THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF TIHE COLLEGE OF ENGINEERING A RAMAN SPECTROSCOPIC STUDY OF SOME LEWIS ACID - BASE SYSTEMS John Raymond" Moyer A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan. 1958 September, 1958 IP-317

Doctoral Committee: Assistant Professor Robert C. Taylor., Chairman Associate Professor Lee 0. Case Assistant Professor Samuel Krimm Associate Professor Robert W. Parry Associate Professor Wyman Ro Vaughan

ACKNOWLEDGMENTS The authaor wishes to- thank Dr. Robert C. Taylor for his.advice and encouragement during this instigati This research was partially supported through several generousgrants: The Minnesota Mining and,Manufacturing Company Fellowship, The UnLtion Carbide and Carbon Corporation Summer Fellowship, and The Deprt-.men -of Chemistry Summer Fellowship. ii

ACKOWITEDGMNTS The.authr wishes to thank Dr, Robert C. Taylorfoforhis.advice and encouragement during this. investigation. This research was iartially supported through several generous grants: The Minnesota Mining and Manufacturing Company Fellowship, The Union Carbide and Carbon Corporation Summer Fellowship, and The Depart-.menr -of Chemi:stry Summer Fellowship. ii

TABLE OF CONTENTS Page ACKNOWLEDGMENTS................................... ii LIST OF TABLES............*.@,........o*...*o.....@... o... iv LIST OF FIGURES...............,............,,......... v INTRODUCTION.. O O *. O * 0 0 0 O......................a. *. * O * 0. a *. a o 1 STATEMENT OF THE PROBLEM.......................... 8 HISTORICAL BACKGROUND............................. 11 A BRIEF REVIEW OF VIBRATIONAL THEORY..........*.............ooo, 14 INFRARED AND RAMAN SPECTROSCOPY....,.............. * 20 EXPERMENTAL PROCEDURES........................................26 Preparation of materials........................, 26 Preparation of samples...-.........** o o, o. 28 Spectrographic equipment.................. 32 EXPERIMENTAL RESULTS AND THEIR INTERPRETATION.......o,.....,,o. 39 Raman Spectra of Pure Components.....,.....,,., 39 Systems Showing no Interaction in the Liquid Phase.... 41 Systems Showing Interaction but Not Complexation in the Liquid Phase,........ *.. * * *.. 41 Systems Showing Complexation.......O.......,......o.,o. 43 The Systems: Boron Trihalide - Dimethyl Ether.....,.. 46 The Spectra of DME:BX3 in Ether-Rich Solutions....... 63 The Spectra of DME:BX3 in Acid-Rich Solutions......... 67 The Effect of HC1 Upon DME:BC135........,.....o.. 68 SUlMMARY.,..............,........o...*... 69 BIBLIOGRAPHY....... a.,...., ** 73 APPEIDIX.. *....................................... 76 iii

LIST OF TABLES Table Page I Vibrational Frequencies of A12Br6.....,........... 40 II Vapor Pressure of An Equimolar Mixture of S02 and DME. o. o...........*......... *oo..o...l 0 l.... 41 III The Effect of S02 and HC1 Upon the Skeletal Frequencies of DME,............................. 42 IV The Raman Spectrum of MeBr:AlBr3.......... o. 47 V Typical Values of the C-F Stretching and the CF3 Deformation Freqienacies............................ 50 VI The Vibrational Spectra of BF- and CF4..,...o... 51 VII The Raman Spectrum of (CH) 20 BF3.................. 53 VIII Typical Values of the C-C1 Stretching and the CC13 Deformation Frequencies..................5......0 56 IX The Raman Spectrum of (CH3)20:BC13................. 58 X A Comparison of the Skeletal Frequencies of DME, DME:"BC13, and DME:BF3....................... 59 XI A Comparison of the B-F Stretching and the BF Deformation Frequencies of BF3, DME:BF3, and BF.... 62 XII A Comparison of the B-Cl Stretching and the BC15 Deformation Frequencies of BC13 and DME:BC13...... 63 iv

LIST OF FIGURES Figure Page 1 Internal Coordinates of a Pyramidal XY3 Molecule..... 17 2 Vacuum Line Used in the Preparation of Raman Sample s....... o o o 30 3 Apparatus Used in Sample Preparation................... 31 4 Raman Light Source Assembly............................ 34 5 Sample Tube Holder...................... 37 6 Raman Spectrum of AliBr6.................... O.o....... 40 7 Raman Spectra of CH3r'A1Br3.......................... 45 8 Raman Spectra of (CH3)20: BF3.......................... 48 9 Raman Spectra of (CH3)20:BC13......................... 55 10 The Mass Dependence-of VXY'or Various ZYZ Angles.. 61 11 0-1200 Region for Various BF3 - (CH3)20 Ratios....... 65 12 300-1000 Region for Various BC13 - (CH3)20 Ratios.... 66 V

INTRODUCTION The study of aci4-base phenomena has led to some of the most vigorous controversies in the field of chemistry. Theories proposed by eminent workers have gained acceptance only to be discarded as subsequent.experimental evidence pointed out omissions.or contradictions. Accumulating experimental evidence has always seemed to indicate a much greater complexity in both reactions and products than had been suspected earlier. As a result the theoretical treatments have tended to become more and more general until they now are closely related to the electronic nature of chemical interaction. The one-element theories of Lavoisier and Davy served adequately through most of the nineteenth century although the type.of experimental evidence which would lead ultimately to the waterrion theory was initiated by Faraday as early as 1834. Work by van't Hoff, Ostwald, and.others upon aqueous solutions.of electrolytes led Arrhenius to a concept which is useful even today as an aid to understanding the behavior of aqueous solutions. Subsequent theories tended to be closely related to the Arrhenius theory but of broader scope. In Franxlin's theory of solvent systems emphasis is placed not just upon water and its ions. An acid, according to him, was defined as a solute giving rise to a cation characteristic of the solvent, and a base was defined as a solute giving rise to a characteristic anion. A typical neutralization reaction in liquid N204 might be: -.1

-2NOC1 + AgN03 = AgCl-+ N204 where NOC1 is an acid because of its ionization into NO+ and AgN03 a base due to its ionization to give N03. Because of its ability to relate acid-base behavior in numerous solvents, the theory of solvent systems offered a real advantage over the water-ion concept. It suffered from disadvantages present in the earlier theory in that it was still limited to solvent systems and overemphasized the ionic character of neutralization processes. These weaknesses were actually the strengths of the protonic concept which Bransted and Lowry advanced in 19235. Since they defined acid and bases as proton donors and acceptors respectively, the neutralization process was no longer dependent upon the reaction medium or upon the mechanism of proton transfer. Many species, both molecular and ignic, which had long been known to affect acids after the fashion of alkalis could now be accepted as bases. However, an equal number of species were excluded from classification as acids because they did not contain hydrogen, despite their acid-like effect upon alkalis. This constituted a serious omission. The electronic theory of G. N. Lewis was actually very closely akin to the Bransted-Lowry concept. Instead of basing acid-base phenomena entirely upon proton transfer, Lewis saw this to be only a special case of the more general occurrence, electron sharing. Actually there are three ways in which two atoms might apportion electrons. A complete transfer of one or more electrons would lead

to the familiar electrovalent or ionic bond. A contribution of one electron by each to make a pair which is then shared by both gives the covalent bond so common in organic chemistry. Finally, one atom may have a pair of electrons which it can donate to form a dative bond. It was this ability to accept an unshared pair of electrons that Lewis found to be common to all acids. Lewis developed his theory by first listing all the species which exhibited a characteristic set of "acid" properties. Luder and Zuffanti(l) describe these four "phenomenological criteria" as: I. Neutralization. Acids and bases may combine more or less rapidly with each other. II. Titration with Indicators. Acids and bases may be titrated against each other by the use of substances, usually colored, known as indicators. III, Displacement. An acid or base will in general replace a weaker acid or base from its compounds. IV. Catalysis. Acids and bases frequently act as catalysts. Any substance displaying all four of these characteristics was then classified as an acid or base, depending upon its similarity to a known acid, hydrogen chloride, or the known bases, the alkalis. While the list of bases turned out to be identical with that under the Bransted-Lowry theory, the compounds designated as acids included not only proton donors and the acids of the solvent systems theory, but also many more which had never before been recognized as such by any formal theory.

-4According to Lewis' electronic theory the neutralization process involved formation of a dative bond between the base and the acid, where the acid was any molecule, radical, or ion capable of accepting an electron pair or pairs, with the base playing the role of the electron pair donor. This brought many seemingly unrelated reactions into the realm of neutralization reactions, since any case of coordination now represented acid-base behavior. This more generalized extension of the theory proved a stimulus to the new areas included, catalysis and coordination chemistry, for example. The catalytic function of acids and bases was noted by Lewis among his "phenomenological criteria." The nature of this function can often be shown to be the result of the formation of an unstable intermediate by the catalyst through donation or acceptance of an electron pair. The complex CH3Br:.AlBr3 discussed in this research appears to be one such case. The contribution of Lewis' electronic theory to the field of coordination chemistry has been of a much broader scope. The difficulties which were present in this area at the time the Lewis theory was being formulated, stemmed from failure to understand the wide variations in binding forces and their obvious interrelation. The situation can perhaps be best illustrated by example. The divergencies in binding forces can be seen in ammine complexes like Ag(NH3)2Cl and Co(NH3)6C15. The former is ionic, readily and reversibly dissociating.on heating or in solution, while the latter loses only one mole of ammonia at 175~C and retains all six moles in sulfuric acid solution.'The interrelation

-5of the two types of bonds is seen in Co(NH3)5Cl13 in which only two chloride ions can be precipitated by aqueous silver nitrate. Werner, in his early work on the nature of coordination, postulated two different valence types. The first, designated "principal" or "primary" valences, described the customary affinity bonds which are now defined in terms of combining capacity with hydrogen. His "auxiliary" or "secondary" valences wereoriginally conceived as being quite different since, unlike the primary valences, they did not allow ionization. These fundamental postulates met widespread criticism and general acceptance of Werner's theories was postponed for many years. Werner himself recognized a connection between his primary and secondary valences and finally concluded that no clear-cut distinction between them was possible. In the light of current valence theory it can be seen that Werner's "primary" valences involved either electrostatic or covalent bonds, and hence showed a wide variation in properties; his secondary valences were the dative bonds encountered in Lewis acid-base interaction. The field of Lewis acid-base interactions now is considered to be but a specialized segment of the much broader field of coordination chemistry. The older question of what is an acid no longer seems to be compelling in the light of Lewis' behavioristic approach. Current interest has turned to an examination of the nature of the acid-base interaction and its products. In general, this is achieved through investigations of acid-base systems to determine the number and nature of the complexes formed.

-6Nearly every available physico-chemical technique has been employed to gain information about the nature of acid-base interaction. Thermodynamic properties of the more stable complexes have been evaluated.. Heat capacity data have revealed something of the internal energy of the complexes while entropies or free energy changes can be a measure of the spontaneity of their formation and permit calculation of equilibrium constants. Even very weakly bound aggregations have been detected using such classical methods as phase studies and conductivity measurements. Cryoscopic methods have the additional advantage of giving stoichiometry as well as melting.points. Other techniques, more recently developed, make available structural information about the complexes. Knowledge of geometrical configuration, an indirect contribution to an understanding.of the bonding forces involved, can be obtained by the methods of x-ray and electron diffraction. Configurations can also be inferred from dipole moment measurements and reactivities can be better understood in the light.of such data. Reference has been made to one convenient method of detecting the occurrence of complexation, that of ultraviolet spectroscopy. Since the electronic part of the complex is most directly involved in the process of ultraviolet absorption, combinations of too fleeting existence for observation by other means can be disclosed. Functional groups can be identified in many cases despite the limited view of the overall interaction which is provided by this technique. The application of group theory to the field of vibrational spectroscopy has permitted a much broader use of the techniques of infrared

-7and Raman spectroscopy to gain information regarding configuration and binding forces in the complexes resulting from acid-base interaction. Nuclear magnetic resonance, which gives a measure of the electron distribution about the nucleus, has not been applied to any extent in this field but appears a promising tool for investigation.

STATEMENT OF THE PROBLEM Earlier work in this laboratory(2) on the nature of the Lewis acid-base interaction centered about the HC1 - dimethyl ether system. The technique employed was that of vibrational spectroscopy which permits direct observation of frequencies associated with certain molecular motions. This feature makes it particularly useful for general exploratory work in detecting complex formation through changes in the characteristic spectra of the reacting components. In addition to establishing the existence of stable complexes, the spectral results also supply information about the nature of these complexes. Of the two methods available for the examination of vibrational spectra, the technique of Raman spectroscopy was chosen over infrared in the previous work principally because it permits use of all-glass apparatus and cells. Other materials, such as the rock salt required for infrared cells, are difficult, if not impossible, to use in view of the reactivity and properties of the reagents. The Raman technique also allows observation of a broader spectral region and is more conveniently adaptable to a wide range of temperatures. Since this research was instituted as an extension of Vidale's work, these same advantages make the Raman technique preferable. The HC1 - dimethyl ether system was found to contain at least three discrete complexes. Besides a 1:1 molecular complex there were two others which appeared to be ionic in nature, resulting from proton transfer accompanied by further complexatiQn on the chloride. However, complete characterization

-9of these ionic species was not possible because of inherent spectroscopic limitations of the HC1 molecule. Although HC1 has one frequency and its shifts could be used to detect complex formation, the solvated C1- ion had frequencies which were either too low in magnitude or of too low intensity to give any information as to the nature of the anion(s) formed. This inability to obtain spectra of the anions prevented complete characterization of these higher order complexes, although much information could be gleaned from shifts in the DME skeletal frequencies. In the present work a more complex Lewis acid was sought which, in mixtures with DME, might give rise to another series of complexes in which both anionic and cationic fragments could be characterized. Such a system might well provide information on the stoichiometry of the HC1 - DMIE complexes in addition to being valuable as another example of acid-base interaction. Conductivity measurements and other work by Kraus and coworkers (3y4) have indicated the possible existence of more than one complex in the aluminum bromide - DME system. Also, Brown's(5) recently proposed single bridge structure for compounds of the boron halides suggests the possibility of higher order complexes in their systems with some of the stronger Lewis bases such as DME. From the experimental standpoint, these strongLewis acids are well suited to an investigation of the type contemplated. Their low boiling points permit the use of high vacuum techniques for both purification and sample preparation. The acids themselves have only three vibrational frequencies active in the Raman effect which would not unduly complicate the spectra and would minimize the chance

-10of a superposition of bands of the pure acid and its complexes. In addition, these halides are sufficiently stable species that cationic fragments resulting from complexation should be characterizable. An extension of the scope of the investigation could be achieved by substituting a weaker acid for the boron and aluminum trihalides and a weaker base for DME. Sul,fur dioxide is suitable for comparison with the stronger acids because it has all of the experimental and spectroscopic advantages of the latter and evidence is available that it forms a 1:1 addition compound with DME in the solid phase. Complexation of strong Lewis acids with the relatively weakly basic alkyl halides has been indicated by H. C. Brown(6) as a result of thermal analyses of several aluminum bromide-alkyl halide systems. Most of the latter are sufficiently volatile to permit the use of vacuum techniques. From the spectroscopic viewpoint, the alkyl halides seem satisfactory also, since their only band in the 0-1200 cm.- region, in which all of the bands of the strong acids fall, is the R-X stretching frequency.

HISTORICAL BACKGROUND Etherates of the boron halides were first observed in 1846 when Ebelman and Bouquet(7) noted the formation of white, crystalline solids on mixing boron trichloride and several alkyl ethers. They reported that the solid evolved HC1 and left a solid residue which they called " methyl protoborate." Much of this work was repeated in 1889 by Gatterman(8) He correctly postulated the formation of a 1:1 addition compound when boron trichloride is added to ether. He also noted the solubility of the complex in ether. As a result of a vapor pressure-composition study of the system BC15 - dimethyl ether (referred to hereafter as DME), Wiberg(9) confirmed the stoichiometry of the complex and noticed its ready decomposition into methyl chloride and a series of esters. In succeeding papers(lo), Wiberg identified these esters and listed their physical properties as well as those of DME:BC13. No thermodynamic or spectroscopic investigations have.3 been reported. Compound formation in the BF3-DME system was first reported as a result of the vapor pressure lowering on mixing(11'12) However, the first attempt to identify the product was a thermal analysis study by Germann and Cleaveland(l3) which showed a maximum melting point at -90C for a concentration corresponding to an equimolar mixture. Considerable effort has been expended upon the study of the BF3-DME system, probably because of its marked Catalytic properties, and -11

-12numerous thermodynamic data are available such as the heat of formation of DME:BF3 (13.3 kcal,/mole)(l4). Its ultraviolet spectrum has been examined by Dundermann and Bauer(15) and an electron diffraction study has been carried out by Bauer and coworkers(l6) The ultraviolet spectrum indicated only a broadening of the characteristic ether absorption band and does not appear to offer any significant information. The diffraction study, however, and the concurrent dipole moment measurement led Korshak (17) and Lebedev'' to an explanation of its catalytic action. A Raman spectroscopic investigation of DME:BF3 was also attempted by Dundermann and (15) Bauer(l5) in connection with the electron diffraction investigation, but no attempt was made to assign the observed bands. This study was hampered by the presence of fluorescent decomposition products in the samples and by inadequate filtering of the exciting radiation. As a,result, continuous background radiation prevented attempts to detect the weaker bands and some confusion from the overlapping of spectra excited by the 4046 A and 4358 A mercury lines was reported. They did note, however, a complete disappearance of all spectral lines associated with BF3, and of all skeletal motions of dimethyl ether. As a result they were able to conclude that a very strong interaction takes place leading to the formation of a stable 1:1 molecular addition complex. It appears that some spectroscopic work on BF3-etherates has been done by Bues(l8) as part of a doctoral thesis in Germany. A footnote by Goubeau and Lucke(l9) makes reference to a stretching frequency of an unspecified etherate of BF3. Unfortunately a copy of the dissertation was not available.

-13The discovery of complex formation in the AlBr53-MeBr system came as part of a series of investigations by H. C. Brown and coworkers into the nature of catalysis in the Friedel-Crafts reaction,(20). Vapor pressure - composition curves, molecular weight data, and vapor pressure - temperature plots have given conclusive evidence for the existence of MeBr: AlBr3 and MeBr:Al2Br6 in the solid state. Compound formation in the SO2 - DME system was first suggested in 1907 by Briner and Cardoso(21) as the result of a vapor pressure. composition study. A phase diagram reported by Baume(22223) showed the existence of a stable 1:1 complex in the solid state. However, a Raman spectroscopic investigation by Wolkenstein(24) revealed only slight shifts in several of the DME bands and no changes whatever in the $02 frequencies. No thermodynamic data have been reported. Martin and Hicks(25) detected a maximum freezing point in the BC15 - ethyl chloride system at a concentration corresponding to the formation of EtCl:2BCl3. Because of the flatness of the maximum they predicted marked dissociation in the liquid phase. This same study reported only a simple eutectic point in the BC13 - methyl chloride system. No spectroscopic or thermodynamic data can be found. For the remaining systems, BX3 - methyl bromide, no information indicating.or denying the existence of complexes has been reported.

A BRIEF REVIEW OF VIBRATIONAL THEORY This topic is given more thorough treatments.in several current texts (26'2728) Only a small portion which is pertinent to the current investigation will be discussed below. From the purely theoretical standpoint, one should be able to obtain all of the internal parameters of a molecular system from it ~Schrodinger's amplitude equation for a conservative system of point particles. For a system consisting of a total of n electrons and nuclei this may be written.in the following form: n! 1/mi Vi2. + 8n2/h2 (W.V) r = 0 i=1 where mi is the mass of the ith particle,.2 is the Laplace operator in the coordinates of the ith particle, 4 is the wave functi'on.of the molecule, W is the total energy of the molecular system, and V is its potential energy. Solution of the wave equation has so far been possible for only the very simplest molecules. For such a complicated system as a polyatomic molecule, simplifying assumptions are required and only approximate solutions are possible even then. Born and Oppenheimer(29) have pointed out oneisuch assumption which permits a separation of variables into nuclear and electronic components. They argued that because of the tremendous mass differences between the electrons and the nuclei, a solution could be obtained for the electrons alone, assuming a fixed configuration of the nuclei. By then substituting -14

a characteristic electronic energy value, a function-of the nuclear coordinates, into the complete wave equation one could obtain an equation for the vibration of the nucleiN Z l/Mi vI t Ih2 [W -V1= O where N is the number of nuclei, Mi is the mass of the ith nucleus, W is the total energy of the. molecule and V its potential energy. While Born and Oppenheimer have shown that this substitution will permit a separation of the wave function into a product of an electronic and a nuclear wave function, no one has been able to obtain an exact solution for this characteristic electronic energy expression for a polyatomic molecule, Thus a complete solution of the vibrational problem from first principles is not feasible. Since this same electronic energy function appears in the nuclear equation as V, the potential energy, it is necessary to develop such a; function empirically in order to obtain.a solution, If one assumes that in the course of a vibrational motion, the displacement of the atoms from their equilibrium positions is very small in comparison to the internuclear distances, V may be expanded as a Taylor series in powers of the displacement coordinates of the nuclei. Elimination of the first power terms can be justified through the choice of the zero value for energy, while the cubic and other higher order terms can be assumed not be contribute significantly because of the small displacements from the equilibrium positions. The resulting expression has only quadratic terms and is identical

-16with the expression obtained if Hooke's Law (harmonic oscillator) forces are assumed between the nuclei. With the potential energy of the molecular system so expressed, it is convenient to effect a linear transformation to a new set of displacement coordinates, called normal coordinates, which are linear combinations of stretching, bending, and twisting motions within the molecule. As a result of this change from external to internal coordinates, six coordinates (five, for linear molecules) which describe translational and rotational motions of the molecule may be factored out and the nuclear wave equation becomes separable into 5N - 6 (or 3N-5) one-dimensional wave equations, each expressed in terms of one of the normal coordinates. While any actual motion of the molecule may be quite complicated, it may be considered as resolvable into components which are the normal coordinates. For each of the normal coordinates there are associated energy levels corresponding to the -absorption of one or more quanta of energy into that particular vibrational mode. If one considers only'"fundamental" frequencies; that is, the frequencies due to transitions between the lowest energy level of the molecule, or 1"ground" level, and the first excited level of only one normal coordinate, there will be 3N-6 (or 5) such frequencies.observable in the vibrational spectrum. It is general practice to associate each of these frequencies with some readily visualizable motion within the molecule in terms of bond extensions and angle deformations. Perhaps this is better described in terms of a specific example. The vibrational motion of-pyramidal XY3 molecules such as NH3 and PF3 can be shown to be resolvable into six components, or normal

-17l / I NI Figure 1. Internal Coordinates of Pyramidal X Y3 Molecules.

coordinates. In terms of the bond extensions and angle deformations, one may write the following symmetry coordinates which have approximately the form of the normal coordinates: S1 = N1(A r1 + A r2 + A r3) S2 = N2(Aa, 1 +~ 2+ A c 3) S = N3(A r2 - r3) s4 = N4(A2 - Aa3) S5 = N5(2Ar - A r2 - A r3) S6 = N6(2A1l - a O2 - A 3) The quantities A ri and A O~i which describe the internal motions are illustrated in Figure 1 while Ni is a normalizing constant. In discussing the motions described by the normal coordinates which are approximated by S1 and S2, for example, are customarily referred to as symmetrical X-Y stretching and symmetrical Y-X-Y deformation motions while S3, S4, S5 and S6 are spoken of as their asymmetrical counterparts. However, because the symmetry coordinates usually only approximate the normal coordinates, there may be a slight amount of angle bending in the stretching motion, and vice versa, but frequently this amount is sufficiently small so that it can be ignored in qualitative discussions. There are several factors which may serve to alter the observed number of vibrational frequencies from the 3N-6 (or 5) which are expected. One of these arises from molecular symmetry. As a consequence of the latter, the frequencies associated with two or more normal coordinates, and hence wave functions, have exactly the same energy value. In the example above, it is found that S3 and S5 give rise to identical frequencies, as do S4 and S6. As a result, one member of each "degenerate pair" of coordinates may be disregarded and the molecular system may be

-19described using four rather than six normal coordinates for the purposes of numerical calculation of the frequencies. When two vibrational modes have exactly the same frequency, they, of course, cannot be distinguished experimentally and the number of bands observable thus becomes less than 3N-6. An experimentally similar situation may arise when the energy differences between the ground level and two fundamental levels happen to be very nearly, although not exactly identical. Whether the associated spectral bands can be resolved under these circumstances depends upon the resolving power of the measuring instrument. If they cannot, the result is referred to as an "accidental" degeneracy. A spectrum showing too many bands can result from the appearance of overtones, combination bands, and difference bands. Overtones result from transitions between the ground state and excited levels other than the first, or fundamental level. The result is the appearance of new bands at some multiple of the fundamental frequency, Combination bands on the other hand, arise from simultaneous transitions between the ground level and two or more fundamental levels, while difference bands involve transitions only between fundamental levels. The frequencies of these bands are approximately the sum and difference, respectively, of the fundamental frequencies concerned,

INFRARED AND RAMAN SPECTROSCOPY In order to observe vibrational spectra one customarily irradiates a sample with electromagnetic radiation having at least the minimum energy per photon to cause a molecule to move from the ground state to a first excited level when one photon is observed. This establishes a lower limit to the frequency usable and is represented by infrared radiation with a wave length somewhere around one hundred microns. If radiation of higher frequency is used, a more complex interaction may take place, This may involve transitions to much higher vibrational levels or to higher electronic levels, or both. An upper limit to the frequency usually is determined by the dissociation energy of the weakest bond in the molecule. This generally is found in the ultraviolet region and is the basis of reaction initiation through ultraviolet irradiation. In infrared spectroscopy the compound is irradiated with a continuous spectrum of infrared radiation covering a sufficient energy range to encompass transitions from the ground state to all of the fundamental vibrational levels. In passing through the sample, those photons whose energy corresponds to an energy of transition of the molecule interact strongly with a relatively high probability of being aIsorbed, The transmitted radiation is thus depleted and if analyzed for missing frequencies, the energy differences between various molecular levels can be established. In Raman spectroscopy, monochromatic radiation of much greater energy than that required for vibrational transitions, usually visible ligkht, is passed through the sample. This results in an interaction leading

to -a scattering of light by the molecule itself. In the vast majority of cases the molecule returns immediately to its initial state and the light scattered has exactly the same frequency as the exciting source (Rayleigh scattering). Occasionally however, the molecule may return to a vibrational level other than the one it was in initially. Light scattered in this case will differ in frequency from the Rayleigh scattering by the energy difference between the initial and the final level. Since most of the molecules participating in this scattering process are initially in the ground state, most of the scattered photons will have less energy than originally. However, same of the molecules at all times are in excited vibrational levels, the numbers being distributed in accord with the Boltzman distribution law. When these molecules are involved in the scattering process, they can impart this excitation energy to what would otherwise be Rayleigh scattered light, leading to the emission of a frequency higher than that of the exciting light. On analyzing the total radiation scattered by the sample, one finds an intense band at the frequency of the exciting light and much weaker bands symmetrically placed on either side of it. Those on the low frequency side are known as "Stokes" lines and those of higher frequency as "anti-Stokes". The intensities of the anti-Stokes lines fall off very rapidly with increasing peparation frQm the Rayleigh scattering since the population of the excited molecules responsible for these. lines rapidly declines as the magnitude of the excitation energy increases. It would seem that one could obtain identical information through the use of either infrared or Raman spectroscopy since both give information

-22on molecular vibrational levels. For molecules having little or no symmetry this is roughly the case, but in many highly ordered molecular structures striking differences in transition probabilities, and hence intensities, arise because of the differing mechanisms of reaction with the incident radiation in the two techniques. Infrared absorption occurs through changes in the molecular dipole moment as the nuclei are displaced from their equilibrium position. In any molecule the instantaneous value of the dipole moment can be expressed in terms of the normal coordinates: M +MO + Xj + 1/2 LX q q qkqq +.. k k,2 where Mo is the permanent moment corresponding to a non-vibrating molecule, and qk and q are any two of the 3N-6 normal coordinates. If only one ormal'vibration takes place, the atoms oscillate in such a way that the normal coordinates may be represented by an expression describing simple harmonic motion about the equilibrium position: qk= Ak cos (2vkY t + 51) where Ak is the amplitude, Vk the vibrational frequency, andl8 a phase constant. Upon substituting for qk and q2 one obtains for the dipole moment: M =Mo + Ak cos (2g vkt + 81) k + l/4 )2M AkA2 {coS42gt(vk+ V~) + 1 a] + cos[2tt(vk- V2) + 81- 2]} + +..

-23The term (NM/&qk) governs the appearance of fundamental frequencies. If the term is zero, that is, if the molecule undergoes no change in dipole moment as a result of the normal vibration described by qk, the corresponding fundamental will not appear, or is said to be "infrared inactive" The value of the term (a2M/6qk.qe) determines the activity of overtones, combinations, and difference bands. Non-zero values for 2 = k permits overtones, while non-zero values for I ~ k allow observation of vk + vI and vk - vi. Raman scattering depends upon a change in the amplitude of the dipole moment induced by the incident radiation. The magnitude of the induced moment, P, may be expressed as a product of the polarizability of the molecule, a, and the electric field vector, E, of the incident radiation of frequency v,:according to the equation P -= E. As with the permanent dipole moment, the instantaneous value of a may be expressed in terms of the normal coordinates. a= a+ X ( k+l/2 X q k + -.. k k, With the electric field expressed in the form: E m Eo cos 2itvot, the expression.for P becOmes: P =.E0o cos 2itvot + 1/2 EoAk k (.C >{cos[2:t(t0+ Vk) + 8k] + cos[ 2:tt(v0o- k) - k]i + 1/8 EoAkAk fZ (I {cos [2tt(vo+ vk+ v2) + ok+l + cos[[2ct(v,- kk- v2)-Sk-5] + oos[2ict(vo+ vk- v2) + Bt-~] + CoS[2ict(v0- Vk+ V2)- + v]y}

Looking at the equation term by term, one finds Rayleigh scattering predicted whenever a non-zero value for ao, the polarizability of the nonvibrating molecule exists. Similarly a change in the induced moment as a result of the vibrational motion described by qk makes the associated frequency'Raman active". Overtones, combinations, and difference bands are governedby a non-zero value for (62a/3qkbqe). Raman scattered light which is observed at right angles to the incident beam is found to be partially polarized. According to classical theory isotropic molecules should scatter completely polarized light at right angles to the direction of the exciting radiation. Thus any depolarization results from the anisotropy of the molecule and vibration4l motions which destroy some or all of the molecule's symmetry will give rise to vibrational bands which show a maximum depolarization. The depolarization ratio is determined by twice recording the vibrational spectrum of the compound, each time rejecting all components of the scattered light except that which is polarized first parallel, and second, perpendicular toi the direction of the incident beam, Taking as the x-axis the direction of the propagation of the incident light, this depolarization ratio may then be defined as the ratio of the intensity of the scattered light whose electric vector is perpendicular to the yz plane, to that polarized parallel to this plane. Vibrational bands showing the maximum depolarization ratio, 6/7 may then be attributed to normal vibrations which destroy some or all of the molecule's symmetry while for vibrational motions which retain all symmetry elements, depolarization ratios ranging from zero to slightly less thawn 6/7 are found. since the

-25symmetry properties of the normal coordinates are known beforehand, knowledge of depolarization ratios can greatly facilitate the assignment of spectral bands to particular vibrational modes. In practice it is found that equivalent results can be obtained if the sample is irradiated with light which is incident in the Xy plane. With this arrangement two exposures must be taken; one with the incident radiation polarized in a plane parallel to the z-axis and another using light polarized perpendicular to it. Qualitative results can be obtained by comparing the spectrum resulting from normal incident light with that from light polarized parallel to the z axcis.

EXPERIMENTAL PROCEDURES Preparation of Materials The compounds used in the preparation of samples were: a. Aluminum bromide b. Boron trichloride c. Boron tritfluoride d. Dimethyl ether e. Ethyl chloride f. Methyl bromide g, Methyl chloride h. Sulfur dioxide Aluminua' bromide was synthesized from C. P. grades of aluminum turnings and bromine. The former was placed in a pyrex tube which was itself placed inside a tube furnace. Temperature control was maintained by inserting a thermocouple into the furnace and manually adjusting the voltage across its terminals by means of a Variac. After an initial incubation period during which the aluminum remained in contact with bromine vapors at 2500C, the reaction proceededunder its own heat. A temperature of about 55 ~C was maintained through intermittent dropwise additions of bromine while the system was flushed with dry nitrogen to remove aluminum bromide from the reaction site as fast as it was formed. The product was triple distilled to remove the last traces of bromine before storage in weighed amounts in fragile glass bulbs. All of the gaseous reagents not specifically stated as purchased elsewhere were obtained in cylinders from The Matheson Company, Inc. Most -26

-27of these gases had been analyzed by the manufacturer and these results have been included in the discussions of their purification. Although the boron trichloride had a minimum guaranteed assay of 98.5% BC13, it gave on condensation a yellow liquid phase containing copious quantities of a white solid which melted and dissolved at about O~C. A careful fractionation separated BC13 from the solid and removed most of the yellow color. Complete decolorization was obtained by the action of freshly activated Norite upon the liquid at -78~C. According to the manufacturer, a typical analysis of the BC13 used showed the following impurities: Si as SiC14............0.......... 0.007% S as S2C12.......2...%.......... 0.2% C as COC12.............. 0.8% Free chloride as C12i,............. 0.2% Only the most intense band of phosgene was detected in any spectra taken of systems containing BC13, and that as a weak line in a ten-fold overexposure of a sample of BC13 which had been fractionated but not treated with Norite. The boron trifluoride used contained 98.5% BF3, with SiF4 and S02 presumably as the only principal contaminants, each present to the extent of 0.5%. Because of the proximity of their boiling points, a separation by fractionation was extremely difficult and since the impurities are both tmuch weaker acids than BF3, it was decided to use the material after only a single trap-to-trap distillation. None of their characteristic bands ever appeared in any of the many spectra recorded of systems containing BF5 nor in the spectrum of BF3 itself. Analyses of DME indicated a purity of not less than 99.5% with the bulk of the impurity being moisture. Further purification was attempted,

-28however, by pAgsage of the gas through a column containing freshly activated alumina. The alumina was then regenerated and the process repeated. Methyl bromide (MeBr hereafter) had.a stated purity of 99.4%, the balance consisting of moisture and orgaiic chlorides. It was condensed into a second cylinder containing 15 grams of aluminum bromide. After standing at room temperature for about a week, the gas was scrubbed by water in a 150 cm. scrubbing tower packed with short pieces of glass tubing. It was dried by passage first through a column containing calcium chloride and then through a aecond containing phosphorus pentoxide suspended upon glass beads. The MeBr was finally recondensed into a stainless steel cylinder for storage until used. The methyl chloride had a stated purity of 99.5%, with moisture and residue as the only impurities. Therefore, it was used after only a single trap-to-trap distillation. Neither of the methyl halides showed bands attributable to impurities even in prolonged exposures. Ethyl chloride was purchased from The Ohio Chemical and Surgical Equipment Company. No data.on purity were available, so the liquid was carefully fractionated. The first and last cuts were rejected while the middle fraction was then used. Since The Ansul Chemical Company's sulfur dioxide was sold as 99.9% pure, it was given only a single trap-to-trap distillation before use. Prolonged exposures of the liquid revealed only the bands of the pure compound. Preparation of Samples Because of the extreme reactivity of the acids with atmospheric moisture, it was necessary to handle the compounds in a closed system. A

diagram of the vacuum line used both in the piurification of materials and in sample preparation is shown in Figure 2. The gaseous reagents were admitted through any of the stopcocks labeled "A". The presence of non-condensable gases often required a preliminary cpndensation and freezing in a pump-through trap, "B". The reagent was then permitted to warm up and expand back into the calibrated volume. By adjusting the pressure within this volume exact mole ratios having a desired total liquid volume could be obtained. Recondensation in trap B before distillation into the mixing bulb attached to the ground glass joint at C aided in the removal of entrained stopcock grease, The two mixing devices used in sample preparation are shown in Figure 3. The upper one (A) was used solely for the preparation of solutions of the systems AlBr3-MeBr and AlBr3-MeBr-HBr. Prior to use the bottom of the apparatus was cut open, permitting introduction of a small weighed bulb of AlBr3. The apparatus was then sealed, evacuated, and shaken vigorously to break the bulb. The other components of the system under investigation were introduced and solution effected., By inverting the apparatus the solution could be filtered through a sintered glass frit of "ultrafine" porosity into the Raman tube. In most of the systems investigated, all of the reactants could be distilled. In such cases the bottom apparatus (B) was used. The gases were distilled from trap-to-trap until they were finally condensed atop one another in the final bulb; the preceding bulbs were then removed by sealing off. This series of trap-to-trap distillations was designed primarily to remove entrained stopcock grease. The final trap was then

-30TO TRAPS AND PUMPING SYSTEM _,, A TO MANOMETER C GROUND GLASS JOINT Figure 2. Vacuum Line Used in Sample Preparation.

-31GROUND GLASS RAMAN TUBE JOINTS A RAMAN TUBE GROUND GLASS JOINT Figure 3. Apparatus Used in the Preparation of Raman Samples.

-32permitted to warm up to dry ice temperature and the interaction, often quite violent, took place. In many cases a prolonged swirling of the trap while in an appropriate cooling bath was necessary to effect solution, but in some a solution was obtained at once. The trap was then inverted so as to pour its contents into the Raman tube, which was then sealed off. Spectrographic Equipment Very little of the exciting radiation gives rise to Raman scattering and only a minute fraction of this scattered light can be collected and analyzed because of physical limitations. Thus thte Raman spectroscopist must endeavor to obtain a monochromatic light scource of maximum intensity, collect as much of the Raman scattered light as possible, and yet prevent as much as possible of the exciting radiation which is not actually scattered by the sample from reaching the spectrograph. These requirements are approached by a variety of techniques. The most satisfactory light source currently available is the helical Toronto-type low pressure, low temperature maecury arc. Most of the energy released from mercury's electronic transitions is concentrated in a few relatively intense bands. Those in the visible range are well separated, thus facilitating the task of obtaining monochromatic light. Continuous background radiation and pressure broadening of the exciting line are minimized through the use of large water cooled pools of mercury for electrodes which favor a low pressure in the arc. If the cold finger, by which the cooling is accomplished, extends well above the surface of the mercury pool into the arc proper, transferal of the mercury from one

-33electrode to the other during operation is greatly reduced. The arc is in the form of a helix made of 32 mm. Pyrex tubing which is wound into three full turns. The arc is operated on 220 volts and normally draws about 24 amps. Despite the fewness of the spectral lines from the mercury arc, it is desirable to utilize light filters to simplify interpretation and reduce the overlapping of spectra. Although there are very few intense bands in the visible range of mercury's spectrum, there are a number of very weak bands. These may obscure Raman bands unless removed from the incident light. Failure to reduce the intensity of the other intense lines can lead to overlapping Raman spectra, Spectra from the 4358 A. mercury line, which was used exclusively, were isolated through the simultaneous use of two filter solutions as shown in Figure 4. A nearly saturated solu~ tion of sodium nitrite in water was circulated through the outer filter jacket, This served two purposes: it reduced the intensities of the mercury lines at 4046 and 4078 A. to the point where Raman scattering resulting from them was no longer detectable, and because of its continuous circulation through a heat exchanger, it served to keep the heat of the arc away from the inner filter Jacket and the sample. The inner filter jacket was sealed and easily removable, Several of these filter jackets were available, each filled with a solution adapted to some special purpose. Generally a solution of ethyl violet in isopropyl alcohol was used. This removed the multitude of weak mercury lines in the 4700 to 5000 A, region and reduced the intensity of the 4916 A. line to the point where Raman scattered light arising from it could not be detected

-34DEWAR FLASK INNER FILTER OUTER FILTER JACKET JACKET 0 0 LIGHT SHIELD AIR INLET 5 CM. Figure 4. Raman Light Source Assembly.

-35When -spectral bands of very low frequency were being studied, it was helpful to substitute a solution of praeseodymium chloride. This served. to reduce the background radiation in the region close to the 4358 A. exciting line. During polarization studies, the inner filter jacket was replaced by collimating baffleso These helped approximate the theoretical requirement that polarized spectra be observed at right angles to the plane of the incident radiation. Spectra of samples at very low temperatures were obtained through the insertion of a specially constructed. Dewar flask into the space inside the inner filter jacket, The temperature of the interior could be maintained anywhere in the range of 0 to -55~C by passing precooled air into the air inlet at the bottom. This stream of air was first dried by anhydrous calcium chloride and then cooled as it passed through a copper coil which was immersed in a dry ice - isopropanol slush~ For lower temperatures it was necessary to pass nitrogen gas through the copper tubing, In this way temperatures down to -130 C could be maintained for relatively long periods of time. A rough control of temperature was possible by regulation of the rate of flow of the gas, but a more satisfactory continuous monitoring and control system was available. This consisted of a thermocouple in the air stream alongside the sample tube connected to a Leeds and Northrup strip chart recorder. The temperature of the air stream could be adjusted on demand by a small electrical heater in the stretam immediately following the cooling coil. In normal operation, a fluctuation of about five degrees was observed in the air stream in the area.of the sample tube. There was in

-36addition a thermal gradient of approximately ten degrees along the length of the Raman tube. Alignment of the optical system was greatly facilitated through the use of the sample holder shown in Figure 5. This fitted snugly inside the Dewar flask, which was itself firmly mounted in the light source assembly. The sample holder was fitted with a series of adapters in order to permit use with a variety of sizes of Raman tubes. It could also be used to support a wrapping of Polaroid. film for polarization studies. In order to minimize the amount of stray light reaching the spectrograph, only the Raman-scattering perpendicular to the plane of the exciting light was analyzed. This was accomplished by using as a sample tube a cylinder of Pyrex tubing whose lower surface consisted of a plane window. By illuminating only the central segement of the tube and blackening the curved outer surfaces above and below this section, most of the incident illumination could be excluded from the Raman and Rayleigh scatters ed light. In addition, extreme care was exercised to exclude from the sample tube any dust or other foreign matter which might cause Tyndall scattering. This was done by steaming out the interior of the tube before use, careful purification of all sample materials, and especially by filtering samples whenever possible. The Raman scattered light was analyzed with a Gaertner two-prism spectrograph having a dispersion of about 180 cm, 1 per millimeter in the blue region. The spectra were recorded photographically using Eastman-Kodak 103a-J plates with an antihalation backing. This particular combination of emulsion

-37SAMPLE TUBE GUIDES SUPPORT POST I i i 5 CM. LENS COMPARTMENT X CORK CUSHION Figure 5. Sample Tube Holder.

-38and. sensitivity was chosen because of high response to light of low intensity in the blue region. The plates were developed for four minutes at 20~C in D-19 solution and fixed in F-6. Most of the plates so obtained were measured using two techniques. The first of these was the use of a comparator which measured directly the distances.of the.Raman lines along the plate and could be read to the nearest 0O.001 mm. The frequencies of the Raman bands could then be calculated from a dispersion curve based on the spectrum of argon. In the other method, microphotometer tracings, representing about a twenty-six fold enlargement of the plates, were obtained with a Leeds and Northrup microphotometer. The distance between peaks could then be measured by means of a modified cathetometer and the displacements from the exciting line calculated from -a second dispersion curve, again based upon the argon spectrum.

EXPERIMENTAL RESULTS AND THEIR INTERPRETATION Raman Spectra of Pure Components Since complex formation was to be detected through changes in the vibrational spectra of the pure components, it was necessary that these spectra be available. Although all had been reported in the literature, they were recorded once again primarily for the purpose of detecting impurities. The spectrum of aluminum bromide only will be reported here briefly since for the other molecules both the number of bands and their frequencies has been well established by earlier workers. Subsequent to the time that Rosenbaum(30) and Gerding and Smit(31,32 reported the Ruman spectrum of liquid aluminum bromide, the bridge structure has been proven conclusively. From group theoretical considerations it can be shown that of the 18 fundamental frequencies expected for this model only 9 are active in the Raman effect, while 8 others are infrared active and one is inactive in both. Thus a complete assignment of the vibrational frequencies of aluminum bromide would require knowledge of its infrared and Raman spectra, including its polarized Raman spectrum. In addition, it would be desirable to have similar data for the other aluminum halides in order that the assignments be unequivocal. The only such data which were available for the aluminum halides prior to this research were their Roman spectra obtained using natural incident light. Because of the magnitude of the undertaking and the experimental difficulties certain to be encountered, especially in the recording of infrared spectra, it was decided to make no attempt at assigning the vibrational frequencies of aluminum bromide. The Ramdn spectrum of aluminum bromide is shown in Figure 6 and the frequencies 39 —

obtained are compared in Table I with the results of the earlier workers, I00 200 300 400 500 CM-' Figure 6. The Raman Spectrum of Al1Br6. TABLE I VIBRATIONAL FREQUENCIES (IN cm.-l) OF A12Br6 Rosenbaum Gerding and Smit This Research 67 67 73 79.3 78 112.8 112 11l 140.3 140 140 185 176 183 208.2 204 209 223 221 223 291 407 407 402 440 488 491 489

-41Systems Showing No Interaction in the Liquid Phase Spectra of equimolar solutions of BC13 and methyl chloride, BC13 and. methyl bromide, and BF and methyl bromide were recorded at -550C in the first two cases and.at -900C in the last. All showed only the superimposed spectra of the pure components with no frequencies shifted. Systems Showing Interaction But Not Complexation in the Liquid Phase.BC13-Ethyl Chloride At -55~C the spectrum of a mixture containing two moles of BC13 per mole of ethyl chloride was that of the reagents. However, at =100~C the C-C1 stretching frequency, which occurs at 655 cm-l in ethyl chloride, -1 appeared at 642 cm. This indicates a slight weakening,of the C-C1 bond due to the action of BC13. None of the frequencies of BC1l were shifted, none of the other frequencies of ethyl chloride were affected, and no new bands were found, Thus the interaction appears to be a solvent effect rather than a case of complexation. SOp-DME A brief temperature-pressure study of an equimolar mixture of 52 and DME was conducted to see if any pronounced interaction could be detected. The results are shown in TaDle II, in which the observed pressures are compared with those of an ideal solution. TABLE II VAPOR PRESSURE (IN mm.) OF AN EQUIMOLAR MIXTURE OF SO2 AND DME t(~C. P Pideal sol. Pobs. -78 25 5 -23 580 265 O0 1500 749

-42This pronounced vapor pressure lowering is certainly indicative of interaction, but not necessarily of complexation. The Raman spectroscopic investigation by Wolkenstein(24) was conducted at room temperature with a sample containing sulfur dioxide and DME in the ratio 3:2. This work has been extended in the present case by examining l:l mixtures at -25 and -55OC. The study of HC1-DME complexes by (2) Vidale and Taylor showed the sketetal frequencies of DME to be a very sensitive indicator of cqmplexation; that is, these frequencies were markedly shifted, usually to lower values, by interaction with the acid. The effect upon these frequencies by SO2 is given in Table III. TABLE III THE EFFECT OF SO2 AND HC1 UPON THE SKELETAL FREQUENCIES (IN cm.-l) OF DME C-O asym, C-O sym. COC deformastretch stretch tion Pure DME 1095 920 42Q DMlE-SO2 mixture (3:2) at room temperature * 1088 908 411 D/ME-S02 mixture (1:1) at -25~C. 1086 910 417 DME-$02 mixture (1:1) at -559C. 1087 904 420 DME-HCl mixture (1:1) at -909C. 1077 886 420 However, the frequencies of SO2 itself in these solutions are exactly the same as those of the pure liquid. The smaller shifts in the skeletal

-43frequencies of DME together with the lack of an effect on the S02 frequencies is interpreted only as an indication of a slightly temperature dependent solvent effect occurring in the liquid phase. The situation further differs from the DME-HC1 case in that two sets of frequencies corresponding to complexed and uncomplexed DME were never observed. Thus both DME:S02 and EtCl:2BC13 reported from phase studies appear to be crystal lattice compounds which exist only in the solid state. Systems Showing Complexation Aluminum Bromide - Dimethyl Ether. Products resulting from the interaction of ALBr3 and DME were found to be ill-suited to Raman spectroscopic study because of their slow deqomposition into products which fluoresce in the visible region. The rate of this decomposition increased so rapidly with temperature that the Raman spectrum of the molten complex could not be recorded despite its low melting point (47~C*), Since examination of the solid appeared out of the question using the Raman technique, the possibility of studying the complex in, solution was considered. Of the many proposed solvents which were examined only CS2, S02 and MeBr appeared to offer sufficient promise to warrant closer study, Sulfur dioxide instantly dissolved ether-rich mixtures but greatly accelerated their decomposition while neither the 1:1 complex nor AlBr3 was significantly soluble in $02. Despite the very low solubility reported by Foster and Kraus(33) for DME:AlBr3 in MeBr (0.3 moles per liter at -78~C), it was hoped that increased solubility at higher temperatures might compensate for the accompanying increased rate of decomposition. Unfortunately, such was not

-44the' case. Lastly, carbon disulfide itself was found to decompose under the intense light of the Toronto arc and was eliminated.. Since the possibility of further work -appeared. discouraging,, the investigation -of the system was discontinued. Aluminum Broide - Mth Bromide The same general solubility problems encountered in the A1Br3.-DIE system -were present here since the 1:1 ratio marks the upper limit of solubility of AlBr3 in MeBr., No suitable solvent could be found for acid-rich mixtures, but MeBr served.as an excellent solvent for the 1:1 complex. As in the case of DME:AlBr3, these solutions slowly decomposed to give products which fluoresced in the visible region, However, the rate of decomposition was much slower and satisfactory spectra were recorded. Polarized spectra,requiredmuch longer exposures and were therefore more seriously affected by this decomposition. As a.result, the assignment of the vibrational bands is hampered by the necessarily underexposed polarize.spectrum. The spectra obtained.are shown in Figure 7. Vapor pressure measurements by Brown and Wallace have indicated that in all mixtures of MeBr and AlBr3 in which the MeBr/AlBr3 ratio is unity or greater, the only complex species present is MeBr:AlBr30 This conclusion is confirmed by the small number of vibrational bands in the 0-500 cml region, since a complex containing A12Br6 would be expected to show at least ten bands in this region (see Figure 6) whereas the pyramidal AlBr3 structure would. show a maximum -of seven. Since only five vibrational bands in the 0-500 cmln region could be assigned to fundamental frequencies, the structure by Brown-and Wallace seems quite reasonable,

-45NATURAL INCIDENT LIGHT POLARIZED INCIDENT LIGHT E, 100 200 300 400 500 600CMI RAMAN SPECTRA OF AIBr3C CH3Br Figure 7

The intense band at 595 cm.'l is the C-Br stretching frequency of the solvent, MeBr. The corresponding band in the 1:1 complex.appears about 40 cm.-l below it, which is an energy difference of 0.12 kcal per -1 mole, The intense band at 200 cm. is analogous to the intense, strongly polarized bands found by Gerding(34)35) in several AlC13 complexes at frequencies near the most intense band of pure A12C16. In each case these bands were assigned to the Al-Cl stretching.frequepcy0 The most intense line in the A12Br6 spectrum occurs at 209 cm. This band and the strong, -1 completely polarized band at 200 cm. in MeBr:AlBr3 are clearly Al-Br stretching frequencies. Asymmetric stretching bands are customarily found at slightly higher frequencies, are much broader, and appear with very little reduction in intensity in spectra observed using polarized incident radiation. Despite the reduction in intensity of the broad band at 277 cm.'l in going from natural to polarized incident radiation, this ras been assigned to the asymmetric Al-Br stretching mode because of its contours and position, Because the polarized spectrum was grossly underexposed, the fact that this band appears at all in the latter may be evidence for its unsymmetrical nature, The lower-lying bands have been assigned to defprmations largely through the process of elimination. A tabulation of the frequencies, their polarization, and relative intensities is given in T~ble IV together with a set of tentative assignments, assuming C3v symmetry. The Systems: Boron Trihalide - Dimethyl Ether Raman Spectrum of DME:BF3. The slow decomposition into fluorescent products which interfered so seriously with Raman spectroscopic analyses of sMIE:AlBr3 and MeBr:AlBr3 appeared

-47TABLE IV THE RAMAN SPECTRUM OF MeBr:AlBr3 Frequency State of Tentative (cm,-l) Polarization Intensity Assignment 71 Depolo w AlBr3 wag or rock (e) 102 Depol. m AlBr3 def. (e) 149 _ w AlBr3 def. (a1) 199 Pol. s Al-Br Str. (al) 277 Depol. m Al1Br str. (e) 454 Depol. w 5545 - 101.76 = 452.6 467 w 276.8 + 199.4 476.2 554 Pol1 s C-Br str, (al) here aso., but at a rate sufficiently low that spectra could be recorded using both natural and polarized incident radiation, These are shown in Figure 8. In the discussion of the frequencies of the BF3:D1ME complex, a natural grouping is to consider three sets of bands, (1) those derived from free.DME bands, (2) those derived from free BF3 bands, and (3) those arising as a consequence of the formation of the B-O link. The spectral bands of DME:BF3 which arise frnom motions of the methyl groups should have nearly the same frequencies as the corresponding bands of DME, As result, bands in the 3000, 1450, and 1000-1200 cm:regions are to be expected from C-H stretching, CH3 deformation, and CH3 rocking modes. The effect of complexation should be more pronounced in the case of the so-called skeletal motions, viz. the symmetric and

-48NATURAL INCIDENT LIGHT POLARIZED INCIDENT LIGHT E,, 500 1000 150.0 o200 2500 3000 CM.g RAMAN SPEGTRA OF (CH3)2 O:BF3 Figure 8

-49asymmetric C-O stretching and the deformation of the C-O-C angle. Vidale and Taylor(2) found these frequencies reduced by as much as 10% as a result of complexation with the halogen acids. By analogy, the corresponding bands in DME:BF3 should be a strong, polarized band between 800 and 850 cm.-l and an only slightly less intense, depolarized band in the 9001000 cm. -l region. The C-0-C deformation frequency, occurring at 420 cm.'l in DME would be expected to be shifted downward to perhaps 350-375 cm.1 This band should also be polarized. depolarized neighbor at 921 have been assigned to the symmetric and asymmetric C-0 stretching motions. The only polarized line below 420 cm.o' is found at 326 cm.'l and has been assigned to the C-O-C deformation. The assigning of frequencies contributed to the complex by the BF5 group is rendered more difficult by the scarcity of spectral data of molecules containing this grouping. However, a wealth of data on molecules containing the CF3 group and the same symmetry (a single reflection plane) is available. The C-F stretching and CF3 deformation frequencies of eight such compounds have been assembled in Table V. The polarized bands have been designated (A') and the depolarized by (a") which respectively preserve and destroy the plane of symmetry of the molecule during the course of the vibration. Generalizing from these data, one may attribute to the CF3 group three stretching and three bending frequencies, the former occurring between 1100 and 1300 cm.-l and the latter between 500 and 700 cm.l1. In addition, two of each group of three will be polarized, that is, suffer a reduction in intensity when

-50TABLE V TYPICAL VALUES OF THE C-F STRETCHING AND THE CF DEFORMATION FREQUENCIES (cm l) Compound C-F Stretch CF Deformation Ref. a' a a a' a'' a" CF3CF2C1 1224 1349 1240 560 647. 593 36 CF3CFC12 1247 1292 1218 505 588 559 36 CF3CH2C1 1152 1262 1285 (455) 639 537 37 CF3CHC12 1140 1270 1226 526 671 554 37 CF3CH2Br 1075 1240 1135 526 632 539 38 CF3CH2I 1098 1210 1252 513 626 39 CF3CO2H 1128 1199 1240 566 580 507 40 CF3COC1 1190 1279 1237 584 710 39,radiated with incident radiation which is polarized parallel to the axis the sample tube. Although it is not indicated in Table V, the intensities all of these bands are very low. Except for the higher-frequency CF3 deformation, every band listed was classified by the author as "weak" or "very weak". The method of transferring these values to anticipated frequencies of the BF3 group is based upon a comparison of the frequencies of CF4 and BF4k That this can be dornd, has been demonstrated by Woodward and Nord(41) who compared the vibrational frequencies of GaBr4 and InBr4 with those of the isoelectronic species GeBr4 and SnBr4. A constant difference of about 10%

was noted, with the ions having the lower values. A similar comparison of BF4 and CF4 is given in Table VI. TABLE VI THE VIBRATIONAL SPECTRA OF BF- AND CF4 (cmo-l) V1(A1) v3(F2) v4(F2) v2(E) Ref. BF4 769 984 524 352 42 CF4 904 1265 630 437 27 BF4/CF4 0.85 0.78 o0.83 0.81 Assuming that frequencies associated with the BF3-group will also be approximately 80% of the corresponding CF3-group frequencies, one would expect three stretching frequencies between 900 and 1100 cm.l and three deformations in the region from 400 to 560 cm. 1. It can be seen that there is an overlapping of the B-F stretching and CH3 rocking bands and neither can be assigned with any certainty. On the other hand, the only band in the 400-560 cm,. region is found at 503 cmo.- and must be the intense, symmetric BF3-deformation frequency. In addition to the bands derived from the original molecules forming the complex, there will be six new bands arising because of the loss of translational and rotational degrees of freedom in the separate components. In the case of the R20:BX3 complexes, these new vibrational motions, which are characteristic of the complex itself, may be approximately described as a B-O stretching vibration, rocking motions of the BF3 group both in the plane of the DME and out of the plane, similar rocking motions of the DME portion

-of the molecule in plane and out of plane, and.a twisting of the BF3 end around its axis with respect to the other end of the complex. Goubeau,, (43) and Lucke have reported the B-0 stretching frequencies.of a series of methyl esters of BF3 as falling in the range from 630 to 770 cm.-. Therefore, the 664 cm. band, which is relatively intense and polarized, has been assigned to this motion. The strong, depolarized band at 348 cm.-l has been assigned to a.rocking motion of the BF3 group. Although values for this frequercy were not included in Table V, all of the compounds listed there have a strong band in the region between 250 and 350 cm.-l which is due to such a motion. There is the possibility that this is one of the BF3 deformation frequencies instead., but because of its intensity, the assignment to a rocking motion is more likely. Vibrational frequencies associated with three other motions characteristic of the complex itself, namely the BF torsional motion and the in-plane and out-of-plane DME rocking motions, could not be assigned because all of the observed bands could be accounted for otherwise. A BF3 twisting frequency probably would be of rather low intensity and moreover would be expected at a rather low frequency, probably less than 100 cm.. Failure to observe it must then be attributed tQ a combination of these factors and is not unexpected. The DME rocking motions would. also be expected in this region and may not have been observed becuase of their low intensity. Broadening of the 4358 A. mercury line which occurred in all solutions prepared using BF3 may well have prevented observation.of these bands. A summary of the frequency values, polarizations, intensities, and assignments is given in Table VII,.

-53TABLE VII THE RAMAN SPECTRUM OF (CH3)20:BF3 2. i-. -.',,,.,, ~'.,, " J..: " Frequency Intensity State of Assignment Polariza(cm.) tion 326 mn Pol. C-O-C deformation 348 s Depol. BF3 rock or BF3 deformat ion 503 s Pol. BF3 deformation 664 s Pol. B-O stretch 806 s Pol.: C-O stretch 921 m C-O stretch 1021 m B-F stretch or CH3 rock 1149 w B-F stretch or CH3 rock 1222 vw CHR rock -or CH3 deformation 1262 vw CH3 rock.or CH3 deformation 1271 vw CH3 rock or CH3 deformation 1455 s.Depol. CE3 deformation 2861 m Pol. 2894 s Pol. 2981 n Depol. C-H stretch 3049 m Depol. C-H stretch * These bands may be the fundamentals of symmetric C-H stretching frequencies or the first overtone of the CH3 deformation in strong Fermi resonance with the stretching frequencies.

-54Raman Spectrum -of DE E:BC1 Spectra obtained through the use of both natural and polarized incident radiation are shown in Figure 9. Since methyl chloride was employed as.a solvent, its spectrum-is superimposed upon that of the solute. The symbol "x" indicates bands due to the solvent. As in the preceding section, the spectrum will be considered as being made up of three groups of bands; those arising from BC13, those arising from DME, and new bands attributable to motions of the complex which in its components represented rotational and translational motions. Looking first at DMEI's contributions to the spectrum of the complex, one may argue through the same reasoning as in the BF3 case that the bands occurring in the 1100 - 3000 cml region can be assigned at once to C-H stretching, CHR deformation, and CH3 rocking motions of the methyl group. The intense, strongly polarized band at 845 and its only slightly less intense neighbor at 976 cm."1 are immediately recognizable as the symmetric and asymmetric C-0 stretching frequencies while the C-0-C deformation frequency must be one of the two polarized members of the triplet which covers the region between 350 and 400 cm.,l. In fact, it can only be the band at 371 since -1 the alternative assignment would represent a shift of only 20 cm. from its position in.DME. In view of the corresponding value in DME:BF3 (326 cm-1) a greater shift is indicated. Frequencies of the complex which may be considered as derived from BC3lmay be deduced by examining a series of compounds containing the CC13ygrouping The stretching.and deformation frequeq;ies of such a series are given in Table VIII.

-55x xx NATURAL INCIDENT LIGHT I I I...... I,,......,. POLARIZED INCIDENT LIGHT E, 500 1000 1500 2000 2500 3000 RAMAN SPECTRA OF (CH3)2 O:BC13 Figure 9

-56TABLE VIII TYPICAL VALUES,OF THE C-C1 STRETCHING AND THE CCl1 DEFORMATION FREQUENCIES (cm.-l)......... -,-,1 - ------,:. -I., ~,...... Compound C-C1 Stretch CC13 Deformation Ref. a' a' a" at' a' a" CC13CH2C1 549 747 717 242 34 307 44 CC13CHC12 584 725 773 224 327 327 44 CH2CHCC13 538 792 281 315 436 45 These indicate that C-C1 stretching frequencies occur in the region between 550 and 800 cm., while CC13 deformations are found at lower values, 250 - 400 cm.1. Preliminary studies by Greenwood(49) indicate that the vibrational frequencies of BCl; are about 85% of the corresponding frequencies of CC14. Assuming that frequencies associated with the BC13-group will also be approximately 80% of the corresponding CC13-group frequencies, one would.expect three B-Cl stretching frequencies between 450 and 800 cm. -l.and three BC13 deformations in the region from 200 to 400 cm.-1 The intensities of the C-Cl stretching frequencies show a definite pattern. The lowest frequency, found between 500 and 600 cm. 1, is of the three, usually the most intense band in the entire spectrum and it is also strongly polarized. Such a band in the spectrum of DME:BC13 is found at 494 cm.1l and must therefore be assigned to one of the symmetric B-C1 stretching modes. The only other strongly polarized band in this region appears at 664 cm.-1 and must be attributed to the B-0 stretching motion. The bands

-58TABLE IX THE RAMAN SPECTRUM OF (CRH3)20:BCl3 Frequency Intensity State of Assignment Polariza(cnm.-) tion 178 w DME rock 189 m Depol. BC13 rock 242,m BC13 rock or BC13 deformation 282 w Depol. BC1, rock -or BC13 deformation 358 s Depol. BC13 deformation 371 s Pol. C-O0-C deformation 399 s Pol. BC13 deformation 494 s Pol. B-C1 stretch 664 m Polo B-O stretch 755 m B-C1 stretch 795 w B-C1 stretch 845 s Pol. C-0 stretch 976 s Depol. C-0 stretch 1136 w CH3 rock 1157 vw CH3 rock 1180 m Pol. CH3 rock 1245 w CH3 deformation 1349 m CH3 deformation 1447 s Depol. CH3 deformation C-H stretching frequencies have been omitted because of interference by solvent bands.

-59Conclusions.Drawn From a Comparison of the Two Spectra.Errors made in assigning spectral bands. can often be detected by comparing corresponding bands in a series of related compounds. One such comparison is given in Table X in which the frequencies of the skeletal motions of DME are compared with those of DME:BC13 and DME:BF3. TABLE X A COMPARISON OF THE SKELETAL FREQUENCIES OF DME, DME:BC13, AND DME:BF3 (cm-l) C-0 sym, C-O asym. COC defstretch stretch ormation..DME. 920 1095 420.DME: BC13 845 976 371 DME:BF3 806 921 326 The formation of a strong dative bpnd between oxygen and boron would be expected to cause these frequencies to appear in the complexes at lower -values than.in.free DME, the amount of lowering.depending..upon both the -mass of the coordinating group and its acid strength. Since the two C-0 stretching.and the C-O-C deformation frequencies were found at lower values in DME:BF3 than in DME:BC13 despite the greater mass of BC13, one could then argue that BF3 is a stronger acid than BC13. The observation of B-O stretching frequencies at exactly the same positions in DME:BF3 and D.E:5BCi3, however, is surprising. One would expect this frequency to be lower in the latter case because of the greater mass of BC13 and also because BF3 is generally considered to be a stronger

-60Lewis acid than BC13. Two explanations of this anomaly can be suggested: first, the geometry of the system may be such that the B-O stretching frequency is practically independent of the mass of the halogen atoms, and second, the B-0 stretching force constant may be different in DME:BC13 than in DME:BF3 such that fortuitously it exactly compensates for the mass effect. To study the effect of mass and geometry upon the B-O stretching frequency, a simplified model system, XYZ3, having a 3-fold symmetry axis, was considered and the effect of various ZYZ angles upon its calculated vibrational frequencies was. followed. In order that the model system be as closely related as possible to the dimethyl etherates of the boron halides, the mass of X was chosen to be that of oxygen, Y that of boron, while the mass of Z varied from 19 to 80 atomic weight units. This artificial system is roughly equivalent to a simple.mechnaical model of a_-XB-0 molecule whose force constants are unaffected by molecular geometry. In carrying.out the calculations, a consistent set of force constants was selected and the totally symmetric vibrational frequencies of X3B-O were calculated as a function of the argle variable. The details of these calculations are given in the Appendix, but the effect of the ZYZ angle upon the X-Y stretching frequency is shown in Figure 10 in which the calculated value of the X-Y stretching frequency is plotted. gainst the mass of X for four different values of a, the Z-Y-Z angle, From this figure it can be seen that the B-a stretching frequercy in the dimethyl etherates of the boron halides can be expected to be appreciably mass dependent for all of a series of X-B-X angles ranging from 1060 to 119.0, * Perhaps a larger value should have been chosen but 16 was selected as a first approximation.

900 800 700 T - 600 500< 1 500 400 I, I, a I I I, 0 10 20 30 40 50 60 70 80 Mx Figure 10. The Mass Dependence of VXy for Various ZYZ Angles.

-62These results appear to eliminate the first explanation of the coincidence of the B-O stretching frequencies of DME:BF3 and DME:BC13 as being due to geometrical factors. To obtain positive evidence for the second explanation, one would have to investigate other systems involving other ethers and perhaps use BBr3 and BI3 as acids as well as BC13 and BF3. The second explanation lends credence to Brown's contention that BC1, is the stronger acid(50) Further investigation of this point appears desirable. A rough check of the assignments of the B-X stretching and BX3 deformation frequencies of DME:BF and DNE:BC3 can be made by comparing these frequencies with those of BX3 and BX4. In Table'XI the B-F stretching and deformation frequencies of DME:BF3 are compared with those of BF3 and BF4 TABLE XI A COMPARISON OF THE B-F STRETCHING AND THE BF3 DEFORMATION FREQUENCIES OF BFs, DME:BF3, AND BF4 (cmol) B-F stretch BF3 deformation (sym.) (asym.) (sym.) (asym.) BF3 888 1446 480 691 DME:BF3 (1021) (1222) 348 503 BFZ 767 984 353 524 One would expect these frequencies to approach those of BF4 in complexes with strong Lewis base and with the exception of the stretching frequency at 1021 cm,'l this is observed. As was mentioned earlier, the B-F stretching frequencies are of very low intensity and are overlapped by methyl-group

-63rocking frequencies. Therefore the band at 1021 cm.1 may be incorrectly assigned and may instead be either a methyl group rocking frequency or ~the second, asymmetric, B-F stretch, A similar comparison of DME:BC13 is not possible because of the lack of spectral data for BC1j, but a comparison with BC13 shows the symmetric B-Cl stretching frequency of DME:BC13 also above the corresponding TABLE XII A COMPARISON OF THE B-Cl STRETCHING AND THE BC13 DEFORMATION FREQUENCIES OF BC13 AND DME:BC13 (cm. l) B-C1 stretch BC1 deformation (sym,) (asym.) (sym.. ) (asym.) e,;-,.....,-.':....- -.,BCl5 471 958 243 462 (242)..DME:BC13 494 755 795 (282)358 399 band of BC1l3 In this case, however, the assignment of the band of the etherate ismore certain and thereby, in a sense, justifies the assignment in the BF3 case. Further work with other complexes of BF3 and BC13 with Lewis bases will be necessary before unequivocal assignments can be made for DME:BF3. The Spectra of BX3:DME in Ether-Rich Solutions One of the goals of this research was the study of acid-rich mixtures of a strong Lewis acid and DME in hope of observing higher order,complexes as in the DME-HCl system, However, the possibility, of higher order complexes in which more than one mole of ether was present per mole

-64of acid was not ignored since several workers have reported such complexes in systems involving DME or diethyl ether and strong Lewis acids such as A1X3 and GaX3 (46) TiQ14 and SnC14( ). As a result, a series of Raman spectra of' solutions having various BX3-DME ratios were recorded and are shown in Figures 4l and 12. Since no changes were noted in the C-H stretching.or CH3 deformation regions, these parts have not been included. The presence of excess DJME actually has very little effect upon the spectrum of DME:BF3, ~The only band of the complex which appears to be affected is the B-0 stretching band which is shifted from 664 cm.-1 in the 1:1 mixture to 673 cm. in the excess ether solution. Such a small change can be attributed to solvent effects. The skeletal bands of DME appear at 420, 920, and 1.)95 cm.9l in the mixture, the increased relative intensity of the band at 920 cmo'l being due to a superposition of bands of -the ether and the complex. At a first glance, the effect of DME appears more pronounced in the case of DM8E:BC1 since several newibands appear. Several of these new 3 bands indicate the presence of a significant amount of methyl chloride, however, and since none was used for solvent purposes, this can only have,I (20) come from the decomposition of DME:BC13. Wiberg and Sutterlin have studied this decomposition and have found the products to be methyl chloride and several methyl chloroborates. The additional new bands at 469 and 573 -1 cm. are undoubtedly due to one of these esters, probably BC120CH3, while the other new bands (at 420, 920, and 1095 cm. 1) are due to the DME present in the mixture. Some interaction of DME with DME:BC13 is indicated, however, by shifts in the skeletal frequencies of the latter. The C-0-C deformation

-65(200 800 000 1 200 400 600 800 1000 1200 GM0-1200 REGION FOR VARIOUS BF3-(CH3)20 RATIOS Figure 11

-66I_ (5,1) I: I \ I I. I.../\/ q 1 \I I I, I. I, 300 400 500 600 700 800 900 1000 CM-' 300-1000 REGION FOR VARIOUS BCI3-(CH3)20 RATIOS Figure 12

-1band at 371 cm. appears at 378 cmo with a noticeable gain in intensity. The B-O stretching frequency is shifted downward in this case, occurring at 641 cm.. The asymmetric C-O stretching frequency, on the other hand, has been shifted from 976 to 987 cm. 1. The fact that all of the bonds about the oxygen atom have been affected is, in one sense, a verification of the correctness of the assignments of the bands associated with the DME part of the complex. The Spectra of BX5:pME in Acid-Rich Solutions The addition of excess BF3 to DME:BF3 introduces several additional bands into the vibrational spectrum besides those of the free acid. A band at 690 cm.-l appears in the 3:2 mixture as a shoulder on the B-0 stretching band and finally becomes equally intense in the 5:1 mixture. Similar behavior is shown by the band at 833 cm. 1 which appears in the 3:2 mixture as a weak shoulder on the symmetrical C-0 stretching band. This shoulder gains intensity in solutions containing more acid until it appears equally intense in the 5:1 mixture. The B-0 and symmetrical C-O stretching frequencies are at the same time shifted to higher and higher frequencies, the B-O stretching frequency going from 664 cm-?1 to 667 in the 3:2 mixture, to 673 in the 2:1 solution, and appearing finally at 685 cm, in the 5:1 ratio. Clearly a new species is present, but whether it has come from interaction of DME:BF3 with BF3 or from decomposition of the complex cannot be stated with absolute certainty. However, because of the regular gradations in intensity obtained from freshly prepared samples, the former seems more likely.

-68The spectrum of DME:BC13 in solution with excess BC13 does not indicate interaction.of the complex and the excess acid. The new bands are those of BC13, their overtones, and one of the bands (575 cm.'l) which in ether-rich solutions were attributed to the decomposition of the 1:1 complex. The other may be obscured by the BC13. The B-O stretching frequency appears at 666 vm.-l in the 5:1 mixture and the asymmetric C-O stretch at 984 cm.J1 It is of interest to note that the effect of an excess of either DME or acid upon the dimethyl etherates of the boron halides is upon bonds about the oxygen atom. Although the formation of higher order complexes is suggested only in the. case of BF3:DME with excess BF3, the shifting of the frequencies associated with these bonds suggested that the point of attack of excess component is at the oxygen atom and is not at the first boron atom through formation of a single-bridged structure. The Effect of HC1 Upon DME:BC1 The Raman spectrum of a 1:1 mixture of DME:BC1l and HC1 was recorded to check the possibility that HC1 might complex with DME:BC13, giving perhaps (DME:H)+ and BC14. The spectrum showed that in the liquid phase no interaction occurs. There remains the possibility that if BC13 were added to DME:HC1l a new species might be formed. However, Laubengayer and Smith(51) prepared (Me3N)2SnC14 and admitted BF3 to the system. The resulting interaction resulted in displacement of the weaker acid, $nC14, by BF3 and led to the formation of Me3N:BF3. Drawing an analogy from this instance, displacement of HC1 from DME:HC1 by addition of BC13 seems likely, but a final answer can only come from further experimental work.

SUMMARY Earlier work in this laboratory by Vidale and Taylor(l) involved a study of the interaction of HC1 and dimethyl ether (DME). This system was of particular interest not only as a case of very strong hydrogen bonding but also because of the number of discrete species which are formed as a result of acid-base interaction. The technique employed was that of vibrational spectroscopy and of the two methods available for the analysis of vibrational spectra, the Raman technique was chosen over infrared because it permitted use of all-glass apparatus, allowed observation of a broader spectral region and was more conveniently adaptable to a wide range of temperatures. One portion of this previous investigation which could not be resolved involved the characterization.of all the species present in the DME - HC1 system. The existence of a 1:1 molecular complex was confirmed but the remaining complexes, at least two in number, which appeared to be ionic in nature and resulting from proton transfer, could not be characterized, partly because of inherent spectroscopic limitations of the HC1 molecule. In the present work a more complex Lewis acid was sought which in mixtures with DME, might give rise to another series of complexes in which both cationic and anionic fragments could be characterized. Such a system might well provide information on the stoichiometry of the DME - HCl complexes in addition to being valuable as another example of acidbase interaction. -69

-70The interaction of many strong Lewis acids with DME to give stable 1:1 addition complexes is quite general and complexes containing two moles of ether per mole of acid are not uncommon. Brown's recently proposed single(5.) bridge structure for compounds of the boron halides suggested. additional complexes containing more than one mole of acid per mole of base. Thus, systems of the boron halides offered reasonable promise of giving rise, in mixtures with DME, to another series of complexes whose study would be of considerable interest. Raman spectra have been obtained. of mixtures of the boron halides with DME in which the concentrations of acid/base varied from 5/1 to 1/ at temperatures just above their freezing points. All of these spectra showed the spectrum-of a 1:1 addition complex plus new bands which, in the DME-BC1 system are attributed to decomposition of the 1:1 complex rather than to formation of a higher order complex. In the DME - BF3 system, however, the results indicate a second complex which is found only in acid-rich solutions. In both systems addition of excess acid or base to solutions of the 1:1 complex caused a shifting of several frequencies which is attributed to a solvent effect. A detailed study of the vibrational spectra of the 1:1 comp exes is presented in which the spectra are considered as being made up of three sets of bands: A. those derived from dimethyl ether B. those derived from BX3 C. new bands arising as a consequence of the formation of the B-O bond.

-71Frequencies associated with the methyl group iof DME are not noticeably affected by complexation but the skeletal frequencies appear at lower values in the complexes. In this respect BF3 has a greater effect than BC13. Bands derived from the parent acids are shifted in a manner similar to that observed in the formation of BF4 from BF3(42) Of interest among the frequencies arising from complex formation is the B-O stretching frequencyo Its appearance at 664 cm. l in both DME:BF3 and DME:BC13 is interpreted as indicating greater acid strength for BC13 Spectra of mixtures of strong Lewis acids and weaker bases and of a weaker acid, SO0, and DME are also discussed. In only one system, methyl bromide - aluminum bromide, was a complex species, MeBr:AlBr3, observed in the liquid phase. The vibrational frequencies of this species, considered by some to be an unstable intermediate in FriedelCrafts catalysis, are assigned to various motions of the complex. The question of trends in the acid strengths of the boron halides is perhaps typical of the many unanswered questions which one encounters in the field of molecular addition compounds. Many of these questions can be traqed to a lack of fundamental knowledge of the properties of the compounds involved and to the limited number of experimental techniques which have been applied. The two are related to the nature of the compounds themselves but as a result most of the information currently available is in the form of thermodynamic data obtained from thermal disc sociation studies or calorimetry. The development of spectroscopic techniques has permitted study of a whole new range of systems which had not been tractable previously.

-72It would be of interest to extend the current investigation of acid-base interaction by studying several related systems whose spectra could be analyzed more easily. These proposed systems should have two properties which would facilitate this analysis. Their components should contain a minimum number of atoms and their 1:1 addition compounds should possess a high degree of molecular symmetry, preferably at least a threefold rotation axis in order to reduce the number of observed bands through degeneracy. The advantages of such systems would actually be two-foldo The reduction of the number of observed spectral bands would facilitate the assigning of these bands to particular vibrational motions while the highly symmetrical structure would permit a normal coordinate treatment of the complex and thereby give a measure of the binding forces of the molecule in terms of force constants. An example of such a series can be found in the 1:1 addition complexes of ammonia and the Group III halides. A spectroscopic investigation of these species would not only give information about the binding forces in a series of related compounds but would also serve as a check on the assignments which were made for the dimethyl etherates of BF5 and BC13.

BIBLIOGRAPHY 1. W. F. Luder and. S. Zuffanti, "The Electronic Theory of Acids and Bases," John Wiley & Son4,~ Inc., New York, 1946. 2. G. L. Vidale and R. C. Taylor, J. Am. Chem. Soc. _Q, 294 (1956). 3. W. J. Jacober and C. A. Kraus, J. Am. Chem. Soc. 1, 2409 (1949). 4. R. E. Van Dyke and C. A. Kraus, J. Am. Chem. Soc. _1, 2694 (1949). 5. H. C. Brown, J. Am. Chem. Soc. 7I, 2020 (1957). 6. H. C. Brown and W, J. Wallace, Jo Am. Chem. Soc. o75, 6279 (1950). 7. M. Ebelmen and Bouquet, Ann. chim. phys. (3) 1, 54, (1846). 8. L. Gatterman, Ber, 22, 186 (1889). 9. H. Ramser and E. Wiberg, Ber. 66B, 1136 (1930). 10. E. Wiberg and W. Sutterlin, Z. anorg. allgem. Chem. 202, 1, 22, 36 (1931). 11. V. Gasselin, Bul. soc. chim., (3) 7, 17, (1892). 12. V. Gasselin, Ann. chim. physE., (7) 3, 1,14 (1894) 13. A. Germann and M. Cleaveland, Science 53, 582 (I921). 14. H. C. Brown and R. M. Adams, J. Am. Chem. Soc. 2, 2557 (1950). 15. F. V. Dunderman and S. H. Bauer, J. Phys. Chem. 50o 32 (1946). 16. S. H. Bauer, G. R. Finlay, and A. W. Laubengayer, J. Am. Chem, Soc. y 339 (1945). 17. V. V. Korshak and N. N. Lebedev, Zhur. Obshchei Khim. 20, 266 (1950). 18. W. Bues, Dip.-Arbeit. Universitat Gottingen (1947). 19. J. Goubeau and K. Lucke, Ann., 49 (1952). 20. H. C. Brown and W. J. Wallace, J. Am. Chem. Soc. 75, 6279 (1953). 21. E. Briner and E. Cardoso, Compt. rend. 144, 911 (1907). 22. G. Baume, cbid. 148, 1322 (1911). -73

-7423. Go Baume, J. chim. phys. 12, 216 (1914). 24. M. V. Wolkenstein, Acta Physico Chem. URSS, 7, 313 (1937). 25. D. R. Martin and Wm. B. Hicks, J. Phys. Chem. 50, 422 (1946). 26. L. Pauling and E. Bo Wilson, Jr,, "Introduction to. Quantum Mechanics," McGraw-Hill Book Company, Inc., New York, 1935. 27. G. Herzberg, "Molecular Spectra and Molecular Structure," D. Van Nostrand Company, Inc,;, New York, 1945. 28. E. B. Wilson, Jr., J. C. Decius, and P. C. Cross, "Molecular Vibrations, " McGraw-Hill Book Company Inc., New York, 1955. 29. M. Born and J. R. Oppenheimer, Ann. d. Phys. 84, 457 (1927). 30. E. J. Rosenbaum, J. Chem. Phys. 8, 643 (1940). 31. H. Gerding and E. Smit, Zo physik. Chem. B50, 171 (1941). 32. H. Gerding and E. Smit, ibid. B51, 217, (1942). 33. L. M. Foster and C. A. Kraus, Proc. Nat. Acad. Sci. U.S. 39, 236 (1953). 34. H. Gerding and H. Houtegraf, Rec. tray. chim. 7Y, 15 (1955). 35. H. Gerding, Rec. tray. chim. _75, 589 (1956). 36. J. R. Nielsen, C* Y. Liang, R. M. Smith, and D. C. Smith, J. Chem. Phys. 21, 383 (1953). 37. J. R. Nielsen, C. Y. Liang, and D. C. Smith, J. Chem. Physo 21, 1060 (1953). 38. R. Thiemer and J. R. Nielsen, J. Chem. Phy, 27, 887 (1957). 39. To R. Riethof, Dissertation; Abstracts 14, 1564 (1954). 40. R. E. Kagarise, J..Chem. Phys.,_ 519 (1957). 41. L, A. Woodward andF A. A. Nord., J. Chem. Soc. 1955, 2655. 42. J. Goubeau and W. Bues, Z. anorg. u. allgem. Chem. 268, 221 (1952). 43. J. Goubeau and K. E. Lucke, Ann. 5, 37 (1952). 44. J. R. Nielsen, C. Y. Liang, and L. W. Daasch, J_. Op. Soc. Amer. 43, 1071 (1953).

-7545. E. P. Shull, J. Chem. Phys. 2, 399 (1957). 46. R. E. Van Dyke and H. E. Crawford, J. Amer. Chem. Soc. 72, 2829 (1950). 47. P. M. Hamilton, R. McBeth, W. Bekebrude, and H. H. Sisler, J. Amer. Chem. Soc. y5, 2881 (1953). 48. A. W. Laubengayer and W. C. Smith, J. Amer. Chem. Soc. 76, 5985 (1954). 490 N. NO Greenwood, (private communication). 50. H. C. Brown, J. Amer.Chem. Soc. 8, 2173 (1956) o. 51. A. W. Laubengayer and W. C. Smithp J. Amer. Chem. Soco 76, 5985 (1954).

APPENDIX The method which was employed to study the effect of geometry upon the vibrational frequencies of the XYZ3 model system is that outlined in Wilson, Cross, and Decius Only a small portion which contains a slight modification of Wilson's method will be discussed below. It can be shown that the potential and kinetic energies of vibration may be written in matrix form. 2V = x f x (A-l) S -X g9 —r (M 2) where x is a column vector of internal displacement coordinates, a is the transposed (row) vector, f is a matrix of force constants, and g 1 is a kinetic energy matrix whose elements are expressed in terms of masses, bond lengths and angles. The form of the secular determinant obtained from the energy expressions given in (A-l) and (A-2) is: If - glxl 0 (A-3) where k = 4-2v2o By multiplying (A-3) through by the determinant g, the inverse kinetic energy matrix, and rearranging, one can obtain the secular determinant in the formIfg - IXj = ~ (A-4) The f-matrix is symmetrical about its diagonal, and has for its elements two types of valence force constants; the "principal" valence force constants lie along the diagonal while the "interaction'" terms fill out the rest of the matrix. Often the magnitudes of these interaction terms are sufficiently small that they may be equated to zero; -76

-77in the XYZ3:ase only two such terms were retained. The f-matrix which was employed in the XYZ3 system is shown below with half the interaction terms omitted because of symmetry. kr 0 0 0 ~0.0 0 0 0 0 kr 0 0 0 0.0.0.0 0 kr 0 0 00 0 0 XU 0 O O O O (A-5) k 0 0 0 ka o o o o k: 0 0.0 o o 0 k O ki The method by which the g-matrix elements are determined is given by Wilson. Since the elements in terms of atom masses and internal coordinates are fairly complex expressions, the matrix is represented symbollically -below, the rows and columns being labeled with the appropriate internal displacements coordinates so that the notation may be understood, Terms having identical subscripts are the same because of the symmetry of the model. In addition, symmetry of the matrix has led to omission of half of the interaction terms, as in the case of the f-matrix.

-781 r2 r3 D cl c02 cj3 1 P2 B3 rl gll g12 g12 g14 g16 g15 g15 g18'g19 g19 r2 gll gl12 g14 g15 g16 g15 g19 g18 g19 r3 gll'g14 g15 g15 g16 gl9 g19 g18 D g44 g19 g19 g9ig g48 g48 g48.1 | g55 g56 g56 g59 g58 g58 (A-6)'~2 -tt ~,g55 g56 g58 g59 g58 Ol3 g55 g58 g58 g59 P1 g88 g89 g89 PB2 g88 g89.3 g88 Since the expressions for the g-matrix elements have been tabulated in Appendix VI of Wilson, Cross, and Decius, they will not be repeated here. However, a translation of the notation employed above into that used by Wilson is given: 2 2 2 1 gll= grr g18 =grC g56= g= p(l) 1 1(1 g58 = 2(l 912 = grr g19 rCPl) 58 1 2 g4 = gr g44 g g gq(l) g15 gr g48 = gr g88 = g-pcp 16 g11) 3...6 9rl 55 Tqxp g89 gp_-(l) Wilson has shown that equation (A-4) can be further simplified by the use of symmetry coordinates rather than internal coordinates. For

-79pyramidal molecules such as XYZ3 the change from the internal coordinates ri, D, Cij, and i to symmetry coordinates, Si, may be expressed as Si = Uxi. (A-7) where U, the transformation matrix has the form AL35 11/3 1/13 0 0 o 0 0 O O 0 0 0 0 0.0 0 U = 0 0 o0 l/ f l /4 lfc b/fc b/c b/ c (A-8) O 1/N2 -1/\2 0 0 0 0 0 0.0 0 0 0 1/~T2 -11/Nf2 0 0 o O 0 0 0 0 0 l/2 -1/Nf2 with b = 3 cos P/cos a/2 and c = 3(1 + b2). In terms.of the symmetry coordinates the secular equation may be written: IFG - xII = O (A-9) where F = U f U and G = U g U (A-10) The F and G matrices obtained according to (A-10) are not only symmetrical but' are also factored into two 3 x 3 matrices as shown below for the F matrix. F11 F12 F13 F12 F22 F23 0 F13 F23 F33 44 45 46 0 F45 F55 F56 46 56 66

-80in which the upper block corresponds to totally symmetric (A1) vibrations and the lower section to antisymmetric (E) vibrations. The product, the FG matrix, is similarly factored, but is not symmetrical. For the purposes of these calculations only the totally symmetric frequencies were considered. The frequency values together with the values of the F and G-matrix elements are given below: vl = 1020 cm,, v2 = 664 cm.-1 V3 = 554 cm7.l Fl = 7.34 F22 = 2-29 F12= F21 = 05 F23 = F32 = -1.0 13 31 33 F1G = F1 0 F3 = 1.75 G1= gll+ 2g12 Gl2 G21= 3 g14 G13 = G31 = (1 /1 + b2)(2gl5 + gl6 + bgl8+ 2bgl9) G22 = 44 G23= G32= (3/sc)(gl9 + bg48) 33= (1/1+ b2)(g55 + 2956 + 4bg58 + 2bg59 + b2g88 + 2b2g89) Since it was assumed that the binding forces were to remain fixed in order that the effect of various masses and angles could be determined, the same F-elements were used throughout. The numerical values of the G-elements are:

-81- = 106~ Mz Gll G12 G13 G22 G23 G33 19 0.09032 -0.06083 -0.07302 0.15330 0o12762 0.41295 40 0.06270 O. I 1t 0.38231 60 0.05437. o.37307 80 0.05020 tt " t 0.36845 l= 109~ 28' 19 0.08290 -0.05241 -o0.0o8584 0.15330 0.14600 0.29646 40 0.05528 I" t It It 0.26868 60 0.04695 f f t 0.26031 80 0.04278 " t f 0.25610 a= 1140 19 O. 06956 -0.03921 -0.05797 0.15330 0.13802 0.23929 40 0.04194 0.21784 60 0.03361 O 0.21136 80 0.02944 0 " " " O. 20813. = 119~ 19 0 o5538 -0.01583 -0.01264 0.15330 0.10069 0.06842 40 Q0.02776 " " 0.06257 60 0.0o1943t it tt if 0o06081 80 0.01526 It ft I t It 0.05992 The frequencies calculated from the roots of the secular equation for the X-Y stretching motion (v2) have already been given in Figure 12 but are

included with the Y-Z stretching frequency (vl) and the ZYZ,deformation frequencay (V3) in the tables below. c.= 10o6.= 1090 28' Mz v1 V2 V MX 1 V V 19 o1048 933 56- 19 10~1 655 556 40 963 822 500 40 916 554 479 60 944 787 460 60 890 523 399 80 936 770 434 80 878 517 349 = 1140 =1190 Mz 1V2 3 Vz V 2 V 3 19 910 665 522 19 816 656 65 40 796 550 453 40 652 570 imag. 60 775 510 396 6o 624 505 imag. 80 767 496 551 80 620 456 imag.