Search

FT-IR transmission spectra of liquids in well-made and firmly held cells with KBr or NaCl windows are usually very reproducible except that their baselines often show unexpected variations. To obtain absolute absorption intensities from these spectra such baseline differences must be corrected. The problem is illustrated with absorbance spectra of the 1036 cm−1band of benzene and the absorption index spectra calculated from them via the National Research Council of Canada program 46. Distinction is made between the experimental absorbance spectrum, the ideal experimental absorbance spectrum, and the absorbance spectrum, and a soundly-based method to correct the baselines is presented. We describe a modification of the NRC program 46 that effects the correction and calculates, on a laboratory computer, the complex refractive indices from a transmission spectrum of a liquid.The method is applied to 15 transmission spectra of the 1036 cm−1band of benzene. It improves the agreement between the 15 peak absorption index values obtained from these spectra from 4% to 2.5%, and improves the agreement between the baseline absorption index values from 5% to 0.01%, all percentages being of the peak value. A table of the average real and imaginary refractive indices and molar absorption coefficients is given. The average peak absorption index value and the area under the band agree closely with those obtained in 1980, and with earlier values from transmission or ATR measurements. These areas are all distinctly lower than those calculated from early measurements of dispersion. These results form the basis for a possible recommendation to the International Union of Pure and Applied Chemistry for a secondary intensity standard. We encourage others to measure this band to help ascertain that systematic errors do not significantly influence our results.To summarise our numerical results at 25 ± 1 °C and 1.0 cm−1nominal resolution: The average peak absorption index is 0.0710 with a 95% confidence limit of 0.0004, which is within 0.0001 of the less precise value measured in 1980 with a dispersive instrument that was calibrated against a primary intensity standard. The area under the absorption index band between 1095.8 and 912.6 cm−1is 1.358 cm−1with a 95% confidence limit of 0.009 cm−1, and that under the molar conductivity band is 3.69 ± 0.03 km/mol. These areas convert to 15.64 ± 0.1 km/mol and 6.79 ± 0.05 km/mol for the more commonly used measures of integrated absorption, the areas under the Naperian and decadic molar absorption coefficients, respectively.Keywords: infrared, absorption intensity, optical constants, FT-IR spectroscopy, benzene (liquid), intensity standard.

Accurate infrared absorption intensities of liquid chlorobenzene at 25°C are presented. Their accuracy was estimated from the agreement between the intensities measured by different spectroscopists using different instruments in different laboratories and by different spectroscopists using the same instrument in the same laboratory. The spectra from different spectroscopists have been averaged, unweighted, to give intensity spectra of chlorobenzene that are presented as the best available. The results are presented as graphs and tables of the molar absorption coefficient, Em(ν˜), and the real and imaginary refractive indices, n(ν˜)and k(ν˜), between 4800 and 450 cm−1. The peak heights and the areas under the bands in the absorption index (imaginary refractive index) spectrum are reported, as are areas under the molar absorption coefficient spectrum. Absorption index, k(ν˜), and molar absorption coefficient, Em(ν˜), values are believed accurate to an average ±2.4% at the peaks of bands with kmax> 0.002 and ±3.3% at the peaks of bands with kmax< 0.002. In the baseline k(ν˜)is accurate to ∼ ±5% above 3000 cm−1and ∼ ±2.5% below 3000 cm−1. The areas under bands in k(ν˜)and Em(ν˜)spectra for which kmax> 0.002 are accurate to ±1.3% on average. The real refractive index, n(ν˜), values are believed to be accurate to ±0.2%.

This paper presents accurate infrared absorption intensities of liquid toluene at 25°C. The accuracy is estimated from the agreement between the intensities measured by different spectroscopists using the same instrument in the same laboratory and also by different spectroscopists in different laboratories using instruments made by different manufacturers. The average agreement between integrated intensities over specific wavenumber ranges is about ±1.8%. The spectra from the different laboratories have been averaged, unweighted, to give intensity spectra of toluene that are presented as the best available. The use of data from different instruments in different laboratories has included the influence of systematic instrumental errors, so that the precision of the intensity data presented should be a better approximation to its accuracy than would be the case from an extensive study by one person on one instrument. The results obtained agree with the only measurements that have been made against a primary standard, the estimated accuracy of which is about 6%. The results are presented as graphs and tables of the molar absorption coefficient spectrum and the real and imaginary refractive index spectra between 6500 and 435 cm−1. The peak heights and the areas under the bands in the imaginary refractive index (i.e., absorption index) and molar absorption coefficient spectra are reported. The absorption index, k(ν˜), and molar absorption coefficient, Em(ν˜), values are believed to be accurate to an average ±2.5% at the peaks of 39 strong, medium, and weak bands and ±1.9% at the peaks of 51 very weak bands below 4100 cm−1. Above 4100 cm−1, 11 very weak bands have an average accuracy of ±1.3%. The baseline k(ν˜)values are accurate to between ±3 and ±10%. The areas under bands or band groups in k(ν˜)and εm(ν˜)spectra are accurate to 2.4% on average, or 1.2% for strong, medium, and weak band groups between 3150 and 775 cm−1with 0.002 < kmax< 0.112. The real refractive index, n(ν˜), values are believed to be accurate to 0.2%.

This is the last of four papers that present the detailed measnrements and results that led to the acceptance by the International Union of Pure and Applied Chemistry of Secondary Infrared Absorption Intensity Standards for liquids. In this paper accurate infrared absorption intensities of liquid dichloromethane at 25°C are presented. The accuracy was estimated from the ±1.5% average agreement of integrated intensities over specific wavenumber ranges between spectra measured by five spectroscopists in four laboratories. The use of data from different instruments in different laboratories has significantly included the effect of systematic instrumental errors. The spectra from the different spectroscopists have been averaged, unweighted, to give intensity spectra of dichloromethane that are presented as the best available. The results obtained agree with the only measurements that have been made against a primary standard, the estimated accuracy of which is about 5.5%. The spectra of the molar absorption coefficient and or the real and imaginary refractive indices are reported as tables and graphs between 6500 and 800 cm−1. Also reported are the peak heights and the areas under band groups in the molar absorption coefficient and imaginary refractive index spectra. The imaginary refractive index, k(ν˜), and molar absorption coefficient, Em(ν˜), values are believed to be accurate to an average ±2.3% over the 36 measured bands. The baseline k(ν˜) values are believed to be accurate to ∼8% below 6000 cm−1, ∼1% below 4500 cm−1, and ∼25% above 6000 cm−1, where the absorption is extremely weak. The areas under band groups in the k(ν˜) and Em(ν˜) spectra are believed to be accurate to 1.5% averaged over the 15 measured band groups and to 1.0% over the 10 band groups below 3600 cm−1. The real refractive index, n(ν˜), values are believed to be accurate to 0.2%.

This paper presents the results of a study to obtain accurate infrared absorption intensities of liquid benzene at 25°C. To achieve this we have determined the agreement between the intensities measured by different spectroscopists using the same instrument in the same laboratory and also by different spectroscopists in different laboratories using instruments made by different manufacturers. The agreement between integrated intensities over specific wavenumber ranges has been found to average about 2%. The spectra from the different laboratories have been averaged, unweighted, to give intensity spectra of benzene that are presented as the best available. The use of data from different instruments in different laboratories has reduced the influence of systematic instrumental errors, so that the agreement presented should be a better approximation to the accuracy of the intensities than would be the case from an extensive study by one person on one instrument. The intensity data presented agree with the only measurements that have been made against a primary standard, the evaluated uncertainty of which is about 6%. The results are presented in both graphic and tabular forms as spectra of the molar absorption coefficient, Em(ν˜), (also called the molar absorptivity, ε, formerly the extinction coefficient) and the real and imaginary refractive indices, n(ν˜) and k(ν˜), between 6225 and 500 cm−1. The peak heights and the areas under the bands in the imaginary refractive index spectrum are reported, together with the peak heights and the areas under the bands in the molar absorption coefficient spectrum. Imaginary refractive index, k(ν˜), and molar absorption coefficient, Em(ν˜), values are believed accurate to an average ±2% at the peaks of stronger bands, ±3.5% at the peaks of the weaker bands below 4100 cm−1, and ±2.5% above 4100 cm−1. The baseline k(ν˜) values are accurate to ∼10% above 1200 cm−1and 1% to 5% below 1200 cm−1. The areas under bands or band groups in k(ν˜) and Em(ν˜) spectra are accurate to 2% on average, or 1.5% when measured above a baseline for calibration purposes. The real refractive index, n(ν˜), values are believed accurate to 0.2%.

The finite impulse response operator (FIRO) method, developed recently by Jones and Shimokoshi [Applied Spectroscopy 37, 59 (1983)] has been applied to the deconvolution of Mossbauer spectra. The overlapped bands of the Mossbauer spectra of iron oxides, formed in the air oxidation of an aqueous suspension of Fe(OH)2, have been deconvolved by the FIRO method in order that the correct positions and intensities of the lines might be defined. The peak positions so determined can be used with good reliability as initial estimates for the customary method of profile determination by a least-squares fitting routine.

The directly observable information in a digitally encoded spectrum can be selectively enhanced by deconvolution with an appropriate filter operator. This may be carried out on the frequency (wavenumber) domain spectrum or on its Fourier transform. Depending on the choice of the filter, one may achieve: (1) reduction in the random noise, (2) reduction of systematic instrumental errors, (3) enhancement of the resolution. These general deconvolution procedures are briefly reviewed in terms of this classification. Recent work at the National Research Council of Canada at Ottawa has emphasized the application to band resolution enhancement (self-deconvolution), principally with respect to Fourier transform infrared spectra. Experience in our laboratory with infrared data generated on dispersion spectrometers is described. Our results confirm those reported by Kauppinen et al.at Ottawa, but the need for caution in applying self-deconvolution in routine laboratory measurements is stressed.

The method of self-deconvolution is applied to the resolution enhancement of electron spin resonance spectra (ESR). The component lineshapes of ESR spectra are basically Lorentzian, but they often carry strong Gaussian perturbations. Deconvolution with a finite impulse response operator filter to enhance the resolution of Lorentzian and Gaussian ESR band envelopes is discussed and a more general function is derived for Gauss-Lorentz convolute (Voigt) lineshapes. The method is applied initially to the simulated isotropic ESR spectrum of the ethyl radical in the liquid state and to a simulated anisotropic spectrum of a crystalline powder. Application to real spectra is illustrated by the anisotropic spectrum of bis(acetylacetonato)-copper(II) [Cu(AcAc)2] in ethanol solution at 140 K in order to resolve the hyperfine lines. The perpendicular component of this spectrum is notable for its poor resolution.

This paper completes the report of our FT-IR measurements of transmission spectra of organic liquids. These measurements were undertaken in order to estimate the photometric accuracy of our FT-IR spectrometer for transmission spectra by comparing the results with those measured previously [Applied Spectroscopy 34, 657 (1980)] by a calibrated dispersive spectrometer. We have studied 97 bands of chlorobenzene, toluene, and dichloromethane, in 500-μm cells and 54 bands of benzene, chlorobenzene, toluene, cyclopentane, and dichloromethane, in cells 11 to 53 μm thick. The average value of the imaginary refractive index at the band peak, kmax, is reported for each band and compared with the calibrated value. When averaged over the 184 bands that are reported in this work and our earlier paper [Applied Spectroscopy 39, 401 (1985)], our measurements agree with the calibrated values to about 3% of kmax, and have a 90% confidence limit of about 0.7 to 1% of kmax, The calibrated values themselves have an estimated error of about 6%, so to obtain better knowledge of the accuracy of our measurements we need better standards and we need measurements to be made on other FT-IR instruments. We plan to publish simplified procedures that will encourage others to make such measurements. Our 90% confidence limit includes effects due to the use of different cells, daily realignment of the fixed mirror of the interferometer, realignment of the optics of the detector and sample compartments, change of beamsplitter, and a day-to-day variation of uncertain origin, in addition to the error sources that always contribute to typical analytical precision. Our data do not show these additional error sources to be systematic, so it is appropriate to include their effect in the evaluation of the confidence limit. We believe the main source of error to be the effect of the infrared cell on the path of the light beam through the instrument. The day-to-day variation of uncertain origin is thought to be due to very small changes in the beam path through the instrument, which in turn cause the effect of the infrared cell on the beam to vary.

Extensive purification of an aldosterone fraction obtained after two chromatographic separations (Zaffaroni's propylene glycol/toluene system and Bush C system) from crude neutral extract is described. Highly purified aldosterone was isolated in four instances. At least four other compounds which may be related to aldosterone were found in this fraction. A procedure is proposed for the isolation and determination of highly purified aldosterone.

The isolation and the partial chemical and physicochemical characterization of a chloroform-soluble substance obtained from human urine and citrous fruit juices are presented. The urinary compound, designated as compound III, is present mostly as a conjugate hydrolyzed by animal β-glucuronidase whereas it is in free form in fruit juices. It is present in the urine of normal males and females and in a number of pathological conditions. Its urinary excretion is significantly increased as a result of diets rich in potassium (250 meq./day). In various paper chromatographic systems, the compound has a mobility similar to that of aldosterone. Detailed study of its chemical and physicochemical characteristics, including oxidative degradation and infrared and elementary analysis, suggests a steroidal or sesquiterpinoidal structure. However, at the present time, the data obtained do not permit the final elucidation of the chemical formula of compound III.

A reflecting microscope designed for use in front of the entrance slit of a Perkin-Elmer single-beam infrared spectrometer is described. It is shown that the heating effect of the condensed beam on the sample is not significant except in the case of compounds which absorb strongly in the 3-micron region of the spectrum, and then only if the sample is unsupported. The instrument is being applied to the study of small quantities of material in solution and to single crystals. Spectra adequate for identification purposes may be obtained from less than 10 micrograms of a sample in solution. The minimum sample size is limited at present by the technique of sample transfer and if this can be improved, spectra could be obtained with about 1 microgram. Preliminary results using polarized radiation in the study of fatty acid crystals are also presented.

A method is described for the spectrographic analysis of conjugated proteins, formed by the interaction of serum proteins with the isocyanates of certain polynuclear aromatic hydrocarbons. The method makes possible the determination of the number of hydrocarbon groups introduced into each protein molecule, provided the hydrocarbon chromophore has maxima at wave-lengths greater than 3250A. At wave-lengths less than 3250A, where the chromophoric groups of the normal protein also absorb, the total absorption is shown not to be equivalent to the sum of the absorptions of the several chromophoric groups involved. The spectra of binary solutions of the unconjugated protein and a simple, water soluble, compound containing the hydrocarbon chromophore, exhibit deviations from the calculated summation spectra which are of a similar character to those exhibited by the conjugated proteins. It is shown that the scattering of light by the protein molecule plays no direct part in this effect. The absorption spectra of mixtures of l-tyrosyl-l-tyrosine and horse serum albumin do not exhibit abnormal behavior; the observed spectra are closely similar to those calculated from the summation of the spectra of the two solutes. Although the abnormalities in the spectra of protein solutions do not appear to be of a general character, attention is drawn to the possibility that effects of this nature may occur in other solutions containing proteins. Care must be exercised, therefore, in the application of spectrophotometry and colorimetry to the analysis of substances in the presence of proteins.

Brief descriptions are given of twenty two modular computer programs for performing the basic numerical computations of absorption spectrophotometry. The programs, written in Fortran IV for card input and output, are available from the National Research Council of Canada. The input and output formats are standardized to permit easy interfacing to yield more complex data processing systems. Though these programs were developed for ir spectrophotometry, they are readily modified for use with digitized visual and uv spectrophotometers. The operations covered include ordinate and abscissal unit and scale interconversions, ordinate addition and subtraction, location of band maxima and minima, smoothing and differentiation, slit function convolution and deconvolution, band profile analysis and asymmetry quantification, Fourier transformation to time correlation curves, multiple overlapping band separation in terms of Cauchy (Lorentz), Gauss, Cauchy-Gauss product, and Cauchy-Gauss sum functions and cell path length determination from fringe spacing analysis.