Analytical Applications of Polarimetry, Optical Rotatory Dispersion

A Circular Dichroism Spectrophotometry Experiment. Ana M. V. Cavaleiro and Júlio D. Pedrosa de Jesus. Journal of Chemical Education 2000 77 (9), ...
0 downloads 0 Views 14MB Size
Analytical Applications of

Polarimetry, Optical Rotatory Dispersion, and Circular Dichroism Neil Purdie and Kathy A. Swallows Department of Chemistry Oklahoma State University Stillwater, OK 74078-0447

Chiroptical methods are optical methods that can differentiate between two enantiomers. They include polarimetry, optical rotatory dispersion (ORD), and circular dichroism (CD). Detection is based on the interaction between a chiral center in the analyte and the incident polarized electromagnetic radiation. Early applications dealt primarily with elucidation of molecular structures, particularly of natural products for which a technique capable of confirming or establishing the absolute stereochemistry was critical. Only recently has significant attention been given to the application of these techniques to analytical chemistry. Although many requirements must be considered when analytical methods 0003-2700/89/0361-077A/$01.50/0 © 1989 American Chemical Society

or procedures are being developed, the sometimes mutually exclusive properties of analytical selectivity and breadth of application are particularly important. Analytical selectivity depends on the structural properties of the analyte and the ability of the chosen detector to

REPORT differentiate the analyte from a potentially large number of interferences. One approach to analyzing a complex mixture is to eliminate the problem of interferences altogether by performing a total separation. Nonselective detectors of broad general applicability such as refractive index or UV absorption detectors can then be used. It is seldom essential to identify and quantitate every component of a mixture, however, which is helpful because total separation often is not possible. At the other

extreme, the ideal selective detector would operate well without any sample separation at all and despite the interferences. Reality is compromise. How many molecular properties are necessary to achieve an acceptable level of selectivity? If only one property is necessary, separation is essential unless a more sophisticated procedure that is either time or phase sensitive is used. If three or more properties are necessary, the number of potential analytes is greatly diminished. The optimum number appears to be two. Thus of the chiroptical methods, we strongly advocate the use of CD, which measures both rotation and absorbance simultaneously. This REPORT will describe the use of chiroptical phenomena for analytical detection, particularly for systems for which separation is not a prerequisite to the identification process. Chiroptical phenomena Numerous excellent treatises on the physical phenomena of chirality and

ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989 · 77 A

REPORT the manifestation of its interaction with polarized light are available (1-3). For chemical analysis, an elementary understanding of the nature of the in­ teractions and their relationships as well as the dependence of the experi­ mentally measured parameters on the concentration of the optically active species (4) is sufficient. Polarimetry and ORD both deter­ mine the extent to which a beam of linearly polarized light is rotated on transmission through the medium con­ taining the chiral sample. The two techniques are entirely equivalent for nonabsorbing chiral species and differ only in that ORD yields a spectral re­ sponse whereas polarimetnc measure­ ments usually are restricted to a limit­ ed number of preselected wavelengths. The degree of rotation is dependent on the rotatory strength of the chiral center, the concentration of the chirophore, and the path length. The older literature (1-3) uses obtuse terms and unusual concentration units when dealing with solution media, such as 2

[Φ] = ΙΟ" Μ [α] and [a] = 100 a/(c'd)

(1)

where a, [a], and [Φ] are the rotation, specific rotation, and molar rotation, respectively; M is the molar mass; d is the sample path length; and c' is the solute concentration, expressed either as a percent or as g/dL. Combining these equations results in an equation that is totally analogous to the BeerLambert law, namely a = [$]dc'. IUPAC recommends retaining the con­ centration unit g/dL because of the wealth of information in the literature; we disagree with this recommendation, however, and have chosen to use moles per liter in our work to make the linear relationship equivalent to the absorbance-concentration relationship. The units of [Φ] therefore are degree cm 2 /

dmol. (This is mentioned only in case someone compares data from our work with data on the same systems taken from the older literature, where they will find a difference in the order of magnitude reported for [Φ].) Experi­ mental values for a are usually on the order of millidegrees (m°) unless laser sources are used, in which case microdegrees can be measured. In the ab­ sence of absorption, the plain ORD spectrum changes monotonically with wavelength. This change can be either positive or negative (see Figure la). For chiral media that absorb the po­ larized light beam, anomalous rota­ tions in the ORD spectrum are pro­ duced if the chiral center and the chromophore are structurally adjacent to each other in an arrangement called a chirophore. This anomalous behavior is referred to as the Cotton effect (5) and is limited to the wavelength range of the absorption band, where it is seen superimposed on the monotonically changing plain curve. The anomaly takes the form of a sigmoidal curve with peak and trough extrema whose wavelength values are roughly bisected at an intermediate crossover point at which the rotation is zero, as shown in Figure lb. In the simplest sense, where a single Cotton effect exists, the height between the extrema can be used for quantitative measurements. ORD has not been extensively pursued as a worthwhile method for analytical work because of a lack of specificity in differ­ entiation and because of the uncertain­ ty in defining the baseline, which is the undeveloped part of the plain curve un­ der the Cotton band. CD is the most sophisticated of the three chiroptical methods in that the rotation and absorbance measure­ ments are made simultaneously. Lin­ early polarized light consists of two beams of circularly polarized light propagating in phase but in opposite

rotational senses. In chiral media the beams are phase differentiated because they "see" two different refractive in­ dexes and consequently travel at dif­ ferent speeds, a phenomenon that pro­ duces the rotation effect. If a chromophore is present, differ­ ential absorption also occurs that causes the transmitted beam to become elliptically polarized, as shown in Fig­ ure 2. The long axis of the ellipse is rotated from the original plane of po­ larization by the angle a, as before, and the ellipticity is defined as Ψ = t a n - 1 (OA/OB). Both terms are wavelength dependent and combine to produce the CD spectrum. The perimeter of the el­ lipse at each wavelength is the trace of the diagonal of the parallelogram whose adjacent sides are the two ab­ sorption vectors. Therefore Ψ is direct­ ly related to the absorbance difference, represented as the difference in vector lengths. Mathematically, the relation­ ship is given by the expression Ψ = x(eL - e R )A, or Ψ = πΔί/λ

(2)

where £ is the molar absorptivity; «L and €R are the molar absorptivities for left and right circularly polarized light, respectively; and λ is the wavelength at which the ellipticity is measured. In the application of CD to analytical problems, the two physical properties of chirality and absorption provide the information necessary for qualitative and quantitative determinations, re­ spectively. Because the change in mo­ lar absorptivity, Δε, is defined as («L €R), CD spectra can have positive and

Figure 2. Production of elliptically po­ larized light in CD. Figure 1. Typical chiroptical curves. (a) Plain ORD curve, (b) ORD curve with a single Cotton effect, and (c) CD curve with a single Cotton curve and a positive maximum. The sodium D-line is the usual wavelength for polarimetnc measurements.

78 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

OP is the direction of polarization of the incident beam, OL and OR are the absorption vectors for left and right circularly polarized beams, and a is the angle of rotation.

REPORT

Figure 3. A conventional CD instrument. LCP and RCP are the left and right circularly polarized light beams, respectively.

negative variations from the baseline as well as wavelengths where Ae = 0 (crossover points), as shown in Figure lc. Because there is no CD signal at wavelengths where there is no absorp­ tion by the analyte, the baseline is easi­ ly defined. Thus CD is superior to ORD for analytical applications. In the earliest CD instruments the experimental parameter measured was in fact the ellipticity of the transmitted beam. Modern instruments, however, are modified absorption spectrophoto­ meters that measure the differences in the absorbances of the two beams as a function of wavelength. The ordinate on the CD spectrum is still reported as an ellipticity, however, in deference to the historical development of the chiroptical methods. Because absorption is being measured, the data conform with the Beer-Lambert law. In CD the pro­ portionality constant is 0M, the molar ellipticity, which is related to ΔΕ by the equation ΘΜ = 3300 Δί. Once again, be­ cause we have defined the chiroptical property relative to mole/L and not to mg/dL, the orders of magnitude of our data and the literature data for the same system will differ. The necessity of a chiral center in the analyte molecule makes CD a much more selective method than straight­ forward absorption spectrophoto­ metry. Broad featureless bands in ab­ sorption spectra are often separated into more than one Cotton band in a CD spectrum, and the bands are fre­ quently of opposite polarity, enhancing the identification capabilities of CD over absorption. In addition, because the direct correlation is between the difference in the molar absorptivity, Δ«, and the molar ellipticity, 0M, for a

given analyte, the strongest CD signals are not necessarily associated with strong absorbers. This is especially use­ ful at wavelengths longer than 240 nm, where molar absorbances are small compared with the strong bands ob­ served in the far UV, yet CD bands are sufficiently large. Analyte concentra­ tions are typically 10~4 M or less for wavelengths longer than 240 nm. CD signals in the far UV can be very large, but the signal-to-noise quality is poor whenever strong absorbers are present. Because Ae is much smaller than the average e value, the CD signal is actual­ ly a very small millivolt quantity riding on top of a relatively large value. De­ spite the large difference in signal size, detection limits are 0.1 μg/mL for analytes with ΘΜ values of ~200 m 0 / mol-cm at the band maxima. This can be improved by introducing fluores­ cence detection (6), which is, in fact, commercially available. (The use of fluorescence, however, introduces the need for a third structural requirement in the analyte molecule, which effec­ tively decreases the applicability of CD.) Instrument operating conditions and solution concentration variables are chosen to give the optimum ratio of CD to the total absorption, bearing in mind that a mixture may contain sever­ al absorbing species. When fluores­ cence is the detector of choice, more serious interferences from the emis­ sions of CD-passive fluorophores can be expected, conceivably canceling any gains made in lowering the level of de­ tection. To predict if an analyte will be opti­ cally active and thus can be determined by CD, the presence of chirality and absorbance must be confirmed. (One

80 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

must remember, however, that al­ though the molecular structure may suggest that chirality is present, the substance may only be available as a racemic mixture and therefore unde­ tectable.) Although these requirements may seem to make CD too selective for prac­ tical analytical use, the applicability of CD can be increased by adding the missing molecular property by in situ derivatization. One can make an achiral absorbing analyte CD active by reacting it with a chiral partner, prefer­ ably one that is nonabsorbing, and a chromophore can be introduced in a way that either does or does not affect the overall chirality of the molecule. These CD-induction reactions should not be considered exclusively as possi­ ble pre- or postcolumn derivatization reactions in chromatographic applica­ tions using CD detection. They are in­ stead intended to be so specific that they can be used for analysis of unseparated mixtures. A block diagram of a conventional CD instrument is shown in Figure 3. The small signal intensity requires that the incident power be very large; a 500-W Xe lamp is usually used. This source must be water cooled and oxy­ gen must be removed from the instru­ ment to reduce the production of ozone, which is detrimental to the op­ tics. The volume of the typical 1-cm pathlength cell is about 3.5 mL; small­ er path length cells, which may require focusing of the incident beam, are available for analyses t h a t require smaller volumes. LC detection

Chiroptical detectors used in liquid

StartYourOtm Analytical Chemistry Reference Libmty

The ANALYTICAL CHEMISTRY Library Case . . . a handsome and convenient way to keep your ANALYTICAL CHEMISTRY issues protected and easy to reach. Each specially-designed library case holds up to seven issues—one-half year's subscription of ANALYTICAL CHEMISTRY. Designed to keep your ANALYTICAL CHEMISTRY issues at your fingertips, the ANALYTICAL CHEMISTRY Library Case is perfect for home, office, or library. The Library Case is available in two styles: the slipcase or the binder. Both styles are covered in attractive navy leatherette with embossed gold lettering and both are reasonably priced. You pay only $6.95 for the slipcase and $8.50 for the binder. Order your ANALYTICAL CHEMISTRY Library Case today! Your satisfaction is unconditionally guaranteed or your money will be immediately refunded. Simply fill in the coupon below and mail in with your payment.

Please send me: ANALYTICAL CHEMISTRY Library Slipcase(s) at $6.95 each ANALYTICAL CHEMISTRY Library Binder(s) at $8.50 each I am enclosing a check or money order for $ . * (Make payable to Jesse Jones Industries. Add $2.50 per case for orders outside the U.S.) Name Address City/State/Zip 'Payment must accompany all orders and must be made in U.S. currency only. Orders from outside the U.S. add $2.50 per case. Mail to: Jesse Jones Industries, P.O. Box 5120 ANALY-C, Philadelphia, PA 19141

REPORT chromatography (LC) are primarily single-wavelength detectors. Because of the constraints of both signal size and small elution volumes, lasers are the most suitable light sources for these detectors. Yeung and co-workers have described both polarimetric and CD detectors for LC (6, 7). Stopped-flow CD spectral detection for LC has been described both by development scientists from JASCO, Inc. (8) and by Westwood, Games, and Sheen (9). We are unaware of any real-time, on-thefly, full-spectrum CD detection in chromatography, although it is not beyond development if the need could ever be justified. Because chiroptical detectors respond to anything that is chiral, they are particularly useful in studying substances of natural origins, and their use can complement the more common chromatographic detectors in studying complex mixtures. The majority of the applications developed to date have involved laboratory preparations; the number of real samples investigated is very small. If total separation of a mixture is possible, polarimetry is the chiroptical detector of choice. It has been used, for example, to identify and quantitate structurally related carbohydrates in a common mixture (10). Polarimetric instrumentation is comparatively inexpensive both to purchase and to operate, and the detector responds equally well to absorbing and nonabsorbing analytes. Furthermore, the effectiveness of a polarimetric detector has been demonstrated in both the direct mode, where the rotation caused by the analyte is measured (7), and in the indirect mode, where the change in the measured background rotation for an optically active mobile phase is used to quantitate the analyte (11). In the latter example the analyte itself need not even be chiral. Limiting CD detection to a singlewavelength measurement reduces it to no more than a very expensive polarimeter with a much narrower range of application because the nonabsorbing chiral compounds are now transparent to the detector. If, however, separation is not complete, then the differentiation capability of the full-range CD detector becomes necessary. As the need arises, convenient instrumentation for fast-scan measurements may become available. Early attempts to develop such devices have been described in the literature (12,13), but a lack of demand is probably responsible for their absence from the laboratory. Enantiomeric differentiation

Enantiomeric differentiation is a two82 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

level problem. If the identity of the substance is known and only one isomer is present, then the sign of the rotation easily establishes its stereochemical identity, in which case polarimetric detection is enough. If both enantiomers are present, which is the rule rather than the exception, the analysis takes on a different dimension: the determination of the enantiomeric excess (EE) or optical purity. There are two general chromatographic approaches to enantiomeric differentiation. In the first approach the enantiomers are converted into diastereoisomers, which have different retention properties and thus can be physically separated. The derivatization to diastereoisomers can be done in either a precolumn reaction or in situ on the stationary phase of the column. For precolumn derivatization, any one of the well-known diastereoisomerizing reagents can be used (e.g., optically active amines for the separation of acid enantiomers). A convenient derivatization reaction cannot always be found, however. The in situ derivatization can be done only if the stationary phase is chiral; a variety of chiral stationary phases are available (14,15). The real thrust of this work, however, is to find effective and convenient ways to enhance optical purities. The actual determination of the EE is of secondary importance. A chiroptical detector is required to establish the elution order, but once this is done it can be replaced by a general-purpose detector. Polarimetry is the best choice for determination of enantiomeric enrichment at the exploratory level where eluted volumes are small. When chromatographic procedures are developed to the point where large-scale separations are possible, CD is the better detector, because differences in the full spectrum of the analyte compared with that of the standard signify the presence of a coeluted chiral interference. The second chromatographic approach is to determine the EE without first preparing diastereoisomers. Measuring the EE or optical purity without either a derivatization step or chiral stationary phase separation is actually easier than might be imagined. Because nonderivatized racemic mixtures will coelute from conventional LC columns, neither a conventional detector nor a chiroptical detector alone is adequate to determine the EE. If the detectors are placed in series, however, a quantitative distinction can be made. Data from either an absorbance or RI detector provide the sum of the concentrations of the two isomers and the signal from the chiroptical detector—

ENVIRONMENTAL SCIENTISTS AND ENGINEERS...

REPORT

ORDER YOUR COPIES OF THE MOST REQUESTED SPECIAL REPORTS FROM ES&T

which is either the rotational differ­ ence, («(+) — «(-)), for the polarimetric detector or the difference in ellipticity, (Ψ(+) — ^(-)), for CD—provides infor­ mation from which to calculate the concentration difference (16, 17). The concentration of each isomer is then readily obtained from the simulta­ neous solution of these equations. In many situations where CD is the detector of choice, its selectivity is so great that it can be used as a stand­ alone detector and will provide the con­ centration difference information without separation (18). This is espe­ cially important whenever eluted vol­ umes are small, because of the small CD signal. An aliquot is injected onto a conventional column, and the total concentration of both enantiomers is measured using absorption detection. The concentration difference is calcu­ lated simultaneously from the CD spectrum of another aliquot of the unseparated mixture. The EE is then cal­ culated, as described earlier. Whichever method is used to deter­ mine the EE, the quality of the results depends entirely on the optical purity of the standard materials. These can never be considered to be either 100% pure or of equivalent purity. The abso­ lute purity argument is somewhat moot in that the instrumental or separation methods are limited by their resolution capabilities. Theoretically, for spectro­ scopic analyses it is necessary to have only one of the isomers for instrument calibration provided that diastereoisomerization is not a prerequisite to the determination, as it is in NMR. In chiroptical methods it would be nice to have both isomers of equivalent optical purity as an internal check of the cali­ bration, but this is an unlikely if not impossible expectation. Reports of enantiomeric ratio determinations should dispel the idea that the determi­ nation is made against an absolute standard of purity and emphasize the fact that the ratio is relative to the pu­ rity of the best available standard ref­ erence material only.

Introducing three series made available by ES&T covering major areas in environmental science. You'll want to order all three!

Cancer Risk Assessment (32PP)

T

his special report explores the various scientific topics of public concern to regulate chemicals in the environment. Among the topics you'll find covered to help you manage the risk and communicate the results of risk assessment are: • Being more realistic about chemical carcinogenesis • Exposure assessment • Cancer dose-response extrapolations • Physiological pharmacokinetic modeling • And more . . .

Wastewater Treatment (48PP)

U

pdate your knowledge of the latest processes for the treatment of wastewater with this in-depth report. You'll find topics such as these . . . • Removing particles in water and wastewater • Removing dissolved organic contaminants from water • Removing dissolved inorganic contaminants from water • Anaerobic wastewater treatment • To name a few . . .

Computer Usage in Engineering (13PP)

U

tilize the power of the computer to bridge the gap between engineers and public decision-makers with this report that briefs you on the computer basics and then discusses new trends and applications, such as artificial intelligence, and their potential impact on environmental engineering. You'll also see how personal computers might change the practice of environmental engineering in the future, and much more. _^^/U

ALL SERIES MUST BE PRE-PAID BY CHECK OR MONEY ORDER.

TO ORDER, send check or money order with name, address^ telephone number, and quantity of each series to: American Chemical Society P.O. Box 57136 West End Station Washington, DC 20037 (202) 872-4363 RATES FOR EACH SERIES lcopy $10.00 ea. 2-5 copies $ 5.00ea. 6 and above $ 4.00 ea. Foreign payment must be made in U.S. dollars by international money order, UNESCO coupons, or U.S. bank draft. Foreign orders add 20% of total order cost for shipping'' handling. Please allow 4-6 weeks for delivery.

Detection limits The detection limits obtained with chiroptical detectors are equivalent to those obtained with absorbance detec­ tors for bulk measurements. In conven­ tional chromatographic systems, nano­ gram detection limits are common­ place; in state-of-the-art detector development, picogram or even femtogram limits have been reported (7). The ability to detect such small quanti­ ties is of critical importance only when the physical size of the sample is limit­ ed, as is the case in microbore LC, where the eluted volumes are exceed-

84 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

ingly small. If the sample size is not limited, the simple alternative is to scale up the experiment. Chiroptical methods are not yet competitive within the lowest of these achievable detec­ tion ranges, showing a typical cutoff in the nanogram-per-milliliter range, un­ less fluorescence is used for signal en­ hancement or laser sources are used (6, 7). When sample sizes are not a prob­ lem and a typical working sample vol­ ume is a few milliliters, detection limits on the order of micrograms per millili­ ter are readily achieved using CD.

Analytical applications The answer to the question of how se­ lective chiroptical methods really are appears to be that polarimetry and ORD have little to no potential as se­ lective detectors. They do have specific redeeming applications, but they are truly functional only when all interfer­ ences have been removed. CD is in the same category as long as its use is limit­ ed to single-wavelength chromato­ graphic detection. The applications de­ scribed here therefore are primarily those utilizing the selectivity capabili­ ties of full-spectrum CD analyses in which neither separation nor derivatization steps have been performed. Some instances in which derivatization is necessary are included. Our experience shows that the most useful wavelength range is from 240 to 400 nm, which encompasses the transi­ tions from the aromatic ring and unsat­ urated ketone chromophores. Few sub­ stances are CD active in the visible range, and at wavelengths shorter than 240 nm, signal-to-noise ratios are sig­ nificantly decreased because of the ex­ tremely intense absorption bands. In addition, CD bands are observed to be broad and featureless and usually of only one sign, resulting in spectra that are the same as the corresponding ab­ sorption spectra. Forensic science. The use of chir­ optical detectors in forensic science generally is limited to drug analysis. Many abused substances are of natural origin; some are synthetic. If used in LC, the function of the detectors is to either establish or confirm the stereo­ chemistry of the material, which is sig­ nificant in states where laws defining controlled substances fail to address the chirality variable. Because in some states only natural sources of cocaine are controlled, a racemic laboratory preparation, although still physiologi­ cally active, is not included on the con­ trolled substances list. The presence of L-cocaine chirality must often be verified for conviction. This can be done polarimetrically after the drug has been completely séparât-

ed from all of the other optically active components added as bulk materials (19). The absence of rotation for the proper eluted fraction is clear evidence that the material is racemic and there­ fore synthetic. Because such separa­ tions are difficult, distinction is often made using the subjective microcrystalline tests instead. Using CD, howev­ er, L-cocaine can be confirmed objec­ tively in alcohol extracts of the confis­ cated materials without separation (20). Other controlled substances, includ­ ing LSD (21), opiates, (22), cannabinoids, (23), and phenethylamines (24), have been conveniently identified and quantitated in the same way. Potential interferences are reduced to include only those that are also CD active. Thus although some of the extracts are intensely colored, the only effect this has on the analysis is to significantly reduce the signal-to-noise ratio and therefore increase the detection limit. When the extract contains more than one CD-active analyte (22,23), the con­ centration of each is determined by mathematically fitting the spectrum of the mixture using the sum of weighted contributions from the standard refer­

ence spectra for each of the compo­ nents. Determination of poisons is another area in which CD can be of value in a forensic context. Colchicine, strych­ nine, brucine, and curare alkaloids all produce excellent CD spectra when dissolved in dilute HC1 (18), and each of these can be identified without a separation step. The presence of these substances can be confirmed by a sim­ ple solvent extraction of the material ingested by the victim or examination of stomach contents. Phencyclidine (PCP) and its analogs and the barbiturates are inherently achiral and seemingly not detectable by CD unless chirality is induced, for example, by forming a molecular asso­ ciation complex with structurally orga­ nized media such as the cyclodextrin sugars. The intensity of the induced signal generally is quite small—typi­ cally one-tenth of the inherent molar ellipticity of a structurally equivalent molecule—which naturally raises the detection limit, β-cyclodextrin (jS-CD) can be used to determine both the PCP group and the barbiturates (25) be­ cause the sugar does not absorb in the range of the chromophores in the ana-

lytes. The CD signal is proportional to the concentration of the complex only, and the formation constant for its mo­ lecular complex with 0-CD must be ob­ tained to determine the analytical con­ centration of the target material. Great care must be exercised when analyzing mixtures where CD activity is induced because of the potential for competing association reactions. Thus far little has been done in the analysis of body fluids, although this area has tremendous potential. How ef­ fective CD might be depends on both the problems with interferences and the detection limits. The most signifi­ cant interferences in body fluids would be from proteins, which have strong CD signals in the wavelength range 180-220 nm. This is outside the favor­ able range of 220-400 nm, so protein interferences are less of a difficulty than might have been expected. Solvent partition, volume reduction, and derivatization are commonly used in preparing a biological sample to in­ crease the analyte concentration and enhance the detection limit. The detec­ tion limit for CD should be comparable to that obtained by LC with UV detec­ tion, but for full-spectrum analysis

When Buying TitraLab™ be Sure to Check your Golf Clubs TitraLab™ Titration Systems ensure you more time away from tedious routine tasks. TitraLab automatically switches be­ tween electrodes and titrant/reagent burettes according to your methods. Time is saved and errors are eliminated.

Live titration curves on the screen make method development faster, and errors during routine operation can be seen immediately. Time for instrument set-up is kept to a minimum by guided dialogue in plain English, German or French.

RADIOMETER

Check with your Radiometer prod­ uct specialist to determine which TitraLab system will suit your needs, and find out about Radiometer's application support. If you need precision and reliability in titration, be sure to contact your local Radiometer office and experi­ ence the convenience of TitraLab yourself.

-when you need to be sure ...

COPENHAGEN R A D I O M E T E R A N A L Y T I C A L A/S KROCSHOJVEJ 49 • DK-2880 BACSVAERD • DENMARK See us at the Pittsburgh Conference'89, Booth numbers 2500, 2513, 2515, 2517. CIRCLE 138 ON READER SERVICE CARD ANALYTICAL CHEMISTRY, VOL. 6 1 , NO. 2, JANUARY 15, 1989

85 A

REPORT larger original samples may be necessary. In preliminary urinalysis studies good results have been obtained for tetracycline (4) and morphine. Complications from CD-active metabolites may require that the specimen be separated using conventional LC before the full spectrum is run. Clinical chemistry. This area is a natural continuation of the analysis of biological specimens discussed above, but here the emphasis is on metabolic disorders. LC-polarimetry can be used to study sugars in urine (26) as well as cholesterol and its ester derivatives in human serum (27). All of these compounds are present in relatively large concentrations and have strong rotational strengths. Derivatization with a fluorophore (e.g., dansylation) can be used to improve the detection limit for amino acids (28), whereas enzyme and hormone levels are probably at the limit of polarimetric detection. Other potential analytes include the sterones, some of which have been studied using ORD (29) and CD (30). If these compounds are present in sufficient quantity, they could conceivably be determined in urine. Because of a general lack of chromophores among the sugars, amino acids, and sterols, the application of CD to their detection is limited unless a derivatization reaction is used to introduce the necessary chromophore. Color induction is already a standard procedure in clinical chemistry when absorption is the analytical method of choice. Although these same color induction reactions may be expected to produce a CD-active derivative, this is not generally the case, and specific derivatization reactions that yield CD-active products are necessary. Cholesterol, for example, contains an isolated double bond and is inherently CD active with a maximum around 200 nm—too close to the wavelength limit of the instrumentation to be of practical use in analysis. We have found, however, that chloroform extracts of human gallstones produce several strong Cotton bands in the 300-600 nm range, as shown in Figure 4. We have attributed these bands to some molecular complexation interaction between cholesterol and bilirubin (which absorbs in this range but is CD inactive) because the intensities of the bands vary directly with the darkness of the stones. This is supported by the reported CD induction reaction of bilirubin with albumin (31). This result led us to a fuller study of serum cholesterol. There is considerable concern in the clinical chemistry community about the lack of accuracy and reproducibility among laboratories reporting blood cholesterol levels and

Figure 4. CD spectrum of the chloroform extract of a human gallstone.

the distribution of the total cholesterol among the various high (HDL), low (LDL), and very low (VLDL) density lipoproteins. Automated procedures are commercially available that are based on absorption and give results for total cholesterol directly and HDL cholesterol levels after the selective precipitation of the low-density lipid fractions. VLDL is estimated to be onefifth of the total triglyceride (TGL) level. LDL, recognized as the principal culprit responsible for the formation of arteriosclerotic deposits, cannot be determined directly and is calculated instead from the mass balance relationship [LDL] = [total] - [HDL] - 0.2[TGL]. Of course the errors in each of the three independent measurements are cumulative, which is the source of the poor precision in the analyses. Using a novel chromogenic chemical reaction and full-spectrum CD detection, we have successfully measured the total low-density lipid (VLDL + LDL) fraction directly (32). HDL levels are also obtained from the same experiment from data taken at a different band maximum, as shown in Figure 5. Except for the earliest preliminary experi-

ments in which it was confirmed that the 525-nm Cotton band was caused exclusively by low-density fractions, the customary low-density lipid precipitation step is not a part of the routine analysis. The selectivity of the CD detector is so great that no spectral distortions are observed, even in the presence of hemolysis or extreme hyperlipidemia (TGL levels greater than 400 mg/dL). Ketosteroids are amenable to direct CD detection (30), which brings up the interesting question of whether the method might be useful for the detection and determination of anabolic steroids in urine. Preliminary studies have been made on three anabolics: methandrostenolone, methandriol, and stanozolol (32). The first, a ketosteroid, has a CD spectrum similar to testosterone but the bands are displaced to longer wavelengths and differentiation is, in fact, possible. Methandriol is a structural analog of cholesterol and combines with the reagent used in the cholesterol test to produce a colored product that has a completely different CD spectrum from that shown in Figure 5. Its presence in urine should thus be detectable after sample concentration. Stanozolol is CD inactive and does not react with the "cholesterol reagent." A review of the chemistry of the compound may suggest suitable reagents for color induction. Thus far no definitive tests have been made to determine the limits of detection for these substances using CD. Pharmaceutical chemistry. Chiroptical detectors can be used in pharmaceutical applications for quality control and for determination of optical purity or EE. Polarimetry and ORD applications as before are limited to LC detection, for example, the neomycins (33), amino acids (28), and 17-ketosteroids (29). CD, however, has been successfully applied to direct determination of numerous products, including opiates; quinine and quinidine (easily distin-

Figure 5. CD spectrum of the colored product of the cholesterol reaction. Curve (a) represents the total lipid cholesterol, whereas the shaded area is the spectrum after addition of the LDL precipitating agent and is thus representative of the HDL fraction only.

86 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

Figure 6. CD spectra of (a) penicillin-V, (b) cephalosporin, and (c) penicillin-V broth (8.07 mg/mL) in aqueous pH 5.4 buffer.

guishable from each other because the Cotton bands are of opposite polarity); reserpine; tetracyclines; penicillins and cephalosporins (whose CD spectra bear no mutual similarities [see Figure 6] in comparison with the absorption spec­ tra); and vitamins (4). Common addi­ tives such as aspirin, caffeine, and most simple sugars are noninterfering (4). This list is not especially long, partly

because the point clearly has been made that the task of using CD for quality control, without chromato­ graphic separation, is elementary and particularly reliable. The method is easily extended to each new applica­ tion because its role is simply that of a detector, and extraction procedures and workup for any given sample are the same as they are for any other

method of detection. A second group of pharmaceuticals, including atropine, meperidine, dilantin, the barbitals, and the diazepams (4) are achiral. These drugs are either inherently achiral (e.g., dilantin), or they exist as exact racemic mixtures (e.g., atropine), but each group can be assayed using /3-CD-induced CD if they contain a suitable chromophore. Race­ mic mixtures can be included despite the fact that the enantiomers may in­ teract differently with β-CD, and what is actually recorded is the net induced signal. This differing interaction is, of course, part of the reason that enantio­ mers can be partially separated on the chiral cyclodextrin stationary phases (15). The need for two detectors in the determination of EE or optical purity was discussed earlier. Some early ex­ amples combined UV or RI with polarimetric detection, and cocaine and co­ deine (19), epinephrine (34), and Dand L-penicillamine (35) were investi­ gated in this manner. We have successfully used UV and CD in series for prepared mixtures of (R)- and (S)-nicotine in which solu-

Graphics for Chemical Structures

T

his exciting book presents the lat­ est information from recognized experts in the field of chemical structure handling on microcomputers... and includes the newest devel­ opments in related sys­ tems on both mini- and mainframe computers. The book begins with an excellent overview of the evolution of computer graphics—and will introduce you to the four general types of software packages currently available · scientific word processing packages • packages that allow graphics entry of chemical structures but do not allow substructure searching or interfacing with other systems · pack­ ages that allow graphics structure entry and substructure searching and interfacing with other systems · packages designed to act as front ends to molecular modeling systems. This volume continues with discussions and comparisons of various software packages for the chemist. Information managers, information scientists, systems analysts, bench chemists, physical chemists, and PC enthusiasts will find this book a vital, necessary reference. Wendy A. Warr, Editor ACS Symposium Series No. 341 LC 87-3575 ISBN 0-8412-1401-8

176 pages (1987) Clothbound US & Canada $44.95 Export $53.95

- when you need to be sure..

RADIOMETER COPENHAGEN R A D I O M E T E R A N A L Y T I C A L A/S KROCSHOJVEJ 49 • DK-288Q BACSVAERD • DENMARK

Order from: American Chemical Society, Distribution Office Dept. 58 1155 Sixteenth St., N.W., Washington, DC 20036

CIRCLE 139 ON READER SERVICE CARD

or CALL TOLL FREE θ 0 0 " Ζ Ζ 7 θ 5 ί ) Ο and use your credit card!

ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989 · 87 A

REPORT

Figure 7. (a) UV spectra of an S-(—)-nicotine standard, (b) UV spectra of a smokeless tobacco 2-propanol extract, (c) CD spectrum of the nicotine standard, and (d) CD spectrum of the tobacco extract.

tions of the natural isomer were deliberately spiked with aliquots of the other (18). Subsequently, leaf extracts were spiked with the unnatural isomer and the EE was determined using conventional LC. The total nicotine concentration is measured by LC using an absorbance detector, and the CD spectrum of either an aliquot of the unseparated mixture or the eluate from a conventional LC column gives the data from which to calculate the concentration difference. The concentration of each is obtained by the simple solution of the simultaneous equations. The limit to the measurement of enantiomeric excess is on the order of ± 1 % with one exception: It is exact for the racemic mixture because the net CD signal will be zero. All of this discussion presumes that the identity of the compound is known. Agriculture and food science. The use of LC with combined UV and either polarimetric or CD detection for the identification of enantiomers of the pyrethroid insecticides (17) is an excellent illustration of the potential difficulties in attempting to make a full analysis of stereoisomers in a complex mixture extracted from a natural source. The analytical focus usually is on the commercially important constituent^), and the other ingredients are looked upon as undesirable interferences that complicate the assay. Because they are very complex, plant and food extracts are taken through a progression of simplifying separations that usually terminate with a definitive GC

or LC step. Even then, a complete analysis is not guaranteed. For example, the CD study of pyrethrins also included an investigation of the Amaryllidaceae alkaloids in both the single-wavelength and stopped-flow detection modes (9), and even then the metabolites associated with a number of the peaks could not be identified. Sugars have been separated from prepared laboratory mixtures and from urine extracts by conventional LC and identified using either RI or UV detectors in series with a polarimeter, with and without laser illumination (16,26). Sensitivity can be enhanced by first complexing the sugars with molybdate ion (36). Although the complex is colored, the possibility of changing the detection system from polarimetry to CD was not explored. Ketoses can be determined directly using CD (37), but patience is necessary to ensure that the time-dependent mutarotation process is complete. After demonstrating the extreme selectivity of the full-spectrum CD detector for moderately complex systems, we applied it to the study of plant extracts. There was plenty of information to suggest that it would function very well as an LC detector, because the polarimeter does (9,17), but how would it perform without sample separation? Two things must be borne in mind in this context. First, the extract will contain many compounds, and many of these could be CD active. But unless the product of the molar ellipticity, #M, and the molar concentration exceeds the

88 A · ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989

minimum detection limit of the instrument, the spectrum of that analyte will not be observed. This reinforces the selectivity capabilities of the technique even further, and CD then focuses on the commercially important component. Second, there are no good standard reference materials for most biological samples, so the results can only be compared with data reported from other analytical methods. Most often a result is considered to be acceptable if it falls within broadly defined limits. The study of nicotine in tobacco (38) illustrates these points. Many nicotine analogs and other organic bases are extracted along with the parent substance but are present in such relatively low percentages that the CD spectrum for the extract is an exact replica of the spectrum for pure nicotine, as shown in Figure 7. The estimated assay of 1-2% nicotine by weight of dry leaf is within the realm of the accepted figures for Nicotinia tabacum, and the direct CD method of analysis is considered to be satisfactory. The identity of the counterion in the salt form would not interfere with the CD detection, and the method would be a convenient alternative for the assay of nicotine in agricultural insecticides. Other equally successful studies have involved assays of hops (humulone) (4); belladonna (atropine) (4); marihuana (THC and CBD) (23); opium (morphine, codeine, thebaine, noscapine) (22);Rauwolfia serpentina (reserpine) (39); fermentation broths (penicillin-V) (40); cattle feed (aureo-

mycin (18) (Figure 8); and fresh vegeta­ bles, fruit, and spices (vitamins C and D) (18). Standard separation proce­ dures were used in every case; in some instances aqueous and organic phases were investigated separately. The assay of food glycosides is a rela­ tively unexplored area with some inter­ esting possibilities for the application of chiroptical detectors. Structurally these compounds meet the require­ ments for CD activity by having the chiral center in the sugar moiety and an aromatic chromophore in close juxta­ position; the connection between the two parts is through either carbon (cyanogenics), nitrogen (nucleosides and nucleotides), oxygen (saponins and flavones), or sulfur (glucosinolates). The magnitude of the CD signal will depend on how adjacent the nearest chiral center on the sugar is to the chro­ mophore, and it may become necessary to enhance the signal for the weakest cases. Polarimetric work on neomycins was mentioned earlier (33), and an ac­ curate CD assay of riboflavin was in­ cluded in a recent general investigation of the vitamins (18). A cursory review of the literature indicates that the lev­ els of riboflavin in meat and fresh vege­ tables are comparable to those in dis­ pensary doses of vitamins already in­ vestigated, and CD detection without chromatographic separation should be possible. References (1) Crabbe, P. ORD and CD in Chemistry and Biochemistry: An Introduction; Aca­ demic Press: New York, 1972. (2) Charney, E. The Molecular Basis of Op­ tical Activity; Wiley: New York, 1979. (3) Mason, S. F. Q. Rev. Chem. Soc. 1961, 15, 287. (4) Purdie, N. Prog. Anal. At. Spectrosc. 1987,10, 345. (5) Cotton, A. Compt. Rend. 1895, 120, 989-1044. (6) Synovec, R. E.; Yeung, E. S. J. Chromatogr. 1986, 368, 85. (7) Synovec, R. E.; Yeung, E. S. Anal. Chem. 1986,58,1237 A. (8) Takakuwa, T.; Kurosu, Y.; Sakayanagi, N.; Kaneuchi, F.; Takeuchi, N; Wada, Α.; Senda, M. J. Liq. Chromatogr. 1987, 10, 2759. (9) Westwood, S. Α.; Games, D. E.; Sheen, L. J. Chromatogr. 1981,204,103. (10) DiCesare, J. L.; Ettre, L. S. Chroma­ togr. Rev. 1982,220,1. (11) Yeung, E. S. J. Pharm. Biomed. Anal. 1984 2 255 (12) Anson, M.; Bayley, P. M. J. Phys. Ε 1974, 7, 481. (13) Hatano M.; Nozawa, T.; Murakami, T.; Yamamoto, T.; Shigehisa, M.; Kimura, S.; Kakakuwa, T.; Sakayanagi, N.; Yano, T.; Watanabe, A. Rev. Sci. Instrum. 1981,52, 1311. (14) Pirkle, W. H.; Finn, J. M. In Asym­ metric Synthesis; Morrison, J. D., Ed.; Academic Press: New York, 1983, pp. 87124. (15) Armstrong, D. W. J. Liq. Chromatogr. 1984, 7, 353.

Figure 8. CD spectra of antibiotics in cattle feed. (a) Chlortetracycline in 1.0 M in pH 5.6 buffer and (b) aureomycin in pH 5.6 buffer extract of cattlefeed crumbles.

(16) Boehme, W. Chromatogr. Newsl. 1980, 8,38. (17) Meinard, C; Bruneau, P.; Perronnett, J. J. Chromatogr. 1985, 349, 109. (18) Purdie, N.; Swallows, Κ. Α., unpub­ lished results. (19) Palma, R. J.; Young, J. M.; Espenscheid, M. W. Anal. Letters 1985, iS(B5), 641. (20) Bowen, J. M.; Purdie, N. Anal. Chem. 1981,53,2237. (21) Bowen, J. M.; McMorrow, Η. Α.; Pur­ die, N. J. Forensic Sci. 1982,27, 822. (22) Han, S. M.; Purdie, N. Anal. Chem. 1986,58,113. (23) Han, S. M.; Purdie, N. Anal. Chem. 1985,57,1068. (24) Bowen, J. M.; Crone, Τ. Α.; Head, V. L.; McMorrow, Η. Α.; Kennedy, R. K.; Purdie, N. J. Forensic Sci. 1981, 26, 664. (25) Han, S. M.; Purdie, N. Anal. Chem. 1984, 56, 2822, 2825. (26) Kuo, J. C; Yeung, E. S. J. Chroma­ togr. 1981, 223, 321. (27) Kuo, J. C; Yeung, E. S. J. Chroma­ togr. 1982, 229, 293.

(28) Reitsma, B. H.; Yeung, E. S. Anal. Chem. 1987,59,1059. (29) Gergely, Α.; Szasz, G. Acta Pharm. Hung. 1983, 53, 280. (30) Gergely, Α.; Szasz, G.; Soos, J.; Fresenius, Ζ. Ζ. Anal. Chem. 1986,323,157. (31) Berglof, J.; Grahnen, Α.; Gjoholm, I. Clin. Chem. Acta 1975,62, 169. (32) Murphy, L. H.; Purdie, N., unpub­ lished results. (33) Gergely, Α.; Papp, O.; Szasz, G.; Vamos, J.; Bacsa, G. Acta Pharm. Hung. 1980 50 98 (34) Scott', B. S.; Dunn, D. L. J. Chroma­ togr. 1985, 319, 419. (35) DiCesare, J. L.; Ettre, L. S. J. Chroma­ togr. 1962,252,1. (36) Dokladolova, J.; Upton, R. P. J. Assoc. Off. Anal. Chem. 1973, 56, 1382. (37) Hayward, L. D.; Angyal, S. J. Carbohydr. Res. 1977, 53,13. (38) Atkinson, W. M.; Han, S. M.; Purdie, N. Anal. Chem. 1984,56, 1947. (39) Swallows, Κ. Α.; Purdie, N. Pharm. Res., in press. (40) Purdie, N.; Swallows, K. A. Anal. Chem. 1987,59,1349.

Neil Purdie (right) is professor and head of the Chemistry Department at Oklahoma State University. He obtained a B.Sc. degree and a Ph.D. from the University of Glasglow, Scotland, and joined the faculty at Oklahoma State in 1965. His research interests include relaxation kinetic studies of metal-ligand complexation reactions and analytical detection using chiroptical methods. He is also involved in the development of matrix materials for the controlled release of pharmaceutical products and agrichemicals. Kathy A. Swallows (left) is a fourth-year Ph.D. student in chemistry at Oklaho­ ma State University. She obtained a B.S. degree from Central State University in Edmond, OK. Her research interests include the development of controlled re­ lease materials and the measurement of optical purities. ANALYTICAL CHEMISTRY, VOL. 6 1 , NO. 2, JANUARY 15, 1989 · 89 A