 (1986) 198. 7. B. Newton and R.F. Foery, J. Anal. Toxical., 8 (1984} 129. 8. B. Rambeck and J.W.A. Meijer, Ther. Drug Monitor, 2 (1980) 385. 9. J.T. Burke and J.P. Therot, J. Chromatogr., 340 {1985) 199. 10. J.W.A. Meijer, B. Rambeck and M. Riedmann, Ther. Drug Monitor, 5 (1983) 39. ll. H.T. Badings, c. De Jong, R.P.M. Dooper and R.C.M. De Nijs, Progress in Flavour, Elsevier Science Publisher, Amsterdam, 1984. 12. N. Fehringer, J. High Resolut. Chromatogr. Chromatogr. commun., s (1985) 98. 13. B. Lau, Chemosphere, 14 (1985) 799. 14. J.C. Duinker, D.E. Schulz and G. Petrick, Anal. Chem., 60 (1988) 478.

-111- 15. M.N. Hasan and P.C. Jurs, Anal. Chem., 60 (1988) 978. 16. C.A. Cramers, G.J. Scherpenzeel and P.A. Leclercq, J. Chromatogr., 203 (1981) 207. 17. C.P.J. Schutjas, E.A. Vermeer, J.A. Rijks and C.A. Cramers, J. Chromatogr., 253 (1982) 1. 18. C.P.J. Schutjes, Ph.D. Dissertation, Eindhoven University of Technology, Eindhoven, The Netherlands, 1983. 19. A. van Es, Ph. D. Dissertation, Eindhoven university of Technology, Eindhoven, The Netherlands, 1990. 20. C.A. Cramers and P.A. Leclercq, CRC Critical Reviews in Anal. Chem., 20 (1988) 117. 21. R.L. Henry, G.A. Baney and C.L. Woolley, in P. Sandra (Ed.), Proc. of llth International Symposium on Capillary Chromatography, Monterey, CA, USA, May 1990, Hüthig, 1990, p. 39. 22. D.H. Desty, A. Goldup and W.T. Swanton, In N. Brenner et al. (Eds.), Gas Chromatography, Academie Press, New York, 1962, p. 105. 23. P. Sandra and M.V. Roelenbosch, Chromatographia, 14 {1981) 345. 24. G. Gaspar, c. Vidal-Madjar and G. Guichon, Chromatographia, 15 (1982) 125. 25. J. Phillips, L. Dershingline and Ru-Po Lee, J. Chromatogr. Sci., 24 {1986) 296. 2 6. M. Proot and P. Sandra, J. High Resolut. Chromatogr. Chromatogr. Commun., 9 (1986) 618. 27. Th. Noij, E. Weiss, T. Herps, H. van cruchten and J. Rijks, J. High Resolut. Chromatogr. Chromatogr. Conunun., 11 ( 1988) 181. 28. A. van Es, J. Janssen, c. Cramers and J. Rijks, J. High Resolut. Chromatogr. Chromatogr. Commun., 11 (1988) 852. 29. J.P.E.M. Rijks, H.M.J. Snijders, A.J. Bombeek, H.E.M. van Leuken and J.A. Rijks, in P. Sandra (Ed.), Proc. of 13th International Symposium on Capillary Chromatography, Riva del Garda, Italy, May 1991, Hüthig, 1991, p. 35. 30. S.R. Lipsky and M.L. Duffy, J. High Resolut. Chromatogr. Chromatogr. commun., 9 (1986) 376. 31. S.R. Lipsky and M.L. Duffy, J. High Resolut. Chromatogr. Chromatogr. commun., 9 (1986) 725. 32. P. Sandra and F. David, in P. Sandra (Ed.), Proc. of llth International Symposium on Capillary Chromatography, Monterey, CA, USA, 1990, Hüthig, 1990, p. 382. 33. s. Trestianu, G. Zilioli, A. Sironi, c. Saravalle, F. Munari, M. Galli, G. Gaspar, J .M. Colin and J .L. Jovelin, J. High Resolut. Chromatogr. Chromatogr. Conunun., 8 (1985) 771. 34. w. Blum and L. Damasceno, J. High Resolut. Chromatogr. Chromatogr. Commun., 10 (1987) 472. 35. R.L. Firor and R.J. Phillips, in P. Sandra (Ed.), Proc. of 9th International Symposium on Capillary Chromatography, Monterey, CA, USA, May 1988, HUthig, 1988. 36. w. Blum, W.J. Richter and G. Eglinton, J. High Resolut. Chromatogr. Chromatogr. Commun., ll (1988) 148. 37. Y. Takayama, T. Takeichi and s. Kawai, J. High Resolut.

-112- Chromatogr. Chromatogr. Commun., 11 (1988) 732. 38. J. Curvers and P. van den Engel, J. High Resolut. Chromatogr. Chromatogr. Com:mun., 12 (1989) 16. 3 9. J. Bui j ten, Ph. Mussche and J. Peene, in P. Sandra (Ed.), Proc. of lOth International Symp. on capillary Chromatography, Riva del Garda, Italy, May 1989, Hüthig, 1989, p. 962. 40. W. Blum and G. Eglinton, J. High Resolut. Chromatogr. Chromatogr. Commun., 12 (1989) 290. 41. J. Buijten, c. Duvekot, J. Peene, J. Lips and M. Mohnke, in P. Sandra (Ed.), Proc. of llth International Symp. on capillary Chromatography, Monterey, CA, USA, May 1990, Hüthig, 1990 1 p. 91. 42. J. Buijten, c. Duvekot, J. Peene and Ph. Mussche, International Chromatography Lab., 2 (1990) 5. 43. L. Damasceno, J. cardoso and R. Coelho, in P. Sandra (Ed.), Proc. of 13th International Symp. on Capillary Chromatography, Riva del Garda, Italy, May 1991, Hüthig, 1991, p. 155. 44. J. Buijten, c. Duvekot, w. van der Poel, Ph. Mussche and M. Mohnke, in P. Sandra (Ed.), Proc. of 13th International Symp. on Capillary Chromatography, Riva del Garda, Italy, May 1991, Hüthig, 1991, p. 165. 45. P. Husek, J.A. Rijks, P.A. Leclercq and C.A. cramers, J. High Resolut. Chromatogr., 13 (1990} 633. 46. P.A. Tice and J.D. McGuinness, Food Addit. Contam., 4 (1987) 267. 47. O.A. Teirlynck and M.T. Rosseel, J. Chromatogr., 342 (1985) 399. 48. M.W. Raynor, K.O. Bartle, I.L. Davies, A. Williams, A.A. Clifford, J.M. Chalmers and B. W. Cook, Anal.. Chem., 60 (1988) 427. 49. P.J. Arpino, O. Dilettato, K. Nguyen and A. Bruchet, J. High Resolut. Chromatogr., 13 (1990) 5. 50. H.P. Burchfield and E.E. Storrs, Biochemica! Applications of Gas Chromatography, Academie Press, New York, 1962.

-113- CBAP'l'ER 6 l\fiCELLAR ELECTROKINETIC CAPILLARY CHROMATO­ GRAPHY - THE INFLUENCE OF DIFFERENT COLUMN MODIFICATIONS ON THE ELECTROOSMOTIC FLOW

A systematic investigation was carried out to study the electroosmotic flow (EOF) in micellar electrokinetic capillary chromatography (MECC). Fused silica capillaries were treated with different reagents to investigate the parameters which influence . the magnitude and reproducibility of the EOF. The resul ts from untreated and treated columns were compared under the same experimental conditions. Furthermore, experiments with and without sodium dodecylsulfate (SDS) addition to the buffer were carried out in order to elucidate the effect of this surfactant on the EOF. Finally, the mechanism behind the effects of these modifiaations on the EOF and the stability of the coatings on the column wall are discussed.

6.1 IN'l'RODUC'l'ION

The term 11 electrochromatography11 {EC) was first used by Berraz in 1943 (l) referring to a primitive form of paper electro- phoresis. The whole process was described even earlier by stain [2], though, he did not use this nomenclature in his research.

At present, i t can be considered as a group of separation methods in which the flow of the mobile phase is driven through the column by electroosmosis {or electroendosmosis). Generally, an EC-separation is accomplished by the combination of electro-phoretic and chromatographic selectivity mechanisms. Adopting the classification suggested by Knox and Grant [3], electrochromato-graphy can be divided into four major groups: micel lar electro-kinetic capillary chromatography (MECC) [ 4-11] , capillary electroosmotic

chromatography (CEC) [12-14], capillary gel electrophoresis

-114- (CGE) [15-21), and càpillary zone electrophoresis (CZE) (13, 22-26].

After the pioneer's works [22, 23] of using narrow bore tubes in zone electrophoresis, CZE became a powerful tool in the separation of ionized solutes [23). Very high efficiencies, usually exceeding several hundred thousands theoretical plates can be obtained. Two major contributions to that may be distinguished (3, 23]. First, due to the plug like flow profile in the column the peak broadening process of the solute zones is in principal only determined by axial diffusion of the solutes in the buffer. This condition is met if no solute-wall interactions occur. Second, the currently applied small diameter columns dissipate very efficiently the generated joule heat in the column, so preventing the disturbance of solute bands.

Complementary to CZE , which can separate only charged and water soluble solutes, MECC can be used for the separation of uncharged compounds, even when they are hardly soluble in water [27). MECC, introduced in its present form by Terabe et al. [4, 5), develops rapidly and a lot of attention is paid to its application. It must be mentioned, however, that the potentials of micellar agents in separation technology were earlier recognized in liquid chromatography as micellar or soap chromatography [28-

30). As mentioned above, in contrast to CZE, also uncharged compounds which are almost insoluble in water, can be

-115- separated by MECC. In fact the separation in MECC is achieved by the partitioning of the compounds over two phases, the electroosmotically driven hydro (organic) mobile phase and the micellar phase. It is a true chromatographic method since the separation of the solutes depends on the different partition of the solutes between two phases which move at different velocities [5]. MECC can be performed in similar apparatus as usually applied in CZE.

In electrically driven liquid chromatography (CEC), packed and open capillary columns can be used. The buffer is driven through the column by electroosmosis rather than by pressure. Partitioning of solutes between the mobile and stationary phases occurs in the usual way and promising results have been reported recently (3, 13, 31, 32].

A predictable and reproducible electroosmotic flow (EOF) through the capillary column is essential for MECC and CEC techniques. Therefore it is important to control the magnitude and the reproducibility of the EOF in order to obtain reliable resul ts. The EOF is the movement of a conducting liquid relative to a stationary surf ace of an isolator, which results when an electrical field is applied along this liquid/solid system.. At the interface between the solid and the liquid phase an electrical double layer exists (Figure 6.1). In the double layer a specific zeta-potential can be distinguished as one of the factors determining the EOF. The expression for the potential distribution in the electrical double layer has been described (33-37} and lts relation to the EOF was derived

-116- by von Smoluchowski [38).

Factors influencing the EOF include, the physical and chemical properties of the column wall, the composition and concentration of the buffer, the pH value and percentage of organic modifiers in the mobile phase and the eventual in-situ coating effects of the column wall by additives to the buffer.

Stern layer dlffuae layer 1 1 1 î 111.: 1 r 1 1 LE-- plane ofrshear 1 1 1 1 1 1 i I+ 1 1 1 1 1 i 1 A + IE li 1 - 1 1 1 1 + 1 ."':.. --- + r 1 + 1 1 1 1 1 + 1 - 1 + 1 + 1 ~------i- 1 1 + 1 1 + - 1 1 1 1 + l 1 + 1 1 1 1 aolld - 1 1 bulk 1 + 1 1 1 aolutlon - + 1 1 11ë;;;-; ------T------1 l - + l 1 lmmoblle 1 moblle reg Ion reg Ion ---+ dlstance x Fiqure 6.1 Schematic diagram of the electrical double layer at a solid/liquid interface.

A number of reports bas been published about the factors which may influence the EOF. Organic solvents such as methanol, 2- propanol and acetonitrile were investigated ~s additives to the buffer solution to affect the EOF [39-48]. Both decreased EOF [ll, 45, 48) and increased EOF (41] have been observed after the addition of acetonitrile. Also silanization by trimethylchlorosilane (TMCS) of the column surface was used

-117- to reduce the EOF (10, 23]. Polymers and a number of other coatings were applied by different authors [ 49-55] and a decreased EOF was generally observed. Terabe et al. [ 56) investigated the effects on the EOF of hydrophobic and hydrophilic coatings. The EOF was increased in MECC by coating with polymethylsiloxane. This was explained in terms of the adsorption of sodium dodecylsulfate (SOS) to this coating at the column wall. A decreased EOF was obtained in a carbowax 20M modified column which could not be explained by the adsorption of SOS. Lux et al. [57] also studied the influence of the same types of coatings. Similar results were achieved as described by Terabe [56], it was also found that the effect of the coating on the EOF increased with increasing film thickness until the film thickness is larger than 40 nm and a constant EOF value is reached. The observed changes of the electroosmotic flow by the column coatings were attributed to the changes of the f-potential in the electrical double layer at the column wall, without further explanation.

In some cases confusing results of the observed EOF values were reported and explained mainly in terms of the changes of the f-potential in the electrical double layer at the column wall (Fig. 6.1). The question, however, why the f-potential changed in a specific situation by the different modifications of the columns surface remains unanswered in most cases.

The aim of this chapter is to study these essential questions by a systematic investigation of the influence of different coatings on the EOF and to correlate that to the

-118- physical/chemical properties of the column wall. FUsed silica capillaries were treated with a variety of reagents to investigate their effects on the magnitude and reproducibility of the EOF. The results of untreated and treated columns were compared under the same experimental conditions. Furthermore, experiments with and without SDS addition to the buffer were carried out in order to elucidate the effect of this surfactant on the EOF. Finally interaction mechanisms at the liquid/solid interface of the column wall are discussed to explain the influence of the different modifications.

6.2 THEORY 6.2.1 Electrical double layer

An electrical double layer occurs at the interface between a solid and a liquid phase which is a feature of all surfaces i:m:mersed in a liquid [37). The surface of silica, glass and Teflon tubes carry negative charges. In the case of silica, the hydrated surface consists of a. o. silanol groups which can be ionized and are balanced by positive charges (hydrated cations) from the buffer solution. Some of these positive charges firmly held on the surface by electrostatic force form a "compact layer" or Stern layer [33). Part of the remaining counterions diffuse away from the compact layer because of the thermal motion and form a loosely held layer indicated as the "diffuse layer" [34). The concentration of the diffuse layer approaches finally the bulk value, as the distance from the wall goes to infinity [34, 37). The potential drops linearly in the Stern layer and exponentially within the diffuse layer

-119- as depicted in Figure 6.1. An expression for the potential distribution in the diffuse layer in the solution was given by Gouy [35] and Chapman [36] for a binary electrolyte system with the valency z+ = z-:

( 6 .1) and X = ( 2 * e2 * no * z2)~ (6.2) t 0 * tr * k * T where ir: the electrical potential, x: the distance from the column wall, x: the Debye-Hilckel parameter, k: the Boltzmann constant, T: the absolute temperature, e: the electronic charge, n°: the electrolyte concentration in the bulk solution, eo: the permittivity of vacuum, er: the dielectrical constant of the buffer, z: the valency of the electrolyte.

A similar result is obtained by the Debye-Hilckel theory, in which, l/x is called the "thickness of the electrical double

layern, ó: 1 (6.3) x

ó is inversely proportional to the square root of the ionic strength of the electrolyte solution (see eq 6.2).

6.2.2 t-Potential and Electroosmotic Flow (EOF)

When an external electrical field is applied across a capillary column filled with the electrolyte solution, the excess counterions within the diffuse layer will move towards the appropriately charged electrode, dragging the solvating

-120- liquid along with them. This movement of solution is known as electro(end)osmosis and its direction and magnitude depends both on the properties of the capillary wall and the nature and concentration of the buffer (14, 58). Buffer ions, which are tightly adsorbed in the Stern layer will not move under the influence of an electrical field. A boundary between the Stern layer and the diffuse layer is assUllled to exist if the bulk solution f lows tangentially to the surface and is indicated as the "plane of shear". The corresponding potential at this shear is called the electrokinetic potential or zeta-potential, denoted by f. The question of localizing the shear plane remains open because of the absence of independent methods of determining the r­ potential which can be determined only by experiments on the basis of various electrokinetic measurements (59). The Smoluchowski equation (eq 6.4) can be used to describe the relation between the electroosmotic Velocity, Veoi and the f­ potential (38]. This equation, however, is restricted to capillary systems with inner diameters considerably higher than the double layer thickness.

where E: the applied electrical field strength, ~= the viscosity of the buffer, µe0 : the coefficient of the electroosmotic flow.

The assumption is made here that Er and ~ have the same value in the double layer as in the bulk solution.

-121- It is clear that a charged sheath around a core of uncharged liquid is formed within the electrical double layer. Opposite to the parabolic flow profile in pressure driven separation systems, the flow profile in electrically driven systems within the core is flat by the applied electrical field. Because there is no charge unbalance within the core. Furthermore, since the sheath itself is so thin that the flow profile within the column as a whole is a near perfect plug flow (see Figure 6.2). This plug like flow profile reduces the band broadening process that occurs as a solute travels along the tube, so resulting in very high efficiencies.

-u

© 1 J-u 0

Fiqure 6.2 Schematic diagram of the flow profile in open tubes under pressure driven (upper) and electroosmotic driven (lower).

several methods have been applied to measure the EOF. A simple way is measuring the migration time of an electrically neutral marker substance to calculate the EOF [ 4, 5, 14] • This substance is transported only by the liquid flow and should not show any electrophoretic mobility or any adsorption effect. A similar problem is encountered with this method as in liquid chromatography to find a good zero retention time (to) marker. The magnitude of the streaming potential, generated by applying a pressure gradient across the length of the

-122- capillary, can also be used to estimate the EOF [37, 60, 61]. The following equation correlates the streaming potential, Estr to the r-potential of the column surface and thus to the EOF:

(6.5) where áp: the pressure drop across the tube, X: the specific conductivity of the electrolyte solution.

This method is limited to situation where the surface conductance is negligible compared to the specific conductance of the electrolyte solution and a laminar flow profile is built up within the capillary. Altria and Simpson [58, 62) introduced a weighing method by using a microbalance to measure the electroosmotic volumetrie flow of the electrolyte solution. This technique may approach most closely the real experimental conditions because of the direct determination of the flow. However, only time-averaged results can be obtained by this method [61).

Another method based on the recording of the current during CZE operation was proposed by Huang et al. [ 63) . In this method different concentrations of the same electrolyte solution were used while monitoring the changes in the electrical current during this CZE operation. The displacement of one solution by another solution in the tube can be established and from that the EOF can be calculated. The difference between the two concentrations of the solutions may not be too large, otherwise, the j"-potential is changed during

-123- the operation. Current disturbances may result in inaccurate EOF measurements using this technique.

Also, colloidal particles, of similar materials as used for the manufacturing of the capillary, were applied to determine the EOF [52]. However, in general it is not recommended to compare or to predict the EOF behaviour in an open tubular capillary with EOF value obtained from colloidal particles because of the dif f erence between this method and the real experimental conditions.

6.2.3 Factors Influencinq the Electroosmotic Velocity

Glass, fused silica and Teflon capillary columns, which are predominantly used in chromatography and electrophoresis, have negative charges on the inner surface when they are filled with an aqueous solution [37]. When an electrical field is applied to the tube, the counterions in the diffuse layer move to the negative electrode dragging the liquid with them and so accomplishing the electroosmotic flow (EOF). The charge density of the surface can be influenced in different ways and also the sign of the charge at the column wall can be reversed.

The control of the EOF in electrochromatography can be subdivided in two major groups: I. dynamic modification of the column wall by addition of specific reagents to the mobile phase, resulting in general in a reversible coating of it.

-124- II. chemical treatment of the column wall, resulting principally in the irreversible modification of it.

I. As mentioned earlier, organic solvents such as methanol, ethanol, 2-propanol and acetonitrile can be added to the buffer solution to influence the EOF [10, 11, 39-48, 64, 65]. Contradictory results were observed with the addition of acetonitrile [41, 45, 48). More generally, the mechanism of the effects on the EOF caused by addition of organic solvents is still not quite clear. The addition of surfactants to the mobile phase is also used to influence or to reverse the electroosmotic flow. Decreased EOF values were obtained by addition of mono- or.diamines [25, 66-69] to the electrolyte solutions. Simultaneously the adsorption of proteins to the column wall was suppressed. Divalent amines were found to be more effective than monoamines in this particular case {69). Polyethylene glycol (70), Triton-x (71), and Brij 35 [72) were also used to decrease or eliminate the EOF by increasing the viscosity of the buffers.

By applying cationic surfactants, such as dodecyltrimethyl­ ammonium chloride (DTAC), tetradecyltrimethylammonium bromide (TTAB), cetyltrimethylammonium chloride (CTAC) and cetyltri­ methylammonium bromide (CTAB), not only the magnitude of the EOF was changed, also the direction of the EOF was reversed [8-10, 25, 62, 73-76]. This is caused by the adsorption of the cationic molecules to the column wall, which consequently changes the polarity of the f-potential in the electrical

-125- double layer. Electroosmosis can be completely eliminated by the addition of s-benzylthiouronium chloride to the buffer according to Altria and Simpson (58].

II. By chemical modification methods, generally a thin layer of a coating is formed on the inner column surface. Two different aims may be pursued here. One is, e.g. in the case of the application of fused silica columns, to cover or shield the surface silanols to control the polarity and the magnitude of the EOF. Second also a specific modification can be applied to suppress or eliminate the interaction between the solute molecules and the column surface. This is special important for the separation of biopolymers such as proteins and other biomolecules. Jorgenson and Lukacs treated columns with trimethylchloro­ silane (TMCS} and observed a decreased EOF (23]. Similar results (10] were obtained with acrylamide (77], polyacryl­ amide [49, 51], octadecylsilane [40], and aminopropyltri­ ethoxysilane [78] coated columns. Due to the different hydrophobic and hydrophilic properties of the coatings, mutual differences in the observed EOF values may be expected. An increased EOF was obtained in MECC applying a column coated with polydimethylsiloxane [56, 57]. Here, this was explained by the adsorption of the surfactant to the column wall (56]. Moreover, also the EOF was found to depend on the film thickness of the coating and to reach a constant value after the film thickness is larger than 40 nm (57]. The hydrophilic coatings resulted in decreased EOF values and in some cases in the reversal of the EOF direction. Glycol [50), glycerol

-126- [55], polyethylene glycol (PEG) [52, 79] and Carbowax 20M [56, 57] coated columns showed suppression effects on the EOF. On polyethyleneimine coated capillaries [80], the EOF is reversed because of the positive charge of the coating when it is in contact with aqueous buffers.

Recently, a method was suggested to control the electroosmotic flow during an EC separation [64]. Here the external applied potential is vectorially coupled with the potential inside the capillary. In this way, the r-potential can be controlled and changed by a definite positive, or negative value. Therefore, the r-potential can be changed at any time during the analysis to achieve improved separation results. However the problem is that the instrumentation f or this technique is complicated and not yet commercially available.

The concentration [81] and the pH value of the buffer solution in EC separations are the most important werking parameters to influence the EOF [58, 82, 83]. This is especially true for fused silica and/or glass capillary columns because the dissociation of the surface silanols strongly depend on the pH of the buffer solution (see Figure 6.3).

6.3 EXPERIMENTAL

6.3.1 Equipment and reagents:

A home-built instrument system consisted of a Perspex box with a pressure system for rinsing the capillary; an UV detector (Unicam Analytica! Systems, Cambridge, UK) for on-column

-127- detection; an HCN 35-35000 High voltage power supply (FuG, Elektronik GmbH, FRG). A Tulip AT compact 3 computer connected to a home made interface Multilab-TS allowed control of electromigrative sample introduction and adjustment of the separation voltage as well. Acquisition and calculation of the data were performed with the software program Caesar (B-Wise Software, Geleen, The Netherlands).

14 c

12

10 I î 8 l... • 1 :l

2

0 .__~ _ _,__....__,___,'----'---'--"'-----' 2 3 4 6 fl 7 8 9 10

pH

Fiqure 6.3 Schematic diagram for the effect of pH on electroosmotic flow. Data for line 1 and line 2 are from ref. ( 82] and ( 58) respectively.

The separations were performed in fused silica capillaries (350 µm o.d., 50 µm i.d., Siemens, FRG) of 67 cm total length, with an optical window at 50 cm from the injection side for on-column detection. Samples were introduced into the column by the electromigration technique applying 5 kV for 2 sec. The separations were carried out at 20 kV using methanol as the

-128- dead time marker and benzyl alcohol for the measurement of peak area.

Polymethylhydrosiloxane (PMHS) and polymethyloctadecylsiloxane (PMODS) were purchased from Petrarch systems (Bristol, PA, USA). Polydimethylsiloxane (PDMS) and Carbowax 20M were obtained from Chrompack (Middelburg, NL). These reagents were used without any further purification.

6.3.2 Procedures:

The capillary column was pretreated according to the procedure developed in our laboratory. [84]. For PMHS deactivation, the column was dynamically coated with 15% (v/v) PMHS solution in pentane (according to Bartle (85), generating a film thickness of 40 nm). After flushing the column with He for 30 min, it was sealed by a micro torch and heated in a GC oven from 40°C to 300°C at 8°C/min and kept at 300°C for 2 h. Then, the column was rinsed with l ml dichloromethane and l ml methanol respectively. For ether modifications, the capillaries were dynamically coated with 15% (v/v) PMODS and statically coated with 0.8% (v/v) PDMS and 0.5% (v/v) carbowax 20M solution respectively. 0.5-3% (w/w to the stationary phases) dicumyl peroxide (DCP) was added to the coating solution f or immobilization of the coatings. curing was carried out by heating the sealed column from 40°C to 1ao 0 c at 8°C/min and then 180°C for 80 min.

Afterwards, the modified capillaries were rinsed with l ml dichloromethane and l ml methanol respectively.

-129- Before use in the MECC experiments all the capillaries ware washed with water (30 min) and subsequently with the same buffer solutions to be used in the experiments (20 min) , prior to the separations.

The buffer solutions used in this study consisted of:

(1) 0.02 M sodium phosphate aqueous solution, pH 7.

(2) as (1) with addition of SOS (0.05 M), pH= 7.

(3) as (1) adjusted to pH 1 with concentrated HCl

solution.

(4) as (1) adjusted to pH 9 with concentrated NaOH

solution.

6.4 RESULTS AND DISCUSSION

6.4.1 Reproducibilities of the EOF data and peak areas

The reproducibility of the EOF expressed in migration times

and peak areas measured for some coated and untreated columns

are presented in Tabla 6.1. A rinsing step which included

washing the column with water for 10 min was involved between

soma of the analyses in order to investigate the influence of

rinsing on the reproducibility.

From the data in Tabla 6.1, it can be concluded that improved

reproducibility of the EOF and peak areas was obtained by

rinsing the capillary column with water between each analysis.

-130- Table 6.1 Reproducibility of migration times and peak areas, buffer (1), UV detection at 210 nm.

Col. 1 migration time (min) peak areas (V*S) No. (a) (b) (a) {b) s a a/s s a a/s s a a/s s a o/s 1 3.20 .04 1.2 3.35 .os 2.3 5.75 .29 5.0 6.16 .48 7.8 3 4.02 .11 2.7 4.32 .26 6.0 9.58 .44 4.5 10.7 .86 8.0 4 4.92 .11 2.1 5.37 .27 5.0 14.5 .42 2.9 15.6 .92 5.9 Column No.: 1 = PDMS coated column; 3 = untreated column; 4 = PMODS coated column; s = average value, a = standard deviation, a/s = relative deviation, %; number of experiments n = 3; (a) with rinsing step; (b) without rinsing step.

6.4.2 J:nfluence on the EOF of the surface Modification of the Columns

1. Without addition of SDS to the buffer:

The EOF data together with the standard deviations for untreated and modified capillary columns without addition of sos to the buffer are summarized in Figure 6.4 and Table 6.2.

Table 6.2 Comparison of EOF data with and without addition of SDS to the buffer solution, UV detection at 210 nm.

col. EOF (mm/sec) No. treatment buffer (1) buffer (2) s a a/s s a a/s 1 2.60 0.03 1.15 2.78 0.04 1.44 PDMS 2 2.25 0.07 3.11 2.39 0.06 2.51 PMHS 3 2.07 0.06 2.90 1.82 0.06 3.30 untreated 4 1.69 0.04 2.37 1.54 0.05 3.25 PMODS 5 1.04 0.03 2.88 0.93 0.03 3.23 Carbowax 20M

-131- From the data in Table 6.2 it may be clear that compared to the untreated column:

{l) The EOF is increased by the surface modification of PMHS and PDMS coatings, which is in agreement to the observation in (56, 57]. (2) A decreased EOF is observed for the PMODS and Carbowax 20M coatings.

3 1 "' without SOS û " Cl I ~ 6 2 "' s Cl u. 0 Cl w "' 0 wit! SOS

o~~~~~""-~~~~-'-~~~~-'-~~~~-' 1 2 3 4 5 column No.

Figure 6.4 EOF values in untreated and differently modified capillary columns, column No. see Table 6.2.

2. With addition of SDS to the buffer:

The EOF data together with the standard deviations for untreated and modified capillary columns with SOS in the buffer solution are included in Tabla 6.2. Several conclusions can be drawn from the data in Table 6.2: (1) In contrast to the situation without SOS in the buffer, the EOF in the untreated column is decreased. (2) The EOF is increased for PMHS and PDMS treated columns

-132- compared with the results without SOS in the buffer. (3) A significantly decreased EOF is observed for PMODS and Carbowax 20M modified columns after SOS was added to the buffer. These conclusions will be discussed in paragraph 6.4.3.

Theoretically, an increased EOF will shorten the analysis time with the sacrifice of resolution. On the other hand, the

resolution can be enhanced by decreasing the EOF, simultaneously, the analysis time is prolonged. This was confirmed by ref. [57] and our experiments. A detailed discussion on the relation of analysis time and resolution is given in Chapter 7.

6.4.3 Disoussion on the Znfluenoe of the surfaoe Treatment and Addition of SDS on the EOP

When an untreated fused silica capillary column is filled with the buffering electrolyte, the dissociation of the surface silanols is the main generation source of the ,S-potential. The rate of dissociation of surface silanols depends strongly on the pH value of the buffer solution. In the case the colUEn is coated with hydrophobic and/or hydrophilic coatings, different effects on the EOF, compared to a bare unmodified column, can be expected. In general, a coating at the wall of a capillary column is supposed to eliminate and/or shield the active sites (silanols) at the column surface. The accessibility of the silanols on the column wall in contact with the water molecules of the electrolyte solution depends a.o. on the

-133- efficiency of the coverage. Both the dissociation of the residual silanols and the specific properties of the coating may contribute to the r-potential in the electrical double layer. To explain the effect of the silica substrate of the column in controlling the EOF, the hypothesis of Tanabe et al. [86- 88] can be applied. This model was developed to describe the surface acidity which was caused by the charge distribution in binary metal oxides. When an inorganic oxide like silica is contaminated with metal impurities it can be considered as a mixture of oxides on molecular scale. These impurities may cause charge imbalances leading to different contribution of the positive or negative charges across the silica surface of the column wall. In turn this may influence the EOF in untreated silica columns at least. Depending on the purity of the silica substrate and the efficiency of an eventual modification of the column wall, this effect will become manifest. In addition to this effect also the charge distribution that is the polarity of the column coating plays an important role. For PMHS, PDMS and PMODS modified columns, the column wal! is covered with a layer of a polysiloxane backbone, of which the Si-o-si bridge is a major component (Fig. 6.5). Considering the electronegativity of the elements involved in the chemica! structure of the column surface given in Table 6.3, it can be assumed that the two positive charges of the upper Si atom are distributed to four bands, ie, +(2/4) ö of a valence unit per bond. While the two negative charges of the o atom are distributed to two bonds, i.e. one negative charge per bónd.

-134- The total charge imbalance is (+2/4-2/2) ó * 2 = -1 ó. This may increase the local negative charge of the column wall. When in contact with aqueous buffer solutions this results in an influence on the EOF. Me 1 (+2/4 - 2/2) ll R "H, Me, C1s R-Si-0- 1 charge difference _____ t0 ______(2/4·2/2) 5 • 2 = -1 ll _

-O-Si-0- column wall 1 0 1 Figure 6.5 Schematic representation of the model structure of polysiloxane modif ied surface.

A more general explanation for the influence of the wall coating on the EOF lies in the structure shown in Figure 6.5. By the partial ionic character of the Si-o bond [89, 90] and the electron donating properties of the alkyl groups, each repeating unit in the backbone will gain an excess of electron density. This makes them behave as a negatively charged centre (spot) in contact with the electrolyte solution. Therefore the EOF is strongly influenced in this way.

Besides the electronegativity, also the steric effects have to be taken into account to explain the differences between these three coatings. The PMHS modification shows an increased EOF compared with the bare silica column, but smaller than the corresponding value in the PDMS modified column. This may be attributed to the weakening effect of the H atom in the PMHS

-135- coating to the charge distribution of the siloxane bridge. Moreover, the film thickness of the PMHS coating is smaller (40 nm) than that of the PDMS coating (100 nm), which also may influence the EOF.

Table 6.3 Electronegativity of the related elements

atom H c 0 Si electronegativity 2.2 2.55 3.44 1.90

For the PMODS coated column, the electron donating properties of the methyl and c18 groups are about the same. The steric effect of the c1s ligands, however, may be dominant in shielding the charged centres at the column wall resulting in a decreased EOF.

In general the siloxane backbone will show a ó+ / ó- charge distribution due to the difference in the electronegativity between the involved elements. In addition this charge distribution will be influenced by electron donating or withdrawing groups.

(CH 2CH 20t H 1 n 0 ------~------0-Si- 0- column wall 1 0 1

Ji'iqure 6.6 Schematic representation of the model structure of Carbowax 20M modified silica surface.

-136- The Carbowax 20M coating by which the column is covered with a layer of polyethylene glycol (see Fig. 6.6) shows a different effect on the EOF. The surface silanol groups are supposed to be shielded from interaction with the electrolyte. Here the electronegativity difference between carbon and oxygen atoms is less than that between silicon and oxygen atoms (see Table 6.3). Therefore, the &+/0-charge distribution is less pronounced. In addi tion, the c-oH groups are more difficult dissociated than silanol groups, in other words, the acidity of silanol groups is higher than that of carbinol groups [91]. Therefore, a reduced EOF is observed using carbowax 20M coatings.

In general, when the column coating is very thin (df < 40 nm), the shielding of the silanols may be not complete which will contribute to the electroosmotic flow [57].

The adsorption of SOS to the column surface will influence the f-potential and normally an increased EOF can be expected. For untreated columns, however, this adsorption may be hindered by the electrical repulsion of the surface silanols. In the case of the untreated column, a decreased EOF was observed after the addition of SOS to the buffer. This may be attributed to the increased viscosity of the electrolyte solution in the presence of sos. Eof is inversely proportional to the viscosity of the electrolyte solution ( eq 6.4). An increase in viscosity by the addition of 0.05 M SDS to the buffer was estimated to be about 10% using the Einstein equation [92, 93]. This is in agreement with the decrease of

-137- the EOF by about 10% of column 3 in Table 6.2.

Taking into account a 10% decrease in the EOF due to the increase of the viscosity of the buffer, the data in Table 6.2 show that SOS may adsorb to the PMHS and POMS coated surface.

This will be due to the hydrophobic interactions between these coatings and the SOS. The enhancement of the EOF by the adsorption overrules the decreasing effect of the viscosity, so a net increased EOF was observed. For the PMOOS coating no net EOF increase or decrease was observed, indicating that very little or no SOS was adsorbed. In contrast to the PMHS and POMS coatings here probably the occurrence of a Oonnan exclusion effect prevented SOS adsorption at the column coating. The adsorption of SOS to the Carbowax 20M modified column wall may be negligible since this material is a hydrophilic coating, showing less or no hydrophobic interaction with SOS.

6.4.4 stability of the coatings

The PDMS and Carbowax 20M coated columns were selected to study the stability of the coatings towards the applied

electrolytes. Also the effects of these possible instability of the coating on the EOF were measured. A sequential

procedure was set up to test the coating stability and consisted of:

(1) EOF is measured with buffer (1) and (2) (see

experimental section), respectively, in each of the columns at the start of the procedure;

(2) next the columns were rinsed with buffer (3) for 1 day;

-138- (3) followed by a rinsing step with buffer (4) for 1 day; (4) finally, the columns were rinsed with water and again the EOF was measured as in step (1) under the same conditions.

This procedure was repeated during 2 weeks. Also, the column activity and the film thickness were determined by GC measurements before and aftar this rinsing procedure. The results of these measurements are summarized in Fig. 6.7. It can be seen that the EOF was decreased by 15% of the starting value with PDMS coated columns after a 2 weeks rinsing procedure. The EOF of the carbowax 20M modified column is decreased by 5% of the starting value after this rinsing test. This latter coating shows a higher st~bility than the PDMS coating under the applied experimental conditions.

3.00 wlth SOS ! POMS modlfied 0 A 0 A D D A A D î A 2.10 1.20 ·"·r·o• Carbowax 20M modlfied

A 4

D o 0 t D 0 D wlth SOS 0.70 .___ __.__ ...... _ __...._ __ ..._ __ .__ _ __,_ ___, 0 2 4 a 8 10 12 14

days :riqure 6.7 Stability test of PDMS modified. column and Carbowax 20M modified column, experimental conditions as in Fig. 6.4.

It also turned out that the film thickness of both coatings was decreased. For the PDMS modified column the EOF value is

-139- approaching the value of an untreated capillary column. GC measurements by comparing the capacity factors of tetradecane, under the same conditions, before and after this rinsing test show that the film thickness of PDMS and Carbowax 20M coated columns is decreased by 15% and 7% respectively. The activity of both columns to the polar components in the test mixture [84] is significantly increased.

6.5 CONCLUSIONS

From the collected data it is clear that rinsing of the fused silica capillary column between analyses is important to obtain reproducible results. Taking precaution with the experimental conditions a repeatable EOF value with a relative standard deviation of about 3% can be obtained throughout the experiments with the home built instrument. Different effects of the several coatings on the electro­ osmotic flow were observed and explained with respect to their specific influences on the f-potential. Enhanced EOF were obtained by PMHS and PDMS surface modification of the columns compared to untreated columns. This is due to the increased charge distribution across the polysiloxane backbone of the coatings, which increases the f-potential in the electrical double layer and hence the EOF. The PMODS coated column suppresses the EOF because of the

steric effect of the c18 ligands which may be dominant in the shielding of the charged centres at the column wall. A lower electroosmotic flow was observed for Carbowax 20M

-140- ' coated columns. This might be due to the less negative charge distribution of the oxygen atom from the c-o bond in the polyethylene glycol coating. In addition, the C-OH groups are more difficult dissociated than the surface silanol groups. The adsorption of SDS to the wall of a bare silica column is hindered by the electrical repulsion of the ionized silanols. The observed decrease of the EOF in this case can be attributed to the increase of the viscosity of the electrolyte solution by the addition of SDS. The enhancement of the EOF in PMHS and PDMS modified columns indicate the adsorption of SDS to the wall. For the PMODS coated columns we assume that by a Donnan exclusion effect the SOS is prevented to adsorb at the coating. consequently no significant change in the EOF by the addition of SDS was observed. Since the Carbowax 20M coating shows little or no hydrophobic properties SOS will not be adsorbed. Therefore, again no significant change in the EOF could be observed. Finally, the stability of the Carbowax 20M modification is superior to that of the PDMS coating under the applied experimental conditions.

6.6 REFERENCES l. G. Berraz, Anales Asoc. Quim. Arg., 31 (1943) 96. 2. H.H. Strain, J. Amer. Chem. Soc., 61 (1939) 1292. 3. J .H. Knox and I.H. Grant, Chromatographia, 24 (1987) 135. 4. s. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya, and T. Ando, Anal. Chem., 56 (1984) 111. 5. s. Terabe, K. Otsuka, and T. Ando, Anal. Chem., 57 {1985) 834. 6. M.J. Sepaniak and R.O. Cole, Anal. Chem., 59 (1987) 472. 7. s. Terabe, K. otsuka, and T. Ando, Anal. Chem., 61 (1989) 251. 8. H. Nishi, N. Tsumagari, and s. Terabe, Anal. Chem., 61 (1989) 2434. 9. D.E. Burton, M. Sepaniak, and M. Maskarinec, J.

-141- Chromatogr. Sci., 25 (1987) 514. 10. A.T. Balchunas and M.J. Sepaniak, Anal. Chem., 59 (1987) 1466. 11. J. Gorse, A.T. Balchunas, D.F. Swaile, and M.J. Sepaniak, J. High Resolut. Chromatogr. & Chromatogr. Commun., 11 (1988} 554. 12. v. Pretorius, B.J. Hopkins, and J.D. Schieke, J. Chromatogr., 99 (1974) 23. 13. J.W. Jorgenson and K.O. Lukacs, J. Chromatogr., 218 (1981) 209 14. T.S. Stevens and H.J. Cortes, Anal. Chem., 55 (1983) 1365. 15. A. Cohen, A. Paulus, and B.L. Karger, Chromatographia, 24 (1987) 15. 16. A. Cohen and B. Karger, J. Chromatogr., 397 (1987} 409. 17. A. Cohen, D. Nagarian, J. Smith, and B. Karger, J. Chromatogr., 458 (1988) 323. 18. T. Kaspar, M. Melera, P. Gazel, and R. Brownlee, J. Chromatogr., 485 (1988) 303. 19. R. Nelson, A. Paulus, A. Cohen, A. Guttman, and B. Karger, J. Chromatogr., 480 (1989) 111. 20. A. Guttman, A.S. Cohen, D.N. Heiger and B.L. Karger, Anal. Chem., 62 (1990) 137. 21. H. Dressman, J.A. Luckey, A.J. Kostichka, J. D'Cunha and L.M. Smith, Anal. Chem., 62 (1990) 900. 22. F.E.P. Mikkers, F.M. Everaerts, Th.P.E.M. Verheggen, J. Chromatogr., 169 (1979) ll. 23. J.W. Jorgenson and K.O. Lukacs, Anal. Chem., 53 (1981) 1298. 24. J.W. Jorgenson and K.O. Lukacs, Science, 222 (1983) 266. 25. H.H. Lauer and D.McManigill, Anal. Chem., 58 (1986) 166. 26. E. Grushka, R.M. McCormick, J. Chromatogr., 471 (1989) 421. 27. S. Terabe, Trends in Anal. Chem., 8 (1989) 129. 28. W.L. Hinze, Annali di Chemica, 77 (1987) 167. 2 9 . J. G. Dorsey, M. T. De Echegaray and J. s. Handy, Anal. Chem., 55 {1983) 924. 30. o.w. Amstrong, Separation and Purification Methods, 14 (1985) 213. 31. J.H. Knox and I.W. Grant, 16th International Symp. on Column Liq. Chromatogr., Washington, USA, 19-24, June 1988. 32. J.H. Knox and I.H. Grant, Chromatographia, 32 (1991) 317. 33. o.c. Grahame, Chem. Rev., 41 (1947) 441. 34. J.O'M. Bochris and A.K.N. Reddy, Modern Electrochemistry, Planum Press, New York, 1970. 35. G.J. Gouy, J. Phys. Radium, 9 (1910) 457. 36. D.L. Chapman, Phil. Mag., 25 (1913) 475. 37. R.J. Hunter, Zeta Potential in Colloid Science, Academie Press, London, 1981. 38. M. von Smoluchoski, Bull. Int. Acad. Sci. Cracovie, 1 (1903) 182. 39. J.C. Rijenga, G.V.A. Aben, Th.P.E.M. Verheggen and F.M. Everaerts, J. Chromatogr., 260 (1983) 241. 40. T. Tsuda, K. Nomura and G. Nakagawa, J. Chromatogr., 248 (1982) 241. 41. s. Fujiwara and s. Honda, Anal.Chem., 59 (1987) 487.

-142- 42. M.M. Bushey and J.W. Jorgenson, Anal. Chem., 61 (1989) 491. 4 3 • M.M. Bushey and J. W. Jorgenson, J. Microcol. Sep. , 1 (1989) 125. 44. B.B. vanorman, G.G. Liversidge, G.L. Mcintire, T. Olefirowicz and A.G. Ewing, J, Microcol. Sep., 2 (1990) 176. 45. A.C. Powell and M.J. Sepaniak, J. Microcol. Sep., 2 (1990) 278. 46. A.T. Balchunas and M.J. Sepaniak, Anal. Chem., 59 (1987) 1466. 47. J.A. Lux, H.F. Yin and G. Schomburg, Chromatographia, 30 (1990) 7. 48. c. Schwer and E. Kenndler, Anal. Chem., 63 (1991) 1801. 49. s. Hjerten, J, Chromatogr., 347 (1985) 191. 50. J.W. Jorgenson, Trends in Anal. Chem., 3 (1984) 51. 51. s. Hjerten, K. Elenbring, F. Kilar, J.L. Liao, A.J. Chen, c.J. Siebert and M.D. Zhu, J. Chromatogr., 403 (1987) 47. . 52. B. Herren, s. Shafer, J. Alstine, J. Harris and R. Snyder, J. Colloid and Interface Sci., 115 (1987) 46. 53. G.J.M. Bruin, J.P. Chang, R.H. Kuhlman, K. Zegers, J.C. Kraak and H. Poppe, J •. Chromatogr., 471 (1989) 429. 54. K.A. Cobb, v. Dolnik and M. Novotny, Anal. Chem., 62 (1990) 2478. 55. R.M. McCormick, Anal. Chem., 60 (1988) 2322. 56. s. Terabe, H. Utsumi, K. Otsuka, T. Ando, T. Inomata, s. Kuze and Y. Hanaoka, J. High Resolut. Chromatogr. & Chromatogr. Commun., 9 (1986) 666. 57. J.A. Lux, H.F. Yin and G. Shomburg, J. High Resolut. Chromatogr., 13 (1990) 145. 58. K.D. Altria and C.F. Simpson, Chromatographia, 24 (1987) 527. 59. s.s. Dukhin and B.V. Derjaguin, in E. Matijevic (Ed.), Surf ace and Colloid .Sci. , Vol. 7, John Wiley & sons, Ine., New York, 1974, Chap. 2. 60, R.A. van Wagenen and J.D. Andrade, J. Colloid Interface Sci., 76 (1980) 305, 61. A.A.A.M. van de Goor, B.J. Wanders and F.M. Everaerts, J. Chromatogr., 470 (1989) 95. 62. K.D. Altria and C.F. simpson, Anal. Proc.,· 23 (1986) 453. 63. X. Huang, M.J. Gordon, and R.N. Zare, Anal. Chem., 60 (1988) 1837. 64. c.s. Lee, w.c. Blanchard, and C.T. Wu, Anal. Chem., 62 (1990) 1550. 65. T. Hanai, H. Hatano, N. Nimura, and T. Kinoshita, J. High Resolut. Chromatogr., 14 (1991) 481. 66. F.S. stover, B.L. Haymore, and R.J. McBeath, J. Chromatogr., 470 (1989) 241. 67. v. Rohlicêk and z. Deyl, J. Chromatogr., 494 (1989) 87. 68. T. Tsuda, Y. Kobayashi, A. Hori, T. Matsumoto, and o. Suzuki, J. Microcol. Sep., 2 (1990) 21. 69. J.A. Bullock and L.C. Yuan, J. Microcol. Sep., 3 (1991) 241. 70. An o. Tran, s. Park, P.J. Lisi, O.T. Huynh, R.R. Ryall, and P.A. Lane, J. Chromatogr., 542 (1991) 459. 71. F. Foret, s. Fanali, L. Ossicini, and P. Bocek, J.

-143- Chromatogr., 470 (1989) 299. 72. H.T. Rasmussen, L.K. Goebel, and H.M. McNair, J. Chromatogr., 517 (1990) 549. 73. T. Tsuda, J. High Resolut. Chromatogr. & Chromatogr. commun., 10 (1987) 622. 74. x. Huang, J.A. Luckey, M.J. Gordon, and R.N. Zare, Anal. Chem., 61 (1989) 766. 75. J.P. Liu, J.F. Banks, Jr., and M. Novotny, J. Microcol. Sep., 1 (1989) 136. 76. T,. Kaneta, s. Tanaka, and H. Yoshida, J. Chromatogr., 538 (1991) 385. 77. K.A. Cobb, v. Dolnik and M. Novotny, Anal. Chem., 62 (1990) 2478. 78. Y.F. Maa, K.J. Hyver, and S.A. Swedberg, J. High Resolut. Chromatogr., 14 (1991) 65. 79. G.J.M. Bruin, J.P. Chang, R.H. Kuhlman, K. Zegers, J.C. Kraak and H. Poppe, J. Chromatogr., 471 (1989) 429. 80. J.K Towns and F.E. Regnier, J. Chromatogr., 516 (1990) 69. 81. s. Fujiwara and s. Honda, Anal. Chem., 58 (1986) 1811. 82. K.D. Lukacs and J.W. Jorgenson, J. High Resolut. Chromatogr. & Chromatogr. Commun., 8 (1985) 407. 83. H.J. Issaq, I. z. Atamna, G.M. Muschik, and G.M. Janini, Chromatographia, 32 (1991) 155. 84. Q. Wu, M. Hetem, C.A. Cramers and J.A. Rijks, J. High Resolut. Chromatogr., 13 (1990) 811. 85. K.D. Bartle, Anal. Chem., 45 (1973) 1831. 86. K. Tanabe, T. Sumiyoshi, K. Shibata, T. Kiyoura and J. Kitagawa, Bull. Chem. Soc. Jap., 47 (1974) 1064. 87. K. Tanabe in J.R. Anderson and M. Boudart (Ed.) catalysis, Science and Technology, Springer-Verlag, Berlin, 1981, p. 231. 88. J. Nawrocki, Chromatographia, 31 (1991) 177. 89. T.C. Kendrick, B. Parbhoo and J.W. White, in s. Patai and z. Rappoport (Ed.), The Chemistry of Organic Silicon Compounds, John Wiley & Sons, New York, 1989, p. 1351. 90. L. Pauling, The Nature of the Chemical Bond, 3rd Ed., Cornell University Press, 1960, Chap. 3. 91. H. Kwart and K.G. King, d-Orbitals in the Chemistry of Silicon, Phosphorus and Sulfur, Springer-Verlag, Berlin, 1977, Chap. 4. 92. A. Einstein, Ann. Phys., 19 (1906) 289. 93. s. Saito, J. Colloid Interface Sci., 24 (1967) 227.

-144- CHAPTER 7

THE SEPARATION OF HERBICIDES BY MICELLAR ELECTROKINETIC CAPILLARY CHROMATOGRAPHY

A method was developed tor the separation by micellar electrokinetic capillary chromatography (MECC) of a number ot herbicides consisting ot chlorophenoxy acids. Sodium dodecyl sulphate (SDS), Brij 35, cetyltrimethylammonium bromide (CTAB) and methanol were applied in the buffering buffer to 1nvestigate their eftects on the separation of the herbicides. SDS combined with Brij 35 as micellar agents were tound to provide the best overall separation of these components. Besides that, the di:fterences in the electroosmotic flow (EOF), due to the different compositions of the various buffers, were correlated with the viscosities thereot. The theoretical calculations and the experimental results of the viscosity measurements were in good agreement with the observed shifts in EOF.

7.1 INTRODUCTIOH

The widespread use of chlorinated phenoxy acid herbicides in agriculture has led to an increased pollution of soil, water and food by their residues and their phenolic metabolites. Their high toxicity to human and aquatic orqanisms [l], makes the development of sensitive methods for the analysis of this class of compounds essential. A number of methods, such as spectrophotometr ic [ 2, 3) , differential pulse polarographic [4) and gas chromatographic methods [5-18), have been reported for the analysis of herbicides. A streng disadvantage of GC methods is that derivatization of the target compounds is necessary. This is time consuming and may introduce errors in the analytical procedure.

High performance liquid chromatography (HPLC) and HPLC

-145- combined with mass spectrometry (LC-MS) have been used also to determine these compounds (19-28]. In this case no derivatization is needed prior to LC separation. The coupling of LC with MS can provide both the speed of analysis and the specificity for non-volatile compounds such as chlorophenoxy acids herbicides. Enrichment and clean-up procedures are nearly always necessary for the HPLC analysis of the chlorophenoxy acids in surface water, urine, soil or sediment. Liquid-liquid extraction [9, 29, 30], solid-phase extraction (11, 22, 25, 31-34) and the enrichment on a precolumn (33, 25, 35-37) or a supported liquid membrane (38-40] can be applied in the sample pretreatment procedure. Recently, a sample pretreatment technique using isotachophoresis as the sample clean-up and enrichment step prior to the LC analysis was developed [41]. Promising results were obtained by this method. Capillary zone electrophoresis (CZE) and micel lar electrokinetic capillary chromatography (MECC) may be also attractive techniques for the analysis of the chlorophenoxy acids. Cue to their outstanding separation efficiency, high speed and potential for automation, CZE and MECC have already been used to separate a large variety of compounds. Applications include amino acids [42, 43], peptides [44-46), proteins (47, 48), priority pollutants (49], vitamins [50-52] and drugs (53-55]. In this study, a method for the separation of a number of chlorophenoxy acid herbicides by MECC was developed. The effects of some surfactants and methanol, in various concentrations added to the buffer were investigated. The

-146- herbicides were detected by a multi-wavelength UV detector. Processing of the detector signals allowed three-dimensional plots of the data, peak confirmation and peak purity evaluation via comparison of spectra. Finally also, the correlation between the variations of electroosmotic flow (EOF) and viscosity changes due to the different buffer compositions will be discussed.

7.2 THEORY

CZE is a one phase process where the separation is based on the differences in electrophoretic mobilities of the solutes

[56, 57). In MECC [58 1 59] 1 however, two phases are distinguished, a hydro (organic) and a micellar phase or pseudo-stationary phase. These two phases are prepared and or formed in the buffer solution containing surfactants, which are added above their critical micellar concentration (CMC).

In Figure 7. 1 a schematic presentation of a micelle and a solute partitioning over the mobile and the pseudo-phase is given. Micelles are dynamic structures and are characterized a.o. by their aggregation number. Both neutra! and charged compounds can be separated by MECC according to their dif f erences in hydrophobicities and electrophoretic mobilities. The electroosmotic velocity, 1.1eo, which was discussed in detail in Chapter 6, in a capillary column is essential for MECC. It can be described by [60):

(7 .1)

-147- where eo: the permittivity of vacuum, Er: the dielectric constant of the buffer,

~= the viscosity of the buffer, r1: the zeta potential at the column wall, E: the external applied electrical field strength.

The negative sign means that when t is negative, the liquid flow is towards the cathode [60). Typical ueo values are in the range of 1-3 mm/sec. The migration time of an unretained solute, t 0, is given by:

(7.2) where 1 is the column length from the injection point to the detector cell.

Figure 7.1 Schematic illustration of micelles and the partition of the solutes

Besides the EOF in a specif ic MECC separation system also a distinct electrophoretic mobility of, at least charged micelles, can be distinguished, resulting in the migration of the micelles. The electrophoretic velocity, Vep, of a charged

-148- micelle then is given by:

\) = 2 (7. 3) ep 3 where f 2 is the zeta potential at the micelle.

Here, it is asswned that € and ~ have the same values in the electrical double layer and in the bulk solution. The overall migration velocity of a micelle, Vmci under fixed experimental conditions then follows from:

(7 .4)

Substitution of eq 7.1 and eq 7.3 into eq 7.4 gives a more detailed insight in the parameters which determine. the velocity of a micelle.

(7. 5)

In order to obtain satisfying separation, the velocity vectors of Veo and liep are preferably opposite and differing in their absolute values. The migration time, tmc1 of a neutral compound completely retained in the micelle phase follows from: 1 t = (7. 6) l1IC "- When a specific neutral compound is solubilized by the micellar phase an expression for the capacity factor, k, can be described: k= (7. 7) where tR: the migration time of the compound.

-149- From this a general expression for the solute velocity in MECC can be given:

\) = __k_ * 1.l + __1_ * 1.leo (7. 8) 8 1 + k me 1+ k where v8 : the velocity of the compound, k: capacity factor of an electrically neutral compounds.

It should be noted here that superimposed on the above mentioned migration processes, charged compounds also move electrophoretically along the capillary tube. The corresponding velocity of such charged species equals:

where v8 (i): the velocity of the charged solute, Vep(i): the electrophoretic velocity of the charged solute in the aqueous phase, k1: capacity factor of the charged solute.

Non-ionic solutes partition between the two phases and elute

in between the range of the migration time of t 0 and tmc· The migration time of a neutral solute is described by [59]:

l+k tR = ----/.,----*-k..- * to (7.10) 1 + (to tmc) As usual in elution chromatography the following resolution equation for a pair of neutral solutes can be derived:

a - 1 * .....,...._k_a_ (7 .11) a ka + 1 where ki and k2: the capacity factors of two neighbouring peaks, N: the number of theoretical plates,

-150- a: the separation factor (equal to k2/ki)·

If two peaks are migrating close together, i.e. k1 ~ k2 ~ k, eq 7.11 can be approximated by:

m IX - 1 * k * 1 - tol tmc R "' i::. * ------(7.12) s 4 a k + 1 1 + (t01tmc) * k

The product of the last two terms in eq 7.12 is defined as f(k) and bas a strong effect on the resolution.

f(k} = _k_ * 1 - t 0 /tmc (7.13) 1 + k 1 + ( t 01 t.mc) * k Figure 7.2 shows the relationship between f(k) and k for some different values of t 0 /tmc· In practical situations to/tmc ranges usually between 0.15 and 0.5.

By assuming that N and a are independent of k [61), the dependence of the resolution on kis restricted to f(k). The maximmn resolution can be derived from eq 7 .12 by setting d{f(k)}/dk = O and yields the optimum value of k:

2 kopc(Rs,max) = ( tt: r/ (7 .14) As can be seen in Figure 7.2, maximum resolution in practical situations is usually obtained for capacity factors between 2 and 5.

This relation indicates that the best resolution of neutral solutes can be achieved by adjusting the working parameters to give an average capacity factor of Ctmc/t0 )l/2,

-151- 1.00

0.80

0.60

0.40

0.20

0.00 '--~~~...... _~~_._._...... ,__~~-=Cl:lll 0.1 10 100

capacity factor. k

Figure 7.2 Dependance of f(k) on the capacity factor, k. The numbers on the plot are the t 0/tmc values.

By substituting the kopt(Rs,max> from eq 7.14 into eq 7.10, the analysis time at maximum resolution can be obtained:

(7 .15)

Finally, the result for Rs,max under the condition of operating at the optimum capacity factor kopt is:

R = ./N * ~ * tmc - to (7.16) s,max 4 (X { ti/2 + t~P )2 In conventional chromatography, Rs,max is obtained when k-ooo:

«. - 1 R s,max = :/J!. (7. 17) 4 *---IX

Comparing eq 7.16 with eq 7.17, it can be seen that Rs,max in MECC is 0.17 to 0.74 times value of Rs max in conventional 1 chromatography (assuming t 0/tmc is between 0.15 and 0.5). This

-152- difference however can be compensated by the superior efficiency of MECC which also contributes to resolution. In the case of a normal standard 15 cm LC column, N is 10,000 and

'V"N is 100. For a MECC system, 100, 000-200, 000 theoretica! plates are easily obtained, 'V"N is between 316 and 447. Taking account the contribution of efficiency, Rs,max in MECC is 0.5 to 3 times that in conventional chromatography.

Under conditions optimized for efficiency, f(k) is the most important parameter for resolution optimization. Two possibilities can be considered, that is the optimization of capacity factor and the optimization of the elution range (window). Again, these two possibilities are reflected by eq 7.14 and eq 7.16 respectively.

When tmc/to is infinite, Rs,max is given by eq 7.17. However, at the same time, tR...co also (see eq 7 .15). In practical situations, tmc/to lies between 3 and 6, then for the maximum

resolution, tR is between 1.73 t 0 and 2.45 to.

7 • 3 EXPERI:MBN'l'AL 7.3.1 Chemicals

2-(2,4-dichlorophenoxy)-propionic acid (2,4-DP), 4-(2,4-di­ chlorophenoxy) butyric acid (2,4-DB), 4-chloro-2-methyl­ phenoxy-acetic acid (MCPA), 2-(4-chloro-2-methylphenoxy) propionic acid (MCPP), 4-(4-chloro-2-methylphenoxy) butyric acid (MCPB), 2,4,5-trichlorophenoxy acetic acid (2,4,5-TPA), 2-(2,4,5-trichlorophenoxy)-propionic acid (2,4,5-TPP), 3,6-

-153- dichloro-2-methoxy benzoic acid (dicatnba), 2,3,6-trichloro­ benzoic acid (2, 3, 6-TB) and 3-isopropyl-2, l, 3-benz:othiadiazin- 4-on-2, 2-dioxide (bentazon) were purchased from Schmidt (Amsterdam, The Netherlands).

SodiUlll dodecylsulphate (SOS) and cetyltrimethylammonium bromide (CTAB) were obtained from Aldrich Chemica! Co. (Milwaukee, WI USA). All the other reagents and solvents were purchased from Merck (Merck AG, Darmstadt, FRG) • HPLC grade water prepared by filtering deionized water through a Milli-Q Water purification system (Millipore Corp., Bedford, USA) was used for the preparation of the buffer solutions and samples.

The buffer solutions and a standard mixture of the herbicides were prepared by dissolving relevant amounts of the reagents in the deionized water. Buffers and samples as well were filtered over 0.45 µm Millex-GS filters (Millipore) prior to use.

7.3.2 Equipment and Procedures

The experiments were carried out on a Spectra Physics iooo™ capillary electrophoresis system equipped with a multi­

wavelength scanning UV detector (Spectra-physics Ine., San Jose, CA, USA).

Automated capillary rinsing, sample introduction and execution of the electrophoretic runs were controlled by an IBM personal

system/2 computer (Model 70 386) with software version 1.3.

-154- Multi-wavelength spectral data were collected, stored and processed on the same computer. The high speed polychrome mode by scanning from 200 nm to 300 nm at 5 nm intervals was employed in these experiments. This permitted three dimensional platting of the electrochromatograms, peak confirmation and peak purity evaluation via the comparison of absorption spectra. sample introduction on this equipment could be performed either by the electromigration or the vacuum technique. Here vacuum injection was applied throughout the experiments to avoid sample discrimination. This was performed by applying the preset vacuum at the cathode side during 1 sec (typically about 2 nl fora 50 µm i.d. column). Fused silica capillary columns with 50 µm i.d., total length 44 cm, effective separation length 37 cm (Siemens, Germany) were used for all the experiments. Prior to use new capillary columns were rinsed with 0.1 N sodium hydroxide solution for 20 min and subsequently by water and the buffer solutions, to be used in the experiments, for 60 min and 20 min, respectively. Thereafter, conditioning for each experiment was effected by rinsing the capillary with water for 5 min and with the buffers for 10 min. In all the experiments a constant voltage of 15 kV was applied and the temperature of the separation system was kept at 25°C. The EOF value was determined by the migration time of methanol, which can be seen from the disturbance of the baseline. The viscosity of some of the buffers were measured on a Rheometrics Fluids Spectrometer (model RFS II, Rheometrics Ine., NJ, USA) at 25°C.

-155- 7.4 RESULTS AND DISCUSSION

7.4.1 The Effect of surfactants on the Separations

From the chemical structures of the herbicides under study shown in Table 7.1 and some preliminary experiments it was concluded that three pairs of components in the mixture were difficult to be separated. They are assigned as: pair a, consisting of compound 8 and 10; pair b, consisting of compound 1 and 2; pair c, consisting Of compound 6 and 7.

Table 7.1 Chemical structures of the investigated herbicides.

/Cl 0 c•-o~- .c"''oH 'R 1. R •H, R •CH , R •Cl: 2,4·DP 1 3 3 2 8. R - OCH 3 : dlcamba 2. R •H, 11 •R •CH : MCPP 9. R • Cl : 2,3,6-TB 1 2 3 3 3. R •R •Il, R •Cll : MCPA 1 2 3 3 4. R •R •Cl, R •H: 2,4,S·TPA 1 2 3 5. R •R •Cl, R •CH : 2,4,$·TPP 1 2 3 3

0 ·O·CH ·CH ·CH -c"' Cl·o~_ 2 2 2 'oH 10. bentazon R 8. Il • Cl: 2,4·08

7. R • CH3 : MCP8

At the selected pH value of the buffers used in this study, all compounds are partially or totally dissociated. Therefore, they carry a negative charge and their migration is governed

-156- by the electroosmotic flow, the electrophoretic mobility and eventually by their interaction with the micellar phase in case MECC is applied. Very high efficiencies {average plate numbers about 120,000 for each compound) were obtained throughout the experiments. In spite of that it was hardly possible to achieve satisfactory separations by using phosphate buffer solutions only. This due to the very similar chemical structures of the target compounds. Therefore, a separation based on MECC was developed.

0 0.05 0.1 concentration of SOS {M)

'i>igure 7.3 Plot of resolution versus concentration of SDS. The experiments were conducted with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: 0.02 M phosphate solution. T = 25°C. (a) bentazon and dicamba; {b) MCPP and 2,4-DP; (c) MCPB and 2,4-DB.

The solubilization of the compounds by specific micelles will contribute to the migration behaviour and thus to a selective separation. · The results obtained at different SDS

-157- concentrations for the three pairs of the components are presented in Figure 7.3. The resolution of the three pairs increases with the concentration of SOS. This is due to the increased hydrophobicity of the components 8, 1 and 6 in the pairs of a, band c respectively (Table 7.1). So an increased interaction of these compounds corresponding to the concentrations of micelles could be expected. A typical three­ dimensional electrochromatogram of the 10 herbicides under study is presented in Figure 7.4.

4 6 wavelength AU (nm) 200 21.0 O.OOEIO

0.0060 240

0.00S9 :öl?O 0 .0019 :::eo

300 -0.0001 ' 1 tS. 150 "'1"~ l.O 7'" •• 90 e"''70 -> "'eso Migration time (min)

Figure 7.4 Three dimensional electrochromatograms of a standard mixture of ten chlorophenoxy acids. The experiments were conducted with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: 0.1 M SOS and 0.02 M phosphate solution. T = 25°C. Peak numbers see Table 7.1.

At a concentration of 0.1 M SOS, the resolution of 2,4-DB and MCPB approaches unity. However, the two peaks are asymmetrie.

-158- A further increase in SDS concentration will increase the electrical current to intolerable values. This will cause additional peak broadening due to the temperature gradients inside of the column.

It has been shown that the addition of Brij 35 (non-ionic surfactant) to an anionic surfactant system may improve the resolution in MECC [62). A distinct advantage of using Brij

35 is that a high concentration of Brij 35 can be used without influencing the current in the electrophoretic system. Brij

35 was employed in this investigation in combination with SDS to study its effect on the separation of the chlorinated phenoxy acids. The results obtained with various concentration of Brij 35 is shown in Figure 7.5. A slightly increased resolution for pairs a and b was observed at ·increasing concentrations of Brij 35. This indicates that the interaction between these solutes and Brij 35 is less effective. The resolution of pair c, however, was almost proportionally increased with the concentration of Brij 35. compared with the other compounds, 2,4-DB and MCPB have relatively longer molecular chains and Brij 35 is a linear molecule. This may result in strenger interaction between these two solutes and

Brij 35,

-159- 3.00

2.50

2.00 c c: .Q ;; ö f" 1.50

a t 1.00

b

0.50 0 2 3 4 5

conc. of Brij 35 {mM)

Figure 7.5 Plot of resolution versus concentration of Brij 35. The experiments were conducted with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: 0.1 M SOS in 0.02 M phosphate solution. T = 25°C. (a) bentazon and dicamba; (b) MCPP and 2,4-DP; (c) MCPB and 2,4-DB.

Further improvement in the resolution of the three pairs of the solutes can be obtained by increasing the concentration of Brij 35. However, the migration times change as well resulting in the overlapping of compounds 3 and 5 and deteriorate the overall separation performance. CetyltrimethylaI!lll\onium (CTAB) was demonstrated to be effective for the separation of a number of negatively charged compounds [63]. It was also applied in this work to evaluate its effect on the separation of the chlorophenoxy acids. An example of the potential of CTAB to separate acids and related compounds is presented in Figure 7.6.

-160- AU 3 1 2 1 1

l 1 1 1 1 1 t i i 1 1 1 L~ J 1 ~ ~ 0 2 3 4 5 · Migration time (mii:l.)

Figure 7.6 Electrokinetic chromatogram of a three components mixture .. The experiments were.conducted with a voltage of -15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: o. 01 M CTAB in o. 02 M aqueous phosphate solution. T == 25°C. Peak identification: l==isonicotinic acid; 2=nicotinic acid; 3=nicotinamide.

The application of CTAB did not result in a sufficient separation for the chlorophenoxy acids. The target compounds are very alike and interact probably in a similar way with CTAB.

Optimization of the resolution by the capacity factor was discussed already in the theoretica! section. It treats the separation conditions concerning one pair of closely migrating compounds. It is more complicated to separate three pairs of difficultly to be resolved solutes because of their similar dependence on the experimental parameters. To achieve a satisfactory separation of the target compounds a compromise has to be made in the composition of the buffer. The best overall separation of the herbicides under study was achieved

-161- by the application of 0.02 M phosphate buffer solution containing 0.1 M.SDS and 3*10-3 M Brij 35 (Figure 7.7).

21 4 7 6 wavelength

0. O:l.00 "'~ ( 310 0.001!10 2:20 230 0.0060 240 :öltlO 1 0.0039 1 260 270 i ::eo 0. 00:!.9 / 390 300 7.':i.o 7 .'.:>o e .'7o .,. .'tso Migration time {min)

Fiqure 7.7 Three dimensional electrochromatograms of a standard mixture of ten chlorophenoxy acids. All experiments were carried out with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: 0.1 M SDS and 3*10-3 M Brij 35 in 0.02 M phosphate solution. T = 25°C. Peak numbers see Table 7.1.

7.4.2 The influence of Methanol on the Separation

It has been show that a small percentage of organic modifiers can also improve the separation efficiency in MECC [64, 65]. The resolution of the three pairs of compounds with addition of various portions of methanol to the buffer are depicted in Figure 7.S.

An improvement in resolution for the pairs a and b was observed with increasing amounts of methanol. A remarkable enhancement in the resolution of the pair c was achieved by increasing the methanol percentages.

-162- The improvement in the resolution can be explained by tl:Î.e increase in the elution range (window) as a result of the addition of the modifiers [64). In fact, a smaller value of to/tmc is obtained by the addition of methanol. From eq. 7.13, and Figure 7.2, it is evident that f(k) increases with decreasing values of to/tmc· consequently, increased resolutions can be expected in the case of decreasing values of to/tmc (cf. eq 7.12).

3.00 c ( A)

2.50

i:: .....0 .... 2.00 ::! ..-1 0 tl) a.I ... 1.50

1.00

o.so 0 5 10 15 20 0 s 10 15 percentage of methanol

Figure 7.8 Plot of resolution versus methanol percentage. (A) O. 02 M phosphate buffer with 0.1 M SDS; (B} 0.02 M phosphate buffer with 0.1 M SOS and 3*10-3 M Brij 35. All experiments were conducted with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm. T = 25°C. (a) bentazon and dicamba; (b) MCPP and 2,4-DP; (c) MCPB and 2,4-DB.

The migration times of the ether four components (peaks 3 1 4, 5, and 9) in Table 7.1 are also influenced by the addition of methanol to the buffer solution. As in the case of the

-163- addition of Brij 35 an increase in migration time is observed for these compounds. This has a negative effect on the overall separation of the sample, e.g. compound 3 starts to overlap compound 5 {Figure 7.9).

wavelength AU (run) 200

Figure 7.9 Three dimensional electrochromatograms of a standard mixture of ten chlorophenoxy acids. The experiments were carried out with a voltage of 15 kV across 44 cm * 50 µm i.d. column, effective length 37 cm, buffer: 0.1 M SOS and 3*10-3 M Brij 35 and 10% methanol in 0.02 M phosphate solution. T = 25"C. Peak numbers see Table 7.1.

With respect to the qualitative and quantitative analysis, multi-wavelength detection allowed three-dimensional plotting of the electrochromatograms (Figure 7.4 and 7.7) showing the relationships of absorbance vs. migration times vs. wavelength.

The absorption spectrum of each component can be extracted from the collected data points indicated as time slices (Figure 7 .10). This provides additional possibilities for the identification of a compound.

-164~ AU

aJ,o wavalength (nm)

AU

wavelength ( lllll)

Figure 7.10 Absorption spectra obtained as a time slice from the data in Figure 7.7. (A) Pairs a and b; (B) pairs band c.

Figure 7.10 shows that in contradiction to compounds s and 10 (pair a), the spectra of compounds 1 and 2 (pair b) and compounds 6 and 7 (pair c) are quite similar. This indicates that the sufficient separation of these chlorophenoxy acids is mandatory to achieve reliable quantitative and qualitative analytica! results.

-165- As an example the results of migration times and peak areas for 6 compounds (the three pairs} are presented in Table 7.2.

Table 7.2 Repeatability of migration times and peak areas.

Peak Migration times Peak areas No. s (1 u /s (%) s a u/s(%} 10 5.67 0.05 0.9 3613 44 1.2 8 5. 72 0.05 0.9 8232 199 2.4 2 5.81 0.05 0.9 13470 193 1.4 l 5.87 0.06 1.0 14217 311 2.2 7 8.11 0.09 1.1 14772 209 1.4 6 8.32 0.11 1.4 14148 247 l. 7 conditions: 0.02 M phosphate containing 0.1 M SDS and 3*10-3 M Brij 35 buffer solution. Sa111.ple concentration about o. 2 mg/ml for each compound, constant voltage 15 kV, column: 44 cm * 50 µm i.d. column and effective length 37 cm. T = 25°C. Peak numbers see Table 7. 1. s-average value, a-standard deviation, a/s (%)-relative standard deviation, number of experiments n = 3,

It was observed that the electroosmotic flow altered with the various buffer solutions used in the experiments. This was examined with respect to the contribution of the viscosity changes of the buffer solutions to the EOF. The viscosity of a dilute suspension of rigid particles in a continuous medium can be approximated by the Einstein equation (66, 67). The result, restricted to the case of small volume fractions of spherical particles, is given by:

11 = 11 0 * ( 1 + 2. 5 * cf> ) (7. 18)

where ~: the viscosity of the suspension, ~ 0 : the viscosity of the medium, ~: the volume fraction occupied by the particles. -166- This equation can also be applied for s'olutions containing micelles which are relative large compared to the solvent molecules [68]. Therefore it can be applied here to estimate the viscosities of the micellar buffer solutions.

The ratio of ~/~o indicates the relative change in viscosity of the solutions with and without micelles and can be determined if t;/> is known. several methods can be used to calculate the volume fraction, t;f>. One of the methods in based on the following equation:

(7. 19) where c is the concentration of the micelles in the buffer solution and Vi is the average mole volume of the micelles.

The viscosity change of a 0.02 M phosphate buffer solution by addition of 0.05 M SDS can be estimated as following. The average radius of an SDS micelle is 2. 5 nm and the aggregation number in this case is 62. The average mole volume of the SDS then, is O. 64 lf:mol and from this t;/> can be calculated as

0.032. Therefore, ~/~o = 1 + 2.5 * t;/> = l.OS, or a 8% increase in the viscosity of the SDS buffer solution compared to the original phosphate buffer solution.

The corresponding experimental results of the viscosity measurements of 0.02 M phosphate and 0.02 M phosphate containing O. 05 M SOS buffer solution show viscosi ties of 0.995 and 1.096 e.p. (25°C}, respectively. The ratio of these two values (l.10) is in reasonable agreement with the value

-167- calculated from the Einstein equation.

Assum.ing that with the exception of " all the variables in eq 7.1 are constant, under defined experimental conditions, then

Cveo)ph/(ueo)sos = "sos/"ph· Where (veo>ph and (ueo)sos are the electroosmotic velocities in phosphate and SOS buffers, "SOS and ?jph are the viscosity of the corresponding buffers respectively. The experimentally observed (1.1eo>ph/(11e 0 )sos value is about 1.09, which is in fair agreement to the value of the measured ?!sos/77ph value.

7.5 CONCLUSIONS

The selectivity of the MECC technique is demonstrated in this study for compounds which have similar chemica! structures. Tuned selectivity in the separation was obtained by optimizing parameters such as the concentration of SOS and Brij 35 and the percentage of methanol in the buffers. SDS combined with Brij· 35 as the micellar agents provided satisfactory separations and relatively short analysis time. Also, high efficiencies (more than 100,000 theoretica! plates were obtained for the separation of these chlorinated phenoxy acids herbicides under MECC conditions. Relative standard deviations for migration times are about 1% and the repeatabilities for peak areas are approximately 2% for all the investigated compounds. The multi-wavelength detection provided useful additional information. The characterization of the sample zones by their migration times and adsorption spectra proved to be a powerful

-168- tool for solute identification. Experimentally measured viscosity data were in good agreement with the calculated results from the Einstein equation. The differences in the viscosities match well with the observed changes in the EOF. MECC offers great selectivity and flexibility. Therefore the possibilities of this technique in other application areas should be further investigated.

7.6 REFERENCES 1 R.W. Bovey and A. Young, The Science of 2,4,5-T and Associated Phenoxy Herbicides, Wiley, New York, 1980, p. 301. 2. N. Gorden and M. Beroza, Anal. Chem., 24 (1952) 1968. 3. s. Gottlieb and P.B. Marsh, Ind. Eng. Chem., 18 (1946) 16. 4. A. Lechien, P. Valenta, H.W. Nürnberg and C.J. Patriache, Fresenius z. Anal. Chem., 306 (1981) 156. 5. W.H. Gutermann and D.J. Lisk, J. Agric. Food Chem., 11 (1963) 301. 6. S.F. Howard and G. Yip, J. Assoc. Off. Anal. Chem., 54 (1971) 970. 7. s.u. Khan, J. Assoc. Off. Anal. Chem., 58 (1975) 1027. 8. A.S.Y. Chau and K. Terry, J. Assoc. Off. Anal. Chem., 59 (1976) 633. 9. H. Agemian and A.S.Y. Chau, J. Assoc. Off. Anal. Chem., 60 (1977) 1070. 10. R. Bailey, G. LeBel and J.F. Lawrence, J. Chromatogr., 161 (1978) 251. ll. J.J. Richards and J.S. Fritz, J. Chromatogr. Sci., 18 (1980) 35. 12. M.L. Hopper, J. Agric. Food Chem., 35 (1982) 265. 13. H. Roseboom, H.A. Herbold and C.J. Berkoff, J. Chromatogr., 249 (1982) 323. 14. E.G. Cotterill, Analyst (Londen), 107 (1982) 76. 15. F.I. Dauska, J. High Resolut. Chromatogr. & Chromatogr. commun., 7 (1984) 660. 16. B. Blessington, N. crabb, s. Karkee and A. Northage, J. Chromatogr., 469 (1989) 183. 17. T. Tsukioka and T. Murakami, J. Chromatogr., 469 (1989) 351. 18. E.De Felip, A.Di Domenico and F. Volpi, J. Chromatogr., 489 (1989) 404. 19. F. Eisenbeiss and H. Sieper, J. Chromatogr., 83 (1973) 439. 20. J.F. Lawrence and D. Turton, J. Chromatogr., 159 (1978) 207. 21. P. Cabras, P. Diana, M. Meloui and F .M. Pirisi, J.

-169- Chromatogr., 234 (1982) 249. 22. P. Jandera, L. Suoboda, J. Kubát, J. Schvantner and J. Churácek, J. Chromatogr., 292 (1984) 71. 23. J.A. Apffel, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr., 312 (1984) 153. 24. R.D. Voyksner, J.T. Bursey and E.D. Pellizari, J. Chromatogr., 312 (1984) 221. 25. M. Äkerblom, J. Chromatogr., 319 (1985) 427. 26. S.H. Hoke, E.E. Brueggeman, L.J. Baxter and T. Trybus, J. Chromatogr., 357 (1986) 429. 27. B. Blessington and N. crabb, J. Chromatogr., 454 (1988) 450. 28. B. Blessington and N. Crabb, J. Chromatogr., 483 (1989) 349. 29. N.P.Hill, A.E. Macintyre, R. Perry and J.N. Lester, Intern. J. Environ. Anal. Chem., 15 (1983) 107. 30. V. Lopez-Avila, P. Hirata, s. Kraska and J.H. Taylor , Jr., J. Agric. Food Chem., 34 (1986) 530. 31. s. Mierzwa and S. Witek, J. Chromatogr., 136 (1977) 105. 32. J.J. Richards, e.o. Chriswell and J.S. Fritz, J. Chromatogr., 199 (1980) 143. 33. R.L. Smith and D.J. Pietrzyk, J. Chromatogr. Sci., 21 (1983) 282. 34. M.J. Bertrand, A.W. Ahmed, B. Sarrasin and V.N. Mallet , Anal. Chem., 59 (1987) 1302. 35. R.B. Geerdink, J. Chromatogr., 445 (1988) 273. 36. R.B. Geerdink, C.A.A. van Balkom and H.J. Brouwer, J. Chromatogr., 481 (1989) 275. 37. R.B. Geerdink, A.M. V .c. Graumans and J. Viveen, J. Chromatogr., 547 (1991) 478. 38. G. Audunsson, Anal. Chem., 58 (1986) 2714. 39. G. Audunsson, Anal. Chem., 60 (1988) 1340. 40. G. Nilvé, G. Audunsson and J. Jönsson, J. Chromatogr., 471 (1989) 151. 41. P.J .M. Hendriks, H.A. Claessens, Th.M. Noij and F .M. Everaerts, 15th International Symposium on Column Liquid Chromatography, Switzerland, June 3-7, 1991, p95/l. 42. N.J. Dovichi and Y.F. Cheng, Amer. Biotechnol. Labs., 7 (1989) 10. 43. K. Otsuka, s. Terabe and T. Ando, J. Chromatogr., 332 (1985) 219. 44. R.M. Caprioli, W.J. Moore, M. Martin, B.B. Dague, K. Witson and s. Moring, J. Chromatogr, 480 (1989) 429. 45. P.D. Grossman, J.C. Colburn, H.H. Lauer, R.G. Nielson, R.M. Riggin, G.S. Sittampalam and E.C. Rickard, Anal. Chem., 61 (1989) 1186. 46. B.L. Karger, A.S. Cohen and A. Guttman, J. Chromatogr., 492 ( 1989) 585. 47. G.J.M. Bruin, J.P. Chang, R.H. Kuhlman, K. Zegers, J.C. Kraak and H. Poppe, J. Chromatogr., 471 (1989) 429. 48. K.A. Cobb, B. Dolnik and M. Novotny, Anal. Chem., 62 (1990) 2487. 49. C.P. Ong, C.L. Ng, N.C. Chong, H.K. Lee and S.F.Y. Li, J. Chromatogr., 516 (1990) 263. 50. s. Fujiwara, s. Iwasa and s. Honda, J. Chromatogr., 497 (1988) 133. 51. H. Nishi, N. Tsumagari, T. Kakimoto and S. Terabe, J. Chromatogr., 465 (1989) 331.

-170- 52. C.P. Ong, C.L. Ng, H.B. Lee and S.F.Y. Li, J. Chromatogr., 547 (1991) 419. 53. H. Nishi and s. Terabe, Electrophoresis, 11 (1990) 691. 54. w. Thormann, P. Meier, c. Marcolli and F. Binder, J. Chromatogr., 545 (1991) 445. 55. P.G. Pietta, P.L. Mauri, A. Rava and G. Sabbatini, J. Chromatogr., 549 (1991) 367. 56. F.E. Mik.kers, F.M. Everaerts and Th.P.E.M. Verheggen, J. Chromatogr., 169 (1979) 11. 57. J.W. Jorgenson and K.D. Lukas, Anal. Chem., 53 (1981) 1298. 58. s. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya and T. Ando, Anal. Chem., 56 (1984) 111. 59. S. Terabe, K. Otsuka and T. Ando, Anal. Chem., 57 (1985) 834. 60. R.J. Hunter, "Zeta Potential in Colioid Science". Academie Press, Londen, 1981. 61. J.P. Foley, Anal. Chem., 62 (1990) 1302. 62. H.T. Rasmussen, L.K. Goebel and H.M. McNair, J. Chromatogr., 517 (1990) 549. 63. H. Nishi, N. Tsumagari and s. Terabe, Anal. Chem., 61 (1989} 2434. 64. A.T. Balchunas, M.J. Sepaniak, Anal. Chem., 59 (1987} 1466. 65. J. Gorse, A.T. Balchunas, D.F. Swaile, M.J. Sepaniak, J. High Resolut. Chromatogr. & Chromatogr. Commun., 11 (1988) 554. 66. A. Einstein, Ann. Phys., 19 (1906) 289. 67. A. Einstein, Ann. Phys., 34 (1911) 591. 68. R.H. Stokes and R. Mills, Viscosity of Electrolytes and Related Properties, Pergamon Press, London, 1965, Chap. 5.

-171- SUMMARY

Chromatography is of extraordinary practical importance in both analytica! applications and preparative separations. No ether method of instrumental analysis can match its efficiency, speed, reproducibility, and versatility. Chromatographic methods are often the only way to analyze target compounds in different matrices. Biochemistry is another important area that benef its from the application of chromatography and the complementary technique of capillary electrophoresis.

In gas chromatography and electrophoresis the best performance is obtained on capillary columns. It is widely accepted that fused silica is the material of choice for capillary separations. For their application in chromatography it is essential that the column properties are determined by the bulk properties of the stationary phase and not by contributions of residual active sites on the column wall. Deactivation of the column wall and compatibility of the modified surface. with the stationary phase to be coated is essential for obtaining reproducible analytica! results. Constant surface properties are also essential in obtaining a controllable and reproducible electroosmotic flow in electrochromatography (EC).

This thesis describes some recent developments in the preparation of capillary columns for gas chromatography (GC) and EC. The emphasis is on deactivation of the column wall and

-112- crosslinking of the stationary phase. The experimental study of the deactivation process is supported by solid state 29si NMR measurements on fumed silica (Cab-O-Sil) as model substrate. Columns prepared according to the optimized procedures are evaluated both in capillary GC and in EC. chapter 2 describes the synthesis of a series of vinyl modified polymethylhydrosiloxane deactivation reagents containing different functional groups. After deactivation some vinyl groups are left to facilitate crosslinking and/or chemical bonding of the stationary phase. The optimization of the deactivation reaction is investigated and evaluated by GC with a dual column system. Excellent deactivation of the fused silica surface is obtained with the reagents studied. Relatively high temperatures and/or long reaction times are needed for diphenyl and bis-(cyanopropyl) substituted reagents compared to methyl substituted deactivation reagents. Physisorbed water on the inner surface plays an important role in achieving a thin and densely crosslinked layer on the column wall. compared to methyl deactivated columns an increase of the Kováts retention indices of polar test compounds is observed. This indicates a successful incorporation of polar functional groups in the deactivation layer.

The effect of the silylation temperature and the reaction time on the deactivation of silica surfaces is investigated using

-173- solid state 29si NMR spectroscopy (Chapter 3). The reaction between Si-H and Si-vinyl groups by thermal treatment is confirmed by NMR experiments using fumed silica as model substrate. The existence of vinyl groups on the silylated cab­ o-sil samples, however, is not evident, due to the low amount

of vinyl groups in the copolymer and the lack of sensitivity

of the 29si NMR technique. The spectra obtained show that the

deactivation layer is stable up to at least 360°C.

The preparation of non-polar and medium polar capillary

columns based on deactivation reactions described in previous

chapters is presented in Chapter 4. The crosslinking and/or chemica! bonding of various stationary phases by thermal

treatment or azo-tert-butane (ATB) initiation is investigated.

The immobilization of vinyl modified non-polar stationary

phases can be achieved by thermal treatment. This results in

very inert columns and has minimal ef f ects on the retention behaviour. To obtain a sufficiently high degree of

immobilization of non-vinyl containing stationary phases, a

combination of ATB initiation and a vinyl modified

deactivation layer is required.

Non-polar narrow bore capillary columns down to approximately

20 µm i.d. could be prepared using the procedure described in

this chapter. Highly inert and efficient columns were obtained.

In chapter 5, the applicability of the prepared non-polar and

medium polar fused silica capillary columns is illustrated by

different examples. High speed GC separations are performed

-174- on 'narrow-bore capillary columns with inner diameters of 20 µ:m. Columns with high thermal stability were successfully prepared for the analysis of less volatile compounds, such as poly:mer additives, glycerides and polyethylene glycol 400. The successful separation of some underivatized steroids, drugs, priority pollutants and derivatized amino acids de:monstrates the high degree of inertness of the columns prepared according to the methods described here.

In chapter 6, a systematic investigation is carried out to study the electroosmotic flow (EOF) in micellar electrokinetic capillary chro:matography (MECC). The chemical nature of the modification layer is shown to have a strong influence on the :magnitude of the EOF. This can be explained by studying in detail the specific influence of the modification layer on the r-potential. The results from untreated and treated columns were co:mpared under the sa:me experimental conditions. Further:more, experiments with and without sodiu:m dodecylsulfate (SDS) addition to the buffer were carried out in order to elucidate the effect of this surfactant on the EOF. Under the proper experimental conditions reproducible EOF values with a relativa standard deviation of about 3% can be obtained using a home-built instrument.

In chapter 7, the selectivity of MECC for the separation of herbicides consisting of chlorophenoxy acids is de:monstrated. Tuned selectivity was obtained by optimizing parameters such as the concentration of SDS and Brij 35 and the percentage of

-175- methanol in the buffer. SOS combined with Brij 35 as micellar agents were found to provide the best overall separation of these components. The analysis of the herbicides was carried out using multi­ wavelength detection. The characterization of the sample zones by their absorption spectra and migration times proved to be a powerful tool for solute identification. In this particular MECC separation system, high efficiencies (more than 100,000 theoretical plates) were obtained. Relative standard deviations for migration times of the components were about 1% and for peak areas approximately 2%.

-176- SAMENVATTING

Chromatografie is van uitzonderlijk belang in het analytisch laboratorium, met name in combinatie met spectroscopische identificatietechnieken, in het bijzonder massaspectrometrie. Preparatieve chromatografie wordt meer en meer toegepast voor de scheiding van produkten met hoge toegevoegde waarde in de farmaceutische en biotechnologische industrie. Geen enkele andere analysetechniek benadert het scheidend vermogen, snelheid van analyse en de nauwkeurigheid van kwantitatieve resultaten. De verscheidenheid aan toepassings­ mogelijkheden is ongeëvenaard. Vaak zijn chromatografische methoden de enige mogelijkheid om "doelcomponententt te bepalen in ingewikkelde matrices, b.v. urine, bloed, water en bodemmonsters. Milieuresearch en monitoring is meestal alleen mogelijk door chromatografische spor·enanalyse uitgevoerd met zeer hoge gevoeligheid en scheidend vermogen. Biochemie en biotech­ nologie zijn twee andere belangrijke gebieden waarvan de vooruitgang mede afhangt van de toepassing van chromatogra­ fische en de verwante.en aanvullende techniek, elektroforese. Zowel voor gaschromatografie als elektroforese geldt dat met capillaire kolommen de beste resultaten worden verkregen. Sinds 1979 worden voornamelijk de door Dandeneau geïntroduceerde 'fused silica' kolommen gebruikt. Het is van wezenlijk belang dat de chromatografische eigenschappen van een kolom worden bepaald door de toegepaste stationaire fase en niet door aktieve plaatsen op de kolomwand. Daartoe wordt voor het opbrengen van de stationaire

-177- fase de kolomwand gedeactiveerd. Voor de spreiding van de stationaire fase op het gedeaktiveerde kolomoppervlak is het van groot belang dat de polariteit van de wand is aangepast aan die van de op te brengen laag. Ook bij de toepassing van 'fused silica' kolommen in de electrochrOlllatografie (EC) zijn konstante oppervlakte- eigenschappen essentieel, zij bepalen namelijk de grootte en richting van de elektro-osmotische stroming. Dit proefschrift beschrijft nieuwe methoden voor de bereiding van 'fused silica' capillaire kolommen voor zowel gaschromato­ grafie als micellaire elektrokinetische capillaire chromatografie (MECC).

In Hoofdstuk 2 van dit proefschrift wordt de synthese van een aantal vinylgemodif iceerde polymethylhydrosiloxaan deacti­ veringsreagentia besproken. De bij deactivering met dit reagens in de deactiveringslaag ingebouwde vinyl-groepen vereenvoudigen de crosslinking en het chemisch verankeren van de stationaire fase. De kinetiek van de deactiveringsreacties wordt bestudeerd en geoptimaliseerd met twee-dimensionale capillaire GC. Het blijkt dat deactivering met diphenyl en bis-{cyanopropyl) gesubstitueerde reagentia hogere deacti­ veringstemperaturen en/of langere reactietijden vereist dan reacties met methyl gesubstitueerde reagentia. Verder blijkt gephysisorbeerd water op het kolomoppervlak een essentiële rol te spelen in het deactiveringsproces. Met de bestudeerde reagentia kan een hoge mate van inertheid van het fused silica oppervlak bereikt worden.

-178- In Hoofdstuk 3 worden vaste stof 29 si NMR technieken toegepast voor de bestudering van de invloed van de silylerings­ temperatuur en de reactietijd op de deactivering van fused silica oppervlakken. Met NMR experimenten aan het model substraat 11 fumed" silica (Cab-O-Sil) kan de reactie tussen si-H en Si-vinylgroepen bij verhoogde temperaturen gevolgd worden. De aanwezigheid van vinylgroepen op de gesilyleerde Cab-O-Sil monsters kan met NMR door het percentage vinylgroepen in het uitgangsmateriaal en de relatief lage gevoeligheid de 29si NMR, niet onomstotelijk bewezen worden. Wel kan worden aangetoond dat de verkregen deactiveringslaag stabiel is tot een temperatuur van tenminste 360°C.

Het onderwerp van Hoofdstuk 4 is de bereiding van niet-polaire en zwak-polaire capillaire kolommen gebruikmakend van de deactiveringsreagentia bestudeerd in de voorgaande hoofd­ stukken. Een belangrijk onderwerp beschreven in dit hoofdstuk is de vergelijking van thermisch geïnitieerde en azo-tert­ butane geïnitieerde crosslinking en chemische verankering van diverse stationaire fasen. Thermisch geïnitieerde crosslinking van vinyl-gemodificeerde stationaire fasen geeft zeer stabiele en inerte kolommen en heeft een verwaarloosbare invloed op de retentiegeigenschappen van de stationaire fase. Voor het verkrijgen van een vergelijkbare deactivering op een vinyl vrije stationaire fase is een combinatie van ATB initiatie met een vinyl gemodificeerde deactiveringslaag vereist. De in dit hoofdstuk beschreven procedures bleken bij uitstek geschikt voor de bereiding van narrow-bore capillaire kolommen met diameters tot ongeveer 20 µm.

-179- In Hoofdstuk 5 wordt de toepasbaarheid van de kolommen, bereid via de procedures beschreven in de voorafgaande hoofdstukken, aan de hand van een aantal applicaties geïllustreerd. In dit hoofdstuk worden enkele zeer snelle scheidingen uitgevoerd met 20 µm kolommen. Verder worden de bereide kolommen succesvol toegepast voor de hoge temperatuurscheiding van zeer hoog­ kokende componenten zoals polymeer additieven, glyceriden en polyethyleen glycol 400. De goede resultaten verkregen bij de scheiding van ongederivatiseerde steroïden, alkaloïden en de zogenaamde 'priority pollutants' is een bewijs voor de hoge mate van inertheid van de kolommen.

In hoofdstuk 6 is het onderzoek beschreven naar de effecten van verschillende coatings van capillaire kolommen op het elektroosmotisch debiet in micellaire elecktrokinetische capillaire chromatografie (MECC). Gebleken is dat het physisch-chemisch karakter van de kolomwand een sterke invloed heeft op de grootte van de electroosmotische snelheid (EOF) in de kolom. Een verklaring hiervoor kan worden gegeven door de polariteit van de verschillende niet-gecoate en gecoate kolommen in beschouwing te nemen. Daarnaast is ook het effect bestudeerd van de toevoeging van het surfactant natriumdodecylsulfaat (SOS) aan de buffer op de EOF. In dit deel van het onderzoek werd aangetoond dat onder de toegepaste experimentele kondities reproduceerbare EOF-waarden met een relatieve standaardafwijking van 3% te realiseren zijn.

-180- In hoofdstuk 7 wordt, aan de hand van de scheiding van herbiciden, de toepassing van verschillende surfactanten in de buffer en de invloed hiervan op de selektiviteit in MECC behandeld. Daarnaast is ook de rol van de toevoeging van methanol aan de buffer op de selektiviteit onderzocht. De effecten van verschillende concentraties van de surfactanten SDS en Brij-35 en de organische modifier methanol in de buffer werden bestudeerd. Gebleken is dat combinaties van de verschillende typen en concentraties van deze additieven aan de buffer de selec­ tiviteit in MECC sterk kunnen verbeteren. Met betrekking tot de kwalitatieve en kwantitatieve analyse bleken de relatieve standaardafwijkingen van migratietijden en piekoppervlakken bij de scheiding van deze herbiciden, respectievelijk 1 en 2%. samenvattend, blijkt MECC een potentieel krachtige scheidings­ methode door de mogelijkheden tot beïnvloeding van de selektiviteit in combinatie met de hoge efficiency, die met deze techniek te realiseren is.

-181- ACDOWLEDGEMENT

At the moment of releasing of this thesis, I feel obliged to acknowledge all of thosepeople, without whose contributions, direct or indirect, I would have been unable to finish this work.

First I would like to express my gratitude to Prof.Or. Carel cramers, who gave me the opportunity to carry out my doctorate research at the Eindhoven University of Technology. It is his liberal way of guidance and his encouragement throughout the investigations that led to the completion of this thesis. Special thanks are going to Jacques Rijks sr. and Henk Claessens for their permanent support. I additionally thank Jacques Rijks sr. for his help in introducing me to the "world of chromatography" • His help in personal matters is kindly appreciated.

I thank Hans-Gerd Janssen, Martin He tem, Robert Houben, Gerard Rutten, Jacek Staniewski, Andrew van Es, Dana Conron, Hans Mol, Andrea Houben, Pieter Hendriks, Marion van straten, Janet Eleveld, Hans Verhoeven, Pavel Coufal, Pim Muijslaar, Marc Bergman, Door Platje, and Gerlie Heemels for the many stimulating discussions on separation sciences. Jacques Rijks, Hans-Gerd Janssen, Henk Claessens, Jan de Haan are further acknowledged for their help in revising the first concept of this thesis and the published papers.

I am obliged to Denise Tjallema, Hans van Rijsewijk, Jack

-182- Rijks jr., Henri snijders, Anja Huygen, Anton Bombeeck, Piet Leclercq, Harry Maathuis and Dick Kroonenberg for their kindness and help. Special thanks to Huub van Leuken for the construction of the electrophoresis system.

I cordially thank Rongrong, my wife, and Fanda, my daughter for their patience and their unreserved support for my work.

Finally, I acknowledge all members and students of the Laboratory of Instrumental Analysis of Eindhoven University of Technology, who in one way or another contributed to my work and my stay in The Netherlands.

Quanji Wu

-183- ~ 1 CURRICULUM VJ:Tll

3 $" The author was born in Liuzhou, Guangxi Province, China, on December 2, 1956. He finished his middle school education in 1973 and entered the Guangxi University in 1978, after 4 years study, he obtained his Bachelor degree. He continued his study in 1978 at the Lanzhou Institute of Chemica! Physics, belonging to the Chinese Academy Science. There he obtained his Master degree in ~ In 1987 he came to the Eindhoven University of Technology in The Netherlands, where he started his Ph.D. research work which led to this thesis.

-184- PtJBLICATIOHS

1. Q.J. wu and W.L. Yu, Analytical Instruments (Chinese), 1987, (1) 35 (Ch.). "Characterization of Hall Electrolytic conductivity detector for GC 11 • 2. Quanji Wu, Q.Y. Ou and W.L. Yu, Fenxi Huaxue (Chinese), 1987, 15(5) 390 (Ch.). "The selectivity and the response factor of Hall electrolytic conductivity detector for GC 11 • 3. M. Hetem, Q. Wu, J. Rijks, L. van de Ven, G. Rutten, J. de Haan and C. Cramers, Proc. of 17th International Symposium on Chromatography, Vienna, Austria, september 1988, I, P32. "Deactivation of silica surface with polyhydrosiloxanes: a quantitative evaluation by chromatography and 29si NMR". 4. Q. wu, C.A. Cramers and J .A. Rijks, Proc. ·of lOth International Symposium on capillary Chromatography, Riva del Garda, Italy, May 1989, P. 188. "Preparation of capillary columns with vinyl modified polyhydrosiloxane deactivation reagents-characterization and optimization of the deactivationtt, 5. Q. Wu, M. Hetem, C.A. Cramers and J.A. Rijks, J. High Resolut. Chromatogr., 13 (1990) 811. 11 Preparation of thermally stable phenylpolysiloxane fused silica capillary columns-optimization and evaluation of the deactivation by capillary GC and solid state 29si NMR". 6. Q. Wu, C.A. Cramers and J.A. Rijks, Proc. of 13th International Symposium on Capillary Chromatography, Riva del Garda, Italy, May 1991, P. 176. "Some aspects of crosslinking and chemical bonding of stationary phases in the preparation of non-polar and medium polar GC capillary columns". 7. G.A.F. Rutten, Q. Wu, J.A. Rijks and C.A. Cramers, Proc. of 13th International symposium on Capillary Chromatography, Riva del Garda, Italy, May 1991, P. 143. "Deactivation of fused silica capillaries with polydi­ phenylvinylmethyl-hydrosiloxane copolymer. Optimization of reaction conditions using a new modification of intermediate test". 8. Q. Wu, H.A. Claessens and C.A. cramers, "The influence of surface treatment on electroosmotic flow in micellar electrokinetic capillary chromatography". Chromatographia, in press. 9. Q. Wu, H.A. Claessens, C.A. cramers, "The separation of Herbicides by micellar.electrokinetic chromatography". Submitted for publication.

-185- STELLINGEN

1. The contribution of the deactivation layer to the retention behaviour of solutes cannot be neqlected.

-This thesis, chapter 2.

2. The ratio of the mobility and the diffusion coefficient of a solute in the equation for plate num.ber N = µE / 2D is not an intrinsic property of the solute, independent of buffer properties.

-J.W. Jorqenson and R.D. Lukacs, Anal. chem., 53 (1981) 1298. -H.'1'. Rasmussen and H.M. McNair, J. chromatoqr., 516 (1990) 223. -H.J. Issaq, I.Z. Atamna, G. M. Muschik and G.M. Janini, chromatographia, 32 (1991) 155.

3. The statement that there are no exposed Si-H bonds on the column surface after deactivatinq fused silica capillary columns with polymethylhydrosiloxane is incorrect.

-C.L. woolley, R.E. Markides and M.L. Lee, J. Chromatoqr., 367 (1986) 23. -M. He tem, G. Rutten, B, Vermeer, J. Rijks, L. van de Ven, J. de Haan and c. cramers, J. chromatoqr., 477 (1989) 3.

4. The conclusion that in micel lar electrokinetic chromatography the effect of a temperature gradient on plate height is negligible is based on a series of incorrectly derived equations.

-s. Teral:>e, R. otsuka and T. Ando, Anal. Chem., 61 (1989) 251. -s. Teral:>e, K. Otsuka and T. Ando, Anal. Chem., 57 (1985) 834.

5. Capillary columns coated with polydimethylsiloxane, a non-polar stationary phase, result in an increased electroosmotic flow in micel lar electrokinetic chromatography. This is not expected because silanol groups on the column surface are covered by the non­ polar coating.

-This thesis, chapter 6.

6. Hanai's observation that the migration rate of fructose increases with the concentration of sodium chloride is in contradiction with theory.

-T. Hanai, H. Hatano, N. Nimura and T. Kinoshita, J. High Resolut. Chromatoqr., 14 (1991) 481.

7. Quantitative 29si CP-MAS NMR measurements of derivatized silica surfaces require more precautions than published by Köhler c.s. 1). In fact, the cross-polarization (CP) characteristics should be measured and accounted for explicitly.

-J. Köhler, D.B. Chase, R.D. Furlee, A.J. Vega and J.J. Rirkland, J. Chromatoqr., 352 (1986) 275. -B. Pfleiderer, K. Albert, E. Bayer, L. van de Ven, J. de Haan and c. cra.mers, J. Phys. Chem., 94 (1990) 4189.

8. The contribution of the term describing resistance to mass transfer in the liquid phase cannot be neglected in vacuwn outlet capillary gas chromatography.

-N. Bel.lier and G. Guiochon, J. Chromatoqr. sci., 8 (1970) 147.

9. A linear behaviour of the logarithm of the capacity ratio of a solute vs. the percentage modifier is not observed in supercritical chromatography even if the action of modifiers is solely due to interactions between the modifier and the stationary phase.

-L.J. Mul.cahey and L.T. Taylor, J. Hiqh Résolut. Chromatoqr., 13 (1990) 393.

10. Modification is an important factor in science and technology but also in society. This is clearly reflected by the food served in most Chinese restaurants in The Netherlands, which is quite different frorn the original.