which then undergoes self-condensation. It can be seen by examination Of Lilar that the steric requirements for formation of the keto1 of @-naphthol are more severe than for the a-isomer. The 6-methyl substituent of 6-methyl%pyridine aldehyde may hinder the formation of a cyclic complex between the organic and inorganic moieties and thereby inhibit the reaction. 2-Pyridine aldehyde has no such hindrance and gives a positive test. An additional factor not connected with the proposed mechanism may be responsible for the negative reactions of the carbohydrates and cyclic p-diketones. Both of these types of compounds may be degraded by t’he strongly alkaline test solution. The carbohydrates may yield lactic and saccharink acids (9) and the cyclic p-diketones may undergo the alkaline cleavage reaction (6) prior to self-condensation.
ACKNOWLEDGMENT
The authors thank the Climax Molybdenum Co. for samples of Some of the heteropoly acids. LITERATURE CITED
( 1 ) Billman, J. H., Borders, D. B., Buehler, J. A., Ind. A d . Sei. 65, 68 (1955). ( 2 ) Boltz, D. F., Lueck, C. H., “Colorimetric Determination of Non-metals,’.’
g5i;F.l$,eg;;tf‘;, ;2ijy i n F t t , n c e j
( 3 ) Burnstein, S., ANAL.CHEW 25, 422 (1953). ( 4 ) Emelius, J.7 Andemon, J. s., “ m d e r n Aspects of Inorganic C h m istry,” 3rd ed., p. 335, Van Nostrand, New York, 1960. (5) Folin, O., Denis, W., J . B i d . Chem. 12,239 (1912); Ibid., p. 245. ( 6 ) FLlSon, R. c., “Advanced Organic Chemistry,” PP. 462-5, WileY, New Yo&, 1950. ( 7 ) Hassall, C. H., O V . Ihctions 9, 73 (1957). (8) Heard, R . D. H., Sobel, H., J . Biol. Chem. 165, 687 (1946). ( 9 ) Hickson, J. L., “The Encyclopedia of Chemical Technology,” R. E. Kirk,
D. F. Othmer, eds., 1-01. 13, p. 239, Interscience, New York, 1954. (10) Jean, M., Ann. Chim. (12) 3, 470
(1948). (11) Jean, M., Chim. Anal. 37, 125 (1955); Ibid., p. 163. (12) Lo, C., Chu, T. J., IXD.ENG.CHEM., ANAL.ED. 16, 637 (1944); Folin, O., Wu, H., J . Biol. Chem. 41, 367 (1920). (13) STcElvain, S. bl., “The Charyterization of Organic Compounds, Macmillan, pp. 44-79, New York, 1953. (14) North, E. O., “Inorganic Syntheses,” 5 ~ ~ I, 1 . H. s. Booth, ed., pp. 129-32 McGraw-Hill, Yew York, 1939. . (15) Shriner, R. L., Fuson, R. C., Curtin, D. Y., “The Systematic Identification of Organic Com ounds,” 4th ed., pp, 63-90, Wiley, 8ew York, 1956. (16) Souchay, P., Faucherre, J., Bull. Soc. Cham. France (Memoires) 18, 355 (1951): (17) Strickland, J. H. D., J . Am. Chem. Soc. 74, 862 (1952). (18) Vignoli, L., Crietau, B., Pfister, A., Ibid., 78, 392 (1956). (19) Wadlein, C., Pllellon, 11. G., ANAL. CHEM.25, 1668 (1953). (20) Wu, H . , J . Biol. Chem. 43, 189 (1920).
RECEIVEDfor review >lay 14, 1962. Resubmitted November 3, 1964. Accepted December 4, 1964.
EfFects of Interference Fringes in Infrared Absorption Cells JOHN U. WHITE White Development Corp., Stamford, Conn.
WILLIAM M. WARD Applied Physics Corp., Monrovia, Calif.
b When the index of refraction of an absorption cell is not exactly equal to that of the sample contained in it, interference fringes are formed inside the sample. Their intensities may become analytically significant when the relative index of refraction between the two exceeds about 1.05. Ways of correcting for them are discussed; their effects can be eliminated by suitable choice of the reference points for intensity measurements by the base line method.
I
application of infrared spectroscopy to analysis, interference fringes have long been used to measure the thickness of liquid absorption cells (9). Such fringes are also observed when slightly absorbing liquids are measured in cells with high index windows like silver chloride or germanium. I t is less generally recognized that they may have significant effects in lower index cells containing common organic materials. N THE
268
ANALYTICAL CHEMISTRY
The intensities of fringes in absorbing dielectric layers have been calculated in detail for the case of evaporated dielectric coatings (3) and for the determination of integrated absorption coefficients (6). I t may be shown that to a first approximation the relative peak to peak intensity of the interference fringes, A I / I o , is given by:
lo
=
4RTa
(1)
where T is transmittance and R is reflectance given by the Fresnel equation:
where n is the relative index of refraction between that of the window, n,,, and that of the sample, n,. Typical values might be n, = 1.500, n, = 1.427 or 1.357, giving relative indices of 1.0513 and 1.1053 with reflectivities of 0.000625 and 0.0025, respectively. The cor-
responding peak to peak fringe heights with T = 1 are 0.0025 and 0.01 I o . With T less than 1, the height of the fringes falls off as T 2 ,generally making them negligible at 50% absorption or more. With T less than 1, the height of the fringes falls off as T 2bccause the doubly reflected beam makes two extra passes through the absorbing sample. Even after the fringes have become negligible, the reflections a t the windowsample interfaces are still present. When the relative index of the sample is large, as in the 675-cm.-l band of benzene, the losses may be significant. Their effect is then to make the band appear too dense and sometimes distorted. The general practice of avoiding weak bands for analytical points usually avoids interference fringes, though with considerable reduction in the range of absorbance measurable without changing cell thickness. Empirical Deer’s law calibrations take account of the reflection loshes at lower transmittance. The intensity expected for the fringes
in typical cases may be estimated from the dispersion curves published for benzene ( 8 ) ,carbon tetrachloride ( I , 5 ) , carbon disulfide ( I , 5 ), bromoform (7) , cyclohexane (51, acetone ( 5 ) , n-heptane ( 4 ) , perfluorotrimethylcyclohexane ( 5 ), potassium bromide (IO), and rock salt ( d ) , which have been plotted in Figure 1 together with lines indicating the refractive indices for which the peak to peak fringe intensities at 100% transmittance are 0.25 and 1% in a KBr cell. A few of these compounds fall within the 0.25% lines where fringes might be considered negligible. Many others have more than 1% fringes. I n regions of anomalous dispersion like the 15-micron benzene band, the reflectance may be as great as 4%, and large distortions may be expected if the spectrum is measured in a thin cell. Even a 10% benzene solution in carbon disulfide should show interference effects greater than 1% a t 14.9 microns in a rock salt cell. To reduce the magnitude of the fringes, the best method is to use windows of the correct refractive index. I n solutions where the refractive index is known, some improvement can be made in this direction. Thus for dilute carbon disulfide solutions in the 8- to 14-micr~n region KBr windows give appreciably smaller fringes than KaCl. K I with an index of 1.62 would give essentially none. Beyond 15 microns KBr is an excellent match for carbon tetrachloride. I n the short wavelength region, rock salt is a slightly better match for carbon tetrachloride than K B r ; fluorite and BaF2 are both much better than KBr or KaC1. For acetone the refractive index is so low that AUOrite or BaFz cells should be used with it whenever possible. Some increase in the permissible amount of mismatch between the refractive indices of the windows and of the sample is achievable by the use of two different window materials in the absorption cell. With a combination like K I and BaFz windows, no fringes result when the sample matches either one. The peak to peak fringe heights are less than 0,5y0over a slightly larger range than for rock salt. I n practice the very limited choice of materials precludes more than small improvements. The existence of interference fringes inside the sample is usually concealed by absorption bands. They may be identified by varying the cell separation at fised wavelength. I n the absence of interference the absorbance changes linearly; in its presence there is a superimposed fringe pattern. The absorption cell used for these measurements had the special feature that the windows came together without turning. Scratches were thus eliminated when the cell was closed to con-
Figure 1 .
Indices of refraction of various materials
Dashed lines porallel to curve for KBr represent extreme values for which peak to peak heights of interference fringes ore less than 0.25 end 1 in KBr cell
%
----- . . -. --......,.,. --- -
Potassium bromide Sodium chloride Benzene Carbon tetrachloride Carbon disulfide Bromoform Cyclohexane Acetone Perfluorotrimethylcyciohexane
tact, and the resulting use of the same parts of the windows at all thicknesses greatly reduced the effects of window imperfections. The cell had Teflon seals in a stainless steel cylinder and a scale calibrated in 5-micron units readable and reproducible to 1 micron. All measurements were made in the Cary-White Model 90 infrared spectrophotometer (11). Table I lists the fringe heights and spacings measured at a fixed wavelength of 16 microns for some combinations of windows and samples. The empty cell gave 9% with a spacing of 8 microns. The CS2 sample in salt windows gave 0.6%, which dropped to 0.1% in KBr windows and 0.25% in a cell with one window of each kind. Similarly, cyclohexane was better in the cell with mixed windows than in the KBr cell. The fringe spacings for CS? were consistent with its published refractive index of 1.60 at a wavelength of 15 microns. The fringe spacings for cyclohexane were consistent with a refractive index around 1.4, which value might be expected from its measured one of 1.41 a t 6 microns. The fringes may also be observed by making scans of the same spectrum a t increments of thickness equal to about a quarter of a wavelength of the light inside the sample. Figure 2 shows tracings of three scans of cyclohexane between KBr windows in the 8- to 10micron region at thicknesses of 103.3, 105.0, and 106.2 microns. T h e solid curves corresponding to the two extreme
settings are parallel throughout the region, showing that the 2.9-micron difference in thickness is very close to a half wavelength of the radiation inside the sample. The phases of their interference patterns are essentially the same at every point. The intermediate spectrum corresl?onds to about a quarter of a wavelength difference in thickness. It is half a wave out of phase with the others and oscillates up and down compared to them. The intensity of the absorption in this region separates the three spectra enough to make the outer two act as a n envelope for the middle one. The approximate points of tangency are marked because of their significance in base line absorbance measurements. T h e upward marks indicate maxima in the interference pattern of the middle
Table 1. Interference Fringes in Absorption Cells at 16 Microns
Sample None
cs2
Cyclohexane
Windows 2 KBr
2 NaCl 2 KBr 1 KBr, 1 NaCl 2 KBr 1 KBr, 1 NaCl
Peak to peak height,
Fringe spac!ng,
9
8 5 5
yo
0 6 0.1
0 2 0 2
microns
5
6
