ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Final Report CHEMISTRY OF BORON HYDRIDES AND RELATED HYDRIDES WADC Technical Report 56-318 R. W. Parry R. C. Taylor To C. Bissot G. Kodama D. H. Campbell D. R. Schultz A. E. Emery S. G. Shore P. R, Girardot Jo T. Yoke III Project 1966 WRIGHT AIR DEVELOPMENT CENTER, U S.o AIR FORCE CONTRACT AF 33(616)-8, Ee O R-464 Br-1 June 1956

FOREWORD The work reported herein was conducted under Contract AF 33(616).-8 with the United States Air Force, the sponsoring agency being the Aeronautical Research Laboratory of the Wright Air Development Center, Air Research and Development Command. Direct technical responsibility for the investigation was vested in the Chemistry Research Branch of the Aeronautical Research Laboratory under the supervision of Dr. Lloyd A. Wood. Lt. David A. Fischer served as technical representative of the Government during much of this work. The authors wish to express their sincere appreciation to the scientists, officers, and men of WADC who provided an ideal situation for a productive scientific investigation. The work was conducted in the Department of Chemistry of The University of Michigan, through the Engineering Research Institute of the University. Administrative assistance by Dr. L. C. Anderson, Chairman of the Department of Chemistry, and W. E. Quinsey, Project Representative for the ERI is gratefully acknowledged. Last but not least, the authors wish to express to Prof. Kasimir Fajans of The University of Michigan their appreciation for his inspiring and iconoclastic views which served to promote this work in many ways. This investigation has been stimuto a great degree by the "Quanticule Theory." WADC TR 56-318 iii

TABLE OF CONTENTS Page LIST OF TABLES vi LIST OF FIGURES x ABSTRACT xiii I. INTRODUCTION 1 II. METHATHESIS REACTIONS OF BOROHYDRIDES IN LIQUID AMMONIA 2 A BACKGROUND 3 B. THE PREPARATION AND PROPERTIES OF HEXAMMINE COBALT (III) BOROHYDRIDE, HEXAMMINE CHROMIUM (III) BOROHYDRIDE, AND AMMONIUM BOROHYDRIDE 4 III. THE STRUCTURE AND CHEMISTRY OF THE DIAMMONIATE OF DIBORANE 16 A. BACKGROUND 17 B. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (I). EVIDENCE FOR TEE BOROIYDRIDE ION AND FOR THE DIHYDRIDO-DIAMMINE-BORON (III) CATION 19 C. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (II). THE PREPARATION OF AMMONIA BORANE 31 D. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (III). THE REACTIONS OF BOROHYDRIDE SALTS WITH LITHIUM HALIDES AND ALUMINUM CHLORIDE 43 E. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (IV). THE REACTION OF SODIUM WITH LEWIS ACIDS IN LIQUID AMMONIA 54 F. A TRACER STUDY OF THE REACTION BETWEEN SODIUM AND THE DIAMMONIATE OF DIBORANE 68 G. MOLECULAR-WEIGHT MEASUREMENTS IN LIQUID AMMONIA. THE MOLECULAR WEIGHTS OF THE METHYLAMINE BORANES, THE DIAMMONIATE OF DIBORANE, AND AMMONIA BORON TRIFLUORIDE 82 H. THE RAMAN SPECTRUM OF THE BOROHYDRIDE ION AND EVIDENCE FOR THE CONSTITUTION OF THE DIAMMONIATE OF DIBORANE 92 IV. UNUSUAL COMPOUNDS RESULTING FROM THE BORANE GROUP 102 A, BACKGROUND 103 B., PREPARATION AND PROPERTIES OF TRIMERIC N-METHYL-AMINOBORANE 104 CO THE PREPARATION AND PROPERTIES OF PHOSPHORUS TRIFLUORIDE —BORANE AND PHOSPHORUS TRIFLUORIDE —BORANE-d3 108 WADC TR 56-318 iv

TABLE OF CONTENTS (Conc luded) Page V. THE REACTIONS OF DIBORANE WITH HYDROXYLAMINE AND ITS METHYL DERIVATIVES 117 A. THE PHYSICAL AND CHEMICAL PROPERTIES OF THE HYDROXYLAMINES 118 B. THE REACTION OF HYDROXYLAMINE AND ITS N-METHYL DERIVATIVES WITH DIBORANE 134 C, THE REACTION OF O-METHYLHYDROXYLAMINE AND ITS N-METHYL DERIVATIVES WITH DIBORANE 147 VI. THE REDUCTION OF CHLOROPHOSPHINES WITH LITHIUM ALUMINUM HYDRIDE AND WITH LITHIUM HYDRIDE 161 APPENDIX. THE ISOTOPE EFFECT IN THE TRACER STUDIES OF THE "DIAMMONIATE OF DIBORANE" 171 WADC TR 56-318 v

LIST OF TABLES Section I IB Page I, Effect of pH of Original Solution on the Composition of the [Co(NH3)6] F3 Crystals 8 II. Analysis of Borohydrides 11 Section III-B I. Reaction of the Diammoniate of Diborane with Sodium Borohydride or Ammonium Bromide and of Sodium Borohydride with Ammonium Bromide 21 II. X-Ray Powder Patterns for H2B(NH3)2Br and NH4Br 23 III. The Reaction in Liquid Ammonia Between Na and the Product Obtained from the Action of NJH4Br on B2H6*2NH3 24 Section III-C I. The Reaction Between the "Diammoniate of Diborane" and AmmoniumChloride Slurried in Ether 34 II. The Reaction Between Lithium Borohydride and an Ammonium Salt Slurried in Ether 35 III. o Comparison of the Properties of Ammonia-Borane with Those of the "Diammoniate of Diborane" 36 Section III-D I. The Reaction Between the "Diammoniate of Diborane" and Aluminum Chloride in Ether 46 Section III-E I. The Reaction of Ammonium Carbamate with Sodium in Liquid Ammonia 56 II. A Comparison of the Results of Schlesinger and Burg with Those from This Laboratory 58 III. The Reaction of the Diatmonniate of Diborane with Sodium in Liquid AmmOnia 60 WADC TR 565318 vi

LIST OF TABLES (Continued) Section III-E (Conttd) Page IV. The Reaction of Sodium with Diborane-Ammonia Systems from Which Ammonia Was Never Removed 60 Section III-F I. A Tracer Study of the Reaction of the "Diammoniate of Diborane" with Sodium in Liquid Ammonia 70 II. A Tracer Study of the Reaction of Sodium with Solutions of Diborane in Ammonia 72 III. A Tracer Study of the Reaction of Mistreated Diammoniate of Diborane" with Sodium in Liquid Ammonia 74 IV. Ammonolysis Reactions 75 Section III-G I. Molecular Weights of Urea 84 II. Molecular Weights of NH4Br, NH4BF4, and NaBH4 in Liquid Ammonia 86 Section III-H I. Raman Frequencies (in cm-1) and Assignments for Lithium, Sodium, Potassium, and Ammonium Borohydrides and for Lithium and Potassium Borodeuterides Dissolved in Liquid Ammonia 96 II. Raman Frequencies (in cml ) of the biammoniate of Diborane Dissolved in Liquid Ammonia 98 Section IV-B I. Analysis for (CH3NHBH2)3, % 105 II. Interplanar Spacings (d Values) and Relative Line Intensities of Trimeric N-Methylaminoborane 107 WADC TR 56-318 vii

LIST OF TABLES (Continued) Section IV-C Page I. Comparison of the Properties of Carbon Monoxide —Borane and Phosphorus Trifluoride -Borane 112 II. Data on Reactions Between Phosphorus Trifluoride and Diborane 114 III. Vapor Pressures of Phosphorus Trifluoride-Borane 115 IV. Vapor Pressures of Phosphorus Trifluoride -Borane-d3 115 Section V-A I. Melting Points of Hydroxylamine and Its Methyl-Substituted Derivatives 122 II. The Vapor Pressure of O-Methylhydroxylamine 124 III. The Vapor Pressure of O,N-Dimethylhydroxylamine 125 IVo The Vapor Pressure of O,N,N-Trimethylhydroxylamine and NMethylhydroxylamine 126 V. The Vapor Pressure of N,N-Dimethylhydroxylamine 127 VI. Melting Points, Boiling Points, Heats of Vaporization, and Trouton Constants of Methyl-Substituted Hydroxylamines 127 VII. pKa Values for Ammonia, Hydroxylamine, and Methyl-Substituted Hydroxylamines 129 VIII. A Comparison of the "Trouton Constant" at a Pressure of 200 mm with the Hildebrand-Trouton Constant at a Uniform Vapor Concentration of 0.00507 Mole/Liter for a Series of Amines 133 Section V-B I. Summary of the Properties of the Borane Addition Compounds of Hydroxylamine and Its N-Methyl Derivatives 135 II. Reaction between N,N-Dimethylhydroxylamine and Excess Diborane 139 III. Catalytic Decomposition of Pure N,N -Dimethylhydroxylamine -Borane by Diborane at 25~C 139 WADC ITR 56-318 viii

LIST OF TABLES (Concluded) Section V-C Page I. Summary of the Properties of the Borane Addition Compounds of Hydroxylamine and its Methyl-Substituted Derivatives 152 II. Characterization of t-Monoaminodiborane 159 Section VI I. Vapor Pressures of Mono- and Diethyl Phosphines 165 II. The Reaction of Phosphines and Amines with LiA1H4 169 WADC TR 56-318 ix

LIST OF FIGURES Section II-B Page 1. Apparatus for the preparation of hexammine metel (III) borohydrides. 10 2. Molecular weight of the "diammoniate of diborane" as compared to that for the decomposition product from NH4BH4. 14 3. Molecular-weight data for NH4Br, NH4BF4, and NaBH4 in liquid ammonia as a function of solution concentration. 15 Section III-C 1. The reaction between ammonium chloride and the "'diammoniate of diborane" in an ether slurry. 33 2. Apparatus for the reaction between ammonium chloride and the "diammoniate of diborane" in an ether slurry. 40 Section III-D 1. The decomposition of the "diammoniate of diborane" in the presence of a lithium halide. 45 2, Apparatus for the preparation of solutions of aluminum chloride in ether. 49 5. Reactor tube for the decomposition of the "diammoniate of diborane" in the presence of a lithium halide. 51 4. Apparatus for the reaction between the "diammoniate of diborane" and aluminum chloride in ethero 52 Section III-E 1.o Bulb crusher for reactions with sodium in liquid ammonia. 65 Section III-F 1. Sodium still. 78 2. Schematic diagram of system for tracer studies. 80 WADC TR 56.318 x

LIST OF FIGURES (Continued) Section III-G Page 1. Molecular-weight apparatus. 85 2. Determination of the molecular weight of NH3BF3 in liquid ammonia. 87 3. Molecular weights of the methylamine boranes in liquid ammonia. 89 4. Molecular-weight measurements on the "diammoniate of diborane." 91 Section IIX-H 1. Apparatus for purification and loading of samples in a Raman tube. 94 2. The B-H stretching region in the Raman spectra of some borohydrides dissolved in liquid ammonia. Tracing 1, LiBH4; 2, NaBH4; 3, the diammoniate of diborane. (Frequency in cm-1) 99 Section IV-C 1. Decomposition of OCBH3 and PF3BH3 at room temperature. 110 Section V-A 1. The pKa values for CH3, OH, and OCH3 substituted amines. 132 Section V-B la. Mole ratio of hydrogen to H2NOHBH3. 137 lb. Decomposition curves of hydroxyl- and methyl-substituted hydroxylamine boranes. 137 2. Special weighing unit for hydroxylamine. 143 3. Reaction of diborane with hydroxylamine at -1120C. 146 WADC TR 56-318 xi

LIST OF FIGURES (Concluded) Section V-C Page 1. Decomposition of NH20CH3'BH3 at 550C. 150 Section VI 1. Apparatus for production of phosphines by reduction of chlorophosphines with LiALH4. 167 Appendix 1. An approximation of ao for the reaction of B2D6'2ND3 with sodium in ammonia-d3. 174 WADC TR 56-318 xii

AB STRACT The types of coordination compounds formed by diborane with various Lewis bases have been explored and the reactions of such coordination compounds have been delineated. The bases studied were: ammonia, methylamine, dimethylamine, trimethylamine, phosphorus trifluoride, nitrogen trifluoride, carbon monoxide, hydroxylamine, N-methylhydroxylamine, N-N dimethyl hydroxylamine, O-methyl hydroxylamine, O,N-dimethyl hydroxylamine, and trimethyl hydroxylamine. The hexammine metal (III) borohydrides have been prepared and it has been shown that the "diammoniate of diborane" has the structure [H2B(NH3)2]+ (BHI4)'. The decompositions of the borane adducts of the methyl hydroxylamines have been studied in some detail, and mechanistic arguments for the decompositions have been considered. The reduction of phosphorus trichloride by lithium aluminum hydride has been studied. Results of the diborane reaction are interpreted in terms of symmetrical and nonsymmetrical cleavage of the double-bridge bond in diborane. WADC TR 56-318 xiii

I. INTRODUCTION Boron compounds, particularly the hydrides, have been recognized as theoretical puzzles for'many years. Recent interest in the large-scale synthesis of certain boron-hydrogen compounds has transformed the theoretical puzzle into a large-scale practical problem. Methods used in developing chemical processes usually owe their success to an application of appropriate theoretical guides. For example, a truly amazing theoretical framework guides the organic chemist in his work. On the other hand, conventional chemical theories, particularly those developed for carbon chemist:ry, usually fail miserably when applied to boron hydrides. As an example, consider the simplest of the free boron hydrides, diborane' —B2H6. The organic chemist pictures a covalent chemical bond as arising from the sharing of an -electron pair, yet B2H6 has more apparent bonds than electron pairs. It is frequently referred to as an "electron-deficient" molecule. Such electron deficiency has been treated extensively in theoretical papers but the results offer very little help in correlating the reactions of boron hydrides. In general, the reactions of carbon cannot be transferred to a study of boron hydrides; hence, it is now painfully apparent that the laws and generalizations correlating the chemistry of boron compounds must be developed independently. The studies described herein were conducted in the hope that a useful reaction pattern for certain boron compounds might be evolved. A number of seemingly separate studies have been conducted during the course of this work, yet all are related through their contribution to an understanding of the reaction pattern of boron hydrides. It appears to be convenient to discuss the various facets of the work as individual units and then to correlate the pertinent observations. For this reason the report will take the form of a series of separate papers, all of which are germane to the announced objective. The credit and -responsibility for each set of data are also indicated more precisely using this pattern. -WADC TR 56-518 1

IIo METATHESIS REACTIONS OF BOROHYDRIDES IN LIQUID AMMONGA WADC TR 56-318 2

A. BACKGROUND Ever since the first description of aluminum borohydride by Schlesinger and his co-wo.rkersl the synthesis, by metathesis reactions in suitable solvents, of new and potentially interesting borohydrides has. seemed worthy of study. It should be noted that the borohydride ion has much in common with the better known complex fluoroborate'ion although several important differences do exist. The stability of the alkali metal borohydrides increases as the size of the cation increases. On the other hand, it is normally assumed that -the presence of protonic or acidic hydrogens in the cation reduces the stability of the borohydride. For these reasons it seemed to be relevant to study the stability characteristics of the hexaimmine metal (III) borohydrides. The hexammine metal (III) cation is large, yet it contains potentially acidic hydrogens attached to the coordinated ammonia. The remote possibility of putting the borohydride ion in the coordination sphere of the Co (III) seemed worthy of consideration and the possibility of obtaining interesting metal-nitrogen-boron polymers offered some additional justification for the study. 1. H. I, Schlesinger, R. I, Sanderson, and A. B. Burg, J. Am. Chem. Soc., 61, 536 (1939). WADC TR 56-318 3

B. THE PREPARATION AND PROPERTIES OF HEXAMMINE COBALT (III) BOROHYDRIDE, HEXAMMINE CHROMIUM (III) BOROHYDRIDE, AND AMMONIUM BOROHIHYDRIDE (R. W. Parry, D. R. Schultz, and P. R. Girardot) Abstract The methods for the preparation of hexammine cobalt (III) borohydride, hexammine chromium (III) borohydride, and ammonium borohydride are described. Hexammine chromium (III) borohydride is a stable compound up to 60'C, the cobalt compound decomposes at 250C under high vacuum, and amrnonium borohydride is unstable above -40~C. The thermal decomposition of these compounds, as well as their decomposition in aqueous solution, has been studied. The synthesis of penetration-type metal ammine borohydrides may be achieved by metathesis reactions in liquid ammonia of the general type MF +: xNaBH4 ix NaF + M(BH4 )x -45~C This reaction has been effected for those cases in which M is the hexammine chromium (III) ion, the hexammine cobalt (III) ion, and the ammonium ion, The resulting new compounds, hexammine chromium (III) borohydride and hexammine cobalt (III) borohydride, are described herein. Although ammonium borohydride was mentioned briefly by Armstrong1 in a previously classified industrial report, little information on its properties was available. In a study of the "diammoniate of diborane" the properties of this compound became important, and a more thorough investigation was warranted. From the philosophical standpoint ammonium borohydride is interesting since it represents borohydridic acid in liquid ammoniao Ammonium Borohydride.-If the reaction between ammonium fluoride suspended in liquid ammonia and sodium borohydride is carried out at -400~C, or below, and this temperature is carefully maintained during all filtration and transfer operations no noncondensable gas is evolved. Pure ammonium borohydride may be obtained from the reaction as a white crystalline solid~ Above about -40~C slow decomposition of the solid begins with liberation of hydrogen. At 25~C one sample of ammonium borohydride was 50% decomposed in about:six hours. The Raman spectrum of a liquid ammonia solution of ammonium borohydride compared favorably in every detail, including the shape of the envelope and intensity of lines, with the spectrum of a liquid ammonia solution of sodium boroh~ydride. A detailed consideration of the Raman spectra of a number of borohydrides in liquid ammonia will be deferred to later publications;2 however, the data now available offer convincing evidence of the presence of the borohydride group in the compound. The molecular 1o Aerojet Engineering Corp., Biennial Report Noo 420, January 26, 1950; RTM-66 May 26, 1950. 2. Ro C. Taylor, Do Ro Schultz, and A, E. Emery, this Report po 92. WADC TR 56-318 4

weight of the compound in liquid ammonia solution was measured by vapor-pressure depression. Some difficulties were encountered in the experiment due to decomposition and other factors; however, the data are sufficiently precise to indicate that the molecule is monomeric and undissociated in 1-molar liquid ammonia solution; the apparent molecular weight was about 34 as compared to a theoretical value of 32.89. Other salts such as sodium borohydride and ammonium bromide likewise exist as undissociated ion pairs3 in 1-molar liquid ammonia solution. At room temperature ammonium borohydride loses one mole of hydrogen per mole of salt to give a relatively stable material of empirical composition BNHe. An earlier report4 on this substance, which has the same empirical formula as the "diammoniate" of diborane, indicated that it could not be the classical "diammoniate." It was reported that the new solid did not yield borazene on thermal decomposition while the classical "diammoniate" gives borazene in yields approaching 50%. In the present investigation a portion of the BFNH6 produced from ammonium borohydride was pyrolyzed at 1800C for 2-1/2 hours to give a 15% yield of recovered borazene. The very small scale of the reaction undoubtedly contributed to the low yield (i.e.,.04 mmole of borazene recovered). The isolation of borazene indicates that the BNH6 could be the same species as the classical "diammoniate of diborane." Molecular-weight measurements in liquid ammonia for the compound BIEH6 compared favorably with values for samples of the "diammoniateo" Values were not in agreement with results found for the recently discovered monomer, H3NBH3.5 In. evaluating the molecularweight data certain anomalous features of the vapor-pressure depression measurements on systems of this type prevented one from using the evidence as conclusive proof of the identity of the two materials. Additional evidence was sought in a study of Raman spectra. The Raman spectrum of (H3NBH3)n obtained by the decomposition of ammonium borohydride was almost the same2 as the spectrum of the "diammoniate of diborane." Results are discussed elsewhere (see po 82)~ Hexammine Cobalt (III) Borohydride and Hexammine Chromium (III) Borohydride.A metathesis reaction in liquid ammonia between a suspension of the ammonia insoluble hexammine metal (III) fluoride and ammonia soluble sodium borohydride produced ammonia insoluble sodium fluoride and ammonia soluble, yellow complexes of composition [M(NH3)61(BH4)3. The reaction appeared to go to completion at -63.5~C in about two hours, but longer reaction times were generally used. NaF was identified by fluoride analysis. Ordinarily an excess of M(NH3)6F3 was used in the reaction in order to remove all of the ammonia soluble sodium borohydride. When reaction was complete, the insoluble NaF and excess [M(NH3)6]F3 were filtered from the system to leave a relatively pure solution of [M(NH3)6](BH4)3 in liqu.id ammonia. After evaporation of the solvent, long yellow needle-like crystals of the pure complexes were obtained. Analysis of the yellow crystals indicated the formulas [Co (NH3)6](BH4)3 1.0 NH3 and [Cr(NH3)6](BH4)3 0.5 NH3 3. See page 82 of this report, 4. Aerojet Engineering Corp. Annual Report No. 367, March 11, 1949. 5. S. G. Shore and Ro Wo Parry, J. Amo Chem. Soc, 77, 6084 (1955)o WADC TR 56- 318 5

Since attempts to remove excess ammnonia always resulted in some darkening and decomposition of the solid, particularly for the cobalt complex, the excess ammonia was not removed. The Thermal Decomposition of the Hexammine Metal (III) Borohydrides of Cobalt and Chromium.-Dry hexammine cobalt (III) borohydride loses ammonia and undergoes irreversible decomposition when the ammonia pressure is reduced below 50 mm at 250~Co In contrast, the dry hexammine chromium (III) borohydride is remarkably stable at 250~C even under high vacuum; the decomposition under these conditions was less than 2% as measured by hydrogen evolution. When the chromium complex was heated in vacuum to 6o0C, its decomposition became comparable to that of the cobalt complex at 250~C Data now available indicate that the instability of the cobalt. complex is to be associated with the strong reducing properties of the borohydride group and with the fact that [Co(NH3)6]+3 is reduced more easily than its chromium counterpart. The latter fact is indicated by the value for the standard electrode potential of the oxidation reaction in aqueous solution: [Co((NE3)6]+2 -> [Co(NH3)61+3 + e Eo -0l1 The value for the corresponding chromium couple is not available, but qualitative considerations indicate that it must be more positive than +0.5 since ammoniacal solutions of Cr(NH3)6+2 liberate hydrogen rapidly from water in the presence of finely divided platinum. As would be expected, the reducing properties of the borohydride group cause a different type of thermal decomposition than is observed with the hexammine complexes containing a nonreducing anion. The hexammine cobalt (III) fluoroborate, which is formally analogous to the complex borohydride, decomposes in accordance with the e quation7 6 [Co(NH3)e] (BF4)3 A 6 Co F2 + 12 BF3NH3 + 6NH4 BF4 + N2 + 16 ITH3 The thermal decomposition of the dry hexammine metal (III') borohydrides differs from that described above in that the borohydride group is oxidized instead of the ammonia. Otherwise certain elements of similarity may be noticed, Although.the decomposition is complex, the equation which best describes the main process is [M(mRE3)6](BH4)3 > Ma Ba+b Nb Hx + 2(BH3NH3)x + NH3 + H2 solid Black mixture apparently of MB and NBHx The hydrogen and ammonia were identified positively as gaseous decomposition products. 6, R. W, Parry and Do Ao Bermnan, to be published~ 70 Baly, Von Gunther, and Zinser, Z. anorg. allgemo Chem., 221, 225-48 (1935). WADC TR 56-318 6

The compound (H3BNH3)X was extracted from the solid residue with liquid ammonia and the empirical formula was established by analysis. During the slow decomposition of [Cr(NH3)6](BH4)3 two relatively large clear crystals of solid formed in the cold trap through which gases were removed. Although these crystals were not identified at the time, more recent experience with the new crystalline compound H3NBH3 suggests that the crystals were probably this monomeric species. The bulk of the material of composition (H3NBH3)X remained in the reaction vessel, however, and was apparently the diammoniate of diborane. The overall composition of the ammonia insoluble, solid black residue suggested a mixture of metal boride and polymeric BNHx, where x ranged from O to 4. Unequivocal identification of separate phases in this solid was never achieved. The Reaction Between Water and the Hexammine Metal (III) Borohydrides of Chromium and Cobalt — The hexammine cobalt (III) borohydride dissolved in water to give a yellow solution which decomposed rapidly. A black ferromagnetic precipitate was obtained which gave cobalt (II) ion when acidified. The same black solid was also observed when sodium borohydride was added to a hexammine cobalt (III) chloride solution, Hydrogen was always liberated, The chromium complex borohydride reacted very slowly with water to liberate a small amount of hydrogen, but no precipitate appeared during this period. In the absence of air, the blue chromous ion was produced upon acidification. It is consistent with the above observation to note that hexammine chromium (III) chloride and sodium borohydride gave no apparent reaction in water solution except for the very slow evolution of hydrogen. Reaction ensued, however, when acid was added. Data for the overall hydrolysis reactions of both complex borohydrides in acid solution are consistent with the following equation: [M(NH3)6](BH4)3 + 5H30+ + H20 - M+2 + 3BO2 + 6NH4+ + 11-1/2 H2 The role of H+ in promoting reduction of the chromic complex is apparent from the equation. Experimental. - a, Materials, 1o Ammonium fluoride was Baker's Analyzed gradeo NH4HF2 was of course eliminated when the solid was placed in liquid ammonia, 2, Sodium borohydride was obtained from Metal Hydrides, Inc., and recrystallized twice from liquid ammonia before use. 3. Hexammine cobalt (III) fluoride-the pure complex was prepared by the following reactions: [Co(NH3)6]Cl3 + 3/2 Ag20 + 3/2 H20 OC 3AgCl4, + [Co(NH3)e] (0H)3 WADC TR 56-5318 7

washed thoroughly with distilled water, Hexammine cobalt (III) chloride (0,1 mole) from laboratory stock was placed in a mortar which had been cooled in ice water, and wet Ag20 filter cake was added, The mixture was ground for about one hour, During this time the mortar was placed in an ice-water bath. The cold mixture was filtered and washed on a BUchner funnel, The filtrate was transferred to a polyethylene beaker, and reagent grade HF was added until the pH of the filtrate was seven, as measured by a Beckman, Model H-2 pH meter. The neutral solution was placed in a vacuum desiccator over concentrated sulfuric acid and peireodically evacuated, After drying about one week the crystals were filtered on a sintered glass frit using a drying train to exclude moisture, The solid was washed with a very small amount of cold distilled water, and finally with absolute alcohol. The crystals were dried overnight at 105~C, The solid product was identified as [Co(NEH3)6] F3 by measuring the absorption spectrum of its aqueous solution and by analysis. The spectrum was that of the Co(NH3.3)6+++ ion, The analysis of the solid showed Co-26,9%, NH3-4606%, F-2604%, and HF0.130% calculatedfor [Co(NH3)6] F3: Co-27,02%, NH3-46,84%, and F-26o13%o Since the neutral salt [Co(NH3)6]F'3ishygroscopic and very soluble in water, while the acid salt [Co(NH3)6](EF2)3 is insoluble in acid solution, pH must be carefully controlled in preparing the neutral salt to prevent bifluoride contamination, The effect of pH on the composition of the resultant salt is shown in Table I. TABLE I EFFECT OF pH OF ORIGINAL SOLUTION ON THE COMPOSITIO1N OF THE [Co(NH3)6] F3 CRYSTALS pH % fF- in Salt* pH %.F'in Sal-t* 4,6 12,7 705. o036 507 0,52 8o9 0o, 70o 0.o 39 *Estimated by titration of salt to pH 700 using.naOH. 4, Hexammine chromium (III) fluoride9 which has rot, been previously reported, cannot be prepared in the same manner as the cobalt comolex because the hydroxide is too unstable, The following procedure does, however, yield the pure salt: [Cr(NIH3)6]C13 + 3AgF A [Cr(N13)6] F3 + 3AgCl 0 Hexarmmine chromium (III) chloride (00354 mole) was dissolved in water in a polyethylene beaker0" Silver fluoride solution was prepared in a separate polyethylene beaker by adding excess Ag20 to a dilute (6N) solution of EFo The excess Ag20 was filtered off and the silver fluoride solution was added. carefully to the solution of the chromium complex until no excess Ag+ or C!V WADC TR 56-318 8

could be detected. The silver chloride was filtered and the filtrate (pH = 6.3) was placed in a vacuum desiccator over concentrated sulfuric acid for one week. The crystals formed were filtered on a sintered glass frit with exclusion of air and dried in a vacuum desiccator over NaOH. The following analysis was obtained: Found Theory Found Theory Cr 24.1 24.63 F 27.0 26.99 NH3 47.0 48.38 HF 1.0 --- 5. Liquid ammonia-commercial tank ammonia (Mathesson) was dried over sodium then distilled into the vacuum line. b. Procedure for the Metathesis Reaction. The metathesis reaction could be carried out at any desired low temperature by means of the apparatus shown in Fig. 1. Stoichiometric amounts of the appropriate dry fluorides8 and sodium borohydride were weighed out in a controlled-atmosphere gloved box and placed in tube T. The remaining portion of the apparatus had previously been attached to the high-vacuum line at the ~ 14/35 joint. Final assembly of tube T to the rest of the apparatus was made as rapidly as possible to avoid contact with the atmosphere. After evaluation, ammonia was condensed into tube T and the reaction was carried out at -65~ to -45~C, The mixture was agitated by an electromagnetic plunger-type stirrer which was actuated. 60 times per minute. The reaction was allowed to proceed for 3 to 5 hours, and the mixture was then frozen with liquid nitrogen. The apparatus was inverted by turning through 180~ about the 9 14/35 joint. At this point a metal can with a hole cut in the bottom was placed on the inverted tube T in place of the solenoid. The hole in the can was just large enough to permit a snug fit around the tube and was placed 1-1/2 inches from its end. Dry Ice and i-propyl alcohol were placed in this can and in the cup on R. The remainder of the tube T was wrapped with glass wool. If necessary, additional cooling of the tube could be effected by pouring a small amount of liquid N2 over this glass wool. The solid reaction mixture gradually melted, ran down the sides, and was held on the filter frit in R. As the pressure increased a Dry Ice i-propyl alcohol bath was placed around section S and the solution filtered through the frit. Effective washing of the precipitate and the sides of the reactor could be accomplished by placing an empty Dewar flask around S, opening the stopcock in the by-pass on R, and adding more Dry Ice to the can on T. Then the stopcock on the by-pass was closed and when S was cooled, the solution would filter again. Usually five washings were required to extract all of the product. The solvent ammonia was carefully evaporated, leaving the complex borohydride in S. The apparatus was finally "filled" with dry nitrogen and taken off the line. The products were removed and handled in the dry box. The observed and theoretical compositions for the compounds assumed are summa8. The actual apparatus used to prepare NH4BH4 differed somewhat from that described here, but the essential principles of operation are the same. Details of the other apparatus are available in the doctoral dissertation of D. R. Schultz, Univ. of Mich,, Ann Arbor. WADC TR 56-518 9

* 24/40 Medium Filter A'- R 1 14/35 1 24/40 Solenoid Position for Filtration Fig. 1. Apparatus for the preparation of hexammine metal (III) borohydrides. WADC TR 56-318 10

rized in Table II. Standard methods of analysis were used with minor modifications. Details are available elsewhere. o9 TABLE II ANALYSIS OF BOROHYDRIDES C NH3 % B Hydridic H a c' % Cr ~ Compound. Obs. Theor.* |Obs, I Theor.* Obs. Theor]* |Obs. |Theor * Obs.| Theor. * NH4 BH4** 51.6 51e79 31.4 32 91 9o10/ 12. 20 -- [Co(N3)6 ](BH4)3 1.0 NH3 53o8 53, 5 14,2 14,6 5,0 5o4 26,6 26,4 [Co(NH3)6 ](BH4)3 0.5 NH3 53.3 53.3 15.6 157 57 5.8 -- - 25.3 25.1 *Theoretical value for compound in left-hand column. **The identity of this compound was also confirmed by the stoichiometry of its formation from a known weight of NaBH4 and.an excess of NH4F which could be filtered off. For 2.45 mmoles of NaBH4, 2,48 mmoles of H2 were evolved in the reaction NH4BH4 ~ H2 + (NH3BH3)a For 10.8 mmmoles NaBH4, 10o77 mmoles of H2 were obtained.!/Values for active hydrogen on B2H6 ~ 2NH3 also low by method used. c. The Thermal Decomposition of Hexammine Metal (III) Borohydrides of Cobalt and Chromium. Thermal decomposition of-a sample was carried out in the vessel which had been used earlier for synthesis (Fig. 1). The sample, placed in section T, was frozen with liquid nitrogen and the system was evacuated. Decomposition was accomplished by increasing the temperature or by pumping on the solid. Exit gases were passed through a trap cooled with liquid N2 to remove condensable products. Hydrogen was identified by measuring its molecular weight; ammonia was characterized by molecular-weight and vapor-pressure measurements. After essentially complete gas removal, liquid ammonia was condensed into the tube T and mixed with the solid black residue. The ammonia soluble portion was filtered through the frit R and the residue was washed with liquid ammonia. After removal of the ammonia the residues were removed in the dry box for 9. R, Wo Parry, D. H.e Campbell, De R. Schultz, So G. Shore, To C. Bissot., and R. C. Taylor, "Chemistry of Boron Hydrides and Related Hydrides," Annual Report for 19539 Engo Res. Insto, Univ.o of Mich., Proj. 1966-1-P (May, 1954). WADC TR 56-318 11

analysis. The colorless ammonia soluble residue contained boron and nitrogen in a ratio of 1 to 1,08. H2 was the only other element present. The solubility of the residue in liquid ammonia and its B/N ratio suggested the material (BNHs)x. The ammonia insoluble residue consisted of 4.40 matoms H, 0o879 matoms Cr, 2.42 matomns N, 3.29 matoms B. Such a mixture would correspond to.18 mmole CoB and 0.22 mmoles (BNHx)n. For the chromium salt, the ammonia soluble portion was stained with a trace of undecomposed [Cr (NHs3)6](BH4)3o After correcting for this contaminant about 0.35 mmole of BNHI6 was ~ound. B/N ratio found was 1.02. The ammonia insoluble fraction contained 0.88 matoms Cr, 2.42 matoms N, and 3.29 matoms B. Such a mixture would correspond to 0.88 mmoles CrB and 2o.41 mmoles (BNHx)n where x = 0 to 4. do Thermal Decomposition of Ammonium Borohydrideo Ammoniumn borohydride was prepared in a tube attached to the vacuum system. The white crystalline solid, obtained by evaporation of the solvent ammonium decomposed slowly just above -40~C and quite rapidly at 00C; upon loss of H2 the crystalline solid became an amorphous mass. The system was filled with dry N2 and the apparatus was transferred to the dry box for dismantling. A 25.6-mg sample of the decomposition product was transferred in the dry box to a 20-mm, 8-4nch tube equipped with an inner Standard Taper joint. The tube was opened and stoppered, removed from the dry box, then quickly fastened to the vacuum system. After evacuation, the tube was enclosed in an electric furnace made by wrapping four feet of No. 22 nichrome wire over an asbestos coating on the tube. The wire was covered with more asbestos. Pressure in the reaction vessel was followed by a manmeter and temperature by a thermocouple. The sample was heated for 2-1/2 hours at 180~C. The pressure developed rapidly during the first half hour then leveled off. The sample was subsequently frozen with liquid nitrogen and the noncondensable gas was Toeplerized and measured, (1o53 mmoles), A molecular-weight value of 2.18 by vapor density indicated H2.' A volatile product was vaporized from the reaction tube and fractionated in the vacuum line. Its vapor pressure and molecular weight, tabulated belows: indicate borazene. ~Vapor Pre Sure T~C Observed Borazene -78.5 0.5 mm 0.5 mm Molecular weight by vapor -63.5 1.0 mm 1.0 mm density (obs.) = 77 -45.2 50 mm 5.0o mm -22.9 23.5 mm 23.5 mm Theory for borazene = 80~5 e o. Molecular-Weight Measurements, Molecular weight was determined by measuring vapor-pressure depression in liquid ammonia solutions. Apparatus and general procedure are described elsewhere * Special techniques were required for ammonium bor<ohydride since it decomposes above WADC TR 56-318 12

-400~C The sample was prepared as described earlier, except that the receiver on the filtration apparatus (Fig, 1) was replaced by the sample bulb used for' molecularweight determinations. The product was washed carefully into the bulb and the solvent ammonia was removed slowly at -45~C to prevent spattering. The decomposition at this point, as measured by H2 evolution, was less than 2%. The vacuum apparatus was filled with dry nitrogen and the chilled sample vessel was transferred rapidly to the molecular-weight apparatus, The ammonium borohydride was then frozen with liquid nitrogen, the system was evacuated then known amounts of solvent ammonia were condensed on the sample and the vapor-pressure depressions were measured. Temperature was always maintained below -40~Co After making measurements on the ammnonium borohydride, the solvent ammonia was removed and the sample was warmed to room temperature~ The hydrogen evolved was measured. The value (2,11 mm H2) permitted an estimate of the amount of ammonium borohydride present, Following H2 evolution known amounts of ammonia were condensed on the resulting BNH6 residue, and molecular-weight measurements were made. After completion of the molecular-weight measurements, ammonia was removed under high vacuum at room temperatures, the apparatus was filled with nitrogen, transferred to a dry box, and the sample bulb was weighed (76,0 g solid)o. Hydrolysis of the sample and subsequent Kjeldahl analysis (29.6 mg N) indicated the amount of BNH6 present. The value (2,11 mm) confirmed the earlier estimate of NH4BH4 based on H2 evolution. The difference in total weight based on analysis and direct weighting (6.7 mg) was attributed to NaBH4, By use of this value and the previously determined curve for the molecular weight of NaBH4 in liquid ammonia, the results were corrected for the small NaBH4 contamination. Results are summarized in Figs, 2 and 30 In Fig. 3 the curves for NH44BEF4 and NH4Br are shown for comparison. The amazing rise in apparent molecular weight as the solution approaches infinite dilution leaves some uncertainty as to the real value for the molecular weight but the most reasonable extrapolations consistent with data on other compounds indicate that NH4BH4 was undissociated and monomeric, More extensive data and treatment of molecular-weight measurements are given elsewhere. WADC TR 56-318 13

i X X Decomposition product of NH4BH4 100 13 | \ | | o Value for "classical U t diammoniate of diborane," o-\ This investigation. 90 Ho * I I ~ Value for classical diamrnmoniate from Stock and Pohland. 80 70 60 40 30 1.0 2.0 3-0 4.0 *FORMULA WEIGHTS OF SOLUTE PER 10009g AMMONIA Formula weight= weight of one H3NBH3 unit. * Actual molality is one-half of above values Fig. 2. Molecular weight of the "diammoniate of diborane" as compared to that for the decomposition product from NH4BH4.

N H4 e r 100 so,~ ~~~ L X NH4 BF4 90 80 ui( 70 060 50 4I0 5 1.5 2.0 30 o 0.5 O 1.5 2.0 FORMULA WEIGHTS OF SO!LUTE PER 10OOg NH3 Fig. 3. Molecular-weight data for NI4Br, NH4BF4, and NaBH4 in liquid ammonia as a function of solution concentration. WADC TR 56-318 15

III. THE STRUCTURE AND CHEMISTRY OF THE DIAMMONIATE OF DIBORANE WADC TR 56-518 16

A. BACKGROUND In many of its reactions the molecule B2H6 cleaves smoothly to give two BH3 groups which then react independently with Lewis bases to give regular coordination compounds such as H R -- BH H Reactions of the foregoing type are considered normal for-the diborane molecule. It was then somewhat surprising for Stock and Pohland1 to find that the reaction between the Lewis base ammonia and the Lewis acid, diborane, did not follow this normal pattern but gave instead a compound whose molecular weight in liquid ammonia corresponded to that of a dimer (H3BNH3)2. In order to rationlize this unexpected dimeric formula, Stock2 assumed that two of the hydrogen atoms of diborane are acidic in character. Such an assumption leads to the representation of the diammoniate as an ammonium salt, (NH4)2 B2H4. Schlesinger and Burg5 questioned the formulation of Stock and proposed instead the formula NH4(H3BNH2BH3). This new representation, showing only a single ammonium ion and requiring no acid hydrogens for diborane, was based on the fact that only one equivalent of hydrogen was liberated when they allowed sodium to react with B2H6e'NH3 in liquid ammonia at -77~C, Additional evidence bearing on the question of acid hydrogens in diborane was presented by Burg,4 It has been shown that when the Bronsted-Lowry acid NH4C1 is dissolved in liquid ammonia, a rapid interchange of protons between the acid and the solvent molecule occurs.5 Burg conducted experiments with ND3 and B2H6&2NH3 which showed that H-D interchange occurs only with the hydrogen of the ammonia and not with that of the B2H6 in the diammoniateo He interpreted his results as proof of the assumption that the hydrogens of diborane have no acidic character. Insofar as exchange experiments can be accepted as a criterion of acidic character, Burg's evidence is convincing, although his isotope analysis was based only on differences in vapor pressure between ND3 and NH30 1o Ao Stock and E. Pohland, Bero, 58, 657 (1925). 2. A. Stock. Hydrides of Boron and Silicon. Ithaca: Cornell Univ. Press, 1933. 3. H. I. Schlesinger and A. Burg, J. Am. Chem. Soc., 60, 290 (1938). 4. A. B. Burg, ibid, 69, 747 (1947). 5. C. J. Nyman, Si-Chang Fung, and Ho W. Dodger, ibid., 72, 1033 (1950). WADC TR 56-318 17

Still another possibility has been considered in recent times. Wiberg, Bolz, and Buckheit6 and others7 have suggested that the compound is indeed a monomer, H3BNH3, and the salt-like character and dimeric nature in liquid ammonia are due to dipole-dipole attraction and hence to systematic errors in molecular-weight measurement in liquid ammonia. The enigma presented by the diammoniate of diborane offered a real challenge to anyone attempting a systematic treatment of boron hydride chemistry. In an effort to rationalize the chemistry and structure of this substance, a detailed study of its nature has been conducted. Results are summarized in the following papers. 6. E. Wiberg, A. Bolz, and P. Buckheit, Z. anorg. Chem., 256, 287 and footnote p. 301 (1948). 7. L. J. Agronomow, J. Chim. gen., 9 (71), 1389 (1939); Chem. Zbl., 194I1 1 2362; Gmelints Handbuch der anorganischen Chemrie, 8. Auflage. System 13, pp. 100-101 WADC TR 565318 18

B. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (I). EVIDENCE FOR THE BOROHIYDRIDE ION AND FOR THE DIHYDRIDO-DIAMMINE>BORON (III) CATION (D. R. Schultz and R. W. Parry) Abstract New chemical evidence is cited to support the following points: (1) the proposed existence of an ammonium ion in the diammoniate is inconsistent with its chemical properties; (2) there is strong chemical evidence for the borohydride ion in the diammoniate of diborane; (3) data available are consistent with the formula [H2B(NH3)2](BH4). The most widely accepted present-day structure for the diammoniate of diborane, B2H6-2NH3, is the ammonium model of Schlesinger and Burg,l NH4(H3BNH2BH3). The principal evidence for this formulation was the stoichiometry of the reaction between the diammoniate and sodium in liquid ammonia. Na + NH4(H3BNH2BH3) - Na(H3BHN2BH3) + 1/2 H2 + NH3 (1) Evolution of only one-half mole of hydrogen per mole of diammoniate was cited as evidence for the presence of a single ammonium ion per molecule.1 It is only fair to note that the foregoing ammonium model was proposed before the discovery of the borohydride ion and represented the structure which was most consistent with facts then available. After discovery of the borohydride ion by Schlesinger and his co-workers2 a new structure was equally plausible, [H2B(NH3)2](BH4). This moiety was mentioned briefly in the literature of 1947 but was rejected without serious examination.5 A more careful experimental study of the diammoniate has revealed that the stoichiometry of the sodium reaction is very sensitive to the experimental conditions, hence the ammonium model still leaves many facts unexplained. A more complete report of this work is contained elsewhere.4 In this paper additional independent chemical evidence pertinent to the structure of the diammoniate (prepared under very carefully controlled conditions) is examined in relation to the two foregoing models-the ammonium structure, NH4(H3BNH2BH3), and the borohydride structure, [H2B(NH3)2](BH4). 1. H. If Schlesinger and A. B. Burg, J. Am. Chem. Soc., 60, 290 (1938). 2. H. I. Schlesinger, R. J. Sanderson, and A. B. Burg, ibid,, 6i. 536 (1939); 62, 3421 (1940); A. B. Purg and Hi. I. Schlesinger, ibid.; 62, 3425 (1940); TF. I. Schlc in.; r and I I. Ct. ierown.ibid., 62, 3429 (194g0). 3. A, B, Burg, ibid., 69, 747 (1947). 4. S. G Shore and R. W. Parry, this Report, p. 54. WADC TR 56-318 19

The Reactions of the Diammoniate of Diborane with Ammonium Bromide and Sodium Borohydride. -Although the stoichiometry of the reaction between sodium and the diammoniate of diborane in liquid ammonia is consistent with the presence of an ammonium ion in liquid ammonia, it is important to note that such a reaction is definitely not proof of the existence of an ammonium ion in the solid state. Rather it is evidence for the presence of any Bronsted-Lowry acid which can oxidize the active sodium metal in liquid ammonia solution. For example, the acid behavior of aluminum sulfate in water solution (e.g., reaction with carbonate to liberate C02) is not evidence for the existence of hydronium ion, H30+, in solid aluminum sulfate hydrate, but is, instead, evidence for the existence of water molecules whose proton-donor qualities have been enhanced by coordination of the water molecule to an ion of relatively high field strength. The proton-donor properties of ammonia can likewise be increased by coordination. An independent, though still not conclusive, test for the ammonium ion would be its reaction with borohydride at room temperature to give hydrogen. The validity of this test is indicated by the earlier data on the instability of ammonium borohydride.5 The overall reaction should be (NH4+)(H3BNH2BH3-) + 2NaBH4 -4 2(Na+)(H3BNH2BH3 ) + 2H2. (2) To test for the ammonium. ion, sodium borohydride was added to the diammoniate in two separate experiments, under conditions which resulted in vigorous hydrogen evolution from mixtures of authentic ammonium salts and NaBH4. Results in Table I show that hydrogen evolution did not proceed as demanded by the ammonium formulation. Since the formula (NH4)(H3BNH2BH3) does not permit easy treatment as an ur,.ionized solid structure, one would assume that the behavior of an ammonium ion, if present, should -be typical of other authentic ammonium salts.. The alternative borohydride structure suggests that the diammoniate should react with ammonium bromide rather than with sodium borohydride. The results of such an experimental test are summarized in Table I, they indicated a stepwise liberation of hydrogen which, as expected, eventually approached two moles of hydrogen per mole of diammoniate. In terms of the borohydride model the expected reaction. iy be written as [H2B(E3)Ia]1(BH4) + 2NH4Br - 2[H2B(NH3)2]Br + 2H2T 1 (3) The Isolation and Properties of [H2B(H3)3)2]Br. -Although several features are distasteful, one might still rationalize the reaction between ammonium bromide and the ammonium model of the diamnmoniate of diborane in terms of the following equation which is based on the suggestions of Stock: (NH4) (H3BNH2BH3) + 2NH4Br --- 4 (H2BrBNH2BBrH2) + 2H2 + 21NH3. (4) The above process involves the inconsistency of a reaction between the ammonium salt anion and the ammonium cation in ammonium bromide. More significant, however, is the composition of the product formed. The product expected from the borohydride structure 5. R. W. Parry, D. R. Schultz, and P. R. Girardot, this Report, p. 4. WADC TR 56-318 20

TABLE I REACTION OF THE DIAMMONIATE OF DIBORANE WITH SODIUM BOROHYDRIDE OR AMMONIUM BROMIDE AND OF SODIUM BOROHYDRIDE WITH AMMONIUM BROMIDE Run No. 1 Run No. 2 Run No. 3 A. Diammoniate Reactions Ratio NH3/B2H6 in prep 1.99 2.00 2.01 Temp. x s NH3 removed -450~ -785~0C -45~C B2H6'2NH3 (mmoles) 0O97 2.11 1.735 NaBH4 (mmole s) 2.7 4.35 o NH4Br (mmoles) 0 0 3.47 NH3 (ml liquid) 5 5 5 Reaction temp. 25~ 45~ 25~ Reaction time 5 hr 3 hr 5 hr H2 evolved (mmoles) o.o69 0.166 3.41 Mmnoles H2/mm B2H6o 2NH3 0.071 0.079 1.965 Bo Sodium Borohydride Reactions NaBH4 (mmoles) 3.98 NH3Br (mmoles) 6o95 NH3 (ml liquid) 5 Reaction tempo 25~0 H2 evolved (mmoles) 6.70 H2/NaBH4 1.68 WADC TR 56-318 21

is [H2B(.NH3)2]Br (see Equation 3). The total amount of boron in the product is determined by the amount of B2H6 originally -used.. This value also fixes the amount of active hydrogen remaining in the product if appropriate correction is made for the gaseous hydrogen evolved. It will be noticed, however, from Equations 3 and 4 that the principal point of difference in the stoichlometry of the two overall processes is the evolution of two moles of ammonia in the reaction of the ammonium salt, but the evolution of no ammonia in the reaction of the borohydride salt, In consequence, the molar N/Br ratio expected in the case of the ammonium salt would be 1 to 13NH4(BrH2BNH2BBrH2.). On the other hand, the N/Br ratio expected in the case of the borohydride salt would be 2 to 1; —[H2B(NH3)2]Br. Analysis of the product of the ammonium bromide "diammoniate" reaction gave a molar ratio of 2.02 NH3 per bromide. The observed percentage of N was 22.1 as opposed to theoretical values of 22.09 for [H2B(NH3)2]Br and 12.75 for (NH4)(BH2BrNH2BH2Br). The observed bromide value wras 62.3 as compared to 63.01 expected for H2B(NH3)2Br and 72.70 for the ammonium salt formulation. Both hydridic hydrogen and boron were low when the sample was hydrolyzed with 20% H2S04 at room temperature. Values are: observed H- - 0.27%, theoretical for [H2B(NH3)2a]Br = 1.59, theoretical for (NH4)(H2BrBNH2BBrH2') = 1.84; observed B = 6.77%, theoretical for [HaB(NH3)2]Br = 8.54, for ammonium structure = 9.86o Analysis of the microcrystalline compound, [H2B(EH3)2]Br, by x-ray diffraction indicated a complicated structure with interplanar spacings as shown in Table II. The "d" values indicate absence of ammonium bromide, hence the possibility of an H2BNH2-N:H4Br mixture instead of [H2B(NlH3)2]Br appears small. This is particularly true since it is already known that the H2BNH2-NaBH4 and H2BNH2NaBr mixtures show the x-ray powder patterns for NaBH4 and NaBr, respectively, and a mixture of H2BNH2NH4C1, resulting from the decomposition of [H2B(NH3)2]C,1, shows the lines for iTH4CI clearly. In view of the foregoing evidence it appears that a serious alteration of the NH4Br lattice by polymeric H2BNHf2 would require totally unexpected and unprecedented behavior, The compound. [H2B(NH3)2]Br is a white- solid, soluble in ammonia, and is slowly hydrolyzed. by water. It reacts with sodium in liquid ammonia as shown in the next section, It can be prepared by the reaction between NaBH4 and NH4Br (see Table I). The Reactions of the Diammoniate of Diborane and of LHB(NH )2iBr with So'ium in Liquid Ammonia. -Carefully prepared B2E3-F 2NH3 reacts rapidly, with sodium in liquid ammonia to liberate one-half mole of hydrogen per mole of "diammon-~liatet," Such a process is consistent with either model for the diammoniatee N4 (H3BN.H2BH3) + Na ---- Na(H3BNH2BH3) + 1/2 H2 (5) [H2BNH3)2](BH4) + Na -— 4 NaBH4 + H2BNH2 + NH3 + 1/2 H2. (6) Again, however, the nature of the reaction products provides a means for differentiation. In the ammonium case the solid residue- should contain the complex salt Na('H3BNH2BHs3), whereas in the borohydride model the solid should contain NaBH4. X-ray powder data of Schaeffer, Adams, and Koenig6 indicate that NaBH4 is indeed present as a major component of the solid phase. 6. G' W. Schaeffer, M. D. Adams, and F. J. Koenig, J. Am. Chem. Soc., 78, 725 (1956). WADI)C TR 56 -318 22

TABLE II X-RAY POWDER PATTERNS FOR H2B (NH3)2Br AND NH4Br d Value Relative d Value Relative (Angstroms) Intensity (Angstroms) Intensity A. Interplanar Spacings (d Values) for [H2B(NHI3)2]Br from Debye-Scherrer Powder Patterns 5.27 vs 1.80 vw 4.60 w 1.71 m 4.00 w 1.65 m 3.70 m 1.62 m 3.42 vs 1.55 w 2.88 vs 1.45 v 2.61 s 1.43 vw 2.52 s 1. 38 m 2.26 s 1.33 vw 2.07 s 1.31 v 1.93 w 1.26 w 1.85 w 1.21 w B. Principal Lines for NH4Br (compare above) 2.832 vs 1.639 m 2-303 w 1.413 w 1.991 w 1.327 w 1.785 m 1.241 w The reaction of the complex dilydrido-diammine boron (III) cation [H2B(NH3)2]+ with sodium in liquid ammonia should be independent of the anion of the salt, hence by analogy to Equation 6 one might expect the following reaction for the bromide salt: [H2B(INH3)2](Br) + Na - NaBr + H2BNH2 + NH3 + 1/2 H2. (7) As shown in Table III, the reaction for the dihydrido-diammine boron (III) cation proceeded as expected and NaBr was identified in the solid product by x-ray powder methods. Finally, brief mention should be made of an earlier argument3 to the effect that in [H2B(NH3)2] "the strong proton-donor character of the quaternary nitrogen atoms should render the proton-sensitive BH4- incapable of existence." Coordination theory indicates that the proton-donor characteristic of the nitrogen in [H2B(NH3)+] should be less marked than those in ITHe4. Furthermore, the existence of quaternary nitrogens WADC TR 56-318 23

TABLE III THE REACTION IN LIQUID AMMONIA BETWEEN Na AND THE PRODUCT OBTAINED FROM THE ACTION OF NH4Br on B2H6'2NH3 (A Comparison of Observed Stoichiometry with That Expected from Two Models of the Diammoniate) Observed Theory for Theory for [H2B (NH3 ) 2 ] Br NH4 (H2BrBNH2BBrH2) Mmoles salt on weight basis -- 1.04 0.60 Matoms Na -- 1.04 1.04 H2 evolved in primary reaction (mm) 0.55 0.52 0o30 H2 evolved in secondary reaction with ammonia(l) 1.06 1.04 0.52 (based on Na present) (1) H2BNH2 + NH3 - HB(NH2)2 + H2 or 2(NH4)(BrBH2NH2) + 2Na - 2Na(BrBH2NH2) + NH3 + H2 in the stable compound [Cr(NH3)6](BH4)3 offers experimental contradiction to the above argument. Finally, sodium borohydride was crystallized from liquid NH3 with [H2B(NH3)2]Br and no hydrogen was evolved. Comparable experiments with NH4Br reSulted in rapid H2 evolution. The Precipitation of Hexammine Magnesium (II) Borohydride from a Liquid Ammonia Solution of the Diammoniate of Diborane —One of the simplest tests for the presence of a borohydride ion in the liquid ammonia solution of the diammoniate of diborane would be the precipitation of ammonia insoluble hexammine magnesium (II) borohydride. When the diammoniate of diborane was mixed with a liquid ammonia solution of magnesium thiocyanate, hexammine magnesium (II) borohydride,[Mg(NH3 )6 ] (BH4)2,precipitated. A comparable reaction between NaBH4 and Mg(SCN)2 gave a product whose x-ray powder pattern was identical to that obtained from the diammoniate of diborane. The equation for the precipitation is [Mg(NH3)6](SCN)2 + [H2B(NH3)2](BH4) li4 [Mg(NH3)61(BH4)2 + H2B(NH3)2(SCN) * (8) NH3 The process appeared to be complicated by a secondary reaction involving the thiocyanate and the borohydride in solution, since the solution darkened a great deal during the course of the experiment. WADC TR 56-318 24

Explanation of the foregoing results in terms of an ammonium model would require relatively rapid equilibrium shifts. The Hydrolysis of the Diammoniate of Diborane. —One of the observations made repeatedly in this laboratory and confirmed elsewhere7 was that hydrolysis of the diammoniate of diborane by 20% sulfuric acid at room temperature liberates approximately 2/3 of the hydridic hydrogen required by the formula B2H6 2NH3. Actual values for five separate experiments with temperatures up to 60~C show a hydridic hydrogen of 66.1 + 6%. High-temperature, sealed-tube reaction is required for complete hydrolysis and quantitative liberation of the boron and hydridic hydrogen. Such observations are consistent with a borohydride model inasmuch as 2/3 of the hydrogen (e.g., that present in borohydride) should be liberated easily while the remaining 1/3 (present in the complex cation) should hydrolyze with more difficulty. Difficult hydrolysis of the cation [H2B(NH3)f] was confirmed by studies onthebromide salt'[H2B(NH3.).)2"]Br Discussion.-The foregoing arguments offer support for a borohydride formulation of the diammoniate [H2B(NH3)2](BH4) as opposed to the currently accepted ammonium model, (NH4)(H3BNH2BH3). An objective examination of the question demands, however, that data of previous investigators, particularly the earlier evidence for the ammonium model, be examined carefully in light of currently available information. Five arguments were originally cited in support of the ammonium model.1 Each of. these (underlined below) can now be shown to be consistent with the borohydride representation. 1. The product (diammoniate of diborane) reacts with sodium in liquid ammonia at -770 to. produce one equivalent of hydrogen per mole of diborane involved; it thus seems to contain one ammonium ion per pair of boron atoms. It has been shown that oxidation of sodium ion in liquid ammonia solution with liberation of hydrogen is not a specific test for the ammonium ion. Tests' on the compound [H2B(NH3)2]Br show that the dihydrido-diammine boron (III) cation is likewise capable of reaction with sodium in an exactly analogous fashion. It is probable that the compound H2B(NH2)(NH3) formed in the reaction loses ammonia to give polymeric (H2BNH2). This explains the analytical results of Schlesinger and Burg on their product.. 2. A slow secondary reaction of the product with ammonia and sodium, yielding barely 40% more hydrogen, is easily explained by assuming that ammonia removes a BH3 group from the above structure (ammonium model) by a reversible reaction, producing (NH4)(BH3NH2). This explanation is supported by the fact that trimethylamine reacts with the diammoniate of diborane to give borine trimethylamine. Neither reaction is easily explained by other structures (i.e., structures other than the ammonium model). 7. W. H. Schechter, R. M. Adams, and C. B. Jackson, "Boron Hydrides and Related Compounds,t" Callery Chemical Co., Contract NO a(s) 10992 (1951), Callery, Pa. WADC TR 56-318 25

4,6 The question of excess hydrogen evolution is considered elsewhere' and is explained easily and quantitatively on the basis of the borohydride model. On the other hand, recent isolation of the compound H3BNH3,8 which presumably gives (NH4) (BH3NH2) in liquid ammonia, shows definitely that it does not exist in labile equilibriun with the diammoniate; the postulated equilibrium is untenable. The fact that trimethylamine reacts with the diammoniate of diborane to give low yields of H3BNR3 does not support the ammonium model, particularly when one realizes the supposed salt Na(H8BNHaBH3) gives no reaction with trimethylamine. It is known that NaBH4 and H2BNH2, the actual components of this mixture, do~, not react with N(CH3)3. On the other hand, the BH4- ion in the field of the acidic, polarizing cation H2B(NH3)t should react slowly with N(CH3)3 to give low yields of H3BNR3. Such an argument is eminently reasonable when one considers the effect of cation field strength on the reactions of borohydrideso Both beryllium and aluminum borohydrides react with trimethylamine. The borohydride of'the complex cation [H2B(NH3)+] would probably fall in between the beryllium and alkali metal borohydrides in terms of borohydride polarization. 3. The reaction of the new compound (CH3)20BH3 with ammonia and sodium produces the salt NaBH3NH2.:iThe negative ion of this salt is considered: to be an intermediate, step in the formation-.of'the abovestructure (ammonium structure ). This compound has no bearing on the structure since it is a substituted borohydride comparable to NaBH3(OCH3). It appears to result from NH3BH3 and sodium in liquid ammonia; it has been shown that 7H3BH3 is not in labile equilibrium with the diammoniate in anhydrous liquid ammonia,8 hence this salt has little relationship to the structure of B2H*2NH3. 4. The salt NaBH3NH2 strongly absorbs diborane. This fact justifies the assumption that the negative ion of the above structure is easily formed by the addition of BH3 to the BH3NH2- ion and is capable of existence. The reaction involved in this argument is not one of formation of (H3BNH2BH3S), but rather it appears to be similar to the acid-base reaction between diborane and NaBH30CH3:9 NaBH3(0CH3) + 1/2 B2H6 -. NaBH4 + BH2(OCH3) (9) NaBH3NH2 + 1/2 B2H6 --- NaBH4 + 1/n(BH2NH2)n. (10) it has been shown6 that the supposed salt Na(BH3NH2BH3) is in reality a mixture of NaBH4 and BHPNH2. There is absolutely no evidence at the present time for the ion (H3BNH2BH3-)o 8. S. G. Shore and R. W. Parry, this Report, p. 31. 9. H. I. Schlesinger and H. C.o Brown, et al., J. Am. Chem. Soc., 75, 186 (1953). WADC TR 56-318 26

5. The new compound B2H7N, having a structural skeleton B-NB, is easily prepared from the diammoniate of diborane. The linear structure, originally attributed to B2H7N, is now known to be incorrect. There is strong evidence10 for a bridge structure, /(I\H2K)\ H2B /BH2, which could easily result from the interaction of the complex cation [H2B(NH3)+] with BH3 groups on B2H6 molecules at the known high temperature (60'C) of the reaction. Alternatively, NH3 might be displaced at high temperatures to react directly with B2H6. The original argument is equally good for the borohydride formulation and in no sense supports the ammonium model. The new borohydride model also permits easy rationalization of earlier disturbing observations of Stock.11 He found that gaseous HC1 reacted with the diammoniate of diborane to yield hydrogen and diborane. Although these products were isolated in relatively large yield, they were considered as products of a side reaction since they were inconsistent with models for the diammoniate then available. The products are those expected from the borohydride model and have a formal resemblance to the reaction with the acid NH4Br in liquid ammonia; the reaction is directly analogous to the reaction of alkali metal borohydrides with gaseous HCl: 2[H2B(NH3)2](BH4) + 2HC1 —- B2H6T + 2H2 + 2[H2B(NH3)2]C1 2M (BH4) + 2HC1 -. B2H6t + 2H2 + 2M C1l All of the foregoing properties of the diammoniate offer strong support for a structure containing a borohydride ion. The known instability of ammonium borohydride as well as other chemical properties of the diammoniate argue strongly against inclusion of ammonium and borohydride ions in the same molecule. A proposed6 stabilizing effect due to H2BNH2 has little experimental support. On the other hand, the new cation [H2B(NH3)2], which can be considered as an ammonia complex of boron (III), explains all properties formerly attributed to the ammonium ion and is itself consistent in every detail with the chemistry of the diammoniateo Experimental. - a. Reagents. 1. Ammonia-The Matheson Co., Inc. — stored over and distilled from sodium before use. 10. K. Hedberg and A.o J. Stosick, J. Am. Chem. Soc., 74, 954 (1952). 11. A. Stock~ Hydrides of Boron and Silicon. Ithaca: CornellUniv. Press, 1933. WAIDC TR 56-318 27

2. Ammonium bromide —Baker and Adamson, reagent grade. 3. Sodium borohydride -Metal Hydrides, Inc.-recrystallized from liquid ammonia. 4. Diborane-prepared by reaction between boron trifluoride etherate and lithium aluminum hydride.12 5 Boron trifluoride etherate —Baker and Adamson, technical grade BF3etherate was distilled just before use and the middle fraction used. 6. Lithium aluminum hydride —Metal Hydrides, Inc. 7. Magnesium (II) thiocyanate. The following reactions were employed in synthesis:13 (a) MgCO3 + 2HI -—. MgI2 + H20 + C02 (b) AgN03 + NH4SCN -Ag(SCN), + NH4NO3 (c) 2AgSCN + MgI2- Mg(SCN)2 + 2AgII An excess of AgSCN in step (c) produced the soluble complex Mg[Ag(SCN)2a]2 The complex was broken and the silver removed from solution by saturating the system with H2S. Ag2S was filtered off and excess H2S removed by boiling. Thiocyanic acid remaining in the solution was neutralized with MgC03. The pertinent equations' are: (d) Mg[Ag(SCN)2]2 + H2S — A A2S + Mg(SCN)2 + 2HSCN (e) 2HSCN + MgC03 -— 4 Mg(SCN)2 + H20 + C02 The Mg(SCN)2 was crystallized from neutral solution by evaporating the water at 35~C in a current of air. The product -was recrystallized from distilled water and dried at about 50~C in the air. The air-dry product was very soluble in liquid ammonia; coordinated water around the Mg (II) cation was replaced by ammonia in the solution. The water in the system resulted in some H2 evolution when the Mg(SCN)2 solution was added to an ammonia solution of borohydride, but this side reaction did not interfere with the precipitation of Mg(NH3)6(BH4)2o Mg analysis: theory for Mg(SCN)2.5H20 = 10.45%; observed = 10o45-11.05% Mg. 8. The: diammoniate of diborane-The procedure of Schlesinger and Burg was used, except that known variations in the product could be introduced by removing excess ammonia at temperatures other than -78.5~C. Such details are included in experimental tables, 12. J. Shapiro, H. G. Weiss, MO Schmick, S. Skolnik, and G. B. Lo Smith, Jo Am. Chem, Sec., 74, 901 (1952). 13. D. R. Schultz, Ph.D. dissertation, Univ. of Mich., Ann Arbor, 19540 WADC TR 56-318 28

b. Reaction of the Diammoniate of Diborane with Sodium Borohydride or Ammonium Bromide, A mixture of the diammoniate of diborane and NaBH4 or NH4Br was dissolved in liquid ammonia and held, with continuous stirring, at low temperatures ranging from -78.5~ to -33. Hydrogen evolution was negligible. Solvent ammonia was then sublimed off and the solid mixture was allowed to warm to room temperature. In one case the mixture was warmed to 45~C. Hydrogen was removed with a Toepler pump; the residue was redissolved in ammonia; after stirring, ammonia was again removed, the system warmed up, and H2 measured. This cycle was repeated 3 to 5 times. c. Reaction of the Diammoniate of Diborane with Sodium in Liquid Ammonia. The reaction vessel was a 25-mm Pyrex tube with a 14/35 T joint sealed at right angles near the top of the tube. A side arm making about a 60~ angle with the lower side of the tube had a 24/40 ~ inner-outer joint combination so materials could be placed in the side arm. A glass bulb (about 15 mm in diameter) containing about 230 mg Na was placed in the side arm together with a magnet in a heavy glass case. After evacuation and drying of the system, the diammoniate was prepared in the conventional fashion; ammonia was removed to check the stoichiometry; ammonia was returned to the system and frozen; then the reaction tube was tipped around the 14/35 T joint until the sodium bulb fell into the reaction vessel. The magnetic hammer, magnetically controlled from the outside, was held back, then was used to break the sodium bulb; the ammonia was allowed to melt and the reaction conducted at -78.o5~C. The reaction between sodium and [H2B(NH3)2]Br was carried out in the same apparatus, using the same general technique. d. The Reaction Between B2H6-2NH3 or NaBH4 and Magnesium Thiocyanate in Liquid Ammonia. Attempts to scale-up the preparation of the "diammoniate" by increasing the amounts of reagents were generally unsuccessful. Accordingly, four batches of the compound were prepared (-78o5~C) and stored under dry nitrogen at -78. 5~C for from 4 to 48 hours. The combined sample (0.768 mmoles) was loaded in a dry box into one arm of an inverted Y-tube reactor; magnesium thiocyanate (2.5 mmoles) was placed in the other leg and the reactor was attached to a special vacuum-line filter assembly. Dry ammonia was then condensed in each leg and stirred; the solutions were mixed; then the precipitate was filtered off and washed eight times in the vacuum system with liquid ammonia. About.2 g of precipitate was obtained from the diammoniate. Analysis of the precipitate showed: Mg1.19 (NIH3)6.24 (BH4.07)2 The slightly high value for Mg indicated that some of the Mg(SCN)2 or some basic Mg (II) salt was contaminating the product, but the x-ray data left no doubt that the product was identical to the material obtained from NaBH4 and Mg(SCN)2 which gave an analysis indicating: Mg(NH3)5.7 B1.91 H8.o. The pattern also WADC TR 56-5318 29

checked with a sample of Mg(NH3)6(BH4)2 generously supplied by Callery Chemical Co. e. Analytical Methods. 1. Magnesium was precipitated from ammoniacal solution using 8-hydroxyquinoline l14 2. Boron was converted to boric acid and titrated with NaOH in the presence of mannitol, using a Beckman pH meter.l4 3. Nitrogen was determined, using a standard micro-Kjeldahl procedure. 40 Evolved hydrogen was always identified by gas density. f. X-ray Methods. A GE model XRD-l instrument was used with a camera of 57-mm diameter. Cu-Kx radiation was employed. Exposure times were one-half to one hour. 14. W Fo Hillebrand, G. E. F. Lundell, W. A. Brightj and J. J. Hoffman, Applied Inorganic Analysis. New York: John Wiley and Sons, Inc., 1953. WADC TR 56-318 30

C. CHEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (II), THE PREPARATION OF AMMONIA BORANE. (S, G. Shore and R. W. Parry) Abstract The momomeric compound H3NBH3 has been prepared from the ".diammoniate of diborane" and from lithium and sodium borohydrides, The reaction provides additional support for a borohydride formulation of the diammoniate, [H2B(NH3)21](BH4)-, and provides unequivocal evidence against all attempts to formulate the diammoniate as H3NBH3. The "diammoniate" and H3NBH3 do not exist in labile equilibrium. In a previous paper an alternative borohydride structure for the'diammoniate of diborane,"' [H2B(NH3)2]+(BH4r, was supported by chemical evidence. By making use of the fact that ammonium borohydride which has been prepared in liquid ammonia decomposes with evolution of hydrogen, Schultz and Parryl were able to prepare a bromide salt of the "diammoniate" cation. A chloride salt has been prepared in the same fashion. The reactions are: NH4C1 + [H2B(NH3)2]+(BH4)- [H2B(NH3)2] (Cl) + NH4BH4 (1) Solvent ammonia sublimed away. MN4BH4 + i4C(1 9above [H2B(NH3)2] (Cl)- + 2H2 Solvent added then removed again. By repeating these two steps alternately, the following overall reaction was observed: 2NH4C1 + [H2B(Nm)2]+(BH4 - -2[H2B(NI3)21+(Cl) + 2H2 The successive treatments with solvent ammonia were essential in order to bring ammonium and borohydride ions into reaction range so that the interaction between a protonic hydrogen of the ammonium ion and a hydridic hydrogen of the borohydride ion could occur. The resulting chloride salt proved to be a definite crystalline substance, As an alternative synthetic route to the: chloride salt, the heterogeneous, roomtemperature reaction between ammonium chloride and the "diammoniate of diborane" in a diethyl ether suspension seemed to offer interesting possibilities, especially since such a reaction would be less tedious to undertake than the ones cited above. The results were entirely unexpected, for the reaction in ether suspension led to the synthesis of the long-sought monomeric compound ammonia-borane, H3NBH3. The Reaction Between the "Diammoniate of Diborane" and Ammonium Chloride in an Ether Slurry. -Under strictly anhydrous conditions and in the absence of all''ktraneous proton sources the reaction in an ether slurry at about room temperature between ammonium chloride and the "diammoniate of diborane" was negligible. It was found, however, that by altering the ether environment through the addition of a small quantity 1 D. R., Schultz and R. Wo Parry, this Report, po 19 WADC TR 56-318 31

of anhydrous ammonia, the rate of reaction increased markedly. The rate was followed by measuring hydrogen evolution as a function of time. Results for one run are indicated in Fig.; 1. The dramatic break in the curve resulting after the addition of ammonia is proof of the efficacy of the catalyst. Data for several runs are summarized in Table IL It is significant that even a twofold excess of ammonium chloride in ether did not produce more than one mole of hydrogen per mole of the "diammoniate " After an appropriate amount of hydrogen had been produced, solutions were filtered and ether was distilled from the filtrate. The crystalline ether soluble material Isolated was analyzed and a correction was applied for the small amount of ammonia used to accelerate the reaction, The corrected analytical ratios were: hydridic hydrogen/boroon ntrogenoron Cboron = 0.0 Yields of 13BNHE3 ranged up to 80% of theory, An x-ray powder examination of the ether insoluble residues from typical ether runs indicated only a trace of the compound [H2B(NH3)2]+(Cl) which was expected from the equation [H2B(N T-3)21]+ +I 4Cl \ [H2B(NH3)21+(Cl) + H3NBH3 + H2 T'he principal crystalline material remaining in the -ether insolublale residue was ammonium chlorideo Since the reactants in the several experiments studied. were in one-to-one stoichiometric quantity, the absence of significant amounts of chloride salt plus the presence of ammonium chloride in the reaction residues suggest that ammonium chloride was consumed and then regenerated, perhaps through the decomposition of the chloride salto [H2B(NHi3)2]'+(C1) —- 4C1 + 1/n (H2BNH2)n The apparent instability of the chloride salt in the ether slurry is'consistent with an earlier observation that the sample of chloride salt prepared by Schultz'sl procedaure displayed some evidence of decomposition also (presence of some ammonium chloride). T3he bromide salt prepared by Schultz. on the other hand, appeared to be mohre stable since no ammonum b3romide was detect@ed in the sample. Such observations are cotsisteit with general acid-base thee.o-y (i.e., HCl, the weaker acid, forms a less stable salt than.fBr). The increasing stability of the phosphonium halides'llustrates this relationshipe P4C1 < PH4Br < PH4I The Reaction Between Lithium Borohydride and AQmonium Salts in an Ether Slurry:The foregoing process for the snmthesis of HsBSE3 from the "diammoniate of diborane" was interpreted in terms of the reaction of a borohydride ion. Such an interpretat;ion suggests that other boarohydrides should react siml1arly with NH4C1 in ether slurry. M3ABH4 X - M + E3BX M3 + 3B + 12 As a preliminary experiment, the reaction between sodium borohydride and ammonium WADC TR 56-318 32

1.2 1 1.00.8 I M -r 0.6 I z com 1 rHRun 3, table I 0.4 0.2 Smell amount of N H3 added 0 20 40 60 80 I 00 HOURS Fig. 1. The reaction between ammonium chloride and the "diammoniate of diborane" in an ether slurry. WADC TR 56-318 33

TABLE I uo TETHE REACTION BETWEEN THE "DIAMMONIATE OF DIBORANE" AND AMMCONIUM CHLORIDE SLURRIED IN ETHER co Mmoles iH4C 1 Ml Time-hr H2 (cumulative) [BR2(Ni,) i+-[BH2(I-BNH3)2]t(Bi4)' Ether (cumulative) [BH2(NH3)2]+(BH4)' 1 2., 02 1.86 5 0 3 mmole 22 0.25 NH3 added - > 40 1.17 2 2o.17 0.99 8 053 mmo le 12 0 55* NH3 added 32 0o.85 3 2.12 1.84 7 0 19 rmmole 12 0.07 NH3 added. 32 o.89 54 lo 04 99 1.o06 4 1.93 0.97 10 0o 15 mrmole 36 0.99* NFI3 added 48 1. 02 *Since the "diammoniate" was in excess, these values represent H2/NH'4C1.

chloride was studied in diethyl ether. Even when small amounts of ammonia were used to accelerate the gas evolution, the production of hydrogen was negligible. On the other hand E. A. Alton, Jr., of this laboratory was able to prepare H3NBH3 in small yields through the reaction between NIH4Br and NaBH4 in Ansul 141 polyether. The polyether is a solvent for NaBH4. It was also found that LiBH4, which dissolves in diethyl ether, will react relatively rapidly with a diethyl ether slurry of ammonium chloride or sulfate at about room temperature to produce H3NBH3. The purest product is obtained from ammonium sulfate and LiBH4. Schaeffer and Anderson2 have used a similar type of reaction to prepare (CH3)3NBH3. Results of several experiments are summarized in Table II. Although the theoretical yield of hydrogen was readily produced in each experiment with LiBH4, the yield of ether soluble ammonia-borane did not ever exceed 50% of theory. The "diammoniate of diborane" is empirically H3NBH3, but ether insoluble, hence it is a possible alternative product. Its reaction with NH4C1 is very slow in the ether containing no NH3o TABLE II THE REACTION BETWEEN LITHIUM BOROHYDRIDE AND AN AMMONIUM SALT SLURRIED IN ETHER Mmoles + Ml Time, % Yield Ether Run LiBH4 Me alt Ether hr H2/LiBH4 Soluble RH3BH3 1 2.08 2,10 ENH4C1 15 25 0.o99 33% 2 14.2 16.8 (NH4)2S04 (a) 50 6 (b) 40% 3 19.1 22.4 (a) 50 4.5 0o 98 44% 4 23.0 26.0 (a) 50 7 (b) 47% (a) Ammonium sulfate was used in preference to ammonium chloride in later experiments because the product lithium sulfate is even less ether soluble than lithium chloride.o (b) The extent of reaction was followed, qualitatively, by observing the rate of hydrogen evolution through a mercury bubbler. Analytical ratios for ether soluble product from Runs 1 and 2. N/B, 0.98, 0.98; H(hydridic)/B, 2.93, 2.98. Analysis of ether soluble product from Run 53 Calculated for H3NBH3: H(hydridic), 9,79; B, 35.0; N, 45.4. Found: H(hydridic), 9.73; B, 35.1; N, 45.6. The Properties of Ammonia-Borane. —Even though ammonia-borane and the "diammoniate of diborane" are both of empirical formula H3NBH3, their properties show striking con2, G. W. Schaeffer and E. R. Anderson, J, Am Chem., So. 7_,- 2143: ('1949). WADC TR 56-318 35

trasts. Illustrative data are summarized in Table III Molecular-weight determinations reported elsewhere3 indicated a monomeric formula for the compound, This has since been confirmed for the solid state also by two independent evaluations of xray powder patterns4 TABLE III COMPARISON OF THE PROPERTIES OF AMMONIA-BORANE WITH THOSE OF THE "DIAMMONIATE OF DIBORANE" H3NBH3 H[3NBH2IH3]+(BH4) 1. Crystalline material which gives a 1. Apparently amorphous material which definite x-ray powder pattern, gives no x-ray powder pattern. 2. Soluble in ether. 2. Insoluble in ether. 3. Molecular weight in ether and 3. Molecular weight in ammonia dioxane corresponds to above corresponds to above formula. formula. 4. Not readily hydrolyzed by distilled 4. Readily hydrolyzed by distilled water. water 5. Slowly splits out hydrogen at room 5. Stable to 80~. Splits out hydrogen temperature. very vevery slowly at room temperature. 6. Reacts with sodium in ammonia to 6. Reacts with sodium in ammonia to produce produce H31riHH3 [H3N1BH2M3 I+ (BH4 Amonnia-borane has a negative temperature coefficient of solublity in- diethyl ether from about -78~ to 25~C. Clear, rigorously anhydrous ether solutions are stable at room temperature to the extent that only a small amount of precipitate appears after standing for several days, The precipitate is probably the "diaamoniate of diborane" and some polymeric (H2NBH2)n, since trace quantities of hydrogen were liberated, In the presence of trace quantities of moisture, the ether solution becomes very unstable and solid material is precipitated quite rapidly. Ammonia-borane is soluble in and stable in anhydrous liquid ammonia. On the basis of x-ray powder pattern intensities there was no detectable conversion to the 35 So Go Shore and R. Wo Parry, J. Am. Chem. Soc., 77, 6084 (1955). 4. E. L. Lippert and W. No Lipscomb, ibid., 78, 503 (1956); E. W. Hughes, ibid., p. 50O WADC TR 56-318 36

"diammoniate" when a sample stood for 30 hours at -78~ and an additional 18 hours at -45~ in liquid ammonia. Reaction of ammonia-borane with sodium in liquid ammonia liberated one equivalent of hydrogen per mole of ammonia-borane. The stoichiometry of the process suggests the reaction H3NBH3 + Na - NaH2NBH3 + 1/2 H2 The sodium amidotrihydrido borate (III) has not been isolated in pure form. Solid ammonia-borane appears to undergo slow conversion to the "diammoniate of diborane" at room temperature. Very slow loss of hydrogen at room temperature occurs also. A sample which stands at 25CC in a dry atmosphere for a period of even five minutes will not redissolve completely in ether; the diammoniate and (H2NBH2)n are the most probable components of the precipitate. Solid ammoniaborane sublimes with difficulty under high vacuum at room temperature; the purity of the sublimate was not determined. The density of the solid, determined from x-ray powder data,4 is 0.74 g/cm3. Discussion. -Data summarized in this and other papersl support the borohydride formulation for the product produced in the reaction between ammonia and diborane [H2B(NH3)2]+(BH4)'. This reaction of diborane with ammonia differs from reactions with mono-, di-, and trimethyl amines in that the former process appears to involve nonsymmetrical cleavage of the hydrogen bridges in diborane, H. H H H NH3+ IH B + 2NH3 --— > B N Hj B EH H [E N3_ H H whereas the latter reactions involve symmetrical cleavage of the bridge to give monomeric products, H Hz H 7- \ + 2NH2R(etc.) 2[H3BNH2R] H H The conversion of the classical "diammoniate" to H3BNH3 through its reaction with NH4C1 makes untenable all earlier arguments6 which supported the hypothesis that the "diammoniate" has the structure H3NBIH3, but shows abnormally high molecular weights in liquid ammonia as a result of dipole association. Comparative properties of H3NBH3 and the "diammoniate" demonstrate unequivocally that two separate compounds are being considered. The formation of H3NBH3 from LiBH4 and from the "diammoniate" by comparable reactions adds strong support for a structure containing 5. R. W. Parry, G. Kodama, and D. R. Schultz, this Report, p. 82. 6. E. Wiberg, A. Bolz, P. Buchkeit, Z. anorg. Chem., 256, footnote 5, p. 287, and footnote 41, p. 301 (1948); Gmelin's Handbuch der anorganischen Chemie, 8. Auflage. System 13, 1954, pp. 100, 235. WADC TR 56 —318 37

the borohydride ion in the solid state, Ammonia-borane,: on the other hand, shows properties which are consistent with those of the methyl substituted amine-boranes. It now becomes of interest to examine the experimental procedures -of this investigation and compare them with earlier procedures which have- invariably led to the "diammoniate of diborane." Work of Schultz and Parryl and of Taylor' Schultz, and Emery7 indicated that.the product obtained,by the solid-phase decomposition of ammonium borohydride was the "dianmnoniate of diborane," rather than ammonia-borane, (H3NBH3). In an earlier study Schlesinger and Burg8 added ammnonia to solid (CI3)2 OBH3 in an attempt to form EsNBH3, but the "diammoniate." was produced instead, In the normal procedure for preparing the "diammoniate," solid ammonia picks up diboraneo All' th'e foregoing processes which produce the "diammoniate" have the reaction of a solid phase in the absence of solvent as a common factor. The principal feature which differentiates the process for preparing ammonia-borane is the presence of'ether during the reaction between an ammonium salt and a borohydride salto The role of a solvent in moderating or:altering the course of the reaction is not unusual. The- reaction of hydrazine and diborane9 and the reaction PC13 and LiAlH410'are both strongly dependent on the presence of.ether as a solvent. Although reaction takes place in the absence of solvents products differ from those obtained in ether. The role of ether in the present process is reasonably certain:, but its mode of action is still unknown. The facts can be interpreted in terms of the following hypothesis. If the decomposition of ammonium borohydride to the "diarmoniate of diborane" can be considered to take place through either of the following reaction schemes, 2 N44BH4 -.2 H23N1BH3 + 2 H2 2 13NBH3- BBH)2or:14:.-> EH3IBT3 + H2 H3NBH3 + MN4BH4 > [IBH2(:N3) 21 (B4) + 2 then the role of ether in an:ether slurry of an ammonium salt and a borohydride salt might be twofold. The ether might serve as a heat transfer medium preventing localized heating at reaction sites by dissipating the heat of the: initial exothermi re7. R. C. Taylor, De R.. Schultz, and A. Ro Emery, this Report, p. 92. 8. H, Io Schlesinger and A. B. Burg, J. Am. Chemr Soc., 60, 290 (1938). 9. Ho I. Schlesinger and M. J. Steindler, ibid.,~''_ 759 756-(1953). 10. J. T. Yoke III, G. Kodama, and R. Wo Parry, this Report, p, 162. WADC TR 56-518 38

action between protonic and hydridic hydrogen. Therefore, the possibility of ammoniaborane possessing the energy of activation required for conversion to the "diammoniate of diborane" is diminished. Secondly, ether is a solvent for ammonia-borane. Therefore, ammonia-borane is removed from the reaction sites as soon as it is formed and the possibility of participating in conversion to the "diammoniate" is dimished due to dilution. These reasons are consistent with the observation that the reaction of ammonium and borohydride salts in an ether slurry produces ammonia-borane, while the decomposition of solid ammonium borohydride produces the "diammoniate of diborane" exclusively. Experimental. - a, Materials. 1. Ammonia-commercial tank NH3 was dried and stored over sodium metal in the vacuum system. 2. Ammonium chloride and ammonium sulfate-reagent grade NH4C1 and (NH4 )2 S04 were dried at 80~ for two hours prior to their introduction into the vacuum system. 3. Diethyl ether-reagent grade (C2H5)20 was dried and stored over lithium aluminum hydride in the vacuum system. 4. Lithium borohydride and sodium borohydride- LiBH4 and NaBH4 (supplied by Metal Hydrides, Inc,) were purified by extraction with liquid ammonia at -65~ to -75~. The extractor was similar; to the vacuum-line filter of Schultz and Parryl and was operated in the same manner. The bulk of the solvent ammonia was removed by sublimation as the system was slowly warmed from -78~ to -40~o Remaining traces of ammonia were pumped away at room temperature. b. The Reaction Between the Diammoniate of Diborane and Ammonium Chloride in an Ether Slurry. The apparatus in which this work was carried out is depicted in Fig. 2. It was similar in appearance and operation to that described earlier except that the vacuum-line filtering device contained a drip-tip and the magnetically actuated stirrer was of spiral construction in order to insure efficient stirring. All the stopcocks and joints in the system which had to be rotated were greased with Dow-Corning High Vacuum Grease. All the joints which did not have to be rotated were sealed with DeKhotinsky cement. In a typical run,ll "diammoniate of diborane"12 was prepared and isolated in the reaction tube of the apparatus. The system was then filled with dry nitrogen and the tube was removed from the line; it was charged with a known 11. See Table I for the actual amounts of materials used in the individual runs. 12. S. G. Shore and R. W. Parry, this Report,p. 54. WADC TR 56-318 39

RECEIVER TUBE, 25 mm PYREX GLASS FR IT, MEDIUM POROSITY \ 24/40 DRIP-TIP TO MERCURY BUBBLER TO MAN IFOLD. 24/40 DRIP TIP'-///SOLENOID REACTOR TUBE, 25 mm PYREX Fig. 2. Apparatus for the reaction between ammonium chloride and the "diammoniate of diborane" in an ether slurry. WADC TR 56-318 40

amount of ammonium chloride and returned to its original position as rapidly as possible. The vacuum system was evacuated and ether was distilled into the reaction tube. A small quantity of anhydrous ammonia was distilled in also. Hydrogen evolution was initiated by maintaining the reaction tube at about room temperature and stirring its contents vigorously. In order to prevent the condensation of ether in other parts of the system, the temperature of the reactor tube was actually maintained at just below room temperature by immersing it in a beaker of water.13 The extent of reaction was determined at various intervals by quenching the reaction tube in liquid nitrogen and measuring volumetrically the quantity of hydrogen produced. Upon completion of the reaction the contents of the reactor tube were filtered and extracted with ether. The filtration and extraction were carried out at about -75~ in order to take advantage of the negative temperature coefficient of solubility of ammonia-borane. Crystalline ammonia-borane was obtained from the filtrate by distilling away the etheras the receiver tube was warmed from -70~ to -20~. c. The Reaction Between Lithium Borohydride and Ammonium Salts in an Ether Slurry. The reaction with amnonium chloride was carried out in the apparatus mentioned above. The reaction with ammonium sulfate was carried out on a tenfold larger scale and it was therefore necessary to modify the apparatus slightly. In this case the reactor consisted of a 100-nl, round-bottom flask which was fitted with a standard taper 24/40 drip-tip. This reactor flask was charged with a magnetic stirring bar and weighed quantities of lithium borohydride and ammonium sulfate14 in the protective atmosphere of the "dry box." The flask was then transferred to the vacuum system as rapidly as possible. After thoroughly evacuating the system, anhydrous ether was distilled into the flask. Hydrogen evolution was initiated by maintaining the flask at about room temperature and stirring its contents vigorously. The extent of reaction was followed either by measuring, at various intervals, the quantity of hydrogen produced or by qualitatively observing the rate of hydrogen evolution through a mercury bubler. After the theoretical amount of hydrogen had been produced, or after the rate of hydrogen evolution had decreased appreciably, the contents of the flask were filtered and extracted with ether at -75~. Crystalline ammonia-borane was isolated by distilling the bulk of the ether from the filtrate as it was slowly warmed from -70~ to -40~o. The remaining traces of ether were pumped from the solid as it was slowly warmed from -40~ to -20~. For further purification, samples of solid ammonia-borane were placed on the frit -of the vacuum-line filter and re-extracted with ether at -75~. d. The Reaction Between Ammonia-Borane and Sodium in Liquid Ammonia. Through the use of standard techniques,12 a 0.96-mmole sample of ammoniaborane was allowed to react with a large excess of sodium in about 5 ml of li13. All the reactions in this investigation which are described as carried out at room temperature were actually carried out -at slightly below room temperature through this technique. 14. See Table II for the actual amo unts of materials used in the individual runs. WADC TR 56-318 41

quid ammonia at -78'~ The reaction produced 0O96 H/HSNBH3 in 16 hours. Just a trace quantity of hydrogen was produced within the next 10 hours at -78o, e. Analytical. 1. X-ray powder analyses —Crystalline products were identified by means of x-ray powder techniques which used nickel filtered copper KC9 radiation. Specimens were sealed into thin-walled glass capillaries of 0.3-mm diameter (supplied by the Cain Specialties Co. of Chicago). A 5.7-cm Debye-Scherer camera was used for routine analyses. Exposures were of the order of 1-2 hours at 30 kv and 15 ma. For the characterization of ammonia-borane and the examination of complex patterns produced by reaction mixtures, a high-resolution, 11.4-cm Debye-Scherer camera was used. Exposures with this camera were of the order of 5-7 hours at 30 kv and 15 ma. 2. Chemical analyses-Boron was determined by titrating boric acid with sodium hydroxide in the presence of mannitol. Hydridic hydrogen was determined through the sealed-tube hydrolysis of the sample at 100~ followed by the volumetric determination of the hydrogen produced. Nitrogen was determined by the Kjeldahl method. The procedures for these analytical methods are discussed by Bissot o15 5. Molecular-weight determinations-The determination of the molecular weight of ammonia-borane in dioxane, by freezing-point depression, was carried out in a Beckmann-type freezing-point apparatus. The molecular-weight determination of ammonia-borane in ether by vapor-pressure depression, was carried out in equipment similar to that described by Parry Kodama, and Schultz,5 ex-:: cept that a mecury-filled differential manometer was used. The molecular-weight cells were thermosta;ted at 26~ and the temperature was held constant within + 0.05~. 15. D. H. Campbell, T. C. Bissot, and R, W. Parry, this Report, pa 145. WADC TR 56-318 42

D. CHEMICAL EVIDENCE FOR TIE STRUCTURE OF THE DIAMMONIATE OF DIBORANE (III). THE REACTIONS OF BOROHYDRIDE SALTS WITH LITHIUM HALIDES AMD ALUMINUM CHLORIDE (SO G. Shore and R. W. Parry) Abstract It is shown that LiBr, LiCi, and LiBH4 bring about slow evolution of H2 from a diethylether suspension of B2H6o2INH3. Results are interpreted in terms of polarization of the borohydride by the Li (I). A diethyl -ether solution of AlC13 will react with an ether suspension of BpH2 ~2NH3 to give hydride-halide interchange. B2Heo22NH3 will dissolve slowly in an ether solution of A1C13. One of the important pieces of evidence pertinent to the structure of the "diammoniate of diborane," B2H6e2NHI3, is the instability of ammonium borohydride. Such instability'justifies the reaction between the "diammoniate" and ammonium salts if a borohydride ion is assumed in the formula. By the same token, the previously reported lack of reaction between the "'diammoniate" and sodium borohydride argues against the presence of an ammonium ion in the "diammoniate."1 In connection with reactions of the latter type, an observation of Schaeffer2 to the effect that LiBH4 in ether reacted slowly with an ether slurry of "diammoniate" to give off hydrogen seemed to merit additional study. The Decomposition of the Diammoniate of Diborane in the Presence of Lithium Salts.0-One of the interesting features of the "diammoniate of diborane" is the fact that it is relatively stable to 80~, even though its most likely structure is [H2B (NH3)2]+(BH4)jo In the past, before Schultz and co-workers demonstrated that the "diammoniate" contains a borohydride ion,3 it was argued that "such a structure is unreasonable on the grounds that the strong proton-donor character of the quaternary nitrogen atoms should render the proton-sensitive BH4- incapable of existence.t Actually, in view of newly available evidence, the stability of such a structure can be justified, for hexamminechromium (III) borohydride, [Cr(NH3)6](BH4)3, is stable up to 600.1 Since such a compound is stable, the existence of [H2B(NH3)2]+(BH4)cannot be denied on the grounds cited above. Both the chromium salt and the "diammoniate" are stable because the cations Cr(NH3)*++ and H2B(NH3)+ are relatively large. They therefore have low charge densities and are of low-polarizing character. It is well known that the proton sensitivity of the borohydride ion increases 1. D. R. Schultz and R. W. Parry, this Report, p. 19. 2. G. W. Schaeffer, private communication. 3. D. R. Schultz and R. W. Parry, this Report, p. 19; Ro C. Taylor, D. R. Schultz, and Ao R. Emery, ibid., po 92. 4. A o B. Burg, J. Am. Chem. Soc., 69, 747 (1947). WADC TR 56-318 43

with the polarizing character of its associated cation. Thus, for example, potassium borohydride can be crystallized from the protonic solvent water; but lithium borohydride reacts vigorously with the protons of water; and aluminum borohydride reacts explosively with water. From arguments of this type one might rationalize hydrogen evolution from the ether slurry of the "diammoniate" and lithium borohydride on the basis of a reaction between the proton-sensitive borohydride ion in lithium borohydride and the weakly protonic hydrogens attached to the nitrogens in the dihydrido-~diammine boron (III) cation. In other words, the lithium (I) influenced the borohydride to such an extent that its hydridic hydrogens reacted with the.protons of the cation, e.g., Diethyl Ether l/n[H2B(NH3)2] (X)n + LiBH4 ----- l/n(Lin + H2 + +l/n(H2NBH2)n + H3NBH3 Room Temperature Xn- any anion. If the foregoing arguments are valid, the "diamxoiate of diborane" should evolve hydrogen in the presence of any lithium salt, since the presence of the Li (I) would increase the proton sensitivity of the "diammoniate's" borohydride ion. Diethyl Ether rH2B(iNH3)2I+(BH4Y) + LiX -— 4 l/n(Ha2BH2)n + H2 + H3NBH3 + LiX Room Temperature An experimental study proved to be in accord with this supposition. It was found that the addition of a lithium halide to an ether slurry of the "diammoniate of diborane" caused the decomposition of the "diammoniate." Of the several experiments carried out, the most significant were those in which it was first established that a particular sample of the:"diammoniate" is only slightly decomposed upon slurrying in ether at room temperature, but that upon addition of a lithium halide to the slurry, the rate of decomposition, evidenced by hydrogen evolution, is increased very markedly. The results of such experiments are summarized in Fig. 1. To avoid complications in the preceding study, it was important to exclude, rigorously, protonic substances such as water or ammonia. Extreme precautions were taken in the drying and handling of the lithium salts. That such precautions were successful in avoiding contamination is supported by the following facts1. Hydrogen evolution was slow compared to normal protolysis due to traces of water or ammonia. 2. Amonia-borane and hydrogen were produced in almost a one-to-one ratio as demanded by the equation above. If traces of water had been present, the ether soluble ammonia-borane would have formed an ether insoluble precipitate, thereby giving a much smaller yield of ammonia-borane.5 5. SO G. Shore and Ro Wo Parry, this Report, p. 31o WADC TR 56-318 44

0.8 IN _. 0.6. z I In all the experiments about 7 ml of ether, about 0.5 g of lithium halide, and about 2 mmoles of "diammoniate" 0.2 / were used. The lithium halides were LiBr added dried in an oven at 1500 for at least 20 hr and then heated in vacuo for an additional 12 hr. Li Cl added 0 20 40 60 80 100 HOURS Fig. 1. The decomposition of the "diammoniate of diborane" in the presence of a lithium halide.

Since the +dirammoniate+ is quite insoluble in ether, and since the lithium salts used are only slightly soluble in ether,'the rate of decomposition of the "diammoniate" is probably dependent in part on the surface areas of the salts used. It seems probable, however, that the somewhat greater catalytic activity of lithium bromide as compared to lithium chloride is related to the greater solubility in ether of the former compound. The Reactions of Borohydrides with Anhydrous Aluminum Chloride in Diethyl Ether.On the basis of the foregoing arguments and results it was thought that the strongly polarizing aluminum (III) ion of aluminum chloride would be more effective than the lithium (I) of a lithium halide in accelerating hydrogen evolution from [H2B(NH3)2] (RH4). However, contrary to the original expectation, the production of hydrogen was negligible when traces of water were scrupulously avoided. Results of two representative runs are sumnmarized in Table I. TABLE I THE REACTION BETWEEN THE T"DIAMMONIATE OF DIBORANE " AND ALUMINUM CHLORIDE IN ETHER Mmole s Mi Mmoles Time, H2 Run[ [H2B(NH3)2]+(BH4)j' Ether AiC3 hr [H2BON )2]B 1 1.93 10 ca. 6 0,3 0.09 12~ 0.11 2 2.88 5 ca. 10 0.3 0.01 - 36 0.05 In each run reaction took place as Soon as the aluminum chloride was, added to the ether slurry and within 20 minutes all the ether insoluble "diammoniate" had been converted into ether soluble compounds. The resulting clear solution proved to be unstable, for within eight hours Athe presence of a fine gelatinous precipitate became noticeable. The precipitate which was isolated by filtration proved to be amorphous (determined by x-ray powder techniques); it was polymerized aluminum hydride etherate. An x-ray powder examination of the solid obtained by removal of ether from the filtrate revealed the absence of familiar materials such as ammonium chloride, dihydrido-diammine boron (III) chloride [H2B(NH3)2a]+(C1)-, and ammonia-borane. It was possible, however, to sublime with difficulty from the residue a small amount of solid boron trichloride etherate. The eaxp-erimental facts strongly suggest that the borohydride ion of the "diammoniate of diborane" was destroyed through hydride-chloride interchange. If this were the case, a simpler'borohydride such as sodium borohydride might act similarly under comparable conditions. WADC TR 56-318 46

Diethyl Ether 3NaBH4 + 4AIC136 ) 3NaC1 + 4AiH3 + 3BC13 Room Temperature Schlesinger and co-workers7 have already established that the high-temperature reaction between solid sodium borohydride and dry aluminum chloride produces aluminum borohydride, but the experimental conditions of the present study were completely different. The more significant differences were (1) the use of an ether slurry, (2) reaction was carried out at room temperature, and (3) the rigorous exclusion of all traces of moisture from the starting materials and the reaction system. In a typical experimental test a five-to-one molar excess of aluminum chloride was added to an ether slurry of sodium borohydride at room temperature. The reaction mixture was stirred for about 38 hours to insure complete reaction; only a trace of hydrogen was evolved during this period. The precipitate from the reaction mixture contained sodium chloride and polymerized aluminum hydride etherate. The residue obtained by evaporating ether from the filtrate contained boron trichloride etherate, which was removed for identification by vacuum sublimation at room temperature. The results strongly favor the equation shown above. Discussion. -Although the lithium halides followed the reaction pattern which was expected from theoretical arguments, the reaction of aluminum chloride was anomalous in the sense that little hydrogen was produced; hydride-chloride exchange took place instead. The strong solvation of aluminum chloride probably plays an important role in determining the ultimate course of the reaction: Ether 3MBH4 + 4AC1l3 ---- 3MC1 + 3BC13 + 4A1H3 Room Temperature which simply parallels the well-known reaction between lithium aluminum hydride and aluminum chloride:8 Ether 3LiAlH4 + ALC13 - 3LiC1 + 4A1H3 Room Temperature Thus for the reaction involving the!?diammoniate of diborane" one would expect Ether 3[H2B(NH3)2]+(BH4)- + 4AlC13 — + 3[HB(NH3)2+(Cl)- + 3BCL3 + 4AlH3 Room Temperature 6. A1H3, AlCl3i and BC13 are actually etherates, but for the sake of simplicity the. ether is omitted from the formulas. 7. H. I. Schlesinger, H. C. Brown, and E. K. Hyde, J. Am. Chem. Soc.j 75, 209 (1953). 8. A. E. Finholt, A. C. Bond, Jr., and H. I. Schlesinger, ibid..> 69, 1199 (1947). WADC TR 56-318 47

However, there was no evidence for dihydrido-diammine boron (III) chloride or its decomposition product ammonium chloride. Apparently the [H2B(NH3)]+ cation underwent exchange also. One might expect such exchange to produce H3NBC13, which is unstable and undergoes decomposition to form boron amide, B(NH2)3,9 and ammonium chloride. But the absence of ammonium chloride from the reaction mixtures negates this.possibility. Therefore, the exchange reaction involving the cation of the "diammoniate" most likely involves the formation of boron trichloride etherate, as in the case of the borohydride ion. Finholt and co-workers8 have shown that Ether 4ALH3 + 3BC13 - 4 Al(BH4)3 + 3A1C13 Room Temperature What apparently happened in this investigation was a reversal of this reaction due to the use of excess aluminum chloride. The destruction of both the cation and anion of the "diammoniate of diborane" by hydride-chloride exchange to produce products which contained hydrogens of a less active nature accounts for the small amounts of hydrogen gas produced. Experimental. - a. Materials. 1. Diethyl ether and sodium borohydride —the purification of these substances is described in the preceding paper.5 2. Aluminum chloride —AlC13 is extremely hygroscopic and readily hydrolyzed. It was therefore purified and prepared for reaction in the following manner. Reagent-grade aluminum chloride was placed in the apparatus depicted in Fig. 2, which was then sealed off and evacuated. By pumping on the system continuously and applying a soft flame to the apparatus, pure aluminum chloride was sublimed past the thin glass-wool plugs and through the constriction into the u-trap, which was immersed in ice water. Then about 8 ml of anhydrous ether was, vacuum distilled into the u-trap, which was now cooled to about 78~. The stopcock was closed and the apparatus was removed from the vacuum system. The u-trap was slowly warmed to room temperature and through appropriate tipping of the apparatus the ether was allowed to dissolve small portions of aluminum chloride at a time. Care was taken to prevent the vigorous solution of aluminum chloride since this can result in ether cleavage. After the aluminum chloride was: completely dissolved, the apparatus was tilted so that a portion of the solution was poured into one of the bulbs. The solution in the bulb and the solution remaining in the u-trap were then frozen and the bulb was sealed off. In the same manner the remaining bulbs were filled and sealed off, one at a time. The amount of aluminum chloride in each of the bulbs was determined, approximately, by analyzing the contents of one of the bulbs. 9. A. Joannis, Compt. rend., 135, 1106 (1902). WADC TR 56-318 48

T [- 12/30 TO MANIFOLD SEALED AFTER INSERTION OF AIC13 SEALED AFTER SUBLIMATION OF Al Co3 INTO o TRAP Al C13 THIN "GLASSWOOL" PLUGS Fig. 2. Apparatus for the preparation of solutions of aluminum chloride in ether.

5. Lithium chloride and lithium bromide-LiCl and LiBr are very hygroscopic materials. They were therefore treated in the manner described in Fig. 1 in order to insure the complete removal of all traces of water. b, The Decomposition of the Diammoniate of Diborane in the Presence of a Lithium Halide The apparatus in which these experiments were carried out is depicted in Fig. 3. All the stopcocks: and joints which had to be rotated were greased with Dow-Corning. High Vacuum Grease. All the joints which did not have to be rotated were sealed with DeKhotinsky Cement. Initially, the side arm containing the - lithium halidel0 was placed'!in an oven and a standard taper cap was put in its place on the apparatus. The system was thoroughly evacuated and the "diammoniate of diborane" was prepared and isolated in the reactor tube, according to the standard procedure 11 Since the purpose of these experiments was to cause interaction between the protonic and hydridic hydrogens of the "diammniate,' great care was taken to remove all traces of aecesis ammonia from the system. After the apparent removal of all of the ammonia, the solid "diaammoniate" was pumped on for an additional'2 hours at room temperature Then dry nitrogen gas was flushed through the system at atmospheric pressure and the tube containing the lithium halide was transferred. while it was still warm, from the oven to its position on the apparatus, The system was then re-evacuated -and the halide was dried at 1000 under vacuum for 12 hours; anhydrous ether was- then vacuum distilled into the reactor tube. Through appropriate manipulation of the apparatus- the lithium halide was dropped upon the "diammoniate"'-ether slurry. The decomposition.was carried out under vigorous stirring at about room temperature,12 The rate of decomposition was followed by measuring the amount of hydrogen given off. After appreciable decomposition had taken place, the solution was: filtered at -75~0 Ether soluble ammonia-borane was isolated by vacuum distilling ether from the filtrate c. The Reactions of Borohydrides with Aluminum Chloride in Diethyl Ether. The apparatus in which these experiments were carried out is depicted in Fig. 4.`'Diaiumoniate" of diborane was prepared and isolated in the reactor tube,13 After the apparent removal of all- the ammonia, the "solid diammoniate" was pumped on for a an additional 12 hours at room temperature. Then anhydrous ether was vacuum distilled into the reactor. Sublimed aluminum chloride in ether solutio~n was- contained in the bulb suspended above the ether slurry. By rotating the arm of the bulb crusher, the bulb was crushed and its contents dripped into the ether slurry which was maintained at about room temperature. Immediiate reaction ensued and with vigorous stirring all the solid "diammoniate" disappeared within 20 minutes, Within eight hours of continuous stirring at 10. See Table I for the actual amounts of material used in. the individual runs. 11. St G. Shore and Ro W. Parry, this Report, po 54~ 12. See the preceding paper for experimental conditions, 13. See Table I for the actual amounts of material used in the individual runs WADC TR 56-518 50

TO VACUUM-LINE FI LTER i 24/40 DRIP-TIP SOLENOID 25-mm PYREX TUBE 14/35 DRIP-TIP LITHIUM HALIDE Fig. 3. Reactor tube for the decomposition of the "diammoniate of diborane" in the presence of a lithium halide. WADC TR 56-318 51

HEAVY- WALLED PYREX TUBE, 19ram. MANIFOLD IN DE N TAT ION AI C13 IN ETHER BULB-CRUSHER ARM /:: 29/42 DRIP-TIP L2 402X0 SOLENOID REACTOR TUBE, 25 mm PYREX Fig. 4. Apparatus for the reaction between the "diammoniate of diborane" and aluminum chloride in ether. WADC TR 56-318 52

room temperature, a fine gelatinous precipitate, polymerized aluminum hydride etherate, appeared. The reactor tube was cooled to about -70~ and the bulk of the ether was distilled away as the reactor was slowly warmed from -70 to -40~. The remaining traces of ether were pumped away as the system was warmed to room temperature. It was possible to sublime, with difficulty, small quantities of boron trichloride etherate in vacuo at room temperature into a receiver tube for identification. The system was then flushed with dry nitrogen, and the reactor tube containing the remaining residue after distillation and sublimation was transferred as rapidly as possible to a vacuum filter such as the one depicted in Fig. 2. The system was re-evacuated and 5 ml of anhydrous ether was vacuum distilled in upon the residue, most of which dissolved at room temperature. The slurry was filtered and the precipitate of polymerized aluminum hydride etherate was extracted with ether at room temperature and removed from the system for identification. The same procedure was employed for the reaction between sodium borohydride and aluminum chloride in ether, except that the sodium borohydride was added to the reactor tube in the protective atmosphere of the "dry box." The tube was then transferred to the vacuum system as rapidly as possible. d. Analytical. 1i. X-ray powder analyses-Reaction products were examined for the presence or absence of known crystalline materials such as ammonia-borane, ammonium chloride, dihydrido-diammine boron (III) chloride, and sodium chloride. The equipment and techniques are described elsewhere (p. 42). 2. Chemical analyses-The products aluminum hydride etherate and boron trichloride etherate which were produced in the reactions involving aluminum chloride did not lend themselves to physical analysis and, because only small quantities of material- could be isolated from the complex reaction mixtures, could not be identified by direct quantitative chemical analysis. It was necessary to rely on observations of the characteristic properties of these compounds. In this investigation it was shown that the fine gelatinous material which precipitated slowly from ether solution was amorphous. Qualitative analysis showed that it contained only aluminum and hydridic hydrogen. It is known that aluminum hydride etherate in ether solution slowly precipitates, forming a material of indefinite composition.8 The product which was sublimed in vacuo from the aluminum chloride reaction mixtures slowly split out ethyl chloride and melted at about 55~. Qualitative analysis showed that it contained boron and chlorine. It is known that boron trichloride etherate can be sublimed in vacuo at room temperature and that it slowly splits out ethyl chloride. It has a melting point of 560.14 14. E. Wiberg and W. Sutterlin, Z. anorg. Chem., 202, 22 (1931). WADC TR 56-318 53

E. CEEMICAL EVIDENCE FOR THE STRUCTURE OF THE DIAMMONIATE OF DIBOR ANE (IV). THE REACTION OF SODIUM WITH LEWIS ACIDS IN LIQUID AMMONIA (R. W. Parry and S. G. Shore) Abstract The reactions between sodium and- four Lewis acids in liquid ammonia have been studied. HCN, H3B03, CO2, and B2H6 have been considered. The complexity of the system increases in the order giveno The sensitivity of the B2H6-NH3 system to experimental details is interpreted by an extension of Werner's coordination theory. The ion [HB(NH[3)3]++ is postulated to explain observed facts. Introduction. —Past investigations have demonstrated that the stoichiometry of the reaction of the "diammoniate of diborane" with sodium in liquid ammonia is very sensitive to the conditions imposed. Only "diammoniate" which has been carefully prepared under a specific set of conditions will react with sodium in liquid ammonia to produce one equivalent of hydrogen per mole of diborane in a short period of time. A slow secondary process yields small quantities of additional hydrogen as a function of time.1 However, if the "diammoniate" has been mistreated, either through the rapid addition of diborane to ammonia or by maintaining carefully prepared "diammoniate" at the relatively high temperature of -40~, then the resulting material has been shown to react with sodium to produce up to 1.3-1.4 equivalents of hydrogen per mole of diborane in short period of time.l 2 In contrast to such observations, an early investigation in this laboratory,3 which seemingly followed the same careful procedure of 1 Schlesinger and Burg, could not obtain much more than one-half an equivalent of hydrogen per mole of diborane. Furthermore, it was observed that sodium was used up without further hydrogen evolution taking place. From these observations it is obvious that the. ammonia-diborane-sodium system is very complex. In order to obtain a better understanding of the system, it was thought that a study of the generalized reaction between sodium and Lewis acids in liquid ammonia was in order, starting with a simple reaction and working up to the complicated case involving diborane. The acids chosen in addition to diborane were hydrogen cyanide, boric acid, and carbon dioxide. Hydrogen Cyanide in Liquid Ammonia.-Hydrogen cyanide reacts with ammonia to form ammonium cyanide. A previous investigation has shown that ammonium cyanide reacts with an active metal in liquid ammonia to produce one equivalent of hydrogen per mole of cyanide. This simple reaction was confirmed. It was found that ammonium cyanide 1. H. I. Schlesinger and A. B. Burg, J. Am. Chem. Soc., 60, 290 (1938). 2. W. L. Jolly, Univ. of Calif. Radiation Laboratory, Livermore Site, Livermore, Calif., Contract No. W-7405-eng-48, UCRL-4504. 3. R. W. Parry, P. R. Girardot, et alo, Final Report, Chemistry of Boron Hydrides and Related Hydrides, Univ. of Mich., Eng. Res. Inst., Project M966, Uo S. Air Force, Contract AF 33(616)-8, EO0R. —464 Br-l, 1952. 4. H. H. Frank and C. Freitag, Z. angew. Chem., 39, 1430 (1926). WADC TR 56-318 54

and sodium in ammonia at -780 produce hydrogen gas according to the equation NH4CN + Na _>1/2 H2 + NaCN Boric Acid in Liquid Ammonia. -ITt was shown that the reaction between boric acid and liquid ammonia forms a tetraborate and in terms of acid hydrogens can be represented as 7sH3 + 4H3BO3-. (N+)2BO407 + 5NH140H In the light of the simple ammonium cyanide case and other well-known reactions of ammonium salts,5 it was expected that the ammonium tetraborate and ammonium hydroxide would react with sodium in liquid ammonia according to the following equations: 2Na + (NH1)2B407 - Na2BO07 + H2 5Na + NOH40H 5NaOH + 5/2 H2 Thus the total hydrogen evolved was expected to be in the ratio of 1.75 equivalents of hydrogen per mole of boric acid used (7H/4[H3BO3). The actual results, which are given below, were in good agreement with this value and offer strong support for the reaction scheme. Ammonia (ca. 5 ml) was condensed upon a sample of boric acid (1.61 mmoles) and the system was maintained at -52~ for two hours. Reaction with a large excess of sodium at -52~ produced 1,81 equivalents of hydrogen per mole of boric acid in 10 hours. A negligible quantity of hydrogen gas was produced within the next 10 hours at -52~, In order to avoid the possibility of secondary reactions, another experiment was carried out at a lower temperature. Ammonia (ca. 5 ml) was condensed upon boric acid (1.87 mmoles) and the system was maintained at -64~ for half an hour, Reaction with a large excess of sodium at -64~ produced 1.74 equivalents of hydrogen per mole of boric acid in 18 hours, Further hydrogen evolution was negligible. Carbon Dioxide in Liquid Ammonia,-Carbon dioxide reacts with ammonia to form ammonium carbamate, which has been shown to react with sodium ig liquid ammonia to give off one equivalent of hydrogen per mole of carbon dioxide. 0 0 N4I0CNH2 + Na 1/2 H2 + NaOCNH2 It was possible to reproduce this observation, but the system appears to be more complicated than was previously indicated, Therefore, the results are tabulated in Table I and are discussed further, In Run 1 ammonia in excess was added to solid carbon dioxide and allowed to melt. Then the excess ammonia was sublimed away at -78~, leaving behind solid ammonium car5. G, W. Watt, Chem. Revs,, 46, 317 (1950). 6. E. Rengade, Bull, Soc. Chim., Paris, 31 565 (1904).o WADC TR 56-318 55

TABLE I THE REACTION OF AMMONIUM CARBAMATE WITH SODIUM IN LIQUID AMMONIA Reaction Reaction Time H Evolved C02 Na Run ( a) Comments mmoles Temp. (cumulative) (cumulative) mmoles C02 ~CCO2,, hr c c02... 1 Excess Nss3 was 2 57 10-15 -78 3 1.06 sublimed away to -64 6 1.o06 isolate NH4C02NH2. Then the NH3 was returned.. 2 Excess NH3 was 1.03 10415 -78 1 0 19 never removed. -78 3 o0.48 -78 7 0 53 -78 17 o 58 -78 39 0.71 -78 51 o.86 -78 75 o091 3 Excess NH3 was 24.2 1 01 -78 24 0.35. never removed. -78 48 o060 -78 72 0.78 -78 96 0. 88 -78 120 o.89 -64 126 0.92, -64 132 0.95 -64 144 1.00 (a) All reactions were carried out in about 5 ml of solution. bamateo Upon returning the ammonia to the solid and introducing sodium metal, the reaction at -780 was in accordance with the equation above. One equivalent of hydrogen per mole of carbon dioxide was evolved. in less than three hours. Little evid-ence for further reaction was noted. In Run 2 ammonia in excess was added to frozen carbon dioxid.e and in Run 3 carbon dioxide was added to frozen.ammonia in excess. In both cases the excess ammonia was never removed, so the solid carbamate was never really isolated. and identified. In both cases abnormally long reaction times were required toproduce one equivalent of hydrogen per mole of carbon dioxide. Although armmonia and carbon dioxide are kniown to react readily with each other to form ammonium carbamate, Hughes and Soddy7 found that if the two substances 7. R. E. Hughes and F. S'oddy, Chemo News, 69, 138 (1894). WADC TR 56-318 56

are thoroughly dried, not the faintest sign of interaction occursp even after the gases have been in contact with each other for 24 hours at room temperature. Since the experiments were carried out under strictly anhydrous conditions, the very slow hydrogen evolution of Runs 2 and 3 suggests that little ammonium carbamate was present at the time that sodium was added to the ammonia solutions. Rengade,6 who first studied this system, also noted the very slow evolution of hydrogen from a sodium ammonia solution into which he introduced carbon dioxide under anhydrous conditions. He assumed that ammonium carbamate had formed but did not react since it is only slightly soluble in ammonia. However, his explanation seems to be unlikely since once the carbamate is isolated as the solid (Run 1), it reacts relatively rapidly with sodium in ammonia. Although the nature of carbon dioxide in ammonia is unknown, it probably is analogous to carbon dioxide in water. It has been reported that for the most part carbon dioxide is simply dissolved in water. According to Mills and Urey,8 only about 1% is hydrated. Furthermore, it is well known that the reaction between water and carbon dioxide is not instantaneous. The rate of hydration is dependent on catalysts which affect the pH of the solution. From available information, one may represent the formation of carbonic acid as 0 0 2H20 + O=C=O z H20O + H20 -" (H30)+ (OCOH) in which the slow step is considered to be hydration. Similarly then, in the absence of catalysts, carbon dioxide may be simply dissolved in liquid ammonia, with conversion to ammonium carbamate taking place very slowly. 0 0 2NH3 + O=C=O -N H3NCO + NH3 - (NfI4) (0 sNH2) The experimental observation that removal of excess ammonia from the carbon dioxide —ammonia system seems to facilitate ammonium carbamate formation (Run 1) might be attributed to the concentration of trace quantities of catalytic material such as water. Rengade6 also found it necessary to remove excess ammonia to be certain that the reaction with sodium would take place. Diborane in Ammonia. —In Table II typical experimental results of Schlesinger and Burg1 are compared with results from this laboratory3 for the study of the reaction between sodium and carefully prepared diammoniate of diborane in liquid ammonia. Careful examination of the experimental procedures revealed a subtle difference similar to that described for carbon dioxide in ammonia. This difference is believed to account for the difference in observations. Schlesinger and Burgl prepared and isolated the diammoniate through the slow low-temperature addition of diborane to ammonia, followed by the removal of excess ammonia by sublimation. The remaining solid was the diammoniate; it was shown to contain ammonia and diborane in a two-to-one mole ratio. Then the diammoniate was 8. G. A. Mills and H. C. Urey, J. Am. Chem. Soc., 62, 1019 (1940). WADC TR 56-318 57

TABLE II A COMPARISON OF THE RESULTS OF SCHLESINGER AND BURG WITH THOSE FROM THIS LABORATORY Reaction Reaction B2H6 Na Tempo H Evolved ~~~~Run -— ~Temp.Time mmoles B2H6 0C hr B2H6 1 (S B) 0 930 1.07 -77 0.5 0.98 2 (S-#B) 0.872 1.05 -77 0.5 0.99 1 2.3 1.0 -81 1.8 0.57 2 0.511 1.02 -80 to -50 2.7 0,64 redissolved and its reaction with sodium in ammonia produced one'equivalent of hydrogen per mole of diborane. In the other investigations diborane was added to ammonia under the conditions prescribed by Schlesinger and Burg. However, since the diborane was readily taken up by the ammonia, it was assumed that the diammoniate had been formed; therefore, the excess ammonia was not removed to confirm the formation of the diammoniate, but sodium was simply added to the solution. In this case only about one-half an equivalent of hydrogen per mole of diborane was produced, but nevertheless one equivalent of sodium was used up. The blue color of sodium in ammonia was discharged without further evolution of hydrogen. A detailed study confirmed that the difference in procedure is responsible for the difference in results. Dependence of Results on Experimental Procedure.-Whereas the ultimate product of the- carbon dioxide- ammonia interaction is ammonium carbamate, the diborane-ammonia reaction seems to be more complicatedo Many details of preparation must be closely observed. Therefore, before the results of this section can be considered, it is first necessary to describe the simple, but rather specific, procedure for preparing pure diammoniate of diboraneo A thin film of ammonia is condensed along the walls and bottom of a cylindrical reactor tube. A band of diborane is- then condensed from one to several inches above the ammonia and the entire reactor is slowly warmed from -1400 to about -80 in a period of about eight hours. During this time the diborane melts (-165.5~) and adds, usually in the gas phase at a temperature near -120~, to the surface of the solid ammonia. Addition is evidenced by a marked decrease in the pressure of the system. Since the addition depends on the surface available, the ammonia is in excess by at least a four-to-one mole ratio. When excess ammonia is removed by sublimation at -78~, the solid, nonvolatile diammoniate remains. WADC TR 56-518 58

As a preliminary experiment, diammoniate, prepared and isolated in the foregoing fashion, was redissolved in liquid ammonia (ca. 5 ml) and allowed to react with sodium (Na/B2H6 = 2) at -78~. In less than one hour one equivalent of hydrogen per mole of diborane had been produced. In a second study diborane was added to ammonia, but the excess ammonia was never removed to isolate the diammoniate. At the end of one hour upon reaction of the ammonia solution (ca. 5 ml) with sodium (Na/B2H6 = 1.7), only 0.58 of an equivalent of hydrogen per mole of diborane had been produced. The total amount of hydrogen which had been given off by the end of the second hour was 0.72 of an equivalent per mole of diborane. Thus both sets of observations reported in Table IITwere repeated.. In order to follow the evolution of hydrogen more carefully, the reactions were slowed down by using a 10-15 —fold excess of sodium.9'10 The new results are presented in Tables ITI and IV. Table III presents results equivalent to those of Schlesinger and Burg.l Diammoniate was prepared and isolated; then it was redissolved in ammonia and allowed to react with a large excess of sodium at -78~. One equivalent of hydrogen per mole of diborane was given off in about three hours. Table IV represents successful attempts to reproduce the earlier work in this laboratory.3 Diborane was added to ammonia in the above fashion, but the excess ammonia was never removed from the system. The reaction of such a system with sodium at -78~ produced less. than one equivalent of hydrogen per mole of diborane in the normal reaction time of about three hours. A much longer period of time was required to produce one equivalent of hydrogen per mole of diborane. Professor G. W. Schaeffer of St. Louis University was informed of these observations and was able to repeat them by using the same procedure.11 In addition to the fact that ammonia was never removed from the systems of Table IV, it was necessary to condense the diborane ring at least three inches above the ammonia in the preparative procedure and to age the system at -78~ for several hours after the diborane had been taken up by the ammonia. If all three conditions were not carefully imposed, the results were inconsistent, giving the stoichiometry of Schlesinger and Burg in some cases, and the. slow evolution Of this laboratory in others. All the above conditions which appear to be conducive to consistent results are designed to avoid localized heating effectso The effect of sudden warming was demonstrated in one particular experiment in which diborane had been apparently taken up by ammonia, but the temperature of the system had not been above -100~; upon suddenly thermostating this system at -78~, a vigorous reaction occurred during 9. The stability of a sodium-ammonia solution increases with increasing concentration of alkali metal,10 which indicates a decrease in the reducing rate of the dissolved metal. The slowing of the reaction by using a large excess of sodium is probably due to the properties of the concentrated solution and to a slightly different experimental procedure which the concentrated solution makes possible. 10. C. A. Kraus, J.o Amo Chem. Soc., 43, 749 (1921); W. C. Johnson and A. W Meyer, ibid., 54, 3624 (1932)o 11. Private communication. WADC TR 56-318 59

TABLE III TIHE REACTION OF THE DIAMMONIAIE OF DIBORANE WITH SODIUM IN LIQUTI) AEONIA Reaction Time H Evolved Run(a) Conmments B2H6 (cumulative) (cumulative) mmoles hr B2H6 1 In all cases NH3 was 1.69 1 0.73 removed to identify 2 0.89 [H2B(NH3)2]+ (BH4) 3 0.99 The NH3 was then re- 16 1.00 turned. 2 1.90 3 0.97 (a) In all cases Na and B2H6 were used in a 10-15 mole ratio in about 5 ml of solution at -78 ~. TABLE IV THE REACTION OF SODIUM WITH DIBORANE-AMMONIA SYSTEMS FROM WHICH AMMONIA WAS NEVER REMOVED Time System Was Reaction Time H Evolved Run(a) Height of B2H6 Aged at -78i B2 Reaction Time ve) Above NH3 in. After B2H6 Had Been mmoles (cumulative) (cumulative) Taken Up by NH3 B2h6 1 3 8 2.02 1 0.43 2 0.51 12 o.63 21 0.71 33 0.72 2 3 8 2.02 1 o.66 3 0.76 7 o082 17 0.90 41 0.96 65 ie1.00 3 6 6 1.21 1 0o71 3 90.82 7 0.87 20 0.93 44 0.399 68 1.00 4 5 8 1.40 1 0.72 3 0.80 7 0087 16 0.92 (a) In all cases- Na and B2H6 were used in a 10-15 mole ratio in about 5 ml of solution at -78~. WADC TR 56-318 60

which solid products were scattered throughout the reactor.'This observation plus the slow evolution of hydrogen in the reaction with sodium suggests that the mere taking up of diborane by ammonia, if heating effects are carefully avoided, does not necessarily result in the formation of the diammoniate of diborane. The condition that diborane be condensed several inches above the ammonia facilitates gas-phase addition of diborane and avoids or lessens the possibility of liquid diborane coming in contact with ammonia to cause localized heating. The aging process at -78~, once the diborane has been added to ammonia, maintains the sysn tem at just below its freezing point (-77o7~), where ammonia has a vapor pressure of about 45 mm. Under such circumstances the slow solvation or solution of adsorbed diborane can take place with the heat evolved being properly dissipated so that the possibility of bond rupture to form the diammoniate i's diminished. Once the system is properly aged in this manner, it can apparently become liquid without conversion to the diammoniate taking place. However, if the system is not aged, inconsistent results in the reaction with sodium are obtained. Discussion. —It is evident that the addition of diborane to ammonia through various procedures does not result in the formation of a unique substance. By considering the possible substances formed as coordination compounds in which ammonia is coordinated to boron (III), all available information can be correlated and interpreted in a reasonable fashion. In those studies which repeated Girardot8s original observations, it is believed that the slow, low-temperature addition of diborane to ammonia resulted in the ads sorption of diborane on the surface of the solid ammonia, without cleavage of the hyi drogen-bridged bonds. Evidence for this is offered in the observations of Schaeffer and co-workers,l2 who report that upon the addition of diborane to ammonia, if the system is warmed from -140~ to -80~ in four hours instead of eight hours as in this investigation, it is then necessary to age the system at 78~ for at least an addie tional two hours. Otherwise it is possible to distill small quantities of diborane from the system. Further evidence is offered in the previously described experiment from this investigation, in which after the initial low-temperature addition of diborane to ammonia the sudden raising of the temperature to -78~ caused a violent reaction to take place. After the diborane was taken up by the ammonia, it is believed that the aging process at -78~ permitted slow solvation or solution to occur without the rupture of both of the hydrogen-bridged bonds. The nature of diborane in ammonia is unknown (henceforth it will be referred to as a solution of diborane in ammon.a), but it may be simply dissolved in the ammonia, or one hydrogen-bridged bond may be ruptured and it may exist as H3B-H-BH2. The latter species may be stable Ns3 at low temperatures, The following reaction probably took place upon the addition of sodium.to the solution of diborane in ammonia: B2H6 + 2Na + NH3 — ) Na2BH3 + 1/x (3sNBHs3)n 12o G. W. Schaeffer, M, D. Adams, and F o J. Koenig, J. Am, Chemo Soco, 78, 725 (1956). WADC TR 56-318 61

H H3B BXH + 2Na > H3BNH3 + Na2BH3 NH3 The latter equation would suggest a one-to-one molar ratio for the products. The actual separation and identification of these products was not attempted because of the complexity of the reaction mixtures due to secondary reactions. However, this proposed reaction is analogous to the observation of Burg and Campbell13 that tetramethyl diborane, B2H2(CH3)4, is split symmetrically upon contact with sodium in liquid ammonia.. B2H2,(CH3)4 + 2Na + NH3 -4 Na2BH(CH3)2 + H3NBH2(CH3)2 This system is less prone to secondary reactions; it was possible for them to identify the products unequivocally. When the theoretical amount of sodium was used, it's characteristic blue color was completely discharged. The formation of disodium-borane, Na2BH3, accounts for a similar observation in this laboratory.3 In analogy to Bdimethyl ammonia-borane, H3NBH(CH3)2, the material which is designated (H NBz3)n should be monomeric H3NB3H3, ammonia-borane. Although such a compound exists,lf the possibility of the formation of some diammoniate, empirically (H3NBH3)2, must be considered also. The presence of either material accounts for the slow evolution of one equivalent of hydrogen per mole of diborane, which was noted in this investigation. If the diammoniate were formed according to the equation above, then it's reaction with additional hydrogen would produce only one-half an equivalent of hydrogen per mole of diborane. An additional hydrogen would come from slow secondary processes. If anmonia-borane were formed according to the equation above, its reaction with sodium would produce one equivalent of hydrogen per mole of dibor-ane; this reaction has been shown to be slower than the reaction between the diammoniate and sodium.l4 When excess ammonia is removed from a solution of diborane in ammonia, it can be said, very definitely, that the remaining solid will be the diammoniate of diborane. It appears as though removal of ammonia actually causes the conversion to the diammoniate. Of course, in preparing a solution of diborane in ammonia it is virtually impossible to avoid the formation of some diammoniate also, but upon removal of ammonia the diammoniate is present in quantitative or near quantitative yield relative to the amount of diborane used. This was determined throughout the course of this investigation, in which the diammoniate was prepared many times by the slow, low-temperature addition of diborane to ammonia, followed by the removal of excess ammonia. One might postulate that a solution of diborane in ammonia is actually H3B-H-NH2 NH3 in ammonia and that the removal of ammonia could cause conversion to the diammoniate. 135 A. B. Burg and Go Wo Campbell, Jr., J. Am. Chem. Sock, 74, 3744 (1952). 14. S. Go Shore and Ro W. Parry, this Report p. 31. WAlDC TR 56-318 62

The existence of such a species would be consistent with the fact that Burg and Campbell13 were able to obtain disodium-borane and B-dimethyl ammonia-borane in equimolar quantities from tetramethyl diborane. On the other hand, the role of solvent removal in the formation of ammonium carbamate from ammonia and carbon dioxide has already been noted and tentatively attributed to the concentration of trace quantities of a catalyst; water.or hydrolysis products are possibilities. A similar explanation might be invoked in the present situation, especially since traces of water, or hydrolysis products, cause ammoniaborane to precipitate from ether solution as (H3NBH3)n..15 It should be emphasized, however, that if such substances are truly responsible for the conversion, then only a very minute quantity is sufficient for catalysis, because the experiments with carbon dioxide and diborane were carried out under anhydrous conditions. For reasons previously cited,l5 the diammoniate of diborane is considered to be the coordination compound [H2B(NH3 )21+(BHE4)-o This formulation applies only to material which has been prepared through the slow,. low-temperature addition of diborane to ammonia, followed by the removal of excess ammonia. Its primary reaction with sodium in ammonia is represented as [H2B(NH3)2] +(BH4)- + Na -- 1/2 H2 + H2BNH2 + NaBH4 The slow secondary reaction1 in which additional hydrogen is given off was confirmed as a matter of course in this investigation. The source of the secondary hydrogen is the material of empirical composition H2NBH2, It probably consists of various substances, any one of which would react slowly with sodium, It is apparent that the handling of diammoniate under relatively severe conditions causes its conversion to another substance. Schlesinger and Burgl found that by maintaining the diammoniate at -4-0~ for 20 hours the final product had about the same empirical composition as the diammoniate, but that its properties were different. It retained ammonia more strongly than ordinary diammoniate; the solid split out hydrogen more readily than the diammoniate; and its reaction with sodium in liquid ammonia at -78~ produced around 1.3 equivalents of hydrogen per mole of diborane in a short period of time. Jolly2 noted similar results from a material he obtained by allowing liquid diborane to react with ammonia at -78~. An explanation of these observations can be given by assuming that under such relatively severe treatment ammonia replaces a third hydridic hydrogen to form the coordination compound [B(ONHI3)3]+2(BH4.)2- in which three ammonia molecules are coordinated to boron (III). Such a compound should retain ammonia more strongly than the diammoniate because the hydrogens of its ammonias are more acidic than those of the diammoniateo For the same reason it should split out hydrogen more readily than the diammoniate. The reaction of such a compound with sodium in ammonia can be represented as [HB(NI3)3]+2(BH4)2- + 2Na _> H2 + 2HB(NH2)2 + 2NE3 + 2NaBH4 15. See the preceding papers in this report. WADC TR 56-318 63

The hydrogen produced is equal to 1.33 equivalents per mole of diborane, Thus the expected properties are consistent with the experimental observations. Experimental. - a. Materials. 1. Ammonia-commercial tank 1H3 was dried and stored over sodium metal in the vacuum system. 2. Boric acid —reagent-grade H3B03 was dried by pumping on it for several hours in vacuo at room temperature. 3. Carbon dioxide-CO2 gas was obtained from Dry Ice. It was dried by passing it several times through a trap which was immersed in a CS2 slush bath (-1119~0). 4. Diborane —B2H6 was prepared by the reaction between LiAlH4 and excess BF3 etherate in ether solution.l6 It was purified at low-temperature trap-totrap distillation at -145~ until its vapor pressure was 225 mm at -111,.06 5. Sodium-sealed, evacuated, thin-walled glass bulbs 1.5-2 cm in diameter and containing 0.5-0.75 g of sodium metal were prepared by standard techniques.17 b. Procedures. All the reactions were carried out in a bulb crusher such as the one depicted in Fig. 1. In a typical study a thin-walled, evacuated, glass bulb containing sodium metal was placed in the crusher, which was then put in place on the vacuum line and evacuated. The materialsto be studied were then prepared in the reaction tube of the bulb crusher as follows: 1. Carbon dioxide in ammonia-A known amount of carbon dioxide was condensed an inch or so above an excess of frozen ammonia which was at the bottom of the tube, and the system was warmed Prom about -140l to -80~ in a period of about eight hours. In other experiments the positions of the ammonia and carbon dioxide were reversed. The excess ammonia was sublimed away at -.78~ when the -formation of ammonium carbamate was desired. See Table I for the actual amounts of reactants which were used in the individual runs, 2. Boric acid in ammonia-A known amount of finely divided boric acid was placed in a glass bucket which was lowered on a thread into the bulb crusher, When the bucket was on the bottom of the reactor tube of the bulb crusher, it was tipped by pulling a second string which was fastened to its bottom. The boric acid was spilled onto the bottom of the reactor tube and the bucket was 16. i. Shapiro, TI. G. Weiss, M. Scbmick, S. Skolnik, and G. i. L. Smith, J. Am, C:iem Soc., 74, 901 (1952). 17. R. S. Sandersono Vacuum Manipulation of Volatile Compounds. New York: John Wiley and Sons, Inc., 1948, p. 72. WADC TR 56-318 64

ARM INDENTATION HEAVY-WALLED TUBING INDENTATION T o IN TUBING co SODIUM-FILLED BULB 24 REACTION TU B E TOP VI EW TO MANIFOLD 250 mm 5mrm TUBING REACTION TUBE SIDE VIEW 220 mm Fig. 1. Bulb crusher for reactions with sodium in liquid ammonia.

withdrawn. Then the bulb crusher was loaded with a sodium-containing bulb and was transferred to the vacuum system -and evacuated. Ammonia was then condensed upon the boric acid. See page 55 for the conditions of the individual runs. It was shown that the action of ammonia on boric acid produces a tetraborate. Liquid ammonia was condensed upon a small quantity of finely divided boric acid. Then the excess ammonia was sublimed away at -78o~ Final traces of volatile material were removed by pumping on the remaining solid for about an hour at room temperature. The nonvolatile solid was analyzed and the percentage composition of nitrogen and boron indicated that the final residue was NH4HB407.5/2 H2( Anal. Calcd. for INHB407 ~ 5/2 20 e N, 6~39; B, 19.73. Found: N, 69.40; B, 19.6. In terms of acid hydrogens, the reaction between boric acid and ammonia can be represented as 7NH3 + 4H3B03 ~- 4 (iNH )2B407 + 5NH440H 3. ]Diborane in ammonia —A thin film of ammonia, about two inches high, was condensed along the walls and. bottom of the reactor tube of the bulb crusher. Diborane was then condensed in a solid ring at least two inches above the ammonia, which was in excess by more than a four-to-one.mole ratio. The reactor was then immersed in a Dewar flask containing Phillips commercial methyl pentanes which had been previously cooled to 140~o The bath was allowed to warm at the rate of about 5~ per hour. During this period of warming, the vapor pressure of the diborane in the system rose and then fell off sharply at some point in the temperature range from -125~ to -115~, indicating that the diborane had been taken up by the ammonia.'When the temperature of the system had reached -80~, the ammonia was allowed to melt and rum down the walls to the bottom of the reactor tube. The reactor tube was then thermostated in a Dry Ice-isopropanol slush at -78~ and all the excess ammonia was sublimed away, leaving behind solid diarmmoniate of diboraneo In most cases the exact amount of ammonia which was put in and removed from the reactor tube'was known. In each such case it was found that ammonia a~nd diborane had reacted in a two-toone mole ratio. The -ammonia was then returned to the system by condensing ammonia, in rings, along the wall of the reactor- tube and then allowing it'to melt, thus washing any solid diammoniate on. the walls down to the bottom of the tube. The preparation of a solution of diborane in ammonia was carried out in essentiaily the same manner, except that the diborane was condensed at least three inches above the ammonia, and after the addition of diborane to ammonia the system was aged at -780 for an additional 5-8 hours. The excess ammonia was never removed from the system. See Tables III and IV for the actual amounts of reactants which were'used. in the individual runs. After the material to be studied was prepared in the reactor tube and a sufficient quantity of ammonia was present, the ammonia solution was frozen and WADC TR 563518 66

the sodium-containing bulb was crushed by rotating the armlof the bulb crusher. The crushed bulb was dropped onto the frozen solution by rotating the crusher around its axis. Sufficient ammonia was then condensed upon the crushed bulb so that the total volume of solution when the ammonia melted would be about 5 ml. When a large excess of sodium was used, the reaction was initiated by warming the system to a point where a portion of the ammonia melted and dissolved the sodium. The system was then thermostated in a Dry Ice-isopropanol slush at -78~.18 The sodium-ammonia solution melted the remaining frozen ammonia and thus it was possible to bring about solution without raising the temperature of the system significantly above -780 for any appreciable perioda of time. When a small amount of sodium (stoichiometric quantity or small excess) was used, it was generally necessary to obtain complete solution before thermostating the system at -78~; otherwise the regaining frozen ammonia would not melt because the quantity of sodium in the system was insufficient to cause an appreciable freezing point depression for the ammonia. This difference in procedure may account in part for the observation that when a large excess of sodium i-s used (p. 59) the reaction time is greater than when an equivalent amount.of sodium is used, since in the latter case the system has been subjected to somewhat higher temperatures and localized heating effects. At any rate, a consistent set of results is obtained from either procedure. 18. This does not apply to the reactions with boric acid in ammonia which were carried out at a higher temperature (p. 55). WADC TR 56-318 67

F. A TRACER STUDY OF THE REACTION BETWEEN SODIUM AND THE DIAMMONIATE OF DIBORANE (S. G. Shore, P. R. Girardot, and R. W. Parry) Abstract A tracer study of the reaction between sodium and the "diammoniate of diborane" in liquid ammonia, using deuterium as the labeling element showed that in the low-temperature reactions with sodium only nitrogen-hydrogen bonds were broken; no boron-hydrogen bonds were ruptured. In the high-temperature ammolysis reaction involving the residues of the sodium reaction, a boron-hydrogen and a nitrogen-hydrogen bond were broken. A rather high separation factor for the isotopes of hydrogen is suggested by the data. Introduction. —It appears that the so-called "diammoniate of diborane," [H2B(NH3)2]+(BH4)- is one of several possible coordination compounds of boron (III) in which the hydride ion and the ammonia molecule act as coordinated ligands. The existence of ammonia-borane, H3NBH3, is well established.1 Furthermore, evidence has been presented2 that the addition of diborane of ammonia through various procedures does not result in the formation of a unique substance. It has been pointed out2 that the formation of the "diammoniate" is guaranteed only by a very specific preparative method. Under slightly different conditions it appears to be possible to prepare a solution of diborane in ammonia. Under more vigorous conditions there is evidence for the formation of [HB(NH3)3]+2(BH4)-2. In an exchange study of the "diammoniate" in ammonia, using deuterium as a tracer, Burg3 demonstrated that H-D exchange between the solute and solvent occurs only between those hydrogens which were originally bound to nitrogen. His results indicate that the hydrogens which are bound to boron in the "diammoniate" are not sufficiently acidic to form an ammonium salt, since it is known4 that rapid proton exchange takes place between an ammonium ion and ammonia, even at -60~. Burg's work refutes the early diammonium formulation, (NH4)2B2H4, of Stock.5 Insofar as exchange experiments can be accepted as a criterion for acidic character, Burg's evidence is convincing; however, his reaction conditions were mild. A more severe test for acidic character would be to determine whether boron-hydrogen bonds, nitrogen-hydrogen bonds, or bonds of both types are ruptured in the reaction i. S. G. Shore and R. W. Parry, this Report, p. 31; E. W. Hughes, J. Am. Chem. Soc.,, 78, 502 (1956); E. L. Lippert and W. N. Lipscomb, ibid., 503 (1956). 2. R. W. Parry and S. G. Shore, this Report, p. 54. 3. A. B. Burg, J. Am. Chem. Soc., 69, 747 (1947). 4. C. J. Nyman, Si-Chang Fung, and H. W. Dodgen, ibid., 78, 725 (1956). 5. A. Stock. Hydrides of Boron and Silicon.. Ithaca: Cornell Univ. Press, 1933. WADC TR 56-318 68

of the."diammoniate" with sodium in liquid ammonia. Since a knowledge of the acidic character of the hydrogens in the "diammoniate of diborane" and its related substances is of importance in testing proposed models of their structures, tracer studies of their reactions with sodium in liquid ammonia were undertaken. A Tracer Study of the Reaction of the "Diammoniate of Diborane" with Sodium in Liquid Ammnonia.-Since Burg3 has demonstrated that hydrogen which is bound to boron does not exchange with hydrogen which is attached to nitrogen, the following species of the "diammoniate" may be regarded as being stable in liquid ammonia-h3 or liquid ammonia-d3: B2De 2NH3, B2D6' 2ND3, and B2H6 2ND3. 6 When the classical "diammoniate of diborane" has been prepared under carefully prescribed conditions, and when it has been always maintained below -700, it will react with sodium in liquid ammonia to produce one equivalent of hydrogen per mole of "diammoniate." 2 The species which are listed above were allowed to react with sodium in liquid ammonia; the results are summarized in Table I. The complete absence of deuterium7 in the gas evolved from the reaction between sodium and B2De.2NH3 in ammonia-h3 (Run 2) confirms Burg's contention that the hydrogens which are bound to boron in the "diammoniate" have no acidic character. Even under the comparatively vigorous conditions of this reaction, no boron-hydrogen bonds were broken. However, the reaction between B2H6 2ND3 and sodium in ammoniad3 (Run 3) produced gas which was only 77% deuterated. This result could not have arisen from contamination of the system because of the elaborate experimental procedures which were employed. Therefore, the possibility of B-H bond rupture in this particular reaction was considered. If the protium which was given off in this reaction did come about through the rupture of B-H bonds, then one would expect B2D6' 2ND3 to give off gas which is much richer in deuterium. However, the gas which was produced in the reaction between B2D6e-2ND3 and sodium in ammonia-d3 was about 78% deuterated (average of Runs 4 and 5). Since the diborane-d6 which was used in this latter case was about 96% deuterated, and since the hydrogen which is bound to boron does not exchange with hydrogen which is bound to nitrogen, 3 there was insufficient protium in the form of B-H bonds to account for the composition of the gas. There6. For the sake of simplicity, isotopically substituted "diammoniate" will be represented by its empirical formula. 7. In discussing the tracer studies, the term hydrogen will refer to the elements of atomic number one. Protium and deuterium will refer to its isotopes of mass one and two, respectively. 8. In order to be certain that any protium enrichment in the evolved gas did not arise from the contamination of the system, control experiments were carried out concurrently with the regular tracer studies under identical conditions of handling and analysis. The control experiments consisted of a tracer study of the reaction of sodium with ammonia-d3 in the presence of a catalyst. Na + ND3 Fe23-PtO2 1/2 D2 + NaND2 There was little, if any isotopic fractionation. The vapor pressure of the ammonia-d3 indicated that itwas about 97% deuterated, and the samples of gas which were obtained from tio separate control experiments were 95.9 and 95.1% deuterated. WADC TR 56-318 69

TABLE I A TRACER STUDY OF THE REACTION OF THE "DIAMMONIATE OF DIBORANE'" WITH,.SODIUI.. IN LIQUID. AMMONIA Reaction Time, Hydrogen Equivalents Isotopic Diborane, Run Solvent Solute (cumulative) Evolved (cumulative) Analysis, mmoles hr diammoniate, immoles Total % D 1 Ns3 B2H6 2NH3 1.69 1 0.73 0.0 2 o.89 0.0 3 0.99 0.0 2.NH3 B2D6 2NH3 1.87 1 0.70 0. 0 2 0.77 0.0 6 1.00 0.0 3 ND3 B2H6' 2ND3 1.92 2 0.77 74.3 4 0.91 77.1 8 1.00 14 1.02 4 ND3 B2D6 ~ 2ND3 1.95 1 0.51 72.5 2 O.56 73.4 4 0.62 75.7 20 o.86 80.3 44 o.98 81 5 ND$ B2D6 2ND3 2.07 2 0.51 69.6 5 0.57 72.3 27 0.79 74.9 33 0.83 75.1 56 0.92 75.7 6 ND3 B2D6 ~ 2ND3 1.91 1 0.62 2 0.71 4 0.75 16 0.88 36 1.00 Notes:.' (a) In all cases sodium and diborane were used in a 10-15 mole ratio in about 5 ml of solution at -78~. The concentration of the "diammoniate" was about 0,01 M. (b) In an attempt to decrease the reaction time, solid B2D6'2ND3 was aged for 22 hr at -78~~ (c) The diborane-d6 was about 98% deuterated and the ammonia-ds was about 96% deuterated. WADC TR 56-318 70

fore, the most reasonable interpretation of these observations is that the "diammoniate of borane" reacts with sodium through the rupture of nitrogen-hydrogen bonds only. Since the ammonia-d3 was about 96% deuterated, it contained N-H as well as N-D bonds. The fact that the evolved gas from B2H6 2ND3 and B2D6-2ND3 was greatly enriched in protium is attributed to an isotope effect in which the N-H bond was preferentially ruptured. In view of the foregoing discussion one might expect'"diammoniate" which contains ammonia-d3 to react more slowly with sodium than "diammoniate" which contains ammonia-h3. The results are roughly consistent with this expectation. In fact, the reactions of B2D622ND3 with sodium were much slower than any of the other reactions, even when a deliberate attempt was made to increase the rate of hydrogen evolution (Footnote b, Table I). The reason for the much slower rate is not immediately obvious since differences are greater than expected from theory, but the observations appear to be real. A Tracer Study of the Reaction of Sodium with Solutions of Diborane in Ammonia.It has been pointed out7 that under certain conditions it appears to be possible to prepare a solution of diborane in ammonia, which may be diborane simply dissolved in ammonia, or may be H3B-H-BH2 NiH3 This solution produces hydrogen upon reaction with sodium, but much more slowly than the "diammoniate." Tracer studies of the reaction with sodium were carried out and the results are presented in Table II. It was found that the reaction between sodium and a solution of diborane-d6.in ammonia-h3 produced essentially pure protium (Run 3), but the reaction between sodium and diborane-h6 in ammonia-h3 produced hydrogen gas which was about 74% deuterated (average of Runs 4, 5, and 6). These observations, with respect to the isotopic composition of the gas evolved, are in complete accord with the tracer studies of comparably deuterated "diammoniate of diborane," and indicate that the rupture of nitrogen-hydrogen bonds is the only source of hydrogen in the reaction. These results then are consistent with the previously Stated belief2 that the source of hydrogen in the reaction is either the "diammoniate" or. ammonia-borane, both.-of which are similar in that they contain coordinated ammonia. A Tracer Study of the Reaction of "Mistreated Diammoniate of Diborane"' with Sodium in Liquid Ammonia.-'Mistreated diammoniate of diborane" is a substance which'has the same empirical composition as the "diammoniate of diborane" but is prepared -either through the rapid addition of diborane to ammonia or by maintaining carefully prepared "diammoniate of diborane" at temperatures above -70~ for several hours or more.2 The resulting'product is believed to be [HB(NHs )3]+2(BH4)2-, since it reacts with sodium in liquid.ammonia to produce around 1.3 equivalents of hydrogen per mole of diborane used in its preparation. This model suggests that only nitrogen-hydrogen bonds should be broken in the reaction with sodium, and the hydrogen which is given off in excess of one equivalent per mole of diborane should not come from.the interaction of protonic and hydridic hydrogen, but rather from the direct sodium-cation interaction. WADC TR 56-3518 71

TABLE II A TRACER STUDY OF THE REACTION OF SODIUM WITH SOLUTIONS -OF DIBORANE IN AMMONIA Reaction Hydrogen Equivalents Isotopic Run Solvent(a) Solute Diborane, Sodium Reaction Time, Solute Temp.,Evolved mmoles - Diborane hr mC diborane, mmoles total % D 1 NH3 B2H6 0o51 1.0 1.8 -81 0.57 0.0 2 NH3 B2H6 2.3 1.02 2.7 -80to -50 0.64 0.0 3 NI3 B2D6.(b) 1.55 1.00 o.83 -811 0.59 0.25 4'NrD3(b) B2H6 0.97 1.01 1.7 -79 0.50 70.0 <- 5 ND3 B2H6 1.57 0.92 0.67 -79 0.59 76.3 P0 6 ND3 B2H6 1.96 10-15 4 -78 0.55 74.6 4.5 weeks(c) -78 1.04(d) 75.3(d) (a) About 5 ml of solution was used in all the reactions. (b) The ammonia-d3 was about 96% deuterated and the diborane-d6 was about 98% deuterated. (c) An interval somewhere between 1 and 4-1/2 weeks may be adequate; 4-1/2 weeks was simply a matter of experimental convenience. (d) These values represent the total amount of gas given off and the deuterium content of that gas.

The results of this investigation plus the more recent observations of Jolly9 are summarized in Table III. In agreement with the model which is proposed for the "mistreated diammoniate,"' the results show that all the hydrogen which is given off in the reaction with sodium, even that which is in excess of one equivalent per mole of diborane, arises from the rupture of nitrogen-hydrogen bonds. Ammonolysis Reactions. —Schaeffer, Adams, and Koenig10 report that the nonvolatile products from the reaction of the "diammoniate of diborane" with sodium in liquid ammonia are sodium borohydride and polymeric aminoborine, (H2NBH2)n. They observed that this residue reacts with ammonia at temperature above -78~; hydrogen gas is produced. This reaction was noted also in this laboratory. After the reaction with sodium had taken place, all the solvent ammonia was sublimed away at -78~ and the reactor was warmed to room temperature. Only small quantities of hydrogen were split out from the solid residue; however, when gaseous ammonia was introduced above this solid, hydrogen evolution at room temperature was rapid. The maximum amount of hydrogen which was obtained from the reaction with ammonia was 3.10 equivalents per mole of diborane, or a total of 4.10 equivalents including the quantity of gas which was given off in the reaction with sodium. A tracer study of the reaction between ammonia and the residue from the "diammoniate'"-sodium reaction demonstrated, unequivocally, that hydrogen is produced through the interaction of acidic hydrogen which is bound to nitrogen and hydridic hydrogen which is bound to boron. The results are summarized in Table IV. It should be noted in Run 1 that ID from N-D bonds reacted with H from B-H bonds to form HD which was identified through the use of a mass spectrometer. Furthermore, it is of interest to point out that the residue from the "mistreated diammoniatetsodium reaction undergoes ammonolysis also. The hydrogen which was given off in Run 3 was 50%4 deuterated,11 presumably ED. Discussion. —The results of this investigation are in accord with the models which have been proposed for the "diammoniate of diborane" and "mistreated diammoniate of diborane." The reactions with sodium in liquid ammonia take place through the rupture of nitrogen-hydrogen bonds only, and may be represented as2 [H2B(NH3)2]+(BH4)- + Na - 1/2 H2 + NaBH4 + H2NBH2 + NH3 [HB(NH3)3]+2(BH4)2- + 2Na ) H2 + 2NaBH4 + (H2N)2BH + 2NH3 The production of hydrogen in the reaction of sodium with a solution of diborane in ammonia comes from the rupture of nitrogen-hydrogen bonds, the sources of which are believed to be the diammoniate and/or ammonia-borane.2 9. W. L. Jolly, Univ. of Calif. Radiation Laboratory, Livermore Site, Livermore, Calif., Contract No. W-7405-eng-48, UCRL-4504. 10. G. W. Schaeffer, M. TD. Adams, and F. J. Koenig, J. Am. Chem. Soc., 78, 725 (1956) 11.- This sample was analyzed by a thermal conductivity technique which gives the total WADC TR 56-318 73

TABLE III A TRACER STUDY OF TEE REACTION OF "MISTREATED DIAMMONIATE OF DIBORANE"(a) WITH SODIUM IN LIQUID AMMONIA Reaction Time, Reaction Hydrogen Equivalents Isotopic Source NH3, ml B21D6, mmoles - (c umulative) Temp., Evolved (cumulative) Analysis, Bj2DLJ6 hr ~C diborane, mmoles total % D This 5 2.06 10-15 O 75 -78 o. 65 0. investigation 3 -78 o.94 0.0 21 -78 1.13 0.0 (a) In this investigation diborane was added to ammonia in the usual fashion for preparing the diammoniate but the excess ammonia was removed at -450 and the solid was stored at -450 for five hours. Jolly 15 1A42 4 3 0.33 -64 1o28 0.0 (a) Jolly prepared mistreated diammoniate through the rapid addition of diborane to ammonia above -78~

TABLE IV AMMONOLYSIS REACTIONS Isotopic Composition of Run Residue From: Ammonia Gas Produced,-:'otal % D 1 B2H6 ~ 2ND3 ND3 49.4 88.2% HD 2 B2DB6 2ND3 ND3 97 3 Mistreated NH3 47 B2D6 - 2NH3 The isotope effect which occurred when ammonia-d3 was the coordinated ligand (e.g., B2D6-2ND3) arose from the fact that since the ammonia-d3 was not completely deuterated, it contained N-H bonds which were ruptured preferentially over the N-D bonds. Since all the nitrogen deuterium bonds in the system, including solvent ammonia-d3, were not ruptured in the reaction with sodium, the evolved gas was en12 riched in protium. The calculation of an instantaneous separation factor, ae seemed to be attractive, but the complexity of the system and the large errors in a% which are introduced by the comparatively small errors in hydrogen analysis have made its value very uncertain. The best estimate places it in the range of 7 to 15. Details concerning its estimation and the complexity of the system are summarized elsewhere. 1 The presence of the isotope effect implies that the rate-determining step in the process is the rupture of the nitrogen-hydrogen bond. In contrast to the regular tracer studies, the control reaction (Footnote 8) in which sodium reacted with ammonia in the presence of a catalyst showed little evidence of an isotope effect, which implies that the slow step of the reaction is not the rupture of the nitrogen-hydrogen bond. Schaeffer, Adams, and Koenigl report that ammonia attacks polymeric aminoborine in the ammonolysis reaction of the nonvolatile residue from the sodium-diammoniate reaction. They represent the reaction, empirically, as H2NBH2 + XH3 >- BH(NH2)+ex'+ X2 This representation is in accord with the tracer study. Although they make no mention of it, their results indicate that the ammonia which reacts with the aminoborine is coordinated to metal cation of the metal borohydride which is also formed in the metal-diammoniate reaction. The effect of coordination is to increase the acidic character of hydrogen which is bound to nitrogen in the ammonia, thereby facilitating the attack on hydridic hydrogen by acidic hydrogen. 12. H/D hydrogen evolved O = H/D ammonia-ds at zero extent of reaction. 13. S. G. Shore, doctoral dissertation, Univ. of Mich., Ann Arbor, 1956. WADC TR 56-318 75

Experimental. - a. Materials. 1. Ammonia-Commercial tank T[i3 was dried and stored over sodium metal in the vacuum system. The NDs which was used in the tracer studies was prepared through the hydrolysis of Mg3N2 by D20. Mg3N2 was prepared by passing pure, dried nitrogen gas for 12 hours over 20 g of highest purity Dow magnesium metal which was heated to 650~ in a stainless steel tubular furnace. The Mg3N2 was then transferred as rapidly as possible to a 125-ml round-bottom flask which was connected to the vacuum system. About a milliliter of D20 (ca. 99 deuterated) was distilled onto the Mg3N2 which was maintained at -78~. Vigorous reaction occurred upon warming the flask to room temperature, but the volume of the system was large enough to accommodate all the ND3 which was produced without having the final pressure exceed one atmosphere. The ND3 was sublimed from the system at -78~. The process of adding milliliter quantities of D20 was repeated until all the Mg,.N2 was used up. The NDS was purified by further sublimation at -78~ and by drying it over distilled sodium metal. Its vapor pressure indicates that it was about 97%0 deuterated. Observed vapor pressure (corrected). 110,2 mm at -63.5~; 354.2 mm at -45.2~o Reported15 vapor pressure (corrected) for ND,, 98*0 deuterated; 109.7 mm at -63.5~; 353.8 mm at 745.20. The hydrogen which was obtained from the control reactions Fe2O3 - PtO2 Na + ND3.- 1/2 D2 + NaND2 was between 96 and 95* deuterated. The latter value probably represents a lower limit on the purity of the ND,, 2. Diborane-B2He was prepared through the well-known16 procedure of adding (C2H5)20OBF3, in excess, to an ether slurry of LiAlH4. BDe6 was prepared in a different manner. About 4 g of AlC13 (obtained from a freshly opened bottle) was dissolved in 30 ml of anhydrous ether in a dropping funnel. The dropping funnel was connected to a standard diborane generating apparatus,17 and the solution was allowed to 14. Supplied on allocation by permission of the Atomic Energy Commission. 15. I. Kirschenbaum and H. C. Urey, J. Chem. Phys., 10, 706 (1942). 16. I. Shapiro, H. G. Weiss, M. Schmick, S. Skolnik, and G. B. L. Smith, J. Amo Chem. Soc., 74, 901 (1952). 17. P. R. Girardot, doctoral dissertaion, Univ. of Micho, Ann Arbor, 1952. WADC TR 56-318 76

drip slowly into a continuously stirred slurry containing 50 ml of anhydrous ether, 1 g of 200-mesh LiD, and a trace quantity of LiAlD4,l8 which served as a catalyst. The system was flushed continuously with dry nitrogen. After one and one-half hours of continuous stirring, 10 ml of distilled (C2H5)20BF3 [prepared by passing BF3 into anhydrous (C2H5)20] was added slowly, by means of the dropping funnel, to the ether slurry. The slurry was stirred for an additional three hours. The B2D6 which was produced was purified in the vacuum system by trap-to-trap distillation at -145~ until it displayed a constant vapor pressure. The yield was about 80% of theory. Its vapor pressure was in excellent agreement with the only reported value for B2D6.19 Observed vapor pressure (corrected). 238.3 mm at -111.9~. Reported vapor pressure (corrected) for B2D6, 98% deuterated: 25383 mm at -1119~0 Further evidence for the purity of the B2D6 was given through its quantitative hydrolysis. B2D6 + 6H20 > 6HD + 3H3B03 The gas obtained was 48.5% Do 3. Sodium-All the sodium which was used in the tracer studies was distilled. The still is depicted in Fig. 1. Metallic sodium was scraped free of its oxide coating and was placed in the side arm of the still. The system was evacuated by gently heating the side arm; the sodium melted and ran into the still pot. The constriction in the side arm was sealed off and after placing a tubular furnace around the pot, the sodium was refluxed in vacuo, with continuous pumping on the system, at 3000 for several hours. Then the column of the still was heated and the metal was vacuum distilled at about 480~o Sodium collected above the constrictions leading to the small glass bulbs (1l5-2 cm in diameter). With gentle heating and tapping, the metal melted and ran through the constrictions into the bulbs. The bulbs were then sealed off in vacuo. 4. Catalytic mixed oxides for the control reaction between sodium and ammonia-d3 —Reagent-grade Fe(NQO3 )3s9H20 was decomposed, thermally, at 300~. The resultant Fe203 was heated at 700~ for two days in order to remove traces of water. The mass was then pulverized. About 10% by weight of PtO2 was thoroughly mixed with the Fe2Os3 and the mixture was heated for another day at 700~O About 0.2 g of the hot catalyst was transferred to the bulb crusher-reactor20 18. Both the LiD and LiAlD4 were obtained from Metal Hydrides, Inc., Beverly, Mass. 19. A. B. Burg, J. Am. Chem. Soc., 74, 1340 (1952). 20. The mode of transfer of catalyst and the bulb crusher-reactor used were identical with those of boric acid, described elsewhere in this report (p. 64). WADC TR 56-5318 77

TO MANIFOLU ASBESTOS BRAID 1/52 SHEEE T ASBESTOS 8 TURNS PER INCH A\//~ (~~ -A t'~~#24 NICHROME,3-4 AMPS. 24/40 SODIUM IS MELTED ANDRUNS INTO STILL POT SEAL OFF Fig. 1. Sodium still. WADC TR 56-318 78

on the vacuum system. The catalyst was heated in vacuo at 300~-400o (at least 24 hours) until it was used. b. Procedures. A schematic diagram of the system in which the tracer studies were carried out is' given in Fig. 2. All the stopcocks and joints which had to be rotated were greased with Apiezon N; all the stationary joints were sealed with Apiezon W. The system was provided with space for two bulb-crusher reactors so that a tracer study and a control experiment could be carried out simultaneously. The procedures which were used in the tracer studies were essentially the same as those which have been used previously,21 except that great care was taken to prevent contamination of ammonia-d3 by H-D interchange which can arise from the presence of impurities containing active protons. In order to minimize errors due to this source, the -system was conditioned as described below. The vacuum system, with the bulb crushers in place, was out-gassed for one day under high vacuum; then a small amount of D20, in the vapor phase, was admitted to the-system and allowed to equilibrate, for one day, with the active hydrogen of the residual impurities. Following equilibration the system was again out-gassed under high vacuum for a minimum of three days. This procedure guaranteed that any traces of impurities would be rich in deuterium and thus would not seriously contaminate the highly deuterated ammonia. Once the system was conditioned it was never opened until a given run was completed. In a typical tracer run a control experiment was set up in one bulb-crusher reactor. This amounted to maintaining a sodium-ammonia-d3 solution in the presence of the mixed-oxides catalyst at -45~. The regular tracer experiment was set up in the second reactor. After the control had produced a significant quantity of hydrogen, it was quenched by immersing the reactor in liquid nitrogen. The hydrogen was freed of traces of ammonia-d3 by passingit through the u,-trapwhich (Fig. 2) was packed with glass helices and immersed in liquid nitrogen. Then the gas was passed into one of the sample bulbs and into the previously calibrated thermal conductivity cell for H-D analysis. After the gas was analyzed, the stopcock on the sample bulb was closed and the system was evacuated. The sample bulb was then sealed off at the constriction. The gas in the sample bulb was submitted to mass-spectrometer analysis in order to cross-check the thermal conductivity method. The same procedure was followed in the regular tracer experiment. At various intervals the.reaction was quenched, the hydrogen was freed of traces of ammonia, and the gas was passed into a sample bulb and the thermal conductivity cell for H-D analysis. Since the volume of the system was known, the amount of hydrogen gas which was produced in the reaction could be determined from PVT measurements. 21. The use of the bulb crusher, the preparation of the "diammoniate" and solutions of diborane in ammonia, and the carrying out of reactions with sodium in liquid ammonia are all described in the experimental section of the preceding paper (p. 54). WADC TR 56-318 79

TO MANIFOLD BU LB CRUSHER-REACTORS H2 STORAGE D2 TORAGE ERCURY BUBBLER THERMAL CONDUCTIVITY CELL GAS -COLLECTING SAMPLE BULBS CAPILLARY BREAK-OFF Q —-TO MANIFOLD 3-mm BORE CAPILLARY MANOMETER TOEPLER PUMP TO WHEATSTONE BRIDGE U-TRAP PACKED WITH GLASS HELICES Fig. 2. Schematic diagram of system for tracer studies. WAIDC TR 56-318 80

c. Analytical. The magnitude of the isotope effect observed made it possible to use a relatively simple thermal conductivity technique for analyzing the H-D gas mixtures. The particular procedure which was used was patterned after that of Peri and Daniels,22 with a few modifications.23 Several of the analyses were cross-checked by mass spectrometer analyses; the agreement was very good. The precision of individual measurements, for the most part, was within 1% D. Considering the good agreement between analyses which were obtained by both mass analysis and thermal conductivity techniques, and considering the general consistency of the results, it is believed that the accuracy of the thermal conductivity technique was well within 3% D in the range 50-100% D. 22. J. B. Peri and F. Daniels, J. Am. Chem. Soc., 72, 424 (1950). 23. A detailed description of the thermal conductivity method is given in the doctoral dissertation of S. G. Shore, Univ. of Mich., Ann Arbor, 1956. WADC TR 56-318 81

G. MOLECULAR-WEIGHT IEASUREMENTS IN LIQUID AMMONIA~ THE MOLEC'ULAR WEIGHTS OF THE METHYLAMINE BORANES, TE DIAMMONIATE OF DIBORAITE, AND AMMONIA BORON TRIFLUORIDE (R. W. Parry, G. Kodama, and D. R. Schultz) Abstract Molecular-weight measurements in liquid ammonia by vapor-pressure depression show' that H3NBF3, MeHH2NBH,3 Me2IHNBH3, and Me3NBH3 are all monomeric in liquid ammonia under conditions comparable to those used. for studying the "'diammoniate of diborane'" The conventional dimeric formula is confirmed for the diammoniate prepared at -78o5~C but a higher value above 80 is indicated for the diammoniate prepared at -450Co The conversion of H3NBF3 to NH4BF4 inwater solution is noted. The original problem associated with the formulation of the ammonia-dib:orane addition compound arose from the early observation of Stock and Pohlandl showing that the product corresponded to the formula B2H6 02N13 in liquid ammonia. Rathjens and Pitzer2 reported that freezing-point depression measurements confirmed the observations of Stock and Pohland,l but their points showed wide scatter and the same disturbing rise in molecular weight with dilution was noted which had been found in the earlier study. Even more disturbing at the present time is the fact thatt it is now known that the technique'which Rathjens and Pitzer used does not produce the so-called classical di.=noniate in liquid ammonia.o3.Although'the d.lammoniate of d.iboran.e' has long been considered to be unique among the boron hydride additi^on compounds, no molecular-weight measurements on H2MeNBH3, HMee2NBH3, Me3NB]3Hs, or He3:NBF3 have been carried out under conditions comparable to those used in the study of the di. ammoniate, hence the generality of the'di-nmmoniate" structure under low.tem-perature.ondirtions remains undetermined. A stuady of molecular weights in.liqui.d. ammonia was undertaken: 1. to check the validifty of 0tock'. s observatio:ns aS to the molecular weight of anxi authentic sample of the`'diammoniate"t 2, to check the molecular ~w~'eights of other betterr-known ionic solids in liquid ammonia in order to provide a, frame of reference for interpretation; and 3. to check the molecular weightts:in liquid ammonia of methylamine borane, dimethylamine borane, trimethylamine borane, and ammonia-boron trifluoride. 1. A. Stock and E. Pohland, Ber., 58, 657 (1925). 2. G. Wo Rathjens, Jr., and K. S. Pitzer, J. Amo Chem. Soc., 71, 2783 (1949)o 3. S Go Shore and Ro W. Parry, this Report, po 54~ WADC TR 56-318 82

These compounds had not been measured under conditions comparable to those used in studying the "diammoniate," hence a real question existed as to their nature in relation to the "diammoniate." Experimental. a. Reagents. The "diammoniate of diborane," methylamine borane, dimethylamine borane, and trimethylamine borane were prepared from carefully purified reagents using standard procedures. Commercial amines (Eastman Kodak Co.) were distilled through a low-temperature fractionating column at 400-mm pressure and dried over sodium metal before use. Commercial NaBH4 (Metal Hydrides) was placed on the sintered glass frit of a vacuum-line extraction apparatus and extracted directly into the molecularweight apparatus with liquid ammonia~ NH4Br, reagent grade, was dried at 80'C for three hours then stored over Mg(C104)2 before use. Ammonia-boron trifluoride was prepared by the interaction of dry NH3 and commercial tank BF3 which had been passed through a B2:03-H2S04 mixture. A oneliter, three-necked flask cooled in ice water served as the reaction vessel. The solid product was placed on a sintered glass frit and leached into the molecular-weight vessel with liquid ammonia. Ammonia analysis on the product gave 20.01% as compared to 20.05% theoretical for H3NIBF3. An x-ray powder pattern of the solid checked the pattern reported by Koenig5 and by Keenan5 but not the pattern reported by Laubengayer and Condike.6 It was also found, contrary to earlier reports, that H3NBF3 was converted almost quantitatively to NH4BF4 under conditions of this experiment by slow crystallization from water at room temperature. Isolation of crystalline NH4BF4 from the solution and precipitation of As04BF4 from the water solution indicated the change. An x-ray powder pattern of the sample taken before and after the molecularweight measurement in liquid ammonia showed that the sample had not changed on standing in liquid ammonia at -400C. No lines for NH4BF4 were found in the product. b. Apparatus and Experimental Procedure. Details of apparatus design were extremely important since very small tem4. The authors wish to thank Dr. Thomas C. Bissot who prepared the R3NBH3. 5. Rev. F. Jo Koenig, S. J., Summary Report XVI, Mathieson Chemical Corp. Project, Subcontract M-3181-14, Dept. of Chem., St. Louis Univ., St. Louis, Mo. 6. A. Wo Laubengayer and G. Fo Condike, J. Am. Chem. Soc., 70, 2275 (1948). WADC TR 56-318 83

perature differences throughout the system caused. serious vapor-pressure differences. The apparatus is shown in Fig. 1l The system could be effectively tested for temperature differences by putting solvent in both tubeso No pressure dief ferential could be detected on the manometer when the system was operating properly. The stirrer could be removed from one of the dried tubes, then the tube capped and weighed. It could then be attached directly to a vacuum-line extractor, and moisture or air-sensitive solutes could be extracted directly into the tube with liquid ammonia. After ammonia was removed, the tube was filled with dry nitrogen, capped, and reweighed. The dry stirrer was then inserted and the, tube was sealed to the molecular-weight apparatus, using Apiezon W wax on. the 14/35 9 joint. Ammonia was added from a flask containing a solution of lithium nitrate in ammonia. By weighing the lithium flask before and after addition of ammonia, the weight of solvent added to the system could be accurately determined. Pressure differences were read using a precision cathetometero Stopeocks were lubricated with silicon grease. c. Calibration of Apparatus. To check the reliability of the procedure on a well-behavred solute, the molecular weight of urea was determined. Values obtained by three different observers using various modif:ications of the apparatus are tabulated below in Table I. TABLE I MOLECLAR WEIGHTS OF URE.A Observer Molal Conc. of Solno Observed Mol Wt A 0.267 60o8 ~ 0.3 B 0.202 60.0 03 0.o382 602.2 0.3 C 0.20 6 + 1 0.30 60 1 0060 60 + l 1.2 62 ~ 1 1o.6 64 + 1 Results and Discussion.a. The Molecular Weights of Well-Kinown Ionic Solids in Liquid Ammonia, Results are presented in Table II. In the concentration range stuLdied NH4Br WADC TR 56-318 84

TO VACUUM LINE w-u-TO MANOMETER i GLS WIO _ / R TO DIFFERENTIAL OIL GLASS WOOL__ /> Q s - MANOMETER OR WIDE-'~v,J Ij BORE DIFFERENTIAL ~.1 fi' ~-~ MERCURY MANOMETER 14/35 1 JOINT TO ELECTRIC STIRRING MOTOR PULSATING ELECTROMAGNET MAGNETICALLY OPERATED STIRRER ~STYRAFOAM 0,~* o ~'''OCEE "| -|'"''"''|' STYRAFOAM. - a' ALUMINUM - FOIL _ _ OVER OVER STYRAFOAM I-GALLON DEWAR FLASK ETHYL ALCOHOL- - SOLUTION SOLVENT (Both at some level in both) SLUSH BATH OF DICHLORETHYLENE Fig. 1. Molecular-weight apparatus. WADC TR 56-318 85

TABLE II MOLECULAR WEIGHTS OF NIH4Br, NH4BF4, AID NaBH4 IN LIQUID AMMONIA Salt Molal Conc. of Soln,...Observed Mol Wt NaBH4 1. 90 28.6 Theor. Mol Wt 1.04 33.9 = 37.8 0.71 35.2 o.56 35.4 o.45 33.0 0.41 34.7 NH4Br 1.52 97.5 Theor. Mol Wt 0.75 98.0 = 97.9 0.50 98.0 0.30 98.5 NH4BF4.1.58 92.0 Theor. Mol Wt 0.80 96.0 = 104.8 0.52 930 lss 0.30 85.0 io ~~0.20 ~75.J ciation and NaBH4 exist in ammonia solution as nondissociated ion pairs. The data indicate that NH4BF4 was undergoing some dissociation in the more dilute solutions. Data on potassium iodide were somewhat erratic and showed an increase in molecular weight with dilution. The results of a number of measurements could be extrapolated to give a value of about 166 at zero concentration, but the precision of measurements in the potassium iodide study was in general low. Further study is needed for this solute. b. The Molecular Weight of Ammonia Boron Trifluoride in Liquid Ammonia. The only previous molecular-weight measurements on NH3BF3 were made in water solution.6 Since at least under some conditions H3NBF3 may undergo conversion to NH4BF4 in water solution, additional molecular-weight data seemed desirable, The possible existence of a species in liquid ammonia, [F2B(NH3)2](BF4), analogous to B2H6o2NH3, seemed to be worthy of investigation at low temperatures. Data for liquid ammonia solutions are summarized in Fig. 2. It was found that solutions with a molal concentration above 0.65 reached equilibrium in less than half an hour and held constant within + 2 molecular-weight units for at least four hours; longer observations were not made. A linear extrapolation of these points to: zero concentration gives a molecular weight of 93 as opposed to a theoretical value of 85 expected for the monomer. These data indicate that H3NBF3 is indeed monomeric in liquid ammonia solution and a structure comparable to WADC TR 56-318 86

380 360 340320 300 280 260-,3240220 0 w 0200z w 180 - rr a. 160 140 120 10080 I-Calculated molecular weight 60 0.1.2.3.4.5.6.7.8.9 1.0 1.1 1.2 1.3 1.4 1.5 FORMULA WEIGHTS OF NH3BF3 PER 1000 G OF NH3 Fig. 2. Determination of the molecular weight of NH3BF3 in liquid ammonia. WADC TR 56-318 87

B2H6o2NH3 is eliminated, An unexplained characteristic of the measurement is clearly revealed, however, by the available data. For solutions in the concentration range 0.30 to 0.56, values were well above the extrapolated curve; the system was slow (two to nine hours) in reaching equilibrium and an uncertainty of about + five molecular-weight units was observed. The uncertainty and time required to reach a steady state both increased with dilution. In solutions more dilute than 0~3 molal, results were completely erratic; in the most dilute solutions no equilibrium value was apparent after more than nine hours. The results were similar to-those observed with KI in dilute solution. Although the exact cause of this behavior is unknown, it is apparently associated with the solute.- arid not with the apparatus proper. Table I shows ihat a well-behaved solute such as urea gives cons-istent...and- reproducible values& -inithe concentral tion range.2 to.4 molall; hence::sys tematic errors due:to pressure measurement, etc, are not the primary cause of such behavior. c. The Molecular Weights of the Methylamine Boranes in Liquid Ammonia. The variation of apparent molecular weight with concentration for the three methylamine boranes is shown in Fig. 3. Again, if uncertainties of dilute solutions are neglected, an extrapolation of the data to zero concentration gives 51 for CH3NH2BH3 as opposed to a value of 45 expected for the monomer, a value of about 61 for (CH3)2NHBH3 as opposed to a value of 59 expected for the monomer, and a value of 79 for (CH3)3NBH3 as opposed to a value of 73 expected for the monomer. The data offer convincing proof that the methylamine adducts of diborane, prepared and measured under conditions comparable to those used in studying the "diammoniate of diborane," are not dimerized but are instead normal addition compounds of the borane group.7 Such observations have a precedent in the study of the metal coordination compounds. Cobalt (II) chloride will pick up ammonia to give the very stable compounds CoCl2 2NH3, CoCl2 4iNH3, and CoCl2.6NH3, yet Yoke and Parry8 found that only one trimethylamine or triethylamine will coordinate with CoC12. It would appear that the extra energy needed to expand the lattice of CoCl2 in order to get more than one molecule of tertiary amine coordinated to the cobalt (II) is greater than the energy released in the coordination to the cation of the second molecule of the amine. In a similar manner it appears that the steric strains associated with the coordination of two methylamine molecules to the small boron (III) cation are greater than or comparable to the energy released in the coordination process for the second amine molecule. At the present time the data seem to be explained most easily by the concept of F-strain: H NH3 R+ R = H or alkyl s \Bh 3 nB rR group H NITH3 stable RH \R unstable, probably due to R J steric hindrance 7. These data do not preclude the possibility of forming a dimeric species (R3NBH3)2 by other techniques. 8. J. T. Yoke III and RF W. Parry, to be published. WADC TR 56-318 88

100 - CH3NH2BH3(Molecular weight 80- calculated 44.90) 60 44.9 I 40 I I I I I I I I (_ 0.1.2.3.4.5.6.7.8.9 1.0 1.1 1.2 1.3 1.4 ~: FORMULA WEIGHT OF CH3NH2BH3 PER 1000 G NH3 D 70 0 (CH3)2 NH B H3(Molecular 58.93(< weight calculated 58.93) Z 50_'. 40 I I I I I I I I I.2.4.6.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 80 72.959 701- (CH3)3N B H3(Molecular weight calculated 72.95) 60.1.2.3.4.5.6.7.8.9 1.0 1.1 1.2 1.3 FORMULA WEIGHTS SOLUTE PER 1000 G NH3 Fig. 3. Molecular weights of the methylamine boranes in liquid ammonia. WADC TR 56-318 89

d. The Molecular Weights of the "'Diammoniate of Diborane" in Liquid Ammonia. The extreme sensitivity to experimental conditions of the ammonia-diborane reaction has already been noted.3 In Fig. 4 data are presented for molecular weights of two samples of the diammoniate, In one case ammonia was removed at -78.5~C to give the so-called classical diammoniate of Schlesinger and Burg.9 Although the values show uncertainty characteristic of this solute, they check fairly well with the earlier measurements of Stock and Pohland1 and justify the simple dimeric formulation and are consistent with the structure QB )NH3) (BH4() for this species. When the ammonia is removed at the higher temperature of -45~C, the sodium reaction indicates that a new species is formed. The formula s e e(BH4 has been tentatively assigned to this compound to account for the stoichiometry of hydrogen liberation.o Molecular-weight measurements on a sample from which ammonia was removed at -45~ showed a higher value than the -78.5~ product. An extrapolation of data for the -45~ case shows a molecular weight of about 82 as compared to a value of 92 expected for the formula written above. Within the limits of accuracy of the evidence, the molecular-weight data support the postulates made to explain the data on the stoichiometry of the sodium reaction. It is recognized that somewhat more precise measurements would be desirable. An effort is currently being made to obtain such additional data on the B2,H-NH3 system. 9. H. I. Schlesinger and A. B. Burg, J. Am. Chem. Soc., 60, 290 (1938). WADC TR 56-318 90

900 Q, 9 I | Theoretical for \ l "Diammoniate of diborane" rc [(NHo Hi++(BH) sample prepared at -450 o'Diammoniate of diborane" sample prepared at -78.5~ 80 * Original data of Stock z3 and Pohland for diammow | \ niote of diborane n- 70 - W -I Theoretical for B2H6. -' 0 6-0 NH3 kO ha. 350 0 40 30 0 1.0 2.0 3.0 4.0 FORMULA WEIGHTS B2H6-2NH3PER 1OOGG NH3 Fig. 4. Molecular-weight measurements on the "diammoniate of diborane."

H. TE RAMAN SPECTRUM OF TIE BOROHYDRIDE ION AND EVIDENCE FOR THE CONSTITUTION OF THE DIAMMONIATE OF DIBORANE (R. C. Taylor, D. R. Shultz, and A. R. Emery) Abstract The spectroscopic work presented on the borohydride ion appears adequate to establish satisfactorily the fundamental frequencies of this ion in liquid ammonia solution although precise values of two of the fundamentals have not been determined. Additional work using isotopically pure boron and perhaps ND3 as a solvent appears desirable in this connection. The results on the diammoniate of diborane give strong evidence for the presence of a borohydride ion in solutions of this substance in liquid ammonia and agree with the formulation which has been proposed in this laboratory for the cation, although they are not conclusive. The evidence from the B-H stretching region however is not in agreement with what one would expect from the presently accepted formulation of Schlesinger and Burg. Data on the decomposition product of ammonium borohydride and the product of the reaction of the diammoniate with ammonium bromide confirm the conclusions based on the chemical work. The usefulness and advantages of liquid -ammonia as a solvent and reaction medium have been amply demonstrated by many workers ever since the early work of Franklin and Kraus before the turn of the century. Experimental difficulties encountered in obtaining vibrational spectra of substances dissolved in this solvent, however, have handicapped workers in their attempts to study the nature of the solute species present. Solvent absorption presents an almost insurmountable barrier to the use of liquid NH3 in the case of infrared, but the situation is quite favorable for Raman spectroscopy, the chief difficulties being experimental in nature and involving the development of satisfactory methods for preparing, clarifying, and transferring solutiOns below the boiling point of the solvent (-33~C)o Very few papers have appeared on -the Raman spectra. of liquid ammonia solutions, and those which have appeared have dealt exclusively with substances forming solutions having less than an atmosphere vapor pressure at room temperature~ The present paper deals with the vibrational spectra of the borohydride and borodeuteride ions and., in addition, presents the results of a spectroscopic study of the diammoniate of diborane, the decomposition product of ammonium borohydride, and the product of the reaction between the diammoniate of diborane and ammonium bromide. All spectra were obtained in liquid ammonia solution at temperatures below -33~O The. Raman spectra of none of these substances have been published previously, although the infrared spectrum of the borohydride ion in the solid state has appeared. The nature of the diammoniate of diborane has long been the subject of discussion and the present results supply, as far as it is known, the first piece of purely physical evidence as to its structure. WADC TR 56-318 92

Experimental. — a. Materials. Sodium and lithium borohydrides were obtained from Metal Hydrides, Inc., and purified by removing all ammonia insoluble material before use. As a check on the purity, all borohydrides were analyzed after purification. Potassium borohydride was prepared by the method published recentlyl and purified as above, while the ammonium borohydride was prepared in solution as described elsewhere (this Report, p 4). Lithium borodeuteride was prepared by the reaction of LiD and B2D6, while the potassium salt was obtained by the reaction of KB(OCH3)4 and B2D6. Both salts were purified and analyzed before use. The method of preparation of the diammoniate of diborane and the other compounds studied has been described elsewhere.2 The ammonia used was the standard Matheson product, which was dried before use by storing over metallic sodium in a small steel cylinder. b. Preparation of Samples. The volumes of the samples examined varied from about 3 to 12 ml and the concentrations from about 1 to 5 molar, according to the amount of substance available. No significant effects of concentration on band shapes or positions were noted. The solution to be examined usually was prepared by condensing ammonia on the dry solid in a small graduated tube and having a mediumn-porosity glass frit in the bottom. Ammonia gas, admitted up through the frit, served both to keep the solution in the bulb and to stir it; if solution took place slowly dry nitrogen could be substituted for the ammonia to prevent excessive dilution. When solution was complete and the volume adjusted to the desired value, the sample was filtered through the frit by nitrogen pressure on the upper surface and transferred. to a second bulb having a 30-mm ultrafine, bacteriological-type, sintered glass disc in the bottom. Any fine suspended particles remaining were removed here as the solution dripped through by gravity. The sample finally was transferred to the Raman tube by nitrogen pressure. Those samples suspected of being contaminated by traces of stopcock grease, which fluoresced badly, were allowed to flow through about a two-cm column of Norite prior to the ultrafine filter. The Raman tubes were approximately 15 cm long with a diameter appropriate to the volume of solution. The system was completely closed to the air during the preparation of the sample and was opened only for a second at the end when the Raman tube was removed and capped. The design was such that the filtering apparatus could be completely immersed in a 100 x l400-mm clear Dewar flask filled with isopropyl alcohol at -35~ or below during all operations. A diagram of the apparatus is seen in Fig. 1. Transfer of the Raman tube from the filling system to the light source could be accomplished sufficiently rapidly so that negligible warming of the sample occurred. Suitable precautions were taken to eliminate the frosting of the sample tube during transfer. 1. M. D. Banus, R. W. Bragdon, and A. A. Hinekley, J. Am. Chem. Soc., 76, 3848 (1954). 2. R. W. Parry and S. G. Shore, this Report, p. 54. WADC TR 56-318 93

TO BLOW-OFF i TO BLO -OFF 2 SUPPORT TO BLOW-OFF 3 TO BLOW-OFF 4 ml tIII14Il\l s ml ~~~~~~~~~~~~~20 25 SPRING-HELD Imm STOPCOCK.. 20 A 12 350mm 12 bi B 30mm FINE 0o ~ FRIT MEDIUM 77 ro FRIT ro 18 RAMAN TUBE C. 30 mm Fig. 1. Apparatus for purification and loading of samples in a Raman tube. WADC TR 56-318 94

c. Spectroscopic Equipment. The Raman spectra were recorded photographically, using the spectrograph and light source described previously.3 In general, exposure times ranged from 15 minutes to 5 hours and several exposures were made on more than one sample of each substance. Measurements were made directly on the plates with a Mann comparator and also on enlarged tracings made with a Leeds and Northrup microphotometer. The numerical values reported represent the average of all measurements. Many of the lines were weak and rather broad and, consequently, their maxima could not be measured as precisely as desired because of the general noise level of the tracings. Figures are given for the estimated probable errors with the data. Results and Discussion — a. The Borohydride Ion. The Raman frequencies observed for lithium, sodium, potassium, and ammonium borohydrides dissolved in liquid ammonia are listed in Table I, as are also the frequencies of lithium and potassium borodeuterides in the same solvent. There appears to be a small but real shift in the maxima of the -bands in the different salts, which may be related to the polarizing ability of the cation. The solvent also affects the position of the bands since in two experiments on aqueous solutions of KBH4, the B-H band maximum was measured at about 2290 cm-l as compared to 2269 cm-' in ammonia, and the shape of the band was altered. No other observations in aqueous media were tried due to the instability of the solutions. Since the hydrogens in BH4- are hydridic in nature, isotopic exchange with the hydrogens of ammonia does not occur at any detectable rate and it was possible to obtain the spectrum of the borodeuteride ion also in NH3. In the following discussion, the data for the potassium salt are used; the shifts of frequencies with cation are sufficiently small so that no ambiguity results. Assignments are based on a tetrahedral ion belonging to the point group Td. This model predicts four fundamental frequencies, one totally symmetric stretching mode (A1), one doubly degenerate mode (E), and two triply degenerate modes, one stretching and one bending (F2). All fundamentals are allowed in the Raman effect, but only the F2 fundamentals are allowed in the infrared. On the basis of its intensity and polarization properties the Al mode (vl) can immediately be assigned to the most intense band in the Raman spectrum, which, for the potassium salt, occurs at 2269 cm"1 in the hydrogen case and 1568 cm-1 in the deuterium. The frequency ratio observed is 1.45 instead of the harmonic value of 1.41, which may indicate either that the anharmonicity effect is not the normal one or that one of the fundamentals is involved in a Fermi resonance. Although as can be seen from Fig. 2, the B-H stretching region contains one or two bands which might be involved in this way, the relative intensities show that any such interaction must be weak and could not account for the size of the shift necessary 35 Go Lo Vidale and Ro C. Taylor, J. Am. Chem. Soc., 78, 294 (1956). WADC TR 56-318 95

TABLE I RAMAN FREQUENCIES (IN CM-') AND ASSIGNMENTS FOR LITHIUM, SODIUM, POTASSIUM, AND AMMONIUM BOROHYDRIDES AND FOR LITHIUM AND POTASSIUM BORODEUTERIDES DISSOLVED IN LIQUID AMMONIA LiBH4 NaBH4 KBH4 NH4BH4 LiBD4 KBD4 Assignment 1 Assignment 2 824 + 5 827 + 5 v4 V4 1202 + 4 1205 + 4 1214 + 4 1205 + 4 862 + 10 v2 v2 2141 +5- 2151+ 5 2155 +5 2146 6 5 2v4 2v4 2239 +- 10... 1662 + 3 1665 + 3 v3 (v2+v4)(for BH4 only) 2260 + 2 2264 + 2 2269 + 2* 2264 ~ 2 1566 + 1 1568 ~ 1 vI V 2394 ~.5 2402-+ 5 2404 + 5 2400 + 5 2v2 V3 *Polarized.

to produce agreement with the harmonic value. In order to assign the band at 1214 cm-1 as the E mode, v2, it is necessary to review the infrared data, since there are two H-B-H bending frequencies to be expected in this general region. Price4 has reported a band at 1080 cm,1 in a mull of NaBH4, while Hornig5 reports the same band at 1121 cm-1 for a solid film of the same salt at liquid nitrogen temperatures. In this laboratory, a value of 1122 cm-1 was obtained for KBH4 by the KBr pellet technique. These data on the solid indicate that the infrared active band v4 lies somewhere in the neighborhood of 1100 cm- and that the 1205 cm-' band consequently is most likely v2. Unfortunately, when liquid ammonia is used as a solvent, the region around 1100 cm- is completely obscured by the very strong v2 band of the NH3 molecule so that the position of v4 in solution could not be established directly in the present work. However, the band appearing at about 2155 cm- can hardly be explained except as the overtone of v4 and its position indicates that the fundamental is probably not far from 1080 cm-1. In the spectrum of the deuterated ion, two bands were observed in the low-frequency region, one at 862 cm-1 and one at 827 cm-l1 The former was quite weak but agrees with the shift expected for v2 and is so assigned. The latter was somewhat more intense and is assigned to v4. The assignment of the remaining F2 fundamental, v3, is perhaps the least certain of the four because of the unexpected complexity of the B-H stretching region. Since this fundamental is infrared active, one might expect some help from infrared results.o Price, in his spectrum of solid NaBH4, shows at least three poorly resolved bands in this region with a main maximum about 2270 cm-. The spectrum of KBH4 in a KBr pellet shows a moderately intense band at about 2278 cm1-, a less intense but well-resolved band at 2210 cmp, and a shoulder at about 2340 cm-'. These bands can be assigned with a high degree of confidence to vs, 2v4, and v2 + v4, respectively. Hornig5 gives a value of 2298 cm-l with no mention of other bands. Although the shift between solid and solution is unknown, it appears that v3 probably lies in the neighborhood of v1o In., the Raman spectrum of the B-H stretching region (Fig. 1), at least four bands are present, including the A1 fundamental, The band at 2155 cm-1 has already been assigned, while the band at 2404 cm-1 appears to be 2v2.a The remaining band, which is present as a shoulder on the 2269 cm-1 peak, is most easily explained as the V3 fundamental~ In a spectrum of NaBH4 taken with polarized light, the intensity of vl was greatly reduced and the band in question, as near as could be determined, was depolarized with its maximum at about 2240 cm-1. Perhaps the chief objection to this assignment is the fact that in most other tetrahedral molecules, V3 occurs at a higher frequency than vl. However, the alternative assignment (Assignment 2, Table I) implies a rather large shift between.the solvent and solution for v3 and requires a rather large anharmonicity correction in the explanation of the 2240 cm-1 band as v2 + v4. 4. W. C. Price, JO Chem. Phys., 17, 1044 (1949). 5o Do F, Hornig, Disc. Far. SOc., No~ 9, P 120 (1950)o WADC TR 56-318 97

In the borodeuteride spectrum, vW is well isolated and v3 is assigned to the maximum of the strong band at 1665 cm-1. The Teller-Redlich product-rule ratio for the F2 class calculated from the masses of the atoms involved is 1.77. Using the figures from Assignment 1, Table I, the approximate frequency product ratio is 1.76, while the ratio from Assignment 2 is 1.89. This provides strong support as to the correctness of the first assignment. b. The Diammoniate of Diborane. Previous work from this laboratory has supported the structure [H2B(NH3)2+] (BH4-) for the classical "diammoniate of diborane"6 rather than the currently accepted ammonium ion formulation of Schlesinger and Burg, (NH4+)(H3BNH2BH3-).7 Since most of the evidence presented is chemical in nature, a Ranan study was carried out to supply additional information The frequencies observed exclusive of solvent bands are tabulated in Table II and a tracing of the B-H stretching region is shown in Fig. 2. TABLE II RAMAN FREQUENCIES (IN CM-') OF THE DIAMMONIATE OF DIBORANE DISSOLVED IN LIQUID AMMONIA Diammoniate.H[4BEH Product from of Decomposition NH4Br-Diammoniate Diborane Product Reaction 772 + 3 733 i 3 773 + 3 839 5 852 + 0lo 882 + 5 884 8 8 - 1209 + 4 1214 + 4 1208 + 4 2140 2265 + 2 s 2263 5 2 s 2321 3 2322 + 3 2322 ~ 3 2403 + 3 2405 3 2407 i 3 2437 + 3 2441 + 3 2437 + 3 Comparison of the data in Table II with those in Table I shows that the quencies of four characteristic bands of the borohydride ion agree with freqi cies in the diammoniate spectrum within the experimental error in determining maxima and the change in position observed with the cation. A more striking comparison is made if one:mentally subtracts the bands at 2321 and 2437 cm1l from the diammoniate spectrum in Fig. 2 and compares the remainder with the borohydride 6. All earlier papers in this report. 7. H. I. Schlesinger and A, B. Burg, J. Am, Chem, Soc., 60, 290 (1938). WADC TR 56-318 98

1/ 002 3 Hg 2000 2400 2800 Fig. 2. The B-H stretching region inthe Raman spectra of some borohydrides dissolved in liquid ammonia. Tracing 1, LiBH4; 2, NaBH4; 3, the diammoniate of diborane. (Frequency in cm'-) WADC TR 56-318 99

spectrum. The very close similarity furnishes strong evidence for the presence of a borohydride ion in the liquid ammonia solution of the diammoniate of diboraneo Conclusions which can be drawn solely from the spectroscopic evidence about the nature of the accompanying cation are more limited. The structure given above for this cation, [H2B(NH3)2+], is isoelectronic with propane. In agreement with this formulation, only two B-H stretching frequencies,not found in the borohydride ion spectrum, are observed. Two frequencies are most consistent with the presence of a BH2 group since a BH3 group would contribute three unless it happened to be present in an ion having C3v symmetry. The latter seems unlikely from chemical generalizations. No information could be obtained about the N-H modes due to the interference of solvent bands. In the lower frequency region, three bands were observed at 772, 839, and 882 cm-. The first was relatively sharp and agrees with what one might expect for a symmetrical stretching mode of the N-B-N skeleton. The latter two were weaker and much more diffuse members of a doublet. No lower frequency which might be assigned to a skeletal bending mode was observed, although this failure may have been caused by experimental difficulties. In propane, the symmetrical and unsymmetrical skeletal stretching modes occur at 867 and 922 cm-l respectivelyo The recorded spectrum thus is reasonably consistent with the new formulation6 but, of course, does not exclude other possibilities for the cation. On the other hand, the structure proposed by Schlesinger and Burg7 for the diammoniate contains six B-H bonds which, judging from the spectra of analogous molecules such as propane and dimethyl ether, would be expected to give a very complicated spectrum in the B-H stretching region. That such a pattern fortuitously would match that obtained by superimposing two additional frequencies on the borohydride spectrum seems extremely unlikely, and to this extent, the spectral evidence does not agree with their proposed structure. c. Decomposition Product of Ammonium Borohydride. Chemical evidence has been given elsewhere9 to the effect that if liquid ammonia solutions of INH4BH4 are evaporated to dryness and the solid allowed to warm to room temperature, the solid evolves hydrogen and leaves a product which has the properties of the diammoniate of diborane. The spectrum of this decomposition product prepared as described was obtained in the present work largely to supply confirmatory evidence. Due to experimental difficulties, the solutions obtained were more dilute' than those of the diammoniate and were not as "clean" optically. However, the frequency values in Table II match those of the diammoniate within experimental error and the band outline in the B-H stretching region 8. G. Herzberg. Infrared and Raman Spectra of Polyatomic Molecules. New York::D van Nostrand andCo;,- -194:-5, po.3561. 9. R. W. Parry, D. R. Schultz, and P. R. Girardot, this Report, p. 4. WADC TR 56-318 100

also agreed very closely. The identity of the decomposition product with the diammoniate thus appears confirmed. d. Product of the Diammoniate —Ammonium Bromide Reaction. As a consequence of the instability of solid ammonium borohydride at room temperature, treatment of the diamoniate with arnmonium bromide under the proper conditions can result in the destruction of the borohydride ion present and its replacement by the bromide ion. In terms of the structure proposed previously, the reaction can be written [H2B(NH3)+](BH) + 2NBr - [H (N (r) + 2H21 Details of this reaction have been given elsewhere10 and the properties of the product described. The Raman frequencies observed for this reaction product dissolved in liquid ammonia are given in Table II. Inspection shows that all values listed also appear in the diammoniate spectrum and, further, that the frequencies of the borohydride ion plus those of the NH4Br reaction product together account for all bands observed in the diammoniate spectrum. The spectral evidence thus is strong that the reaction written above is Correct and that in the reaction product one has the bromide salt of the cation which is present in the diammoniate. Two bands appear to be common to the spectrum of the cation and of the borohydride ion, namely, the bands at-about 1210 cm1 and their overtones at about 2405 cm1l., Apparently the bending frequency of the BHZ group in the cation.occurs at almost the identically same position as v2 of the borohydride ion. Due to the small, and in this case unknown, effect of the cation on the borohydride frequencies plus the inherent experimental uncertainty in measuring the positions of weak and rather broad bands, the two cannot clearly be distinguished. 10. D. R. Schulz and R. W. Parry, this Report, p. 19. WADC TR 56-318 101

IV. UNUSUAL COMPOUNDS RESULTING FROM THE BORANE GROUP WADC TR 56-318 102

A. BACKGROUND Previous investigators have noted a formal resemblance between a B-N system and a C-C system. Since boron has one less electron than carbon and nitrogen has one more electron, the combination is isoelectronic with the C-C unit. From this analogy the cyclic compound borazene HB'BH HN AN H is frequently referred to as "inorganic benzene." The physical properties show a striking resemblance to those of benzene, but the chemical properties differ. For example, all attempts to hydrogenate borazene to obtain an:"inorganic cyclohexane" have failed. In the work described herein, an N, N', N"-trimethyl "inorganic cyclohexane" has been produced. The second compound, F3PBH3, was of interest because of theoretical arguments against a bond between F3P and boron. Such a bond can exist, but it is of limited stability. WADC TR 56-318 103

B. PREPARATION AND PROPERTIES OF TRIMER'C N-METHYL-AMINOBORANE (T. C., Bissot and R. W Parry) Abstract The compound N, N', N Nt-trimethyl "inorganic cyclohexane" or, more properly, trimeric N-methylaminoborane has been prepared and its properties are described. Among the more interesting of the boron-nitrogen compounds are the borazenes. The structure and properties of these compounds have been well characterized and have been compared with those of benzene and its alkyl derivatives. Hitherto there has been no report of a "saturated borazene" of the formula (R2NBR2)3. in which R is either hydrogen or a methyl radical, which would be structurally similar to cyclohexane or its methyl derivatives, The compound (CH3)2NBH2 is known to exist as a dimer at room temperature and as a monomer at higher temperatures, while H2NBE2 Is reported as a high polymer.l However, recent observations in this laboratory indicate that ECH3-NBH2 is trimeric and is in all probability a sixmembered ring composed of ali~eriating boron; and nitrogen atoms. The new compound, tr-imeric N-methylaminoborane, was first isolated in low yield from the decomposition products obtained in the pyrolysis of O,N-dimethylhydroxylamineborane.2 It can be prepared more conviently and in yields of 80 to 90% by heating methylamine-borane at 1000o 1000 3CH3NEL2B.H3 --- 3:2 + (CH31hBH2)3 ) (l) The identity of the compound was proved by chemical analysis, molecular-weight determinations, and its decomposition at 200~ into hydrogen and 1,3,5-trimethylborazene Although this compound does not display the extreme stability toward heat and hydrolysis which has been reported by Burg and his co-workers3 for the trimeric Pdimethylphosphinoborane and the trimeric As-dnmethylarsrinoborapnie, it is very stable for a boron-nitrogen compound containing two active hydrogens per boron. It is unaffected after weeks in contact with moist air and is hydrolyzed only very slowly by cold water. 1. A. Be Burg and C. L. Randolph, Jr., &. Am. Chem. Soc., 73, 953 (1951); E. Wiberg, Ao Bolz, and P. Buckheit, Z. anorgo Chem., 256, 285 (1948); H. Io Schlesinger, et alo, Univo of Chicago, Signal Corps Contract W3434-SC-174 Final Report (1948-r949). 2. A study of the reaction of hydroxylamine and the five possible methyl-substituted hydroxylamines with diborane is being conducted in this laboratory and will be published in the near futureo 35 Ao B. Burg and R. I. Wagner, J. Am. Chem. Soc., 75, 3872 (1953); F o G. A. Stone and A. B. Burg, ibid., 76, 386 (1954)e WADC TR 56-318 104

The pyrolysis of methylamine-borane has been studied by a number of previous investigators,,4 but this initial pyrolysis product never has been characterized. Experimental. — a. Preparation of Trimeric N-Methylaminoborane. In a typical preparation, 3 mmoles of diborane was condensed with liquid nitrogen, together with a small excess of anhydrous methylamine (7 mmoles). The mixture was then allowed to warm up to room temperature over a period of about three hours. A small amount of hydrogen was produced in this preparation of methylamine borine. Air was then admitted to the system and the tube was removed from the vacuum line, stoppered with a calcium sulfate drying tube, and heated on a steam-bath for two hours. The liquid in the tube changed into a crystalline mass after the first hour of heating. This white solid was then placed in a clean tube, which was replaced on the vacuum line. A beaker of boiling water was placed around the bottom of the tube and the trimeric N-methylaminoborane sublimed under high vacuum. The product collected as a ring of fine needles on the walls of the tube just above the surface of the bath. The tube was removed from the line and broken on each side of the ring. The long white needles were then scraped from the walls; yield about 0.22- g or 85%. The analysis of the product is summarized in Table I. TABLE I ANALYSIS FOR (CH3NHBH2)3, % 0Obsd. Calcd. Carbon 28. 00 28.00 Hydrogen 146 02 14 -10 "'Active" hydrogen 4e35 4.70 Boron 25.0 25.23 Nitrogen 32..78 32%66 b. Molecular-Weight Determinations. The molecular weight of the solid was estimated from the freezing-point depressions of benzene and nitrobenzene solutions, using a standard Beckmann apparatus. The results indicated molecular weights of 121 and 141 for two trials in benzene and 134 for nitrobenzene. The molecular weight was also determined by the vapor-pressure lowering of 4. H. I. Schlesinger, D. M. Ritter, and A. B. Burg, J. Am~ Chem. Soc., 60, 1926 (1938); E. Wiberg, K. Hertwig, and A. Bolz,; Z. anorg. Chem., 256, 177(1948); G. W. Schaeffer and E. R. Anderson, J. Am. Chem. Soc., 71, 2143 (1949); A. B. Burg and Lo C. Randolph, ibid., 3451 (1949). WADC TR 56-318 105

liquid ammonia solutions. The apparatus designed by D. R. Schultz5 was modified by using a large slush bath of ethylene chloride as the constant-temperature medium. A molecular weight of 127 was obtained. A trimer of CH3NHBH2 has a calculated molecular weight of 128.7. c. Decomposition into 1,3, 5-Trimethylborazene. A quantity, 0.2364 g of (CH3MTBH2)3 was placed in an evacuated sealed tube and heated at 200~ for four hours. Upon opening the tube, 5.96 mmoles of hydrogen, identified by a molecular weight of 2.1, was found. This gives a ratio of hydrogen per mole of starting material of 3.24. The 1,3,5-trimethylborazene was purified by vacuum condensation, the portion collecting in a trap at -45~ being retained. A 74% yield was obtained based on the equation 2000 (2) (CH3NMBH2)3 -_30 3H2 + (CH3NBH)3 (2) The 1,3,5-trimethylborazene produced was characterized by vapor-pressure measurements over the temperature range 50' to 75~ and by vapor-phase molecularweight measurements based on vapor density (obsd. = 126; theor. = 122.,5). The melting point of the solid was 0 to -10 as compared to -9~ reported by Wiberg, Bolz, and Buckheitl and -7~ to -8~ reported by Schaeffer and Anderson.4 It is believed that the higher value of 0 to 1~0 indicates greater purity of the sample studied and does not negate the characterization of the borazene. do Properties of Trimeric N-Methylaminoborane. The compound may be recrystallized from methyl or ethyl alcohol as long white fibrous needles. An analysis for boron and "active" hydrogen on this recrystallized material demonstrated that the composition was unchanged. In addition to the above solvents, the compound is very soluble in acetone and liquid ammoniao It is moderately soluble in benzene, ether, and chloroform and is insoluble in carbon tetrachloride, petroleum ether, and water. The insolubility in water may be primarily a wetting problemo The density of the compound is 0.90 g/ml as determined by centrifuging a few crystals in a series of mixtures of carbon tetrachloride and kerosene. The crystals would remain suspended in the mixture with a density of 0.90 g/ml. A sample of the solid was allowed to stand exposed to the moist air of the laboratory for three weeks, during which time no noticeable hydrolysis occurred. A portion placed in cold water will float for hours but will disappear after about a day. A 20% solution of hydrochloric acid must be heated to boiling to bring about rapid hydrolysis. e, X-Ray Powder Diffraction Data for Trimeric N-Methylaminoboraneo In Table II are listed the interplanar spacings (d values) and the relative 5. De R. Sch~ultz, doctoral dissertation, Univ. of Mich., Ann Arbor, 1954. WADC TR 56-5318 106

line intensities found to be characteristic of the compound. The powder patterns were taken with copper Kc radiation in a cylindrical camera of 57.3-mm diameter. The low-absorbing glass capillaries used to hold the sample had a diameter of 0.2 mm and a wall thickness of O01 mm. TABLE II INTERPLANAR SPACINGS (d VALUES) AND RELATIVE LINE INTENSITIES OF TRIMERIC N-METHYLAMINOBORANE d I d I d I 7.1 VS 2.68 VW 1.20 VW 4.26 VS 2.29 W 1.88 VW 4.00 MS 2.27 W 1.77 VW 3.45 M 2.19 W 1.70 VW 3006 VW WADC TR 56-318 107

C. THE PREPARATION AND PROPERTIES OF PHOSPHORUS TRIFLUORIDE-BORANE AND PHOSPHORUS TRIFLUORIDE —BORANE-d3 (R. W. Parry and T. C. BTissot) Abstract Phosphorus trifluoride and diborane react under pressure in sealed tubes at room temperature to give the new compound F3PBH3. Some physical and chemical properties of the new compound are described. Its properties show a striking resemblance to those of carbon monoxide-borane. NF3 does not add to the BH3 group, but under appropriate conditions B2HE is oxidized explosively by NF3. The compound Pt(PF3)2C12 prepared by Chatt and Williamsl bears a striking resemblance to Pt(CO)2C12, and the complex Ni(PFs)4 prepared by Irvine and Wilkinson2 is very similar.in properties to Ni(CO)4. This experimental resemblance between the coordinating properties of CO and PF3 suggested the existence.of the compound F3PBH3, which would be analogous to the OCBH3 of Burg and Schlesinger.3 On the other hand, Chatti pointed out that PF3 did not add to A1C13, AlBr3, or BF3, and he invoked theoretical arguments as the basis for the prediction that a stable bond was unlikely between the boron in boron acids of the Lewis type and the phosphorus of phosphorus trifluoride. If, as suggested by Chatt, the acid-base type of reaction did not occur, the possibility of an oxidation-reduction reaction between B2H6 and PF3 still merited consideration. In view of these interesting possibilities, an experimental study was conducted on the. system B2H6-PF3. The formally analogous system B2H6-NF3 has also been examined. The Preparation and Physical Properties of Phosphorus Trifluoride -Borane, F3PBH.. —The high-pressure. reaction between diborane and excess phosphorus trifluoride (8 atmospheres) proceeds slowly over the period of several days to yield the compound F3PBH3 as the primary -product.. On prolonged standing, secondary reactions involving F3PBH3 ensue which give as yet undefined liquid products along with appreciable amounts of hydrogen. The primary phosphorus trifluoride-borane' adduct may also be obtained by displacing CO from carbon monoxide —borane using a smaller excess of PF3 (5 atmospheres). Phosphorus trifluoride —borane is a colorless gas which is -spontaneously in-. flammable in air-. The melting point is -116.1~ + 0.20C. The vapor. pressure of F3PBH3 over the liquid range can be.given by the equation 1. J.'Chatt and A. A-. Williams, J. Chem.. Soc., 3061, (1951); J. Chatt, ibid., 3340, (1949). 2. J. W. Irvine, Jr., and G. Wilkinson,: Science, 113, 742 (1951; G. Wilkinson, J. Am. Chem. Soc., 73, 5501 (1951). 35 A. B. Burg and H. I. Schlesinger, ibid. 59, 780 (1937). WADC TR 56-318 108

Loglo P = -1038.9/T + 7.8061 Since the pure material resembles carbon monoxide-borane in that it will dissociate into B2H6 and PF3, it was necessary to purify the compound before each reading and to achieve temperature equilibrium rapidly. The vapor-pressure data extrapolate to give a boiling point of -61.80C, a heat of vaporization of 4760 calories per mole at the boiling point, and a Trouton constant of 22.5. The molecular weight by vapor density at 250C was 102.7 as compared to a theoretical value of 101.82. The Dissociation Constant of Gaseous F3PBH3 at 25~C, —The equilibrium constant for the reaction 2 F3PBH3 (g) M -a 2 PF3 (g) + B2H6 (g) was estimated by analyzing equilibrium mixtures according to the procedures described in the experimental sections. A value for K(atm) of 1.0 + 0.3 was obtained at 25~C. The uncertainty in the value can be attributed to three factors. First, a number of experimental difficulties were associated with the separation of the large excess of PF3 from the product and the unreacted diborane.4 Second, since no effort was made to thermostat the reaction vessel, normal deviation of room temperature from 250C would have some effect, Finally, secondary reactions involving the splitting out of H2 and the formation of less volatile products complicated the stoichiometry after longer periods of time. Rate of Decomposition of F3PBH13.3-In order to compare phosphorus trifluoride — borane with carbon monoxide —borane, their rates of decomposition at room temperature were compared. The decomposition reaction, 2 F3PBH3 - B2H6 + 2 PF3, produces an increase in pressure, and the rate of pressure change with time can be used as a measure of reaction velocity. Rate data are shown in Fig. 1 along with comparable data of Burg and Schlesinger3 for the OCBH3. The similarity of the rates is striking, particularly in the early phases where side reactions have not complicated the interpretation of pressure measurements. The mechanism for the decomposition of carbon monodixe-borane has been the subject of some recent controversy.5 Unfortunately, the current data do not aid in resolving the question; however, they do suggest that the mechanism which explains 4. The problem is simplified in the case of carbon monoxide —borane since the CO is not condensable with liquid nitrogen, whereas the other components are. Since PF3 boils at -95~C and B2H6 at -920C, chemical procedures had to be employed for analysis. 5. A. B. Burg, J. Am. Chem. Soc., 74, 3482 (1952); S. H. Bauer, paper presented befor the 127th Meeting, American Chemical Society, Cincinnati, Ohio, April, 1955. WADC TR 56-318 109

:J \3] 0 X o X + 0 0 100 200 300 400 500 Time Minutes Fig. 1. Decomposition of OCBH3 and PF3BH3 at room temperature.

OCBH3 will also explain F3PBH3, It is perhaps significant that both CO and PF3 might be classed as "secondary bases" as mentioned by Bauer.5 Reaction of F3PBH3 with Trimethylamine and with Ammoniao -When stoichiometric quantities of F3PBH3 and trimethylamine are mixed, the trimethylamine displaces the PF3 quantitatively to form (CH3)3NBH3 and free PF3. If an excess of trimethylamine is used, the PF3 which is liberated reacts with the excess trimethylamine in a oneto-one ratio to give gases and a solid which have not been fully characterized. When P3PBH3 reacts with excess ammonia over the temperature range of -1280 to -80~C, five molecules of ammonia will be picked up by each molecule of the borane adduct. The reaction product is a white solid which begins to decompose at room temperature with evolution of hydrogen and development of a yellow color. When the solid was heated to about 55~C, 162 moles of ammonia were evolved per mole F3PBH3. It is probable that the original solid product is a mixture. The reaction, which appears to be rather complex, is being studied further. Data suggest rupture of the phosphorus-fluorine bond by the ammonia. The Properties of Phosphorus Trifluorlde-Borane-d3. —The compound F3PBD3 was needed for Raman spectral studies. The deuterated compound had a melting point of -11.ol~ +.1~C. The vapor pressure of the liquid can be given by the equation Loglo P = -1010.8/T + 7.6171 Extrapolating this vapor-pressure data yields a boiling point of -59.80C, a heat of vaporization of 4630 calories per mole, and a Trouton constant of 22. The Reaction Between NF3 and Diborane.-NF3, which is formally analogous to PF3, is known to have essentially no basic properties. It was nonetheless interesting to compare the behavior of NF3 and PF3 in their reaction with diborane. Results can be summarized as follows: (1) under conditions similar to those used in making F3PBH3 and OCBH3 no compound formation between NF3 and B2H6 was observed; (2) in these experiments, mixtures of NF3 and B2H6 were stable in the gas phase even under high pressures; (3) mixtures of NF3 and B2H6 can be violently explosive at temperatures which are low enough to give condensed phases in the system. The conditions necessary to initiate explosive reaction between the two liquids were not clearly defined, but spontaneous explosive reaction was observed only when the contents of the reaction tube were liquef.ed after the gases had been standing for several days under pressure. This explosion was undoubtedly an oxidation-reduction process with N2, HF, and BF3 as probable productso The possibility that the reaction was triggered by very small amounts of impurities in the NF3, such as N20, NO, or NO3F, or by decomposition products of B2H6, remains reasonable. Discussion,,-The striking similarity between the physical properties of phosphorus trifluorlde-borane and carbon monoxide-borane is illustrated by the data in Table I. Even the differences between the deuterated and nondeuterated compounds are comparable. The stability of F3PBH3 is somewhat greater than the compar7_a.le OCBH3 (see K values in Table I). Its rate of dissociation at room temperature also is somewhat lower than the carbonyl adduct. These facts correlate well with the observations of Chatt and Williamsl to the effect that PF3 adducts of the platinum (II) halides are more stable than the corresponding carbonyl compounds. WADC TR 56-318 111

TABLE I COMPARISON OF THE PROPERTIES OF CARBON MONOXIDE-BORANE AND PHOSPHORUS TRIFLUORIDE-BORANE OCBH3 OCBD36 F3PBH3 F3PBD3 Melting Point -137.0 -134.4 -116.o 1 -115.1 Boiling Point -64 -62.2 -61.8 -59. 8 H Vaporization 4750 4760 4760 4630 Trouton Constant 23 23 22 22 Dissociation Constant Katm at 25~C 2.55 -- 1.0 + o3 Percent Decomposed After 1 hr at 25~C 17% -- 14% The initial reaction of F3PBH3 with trimethylamine parallels closely the corresponding reaction involving carbon monoxide —borane. Differences are noticed, however, in secondary reactions because the liberated PF3 reacts directly with trimethylamine whereas CO does not. The reactions of both OCBH3 and F3PBH3 with ammonia are more complex and require further study. The spontaneous inflammability of F3PBH3 in air correlates well with the general observation that the reactivity toward oxidizing agents and moisture of the simple borane adducts is inversely related to the strength of the base to which it is bound. The weak base strength of PF3 produces a highly inflammable adduct. The most reasonable model for the molecule would be an ethane-like structure with a sigma bond between the boron and the phosphorus. Double bonding of the type postulated by Chattl for the metal adducts of PF3 is ruled out' by the absence of d electrons on the boron. Other systems involving addition of PF3 to Lewis acids are under investigation. Further speculation on bonding can best await more complete experimental information. Experimental.9 - a. Preparation of Phosphorus Trifluorideo The phosphorus trifluoride was prepared by the reaction between phosphorus trichloride and zinc fluoride, using a minor modification of the procedure of Chatt and Williams.1 The crude products were passed through a reflux column at -80'C in order to recover unreacted PC13, then the impure PF3 was trapped in a liquid nitrogen trap and purified by fractionation in the vacuum line. b. Diborane Diborane was prepared by the reaction between lithium aluminum hydride and 6. AO B. Burg, J. Am.o Chem. Soc., 74, 1340 (1952)o WADC TR 56-318 112

boron trifluoride etherate in ether solution,7 After purification, by low-temperature fractionation, the vapor pressure at -111.80C was 225 mm. Diborane-d6 was prepared using lithium aluminum deuteride, The LiAlD4 was prepared in situ from lithium deuteride and aluminum chloride. c, The Preparation of F3PBH3, A measured amount of diborane, together with a large excess of PF3, was condensed in a heavy-walled Pyrex bomb tube of 60-ml capacity. The techniques resemble closely the procedures used by Burg and Schlesinger for the preparation of OCBH3A3 The bomb tube was allowed to stand several days at room temperature. Maximum yields were usually obtained after four days; longer periods of time frequently resulted in the formation of secondary products, although the factors responsible for the secondary reaction were never completely delineatedo After standing, the bomb tube was frozen with liquid nitrogen and the tube opened to the line by means of the vacuum-tube opener. Noncondensable gases were removed with a Toepler pump, measured, and identified by molecularweight measurements. The condensable portions were distilled through a trap at -1550C, at which temperature the F31:PBH3 is quantitatively retained. The excess B21H{ and PF3 was measured and analyzed by a sealed-tube hydrolysis, using the amount of hydrogen generated to determine the portion of B2EH in the mixture, The F3HPBH3 was freed from less volatile impurities by distilling it through a trap at -128~C. The pure F3PBH3 showed a vapor pressure of 23.0 mm at -111,80C. The less volatile materials formed from secondary reactions were easily separated from the more volatile components by fractional distillation in the vacuum line. Upon hydrolysis of these materials, only small amounts of hydrogen were liberated, In one case the less volatile materials were separated by fractional condensation and a liquid with a vapor pressure of 5 mm at -80oC was isolated. A measurement of molecular weight by vapor density gave a value of 120, It appeared stable in the vapor phase but polymerized in the liquid phase to give a nonvolatile oil, No hydrogen was liberated on addition of water, indicating that it contained no active hydrogens. However, the compound was destroyed by water. These observations, plus the fact that the secondary reaction results in the liberation of gaseous hydrogen, suggest the tentative formula F2PBF2 for the monomeric liquid, The characterization is still incomplete since only very small amounts of material were available and the conditions which favor its formation have not been completely determined. Data for representative runs are summarized in Table Io, do Analysis and Characterization of the F3PBH3. Hydridic hydrogen in the compound was determined by condensing a weighed sample of gas in the tube and then distilling in an excess of water. The sealed tube was allowed to stand at room temperature for one day, and then the tube was opened to the vacuum line and hydrogen was measured; observed hydridic hydrogen is 2.89*; theoretical for F3PBH3 is 2.97*~. 7, I. Shapiro, Ho G. Weiss, M, Schmich, S. Skolnik, and G. B. L, Smith, J, Am. Chem. Soco, 74, 901 (1952). WADC TR 56-318 113

The PF3 is partially hydrolyzed by the above procedure, and the resulting HF attacks the Pyrex glass enough to make a boron analysis meaningless. Small amounts of PF3 were always obtained after water hydrolysis; a contaminant which appeared to be SiF4 was always-present, as would be expected. The addition compound was hydrolyzed by 40% NaOH as the first step -in obtaining a phosphorus analysis; the P-F bond is cleaved rapidly in alkaline solution.8 The resulting solution was acidified and the phosphorous acld was oxidized to phosphoric by evaporating the solution almost to dryness with aqua regia. The precipitation of the iammonium phosphomolybdate was carried out in the usual manner. Considering the precision of the methods for determining phosphorus in compounds of this type, the result, 29.87% P, is in good agreement with the formulation F3PBH3 (30435% P). TAJLE II DATA ON REACTIONS BET.fWEEN PHOSPH3ORUS TR.i LUOREDE AND DIBORANE Run Number 3 4 5 6 8 Conditions Voldume of bomb, ml 62 5 62.9 63 5 61 61o 7 6 E'nitial mmoles B2H6 3~.28 3 20 3 16 3 22 5.42 Initial mmoles PF3 19 21 18. 18 19. C 17 92 110o. 71 Enitial pressure, atm 8.72 8 30 8. 57 8 36 6.40 Time at 250, days 2 4 7 22 25 Products Mmoles B26 1.2o 4 0.69 0. 27 0. 44 2o 12 Mmoles PPF3 4.65 13.1'7 115 37 120 31 4.87 Mmoles PFS3E3 4.26 5. 01 4. 34 4.73 4.50 Mmoles of less volatile by-products 0.o 0 0 o.8 1.40 0.o 38 0.84 Mmoles H2 0j G03 0o 55 1'29 2.10' Yield PE.3B.3 based on B2H6 65f 78.3% 68.7% 73 5% 41t.5% Apparent equilibri:um constant Katm 5. 73 1.86 0 71 1o 18 0. 98 e. Measurement of Physical Properties of 3BpF3. Conventional techniques were used for vapor-densilty and vapor-pressure measurements (Tables III and I).' Temperatures were measured with ethylene and carbon dioxide vapor-pressure thermometers. The melting point was obtah.:aed in the vacuum iine, using the method of Stock..9 8. H. S. Booth and A. R o Bozarth. J3 Amno Chemo Soco, 61, 2927 (1939)o 9. Ao Stock, Bero, 50, 156 (1917)o WADC TR 56.518 1l4

TABLE III VAPOR PRESSURES OF PHOSPHORUS TRIFLUORIDE-.-BORANE Observed Calculated AP Pressure mm Pressure mm (Cal. - Obs.) -127.7 4.1 (s) -119.7 11.4 (s) -111.8 23.0 23.4 +0.4 -101.5 56.6 57.0 +0.4 -98.1 76.4 74.6 -1.8 -95.7 90.6 89.7 -0.9 -87.6 162.1 161.6 -0.5 -79.9 267 270 +3.0 TABLE ITV VAPOR PRESSURES OF PHOSPHORUS TRIFLUORIDE-BORANE-d3 Observed Calculated AP Pressure mm Pressure mm (Cal. - Obs.) -119.4 8.o (s) -112.0 20.9 22.2 +1.3 -101.5 54.2 53.7 -0.5 -95.6 85.2 84.2 -1.0 -84.1 186 187 +1.0 -79.0 260 259 -1.0 f. Reaction of F3PBH3 with Trimethylamine. One and ninety three hundredths mmoles of F3PBH3 were condensed in a 250ml bulb with 2.31 mmoles of trimethylamine. When the mixture was warmed to room temperature, there was an immediate reaction in which a white solid was produced. The bulb was cooled to -80~C; the more volatile products were distilled out; the bulb was weighed; the trimethylamine-borane was sublimed out; and the bulb was reweighed. A 97% yield (0.1362 g) of (CH3)3NBH3 was obtained. Vacuum-line distillation of the volatile gases yielded 79.5% of the theoretical PF3. This was characterized by vapor-pressure and molecular-weight measurements (obs. 87.6; theor. 87.98). The unrecovered PF3 (0.4 mmole) corresponded exactly to the excess of trimethylamine (0.38 mmole), a fact which suggests that a reaction between PF3 and N(CH3)3 had taken place to yield the unidentified solid and gaseous products. This postulate was checked by allowing a small amount of trimethylamine to react with an excess of PF3. An approximately oneto-one reaction occurred to produce a nonvolatile white solid and a mixture of gaseous products which were qualitatively the same as those observed previously. WADC TR 56-318 115

g. Reaction of F3PBH3 with Ammonia. A quantity of dry ammonia, 14.70 mmoles, was condensed on the walls of the reaction tubeo Then, 2.20 mmoles of F3PBH3 were added and the temperature was raised from -128~ to -80~C over the period of 2-1/2 hours. The excess ammonia was then sublimed from the reaction tube at -80~C. The amount of ammonia recovered was 3.69 mmoles, meaning that 5.0 moles of ammonia react with one mole of F3PBH3. The reaction product was a white solid which began to decompose at room temperature by turning yellow and giving off small amounts of hydrogen. Upon heating the solid to about 55~C, an amount of ammonia equivalent to 1.62 moles of -the original 5.0 was recovered. During the decomposition a small amount of a white solid of very low vapor pressure was collected in a Dry-Ice trap. This white solid burst into flame when the trap was cleaned with concentrated nitric acid. ho Preparation of Nitrogen Trifluorideo Nitrogen trifluoride was prepared by a modification of the method of Ruff.10 A closed electrolysis cell was substituted for Ruff's open cell. The impure NF3 was bubbled slowly through a warm KI solution. Final purification was achieved by low-temperature fractionation in the vacuum line. The gas was passed slowly through a trap cooled to -185C with liquid oxygen. The impurities, principally small amounts of N20, condensed, while the NF3 passed through into a third trap cooled to -2080C with low-pressure liquid nitrogen (maintained at 100-rmm pressure'by pumping through a manostat). Small amounts of dissolved oxygen and nitrogen were pumped from the NF3 in the third trap to give a product of high purity. The molecular weight of the gas, as determined by vapor density, was 71.o2 The theoretical value for NF3 is 71.01. io The Reaction of Nitrogen Trifluoride with Diboranes. The sealed-tube, high-pressure techniques described for the preparation of F3PBH3 were employed. In a typical run 2.61 mmoles B2H6 and 18.84 mmoles NF3 were sealed into a bomb tube and allowed to stand for eight days at room temperature. When the tube was opened to the line, the amount of noncondensable gas was negligibleo A mixed pentane bath at -155~C was placed around the tube and the entire contents distilled into the vacuum system. On further fractionation in the vacuum line, no fraction could be retained in the -.155C trap. Finally, during a transfer distillation a violent explosion occurred. The temperature of the vessel at the time of the explosion was estimated to be about -1700~C, since a liquid nitrogen Dewar had just been removed and another lowtemperature bath was to be substituted. The explosion occurred before the second bath was put in placeo In a second trial under similar circumstances an explosion occurred while the contents of the reaction tube were being distilled into the vacuum lineo 10. 0o Ruff, J o Fischer, and F o Luft, Z o anorgo Chem., 172, 417-25 (1928). WADC TR 56-318 116

V. THE REACTIONS OF DIBORANIE WITH HYDROXYLAMINE AND ITS METHYL DERIVATIVES WADC TR 56-318 117

Ao THE PHYSICAL ANMI CHEMICAL PROPERTIES OF THE HYDROXYIJAMINES (To Co, Bissot, Ro Wo Parry, and Do H. Campbell) Abstract Free hydroxylamine and all its O- anzd N- methyl derivatives have been prepared in pure form. Melting points, vapor pressures, and. heats of vaporization have been determined for the entire series, Values are those expected from considerations of hydrogen. bonding. The basic constants for the series have been determinedo Variations in the series are compared to variations foun.d in the series ammonia, methyl aminesos Although all the methyl derivat-ves of hydroxylamirne have been mentioned in the early literature, the physical and chemical properties of these materials are for the most part either inaccurately described or unknowno Since these properties were important in delineating the conditions for the reaction of the hydroxylamines with diborane, and since the properties of the bases are of interest in their owr, right, a study of the puremeth;yL-substi tuted hydroxylamines has'been conducted. Results are reported herein.. Experiment al. a. The 7,reparation of Hydroxylamine and Its Methyl Derivatives. io Hy'droxylamine-The free'base was liberated from a suspension of hydroxylammoniium chloride in bu~tanol, using sodium butoxide as the base.1 The salt was thoroughly dried at iLC~C before use, Since the free base undergoes rapid decomposition at temperatures above 15'C and is very hydroscopic,'it was stored under anhydrous butanol at -10C lintil'used. Cold samples were filtered rapidly, wash.ed wih ld ai lold anlhydrot.s ethe:r., and. weighed, in a special vacuuxm-jacketed weighing "bottle. 2, O-methyl.hy-droxylamine.eMethoxy'aa-L ne hydrochloride (Eastman Organic Chemicals) was treated with an excess of 50% KOF, in a small one-piece distillation apparatus o The fraction.'boilii.g at 45~ 50~ was distilled onto KOH pellets contained in a:receiver which was immersed in. an. ice-water'bath. The free'base was then decanted into a tube containing BaO and equipped with a T joint. The tube was atftached to the line and the Omethylhydroxylamrxine was allowed to stand in. contact with the BaO for several dayso The CH30,lNH2 prepared. in this manner was contaminated with small amouants of other amines9 chiefly ammoniao These were removed.'by fractional condensation, effected -by holding the first trap at 63~ 9 the second at -79~, and the third at 196o. The desired component was retained in the second trap. l1 C.o D Hurd, in Inorganic Synthesis. Volo Io New York~ McGrawTHill Book Coo., Inc.,o 1939, p. 87. WADC TR 565318 i18

3. 0,N-dimethylhydroxylamine — This methyl-substituted hydroxylamine was prepared by the following reactions: C2H50C(O)C1 + HONH2'HC1 K2C03 > C2H50C(O)NHOH C2HSOC(O)NHOH + (CH30)2SO2 ---— 3 C2H50C(o)N(cH3)OCH3 C2H50C (O)N(CH3 ) 0CH3 a —-e K0H > C2H50C ( O ) OK + CH30NHCH3 CH30NHCH3 + HC1 ---- - CH30NHCH3'HC1 The literature directions2'3 were followed without modification. The hydrochloride salt was recrystallized from absolute alcohol by the addition of dry ether. The melting point was 1150-1160 (literature 115-116 ). The free base was prepared in the following manner. A few milliliters of 50% KOH was frozen in the bottom of a tube which was equipped with a T joint. "The amine hydrochloride was placed on top of the frozen solution and the tube attached to the vacuum line. After evacuating the tube, it was allowed to warm up to room temperature. The contents were mixed by agitating a small iron-cored stirring bar with a magnet. The free 0,N-dimethylhydroxylamine was then distilled into another tube containing BaO and was dried for several days before using. No further purification was required since all fractions had identical vapor pressures. As an extra checkon the identity and purity of the material, the molecular weight was checked by vapor density. The value observed, 61.7 (theory 61.10), was within the experimental error of the measurement. 4. 0,N,N-trimethylhydroxylamine —The trimethyl derivative was prepared as described by Jones and Major,4 using the reaction of methyl iodide with the 0,Ndimethylhydroxylamine. CH3I + 2CH30NHCHs ether) CH30NHCH3HI + CH30N(C3) The ether solution was filtered and phenyl isocyanate was added to remove any unreacted CH30NHCH3 as the N-phenyl-N' -methyl-Nt -methoxy urea derivative. The ether and the free O,N,N-trimethylhydroxylamine were distilled out and dry HC1 passed into the ether solution. The recrystallized hydrochloride salt had a melting point of 122.5~-123.5~. The literature value is 123~. The free base was prepared and dried using the method described for the previous amine. No further purification was required. The molecular weight was measured by vapor density as 76.2 (theory 75.11). 5. N-methylhydroxylamine-The N-methylhydroxylamine was prepared by the following series of reactions, using literature directions: 2. L. W. Jones, Am. Chem. J., 20, 38 (1898). 3. R. T. Major and E. E. Fleck, J. Am. Chem. Soc., 50, 1479 (1928). 4. L. W. Jones and R. T. Major, ibid., 2742. (1928), WADC TR 56-318 119

(')2CO + HON2oHC1 l-NaO- (0)2CNOH + 2H20 + NaCl5 (0~)2CNoH) + (CH30)2S02 (0)2N(o)CH3 + ()6 100~ 07 (~)2CN(O0)CH3 + 120 + HCI - HONHCH3HCl + (0)2C07 HONHCH3 HC1 + CH30Na > CH30H + NaCl + HONHCH3 8 The N-methylhydroxylamine is a solid of rather low volatility at room temperature. The material required a large number of fractional condensations before a portion was obtained whose vapor pressure did not change upon further fractionation, Because of the low vapor pressure, the purity was checked by a nitrogen analysis rather than by molecular weight (observed 29.69% N; theory 29.77% N). 6. NN-dimethylhydroxylamine-This hydroxylamine was prepared by the action of a methyl Grignard reagent upon ethyl nitrate in ether solution.9 After hydrolysis, the amine was removed by steam distillation and collected in dilute HCl. The resulting solution was evaporated to dryness and the crude hydrochloride was purified by two to four crystallizations from an ether-alcohol mixture. The salt was thoroughly dried under high vacuum on the vacuum line. The melting point oboserved was 1035~106~o 9epworth9 listed an mp of 1020, while Cope and co-workers10 reported 106, 50~109 o The free anhydrous HON(0-HB3)2 could not be prepared using the method described'by Hepworth. His procedure involved distilling the amine, together with some water from a flask containing the crude hydrochlqride and concentrated KOR.I The distillate was extracted with ether and. the ethereal solution dried with BaO, After several distillations he obtained a fraction'boiling at 94~5~ to 95o50~ which he thought was -the anhydro'as free base, and for which he reported nitrogen analysis of 22.78 and. 22.82% (theory'22o94%). In the current investigation this proced-Sure was modified somewhat by utSilizing vacuum techniques for the initial separation of the amine after adding an excess of KOH: to the recrystallized hydrochloride salto When a large amou..t of BaO was -used to dehydrate the mixture 5. A.o o Vogel. Practical Organic o!.hemistryo London~ Longmans, Green and C o., Inc o, 1948, p. 704. 6. The structural isomers CB;%3 NO OCH C"N,WO and N 3 are readily separated "by the difference of their solubilities in petroleum ether (30~-50~). The N-methyl compound is slightly soluble, whereas the other is insoluble 7. L. Semper and L. Lichtenstadt, Ber., 51, 933 (1918)o 8. Co Kjellin, ibido, 26, 2377 (1893)o 9o Ho Hepworth, J. Chem. Soco, 119, 255 (1921)o 10. A. Co Cope, To TO Foster, and P. Ho Towle, JO Amo Chemo Soc., 71, 3929 (1949). WADC TR 56-318 120

of water and amine obtained in the distillation, it was found that all the material was absorbed. When a somewhat smaller amount of BaO was used, some of the liquid could be recovered. The melting point was about 0~ and the vapor pressures, measured over the range of 0~ to 88~, extrapolated to a boiling point of 93.5.. However, a molecular-weight determination and a titration of a weighed amount indicated that a significant amount of water was still present in the amine. The weak base strength of the N,N-dimethylhydroxylamine suggested that it might be displaced from its hydrochloride salt by ammonia. This procedure was found to work very satisfactorily. A quantity of the recrystallized HON(CH3)2' HCl was dried at room temperature for several hours under high vacuum. An excess of anhydrous ammonia, dried over sodium, was distilled into the tube and stirred until the solution was clear. The mixture of ammonia and the hydroxylamine was then distilled away from the ammonium chloride and separated by fractional condensation. The HON(CH3)2 was retained in a trap maintained at -80~ with a Dry Ice —isopropyl alcohol bath. The displacement was quantitative since the remaining hydrochloride salt would not reduce silver nitrate. The melting point and boiling point of the anhydrous material were raised considerably, mp 17.7~, bp 100.70. Nitrogen analyses, 22.56 and 22.70% N, are in agreement with the theoretical value of 22.94%. The molecular-weight values of this amine, determined over a range of temperature, show that there is considerable association in the vapor phase. Temperature Apparent 0C Molecular Weight 59.0 73.0 64.8 70.5 69.4 69.2 98.3 64.6 The values approach the theoretical molecular weight of 61.09 at the higher temperatures. This anhydrous base was found to be quantitatively absorbed by alkaline drying agents such as BaO and KOH pellets. The hydroxyl group on this amine is responsible for the similarity of its physical and chemical properties with those of water. Since the amine and water also apparently form a low boiling azeotrope, they cannot be completely separated by ordinary means. b. Physical Properties. 1. Melting points-The melting points of these hydroxylamines divide the group into two categories. The ones with a methyl group on the oxygen are very low melting materials, whereas those with a hydrogen on the oxygen have melting points near room temperature. The data for the O,N-dimethyl- and the O,N,N-trimethylhydroxylamines were determined using the magnetic-plunger method first described by Stock.ll112 A sample of the amine was condensed with liquid nitro11. A. Stock, Ber., 50, 156 (1917). 12. R. S. Sanderson. Vacuum Manipulations of Volatile Compounds. New York: John Wiley and Sons, Inc., 19)8. WADC TR 56-318 121

gen in a ring which supported the glass plunger. A cooling bath of mixed methyl pentanes was cooled well below the melting point of the amine and placed around the melting-point apparatus. A carbon dioxide-vapor-pressure thermometer was used to measure the temperature of the bath. The cooling bath was then allowed to warm up slowly, with constant stirring to keep the temperature uniform. When the pointer dropped, the vapor pressure of the carbon dioxide gave the temperature at which the material melted. For the above two amines the melting points on three separately purified samples were reproducible to within less than a tenth of a degree. The melting points of the N-methyl- and the N,N-dimethwylhydroxylamine were determined visually in the same apparatus that was used to determine vapor pressures. A National Bureau of Standards calibrated mercury thermometer was used to measure the temperature. The results of this investigation and the values recorded by previous workers are collected in Table I. TABLE I MELTING POINTS OF HYDROXYLAMINE AND ITS METHYL-S'UBSTITUTED DERIVATIVES Compound Melting Point Compound Investigator ~C HONH2 33.05 Lobry de Bruyn13 32-33 Brtih114 HONHCH3 42 Kjellin8 15 38.2, 38.7 This investigation HON(CH3)2 17.5 -17.8 This investigation CH30NH2 -86.4 This investigation CH30NHCH3 -97.0 This investigation CH30N(CH3)2 -97.2 This investigation 13. C. A. Lobry de Bruyn, Rec. tray. chim., 10, 100 (1891). 14. J. W. Br'thl, Ber., 26, 2508 (1893). 15. The early observations of Kjellin were apparently rather crude. The boiling point which he reported for HON-IHCH3 was in error by a large factor (Table VI). Extensive purification in this investigation failed to raise the melting point above the reported value. WADC TR 56-318 122

2. Vapor pressures and related constants of hydroxylamine and its methylsubstituted derivatives-When handling volatile compounds in a high-vacuum apparatus, a knowledge of their vapor pressures at various temperatures is essential. Not only are the values necessary in order to design a separation procedure, but they provide a rapid check on the identity and purity of the separated material. No vapor-pressure data were available in the published literature for the methyl-substituted hydroxylamines. In view of these facts, a portion of this investigation was directed toward obtaining these data. The details of the measurement varied with the temperature range which was involved. (a) -80~ to O~. Four different two-phase systems were used as constant-temperature baths in this region. These were slush baths of chloroform (-63.5o~, chlorobenzene (-45.3~), carbon tetrachloride (-22.8~), and ice-water (0~). A Dry Ice-isopropyl alcohol bath was used at -78~. For intermediate temperatures a large bath of isopropyl alcohol was cooled to the desired temperature. This bath was well stirred and the temperature measured with an alcohol thermometer. Since alcohol thermometers are often unreliable, the one used in these measurements was calibrated each time it was used against two-phase slush baths of CC14, C6H5Cl, and H20. A mercury manometer was used to measure the pressures, which were read to the nearest 0.1 mm with a cathetometer. (b) 0~ to room temperature. The desired temperature was achieved by using a large, well-stirred water bath cooled to the desired temperature. A mercury thermometer with 0.01~ graduations, which had been calibrated and certified by the National Bureau of Standards, was used to measure the temperature of the bath. (c) Above room temperature. An imme:rsible v ensimeter designed after the one 16 described by Burg and Schlesinger was used to determine vapor pressures above room temperature. The tensimeter was immersed in a constant-temperature bath controlled to ~ 0.05~ Two National Bureau of Standards calibrated thermometers, -5~ to 50~ and 50~ to 1100, were used with stem corrections applied where they were significant. The vapor-pressure data were fitted to an equation of the form LoglO P = - A/T + B - C log T Data are summarized in Tables II to V. The heats and entropies of vaporization at the boiling point were estimated from the vapor-pressure data. Comparative physical data for all the methyl derivatives of hydroxylamine are summarized in Table VI. 16. A. B. Burg and H. I. Schlesinger, J. Am. Chem. Soc., 59, 780 (1937). WADC TR 56-318 123

TABLE II THE VAPOR PRESSURE:'OF O-METHY]L.YDROXYIAEMNE Empirical Equation Loglo P = -2433.7/T + 23.9284 - 5.3746 Log T Observed Calculated Temperature Pressure Pressure AP ~C Cal. -Obs. mm mm -63.5.75.69 -0.06 -45-.3 3.55 3-76 +0.21 -22.8 20.5 20.7 +0.2 -20.4 24.2 24.4 +0.2 -17.6 29.0 29.2 +0.2 -135.3 38.2 38.5 +o. 3 -10.4 45 6 46.1 +0.5 -6.8 56.4 57.0 40.6 0.0 83.9 83.8 -0O1 9.4 138.8 138.5 -0.3 10.1 143.5 143.6 +0.1 12.6 162.9 162.5 -0.4 14.4 177.3 177.4 +O.1 16.0 192.1 192.6 +0.5 17.7 208.4 208.6 +0,2 19.1 222.8 222.6 -0.2 20.1 233.6 232.7 -0.9 21.1 243.9 245.4 +1.5 24.8 291.1 290.5 -0.6 29.8 362.9 363.7 +0.8 34.9 451 450 -1 40.3 561 561 0 44.8 670 668 -2 48.2 761 765 +4 WADC TR 56-318 124

TABLE III THE VAPOR PRESSURE OF O,N-DIMETHYLHYDROXYLAMINE Empirical Equation Loglo P = -2282.9/T + 22.1065 - 4.7976 Log T Observed Calculated Temperature AP Pressure Pressure ~C Cal. -Obs. mm mn -45.2 6.1 6.0 -0.1 -42.0 8,0 8 0 -36.5 11.8 11.7 -0.1 -31.4 16.8 17.0 +0,2 -27.0 23.1 23.0 -0.1 -22.6 30.6 30.8 +0.2 -18.0 40.8 41.2 +0.4 -13.3 54.9 54.5 -0.4 -7.3 77.8 77.2 -0o.6 -4.2 91.5 92,0 +0.5 0.0 114.4 115.3 +0.9 4.0 141.1 141.3 +0.2 8.0 173.2 173.5 +0.3 12.0 210.9 210.7 -0.2 16.0 253.8 253.7 -0.1 19.7 301.4 301.0 -0.4 23.9 361.5 361.7 +0.2 28.0 430.4 431.0 +0.6 32.9 526.6 527.1 +-0.5 37.5 630.2 631.1 +0.9 42.4 765 762.6 -2.4 WADC TR 56-318 125

TABLE IV A. THE VAPOR PRESSURE OF O,N,N-TRIMETHYLHYDROXYIAMINE Empirical Equation Log10 P = -2296.9/T + 27.3690 - 6.8151 Log T Observed Calculated Temperature AP 00C ~Pressure Pres sure ~ C Cal. -0bso mm mm -78.5 0.9 0.9 o -63.5 3.6 3.9 +0o3 -45.2 17.3 16.8 -0.5 -39.0 26.5 25.8 -0.7 -35.0 35.5 33.7 -1.8 -29.8 47.5 46.8 -0.7 -25.2 61.7 61.5 -0.2 -20.0 81.5 82.9 +1.4 -14.1 113.6 113.4 -0.2 -9.5 144.4 145.8 +1.4 -4.o8 186.0 181.3 -4.7 0,0 226.9 227.7 +0.8 3.9 270.7 271.0 +0.4 8 5 328.0 328.7 +0O7 11.8 579.1 579.8 +0,.7 15o9 448.0 448.1 +0.1 20.0 525.4 524.9 -0.5 23.3 597.5 595.3 -2.2 B. THE VAPOR PRESSURE OF N-METHYLHYDROXYLAMINE Solid Phase-Empirical Equation Log10 P = 13.7/T - 4.261 + 5.63 x 105T2 0.0 0.9 1.0 +0.1 5.0 1.5 1.4 -0.1 10.0 2.0 2.0 0 15.0 2,.9 2.9' 0 20.0 4.2 4.2 0 25.0 6.1 6.1 0 30.0 9.0 9.0 o 55.0 13.5 13.4 -0.1 Liquid Phase —Empirical Equation Log10 P = -2597/T + 9.570 20.0* 5.2 5,2 0 25.0* 7.3 7.3 0 30.0 10.0 10.1 +0.1 35.0* 1359 13.9 0 40.0 19.1 19.o -0.1 45.0 25.8 25.4 -0.4 50.2 34.6 34.7 +0.1 55.0 45.4 45.4 0 60.0 59.8 59.7 -0.1 65.0 76.8 77.8 +1.0 ~Superooled liquid. WADC TR 56-318 126

TABLE V THE VAPOR PRESSURE OF NN-DIMETHYLHIYDROXYLAMINE Empirical Equation Log10 P = -3780.1/T + 39.4125 - 10.2693 Log T Observed Calculated Temperature Pressure Pressure AP ~C Cal.-Obs. mm mm 17.6 13,0 13.0 0.0 19.1 14.5 14.4 -0.1 25.2 21.4 21.4 0.0 29.8 28.5 28.6 0.1 34.7 38.0 38.1 0.1 40.3 - 52.4 52.5 0.1 44.9 67.3 67.3 0.0 50.3 90.1 89.8 -0.3 55.1 114.6 114..2 -0.4 60.2 146..0 146.8 o.8 65.1 1835. 183.8 0.5 70.2 229.7 231.2 1.5 75.1 284.4 285.2 0.8 80.2 351.7 351.2 -0.5 85. o 427.0 426.3 -0.7 90.0 521 518 -3 TABLE VI MELTING POINTS, BOILING POINTS, HEATS OF VAPORIZATION, AND' TROUTON CONSTANTS OF METHYL-SUBSTITUTED HYDROXYLAMINES Melting Boiling Compound Point Point AH Vap. H ~C ~C 760 mm cal/mol e u. HONHCH3 38.5 115.0 11,880 30.6 HON(CH3)2 17.6 100.6 9,670 25.9 CH3ONH2 -86.4 48.1 7,710 24.0 CH30NH (CH3) -97.0 42.3 7,440 23.6 CH3ON(CH3)2 -97.2 30.0 6,410 21.1 WADC TR' 56-318 127

c. Chemical Properties. 1. The base strength of hydroxylamine and its methyl-substituted derivatives-The base strength of this series of compounds in water solution was determined potentiometrically. A dilute solution of the hydrochloride salts was titrated with 0.05 N NaOH and a typical titration curve was obtained. The hydrochloride salts of these amines were carefully purified and dried...A quantity was weighed out: and diluted to obtain a solution which was approximately 0.0025 molar. A 100-ml aliquot was transferred to a four-necked flask, The central neck contained a stirrer; two necks were used for the electrodes, while the fourth was used for the burette. A Beckmann model G pH meter with a combination glass-calomel electrode system was carefully standardized against a potassium acid phthalate buffer. Both the four-necked titration flask and the buffer solution were thermostated at 25 + 0.10. The dilute solution of the amine hydrochloride was then titrated with 0.05 N NaOH from a 5-ml microburette. The pH of the solution was recorded after the addition of each 0.2 ml of base; near the endpoint much smaller increments were taken. The pKa of the amine was then calculated over the middle third of the titration curve by the formula total ml NaOH - ml NaOH pKa = pH + Log. The values calculated over the central portion of the curve rarely deviated by more than 0.005 pK unit from their average. The titrations we're repeated using 100-ml aliquots of the same hydrochloride solution and adding weighed amounts of KC1 to vary the ionic strength of the solution. The averages from these titrations were then plotted as a function of the square root of the ion'ic strength and the extrapolated value of pKa at infinite dilution was obtained. The glass electrode was used in this study because of its simplicity and because of errors arising from the reducing action of hydroxylamine on the noble metal salts of the. usual reference electrode.17 In order to estimate the magnitude of the errors in the procedure, the constant for NH40H was determined using the identical procedure. The value obtained, pKb 4.73, is in reasonably good agreement with the precision value of Bates and Pinching,18 who reported a value of 4.751. In view of the inherent errors in the glass electrode, such as the, asymmetry potential, and the uncertainties in standardizing the pH meter, the error which has been assigned to the pKb values of these hydroxylamines is + 0.03 pK unit. Data are srummarized in Table VII. 17. D. M. Yost and H. Russell, Jr. Systematic Inorganic Chemistry. New York: Prentice-Hall, Inc., 1944, p. 97. 18. Ro G. Bates and G. D. Pinching, Res. Paper 1982, J.o Res. Natl. Bur. Std., 42, 419 (1949) o WADC TR 56-318 128

TABLE VII pKa VALUES FOR AMMONIA, HYDROXYLAMINE, AND METHYL-SUBSTITUTED HYDROXYLAMINES Ionic Average Value of pKb Compound Average at Infinite Strength Apparent pKa Dilution NH3 0.058 0.241 9.310 0.0213 0.146 9.298 0.0025 o.o050 9.277 0.00 0.00 9.27 4.73 + 0.02 HONH2 0.0226 0.150 6.012 0.0075 0.087 5 997 0 0023 0.048 5.985 0.00 0.00 5.97 8..03 + 0.02 HONHCH3 0.0226 0.150 6.002 0.0100 0.100 5.968 0.0034 0.058 5.969 0.0026 0.051 5.983 0.00 0.00 5.96 8.04 + 0.02 HON(CH3)2 0.0626 0.252 5.276 0.0212 0.145 5.246 0.0026 0.051 5.218 0.00 0.00 5.20 8.80 + 0.02 CH30NH2 0.100 0.316 4.674 0.044 0.210 4.652 0.022 0.148 4.639 0.0091 0.095 46o01 0.00216 0.047 4.611 0.00 0.00 4.60 9.40 + 0.02 CH30NHCH3 0.0615 0.248 4.780 0.205 0.143 4.768 0:-00223 0.048 4.752'.0 0.00 4.75 9.25 + 0.02 CH30N(CH3)2 0.0672 0.259 5.746 0.0206 0.144 3.704 0.00223 0.047 3.671 0.00 0.00 3.65 10.35 ~ 0.02

The base strengths of hydroxylamine and methoxyamine have -been reported previously in the literature. The basic ionization constant usually reported for hydroxylamine19'20 is the early value of Winkelbleck,21 6.6 x 10-9, corresponding to a pKb of 8.18. The pKb value, 7.40, given by Ishikawa and Aoki22 appears to be erroneous. The value closest to that found in this investigation (8.03) is that of Kolthoff and Stenger23 who listed the pKb of hydroxylamine as 7.97. The ionization constant for methoxyamine of 2.4 x 10-5 reported by Vodrazka24 corresponds to a pKa of 4.62. This is within the assigned error of the value obtained, pKa 4.60. Discussion.-Brown and co-workers5 have found that the series of methylamines display the same order of base strengths when trimethylborane is used as the reference acid, as they do when one uses the proton. If the inductive effect of the methyl group were the only factor in determining the strength of the base, the order expected would be NH3 <NH2CH3 <NH(CH3)2 < N(CH3)3. The order actually observed with a proton and with trimethylborane as reference acids is NH3 < <.NN(C3 eC-H3' NH(CH3)2. A number of theories have been proposed to explain this order of base strengths of alkyl amines. 2628 The opinion is divided as to whether the inductive effect of the alkyl group is opposed to a steric factor or by solvent interaction. Brown25 has argued that the decrease in the basicity of (CH3)3N is due to a steric strain, operating at the back of the molecule, by the three bulky methyl groups on the small central atom (Bstrain). Trotman-Dickenson26 has noted that the replacement of hydrogens on the amine by alkyl groups reduces the hydration energy for the ammonium ion form, since the bonding between the amine ion and solvent is through the hydrogens of the former. 19. D. M. Yost and H. Russell, Jr. Systematic Inorganic Chemistry. New York: Prentice-Hall, Inc,, 1944, p. 98. 20. Wo Latimer. Oxidation Potentials. New York: Prentice-Hall, Inc., 1938, p. 89, 21. K. Winkelbleck, Z. Phys. Chem., 36, 574 (1901). 22, F. Ishikawa and I. Aoki, Bull. Inst. Phys. and Chem. Research, 19, 136 (1940). 235 I. M. Kolthoff and V. A. Stenger. Volumetric Analysis. Vol. I. 2nd ed. New York: Interscience Publishers, 1942, Table II, p. 284. 240 Z. Vodrazka, Chemo Listy., 46, 208 (1952). 25. H. Co Brown, H. Bartholomay, and M. D. Taylor, J. Am. Chem. Soc., 66, 435 (1944). 26. A. F. Trotman-Dickenson, J. Chem. Soc., 1293 (1949); R. B. Bell and A. F. TrotmanDickenson, ibid., 1288 (1949). 27. R. G. Pearson, J. Am. Chem. Soc., 70, 204 (1948); R. G. Pearson and F. V. Williams, ibid., 76, 258 (1954). 28. Ko S. Pitzer and R. Spitzer, ibid., 70, 1261 (1948). WADC TR 56-318 130

The base strengths of hydroxylamine and its methyl derivatives, which have been measured in this investigation, display the same order as ammonia and its methyl derivatives. In Fig. 1 the pKa values of the methylamine series and the experimental values for the hydroxylamines are plotted as a function of the number of substituents replacing hydrogen on the nitrogen atom. The oxygen atom is somewhat smaller than a methyl group, but it is much larger than the hydrogen atom. Therefore, hydroxylamine is placed in the vertical row with methylamine rather than with ammonia, since the steric relationships would be more nearly comparable. The similarity between the two series of hydroxylamines and the methylamine series is immediately apparent. The inductive effect of the methyl group in going from one to two substituents is small and is zero in the hydroxylamines containing the OH group. In going from two to three substituents there is a large decrease in the base strength for all three series. The electron-withdrawing power of the hydroxyl group is seen to reduce the electron-donating power of the nitrogen by a large amount. The difference between the hydroxyl and the methoxy series is much smaller. On the basis of an inductive effect alone it would be expected that O-methylhydroxylamine would be slightly stronger than hydroxylamine, but the converse is seen to be true. This anomalous effect of the methoxy group is known in other instances. An example is methoxyacetic acid, CH3OCH2COOH (pKa 3.48), which is a stronger acid than glycolic acid, HOCH2COOH (pKa 3.82). The values obtained for these hydroxylamines can be correlated best in terms of the steric-strain model proposed by Brown. If solvation were the important factor, one would expect a different order with the amines containing an OH group, since this hydrogen would contribute to the bonding with the solvent. This does not appear to be the case. Data summarized in Table VI show that the replacement of hydrogen by a methyl group on the oxygen lowers the boiling points and the melting points of the compounds. The removal of the relatively strongly protonic hydrogen on the oxygen eliminates hydrogen bonding through the hydroxyl group. The same effect, though to a much smaller degree, is expected upon replacement of one of the hydrogens on the nitrogen by a methyl group. For comparable compounds the Trouton constant and the boiling point are reduced by substitution of methyl for hydrogen on nitrogen. Although it is frequently assumed that the entropy of vaporization for a liquid under a constant pressure of one atmosphere (Trouton constant) is related to the degree of association in the liquid phase, Hildebrand29 showed that the entropy of vaporization at a constant vapor-phase concentration is of more fundamental significance. The so-called Hildebrand-Trouton constant is the entropy of vaporization at a constant vapor-phase concentration of.O0507 mole/liter. The Hildebrand-Trouton constant for a group of amines is compared with the value of AHvap./T at 200-mm pressure for the same series in Table VIII. It will be noticed that according to the Trouton-Hildebrand constant all the amines except the trimethyl derivatives show a roughly comparable degree of association in the liquid phase prior to vaporization, a fact which may be ascribed to hydrogen bonding. 29. J. H. Hildebrand, J. Am. Chem. Soc., 37, 970 (1915). WADC TR 56-318 131

1200 NH2CH NH( CH3)2 10.00 N(CH3)a NH3 800 pKa 600 NH2H NHCH30H _ N(CH3)20H NHCH30CH NH20CH3 400 N(CH3)20GH3 2.00 NH3 RNH2 R2NH R3N Fig. 1. The pKa values for CH3, OH, and OCH3 substituted amines. WADC TR 56-318 132

TABLE VIII A COMPARISON OF THE "TROUTON CONSTANT" AT A PRESSURE OF 200 MM WITH THE HITDEBRANDT-TROUTON CONSTANT AT A UNIFORM VAPOR CONCENTRATION OF 0.00507 MOLE/LITER FOR A SERIES OF AMINES Temp. for AHvap. Compound AHvap. T" AAvapo Vap. Cone. P for at AH A T of00507 Vap Conc. This Temp. T Ref. calmo.le moleiter mm cal/mole NH3 5,860 216 27..1 200 63 6,041 30.2 (a) CH3NH2 6,770 240 28.2 224 72 6,860 30.5 (b) (CH3)2NH 7,100 252 28.2 236 76 7,330 31.1 (c) (CH3)3N 6,o090 245 24.8 227 72 6,150 27-0 (c) N2H4 10o,230 351 29.1 337 107 10,280 30.5 (d) NH20H 16,600 361 46.0 352 106 17,d266 49.0 (a) NH20CH3 8,400 290 29.0 273 84 8,483 31.0 (e) H20 10,070 340 29.6 325 105 10,202 31.3 (a) CC14(Ref.) 7,680 311 24.7 294 48 7,938 27.0 (a) (a) D. R. Stull, Ind. Eng. Chem., 39, 540 (1947). (b) Aston et al., J. Am. Chem. Soc., 59, 1743 (1937). (c) T. E. Jordan. Vapor Pressures of Original Compounds. New York: Interscience Publishers, 1954, PP. 176 and 178. (d) L. F. Audrieth and B. A. Ogg. Hydrazine. New York: John Wiley and Sons., Inc., 1951, P. 74. (e) This investigation. The value of 49.0 e.u. calculated from the literature data for the vapor pressure of hydroxylamine deserves further consideration. In comparison to the corresponding constants for the closely related water, hydrazine, ammonia, and O-methylhydroxylamine (Table VIII), this value is unusually high. In view of the instability of free hydroxylamine above 100C, it is almost certain that the high value of AS indicates. some decomposition of the sample during the vapor-pressure measurement. Such an error in vapor pressure would result in a low value for the boiling point. The data for the more stable methyl derivatives of hydroxylamine suggest that the true boiling point of hydroxylamine is indeed above the recorded value of 111C. For Nmethylhydroxylamine the value is 115'C and for the! NN1i -dimethylhydroxylamine the value is 100.60C. These values suggest 125~C as a more reasonable boiling point for the pure H2NOH. WADC TR 56-318 15533

B. THE REACTION OF HYDROXYLAMINE AND ITS N-METHYL DERIVATIVES WITH DIBORANE (Do H. Campbell, T. C. Bissot, and R. W. Parry) Abstract Hydroxylamine-borane and N-methylhydroxylamine-borane have been prepared in impure form as solids at -1120C. These lose H2 on warming to room temperature. Pure N,N-dimethylhydroxylamine-borane has been prepared as a volatile liquid, which is relatively stable at room temperature. The decomposition of the borane complex is catalyzed by B2H6 and by the decomposition residues. The role of B2{6 in the decomposition.$is interpreted in terms of a proposed reaction scheme. Many of the reactions of the boron hydrides are assumed to occur through the addition of a Lewis base to the borane group followed by internal decomposition of the adduct. One of the mechanisms suggested for the hydrolysis of diboranel illustrates the transitory existence of the unstable addition complex: B26 - 2BH3,HHJ H3B + H20 - HBOH H l HBOH + H2, etc. H The internal decomposition of such borane adducts is a-question of some interest, yet many of these addition complexes, suggested as intermediates, are too unstable to make their direct study practical. In contrast, the N-methylhydroxylamineboranes seemed particularly suitable for such a study since the boron-nitrogen bonds should be of moderate stability and secondary reactions such as oxidation of the boron by the hydroxylamine and loss of hydrogen through interaction of protonic and hydridic hydrogens in the molecule should proceed with some ease. In addition, a study of the hydroxylamine addition compounds was of some practical interest because of the possibility of B-N-O polymer formation. The formation and decomposition of the borane adducts of hydroxylamine and of its N-methyl derivatives have been studied and the results are interpreted here in terms of a reaction mechanism for complex decomposition. The Reaction of Diborane and Hydroxylamine. -Diborane and solid anhydrous hydroxylamine gave no evidence for reaction over the temperature range -186~ to -96~C; however, after the solid and gas had been standing in contact for several hours at -960C, the system was warmed up slightly and explosive interaction occurred. 1. H. G. Weiss and I. Shapiro, J. Am. Chem. Soc., 75, 1221 (1953). WADC TR 56-318 134

In-order to moderate the reaction, diethyl ether was used as a reaction medium.2 Diborane reacted very slowly with a stirred ether suspension of anhydrous hydroxylamine at -1120C without the liberation of significant amounts of hydrogen. Under these conditions the apprpximate stoichiometry of the process was given by the equation n B2H6 + 2n NH20H >- 2(H3BNH20H2) n In every -un the ratio NH20H/B2He6 deviated somewhat from the theoretical value of 2 and the deviation was always larger than the experimental error of measurement. This was a general observation. applicable to bhydroxylamine and both of its N-methyl derivatives but not to the 0-methyl hydroxylamines.3 See Table I. This fact is considered later. TABLE I SUMMARY OF THE PROPERTIES OF THE BORANE ADDITION COMPOUNDS OF EYDROXYLAMINE AND ITS N-METHYL DERIVATIVES Ratio Approx. Temp. at Compound Melting Volatility Amine/B2H6. which 50% Decomp. Point Point..in Prep.,* in 24 hr, 0C HONH2BHs3 Solid at -78~ None (Decomp.) 1.80 -75 (impure) HONTfECH3BH3 Liquid* at -780 None l 92 15 (impure) HON (CH3)2BH3 2~ - 4~ 6 mm at 250 1.92 25 (impure) 55 (pure) * Sample impure; low melting point may be due to impurities, * Each value is a result of -several runs. The theoretical value is 2. If the temperature of the ether insoluble hydroxylamine adduct of diborane was to rise slowly above -112~C, hydrogen was liberated, probably through the interaction 2. The advantages of diethyl ether as a solvent in some boron hydride reactions have been demonstrated previously. Emeleus and Stone [J. Chem. Soc., 840-1 (1951)] failed to obtain exact stoichiometry for the reaction between diborane and nonsolvated hydrazine, but Steindler and Schlesinger [J. Am. Chem. Soc., 75, 756 (1953)1 obtained the-compound HsBNEI2NH2BH3 with excellent stoichiometry in the presence of diethyl ether. 3. T. C. Bissot, D. H. Campbell, and R. W. Parry, this Report, p, 147. WADC TR 56-318 135

of the hydroxyl hydrogen of hydroxylamine and the hydridic hydrogen of a borane group. The total amount of hydrogen liberated at various temperatures is shown in Fig. 1. The moles of hydrogen lost per mole of hydroxylamine used approached 1 at about 25~C and 2 at about 125~C. The product remaining after the loss of one mole of hydrogen was a white solid which dissolved in water, methanol,. and ethanol with the evolution of small amounts of hydrogen. The material was insoluble in liquid ammonia, ether, benzene, toluene, and 1l4-dioxane. The product remaining after the loss of two moles of hydrogen dissolved only with difficulty in water, but readily in dilute acid or base. The above facts suggest the formula (RH HH\ -BNOBNO) \HHH/ In for the 25~C decomposition product; however, it is difficult to harmonize the above formulation with the data for hydrogen evolution at different temperatures shown in Fig. 1. One would expect that the temperature - hydrogen-evolution curve should show a break after the loss of one mole of hydrogen per hydroxylamine-borane The curve shows a break after loss of only 2/3 of a mole of H2 per HONH2BH3. This anomaly is attributed to the fact that increasing rigidity of the polymer prevented free interaction of the acidic and hydridic hydrogen in the addition complex. This postulate receives support from the observation that in a similar type of decomposition process the presence of a polyglycol ether solvent eliminated the change in rate after loss of 2/3 of a mole of hydrogen.3 As the temperature was raised, the stoichiometry was complicated by the interaction of the protonic hydrogens attached to the nitrogen and the hydridic hydrogens of the- borane group. Finally, direct pyrolysis of the borane group added further complication at temperatures above about 700~C The Reaction of Diborane and N-methylhydroxylamine. —This reaction, conducted only in ether, resembled the reaction of diborane and hydroxylamine except for minor points. The addition product of the N-methylhydroxylamine was a clear ether soluble liquid at -78~C. Above -78~ hydrogen was lost, and the liquid increased in viscosity until the mass solidified near room temperature. No volatile products other than hydrogen were observed until the temperature was raised to 100~C and above. At this temperature methylamine contaminated with traces of ammonia was obtained. About 19% of the total nitrogen in the complex was recovered as methylamine. The polymerization upon loss of hydrogen parallels closely the process described previously for the hydroxylamine-boraneo The residue remaining in the tube after pyrolysis at 220~C was a yellow solid. Upon hydrolysis with dilute HC1 enough additional hydrogen was obtained from the residue to bring the total "hydridic" hydrogen per molecule of complex up to two instead of the expected three. The recovery of methylamine at high temperatures and the loss of a hydridic hydrogen from the boron indicate a shift of the:-oxygen group to the boron and of a hydrogen to the nitrogen. This shift; is considered elsewhere.3 WADC TR 56-518 136

3.6 3.4 3'2 /act 3.20 2.8 2.6 2.4 i2.2 1.4 1.2 - - 120 -80 - 40 0 40 80 120 160 TEMPERATURE, ~C Fig. la. Mole ratio of hydrogen to H2NOHBH3. 1.4 I. HONH2 BH3 2. HONHCH3BH3 1.2 3. HON(CH3)2B H 3 5. CH30NH2BCH3H3 r e I. T.2 0 -120 -80 -40 0 40 80 Temperoture ~C Fig. l. Decomposition curves of hydroxyl- and methyl-substituted hydroxylamine boranes. WADC TR 56-318 137

The Reaction of NN-Dimethylhydroxylamine and D1iboraneo -This addition reaction was conducted with and without ether as a solvent. In contrast to the observations with hydroxylamine, the reaction without a solvent was not explosive and gave stoichiometry comarable to that observed in the presence of ether. As in the two preceding cases, the ratio of NOH(CH3)2 to B2H6 deviated from the expected valve of two; fortunately, however, the isolation of the pure:N,N-dimethylhydroxylamine-borane and a study of its properties gave a reasonable explanation for the observed deviations in the entire series. Tn the decomposition of the two previous compounds no volatile product other than hydrogen separated at temperatures below 1000t. In contrast t'he pure N,N-dimethylhydroxylamine-borane could be distilled from the reaction. mixture at 25~C as a liquid. The liquid had a vapor pressure of 6 mm at 25~C, a melting point of 2 to 4C, and a molecular weight lby vapor density at room temperature of 70 + 6. (Theoretical for HON(CH3)2BH3 is 74.9~) The small weight of the sample, resulting from its low vapor pressure at 25~, accounted for the large experimental uncertainty. This uncertainty plus the known association of N IN-dimethylhydroxylamine in the vapor phase prevented the use of the pressure measurements for evaluating the degree of dissociation. of -the complexo The followin analyti.cal results were obtained for the pure N N-ddimethylhydroxylamine-boraneo Observed Theoretical ydrd: l.dic yl orgen 3.92 4.04 iboron 14o4 i4.44 Nitrogen 1806 18o69 When the pure complex was distilled from the rapidly' decomposing system at room temperature, it turned o-ut to be surprisingly stable. A pure sample standing at 250~n for five da~ys was less than 5% decomposed If, however, the pure liquid was returned to the tube contain-ng the initial decomposition products, rapid evolution of hydrogen began immediately at room temperature. The yield of pure HOIN(CH3)2BH3 obtained from the early preparations, in which excess diborane was used, was found to correlate with. -the amount of undecomposed compo-und calculated from the observed hydrogen evolution. TAhe results are shown in Table IR,* When extra diborane was added to the pure compound, vigorous hydrogen evolution began. Data are summarized in Table ~:YTI: Although small amounts of diborane were used up in the reaction:, most of it was recovered and at least 12 moles of hydrogen were liberated for every mole of diborane consumed. The high ratio indicates that the effect of the diborane was definitely catalytic and no primary stoichiometric process including diborane was involved. On the other hand, the fact that small amounts of diborane were used up I.n a secondary process justifies the poor stoichiometry in all the reactions of diborane and the hydroxylamines. TChe observation that excess B2R6 catalyzes the decomposition suggested that better results would be obtained. if an excess of amine were used rather than an excess of diborane. The prediction was verified. by experiment~ When the amine was WADc TR 56-31.8 138

TABLE II REACTION BETWEEN N:~N-DIMETHYIHYDROXYLAMINE AND EXCESS DIBORANE Run Number Solvent 1 2. Ether None Ether Ratio HON(CH3)2/B2H6 1.94 1.91 1.91 H2/HON (CH3)2 o.43 0.72 0.32 % Complex Not Decomposed* 57 28 68 ~ Pure Complex Recovered 50 30 6 *Based on hydrogen evolution. TABLE III CATALYTIC DECOMPOSITION OF PURE NNDIMET~HYLYDROXYLAMINE-BORANE BY DIBORANE AT 25~C Trial 1 Trial 2 Mmoles B2H6 added per mmole of pure HON(CH3)2BH3.78.14 Moles H2 liberated per mole of pure HON(CH3)2BH3 0.82 (0.5 hr) 1.07 (28 hr) % added B2H6 which was recovered unchanged 91 35 Mmoles B2e6 consumed for each mmole of pure HON(CH3)2.07.09 BH3 decomposed Mmoles H2 released for each mmole B2H6 consumed 16 12 in excess, hydrogen evolution, prior to separation of the pure complex, corresponded to only 2.3% decomposition; when diborane was in excess, hydrogen evolution under comparable conditions corresponded to from 50 to 70% decomposition. Unfortunately, the NN-dimethylhydroxylamine and its borane adduct have vapor pressures which are too close together to permit easy vacuum-line fractionation. Due to mechanical purification losses, only a 70% yield of the pure NN-dimethylhydroxylamineborane was recovered, although better separation procedures would undoubtedly have given higher yields. These observations suggest that pure hydroxylamine-borane and WADC TR 56-318 139

N-methylhydroxylamine-borane might be prepared as stable compounds if a way could be found to minimize contact between the complex and the excess diborane. Unfortunately, the fact that these two amines are solids in ether at the reaction temperature means that diborane must be present in excess since separation of the unreacted amine from the adduct is difficult at low temperatures. The thermal decomposition of the N,N-dimethylhydroxylamine -borane was studied by holding a sample in a sealed tube at 55~C. The liquid changed directly into a white solid without going through various stages of increasing viscosity. Upon opening the tube exactly one mole of hydrogen per mole of complex was found. The formula indicated for the white solid which remained is CH3 H N B CH3 H Since no suitable solvent was found, the value of x was not determined. The [ON(CH3)2 BH2.] polymer underwent further decomposition upon heating to high temperatures. A sample of HON(CH3)2BH3 upon prolonged pyrolysis at 100~ yielded hydrogen equivalent to 1.25 moles H2 per mole HON (CH3)2BH3 plus small amounts of a high-boiling liquid. The type of thermal decomposition which might be expected is explained by the equation 3[-ON (CH3)2BH2-] -1B [N(CH3)2]3 + 3H2 + B203 The coFpound B[N(CH3)2:]3 is reported as a liquid with low volatility at room temperature. In contrast to the behavior of the N-methylhydroxylamine-borane there was no evidence for a shift of oxygen group from nitrogen to boron and of a hydridic hydrogen from boron to nitrogen in the thermal decomposition of the N,N-dimethylhydroxylamine borane. A sample of the N,N-dimethylhydroxylamine-borane was allowed to react with a small excess of trimethylamine. An 80% yield of (CH3)3NBH3 was obtained, and the amine displaced was identified as N,N-dimethylhydroxylamine. Discussion. —The thermal decomposition of the N,N-dimethylhydroxylamine-borane was different in several respects from that of HONH2:BH3 and HONCH3HBH3. In the decomposition of the latter two compounds there was a range of about 100~C between the temperature at which a significant amount of hydrogen was first produced and the temperature at which one mole was collected. Furthermore, a change in rate of H2 evolution was noticed after loss of 2/3 of a mole of hydrogen but not after loss of one mole. On the other hand, pure HON(CH3)2BH3 was relatively stable at 25~C but decomposed rapidly when the temperature was raised just 300. In addition, the rate 4. E. Wiberg and K. Schuster, Z. anorg. Chem., 213, 77 (1933). WADC TR 56-518 140

of hydrogen evolution decreased almost to zero at 550C after the expected one mole of hydrogen had been liberated. This difference can be attributed to" the volatility of the NN-dimethylhydroxylamine-borane and -.tO."~:.the fact that the pure liquid product and the decomposition product do not appear to be mutually soluble. Since they separated as distinct phases, the polymerization which accompained the -loss of hydrogen did not impede the decomposition of the unreacted material, Furthermore, the decomposition of the dimethyl compound was not complicated by the loss of the hydrogen attached directly to nitrogen,. as was the case with the other two complexes. When carefully purified, HON(CH3)2BH3 was'amazingly stable for a compound containing a protonic and hydridic hydrogen in the same molecule. The role of B2H6 in promoting H2 evolution suggests a reasonable explanation for the stability of this pure material. In general, the coordination of the BH3 group to a base reduces the hydridic character of the attached hydrogens, the degree of reduction being almost proportional to the strength of the coordinate link which is formed., For example, a weakly complexed addition compound such as-F3PBR3 or'OCBH3 is almost as reactive as diborane itself and inflames spontaneously in moist air. In contrast, the strongly complexed trimethylamine-borane is stable indefinitely in dry air and is only slowly hydrolyzed by liquid water. on HON(CH3)2BH3 it is logical to assume that the hydridic hydrogens of the borane group are sufficiently deactivated through coordination to prevent their interaction with the weakly acidic hydrogen of the -OH group. However, the unstabilized hydrogens on the diborane are basic enough to react readily with this protonic hydrogen. After formation of the B-O-N bond, the BH3 which was originally attached to the amine could be released to attack another molecule or to reform B2H6: CR3 H C BH + HON BH3 -- H2 + B-O-N-BH3 H CH3 H CH3 H CH3, B-O-N- + BH3 H CH3 It would be logical to assume that the first step in the above process is the coordination of the BR3 to the OH group of the amine, though this is not proved. An explanation for the fact that a slight excess of diborane was always consumed in the reactions of the hydroxylamines with diborane can be based on the above mechanism. It is probable that this process occurs to some extent during complex formation and the second step of the interaction, involving release of the BR3 group, is not always quantitative.'Excellent stoichiometry was obtained in the reaction of diborane and the O-methylhydroxylamines where the above process would be blocked.3 It was also noticed that loss of hydrogen from the pure methoxyamine boranes was not catalyzed by diborane, an observation which is consistent with the above mechanism. The hydroxylamine boranes do not have to shift -any groups to form a boron-oxygen bond. As indicated above, the B-O bond arises directly from the interaction of the protonic hydrogen of the hydroxyl group and the hydridic hydrogen of the borane. This reaction appears to take place smoothly as' the temperature is raised and no true exWAIDC TR 56-318 141

plosions were ever encountered in heating a hydroxylamine-borane after the addition compound was once formed. A sample of the N,N-dimethyl derivative was heated directly up to 100~Co The explosion of the diborane-hydroxylamine mixture at low temperatures can be attributed to sudden vigorous interaction of previously unreacted condensed phases or perhaps to impurities. Hydroxylamine itself will detonate if subjected to sudden intense heating, but the free base is much less subject to detonation if coordinated.3 In Table I are summarized the approximate temperatures at which 50% of the hydroxylamine-borane adducts are decomposed through loss of hydrogen from the original reaction mixtures. The data suggest that the OH hydrogen in hydroxylamine is the most acidic, that in the N-methyl is next, and that in the N,N-dimethyl derivative is least acidic. This interpretation would be consistent with the assumed inductive effect of the methyl groups, The displacement of N,N-dimethylhydroxylamine by trimethylamine is consistent with the fact that the latter is the stronger base.5 Experimental. - a. Materials..lo Hydroxylamine, preparation and handling-Free hydroxylamine was prepared by the reaction of sodium butoxide with hydroxylammonium chloride suspended in butanol.6 The free base decomposes rapidly above 15oC and it picks up water and carbon dioxide avidly. These properties necessitated the use of the special cooled weighing bottle shown in Fig. 2. In such a device the amine was kept cool, yet weighable. The plate-like hydroxylamine crystals were stored in anhydrous butanol at -10~C until needed; they were then filtered off, washed rapidly with cold ether, sucked dry, and transferred to the cooled weighing bottle (Fig. 2)> After weighing, the crystals were dumped into the reaction tube, which was immediately cooled with ice-water; the tube was rapidly attached to the vacuum line and e-vac-uated. The special weighing bottle was tared.. In order to determine the accuracy which can be obtained by handling the hydroxylamine in this manner, weighed samples of the free base were titrated with standard acido Results on several samples indicated that the weight of the sample was reproducibly about 8% above the theoretical weight obtained from the titration. The discrepancy is undoubtedly due to ether adhering to the crystals. An 8% correction was applied to the weights of all samples handled in the above manner. 2. N-methyl and NN-dimethylhydroxylamines-The preparation and properties of these free bases are described elsewhere.5 5. T. C. Bissot and R. W. Parry, this Report, p. 118. 6. C. D. Hurd in Inorganic Synthesis. Vol. I. New York: McGraw-Hill Book Co., Inc., 1939, p. 87. WADC TR 56-318 142

RUBBER COLLAR I''- lIWEIGHING /~~ BOTTLE 2.751IN. SMALL DEWAR FLASK CRACKED ICE Fig. 2. Special weighing unit for hydroxylamine. WADC TR 56-318 143

3. Diborane —Diborane was prepared from LiAIH4 and boron trifluoride etherate as described elsewhere 7 b. General Procedure for Compound Formation. The generalized procedure for conducting reactions is described below. Some minor deviations from this procedure were used in specific cases; such details are available in dissertation formii.-89 1. Diborane in excess-The reaction vessel was a 20-mm Pyrex tube, about 8 inches long, attached to the conventional high-vacuum system by means of a 24/40 joint. The tube was equipped with an automatic, electromagnetically operated, plunger-type stirrer, In a typical run anapproximately 2-to-6mmole sample of the amine was weighed out in. an evacuated micro weighing tube or in the previously described chilled weighing bottle. HON(CH3)2 and HON(CH3)H were vacuum distilled directly into the reaction vessel from the weighing tube; hydroxylamine was transferred as described earliero A l-to-6 —ml aliquot of anhydrous diethyl ether, which had been dried for several days over CaH2, was distilled into the reaction tube and the system was stirred until a solution of uniform suspension was obtained. (When no solvent was used, this step was omitted.) The system was then frozen wi.,th liquid nitrogen and a carefully measured amount of diborane, about 10 to 20% in excess of the stoichiometric amount, was distilled into the tube. The system was then warmed to -112~C by means of a CS2 slush bath and stirring was continued for the required period. The free hydroxylamine, which was insoluble in ether, reacted very slowly and required up to 20 hours for nearly complete reaction. The more ether soluble methyl and dimethyl derivatives gave complete reaction in two hours or less at -1120C. The system was frozen with liquid nitrogen; noncondensable gases were pumped off with a Toepler pump and passed through a packed trap cooled with liquid nitrogen. The total amount of gas was measured and was identified by a molecular-weight determination. The temperature of the original reaction vessel was raised to -78~C; H2 was collected until its rate of evolution approached zero, then it was pumped off and measured; the excess diborane and the ether solvent were distilled out and very carefully fractionated in the vacuum line. The recovered diborane was measured and the amount used in the reaction was calculated. Because of the extremely slow reaction between dib-orane and hydroxylamine, some question as to the stoichiometry arose. For this reason the rate of absorption of B2H6 by an ether suspension of hydroxylamine was studied at -112'Ci 7. I. Shapiro, H. G. Weiss, M. Schmick, S. Skolnik, and G. B. L. Smith, J. Am. Chem. Soc., 74, 901 (1952). 80 To C. Bissot, doctoral dissertation, Univo of Mich., Ann Arbor, 1955. 9. Do H. Campbell, doctoral dissertation, Univo of Mich., Ann Arbor, 19553. WADC TR 56-318 144

At definite time intervals unreacted 3B2H6 was distilled from the isystem, purified, measured, and returned to the reactor. Quantitative removal of B2H6 from an ether solution at -112'C was a slow and difficult operation; hence, a standard separation procedure was used which gave almost complete recovery and when applied to an ether solution of diborane would give an apparent rate of B2H6 absorption somewhat greater than the actual rate, but the ultimate stoichiometry would be unaffected. Data plotted in Fig. 3 show that the rate of complex formation is slow, but the ratio of B2H16 to INH20H approaches the theoretical value of 0O5 as a limit. 2. NIN-dimethylhydroxylamine in excess. —In a typical run a 5-mmole sample of HON(CH3)2 was weighed out and dissolved in 3 ml of diethyl ether. After adding 2.39 mmoles of B2H6, the system was warmed to -112~C and stirred for 45 minutes. The mixture was separated by distillation through a trap at -63eC which retained the compound and the excess amine. In this preparation only 0.11 of a mmole of H2 was obtained.~ The excess amine and the borane complex were separated by fractionation n n the vacuum system. c. Thermal Decomposition of the Borane Adducts. After the excess B2H6 and solvent separated at -780C, the temperature of the complex was raised to.640C; after the rate of hydrogen evolution dropped almost to zero, the gas was removed, purified, and measured. This procedure was repeated at regular temperature intervals up to 220~C. At some of the higher temperatures volatile products such as methylamine or pure N,N-dimethylhydroxylamine-borane separated. These were trapped and purified in the vacuum system and identified by physical and chemical properties. d. Analytical Methodso 1i Hydridic hydrogen-Determined'by heating the sample for several hours in a sealed tube with 6 N HC1. The hydrogen gas evolved was measured and identified in the vacuum system. 2. Boron-A modification10 of the identical pH method of Footell was applied to the microdetermination of boron -;in the presence of nitrogen compounds. The acid solution from hydrolysis was adjusted to a pH value very close to 7; the solution was then saturated with mannitol and the boric acid complex was titrated by bringing the system back to the same pH of 7. Excellent results were obtained when the method was applied to known standards. 3. Nitrogen-Nitrogen was usually determined by the micro Kjeldahl method. The hydroxylamines were reduced to ammonia by digestion with glucose in the presence of a K2SO04-CuS04 catalysto 10. R. W. Parry, D. H. Campbell, D. R. Schultz, S. GO Shore, T. C. Bissot, and R. C. Taylor. The Chemistry of Boron Hydrides. Engo Res, Insto, Univ. of Micho, Annual Report Project 1966-1-P (1954). 11. F. J. Foote, Ind. Eng, Chem. Anal. Ed., 4, 39 (1932). WADC TR 56-318 145

3Ha 0.3 N H-OH 0.5 0.4 B2H6 0.3 _ 0.2 0 2 4 6 8 10 12 14 16 18 20 22 TIME (HRS.) Fig. 3. Reaction of diborane with hydroxylamine at -1120C.

C. THE REACTION OF O-METHYLHYDROXYLAMINE AND ITS N-METHYL DERIVATIVES WITH DIBORANE1 (T. C. Bissot, D. H. Campbell, and R. W. Parry) Abstract The borane adducts of O-methyl, O,N-dimethyl, and O,N,N-trimethyl hydroxylamine have been prepared. Their thermal decomposition and hydrolysis have been studied. A mechanism for the shift of a methoxy group from nitrogen to boron and of a hydrogen from boron to nitrogen is suggested. In another publication2 the reaction of diborane with hydroxylamine and its Nmethyl derivatives was considered. Structural and mechanistic arguments suggested that the chemistry of O-methylhydroxylamine-borane, and its N-methyl derivatives, (CH30NR2BH3, R = H or CH3), should differ appreciably from that of hydroxylamineborane and its N-methyl derivatives, (HORa2BH3). The preparation, properties, and chemistry of the 0-methyl -hydroxylamSi.ne boranes are considered herein. HH O-Methylhydroxylamine -Borane, CH3ONBH a. Preparation and Characterization. In the presence of diethyl ether pure O-methylhydroxylamine and pure diborane react to produce 0Ome-thylhydroxylamirnne-borane. The observed stoichiometric ratios of CH30RHp to B2}E6 for five separate runs were in the range 1.98-201. In the absence of ether the stoichiometry was poor in some cases and hydrogen gas was evolved at low temperatures. The pure white solid addition compound melts sharply at 550~C with rapid evolution of hydrogen gas. Its solubility in ether is appreciable at room temperature but falls off as the temperature is lowered. Some of the solid could be sublimed slowly and with difficulty at 40~C in high vacuum. Decomposition was a competing process. After hydrolysis of CH30:NIIBH3 with dilute hydrochloric acid, the following analytical data were obtained: 1-hydridic hydrogen: obs. 4.63%, theor. 4.96%; borono obs. 17.8%, theor. 17-77%.o The amount of O-methylhydroxylamine hydrochloride in the hydrolysis product was estimated roughly from a titration curve on an aliquot. At least 93% of the nitrogen was still present as CH30NH2. When an ethereal solution of Cf1OI3OH2BH3 was treated with a slight excess of BF35 no B2'i was released. Tn view of earlier mechanistic arguments, it is significant to note that B2BI, which catalyzed hydrogen loss from pure Nmethylhydroxylamine-borane, did not catalyze the decomposition of pure O-methylhydroxylamine -borane. Attempts to displace CH30NH2 from the complex by the stronger base3 N(CH3)3 gave only a 22% yield of (CH3)3NBH3. A competing reaction was loss of hydrogen from the complex, a process which appeared to be catalyzed by the presence of trimethylamineo Thirty-three percent of the original 1. Abstracted fromn doctoral dissertations submitted to the Horace H. Rackham School of Graduate Studies of The University of Michigan by T. C. Bissot and D. H. Campbell. 2. D. H. Campbell, T. C. Bissot, and R. W. Parry, this Report, p. 134. 3. T. Co Bissot and R W Parry, this Report, p. 118. WADC TR 56-318 147

complex decomposed through hydrogen loss. The foregoing observation was of importance in interpreting data obtained in early attempts to synthesize CH301\12BH3s In early synthetic trials the adduct lost hydrogen at low temperatures and a viscous liquid remained after the ether was removed. It was found that a small amount of ammonia, contaminating the O0-methylhydroxylamine, was responsible for the low-temperature hydrogen loss; very pure O-methylhydroxylamine gave a relax tively stable borane adduct.' The ammonia was surprisingly effective; a mixture containing nine mole percent NH3 with ninety-one mole percent CH30NH2 reacted with diborane in ether to give off hydrogen such that the ratio H2/CH30NH2BH3 (theor.) was 0.48 after one hour. After low-temperature hydrogen loss, varying amounts of a weakly complexed, approximately one-to-one mixture of diethyl ether and k-monoamino-diborane, H2NB2H5, were isolated from the system. The production of H2N:B2H5 by such a low-temperature reaction is unexpected. The foregoing data suggest that 0-methylhydroxylamine is effective in catalyzing low-temperature hydrogen loss from the ammonia-borane addition compounds in the presence of excess B2H6. H2NB2H5 is a product.4 On the other hand, both ammonia and trimethylamine were effective in catalyzing hydrogen loss from CH30NH12BH3. This was the only 0- or N-methylhydroxylamine-borane which was shown to be unstable in the presence of these bases. The mode of base action is not clear. b. Pyrolysis of O-Methylhydroxylamine-Borane. The thermal decomposition of CH30NH2BH3 occurred in two ways. If it was heated rapidly to approximately 90~C, a violent detonation resulted. If it was heated slowly, hydrogen was evolved at a slow rate until the solid melted at 55~C0 The rate of H2 evolution increased markedly when the liquid phase appeared; within ten minutes or less at 550C the amount of hydrogen evolved was such that the ratio H2/CH30NH2BH3 amounted to 0.66. At this point the liquid became increasingly viscous and the rate of hydrogen.eolution dropped sharply. After four days at 55~C the last one-third of a mole of H2 was obtained. During heating at 55~C a complex solid mixture of NH3, B(OCH3)s, and in some cases CH30H sublimed from the reaction vessel. The chief component of the mixture was the weakly associated solid H3NB(OCH3)3s In the vapor phase the complex dissociated completely to H3N and B(OCH)3. The volatile solid always contained a slight excess of boron over nitrogen. Hydrogen evolution from the original viscous mass did not stop at one mole but continued at a slow rate. After the loss of one mole-of hydrogen and the evolution of considerable trimethylborate from the system, the residue could be heated safely above 90~C. The viscous mass changed slowly into a white solid as the temperature was raised. After heating to 300~C, the residue was analyzed. The -empirical formula for the solid was BN1.12(0CH3)0.oiHo.3 It is important to 4. All variables important in obtaining good yields of H2NB2H5 from mixtures of NHs, B2H6, and CH30NBH2 were not studied systematically. Yields of HaiBaH5 were rather erratic, varying from good to poor. Catalysis of hydrogen loss was definite o WADC TR 56-518 148

note that more than 90% of the original methoxy content of the compound was sublimed from the system, largely as trimethylborate. The sharp break in the rate of hydrogen evolution after loss of only twothirds of one mole of hydrogen was difficult to correlate with structural or mechanistic arguments for the process. Accordingly, pure O-methylhydroxylamineborane was prepared, and its rate of decomposition at 55~C was studied in solution as a function of dilution. Diethyleneglycol dimethylether was chosen as a suitable solvent. Figure 1 shows decomposition rate data for (1) the pure compound, (2) a 50% solution of the complex in the polyglycol ether, and (3) a 10% solution. The previously unexplained sharp break in rate after loss of only twothirds of a mole of hydrogen disappeared when the solvent was introduced. Such an observation suggests that the break is associated with the rapidly increasing viscosity of the decomposing mass. Reduced mobility of interacting groups reduces the rate of gas evolution. More detailed treatment of the data given below lend further support to such an argument. The rate data of Fig. 1 can be used to differentiate between intramolecular and intermolecular loss of hydrogen from the complex. If the liberated hydrogen were produced by an intramolecular reaction, the process would be first order and one mole of hydrogen would be evolved per mole of borane adduct: H H CH30s /H CH30NBH — > H2 + N-B -,HH H H The resulting unsaturated product would be expected to polymerize. If the hydrogen were produced by an intermolecular process, the kinetics should be higher than first order, but the stoichiometry would be the same as for the first-order process: CH3 CH3 CH3 1H - OH 0 H O H CH3NBi+ HNBH + H2 + H-N- B-NBH HH HH H H HH Further reaction between the monomer and the dimer or polymer should give long chains. Both intra- and intermolecular processes would be complicated by the partial loss of a second mole of hydrogen, since hydrogens are still available on boron and nitrogen. Crosslinking of chains would thus be expected. In Fig. 1 the fraction of complex decomposed, x/a is plotted as a function of time where x is the number of moles of hydrogen at time t and a is the initial number of moles of CH3ONH2BH3. The first-order rate equation can be converted to the form d(x/a) = k(a-x) dt a From this form it is apparent that the fraction decomposed at any time would be WADC TR 56-318 149

co3 \31 H 1.0 x8 0 ~ ~ ~ ~ ~ ~ ~ ~ ~~0 8 aa I0 z \31 o oo jC~~~~~~~~~~~~~~~~U H) \,Y~~~~~~ to.4 X N O O.2 ~~~~~~~10 20 30 4 tO iTime 20 iours 30 40 Fig. 1. Decomposition of NH20CH3'BH3 at 550C.

independent of the volume of the solution if the reaction were first order. The data in Fig. —'-~ thus provide strong evidence against intramolecular loss of hydrogen under the conditions used. A simplified treatment of the solution curves in Fig. 1 gives an apparent order of about 2 for the process. The intermolecular decomposition indicated by the rate data produces polymerization, a result which is consistent with the increasing viscosity of the residue as hydrogen is lost. Increasing viscosity of the system would be expected to retard an intermolecular process but not an intramolecular process. c. Hydrolysis of 0-Methylhydroxylamine-borane..Acid hydrolysis of CH3ONH2BH3 gave the expected hydrochloride salt of the base along with boric acid; in contrast, basic hydrolysis of the complex gave reduction of the O-methylhydroxylamine to ammonia and only two of the expected three hydridic hydrogens of the borane group. A shift of hydrogen from boron to nitrogen and of a methoxy group from nitrogen to boron took place in basic solution. The shift is comparable to that observed in the thermal decomposition proces s. CIg H 0,?N-Dimethylhydroxylamine -Borane, CH30NBIH.HH a. Preparation and Characterization. O,N -dimethylhydroxylamine will react with diborane to produce HH CH30N BH CH3H Ether was acceptable but not necessary as a solvent. In contrast to the case of 0-methylhydroxylamine, excellent stoichiometry and a pure product resulted from the interaction of diborane and the free, nonsolvated base. See Table I. The liquid H CH3ON BH3 CH3 froze at about -220C. Vapor pressure at room temperature could not be measured with the mercury manometer on the system, but the liquid could be distilled at room temperature under high vacuum without decomposition. Analysis gave a B/N/ hydridic H ratio of 1/0.97/2.92. Upon hydrolysis with hydrochloric acid the original amine was recovered and identified as the hydrochloride salt. When an-ether solution of CH30NHCH3BH3 was stirred overnight with an excess of N(CH3)3 at 0~C, an 86% yield of (CH3)3NBH3 was recovered. There was no hydrogen evolution in thisdisplacement reaction, as there was in the similar experiment with CH30Nta2BH3. WADC TR 56-318 151

TABLE I SUMMARY OF THE PROPERTIES OF THE BORANE ADDITION COMPOUNDS. OF HYDROXYLAMINE AND ITS METHYL-SUBSTITUTED DERIVATIVES' Ratio Approx. Temp. Melting at -Which Compound Volatility Amine/B2H at Which Point in Preparation 50: Decomp. in: 24,- hr-,~C HONH3BH3 Solid decomp. Nonvolatile 1.80 -75 HONHCH3BH3 Liquid Nonvolatile 1.92 15 HON(CH3)2BH3 2~ -4 6 mm 25~ 1.92 25 as prep. 55 pure CH30NH2BH3 550 Slightly 2.00 55 volatile 40~ CH30NHCH3BH3 -230'to.21 Slightly 1 97 65 volatile 25 ~ CH30N(CH3)2BH3 -.16.5 3.8 mram.26 1.99 90 b.. Pyrolysis.'of 0,N-Dimethylhydroxylamine-Borane. CH30NHCH3BH3 is more stable than the 0-monomethylhydroxylamine adduct. Only traces of H2 were liberated in its preparation and storage at room temperature. As in the previous case, both explosive and nonexplosive decomposition were observed. Explosive decomposition, brought about by heating the compound rapidly.to 100~C, is obviously a very complex process. The stoichiometry for one sample was obtained from an overall analysis of explosion products.. The process is approximated by the following empirical equation: CH30NHCH3BH3 - 3 2.o3 I2 + 1.1.CH4 + 0 03 C2H6 +.0o8 N2 + 0,21 HCN + solid of approx. composition BN6o.5OHo 6 + o.6 c The quantitative features of the above equation probably vary from one..explosion to another, but the qualitative features are of interest. Hydrogen is to be expected in view of the ease with which hydrogen evolution occurs from compounds of this type; nitrogen is a no nitroge normal explosion product of nitron ncompounds The appearance of methane and ethane suggests methyl radicals as an intermediate; their reduction gives methane, their combination gives ethane.. The appearance of HCN-was rather unexpected, but can be rationalized in terms of the original WADC TR 56-318 152

HCN linkage in the molecule. The nonexplosive decomposition which results from storage of the compound at 650C for several hours differs markedly from the explosive decomposition. The overall stoichiometry is approximated by the following equation: COH3 pCH3 CH3 3- BOCH + (HNBH)x H OCH3 OCH3and CHS H I H 650 and —N — B 3 He-N —BH ~ small amount Several H H CHs H Hours H JCH3 0H3-N B\-ocH3 ( 3')x CH —N B0OCH + (CH3sNBH)x OCH3 The compound (CH3NHBH2)3, first prepared as a definite compound by the above process, has been described elsewhere.5 It is significant that the amine bound to trimethylborate was largely dimethylamine. Much smaller amounts of methylamine were found. A shift of methyl groups from one nitrogen to another in a controlled disproportionation reaction is implied. c. Hydrolysis of 0,N-Dimethylhydroxylamine —Borane. Hydrolysis reactions followed closely the pattern set in the hydrolysis of O-methylhydroxylamine-borane. Acid hydrolysis produced boric acid and the salt of O,N-dimethylhydroxylamine. Hydrolysis with 50% KOH in a sealed tube at 100~C produced a shift of methoxy and hydrogen to give methylamine as the principal basic product (27% yield) in a complex mixture of amines. Pure CH3ONHCH3 was not decomposed when heated under similar conditions with 50% KOH. Only two of the three hydridic hydrogens on the borane group were ever released as H2. O.,N, N-Tr ime thylhydroxylami ne -Borane.a. Preparation and Characterization. CH3ON(CH3)2 reacted with diborane without a solvent to give CHSON(CH3)2BH3. Stoichiometry of the process was excellent. The pure addition compotuid melts at -16.50 to -16.0~C and has a vapor pressure of 3.8 mm at 25~C. An equimolar mixture of CH3ON(CH3)aBH3 and trimethylamine., N(CH3)3, reacted as indicatep CH30N(CH3)2BH3 + N(CH3)3 --— > (CH3)3NBH3 + CH3oN(CH3)2 Ninety-eight percent of the original CH3ON(CH3)2 was recovered. No H2 was evolved. b. Pyrolysis of O,N,N-Trimethylhydroxylamine-Borane. In contrast to the behavior of O-methyl and O,M-dimethylhydroxylamine boranes, 5. T. C. Bissot and R. W. Parry, J. Am. Chem. Soc., 77, 3481 (1955). WADC TR 56-318 153

O,N,N-trimethylhydroxylamine-borane did not undergo an explosive decomposition when heated rapidly to 1000C in a sealed tube. At this temperature slow decomr position took place. All products were volatile. The molar ratio of hydrogen produced to O0N,N-trimethylhydroxylamine-borane used was 0.64. Trimethylborate, B(OCH3)3, was isolated as one of the products. The remainder of the volatile components were fractionated extensively, but no pure compounds could be separated. c. Hydrolysis of O,N,N-Trimethylhydroxylamine-Borane. Hydrolysis of CH30N(CH3)2BH3 with 50% KOH for two hours at 1000 in a sealed -tube gave::no.evidenvce for:':a shift'of -methoxy or hydrogehf.as- had beena:observed in the two preceding cases. About 92% of the original base and 90% of the hydridic hydrogen in the compound were recovered after hydrolysis. Discussion. - a. Loss of Hydrogen from the Borane Adducts of the Methylhydroxylamine Derivatives. The role played by both the complexed amine and the free diborane in the loss of hydrogen from the amine-borane complex is of interest in connection with mechanistic arguments suggested earlier.2 If one assumes that hydrogen evolution from the impure complex results from the interaction of a protonic hydrogen on the amine and a hydridic hydrogen on the attacking borane group, then the temperature for loss of hydrogen from the complex should correlate with the reNl ative acidity of the hydrogen atoms on the amine. In Table I selected chemical properties of all hydroxylamine-borane addition compounds are summarized. The temperatures at which 50% decomposition of the impure complex occurred in 24 hours are listed. As expected, the 0-methylhydroxylamine-borane adducts are more stable than the compounds containing a free hydroxyl group. Furthermore, if one assumes the usual electron-donating power attributed to alkyl radicals, the inductive effect of the methyl groups would reduce the acidic character of the remaining hydrogen atoms on the amine whenever a hydrogen is replaced by an alkyl radical. In general, the deconipo sition temperature of the amine-borane adducts increases where a methyl group replaces a hydrogen attached to the nitrogen. Other factors such as entropy of activation also contribute to this result. The poor stoichiometry observed in the reaction between diborane and those hydroxylamines which contain a free -OH group has been attributed to interaction of the -OH and borane groups.2 This argument finds support in the current study since stoichiometry was always excellent in the reaction between diborane and pure O-methylhydroxylamines where no free -OH group was present. 6. The relative acidic character of the secondary hydrogens on the amines is not necessarily the reverse of the primary basic strength of the amines as measured in water.3 WADC TR 56-318:/

b. The Hydride-Methoxy Shift in Pyrolysis. The pyrolysis of RH RH CH30NBH and HONBH RH RH (R = methyl or H) produced striking contrasts in behavior. Rapid heating of the 0-methyl and 0,N-dimethylhydroxylamine boranes invariably led to violent explosion. Rapid heating of hydroxyl and N-methylhydroxylamine boranes led to rapid H2 evolution, but no violent explosion was ever observed.7 The explosions in the former case appeared to be triggered by the sudden exothermic shift of methoxy groups from nitrogen to boron. Such shifts were never necessary in the latter case. Slower, more gentle heating of the 0,N-dimethylhydroxylamineborane adducts led to a controlled shift of methoxy groups from nitrogen to boron and of hydrogen from boron to nitrogen. Loss of hydrogen from the compound always preceded the low-temperature shift, Prior to the loss of hydrogen from the borane adduct, acid hydrolysis always yielded the original O-methylhydroxylamine and three molecules of hydrogen per borane group. After loss of hydrogen, B(OCH3)3, ammonia, methyl and dimethyl amines were obtained. Clearly, group transfer had occurred after or concurrent with hydrogen loss. The difficult pyrolysis of CH30N(CH3)2BH3 below 100~C also suggested that hydrogen loss was an important initial step in the shifting of groups. These facts imply that in those cases where a hydrogen is still attached to the nitrogen of a methoxyamine the pyrolysis proceeds by an initial splitting out of hydrogen from boron and nitrogen followed by rapid CH30 group transfer to the newly opened site on the boron. In explosive decomposition rapid heating produces more violent fragmentation, probably involving breaking of B-H and O-CFI3 bonds to give free radicals and atoms. This mode of decomposition undoubtedly involves a higher energy of activation, hence higher localized temperatures. Temperatures may build up rather rapidly in small volumes of the solid as a result of the exothermic low-temperature process, hence the lowtemperature shift in the 0-methyl and 0,N-dimethylhydroxylamine boranes could be easily converted to the explosive decomposition process if heating were rapid and heat were not dissipated. Although it was found that explosive decomposition of pure methoxyamine could be initiated by both spark and heat, it was shown that attachment of a nonoxidizable Lewis acid such as 3F3 to the base decreased its sensitivity to explosion. The compound CH30NH2BF3 was prepared in benzene as a white solid melting at 86~-88~C. The compound was heated in a sealed tube to 3000C. There was no explosion, but at the higher temperature the solid turned dark. This evidence indicates that the explosive decomposition is associated with the reducing character of the BH3 group and not alone with the properties of the coordinated amine. The following equation was written for the low-temperature decomposition of 7. D. H. Campbell, T. C. Bissot, and R. W. Parry, this Report, p. 134. WADC TR 56-318 15

O, N-dimethylhydroxylamine -borane: 3 H(CH30)CH3NBH3 653 ~ (CH3)2HNB(OCH3)3 + l/x(HNBH)x + (HCH3NBH2)3 + 3H2 Several Hours Data for the pyrolysis of 0-monomethylhydroxylamine-borane suggest a similar process, but the trimeric compound (H2NBH2)3, the inorganic analog of cyclohexane, was not-isolated despite several experimental attempts. A polymerization beyond the trimer stage was probably responsible for the difference. The expected equation would be 3(CH3O)H2NBH3 -5. ) H3NB(OCH3)3 + 2/n(H,.5-xNBH1.5-x)n + (3+x) H2 Several Hours The O,N,N-trimethylhydroxylamine-borane also underwent the methoxyl-hydride shift to give products which were rather similar in a formal sense to those outlined above: 3(CH30)(CH3)2JBH3 -* (CX3)2HNB(OCH3)3 + 2[(CH3)2NBH2] + 2H2 Although all the products of the reaction were volatile, only trimethoxyborane could be isolated and characterized as a pure compound. Despite extensive fractionation, the expected compound [(CH3)2\VBH2] could not be separated. The explanation is found in a report by Burg and Randolph8 to the effect that N,Ndimethylamine-borane, in addition to participating in' a monomer-dimer equilibrium, can enter into the reaction: 3/2[(CH3)2NBH2]2 -- - [F(C3)2N]2BH + (CH3)21 B2IH5 They reported that the mixture could not be separated into its components by conventional trap-to-trap distillation. The vapor pressure of the decomposition product in this investigation was in the same range as that expected for the mixture of the above compounds. Despite the formal similarity, the actual decomposition process, the pyrolysis of O,N,N-trimethylhydroxylamine-borane must differ significantly from that suggested for the O-methyl and 0ON-dimethyl derivatives. The absence of hydrogen attached to nitrogen renders impossible a low-temperature hydrogen loss followed by a shift of a methoxy group. Any mechanism which would impose a coordination number of five on boron or nitrogen is equally distasteful. Since the decomposition occurred at high temperatures where there would be partial dissociation of the complex to give free BH3 groups, it is suggested that the mechanism involved BH3 groups. A temporary coordination of BH3 with the oxygen on the amine would give an unstable intermediate which could shift a hydrogen over to the nitrogen'while the boron retained the methoxyl group. A disproportionation of methoxyborane to trimethoxyborane and diborane would give the ob8. A. B. Burg and C. L. Randolph, J. Am. Chem. Soc., 73, 953 (1951). WADC TR 56-318 L56

served products. The rate for such a process, involving a high-energy intermediate, would be slow compared to the shift over the unsaturated intermediate, and would not be expected to achieve explosive proportions even at 1000C. Such was the case. c. The Methoxyl-Hydride Shift During Hydrolysis with a Strong Base. Hydrolysis of CH30NHEBH3 and CH30NHCH3BH3 by 50% KOH invariably resulted in transfer of the methoxyl group to the boron and of the hydrogen to the nitrogen. In contrast, hydrolysis of CH30N(CH3)2BH3 simply liberated the base unchanged and produced hydrogen and borates from the BRH3 group. No shift of the methoxyl group was ever detected in the latter case. Clearly, the hydrogen attached to the nitrogen played a dominant role in group transfer. The following mechanism is suggested. In a strongly alkaline solution the equilibrium shown below would be displaced to the right: BH3F BH3 CH30-N-H = Hd + CH3 ON A - R l The resulting anion could then rearrange by shift of a hydrogen to the nitrogen and of a methoxyl group to the boron. Since the hydrolysis of such B-H bonds in alkaline solution is a rather slow reaction, the reduction of the amine would occur before the original compound was decomposed. For the compound BH3 CH30-N-CH3 CH3 acid ionization is impossible and no shift would be expected. Experimental facts support the expectation. d. Physical Properties of the Methyl-Substituted HEydroxylamine Boranes. Although the boiling points, melting points, and heats of vaporization for the methyl-substituted hydroxylamines are those expected from considerations of the hydrogen bond, the physical properties of the hydroxylamine boranes cannot be treated so easily. Data are summarized in Table I. The melting points, for example, do not show any definite trends. Although hydrogen bonding directly to the nitrogen is blocked by coordination of the borane group, bridging to the oxygen atom is still possible. The relatively low vapor pressures of the methylsubstituted hydroxylamine boranes and the slight increase in volatility with alkyl substitution on the nitrogen suggest that hydrogen-bond formation is still of some small importance for these compounds. On the other hand, the fact that HON(CH3)2BH3 is more volatile than CH30N(CH3)2BH3 is contrary to results expected WADC TR 56-318 157

from hydrogen-bonding considerations. A relatively large dipole moment for these compounds is indicated. Experimental. -- a. Reagents. 1. The O-methylhydroxylamines were prepared and purified as described earlier.9 2. Diborane was prepared and purified as described earlier.10 3. Ether-KReagent-grade diethyl ether was stored over CaH2 for several days and distilled under low pressure before use. Commercial diethyleneglycol dimethylether was handled in a similar fashion. b. Preparation and Reactions of Compounds. The general procedure for conducting reactions of this type has been described earlier.l0 Ether as a solvent was necessary for the preparation of the O-methylhydroxylamine-borane, but not for the O,N-dimethyl or O,N,N-trimethyl derivatives. It has been noted elsewhere that ammonia or trimethylamine as a contaminant in the O-methylhydroxylamine makes the resulting borane adduct lose hydrogen. When 9% ammonia was added to carefully purified 0-methylhydroxylamine and the reaction with diborane was effected as described above, the resulting mixture of CH30NH2BH4 and B2H6'2NH3 changed into a liquid at room temperature and evolved hydrogen. After only one hour, the ratio H2/CH30NH2BH3 was 0.48. A weakly complexed 1:1 liquid mixture of H2NB2H5 and (C2H5)20 was distilled from the system. Although the complexing of H2NB2H5 with (C2H5)20 has not been reported previously, it is not unexpected since amnmonia and trimethylamine adducts of H2NB2H5 are known. If the 1:1 mixture is completely dissociated in the vapor state, it would have a molecular weight of 58.4. The observed values determined by vapor density, were 61.5 and 63.7. Essentially pure H2NB2H5 was recovered from the mixture by complexing the ether with either A1Ci3 or BF3. Both worked satisfactorily. A 7.7-mg sample of aminodiborane purified by this technique was characterized on the basis of the evidence listed in Table II. Conventional high-vacuum techniques were used for effecting reaction between the methoxylamine-borane adducts and reagents such as BF3 or NR3. Details are available elsewhere.ll 9. T. C. Bissot, R. W. Parry, and D. H. Campbell, this Report, p. 118. 10. Do H. Campbell, T. Co Bissot, and Ro W. Parry, this Report, po 134, 11. T, C. Bissot, Phn.D dissertation, Univ. of Mich, Arnn Arbor, 1955. WADC TR 56-318 158

TAKLE ITI CH1RACTERIZATION OF a -MONOAMINODIBORANE Observed Literature (a) Molecular Weight 46.6 42.7 Vapor Pressure -23 7.8 mm 7.6 mm 0 32.4 mm 32.3 mm 18.9 85.4 mm 86.8 mm Melting Point Above -78 -66.4~ B/N/Hydridic H Ratio 1/1.99/4.86 1/2/5 (a) H. I. Schlesinger, D. M. Ritter, and A. B. Burg, J. Am., Chem. Soc., 60, 2297 (1938). co Low-Teempe.rature Decomposi tion of Compounds. Samples wesre mnaintaired in the preparation vessel at the desired temperature and products were isolateal and characterized. The most difficult phase of the operation was thle resolution of the complex solid mixture of NH3, B(OCH3)3, and, on occasion, (H-i3OH which resulted. Ammronia was identified by a potentiometric titration curve. Upon hydrolysis of thkle mixture.se no [Yydridic hydrogen was found. Boron analysis indicated 9.22% B; nit-rogen found was 7.e52%. Thlis is equivalent to 88.4% B(OCH3)3 and 9.15% NRH3, a, mix-ul:re iwh.ici. siould hrave a molecular weight of 70.0 when completely dissociated. The o'1bserved molec'ular weight by vapor density was 70.3. Synthe'-ic`th.ixt.ures`' of 1d3N.(0CH3)3, and CH30H were identical to the solid mixtures in term.s of physical appearance, vaporpressure, and dissociation pressures in -the vapor phta~se. In the mix-ttre resulting fromr the decomposition of CH30NCH3HBH3, the mixture of R2HN,!(W:YK"3)3, and CH30HT was separated by tying up the amines and alcohol with excess diborane, then separating the B(OCH3)3 and B2H6 by vacuum-line distillation. T-he purified,(0CH3)3 was characterized by vapor pressure at four temperatures, by melting point (-29.0~ obs. and -29.3O literature), by molecular weight (114.6 vs 103.9 literature), by absence of active hydridic H, and by boron analysis (10.05% vs 10.41% t3heoxry). d. Explosive Decomposition0 Explosive decomposition was conducted by heating a well-annealed bomb tube WADC TR 56-318 159

containing the sample to 1000~. The tube was opened to the vacuum line by means of the vacuum-tube opener. Noncondensable gases such as H2, N2, and CH4 were pumped off and separated by conventional techniques. C2H6 and HCN in the condensable products were separated by fraction distillation in the vacuum line. The HCN was characterized by molecular weight, melting point, vapor pressures, and reaction with silver nitrate. e. Analytical Methods. These have been described elsewhere (this report, p. 145). WADC TR 56-318 160

VI. THE REDUCTION OF CHLOROPHOSPHINES WITH LITHIUM ALUMINUM HYDRIDE AND WITH LITHIUM HYDRIDE (J. T. Yoke, G. Kodama, and R. W. Parry) WADC TR 56-318 161

Ab stract Alkyl phosphines have been prepared by the reduction of alkylchlorophosphines with lithium aluminum hydride. Coordination compounds of the general form LiAl(PHR)4, LiAl(PR2)4, and [LiAl(PR)2]n result from the interaction of alkyl phosphines with LiAlH4. Because of complex formation, direct distillation of alkyl phosphines from the reaction mixture is retarded. Simple PH3 can be distilled directly from the mixture of PC13 and LiAlH4 since it escapes at low temperatures before complexing. Lithium hydride reacted with PC13 to give large amounts of elemental phosphorus and less than 10% PH3. The reduction of boron or silicon halides with LiAlH4 represents an effective method for the synthesis of hydrides of boron or silicon.1 The corresponding reduction of phosphorus trichloride to give phosphine was studied briefly by Paddock2 and the preparation of C6H5PH2 from C6H5PC12 was repor td by Horvat and Furst3 while this investigation was in progress. The reduction of alkyl chlorides of phosphorus to give alkyl phosphines and their lithium aluminum complexes has been studied in this investigation., The Reduction of Phosphorus Trichloride. —In the presence of diethyl ether or tetrahydrofuran, PC13 was reduced rapidly at -60~C to give recovered yields of PH3 ranging from 60 to 80%. An earlier observation of Paddock2 to the effect that the reaction would not go properly at room temperature in the absence of a suitable solvent was confirmed. In the preparation of B2H6. LiH can be substituted for LiAlH4 to reduce BF3. The corresponding substitution of LiH for LiA1H4 in the reduction of PC13 produced only about a 6% yield of PH3- The major reaction was 3LiH + PC13 > P + 3/2 H2 + 3LiCl More than 70% of the theoretical phosphorus was consistently recovered. Variation of the temperature of the reaction from -60 to 250C and the presence or absence of LiAlH4 produced no major change in products. The Reduction of Monoalkyl Phosphorus Dichloridde-s r.The reduction'of "tPCl2 with LiAlH4, using the procedure outlined above for the reduction of PC13, gave very low yields of directly recoverable product. With LiAIH4 the alkyl phosphine 1. A. E. Finholt, Ao C. Bond, and H. I. Schlesinger. J. Am. Chem. Soc., 69, 1199 (1947). 2. No L. Paddock, Nature, 167, 1070 (1951). 5O Ro J. Horvat and A. Furst, ibid., 74, 562 (1952). WADC TR 56-3518 162

underwent a secondary reaction to give hydrogen and a mixture of lithium tetrakis ethyl monohydrogen phosphido aluminate (III) and lithium bis ethyl phosphide aluminate (III). The ratio was dependent upon conditions, all of which have not yet been completely delineated. LiAlH4 + 4H2PC2H5 -- LiA1(PHC2s5)4 + 4HI2 LiAlH4 + 2H2PCa2H5 - LiAl(PEt)2 + 4H2 The alkyl phosphine could be released from either of the solid products by hydrolysis. Heating the monohydrogen coordination compound brought about release of a portion of the coordinated phosphine through a disproportionation type of reaction: LiAl(HPEt)4 --- > LiAl(PC2H5)2 + 22PC2E H5 The reaction between ethyl phosphine and LiAl1AH4 was very similar to that observed with a primary amine and LiA1H4 (Refo 1, p0 162, and Table II, p. 169) except that the amine reaction was much faster than the phosphine process under comparable conditions. This appears to be somewhat strange since the phosphine is usually considered to be a somewhat stronger acid than the amineo In the reduction of PC13 with LiAlAH4 appreciable amounts (60-80% yields) of PH3 could be distilled directly from the reaction mixture and no evidence of complex formation was noted. In the reduction of chlorophosphines only small amounts (0625%) of alkyl phosphine could be distilled and complex formation was indicated by hydrogen evolution. This difference is attributed to the fact that PH3 is volatile at the low temperature of the reduction. (-63~ ), hence further reaction with LiAlH/4 is difficult. On the other hand, the EtPH2 produced from EtPCl2 is not volatile at -63~C and is retained in the ether solution where loss of hydrogen occurs. The Reaction of Methyl Phosphines with LiAlH4, —Although attempts to synthesize methyl phosphine by reduction of a product believed to be MePC12 were inconclusive due to experimental difficulties, the direct reaction between methyl phosphine and LiAlH4 was studied in some detail0 Results summarized in Table II, p. 169, support representation of the process by the equation H2PMe - LiAlH4 LiAl(HPMe)4 + 4H2 The presence of some of the bis product, LiAl(PM3)2, is also suggested by the data. A comparable reaction involving diethyl monohydrogen phosphine gave evidence for the formation of LiAl(PMe2)4, Table II, po 169. The complex was destroyed by water to liberate the original phosphineo The Synthesis of Mixed Alkyl Chlorides of Phosphorus. -The use of the foregoing reduction process for the preparation of phosphines is limited in practical application by the comparative unavailability of the aLkyl chlorophosphines. Ethyl dichlorophosphine, EtPC12, can be prepared by the alkylation of PC13 with tetraWADC TR 56-318 163

ethyl lead.l Normal propyl dichlorophosphine was also prepared by an extenslon of the tetraethyl lead'alkylation procedure. In an attempt to prepare methyl dichlorophosphine from Pb(Me)4 and PC13, a precipitate of PbCl2 was obtained in the original step, as expected, but difficulties encountered during the reduction step prevented isolation and characterization of MePH2 which was needed for positive identification of methyl phosphorus (III) chlorideo Fox5 suggested the use of dialkyl cadmium as an alkylating agent for the preparation of alkyl dichlorophosphines. The procedure worked wello On the other hand, the Grignard reagent as conventionally handled in diethyl ether gave no EtPC12 when equal molar quantities of MgRX and PC13 were allowed to react. The Vapor Pressures of Mono- and Diethyl -Phosphines. -Since vapor-pressure data for mono- and diethyl phosphines have not been recorded in the literature, data for carefully purified samples were obtained, For ethyl phosphine the vapor pressure is given by the equation log1oP = - 1766.2 - 3.5850 T + 17.7567 T The molar heat of vaporization at the normal boiling point of 200C is 6000 caL. For diethyl phosphine the equation is log oP 2359.6 - 4.5931 log T + 21.1877 T The molar heat of vaporization at the normal boiling point':of 860C is 7520 cal0 Trouton's constants of 20.5 cal/mole x deg for monoethyl and 21.0 cal/mole x deg for diethyl phosphine are normal for-nonassociated liquids.: Data are- given in Table I. The Reaction Between Ethyl Phosphine and an Ether Solution of Aluminum Hydrideo-Although ethyl phosphine reacted rapidly with hydridic hydrogens of LiAiH4 to liberate gaseous hydrogen, it was found that EtPH2 would not react with a solution of aluminum hydride in ether at room temperature. It is well known that coordination to a strong uncharged base such as N(CHI3)3 of a species containing hydridic hydrogen such as BH3 reduces the hydridic character of the hydrogen. For example, trimethylamine-borane shows very little hydridic character, even less than LiBH4. An-extrapolation.of this argument suggests that the strong coordination of solvent -ether to the aluminum hydride reduces the hydridic character of the hydrogens to such an extent that reactionvwith the protonic hydrogens of phosphine is not observed at room temperature. On the other hand> the corresponding reaction with the more hydridic LiAlH4 proceeds easily at -65~0c 4. M. S.Kharasch, E, Vo Jensen, and S. Weinhouse, J. Org. Chem., 14, 429 (1949). 5. Ro B, Foxy JO Am. Chemo Soco., 72, 4147 (1950). WADC TR 56-318 164

TABLE I VAPOR PRESSURES OF MONO- AND DIETHYL PHOSPHINES A. Monoeth!yl Phosphine PEtH2 B. Diethyl Phosphine PEt2H Temp. Pressure Pressure Temp. Pressure Pressure C0 mmm Hg mm Hg mm Hg A (obs. ) (calc. ) (obs.) (calc.) -83.6 1.9 1.9 0 -35.3 2.1 2.26 -.1 -63.5 10.1 10.3 +.2 -22.6 5.8 5.67 -.1 -45.2 36.4 36.1 -.3 0.0 23.9 22.9 -1.0 -35.3 61.7 65.1 +.4 5.5 31.2 31.0 -.2 -22.6 128.2 128.4 +.2 31.5 108.3 108.4 +.1 0.0 363.5 360.6 -3.5 42.4 170.7 170.8 +.1 5,2 446.7 445.1 -1.6 51.6 244.6 243.7 -.9 16.9 690.2 692.2 +2.0 52.9 255.3 255.8 +.5 61.4 343.7 347.0 +3.3 71.4 484.0 485.4 +1.4 75.1 549.1 546.6 -2.5 80.6 653.8 648.2 -5.6 log P = -1766.2/T - 3.5850 log T + 17.7567 log P = -2359.6/T - 4.5931 log T + Boiling Point = 20~C from equation 21.1877 AHvap at 760 mm = 6000 cal at 293~K Boiling Point = 860~C from equation ASvap at 760 mm = 20.5 cal/mole x deg. AHvap at 760 mm = 7520 cal at 293~K ASvap at 760 mm = 21.0 cal/mole x deg WADC TR 56- 318 165

Experimental. - a. Reagents Analytical reagent grade Mg. C2H5Br, PC13, and AlC13 were used without further purification. LiAlH4 and LiH were commercial products obtained from Metal Hydrides, Inc. Tetrahydrofuran (Eastman Kodak) was distilled from mineral oil through a Snyder column, dried over sodium, and refractionated. Boiling point = 63.0-63.3~C uncorrected. bo The Reduction of Phosphorus Trichloride with Lithium Aluminum Hydrideo The reaction vessel was a 500-cc three-neck ~ flask. One neck was connected through a vacuum stopcock to a conventional high-vacuum fractionating system. The center neck was equipped with a cold finger, the top of which led to a column packed with glass helices and surrounded by a jacket for a cooling bath. This column led to a train consisting of a similarly packed U-trap, a mercury bubbler, a stopcock, and a second U-trap which could be opened on the end to a Toepler pump or the vacuum fractionation system. See Fig. 1. A stream of nitrogen was passed through the apparatus. Slush baths at -75~C were placed around the flask, the packed column, and the packed Utrap. The final U-trap was immersed in liquid nitrogen. In a typical run about.012 mole of PC13 (1 cc liquid) in 25 cc of diethyl ether was added dropwise with magnetic stirring to about 1 g (.026 mole) LiAIH4 in 75 cc of diethyl ether (temp -75~ to -65~C). After addition was complete, the cooling bath was removed from the reaction flask. The highly volatile PH3 product passed through the train to the final trap. After purification in the highvacuum line, the PH3 was characterized by vapor pressure and molecular weighto Yields of recovered PH3 ranged from 55 to 70% of theory, based on the equation 3LiA1H4 + 4PC13 - 4iPH3 + 3LiC1 + 3A1C13 ce The Reduction of Phosphorus Trichloride with Lithium Hydride. The apparatus and technique were roughly the same as described above. About.013 mole PCl3 in 20 cc of tetrahydrofuran was added dropwise to.0315 g of 200-mesh LiH suspended in 75 cc of tetrahydrofuran under N2o- Vigorous stirrifig was achieved by a magnetic stirrero The PH3 formed was about 6% of the theoretical based on the equation 3LiH + PCl3 - P3 + 3L+ C1 l Red phosphorus formed in the flask was filtered on a ground-glass crucible, dried, and weighed-. Yield was 72%, based on the equation 3LiH + PC13 -- P+ 3/2 H2 + 3LiCl. WADC TR 56-318 166

TO VACUUM LINE 10/3o GLASS HELICES 14/35 943 TOEPLER PUMP AND GAS BURETTE 1 24/40 I.~~~~~~~~~H MAGNETICE H STIRRING BAR Fig. 1. Apparatus for production of phosphines by reduction of chiorophosphines with LiAIH4. WADC TR 56-318 167

do The Reduction of Ethyl Dichlorophosphine with LiAlH4o About 5 g (,04 mole) of ethyl dichlorophosphine was distilled under vacuum into the reaction flask along with 25 cc of tetrahydrofuran which had been distilled directly from LiA1H4, A chloroform slush bath (-63~C) was placed in the cold finger and the packed column was surrounded with powdered Dry Ice, The reaction flask was kept at a low temperature by refluxing of the solvent from the cold finger in the evacuated system. A large excess of lithium aluminum hydride (7.'0 g)' was stirred with 75 cc of dry tetrahydrofuran. Since all the hydride does not dissolve (impure), the mixture was centrifuged and the supernatant liquid transferred with a dry hypodermic syringe to the dropping funnel (under N2). The solution was then added dropwise with magnetic stirring to the reaction flask. Condensible and noncondensible gases were separated, identified, and measured. Usually very little phosphine was distilled from the system at the end of the run, but the alkyl phosphine was easily obtained from the solid residues by hydrolysis, using a 20% solution of water in tetraethyleneglycol diethyl ether. e. The Direct Reaction Between Methyl Phosphi.nes and L:A1lI4 and Methyl Amine and LiAIH4, Commercial Ansul polyether Noo 141 was dried by warming with sodium wire overnight. It was distilLed under reduced pressureo LiAI4 was stirred in the ether and the mixture centrifuged to get a clear solution. Concentration of the LiAIH4 solution was estimated by hydrolyzing 1 cc of the solution and measuring evolved hydrogen, The reaction vessel for the process was essentially the same as that described earlier. The LiAIE4 solution (about 1.o5 cc of about 1 molar solution) was placed in the flask. After cooling the sol.jti',:on to -780C and evacuating, stopcocks were closed and the system held at room temperature for 1/2 hour. Traces of evolved hydrogen were removed, then a measured amount (in vacuum line) of PH2(CES3),, P7:(C:H3)2, or N1E2CH3 was condensed into the vessel from the vacuum system. Ice water was placed in the cold finger and the reaction vessel was allowed to warm to 0~C. Near OC hydrogen evolution was quite- rapid. At intervals hydrogen was pumped from the frozen system and measured. Any amine or phosphine distilling from the system was returned. When gas evolution was complete about 3 or 4 cc of water was condensed into the flask and the system warmed. The liberated H2 and P(CH3)2H or amine were purified in the vacuum system and measured. Detailed data are summarized in Table IXo fo The Preparation of Alkyl Dichlorophosphines. Fifty grams of tetraethyl lead (0.15 mole) and 68.7 g (0.5 mole) of PC13 were refluxed for 30 hours in a nitrogen atmosphere. Distillation of the reWADC TR 56-318 168

Q xii 0C) TABLE II THE REACTION OF PHOSPHINES ANT AMINES WITH LiA1H4 Original Mmoles Heating Hydrolysis Observed Experimental Ratios Mmles Mmoles Amine or LiAPhospPrincipal Solid Products Rt LihE4 Mmoes11 Phosph. Mmolets M oles moles M- PicplSldPout ai Phosphine orpmn From Evolved Phosp la. 112sMmoles Phos M Initial 12 Phosph.(Heat) Phosph. Indicated by Stoichiometry H2PCH3 (Heat) Mixed Total 12 Recover Ror2 Out ed 112 Total Thosl~a. Phosph. (Hyd) LA1H4 H2PCE3 (Hyd) Mixed Recovered. Recovered H2 11.09 1.18 4.54 3.91 ---- ---- 3.85 0.22 1.16 3.51 65* LiAl(MeHP)4 and 355% LiAl(MleP)2 CH3PH2 10.81 1.06 4.14 4.02 1.435a) 0.21 2.50 0.09 1.03 0.57 5.90 95% LiAl(MeHP)4 and 0.9 Cc) 5% LiAI(MeP)2 9.17 1.59 6.52 5.90 2.50 0 3.36 0.07 1.07 0.69 3.71 86% LiAl(MeHP)4 and H 1 I I ~14% LiAl(MeP)2 0.75 I 11. 54'' about 1.435 s. 6 28(d) ---- 4.57 0.21 0.88 (d) 4.5 LiAl(PMe2)4 (CH3)2PH1.(e 11.72(e) 0.94 3.26 3.24 ---- ---- 2.82 0.51 1-00(e).4 LiAl(PMe2)4 and 3 -LiAlH2(PMe) CHII3NH2 )0.83(f) 11.65 1.92 7.62 7.16 151(g) m.6 1.2+(h) 1.o6 o.49 3.7 85% LiAl(HNMe)4 and 0.74 15% LiA1(NR)2 (a) Heated 40 min at 500 to 600C. (b) Based on assumption that reaction LiAl(HPCH3)4 - LiAl(PCH3)2 + 2H2PCH3 goes to completion. Cc) Heated to 100lC for 2.5 hr. (d) Original R2EP contained some NH3 as impurity. (e) Pure sample. (f) 10000 for 1 hr. (g) 2250C for 1.5 hr. (h) Remainder of methyl amine was not separated from 120 in recovery. More present.

action mixture through a Widmer column gave a 46% yield (26.8g) of EtPC12; bp 111-112~C, uncorrected, and density 1.19 g/cc. In an unsuccessful attempt at Grignard synthesis, 0.5 mole of EtMgBr in 100 cc of diethyl ether was added dropwise with stirring under an N2 atmosphere to 68.7 g PC13 in 100 cc ether at 0~C. No C2H5P(OH)2 or C2H5PO(0H)2 could be isolated by ether extraction of the hydrolyzed reaction mixture, hence:no..EtPC12 was formed. n-Propyl dichlorophosphine was prepared when 9.5 g (0.025 mole) (n-C3H7)4Pb and 6~55 cc (0o075 mole) PC13 were refluxed for 36 hours in a nitrogen atmosphereo The volatile material was fractionated in the vacuum system. A 60.9% yield of n-C3H7PC12 was obtained. go Reaction of Aluminum Hydride with Ethylphosphine. A solution of A1H3 in ether was prepared in the reaction vessel of the earlier system by following the method of Wiberg,"Graf, and Uson in which stoichiometric amounts of standard solutions of AlC13 and LiAlH4 were mixed. An excess of C2H5PH2 was distilled into the flask by chilling it and its:contents with liquid nitrogen. The temperature of the system was allowed to rise gradually to room temperature. No H2 was evolved and all the added C2HsPH2 could be recovered. No protonic-hydric hydrogen interaction or complex formation occurred under the conditions of this experiment. Acknowledgement We wish to thank the Ethyl Corporation of Detroit, Michigan, for the samples of Pb(n-C3H7)4 and Pb(CH3)4 which were generously provided for this study. 6. E, Wiberg, H. Graft, and R.'Uson, Z. amorg. und allgem. Chem., 272, 221 (1953). WADC TR 56-318 170

APPENDIX THE ISOTOPE EFFECT IN THE TRACER STUDIES OF THE "DIAMMONIATE OF DIBORANE" The original intent of the tracer studies was to re-examine Girardot's work concerning nitrogen-hydrogen, boron-hydrogen bond rupture. This has been accomplished. Because of the now apparent complexities of the systems which employed ammonia-d3, the data obtained do not necessarily lend themselves to a quantitative treatment of the isotope effect involved. However, the available information does make it possible to present a reasonable description of the events occurring in these systems. Consider the distribution of protium when B2D6 2ND31 was prepared from and dissolved in ammonia-d3. Since the solvent was about 96% deuterated, the principal species in solution were ND3, ND2H, D D D D H IND3 and LB XD3 + [D D/ \ND2H D H-D exchange occurred through only those hydrogens which were bound to nitrogen; equilibration was rapid.2 D ND1 D D D /ND2H+ FD\7 ND2H + /B\ = ND3 + ] B D ND_ D ID D ND3 D D Furthermore, Burg's work2 indicates that the isotopic separation factor a, which is a distribution coefficients for the exchange reaction, should be about unity. Therefore, prior to the addition of sodium, this system had reached or approached equilibrium and the nitrogen-hydrogen bonds of the "diammoniate" probably contained protium and deuterium in the ratio of about 4H to 96D. 1. B2D6.2ND3 was used for the purpose of illustration. Since the boron-hydrogen bonds are apparently inert, with respect to hydrogen exchange, the above illustration applies to B2H6.2ND3. 2. A. B. Burg, J. Am. Chem. Soc., 69, 747 (1947). 3. ex = H/D(B2D6.2N3) H/D(ND3) WADC TR 56-318 171

Nitrogen-hydrogen bonds were ruptured when sodium was allowed to react with the solution at -78~. The principal isotope effect probably occurred at this point. It is believed that the rate-determining step of the reaction involved the rupture of the nitrogen-hydrogen bonds and that the N-H bond took preference over the N-D bond>4 Since only a small fraction of the total number of nitrogen-hydrogen bonds in the system w6re ruptured, the evolved gas was considerably enriched in protium; it contained more than 4H to 96D. Because of this, the unreacted "diammoniate" should have been enriched in deuterium as the reaction progressed; however, since the solvent ammonia-d3 was in large excess (solution ca. 0.01 M), and since the system establishes equilibrium readily, there was a tendency for the unreacted "diammoniate" to contain a constant, or nearly constant, protium-deuterium distribution in the nitrogen-hydrogen bonds. The reaction with sodium may be represented as D _ + D D ND2H + B ND + ] ND3 [B + + Na Na D -...., —.-.. H + some D D2, HD, t H Because the gas which is evolved is so greatly enriched in protium, an estimation of the instantaneous separation factor a0o seemed to be attractive. However, there were several points to consider. First of all, although the deuterium content of the hydrogen evolved in the tracer reactions and the deuterium content of the ammonia-d3 were known to a degree of accuracy which was certainly sufficient for the principal purpose of the tracer investigation, the isotopic ratios of protium to deuterium were not known to a comparable degree of accuracy. Thus, for example, the ammonia-d3 was considered to be about 96% deuterated, and most likely was 96-97% deuterated. It can be seen that the H/D ratio for 96% deuterated ammonia-d3 is 35% greater than that for 97% deuterated ammonia-d3. 4/96 3/97 Also, although the instantaneous separation factor is most accurately determined from data which are obtained after a small extent of reaction, in this investigation tracer data were obtained after an appreciable extent of reaction. In spite of these difficulties it seemed that an estimate of the order of magnitude of a% would still be worthwhile. 4. Owing to the mass difference between H and D, the zero-point energy of the N-D bond is lower than that of the N-H bond. From a simple, qualitative consideration of the vibrational stretching frequencies of these bonds it is seen that the N-H bond is more readily broken than the N-D bond. Also, from a similar type of consideration it is not unreasonable to expect than an:N-XH bbnd would.be' more reactive than an N-D bond. 5 o = H/D(gas) at zero extent of reaction. H/D(B2D6 2ND3) WADC TR 56-318 172

The instantaneous separation factor for the reaction of B2D6 2ND3 with sodium in liquid ammonia-d3 was approximated from the data of the tracer reactions (see p. 70 of this report). Assuming that the protium-to-deuterium distribution in the nitrogenhydrogen bonds of the "diammoniate" was 4H/96D, an apparent separation factor was calculated and plotted for a given fraction reacted.~ This process was repeated, assuming a ratio of 3H/97D. Then the apparent separation factors were extrapolated to zero extent of reaction (Fig. 1). The solid-line extrapolations are drawn through the points of two individual experiments and are intended to give the maximum and minimum values possible for aU. In carrying out these extrapolations greater significance was given to those points representing smaller fraction reacted. The dotted-line extrapolation is drawn through the data from both experiments and represents the best estimate of ao, which appears to be 17 ~ 4. WADC TR 56-318 173

Q20 18 16 1 4 12 Ht- 8 [ T ~< 2 Run 4 Table I (p.70) * ____ ____ ____ r t a a 0 Run 5 Table 1 (p.70) 0 6 1 1 I I 1 I I I " I Calculated assuming initial ratio of H/D in B2D6 2ND3was: 4 p 3/97 (Upper limit) t RunS Run4 2 4/96 ( Lower limit) 4 0 02 0.4 0.6 0.8 1.0 FRACTION REACTED Fig. 1. An approximation of ao for the reaction of B2D6.2ND3 with sodium in ammonia-d3,

UNIVERSITY OF MICHIGAN 3 9015 03695 5410