ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Quarterly Report No. 2 THE CHEMISTRY OF BORON HYDRIDES AND RELATED HYDRIDES April-June, 1956 R. W, Parry Goji Kodama So G. Shore R. CO Taylor Earl Alton Project 2469 WRIGHT AIR DEVELOPIMENT CENTER WRIGHT-PATTERSON AIR FORCE BASE, OHIO CONTRACT NO~ AF 33 (616)-3343 August 1956

The University of Michigan * Engineering Research Institute OBJECTIVE A continuation of fundamental studies on the chemistry of the hydride of boron and related compounds. ABSTRACT The ammonolysis of H2BNH2 has been studied and an explanation of pertinent facts has been suggested. Additional data on the system B4Hlo-NH3 indicate that the highest ammoniate may be B4H0lo-6NH3. The study is being continued. Force constants for the borohydride ion and for the borane carbonyl molecule have been calculated from the Raman spectrum and are reported. Molecular-weight data on the system B2H6'2NH3, prepared at -45~C, have been rechecked and earlier results are confirmed. The compound C13AlPF3 has been prepared and its properties are briefl described. ii

The University of Michigan * Engineering Research Institute I. THE REACTIONS AND STRUCTURE OF THE DIAMMONIATE OF DIBORANE A. MOLECULAR-WEIGHT MEASUREMENTS IN LIQUID AMMONIA FOR SAMPLES OF THE DIAMMONIATE PREPARED AT -45~c In an earlier reportl it was noted that the molecular weight of the diammoniate prepared at -45~ differs from that of the "diammoniate" prepared at -78~C. The desirability of additional confirmatory evidence was indicated. Some of the needed data have been obtained. Earlier results have been confirmed. Diammoniate of diborane B2H6-2NH3 was prepared by adding 1.035 mmoles of diborane to frozen ammonia in the same manner in which normal diammoniate is normally prepared. After the diborane had been taken up by the ammonia, the excess ammonia was sublimed from the system at -450C instead of -780C over a period of 12 hours. Ammonia was then returned and the resulting solution was aged at -45~ for around 18 hours. The molecular weight of the above product was determined from the vapor-pressure depression of the ammonia solution. Results, shown in Fig. 1, confirm the earlier observations of Schultz for this compound and are consistent, within limits, with the formula [HB(NH3)31][BH42 for the product. B. COMPOUND PREPARATION H3BNH3 and the conventional B2H6-2NH3, [H2B(NH3)21[BH4] have been prepared for further study. C. STUDIES ON THE AMMONOLYSIS OF BH2NH2 FROM TEE RESIDUES OF TEE ALKALI METAL B2H6 2NH3 REACTION Reports by Schaeffer, Adams, and Koenig2 and independent observations in this laboratory3 indicated that the solid residues from the alkali metal1. R. W. Parry, G. Kodama, and D. R. Schultz, Final Report, Project 1966, Eng. Res. Inst., Univ. of Mich., June, 1956, p. 90. 2. G. W. Schaeffer, M. D. Adams, and F. J. Koenig, S.J., J. Am. Chem. Soc., 78, 725 (1956). 3. S. G. Shore and R. W. Parry, Final Report, Project 1966, Eng. Res. Inst., Univ. of Mich., June, 1956, p. 73. __ __ __ _ __ __ _ __ __ _ __ __ _ __ __ _ 1

The University of Michigan ~ Engineering Research Institute 90 X 80 -J 0 070 — 5C-mI I I 60 0 1.0 2.0 3.0 MOLALITY (BASED ON H3NBH3) Fig. 1. The apparent molecular weight in liquid ammonia of B2H6.2NH3 prepared at -45~C instead of -78~0C. diammoniate reaction (i.e., BH2NH2 and MBH4) reacted with ammonia at room temperature to give off H2. A tracer study3 showed that such hydrogen arises from the interaction of a hydridic boron-hydrogen unit and a protonic nitrogenhydrogen unit. Schaeffer, Adams, and Koenig2 assigned the equation H2BNH2 + NH3 - HB(NH2)2 + H2 15) B(NH2)3 + H2 to the process, but several features of such a simple, direct process are open to question. In the first place, the ready reaction between the very slightly acidic ammonia and the residue H2BNH2 seems inconsistent with the properties expected for H2BNH2. The formally analogous compound [H2BNHCH 3] will float on a slightly acidic water solution at room temperature with only very slow hydrolysis. Complete hydrolysis is effected only by a boiling 20% hydrochloric acid solution. It is difficult to understand why replacement of a hydrogen on the nitrogen by a methyl group should cause such a decrease in the hydridic character of the hydrogens bound to boron that H2BNH2 would react with ammonia at room temperature, whereas H2BNHCH3 would not react with cold aqueous acid at room temperature.

The University of Michigan * Engineering Research Institute Other observations are equally disturbing. Schaeffer, Adams, and Koenig2 reported that the presence of an alkali metal borohydride promoted the reaction —LiBH4 being very effective while KBH4 was relatively ineffective. NaBH4 occupied an intermediate position. Their interpretation of this observation seems to be based on the relative abilities of LiBH4 and KBH4 to hold ammonia in the reaction zone as the reaction temperature of the system was raised. A number of uncertainties in the foregoing observations and interpretations have prompted a more thorough study of the system H2BNH2-NH3. A mixture of H2BNH2 and KBH4 in liquid ammonia was prepared by allowing [H2B(NH3I)2[BH4] to react with a slight excess of potassium in liquid ammonia. After reaction was complete all excess potassium was removed quantitatively by amalgamating with about 0.7 cc of liquid mercury at -355C. The solution was filtered from the potassium-amalgam and the solvent was removed from the system at -78~. The mixture of pure KBH4-H2BNH2 was transferred under nitrogen to a dry box where it could be weighed. In a separate reaction vessel a sample (1.09 mmoles) of commercial LiBH4 was attached to the vacuum line and dissolved in 1 cc of liquid ammonia. Hydrogen, which evolved very slowly, was measured. After 24 hours a sample consisting of about.07 mmole each of the KBH4-H2BNH2 mixture was weighed out in a dry box and introduced into the system under dry nitrogen. The entire mass was dissolved in liquid ammonia,then the liquid phase was removed and the system was allowed to warm to room temperature in contact with gaseous NH3. Data in Fig. 2 show that hydrogen evolution was extremely slow under these conditions, less than.018 mmole of H2 being evolved from.07 mmole of H2BNH2 in 18 hours at room temperature. If the explanation of Schaeffer, Adams, and Koenig2 were correct, this system should have lost H2 at an appreciable rate since H2BNH2, KBH4, and a considerable excess of LiBH4 were all present. As the data in Fig. 2 show, hydrogen evolution became extremely rapid after an excess of Na metal was introduced into the system; the entire solid was dissolved in liquid ammonia; then the liquid phase was removed and the solids were held at room temperature in contact with gaseous NH3. Hydrogen corresponding to the reaction H2BNH2 + NH3.. -- HB(NH2)3 + H2 was evolved rapidly, while hydrogen corresponding to the second step evolved more slowly. Do DISCUSSION The foregoing observations suggest that it is not LiBH4 which is responsible for the reaction between H2BNH2 and NH3 at room temperature, but the alkali metal. When excess alkali metal was rigorously removed from the system by mercury extraction, H2BNH2 and LiBH4 reacted very slowly with gaseous ammonia at room temperature. Addition of alkali metal accelerated the reaction in a dramatic fashion. The following explanation is consistent with all facts now available, including those from the tracer study, the work of Schaeffer, Adams, and Koenig, and the present investigation. 3

The University of Michigan ~ Engineering Research Institute I4 |Li BH4-V NH'V% BHONH2 (AND KBH4) z AND EXCESS No N 12 o0 ROOM TEMP0 Q.09 c.08 0.07- -I II 80C~ H/BH1-IHgN 3 I>, 2 [LiBH4' - NH3] ELiBH4, NH3 AND 1 o~. BH2NH2 (AND KBH4) I H/BH2NH =2 N.05 u~~~~~~~~ —04 ROOM TEMP b.03 Bur No Cmbl fon th yroe f hi ADDED 0.02ROOM TEMP 01 -780C 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 HR Fig. 2. Rate of H2 evolution in system H2BNH2-MBH4 and alkali metal. It is postulated that the free alkali metal in the system reacts with the H2BNH2 in accordance with the following equation to give a compound comparable to the "base" Na2[HB(CH53)2] described by Burg and Campbell:4 2Na + H2BNH2 —. Na2 [H2BNH21 Burg and Campbell found the hydrogen of their compound, Na2[HB(CH5)2, to be extremely hydridic in character. The compound reduced SiH3C1 to SiH4. They also found some evidence which can be interpreted as indicating ammonolysis of Na2[HB(CH3)2] at temperatures below -500C. In view of these observations the compound Na2[H2BNH2] would be expected to be easily ammonolyzed at room temperature: Na2[H2BNH2J + NH --- Na2[HB(NH2)2] + H2 Na2 HB(NH2)2 + lNH3 Na2EB(NH2)3] + H2 Na2B(NH2)3..;.. 2Na + B(NH2)3 2Na + H2BNH2 -j Na2[BH2.N.2J. 4. A. B. Burg and G. W. Campbell, Jr., J. Am. Chem. Soc., 74, 3744 (1952). 4

The University of Michigan * Engineering Research Institute In accordance with the above sequence a small amount of alkali metal could well be catalytic for the hydrogen evolution. If one recalls the fact that LiBH4 is very proton sensitive, NaBH4 less so, and KBH4 stable even in water solution, it is not difficult to rationalize Schaeffer's observation regarding the relative ammonolysis rates for the residues resulting from the reaction of Li, Na, or K with B2H6'2NH3. The relative proton sensitivity of Li2(H2BNH2), Na2(H2BNH2), and K2(H2BNH2) would decrease quite rapidly from Li to K, hence the rate of the ammonolysis reaction would be expected to fall sharply, as was observed. Attempts to prepare pure K2EH2BNH2] are now in progress. IIo RAMAN SPECTRALI STUDIES A. THE RAMAN SPECTRUM OF H3BCO. FORCE CONSTANTS FOR THE CARBON MONOXIDEBORAJE MOLECULE The vibrational frequencies (Raman) of the BH3CO and BD3CO molecules and the results of a normal coordinate treatment have been given in a previous report.5 The calculations were carried out using the FG method of Wilson based on the use of symmetry coordinates which are linear combinations of the internal coordinates (bond lengths, bond angles, etc.) of the molecule. In ideal cases, the symmetry coordinates are close approximations to the actual normal coordinates. The force constaznts which are obtained most directly in this approach are the symnetry force constants related to the respective symmetry coordinateso Since these quantities are rather difficult to interpret in terms of the chemical behavior of the molecule, which one is accustomed to thinking of in terms of bonds and bond angles, the symmetry force constants are customarily resolved into the simpler valence force constants by making use of the original transformation between coordinates. Frequently there are more valence force constants than symmetry coordinates so that one cannot resolve certain combinations of the valence force constants unless arbitrary assumptions are made. In the previous report,5 the assumption was made that kcxa = k,, and the values tabulated for the valence force constants were derived on this basis. An alternative procedure for resolving all combinations of force constants, which is more satisfactory, makes use of the fact that the sum of the changes in all the angles around an atom bonded to four others must be identically zero; that is, another (redundant) symmetry coordinate can be defined which must be identically zero. In the case of BH3CO, this coordinate is 5 R. W. Parry, G. Kodama, S. G. Shore, R, C. Taylor, and E. Alton, Quarterly Report No. 1, Project 2469, Eng. Res~ Inst., Univ. of Mich., Jan.-Mar.,1956.

The University of Michigan ~ Engineering Research Institute so = (6)1/2 (Au12 + Ac 23 + Au13 + A +1 + 2 3) 0. By including this coordinate in the original transformation from valence to symmetry coordinates and equating the elements of the F (symmetry force constant) matrix belonging to this redundant coordinate to zero, additional relationships are obtained which allow values to be given to all valence force constants. These revised numbers are given in the table below. VALENCE FORCE CONSTANTS FOR THE BH3CO MOLECULE Force Value Constant _ (Millidynes/Angstrom ) kT 17.9800 kR 2.7875 kr 3.1607 kc o0.2799* k5 0.2644* k 0.2744 krr 0.0687 ko o0.0365* k o0.0o442* kaj 3kLB -0 1176* kR 0o.1451 kR=kR -okR -0.0726* k0 o.o647** *revised **error corrected B. THE RAYMAN SPECTRUM OF BH4. FORCE CONSTANrTS FOR TE BOROHYDRIDE ION Using the frequency assignments reported recently6 for the BH4- and BD4 ions dissolved in liquid ammonia, force constants have been calculated using the general quadratic valence force potential function below: 2V = kr XAri + 2krr AriArj + r2ka Aij + 2r2kj XAcijActk +AijAkl + 2r krk 7 Aor iAij + 2r k' A + 2rr A Ajk 6. R. C. Taylor, A. E. Emery, and D. R. Schultz, Final Report, Project 1966, Eng. Res. Inst., Univ. of Micho, June, 1956, p. 92.

The University of Michigan * Engineering Research Institute Principal force constants in the above are indicated by a single index and interaction constants by a double index; those constants relating two coordinates having only a single atom in common are indicated by a prime. The numerical values for the constants are tabulated in Table I and are compared with the appropriate values of the force constants previously reported for the BH3CO molecule.5 The observed and calculated frequencies are given in Table II. TABLE I FORCE CONSTANTS FOR THE BOROHYDRIDE ION IN LIQUID AMMONIA SOLUTIONS (IN MILLIDYNES/ANGSTROM) Force Comparable Constant Value BH3CO Value kr 2.751 3.161 ka 0.241 0.280 krr 0.0763 o0.0o687 kaa -o.o484 0.0365 kcx -0.0471 -0-.1176 k= kr 0.0550 0.0 krc -o. o550 0.0 TABLE II OBSERVED AND CALCULATED FREQUENCIES FOR THE BOROHYDRIDE ION IN LIQUID AMMONIA SOLUTION Frequency Observed Calculated Diff. % Diff. BH4 v1 Al 2264 cm'1 2241 cm-1 -23 -1.024 V2 E 1205 1212 7 0.58 v3 F2 2239 2238 - 1 -0.04 v4 F2 (1085) 1086 1 0.09 BD4- vl Al 1568 1585 17 1.08 V2 E 862 857 - 5 -0.58 V3 F2 1665 1666 1 o0.0o6 V4 F2 827 822 - 5 -0.60 (The frequency enclosed in parentheses was not directly observed as a fundamental but was obtained from the overtone.) Standard deviation of eight frequencies = o06%.

The University of Michigan * Engineering Research Institute The agreement between observed and calculated values in general is quite satisfactory, except possibly for the Al frequencies. The nature of the disagreement in the latter case perhaps is best discussed by noting that the frequency ratio of the hydrogen to the deuterium vibration is 1.45 as compared to the theoretical harmonic value of 1.41. Since the normal effect of anharmonicity is to cause the observed ratio to be less than the harmonic value, it is possible that one of the observed frequencies has been disturbed by resonance with an overtone or combination of the other fundamentals. It is hoped that this point can be clarified by using isotopically pure B10 in the borohydride ions and liquid ND3 as the solvent. It is of interest to note that the B-H force constant is some 10-15% less in the borohydride ion than it is in the carbon monoxide-borane molecule. Since one would expect stronger bonding for a trigonal sp2 hybridization than for the tetrahedral sp3 configuration, the comparison indicates that the borane group in BH3CO retains a significant amount of the trigonal character expected in free borane despite the contribution from the lone pair of electrons on the carbon. This conclusion is in accord with the larger than tetrahedal angle (113~) observed in BH3CO. Somewhat less information can be obtained from the other constants. The positive sign of the interaction constant, krr, indicates that as one B-H bond is stretched, the others tend to contract. This behavior is to be expected if ionic repulsion between the hydrogen atoms is present. However, removal of a hydrogen would leave behind a BH3 molecule with a trigonal sp2 electronic configuration in which one would also expect shorter B-H bonds. Consequently, both the change in hybridization during a B-H stretching vibration and the coulombic interaction of the hydrogen atoms predict the same sign for the interaction term and one cannot determine conclusively which is dominant. Exactly similar arguments with the same conclusion can be applied in the case of the constants, kra and k~. IIIo THE REACTIONS AND STRUCTURES OF THE AMMONIA ADDITION COMPOUNDS OF B4H10 A. BACKGROUNDI Conditions favoring either symmetrical or nonsymmetrical cleavage of the double bridge bond in B4H10 are summarized in an earlier report.5 B. TmE PREPARATION OF B4O Laboratory stocks of B4H10 became exhausted and new stocks have been made by decomposition of B2H6 under pressure. The flow method described

The University of Michigan * Engineering Research Institute by Burg and Stone7 was used as well as the bomb method in which B2H6 under pressure was stored in a tank at room temperature. More B4H10 is still being made. C. THE REACTION BETWEEN B4H10 AND NH3 Additional data on the system ammonia-B4H10 has been sought. A sample of B4H10 was placed in a tube and increasing amounts of ammonia were added. The equilibrium pressure on the system was measured at -78o, -63~, -450, and -35~0C. Many days were allowed to reach equilibrium at each composition-temperature point. Data are presented in Fig. 3. The highest definite ammoniate appears to be B4H10'6NH3 rather than B4H107NH3 as previously reported.5 This study is being continued. Equilibrium is approached very slowly in all cases../ -350C 400,I ep W 200:) I oo -/ -630C — 78oc I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 NH3/B4H1o0' Fig. 3. The ammonia-B4H10 system. 7. A. B. Burg and F.o G. A. Stone, J. Am. Chem. Soc. 75, 228 (1953).

The University of Michigan ~ Engineering Research Institute IV. TIE REACTION BETWEEN A12C16 AND PF3 An earlier attempt by Chatt and Williams8 to prepare an addition compound between PF3 and AlC13 was made by passing PF3 over sublimed A1C13 at a temperature of 2500C. No addition product was reported. More recently Holmes and Brown9 reported that AlCl3 would not combine with PC13 in a lowtemperature study. Earlier work in this laboratoryl0 indicated that B2H6 at room temperature would combine with PF3 under high pressure to give the unstable compound H3BPF3. BF3 gave no evidence for interaction with PF3 under comparable conditions. From these facts it was suggested that those Lewis acids which are dimeric in the free state would combine with very weak acids such as PF3, whereas those which are monomeric Ei.e., BF3, B(CH3)3, etc.] would not. In an effort to test this prediction, the system A12C16-PF3 has been studied, using the methods employed in the B2H6-PF3 study. Very pure resublimed A12C16 (99.9+% A12C16) reacts with PF3 in a bomb tube under 8 atmospheres pressure in accordance with the following equation: A12C16 + 2PF3 ~z 2C13A1PF3 About 4 hours at room temperature was required for reaction. Shorter periods of time or lower pressures gave no reaction. Longer periods of time permitted group exchange on the complex C13AIPF3 --- PC13 + A1F3 ~ PF3 and AlF3 were recovered as products. The identity of the complex CL3AlPF3 was established by determining the amount of PF3 combining with a given amount of A12C16, then displacing PF3 from an ether solution of the complex at -112CC, using trimethylamine. The amount of PF3 recovered was equal to the amount of trimethylamine added, indicating the reaction Cl3A1PF3 + NMe3 - C13AlNMe3 + PF3. A formal report will be written to describe this work in more detail. 8. J. Chatt and A. A. Williams, J. Chem. Soc., 3061 (1951). 9. R. R. Holmes and H. C. Brown, Paper No. 70, Div. of Phys. and Inorg. Chemistry, Meeting of American Chemical Society, Dallas, Texas, April 12, 1956. 10. R. W. Parry and T. C. Bissot, J. Am. Chem. Soc., 78, 1524 (1956). 10