Introduction Cyclodextrins in HPLC Applicationsindex-of.co.uk/Tutorials-2/CHIRAL SEPARATIONS...chromatography was as mobile phase additive in thin-layer chromatography. In the mid-1980’s

Figure 1 Numbering of carbon atoms in the cyclodextrin ringstructure.

Cyclodextrins and Other Inclusion Complexation Approaches

J. Dingenen, Janssen Research Foundation, Belgium

Copyright^ 2000 Academic Press

Introduction

Cyclodextrins are cyclic nonreducing oligosacchar-ides containing from six to twelve glycose units ina C-1 chair conformation, bonded through �-(1,4)linkages (Figure 1). The glycopyranose units arearranged in the shape of a hollow truncated cone. Thelarger opening of the molecule is surrounded bythe secondary (C-2 and C-3) hydroxyl groups, whilethe primary (C-6) hydroxyl groups constitute thesmaller end of the cone.

Since all the primary and secondary hydroxylgroups are located at the outside of the molecule, theexterior faces are hydrophilic. The interior cavity,essentially comprised of methylene linkages andglycosidic oxygen bridges, is relatively hydrophobicin comparison with polar solvents such as water.Furthermore, the glycosidic oxygen bridges producea high electron density, giving the interior of thecavity a slightly Lewis-base character.

The three smallest cyclodextrin homologues arereadily commercial available:

� �-cyclodextrin (cyclohexamylose);� �-cyclodextrin (cycloheptamylose);� �-cyclodextrin (cyclooctamylose);

The basic property of cyclodextrins is their ability toform selective inclusion complexes with a broad var-iety of organic and inorganic molecules.

The formation of inclusion complexes is in generaldetermined by the ability of the guest molecule toclosely Rt the cavity of the cyclodextrin. However, thepolarity of the guest molecule also plays an importantrole.

Inclusion complexes are usually formed in the pres-ence of water or in water mixed with organic modi-Rers.

One of the Rrst effective uses of cyclodextrins inchromatography was as mobile phase additive inthin-layer chromatography. In the mid-1980’s a pro-cess was developed to produce stable cyclodextrinhigh performance liquid chromatographic (HPLC)phases. Nowadays, native �-, �- and �-cyclodextrin,as well as a variety of derivatized cyclodextrin HPLCcolumns are commercially available. Also many cyc-lodextrin-based capillary gas chromatography (GC)columns are on the market. With the growing import-ance of capillary electrophoresis in chiral separations,the use of native cyclodextrin and cyclodextrin deriv-atives as an electrolyte additive steadily increases.

Cyclodextrins in HPLC ApplicationsNative cyclodextrin HPLC columns were deliberatelydesigned to be used in the reserved-phase mode ofoperation, in order to take full advantage of thehost}guest complexation capabilities of the molecule.

In a more recent and somewhat different experi-mental approach, the inclusion properties are sup-pressed by using a non-hydrogen bonding, polar or-ganic solvent (e.g., acetonitrile) as the main compon-ent of the mobile phase. Acetonitrile has the tendencyto occupy the cavity and seems to enhance hydrogenbonding between the hydroxyl groups on the cyc-lodextrin and hydrogen bonding groups on the chiralanalyte.

In this so-called polar organic mode of operation,the addition of small amounts of glacial acetic acidand triethylamine is used as a tool to enhance enan-tioselectivity. On the other hand, the addition ofa hydrogen bonding solvent such as methanol allowsreduced retention of strongly retained molecules. Thistechnique produces some unusual enantioselectivitiesthat certainly enhance the usefulness of native cyclo-dextrin phases.

Furthermore, a variety of cylcodextrin derivativeshas been immobilized on a chromatographic support,which can be used under normal as well as underreversed-phase conditions. The most popular com-mercially available derivatized cyclodextrin chiralselectors are listed in Table 1.

Mobile Phase Design and Parameter Optimization

Because complex stability constants usually havegreater values in water or water}organic organic sol-vent mixtures than in a pure organic medium, nativecyclodextrin columns are predominantly used inthe reserved-phase mode. Ethanol, propanol, 1,4-dioxane, dimethyl sulfoxide, dimethyl formamide,

III / CHIRAL SEPARATIONS / Cyclodextrins and Other Inclusion Complexation Approaches 2335

Table 1 Commercially available derivatized cyclodextrin phases

R Trade name, Advanced Separation Technologies

Cyclobond� I 2000 Ac Acylated

Cyclobond� I 2000 RSP Hydroxypropyl etherCyclobond� I 2000 SP

Cyclobond� I 2000 RN Naphtylethyl carbamateCyclobond� I 2000 SN

Cyclobond� I 2000 DMP 3,5-Dimethylphenyl carbamate

Cyclobond� I 2000 PT p-Toluoyl

methanol and acetonitrile have been used as organicmodiRers. However, methanol and acetonitrile aremost commonly used. It is difRcult to predict in ad-vance which of these two modiRers will produce thebest separation in any given case. In many cases, pHand ionic strength of the aqueous part of the mobilephase are even more important than the choice of theorganic modiRer.

The following factors inSuence enantioselectivity.

Ionic strength With an eluent composed of a mix-ture of 5 millimolar tetrabutylammonium hydrogen-sulfate (TBAHS) in water and methanol in an 80}20volume ratio, experiments were performed on a�-cyclodextrin Astec column (Cyclobond�) and asimilar Merck column Merck (Chiradex�). The onlydifference between both stationary phases is thechemistry used to attach the �-cyclodextrin to thesilica matrix.

In these initial experiments, especially on theCyclobond� column, a poor peak shape for differentproducts was observed. This effect was less pro-nounced on the Chiradex� column but it also occur-red for some products. The origin of the poor peakshapes can have different causes, as for example aninsufRcient shielding of residual silanol groups.Therefore, the effect of the tailing reducer concentra-tion was investigated.

Figure 2 illustrates the effect of the tailing reducerconcentration on the retention factor and the resolu-tion of miconazole (R14889). In this Rgure, the res-olution factor based on the location of the valleypoint between two peaks is used. The resolution para-meter (AuSoK sung (�)) is deRned to evaluate the degreeof separation between two partially resolved enantio-mers. This measuring principle is based on the loca-tion of the valley point between two adjacent peaks.It is a useful and, from a measurement viewpoint,

2336 III / CHIRAL SEPARATIONS / Cyclodextrins and Other Inclusion Complexation Approaches

Figure 2 Influence or tailing reducer concentration on the re-tention factor and valley point resolution �, Cyclobond] ;

, Chiradex] . Experimental conditions: column: 25 cm�4.5 mmID chemically bonded cyclodextrin phase; mobile phase: TBAHSin water at different concentrations}methanol (80}20, v/v); flowrate: 1 mL min�1. Solute:

easy evaluation technique for characterizing theseparation degree between two peaks.

If we analyse Figure 2, it is easy to conclude thationic strength has a strong effect on the chromato-graphic behaviour of the investigated chiral molecule.It is furthermore perfectly clear that the inSuence ofionic strength on the retention factor is more pro-nounced on the Chiradex� than the Cyclobond� col-umn. However, for both types of stationary phasea constant value is observed above a tailing reducerconcentration of 20 mmol. At low TBAHS concentra-tion, the smallest resolution values are measured on

the Cyclobond� column. Also, the ionic strength re-quired to reach the highest resolution value is differ-ent for both columns.

Further experiments dealing with the effect of ionicstrength on peak shape and chromatographic behav-iour have demonstrated that for each individual col-umn type}tailing reducer combination it can be veryhelpful to investigate this parameter thoroughly.However, to save time, we nowadays start our chiralmethod development work on cyclodextrin-basedcolumns with tailing reducer concentrations between30 and 50 mmol, because we have experienced thatwith these values there is in general a good chance ofbeing successful.

Type of tailing reducer The acetic acid salt oftriethylamine is popular as tailing reducer in reversed-phase chromatography. Instead of using acetic acid toadjust the pH value of an aqueous triethylamine solu-tion, we investigated the usefulness of some otherorganic and inorganic acids. At Rrst we did not expectany inSuence of the type of counterion on enan-tioselectivity, but some preliminary experimentsshowed a distinct effect, worthwhile to investigatefurther. Therefore, we examined about 30 productsbelonging to different product classes. As mobilephase, a mixture of 20 vol% of methanol and 80vol% of 50 mmol aqueous triethylamine solution ad-justed to a pH value of 2.5 with respectivelyhydrochloric, hydrobromic, phosphoric, perchloric,sulfuric, triSuoroacetic and oxalic acids was used.

For all the investigated products, the largest reten-tion factors were observed for the triethylamine solu-tion adjusted to the desire pH value with sulfuric andoxalic acids. Also with sulfate as counterion, thelargest number of investigated products was partiallyor completely resolved. Compared to the other acids,with triSuoroacetic and perchloric acid a smallernumber of products were separated. To further inves-tigate the observed effect, we repeated the same ex-periments for a series of azoles, which differed onlyslightly in structure. For these substances, the effect ofthe counterion on resolution is illustrated in Figure 3.See also Table 2.

For this product series, the results conRrm theinitial observations. Furthermore, Figure 3 clearlydemonstrates that even small change in molecularstructure, as for example the number or position ofthe chorine atoms on one of the phenyl groups of themolecule, can have a tremendous effect on the enan-tioselectivity.

Although for the investigated series of azoles pHadjustment with sulfuric acid in general resulted ingood peak shapes and high resolution values, onepeculiarity was observed. Under these experimental


Figure 3 Azoles: effect of the counterion on resolution. Experimental conditions: column: 25 cm�4.6 mm ID Cyclobond� (Astec);mobile phase: 50 mM triethylamine in water adjusted to pH 2.5 with different acids}methanol (80}20, v/v); flow rate: 1 mL min�1,solutes; see Table 2.

conditions, R78575 was strongly retained and bothenantiomers eluted as broad peaks with an irregularshape (Figure 4). On the other hand, pH adjustmentof the triethlmaine solution with the other acidsresulted in normal peak shapes (Figure 5). This isan indication that minor differences in structure, orsmall variations in experimental conditions, can havea tremendous effect on the chromatographic behav-iour of enantiomers.

A comparable set of experiments was subsequentlyperformed using tetrabutylammonium hydroxideinstead of triethylamine as tailing reducer. In this setof experiments, oxalic acid was excluded and per-chloric acid could not be used owing to the formationof an insoluble salt with tetrabutylammonium hy-droxide in the water}methanol mixture. For thewhole test series of 24 different substances, sulfuricacid could be identiRed as the counterion that gener-ated the highest resolution values. From a practicalpoint of view, it is certainly an advantage that sulfateis the anion of choice. For temperatures betweenroom temperature and 503C the corrosion of stainlesssteel (316 or equivalent) is negligible for dilute sulfur-ic acid solutions, while the chemical resistance of thismaterial for chloride ions (even in low concentra-tions), is rather limited.

In conclusion, the anionic part of the tailing re-ducer has a clear effect on enantioselectivity, as illus-

trated previously. The inSuence of the counterion onthe retention factor and resolution of the investigatedcompounds can possibly be related to a difference inability of the counterion to compete with the organicguest molecule for the hydrogen bonding sites of thecyclodextrin host.

To complete the experiments concerning the inSu-ence of the type of tailing reducer, we also found itnecessary to do some tests in which the anionic partof the tailing reducer was kept identical while thecationic part was varied. In these experiments,a 30 mmol solution of tetramethyl-, tetraethyl-, tetra-propyl- and tetrabutylammonium hydroxide solutionwas adjusted with sulfuric acid to a pH value of 2.5and used as mobile phase in an 80}20 volume ratiowith methanol as organic modiRer.

The smallest resolution values are observed fortetramethylammonium hydroxide and the highestvalues for tetrabutylammonium hydroxide, but com-pared with the inSuence of the anionic part of thetailing reducer the differences are far less pro-nounced. However, for the investigated test series, thetailing factor measured at 10% of the peak heightgradually decreases with increasing chain length ofthe alkylammonium hydroxide.

pH Whatever type of silica-based reversed phasesare used, these materials always display some acidic


Table 2 Structures of the investigated azole derivatives

Azoles

Product R

T824 H

R14827 (econazole)

R14889 (miconazole)

R24095 (isoconozole)

R78575

R78690

Figure 4 Peak shape of the different azoles. Experimental con-ditions: column: 25 cm�4.6 mm ID Cyclobond� (Astec); mobilephase: 50 mM triethylamine in water adjusted to pH 2.5with sulfuric acid}methanol (80}20, v/v); flow rate: 1 mL min�1;solutes: see Table 2.

surface area properties owing to residual hydroxylgroups on the silica surface. During the analysis ofbasic substances not only hydrophobic interactionstake place, but also acid}base interactions betweenthese acidic groups and basic functions in the analytecan be expected to occur. This type of interactionoften results in increased retention combined withpeak tailing. One of the possibilities to solve thisproblem is to adjust the pH below the pKa value ofthe sample that has to be analysed. At pH valueslower than pKa!2 the basic function is protonatedand a salt is formed which no longer has basic proper-ties.

Because the cyclodextrin columns are used underreversed-phase conditions, it is certainly worthwhileto investigate the effect of pH on enantioselectivity.For some monobasic molecules, the effect of pH onthe retention factor is depicted in Figure 6.

Figure 6 clearly illustrates that the retention factorof the investigated products follows a typical rever-sed-phase pattern. At higher pH values the degree ofprotonation of the analyte diminishes. As a result thechance for hydrophobic interactions with the station-ary phase increases and higher retention values aremeasured. When the pH is equal to the pKa valueof the sample, the number of protonated andnonprotonated molecules is equal. Small changes inpH around this value will immediately have an effecton the retention factor. Even the effect of small struc-tural changes, causing some differences, in hydropho-bic nature of the nonprotonated solute molecules, canbe clearly observed.

R8110 has the smallest retention factor owingto the presence of a Suorine atom on the phenylgroup, giving the molecule a more polar characterthan the other two members of the test series. Thelargest retention factors are measured for R7405bearing an ethyl group on the ester function attachedto the imidazole ring instead of a methyl group forR7315.

To better visualize the effect of pH variationson enantioselectivity we often use for the graphical


Figure 5 Peak shape of R78575. Experimental conditions: col-umn: 25 cm�4.6 mm ID Cyclobond� (Astec); mobile phase:50 mM triethylamine in water adjusted to pH 2.5 with differentacids}methanol (80}20, v/v); flow rate: 1 mL min�1; solute:

Figure 6 Effect of pH on rentention factor. �, R7315;, R7405; �, R8110. Experimental conditions: column:

25 cm�4 mm ID Chiradex� (Merck); mobile phase: 50 mMtriethylamine in water adjusted to different pH values with sulfuricacid}methanol (80}20, v/v); flow rate: 1 mL min�1; solutes:

representation of experimental data the degree ofprotonation of the solute molecules instead of pHvalues. For monobasic substances the ratio betweenprotonated and nonprotonated molecules at a certainpH value can be easily calculated:

R}NH#3 �R}NH2#H#

Ka"[R}NH2] ) [H#]

[R}NH#3 ]

pKa"pH#log[R}NH#

3 ][R}NH2]

(Henderson}Hasselbalch equation)

pKa!pH"log[R}NH#

3 ][R}NH2]

From:

10[pKa�pH]"[R}NH#3 ]

[R}NH2]

and:

[R}NH#3 ]#[R}NH2]"100


Figure 7 Effect of degree of protonation of the analyte on res-olution. �, R7315; , R7405; �, R8110. Experimental conditions:column: 25 cm�4 mm ID Chiradex� (Merck); mobile phase:50 mM triethylamine in water adjusted to different pH valueswith sulfuric acid}methanol (80}20, v/v); flow rate: 1 mL min�1;solutes:

Figure 8 Effect of pH on the degree of protonation and resolu-tion of a dibasic substance.N, Alpha 0; } } }, alpha 1; - - - -, alpha2; �, TBA-OH; , TEA. Experimental conditions: column:25 cm�4 mm ID Chiradex� (Merck); mobile phase: 50 mMtriethylamine in water adjusted to different pH values with sulfuricacid}methanol (80}20, v/v); and 30 mM tetrabutylammonium hy-droxide in water adjusted to different pH values with sulfuricacid}methanol (80}20, v/v); flow rate: 1 mL min�1; solute;

it is possible to easily calculate for each pH value theratio between protonated and free base. For R7315,R7405 and R8110 the valley point resolution versusthe percentage protonation is depicted in Figure 7.

For the investigated product series, it is clear thatunder the experimental conditions applied above acertain degree of protonation of the solute moleculesthe enantioselectivity strongly decreases. Probablydue to an increase in hydrophilic character of theprotonated molecules, the strength of the hydropho-bic interactions between the analyte and the cavity ofthe cyclodextrin molecule diminishes, with a reducedenantioselectivity as a result. Within the investigatedpH range, triethylamine (pKa"10.72) always re-mains fully protonated. Therefore, competition forhydrophobic interactions with the cyclodextrin cavitybetween the tailing reducer and the solute moleculeshas to be considered as nonexistent and can be ex-cluded as a possible reason for reduced enantioselec-

tivity at higher pH values. Besides investigations onthe effect of pH using triethylamine as basic sub-stance in the aqueous part of the mobile phase, ex-periments have also been performed with differenttetraalkylammonium hydroxides adjusted to thedesired pH value with sulfuric acid. Within thesame pH range, tetrabutylammonium hydroxide andtriethylamine displayed comparable patterns whenthe valley point resolution was plotted against thepercentage of protonation of the solute molecules.

For a dibasic substance (R60844) with respectivelypKa values of 5.4 and 6.7 the protonation patternversus pH, together with the valley point resolutionfor two types of mobile phase additives, is given inFigure 8.

Although for the dibasic product R60844 bothbasic functions remain fully protonated up to a pHvalue of 4, the resolution continuously increasesbetween pH 2 and 4. However, above pH 4 theenantioselectivity rapidly decreases.

For the different product series that have beenexamined we have observed that under the sameexperimental conditions, in general, higher resolutionvalues are obtained with triethylamine than with


Figure 9 Influence of temperature on the retention factor.�, R7315; �, R7405; �, R60844; , R23979. Experimentalconditions: column: 25 cm�4 mm ID Chiradex� (Merck); mobilephase: 30 mM triethylamine hydroxide in water adjusted to differ-ent pH values with sulfuric acid}methanol (80}20, v/v); flow rate:1 mL min�1; temperature range: 1}403C; solutes:

tetrabutylammonium hydroxide as mobile phaseadditive.

In conclusion, the stability of the inclusion com-plexes formed between the solute molecule and thecyclodextrin seem to be dependent on the charge ofthe guest molecule. Therefore, the retention as well asthe degree of separation of molecules bearing anionizable acidic or basic group can be affected bychanging the pH. However, the effect of pH vari-ations is not so easy to predict.

In the experiments performed, some products(R7315, R7405, T824, etc.) display the highest res-olution values when they are present as free base or asa partially charged molecule. Other products(R60844, most of the azoles) reach maximum resolu-tion when they are completely or nearly completelyprotonated. For some products the effect of pH vari-ations within a rage of 2}3 pH units is small, whilefor others extreme effects can be observed. Further-more, for different types of tailing reducers the res-olution versus pH proRles can have a different shape.

Therefore, the different experiments that have beenperformed to investigate the effect of pH on enan-tioselectivity clearly demonstrate that besides host}guest complexation interactions between the solutemolecule and the cyclodextrin cavity, the hydroxylgroups at the outside of the cyclodextrin moleculetogether with reversed phase and other less predict-able types of interactions certainly play an importantrole in the chiral recognition process.

Because small changes in pH can have a tremen-dous inSuence on the enantioselectivity, it is certainlyadvisable to thoroughly investigate this parameterduring method development and optimization workon cyclodextrin columns.

Temperature In equilibrium-based processes, tem-perature plays an important role. For all investigatedcompounds, a very good linear relationship betweenthe natural logarithm of the retention factor and theinverse of temperature is observed. For the productsR7315 and R7405 only small differences in slope aremeasured. Indications that for both products the en-thalpy values for solute}stationary phase transfer arevery similar. The effect of temperature on the valleypoint resolution is given in Figure 9.

Only for R23979 was a valley point resolution ofone measured for the whole temperature range be-tween 1 and 403C meaning that for this product theeffect of temperature variation is of the same magni-tude for both enantiomers. For R60844 and R7315practically no inSuence on the resolution was ob-served between 1 and 153C. Above that temperature,for both products the resolution value starts to de-crease in a similar way. The continuous decrease of

the resolution value with increasing temperature ob-served for R7405 can be considered as a logical be-haviour, because for R7405 and R7315 the smallestretention factors are measures } an indication that thebinding forces between the stationary phase and thesesolutes are smaller than for the other products investi-gated. As a result, the effect of a temperature increaseon the resolution is more pronounced for these com-pounds. However, we did not expect to observe a dif-ferent effect of temperature variations for R7315 andR7405, because for both products only minor differ-ences in enthalpy values were measured. Therefore,temperature variations probably have the same inSu-ence on both enantiomers of R7405, while the enan-tiomers of R7315 are affected in a different way.

In conclusion, temperature variations have an in-Suence on the retention factor as well as on theresolution value. However, the effects observed differfrom one product to another. Therefore, this parameter


has to be investigated on an individual basis duringmethod development and optimization work.

Type and concentration of organic modiVer It isknown that for the reversed-phase analysis of severalalkylbenzenes on chemically bonded cyclodextrincolumns the type of alcohol, together with its concen-tration in the mobile phase, strongly inSuences theretention behaviour of these substances.

To investigate the effect of type of alcohol onenantioselectivity, we performed some experiments inwhich the normally used modiRer (methanol) wasreplaced by the same amount of ethanol 1-propanol.For all the investigated substances, the retention fac-tor strongly decreased with increasing chain length ofthe alcohol used.

The retention factors measured for an ethanol-based mobile phase are about 30}50% lower than thevalues observed for methanol as organic modiRer.A decrease of approximately the same magnitudecould be observed when ethanol was replaced by1-propanol. Therefore, in extreme cases the retentionfactor drops to about 20% of the value measured formethanol, when this solvent is replaced by the sameamount of 1-propanol. With increasing chain lengththe hydrophobicity of the alcohol increase, whichenhances the chance for competition between thesolvent and the analyte molecules to interact with thehydrophobic cyclodextrin cavity.

Because a strong decrease in retention time of thesolutes could be observed when ethanol or 1-pro-panol was used as organic modiRer, these solventsseem to have a greater afRnity for the cyclodextrincavity than methanol. Therefore, they will be moreeffective for displacing strongly retained substancesfrom a cyclodextrin column. The manufacturers ofcyclodextrin columns in fact use this property, be-cause they recommend regenerating their columns bypassing several column volumes of pure ethanolthrough the column, followed by pure water and thenmethanol.

In general, an increasing water content in the mobilephase increases both the retention and the enan-tioselectivity. However, for practical applications theretention factors are generally too long when mobilephases with a high water content are used, and a com-promise has to be found. For our product classes wetherefore use a mixture of 80 vol% of aqueous phaseand 20 vol% of methanol as a typical starting mobilephase composition. If the products are too stronglyretained or do not elute at all, the amount of meth-anol is systematically increased.

Flow rate Due to the very speciRc type of interac-tions, which play a major role in chiral recognition

processes, chiral stationary phases often display slowmass transfer characteristics. Therefore, on chiral sta-tionary phases, Sow rate can have a strong effect onthe enantioselectivity. For difRcult separations,lowering the Sow rate certainly has to be consideredas a tool to enhance enantioselectivity.

Derivatized Cyclodextrins

Native cyclodextrin columns cannot be usedeffectively for the separation of enantiomers undernormal phase chromatographic conditions. On theother hand, different naturally occurring chiral mol-ecules that have been derivatized are extensively andvery successfully used in the normal phase mode ofoperation.

Triacetylcellulose, obtained by heterogeneousacylation of cellulose, was one of the Rrst commer-cially available derivatives. However, the later de-veloped and commercialized aromatic cellulose andamylose derivatives (benzoates, carbamates), com-pared with triacetylcellulose, are much more univer-sally applicable. Owing to the broad applicabilityof the cellulose and amylose derivatives similarcyclodextrin-based stationary phases have beendeveloped.

In our laboratories we investigated the possibili-ties of the derivatized cyclodextrin columns undernormal as well as reversed-phase chromatographicconditions. We also compared these phases with thecorresponding cellulose derivatives.

Our Rrst experiments on the functionalized cyclo-dextrin phases were performed under normal phaseconditions. A series of 42 products covering a broadrange of organic chemistry were investigated on theS-naphthylethyl carbamate, the para-toluoyl and 3,5-dimethylphenyl carbamate cyclodextrin derivative,using n-hexane-2-propanol in different ratios as themobile phase.

The results obtained were rather poor. Of thewhole test series only six products were partially orcompletely resolved on the 3,5-dimethylphenyl car-bamate column. The situation was even worse on thetwo other columns tested. Therefore, we decided toswitch immediately to the reversed-phase mode ofoperation. The initially used test series of 42 productswas Rrst investigated on the three above-mentionedcyclodextrin columns with a mobile phase consistingof 0.5% ammonium acetate in water and methanol ina 30}70 volume ratio. This mobile phase compositionwas chosen after some preliminary experiments,which demonstrated that higher water content causeda tremendous increase in retention times. Comparedwith the results under normal phase conditions, thenumber of products separated and the degree ofseparation were much better on all the investigated


Table 3 Derivatized cyclodextrins versus the correspondingcellulose derivatives under normal phase conditons

Column type Good separa-tion �'0.90

Partiallyresolved a

Total

Number % Number % Number %

Chiralcel OD-R 9 42.9 6 28.6 15 71.4Cyclobond I-DMP 2 9.5 3 14.3 5 23.8Chiralcel OJ-R 13 61.9 5 23.8 18 85.7Cyclobond I-PT } } 2 9.5 2 9.5

aFor the partially resolved peaks, the resolution on the cellulosederivatives is always much higher than on the cyclodextrin deriva-tives.

column types. Because earlier experiments on thenative cylcodextrin phases have demonstrated thatan acidic pH generally results in better separations,a 20 mM tetrabutylammonium hydrogensulfatesolution (pH 2.3) was thereafter used as tailingreducer.

Owing to the well-known reversed-phase effect ofretention decrease with lowering pH values, themethanol content in the mobile phase had to bereduced to 40 vol% instead of the 80 vol% used inthe experiments with ammonium acetate as tailingreducer. Under these experimental conditions, thelargest number of products was separated on theS-naphthylethyl carbamate column, although theresults on the other two columns only differed slight-ly. It was also interesting to observe that some prod-ucts, which could not be separated on native cyc-lodextrin, were completely resolved on one of thederivatized phases. In the next set of experiments,a 20 mM solution of respectively tetramethyl-,tetraethyl- and tetrabutylammonium hydroxide wasadjusted with sulfuric acid to a pH value of 2.5 andused in combination with 70 vol% of methanol asthe mobile phase. The effect of the cationic part of thetailing reducer is not clear. With tetraethylam-monium hydroxide the largest number of productsare partially or completely resolved, but the resolu-tion values are in general somewhat higher withtetramethylammonium hydroxide, although the dif-ferences with the two other alkylammonium hydrox-ides are minimal. Only for one member of the testseries was tetramethylammonium hydroxide requiredto obtain partial resolution.

On native cyclodextrin columns, we could clearlydemonstrate the inSuence of the anionic part of thetailing reducer. Therefore, a similar test was doneon the 3,5-dimethylphenyl carbamate cyclodextrinderivative, using 20 mM tetramethylammoniumhydroxide adjusted to pH 2.5 with respectively triSu-oroacetic, hydrochloric, phosphoric, (d)-camphor-sulfonic and sulfuric acids, (d)-Camphorsulfonic acidhas been deliberately chosen to investigate eventualadditional effects by introducing chirality in the mo-bile phase. After pH adjustment, the aqueous phasewas mixed with methanol in a 30}70 volume ratio.Some products were also tested with a 50}50 mixtureof aqueous phase and methanol. Comparable withthe observations on the native cylcodextrin column,and also on the functionalized cyclodextrin, pH ad-justment with sulfuric acid resulted in the largestnumber of separations. For all the other acids, thenumber of partially or completely resolved productsdropped to about 50% or less of the value observedfor sulfuric acid. However, three products whichcould not be separated with one of the different acids

tested were partially resolved when (d)-camphorsul-fonic acid was used to adjust the pH of the tet-ramethylammonium hydroxide solution.

Comparison of derivatized cyclodextrins and thecorresponding cellulose derivatives Because de-rivatized cellulose and amylose columns are exten-sively used in our laboratories for both analytical andpreparative chromatographic applications, it seemedworthwhile to compare these phases with the corre-sponding cyclodextrin derivatives. At present onlytwo cellulose phases are commercially availablewhich can be used equally well under reversed-phaseand normal phase conditions, namely Chiralcel�OD-R (3,5-dimethlyphenyl carbamate and Chiralcel�OJ-R (para-methylbenzoate) (Daicel, Japan). Wecompared these phases with Cyclobond�-DMP andCyclobond�-PT columns (Advanced SeparationTechnologies), respectively.

The Rrst experiments were performed under nor-mal phase conditions, using n-hexane}2-propanol ina 70}30 ratio as the mobile phase. If products elutedtoo fast with this mobile phase composition, theamount of 2-propanol was reduced to respectively 20or 10 vol%. A total of 21 different products wereexamined. The results of these experiments are sum-marized in Table 3.

From this data it is clear that for the investigatedproduct classes the cellulose derivatives are far su-perior compared to the corresponding cyclodextrinphases when normal phase conditions are applied.We thereafter examined the same product seriesunder reversed-phase conditions using a mixture of70 vol% of methanol and 30% of a 50 mMtriethylamine solution adjusted to pH 2.5 with sulfuricacid. The results of these experiments are summarizedin Table 4.

When we compare this data with the results ob-tained under normal phase conditions, we have toconclude that in the reversed phase mode of


Table 4 Derivatized cyclodextrins versus the correspondingcellulose derivatives under reversed phase conditons

Column type Good separa-tion �'0.90

Partiallyresolved a

Total

Number % Number % Number %

Chiralcel OD-R 2 9.5 10 47.6 12 57.1Cyclobond I-DMP 2 9.5 5 23.8 7 33.3Chiralcel OJ-R 14b 66.7 } } 14 66.7Cyclobond I-PT } } 3 14.3 3 14.3

aFor the partially resolved peaks, the resolution on the cellulosederivatives is always much higher than on the cyclodextrin deriva-tives.

bAll products are fully baseline resolved (�"1.00).

Figure 10 Derivatized cylcodextrin columns versus the corresponding cellulose derivatives. Experimental conditions: column:250 mm�4.6 mm ID (Cyclobond�-DMP, Cyclobond�-PT, Chiralcel� OD-R, Chiralcel� OJ-R); mobile phase: 50 mM triethylamineadjusted to pH 2.5 with sulfuric acid}methanol (30}70, v/v); flow rate: 1 mL min�1.

operation fewer products are separated on the cellu-lose derivatives, although on the Chiralcel� OJ-Rcolumn all separated products are fully baseline re-solved. The smallest resolution value equals 2.5 whilethe largest value is greater than 12.5, while undernormal phase conditions the highest resolution valueobserved equals 6.3.

On the cyclodextrin derivatives a few more prod-ucts are separated but the increase in number iscertainly not spectacular. Furthermore, in many caseswhere partial resolution has been indicated in thetable the chromatograms only showed a small devi-ation in the peak shape, indicating the early beginningof separation.

For a series of products which under comparableexperimental conditions are all very well separated onthe native cyclodextrin column, the results on thedifferent cyclodextrin and cellulose derivatives using50 mM triethylamine adjusted to pH 2.5 with sulfur-ic acid and methanol in a 30}70 volume ratio asmobile phase are illustrated in Figure 10.

Because only one substance of the test series isseparated on the Cyclobond-PT column, and mostof the products are only partially resolved on theCyclobond-DMP column, while all these productsare perfectly baseline resolved on native cyclodex-trine, it is clear that other parameters must play a rolein the chiral recognition process on the derivatizedphases.

As a general conclusion it can be stated that for thetype of substances investigated the derivatized cyclo-dextrin columns are, in both modes of operation(normal as well as reversed phase), less universallyapplicable than the corresponding cellulose deriva-tives.

Hydroxypropyl-�-Cyclodextrin Derivative

A hydroxypropyl-�-cyclodextrin column (experi-mental phase of the Chromatography Research groupof Merck Darmstadt) in the reversed-phase mode ofoperation has been extensively investigated. Theresult on this type of material were completelycomparable with the data obtained on the nativecyclodextrin columns. However, for the whole rangeof products the degree of separation was in general


better on the hydroxypropyl column. To investigatenew products, we therefore always start our experi-ments on a hydroxypropyl-�-cyclodextrin columninstead of using the classical native �-cyclodextrintype of material.

�-Cyclodextrin

A test series of 28 products, which also have beeninvestigated on �-cylcodextrin, have been examinedon a �-cyclodextrin column using a mobile phaseconsisting of 80 vol% of a 50 mM triethylaminesolution in water adjusted to pH 2.5 with sulfuricacid and 20 vol% of methanol. On the �-cyclodextrincolumn, 22 products were partially or completelyresolved. On the �-derivative only nine productscould be resolved.

Compared with the results obtained on the �-cyclo-dextrin column, it is clear that from a general point ofview the � derivative is less suitable for the separationof the investigated product series. Nevertheless, itremains an additional tool that might help to solvea separation problem when experiments on othertypes of cyclodextrin columns fail.

Dynamically Generated Cyclodextrin Phases

Instead of using the commercially available chemic-ally bonded cyclodextrin phases, it is also possible toperform enantiomer separations on a reversed-phasecolumn after addition of cyclodextrin or cyclodextrinderivates to the mobile phase. A number of experi-ments have been performed with hydroxypropyl-�-cyclodextrin as mobile phase additive. This derivativewas initially chosen because it is readily soluble inwater.

To investigate the possibilities of a dynamicallygenerated cyclodextrin phase, a test series of 22products was examined on three different types ofreversed-phase packing material. RP SelectB (Merck), Hypersil BDS (Shandon) and AluspherRP Select B (Merck) were selected as stationaryphases. An aqueous solution containing 50 mMtriethylamine and 50 mM of hydroxypropyl-�-cyclodextrin adjusted to pH 3 with sulfuric acidcombined with methanol in an 80}20 volume ratiowas used as the mobile phase.

Compared with the data obtained on a chemicallybonded hydroxypropyl-�-cyclodextrin column usingthe same eluent, the chemically bonded column gives,in general, better results than the dynamically coatedreversed-phase materials. Furthermore, alumina asstationary phase matrix seems to be less effective.However, in a few exceptional cases the AluspherSelect B column gives as good or even better resultsthan the silica-based materials.

Cyclodextrin Phases in PreparativeChromatographic Applications

The importance of preparative chromatographicenantiomer separations in industry is continuouslygrowing. Therefore it was very interesting to investi-gate the usefulness of cyclodextrin phases in prepara-tive chromatographic applications.

On an experimental hydroxypropyl-�-cyclodextrinphase (Merck Darmstadt), several products whichshowed a good resolution on the corresponding ana-lytical material were investigated on a preparativescale. An example of a preparative chromatographicseparation is illustrated in Figure 11.

On the hydroxypropyl-�-cyclodextrin phase, ingeneral a loading capacity of 2 mg g�1 of pack-ing material was used. However, in some other caseswe were able to load up to 4 mg of product pergram of stationary phase, which from a preparativechromatographic point of view certainly can beconsidered as a reasonable value for this type ofapplication.

Cyclodextrins as Stationary Phasesin Gas Chromatography

For the preparation of cyclodextrin-based capillarycolumns for gas chromatographic applications, twocomplementary methods have been developed andare commercially used. In the Rrst method, alkylatedcyclodextrins are diluted with polysiloxanes and im-mobilized on the inside wall of a fused silica capillary.In the other method, pentyl and hydroxyalkyl-dimethylcyclodextrins are coated on the inside wallof a fused silica capillary. Different well-known chro-matography companies offer cyclodextrin-based cap-illary columns:

� Chrompack: diluted cyclodextrins;� Macherey-Nagel: undiluted n-pentylated or

acylated cyclodextrins;� Advanced Separation Technologies: a broad var-

iety of undiluted cyclodextrin derivatives (per-methylated, hydroxypropyl, triSuoroacylated,butyrylated, dialkylated).

Different types of compounds (amines, epoxides, al-kanes, alcohols, lactones, sugars, etc.) can be separ-ated on cyclodextrin-based capillary columns.

The usefulness of these materials is illustrated bymeans of the separation of the four isomers ofa piperidine derivative. Gas chromatography on cyclo-dextrin-based columns was tried after acetylationwith triSuoracetic anhydride using the followingprocedure:


Figure 11 Preparative chromatographic separation on a�-cyclodextrin columns. Experimental condition: column: 80 mmID dynamic axial compression column (Prochrom); stationaryphase: 500 g 10 �m chemically bonded hydroxypropyl-�-cyc-lodextrin (experimental phase Merck Darmstadt; packing pres-sure 80 bar; mobile phase: 50 mM triethylamine adjusted with50 mM sulfuric acid to pH 2.5}methanol (80}20, v/v); flow rate:150 mL min�1; detection: UV, wavelength 220 nm, range 2.56AUFS; sample size: 1 g dissolved in 50 mL of concentrated sulfur-ic acid; solute:

An advantage of this dervatization procedure wasthat all types of salts (used to investigate the possibili-ties of diastereomeric salt formation as a stereoselec-tive synthesis method) could be acylated withoutprior liberation of the free base.

The Rrst experiments were done on a 15 m�0.32 mm ID Chiralsil-Dex� column (Chrompack).On this type of column, it was only possible to separ-ate the cis and trans isomers.

Thereafter, chromatographic experiments wereperformed on four different cyclodextrin phases fromAdvanced Separation Technologies:

� Chiraldex� B-PH: (S)-2-hydroxypropyl per-methylated �-cyclodextrin;

� Chiraldex� B-DA: (2,6-di-O-n-pentyl)-�-cyclodex-trin;

� Chiraldex� B-TA: (2,6-di-O-n-pentyl-3-O-tri-Suoroacetyl)-�-cyclodextrin;

� Chiraldex� G-TA: (2,6-di-O-n-pentyl-3-O-tri-Suoroacetyl)-�-cyclodextrin.

Chromatograms of the different experiments are il-lustrated in Figure 12 and Figure 13. The largest dif-ferentiation between cis and trans isomers was ob-served on the permethylated hydroxypropyl-�-cyclodextrin column (Chiraldex� B-PH). However,the best separation of all isomers individually wasobserved on the triSuoroacylated �-cyclodextrin col-umn (Chiraldex� B-TA).

Noticeable is the time required to perform ananalysis. Compared with the analysis method on thecrown ether column, a GC analysis is more than sixtimes faster, although one has to take into accountthat on the crown ether column the diastereomericsalt could be analysed as such, while for the gaschromatographic method the product has to be de-rivatized prior to chromatography.

Cyclodextrins in CapillaryElectrophoresis

In the pharmaceutical industry the importance ofcapillary electrophoresis is continuously growing. It isa technique which is increasingly used for determina-tion of the optical purity of intermediates and Rnalproducts. Many different optically pure compounds


Figure 12 Capillary GC analysis on cyclodextrin stationaryphases. Experimental conditions: column: (A) 15 m�0.32 mm IDChiraldex� B-PH ((S)-hydroxypropyl-�-cyclodextrin (Astec)); (B)15 mm�0.32 mm ID Chiraldex� B-DA (dipentylated �-cyclo-dextrin (Astec)); carrier gas: helium (linear velocity 25 cm s�1);temperatures: column 150}2003C (33C min�1), injector 2103C,detector 2103C; detection: FID; injection: 1 �L split (split ratio1/100).

Figure 13 Capillary GC analysis on cyclodextrin stationaryphases. Experimental conditions: column: (A) 15 m�0.32 mm IDChiraldex� B-TA (trifluoroacetyl-�-cyclodextrin (Astec));(B) 15 m�0.32 m ID Chiraldex� G-TA (trifluoroacetyl-�-cyclodex-trin (Astec)); carrier gas: helium (linear velocity 25 cm s�1); tem-peratures: column 150}2003C (23C min�1), injector 2103C, de-tector 2103C; detection: FID; injection: 1 �L split (split ratio 1/100).

can be used to generate the required chiral environ-ment. Cyclodextrins of course are also very suitableas an electrolyte additive.

The different types of cyclodextrins, which aremost frequently used are:

� �-cyclodextrin;� �-cyclodextrin;� �-cyclodextrin;� (2-hydroxy)propylated �-cyclodextrin;� (2-hydroxy)propylated �-cyclodextrin;� Heptakis (2,6-di-O-methyl) �-cyclodextrin;� Heptakis (2,3,6-tri-O-methyl) �-cyclodextrin;� carboxymethylated �-cyclodextrin.


Figure 14 Capillary electrophoresis using a derivatized cyclodextrin as electrolyte additive. Experimental conditions: equipment:P/Ace system 5500 (Beckman); capillary: 50 �m ID uncoated fused silica; total length: 57 cm; length to detector: 50 cm; electrolyte:20 mM Heptakis (2,3,6-tri-O-methyl)�-cyclodextrin 10 mM disodium hydrogen phosphate solution adjusted to pH 2.2 with phosphoricacid; analysis: temperature 253C, voltage #20 kV, inject sample 2 s, detection UV (220 nm).

The usefulness of cyclodextrins as an electrolyte addi-tive is illustrated in the following example. A sub-stance containing three optical centres, which meanseight possible isomers, had to be separated. HPLCexperiments on different types of chiral stationaryphases did not succeed in a complete resolution of themixture. The result of a capillary electrophoresis ex-periment using Heptakis (2,3,6-tri-O-methyl) �-cyc-lodextrin as electrolyte additive is illustrated in Fig-ure 14.

Compared with HPLC, in capillary electrophoresismany more parameters can be varied to improveseparation. Therefore, most of the methods de-veloped on one of the commonly used chiral station-ary phases can be replaced by a capillary electrophor-esis methods, using cyclodextrins or another chiralauxiliary as electrolyte additive.

See also: II /Chromatography: Liquid: Ion Pair LiquidChromatography; Mechanisms: Chiral; III /Chiral Separ-ations: Amino Acids and Derivatives; Capillary Elec-trophoresis; Cellulose and Cellulose Derived Phases;Chiral Derivatization; Gas Chromatogrpahy; Ion-PairChromatography; Ligand Exchange Chromatography;Liquid Chromatography; Molecular Imprints as StationaryPhases; Protein Stationary Phases; Synthetic Multiple

Interaction (‘Pirkle’) Stationary Phases; Thin-Layer(Planar) Chromatography.

Further Reading

Allenmark S (1991) Chromatographic Enantioseparations,Methods and Applications. London: Ellis Horwood.

Beesley TE and Scott RPW (1998) Chiral Chromatography.Chichester: Wiley.

Bender ML and Komiyama M (1978) Cyclodextrin Chem-istry. Berlin: Springer.

Chiraldex GC Handbook, 5th edn (1996) Whippany, NJ:Advanced Separation Technologies.

Coventry L (1989) Cyclodextrin inclusion complexation.In Lough WJ (ed.) Chiral Liquid Chromatography.Glasgow: Blackie.

Cyclobond Handbook (1996) Whippany, NJ: AdvancedSeparation Technologies.

Han SM and Armstrong DW (1989) Separation of enantio-mers and other isomers with cyclodextrin-bonded phases:rules for chiral recognition. In KrstulovicH (ed.) ChiralSeparations by HPLC, Applications to PharmaceuticalCompounds, pp. 208}286. Chichester: Ellis Horwood.

Hinze WL (1981) In: van Oss CJ (ed.) Separation andPuriTcation Methods, vol. 10, p. 159.

Kaiser R (1965) Chromatographie in der Gasphase I. Mann-heim: Bibliographisches Intitut.

Gas Chromatography

V. Schurig, University of Tu( bingen, Germany

Copyright^ 2000 Academic Press

Introduction

The separation of enantiomers (optical isomers) bycapillary gas chromatography on a chiral stationary

phase (CSP) was discovered by Gil-Av and co-workers at the Weizmann Institute of Science, Israel,in 1966. At the outset of this work, according toGil-Av,

this topic was in a ‘state of frustration’. Nobodybelieved it could be done. In fact, people were

III / CHIRAL SEPARATIONS / Gas Chromatography 2349

Introduction Cyclodextrins in HPLC Applicationsindex-of.co.uk/Tutorials-2/CHIRAL SEPARATIONS...chromatography was as mobile phase additive in thin-layer chromatography. In the mid-1980’s

Documents