Matrix-Isolation Infrared Spectroscopy of Organic Phosphates

LISA GEORGE, K. SANKARAN, K. S. VISWANATHAN, and C. K. MATHEWS* Chemical Group, Inclira Gandhi Centre for Atomic Research, Kalpakkam 603 102, India

Matrix-isolation infrared spectra of trimethyl phosphate (TMP), triethyl different conformations.6-1° These can be ideally studied phosphate (TEP), and tri-n-butyl phosphate (TBP), in argon and nitro- with the use of matrix-isolation infrared spectroscopy. gen matrices, are reported for the first time. The peak widths of the Hence we have taken up studies on the matrix-isolation sharpest features in our matrix-isolated spectra are typically 2 cm-', infrared spectroscopy of organic phosphates, to be fol- compared with peak widths of 40 cm ~ seen in liquids for these compounds. Comparison with the vapor-phase spectrum of TMP reported lowed later with studies on the phosphate/diluent inter- earlier indicates that TMP is trapped in two different conformations in action. In this paper, we report, for the first time, the these matrices. Similar spectra were also obtained for TEP. Our matrix- infrared spectra of matrix-isolated trimethyl phosphate, isolated spectra indicate that the intramolecular hydrogen bonding (which triethyl phosphate, and tri-n-butyl phosphate. is believed to be responsible for the lowering of the P=O frequency in the C3~ conformer relative to the Cs conformer in these compounds) is EXPERIMENTAL stronger in TEP than in TMP. In the case of TBP, the peak widths were larger (8-10 cm -z) than those obtained for TMP and TEP. This obser- The matrix-isolation experiments were carried out with vation is probably due to a distribution of conformers that may be trapped the use ofa Leybold AG refrigerator-cooled cryostat, RD in the matrix, as a result of the increased alkyl chain length in TBP. 210, which makes use of a closed-cycle helium compres- Index Headings: Matrix-isolation; Infrared; FT-IR; Organic phos- sor. The minimum temperature attainable with this sys- phates; Trimethyl phosphate; Triethyl phosphate; Tri-n-butyl phos- tem was 12 K at the cryotip, with a thermal stability better phate; Conformations. than 0.2 K. A KBr substrate was mounted on the cryotip, and the sample, together with an inert gas, was deposited on this substrate. The deposition rate was typically 3 to INTRODUCTION 5 mmol h -1, as measured with the use of a mass flow sensor (Brooks 5860). The duration of deposition ranged Organic phosphates, particularly tri-n-butyl phosphate from 30 rain to 2 h. The temperature at the KBr substrate (TBP), find extensive applications as an extractant for was monitored with a carbon resistor. The substrate could actinides in nuclear fuel reprocessing.l They are generally be heated to any desired temperature, with the use of a used together with an organic diluent, such as dodecane, heater mounted on the cryostat. While the substrate was which tailors the physical properties, e.g., viscosity and heating, its temperature was controlled with a Leybold- density, of the organic phase (containing the phosphate Heraeus temperature controller (Variotemp HR 1). and the diluent), to facilitate the solvent extraction pro- The cryostat was housed in a vacuum system, pumped cess. However, the addition of the diluent also alters the with an Edwards Diffstak MK2 series 100/300 diffusion extraction efficiency of the phosphate. It is not clearly pump, having a pumping speed of about 300 L s -t. This understood exactly how the diluent affects the extraction device was backed by a rotary pump (Hind Hivac ED 12), behavior of the phosphate. No firm correlation between with a pumping speed of 200 L min -~. The base pressure any of the physical properties of the diluent (such as dipole in the vacuum system was 1 x 10 -6 mbar, as measured moment or dielectric constant) and the extraction behav- by a Penning Gauge (Hind Hivac). ior of the phosphate/diluent system has been established. 2 Nitrogen and argon (IOLAR Speciality gases, Indian It is believed that the variation in the extraction prop- Oxygen Limited) were used as matrix gases. Though the erties of the phosphate/diluent system could result from levels of impurities were extremely low in these gases the interaction of the phosphate and the diluent. 3-s A (e.g., moisture less than 4 ppm), they were still passed diluent that interacts strongly with the phosphoryl group through a column of molecular sieves (13X). A Leybold- of the phosphate leaves a lower concentration of the free Heraeus precision leak valve regulated the flow of the extractant, resulting in a lower extraction efficiency of the matrix gases during the deposition. A Hastings vacuum phosphate/diluent system. However, a systematic cor- gauge (Model EDNNV-800), incorporated in the gas line, relation between the phosphate/diluent interaction and was used to measure the pressure of the matrix gas. extraction efficiency has not yet been reported. We have Trimethyl phosphate (Merck), triethyl phosphate (To- therefore started a program to study the phosphate/dil- kyo Kasei), and tri-n-butyl phosphate (BDH) were all uent interactions, using matrix-isolation infrared spec- obtained commercially. All the phosphates were purified troscopy. However, before taking up the work on the further by distillation under reduced pressure (0.03 mbar). interactions, we thought it necessary to study the matrix- Each sample was usually distilled at least three times isolation infrared spectroscopy of the pure organic phos- before use, and treated with anhydrous sodium sulphate phates (without the diluent). to remove any trace of moisture. The purified samples Furthermore, organo phosphorus compounds such as were then transferred to a glass sample container in an the phosphates and phosphonates are known to exist in inert atmosphere glove bag, to prevent any moisture up- take by the sample. Before the sample was loaded, the Received 11 May 1993; revision received 9 August 1993. empty sample container was degassed thoroughly in vac- * Author to whom correspondence should be sent. uum (10 6 mbar) for at least 24 h, to remove any moisture

Volume 48, Number 1, 1994 0003-7028/94/4801-000752.00/0 APPLIED SPECTROSCOPY 7 © 1994 Society for Applied Spectroscopy vacuum glass stop-cock, and taken into the glove bag for sample transfer. After the sample was transferred to the sample container, it was connected to the vacuum system through homemade "Veeco-type" joints. The sample was then subjected to a number of freeze-thaw cycles before use. Such laborious techniques for sample handling were adopted because, otherwise, infrared bands due to OH impurities were seen in the spectra. Even after using all fe these procedures, we could only cause a significant re- duction in the intensity of the OH bands, not remove them completely! The required matrix-to-sample ratio (M:S) was obtained by adjusting the temperature of the phosphates (and therefore its vapor pressure), by using slush baths of organic compounds. The temperatures of the slush baths were measured with the use of a platinum resistance thermometer. The matrix gas and the vapors of the organic phosphate streamed out of two separate nozzles (twin jet mode), where they mixed before being deposited C on the KBr substrate. The tips of the nozzles were located at a distance of 28 mm from the KBr substrate. We also W used a single-jet mode in some of our experiments, where CA Z the matrix gas was passed through the sample cell containing the phosphate, before it reached the nozzle. The spectra obtained by using both these deposition modes were basically identical. However, in the case of TBP, it was found that the single-jet mode yielded spectra with a better signal-to-noise ratio. Hence, for TBP alone, a Z B single-jet mode was used, whereas the twin-jet mode was used for deposition of TMP and TEP. Spectra were recorded with a Digilab FTS 15/90 FT- IR instrument, at a resolution of 1 cm-'. Typically, 128 scans were coadded, to obtain good signal-to-noise ratio. After a spectrum was recorded, the matrix was warmed J to 35 K, maintained for 15 min at this temperature, and then cooled back to 12 K. Spectra of the matrix, thus annealed, were then recorded. All matrix-isolation spectra shown in this work were those recorded after the matrix was annealed. All spectra were recorded over the region 4000 to 900 cm '; however, only the region 1330 to 950 cm -~, which encompasses the P=O and P-O-C vibrations and which is relevant for our discussions, has been displayed. Liquid (neat) spectra of the phosphates were recorded by taking a thin film of the sample between ZnSe windows. Vapor-phase spectra were recorded with the use of a commercial gas cell, fitted with KBr windows. The liquid and vapor phase were also recorded with the use of V the Digilab FTS 15/90 spectrometer at resolutions of 1 cm -~, for comparison with matrix-isolated spectra.

1330 950 RESULTS AND DISCUSSION Trimethyl Phosphate (TMP). The infrared matrix-iso- cm-1 lation spectra of TMP, in argon and nitrogen, are shown FiG. 1. Infraredspectra ofTMP: (A) liquid; (B) vapor; (C) in a nitrogen in Fig. 1. In these experiments, the typical matrix-to- matrix (M:S ratio, I000:1); and (D) in an argon matrix (M:S ratio, 1000: sample ratio was 1000:1. Also shown in the figure are the 1). liquid- and vapor-phase spectra, which agree well with those reported in the literature. 1°-'2 In the vapor phase adsorbed to the walls of the glass container. To hasten spectrum, the two bands centered around 1316 and 1291 this process, we intermittently heated the sample con- cm-' have been assigned by Herail to the ~,P=O corre- tainer using a hot air gun. After degassing, the sample sponding to two rotational isomers. Herail assigned the container was stoppered with the use ofa greaseless high- 1316-cm-l band (peak a) to the conformer with the C3

8 Volume 48, Number 1, 1994 symmetry and the one at 1291 cm-~ (peak b) to the con- °" former with C3v symmetry. (The structures of the two conformers are shown in Fig. 2.) In our matrix-isolation experiments, using a nitrogen matrix, we observed two C P 0 C peaks corresponding to the P=O absorption: one at 1287 cm -~ (peak c) and the other near 1305 cm -~ (peaks d), which appears as a doublet. Since the vapor-phase com- position of the conformers is expected to be retained in the matrix, the two peaks seen in our matrix-isolation spectrum can be assigned to the two conformers of TMP, seen in the vapor phase. The peak at 1287 cm -t can be assigned to the C3v conformer, and the doublet near 1305 1 H cm-~ to the C3 conformer. As can be seen, the C3v con- FIG. 2. The C3,. and the C3 conformers of TMP. former has a lower P=O frequency than the C3 conformer, both in the gas phase and in the matrix. Intramolecular hydrogen bonding between the alkyl hydrogens and the possible to rationalize the lower P=O frequency in the phosphoryl oxygen probably leads to a lowering of the case of TEP, compared with TMP. In the case of TMP, P=O frequency in the case of the C3~ conformer. 13-~5 a five-membered ring may be formed, which results in In the argon matrix, the peak near 1285 cm -t appears the intramolecular hydrogen bonding, 2° as shown in Fig. as a multiplet (peaks e), and the one near 1307 cm -~ as 4. However, in the case ofTEP, a six-membered ring may a doublet (peaks JO. This structure of the band was seen be formed, z° This formation brings the hydrogens closer in all our spectra recorded at various matrix-to-sample to the phosphoryl oxygen in TEP than in TMP. The hy- ratios, with the ratios varying from 1000:1 to 30,000:1. drogen bond would, therefore, be expected to be stronger This consideration rules out the possibility that these in TEP than in TMP, and hence results in a lower P=O multiplets could be due to aggregation. On annealing the frequency in TEP. The 1271-cm -~ peak can therefore be matrix, the relative intensity of the peaks in the multiplet assigned to the P=O frequency for a conformer of TEP structure changed, but the structure did not significantly similar to conformer I of TMP (with a C3v symmetry). In simplify. We believe that this observation could be due the vapor, the corresponding peak appears at 1279 cm- to multiplet site splitting in the argon matrix, where the (peak d in Fig. 3). TMP could be trapped in different stable sites. In contrast, As with the 1305-cm -t doublet in TMP, we also ob- the spectra in the nitrogen matrix are much simpler, with served a similar peak at 1305 cm -~ for TEP (peak b in the 1287-cm -~ peak appearing as a single sharp peak. As Fig. 3). This peak actually appears as a doublet before reported in several earlier studies, argon is known to dis- annealing; however, on annealing it collapses to a singlet. play multiplet site splitting, whereas nitrogen generally In analogy with TMP, this peak can be assigned to a yields spectra free from such splittings) 6-~9 The peak near conformer of TEP, similar to conformer II of TMP (with 1305 cm -~, however, appears as a doublet in both ma- the C3 symmetry). Since this conformer does not involve trices, probably due, again, to multiple site effects. It should any intramolecular interactions, the P=O frequency can be noted that the 1305-cm -~ doublet in the nitrogen ma- be expected to be left almost unaltered in both molecules, trix is shifted by about 4 cm-~ to the red, relative to that as observed. in the argon matrix. A third peak, a doublet near 1290 cm -~ (peak c in Fig. The (P)-O-C stretch of TMP appears near 1046 cm -~. 3), has already been assigned to the CH2 twisting mode.t~ From Fig. 1, it can be seen that this band is again simpler The (P)-O-C stretch for TEP occurs near 1042 cm -~. in the nitrogen matrix than in the argon matrix. This In the case of TEP, both nitrogen and argon matrices result may also be due to the site splitting effects in the yield similar spectra, except for two differences. The peak argon matrix. widths, in the nitrogen matrix, are smaller than those Triethyl Phosphate (TEP). Figure 3 shows the infrared obtained in the argon matrix. Second, the P=O frequency spectra of TEP in nitrogen and argon matrices. The liquid- in the nitrogen matrix is red shifted by about 2 cm -~ and vapor-phase spectra are also shown for comparison. relative to that in the argon matrix (1273 cm -~, peak e Unlike in the case of TMP, it can be seen that the matrix- in Fig. 3). However, unlike the case of TMP, where the isolated spectra of TEP in argon do not display multiplet P=O absorption appeared as a multiplet in the argon site splitting effects. matrix, in the case of TEP the P=O absorption (for the The P=O stretch for TEP occurs at 1271 cm-~ (peak C3v conformer) is seen to be a singlet. a) in the nitrogen matrix. This is about 16 cm -~ lower Tri-n-Butyl Phosphate (TBP). Figure 5 shows the in- than what is seen in the case of TMP (1287 cm-~). The frared spectra of TBP in nitrogen and argon matrices, P=O frequency is consistently lower in the case of TEP, together with that in the liquid phase. As already men- when compared with TMP, in every single case--matrix- tioned, the matrix-isolation spectra of TBP were recorded isolated, liquid, and vapor. In the case of TMP (in a with the use of the single-jet mode, where the matrix gas nitrogen matrix) the 1287-cm -~ peak was assigned to the was passed through the sample. In the case of TBP, it C3v conformer. As already mentioned, this conformer gives was found that, in spite of the use of high dilutions (M: rise to the possibility of intramolecular hydrogen bonds S, 6500:1), the peak widths were never nearly as sharp between the alkyl hydrogens and the phosphoryl oxygen. as those obtained with TMP and TEP. The P=O peak, If a similar conformer in TEP is responsible for the peak which appears near 1279 cm -~ (peak a), was about 8-10 at 1271 cm -~ (in the nitrogen matrix), then it becomes cm-1 broad, in both nitrogen and argon matrices. Similar

APPLIED SPECTROSCOPY 9 o ...... H 0 oO "" H

P c

e TMP TEP D FIG. 4. Illustration of the five- and the six-membered ring formation in TMP and TEP, in the type I conformer (C3,.).

since we did not see any improvement in the peak widths, even when the matrix-to-sample ratio was increased to 6500:1. It is possible that, with an increase in the alkyl chain length, the number of conformers trapped in the matrix also increases, in which the alkyl groups are ori- W ented only slightly differently. This pattern is particularly accurate for long alkyl chains, where the rotation about Z the C-C bond can give rise to a distribution of conformers a between the two extreme conformations--syn and anti. z4 I-- C I-- These different conformations would be expected to alter the P=O frequencies only marginally. This factor would, therefore, give rise to a broad peak, in which the different u3 P=O frequencies have not been resolved. It should be Z noted that the P=O frequency in the nitrogen matrix (1277 cm-') again occurs about 2 cm -' to the red of that I-- d seen in the argon matrix (1279 cm-'). The (P)-O-C stretch in TBP occurs near 1030 cm -~, and the peak width is again broader than what we had observed in the cases of B TMP and TEP. In addition to these two peaks, we also observed a peak at 1239 cm -~ (peak b) in the matrix-isolated spectra of TBP. In the liquid spectra, a shoulder appears in this region (c). Earlier workers have observed this shoulder, 3,15 but no assignment has been made. This shoulder in the liquid spectrum is clearly resolved as a peak in the matrix- isolated spectra. It is more prominent in the argon matrix than in the nitrogen matrix. Although we do not have a definitive assignment for this peak, it is our conjecture that this 1239-cm-' peak could be the P=O stretch of a conformer of TBP (Fig. 6) in which the hydrogens on the carbon atoms, C2 and C4 of the butyl group, are both involved in a hydrogen-bonding with the phosphoryl oxygen. In fact, in this process, a six-membered ring is formed, which, being stable, probably stabilizes this con- 1330 950 former. Since there are three butyl groups in TBP, this cn,, 1 process leads to a total of six hydrogens participating in Fro. 3. Infrared spectra ofTEP: (A) liquid; (B) vapor; (C) in a nitrogen the hydrogen-bonding, resulting in a significant red-shift matrix (M:S ratio, 1000:1); and (D) in an argon matrix (M:S ratio, 1000: of the P=O frequency. Such a conformer is possible only 1). if the alkyl chain is at least four carbon atoms long, as in the case of the butyl phosphate. Interestingly, such a peak was not observed in the case of TMP and TEP, where increases in the peak widths of the carbonyl bands of the alkyl chains are less than four carbon atoms long. In ketones and esters, with an increase in the alkyl chain these lower phosphates, only a maximum of three hy- length, have been reported by Coleman and Gordon in their matrix-isolation experiments. 2'-23 They have attrib- drogens can be involved in the hydrogen bonding (Fig. 4). uted this peak broadening to (1) nearest-neighbor interactions and/or (2) trapping of a distribution of conformers CONCLUSION in the matrix, as a result of an increase in the alkyl chain length. We do not believe that nearest-neighbor inter- Matrix-isolated infrared spectra have been recorded for actions are the cause of broadening in our TBP spectra, TMP, TEP, and TBP in nitrogen and argon matrices. The

10 Volume 48, Number 1, 1994 c, a

1 FIG. 6. A conformation of TBP, where the orientation of one of the butyl groups around the phosphoryl oxygen is shown. For clarity, not all the atoms have been depicted.

w V ~ a two compounds remains almost unaltered. This obser- Z vation has been explained on the basis of intramolecular hydrogen bonding, which is stronger in TEP than in TMP. The matrix-isolated spectra of the TMP and TEP are H / clearly sharper and simpler in the nitrogen matrix than in the argon matrix. The peak width of the sharpest fea- Z tures is typically 2 cm -1. However, in the case of TBP <[ c the peaks are not as sharp. This result, we believe, is due to conformational effects. The 1239-cm -1 peak in TBP has been tentatively assigned to the P=O stretch of a conformer of TBP where a total of six hydrogens partic- ipate in hydrogen bonding with the phosphoryl oxygen. Table I lists the frequencies of the P=O and P-O-C vibrations of the three organic phosphates in liquid, vapor, and matrix-isolated conditions. These studies also clearly indicate that, for our future studies on phosphate/diluent interactions, TEP is a good choice among the phosphates, and specifically in a nitrogen matrix, since it yields sharper features under these experimental conditions.

TABLE I. Frequencies (em ') and vibrational assignments for the organic phosphates in liquid, vapor, and matrix-isolated conditions. Matrix-isolated Mode Liquid Vapor Nitrogen Argon Trimethylphosphate P=O 1276 1316 1307,1302 1311, 1307 1291 1287 1292, 1285 1320 950 1282 P-O--C 1038 1060 1061,1048 1068, 1054 1046 1047 cm-1 Triethylphosphate Fi~. 5. Infrared spectra of TBP: (A) liquid; (B) in a nitrogen matrix P=O 1265 1307 1305 1308, 1304 (M:S ratio, 1000:1); and (C) in an argon matrix (M:S ratio, 1000:1). 1271 1273 P-O-C 1034 1049 1050,1047 1050, 1046 1045,1042 1042 two conformers of TMP seen by Herail in the vapor phase 12 Tri-n-butyl phosphate have also been observed in our matrix-isolated spectra. P=O 1281 -- 1277 1279 TEP also displays similar spectra. However, it was ob- 1236(sh)a 1240 1239 served that the P=O stretch of TEP occurs about 16 cm-' P-O-C 1060 -- 1062 1063 to the red of that observed for TMP, for the C3v conformer. 1029 1030 1031 However, the P=O stretch of the C3 conformer for the " Note: sh = shoulder.

APPLIED SPECTROSCOPY 11 14. L. S. Mayants, E. M. Popov, and M. I. Kabachnik, Opt. Spectrosc. 1. H.A.C. McKay, in Science and Technology of Tributyl Phosphate, 7, 108 (1959). Vol. I, W. W. Schulz and J. D. Navratil, Eds. (CRC Press, Boca 15. D. Dyrssen and Dj. Petkovic, J. Inorg. Nucl. Chem. 27, 1381 (1965). Raton, Florida, 1984), Chap. 1. 16. E. L. Wehry and G. Mamantov, Prog. Anal. Spectrosc. 10, 507 2. D. A. Orth, R. M. Wallace, and D. G. Karraker, in Science and (1987). Technology of Tributyl Phosphate, Vol. 1, W. W. Schulz and 17. A. Givan, A. Lowenschuss, K. D. Bier, and H. J. Jodl, Chem. Phys. J. D. Navratil, Eds. (CRC Press, Boca Raton, Florida, 1984), Chap. 6. 106, 151 (1986). 3. L. L. Burger, in Science and Technology of Tributyl Phosphate, Vol. 18. G. Mamantov, A. A. Garrison, and E. L. Wehry, Appl. Spectrosc. I, W. W. Schulz and J. D. Navratil, Eds. (CRC Press, Boca Raton, 36, 339 (1982). Florida, 1984), Chap. 3. 19. S. Li and Y.-S. Li, Spectrochim. Acta 47A, 201 (1991). 4. P. M. Petkovich and B. A. Kezele, Proc. Int. Solvent. Extraction 20. K. Ohwada, Appl. Spectrosc. 21, 332 (1967). Conf., 1971, Vol. 2, (Soc. Chem. Ind., London), 1137. 21. W. M. Coleman III and B. M. Gordon, Appl. Spectrosc. 41, 1163 5. J. R. Ferraro, Appl. Spectrosc. 17, 12 (1963). (1987). 6. O. A. Raevsky, J. Mol. Struct. 19, 275 (1973). 22. W. M. Coleman III and B. M. Gordon, Appl. Spectrosc. 42, 101 7. F. Herail, C. R. Acad. 261, 3375 (1965). (1988). 8. J. R. Durig, J. Mol. Struct. 113, 127 (1984). 23. W. M. Coleman III and B. M. Gordon, Appl. Spectrosc. 42, 666 9. B. J. Van der Veken and M. A. Herman, J. Mol. Struct. 42, 161 (1988). (1977). 24. J. March, Advanced Organic Chemistry." Reactions, Mechanisms, 10. R. A. Nyquist and W. J. Potts, in Analytical Chemistry of Phos- and Structure (Wiley Eastern Limited, New Delhi, 1987), 3rd ed. phorus Compounds, M. Halmann, Ed. (Wiley Interscience, New 25. K. Nukada, K. Naito, and U. Maeda, Bull. Chem. Soc. Japan 33, York, 1972), Chap. 5. 894 (1960). 11. F. S. Mortimer, Spectrochim. Acta 9, 270 (1957). 12. F. Herail, J. Chim. Phys. 68, 274 0971). 13. E. M. Popov, M. I. Kabachnik, and L. S. Mayants, Russian Chem. Revs. 30, 362 (1961).

12 Volume 48, Number 1, 1994