Comparison of Experimental and Theoretical Calibration Curves in Solid-Surface Fluorescence Analysis R. J. Hurtubise Department of Chemistry, University of Wyoming, Laramie, Wyoming 8207 1
Scattering coefficient-layer thickness product terms and absorption coefficient-layer thickness product terms were obtained experimentally from aluminum oxlde and silica gel chromatoplates using fluoranthene as a model compound. The terms were employed to obtain theoretical fluorescence calibration curves from simplified equations derived from a recent theory on fluorescence densltometry. Experimental fluorescence calibration curves were also obtained and compared to the theoretical calibration curves for ranges of linearity, approximate points of first slope change, and relative fluorescence response in the reflected and transmitted modes. The results indicate that the theory should have applicability in fluorescence thin-layer densitometry.
Recordall series 5OOO recorder. The spectrodensitometer contained a high pressure xenon lamp and a miniature quartz-type monochromator that was set at 370 nm for fluorescence excitation or transmission-absorption modes. Ultraviolet cut-off filters were used in both transmission and reflection modes. Wedge monochromators set at 470 nm were employed in the fluorescence reflection and transmission single beam modes. For reflectance measurements the Schoeffel reflectance assembly was adjusted to a constant height above the stage of the instrument. For transmission-absorption measurements, the instrument was zeroed in the double beam mode with the chromatoplate and 7-60 filters (Esco Products, Oak Ridge, N.J.) in the substage filter holder. The filters were used to eliminate fluorescence interference from fluoranthene in absorption measurements. The filters were not used in the measurements with no absorbing component on the chromatoplates ( A , values). The inlet and exit slits for all fluorescence and transmission-absorption readings were set at 2.0 mm and 3.0 mm, respectively. For two aluminum oxide chromatoplates developed in n-hexane, the inlet slit was set a t 1.0 mm t o obtain A . readings. Reagents. Fluoranthene was purchased from Aldrich Chemical Co., Milwaukee, Wis., and recrystallized from ethanol. Ethanol was distilled before use. Aluminum oxide (A1203)chromatoplates and silica gel ( S O z ) chromatoplates were obtained from Brinkmann, Westbury, N.Y. Procedures. For the undeveloped and developed Si02and A1203 chromatoplates, 5 gL of ethanol standard solutions of fluoranthene were spotted for every spot. Each sample was spotted in duplicate and absorbance and fluorescence values for a given amount of fluoranthene were averaged. The amounts of fluoroanthene spotted are given below. For undeveloped chromatoplates the samples were spotted in two rows. The developed chromatoplates were allowed to develop until the n-hexane solvent front reached 10.0 k 1.0 cm.
Solution luminescence analysis is used extensively for the analysis of a variety of samples (1). Solid surface luminescence, whereby luminescence is measured as radiation transmitted or reflected from a solid surface, has been used as a precise and accurate analytical technique in areas such a s pharmacy, biochemistry, medicine, and pollution studies (2-7). Little work has been done on the theoretical analytical aspects of luminescence reflected or transmitted from solid surfaces. Pollak and Boulton (8)used the Kubelka and Munk theory (9, 10) of radiation transfer in scattering media to develop a theory of fluorescence from thin-layer chromatoplates. Pollak (11, 12) further discussed this theory and its potential application to thin-layer chromatography. Goldman (13) also used the Kubelka and Munk theory of radiative transfer in scattering media to develop a theory of fluorescence for thin-layer chromatoplates. Neither Pollak and Boulton, nor Goldman gave any experimental data t o support their theoretical conclusions. There is a definite need in solid surface luminescence analysis for a theory t h a t has been substantiated by experimental data. Presently, experimental conditions are determined by trial and error, and empirical calibration curves are used without sound theoretical basis. In this work experimental data were obtained from thin-layer chromatoplates using fluoranthene as a model compound. T h e data were employed to offer experimental support for the Goldman theory. T h e Goldman theory was investigated because it is more directly related to the Kubelka and Munk theory. Pollak and Boulton (8) and Pollak (11) used the Kubelka and Munk theory and an electrical transmission line with resistive parameters as a model simulating the optical behavior of chromatographic media. The data reported are not sufficient t o substantiate or refute either theory because much more experimental data will be needed to fully develop a useful theory in solid surface luminescence analysis. However, the data d o lend support to the Goldman theory.
For some chromatoplates, reliable absorbance values could not be obtained at 0.025 pg; thus these points were not used in the calculations. Fewer spots were used for Plates 1, 2, 5, 6 because these plates were developed with n-hexane. The fewer spots ensured that there would not be overlap of spots. Transmission-absorption data from SiOs or A1203 chromatoplates with fluoranthene adsorbed on them were obtained by zeroing the instrument with the chromatoplate and then obtaining the resulting absorbance from recorder tracings. Average A0 values were obtained by averaging several A , values determined across the chromatoplate. For these measurements the instrument was zeroed initially with a clean chromatographic glass plate with the chromatographic material removed. Fluorescence data in the reflection and transmission modes were obtained by zeroing the instrument initially with the chromatoplate and then measuring peak heights corresponding to fluorescence from recorder tracings.
EXPERIMENTAL Apparatus. A Schoeffel SD3000 spectrodensitometer equipped with a SDC 300 density computer (Schoeffel Instruments, Westwood, N.J.) was used in all measurements with a Fisher
RESULTS AND DISCUSSION General Comments. Goldman and Goodall (14-16) have described the densitometric application of the absorptiometry
2160
A N A L Y T I C A L CHEMISTRY, VOL. 49, NO. 14, DECEMBER 1977
Plates 1, 2, 5, 6 Plate 3 Plates 4, 7 , 8
0.05 p g , 0.10 u g , 0.15 u g , 0.20 p g , 0.25 u g 0.05 pg, 0.10 pg, 0.15 p g , 0.20 p g , 0.25 u g . 0.30 p g , 0.40 p g_., 0.50 p g , 1.o-ig, 2.0 ;g 0.025 p g , 0.05 fig, 0.10 p g , 0.12 p g , 0.15 p g , 0.2Opg. 0.25 p g , 0.30 p g , 0.40 p g , 0.50 p g
theory of Kubelka and Munk. They based the theory on a one-dimensional approximation to transfer of radiation in scattering media for components on thin-layer chromatoplates that absorb radiation. For fluorescence densitometry Goldman (13)took into consideration both excitation and fluorescent radiation in scattering media and obtained t,wo pairs of differential equations. One pair corresponded to absorption of exciting radiation in the transmitted and reflected directions and the other pair corresponded to fluorescent radiation emitted in the transmitted and reflected directions. After further mathematical treatment of the two pairs of differential equations, Goldman arrived a t two general equations that encompassed a wide range of fluorescer concentrations. One equation represented fluorescence transmitted through a thin-layer chromatoplate and the other equation fluorescence reflected from the surface of a thin-layer chromatoplate. The complexity of these equations almost necessitates the use of a computer to process experimental data, Goldman presented two simplified equations for very low concentrations of fluorescer a t relatively large values of scattering coefficients for exciting and fluorescent radiation. T h e simplified equations should be of use in chemical analysis because under certain conditions they show a linear relationship between fluorescence and amount of fluorescer adsorbed on a scattering material. T h e equations are given below and are the ones investigated in this work.
I'/iocu = l / 3 k X ( 1 - 7 / 3 0 s X * k X ) J + / i o a= 2 / 3 k X ( 1 - 4 / 3 0 s X . k X )
(1) (2)
Where I+ = transmitted fluorescence, io = initial exciting intensity, a = the proportion of absorbed radiation converted into fluorescence, k = absorption coefficient of exciting radiation, s = scattering coefficient of exciting radiation, X = thickness of the scattering medium, and J+ = reflected fluorescence. I n Equations 1 and 2 , the scattering coefficient for fluorescent radiation does not appear because of mathematical approximations. For Goldman's more complex equations, he assumed the scattering coefficient of exciting radiation was equal to the scattering coefficient of fluorescent radiation for graphs of both I+ and S vs. k X . The term k X is proportional to the amount of absorbing compound in the exciting beam (15).
Several assumptions t h a t were made in deriving the equations in the Goldman theory follow: (a) the exciting light beam is parallel and normal to the absorbing component on t h e thin-layer chromatoplate; (b) the reflections and absorptions occur a t infinitesimal distances and are constant over the frequency range of exciting radiation, the area illuminated by the exciting radiation, and the depth of the absorbing component; (c) the exciting and fluorescent radiation travel in the chromatoplate only in the directions perpendicular to the surfaces of the chromatoplate; (d) the chromatographic material does not absorb the exciting radiation. E x p e r i m e n t a l Values of k X a n d s X . Because X always appears in either k X or S X in Equations 1 and 2 , it was not necessary to determine X in this work but only s X and h X . The factor s X was calculated with the following equation (9), T o = l / ( s X + 1) where T o is the transmission of the chromatoplate when k X is zero. TGwas calculated from the equation, A. = log l / T o and A. was determined experimentally. T h e A. values for the two developed A1203chromatoplates were obtained a t an inlet slit setting of 1.0 mm. T h e A . values were somewhat dependent on the inlet slit setting. For example, for A1203plate 2 , average A,, values of 0.98 and 1.07 were obtained a t inlet slit settings of 1.0 mm and 1.5 mm, respectively. However, the A. values obtained across the chromatoplates generally varied a t least this much
especially with A1203chromatoplates. All but two A. values were determined a t a constant inlet setting of 2.0 mm. T h e variation of A,, across the chromatoplates is considered a fundamental limitation of the systems studied. One of the main purposes of this study was t o use a commercial instrument and chromatoplates to establish their capability to obtain acceptable data. More homogenous chromatoplates and possibly more sophisticated instrumentatior, would be needed to obtain highly accurate data. The k X values for fluoranthene were calculated bv successive approximations with a programmable calculator using the general Kubelka and Munk equation for transmission in scattering media (9, 15), which is given below.
T
=
b -_ a sinh ( b s X ) + b cosh ( b s X )
= transmittance of a chromatoplate with an absorbing
in the scattering medium
s+k
a=--
S
-
sX+kX sx
It would be ideal to express k X simply in terms of T and s X ; however, this is quite complicated theoretically. In this work T was calculated from the equation, A = log 1 / T , where A is equal to A. plus the experimental absorbance for fluoranthene. T h e term s X was calculated as described above. Using successive approximations to obtain k X was a relatively simple operation. It was assumed that the chromatographic material did not absorb exciting radiation. Goodall (17) showed that chromatographic material such as silica gel and aluminum oxide absorb very weakly at 340 nm. The excitation wavelength used in this work was at 370 nm. For the situation where chromatographic material would absorb exciting radiation, the absorption would have to be taken into consideration. Polycyclic aromatic hydrocarbons and nitrogen heterocycles are two classes of compounds th,at generally absorb at wavelengths greater than 340 nm and are fluorescent. T h e results in this work should be applicable to these two important classes of compounds and other classes. Calculation of P/i@a n d J+/iotr.Once sX arid k X were obtained, it was possible to calculate I+/ioa and S / i o a from Equations 1 and 2 . With these values and the corresponding k X values, theoretical calibration curves were plotted for reflected and transmitted fluorewence. Experimental fluorescence calibration curves were obtained for fluoranthene in the transmission and reflection modes and several comparisons were made with the theoretical curves. One comparison was the microgram values a t which the theoretical curves and experimental curves approximately first changed slope. These values were obtained at the point of intersection of two extrapolated straight lines that were tangent to the two major portions of the calibration curves (Figures 1 and 2). The k X values from the theoretical calibration curves were converted to microgram values and the comparisons are shown in Table I. The data were obtained for fluoranthene on developed and undeveloped aluminum oxide and silica gel chromatoplates. This was done to see if the chromatographic conditions had any influence on the reflected or transmitted fluorescence. The data show fairly good comparison between theoretical and experimental values for both developed and undeveloped chromatoplates whether the fluorescence was measured in the transmitted or reflected modes. The results indicate with the conditions employed in this work that Goldman's simplified equations can be used to predict approximately when a calibration curve first changes slope in ANALYTICAL CHEMISTRY, VOL. 49, NO. 14. DECEMBER 1 9 7 7
2166
Table I. Comparison of Theoretical and Experimental Approximate Points of First Slope Change in Terms of Microgram of Fluoranthene Transmitted fluorescence Plate 1
2
Developed A1,0, Theor. Exptl.
Plate
0.17
0.12 0.12
Undeveloped A1,03 Theor. Exptl.
3 4
0.18
0.15 0.11
0.17 0.15
Plate
Developed SiO, Theor. Exptl.
5 6
0.16 0.16
0.18 0.20
0.16 0.14
0.22 0.15
Undeveloped SiO, Theor. Exptl.
Plate
7 8
0.15 0.18
0.18 0.18
0.19 0.18
0.13 0.16
Reflected fluorescence 1 2
0.12 0.13
0.14 0.15
3 4
0.21 0.16
0.15 0.14
5 6
7
8
Table 11. Experimental Ratios of Slopes of Calibration Curves of Transmitted Fluorescence t o Reflected Fluorescence Developed Plate 1 2
f
Developed SiO, 5 6
0
a05
0.10
0.1 5
a20
0.25
040
MI C 1 0 0 1 A M
Figure 1. Comparison of experimental (0)and theoretical (A)calibration curves from developed A1203 plate 1 in the fluorescence transmission mode
Lo
2.0
r b '
w
-
t/ 0
0.05
0.10
0.1 5
0.20
0.25
0.30
MICROORAM
Figure 2. Comparison of experimental ( 0 )and theoretical (A)calibration
curves from undeveloped SiO, plate 7 in the fluorescence reflection mode reflected or transmitted modes with acceptable accuracy. Goldman (13),using unsimplified equations for reflected and transmitted fluorescence, made rough graphs of theoretical calibration curves a t low fluorescer concentration. He predicted the range of linearity of reflected fluorescence should be approximately twice that obtained by transmission in terms 2162
0.39 0.36
Plate 3 4
ANALYTICAL CHEMISTRY, VOL. 49, NO. 14, DECEMBER 1977
0.62 0.58
Undeveloped A1203 0.26 0.51 Undeveloped SiO,
7 8
0.91 0.89
of k X . The data in Table I do not support this conclusion. The simplified Goldman equations used in this work to obtain theoretical calibration curves show that the range of linearity of reflected and transmitted fluorescence in terms of k X are relatively close to one another. The experimental values in Table I support this and show that the range of linearity of reflected and transmitted fluorescence are relatively close. Figures 1 and 2 compare two sets of theoretical and experimental calibration curves with their relative intensity values normalized. Normalization was accomplished for the experimental curves by dividing the relative intensity values by the relative intensity value obtained for 0.10 ,ug fluoranthene. For the theoretical curves the calculated relative intensity values were divided by the calculated relative intensity value for 0.10 pg fluoranthene. The relative intensity values at 0.10 pg were chosen because they were on the linear part of the calibration curves and were considered more accurate than values a t lower amounts of fluoranthene. As seen in Figure 1 for an A1203chromatoplate the theoretical and experimental curves are superimposable to about 0.15 pg and beyond that they diverge with the experimental curve showing larger relative intensity values per microgram of fluoranthene. Figure 2 shows the theoretical and experimental curves for a Si02 chromatoplate. As indicated the two curves are superimposable to about 0.10 pg and then they diverge with the theoretical curve showing larger relative intensity values per microgram of fluoranthene. Both experimental curves change slope and then appear to become linear a t higher amounts of fluoranthene. Most of the experimental curves were linear and then changed slope and gradually curved downward. Experimental calibration curves for fluoranthene from 0-2 and 0-4 pg on undeveloped A1203 chromatoplates showed that the curves became almost parallel with the abscissa at about 1pg. One important contribution of the Goldman theory, considering the conditions employed in this work, is the capability of predicting approximately when the calibration curve first changes slope. Further work is needed to explain the nonsuperimposability of the curves. According to Goldman's simplified equations, the theoretical ratio of the slopes of a linear transmitted fluorescence calibration curve to a linear reflected fluorescence calibration . I1 compares the curve is 0.50, namely 1 / 3 k X / 2 / 3 k XTable
Table 111. k X a and sX Values Transmission,
Reflection,
Transmission, Plate kX
Reflection, kX
7130
4/30
SX
sX.kX
sX.kX
0.080 0.086 0.058 0.0014 0.047 0.041 0.048 0.056
0.066 0.071 0.047 0.0013 0.058 0.031 0.034 0.050
8.1 8.6 16 43 2.9 6.6 4.9 4.4
0.15 0.17 0.22 0.14 0.032 0.063 0.055 0.058
1 2 3 4 5 6 7 8
0.071 0.081 0.10 0.073 0.022 0.027 0.022 0.029
The k X values correspond t o the microgram values in Table I.
slopes of the experimental calibration curves obtained in this work. T h e ratio of slopes for calibration curves obtained for the developed A1203and Si02 chromatoplates compare favorably with the theoretical ratio. As indicated the experimental ratios for the developed A1203 are lower than the theoretical ratio of 0.50. T h e experimental ratios for the developed Si02chromatoplates are higher than the theoretical ratio. This shows t h a t the reflected fluorescence from the A1203chromatoplates is relatively greater than the reflected fluorescence from the Si02chromatoplates. The experimental ratio for the undeveloped A1203plate 3 was only 0.26. T h e low value may indicate that the fluoranthene did not penetrate the aluminum oxide chromatoplate to any great extent because with a ratio of 0.26 the reflected fluorescence is substantially greater than the transmitted fluorescence. The ratios for the undeveloped S i 0 2 chromatoplates are close to 1. This indicates that the fluorescent signals measured in the reflected and transmitted modes were approximately the same. With the undeveloped Si02chromatoplates, it was observed that the fluorescence of fluoranthene a t a given amount was much greater than its fluorescence on the developed S i 0 2 chromatoplates. For example, a typical ratio of the slopes of the calibration curves for the undeveloped chromatoplate to developed chromatoplate was 2.27 and 3.51 for reflected fluorescence and transmitted fluorescence, respectively. This suggests t h a t the fluorescence quantum efficiency of fluoranthene is enhanced greatly on undeveloped S i 0 2 chromatoplate by an adsorption mechanism, or other mechanisms, different from the one for fluoranthene when it is developed chromatographically and is also worthy of further investigation. Fluorescence emission spectra for fluoranthene on developed and undeveloped S i 0 2 chromatoplates showed essentially no change in the shape or fluorescence maxima. Thus, the change in quantum efficiency is not related to spectral shifts. Apparently the fluorescence intensity of fluoranthene is so great and the layer thickness of the chromatoplates is such t h a t signals in the reflection or transmission modes do not show a substantial difference in fluorescence intensity. The thicknesses of the A1203and Si02 layers were both 0.25 mm. T h e data suggest that the fluorescence intensity in the reflection and transmission modes is independent over a range of layer thicknesses for strongly fluorescent components. Goldman did not consider these aspects in the development of his theoretical equations and they were not investigated in this work. T h e simplified equations apparently cannot be used to predict the ratio of relative fluorescence intensities in the transmitted and reflected modes for highly fluorescent components over a certain range of layer thicknesses. More work is necessary to establish the layer thickness dependency of the fluorescent signals. Table I11 lists the experimental k X values of fluoranthene when the calibration curves approximately first changed slopes and the experimental S X values for all the chromatoplates.
Also, 7/30 sX.kX and 4/30 sX.kX products were calculated using the experimental k X and sX values and appear in Table 111. As Equations 1 and 2 indicate, the terms 7/ 30 sX.kX and 4/30 sX-kX are important in determining the points at which fluorescence transmission and reflection calibration curves become nonlinear and the relative response in the nonlinear region. As discussed earlier the theoretical and experimental calibration curves for a given chromatoplate became nonlinear at approximately the same microgram (or k X ) value whether fluorescence is measured in the transmission or reflection modes. Equations 1and 2 and the data in Table I11 show that the product 7/30 s X - k X is approximately twice the product 4/30 s X - k X . This information in conjunction with Equations 1 and 2 suggests that the relative change in reflected fluorescence intensity after a calibration curve becomes nonlinear should be greater than the relative change in transmitted fluorescence intensity. This was found to be true experimentally; however, it is doubtful whether this region will be useful analytically compared to linear calibration curves and it was not investigated further in this work. T h e data in Table I11 indicate the 7 / 3 0 s X - k X and 4/30 s X - k X products for Si02 chromatoplates are smaller than the corresponding products for A1203chromatoplates. This indicates that the linear range for Si02 chromatoplates in terms of microgram (or k X ) should be greater than the linear range for A 1 2 0 3 chromatoplates. Generally, the experimental results in Table I support this whether fluorescence is measured in the transmitted or reflected modes. However, the greater linear range for SiO2 is not substantial. Exceptions are the experimental value for plate 7 in the reflected fluorescence mode and the theoretical value for plate 3 in the reflected fluorescence mode in Table I. The Kubelka and Munk theory and thus the Goldman theory assume uniform particle size of the scattering medium. T h e commercial chromatoplates employed probably have a range of particle sizes and more work is needed to determine the magnitude of the effect of particle size as related to the Goldman theory. Kortum (9) has discussed the dependence of scattering coefficient on particle size and states it is nearly iiiversely proportional to the average particle size. Table I11 shows that the S X values for A1203plates 1-4 are greater than the s X values for Si02 plates 5-9. This suggests t h a t the average particle size for the Si02chromatoplates is larger than the average particle size for the A1203 chromatoplates. Particle size appears to be a major factor contributing to the smaller 7/30 sX.kX and 4/30 s X - k X products for S i 0 2 chromatoplates (Table 111).
CONCLUSIONS More work under a variety of experimental and instrumental conditions is needed to establish the validity of Goldman's equations. However, the results of this work indicate that under certain conditions Goldman's simplified equations can be applied to solid surface fluorescence analysis to predict approximately at what point a calibration curve will become nonlinear and the approximate range of linearity of a calibration curve. Also the results show t h a t it is more advantageous t o measure reflected fluorescence because of the greater relative fluorescence response except in those cases where the fluorescence quantum efficiency and layer thickness are such that reflected fluorescence and transmitted fluorescence give approximately the same response.
LITERATURE CITED (1) G. G. Guiibauit, "Practical Fluorescence: Theory, Methods, and Techniques", Marcel Dekker, New York, N.Y., 1973. (2) A. P. Shroff and C,:J. Shaw, J . Chromatogr. S d . , 10, 509 (1972). (3) J. C. Touchstone, Quantitative Thin-Layer Chromatography", WiieyInterscience, New York, N.Y., 1973. (4) E. Sawicki, T. W. Stanley, and W. C. Eibert, J. Chromatogr., 20, 348 (1965).
ANALYTICAL CHEMISTRY, VOL. 49, NO. 14, DECEMBER 1977
2183
(5) G. G. Guilbault, "New Fluorometric Methods of Analysis of Biologically Important Compounds",in Natl. Bur. Stand. (U.S.), Spec. f u b l . , 378, Accuracy in Spectrophotometry a n d Luminescence Measurements, Proceedings of a Conference held at NBS Gaithersburg, Md., March 22-24, (6) (7) (8) (9)
1972. R. A Puynter, S. L. Wellons, and J. D. Winefordner, Anal. Chsm.. 46, 736 (1974). R. M: A. von Wandruszka and R. J. Hurtubise, Anal. Chem., 48, 1784 (1976). V. Pollak and A. A. Boulton, J . Chromatogr., 72, 231 (1972). G. Kortum, "Reflectance Spectroscopy",Springer-Verbg, New York, N.Y., 1969.
(IO) W. W. Wendlandt and H. G. Hecht, "Reflectance Spectroscopy", Interscience Publishers, New York, N.Y., 1966.
(11) V. Pollak, Opt. Acta, 21, 51 (1974). (12) V. Pollak, J . Chromatogr., 133, 49 (1977). (13) J. Goldman, J . Chromsogr., 78, 7 (1973). (14) J. Goldman and R R. Goodall, J . Chromatogr.. 40, 345 (1969) 115) J Goidman and R. R. Goodall. J . Chromatoor..,~32..~ 24 11968). (16) J. Goldman and R. R. Goodall; J . Chromatogr., 47, 386 (1970). (17) R. R. Goodall. J . Chromatogr., 78, 153 (1973).
~~-
RECEIVED for review July 8, 1977. Accepted October 4,1977.
Room-Temperature Phosphorescence of Compounds Adsorbed on Sodium Acetate R. M. A. von Wandruszka and R. J. Hurtubise" Department of Chemistry, University of Wyoming, Laramie, Wyoming 8207 1
The room temperature phosphorescence behavior of a number of compounds adsorbed on sodium acetate was investigated. Comparisons of molecular structures and consideration of reflectance, fluorescence, and infrared spectra allowed the postulation of certain molecular criteria for room temperature phosphorescence. The interactions of p-aminobenzoic acid wlth a sodium acetate surface were considered in detail. The adsorption was found to involve the formation of the sodium salt of p-aminobenzoic acid on the sodium acetate surface, as well as hydrogen-bonding. Only protic solvents could be used. Room temperature phosphorescence measurements of chemisorbed compounds on sodium acetate provided a novel way of surface area determination. I t was also shown that the phosphorescent compounds are adsorbed flatly on sodium acetate. Analyticai findings for several compounds are reported.
Room temperature phosphorescence of adsorbed ionic organic molecules was first reported by Roth ( I ) and later by Schulman and Walling (2, 3 ) . Paynter e t al. ( 4 ) put the phenomenon t o its first analytical use. T h e use of a sodium acetate adsorbent in room temperature phosphorimetry was introduced by von Wandruszka and Hurtubise ( 5 ) . They employed it for the determination of p-aminobenzoic acid (PABA) in vitamin tablets. T h e phosphorescence signal of the adsorbed compound was found t o be insensitive to moisture, and quantitation was achieved with a spectrodensitometer. T h e same approach was later applied to folic acid, p-hydroxybenzoic acid, and benzocaine (6). In this paper, the use of sodium acetate in room temperature phosphorimetry is extended t o yet other compounds. Consideration is also given t o the theoretical aspects of the adsorption process. T h e phosphorescence properties of a variety of compounds adsorbed on sodium acetate were studied. Other surfaces were also investigated. Infrared, reflectance, and fluorescence spectra of adsorbed species, and surface area data for adsorbent and adsorbates were obtained. From the information gathered, generalizations could be made about the molecular requirements of the adsorbed species. A hypothesis for the adsorption process is presented. 2164
ANALYTICAL CHEMISTRY, UOL. 49, NO. 14, DECEMBER 1977
EXPERIMENTAL Apparatus. Phosphorescence measurements were made with a Perkin-Elmer MPF-2A Fluorescence Spectrophotometer and with a Schoeffel SD 3000 Spectrodensitometer as described previously ( 5 ) . Fluorescence spectra were obtained in the same fashion. Reflectance spectra were obtained with the Schoeffel spectrodensitometer in the double beam mode with a blank adsorbent sample in the reference beam. The IR spectra were obtained with a Perkin-Elmer 621 Grating Infrared Spectrophotometer. Reagents. Ethanol was purified by distillation as described by Winefordner and Tin (7). Other solvents were reagent grade and used without further purification. PABA was reagent grade and purified by recrystallization from ethanol. p-Aminohippuric acid was purified by recrystallization from water and folic acid by recrystallization from benzene. p-Aminophenol was purified by washing with ethanol, and terephthalic acid was recrystallized from ethanol. Other compounds were reagent grade and used without further purification. Procedures. The samples for the reflectance measurements were prepared by addition of ethanolic solutions of the compounds to the adsorbent and subsequent evaporation, as described previously ( 5 ) . The amount of adsorbate used for reflectance spectra was 500 ng on 10 mg adsorbent. The spectra were obtained with the spectrodensitometer and taken point-by-point at 5-nm intervals. For the determination of phosphorescence and fluorescence maxima, 200 ng adsorbate was used on 10 mg adsorbent. KBr pellets were used for the IR spectra. They were prepared with a Wilks Mini-press. For the spectra of pure compounds, 200 mg KBr was mixed with 8 mg of the compound. For the spectra of adsorbed compounds, 15 mg of the adsorbed system (sodium acetate plus compound) was used with 200 mg KBr. Of these mixtures, 60 mg was used to prepare each pellet. No more than the indicated amount of sodium acetate-adsorbed compound system could be used in a pellet because larger amounts of sodium acetate gave opaque pellets that could not be cleanly separated from the die. The amount of adsorbed compound on the sodium acetate had to be sufficient to be detected in the IR spectra. However, the systems of interest were those that showed room temperature phosphorescence, so too much adsorbate could not be used because the phosphorescence would be quenched. It was found that an amount of adsorbed compound that was 1.0%, by weight, of the sodium acetate adsorbent, satisfied both criteria. The surface area determination of anhydrous sodium acetate powder was carried out by Quantachrome Corporation, 69 Glen