SPECTROSCOPIC STUDIES OF ALKALI METAL SALT SOLVATION IN DIALKYLSULFOXIDES Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY BRIAN WILLIAM MAXEY 1968 -.—‘..._‘..ll LIBRAR Y Umwcfiiit] This is to certify that the thesis entitled Spectroscopic Studies of Alkali Metal Salt Solvation in Dialkylsulfoxides presented by B. William Maxey has been accepted towards fulfillment of the requirements for . Bhu'Dr degree in__Chemiatry .éfflggtg v/y 490/ Major professor / Date July 12, 1968 0-169 1. airline‘s av. ' m l HDAE & SUNS’ L WPEE'BMJEE- $3M ABSTRACT SPECTROSCOPIC STUDIES OF ALKALI METAL SALT SOLVATION IN DIALKYLSULFOXIDES by B. William Maxey The techniques of far-infrared spectroscopy and nuclear magnetic resonance spectrometry have been combined to give a new and useful approach to the study of ionic solvation. Low frequency infrared bands have been observed in solutions of alkali metal salts in dimethyl-, dipropyl- and dibutyl-- sulfoxides which are attributable to the vibrations of cations in solvent cages. Within a solvent cage, solvent molecules are oriented with the oxygen atom of the sulfoxides in closest proximity to the cation. Isotopic substitution experiments demon- strated conclusively that both the cation and the solvent molecule were involved in the vibrating species. Studies of the 8-0 fundamental stretching mode indicated that the oxygen of the sulfoxide was oriented closest to the cation. The frequencies of the far-infrared bands were highly dependent on the mass of the cation but showed no dependence on the mass of the anion. The intensities were directly ‘proportional to the concentration of the salt employed. Bands due to the cations Li+, NHu+, Na+, K+, Rb+ and Cs+ in dimethylsulfoxide occurred at 430 cm'l, 214 cm'l. l l l l 200 cm' , 154 cm“ . 125 cm' and 110 cm' , respectively. Small shifts from these values were observed for these bands in the other dialkylsulfoxides. B. William Maxey The results of nuclear magnetic resonance mole-ratio studies indicated a 1:2 stoichiometry, of the type M(DMSO)2+, for both lithium ion and ammonium ion in dimethylsulfoxide solutions. Similar measurements on sodium ions indicated either a salvation sphere of ill-- defined stoichiometry or the simultaneous existence of more than one well-defined structure. Assuming a linear-triatomic model for the 1:2 solvates, it was possible to calcuLate force constants for the DMSO-cation bonds. The force )4, constants so calculated are 3.6 x 10 dynes/cm and 2.2 x 10“ dynes/cm, respectively, for the lithium ion and ammonium ion solvates. In addition, solid solvates of dimethylsulfoxide with various alkali metal salts were prepared and char- acterized by elemental analyses, melting points and infrared spectra. Ratios of DMSO to cation for Li+, NHu+, Na+ and + K solvates ranged from 0.7:1 to 4:1. Solvates of rubidium and cesium salts could not be prepared. SPECTROSCOPIC STUDIES OF ALKALI METAL SALT SOLVATION IN DIALKYLSULFOXIDES By Brian William Maxey A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1968 /‘ - To - a/ j (1' J.” "J ““ / /§/ .957 ACKNOWLEDGEMENTS The author wishes to publicly express his sincere gratitude to Professor Alexander I. Popov for his interest, enthusiasm and conscientious guidance which was sustained throughout this investigation. The author also wishes to express appreciation to Dr. John W. Pankey of the Dow Chemical Company who encouraged initiation of this program: to Professor Paul G. Sears of the University of Kentucky for valuable assistance during the early stages of the investigation; to K. L. Treuil and Dr. B. E. Miller for the Raman spectra and to Professor George E. Leroi. William J. McKinney and John L. Wuepper for numerous enlightening discussions. Above all. heartfelt thanks are extended to my wife; without her support the graduate program would not have been initiated, and without her patience, love and under- standing it could not have been completed. ************** ii TABLE OF CONTENTS I. INTRODUCTIONCOC...00.00.000.00...0000...... II. HISTORICAL SECTIONOOOOOCOOOOOOOOOOOOOOOOOOO SOlvationoococo.oocooooooooooooooooooooo SUlfOXideSoocoo...00000000000000.0000... Solid Solvates of Alkali Metal Salts With DimethylsUlfOXIdeooooooooooooooooo Far'Infrared SpBCtTOSCOpYQoooooooooooooo III. EXPERIMENTAL SECTION..................o.... Reagents...00000000000000.0000.ooooooooo sulf011desoooooooooooooooooooocooooo DimethYlsulfoxide (DMSO)oooooooooooo Dipropy18u1f0x1de (DPSO)oooooooooooo Dibuty1su1foxide (DBSO)ooooooooooooo l‘Pentan01oooooooooooooooooooooooooo l-Methyl'2-Pyrr011doneoooooooooooooo Benzene..o.o........................ Alkali metal SBltSoooooooooooooooooo Analytical NethOdSoooooooooooooooooooooo Halogen analyseSoo00.000000000000000 Water determination................. Carbon. hydrogen, sulfur analyses... Melting point determinations........ Preparation Of SOlutionSoooooooooooo Preparation.of Solid DMSO-Alkfili Metal Salt SOlvateSooooooooooo00000000000000. Lithium SOlvateSoooooooooooooooooooo SOdium 801vat380000000000000000.0000 P0t8381um SOlvateSooooo0000000000000 Bubidium and cesium solvates........ Ammonium SOlvateSooooooooooooooooooo InStrumental measurementSooooooooooooooo InStrumentationoo0.0000000000000000. Experimental techniques in the Far-- infrared region.................... Far-infrared spectra of Alkali metal salts in dialkylsulfoxides......... Nuclear magnetic resonance spectra: m013’ratio studieSOOOOOOOOOOOOOOOO. IV. RESULTS AND DISCUSSION..................... Infrared Spectra in DMSOooooooooocoooooo Far-Infrared Spectrum Of d6'DM8000000000 iii Page 1 8 8 15 16 17 TABLE OF CONTENTS - continued VI. VII. VIII. Infrared Spectra of DMSO-Alkali Metal SOlUtlonSoococo.ooooooooooooooooooooooo Infrared Spectra of Dipropyl- and Di- butylsulfoxide SOlutionSooooooooooooooo Infrared Spectra of Solid DMSO Solvates. Overtone Spectra Of DMSOoooooooooooooooo Far-Infrared Spectra of Alkali Metal Salts in Mixtures of DMSO and l-Methyl- 2-Pyrrolldone.......................... Nuclear Magnetic Resonance Spectra of Dimethyl- and Dipropylsulfoxide Solu- tionSooooooooooooo00.000000000000000... Calculation of Force Constants and Frequen01es fOr M(DMSO)2+oooooooooooooo Raman Spectra of LiClOu-DMSO Solutions.. Far-Infrared Spectra of Miscellaneous SOlutionSooooooooococoon-0000000000.... CONCLUSIONS...00.000.000.00.00.00.000.000... PROPOSALS FOR FUTURE INVESTIGATIONS......... Infrared StUdieSoooo0.000000000000000... Raman StUdieSoo0000000000000...00000000. Nuclear Magnetic Resonance Studies...... Miscellaneousoo......................... APPENDICES0.0....O...OOOOOOOOOOOOOOOOOOOOOCO I - Instrument Usage and Maintainance... II - Cells for the Far-Infrared Region... III - Force Constant and Frequency Calcur lationooooooooooooooooocoooooooooooo BIBLIOGRAPHY’OOoooocoo00000000000000.0000... iv Page 32 1+7 51 56 58 TABLE I. II. III. IV. V. VI. VII. VIII. X. XI. XII. XIII. XIV. XV. LIST OF TABLES Hydration Numbers for Alkali Metal Salts.. Hydration Numbers of Alkali Metal Salts... Absorption Bands of Sodium Salts in Di- methy18u1f011deooooooooooooooooooooooooooo Absorption Bands of Alkali Metal Salts in Dimethylsulfoxldeooooooooooocooooooooooooo Comparison of Calculated and Observed Fre- quencies of Alkali Metal Salts in DMSO.... Absorbances of Lithium Bromide Solutions in DMSO0.000000000000000000000000000000000 Effect of Salt Concentration on Spectra in DMSO...O0.00.00.00.00.00.000.000.000... Effect of Anion Concentration on Cation-- DMSO Bands...ooooooooo00000000000000.0000. Absorption Bands of Alkali Metal Salts in DMSO"Benzene Mixtur6800000000000000000000 Absorption Bands of Alkali Metal Salts in Dimethylsulfoxide, Dipropylsulfoxide and leutylsulfoxlde.......................... Calculated and Observed Band Frequencies.. Characterization of Solid Solvates........ Stoichiometries and Melting Points of Solid DMSO-Alkali Metal Salt Solvates..... Infrared Absorption Spectra of DMSO SOlvateSoooooooooooooooooooooooooooooooooo Miscellaneous Spectra..................... Page 10 14 33 35 36 39 41 43 “5 #8 so 52 5# 55 68 LIST OF FIGURES FIGURE 1. 2. 3. 1+, Frequencies and intensities of bands in alkali metal salt - DMSO solutions......... Chemical shift of DMSO protons gs mole-- ratio DMsonX0.0.00.00.00.00.00000000000000 Chemical shift of DMSO protons gs mole-- ratio DMSO:MXoooooooooooooooooooooooooooooo Chemical shift of DMSO protonsng mole-- ratio DPSO:MXOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO vi Page 37 62 6b 66 LIST OF APPENDICES APPENDIX Page I. Instrument Usage and Maintainance.......... 7# II. Cells for the Far-Infrared Region.......... 84 III. Force Constant and Frequency Calculation... 87 vii I. INTRODUCTION "It is regrettable that while electrolyte solutions owe their very existence to ion-solvent interactions, we have so far been able to find out little of a quantitative nature about these interactions." Robinson and Stokes (l) have admirably stated the challenge which initiated the research described in this thesis. This research concerns ion-solvent interactions which are generally described by the term "solvation". Before proceeding it is necessary to define as exactly as possible the meaning of this term. The first definition of the term "solvation" was ad- vanced by Fajans (2) in 1920. His concept was that the energy of interaction between a solute and a solvent mole— cule was entirely coulombic in nature and did not involve electron transfer leading to the formation of covalent bonds. Without the existence ofcovalent bonds, the solvent molecules in the solvation shell were considered to be exchanging con- tinuously with those in the bulk of the solution. This con- cept of solvation included three types of interactions: ion-dipole, dipole-dipole, and hydrogen-bonding. It was an adequate concept for the cations of group IA and IIA of the periodic table and also for many common anions: halogens, nitrates, thiocyanate, perchlorate, etc. Parker (3) prefers a more general definition which in- cludes not only the three interactions mentioned above but also coordinate covalent bonds in which the solvent molecule 1 2 shares a pair of electrons with the solvated ion or molecule. Some investigators refer to the latter interaction as com- plexation and reserve the use of the term solvation for pre- dominantly electrostatic interactions. Any definition of electrostatic solvation must include the fact that the electrostatic field at an ion does not stop at the first solvent shell but must extend to an infi- nite distance in all directions. Bockris (h) suggests that the term "primary solvation" be used in reference to those solvent molecules which are attached so firmly that the ion and its solvent shell move as a single entity during trans- lational motion, while the term "secondary solvation" should refer to all other interactions which are influenced by the electrostatic field emanating from the ion in question. The term "total solvation" would then refer to the sum of both types. A second problem in attempting to unambiguously define solvation arises when the dielectric constant of the sol- vent is sufficiently low to allow ion-ion interactions. The difficulty lies in differentiating among solvated ions, solvated ion-pairs and higher aggregates. If the solvated species is an ion-pair, a problem arises in discerning the nature of the ion-pair itself. There may be several differ- ent types of ion-pairs and, for the current purpose, three distinct species will be considered. These are the contact ion-pair, in which an anion and cation are in direct contact, a solvent-shared ion-pair, in which one solvent molecule separates the two ions, and a solvent-separated ion-pair in 3 which each ion is solvated individually and yet retains an attraction for the other (5). It is clear that a true understanding of the salvation phenomenon requires some knowledge concerning the nature of the entity which is being solvated. Although intensively investigated for many years, such knowledge has proven to be difficult to obtain. The importance of ion-solvent interactions in deter- mining the behavior of ionic solutions has long been recog- nized and their elucidation has been the subject of hundreds of publications over a period of several decades. The wide- spread influence of such interactions is indicated by the large variety of techniques which have been applied to the study of solvation phenomena. Classical techniques include studies of the following properties: refractivity, freezing and boiling points, vapor pressure, surface tension, density, specific heat, solution compressibility, ionic mobility and solubility. Spectroscopic techniques include visible, infrared, Raman,a1 nuclear magnetic resonance and mass spectroscopies. In order to clarify the reasons why a thorough under- standing of ionvsolvent interactions is necessary for in- telligent use of ionic solutions, one rather simple system will be considered - water and sodium chloride. Liquid water has been one of the most highly studied materials known to science and, in a pure form, its behavior is fairly well understood. Its thermodynamic and colligative properties are known which makes it possible to predict with some certainty how this material will behave under a Lt given set of~conditions. Likewise, crystalline sodium chloride has been extensively studied and the same state- ments may be applied to it as were given for water. However, our understanding of the nature of the two chemicals decreases dramatically when they are combined in solution. At ambient temperatures, sodium chloride exists in a highly ordered structure and water in a less ordered array. When mixed, the existing structures are in part broken down and a new set is added. Sodium chloride may not dissociate completely but may form some ion-pairs which can exist in solvated form. The dissociated ions may exist as free ions or may be solvated by one or more molecules of water with these molecules being attached with varying degrees of firmness. In addition, aggregates of cations, anions and solvent molecules may be formed. It is clear that these particles will possess different sizes, masses and electrical characteristics and that each Species will behave differently, both physically and chemically, under a given set of conditions. Considering the complexity of the solution resulting from mixing two simple chemicals, it is not surprising that bulk solution properties do not provide a sound basis for understanding and predicting solution behavior. In order to do so, the properties of the individual species which make up the solution must be understood. The necessity for obtaining detailed information about solution components on a molecular scale should be readily apparent. As will be seen, many of the methods which have been 5 used to study ion-solvent interactions have failed to pro- vide all the data necessary for a thorough understanding of the solvation phenomenon. During preliminary far-infrared spectral investigations of solutions of alkali metal salts, bands were observed which could be attributed to ion-solvent interactions. Since information concerning stereochemistries, bond energies and thermodynamic constants of the vibrating species may be obtainable from the vibrational spectra, these investigations were continued and they form the substance of this thesis. The definition of that portion of the electromagnetic Spectrum known as the far-infrared is an arbitrary one and for the purposes of this presentation will be defined as the range from 666-1“ cm-l (l6-7lh microns). As well as being arbitrary, any definition of this re- gion is also an artificial one in that it is separated from the mid-infrared by reasons of experimental technique rather than as a result of any theoretical distinctions. Most of the applications of far-infrared Spectroscopy are not basi- cally different from those in the mid-infrared. Included in the common chemical applications are studies of the pure rotational spectra of light molecules, intramole- cular vibrations in which the masses of the vibrating Species are large or the force constants binding them are very small, and intermolecular vibrations such as found in hydrogen-bonded systems. In order to detect an ion-dipole vibration by infrared spectroscopic techniques, two requirements must be met. First, 6 the ion-dipole unit must exist for sufficient time to allow at least one complete vibration to take place. Second, a change in the dipole moment must be induced during the course of the vibration. It is clear that the second requirement would be met for all cases allowed by molecular symmetry, whether the first would also be met required experimental verification. In order that the data which were to be obtained would be interpretable in as unambiguous a fashion as possible, it was necessary to choose a reasonably simple system to study. The choice was made to investigate ion-solvent interactions which were predominantly electrostatic. Solutions of alkali metal salts in dimethylsulfoxide, and related solvents, proved to be satisfactory. The inter- actions of alkali metal salts with polar solvents should be predominantly electrostatic. Dimethylsulfoxide is trans- parent across much of the far-infrared region: it is capable of dissolving alkali metal salts to a high concentration: it is purifiable and stable. Ammonium salts were included in the study and, hereafter, the ammonium ion will be considered an alkali metal ion. The purpose of this presentation is to describe the initial development of an experimental technique employing far-infrared spectroscopy and nuclear magnetic resonance Spectrometry which allowed, for the first time, not only qualitative detection of a solvated ion, but also quantita- tive measurements of some of its properties. Although the major effort was concentrated on dimethyl- 7 sulfoxide, dipropyl- and dibutylsulfoxides were also in- vestigated in detail. In addition, preliminary investi- gations were conducted on 2-propanol, l-pentanol, aceto- nitrile and tetrahydrofuran. II. HISTORICAL SECTION Salvation. Bockris, in his excellent review of the classical methods for determining salvation number (4), has grouped the methods into six categories depending on the measured property, namely ionic and solvent transport, activity of solvent, activity of nan-electrolyte solute, ionic entropy, solution density and miscellaneous properties. Discrepan- cies among the values determined by the various methods are, unfortunately, quite large and demonstrate that each method measures some indeterminate combination of primary and secondary salvation. Bockris stated that, in general, activity determina- tians measure total salvation while entropy, density and mobility studies measure primary salvation. Of the numer- ous methods included in his final category only two were thought to be of value. The first was measurement of solution compressibility and the second involved subjection of a solu- tion to supersonic vibrations and measurement of the emf pro- duced by the induced vibration of the ions in the solution. The potential so generated was shown to be a function of effective ionic mass (6). Both of these methods, which measure primary salvation, have been critically reviewed by Robinson and Stokes (7). From the many techniques which have been described for determination of primary salvation numbers, it is possible to choose several which give reasonably consistent results. Hydration numbers for several alkali metal salts as measured 8 9 by three distinctly different_procedures are presented in Table I. By comparison, most of the other techniques give salvation numbers which do not inspire a high degree of con- fidence, e.g. hydration numbers for sodium ion have been reported varying from 10 all the way to 1100. A difficulty common to most of these methods lies in dividing the salvation of the salt between its ions. To do so requires an assumption of a hydration number for a single ion which is then used as the starting point for calculating all others. Another serious drawback is that the classical techniques are incapable of distinguishing among two or more salvate Species which may coexist in solution as they determine only average salvation numbers. Since the appearance of Bockris' review (A), two monographs have appeared which discuss in detail the gen- eral subject of ionic solutions (8,9) and there has been one review of anion salvation in nonaqueous solvents (3). The uses of visible, infrared and nuclear magnetic resonance spectroscopies as well as polarography and conductance tech- niques for salvation studies have been briefly reviewed by Price (11). In salvation studies by spectral techniques, attempts have been made to correlate the extent of salvation with the spectra of solvents and complex ions in all regions from the visible through the far-infrared. Charge-transfer bands of pyridinium iodide complexes have been studied in the visible region and were found to be highly dependent on the nature of the solvent used due to the importance of hydrogen-bonding in salvation of the polar- Table I. Hydration Numbers for Alkali Metal Salts. Ion Method Mobility(a) Entropy(b) Compressibility(c) Li+ 6-7 5 3 Na+ 2-u u u K+ - 2 3 F‘ - 5 5 c1" - 2 3 1' - 0.5 2 a. H. Ulich, Trans. Faraday Soc., g3, 388(1927). b. H. Ulich, Z. Electrochem., 6, 497(1930). 0. A. L. T. Moesveld and H. T. Hardon, g, Phys. Chem., l55.. 238(1931): C.A.. as. 2636(1932). lO ll izable iodide ion (12). Hydration numbers have been determined also by studying changes in the near-infrared Spectrum of a solvent caused by addition of electrolyte (13). Results obtained in water- methanal Solutions were extrapolated to give hydration num- bers of 5.7 and 7.1 for Cr(Cl)3 and Cr(N03)3 reSpectively. Most spectral investigations of salvation have been concerned with infrared studies of the solvent molecule or of complex anions or cations. Many of these studies are of hydrogen bonding between anions and suitable hydroxylic solvents (1h,15,l6,17). Among the alkali metal halides, the strengths of the anion-solvent bands almost invariably fall in the order I" < Br" < 01" < F". This order has been sub- stantiated by a variety of experiments and leads to the reasonable generalization that anions are solvated by hydroxylic solvents with the smaller, higher charge density, anions solvated most strongly. Results of studies of the N-H stretching frequency of aqueous ammonium salts were interpreted on the basis of ion- pair formation at concentrations greater than 2 M with hydro- gen bonding predominating at lower concentrations (18). Changes produced in the near-infrared Spectrum of water by the addition of alkali metal salts were interpreted an the basis of "structure breaking" by polarizable ions such as ClOn-, SCNT, N03‘, 1' and Br' (19). Infrared studies of lithium perchlorate in acetone have been reported (20). Data obtained indicate a 1:2 stoichiometry for the solvated ion, Li(acetone)2+. 12 In recent years, far-infrared Spectroscopy has been used in studies of salvation. Edgell and coworkers have been concerned primarily with alkali metal salts of complex car- bonyl anions and their salvation by solvents of low dielectric constant (21,22,23). They have used far-infrared bands, attributed to the vibration of a solvated ion-pair, and the carbonyl fundamental stretching mode to distinguish between solvated contact ion-pairs and solvent-separated ion-pairs. Raman spectroscopy has been applied extensively to studies of aqueous salt systems in attempts to elucidate the structure of the solutions (2h,25,26). In support of studies in other hydroxylic solvents by other techniques, the conclusions almost invariably reached are that anionic salvation is the predominant salvation reaction in hydrox— ylic solvents due to the ease of hydrogen bond formation. In an interesting study of ionic association of metal nitrates, sulfates and perchlorates it was shown that sul- fates af alkali metals farm solvent-separated ion-pairs (27). As yet Raman Spectroscopy has not been applied exten- sively to studying salvation in nonaqueous systems although a Raman study of electrolytes in methanol gave results in- dicative of Specific interactions between various oxyanions and the solvent. The anion interactions were found to be more important than the very weak cation interactions (28). Mass Spectrometry has been applied to the study of solvates in the gas phase (29). Kebarle's technique allows determination of the number of NH3 molecules associated with an ammonium ion. The method may be applicable to other 13 systems but will be limited by the types of ions which can be studied. During the past few years, the use of nuclear magnetic resonance Spectroscopy in salvation studies has been reported with increasing frequency. A recent critical review compre- hensively covers the pertinent literature up to 1966 (30). As with other techniques, most of the applications have dealt with covalent complexation of transition metal ions rather than with interactions which are primarily electro- static in nature. Although this technique has brought about a substantial advancement in salvation studies, it also suffers from some of the same problems as were found with the classical tech- niques. For example, the most commonly used method, proton magnetic resonance, is of necessity an indirect technique as it measures changes in electron density at a site frequently well removed from the atom involved in the salvation process. The accuracies of the measurements are often disappointing and agreements among the results of various methods applied to the same system are frequently poor. Typical data for aqueous solutions of Group IA salts are cited in Table II and would seem to indicate that nmr techniques do not always distinguish well between primary and secondary salvation. In defense of this approach to salvation studies it should be stated that there are many nmr techniques which provide reasonable values for salvation of transition metal ions and which may provide equally valid results when applied to predominantly electrostatic interactions. Table II. Hydration Numbers of Alkali Metal Salts. Hydration Number(3o) Method 1. Method 2. Method 3. 1 i 1 5 i 2 u 3.6 i 1 3 i 1.2 3.1 6.0a 1 i o.u 2.1 9.9 i 2 - 1.6 14.6 i 2 - 1-0 9.9 i 2 - 1.6 13.2 i 2 2 i 0.5 o 16.2 i 2 - o 21.8 i 2 - o B. P. Fabricand and S. Goldberg, J. Chem. Phys., 33, l62h(1961). S. Broersma, ibid;,jg§, 1158(1958). J. C. Hindman, igig;, fié, 1000(1962). assumed as basis for calculation. 14 15 Recently a paper appeared describing a procedure which applied the classical mole-ratio technique to nmr measurement of sodium salvation by tetrahydrofuran. diethyl ether and tri- ethylamine (10). Salvation numbers of one and four were ob- tained for sodiumtetrabutylaluminate in tetrahydrofuran solu- tion. However, such measurements could not be made for this salt in the other solvents since the interactions were too weak. The importance and value of nmr studies of salvation will grow with time but the state of the art now is in con- stant flux and was summed up well by Hinton and Amis (30), "Chemical shifts are characteristic of nmr measurements resulting from salute-solvent interactions. Theories relating shifts to details of interactions are still not complete and hence the real significances of such shifts are not fully understood". Sulfoxides. Dimethylsulfoxide (DMSO) is a solvent which has attracted extensive attention in the literature in recent years. Its uses as a solvent and reactant, as well as its potential as a useful electrolytic medium, have been recently reviewed (31,32). Indicative of its increasing importance is the fact that the first number of a new journal was devoted exclusively to this highly versatile material (33). Dimethylsulfoxide is characterized by a broad liquid range, 18.N - 189o (32i,a re- latively high dielectric constant, h6.h (3“), a high dipole moment, 3.9D (35), and extensive dissolving power for both covalent and ionic compounds. The solvent possesses a high l6 Troutan constant of 22.9 which indicates that it is an as- sociated liquid (36). Lindberg (3h,37) has shown that its liquid structure is polymeric. The polarization of the oxygen-sulfur bond of sulfoxides places a negative charge on the oxygen atom. For this rea- son, sulfoxides possess high electron donating abilities. Numerous complexes with sulfoxides are formed by coordina- tion at the oxygen atom although a few instances have been reported in which coordination occurs via the sulfur atom (35.38). Other dialkylsulfaxides have not been studied as inten- sively although a recent publication has investigated transi- tion metal complexes with dipropyl- and dibutylsulfoxides (39). From the results of a study of sulfoxide-iodine complexes, Klaeboe (#0) concluded that the complexing ability of sul- foxides is strongly influenced by the inductive effect of the substituent groups. It is expected that the dipole moments of higher dialkyl- sulfoxides would be similar to that for DMSO since the polar- ity resides in the sulfur-oxygen bond (#1). Published data for diethylsulfoxide, 3.88D (#2), and diisobutylsulfoxide, 3.93D (#3), support this assumption. The lowered abilities of the higher sulfoxides to dissolve ionic compounds should be due to an expected decrease in dielectric constant of the liquids as the alkyl chain is lengthened (#1). Solid Solvates of Alkali Metal Salts with Dimethylsulfoxide. Various workers have reported preparation of some solid dimethylsulfoxide solvates of alkali metal salts. No at- 1? tempt to synthesize the complete series of solvates with all of the alkali metals has been carried out and no one has attempted to determine the nature of the solvates other than by obtaining elemental analyses on those few which have been prepared. Kentamaa, (##), who has conducted the most thorough study reported to date, has reported preparation of some lithium, sodium and potassium solvates with salvation numbers ranging from 0.5 to 3.5. His work will be discussed in more detail later. Skerlak, gt_a;. (#5) reported prepara- tion of the adduct KNCS°DMSO by a method not given in the abstract. Physical properties were not cited but it was stated that infrared data indicated coordination through the oxygen atom. Koenig (#6) prepared NaI-ZDMSO by an in- volved reaction and gave the melting point as 112°C. Schlafer (#7) reported preparation of LiCleDMSO as a white crystalline solid by dissolving the salt in hot DMSO then cooling and collecting the crystals which formed. Far-Infrared Spectroscopy. Our first knowledge of far-infrared Spectroscopy came from the experiments of Rubens and coworkers in 1911 (#8). Lacking suitable prisms or gratings, they isolated rather impure long-wave radiation by reflection techniques. The development of high-quality gratings by Randall and coworkers, beginning in 1910, was followed by the introduc- tion of sensitive thermocouples and stable amplifiers in about 1935 (#9). These contributions allowed building high- quality, single-beam instruments which could obtain spectra out to several hundred microns. 18 Prism materials for long-wave radiation were not devel- oped until the 19#0's and 1950's. Of the various crystals now available, cesium iodide, which is tranSparent to 50 microns, is the most generally useful. Beyond 50 microns, gratings must be used although the problem involved with filtering undesired grating orders is severe. High-quality, double-beam recording Spectrophotometers specifically designed for use in the far-infrared region have been available only for the past six years. There are two types of instrumentation currently being used to obtain far-infrared Spectral data, Spectro- photometers and interferometers. Recently, far-infrared interferometry has been discussed by Hurley (50). Several outstanding reviews and bibliographies of the history, apparatus and applications of far-infrared spectrophoto- metry have been published (#9,51,52). Other valuable refer- ences which are not contained in these compilations include one which concerns use of the Perkin-Elmer 301 (53), one which gives a good treatment of infrared spectroscopy in general (5#), and another concerning applications of far- infrared to problems in inorganic chemistry (55). In addi- tion, a continuing review of a portion of the field is presented biannually in the Fundamental Reviews section of Analytical Chemistry with the first references appearing in the 195# issue (56). III. EXPERIMENTAL SECTION REAGENTS Sulfoxides. Alkyl sulfoxides used in this investigation were found to be quite hygroscopic. A gradual increase in water concen- tration, as shown by their infrared spectra, was noted even when the compounds were stored in capped bottles. Storage of the solvents over granulated barium oxide effectively corrected this problem. Dimeth sulfoxide DMSO . J. T. Baker reagent grade dimethylsulfoxide was used throughout this study. Some of the numerous methods of purification found in the literature have been recently re- viewed (57). The solvent was purified by refluxing for several hours over barium oxide followed by slow vacuum distillation through a one meter packed column. The first and last frac- tions, totalling 20% of the batch, were discarded. The middle fraction was collected and subjected to fractional crystal- lizations until a final product was obtained which melted at 18.3 t 0.20 as determined by the capillary tube technique. The best literature value, 18.#5° (58), can be achieved only after very elaborate and tedious purification procedures. A typical purified product reported in the literature usually melts in the 18.3 - 18.#o temperature range. d6-Dimethy1su1foxide was obtained from the Mallinckrodt Chemical Works as 99.5% isotopically pure compound and was used as received. 19 20 . Di ro lsulfoxide DPS Impure DPSO was obtained from the Aldrich Chemical Com- pany. Although the solvent may decompose during distillation, it was found that slow vacuum distillation was possible if care was taken to provide a high vacuum so that boiling oc- curred at temperatures below 55°. Unfortunately some of the impurities distilled at approximately the same temperature as DPSO so that the distillate was never completely pure. However, this procedure was excellent for the removal of the intense yellow coloration of the original material. After distillation, the product was fractionally frozen several times until its melting point was 25.0 - 27.o°. The litera- ture value is 2n.5 -25.5° (59). If the tedious distillation is avoided, it is possible to obtain colorless DPSO of the same melting range by numerous fractional crystallizatians but with an unavoidably high loss of material. Dibut lsu faxide DBS Impure dibutylsulfoxide was obtained from the Aldrich Chemical Company. The liquid was stored over barium oxide for several days then decanted and subjected to numerous fraction- al crystallizatians. The final product was colorless and had a melting range of 32.0 - 33.0°. The literature value is 32° (60). 1-Pentanol. "Baker Analyzed" ].-pentanol was stored for several days over anhydrous calcium sulfate, then decanted and distilled slowly. The first 15% of the distillate contained a water azeotrope and was discarded. The water content of the center 21 fraction was less than 0.005 M. A higher water concentration resulted if the calcium sulfate was not removed from the sol- vent prior to distillation. Barium oxide could not be used as a drying agent due to its reaction with the alcohol. l-Methyl-Z-Pyrrolidone. Practical grade l-methyl-Z-pyrrolidone was obtained from Eastman Organic Chemicals, Rochester, New York, and was vacuum distilled over granulated barium oxide. The water content of the distillate was less than 0.01 M. The freezing point was -l7° and the boiling point was 710 at approximately # mm pressure. Benzene, Reagent grade benzene was distilled from and stored over barium oxide. The boiling point of the purified material was 80°. Alkali Metal Salts. All of the alkali metal salts,.with the exceptions of LiI, NHuI, NDuBr, KPhnB, RbPhuB, and CsPhnB were reagent grade chemicals and were used after drying without further purification. Except where noted all salts were vacuum dried at 60°. Lithium iodide, 98% pure, was obtained from K & K Lab- oratories and was used without purification. It was neces- sary to recrystallize ammonium iodide from a water - ethanol mixture in order to obtain it completely free of yellow color- ation. The tetraphenylborates were prepared by adding a stoichiometric amount of sodium tetraphenylborate to a chloride solution of the desired cation followed by filtration, washing 22 and vacuum drying. dn-Ammonium bromide was obtained from the Isomet Corporation, Palisades Park, New Jersey, as 98% iso- topically pure material and was used without treatment. It was necessary. to dry lithium perchlorate at 1500 in order to completely remove adsorbed water. ANALYTICAL METHODS Halogen Analyses. Analyses for chloride, bromide and iodide were per- formed by the Mohr volumetric method in which titration is carried out with a standardized silver nitrate solution using potassium chromate as an indicator (61). Water Determination. All quantitative determinations of water were carried out by the Karl Fischer procedure (62). Carbon, Hydrogen, Sulfur Analyses. These analyses were performed by the Spang Micro- analytical Laboratory, Ann Arbor, Michigan. Melting Point Determinations. Due to the hygroscopicity of many of the compounds studied, melting point determinations were performed in sealed capillary tubes. Prepnration of Solutions. Many of the solvents employed in these investigations were hygroscopic and care was taken to prepare them in as nearly an anhydrous condition as possible. Exposure of the solvents and solutions to the air during preparation or trans- fer was minimized by making all transfers with a syringe or pipet. Solutions were prepared at room temperature, approxi- mately 22°. 23 PREPARATION OF SOLID DMSO-ALKALI METAL SALT SOLVATES In general, solid solvates of DMSO-alkali metal salts were prepared either by precipitating the addition compound by the slow addition of benzene to a I‘n solution of the salt in DMSO or, in some cases, simply by obtaining a saturated solution of the desired salt in DMSO at approximately 50° and cooling the solution to room temperature. Precipitated solvates were washed with benzene and dried under vacuum. Prolonged exposure to vacuum was avoided since some of the solvates decomposed Slowly under reduced pressure. Washing with acetone or diethylether resulted in decomposition of some of the solvates. Lithium Solvates. Solvates of lithium bromide and iodide were prepared by precipitation with benzene while those of perchlorate and chloride were obtained by cooling saturated solutions. All of the solvates were hygroscopic so that filtration and washing operations had to be carried out in a dry-box. Sodium Solvates. Either of the above methods can be used for the prepara- tion of sodium solvates. In order to determine if solvates of variable composition could be obtained, sodium iodide so- lutions ranging from 0.1 - l n were prepared and the solvate was precipitated by the addition of benzene. In all cases, identical solvates were obtained. Likewise, there was no de- tectable difference in composition among solvates obtained by the two techniques. In the case of sodium perchlorate, 2# benzene precipitation led to the formation of an oil. The sodium nitrate solvate was found to be unstable and gradually underwent conversion to an oil. The same behavior was repor- ted by Kenttfimaa (##) who noted evolution of nitrogen during the decomposition process. The oils were not characterized. Potassium Solvates. Attempts were made to isolate solvates of the following potassium salts: bromide, iodide, thiocyanate, perchlorate and nitrate. Of these, only the iodide and the thiocyanate solvates could be obtained. The iodide solvate was prepared by both techniques but when prepared by benzene precipitation the product was an oil which was converted to crystalline form by cooling to 0°. The thiocyanate solvate could be prepared as a solid only by cooling a saturated solution. The adduct precipitated as flat, oily plates which converted permanently to an oil when ground with a mortar and pestle. Rubidium and Cesium Sglvates. Attempts were made to prepare solvates of the following salts: bromide, iodide, perchlorate, tetraphenylborate (Rb+ only), and thiocyanate (Célonly). In all cases only the pure salts were recovered. Ammonium Sglvates. The following solvates were prepared by cooling satur- ated solutions: bromide, iodide, thiocyanate and nitrate. The bromide and iodide were obtained as white, crystalline solids which underwent slow discoloration at room tempera- ture. The thiocyanate precipitated as flat, oily plates which converted to an oil under pressure. The nitrate was 25 obtained as a mixture of crystals and oil which gradually converted to an oil at room temperature. Precipitation with benzene in all cases gave oils. Only the ammonium bromide oil could be converted to a solid by cooling to dry-ice temperature. The solid ammonium bromide solvates prepared by the two techniques were identical. The oils were not characterized. INSTRUMENTAL MEASUREMENTS Instrumentation. A Varian A-60 nuclear magnetic resonance spectrometer was used to obtain all proton magnetic resonance spectra. All measurements were made at the normal instrument temper- ature Of 37° using tetramethylsilane as an internal standard. A Cary Model 1# recording Spectrophotometer was used to obtain all ultraviolet and near-infrared spectra. Spectra in the region from #000 - 650 cm.1 were obtained on a Beckman IR-5, Beckman IR-7 or Unioam SP-ZOO Spectrophoto- meter. Polystyrene film was used to calibrate each Spectrum. Spectra of solid samples were obtained from mineral oil or hexachlorobutadiene mulls. Raman Spectra were obtained on a laser Raman spectra- photometer which was designed and constructed at this Uni- versity (63). The excitation line was the 51#5 A line of a Spectra-Physics Model l#0 Argon-Ion laser operating at a power of 1.0 watt. A Perkin-Elmer Model 301 Far-Infrared Spectrophoto- meter was used to obtain all of the spectra in the region from 1 l 667 cm- to 32 cm- . The design of this instrument has been 26 described (6#). It was used without major modification from the manufacturers recommendations although numerous minor changes were made if such were necessary to conduct a given experiment. Some of these modifications will be described. All gratings were calibrated using the rotational Spectrum of water vapor. These absorption frequencies have been tabu- lated by several investigators (65.66.67.68). This spectrophotometer is a double beam recording in- strument Specifically designed for use in the far-infrared Spectral region from 666 - 1# cm-1 (15 - 71# microns). Dif- fraction gratings are used as dispersing elements and are used in the first order only. A globar source is used down to 100 cm-1. Since only a very small fraction of its output consists of far-infrared frequencies (69) , it is necessary to extensively filter the beam in order to remove, or render ineffective, the high intensity, high frequency radiation. Below 100 cm-1 a mercury arc lamp is used as it is superior to the globar but it also retains the problem of high inten- sity undesirable radiation. The Model 301 uses reflection filters, transmission filters and crystal choppers in various combinations to reduce undesirable radiation to acceptable levels. The instrument can be flushed with a dry gas since it is necessary to remove water vapor which absorbs strongly be- low #00 cm'l. Instrument construction, optical layout and electronics are unusually well described in the operational manual sup- plied by the manufacturer. For this reason, such details will not be reproduced here. Some suggestions on Special instru- 2? ment usage and maintenance, which are not adequately described in the manual, are presented in Appendix I. Expenimental Teghnigues in the Far-Infrared Region. Conventional infrared sample handling techniques may be used and in some cases are Simplified in the far-infrared region. The selection of solvents is much broader than in the rocksalt region. Spectra of solid materials. can be ob- tained from dispersions in mineral oil since this material is tranSparent throughout the entire region. Due to the longer wavelengths employed, grinding a sample to extremely small particle size is not as necessary as in the mid-infrared region. It has been reported, however, that the physical state of the sample has a greater effect an observed absorp- tion frequencies in this region than in the rocksalt region (#9,?0). The pressed disc technique is well adapted to far- infrared studies with thin discs of KBr usable to 250 curl and those of CSI to about 130 cm"1 (53). Common paraffin wax (71) or polyethylene powder (72) are convenient materials for preparation of pressed pellets particularly for cases in which the sample would react with alkali halides. An ex- cellent paper has recently appeared which reports sampling techniques which avoid use of any matrix material (73). Window materials for the low frequency infrared region are rather limited in number. Salt windows can only be used to 1 200 cm- at which point CSI becomes opaque: CSBr and KRS-5 1 and KBr to #00 cm-l. (TlBr + TlI) are usable to 250 cm— Of these CsI is the most generally useful as its absorption cut-off occurs at the lowest frequency and it is not diffi- 28 cult to handle. Cesium iodide does fog readily but at low frequencies this does not constitute a serious problem. Win- dows of KRS-5 are not attacked easily by water but the high refractive index of the material causes large reflection losses. 1 but its high refrac- Crystal quartz may be used below 120 cm- tive index also causes serious loss of energy due to reflec- tion from the window surfaces (53). Non-crystalline window materials include polyethylene, polypropylene, polystyrene and paraffin. Polyethylene is the most useful material as it possesses only one weak absorp- tion which occurs at 75 cm"1 (53). Polypropylene is a con- venient substitute for studies in this region (53). Poly- styrene and paraffin are reported to be transparent below 300 cm"1 (74). There are, however, some problems inherent to the use of non-crystalline window materials. They are easily contaminated by some classes of compounds, such as polyhalogens and aro- matics. Such contamination may change the Spectrum of the sample and of the window material itself (75). Care must be taken, therefore, to avoid spurious results. Fortunately, the materials are very inexpensive so that a pair of windows need not be used morethan.once. Another problem encountered is one of flexibility which makes it difficult to construct a leakproof cell with reproducible pathlength. Several cell designs have been tried in this laboratory and it has been found that no one type will fulfill all purposes. The various designs which have been used are discussed in Appendix II. 29 Far-Infrared Spectra a; Alkali Metal Salts in Dialkylsulfoxides. As sulfoxides dissolve alkali halides, polyethylene windows were used throughout this investigation, usually with the Barnes demountable cells which are described in Appendix II. All spectra, other than for purposes of grating calibra- tion, were obtained in double-beam operation. Most solution spectra were obtained with an equal thickness of solvent in the reference beam. Those obtained using sheet polyethylene as reference were corrected for absorption due to the solvent. All intensity measurements reported are in absorbence units measured at the frequency of maximum absorption after correction for any background contribution. Solution Spectra of alkali metal salts were obtained from 660 - 90 cm'l. All solutions were investigated in\ 0.1 mm and 0.2 mm cells except for rubidium and cesium salts in DMSO which were studied using 0.05 mm cells. Except in special cases which will be discussed separately, concentra- tion varied from 0.1 - 2 n. Dipropylsulfoxide is a solid at room temperature, mp 25.- 27°, but the instrument was sufficiently warm that crystallization during an experimental run did not occur. Crystallized dibutylsulfoxide, mp 32°, constituted a more serious problem as it remained solid in the instrument. To avoid this problem, solutions were maintained at 50° and the cells were heated to that temperature prior to filling. Crystallization did not occur in the few minutes required to record a Spectrum. 30 Nuclear Magnetic Resonance Spectra: Mole-Ratio Studies. Solutions for mole-ratio study by nmr were prepared by volume from stock solutions containing known concentrations of salt and DMSO in l-pentanol. Studies were made of lithium iodide, lithium perchlorate, lithium bromide, sodium iodide and ammonium thiocyanate. The molarities of the stock salt solutions were, reSpectively, 1.#, 1.#, 0.9, 0.9 and 0.7. The solutions of lithium bromide and ammonium thia- cyanate were nearly saturated. Solubilities of several potas- sium salts were all less than 0.5 n. The solubility of NaSCN was 0.75 M and that for sodium perchlorate was less than 1.0 n. The solubilities of the DMSO-NaI and DMSO-NaSCN solvates, which formed when the two stock solutions were mixed, were much less than the solubilities of the pure salts. IV. RESULTS AND DISCUSSION Infrared Spectra in DMSO. Infrared spectra of dimethylsulfoxide have been studied in detail by several investigators; Horrocks and Cotton (76) have carried out a complete analysis of the vibrational Spec- trum of DMSO. Bands which are of primary interest to this study are the S-0 fundamental stretching mode at 1055 cm-1, two C-S-O bending modes - a symmetric deformation at 382 cm-1 and an antisymmetric mode at 333 cm-1 - a C-S-C deformation at 308 cm-1 and a methyl torsional mode which Horrocks and Cotton predicted at approximately 200 cm-1. They observed neither the torsional mode,as it was beyond the long wave- length limit of their instrument, nor the C-S-C deformation as it was either ir inactive or unresolved from the C-S-O de- formations. Berney and Weber (77) have studied the DMSO spectrum down to 200 cm'1 and have observed all the bands reported by Horrocks and Cotton in addition to observing the C-S-C defor- mation at 308 cm-1 as a weak band partially resolved from the c-s-o deformation at 333 cm'l. The spectrum of DMSO has been obtained from 5000 - 32 cm-1 in this laboratory. Down to 200 cm-1 Berney and Weber's re- sults were confirmed. In addition, a new broad band was ab- 1 which was assigned to dipole-dipole vibra- served at 80 cm- tions among the solvent molecules. Such bands have been ob- served in other polar solvents (78). No other bands were observed below 200 cm-1. No evidence of the torsional mode predicted by Horrocks and Cotton has been obtained. 31 32 Fan-Infrared Sppctpnn of d6-DMSO. The same vibrations are observed as in DMSO but at lower frequencies. The two C-S-O deformations occur at 337 cm'"1 and 308‘cm-1: the C-S-C deformation was not obser- ved; the dipole-dipole band was not investigated in detail but it may be expected to lie somewhat below 80 cm-l. Infrared Spectra of DMSQ-Alkali Metal Solutions. The Spectrum of a l n solution of sodium iodide in DMSO 1 was obtained from 5000 - 80 cm“ and was found to be identi- cal tothat of pure DMSO except for the appearance of a new, broad and intense band at 200 cm’l. This band could be a- scribed to an ion-pair vibration, or to vibrations between an ion-pair and the solvent or to vibrations of solvated cations or anions. Numerous experiments were conducted to elucidate the nature of this vibration. Solutions of several sodium salts were studied and, in each case, only one band appeared which could not be attri- buted to either DMSO itself or to the anion which was present. It can be seen from Table III that the observed band frequen- cies showed no significant dependence on the anion. This fact is made most evident, for example, by comparing the frequency obtained for sodium tetraphenylborate to the frequency for sodium chloride. The positions of the bands are virtually identical even though the anion masses differ by nearly a full order of magnitude. Of the four possible causes for this vibration, the only one which would be independent of the nature of the anion would 1x3 a vibration between a cation and one or more solvent Table III. Absorption Bands of Sodium Salts in Dimethylsulfoxide Salt NaCl NaBr NaI NaNO NaSCN NaPhuB 33 199 199 198 206 200 198 1+ 1+ 1+ 1+ 1+ uwwwwm H' 3# molecules. Spectra of the other alkali metals in DMSO were obtained from 5000 - 80 cm-l. As in the case of sodium salts, only one new band was observed in each solution which was not due to DMSO or to the anion. The frequencies of the new bands are presented in Table IV and a comparison of the observed and calculated values is given in Table V. It can be seen that there is a large and predictable frequency shift as a result of changing the cation but little or no shift upon changing the anion. Band shapes and relative intensities are presented in Figure 1. In the series from Li+ to Cs+, the combination of decreasing absorptivities and decreasing sol- vent tranSparency made it impossible to obtain spectra for several cesium and rubidium salts and for potassium chloride. The complex anions, perchlorate, thiocyanate and tetra- phenylborate, all possess fundamental vibrations which occur in the far-infrared region. These bands were observed in all perchlorate solutions, at approximately 610 cm-1, but were generally unobserved in the other cases presumably due to the low salt concentrations employed. Using both diatomic and linear-triatomic models, and the appropriate equations (Appendix III),the frequencies of bands due to vibrations between DMSO and the other alkali ions were calculated using the known frequency of DMSO-Na+ as a reference. Due to the known high coordinating ability of lithium ion, it was expected that the observed DMSO-Li+ frequency would lie at a much higher energy than that calculated. The agreement among calculated and observed values was good for all other Table IV. Absorption Bands of Alkali Metal Salts in Dimethylsulfoxide Compound Vmax' cm-1 LiCl 429 i 2 LiBr 429 i 2 LiI 429 i 2 LiNo3 429 i 2 LiCloLp 429 i 2 NHuCl 21# i 5 NHuBr 214 i 4 NHAI 214 i' 11 NHuNOB 214 i 4 NHQClOu 214 i 4 NHuSCN 214 i 4 NaCl 199 i 5 NaBr 199 i 5 NaI 198 i 3 111111103 206 i 3 NaClOu 200 i 3 NaSCN 200 i 3 NaPhuB 198 t 3 KBr 153 t 3 KI 153 i 3 KN03 154 i 3 KSCN 153 i 3 RbBr 125 i 4 RbI 123 i 3 RbNO3 125 i A RbClOu 122 i # 081 110 i 4 CSClOu 109 t h 35 Table V. Comparison of Calculated and Observed Frequencies of Alkali Metal Salts in DMSO Cation ‘ucalcda vcalcdb u_gp§3 Li+ 336 347 429 i 2 NH4+ 221 223 214 i 4 Na+ - - 200 i 3 K+ 165 160 153 i 3 Rb+ 132 120 124 i 4 Cs+ 120 106 110 i 4 a. diatomic model b. triatomic model c. all frequencies are in cm . 36 Figure 1 . 50- B 50— /\ 50’— 50- /\ /\ 50 _. 50 a \ \ \ \\\ ///~\\\ 0 1’//PT\ I l I I I I I 500 #00 300 200 100 cm-1 Frequencies and intensities of bands in alkali metal salt - DMSO solutions. Salt concentrations - 1M: pathlength - 1mm A - Lithium salts, B -.Ammanium salts, C - Sodium salts, D - Potassium salts, E - Rubidium salts, F - CeSium salts. 37 38 cations: the linear-triatomic model appears to be the better approximation. These calculations required the assumption that the force constant of the cation-solvent bond did not vary with change of cation. This assumption may not have been com- pletely valid. However, using NDuBr and d6-DMSO it was pos- sible to calculate directly the effect of changing cation and solvent masses since isotopic substitution Should not affect the force constant. Using a linear-triatomic mode1,it was calculated that the 430 cm"1 band of Li+in.DMSO should shift to 416 cm'1 in d6-DMSO. Spectra of LiI and LiBr in d6-DMSO showed the band to occur at #21 i 2 cm-l. Using the same model, it was cal- culated that the 214 cm'1 1 band in NHJ’solutions would shift to 195 cm- in ND; solution. The observed frequency in NDuBr solution was 200 i 3 cm-l. The observed shifts are in reason- able agreement with the predicted values and clearly indicate that both the cation and the solvent molecule are involved in the vibrating species. In order to determine if the band intensities were func- tions of salt concentration, spectra were obtained of known concentrations of lithium bromide using 0.1 mm cells in the region 550 - 370 cm-1. The results are presented in Table VI. Experimental error in the absorbances may be as high as 3 10% for the 0.08 value but less than this for the others. The data indicate that solutions in this concentration range obey Beer's Law. Further evidence of cationic coordination was obtained Table VI. Absorbances of Lithium Bromide Solutions in DMSO Conan, !. Absorbance 0.00 0.00 0.10 0.08 0.20 0.17 0.#0 0.30 0.80 0.59 39 #0 from a study of the fundamental S-O stretching frequency. If the coordination were due to a dipolar attraction of the cat- ion to DMSO, it would be to the negative dipole on the oxygen atom. Such coordination Should result in a Shift of the 8-0 stretch to lower frequencies (38). This shift was observed for solutions with concentrations greater than 1 M (Table VII, Column 3). It is difficult, however, to determine the exact location of the true S-O stretch or to use its experimental frequency to deduce any other information concerning the sol- vates as it is well known that the observed frequency is not due to a pure stretch but is a mixed mode with substantial con- tribution from methyl rocking motions (76). Currier and Weber, in attempting to correlate the shift of the 8-0 stretching frequency with metal-ligand bond strength for transition metal ion-sulfoxide complexes, have found that little significance could be attached to the extent of the observed shift (39). Addition of up to 7% water to the solutions caused a further shift of the 8-0 stretching frequency but induced no changes in the far-infrared bands. If the vibrations were due to solvent-—-ion-pair inter- actions, the observed frequencies should have been dependent on the mass of the anion. Such cases have been observed by Edgell and co-workers in tetrahydrofuran solutions (21) and by Evans and Lo (79) who observed ion-pair vibrational bands in benzene solutions of tetraalkylammonium halides. In the last case, observed spectral shifts in changing the anion were very close to those predicted on the basis of the cor- responding change in the reduced mass of the vibrating system. Table VII. Effect of Salt Concentration on Spectra in DMSO Compound Concn,M 080 “C80 oMOa'b NHuSCN 0.05 1055 330,378 214 " 0.10 1055 --- 214 " 0.25 1054 --- 214 " 0.50 1051 330,378 212 " 1.0 1047 --- 216 " 3.0 1030 --- 212 " 5.0 1025 --- 215 " 7.0 1020 338,386 208broad L1N03 0.10 1055 330.379 429 " 0.50 1054 --- #28 " 1.0 1050 332,380 428 " 2.0 1044 --- 429 " 4.0 1035 333,382 430 Note: All frequencies are in cm-l. a. The range for ammonium thiocyanate is t 3 cm-l. The range for lithium nitrate is i 2 cm'l. b. The symbol VMO refers to the DMSO - cation vibration. #1 #2 Existence of ion-pairs in concentrated solutions of electrolytes in DMSO has been somewhat debated. Sears pp pg. reported that at concentrations up to 7 x 10.3 M of various sodium, potassium and tetraalkylammonium salts, electrical conductance studies indicate complete dissociation of the salts (32). Gasser and co-workers have studied concentrated solutions of lithium chloride, rubidium iodide and cesium iodide in DMSO. Evaluation of activity coefficients of these electro- lytes led the authors to the conclusion that ionic association occurs in solutions with concentrations larger than 1 x 10-2 M but their electrical conductance data did not support these conclusions (80.81.82). Four studies were conducted in attempts to determine if the anion had any effect on the frequencies or intensities of the observed bands. First, the effect of increasing the anion concen- tration without increasing alkali-cation concentration was investigated by the addition of tetraalkylammonium halides to solutions of lithium and ammonium salts of the same anion. If the anion participates in the vibration which gives rise to the far-infrared activity, addition of anion should result in a change in the frequency and/or intensity of the far-infrared band. The data which are presented in Table VIII demonstrate no such effects. The second study dealt with the effect of bulk dielectric constant on the frequencies of the far-infrared bands. It was expected that with a decreased dielectric Table VIII. Effect of Anion Concentration on Cation-DMSO Bands Sample Soln Reference Soln vmax' cm'l Absorbancea 0.5 n LiI DMSO 430 i 2 0.26 0.5 n LiI + + 0.5 n (Pr),+NI 430 - 2 0.25 0.4 n NHuBr DMSO 212 1' 3 0.47 0.4 M NHuBr + 0.4 n (Bu)uNBr 210 - 3 -- 0.4 M (Bu)uNBr Ooh M NHuBr + + 1.0 M (Bu)4NBr 211 - 2 0.48 1.0 M (Bu)uNBr 43 #4 constant the extent of ion-pair formation would increase. Solutions of lithium bromide and sodium nitrate were studied in the appropriate spectral regions. It should be noted that, due to the limited solubilities of the salts in the solvent mixtures, a large excess of DMSO was always present. As can be seen from Table IX no significant frequency changes were observed. Even more significant is the fact that the intensity of the band was found to be independent of the DMSO:benzene ratio. If the observed band was due to an ion- pair vibration, drastic decrease in the dielectric constant of the medium should result in an increase in the concentra- tion of ion-pairs and, therefore, in the intensity of the observed band. In the third study, Spectra were obtained of ammonium thiocyanate solutions (the most soluble of the salts in DMSO) with the concentration varied from 0.05 M to 7.0 M. Thus the mole-ratio of DMSO/NH4+ was varied from about 350:1 to 1:1. It may be assumed with reasonable certainty that a 0.05 M so- lution of NHuSCN does not contain a high concentration of ion-pairs. However, in a 7‘M solution, in which the mole- ratio of DMSO:NHuSCN is 1:1, there is little doubt that the anion and cation must exist in close proximity. The Spectral data which were obtained are given in Table VII. In no case was therea Significant change in the absorption frequency of the observed band at 214 cm-1. This band broadened and shifted very slightly but these changes may have been related to the extremely high viscosity of the solutions rather than to ion- pair formation. As can be seen, however, the 8-0 stretch Table IX. Absorption Bands of Alkali Metal Salts in DMSO--Benzene Mixtures Compd Concn(M) Solventa xb Diel. Ct. (34) “max NaNO3 1.0 DMSO 1.000 46.4 206 NaNO3 0.60 2:1 C6H6-DMSO 0.386 14.5 205 LiBr 0.05 DMSO 1.000 46.4 431 LiBr 0.10 1:1 C6H6-DMSO 0.567 21.4 #31 LiBr 0.10 2:1 C6H6-DMSO 0.386 14.5 433 LiBr 0.05 4:1 C6H6-DMSO 0.222 8.4 429 LiBr > 0.05* 5:1 C6H6-DMSO 0.201 6.9 #25 a. b. * Sa t'd Volume Ratios. Mole Fraction DMSO #5 ,cm- l+ H' l+ H- l+ H- wwwwmww H' l #6 moved to lower frequencies with increasing concentration of the salt while the C-S-O bending modes shifted to higher fre- quencies. These effects indicate increasing coordination of the DMSO. A similar experiment with lithium nitrate, Table VII, gave identical results. In the fourth study, the spectra of SCN' and N03- were investigated as a function of DMSO:salt mole-ratio. In all cases from highly dilute solutions to saturated solutions, in which mole-ratios were 1:1 for DMSO:NHuSCN and approxi- mately 3:1 for DMSO:LiN03, the anion spectra did not under- go significant change. If it can be assumed that a strong interaction between an anion and a metal cation or a solvent molecule will distort the symmetry of the anion, and there- fore perturb its infrared spectrum, it can be concluded that no such strong interactions are occurring in these solutions. Such distortion of the nitrate ion symmetry has been observed in fused lithium nitrate and was attributed to strong inter— action with the cation (83). The apparent lack of measure- able anion salvation is in agreement with results obtained by Prue and Sherringtan (84) which show that DMSO solvates anions only very weakly due to shielding of the positive end of the solvent dipole by the two methyl groups. The results of all four anion studies indicate clearly that the presence or absence of ion-pairs in these solutions does not appreciably affect the DMSO-cation vibrations which are observed. It seems reasonable to assume that in solutions so highly concentrated that there exists no molar excess of DMSO, #7 the number of DMSO molecules intimately associated with a single cation will be less than the number so associated in a solution which contains a large molar excess of DMSO. The similarity of Spectra of such different solutions indicated that all solvate structures possessed identical infrared Spectra. In order to further investigate this possibility, solid DMSO-alkali metal solvates were prepared, and their elemental analyses and infrared spectra were obtained (pp 51-56). The results showed that the cation-DMSO band frequencies were, indeed, independent of the stoichiometries obtained. Infrared Spectra a; Dipropyl- and Dibutylsulfoxide Solutions. Investigations were extended to the above solvents in order to study the effects of decreasing dielectric constant, increasing mass and increasing electron donating ability which is dependent on the inductive effect of the alkyl groups. The importance of the inductive effect in determining the complexing abilities of various sulfoxides has been demon- strated (#0). Just as in the case of DMSO, solutions of alkali metal salts in dipropylsulfoxide (DPSO) and dibutylsulfoxide (DBSO) each gave a single band characteristic of the cation. The observed frequencies are given in Table X. In comparing the Spectra obtained from DMSO solutions to those obtained for the other sulfoxides certain trends may be anticipated. In the DMSO - DBSO series inductive effects will tend to shift a given cation-solvent band to higher frequency. The order predicted from such a consider- Table X. Absorption Bands of Alkali Metal Salts in Dimethylsulfoxide, Dipropylsulfoxide and Dibutylsulfoxide __sL_Co ound .2113; .2199... JESS)... LiCl 429 i 2 416 i 4 426 i 3 LiBr 429 i 2 421 i 4 425 t 3 Lil 429 1 2 420 i‘4 425 t 3 LiNo3 429 i 2 421 i~4 425 t 3 Limo,IL 429 i 2 421 t 4 426 t 3 NH4C1 214 i 5 221 t 4 -- NHnBr 214 i 4 222 t 3 .226 t 3 NHuI 214 i 4 223 t 3 225 t 3 NHhNOB 214 i 4 223 t 3 226 t 3 NHuClOu 214 i 4 224 t 3 226 t 3 NHuSCN 214 t 4 224 t 3 225 t 3 NaCl 199 t 5 "‘ '- NaBr .199 i 5 216 t 5 -- max 198 i 3 219 i 3 219 t 5 NaNO3 206 i 3 218 t 5 219 i 7 N3°1°4 200 i 3 217 t 3 221 i 5 NaSCN 200 i 3 221 t 3 224 t 5 NaPhuB 198 i 3 220 t 7 226 t 5 KBr 153 t 3 -- -- KI 153 i 3 156 t 3 152 t 5 K1103 154 t 3 -- -- KClOu -- 153 * 4 ~- KSCN 153 i 3 154 t 4 155 t 5 KPhuB --- 153 i 3 154 * 4 RbBr 125 t 4 -- —- RbI 123 i 3 123 t 6 -- RbNO3 125 i 4 -- -- RbClOu 122 i 4 -- -- CSI 110 i 4 -- -- c.3010,p 109 i 4 -- -- Note: all frequencies are in cm- #8 #9 ation would be (n-CnH9)280 > (n-C3H7ESO > (CH3)ZSO. This effect should be greatest for Li+ and least for as+ due to their relative charge to mass ratios. The effect of increasing solvent mass will, however, tend to Shift a given cation-solvent band to lower frequency. This effect Should be least important for Li+, due to its very low mass relative to that of a solvent molecule, and most important for Rb+ and Cs+. The expected predominance of the inductive effect for lithium ion is observed for the change from DPSO to DBSO. The lower frequency of the band in DPSO solutions as com- pared to that in DMSO is somewhat unexpected but it may be related to the small size of lithium ion and to as yet un- explained steric effects. Similar unpredictable behavior has been observed in sulfoxide complexes of transition metal salts (39). Using the sodium band frequency in each solvent as a reference, estimations were made of the locations of other cationic vibrations based upon a linear-triatomic approx- imation. The results are presented in Table XI. Due to the known high coordinating ability of Li+, it was expected that the lithium frequencies should lie at much higher energies than those calculated. The deviations observed for the ammonium ion, though not large, are understandable. Although the mass of this cation falls between Li+ and Na+, its large radius and low charge density will decrease the influence of solvent induction substantially over that for Table XI. Calculated and Observed Band Frequencies (cm-1) 9gpgpn DMSO DPSO DBSO v calcd v obs v calcd v abs 9 calcd \) obs Li+ 347 429 384 421 394 425 NHu+ 223 214 244 223 248 226 Na+ -- 200 -- 218 -- 223 K+ 160 153 172 154 174 154 Rb+ 120 124 125 123 125 -- Cs+ 106 110 106 -- 106 -- 50 51 Na+ resulting in low observed values. Agreement among cal- culated and observed values is acceptable for potassium, rubidium and cesium salts. It is interesting to note that the mass and inductive effects, being in opposition as stated above, exactly balance in the cases of K+ and Rb+. For lithium, ammonium and sodium ions, the inductive effect clearly predominates. Insufficient data are available for cesium salts but it would be expected that the mass effect would predominate and a shift to lower energy would be observed. The decreasing dielectric constant encountered in the series from dimethyl- to dibutylsulfoxide might have induced formation of-ion-pairs: however, no evidence to verify such formation was obtained. I frare S ectra a Solid DMSO So vates Melting points and analyses of the solvates are given in Table XII. Although the solvate stoichiometries are unambiguous, in several cases the C:H:S ratios indicate some DMSO decompo- sition. However, the deviations are no more than those found in the literature for many transition metal ion-DMSO complexes. Evidently, slight decomposition of DMSO is not uncommon al- though the reasons for such decomposition are unclear. As in- dicated, the melting points frequently were not sharp, but, except where noted, they were reversible indicating that they were true melting points and did not merely represent decom- position of the solvate. Kenttfimaa's method of preparation consisted of placing the solid salt in contact with DMSO for several weeks fol- maouoamfioo >3 nonmamcd .Q coauflmomfiouop n p .m .mmoH uLmHSB any mcflusmmoe pom Omzn mcHNHHHumHo> m.>m III III III m.mm III III III p or >.m¢ III III III Qm.m¢ III III III p m.am III III III nm.mm III III III fled I mma III mo.>m no.4 m¢.md III mm.mm mo.m HN.ON moa I moa III Od.d¢ H¢.¢ o>.mm III >m.o¢ oa.m mm.mm ema I NMH 0.48 om.mm sm.¢ mm.>a Nw.m¢ NN.MN mm.¢ o¢.>a mod I and m.m¢ om.am >N.¢ mm.md m>.mw mm.am NH.¢ mm.md U ooa III III III III III III III III «NH I or N.m¢ III III III om.N¢ III III III mma I mad .N.m¢ III III III om.mm III III III p 00a III m~.mm N>.m mm.dm III NN.®N mm.m SH.HN mm I mm ~.mm III III III mm.mm III III III Hm I On o.¢m III III III oa.¢N III III III Na 1 mm N.¢m III III III hd.mm III III III p wow x: m m o INS m m o Imal bosom pmumasuamo mmum>aom paaom MO coaumNHHouumuan .HHN oHQmB omzom.s.Hx omzom.zomw omzae.meammz omzon.eoaomz omzom.zommz omzom.m.Hmz omzom.a.ammz omzox.zomemz omzam.m.H+mz oozes.o.amemz omzom.soaoaa omzom.Haa omzom.m.amaq OmSQH.HUHA panamamu 52 53 lowed by filtering and vacuum drying (#4). This method pos- sesses several inherent difficulties: first, there is no proof that all of the salt undergoes conversion to the solvate: sec- ond, over such a long time interval, decomposition of the sol- vent may become more and more severe: third, the extended vacuum drying necessary to remove all residual DMSO may change the stoichiometry of some of the solvates. His anal- ysis consisted of photometric determination of the metal with no C,.,H,-orS analyses so that the extent of the solvent decomposition is unknown. However, solubility data reported in the same paper occasionally indicate much higher solu- bilities than those found in this laboratory. It was stated that some of the solubilities increased with time indicating a slow solvate formation. Such time dependence has not been noted here and it is thought that the time dependency was due to gradual solvent decomposition or pickup of water rather than to slow solvate formation. The stoichiometries and melting points which Kenttamaa obtained are presented in Table XIII. The infrared data for the solid solvates prepared in this investigation are given in Table XIV. By analogy with the Spectra of DMSO-transition metal complexes, it can be inferred that, as for the corresponding alkali metal solutions, coordination in these solvates occurs through the oxygen atom of DMSO. The similarity among the solution and solid state spectra of DMSO solvates is striking. In the Spectra of the solids, as in spectra of the corresponding solutions, only one new Table XIII. Stoichiometries and Melting Points of Solid DMSO-Alkali Metal Salt Solvates (44) 0 Solvate mp, C LiC1°1DMSO partial 200° LiBr'BDMSO #0 - #4 LiI°3DMSO 60 - 70 LiClOu°3.5DMSO 55 - 60 NaBr'ZDMSO partial 105 - 110 NaI°2DMSO partial 128 - 135 NaClOu°3DMSO partial 100 - 105 KI°2DMSO partial 78 - 82 KBr and K0104 did not form solvates 54 Table X IV 0 Compound LiCl-lDMSO LiBr'3.5DMSO LiI-3DMSO LiClOuo3DMSO NH43CN°IDMSO NaBr. la SDMSO NaSCN'ZDMSO NaClOu'3DMSO NaPh4B04DMSO KSCN‘ZDMSO KI.105DMSO DMSO a. ”CH and v50 are coupled in DMSO. 1028 1025 1015 1024 1068m 10303 1055 1030 1030 1030 1055 ”QSQ 373W. 355m 369w. 332w 370w, 334w obscured by 9 MO 384m, 337m 387mo 340m, 3358ho 3293h very complex multiplet 395m9 344m 388m. 383sh. 360W. 337m 385m. 339m 386m! 352", 336W 3918h. 379m, 361'. 333m 380": 335W 383m. 3768h. 346W. 337m 378M. 330m strongest and broadest band. b. Abbreviations: br,broad: v,very. 55 Infrared Absorption Spectra of DMSO Solvates A V MO 408s, #05w 450 453 #40 br 205-139br 188m,br 203br,w 200m,br 1903 1908 1968 l98br l48br 1508, 1308h ‘So taken as the s,strong: m,medium: w,weak: sh,shou1der: 56 band is observed which is not due to DMSO or to the anion present. Its frequency in most cases is very nearly identical to that obtained for the solution solvates and it is com- pletely independent of the anion involved. This new band may be assigned to the metal-oxygen stretching mode. Its frequency is also independent of the stoichiometry of the species giving further indication that it is a virtually pure M-O stretch with little contribution from its surroun- dings. This may be additionally confirmed by consideration of the frequency ratios for the typical alkali metal ions, Na+ and K+z 195/109 = 1.31. For a pure M-O stretch, the expected ratio would be that of the square roots of the molecular weights of the metals involved:(39/23)% = 1.30. The assignment of these bands as being the M-0 stretching mode appears to be justified. The M-O band in the lithium chloride solvate occurs at a measureably lower frequency than that for the other lithium solvates. In addition, the band is split whereas for the other salts the band is not Split. This compound differs from the others also in that its melting point is unusually high occurring at 207°. The reason for these differences is not clear but evidently this solvate forms with a completely different crystal structure and symmetry than do the others. Overtgne Spectra of DMSO. Attempts were made to locate one of the overtones of the fundamental S-O stretching mode since it was possible that the overtones would not have been as complex as was the fun- damental itself (p00). 57 The S-O fundamental stretching frequency is 1055 cm-1; it would be expected that the first overtone of this vibration 1 1 would occur at 2110 cm’ , the second at 3165 cm- and the third at 4220 cm"1 , etc. (disregarding anharmonicity which would tend to lower the observed frequencies.). The over- tones are expected to have much lower intensities than the fundamental making it necessary to investigate them with relatively thick cells. Although the spectrum of DMSO in the region around 2100 cm"1 contained several weak bands, none of them shifted upon addi- tion of high concentrations of salts. Evidentlm.the first S—O overtone is not observable in liquid DMSO; it has, how- ever, been observed in the vapor phase (38). 1 l Spectra of DMSO between 1H,300 cm- and #100 cm- were obtained using 1 mm cells. The.spectrum recorded with a 1 mm cell was consistent both in peak location and relative intensity with the spectrum of any aliphatic hydro— carbon (85). No evidence of bands involving sulfur or oxygen were observed. The Spectrum of a l‘fl lithium bromide solution was no different from pure DMSO except for a new. broad band at 1 5150 cm" which was attributed to a water vibration since addition of water to the solution enhanced the intensity of this peak. The 5150 cm'1 band may provide a nondestructive method for determining water in DMSO and probably also in other sulfoxides. 1 1 Peaks at #013 cm' and 3923 cm' were observed in pure DMSO which were not characteristic of simple hydrocarbon. 58 Upon addition of salt to the solvent, both bands shifted very slightly to higher frequencies. The maximum shift was only 5 cm-1 for 7‘g NHuSCN. The shift is opposite in direction to that expected for an S-O overtone and may be due to combina- tions involving either the methyl rocking modes or the C-S stretching modes both of which shift to higher energy upon addition of salt. Far-Infrared Spectra of Alkali Meta; Salts in Mixtures of DMS and -Meth -2-P olidone. The band due to lithium ion in DMSO occurs at #30 cm”1 while in l-methyl-Z—pyrrolidone it occurs at 399 cm-1. In order to determine if individual peaks could be observed in mixtures of the two solvents, solutions of 0.5 fi_LiC10u were prepared in several mixtures, as well as in the pure solvents. The spectra were obtained from 550 - 350 cm-1 using 0.1 mm cells. The intensities of the bands in the pure solvents were approximately equal. Individual peaks were not observed but the 430 cm-1 peak in pure DMSO shifted continu- ously to lower frequency as the mole fraction of 1-methyl-2- pyrrolidone was increased. For mole fractions of DMSO from 1.0 - 0.7 the shift was small but it increased rapidly when the latter value was reached. Below 0.? the shift was linear with change in mole fraction DMSO. The band did not broaden appreciably in solutions of intermediate mole fraction which indicated that the absorption was a singlet and was not com- posed of two bands representing individual solvates of DMSO and methylpyrrolidone. It is reasonable to assume, based up- on the known behavior of Li+ in several nonaqueous systems, 59 that the DMSO-Li+ solvate and the methylpyrrolidone-Li+ solvate possess the same stoichiometry. If the two solvents were competing equally for the cation, i.e., if the formation constants of the two solvate structures were equal. then the frequency shift would be expected to be linear with change in mole fraction of either solvent. The small observed shift above a mole fraction of 0.7 would seem to indicate that the DMSO was preferentially coordinating to the cation and. therefore, that the formation constant of the DMSO solvate was greater than that for the l-methyl-Z-pyrrol- idone solvate. Nuclear Magpetig Resopgncg Spectra of Dimethyl- and Dipropyl- sulfoxide Solutions. (As stated previously, nuclear magnetic resonance techni- ques have been applied recently to the determination of solva- tion stoichiometries. The simplest case is one in which the nmr signal of a "free" solvent molecule is completely resolved from that of a ”solvating" solvent producing two peaks the areas of which are proportional to the concentrations of the two species. Such optimum conditions occur only if the ex- change of solvent molecules with those in the solvation ens velope is rather slow. Such conditions have been observed in the system A1(C10u)3-DMSO (86). In the case of alkali metal salt solutions the exchange is too rapid to allow the resolu- tion of the two signals, but there is appreciable shift of the DMSO proton singlet as a function of salt concentration. The technique employed in this investigation was a classical mole-ratio study in which the concentration of the 60 alkali salt was held constant and the DMSO conCentration was varied over a wide range. The shift of the DMSO proton signal was measured in each solution. In order for a study of this type to be meaningful, three conditions must be obeyed. An inert solvent must be available which will dissolve all solution components, a measurable parameter must exist which is a function of solvate concentration, and the formation constant of the solvate must be sufficiently large so that a limiting value of the parameter can be obtained. In this case the parameter was the proton nmr signal of a DMSO molecule. Its position in the nmr spectrum is dependent on whether or not it is included in the solvation shell of an ion, e.g., coordination of DMSO to a cation will result in a decrease in electron density around the proton nuclei resulting in a downfield shift of their resonant frequency. Since truly inert solvents did not dissolve appreciable amounts of common alkali metal salts it was necessary to select a solvent with a minimal interaction towards DMSO and the alkali metal ions. l-Pentanol represents a good compromise of the various requirements. It will appreciably dissolve alkali metal salts and, although it may hydrogen-bond to the DMSO, this interaction is much weaker than that between DMSO and a metal cation (3). In addition, the lack of a cation-solvent far—infrared band in solutions of sodium iodide in l-pentanol is evidence that the interaction of sodium ion with this alcohol is considerably weaker than with DMSO (Table XV). 61 Fortunately, the mole-ratio data obtained provided a built-in check of the "inertness" of the solvent and indicated that it was, in fact, sufficiently unreactive that it caused no appreciable error. Figure 2 illustrates the results of the study. Curve A is a mole-ratio plot of the DMSO-LiClOu system in which lithium perchlorate concentration was kept at 0.70h n. Curve B is a similar plot for 0.70 g lithium iodide while Curve C shows the magnitude of the shift induced by l-pentanol over the same range of concentrations of DMSO but without the presence of lithium perchlorate. The shift for pure DMSO is 156 cps downfield from tetramethylsilane. Both with lithium perchlorate and with lithium iodide a clear break at the Dmso/Li+ ratio of 2:1 is evident. It can be seen that although the alcohol has some effect on the proton signals of DMSO, it does not invalidate the indicated stoichiometry. It would be helpful to correct Curves A and B for the DMSO-- pentanol interaction but to do this quantitatively it is necessary to know the formation constant of the DMSO-Li+ solvate. It is possible, however, to make the reasonable approximation that while the DMSO:Li+ mole-ratio is 2:1 or less essentially all the DMSO present in solution is coordinated to the lithium ion. If now Curve C is subtracted from Curves A or B it is seen that the break at 2:1 mole-ratio is accentuated. At mole-ratios greater than 2:1 some of the observed shift will be due to interaction with the alcohol so that a 168 F- 167 — 166 - 165 ~ 161! 163 cps 162 161 160 159 158 Ratio DMSO:MX Figure 2. Chemical shifts of DMSO protons pg mole-ratio DMSO:MX A - L1C10u, B - LlI, C - blank 62 63 "correction" may be legitimately applied. As can be seen, the effect of such a correction will always be to enhance the break and in no way to remove or decrease its clarity. It is thought, therefore, that although 1-pentanol is not an ideal solvent for this type of study, it does not induce a sufficient error to invalidate the conclusions. The maximum solubility which could be attained with any of the ammonium salts was 0.50 g with NHnsCN. The mole- ratio plot is shown in Figure 3 (Curve A). It can be seen that the interaction is strong with a weak inflection at a mole-ratio of 2:1. The decreased clarity of the break as compared to that obtained from Li+ may be indicative of a lower force constant for the ammonium ion-DMSO interaction or it may merely represent increased competition with the alcohol due to the lower concentrations of salt and DMSO employed. The study of sodium iodide was limited to solutions with a maximum concentration of 0.22 h. At higher concentrations a solid solvate DMSO-NaI precipitated. The mole-ratio plot (Figure 3, Curve B) differs significantly from those for Li+ or NHL,+ in that its slope is much less, indicating a strong interaction even at mole-ratios as high as 9:1. It seems reasonable to assume that while sodium ion is sol- vated by DMSO, in this system either the solvate is much less well defined than in the case of lithium ion or more than one well-defined stoichiometries exist simultaneously. The former interpretation is in accord with the usual behavior cps Figure 3 . \‘x‘ \‘° 3‘ I \ L:| ‘ 163 _ an. r 3 c i i e - a a 162 - ~‘ ‘ “a B 161 - \\ ¢ 160 t A 159 '— C l l l I i L 4L l 1 I 2 4 8 10 Chemical shift of DMSO protons pg mole-ratio DMSO:MX A - NHuSCN, Ratio DMSO:MX B "' N31, 6h C - blank 65 of lithium and sodium ions in solution. The same technique was applied to lithium perchlorate and dipropylsulfoxide in l-pentanol. The data which were obtained are given in Figure h. Curve A represents sol- utions with LiClOu, Curve B represents those without LiClOn. As can be seen, the effect of the alcohol is much larger than in the case of DMSO and no break was obtained. Calculation 0rce Constants and Fre uencies or M DMS 2: The results of the nmr investigations indicated a 1:2 stoichiometry, of the type M(DMSO)2+, for both lithium ion and ammonium ion in DMSO solution. The most likely configuration for the 1:2 solvates is linear-triatomic. The far-infrared band for each solvate is evidently the highest energy mode, the asymmetric stretch. Using the linear-triatomic formula (Appendix III) the force constant of the DMSO-metal bond and the frequency of the infrared inactive symmetric stretching mode were calculated. The force constants for lithium and ammonium 4 dynes/cm and solvates are, respectively, 3.6 x 10 2.2 x 10“ dynes/cm. The frequencies for the symmetric stretchs, which may be Raman active, are, respectively, 88 cm"1 and 54 cm’l. Very rough estimations for the infrared 1 and 25 cm'l. active bending mode are, respectively, #0 cm- These latter bands may not be observable since dimethyl- sulfoxide is not highly transparent in these regions. It is interesting to note that if the same equation is used to calculate force constants for the other alkali cations 161; 163 162 161 cps 160 159 158 157 l l OM— 2 lb Ratio DPSO:MX Figure LL. Chemical shifts of DPSO c-protons 1s mole-ratio DPSO:MX A - LiClou. B - blank 66 67, in DMSO, values are obtained which are virtually identical with the one obtained for NH4(DMSO)2+. However, at this time, without stoichiometric data for the other solvates, little significance can be attached to such calculations. Bamap Spectra pf LiCth-DMSQ Sclutiops. Raman spectra of 0.75 5, 1.5 g and 3.0 h lithium per- 1 to 550 cm'l. chlorate solutions were obtained from 20 cm- There were no significant features of the spectra which could not be attributed to DMSO or to the anion. The exciting line was broadened substantially from its normal width so that 1 would have been somewhat any weak Raman lines below 100 cm- obscured. The absence of Raman lines cannot be considered highly significant since the intensities of Raman lines due to vibrations of polar bonds are expected to be low (87). Far-Infirared Spectra pf Miscellaneous Solutiops. Incomplete results of spectral studies involving aceto- nitrile, nitromethane, Zepropanol, l-pentanol, tetrahydro- furan and several tetraalkylammonium salts are presented in Table XV. H .Huso arm as assumes soaposompe omn-oom .Huso mum pm assawms :oapahomp< ommuoom .Hlso mo: pm ESEaHmS Soapghomfld ommuoow .Aompmoaoa pom paosaamaxov HIso mam pm camp xmoz oHHIoom .30H hpaaansaom .onpmHompm oz oaaloow .3oH auaaapsaom .soaudsomns oz oHHnoow .sssdnma oz .soapaaomnm pecan mam» one Moms aho> omaiomm .Qoapahomnm on whom Hdmz .smaq .Hmz omaiomm .maaNms soapahomnm oz onalomm .Hlso mam on also own m>opm soak noapmhomnm ocean aho> omHIomm .Huso cam pm soapasomnm moose use aces osauowm .soapaaomnm wnammoaooo aaaomoum p59 .ssaaxms oz onaaomm .oowsmSond who; woman psobaom 03» one .mhpooam aoflasoo magmas Hsoapsoea teem Has Hmz .Haq .smeq .Hodq ommummm .Haso wmm osm Hass one as modem omaummm .csseeas useeaom aha on sseeocom +aq-omzo Hessoz osmnomm .Huso 00H mamumsaxonaam pm Coapahompm muonpm owalomm meadmom mmmwmmm mppooam ozoosmfiaoomaz .>x canoe Hons Han moqu stasmv staezv Hozsaozv zomsmz 80 CH moaososvonm .m awe awe awe omzm .omzo omzn Hosmpzomna Hosmpcoaia Honmpcmmsa HostOHanm HostOHauN Headachanm HostOHaum HonQOHQIN osmozsomzo m 20 mo pcmbaom 68 V. CONCLUSIONS The results of these investigations indicate that the techniques of far-infrared Spectroscopy and nuclear magnetic resonance spectrometry can be combined to give a new and useful approach to the study of ionic salvation. Low frequency infrared bands in solutions of alkali metal salts in a variety of nonaqueous solvents, of both low and high polarity, are attributable to the vibrations of cations in solvent cages. It is possible to study the effects on these vibrations of dielectric constant, anion and cation masses, anion and cation concentrations, solvent mass and solvent electron donor ability by observing the infrared spectra of appropriate solutions. Additionally, in some cases, the relative solvating ability of two sol- vents may be qualitatively compared. Methods employing nuclear magnetic resonance spectrometry allow in some cases the determination of salvation numbers. These data, in combination with the observed infrared bands and a reasonable assumption concerning the configuration of the solvated species, allow calculation of the force constant of the bond between the cation and the solvent molecule. Such force constants should give an indication of the relative solvating abilities in a series of solvents. Specifically concerning alkali metal salt solutions in dialkylsulfoxides, the far-infrared bands are due to the vibration of the alkali cation in a cage of solvent molecules with the oxygen atom of the sulfoxides in closest proximity to the cation. Although it is clear that the 69 70 nature or the mass of the anion does not influence the vibrational frequency of the sulfoxide-cation bond it cannot be concluded that the anion is absent from the immediate surroundings of the solvated cation. The extent of anionic solvation does not appear to be large, but the nature of this interaction is not yet well understood. Although, in this case, the infrared spectra cannot readily distinguish among several existing cationic sol- vates, the observed frequencies of the cation-solvent bond can be used to calculate the force constant of this bond if the stoichiometry is known. Nuclear magnetic resonance spectrometry appears, at this time, to be the most generally useful technique for determining solvate stoichiometries. A. VI. PROPOSALS FOR FUTURE INVESTIGATIONS Infrared Studies 1. Calculations indicate that the low frequency modes of the solvate structures lie below 100 cm‘l. Although this is not an easy region in which to work, observation of these modes would allow determination of all the force constants of the solvate structures and would contribute evidence concerning the configuration of the solvates. Many polar solvents will have dipole-dipole bands below 1 100 cm- but this problem may be avoided by proper selec- tion of solvent, e.g., dipropylsulfoxide is much more transparent at frequencies below 100 cm"1 than is dimethyl- sulfoxide. 2. With the recent availability of polyethylene cells with rather highly reproducible and stable pathlengths, it may now be possible to perform mole-ratio, or other similar studies, in the far-infrared region providing a suitable solvent is available. The use of cesium iodide windows with a thin covering of polyethylene, to avoid dissolution of the window, should be investigated for this purpose. Windows of KRS-5 should not be as readily attacked as are alkali halide windows and may be well suited to such studies. 3. The observed S-O fundamental stretch in dipropyl- and dibutylsulfoxides does not appear to be as complex as in DMSO. It is possible that shifts of this band in DPSO or DBSO as a result of cationic solvation may be useful in determinations of relative bond strengths between the sol- 71 B. C. 72 vent and various cations. h. Those solutions, such as lithium salts in acetone, which are known to consist largely of ion-pairs should be investigated to determine if, first, the extent of ion-- pairing affects the infrared spectrum of the solvate and second, if a band due to vibration of the ion-pair itself can be located. 5.‘ Studies of anionic solvation in hydroxylic sol- vents have been conducted by measuring shifts of the 0-H frequency as a result of solvation. There may be low fre- quency bands due to the anion-solvent bond in such systems. Although not highly transparent, hydroxylic solvents can be studies by far-infrared spectroscopy. 6. A logical extension of the current investigation is to the study of hydrogen ion solvation. Preliminary investigations of hydrogen ion solvation in DMSO solution of methane sulfonic acid have been reported (91). The interpretation of data was inconclusive and somewhat debatable. Further investigations of this type could yield valuable results. .Raman Studies If any bands due to the solvate vibrations prove to be Raman active, valuable information concerning con- figuration may be obtained. uNuclear Magnetic Resonance Studies 1. Both sodium and lithium nmr need to be investigated as these methods constitute a very direct means for D. 73 studying alkali cation salvation. 2. The numerous methods currently developed for ab- taining transition-metal-ion salvation numbers should be extended to include alkali metal salts if possible. Miscellaneous Studies of solution properties such as viscosity, refractive index and conductivity have historically been used to determine the nature of ionic solutions. These techniques, although limited in scope, may lend signifi- cant support to correct interpretation of the ir and nmr data now being collected. APPENDIX I The following sections are written to be understand- able by anyone familiar with operational details of the Perkin-Elmer Model 301 Spectrophotometer. I. Instrument Usage. A. Optics. Due to the necessity of extensive filtra- tion to remove short wavelength light, the spectral region covered by the Model 301 is divided into numer- ous short spans each with a different combination of optics. At times, the interchange points between two different optical arrangements may be so located as to interfere with a measurement. It should be noted that although the recommended optics are optimum for their respective regions, other combinations will fre- quently suffice for many purposes. For example, reststrahlen filters may always be used at higher than recommended frequencies, although with some energy loss. In most spectral regions, this loss may be off- set through use of the normal instrument adjustments. Thus, if an indicated filter change cannot be accom- plished without disrupting a critical portion of a Spectrum it may be possible to deviate from the es- tablished rules without seriously affecting the qual- ity of the recorded spectrum. However, most such fil- ters may not be used at lower than recommended fre- quencies due to the appearance of chopped second or- der radiation at the detector which will result in recording an inaccurate spectrum. It should be clear 74 75 that in order to know when a given set of optics may be extended beyond its recommended range it is neces- sary to understand the purpose and operation of each optical component used. It is acceptable to insert additional filters into the beam if such are desirable. For example, in the region from 660 - 320 cm-1 silvered mirrors are used and the beam is quite hot, approximately no - u5° at the sample compartment focus. This temperature may cause difficulties in handling volatile liquids and has been known to cause decomposition of some complexes. However, insertion of a sheet of black polyethylene film, such as is available in any garden supply store, will reduce the temperature to ambient levels. A less efficient, but more convenient, procedure for reducing the beam temperature is merely to use scatter plates in place of the silver mirrors. The resultant energy loss may be compensated for by the usual means. Another example concerns use of the choppers outside their recommended ranges. It is possible to use opaque choppers well beyond the indicated limit of 31 microns merely by leaving the an micron cut-on interference filter in the beam. Second order radiation will not appear at the detector until a wavelength of 08 microns is reached. It is B. 76 possible therefore to cover more than 70% of the frequency range accessible to the instrument with only one optics change, a grating, which requires stopping the scan. There are many such alterations that may be made which will not be covered here. The purpose in this section was to point out that the recom- mended procedures are merely recommendations and do not constitute a rigid set of unbreakable rules. However, it should be noted that when operating under any conditions which are less than optimum, it may be necessary to accept a lowered resolution, a decreased signal-to-noise ratio or a slower scan speed. For the additional convenience obtained, such losses may be acceptable in many experiments. Flushing. Although flushing with dry gas should begin ‘ at about #00 cm'l. the operation is normally not 1 is reached. This is due to begun until 320 cm‘ the fact that a major optics change occurs at this point making it a convenient frequency to begin. Liquid nitrogen boil-off has been found to be the best choice for this purpose. Compressed nitrogen may also be used but should be passed through a suitable dessiccant to remove traces of water. In current use for this purpose is a tube con- taining 100 g of phosphorous pentoxide impregnated on an inert support. The dessiccant is available from the Mallinckrodt Chemical Works, and is 77 tradenamed Aquasorb. An airdryer suitable for use with the instrument is also available from the Perkin-Elmer Corporation. C. Scan Speed. It is normally desirable to scan as rapidly as is consistent with obtaining data of suff- icient accuracy for the intended purpose. The maximum scanning speed is normally used for survey purposes. When accurate determination of frequency, intensity, or band contour is desired, slower Speeds will be required. A scan speed which is too fast for accurate measurement will be demonstrated in four possible ways. The recorder pen will lag the actual absorbance changes resulting in recording a band which will be lower in both intensity and frequency than the true values: resolvable bands will blend into a single envelope: the-baseline will show uncompensation effects,i:e., the scan is so fast that the slits will not be able to main- tain constant energy to the detector when scanning across a water vapor band. All of these effects will be lessened by decreasing the scan rate. D. Amplifier Gain. Before recording a spectrum, it is desirable to know what amplifier gain setting will produce desirable slit widths. In order to do so, it first must be decided what will be the maximum tolerable slit width. Second, the frequency at which the system energy is minimum must be located. This is usually done by rapidly scanning the 78 frequency interval of interest by manually turning the grating drum while observing the slit. The proper gain setting is that which will produce the maximum tolerable slit width at the point of minimum system energy. It should be noted that, according to Perkin-Elmer technicians, the amplifier will not function properly below a setting of 8. II. Instrument Maintenance. A. Lubrication. a. Breakers. This subject is adequately covered in the manual. b. Recorder. Lubrication of the Leeds and Northrup recorder is an involved procedure and should be performed monthly. Detailed information is con- tained in the following service manual: Speedomax Type C Recorders, Model S 60000 Series, Leeds and Northrup.Company, #901 Stenton Avenue, Philadelphia ##, Pennsylvania. c. Slit Drive. The'only likely source of difficulty is in the rotating barrel mechanism. If noise develops it should be repaired by a Perkin-Elmer technician. d. vaelepgth Drive. If binding develops it will become evident during manual turning of the drum. To correct this problem, slide right end of instru- ment over the end of the table, remove the bottom plate and lightly all all of the moving parts of the drive mechanism. Remove all excess oil. B. C. 79 CsI choppers. After extended use with the mercury arc source, the side of the crystals nearest the source will become etched, which will result in the chopping of short wavelength light. To check for the amount of stray light, i.e., undesirably short wavelength radiation, which is detected as a result of the fogged chopper, set the instrument up for operation at 100 microns in single beam mode. Block the beam with an opaque material and adjust the transmittance to zero. If the opaque material is removed and a cesium iodide window inserted in the beam, the transmittance will be zero unless higher grating orders are being chopped by the fogged blade and not completely removed by the scat- ter plates and reststrahlen filter. According to the Perkin-Elmer technician, if the transmittance exceeds 2% or 3%, the chopper should be repolished. Polishing service is available from the Harshaw Chemical Company, l9#5 E. 97th St., Cleveland; Ohio, ##106 or from the International Crystal Laboratories, 120 Coit St., Irvington, New Jersey 07111. Current prices from the two com- panies are $35.00 and $17.00, reSpectively, for repolishing one side each of two 3.5" semicircular crystals of cesium iodide. Detector. a. Golpy bulb. It is possible to observe the con- dition of this bulb by inserting a white card between 80 the Golay cell and the ellipsoidal mirror in the de- tector compartment. If the Golay power supply is turned on, a spot of light will be plainly visible on the card unless the bulb is burned out. Proper aiming of this bulb after installation is critical to good detector operation. The directions given in the instrument manual have been found to be, at best, inadequate if not completely incorrect. The adjustment is not difficult but before attempting it for the first time one should have a lesson from a Perkin-Elmer technician. b. Gplay cell. A defective cell usually is demon- strated by a high noise level which will be in op- posite phase in the Direct I and Direct Io channels. See Test Number 2.3-Instrument Zero- in the instrument manual. 0. Photgtube. There is no built-in test to determine the quality of this component. It is, however, a nor- mally longlived tube and should not need replacement for many years. If its operation is suspect the only method for checking it is by replacement with a new phototube. Wavelength accuracy. This should be periodically checked for each grating by scanning a portion of the water vapor spectrum in single beam mode and comparing the result to the spectrum obtained when the grating was calibrated. Two adjustments may cause difficulty. Each grating is mounted in its holder with an angle 81 determined by a set screw on its base which may become loose and allow the angle to change. The entire wave- length drive is held together with a single hex-head screw located under the monochromator section and ac- cessible from underneath the instrument. If this screw works loose, all gratings will simultaneously go out of calibration. Electronics. a. Main Pgwer Supply. This is a Lambda Electronics Corporation Model C-281. Maintenance instructions are not in the 301 manual but are in a descriptive manual available from Lambda Electronics Corporation, 515 Broad Hollow Rd., Huntington, L. I., New York. An accurate and involved procedure for checking over- all operation is in this manual. A rapid check of output voltage may be made by measuring the voltage between a good ground and the +DC terminal on the power supply terminal board. The voltage reading should be within a few volts of 2#0v. b. Globar Power Supply, Mercury Source Ppwer Supply and Control Unit. These units have required no ser- vicing so no comment can be made. The Globar itself should operate at 200 watts. The Variac setting which will give this power input may be determined by mea- suring the ac voltage and ac current at the jacks provided. As the variac setting is increased from zero, with the globar turned on, the power will climb to a value above 200 watts then gradually return to 82 200 watts, i.e., two different Variac settings will produce a 200 watt input with the higher one of the two being correct. At the lower setting, the ballast tubes have not begun to regulate and the power will vary with line voltage fluctuation. The three tubes located at the lower left rear of the elec- tronics rack will be barely glowing when the power is properly adjusted. c. Recorder. This Leeds and Northrup Model G is a standard recorder which will rarely require ser- vicing. The units described in a, b and c have proven to be virtually trouble-free and so constructed that gradual degradation of vacuum tube operation does not seriously affect overall performance. The unit described in section d is, however, more sensitive to tube weakness or failure. d. Pro-amplifier and Amplifier. Among the numerous vacuum tubes in these sections are two which are critical and whose performances are not adequately de- scribed by testing in a commercial tube tester. They must be tested by replacement with tubes which have been shown to work well in the same configuration in the amplifier. One is located in the preamp section in the detector compartment and is type 6072. The other is a 12AT7/6201 located at the right rear of the amplifier chassis. This is the main voltage amplifier tube and its characteristics are critical to proper amplifier operation. At one time, ten new 83 tubes of each type were checked by replacement. Of these only two of each type produced a sufficient amplification to allow the amplifier to meet the specification test described in the instrument manual and several of each operated so poorly that vir- tually no response was obtained under the conditions of the test. The test signal which can be injected into the circuit provides a convenient test for the amp- lifier and, therefore, for the power supply. If the specifications for amplification can be met using this signal in the appropriate test it is safe to assume that all components are functioning well unless other problems are evident, such as ex- cessive noise or intermittent operation. General Electric tubes are supplied with the instrument and, in the two critical amplifier configurations, have been found to be superior to other brands of the same tube type. It should be noted that replacement tubes obtained from Perkin-Elmer, at a substantial premium, are "bff-the-shelf" items and have not beenchecked for proper performance in the Model 301. APPENDIX II Cells for supporting mineral oil mulls are easily made. If reproducibility of thickness is not a requirement, a typical mull cell such as is available from numerous sup- pliers may be used with polyethylene windows in place of salt plates. Demountable liquid cells with 2 mm polyethylene windows and appropriate Spacers may be used if it is necessary to obtain Spectra using known pathlengths. These cells, available from the Barnes Engineering Company, Stamford, Connecticutt, are designed so that uniform pressure is placed on approximately 50% of the window surface. This pressure largely overcomes the usual irregularities associated with the flexible windows. The types of cells available to accommodate liquids are more varied with the particular type chosen beingpde- pendent on the nature of the experiment. Barnes Engineering Company has made available a series of molded polyethylene cells for far-infrared studies in pathlengths from 0.1 - 10 mm. For thicknesses greater than 0.5 mm these cells are acceptable but pathlengths of the cells below 0.5 mm have been found to be highly unreliable. However, for qualitative work or for those cases in which it is not necessary to place an equal pathlength of solvent in the reference beam these cells are convenient. The molded cells would be excellent for volatile liquids if some material could be found which would seal the filling hole. As yet, this has not been satis- factorily accomplished. For capillary films, such as are necessary to obtain spectra of highly hydrogen-bonded systems, use of thin film 8# 85 bags has been reported (88). These cells are excellent for highly volatile liquids but are not of reproducible path- length. A pair of homemade cells has been constructed which work very well for volatile liquids if pathlengths of 2 mm or greater can be used. They consist of two aluminum plates with center circular windows which are sandwiched around two equal sized sheets of polyethylene with no center hole. These in turn are sandwiched around an equal sized poly- ethylene sheet spacer with a center hole. Six holes to accommodate screws are drilled around the window area. The screws provide the means of tightening the entire assembly into a reproducible and fairly leakproof cell. For spacers of 2 mm or greater it is possible to drill two parts through the spacer to the top and bottom of the cavity for easy filling. The filling holes may be well sealed with machined Teflon plugs. In order to obtain spectra of highly volatile liquids, the following cells work well but are not at all reproducible and also are not available in short pathlengths. Using the previously mentioned aluminum sandwich cellaithe polyethylene spacer is replaced by a rubber O-ring of the appropriate size. The screws are tightened partially until the poly- ethylene windows are held tightly against the O-ring. The cell is filled by inserting two needles through the ring, done more easily before thehscrews are tightened, using one as a filler and one as a vent. After filling, the needles 86 are removed. Such a cell will hold even the most volatile liquids for several days at elevated temperatures without loss. For most purposes, the Barnes demountable liquid cells have been found to be convenient and of rather highly re- producible pathlength varying from i 1% to i 10% depending on the pathlength and window material. All common alkali halide windows are commercially available and those of poly- ethylene may be easily made by punching circular windows out of sheet polyethylene using a hydraulic press and a hardened steel die of the appropriate dimensions. The thick- ness of polyethylene to use is a matter of personal pre- ference although 2 mm Sheet has proven to be sufficiently transparent and rigid to satisfy most requirements. For those experiments in which it is necessary to main- tain a constant cell path from one run to the next it is inadvisable to dismantle the cell for cleaning purposes between each run. Although the demountable cells can be reassembled with deviations not exceeding 10% in pathlength, this amount of uncertainty may be intolerable. It was found that the cells could be satisfactorily cleaned without dismantling by flushing with a suitable solvent, usually acetone, followed by flushing with pentane, heating in a 1100 oven for approximately 30 seconds and final drying with dried compressed air. Pentane was used as a final rinse as it is highly volatile, transparent in the far-infrared, and immiscible with sulfoxides. APPENDIX III The force constant and vibrational frequency of a diatomic molecule, XY, is given by Hook's Law (89): s ,v ___ l [kmx + my)ch ch mxmy 1 where v is the frequency in cm- , 10 c is the speed of light, 3 x 10 cm/sec, k is the force constant in dynes/cm, and m are the reSpective atomic weights, mx y and N is Avogadro's Number, 6.023 x 1023. 0 The force constants and frequencies of the two stretching modes of a linear-triatomic molecule, XYZ, are given by the following expressions (90): “2 2 k1 N0 TTV =—_ 1 m 02 y 2m‘1k.N #flszz = l + y l O 2 .xjmy. where v1 and v3 refer to the symmetric and asymmetric stretching frequencies, respectively. For the present purpose, the equations will be used to describe molecular vibrations with X representing a cation and Y a molecule of DMSO: the terms diatomic and triatomic will be retained for convenience. 87 BIBLIOGRAPHY 1. Robinson, R. A., Stokes, R. H., "Electrolyte Sol- utions," Butterworths Scientific Publications, London, 1955, p 62. 2. Fajans, K., Qeut. Physikal. Ges., 21, 709 (1919). 3. Parker, A. J., Quart. Rev. (London), 16, 163 (1962). #. Bockris, J. 0'M., ibid., 3, 173 (19#9). 5. Griffiths, T. R., Symons, M. C. H., Mal. Ph 8., 3, 90 (1960). 60 Debye. Po. J. Chem. Phlgo. l, 13, (1933). 7. Same as l but pp 60-62. 8. Same as l but complete text. 9. Harned, H. 8., Owen, B. B., "The Physical Chemistry of Electrolytic Solutions," 3rd ed, Reinhold Pub- lishing Corp., New York, N.Y., 1958. 10. Schaschel, E., Day, M. C., J. Amer. Chem. Soc., 11. Price, E., "The Chemistry of Nonaqueous Solvents," Vol. 1, J. J. Lagowski, Ed., Academic Press Inc., New York, N.Y., 1966, Chapter 2. 120 Kosower. E0 M0, J. Amer. Chem. SOCO. fig, 3253 (1958). 13. Durst, R. A., Taylor, J. K., J. Res. Natl. Bur. Std., A68, 625 (l96#). l#. Lund, H., Acta Chem. Scand., 12, 298 (1958). 15. Lord, R. 0., Merrifield, R. E., J. Chem. Phys., 21. 166 (1953). 16. Bufalini, J., Stern, K. H., J. Amer. Chem. Soc., 83, #362 (1961). 17. Hyne, J. B., Levy, R. M., Can. J. Chem., 59, 692 (1962). 18. Thompson, W. K., Trans. Faraday Soc., pg, 266? (1967). 19. Worley, J. D., Klotz, I. M., Jp_gpppp_gpy§., #5, 2868 (1966). 20. Pullin, A. D. E., Pollock J., Trans. Faraday Soc., ill. 11 (1958). 88 21. 22. 23. 2#. 25. 26. 27. 28. 29. 30. 310 32. 3#. 35. 36. 37. 38. 39. #0. #1. 89 Edgell, W. F., Watts, A.T., Lyford, J., Risen, W. M. Jr., J. Amer. Chem. Soc., 88, 1815 (1966). Edgell, W. F., Abstracts l53rd Meeting of the American Chemical Society, Miami, Fla., April 9-1# (1967) R-l#9. Edgell, W. F., Lyford, J., and Fisher, J., Abstracts 155th Meeting of the American Chemical Society, San Francisco, Cal., March 31 - April 5 (1968) S-136. Wall, T. T., Hornig, D. F., J. Chem.-Phys ., #5, 3#2# (1966). Wall, T. T., Hornig,.D. F. ibid., #2, 78# (1967). walrafen. G. E., ibid., £&,15u6 (1966). Hester, R. E. and Plane, R. A., Inorg. Chem., 3,769 (l96#). Hester, R. E. and Plane, R. A., Spectochim. Acta, g3_, 2289 (1967). Hogg, A. M., Haynes, R. M., Kebarle, P., J. Amer. Cheg. Soc., 88, 28 (1966). Hinton, J. F., Amis, E. 3., Chem. Rev., 62, 367 (1967). Martin, D., Weise, A., and Niclas, H-J., Apgew. Chem., (EngliSh'Edition), 6, 318 (1967). Sears, P. G., Lester, G. R., and Dawson, L. H., J. Phys. Chem., fig, 1&33 (1956). Kharasch, N., and Thyagaragan, B. S.,"Quarterly Reports on Sulfur Chemistry," 3 (l , (1966). Lindberg, J. J., Kenttamaa, J., and Nissema, A., Suomen Kemistilehti, B3#, 156 (1961). Cotton, F. A., Francis, R., J. Amer. Chem. Soc., gg, 2986 (1960). Douglas, T.B., J. Amer. Chem. Soc., 29, 2001 (l9#8). Lindberg J. J., Kenttamaa, J., and Nissema, A., Suomen Kemistilehti, B3#, 98 (1961). Cotton, F. A., Francis, R., and Horrocks, W. D., Jr., J. Phys. Chem., 63, l53# (1960). Currier, W. F., and Weber, J. H., Inorg. Chem., 3, 1539 (1967). Klaeboe, F., Acta Chem. Scand., 33, 27 (l96#). Sears, P. G., personal communication. 90 1+2. Gur'yanova. E. No. Zhuro Fit. Khim.’ all. “79 (1950). #3. Hammiok, D. L., Williams, R. E., J. Chem. Soc., 1228, 211. ##. Kenttamaa, J., Suomgn ggmistilehti, 33B, 179 (1966). #5. Skerlak, T., Ninkov, B., Sislov, V., Glasnik Drustra Hemicara Technol. SR Bosne Hercegovine, 38, 93 (1963): Co A., _EE: 278b 19 O #6. Koenig, H., Metzger, H., Seelert, K., Chem. Ber., 28, 3712 (1965)- #7. Schlafer, H. L., Schaffernicht, W., Apgew. Chem., 2, 618 19 0 . #8. Rubens, H., Van Baeyer, 0., Phil. Mag., 83, 689 (1911). #9. Bentley, F. F., Wolfarth, E. F., Srp, N. E., Powell, W. R., Spectrochim. Acta, 33, l (1958)- 500 Hurley, W. Jo. J. Chem. Ede, E3, 236 (1966). 51. Bentley, F. F., in "Handbook of Analytical Chemistry," lst ed, L. Meites, Ed., McGraw-Hill Book Co., New York, N.Y., 1963, Section 6, pp 152-168. 52. Wood, J. L., Quart. Rev.. (London), 32, 362 (1963). 53. McDevitt, N. T., Rozek, A. L., Davidson, A. D., Technical Documentary Report ML TDR 6#-l92, Air Force Materials Laboratory, Wright-Patterson Air Force Base, Ohio, l96#. 5#. Whetsel, K., Chem. E . News, 88 (6), 82 (196#). 55. Ferraro, J. R., Anal. Chem., _#_0_ (#), 2#A (1968). 56. Gore, R. C.,Anal. Chem., 88, 12 (195#). 57. Butler, J. N., J. Electroanal. Chemqh33, 89 (1961). 58. Same as 33 but p 81. 59. Addison, C. C., Sheldon, J. C., J. Chem. Soc., 1956, 2705. 60. Searles, S. Jr., Hays, H. R., J. Org. Chem., 83, 2028 (1958). 61. Skoog, D. A., West, D. M., "Fundamentals of Analy- tical Chemistry," Holt, Rinehart and Winston Co., New York. NoYo. 1956, p 2610 62. 63. 6#. 65. 66. 67. 68. 69. 70. 71. 72. 73- 7#. 75. 76. 77- 78. 79. 91 Stone, K. G., "Determination of Organic Compounds," McGraw-Hill Book Company, Inc., New York, N.Y., 1956' p 60 Miller, R. E., Treuil, K. L., Getty, R. H., Leroi, G. E., Michigan State University, Technical Report N00 h(29). may. 1968. Helms, C. C., Jones, H. W., Russo, A. J., Siegler, E. H. Jr., Spegtrochim. Acta, 39, 819 (1963). RObinson. Do We, Jo Opt. 5000 Amo,‘&2, 966 (1959). YOShinta, HO, FUJlta, SO. Minami' SO. oetjen' R. A., Yamada. Y0. lbidO. fig, 315 (1957). Blaine, L. H., Plyler, C. K., Benedict, W. S., J. Res. Natl. Euro Std., 66A. 223 (1962). Narahari Rao, K., Humphreys, C. J., Rank, D. H., "Wavelength Standards in the Infrared," Academic Press, New York, N.Y., 1966. Oetjen, R. A., Haynie, W. H., Ward, W. M., Hansler, R. LO. SOhBuWSOker, KO E0. B611, E0 E0. Jo OEta SOC. Amero, fig. 559 (1952). Miller, F. A., Carlson, G. L., Bentley, F. F., Jones, W. H., S ectrochim. Acta, 38, 135 (1960). YOShlnaga, Ho, Oetjen, Re A., g. QEto $200 Ager., £5. 1085 (1955)- Smethurst, B., Steele, D., Spegtpoghip. Agta,.8g, 2#2 (l96#). SCthng, K. J., Anal. Chem., 2Q. 523 (1966). MCCUbbinS, To Kc Jr., Sinton. We Mo, Jo Opt. 3290 Ager., 52.9.. 537 (1950). Rothschild, W. G., Spectrochim. Acta, 83, 852 (1965). Horrocks, W. D. Jr., Cotton, F. A., Spectrgphip. Ac , .12. 13# (1961). Berney, C. V., Weber, J. H., 33532h_£fl§ag.,,2. 283 (1968). Jakobsen, R. J., Brasch, J. W., J. Amer. Chap, 309., so. 3571 (1964) . Evans, Jo Co. LO. Go Y-So, J. Phys. Chem., £2. 3223 (1965)- 80. 81. 82. 83. 8#. 85. 86. 87. 88. 89. 90. 91. 92 Dunnett, J. 8., Gasser, R. P. H., TranS.Faraday Soc., £1. 922 (1965)- Archer, M. 0., Gasser, R. P. H., ibid., 88,3#51(l966). Crawford, J. M., Gasser, R. P. H., ibid., 83, 2758 (1967). Wilmshurst, J. K., Senderoff, S., J. Chem. Phys., 35. 1078 (1961). Prue, J. E., Sherringtan, P. J., Trans. Faraday Soc., 52. 1795 (1961). Goddu, R. F., in "Advances in Analytical Chemistry and Instrumentation," C. N. Reilly, Ed., Vol. 1, Interscience, New York, N.Y., 1960, pp 362 - 375. Thomas, S., Reynolds, W. L., MM” .1111. 3l#8 (1966). Redlich, 0., Chem. Rev., 39, 3#3 (l9#6). Wood, J. L., Taimsalu, P., Spectrpchim. Acta, 88, 1357 (1964) . Willard, Ho Ho. Herritt. Lo Lo, Dean. Jo A... "Instrumental Methods of Analysis," #th ed, D. van Nostrand Co., Inc., Princeton, New Jersey, 1965, p 12#. Herzberg, G., ”Infrared and Raman Spectra," D. van Nostrand Co., Inc., Princeton, New Jersey, l9#5, p 172. Williams, J. M., Kreevoy, M., J. Amer. Chem. Soc., 89, 5499 (1967).