ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Quarterly Report No. 1 THE CHEMISTRY OF BORON HYDRIDES AND RELATED HYDRIDES January-March, 1956 R. W. Parry Goai Kodanma S. G. Shore R. C. Taylor Earl Alton Project 2469 WRIGHT AIR DEVELOPMENT CENTER WRIGHT-PATTERSON AIR FORCE BASE, OHIO CONTRACT NO. AF 33(616)-3343 May 1956

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN TABLE OF CONTENTS Page OBJECTIVE iii ABSTRACT iii I. THE REACTIONS AND STRUCTURE OF THE DIAMMONIATE OF DIBORANE 1 A. Review of Previous Work 1 B. The Preparation of Salts of H2B(NH) )2 1 II. RAMAN SPECTRAL STUDIES ON H3BCO 1 III. THE REACTIONS AND STRUCTURE OF THE AMMONIA ADDITION COMPOUNDS OF B4H1o 2 A. Background 2 B. The Reaction Between B4H1o and NH3 4 1. Reaction in the Absence of Solvent 4 2. Reaction of the NH3-B4Hlo Addition Product with Sodium in Liquid Ammonia 4 3. Precipitation Reactions in Liquid Ammonia for the NH3B4H1o Addition Product 8 4. Raman Studies on B4H1o Ammoniates in Liquid Ammonia 8 5. The Reaction Between Ammonia and B4Hlo in Diethyl Ether 8 6. The Preparation of NaB3H8 in Diethyl Ether from NaH and B4Hlo 9 IV. THE REACTION BETWEEN [A12C16] and PF3 11 APPENDIX. THE RAMAN SPECTRUM, VIBRATIONAL ASSIGNMENTS, AND FORCE CONSTANTS FOR BH3CO AND BD3CO 12 Introduction 12 Experimental 13 Experimental Results 13 Assignments 13 Normal-Coordinate Treatment 17 Discussion 19 References 22 ii

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN OBJECTIVE The research reported here had as its objective the fundamental study of the chemistry of the hydrides of boron. ABSTRACT Evidence is presented to show that B4H1o reacts with NH3 to form B4H10o7 NH3, B4Hlo.3 NH3, and B4Hlo02 NH3. The product formed is dependent upon experimental procedures. No evidence for B4H10o4 NH3 has been founid. The addition product formed by B4H1o in liquid ammonia at -77~C reacts with excess sodium to give 1/2 mole H2 per B4H20 and the solid NaB3Hs. Results are interpreted in terms of bridge cleavage in B4H10. A Raman Study of H3BCO is described in detail. iii

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN I. THE REACTIONS AND STRUCTURE OF THE DIAMMONIATE OF DIBORANE A. REVIEW OF PREVIOUS WORK All preceding work on the ammonia addition compounds of diborane has been reviewed. The vacuum line for this system has been reconditioned. H3BNH3 has been prepared for further Raman study. Work now proceeding on the ammonia-diborane system is directed toward the larger-scale production of salts of the [H2B(NH3)2]+ cation. B. THE PREPARATION OF SALTS OF H2B(NH3)2 The compound [NIH4][AsF6] was prepared from laboratory stocks of KAsF6* by the following reactions. KAsF6 + CdCl2 + 6NH40H -— >,ICd(NIH3)6[AsF 62 + 2KC1 [Cd(NH3)6[ASF6]2 t C dS + 2NH4AsF6 + 4]E3 H2S The NH4AsF6 was allowed to react with [H2B(NH3)2j7FBH4 according to method l developed by Shultz and described in earlier reports. A solid which gave a powder pattern, definitely not NH4AsF6, was obtained. Further characterization is still incomplete. The same process with NH4PF6 produces a solid which still contains some NH4PF6 lines. Such solids are under more thorough study. II. RAMAN SPECTRAL STUDIES ON H3BCO A normal coordinate treatment of the H3BCO molecule which was carried over from the earlier contract has been completed and a formal report on the Raman spectra of H3BCO and D3BCO has been written. A copy is submitted with H. M. Dess, Doctoral Dissertation, The University of Michigan, Ann Arbor, 1955

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN this report. Permission to submit this work to a technical journal for publication is requested. III. THE REACTIONS AND STRUCTURES OF THE AMMONIA ADDITION COMPOUNDS OF B4H1o A. BACKGROUND Although the structure of the diammoniate of diborane has long been a question for debate, the stoichiometry of its formation has been well established and easily reproduced in any laboratory, if proper precautions were taken. In contrast, both the stoichiometry for the formation, and the structures of the ammoniates of B4H1o have been uncertain. Stock, Wiberg, and Martini* reported that four ammonia molecules add to one molecule of B4H1o to give B4Ho104NH3. Since their observations on stoichiometry have not been tested in other laboratories, the original work must be considered uncertain until it is confirmed. In previous studies of the diborane ammoniates, compounds were identified which suggested that reactions of the diborane molecule involve cleavage of the hydrogen bridges. Such cleavage can be symmetrical or nonsymnetrical, depending upon the conditions of the experiment. If symmetrical cleavage occurs, H3B groups are liberated and their primary reaction is that of a Lewis acid. / Symmetrical Cleavage H H /H B_,,/B + 2NR3 -— > 2H3BNR3 H JH 1H If nonsymmetrical cleavage occurs, the reactions are formally analogous to those observed with some bridge-type coordination compounds and typical Werner coordination compounds of boron result. Nonsymmetrical Cleavage H H H H\ /NH H H B +2NH F B/N B B/ [H/ 2NH / 3 H1HI H H NH31 H Hj Certain reagents such as trimethylamine promote symmetrical cleavage. Molecular-weight studies on the addition compounds of H3B with mono- and dimethylamines indicate that these reagents also promote symmetrical cleavage, Ber., 63, 2927 (1930). _~~~~~~~~~~~

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN - while ammonia promotes nonsymmetrical cleavage under the usual conditions of the interaction. Not only the reagent but the conditions of the experiment seem to be important in determining the type of interaction to be expected. Gas-phase reactions of B2H6 seem to favor symmetrical cleavage as the initial process. Reactions in a solvent such as ether, likewise seem to favor symmetrical cleavage, although with reagents such as ammonia, which strongly favor nonsymmetrical cleavage, reactions in ether may also give the nonsymmetrical cleavage product. Reactions involving solid phases usually give nonsymmetrical cleavage for those reagents such as ammonia which are capable of producing nonsymmetrical bond rupture. The reactions of NH3 and B2H6 produce a variety of products, the formation of which has been associated with localized heating during the reaction process. Formulas which are supported by significant evidence are: BHv XB/ iLBH4I - 7 7 17', HH \ I'H HE HI\ N H -B-"_, 223NL2 and f\' HEH H H H Obviously, the reaction of ammonia with diborane is a complex process which is strongly dependent upon the details of the experimental procedure. An extrapolation of the observations on the ammonia-diborane reaction suggests that the ammonia-B4H1o reaction should be at least as complex as that involving the simpler diborane molecule. From the known structure for B4Hjo one can make a number of predictions as to the expected products of reaction with amines and ammonia. For example, since N(Me)3 is known to favor symmetrical cleavage of double bridges, one might expect interaction as follows: I I H \H Three molecules of H3BNR3 per molecule of B4H1o have been recovered from the foregoing process. Agreement between theory and observation is excellent. *See Final Report from this laboratory for Project 1966, 1956. \H/~~~~~~

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN The more complex possibilities with ammonia are considered in connection with the discussion of experimental results. B. TEE REACTION BETWEEN B4Hlo AND NH3 The B4HLo-ammonia reaction, like the armonia-B2H6 process, is strongly dependent upon experimental conditions. Reactions in ether and in the absence of solvent were studied. 1. Reaction in the Absence of Solvent. —Stock, Wiberg, and Martini* reported that B4H10 and excess NH3, frozen together with liquid air will form B4Ho104NH3 when warmed up to -78~C for one hour. All attempts to reproduce their work have been unsuccessful. There appears to be a reasonable doubt as to the validity and generality of their observations. Data summarized in Table A and Fig. 1 suggest that equilibrium in the reaction is not reached in one hour at -78~C. If ample time is allowed at -780C, the compound B4H1-7NH3 appears to form. This loses four molecules of ammonia when the temperature is raised to -63.5~C. The resulting compound B4H1o-3NH3 appears to be stable up as high as -23~C. Above this temperature hydrogen evolution becomes vigorous. The above conclusions are being tested further by construction of a phase diagram for the system B4HLo-NH3. Symmetrical cleavage of the double-bridge bond in B4H10 would yield BH3 and B3H7 as fragments. These might add a base such as NH3 to give H3BNH3 and other products (e.g., NH3B3H7). Although the conditions used in this NIH3B4H1o reaction were not those which should favor symmetrical cleavage, the residue was leached with diethyl ether in the vacuum system to extract any H3BNH3 from the solid reaction products. No BH3NH3 was detected; it was concluded that symmetrical cleavage did not occur under the conditions of this experimernt 2. Reaction of the N? -B4NHo Addition Product with Sodium in Liquid Ammonia. —To the product from Run No. 1, Table A, 3 ml of liquid ammonia was added; then a bulb containing sodium metal was crushed and added to the frozen system. The temperature was permitted to rise slowly to -780C and then held at this temperature. Hydrogen evolution as a function of time is shown in Fig. 2. Similar data for Run No. 2, Table A, are also shown. It is significant that one-half mole of hydrogen per B4H1o is liberated rapidly, then another one-half mole is liberated more slowly. The product formed in the reaction of sodium with the diammoniate of diborane is sodium borohydride. If one assumes similar unsymmetrical cleavage of the double-bridge bond in the B4H1 o-ammonia system, the reaction should be: Loc. cit. 14.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

TABLE A. SUMMARY OF ALL RUNS ON B4Hlo AND AMMONIA m Temp. Time ibr - / B4HO. Millimoles Reaction Reaction R N.MEillimoles N.atTpTieNH3 NH3 Ratio Millimoles H2 RunNo. B4H0 NHaTRemoval Removal Approached Millimoles B4H m ~ "Start (0c) ( C) (hr) by System 1 1.04 12.58 -196 to 5 -78 8-1/2 7.05 nil -78 -63.5 6 3 to 3.2 nil m -65 3 0.325 4.66 -78 1 -78 24 3.2 0.15 -63 6 2.91 0.15 -45 3 2.83 0.15 -23 8 2.34 0.17 0 25 2.16 0.70 r 25 6 2.16 0.76 4 0.337 5.17 -78 24 -23 3.3 3.03 0.016 25 18 2.14 0.72 70-80 5 2.14 0.81 z 25 24 2.14 0.82 Comparable Data of Stock et al. iS 0.445 4.7 -165 to 1 -75 About 3 4.0 Not Reported -< -78 2S 0.455 4.8 -165 to 1 -75 About 3 4.0 Not Reported -752 3S 0.442 4.6 -165 to I -75 About 3 4.0 Not Reported -78. __~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~7

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN 1.04 Millimoles B4 Ho1 0 12.58Millimoles NH3 Originally I 0O 55-Hr Reaction Time m 8 — Temperature Raised From -78~ To -63.50C -J -780C w z ~ 6t cr. Data Reported By o Stock, Wiberg, and Martini J +: [ar.,3, 2927 (1930)] I 63.50C - 4 1.12 Millimoles B4 Hi1 Second Run: NH3 OW 8.34Millimoles NH3 Distilled Out-63.50C I Ratio = 3.25/1: 0 2- Reaction Time=12Hr+ H2 Evolved Ratio H2 /B4 Hio =.053/1 I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 TIME (HOURS) Fig. 1. Initial runs on stoichiometry of reaction between B4H0 and NH'3 at -78.5~C and below and -63.5~C and below. I.I.9 Run # 2 Run# I Table I.7 0.5:-4 I.0 - I 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 TIME (HOURS) Fig. 2. Evolution of H2 from reaction of Na with liquid-amonia solution B4Hlo (See Table A for preparative data on B4Hlo amnmoniate. )

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN H H [H IIH3 ] |,:B-H | Na > H H+ {ff +2H2 +Na B3H The compound NaB3H8 has been isolated and identified through an independent set of reactions by Callery Chemical Co,* and is a. known species. It is ether soluble. Accordingly, 1.859 mm of B4H1o was condensed in a reaction tube and about 3 ml of ammonia was condensed above it. The system was allowed to warm slowly to dry-ice temperature, then held at -78~C overnight. The system was frozen with liquid nitrogen, sodium was added, then the temperature was raised rapidly to -78~Co. After one-half hour ammonia and evolved hydrogen were removed; 1.17 equivalent of H2 per mole of B4HLo was found. Traces of ammonia were pumped off at room temperature; then the solid product was leached with dry ether in the vacuum line extraction system. Crystals were isolated by evaporation of the ether from the filtrate. If great care was taken to avoid exposure of these crystals to water, their x-ray powder pattern was identical to the pattern for NaB3H8. The pattern for NaB3H8 was generously supplied by Callery Chemical Co. and was checked by an independent preparation of NaB3H8 in this laboratory. The foregoing experiment was confirmed in a second run using an almost identical procedure. Only one significant change in method should be recorded. In the second run ammonia was removed from the solid and B4H0O ammoniate was isolated; then ammonia was added again to the system, followed by sodium addition. Again NaB3H8 was the product when hydrogen evolution was stopped after loss of one-half H2 per B4H10. Yields of NaB3H8 were estimated at 60 to 80G. This was not a trace quantity of material. The foregoing observations support a symmetrical. cleavage of the double-bridge bond and are consistent with earlier observations made on the system sodium-diammoniate of diborane. The source of the second equivalent of hydrogen which results from the sodium reaction is less certain. It was found that the solid product remaining after loss of one mole of H2 per B4H1O did not contain NaB3H8. ProHough, Edwards, and McElroy, J. Am. Chem, Soc., 78, 689 (1956).

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN longed ether extraction provided just a trace of material which gave the x-ray pattern for NaBH4. Then 1.04 millimoles of B4H1o was condensed with a large excess of ammonia. The tube was held at -78oC overnight, then an excess of sodium was added and H2 was evolved. After 48 hours, 2.45 equivalents of H2 per mole of B4H1o had been produced. The ammonia was removed from the system and the solid residue was leached with diethyl ether in the vacuum system. Only a trace of solid was extractable with ether. It should be noted that more NaBH4 could have been present in the solid residue but was not removed by ether extraction. The point was not checked because excess sodium in the solid residue made residue handling difficult. A reaction between Na and NaB3H8 in NH3 to give HE and some NaBH4 was considered as a possible explanation for the foregoing observations. A direct experimental test of the hypothesis was tried. NaB3H8 was prepared from B4Hlo and NaH in ether. Sodium and 3 cc of liquid ammonia were added. No H2 was evolved at -780C over a period of 15 hours. The system was raised to -45~C and held for 44 hours. Still no hydrogen was evolved. 3. Precipitation Reactions in Liquid Ammonia for the NH3-B4H1o Addition Product. —Addition of magnesium salts [Mg(AsF6)2 and Mg(SCN)2] to a liquid-ammonia solution of B4H10 ammoniates did not produce a precipitate of [Mg(NH3)6][BH4]2. Apparently an unsymmetrical cleavage which produces borohydride IH I i H H4BH4 H \B/ \B/ H/\ I / \ H HHTB-H H is not favored as a primary process in liquid ammonia. A trace of still unidentified precipitate was found when Mg(AsF6)2 and KAsF6 were added to liquid-ammonia solutions of B4Ho1 at -780C. It was thought that this might be [IIH2B(NH3)2][AsF6I; however, its x-ray pattern was not the same as that of the product, presumably [H2B(NH3)21[AsF6], resulting from the reaction between [H2B(NH3)21[BH4] and NH4AsF6. The solid is still being studied. 4. Raman Studies on B4H1o Ammoniates in Liquid Ammonia.-These studies are only in a preliminary stage and conclusions are not warranted. 5. The Reaction Between Ammonia and B4H10 in Diethyl Ether. —It is known that ether favors symmetrical cleavage of a double-bridge bond. Accordingly, one might expect the following reaction between ammonia and B4HLO in ether:

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN, H B/' / B + 2NH3 — A H3BNff3 + H3NBs3H7 H, H-B-H H H A solution of B4H10 in ether absorbs NH3 readily. The equilibrium pressure (ammonia + ether) above the solution was measured as the ammonia content of the system increased. Results are summarized in Fig. 3. Although a choice between a compound B4H102NH13 and B4Hlo.2-1/2 NH3 is somewhat arbitrary, the former formula was accepted. It is significant that the ammonia addition product prepared in this manner is completely ether soluble, a fact consistent with the existence of the compounds H3N7BH3 and H3NB3H8; the product prepared by the reaction of NH3 and B4H10 in the absence of ether is not ether soluble, a fact consistent with nonsymmetritcal cleavage of the double-bridge bond. 24 22 - 20 E 18 / l D 106 - / 8 14w 0 < 12() w n 86 1 2 3 4 5 6 7 8 NH3 MOLECULES PER B4 HO10 Fig. 3. The system NH3-B4Hio in diethyl ether solution. 6. The Preparation of NaB3H8 in Diethyl Ether from NaH and B4Hlo. [Method first used by Callery Chemical Co.] —Symmetrical cleavage of the double bridge bond of B4HLo would give BR3 and B3H7. It is known that a sodium hydride slurry in -diethyl ether will not react with B2H6 (BH3 groups) to give ether insoluble NaBH4. Accordingly, such cleavage should give rise to B2H6 from the BR3 groups.

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN On the other hand, B3H7 should react with NaH to give ether soluble NaB3H8. The reaction goes as expected: B4Hzo + NaH diethyl ether NaB3H8 + 1 B2H (in ether) 2 NaB3H8 was isolated and identified by the powder pattern reported below. The corresponding pattern of Callery Chemical Co. is given for comparison. Interplanar Spacings in Angstroms Univ. of Mich. Callery Univ. of Mich. Callery. Product NaB3H8 Product NaB3H8 5.-75 w 5.67 m 2.098 wm -- 5.09 m 2.025 wm 2.02 w 4.66 w 4.66 m 1.871 w 1.88 w 3.95 s 5.93 s 1.783 w 1.79 vw 3.66 s 35.67 s 1.373 vw -- 3.31 vs 3.32 vs 1.675 vw 1.68 vw? vw 3.09 vw 1.639 w 1.64 w 2.69 wm 2.71 w 1.55 vw 1.56 vw 2.47 w 2.49 vw 1.497 w -- 2.37 m 2.35 s 1.467 w 1.46 vw 2.31 w? 1.368 vw 1.379 vw 2.18 vw? 1.327 vw 1.318 vw 2.15 m 2.15 m Camera circumference = 180 mm; CaKa radiation. Note: w = weak; m = medium; s = strong; v = very. An observation of great significance was made in studying this reaction. NaH of good quality, prepared by E. I. DuPont, reacted with B4Hlo in ether to give one H2 per B4H10. No NaB3H8 was ever isolated from the system. On the other hand, Nai obtained from the Callery Chemical Co. reacted with B4H1o to give some excess H2 and high yields of NaB3H8. The differences in the solid Nai are Still undefined, but the contrast in the reaction is striking when different samples of NaH are used. 10

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN IV. THE REACTION BETWEEN [A12C016] and PF3 Earlier work in this laboratory suggested that PF3 might react with dimeric Lewis acids such as B2H6 and Az12CL6 to give a coordination compound. H3BPF3 has been prepared and described elsewhere.* Under similar conditions [AiC13]2 reacts with PF3 to give complete halogen interchange. A1C3l + PF3 —. >PC13 + A1F. Although there was no direct evidence for the compound Cl3AlPF3, it is probable that it existed as a reaction intermediate in the exchange process. Work is underway to isolate the complex C13A1PF3 if it forms. R. W. Parry and T C. Bissot, J. A. Chem. Soc., 78, 1524. (1956).

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN - APPENDIX THE RAMAN SPECTRUM, VIBRATIONAL ASSIGNMENTS, AND FORCE CONSTANTS FOR BH3CO AND BD3CO Robert C. Taylor INTRODUCTION Despite the rather active chemical interest in boron hydrides and their derivatives, the amount of spectroscopic work which has been carried out on these compounds has been disproportionately small. This disparity is particularly marked if one compares the data available on vibrational frequencies, assignments, force constants, molecular parameters, and so on for the derivatives of simple hydrocarbons with the corresponding data for the various boron hydride derivatives. Experimental difficulties caused by the high reactivities of the boron compounds are responsible to a large degree for this situation; however, if one is willing to work in condensed phases and at low temperatures, a great deal of spectroscopic information can be obtained which can yield significant values for molecular constants and also provide a basis for comparisons of chemical properties. The class of compounds containing the BH3 group is of particular interest here, from the spectroscopic point of view, as yielding information about the hypothetical simple boron hydride BH3, and from the chemical point of view as an example of complex formation through a Lewis acid-base interaction. In the present work, the Raman spectrum of a simple member of this class, BH3CO, has been obtained, a complete assignment of fundamental frequencies has been made, and a set of valence force constants determined which agrees with the experimental data for four isotopic combinations. It is hoped these data will serve as a basis for comparison with other molecules containing the borane group. Only one previous spectroscopic paper on BH3CO has appeared, a paper by Cowanl reporting the infrared spectrum of the vapor. Five fundamentals reported by him agree with the values found in the present work, two he did not observe, and his assignment of the last appears incorrect. No data for the BD3CO molecule have been found. 12

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN EXPERIMENTAL Both BH3CO and BBD3CO were prepared by reaction of B2aH6 or B2D6, respectively, with CO in a sealed tube at several atmospheres pressures After several days, the tubes were opened and the contents carefully fractionated at low temperatures on the vacuum line. After fractionation, the sample was distilled into the Raman cell which was then sealed off. To reduce thermal decomposition, the vapors were never allowed to come in contact with surfaces warmer than about -50~C during all transfer operations. The spectra obtained showed no bands attributable to diborane or CO which would be produced as decomposition products. The sample of BH3CO examined was about i ml in volume while the BD3CO was about 0.2 ml. During the exposures, the samples were maintained at approximately ~800C, at which temperature the decomposition occurring in the liquid is negligible. A general description of the apparatus and spectrograph has been given previously.2 Exposure times varied from ten minutes to three hours, using Eastman 103a-J plates. Measurements were made with a comparator directly on the plates and also on enlarged. tracings made with a Leeds and Northrup microphotometer. The estimated probable error for most lines reported is approximately 1 cm'lo EXPERIMENTAL RESULTS The experimentally observed frequencies for BH3C0 are listed in Table I, while those for BD3CO are listed in Table II. Tracings of spectra of the two substances selected to show the fundamentals most clearly are shown in Figs. A-1 and A-2. The agreement between the frequencies here reported and those found previously in the infrared of the vapor is very satisfactory, the differences at most amounting to a few cm-l and being well within the normal shift in frequency observed in the transition from vapor to liquid. Several overtones and combinations were observed on some of the longer exposures on BH3CO which are not shown in the figure. No bands attributable to diborane or CO were observed in any of the spectra, indicating a fairly high purity for the compounds. However, a weak band was observed at 2411 cm`: in the spectrum of the deuterated compound, which indicates a small amount of hydrogen to be present. ASSIGNMENTS The BH3CO molecule has C3v symmetry which predicts eight fundamentals, all active in the Raman effect, which are either totally symmetric (Al) or doubly degenerate (E). Previous work on boron compounds has shown that B-H stretching frequencies fall in the range between 2000 and 2600 cm"1. Three frequencies appear in this range in the BH3CO spectrum. Deuterium substitution affects only two, however, and on the basis of their polarization characteristics the band at 2380 cm-I is assigned as vl and the band at 2434 cm as V5. 13

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN TABLE I. OBSERVED RAMAN FREQUENCIES OF LIQUID BH3CO AT -800c Band Position Ban Posito Intensity Assignment (in cm-F) 517 m V8 - e fundamental 632 vvw 2v8 692 w (pol.) v4- al fundamental 705 ~- 2 vw v4 - B10 isotopic species 816 w V7 - e fundamental 1073 s (pol.?) V3 - al fundamental 1101 m ve- e fundamental 1133 w V7 + V8 1626 - vvw 2v7 1761 vvw v3 + v4 1887 vvw v6 + v7 2129 vw 2v3 2169 s (pol.) vz - al fundamental 2380 s (pol.) - a, fundamental 2434 s v5 - e fundamental 2703 vvw V1 + v8 TABLE II. OBSERVED RAMAN FREQUENCIES OF LIQUID BD3CO AT -80oC Band Position Band Position Intensity Assignment (in cm-') 264 w v8 - e fundamental 619 i V4 - al fundamental 706 w 7 - e fundamental 808 mn vs - e fundamental 860 m v3 - a1 fundamental 881 w ~- -Blo isotopic species 991 w? 1678 s vl - al fundamental 1749 s 2V3 1777 w 2vj - B10 isotopic species 1825 s 5 - e fundamental 2169 s V2- a1 fundamental 2411 w B-I stretch.,,,,,,,,,.,. _..,..........,.,, L.., _ j, _,,.,, J. _,

ENGINEERING RESEARCH 1INSTITUTE * UNIVERSITY OF MICHIGAN Hg I I I I BH3C0 0 1000 2000 CM'-' Fig. A-1. The Raman spectrum of liquid BH3CO at -80~C (1-ml sample). Hg BD3GO I I I I I 0 1000 2000 CM' Fig. A-2. The Raman spectrum of liquid BD3CO at -800~C (Oo2-ml sample), In the BD3CO spectrum, the asymmetric frequency v5 occurs at 1825 cm, but the position of vl cannot be determined exactly because of Fermi resonance with the overtone of the fundamental at 860 cm-z. The two members of the Fermi doublet occur at 1678 and 1749 cm-l. The latter is assigned as the overtone and the former to the fundamental on the basis of the B10 satellite appearing on the high-frequency side of 1749 at 1777 cm-1, Comparison of the intensities of the two bands indicates that the coincidence between the overtone and 15

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN the fundamental is very close and consequently the unperturbed value of v, probably is not far from 1700 cm-l on the high-frequency side. The frequency vl was not observed in the infrared spectrum of the hydrogen compound1 but its predicted value agrees with that given above. The third band in the 2000-cml| region is immediately identified as the C-0 stretch both from its nearness to the carbon monoxide frequency at 2143 cma1 and from the fact that deuterium substitution does not shift its position. The assignment of the band is further confirmed by its polarization characteristics. The situation with regard to the B-H bending modes is somewhat more complicated. In the hydrogen compound, a triplet is observed in the 1100-cmm1 region with maxima at 1073, 1101, and 1133 cm-', the first being the most intense and probably polarized. In the BD3C0 spectrum, two bands of approximately equal intensity appear at 808 and 860 cm, the latter having a weak satallite on its high-frequency side at 881 cm-'. Since polarization measurements were not made on the deuterated spectrum, the product rule plus the resuits of the normal coordinate treatment were necessary to arrive at a satist factory assignment. Fortunately, the dimensions and moments of inertia of the four possible isotopic molecules of C3v symmetry, B H3C0, B10D3CO, B 13H3CO, B1~D3C0, have been determined from microwave results3 so that the theoretical product ratios can be calculated with no assumptions. The closest agreement with the theoretical values is obtained by assigning v3 to the 1073-cma and v6 to the 1101-cml band in the hydrogen compounds and v3 to the 860-cm-"' and v6 to the 808-cml1 band of the deuterium species. Confirmation for the interchange in the relative positions of the two bands in the deuterium case is found in the normal coordinate treatment. This predicts that the AL frequency of the isotopic B10 molecule is 22 cm-1 higher than the A! frequency of the molecule containing the more abundant B1l isotope. In the case of the E frequencies, however, the difference amounts to only 5 cmi1, a separation that would not be resolved under the present circumstances. The presence of a weak satellite 21 cm'l higher than 860 cmA1 is therefore accepted as evidence that the latter band is actually the A, band, the satellite being assigned as v3 of the Bo1 isotopic species. In the hydrogen compound, the BlO isotope shifts are calculated to be +12 and +4 cm, respectively, for v3 and V6e Since the observed spacings between the members of the triplet are 28 and. 32 cm 1 it appears that neither can be easily assigned to the B~O species. However, the combination of the two E modes at 816 and 317 cm has a calculated value of 1133 cm-1 and the correct symmetry to resonate with the E fndamental and borrow sufficient intensity to appear as a weak band. The band at 1101. cm-' accordingly is assigned as v6e In the infrared work, v6 was assigned to a band at 1392 cm"'. No band at this position was observed in the Raman spectrum and. it appears that the infrared band most likely i 3 8whc foteBradtis v + v whicalclulated at 1390 cm1l. The infrared band at 1105 cm'l was assumed to be vs3. The error in the infrared assignments thus appears to arise from the failure to resolve the three bands observed in the Raman spectrum in this region.'The remaining fundamentals may all be classed as skeletal modes The only polarized, 16(

ENGINEERING RESEARCH INSTITUTE ~* UNIVERSITY OF MICHIGAN fairly intense band left occurs at 692 cm-1 in the BH3CO spectrum and shifts to 619 cm-1 upon deuteration. This is assigned to v4 in agreement with the infrared results. A satellite was observed at 705 cm-l in the more intense exposures on the hydrogen compound and on the basis of a calculated shift of +16 cm-1 from the force-constant treatment is assigned to v4 of Bl~H3CO. The corresponding shift in the deuterated molecule is calculated to be only 5 cm 1 and accounts for the failure to observe a satellite to the 619-cm'l band. The two fundamentals v7 and v8 can be considered as bending motions of the axial chain of atoms. The second, v8, the B-C-O deformation, is to be expected at a rather low frequency in view of the masses of the atoms involved. It consequently is assigned to the moderately intense depolarized band at 517 cm 1 in the BH3CO spectrum. This fundamental was not observed in the infrared work but its position was predicted quite accurately. In the deuterated spectrum it appears at 264 cm1. The last fundamental, v7, which is most simply described as a BH3 rock, is assigned to 816 cml partly by a process of elimination and partly from the infrared evidence. The corresponding band at 706 cm-1 in the deuterated compound is rather weak but the correctness of the assignment is substantiated by the product-rule calculations. NORMAL-COORDINATE TREATMENT Cowan1 carried out a normal-coordinate treatment of the BH3CO molecule based on the results of his infrared study and obtained a set of force constants which produced reasonably satisfactory agreement with his assignments. However, in view of the incorrect assignment for v6 and the fact that data on the deuterated molecule were not available, it would appear that a better approximation can now be obtained. Since his equations did not include interaction force constants, the molecule was reanalyzed using the FG method of Wilson and the following symmetry coordinates: Al species: S, = AT S2 = AR S3 = 31/ (Arl + Ar2 + Ar3) S4 = 1/ 4 (Ao,2 + A0C23 + A~C31 - A, - A2 - A 3) E species: S5 = 1/H (Ar2 - Ar3) Se = 1/H (A~2 - A3) S7 = 1/H (AbC31, - Cl Ss = Ax17

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN In terms of the molecule parameters, T refers to the C-0 bond, R to the B-C bond, ri to the ith B-H bond, Cij to the H-B-H angle between ri and rj, Bi to the ith H-B-C angle, and 6x to the B-C-0 angle. The equilibrium values for these parameters taken from the microwave work3 are as follows: 0 0 0 T = 1.131 A, R = 1.540 A, r = 1.194 A, c = 113052', P = 104037', and 6 = 180~. The elements of the inverse kinetic-energy (G) matrix were evaluated from the tables of Decius and the note by Ferigle and Meister.5 As a check on the correctness of the equations, the force constants of Cowan were substituted into the secular equation and the roots were found to agree with his calculated values with an average deviation of about 0.5 cml', an amount consistent with accumulated rounding-off errors. It was found possible to match the frequencies of the hydrogen compound to within 0.5% by modifying Cowan's force constants somewhat and introducing two interaction constants, krr and keg. However, this set of constants reproduced the frequencies of the deuterium compound very poorly and was discarded. The final set obtained reproduces the sixteen frequencies of the B11H3CO and B11D3CO molecules with a standard deviation of 0.5% from the observed values. The, calculations actually were carried out in terms of the symmetry force constants, Fi, of which ten were required to produce the above fit according to the following potential function: 2 2 2 2 2 2V = F!S12 + F2S2 + FS + rF4S4 + F5S52 + r2F6S62 + r2F7S72 + RT FsSs + R F24S2S4 + R 2F68S6S8 Since there are four product-rule relations, twelve of the frequencies are independent, making the problem slightly overdetermined. The potential energy may also be expressed in terms of the valence force potential constants which are related to the preceding symmetry force constants. 2V = kTAT2+kRAB2 + kR2R + krr riAr + rr 2k Aaij2 Z Z + r 2ki Ai A 2k + RT kAbSx2 + R kRB Z ARAfi + R 2k 7 AiAbx ~ To calculate the eleven constants in the latter equation from the ten in the preceding, one assumption is necessary. The most convenient one is to equate ko~ to k5. Both sets of constants are given in Table III. The observed values of the fundamentals are compared with the calculated values in Table IV. Since two frequencies attributed to B10 molecules were observed, the calculations were extended to include the two boron isotopic species, the calculated 18

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN TABLE III. FORCE CONSTANTS FOR THE BH3CO MOLECULE Symmetry Value Valence Value Force ) Force (Mllidynes Constant (Millidynes/ ) Constant F1 17.9800 kT 17.9800 F2 2.7875 kR 2.7875 F3 3.2980 kr 3.1607 F4 0.7057 ka 0.4013 F5 3.0920 kp 0.3782 F6 0.2203 k 0.2744 F7 0.2434 krr. 0687 F8 0.2744 kmn = keg 0.1579 F24 -0.1778 kRB 0.1451 F68 0.0793 ki 0.1121 values also being listed in Table IV as the shift in frequency from the calculated value of the respective B1l frequencies. As an additional check on the calculations, the B10 product-rule ratios were calculated from the experimentally observed B11 frequencies and B10 frequencies obtained by adding the calculated shifts to the B11 values. The agreement as shown by the numerical values in Table V is quite satisfactory. DISCUSSION Valence force potential constants can be used in a number of ways to obtain information about the electronic structure of molecules. In general, however, some caution must be used in making comparisons between different molecules since the validity of such comparisons may be affected by a number of factors. For example, the type of potential function employed affects the magnitude of the calculated force constant, as does also the number of assumptions made regarding interaction constants. Other sources of disagreement may arise from anharmonicity effects, the inadequacy of the potential function used and the closeness of fit obtained, assuming, of course, that the correct assignments have been made initially. Obviously, the most meaningful comparisons are those made within a group of fairly closely related molecules of the same symmetry type for which the data have all been analyzed in the same way. Conclusions arrived at under other circumstances may still be of considerablevalue but should be considered qualitative, or at best semi-quantitative, and subject to possible revision. 19

m z z Z TABLE IV. COIPARISON OF OBSERVED AND CALCUIATED VALUES OF THE FUNDAMENTAL FREQUENCIES m m FOR THE VARIOUS ISOTOPIC SPECIES OF THE BH3CO MOLECULE z B'lH3CO B"D3CO'' | B'0H3CO | B'0D3CO Fundamental Calc. Calc. m Ohs C~~~~~~alc. Dif alc Df. Obs. Caic. Diff. % Obs. Caic, Diff. % Shift Obs. Shift Obs. m A1 vz 2380 2380 0 0 (1700) 1703 - - +2.5 - +4.3 - f~ v2 2169 2169 0 0 2169 2169 0 0 0.4 - 0.4 - I vs 1073 1070o -3 0.3 860 863 +3 0.3 12.1 - 22.5 21 + 2 _ v4 692 692 -1.1 619 621 +2.3 16.0 13 2 4.8 Z V7 816 816 0 0 706 704 -2 3 6.3 - 3.2 - Vs 317 315 -22 2.6 264 265 +1.4 0 -.1 - Note: Standard deviation for 14 frequencies = O.Slo. < hD~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r The data enclosed in parentheses are for fundamentals disturbed by Fermi resonances and are estimated x values. - 0 O II z ws 317 31 -2.6 26 265 +1..0 -. 1 g

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN - TABLE V. COMPARISON OF PRODUCT-RULE RATIOS FOR VARIOUS ISOTOPIC COMBINATIONS OF THE BH3CO MOLECULE Isotope Isotopes Symmetry Frequency Product Ratio Held Constant Interchanged Class Theoret,* Calc.** Devo Bgl H/D Al 1.931 1.953 1.1% Bll H/D E 2.513 2.522.4 Bl0 H/D Al 1.929 1.951 1.1 B10 H/D E 2.498 2.507.4 H B"l/Bl~ Al 1.036 1.036.0 H B"Z/Bz~ E 1.017 1.017.0 D Bl"/Bl~ Al 1.037 1.037.0 D Bgl/Blo E 1.024 1.023 -.1..... - -:, -,,.._. -.. _:.r....-.,,, _ ~,,, _.... Calculated from moments of inertia of Ref. 3 and the masses involved. The frequencies for the Bll molecules were those observed; the frequencies for the'10 molecules were obtained by adding the shifts determined in the force-constant treatment to the experimental values of the B11 frequencies. Of the various types of potential constants employed in the valency force field, the bond-stretching force constants are the least affected by the factors mentioned above and can be interpreted in much the same way as bond energies and bond lengths as giving an indication of the electron density concentrated in a given bond. In the carbon-monoxide-borane molecule, one of the interesting observations is the small effect which the presence of the borane group has on the C-O bond. The force constant of the C-O bond in carbon monoxide gas calculated from the observed infrared frequency of 2143 cmX1 is 18.5 md/A. This is decreased only to 17.98 md/l in BH3CO, whereas in nickel carbonyl the C-O force constant is 15.9 md/A6 and in carbon dioxide it is 15.5 md/X. The B-C bond, however, appears about normal, the value of 2.79 md/A being only slightly less than the value 2.9 md/A given by Badgerss rule, and the bond length is in good agreement with the sum of Pauling's covalent-bond radii. The force constant for the B-H bond of 3.16 md/A is slightly less than the value 3.42 md/A given for the B-H bond in diborane.7 These comparisons, if significant, lead to some very interesting speculations regarding the electronic structure of the BH3CO molecule and to some unexpected conclusions about the BH3CO group. Since there is some uncertainty involved, a more detailed discussion will be postponed until the analyses of two or three other molecules containing the BH3 group have been completed. 21

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN The rest of the force constants calculated for the BH3CO molecule either involve the bending of bond angles or are interaction constants. The bending constants are quite reasonable in magnitude, although the H-B-H constant of 0.401 md/A is nearly twice as great as the corresponding value of 0.243 md/A reported for diborane.7 It was found that the calculated frequencies were rather insensitive to most of the possible interaction force constants but quite sensitive to others. In the final set listed in Table III, none of the insensitive constants appear. If the symmetry force constants are examined instead of the valence force constants, it will be seen that the F matrices are nearly diagonal, only one off-diagonal element appearing in each case. In the Al matrix, the off-diagonal element is F24, linking the symmetrical bending coordinate and the B-C stretching coordinate, while in the E matrix it is F68, connecting the BH3 rocking coordinate with the B-C-0 deformation. In both cases, the two lowest frequencies in the respective symmetry classes are involved. In conclusion, it may be pointed out that the experimental productrule ratios for the hydrogen-deuterium substitution listed in Table V are greater than the theoretical values, even though the differences are not large. Exact agreement, of course, is not to be expected since the observed rather than harmonic values have been used in calculating the ratios. However, the normal effect of anharmonicity is to cause the experimental ratios calculated in this way to be less than the theoretical.8 In view of the weight of other evidence, plus the fact that all experimental ratios are greater than the theoretical, it does not seem probable that there is an error in the assignments. The difference, therefore, may also reflect the specific electronic structure of the molecule or the BR3 group and if so, should be found in other molecules containing the borane group. Acknowledgment: The author is indebted to Dr. Tom Bissot and Mr. Coran Cluff for the careful preparation of the compounds used in this study. REFERENCES 1. R. D. Cowan, J. Chem. Phys., 18, 1101 (1950). 2. G. L. Vidale and R. C. Taylor, J. Am. Chem. Soc., 78, 294 (1956). 3. Gordy, Ring, and Burg, Phys. Rev., 78, 512 (1950). 4. J. C. Decius, J. Chem. Phys., 16, 1025 (1948,). 5. S. M. Ferigle and A. G. Meister, J. Chem. Phys., 19, 982 (1951) 22

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF, MICHIGAN REFERENCES (Continued) 6. B. L. Crawford, Jr., and P. C. Cross, J. Chem. Phys., 6, 525 (1938). 7. R. P. Bell and H. C. Longuet-Higgins, Proc. Roy. Soc. London, A183, 357 (1945). 8. F. Halvorsen, Rev. Mod. Phys., 19, 87 (1947). 23