Creator:C.E. Weir, A. Van Valkenburg, and E.N. Bunting Date Created:October 5, 1959 Place Created:Maryland Keywords:infrared absorption of solids Context:aticle from Spectrochimica Acta ************************************************** Spectrochimica Acta, 1960, Vol. 16, pp. 58 to 73. Pergamon Press Ltd. Printed ill Northern Ireland Studies of infrared absorption spectra of solids at high pressures E. R. Lippixcott Department of Chemistry, University of Maryland and C. E. Weer, A. Van Valkenburg and E. N. Bunting National Bureau of Standards, Washington, D.C. (Received 5 October 1959) Abstract—Infrared spectra of solids were studied with a diamond pressure cell in the wavelength range 5-15 fi at pressures between 1 atm and 50,000 atm. The calibration of the cell at the 14,000 atm transition of NaN02 is described. Spectra were studied for aromatic organic compounds, inorganic hydrates, and ammonium halides. In general, band shifts produced by pressure were to higher frequencies and at most 10 cm_1/10,000 atm. Many bands exhibited large changes in intensity. Occasionally bands increased in intensity or were unaffected but in general a decrease in intensity was observed at elevated pressure. Representative spectra are given, one to a pressure of 160,000 atm. Suggestions for the causes of the frequency shifts are given. Introduction In an earlier paper [1] a diamond high-pressure cell was described which was used to obtain infrared spectra in the wavelength region 1-15 fi to pressures of 30,000 atm. The use of this high-pressure cell was illustrated with studies on calcite. Since then, studies have been made on a wide variety of solid substances, both organic and inorganic. The purpose of this paper is to describe some of these results. The infrared spectrum of a substance shows a number of changes on application of pressure; these include shifts of absorption bands to both higher and lower frequencies from the positions of the band at 1 atm, the occurrence of new bands, the splitting of degenerate bands arising from a change in selection rules, and changes in apparent band intensity. The shifts in absorption bands observed to date range up to a maximum of ±10 cm-1 per 10,000 atm pressure with greater shifts sometimes occurring for substances involving systems of hydrogen bonds. With the exception of these latter systems, shifts of frequencies on application of relatively low pressure have not been studied in detail in our work to date. Similarly, the splitting of absorption bands and the appearances of new bands seem to be relatively rare occurrences at pressures below 20,000 atm. However, changes in apparent intensity of many bands were observed at relatively low pressures, i.e. 10,000 atm, -with larger changes occurring at higher pressures. Since the magnitude and direction of these intensity changes were unexpected, they will be illustrated and discussed in considerable detail. The change of intensity is specific both with respect to the nature of the substance and the mode of [1] C. E. Weir, E. R. Lippincott, A. vai* Valkenbubg and E. N. Bunting, J. Res. Nat. Bur. Standards A 63, 55 (1959). 58 Studies of infrared absorption spectra of solids at high pressures vibration involved. Examples will be given in which the intensity decreases, some in which it increases, and others in which it does not change more than the experimental error. However, the greatest number of cases observed to date have been those which show a decrease in band intensity. The diamond pressure cell has been used to pressures as high as 160,000 atm and in the wavelength range 1 fx-'M) /a. One example of spectra taken at the higher pressure will be presented and discussed. The data at the longer wavelengths are incomplete at the present time. Experimental The pressure equipment consists of two type II diamonds which are transparent from 1 to 4 /j, and from 5-2 ji to beyond 30 fi [1], The specimen is compressed between two flat surfaces ground on the diamonds. The incident beam traverses both diamonds and the specimen parallel to the direction of stress. Pressures have been calculated as force per unit area. Since the sample is held between the diamonds by frictional forces resisting flow in a thin specimen, there is a question as to whether the applied pressure is hydrostatic. Fortunately, the thickness of the specimen retained after extrusion ceases is usually just adequate to obtain a reasonably characteristic spectrum. For organic compounds this thickness is sometimes too small to observe weak bands, while for some inorganic salts it appears somewhat too large in the case of strong bands. The question of the hydrostatic nature of the pressure and the magnitude of the pressure gradients near the edges of the specimen makes it important that the pressure calculated as force per unit area should be checked by a calibration based on a phase change for a given substance at a known hydrostatic pressure. This type of calibration is surprisingly difficult because most solid-solid transitions appear to produce such small discontinuous changes in the infrared spectrum that they are difficult to identify with present experimental techniques. The calibration point which has been most useful to date has been that for a transition in NaN02 which is found to take place at a hydrostatic pressure of 14,500 atm by a linear interpolation of Bridgman's data [2]. This transition may be studied in the infrared spectrum by observing the behavior of the 825 cm-1 (12-1 fj,) absorption band of NaN02 corresponding to the nitrite ion bending mode of vibration. The frequency shifts discontinuously to near 855 cm-1 (11-7 [i) when this transition takes place. An example of this is illustrated in Fig. 1. In these experiments the pressure was raised to 14,000 atm, lowered a few hundred atmospheres, raised again to 14,000 atm, and the process repeated several times, always approaching the transition from the low pressure direction. This transition served to check the pressure calibration in terms of the applied load and its sharpness furnished evidence that excessive pressure gradients did not exist in the specimen. The absence of appreciable gradients provides evidence that the pressure is reasonably hydrostatic in nature. The transition was sometimes difficult to initiate but once obtained it was readily reversible. By following the transition with different sets of diamonds it [2] P. W. Bbxdgman, Proc. Am. Acad. Arts. Sci. 72, 45 (1937). 59 E. . Lippincott, C. E. Wei, A. van Vand E. N. Bting was possible to obtain information as to the condition of the diamond surfaces. In some instances the transition was found to occur over a narrow pressure range (±300 atm) with only a single phase present at any time, while in other cases the 855 and 825 cm-1 bands were present simultaneously over a rather wide pressure range indicating the presence of a considerable pressure gradient. This latter situation usually occurred when the diamond surfaces had deteriorated markedly wavenumber.cm" 1 900 800 wavelength,microns Fig. 1. Behavior of the 825 cm-1 nitrite ion bending frequency for the transition occurring in NaN02. with the presence of a number of surface flaws or with diamonds of very small surface area. In such cases it was desirable to regrind the diamond surfaces as described in [1], Sample handling was accomplished by the techniques described previously [1]. In a typical run, the piston with one of the diamonds was inserted in the bearing and a small quantity of specimen placed on the surface with a spatula. The other piston was inserted and the thrust plate screwed into place. The pressure was then raised to a few thousand atmospheres to produce a film between the diamond faces. The pressure cell was then placed in the focal point of the beam-condensing unit and its position adjusted in the beam to produce a maximum transmission at a wavelength where the specimen had no absorption bands. The pressure was then raised and lowered until no further evidence was found for extrusion of the sample from between the diamond surfaces. The pressure was then raised to the maximum pressure desired and the spectrum run with subsequent runs at successively lower pressures. When the lower pressures were reached, the whole process was repeated by starting at the higher pressures. In most cases the changes which occurred in the spectrum were reversible when run under these conditions. 60 Studies of infrared absorption spectra of solids at high pressures In a few experiments there was evidence of irreversible changes in the spectrum occurring at the highest pressures (50,000 to 160,000 atm) in that new bands appeared which remained when the pressure was reduced to 1 atm. These irreversible changes may have been caused by a chemical reaction. Examples of these spectra will not be reported until they are investigated further. Since the diamond faces had an area of from 1 to 2 X 10-4 in2, only a portion of the incident i.r. beam could be accepted by the cell. Thus, it was necessary to restrict the reference beam to permit utilization of the full scale of the recorder. This was accomplished by placing a suitable screen (or perforated sheet-aluminum) in the reference beam. The low available energy made it necessary to operate at high gain and to use a slow scanning speed. Scanning speeds used varied from 0-08 to 0-5 /i/min with a slit program usually from three to four times that for the standard program of the instrument. In studying a given specimen, all instrumental settings were left unchanged. Because of the use of high gains, some of the recorded spectra showed considerable noise. Errors in reading positions of absorption bands may occur because of the finite rate of scan and the limited amount of energy available. In cases where the exact position of a band was desired, the scanning speed was reduced, or the position was obtained from a manual setting. Since most of the work reported here is not concerned with the quantitative behavior of the frequency shifts, no special effort was made to measure the positions with greater accuracy than that available from the normal scanning rate of the spectrum. A number of sources of error may affect the intensities of the recorded bands. Among these are the errors associated with the finite slit width, and scanning speeds which are accentuated by the limited amount of energy passing through the small specimen. For a given run under a fixed set of instrument settings, there is the question of error associated with extrusion of the specimen. Also, the material may become increasingly more transparent to infrared radiation at higher or lower pressures depending on the condition of the specimen film. Initially, the powdered specimen is translucent to radiation with considerable scattering. This scattering is usually reduced as the pressure is applied but sometimes increases at higher pressures, presumably because of some sort of phase change. In general, bands tend to broaden at elevated pressures. The finite slit width used would act to increase the apparent intensity of the pressure-broadened bands. Inasmuch as the intensity of most bands decreases at elevated pressures, this source of error produces an effect opposite to that observed. The effect of the smaller amount of scattering at the higher pressure would be to furnish more energy to the detector with the result that the apparent band intensity would appear slightly greater. In actual practice most, but not all, bands show a decrease in intensity at higher pressures and thus the effect of this source of error is opposite to the observed behavior. The effect of extrusion of specimen from between the diamond surfaces would produce a decrease in band intensity. However, a loss in intensity through extrusion should not be reversible for the type of materials studied here. The extrusion is generally observed in all experiments as an irreversible, time-dependent, decrease in intensity of all bands on the initial application of pressure. 61 E. . Lippncott, C. E. Wei, A. van Vand E. N. Bting In practice, quantitative intensity measurements were not considered feasible until the flow of specimen from between the diamond surfaces had ceased as shown by reversibility of the band intensities at different pressures. In addition, quantitative data were usually obtained by starting at the maximum pressure with subsequent measurements at successively lower pressures. Additional loss of material would be minimized by this technique. Since the band intensities usually increased at the lower pressures, it would be necessary to postulate a flow of extruded specimen into the pressure cell. Furthermore, a few spectra have been obtained which show that in a given specimen some bands increase in intensity, some decrease, and others show no change as the pressure increases. Such spectra show unmistakably that there is, at the very least, a marked relative intensity change between such bands. In addition, it is believed that the present data illustrate the behavior of the absolute band intensities as a function of pressure. For obvious reasons no great precision is to be expected in an individual quantitative measurement of absolute intensity but data on the pressure dependence are believed to be qualitatively correct. The apparent band intensities were calculated from plots of log /„// vs. wave-number (or wavelength) followed by a graphical determination of the area. All results are given in ratios of the intensity at a given pressure to that at the lowest pressure recorded. In this manner, many problems associated with the computation of integrated intensities have been minimized or cancelled. Intensity ratios calculated from the integrated expression [3] for band intensities A =[Klcl] Arjln (/„//) gave results in essential agreement with those obtained graphically. In this equation Avi is the band half-width, K is a proportionality constant, A the band area, c the concentration in moles per/liter and I the path length in centimeters. In these experiments the product cl, which is a measure of the quantity of material in the beam, is considered to be constant. Results The following are examples of infrared spectra taken at high pressures for some selected substances. Specimens were obtained from chemically pure analyzed stock supplies. No attempt at further purification was made. The structures of these substances are usually quite complicated and in general it is not possible to interpret the changes which are occurring under pressure in terms of changes in structure. Benzoic acid The infrared spectrum of benzoic acid was studied extensively to calculated pressures of 50,000 atm. The results of one typical run are given in Fig. 2. These spectra were taking consecutively on a sample after it had been established that no further extrusion was taking place. The order of recording the spectra at the different pressures was 46,000, 8000, 48,000, 3000 atm, respectively. The spectra [3] D. A. Ramsay, J. Am. Ghem. Soc. 74, 72 (1952). 62 Studies of infrared absorption spectra of solids at high pressures have been displaced vertically to prevent confusing overlap of the bands. These spectra serve to illustrate two points. First, the relative shifts of most infrared absorption bands with pressure are small. The shifts here are either to higher frequencies, or are zero within experimental error. For example, there is a definite WAVENUMBER, CM"1 6 7 a 9 10 II 12 13 |4 15 WAVELENGTH, MICRONS Fig. 2. Infrared spectrum of benzoic acid at calculated pressures 46,000, 8000, 48,000 and 3000 atm, respectively. shift of a few wavenumbers associated with the 720, 810 and 930 cm-1 bands, respectively. The 720 and 810 cm-1 bands correspond to out-of-plane C—H bending vibration while the 930 cm-1 band is an 0—H out-of-plane bending vibration for the acid dimer structure. Secondly, these spectra illustrate that definite changes of intensity occur on the application of pressure and that these changes are specific with respect to different modes of vibration. In particular, some bands show little, if any, change of intensity on the application of pressure while others show large changes in both maximum absorption and apparent band 63 E. . Lippincott, C. E. Wei, A. van Vt and E. N. Bting area. The hands which show a marked decrease in intensity are 810, 930, 1000, 1030, 1130, 1195, 1305, 1435 and 1600 cm"1, respectively. The band areas for the bands at 810, 930 and 1195 cm-1 have been measured graphically in terms of the ratio of the area at high pressure to that at low pressure. The band areas for these peaks decrease by factors of 0-67, 0-5 and 0-67, respectively. Because of the uncertainty of locating the background, no area measurements were made for the other bands. However, it is clear that the decrease is rather large for the 1000 cm-1 band and for the aromatic frequencies near 1600 cm-1. The least change of intensity seems to occur for the bands at 1455, 1335 and 1070 cm-1, respectively, but quantitative measurements are not possible because of overlapping. The carbonyl band near 1700 cm-1 cannot be measured quantitatively because it occurs on the shoulder of the strong diamond band where there is only a limited amount of energy. The available data indicate that it shifts to lower frequencies. The bands at 690 cm-1 may increase in intensity but again no reliable measurements could be made because of the shortage of energy. It should be noted that the bands associated [4] with the C—0 single bond stretching mode and the coupled O—H bending C—0 stretching mode (1195 and 1305 cm-1, respectively) both show large decreases in intensity. In some spectra taken at pressures greater than 48,000 atm, there was evidence for a phase change as indicated by the appearance of a band near 860 cm-1. However, the change was sluggish and could not always be reproduced. Succinic acid The infrared spectrum of succinic acid has been studied to pressures of 50,000 atm. The results of one series of spectra are given in Fig. 3. These curves are spectra taken consecutively on a sample at pressures of 44,000, 6000 and 44,000 atm, respectively, after it was established that a fixed quantity of sample was remaining between the diamond faces. Most of the absorption bands show slight shifts to higher frequencies on application of pressure. The 930 cm-1 hydrogen-bonded out-of-plane 0—H deformation frequency gives a relatively large shift of nearly 40 cm-1 to higher frequencies. The carbonyl band near 1700 cm-1 appears to undergo a small but definite shift to lower frequencies. The behavior for these two frequencies may be understood in terms of the effect of pressure on the hydrogen bond structure 0 • • • • H—0 / \ R—C C—R \>-H;..;