660 Analyst, October, 1968, Vol. 93, &5. 660-662 The Rapid Determination of Benzil, Benzoin and Hydrobenzoin in Mixtures by Quantitative Infrared Spectrophotometry BY P. R. FALKNER, G. DAVISON AND G. B. STOKER (John Dalton College of Technology, Manchester) From the infrared spectra of mixtures of benzil, benzoin and hydrobenzoin in solution in chloroform, benzoin and hydrobenzoin are determined by means of their absorption bands a t 3460 and 3590cm-1, respectively. Benzil is determined from the carbonyl absorption at 1680 cm-l after subtracting the contribution of benzoin to the absorbance at this frequency. The most suitable concentrations are 0.1 to 1.0 per cent. for benzil and benzoin and 0.1 to 2.0 per cent. for hydrobenzoin. IN a study of the reduction of aromatic diketones, a method was required for the rapid deter- mination of benzil (C,H,.CO.CO.C,H,), benzoin (C,H,.CHOH.CO.C,H,) and hydrobenzoin (C,H,.CHOH.CHOH.C,H,). Several methods have been described in the literature for the determination of these substances singly, including titrimetric,l,2 gravimetric,2 s3 p~larographic~ and fluorimetric3 methods for benzil, and colorimetric methods5p6 for benzil and benzoin, but none of these appeared to be readily adaptable to analysis of three component mixtures.Ultraviolet spectral results have been reported for benzil,' benzoin8 and hydroben~oin,~ but because of overlapping of absorption peaks in a mixture, ultraviolet spectroscopy did not appear to be suitable for the present purpose. Although infrared spectral results have previously been reported for benzil,1° benzoinll and hydrobenzoin,l2 no quantitative study of mixtures of these substances appears to have been undertaken. It was therefore decided to investigate the possibility of quantitative analysis of benzil, benzoin and hydrobenzoin mixtures by infrared spectrophotometry. EXPERIMENTAL Infrared spectra were recorded by means of a Hilger & Watts Infrascan H900 double-beam spectrophotometer, with a grating and sodium chloride prism.A 1.00-mm path length, semi-permanent cell with sodium chloride windows was used throughout. To reduce, as far as possible, hydrogen-bonding effects for benzoin and hydrobenzoin, dilute solutions (about 0-1 to 2-0 per cent. w/v) of the materials in purified chloroform were used. The solvent used was analytical-reagent grade chloroform, which contains about 2 per cent.of ethanol as stabiliser. This was washed three times with an equal volume of water, dried over a mixture of anhydrous analytical-reagent grade sodium sulphate and analytical-reagent grade calcium chloride, distilled between 60.0" and 60.5" C and stored in amber bottles in the dark. Chloroform thus purified gave no detectable infrared absorption at 3460 cm-1 attributable to the hydroxyl group of ethanol. A small absorption at 3590 cm-1 remained, but the percentage transmission was shown by repeated checks to be constant, and therefore not liable to interfere in the procedure described below. Absorbance measurements were made by using the "Baseline Technique," vix., con- struction of a tangent from the troughs immediately on either side of the required absorption peak, followed by measurement of the vertical distance from the peak to the base-line or tangent. Only 500-ml amounts were prepared at one time.0 SAC and the authors.FALKNER, DAVISON AND STOKER 661 CALIBRATION PROCEDURES AND BASIS OF QUANTITATIVE ANALYSIS- Initially, separate calibration graphs were prepared for each component in the concen- tration range about 0.1 to 1.5 per cent. The peaks used were: for benzil, the carbonyl stretch- ing absorption at 1680cm-1; for benzoin, both the carbonyl peak at 1680cm-l and the hydroxyl stretching absorption at 3460 cm-1; and for hydrobenzoin, the hydroxyl stretching absorption at 3590 cm-1. It will be noted that the hydroxyl peaks of benzoin and hydro- benzoin are separated by 130cm-1 and are thus readily distinguishable.In all instances plots of absorbance against concentration were linear, and passed through or close to the origin. The reproducibility of the absorbance measurements was checked by running repeated spectra. Thus in a typical series of eight measurements of the absorbance a t 1680 cm-1 given by 0.1 per cent. w/v of benzil, a mean value of 0.214 was obtained, with a standard deviation of 0.0015 and a coefficient of variation of 0-70 per cent. Standard mixtures were then prepared containing benzil and benzoin in the concentra- tion range 0.1 to 1.0 per cent. w/v and hydrobenzoin in the range 0.1 to 2.0 per cent. w/v. These concentration ranges were chosen primarily to give reasonable percentage transmissions on the recorded spectra.In particular, the intense carbonyl absorption at 1680cm-l renders the use of higher concentrations of benzil and benzoin unsatisfactory. With the standard mixtures, calibration graphs were plotted for benzoin and hydrobenzoin from the absorbances at 3460 and 3590 cm-1, respectively; these graphs were almost identical with those for the single components. The 1680 cm-1 absorbance, however, now represents the combined contributions of benzil and benzoin. It is therefore necessary to use an indirect procedure for the benzil calibration. As in the spectrum of benzoin the absorbances at 1680 cm-1 and at 3460 cm-1 are linearly related to concentration, and hence to each other, it is a simple matter to calculate the contribution of benzoin to the 1680 cm-l absorbance from the measured absorbance at 3460 cm-l.Thus by subtraction of this contribution from the total absorbance at 1680 cm-l, the absorbance caused by benzil at this frequency can be obtained. As a check on the validity of this procedure, the strict additivity of the contri- butions of benzil and benzoin to the total absorbance at 1680 cm-1 was confirmed by using a series of binary mixtures of these substances. From a series of runs on the standard mixtures, linear calibration graphs passing through the origin were obtained in all instances. The actual absorbances corresponding to a con- centration of 1.00 per cent. w/v of each component were as follows. Hydrobenzoin at 3590 cm-l .. .. .. .. .. .. . . 0.472 Benzoin at 3460 cm-1 .. .. . . .. .. .. .. . . 0.160 Benzoin at 1680 cm-l, in absence of benzil . . .. .. .. . . 1.605 Benzil at 1680 cm-l, after allowance for contribution of benzoin, when present 2.250 As a measure of the accuracy of the method, several synthetic mixtures were treated as unknown samples, and the results of these analyses are shown in Table I. TABLE I DETERMINATION OF BENZIL, BENZOIN AND HYDROBENZOIN IN SYNTHETIC MIXTURES Benzil, per cent. w/v taken found 0.10 0.10 0.10 0.11 0.20 0.21 0.25 0.26 0.30 0.29 0.30 0.28 0.40 0-39 0.40 0-40 Benzoin, taken 0.10 0.20 0.40 0.10 0-30 0.40 0.10 0-20 per cent. w/v found 0.09 0.19 0.39 0.09 0.30 0.37 0.09 0-18 H ydrobenzoin, per cent. w/v taken found 0.25 0-27 0.80 0-82 0.40 0.40 0.25 0.26 0.40 0.40 0.80 0.84 0-60 0.63 0.60 0.62 DISCUSSION AND CONCLUSIONS The method as described was found to be rapid and convenient once the initial calibra- The optimum concentrations were within the range 0.1 to 1.0 More tions had been carried out.per cent. w/v for benzil and benzoin and 0.1 to 2.0 per cent. w/v for hydrobenzoin. concentrated mixtures should, of course, be diluted to these concentration ranges.662 FALKNER, DAVISON AND STOKER The values for percentages found and taken agreed, in general, to within t0.02 per cent. a t the lower concentrations and k0.05 per cent. at the higher concentrations, which was satisfactory for a rapid method for determining all three components. Greater deviations occurred occasionally, particularly when one component , not necessarily the one being determined, was present in excess of the recommended concentration range.The deter- mination of benzil, by the indirect procedure involved, is particularly susceptible to error if the concentration of benzoin is substantially greater than that of benzil. In spectroscopic methods of this type, errors can arise because of physical or chemical interactions in the solution, or because of uncertainties in the measurement of absorbances. In the present instance, although intermolecular hydrogen bonding is possible between -OH of benzoin and -OH of hydrobenzoin, or between these -OH groups and the >CO of either benzoin or benzil, it is unlikely to be pronounced at such low concentrations. Furthermore, such hydrogen bonding would be expected to cause an appreciable displacement of the absorption peaks, which were constant to within k2 cm-l throughout.Such errors are likely to be greatest with low percentage transmissions, which may be found par- ticularly at the intense 1680 cm-l peak, and when the spectral background causes uncertainties in the construction of the base-line for absorbance measurement. Therefore, calibration with standard mixtures of composition similar to those to be analysed is regarded as essential. It is possible that greater accuracy could be achieved by means of a preliminary separa- tion of all three components, e.g. , by thin-layer chromatography and subsequent elution. This aspect is currently being investigated, but any gain in accuracy will inevitably be accompanied by a substantial increase in the time required for the analysis. The authors thank the directors of Messrs. Beck, Koller & Co. for permission to use their Infrascan H900 double-beam spectrophotometer in connection with this work. Uncertainties in measurement of absorbances are a more likely source of error. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. REFERENCES Bottei, R. S., Analytica Chim. Acta. 1964, 30, 6. Ruzhentseva, A. K., and Metsler, A. A., Zh. Analit. Khim., 1950, 5, 160. Sass, S., and Goldenson, J., Analyt. Chem., 1951, 23, 540. Rogers, W., jun., and Kipnes, S. M., Ibid., 1955,27, 1916. Nagai, Y., Kagaku., 1954, 24, 417. Vonesch, E. E., and Guagnini, 0. A., An. Asoc. Quim. Argent., 1955, 43, 62. Bottari, F., and Carboni, S., Gazz. Chim. Ital., 1957, 87, 1281. Rumpf, P., and Gillois, M., Bull. SOC. Chim. Fr., 1955, 1348. Kland-English, M. J., Summerbell, R. K., and Klotz, I. M., J . Amer. Chem. Soc., 1953,75, 3709. Blout, E. R., and Abbate, M. J., J . Opt. Soc. Amer., 1955, 45, 1028. Luttke, W., and Marsen, H., 2. Elektrochem., 1953, 57, 680. Sadtler Standard Spectra, No. 8333, Sadtler Research Laboratories Inc., Philadelphia, P.A. Received March 4th, 1968 19104, U.S.A., 1966.