discharge medium ( 7 ) , the admission of small amounts of water vapor to the detector produced results similar to those encountered with electronattaching compounds. The lack of response noted for NH8 may not be due t o insensitivity of the detector to it, but rather t o its never reaching the detector because of reaction with the brass fittings and union t h a t were used in the gas line carrying the sample vapors. The response to paraffins falls on a curve which rises sharply from butane to pentane. I t is expected that the curve will level out for higher members of the series. Qualitatively, the generalizations about ionization potentials noted above apply to the response of the photionization detector. Isobutane has a slightly higher response than n-butane. The detector’s sensitivity is neopentane < cyclopentane < n-pentane. Butene-2 produces a much larger response than the saturated butanes. Comparison of the results with the three discharge gases reveals certain anomalies. I n He, abnormally large response is found for acetone and butene -2. I n Nz, neopentane gives a n unusually large signal. Further study of the response of the detector should account for or correct these results.
CONCLUSIONS
The photoionization detector is a quite versatile device, with good sensitivity and linearity, and with potential for extreme selectivity. The operating and constructional parameters are not critical. Future studies should be concerned with investigation of a n external source and vacuum monochromator, utilizing LiF or CaFz windows. Several advantages to this arrangement are anticipated. First, the problems associated with the use of a glow discharge in the detector would be eliminatedthe sample could not interact with the photon source, the source-generated background signal would be eliminated, and a more intense energy could be applied t o the sample. Second, the potential extreme selectivity of the device could be realized by using the monochromator. Third, the detector would also be of use as a n instrument for making precise physical measurements. Fourth, by the inclusion of a photocell in the chamber, the detector could be used as a nondestructive far ultraviolet monitor for aromatic hydrocarbons. LITERATURE CITED
ics and Electron Physics,” Vol. VII, p. 399, L. Marton, ed., Academic Press, New York, 1955. (3) Knewstubb, P. F., Tickner, A. W., J . Chem. Phys., 37, 2941 (1963). (4) Knewstubb, P. F., Tickner, A. W., Zbzd., 36, 674 (1962). (5) Zbid., p. 684. (6) Lovelock, J. E., ANAL.CHEM.33, 162 (1961). ( 7 ) Lovelock, J. E., K’ature 188, 401 (1960). (8) Nicholson, A. J. C., J . Chem. Phys. 39, 454 (1963). (9) Price, W. C., “Advances in Spectroscopy,” Vol. I, p. 56, H. W. Thompson, ed., Interscience, New York, 1959. (10) Price, W. C., Bralsford, R., Harris, P. V., Ridley, R . G., Spectrochim. Acta 14, 45 (1959). (11) Riley, B., “Gas Chromatography,” p. 81, R . P. W. Scott, ed., Butterworths, London, 1960. (12) Robinson, C. F., Brubaker, W. M., U. S. Patent 2,959,677, November 8, 1960 (filed May 2, 1957). (13) Roesler, J. F., ANAL.CHEM.36, 1900 (1964). (14) Steiner, B., Giese, C. F., Inghram, M. G., J . Chem. Phys. 34, 189 (1961). (15j Walker, D. C., Back, R. A., Zbid., 37, 2348 (1962). (16) Watanabe, K., Nakayama, T., Mottl, J., “Final Report on Ionization Potential of Molecules by a Photoionization Method,” Dept. of Army Project No. 5B 99-01-004, 1959, Department of Physics, University of Hawaii.
(1j Cobine, J. D., “Gaseous Conductors,”
Dover, Xew York, 1958.
(2) Goldstein, L., “Advances in Electron-
RECEIVED for review June 8, 1964. Accepted November 5, 1964.
Semiquantitative Determination of Impurities in Bisphenol A by Circular Paper Chromatography N. H. REINKING and A. E. BARNABEO Plastics Division, Research a d Development Department, Union Carbide Corp., Bound Brook, N. 1.
b A semiquantitative method for the determination of the principal impurities in commercial bisphenol A utilizes reversed-phase circular paper chromatography and is especially useful in analyzing samples containing minor amounts of impurities. As little as 0.03% of an individual impurity can b e detected in bisphenol A by this method. The components are separated on circular filter paper impregnated with tricresyl phosphate, using an aqueous solution of trisodium phosphate as the eluent. The chromatogram is sprayed with a diazonium salt solution to develop color. The concentration of each impurity in an unknown can b e calculated by determining its respective extinction point and using the previously established sensitivity value.
T
HE principal impurities normally associated with commercial bisphenol A have been reported ( f , 4 ) as Dianin’s compound (4-p-hydroxyphenyl - 2,2,4 - trimethylchroman), the 2,4‘ isomer [2-(o-hydroxyphenyl)2 - ( p - hydroxyphenyl)propane], and a trisphenol [2,4-bis(a,a-dimethy1-4hydroxybenzy1)phenol. Work in this laboratory has confirmed these findings and, additionally, has revealed the presence in trace amounts of several other impurities. I n certain reactions, the presenc,e of any impurity in bisphenol A is objectionable; in other cases, the amounts of individual impurities present are of concern. Of the methods generally available for identification and measurement of impurities, the most useful, in this case, have been found to be gas
liquid and paper chromatography. Tominaga has reported on the direct analysis of bisphenol A using gas liquid chromatographic analysis (6). Gill has also reported on a quantitative gas liquid chromatographic method after acetylation of all reactive hydroxyl groups in the bisphenol d sample ( 3 ) . Paper chromatographic methods have been reported by Anderson, Carter, and Landua ( I ) , who used a dual chromatographic scheme to separate and measure impurities and by Challa and Hermans ( 2 ) , who employed a single one-dimensional chromatographic method. Whereas Anderson, Carter, and Landua (1) and Challa and Hermans ( 2 ) require preconcentration of impurities when present in low concentration, work in this laboratory has resulted in a simple analytical procedure using reversedVOL. 37, N O . 3, MARCH 1965
* 395
Table 1.
Physical Properties of Principal Components of Commercial Bisphenol A
Compounds Pp-Hydroxyphenyl-2,2,4 trimethylchroman 2-(o-Hydroxyphenyl)-2-( phydroxypheny1)propane 2,2-Bis(p-hydroxypheny1)propane 2,4-Bis(aJrudirnethyl 4-hydroxybeneyl) phenol
Trivial name Dianin’s compound, codimer 2,4’-Bisphenol A
$::s.i’b. 156
Boiling pointa t 0.25 mm., oc. 165-1 70
108.5
170
4,4’-Bisphenol A
157
175
Trisphenol I
193
240-245
phase paper chromatography, which is particularly effective for estimating individual phenolic impurities in concentrations as low as 0.1% or less. GENERAL PROCEDURE
This method employs circular filter paper, impregnated with tricresyl phosphate, and an aqueous solution of trisodium phosphate as the eluting agent. The completed chromatogram is sprayed with a diazonium salt solution to develop color. The various phenolic derivatives are separated effectively by this system and appear as concentric rings of distinctive color and R, value. This well-known qualitative technique has been converted into a semiquantitative method by the following rather simple and straightforward method. The sensitivity of the chromatographic procedure to each of the principal impurities was established by chromatogramming successively more dilute solutions of each compound until its characteristic band could no longer be observed. By this means, the minimum amount of each impurity which could be detected was established. Unknowns were analyzed by a similar procedure using successively more dilute solutions until the extinction point was bracketed for each impurity. The concentration of each impurity in the original sample could then be calculated using the previously determined sensitivity value for that particular impurity. This surprisingly simple procedure is satisfactorily reproducible and sensitive when carried out under specifically defined conditions.
glass aspirator was used to spray the developed chromatograms. Reagents. Chemicals required were tricresyl phosphate, 0.125M aqueous trisodium phosphate solution, benzene, methanol, and p-nitrobensenediazonium fluoborate (Eastman P-7078). The compounds used for reference were 4,4’-bisphenol A, 2,4’-bisphenol A, trisphenol I , and Dianin’s compound. T h e latter were isolated from commercial bisphenol A by fractional crystallization and distillation, purity being established by paper chromatographic examination, infrared examination, and melting point determination. Properties of the pure compounds are shown in Table I. Procedure. The circular filter paper, with a 3 mm. X 4 cm. wick cut a t the center, was impregnated with tricresyl phosphate by immersion in a 9.1% (by volume) solution of this reagent in benzene for 5 seconds. It was then air-dried, taking care that the concentration of tricresyl phosphate was uniform throughout the paper. Two microliters of a methanolic solution of the sample was spotted 5 mm. below
the junction of the paper and the wick by means of a suitable pipet. The paper was then placed between two glass plates, the lower one having a center hole. The wick was allowed to protrude through the opening into the eluting solvent (0.125M Na3P04). The chromatogram was removed from the plates when the solvent boundary extended exactly 100 mm. from the center of the paper. After air drying, the chromatogram was developed with a saturated methanolic solution of p nitrobenzenediazonium fluoborate, applied as a spray. The most satisfactory elutions were obtained when the chromatograms were prepared in a constant temperature-humidity (25’ C. and 50y0 R.H.) chamber. The sensitivity of the procedure to the various impurities was determined by chromatogramming successively smaller amounts-Le., more dilute solutions-of the known compounds until the respective bands disappeared. Below a 4y0 concentration of a given impurity in 4,4‘-bisphenol A, the sensitivity of the analysis was found to be unaffected by the total concentration in bisphenol A and equal to the sensitivity determined for the impurity alone. The extinction points (minimum detectable amount in grams) for the major impurities under such conditions were: Dianin’s compound, 8 X 10-8; trisphenol I, 4.8 x 2,4’-bisphenol A, 6.4 X Unknowns were analyzed by this same procedure, the sample size being progressively reduced (by diluting the principal solution) until the extinction point had been closely bracketed. The sensitivity values were used, along with the “unknown” sample concentration a t the extinction point, to calculate impurity concentrations in the original sample. For greater convenience,
EXPERIMENTAL
Apparatus. Whatman No. 1, 24.0cm. circular filter paper, conditioned a t 50% relative humidity, was employed in all cases. Using a specially prepared die, a wick 3 mm. x 4 cm. was cut from the center of the paper. To support the paper during elution, two glass plates (9 X 9 X inch), the lower with a center hole to allow passage of the wick, were used. Small Petri dishes were employed as a reservoir for the eluting solvent. A 396
ANALYTICAL CHEMISTRY
Figure 1 . Correlation of sample concentration with impurity levels in bisphenol A as determined by semiquantitative paper chromatographic analysis
Table 11.
Phenolic Compounds Present in Commercial Bisphenol A
Identity or probable identity
R,
Structure
“9
Unknown Dianin’s compound
0.03 0.04
Color of band.
Purple Red
C’
Unknown Unknown * Trisphenol I
2,4’-Bisphenol A
H
O
q
T
0.07 0.08
0.13
Lavender Green Red
0.18
Purple
0.33
Pink
OH
Trisphenol I I
Table V. Repeated Analyses of Synthetic Bisphenol A Mixtures Containing One Impurity
Weight,
7%
2,4‘ isomer 4 > 3 . 8 3.8 3 . 8 2 < 2 . 2
Found, wt. yo 2,4’ isomer Trisphenol I
0.72-0.88
0.65
... ...
Table VI. Repeated Analyses of an Authentic Bisphenol A Sample
0.72-0.88 4,4‘-Bisphenol A
Trisphenol I
along with the corresponding structures, where known, are shown in Table 11. Because the individual impurities in bisphenol A rarely exceed 4’%, attention was concentrated on a procedure for detecting impurities a t or below this concentration. Particular emphasis was placed on the detection of the principal impurities associated with bisphenol A when present in less than 1% concentration. It is obvious that, in using the bracketing procedure described, the accuracy of a given series of analyses depends to some extent on how closely one desires to bracket the extinction point. I n general, it was found that the precision shown in Table 111 could be obtained easily without a n involved or extensive bracketing procedure. Data for a series of repeat analyses of synthetic mixtures by an experienced operator are shown in Tables IV and V. Repeatability of the procedure is indicated in Table VI, which shows the results of repeat analyses of a n authentic bisphenol A sample.
Repeated Analyses of a Synthetic Bisphenol A Mixture
2,4‘ isomer
Trisphenol I
Dianin’s
1 0
0.8
0.2
>0.94 0.94 0.94 0.71 0.71 0.71 0.18 0 . 1 8 0.18
