Direct analysis of thin-layer chromatography spots by diffuse

ACS Legacy Archive ... Thin-layer and paper chromatography .... Report on the Nondestructive Examination of Ballpoint Pen Ink on Questioned Documents ...
0 downloads 0 Views 588KB Size
Anal. Chem. 1984, 56,2935-2939

Table VI. Comparison the Effect of Stereochemistry on Retention Index and log P of Different Ring Systems ring system

AI

A log P

bicyclononanes bicyclooctanes bicycloheptanes

+57" +67 +34*

+0.29 +0.34 +0.17c

nFrom Model 11. *Estimated from loa P. CFromref 12.

16 was smaller than the other two compounds. The N-benzyl derivative was also observed to have a higher AI value than that for the N-phenethyl derivative even though this latter substituent had a higher K value. The effect of stereochemistry on the lipophilicity of the azabicyclononanes and azabicyclooctanes could also be generalized to include aminobicycloheptanes as well (Figure 2). The log P values (measured by the classical shake-flash method) have been reported for 24 isomeric bicycloheptanes and it was found in all cases that the isomer with the phenyl groups located trans to the amino group was more lipophilic than the other isomer (12). For compounds 21a, 22a, and other analogues with the same stereochemistry, it was found that the log P values were on the average 0.17 units higher than the other isomer. From a comparison of the effect of stereochemistry on the lipophilicity of all four types of ring system shown in Figure 2, one could generalize a rule that would encompass all of the 27 compounds in the present study and the 24 compounds reported by Pleiss and Grunewald (12). In general, the isomer with the phenyl group closest to the retroposition of the acarbon atom of the amine will have the higher retention index or log P value. Not only is the sign of the effect the same in all of the cases but the magnitude of the effect is remarkably

2935

constant (Table VI) considering the wide variety of structures. A possible explanation for this rule may reside in the proximity of the polarizable electrons of the phenyl group and the electrons that are delocalized from the nitrogen to the acarbon. Perhaps the proximity of these two groups increases the organization of the adjacent water molecules which drives the compound into the other phase by an entropy effect. Registry No. 1, 92643-18-0; 2, 92643-19-1; 3, 92643-20-4; 4, 42471-68-1; 5, 92643-21-5; 6, 92643-22-6; 7, 57209-56-0; 8, 57209-59-3; 9, 57209-58-2; 10, 46849-36-9; 11, 15378-99-1; 12, 14948-73-3; 13, 13493-41-9; 14a, 76754-14-8; 14b, 76754-19-3; 15a, 63930-11-0; 15b, 63930-12-1; 16a, 76754-15-9; 16b, 71258-48-5; 17a, 76754-16-0; 17b, 76754-20-6; 18a, 76754-17-1; 18b, 76754-21-7; 19a, 71258-35-0; 19b, 71258-43-0;20a, 76754-18-2;20b, 71394-80-4;21a, 86943-79-5; 21b, 86992-69-0; 22a, 62624-27-5; 22b, 58742-05-5. LITERATURE CITED (1) Baker, J. K.: Ma, C. Y. J . Chromatogr. 1979, 169, 107. (2) Baker, J. K. Anal. Chem. 1979, 51, 1693. (3) Baker, J. K.; Cates, L. A.; Corbett, M. D.: Huber, J. W.: Lattin, D. L. J . Liq. Chromatogr. 1982, 5 , 829. (4) Smith, R. M. J . Chromatogr. 1982, 236, 313. (5) Smith, R. M. J . Chromatogr. 1982, 236, 321. (6) Smith, R. M. Anal. Chem. 1984, 56, 256. (7) Baker, J. K.; Skelton, R. E.; Riley, T. N.; Bagley, J. R. J . Chromatogr. Sci. 1980, 18, 153. (8) Baker, J. K.; Fifer, E. K. J . Pharm. S d . 1980, 69,590. (9) Ma, C. Y.; Elsohly, M. A.; Baker, J. K. J . Chromatogr. 1980, 200,

163. (10) Baker, J. K. J . Liq. Chromatogr. 1981, 4 , 271. (11) Martin, Y. C. "Quantitative Drug Design"; Marcel Dekker: New York, 1978. (12) Pleiss, M. A.; Grunewald, G. L. J . Med. Chem. 1983, 26, 1760. (13) Salva, Paul Ph.D. Thesis, University of Connecticut, 1983. (14) Reamer, Marie M.S. Thesis, University of Connecticut, 1979. (15) Riley, T. N.; Bagley, J. R. J . Med. Chem. 1978, 22, 1167. (16) Unger, S. H.; Chiang, G. H. J . Med. Chem. 1981, 2 4 , 262.

RECEIVED for review June 4,1984. Accepted August 27, 1984.

Direct Analysis of Thin-Layer Chromatography Spots by Diffuse Reflectance Fourier Transform Infrared Spectrometry Gary E. Zuber,* Richard J. Warren, Peter P. Begosh, a n d Ellen L. O'Donnell SmithKline & French Laboratories, P.O. Box 7929, Philadelphia, Pennsylvania 19101

This paper presents a method for obtaining infrared inforrnation from microgram amounts of material in sku on thln-layer chromatographic (TLC) plates using diffuse reflectance Fourier transform infrared (FTIR) spectrometry. TLC plates are directly Inserted into the spectrometer sample chamber. A reference laser aids in spot alignment, and plate contrlbutions to background are subtracted from the resultant I R spectra by using the FTIR data system. A variety of different materials on varlous cornmerclally available plate types were analyzed. The technique allows for both spot identlficatlon when references are available and provides functlonal group information useful in the subsequent characterization of unknown impurity spots. Spectral quality, I R band positlon, and the size of the useful I R spectral wlndow can vary somewhat depending on the plate type used. Infrared spectral subtraction methods can also be employed to obtaln infrared spectra from overlapping spots not totally resolved by TLC. Reproducible results can often be obtalned in 30 min or less.

Thin-layer chromatography (TLC) is a very simple and useful separation technique that suffers from a lack of identity-indicating information. Although nonspecific, comparative procedures such as R, values and spray reactions are widely used, their utility in the area of unknown TLC spot identification is somewhat limited. An alternative to these methods is the chromatographic isolation of the material of interest from the TLC plate with subsequent identification by spectral analysis. This time-consuming procedure, however, often encounters certain problems. As changes in the chemical environment are made to bring about isolation, the unknown material will often decompose, change physical form, or undergo further reactions. Consequently, the spectral results arising from these isolates can tend to be misleading. A more desirable approach would be the ability to obtain infrared spectra in situ from TLC unknown spots and then to use this information in the subsequent characterization of the material. A number of different works have been pub-

0003-2700/84/0356-2935$01.50/00 1984 American Chemical Society

2036

ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984

lished in this area. Percival and Griffiths have developed an in situ technique which requires the preparation of the TLC plate on an infrared transmitting material, such as silver chloride (1, 2). Although good spectra are obtained, the method is limited to the use of these specially prepared plates. Fuller and Griffiths did Qomepreliminary work with diffuse reflectance infrared Fourier transform (DRIFT) spectrometry to directly analyze spots punched out from aluminum-foilbacked silica gel TLC plates (3, 4). The results obtained, however, from this nonoptimized method were far from ideal with poor background compensation resulting. Lloyd et al. used Fourier transform infrared photoacoustic spectroscopy for direct TLC spot analysis (5). Photoacoustic infrared often results in good in situ TLC/IR spectra; however, the availability of such an attachment is limited to very few laboratories. Reflectance spectrometry has been extensively used to obtain infrared spectra from various highly scattering surfaces (6-8). It has also been widely used for the measurement of ultraviolet-visible spectra from in situ TLC spots (9-11). In this paper, we will demonstrate a method for obtaining infrared spectra of compounds absorbed on TLC plates using diffuse reflectance Fourier transform infrared spectrometry. The technique is an extension of the earlier work of Fuller and Griffiths (3);however, small TLC plates are used and great care is made to maintain surface uniformity. The preparation and handling of both sample and background TLC plates have been found to be the key to the success of the method. Measurements have been optimized in order to obtain the lowest limits of detection and reproducibility. A variety of different TLC plates have been found to be useful.

EXPERIMENTAL SECTION Samples were prepared by using a micropipet to spot solutions of the sample and solvent onto TLC plates. Samples were usually dissolved in acetone, chloroform, or methanol prior to spotting. A variety of different solvent systems were used for plate developing. For instance, the aspirin, salicylic acid, and caffeine separation was accomplished by using a mixture of ethyl acetate, chloroform, methanol, and formic acid (4010:2:1.5), while separations on reverse-phase TLC plates made use of methanol and water mixtures. Although solvent system choice is extremely important to achieve the desired chromatographic separation, it appears to have no significant effect upon the resultant TLC/ FTIR spectra providing the plates are properly dried. The following TLC plates were used: Analtech Silica Gel GF (gypsum binder with fluorescent indicator) and G (250 pm; 2.5 X 10 cm), Analtech Avicel (microcrystalline cellulose; 250 pm; 20 X 20 cm), Analtech Florisil (activated magnesium silicate; 250 pm; 5 X 20 cm), E. M. Reagents Silica Gel 60 F-254 (pore diameter 60 A with fluorescent indicator; 250 pm; 5 X 20 cm), Baker-flex (flexible sheets; 5 X 20 cm) (diethy1amino)ethyl (DEAE) cellulose, 10% acetylated (AC-10) cellulose, and aluminum oxide (IB) with inert binder, and Whatman reverse phase F-254 (with fluorescent indicator; 200 pm; 2.5 X 7.5 cm). A Nicolet (Madison, WI) 6000 FTIR spectrometer equipped with a liquid nitrogen cooled mercury cadmium telluride (MCT) detector and a diffuse reflectance attachment was used to obtain all of the spectral results. Computations were performed on a Nicolet IR-80 data processor with results stored on either a Control Data Corp. (CDC, Minneapolis, MI) cartridge disk drive or a CDC PA 5A1 fixed storage drive (160 Mbyte). Dry air from a Deltech (New Castle, DE) Model GC compressed air dryer was used to purge the instrument. The interferometer was operated at a mirror velocity of 0.586 cm/s with 2000 scans or less necessary to produce a good quality TLC/FTIR spectrum. The diffuse reflectance device (“Praying Mantis” model operating in the downward-looking mode) was obtained from the Harrick Scientific Corp. (Ossining,NY). The device was mounted within the FTIR’s 21 X 28 X 52 cm sample chamber. During sampling, the plates were placed upon two variable height 13 mm sampling cups as provided by Harrick. The plates were slipped beneath the IR beam for analysis. Plate heights could then be adjusted to obtain maximum signal output and t o optimize the

collection efficiency from the TLC plate surface. In most cases, the small plates sizes (2.5 X 10 or 2.5 X 7.5 cm) allowed for direct insertion of the plate into the FTIR’s sample chamber with minimal or no glass cutting and manipulation required. When it became necessary t o use large plates, care was taken to make plate alterations as remote as possible from the locations of the TLC spots. All TLC plates were analyzed at room temperature and each chromatographic plate was thoroughly dried prior to IR analysis to eliminate contributions from residual solvents. Plates were dried with a hot air gun or air-dried in a hood depending upon the thermal stability of the sample. In general when dealing with drug substances, as we were, each spot was located with a portable UV light which in turn allowed for the plate to be positioned so that the infrared reference laser hit the center of each spot. Spot diameters varied from 2 to 8 mm and the diameter of the infrared beam focused on each TLC plate was approximately 1mm. When non-UV absorbing compounds were analyzed, duplicate sample plates required developing. This allowed one plate to be treated with the appropriate spray reagents. This then served as a reference spot locator for the other untreated plate which was subjected to IR analysis. A final but extremely important requirement was the obtaining of a representative blank TLC plate which matched the sample mixture plate as closely as possible. This was accomplished by developing two plates at the same time (one containing the sample while the other was the same in every way except for the sample). Diffuse reflectance infrared spectra were obtained on the blank as before with a laser reference aligned on the plate at an R corresponding to the position of the mixture spot under consideration. After both sample and blank spectra are obtained, IR subtraction is used to remove the background interferences from the sample. All diffuse reflectance infrared spectra were calculated by measuring the single-beam reflectance spectrum of the sample and ratioing this against the single-beam spectrum of a reference mirror. Spectra were all obtained at a resolution of 4.0 cm-’. Typical analysis time was approximately 30 min with an additional 15 min required for TLC plate development and preparation. Prior to actual infrared analysis, the TLC plates were subjected to about 5 min of purge within the IR sample chamber. The exact length for this stabilizationperiod appeared to have no significant influence on the resultant spectra. All of the pharmaceuticals analyzed were of high purity, and the mixtures studied were of known composition. All other chemicals were high-grade commercial products and were used without further purification. The normal concentration of sample studied by TLC/FTIR was about 10 pg/spot. The actual detection limit is about l wg, although it is important to realize that this limit may vary from compound to compound and from plate to plate since it is dependent on the IR absorptivity of the various functional groups.

RESULTS AND DISCUSSION TLC/FTIR spectra were obtained by using diffuse reflectance on nine different types of commercially available TLC plates. Figure 1 illustrates the background infrared spectra of Silica Gel GF (a) as compared to that of Avicel (b). As can be seen, a number of differences are observed between each plate type. One of the most important contrasting features is both the size and position of the useful infrared spectral window. With all of the plates tested, infrared regions were found to exist where the plate material absorbed so greatly that only minimal information would remain after subtraction. Consequently, only the areas having low to moderate absorption were of great use. For example, silica gel absorbs only moderately in the infrared regions from 4000 to 3700 cm-’ and from 3100 to 1650 cm-’. Within these two small IR windows, background has very little influence, detection limits are quite low, and sensitivity to appropriate functional groups is high. Although infrared spectra are obtainable from silica gel TLC plates from 4000 through 800 cm-l, it is important to remember that the IR regions from 3700 to 3100 cm-’ and 1650 to 800 cm-l are regions of low sensitivity due to the high absorptivity of the background silica gel. This is especially true of the region from 1300 to 1100 cm-’ were silica gel has

ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984

22.31\ \ 8 z

E W

(A)

1361

(A)

2937

14.710.9-

W LL

;

7.1-

-.5 3’3)



14

-.2

j

I

4000 3610

I

I

3220

2ex)

I

1

2440 2050

1

1660

I

1270

I

am

1

I

1

I

I

1

I

I

I

spectra of (A) Silica Gel GF and (B) Avicel.

its most intense IR absorption band. Avicel, however, has low IR absorption from 4000 to 3550 cm-l and from 2700 to 1500 cm-’ and, as a result, Avicel should and does have better limits of detection for carbonyi-containing materials (IR region 1800-1600 cm-’) than does Silica Gel GF. Another important consideration is that compounds will bind differently depending upon the chromatographic plate chosen. This is illustrated by Figure2 which shows the TLC/FTIR spectra of acetaminophen on Silica Gel GF (a), Avicel (b), and aluminum oxide (c). Each spectrum can be seen to be somewhat different because acetaminophen is absorbed on each plate in a different way. Thus, the same compound may give different TLC/FTIR spectra when present on different TLC plate materials with infrared band position and intensity variations common. These variations in band heights are also greatly influenced by TLC substrate absorption. This makes comparison of spectra from different plate types difficult. The TLC plate layer thickness (200and 250 pm) and particle size (3 to 30 pm) were found to have no significant influence on the TLC/FTIR results. A final factor is the nature of the plate surface itself. Certain plate surfaces (e.g., Florisil) are easily disturbed and can result in significant nonuniformities during even minor plate alterations. In all ill be used of the following TLC/FTIR studies, Silica Gel GF w because it is the chromatographic material of choice in our laboratories. An important advantage of the in situ analysis of TLC spots is that spot elution and associated problems can be avoided. Figure 3 illustrates a transmittance FTIR speetrum of the pharmaceutical 2-[2-(acetyloxy)-l-oxopropoxy}-N,N,N-trimethylethanammonium 1,5-napthalenedisulfonate(Aclatonium napadisylate). This spectrum was obtained on a KBr disk with the two infrared absorbances corresponding to the two characteristic ester carbonyls observed at 1754 and 1725 cm-’. This compound was then subjected to routine thin-layer chromatography on Silica Gel GF plates. The resultant TLC

w LL

I

$ 45-

i

4

490

WAVENUMBERS (cm-’)

Figure 1. TLC/FTIR

cc7

?H

‘I

NH-&-CH,

I34 I I I I 4000 3660 3320 2980 2640 2300

I

I

I

I

1960 1620 1280 940 WAVENUMBERS (cm-’)

Flgure 2. In situ TLC/FTIR spectra of acetaminophen on (A) Silica Gel GF, (B) Avicel, and (C) aluminum oxide.

I

19.0 I I I I 1 I 4000 3640 3280 2920 2560 2200 1840 WAVENUMBERS (ern-')

Figure 3.

I

I

1

1480

1120

760

Transmittance FTIR spectrum of Aclatonium napadisylate.

spots were analyzed both in situ and after removal from the plate. The transmittance infrared spectrum of the isolated drug substance run as a KBr disk can be seen in Figure 4. It can be clearly observed that the expected carbonyl absorbances associated with this compound are now absent indicating that sample decomposition has occurred. This resultant spectrum along with other spectral data obtained from the isolated decomposed material later indicated that the drug had undergone hydrolysis during isolation. Alternative iso-

ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984

2938

$ 1 45 6

3

6 7 1 I I 4000 3640 3280 2920 2560 2200 1840

I

I

I

1480

1120

760

WAVENUMBERS (ern-')

4000 3610 3220 2830 2440 2050

1660 1270

680

490

WAVENUMBERS (cm”)

Figure 4. Transmittance FTIR spectrum of Aclatonium napadisylate after isolation from Silica Gel GF. 9 97, I

i

S 411

I

1

I

1900

1720

I

I

154 I

4000 3640 3280 2920 2560 2200 1840 WAVENUM BERS (ern-")

I

I

1

1480

1120

760

(A) 961 I

I

12 I I I I 71-74000 3610 3220 2830 2440 2050 1660 1270 660 ~

490

WAVENUMBERS (cm-’)

(E)

11 67

I

CHI

1660 WAVENUMBERS ( c m - ’ )

32’20 2830 2440 2050

14

I

&OOCCH3 I

4000 3610

I

I

I

I

3220 2630 2440 2050

I

I

1660

1270

7 - 1 880

490

WAVENUMBERS (cm-’)

Flgure 5. I n situ TLC/FTIR spectrum of Aclatonium napadisylate on Silica Gel GF.

14 4000 36110

3d

I/

COOH

1i70

880

4’h

Figure 6, Transmittance/KBr FTIR spectrum (A) of caffeine and in situ TLC/FTIR spectrum (B) of caffeine on Silica Gel GF.

Figure 7. Transmittance/KBr FTIR spectrum (A) of aspirin and in situ TLC/FTIR spectrum (B) of aspirin on Silica Gel GF. lation attempts using different extraction solvents gave similar results. The in situ TLC/IR spectrum for the compound appears in Figure 5. Although many infrared band changes are obvious in comparison to the KBr/transmittance reference spectrum, both ester carbonyl absorbances are still clearly observable (overlapping bands at 1754 and 1739 cm-l) indicating that the molecule is still intact. Figures 6-8 illustrate both the KBr/transmittance (A) and TLC/FTIR spectra (B) of the pharmaceutical compounds caffeine, aspirin, and salicylic acid, respectively. The reflectance spectra (B) were the result of the in situ infrared analysis of the components of an aspirin-caffeine pharmaceutical formulation after separation on Silica Gel GF. The salicylic acid results from the known hydrolytic degradation of the aspirin. Less than 10 Mg of each component was necessary to produce high quality spectra. These three compounds (two of which are very similar) gave spectra which were significantly different from one another. Each gave a spectrum which, although not as detailed as that obtained with transmittance IR, did possess infrared absorbance bands which were both assignable to structure and similar to known characteristic infrared functional group frequencies. Aspirin shows ester and acid carbonyl bands at 1742 and 1706 cm-’, respectively. Caffeine possesses infrared bands at 1705 and 1663 cm-l corresponding to the two amide carbonyls which are present. Finally, salicylic acid possesses a single acid carbonyl band a t 1670 cm-l. In addition to these obvious characteristic features, each spectrum contains some aromatic absorption bands resulting from the pharmaceutical being analyzed as well as HO and/or NH absorbances which are believed to be primarily due to uncompensated plate background contributions. The infrared region between 1300 and 1250 cm-l was also not found to be very useful because with each compound run on silica gel type TLC plates a rather large

ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984

By use of this leading edge/trailing edge subtraction method on these two components, TLC/FTIR spectra were obtained of each having similar quality to those spectra obtained using the previously mentioned ideal developing solvent system.

(A)

64 9-

2939

CONCLUSIONS

I

4000 3610

I

I

I

I

I

3220 2830 2440 2050 WAVENUMSERS

I

1660 1270

I

1

880

490

(cm-’)

(6)

12 9 1

@OH

2 14 I I I I I I 4000 3610 3220 2830 2440 2050 1660 WAVENUMBERS (ern-')

I

I

1

1270

880

490

Flguro 8. Transmittance/KBr FTIR spectrum (A) of sallcylic acid and In situ TLC/FTIR spectrum (B) of salicylic acid on Sllica Gel OF.

peak was found which is indicative of the expected Si-0 stretching vibration. During some of the TLC/FTIR work already discussed, occasionally TLC plates were developed where unresolved spots were observed. In these cases, infrared spectral subtraction techniques were employed using available references and/or leading edge/trailing edge results to obtain difference infrared spectra representative of the individual components of the overlapping materials. For example, the initial separation system for aspirin, salicylic acid, and caffeine on Silica Gel GF plates made use of a chloroform, methanol, and formic acid (9O:lOl) solvent system for development, In this system, the resultant aspirin and salicylic acid TLC spota overlapped.

The above results clearly demonstrate the utility and numerous advantages of the practical in situ infrared analysis of TLC spots using diffuse reflectance FTIR. With this method, infrared spectra can be quickly and easily obtained directly from thin-layer chromatography plates. Consequently, TLC elution-induced decomposition is avoided allowing for both spot identification when references are available and providing functional group information useful in the subsequent identification of unknown impurities. It is important to note here that IR spectral comparison to references requires that TLC/FTIR spectra of these standards be available since differences often exist between TLC/IR reflectance spectra and transmittance spectra. The method appears applicable to a wide variety of different commercially available TLC layers. The infrared results obtained from different TLC plate types can vary somewhat and this can be an important consideration when choosing the type of plate necessary to bring about a particular separation. Finally, it was observed that infrared beam positioning within overlapping spots and infrared subtraction techniques can allow for the analyses of chromatographically unresolved TLC materials, thus, providing a potential source for enhancing the resolving power of thin-layer chromatography.

LITERATURE CITED (1) Percival, C. J.; Grifflths, P. R. Anal. Chem. 1975, 4 7 , 154-156. (2) Gomez-Taylor, M. M.; Grlfflths, P. R. Appl. Spectrosc. 1977, 37, 528-530. (3) Fuller, M. P.; Griffiths. P. R. Anal. Chem. 1978, 5 0 , 1906-1910. (4) Fuller, M. P.;Grlfflths, P. R. Appl. Specfrosc. 1980, 3 4 , 533-539. (5) Lloyd, L. 6.; Yeates, R. C . ; Eyring, E. M. Anal. Chem. 1982, 5 4 , 549-552. (6) Fuller, M. P.; Griffkhs, P. R. Am. Lab. (falrfield, Conn.) 1978, 70 (lo), 69-80. (7) Maulhardt, H.; Kunath, D. Talanta 1982, 29, 237-241. (8) Chase, D. 6.; Amey, R. L.; Holtje, W. G. Appl. Spectrosc. 1982, 36, 155-1 57. (9) Frodyma, M. M.; Frei, R. W.; Wllllarns, D. J. J . Chromatogr. 1964, 13, 61-88. (IO) Kirk, A. D.; Moss, K. C.; Valentln, J. G. J . Chrornafogr. 1968, 36, 332-337. (11) Ebel, S.;Kussmaul, H. Chromafographia 1974, 7 (4),197-199.

RECEIVED for review June 11,1984. Accepted September 10, 1984.