A VARIABLE TEMPERATURE NUCLEAR MAGNETIC DOUBLE RESONANCE STUDY OF AQUEOUS ALKYLAMMONIUM IONS BY Bruce Erickson Wenzel A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1969 (Q Copyright by BRUCE ERICKSON WBNZEL J1970' To my parents, who taught me the importance of an education and a profession, to my sisters, who encouraged me and showed me the way, to my wife, who stood by me and advised me, and to my progeny, who made everything worth- while and necessary, I dedicate this thesis. ii ACKNOWLEDGMENTS It is with sincere appreciation that I acknowledge the counsel of Professor Max T. Rogers, under whose direc- tion these investigations were conducted. Thanks are expressed to Dr. Ruth R. Benerito, Dr. Joseph H. Boyer, Dr. John J. Eisch, and Dr. Irving Siegel for encouragement, inspiration, and advice. iii TABLE OF CONTENTS Chapter Page I . INTRODUCTION ........................ 1 II. HISTORICAL REVIEW . . . ................... 4 A. NMR Studies of Proton Exchange Involving Substituted Ammonium Ions .............. 4 Ammonia .......................... 4 Ammonium Salts ...................... 10 Methylammonium Salts .................. 13 Dimethylammonium Salts ................. l9 Trimethylammonium Salts ................. 21 Other Alkylammonium Salts ................ 27 Other Nitrogen Compounds ................ 29 B. Nuclear Relaxation and Quadrupolar Effects ...... . 31 General Theory of Nuclear Relaxation ........... 31 Relaxation for Nuclei with a Quadrupole Moment . . . . . 39 Nuclear Relaxation Times of Nitrogen Compounds . . . . 43 Viscosity and Temperature Effects on T1 of Quadrupolar Nuclei ................. . 50 Activation Energies for the Relaxation Process ...... 58 Solvent Effects . . . . . . . . ...... . ....... 60 C. Nitrogen Chemical Shifts ................. 72 D. Nuclear Spin-Spin Coupling ................ 81 General Observations ................... 81 Mechanism of Spin-Spin Coupling ........... . 86 Spin Coupling in Nitrogen Compounds ......... . 88 Proton-Proton Coupling Constants ........... 89 14N-—Proton Coupling Constants ............ 98 15N Proton Coupling Constants ............ 107 iv Chapter Page E. Nuclear Magnetic Double Resonance Spectroscopy ...................... 1 14 III. EXPERIMENTAL ......................... 137 A. Purification of Compounds ............... . 137 B. NMR Spectrometers .................... 141 C. Homonuclear Decoupling-Variable Temperature Experiments ...................... 143 D. Heteronuclear Decoupling Experiments .......... 150 E. Viscosity Measurements ................ . 165 IV. RESULTS ................ . ........... 167 A. Spectra and Plots of Data ................. 167 B. Data Analysis by Digital Computer ............ 186 C. Data .......................... . 190 V . DISCUSSION .......................... 216 A. N-H Lineshapes in the Proton Spectra of Substituted Ammonium Ions ............. . 216 Contributions to Linewidths from Proton Exchange . . . . 216 Contributions from 14N Spin-Lattice Relaxation and Viscosity .................... . 222 Spin-Spin Coupling Constants ............. . 233 B. Heteronuclear Decoupling ............... . 236 NH Proton Lineshapes ................... 237 Long-range lH-MN Spin-Spin Coupling ........ . 239 Nitrogen Chemical Shifts ................. 240 Chapter Page C. Solvent Effect on Proton NMR Spectra of Ammonium Salts .................... 251 Changes of 1,](1H-14N) with Solvent and Temperature ...................... 2 5 1 General Effects on NH Lineshapes ............ 255 Ammonium Ion in Acidic (protic) Solvents . . . . . . . . . 256 Ammonium Ion in Non-Acid (aprotic) Solvents ..... . 260 VI. SUMMARY ........................... 269 VII. RECOMMENDATIONS .................... . 273 BIBLIOGRAPHY ............................ 276 APPENDICES ............................. 291 A. Program QUADRELX .................... 292 B. Program NMR PIT8 ................... . 297 vi 10. 11. 12. 13. 14. LIST OF TABLES Reported nitrogen chemical shifts (ppm). Three-bond proton-proton coupling through nitrogen of the type 3J(R-N§-Q§3). . . . . Three-bond proton-proton coupling through nitrogen of the type 3J(flC-N§+). . . . . Proton-proton coupling constants for the ethyl group. . . . . . . . . . . . . . 1H—14N coupling constants across one bond. 1H-14N coupling constants across two bonds 1H-14N coupling constants across three bonds 1H-lsN coupling constants. . . . . . . . . Concentrations and pH values for solutions of substituted ammonium salts used in this work I O O I O I I O I O O O D O O I O O 14N relaxation times for various J values. Results of computer least-squares fitting of ammonium proton lineshapes . . . . . . . . Variation of 14N relaxation time with tem- perature for substituted ammonium ions . Results of least-squares fitting of relax— ation time-temperature data. . . . . . . Results of least-squares fitting of viscos- ity-temperature data - . . . . . . . . . . vii Page 90 93 96 100 103 105 109 138 187 190 192 195 196 easured solution viscosities. . . . . >in-1attice relaxation rates and calculated solution viscosities . . . . . . . . . . . . asults of least—squares fitting of data tc the equation l/T1( 4N),1§.W/T. . . . . . . . elaxation rates calculated for three selected values of viscosity/temperature. . . . . . . roton coupling constants and chemical shifts for alkylammonium ions . . . . . . . . . . . easured nitrogen and proton NH chemical Shifts O O O O O O O O O O O O 9 0 O . 9 . . .trogen chemical shifts for substituted amines . . . . . . . . . . . . . . . . . . . .trogen chemical shifts for substituted ammo- nium ions. . . . . . . . . . . . . . . . . .trogen chemical shifts for nitriles and isonitriles. . . . . . . . . . . . . . . .trogen chemical shifts for amides. . . . .trogen chemical shifts for nitro compounds . easured viscosities for ammonium salts in various solvents . . . . . . . . . immary of NMR results for ammonium salts in various solvents and at various temperatures lYSiCll properties of solvents. . . . . . . . Llculated linewidths for central NH proton of aqueous monomethylanmonium ion . . . . . . r(IR-141V) for ammonium salts in various sol— vents. . . . . . . . . . . . . . . . . . . . mange of NH proton chemical shift with tem- perature . . . . . . . . . . . . . . . . . . viii Page 197 198 200 200 201 204 205 206 207 208 209 209 210 214 218 253 266 LIST OF FIGURES Figure Page 1. Detail of Dewar jacket, 60 MHz receiver coil nsert, and 4.33 MHz transmitter coil for H-14N decoupling experiments. . . . . . . . 156 2. Detail of modified probe for variable temper- ature decoupling experiments showing rela- tive position of supporting acessories . . . 157 3. Photograph of the experimental heteronuclear double resonance system of Figure 2. . . . . 158 4. Block diagram of electronic configuration for heteronuclear double resonance experiments . 160 5. Complete proton resonance spectra of a few alkylammonium chlorides in aqueous solution. 168 6. NH proton spectra for a series of alkylammonium salts. I I I I I I I I I I I I I I I I I I I 169 7. NH lineshapes calculated for various values of 7] I lOnJTl from Pople's Equation. (Ref. 9) . . 170 8. Effect of temperature on the NH resonance of monomethylammonium and monoethylammonium ions I I I I I I I 0 I I I I I I I I I I I I 171 9. NH resonance of monomethylammonium ion showing (A) single resonance spectrum (3) methyl pro- ton decoupling and (C) nitrogen decoupling . 172 10. NH resonance of monomethylammonium ion showing variation with the frequency used to decouple the alkyl protons. . . . . . . . . . . . . . 173 ix ‘- resonance of monomethylammonium ion showing Iariation with the frequency used to de- :ouple nitrogen. . . . . . . . . . . . . . . :iable temperature study of NH proton reso- iance of monoethylammonium ion while de- :oupling nitrogen. . . . . . . . . . . . . . :ail of the methylene proton spectrum of the series of ethylammonium ions showing the .ine broadening. . . . . . . . . . . . . . . i spin-lattice relaxation times‘xs. recipro- :al of absolute temperature for the alkylam- nonium salts . . . . . . . . . . . . . . . . l relaxation rates 1;. viscosity/temperature for the alkylammonium salts. . . . . . . . . .culated and experimental nitrogen chemical shifts for the alkylammonium salts . . . . . proton chemical shifts for the alkylammonium ialts I I I I I I I I I I I I I I I I I I I I :rogen chemical shifts ya. 0P values (Grim) for amines and substituted ammonium ions. Phe chemical shifts are relative within each series . . . . . . . . . . . . . . . . . . . .ative nitrogen chemical shifts and op val- 1es for nitriles and nitro-compounds . . . . .ative nitrogen chemical shifts and up val- ies for amides . . . . . . . . . . . . . . . >ton NMR lineshapes for ammonium salts in rarious solvents at selected temperatures. lnge of NH proton chemical shift with tem- perature for ammonium salts in several dif- ferent solvents. . . . . . . . . . . . . Page 174 175 176 177 178 179 180 181 182 183 184 185 I . INTRODUCTION The proton magnetic resonance spectra of the major- ity of organic compounds containing nitrogen yield only broadened signals for hydrogens attached directly to ni- trogen or removed two and three bond lengths from it. In a few: exceptional compounds, however, triplet structure is observed due to spin-spin interactions between the protons (I a 1/2) and the nitrogen (14N, I - l). The broadening of proton signals for the majority of nitrogen-containing compounds may be a result of hy— drogen bonding with nitrogen through the lone-pair elec- trons or of intermediate rates of exchange of protons between the nitrogen and the solvent. Another cause of broadening is the nitrogen nucleus itself which possesses an electric quadrupole moment that may interact with surrounding electric fields. Where exchange is slow a symmetrical electric field leads to relatively slow spin-lattice relaxation and a symmetrical (lxltl) triplet in the N—H proton spec- trum from coupling with 14” (5:1): an unsymmetrical 2 electric field, on the other hand, lead: to rapid spin-lattice relaxation which collapses the triplet to a broad or sharp single line. This thesis is primarily concerned with ef- fects of the nitrogen nucleus on the proton resonance spectra of nitrogen compounds and with nitrogen NMR spectra themselves. The effects of symmetry around nitrogen and of temperature on the nitrogen spin-lattice relaxation process have been studied. A set of compounds with grad— ually increasing asymmetry of the electric field at the nitrogen atom have been chosen for study: these are the mono—, di— and trialkyl ammonium chlorides. Increasing temperature has two possible effects on the proton N—H lineshapes depending on whether exchange or quadrupolar spin—lattice relaxation processes;xedominato. The effect of increased proton exchange rate between ni- trogen and solvent is to cause a collapse of multiplet structure in the nuclear magnetic resonance spectrum, e.g. a triplet would broaden and in the limit collapse to a single broad line which would then become narrower at very high exchange rates. Conversely, for quadrupolar nuclei in nonspherically symmetric electric environments, in— creasing temperature causes relaxation times to increase and relaxation rates to decrease. One purpose of this If; . . In 3 thesis is to determine which process governs the proton NH lineshapes in NMR spectra of the mono-, di-, and tri- alkyl ammonium chlorides in aqueous solution at low pH. Another method of studying the NH lineshapes in order to determine the effect of the nitrogen is by means of heteronuclear decoupling, i.e., observing the proton resonance spectrum while irradiating nitrogen at its resonance frequency with high RF power. An additional result of the double resonance study is to show the exis— tence of nitrogen spin coupling with protons two and three bonds distant. II. HISTORICAL REVIEW A. NMR Studies 9f Proton Exchgnge Inyglying Substituted Ammonigm Igng Meals The earliest nuclear magnetic resonance (NMR) work on amines was the kinetic study of proton exchange between methylamine and water performed by observing changes in the hydrogen resonance spectrum produced by changes in hydrogen ion and amine concentration. (1,2) This work was predated by that of Ogg (3,4), who reported the effect of proton exchange rates on the liquid ammonia spectrum. It is appropriate to begin any dis— cussion of the kinetics of proton exchangu in amines with a brief review of the NMR work on ammonia and the ammonia ion. In their study of the effects of electrolytes on the proton chemical shift of the solvent water, Gutowsky and Saika (5) mentioned that solutions of ammonium hydrox- ide and ammonium chloride gave single lines whose shifts were concentration dependent as a result of proton exchange. 4 Gutowsky and Fujiwara (6) then showed from the change of proton chemical shift that there is Significant associa— tion of ammonia with water in dynamic equilibrium and that the structure can neither be adequately represented as free ammonia mixed with water nor as ammonium ion and hydroxide ion. Ogg showed (3) that, first. there is a marked chemical shift on going from gaseous to liquid ammonia which he attributed to hydrogen bonding in the liquid phase. Secondly, he showed that a trace of water causes rapid proton exchange which collapses the triplet resulting from spin-spin interaction with the 14N nucleus (I=l). (It should be pointed out here that there is some dispute as to the trace quantity of water and conditions necessary for proton exchange in liquid ammonia to occur. Nevertheless, these questions are deferred for the present). In conclusion then, although the systems observed were Similar, the approaches and results were quite different. Whereas Gutowsky gt, 1. (5,6) observed chemical shift changes for the composite single proton line (NH3, NH4+. OH-,H20) by quantitatively varying the concentration of ammonia or ammonium salts, Ogg (4) noted the gradual collapse of the proton triplet by quantitatively varying the speCies present either in dry liquid ammonia ‘ + . . + (NHZ or NH4) or in aqueous ammonia (H3O+ thus NH4). 6 Whereas Gutowsky st, 314” on the one hand, only speculated that dynamic processes were occurring, Ogg (4) was the first to show what species contributed and that it was possible to use NMR for obtaining kinetic data on ammonia and amines. Considering all reports on amines and aqueous ammonium salts since the first work of Ogg and Gutowsky _J; ngL, there have been only developments of the basic ideas outlined above. The developments have followed three paths, first, improvements in chemical techniques have permitted more accurate values for the concentrations of various species in the systems observed. Secondly, im- provement in magnetic resonance spectrometers has yielded more accurate values of the NMR parameters - chemical shifts, spin-spin splittings,half-height widths and spin relaxation times, as well as more accurate representations of line shapes. The third path has been the development of models for the system and mathematical treatment of the measurable quantities to yield reaction rate constants. Ogg and Ray (7) were the first to clearly demon- strate that the unequal line widths and line heights for the proton triplet of pure dry liquid ammonia was a result of relaxation from quadrupolar interaction with the 14N 7 A sample of pure dry ammonia enriched 99.8% with uadrupolar 15N nucleus was prepared in the same natural ammonia and yielded a proton spectrum con— f a very sharp and narrow doublet. At about this , Roberts (8) showed that this was generally true ge number of amines and nitrogen compounds, i.e., ening of proton resonances was due to 14N quadru- xation rather than hydrogen bonding or rapid proton As a result of these reports, Pople (9) presented based on nuclear quadrupole spin-lattice relaxa- ing the broadening of multiplet components, which y reproduced the line shapes observed. However, a iiscussion of relaxation effects is postponed until :tion II B). irchall and Jolly (10, 11) have used liquid ammonia ant to measure the pK for weak acids such as malono— :yclopentadiene and 2,4—dinitroaniline. They t the triplet in the proton spectrum of the solvent allapsed in solutions of the stronger acids 9K > 15) and attributed this to increased concen— E ammonium ion resulting in the process: NHZ+NH39NH3+NHZ (1) nmediate concern here, they showed that Ogg's 8 position, that only a trace of water was necessary rapid ammonia exchange and resultant triplet col— 5 incorrect since the system consisting of water in liquid ammonia can easily be buffered with adiene and cyclopentadienide ion to produce a par— llapsed ammonia triplet and a separate sharp water sonance. water must then be a weaker acid than adiene, and Ogg's suggested rate constant for the H20 + NH3 3 0H“ + NH: (2) ect by several orders of magnitude. This is fur- orted by Alei and Florin (12), who showed that the mmonia triplet is only partially collapsed with r present and that separate water and ammonia sig— possible even up to a concentration of 17 M water. that special care is taken to remove any NHZ Alei and Florin suggest that NH: may be formed reaction of ammonia with protons in the glass They showed that ammonium ion concentrations from 0-2 M are sufficient to cause rapid exchange so an average broadened single line for the H20 and n resonances is produced. J. Swift‘s; al., using careful chemical ‘— ll 9 techniques and a more accurate mathematical method of deter— mining proton line shapes, have reported a more precise value of the rate constant for amide-ammonia exchange (13): §H§+m3sgfl3+m§ (3) as well as for the ammonium-ammonia exchange (14 and 15) given by Equation (1). Perhaps it is worthwhile at this point to mention the suggestion of Powles and Strange (16) that pulsed radio frequency or spin-echo methods be used to study the proton exchange process. They have shown that for dry liquid ammonia a simple echo decay pattern, without modulation from spin—coupling with the 14N nucleus, is obtained for the protons. Thus, in a system containing amide or ammon— ium ion, the exchange effects on the echo pattern would be quite straightforward and tractable. Also, whereas in the steady-state method the experimenter has at his dis— posal only the line shape as a function of concentration and temperature, in the spin-echo method the experimenter has these and the advantage of another variable, the pulse spacing, plus freedom from heteronuclear coupling with the 14N nucleus which is a nuisance in the steady-state method. The reason for mentioning the spin—echo method is that although it has an obvious superiority over the :3 . II" '1'! i . xi , arrifi~ I3 LNWQJI.V. . i. 5 . i a. . . . .4l'u ‘ , . «v. Hui(1|., illnlmi.((l,. 10 ate method for studying ammonia exchange, and it was mentioned in 1962, it has not yet been Salts was mentioned before, the first reported work on salts was by Gutowsky gt aly, (5,6) who observed ra of aqueous ammonia or ammonium salts at various tions, but with no regard to pH. Under these 3 only a single proton resonance is obtained a composite of those for several species and is act to exchange averaging of the chemical shifts. and Saika (5) gave a theory to support their in— Lon and made a rough estimate of the proton life— porting 10'4 sec or less. Ogg (4) was the first 1at the exchange rates between NH4+ and solvent be decreased by acidification so that separate sonances were observable for ammonium ion and ant water. McConnell and Thompson (17) gave a etchy theoretical justification for Ogg's obser— 1 the case of acidified aqueous ammonium ion, but sented all the details. The first detailed study 1m ion in aqueous solution was by Meiboom §t_g1; 3y carefully changing concentrations and pH, were ll able to obtain kinetic rate data from changes in the spectra. They found that the kinetics could be accounted for by the following mechanisms: + k4 + R3NH + H20 .74 R3N + H30 (4) + k-4 ( ) - 5 + + R3.IiH + NR3 kg R3_N_ + HNR3 (6) + H H + R3NH + OH + NR3 k1 R3N + H0 + HNR3 (7) where, for the ammonium ion, R-H. Under their conditions, 1.5prf2.5, Reaction (5) is negligible and k4 of (4) can only be estimated: k6 and k7 (of 6 and 7) were determined, but are not given here as the values have since been revised. Grunwald gt_a1;, (l9 and 20) by extending the pH range and by controlling the ionic strength of the solutions were able to determine accurately k4, k_4 and k5 and estab— lish that these were diffusion controlled. By means of deuterium exchange they were able to determine the mean lifetime of the ammoniadwater hydrogen bond and estimate its rate of rupture by both diffusion and rotation. Finally, Connor and Leewenstein (21) obtained rate con- stants as functions of temperature, and thus obtained acti- vation energies for Reactions (4) and (6). Most recently, Grunwald and Ku (22) report a new value for R7. Kinetics of ammonium—solvent proton exchange in a 12 'aqueous solvents has also been investigated. The 'oblem is that the exchange mechanisms are usually ted by ion-association, which is promoted by sol- : low dielectric constant. Grunwald.gt_al; (23 and studied the proton exchange of ammonium acetate in acetic acid and report that the proton is exchanged the ion pair and the acetic acid carboxyl group. ‘ to calculate a rate constant, values of the pin-spin coupling constant and the 14N spin-lattice on time were required. It is important to point their values for these were obtained at 0°C or d used in calculations of rate constants from Perhaps it should be mentioned that the exchange the case of ammonium chloride in glacial acetic too slow to measure. he studies by Alei and Florin (12). and by Swift (13, 14 and 15) of the ammonium ion-ammonia proton were mentioned above. As these studies were con- n liquid ammonia or solutions composed chiefly of they can be included with the studies of ammonium ange in non-aqueous solvents. As mentioned before. Florin presented data for ammonium iondwater- exchange in liquid ammonia, which they compared 13 ‘with the aqueous solution studies of Meiboom (18) and Grunwald (19 and 20). Swift and Clutter (l4 and 15) studied the ammonium ionsammonia exchange for ammonium halides added to liquid ammonia and also the ammonium iondwater-ammonia exchange. They showed that exchange is due to ammonium ion, which is generated from reaction with water, reacting with the solvent ammonia as given by Equation (2): thus they did not obtain specific rate constants for direct water-ammonia exchange. Methylgmmgnium Salts Turning now to the alkylammonium studies, let us consider methylamine and aqueous methylammonium ion. Kine- tic studies of this system by NMR (1,2) were the first on alkylammonium ions. Grunwald.g;,a1¢_(l,2) studied proton exchange in aqueous methylammonium chloride in a concen- tration range from 0.272 M to 4.47 M and pH range from 3 to 5. At low pH the exchange is slow and three sets of lines are apparent in the proton spectrum: the methyl hydrogens appear as a quartet from spin-spin coupling with the three ammonium ion protons, the solvent water hydro- gens yield a single sharp line, and the ammonium ion protons produce a large triplet, from coupling with 14N (I-l), each component being broadened by the 14N quadrupole 14 n. There is also a barely discernable coupling methyl hydrogens. At high pH exchange between amine is rapid so that only two lines are ob— ne from the methyl hydrogens (showing no spin because proton residence time on the nitrogen is ) and the other a composite of hydrogen signals e protons, ammonium ions, water and hydrogen ions. ulated that the mechanism for exchange followed 4-7, with R3 - CH3H2. Rate constants were de— from the change in shape of the methyl signal by experimental spectra with theoretically calculat- hapes. They showed that Reaction 4 is negligible is very small. Thus, the total rate of exchange to follow the equation: te [H+] H3W3 1 W=[H“'] [024“]; KAI-[H+] [CH3NH3+] . .k [n+1 skSKw-i- (k6+k7) KA [CH3NH3'I'] (8) ct of mean values of kllgyuzig [CH3NH3+] they ob- and (k6+k7) and found that k5 contributes only a nt to the overall rate since it is small. From ening of the water signal, an estimate of k7/(k6+k7) ained. The broad annnonium ion'UWplet 1y as an order-of-magnitude check on the overall since it is observable only up to pH = 4.0 and 15 at high methylammonium ion concentrations. It is worth-while at this place to depart from the discussion of kinetics and mention something in detail about the broad ammonium ion triplet. Grunwald gt a1; (1,2) readily admitted that the chief cause of broadening of the triplet components was spin—lattice relaxation by the quadrupolar nitrogen nucleus and they made an estimate of the 14N spin-lattice relaxation time of 0.02 sec (25). That relaxation was the chief cause of broadening was very clearly demonstrated by Ogg and Ray (26), who presented the spectrum of acidified aqueous methylammonium chloride en— riched 65% with the non-quadrupolar l5N isotope in which the -NH3+ resonance was a very sharp quartet from spin—spin coupling with the methyl hydrogens. Unfortunately, Ogg and Ray did not report the concentrations used, nor did they study the collapse of the --15NH3+ peak with changing pH, so that no direct comparison with the work of Grunwald (2) could be made. In summary, then, although it was realized and substantiated very early that the line shape for the ammonium proton resonance was chiefly governed by nitrogen spin-lattice relaxation, no study of the relaxation process itself was undertaken. Returning now to the discussion of exchange kinetics l6 ylammonium salts, recent work has chiefly added cation to the earlier studies (1 and 2) to obtain nsight into the mechanisms. Grunwald and co- (27) extended their earlier studies by obtaining ct values for k5 and k of processes (6) and (7). 7 stants were obtained from the methyl group line bserved over the range of concentrations of monium chloride from 1.7 M to 8.1 M, and of pH to 8.0. Using these data in conjunction with ge in shape of the water line, k6 and k7 were ed separately, and using viscosity data were extra- to the viscosity of pure water. The ammonium ion again was used only to confirm the mechanism in the from 2.5 to 4.0. In another study (20) the mean of the amine—water hydrogen bond was determined dies of deuterated species in very strongly acidic s. hanges in the line shapes of the methyl quartet i to obtain exchange rates 1;, temperature, while nonium ion and hydrogen ion concentrations were stant. From plots of these rates 1;, temperature, nd Loewenstein (21) obtained the activation energy combined process represented by k6 + k7, k4 and k5 all and thus neglected. They noted that the value 17 was surprisingly low (LE R:l6.0 cal/mole) and that within experimental error the ratio k6/'k7 was temperature indepen- dent. They observed that the ---NH3+ triplet sharpens with increasing temperature, and attributed this to the change in correlation time (TC) of the nitrogen which affects T1 for the 14N nucleus. [-1 Unlike ammonia and ammonium salts, discussed above, methylamine proton exchange was studied in a number of sol- vents. In general, the interaction of alkyl amines or alkyl-ammonium salts with solvents may be formulated as in the following reactions, which may occur concurrently or stepwise depending on the gegen ion and solvent properties: Ionization (ion-pairing): Ki + R3N 4-1-12: R3N ’ H82: R3NH-2‘ (9) Homo-ion association: +~ + — nRBNH‘Z eh (33m -z )n (10) Direct ion-pair solvent transfer: _ RH - + 21H + 113N1122 :3 Zl'HNR3 + H22 (11) Ion-dissociation: K d + - .A , + . ‘ R3NH-8 +np v. (R3NH Dx) + (2 DY) (12) .After dissociation all the usual proton exchange processes are possible [Equations (4) through (7)]. It has been pointed out by Bruckenstein (28) that base dissociation occurs in two major steps, Equations (9) and (12), especially 18 in the case of a weak base such as retnylanine in a solvent of low dielectric constant. Equations (10) and (11) describe additional processes occurring. Grunwald and Price (23 and 29) studied methylamine (0.02 M-0.5 M) in glacial acetic acid finding that it was largely converted to ion pairs, Equation (9), and that the exchange followed Equation (11). Kinetic rate data were obtained from the broadening of the carboxyl proton peaks, with which only the NH protons are exchanging, by means of spin-echo measurements. Quite interestingly, methylammonium chloride forms higher aggregates, Equation (10), in glacial acetic acid and proton exchange was too slow to measure. In addition, they noted that the line shape for the NH+ proton resonance,being governed by 14N spin-lattice relaxation, is concentration dependent. Thus, for glacial acetic acid solvent, the gegen ion determines which process controls the NH line shape. Proton exchange rates for methylammonium salts in the solvents methanol and t-butanol, were obtained by Cocivera and Grunwald (24, 30, 31, and 32). Methanol, with a very high dielectric constant, behaves very much like water with exchange described by Equations (6) and (7) accounting in large part for the overall rate (30 and 31). On the other hand, t—butanol induced quite noticeably 19 different results. Owing to its low dielectric constant the salts exist largely in the form of ion pairs, as evidenced by the change in NH chemical shift with gegen ion. As a result of the close proximity of the gegen ion, an asymmetric electric field exists around the nitrogen nucleus producing rapid l4N spin-lattice relaxa- pi tion and, consequently, the NH+ proton peak is not split into a triplet by 14N as it is in water and methanol solutions. Oddly enough, though, the NH proton peak is split into a quartet from spin—spin coupling with the methyl protons since exchange is slower than in water. The exchange process in t-butanol was found to follow a mechanism analogous to that in water and methanol (Equa— tion 7), and formulated as: kl + _ 2 HO + - OH - R3NH 2 + (t But)“ 4- gay—7 R3N + (PB/nth: + 2 H1133 (13) (for methylamine R3 - (CH3) H2 ) k7 for water was found to be about twenty-four times faster than k2‘ for t-butanol. Dimethylammonium Salts Of the alkyl amines studied by NMR, the dimethyl is the one to which the least attention has been devoted. The most likely reason for this is that in aqueous solu- tions its exchange mechanisms and corresponding rate constants are very similar to those of methylamine and both 20 of these are very different from trimethylamine. The kinetic analysis of protolysis of dimethylammonium ion in aqueous solution in the concentration range from 0.23 M to 4.68 M and the pH range between 3 and 5 was carried out by Loewenstein and Meiboom (33). They noted that k6 of Reaction (6) is measureably slower and k7 of Reaction (7) is measureably faster, although of the same order of magnitude, than for the methylammonium ion. Again k4 and k5 ‘were found to be small with k6 and k7 being the major contributors to the overall exchange rate. In addition, they noted that the NH resonance is a broader triplet than observed for the monomethyl derivative and ascribed l4N nucleus. this to increased quadrupole relaxation from the Exchange kinetics in non-aqueous solvents were studied only in alcohols. Although not stated specifically, it is implied by Cocivera and Grunwald (30 and 31) that in methanol the dimethylammonium ion behaves very much like the monomethylammonium ion, showing no measureable dif- ference in exchange rates from water. t—Butanol was the only other solvent investigated (31 and 32), and there is definitely a measureable difference from the aqueous solution value. As mentioned before, t—butanol promotes ion pairing and the rate of proton exchange was found to vary with gegen ion. Studies with added tetraethylammonium 21 salts (32), gave a significant'decrease in rate which was attributed to increased solution viscosity and to the for— mation of less reactive aggregates. Trimethylammgnium Salts There has been much more attention devoted to the proton exchange kinetics of trimethyl— and monomethyl- ammonium salts than any others. Quite typically, trimethyl- ammonium was studied firstin aqueous solutions as reported by Loewenstein and Meiboom (33). Trimethylammonium chloride was studied in the concentration range from 0.26 to 2.3 Ni and aqueous pH range from 3 to 5. Exchange reactions given by Equations (4) through (7) were considered and, as before, k5 could not be measured experimentally, only an upper limit being assigned. The rate constants k7, k6' and k4 'were measured, as they had been for the mono- and dimethyl salts, and k7 was found to be lowest for the trimethyl compound. Unlike the mono- and dimethylammonium ions k6 proved almost unmeasurable for trimethylammonium ion, whereas k4 became measurable for the first time. No correlation of the pKA's was found with any of the rate constants for the mono-, di- and trimethylammonium series. In a later study, Grunwald (34) obtained proton exchange rate data for aqueous trimethylammonium chloride 22 in dilute solution (0.15 M 5 conc. s 0.65 M) and at low pH values (1.58—3.32). Activation parameters were ob- tained from rate data in the temperature range 35° to O 80 . Over this temperature range k6 a 0.1 k and k4 was 7 shown to be diffusion controlled. In his most recent study Grunwald (22) has aimed at getting a better insight into the mechanism governing k6' Equation (6). This was accomplished by varying the ammonium ion (R3NH+, R-CH3 or H) and the amine (R3N, R-CH or H), and measuring the rate 3 constants. From this study it was hypothesized that Reaction (6) can be broken up into three steps and that the rate constant k6 is governed by the number of methyl groups. Additional methyl groups reduce the speed at which the am- monium ion and the amine molecule become nearest neighbors. Grunwald.gt,al. (20), by studies of exchange rates of deuterated trimethylammonium ion in sulfuric acid-water solutions.determined the mean lifetime of the hydrogen bond between water and trimethylamine. This proves to be longer in the case of trimethylamine than for ammonia or monomethylamine. They presented two alternative hypotheses to explain this, but their data could support either. These rates depend on k4, Reaction (4), and it was noted that at high acid concentrations, pH < 2, this rate constant decreases markedly. Thus, Grunwald (35) showed that k4 23 can be subdivided into a two-step mechanism in which the first step is the equilibrium described by KA' Also, the acid dissociation in aQueous media precedes in such a way that the connection to the previously covalent proton remains intact for a measurable period. In a recent and more detailed investigation (36), it was shown that the rate of breaking the trimethylamine-water hydrogen bond measures the rate of diffusion of the water molecule from the amine into bulk solvent. Using the deuterated ammonium salt(0.44lm i [(CH3)3ND+] IA 1.76 M) in D20 with deuterium chloride (.08 M ‘5 [0+] 5 5 x 10-5) Day and Reilley (37) determined rate constants for deuterium exchange by fitting the methyl line shapes. They found that transfer rates for deuterium between amine and am— monium ion fit the mechanisms advanced by Grunwald gt al. for the corresponding proton transfer rates. 0n the other hand, Fraenkel and Asahi (38) measured exchange rates in strong acid solution (concentration of D2804 from 31% to 81% in D20) by mixing quantities of trimethylammonium + nitrate ( (m3)3NH N03“) and integrating the methyl signals with time. .Measuring the rates over the temperature range from 40° — 103°C yielded values for the activation parameters. Their rates and mechanisms for strong acid solutions agree with those of Grunwald gt il- (20, 34, and 35). 24 The order of the reaction with respect to water, Equation (7) , was shown to be one from a proton spin-echo study of the water resonance line in buffered trimethylamine-trimethylammonium salt solutions using water enriched with 17O at various concentrations . Using the appropriate mathematical relationship to treat the data, Luz and Meiboom (39) showed that one water molecule is involved in Reaction (6) . Although the studies carried out in deuterium oxide and deuterated sulfuric acid which were just discussed might be categorized as being performed in a non-aqueous system proton exchange involving trimethylamine in more conventional non-aqueous solvents will now be considered. A study of proton exchange in the system trimethylamine-glacial acetic acid has been reported by Grunwald and Price (23) . In glacial acetic acid, trimethylamine was found to first react forming ion-pairs, Equation (9) , and then to exchange by means of transfer of acetic acid between the amine, the solvent shell of the amine, and the bulk solvent, Equation (11) . They found that exchange rates for trimethylammonium ion were slower than for ammonium ion but faster than for monomethylammonium 1011- Concentrations of trimethylammonium acetate ranged from 0.0207 to 0.1520 M and a temperature interval from 15° to 750 was used . 25 Interestingly, they noted that the rate of exchange for trimethylammonium choride in glacial acetic acid is negligibly slow, and because of this they obtained a "moderately precise“ (130%) value for the 14N relaxation time T1‘ From T1 for the chloride salt they estimated T1 for the acetate salt, since this value is necessary to describe the broadening of the carboxyl proton line shape, from which the exchange rates were determined. Unlike menomethyl- and dimethylammonium salts, the trimethylammonium salt in methanol yielded proton exchange rates measunflfly different from those in water (24, 30 and 31). In a recent detailed report Grunwald (40) shows that the kinetics follow two mechanistic paths, the first being acid dissociation, an equilibrium analogous to Equation (4). The second path of exchange involves the solvent and is analogous to Equation (7). Concentra- tions of trimethylammonium salt ranged from 0.001 M to 0.55 M and HCl from zero (excess amine) to 0.51 M. The kinetic data indicate that in the buffered system (excess amine) the second path plays a major role in the overall rate, whereas in the presence of excess HCl the second path is not the dominant reaction, the acid dissociation being 14 more important. He also noted that the NH spin-spin interaction is well resolved at 50°. t—Butanol, owing to 2 0“ its low dielectric constant, causes trimethylanmouium salts to exist predominantly as ion pairs, Equation (9)- This is evidenced most obviously by the dependence of the NH pro— ton chemical shift on gegen ion, a difference of 0.78 prnn betvveeri trifluoroacetate and tosylate being noted (31). Exchange rates were also dependent on the gegen ion and could be attributed to the process described by Equa- tion (13)(24, 30 and 31). The exchange rate for the to- sylate was seven times faster than that for the chloride in t—butanol, and comparing k7 for water (Equation 7) to k'2 for t-butanol, it was 560 times faster than the tosyl— ate and 3500 faster than the chloride. Unlike mono- and dimethylammonium salts, addition of quaternary alkylammon- ium salts to the solutions of trimethylammonium salts caused no major change in the exchange rates because the formation of aggregates greater than pairs (i.g. n > 1 in Equation 10) is not possible (32). Cocivera noted (31) that the NH proton peaks for the trimethylammonium salts in t-butanol exhibit fine struc— ture not observed in aQueous solutions. This is a result of ion pairing: the closeness of the gegen ion generates an electric field which causes rapid relaxation of the nitrogen nucleus and thus prevents any l4N coupling, 'while at the same time exchange is slow so that the alkyl protons can couple with the NH proton. It was pointed 27 out before that a similar effect was found for mono- methylammonium salts in t—butanol. other Alkylgmmonigm Sglts The only other simple alkyl amine studied in detail is triethylamine, reported by Ralph and Grunwald (41). In the pH range 5-8 the kinetically significant reactions are given by EQuations (5) and (7), whereas at pH < 1 they are those given by (4) and (9). Although not stated specifically, direct transfer of protons to the base must be insignificant since no mention of anything similar to Process (6) is made. Comparing these rates to the methylamines, k7 for triethylamine is much smaller than expected and this is attributed to steric hindrance from the bulky ethyl groups. At low pH, where exchange is slow, the deuterated triethylammonium ion in water was studied to determine kH of Reaction 11. Grunwald and Ralph (36) considered the series of amines (ammonia, monomethyl-, and trimethyl-, and triethylamine) but found no correlation between kH and basicity, KB' but a systematic decrease of kH with number and size of alkyl substituents. Two additional points should be mentioned. First, exchange rates were determined by spin—echo measurements of the dominant solvent water line, i,g. 28 changing exchange rates causes broadening of the water resonance. The second point is that parameters necessary for treating the spin-echo data are the 14N-H spin—spin coupling value and the 14N spin-lattice relaxation time. The values used in the triethylammonium study (41) were not actually measured, but were estimated from those for trimethylammonium ion. This procedure was justified on the basis that the kinetic results are not sensitive to small errors in these parameters. Reviewing the results of a number of workers over a period of more than a decade, Grunwald, in a recent publication (22), has pointed out some general relationships. The effects of alkyl substitution on the various rate constants in water are quite different. Thus, k7 for Equation 7 changes very little, being the same order of magnitude for all. Although k7 showed no correlation with pKa (33). a measure of the proton donating capacity of the alkyl ammonium ion, there is a correlation with KB, the base dissociation constant, and with the number of methyl substituents, the triethylamine being an exception (41). For both k6 (Equation 6) and RH (Equation 11) there is a change of two orders of magnitude, and a pro- gressive decrease in rate with increasing number and size of alkyl substituents, on going from ammonia to 29 triethylamine (22, 36). These kinetic results, then, lead to speculation as to the reason for lack of correlation. One explanation given by Brown (42). which seems very likely;is that the decrease of kH and k and the lack 6! of correlation of k7 in all cases, is due to increased steric Jhindrance or strain with increasing number and size of alkyl groups. Other Nitrogen Compgunds In concluding the review of kinetic work, it is appropriate to point out the work on alkylammonium salts by Swain and coaworkers (43, 44, and 45), who used tracer techniques. Although they did not report actual values for the rate constants, they nevertheless did obtain relative rates for the ammonium ion,the series of tethyl-substituted ammonium ions,and aniline. They also pointed out the chief mechanisms for exchange, which are the same as those mentioned above. They also studied exchange in acetic acid and methanol, as well as a number of solvents which have not as yet been employed in the NMR method, including dimethylformamide, toluene, ethylene glycol, and formic acid. While Swain g; al., did not obtain the detailed information that the NMR investigators did, their work predated (43) the first NMR reports (1 30 and 2) and pointed the way for that work as well as for the later studies in other solvents (23, 24, 29, 30, 31, and 32) because their reports showed that exchange rates were in the correct range to be observed by the NMR spectroscopic technique. While it is true that the methylamine series discussed above has been studied most extensively, and in far greater detail than all other amines, one should not conclude that these are the only amines for which proton exchange between amine and solvent or between amine and ammonium ion have been studied by the NMR tech- niques. For example, proten exchange between glacial acetic acid and tris-(hydroxymethyl)methylammonium acetate has been studied (23). Steinblatt investigated sites and mechanisms of pretonation for glycylglycine (46) and triglycine (47), reporting the rates of exchange in aqueous solutions. Grunwald and Cocivera (24) considered p—toluidinium ion in methanol solution, and Grunwald and Puar (48) reported exchange rates and mechanisms for N,N-dia1kylanilinium tosylates (dialkyl = dimethyl, diethyl, and di—n-propyl) in acetic acid. The dissocia— tion constant of the dibenzylmethylamine—water hydrogen bond has been found (36). Although not measuring specific rates, Alei and Florin (12) have observed that the 31 exchange of water with the amine protons of ethylenediamine is so rapid over the entire range of composition of water- ethylenediamine solutions that steady-state methods cannot be employed, but presumably one could use the spin—echo method (23, 29, and 48). Most recently, the steady—state method has been employed by Sudmeier and Occupati (49) to study proton exchange in aqueous N,N'-dimethy1piperazine hydrochloride. All the compounds and systems discussed above have been illustrative of the types of compounds, other than simple alkylamines, which have been studied by NMR. No attempt at a complete review of all proton ex- change research has been made, for this would be placing undue emphasis on that aspect of the NMR work. B. Nu 1 R ion and u d Effects General Theory of Nuclear Relaxation Nuclei posssessing a magnetic moment, when placed in a static magnetic £191d (Ho), tend to be aligned in specific orientations with respect to this field. The natural tendency is for the nuclei to be aligned parallel to the static field, for that is the condition of minimum poten- tial energy. However, as a result of random molecular 32 translational and rotational motion (i.g. Brownian motion) at room temperature, each nucleus will experience a rapidly fluctuating magnetic field produced by neighboring nuclear magnetic moments. Since the energy produced by these fluc- tuations far exceeds the energy difference between parallel and antiparallel alignments, there is only a small excess of nuclei in the lowest energy state. For simplicity, consider a set of nuclei with a spin quantum number of 1/2. These nuclei have a symmetrical charge distribution which is spherical and possess a magnetic dipole moment. In the absence of a magnetic.field there will be equal populations in the two spin states but if these nuclei are suddenly placed in a strong static magnetic field (Ho), one is then concerned with how long it will take for the populations of these nuclear spin states to reach equilibrium as described above, that is, with a small excess number in the lower state. After equilibrium is reached, if a strong radiofrequency field of frequency w is applied in a direction perpendicular to the static field, transitions between the two levels will occur as energy is absorbed, provided that the Larmor precession condition (14) is met ($.51. a) ' so): = 7H (14) where 7 is the nuclear gyromagnetic ratio. If the 33 radiation field at the precession frequency is sufficiently strong, transitions to the upper state will occur until the two states have equal populations. One is then concerned with how much time must elapse before the equilibrium num- ber in the lower level is re-established, once the strong radiation is removed. Both this latter situation and the former one involve the approach to equilibrium between the nuclear spins and the environment (or the "lattice") by means of energy exchange with thermally induced fluctua— tions, after a stress is either introduced or removed. This establishment of equilibrium, then, is termed relaxa— tion and the characteristic time for all but l/e of the excess spins in the upper state to return to the lower energy level is called the spin—lattice, thermal or longi— tudinal relaxation time (50 and 51). As Sillescu (52) has mentioned, the longitudinal or thermal relaxation time (T1) was first introduced by Bloch (53) as an empirical constant in his famous phenomenologi— cal equations. Bloembergen gt al. (54 and 55) showed that a reasonably good value of the spin—lattice relaxation time could be obtained from a time-dependent perturbation treat- ment of a suitable set of fluctuation functions describing the Brownian motion. Unfortunately, this work was riddled with errors as pointed out by Kubo and Tomita (56). Andrew 34 (57) presents a clearer outline of the important features of Bloembergen's derivation. The best treatments of relax- ation in general are presented by Abragam (58) and Slichter (60). These presentations give the most details, the clear— est explanations and methods for the most general applica- tions. Let us proceed, then, to consider the important features of the theoretical approach to spin-lattice relax- ation time. It should be re-emphasized that the following dis— cussion is limited to the case of nuclei with a spin quan— tum number of 1/2, and as a result of their spherical charge distribution only dipolar interactions are impor- tant. In addition, it is assumed that the dipolar coupling between nuclei is weaker than the coupling with the lattice. This is the characteristic condition of nuclei in the liquid state and thus the chief mechanism for relaxation is Brownian motion. The situation is different in solids since internuclear dipolar coupling is much stronger in the rigid lattice (51 and 60) than in liquids. In any form of spectroscopy, the intensity of absorp- tion or emission of electromagnetic radiation is related to the rate of transitions between energy levels, which is in turn related to the probability of these transitions occur— ring, For an emission process, the lifetime in an excited 35 state is then of concern (61). This is true also with NMR; the major distinction resulting from the relatively small differences in energy between nuclear spin states. Whereas with other forms of absorption spectroscopy, rela- tive populations of upper levels are small, and emission is unimportant, with NMR emission is quite important because of the very small differences in energy between upper and lower levels and the nearly equal populations of these levels. Thus, for NMR the absorption spectrum is closely related to the lifetimes of the excited states and the time necessary to reach the equilibrium populations of upper and lower levels is of prime concern. Pake (50), Andrew (51), and Slichter (60) have shown that the rate of change of the excess number of nuclei approaching thermal equilibrium with the lattice is an exponential function. The spin-lattice relaxation time (T1), then, is the time constant of this exponential function or the characteristic time in which the spin—system exponen- tially approaches the temperature of the lattice. It is shown that there is a simple relation between the spin-lat- tice relaxation time and W, the mean value of the two tran- sition probabilities, from lower to upper and from upper to lower levels: T1 3 1/2 W (1.5) 36 are, the main problem of any nuclear magnetic relax- :heory centers on a method of computing W. The first attempt at calculating a value for the :ansition probability was made by Bloembergen gt gl. 1 55). Their final values of T1 were quite close to measured experimentally, although their general an was in error having incorrect constant coeffi— Kubo and Tomita (S6) re-examined this work ard 1 some of the equations. The correct result, also 3y Andrew (57) is: w a 3/4 74 h2 I (I + 1) [J1 (mo) + 1/2 J2 (2%)] (15) ', h,and I have their usual meaning and J (m) is the . density of the random functions or the intensity of trier spectrum of the fluctuation functions of the >n co-ordinates (iqg., functions which describe the in motion). J (o) is obtained as the Fourier trans- E the correlation function and therefore is: +oo J (m) - S-oo G(T) exp (im) d7 (17) 'relation function, G(7), of the fluctuation func- ,s so named because it tells how these functions at 1e t are correlated to their value at a later time '). Finally, the functions in Equation (17) should ;he same subscripts as those of Equation (16). These .pts denote which of the three fluctuation functions, 37 ' 2, are capable of inducing transitions in a neigh— nucleusy from Equation (16) function 1 at frequency 2 at Zmo, are able to do so, but functiow 0 is non. At the outset this discussion was limited to liquids .clei of spin one-half, whereas Equation (16) has had :rictions placed on I. This seemingly contradictory ,on does indeed have very major limitations when I > ‘irst, the population distribution among adjacent levels must vary exponentially so that a spin tem- 'e may be defined. Second, as will be given in more shortly, certain symmetry requirements must be ful- so that only nuclear dipolar interactions are ve. Finally, in considering the case of nuclei with spin in the liquid state, one should examine another rh presented by Abragam (58) and Slichter (60). They technique of the density matrix which is more ly applicable. There are a number of advantages to iproachx it does not require the concept of a spin ture, and thus is very useful when this concept down, and it is ideally suited for situations in ne resonance is narrowed by the motion of nuclei as ids. It alSC has the advantage that both Tl and 38 icesses are handled in a natural way. The density matrix method of Wangsness and Bloch (59) more qualitative theory of Bloembergen, Purcell and 'ield the same result, namely that the spin-lattice ion time is related to the transition probability alue is determined by the spectral density function, Both methods give the same expression for the 2 1(a) oc ZTC/(1+u,° TCZ) (18) most cases the constant of proportionality is one. relation time, TC, is determined by the random n motion. It is of the order of the time necessary olecule to turn through a radian or to move through nce comparable with its dimensions so that the rela— sitions of the nuclei with respect to the external nd thus the fluctuation functions have changed ably. There have been a number of suggestions as to calculate this correlation time. From methods listed Evans and Richards review (62), as well as from the 1 form proposed by Bloembergen gt, 1. (54). one may I— '— *T2 is the spin-spin or transverse relaxation time, re of the time necessary to destroy phase coherence precessing components of the transverse nuclear c moments. 39 :he general equation: . TC - c WM? (19) 7 is the solution viscosity, k is the Boltzmann con— Lnd T is the absolute temperature. Alternatively. rwrite 67k ' c so that we - c’7/T (20) .5 Equation 6 of the treatment given by Arnold and (63). :ion for N i 0 Mo nt Now we turn to the more complex case, nuclei with a lantum number greater than one-half (I>1/2), which l asymmetrical (non-spherical) nuclear charge dis- .on and possess an electric quadrupole moment. Spin- ) relaxation may result from this quadrupole moment :ting with fluctuating electric field gradients pro- .t the nucleus by various molecular degrees of free- 'his additional mechanism contributes to spin—lattice ion and consequent broadening of signals. When there symmetric molecular electric field about the quad- ' nucleus, the quadrupole coupling may be quite nt. For the special case of spin I - l (qu.l4N) Abragam 5 shown how the quadrupolar effect can be calculated ; to a term in the expression for T1, the spin-lattice 40 relaxation time: 1/T1 - 3/80 (1+a2/3) [(eQ/h) (azv/azz)] {3 (mo) + 4 3 (Zub)} (21) where a is the asymmetry parameter, Q the scalar quadrupolar moment, 83V/322 is the electric field gradient at the nucleus, and 3 (ufiis the Fourier transform of the reduced correlation functiony its constant of proportionality is unity here and from Equation (18) we now have: 3 (w) .- 2Tc/(1+ub2 rcz) (22) From the values of the nitrogen quadrupole coupling constants and the nitrogen relaxation times given by Kemp gt 1;. (64), one finds values of the correlation time of 3x10’11 sec for both of the two extremes, methyl isocyanide with a symmetrical electric field gradient around nitrogen and methyl cyanide with an asymmetrical electric field gradient around nitrogen. This is clear experimental evi— dence that spin-lattice relaxation time for a quadrupolar nucleus is a major function of the field gradient since the correlation times for both extremes are the same and the T1 values differ only because the electric field gradients differ. This value for the correlation time used in con— junction with a value of 3.5x106 rad sec.1 for the nitrogen Larmor frequency leads to wo TC < l/2). Nuclear Relaxation Times of Nitrogen Compounds A review of some results which have been reported for nitrogen compounds will now be given. Considering 44 first molecules with symmetrical electrical fields, the quadrupolar nucleus will have a long spin-lattice relaxa— tion time and consequently spin-spin coupling will be ob- served in the spectrum of the low spin nucleus (I = 1/2). This review will be limited to proton (I a 1/2) spectra of nitrogen (I = 1) compounds since this thesis is concerned with alkyl amines. A 14N spin-lattice relaxation time of greater than one second was estimated from the proton spectrum of aqueous tetramethylammonium ion by Grunwald gt 5;. (25). Independ- ently, the same observation of ll-l-14N coupling for the tetramethyl- as well as the tetraethylammonium ions was made by Hertz and Spalthoff (70), although they gave no explana- tion for the conditions of its existence. Shortly there- after, Anderson._§‘_l. (71) and Bullock g£_gl. (72) reported the observation of 1H-14N coupling in the tetraethylammonium ion noting that it was independent of anion and solvent, but dependent on symmetry around nitrogen since the coupling was absent when one of the ethyls was replaced with some other alkyl group or when the quaternary structure was non- existent, as in the free amine. That the coupling was def- initely with nitrogen was proved by the heteronuclear dou- ble resonance (14N-decoupling) experiment of Anderson gt al. (71). Neumann and Lehn (74) estimated the 14N spin-lattice 45 relaxation time for quaternary ammonium ions to be about ().w2 sec although they later (73) reported a lower limit of .314 nulliseconds (msec) for 14N in aqueous tetraethylammonium larwomide. In an interesting study (73) they estimated T1(14N) + fcar two symmetrical compounds of the series [Et N-(CHZ)- 3 + .. T1 (14N) n 31 msec 6 29 5 Although they did not estimate T1 (14N) , Gassman and Heckert (75) made use of the fact that it is long for quaternary anunonium salts to measure 1H—14N coupling in ten alicyclic alJtyl ammonium and seven alicyclic ammonium salts. Another type of quaternary salt having long 14N relaxation times is tile 4-substituted ethylpyridinium salts reported by Biell- Huann and Callot (78). The 14N relaxation times estimated Eire: T1 (14N) R 0.21 sec -CN + , R ON-.. 0.22 —CF 3 A second class of compounds exhibiting long nitrogen Spin-lattice relaxation times is the isonitriles or iso- cyanides. Kuntz 3; al. (85) estimated Tl (14N) = 0.3 sec e.' 46 from the proton spectra of five compounds using Pople's (9) lineshapes, whereas Kemp 95 al. (64) report Tl (14N) =- 0.35 =I=O .1 sec for CH3—NC from spin-echo measurements. Recently Lamberton gt_al, (86) reported yet another class of compounds for which the electric field gradient around nitrogen is symmetrical enough so that relaxation tiJnes are long and 1H-14N coupling is observed. This class is the dialkylnitramines and the unusual feature is that tflie'nitrogen with the long T1 is in the nitro group rather than the amino group. This was suggested by the magnitude cxf the proton-nitrogen coupling constant. It has been con— firmed by heteronuclear decoupling (87) , the optimum l4N decoupling frequency being in the chemical shift range of Ititro groups rather than amino groups. The second group of relaxation times includes the iJltermediate range where the rate of 14N spin-lattice relax- Eition is beginning to cause coalescence of the proton spec- tIrumtriplet arising from spin—coupling with 14N. For this intzermediate range of relaxation times, from values of T1 ‘flhere the components of the triplet are beginning to broaden to values where they are coalescing, the electric field gradient is much less symmetrical provided the correlation times are all about the same, as they should be for fluid liquids. The first report of a 14N relaxation time for 47 ‘tlais category was by Grunwald gt g1. (25) who estimated tr]_ - 20 msec for aqueous methylammonium chloride. For dry liquid ammonia Pople (9) estimated 141v T1 - 22.2 msec from ‘tlie line widths but noted that this might be slightly high. Irideed, Anderson and Baldeschwieler (88) report a value of 15 .9 msec and the data reported by Swift gt al. (13) essen- tnially agree with this. Lehn and Neumann (73) show the ssmectra of several comounds in a homologous series and spe- czifically the tetramethylene one, Et3N-(CH2)4-§Et3 2 BE, exhibits the characteristic lineshape for this category. They estimate a T1 value of 23 msec. Similarly Biellmann arms Callot (78) estimate the following relaxation times for tfliose 4-substituted ethylpyridinium ions which come in this iJitermediate range of T1 values: T1 (14N) R .15 sec -COOEt + .09 -CONH2 '3 R T1(14N)‘22.78 I“Secwith a smaller error of r 10%. For the same solvent, glacial acetic acid, they reported (23) the following values of T1, noting that due to ion-pairing some are very concen- tration dependent: 49 ‘1‘ 1 (14N) Compound Conditions Conc. Depen— dence 9 - 5 msec NH4OAc 0.11 M < conc. < 0.74 indep. l - 1(:t30%) Me3NHC1 0.6 < conc. < 6 M dep. 1 -5 Me3NHOAc (estimated from the - chloride) 0 . 14 (HOCHZ) 3CNH3OAc 0.9 M _ It: should be noted that these values are estimated from the line shapes and that there may be sizable errors. Similar- 1y. Ralph and Grunwald (41) reported an estimated value of 2 .5 msec for triethylammonium chloride in water. For the remaining compounds in the series, Et3N-(CHz)n -§Et3 Lehn and Neumann (73) report values of 12 (n =- 3) and 4.5 (n =- 2) msec. Similarly for the 4-substituted quaternary ethyl— PYridinium salts, Biellmann and Callot (78) estimated val— ues 01‘- T1 (JAN) ranging from 50 msec to < 10 msec. In view Of other reported estimates and measurements of relaxation times for compounds with asymmetric field gradients, these values seem a little long. Indeed, Kawazoe g; a_l_. (79) eStimate from proton line shapes that T1 < 20 msec for a Ilumber of quaternary alkylammonium salts. Kintzinger and Lehn (65) estimated T1 values of the same order of magnitude, 0.18 msec 14 < Tl < 4.55 msec, from the N line widths of five hetero- cyclic compounds. Finally, for all the remaining compounds 50 reported by Moniz and Gutowsky (66) the values of T1 range from 0.8 msec to 5.8 msec, these being obtained from 14N spin-echo measurements. In addition to these, values of 4- 9e0.8 msec for acetonitrile (64, 89 and 90) and l.24:t0.07 msec for 2-fluoropyridine (89 and 90) have been reported. The various values can now be summarized. When 14N F is in a symmetrical electric field, the 14N spin-lattice relaxation values range from 0.5 sec down to 0.2 sec, Grunwald's estimate (25) that Tl should be longer than a second being much too long. When 14N is in an asymmetric electric field the range is 12 msec to 0.2 msec and values for electric fields intermediate between these two ex- tremes run from 90 msec down to about 15 or 10 msec. Im— Plicit in all these speculations about the field gradients and relaxation times is the assumption that the correlation times are of the same order of magnitude throughout. viscosity and TemperatugEffects on T; gfjugdrupolar N a \uclei We shall now consider the factors influencing thecorrela- tion time and the effect of its variation on the nuclear relaxation times. As Equation (20) shows, the correlation time is directly proportional to the viscosity of the sam— Ple system and inversely proportional to the temperature. Thus, there are two possibilities: first, for isothermal 51 conditions increasing the viscosity raises the correlation time and according to Equation (24) this results in a fast- er relaxation rate and thus a shorter relaxation time. A plot of 1/T1 y_s_. ’7 (Form I) should yield a straight line with a positive slope of 3/8[(eQ/h) (aZV/aZZHZ C/T. A treatment of this type we shall call Form I. The second possibility is to increase the temperature holding viscosity constant which would lower the relaxation time since the correlation time, Equation (20), is inversely proportional to temperature. However, in- creasing temperature also lowers the viscosity which low- ers the correlation time and, therefore, both will operate in conjunction. According to Equation (24) , a lower cor- relation time means a slower relaxation rate and thus a longer relaxation time. A plot of l/Tl lg W/T (Form II) also yields a straight line with a positive slope of 3/8[(eQ/h) (62V/322)]2C. This type of treatment we shall call F'carm II. Evans and Richards (62) reported one of the first studies of the behavior of nitrogen relaxation times with Viscosity, following Form I. They used the 14N magnetic resonance line widths as a measure of relaxation rates 52 and, indeed, the plots of these values w solu~ tion viscosities yield the predicted straight lines. From the slopes, using an average value for the quadrupole coupling constant from microwave data, they were able to obtain values for C for three alkylcyanides. These agree very well with the calculated ones. In a study of the 115:m resonance in indium perchlorate and sulfate solu- tions Cannon and Richards (91) showed that, despite the larger quadrupole moments, these still follow Form I. Grunwald and Price showed that there was a viscosity effect of Form I on the 14N relaxation time for methylam— monium chloride (29) , ammonium acetate, trimethylammonium acetate, and trimethylammonium chloride (23) in glacial acetic acid. Recently, Kintzinger and Lehn (65) , in a qualitative fashion, reported data following both Forms I and II for the nitrogen and hydrogen NMR spectra of five heterocyclic compounds. Studying compounds of other quadrupolar nuclei, Massey _e_t_ al. (81) report qualitatively effects of Form I for quaternary alkyl salts of 75As and 1Zle. They caution that solution viscosity should not be the only criterion of relaxation rate when considering electrolytes in a series 0f solvents, but that solute—solvent and ion—pairing are also very important because they may induce changes in the - ..-.—- w v 53 electric field gradient about the quadrupolar nucleus (82) . Arnold and Packer (92) studied relaxation times as a function of viscosity for the hexafluoroarsenate salts of potassium and a few other cations. 75As relaxation times were obtained from changes in the 19? resonance line shapes, plots of Form I were linear for various solvents and concentrations. Although the results were not conclu- sive, they supported the contribution of ion—ion inter- actions in the relaxation process. In a more recent study (93) of solutions of potassium, silver, and tetra—n-butyl- hexafluoroarsenate salts in acetonitrile and acetone, at treatment of Form II was made. They found that the relaxa- tion process was due to changes in the electric field gra- dient and was short ranged being a function of the cation. They noted that the plots, although linear, did not follow the predictions of Form II, since they intercepted the y— EXis above the origin. They suggested that the slope is cOncentration dependent at lower concentrations, but did not speculate as to the cause. Arnold and Packer (63) have looked into the basic ElSsumption which so many have accepted, namely, that vis— cosity affects the correlation time and thereby the relax— ation rate. Since most of their previous work dealt with aqueous electrolyte solutions, quite naturally these were 54 the focus of their attention. Deviations from the corre— lation time-viscosity proportionality had been noted by others and Arnold and Packer concluded that such deviations were found when ions have a marked effect on the "structur- al equilibrium" of the water. They conclude that the rela— tionship holds, ing. is linear as in Form I and Form II, when the viscosity reflects changes in the local mobility or diffusion of the relaxing nucleus. Turning now to another series of studies, we shall consider those which deal with the temperature effect on the quadrupolar nuclear spin-lattice relaxation time or the effect this will have on the line shape of the resonance for the spin 1/2 nucleus, which is more easily observable. The first report of work along these lines was by Roberts (8), who pointed out that the broadening of a sharp proton singlet or the appearance of a triplet from a broad peak observed on increasing temperature in the case of ten nitrogen compounds was caused by an increase of 14N spin- 1attice relaxation time with temperature. He emphasized that this was unlike the temperature effect on exchange, since increasing temperature then causes increased ex- change rates which first collapse a triplet to a broad singlet and finally lead to narrowing of the single line. Since then, a number of others have observed this ‘ 55 temperature effect on nitrOgen-containing compounds so that the connection is well established nowo Anderson._t.al. (71) noted the effect of a 70° temperature change on (C2H5)3N-(CHZS CH3)+I- in water. At the high temperature, relaxation is sufficiently slow so that triplet structure 14 arising from coupling between N and the CH3 protons of the ethyl groups is observed. Similarly, Lehn and Neumann (73) reported 14 N triplet fine structure for the CH3 pro- tons of the ethyl group of (CZHS)3N-(CH2)4—N-(C2H5)§2 2 BE about 600 above ambient temperature. The data on liquid ammonia presented by Swift _t,§;, (13) clearly shows that the 14N relaxation time increases with increasing tempera- ture For a 650 increase in temperature, Biellmann and Callot (78) show that 14N triplet coupling of the CH3 for the ethyl becomes observable for N—ethylpyridinium ion and becomes much better resolved for N-ethyl-4— trifluoromech- ylpyridinium ion. Along these same lines, Kawazoe gt 1. l4N-methyl (79) presented spectra showing more pronounced proton coupling for tetraethylammonium bromide in chloro- form and for gigfz,6,N,N-tetramethylpiperidinium iodide in water for a 60° temperature increase. Kintzinger and Lehn (65) studied the temperature induced change in the proton spectra of the hydrogen u to the nitrogen in five heterocyclic compounds observing only broadening of the 56 2 line, but noting that this followed the usual relax- time—temperature relationship. The nitramines of 'ton.gt,a;, (86) also follow this relationship. They :ed that the CH3 signal of dimethylnitramine in tetra— >ethane is a sharp singlet at -20° and a sharp triplet 30°. In summary, an increase of spin-lattice relaxa- :ime with increasing temperature has been observed for a variety of compounds but is limited to those which .n asymmetric electrical environment (71). It is important to point out that the spin-lattice (tion time-temperature relationship is not limited to (t is general for all quadrupolar nuclei. To cite a :amples, it has been observed for 2D in D20 solutions in ECEtORE'dg and benzene-d6 (95): for 113 in liquid (rifluoride (83) and dimethoxy—dG-borane, (CD30)2 BH 35 for C1 in perchloryl fluoride (83) and in chloro— 97): and for 75As in potassium hexafluoroarsenate For the quadrupolar nuclei studied so far, increas— emperature always causes slower relaxation rates or ' relaxation times. The theory of the temperature effect will now be lered. From the basic Equation, (24), the relaxation .s a function of two variables, namely, the quadrupole .ng constant and the correlation time. The 57 lation time is inversely proportional to the temper— and directly proportional to the viscosity. Since asing temperature lowers the viscosity this will also te to decrease the correlation time. The question s whether the correlation time is the only thing af— d by temperature change or whether the quadrupole ing constant is also temperature dependent. Turning crowave and pure quadrupole resonance (NQR) spectros- by means of which one can measure the quadrupole ing constant directly, the situation is not clarified. wave data, obtained from studies of gases at low pres— give no support to temperature dependence, whereas ata, obtained from studies on crysralline solids, ates there is a measurable temperature dependence of uadrupole coupling constant (98 and 99). The differ— in the two may perhaps be due to the fact that micro- results refer to a single molecule, the field gradient r being related to molecular axes, whereas NOR data to the entire crystal, the field gradient tensor related to crystal axes. A change of quadrupole ing with temperature in the latter case may represent solid-state effects. Thus, the question in the case 1 studies of liquids and solutions is still unan— 3. Results presented so far seem to indicate that 58 there is indeed a definite temperature effect on the quad- ruple coupling constant for species in solution (63, 93, and 100). Aggiyatign Energies for the Relaxation Process For any thermally dependent process it is possible to obtain an activation energy so one may assume, as did Moniz and Gutowsky (66) . that the correlation time is described by an Arrhenius-type equation: TC . 72 exp (Ea/RT) (25) COmbining this with Equation (24) , taking logarithms of bOth sides and changing signs, one obtains: 109 T1 - -(Ba/2.303 RT)-log (3/8 (eZqQ/n>2.g> (26) A plot of log T1 is. 1/T should be a straight line of slope —Ea/2'303 R. The assumption is that only the correlation tirue is temperature dependent and that it follows the AI‘rhenius equation. However, even if the nitrogen quad— rupole coupling constant, QN, is also temperature depend— ent, one still obtains a straight line for a plot of log Tl Y§. l/T because it can be shown that: 109 T1 - —(2Ea(Q) + sa(c))/2.303 RT — 109(3/8(Q§)2 wg)(27) When, in addition to (25) , the quadrupole coupling is also Considered to be temperature dependent according to the relationship: 59 (equ/h) - 0N - of; exp(Ea(Q)/RT). (28) The only differences between Equations (26) and (27) are that, first, the activation energy obtained from the slope is the sum of the activation energies for quadrupole coupl- ing, 3am), and for correlation, Ea(C) , for the latter, whereas for the former it is simply the activation energy for the correlation process. The second difference between the two equations will be in the interpretation of the value obtained for the high temperature intercept. Thus. implicit in the assumption of Moniz and Gutowsky is the assumption that. Ea(C)>>Ea(Q) which results in, Ea = Ea(C) and ON " QB. Nevertheless, it is possible to obtain a Value for the activation energy of the relaxation process. Based on the assumption outlined above, Moniz and Gutowsky (66) obtained activation energies for nine nitro- gen compounds. The values range from 1.4 to 3.2 kcal/mole and both these and the differences among them were gener- ally compatible with predictions based on the relative sizes and shapes of the molecules. Hertz and Zeidler (94). from spin-echo studies of deuterium compounds, obtained an aCitivation energy of 3.5 kcal/mole for pure D20 and a range from 3.7 to 5.6 kcal/mole for the solvent D20 in six elec- trolyte solutions. Values of 1.71 0.2 kcal/mole for 6O CD3COCD3 and 1.5 t 0.2 keel/mole for C6D6 were reported by Bonera and Rigamonti (95), who obtained the deuterium T1 from the signal height after adiabatic rapid passage. Bacon _e_t 1;, (83) report a 3’5C1 activation energy of 1.0 kcal/mole for perchloryl fluoride and 1.4 cal/mole for 11 113 in boron trifluoride. Also, for B, the value 1.2 t 0.1 kcal/mole has been reported by Boden g_t_ l. (96) for dimethoxy (d6)-borane, (CD3O-)2 BH. Finally, Arnold and Packer (93) found a range of activation energies for 75 As from 2.09 to 4.10 kcal/mole in potassium hexafluoroar- senate at five different concentrations in diethyleneglycol- dimethyl ether (diglyme) . These reported values give some idea of the magnitude of the activation energy for the relaxation process regardless of the exact mechanism, which is not known at present. ‘93 lyent Effggts The last tOpic for consideration is the effect of Variation of solvent on the spin-lattice relaxation of a Quadrupolar nucleus. It may be again pointed out that Changing solvent, and thus varying the viscosity, causes a change in : and, according to Equation (24), a change in the relaxation rate. For a pure covalent compound, the major effect of 61 solvent is to change the correlation time and the relaxation rate as described before. The electric field gradient, being short ranged, is largely unaffected because it is intramolec- ular in origin. Examples of this type of behavior have been given by Kintzinger and Lehn (65) for nitrogen heterocycles in various aprotic solvents. For these solvents, the chief effect is to change viscosity and hence the correlation time. A similar situation is found for low concentrations of elec- trolytes in solvents of high dielectric constant, where sol- vation and ionization are complete. The only change is in viscosity, from solvent variation or from small concentration changes in a given solvent. Examples of this type have been given by Randall and colleagues (80, 81, and 82) for quater- nary methyl and/or ethyl halides of boron, aluminum, nitro- gen, arsenic and antimony in water (a = 80.4 at 20°): also by Kawazoe (79) for tetraethylammonium bromide in water. dimethylsulfoxide (c = 48.9 at 20°), and methanol (e = 32 4 at 20°). Similar behavior was observed for 87Rb for concen— trations of RbI below 10"2 M in dimethylsulfoxide (01450) by Crawford and Gasser (101). The set of conditions where only viscosity effects operate to change the correlation time, and hence the quadrupole relaxation rate,will be called Case I. The warning of Massey, Randall and Shaw (81) is again emphasized here: solution viscosity should not be 62 the only criterion for relaxation rate variation. A detailed examination of the conditions, other than vis- cosity, which lead to solvent effects on the relaxation rate will now be made and will be designated in Case II. The quadrupole coupling constant should be con- sidered as another possible source of variation (Equation 24) which can be affected by changing solvents since the electric field gradient, (an/ofiz) . is dependent on the surrounding structure and hence has a very marked effect can the relaxation rate. This was pointed out earlier for those cases where the electric fields were intramolecular 111 origin. Now electric fields induced by the solvent 3 re examined . Consider an electrolyte, an ionic compound, in a aOlvent of low dielectric constant. Here there can be liuttle doubt about the microstructure. The low dielectric c('Jnstant of the solvent will not favor ionization. there- f(Dre the electrolyte will be present as ion pairs or higher aggregates and, as a result of the close proximity of the ions there will be a definite interaction. Now consider an electrolyte, one of the ions of which contains a quadrupolar nucleus, in a solvent of high dielectric constant. The ions will be completely solvated and separated and the ion of the quadrupolar 63 nucleus, for the sake of simplicity chosen monatomic, 3‘34 79Br",will be in a symmetric electrical environment,nuclear relaxation will be slow and a narrow resonance line will be observed. At extremely high concentrations in a solvent of high dielectric constant, or near saturation in a sol- vent of low dielectric constant, there will be significant ion-ion interaction such as ion pairing. As a result of ‘7 'the ion pairing, the gegen ion will set up an electric field near the quadrupolar nucleus. In the example of the 79Br" ion the electric field caused by its gegen ion will be asymmetrical and faster relaxation with a unider resonance line will be the result. The appearance or disappearance of spin coupling (from <=<>cpling with the quadrupolar nucleus) in the proton Spectrum E>Iwovides a method of determining whether or not ion pairing 153 important. Kawazoe and workers (79) investigated the 8<>lvent effect on quaternary ammonium halides. They rnbted that chloroform (c I 4.8 at 20°) caused a broad— 14N coupling for tetraethylammon- Gning or collapse of the ium, 2,N,N-trimethylpiperidinium and sis-2,6,N,N~tetra- Methylpiperidinium halides compared to aqueous solutions. They pointed out, that there was little observable effect on changing the various halide gegen ions. Randall and co—workers. on the contrary, reported that there was no 64 solvent effect on tetramethyl- and tetraethylammonium halides (80 and 82) whereas there was a very definite effect on the tetramethyl— and tetraethylarsonium and antimonium halides (80, 81, and 82). This solvent effect. in which multiplet-structure is lost in solvents of low dielectric constant. they attributed to ion pairing. .3: In addition to these, Crawford and Gasser (101) rioted considerable broadening of the 87Rb resonance Iline with increased concentration of RbI in DMSO which t:hey attributed to ion—pairing. Their conductivity rneasurements indicate that above 10"2 M the salt exists as ion pairs in DMSO. Next, consider the more complex situation in which some type of chemical reaction between solute and solvent c><=curs. The obvious case is the one we are most concerned aJDout, namely, the pretonation of the nitrogen lone pair electrons. Although they did not appreciate the complexity c3f the situation, Tiers and Bovey (69) were the first to The emphasis of their paper report this type of effect. ‘Was on the collapse of 1H—14N coupling owing to an asymmetric field at nitrogen caused by the molecular structure: however, the solvent they chose was trifluoroacetic acid. Obviously all the amines and amides were protonated to form the tri- fluoroacetate salts, but since the dielectric constant 65 (s I 39.5 at 20°) is rather low ion pairing could be appreciable. As a result, the electric field at nitrogen could have been more or less symmetrical than they assumed but certainly was quite different. Along these same lines, Grunwald and Price (23 and 29) reported a very definite effect of anion on the NH line shape, and hence on 14N relaxation times, for monomethyl— and trimethyl- ammonium salts in glacial acetic acid (e I 6.22 at 25°). From freezing point data, they knew that the salts existed as ion pairs and from the change in NH line shape compared to aqueous solutions they reasoned that the close proximity <3f the anion caused a strong electric field gradient at the rlitrogen. Similarly, Cocivera (31) reported the collapse <>f 1H-14N coupling and change in line shapes for salts of tflae same amines in t-butyl alcohol (s I 12.47 at 25°). Tunis he explained as being caused by the anion being in intimate contact with the ammonium ion so that a strong IElectric field is created at the nitrogen nucleus causing rapid 14N relaxation. Another case in which the reaction of solute and solvent is quite clearcut is that of solutions of nitrogen heterocycles in sulfuric acid. Kintzinger and Lehn (65) pointed out that the protonation of nitrogen in any compound tends to make the electric field around nitro- gen more symmetric. This will be true in any solvent 66 where protonation can occur; however, if the protic solvent has a low dielectric constant ion-pairing will occur, as was described for glacial acetic acid solution, and the gegen ion will again produce an asymmetric electric field at nitrogen. As was described in detail in Section IIA, an additional complicating feature of the particular solvents glacial acetic acid and t-butanol is that they are weak acids and so rapid proton exchange between the nitrogen and solvent occurs. These two complications are eliminated by using sulfuric acid since it is a strong acid and proton exchange between it and the nitrogen is ‘very slow: also, since it has a very large dielectric constant (c In 100.0 at 25°) the electrolytes will be fully d issocia ted . Finally, we come to the most complicated situation, 2Me2H > lMe2H > 3MelH > 4H which 3Qems unusual since it would be reasonable to expect increasing ‘lkyl substitution to have an additive effect on the 14N shift. Details of the concept of substituent additivity effects on chemical shifts have been presented by Maslov (120). From the work of Schmidt 3; 31. (117), Evans and Richards (118) , and Bose 3.3; L. (119), one learns that these workers often did not observe the shift values they reported, but rhther obtained them from the earlier published spectra of 099 and Ray (68) . Indeed.in reviewing the work of Ogg And Ray (68) , one sees that only the spectra are presentedy neither a listing of the chemical shift values are given, 74 nor are details as to how the spectra were calibrated given since the main point of their paper was to establish the 141! line shapes and the effects of quadrupolar relaxation. (Section IIB) . A review of some of the theories of chemical shifts in general will now he presented, after which nitrogen chemical shifts are treated specifically. Magnetic coupling of the surrounding electrons to the observed nucleus arises from magnetic fields originating from the motion of the electrical charges. The fields at: the nucleus caused by orbital motion of the surrounding electrons act either in conjunction with, or opposed to, the iPPlied magnetic field Ho- Thus, the nucleus in a molecule W111 resonate at an applied field different from the reson— ance field necessary for a bare nucleus. Furthermore, as the field at the nucleus is a function of the electronic ntructure about it, there will be a shift in applied field with a change in chemical structure. This may conveniently bQ expressed as: w - 7(Ho—9Ho) = yflo(l-0) (29) lWhore s is the resonance frequency, y is the gyromagnetic ratio, and o is a parameter describing the shift in resonance with respect to the bare nucleus, _i__._e__._ the chemical shielding parameter (121) . The shielding term may be broken up into 75 the sum of three other terms (122): o - MD + up + 5A (30) 59 is a diamagnetic effect from electrons in s-states, given by (121): OD - (q2/3mc2) (1m (31) where q is the charge of the electron, m its mass, and r is the distance from the electron to the nucleus. OP is a paramagnetic effect resulting from 'unquenching' of p-states, given by (121): op - -(2/3:.E) (11 q/mc)2 (m3) (32) where LE is the mean excitatio n energy and the other terms are as defined above. °A includes long-range effects from other electrons in the molecule. GA is very difficult to evaluate since it depends on the orientation of the group producing the effect relative to the observed nucleus and the resultant contribution is often very small. The diamagnetic term acts contrary to the applied field, causes shifts to higher field, and as Slichter (121) has pointed out, its range is small being only.v10 ppm for the F2 molecule, whereas EP is two orders of magnitude larger for the same molecule. Since the diamagnetic shift is caused by 6 bonds, it should be of little importance forrntrogen compounds since these bonds have con- siderable p character. The :p term generallyxnakes 76 'the largest contribution to the shielding in nuclei other than protons, both because its effects are two orders of magni— ‘tude larger than 9D and because,in the case of nitrogen, loonds involving I'electrons are important. It is difficult 'to obtain good wave functions for the molecules and difficult to evaluate themean excitation energy. LE, which is also :necessary for the theoretical calculation. For these reasons, 14N chemical shifts have so far been calculated for only a few compounds. Lambert and Roberts (123) have calculated the diamagnetic contribution for seven nitrogen compounds and have shown that it accounts for only about 1/7 of the observed 15 N chemical shifts. Kent and wagner (124), on the other hand, considering only the paramagnetic contri- 'butions, have gotten quite good agreement between their calculated and experimental 14N chemical shifts for twelve linear triatomic molecules and ions. In order to circumvent the difficulties in choosing I value for the mean excitation energy and calculating the paramagnetic shielding, Witanowski gt alv,have used thbond orders and charge densities to explain 14N chemical shifts for m- and p-dinitrobenzene (113) and for nitriles and isonitriles (114). This approach is typical because, rather then face all the difficulties encountered in making a full calculation of the 14N shifts, the observed values 77 are usually rationalised in semi-quantitative or qualitative ‘terms. There are a few useful qualitative correlations, rnostly stemming from the early work of Holder and Klein (116), of the observed 14N chemical shifts. (a) Since 1013 :resonates at high field, N03“ and N02— at lowest field, and 'various carbon-nitrogen compounds in between, they suggested 'that the electronegativity of the neighboring atoms or group plays a major role in the chemical shift value. Thus, the inore electronegative the substituents attached to nitrogen. the greater will be the deshielding of nitrogen and the :result will be a paramagnetic or low-field shift. Con— ‘rersely, electropositive atoms or groups cause nitrogen to be shielded with upfield shifts resulting. (b) Their second general observation is that since NH4+ has a more ionic bond and resonates at high field whereas the more covalent N03" resonates at low field, bond character must also be of considerable importance. An atom with closed shell, i.e., a noble-gas configuration, possesses zero angular momentum, and as a result, would have no paramagnetic contribution. Thus, the ammonium ion, which fulfills this prerequisite of having nitrogen in an electronic environment most similar to the noble-gas structure, is dominated by the diamagnetic term and resonates at high field. Conversely, as asymmetries 78 in the electronic structure become manifest, orbital an— gular momentum will result, the paramagnetic term will not be negligible and resonance will occur at lower applied field. (c) For a series of structurally similar compounds, Witanowski 33,; m (111, 113. 114 and 125) have shown that the 14N chemical shifts for nitro-, compounds, -m’triles, and -isonitriles are governed by inductive effects, either electron-donating (+1) or electron-withdrawing (—I), which can be estimated by considering the n—bond polarity (114). All of these theories, in general, do agree with the experimentally observed 14N shifts and do give rea- sonable explanations why some resonances are at higher field than others. For specific cases, such as the alkyl ammonium salts we are dealing with here, however, all theories do not offer valid explanations since their pre- dictions often fail to agree with what is observed. The electronegativity of the group bonded to 14N seems to explain gross nitrogen shifts over the entire range observed. as was pointed out above. In the specific case of alkyl nitro compounds, Witanowski.gt al. (125) point out that their results are just the opposite of ‘what is predicted on the basis of electronegativity. In addition, and also w to our observations, Evans and Richards (118) have indicated that electronegativity 79 fails to explain the shifts for the symmetric ammonium ions: the shifts to low field are found to be in the order: 4Et >»4Me > 4H whereas the opposite order should be observed if electro- negativity is of major importance. Ionic character of the bonds to nitrogen cannot be the only factor as shown by Schmidt,et.glé,(ll7), who pointed out that in NH: the NH bond has about 20% ionic character while the N0 bond in N03 has about 12% ionic character: thus, the theory relating 14N chemical shift to ionic character must account for a very large shift, approximately 350 ppm, for such a small change L95, 8%) in ionic character. The theory would be even more hard- pressed for the alkyl ammonium salts (§£¢_Table 1)! as the bonding is very similar for all, one would expect changes for ionic character in the third significant figure and this would have to account for shifts of up to 43 ppm. Nevertheless, it could possibly be argued that the quaternary alkyl ammonium ions (4Et and 4Me) have a more covalent carbon—nitrogen bond than the hydro- gen-nitrogen bond of the ammonium ion (4H), hence, should show a paramagnetic shift as observed. This explanation 80 does not account for the ethylammonium ions consistently resonating at lower field than the corresponding methyl analogues, nor could a prediction for such alkyl substi- tuted ions as butyl, cyclohexyl and neopentyl be made. Witanowski.gt.1;; (113 and 114) have shown that inductive effects adequately explain the 14N shifts for nitro compounds, nitriles and isonitriles but bond orders and charge densities are not always easily obtained. An additional point is that their approach is based on H-bond orders, as paramagnetic shifts and paramagnetic shielding is dependent on the amount of s-bonding between nitrogen and the adjacent atoms. In the alkylammonium ions nébonding is negligible so this approach is not helpful. Finally, let us consider specifically ammonia and the alkyl amines, which are our chief concern. For these compounds, the nitrogen shifts result mainly from the diamagnetic shielding term, the paramagnetic shielding term having only a small second—order effect. The reason is that excited states for ammonia and the amines corres— pond to an n-to* transition rather than to an n-on * transition as in nitrogen-heterocyclics, nitro and nitroso compounds, nitrates and nitrites. Since op is inversely proportional to the excited state energy and since ;E for an n—vo* transition is greater than LE for an n-’H'* transition, 81 it is easy to see that Op for ammonia and amines will be of less importance than it is for other nitrogen contain- ing compounds (118, 119 and 123). As a result of the greater importance of the on shielding term, ammonia and amines appear at the high field end of the nitrogen resonance spectrum. D. Nuclear Spin-Spin Coupling General Obsgryations Under high-resolution conditions (high degree of magnetic field homogeneity) fine structure is sometimes observed in the NMR spectrum. This fine structure is intramolecular in origin like chemical shifts but unlike chemical shifts, which are due to the electronic shielding of the nucleus, this fine structure results from the interaction of the nucleus with neighboring nuclei. Since the orientation of the neighboring nuclear spins may be parallel, antiparallel or inclined at various fixed angles to the applied magnetic field, the local field at the observed nucleus will have values greater than, less than or the same as the applied field. As a result, the resonance of the observed nucleus will be split into two or more lines with separation J depending on the number of allowed orientations of the spins of 82 neighboring nuclei. In turn, the resonances of the neigh- boring nuclei will themselves be split by the same value, J. Since one nucleus (or set of equivalent nuclei) has its spin(s) coupled with another set of chemically shifted nuclear spins, this interaction is called "spin—spin coupling' and the fine structure splitting value J is termed the I'spin-spin coupling constant" or simply the I'coupling constant“ (126 and 127). Throughout the years a large number of coupling constants for various molecules and nuclei have been measured and a number of generalizations can now be made (128 and 129). First, coupling is not observed between equivalent nuclei. For example, the hydrogens of a methyl group do not couple with one another nor do the hydrogens of benzene. Coupling is only observed when the nuclei are magnetically non-equivalent. For example, when there is a chemical shift difference between the nuclei they are obviously non-equivalent. Secondly, when the chemical shifts (measured in units of frequency) are an order of magnitude or more greater than the coupling constant, the number of equally spaced resonance lines is given by: Number of resonance signals a 2 n1 + l (33) where n is the number of equivalent neighboring nuclei and 83 I is the nuclear spin number of the neighboring nuclei. In addition, for nuclei of spin 1/2 the relative intensities of the lines will be given by the binomial coefficients. For example, the resonance peak for a nucleus (or set) with one neighbor (I - 1/2) will be split into an equal-intensity doublet and the peak for a nucleus with three equivalent neighbors (I - 1/2) will be split into a quartet with intensity ratios 1:3:3tl. If I > 1/2 the relaxation time of the quadrupolar nucleus (I > 1/2) must be sufficiently long so that coupling can be observed (cf. Section II B). Under these conditions the number of lines will be given by Equation (33): however, the intensities may not follow predictions and the separations may be less than the true coupling constant, both of these being a function of the relaxation time as shown by Pople (9), Bacon gt 5;; (116). and Suzuki and Kubo (84). As an example of this situation consider the ammonium ion in aqueous solution. The proton spectrum consists of three lines of equal intensity re- sulting from coupling with one l4N nucleus (I - l) and the 14N spectrum consists of five lines with intensity ratios 114162431. It is to be emphasized, that all of the fore- going description applies to the situation where L 3 10 J and such spectra are called first order because some of the 84 spin states are degenerate. The other situation is strong coupling where the degeneracy is removed and this occurs when the chemical shift approaches the same magnitude as the coupling constant. This condition leads to anomalous intensities, spacings, and number of lines. Now there are more lines than predicted by first-order theory (i.g. Equation 33), the intensities do not follow the binomial coefficients and there may be no spacing directly identi- fiable as a chemical shift or coupling constant. In this situation one must employ a complete quantum mechanical analysis of the spin system. A digital computer is then most conveniently used for analysis of the spectrum to determine coupling constants and chemical shifts. A third generalization is that the magnitude of the coupling constant between two nuclei increases with increasing atomic number, e.g. 19F-19F coupling is greater than 1H-1H coupling and 19F-13C coupling is greater than 1H-13C coupling. In addition, for two isotopes (nuclei with the same atomic number) coupling with a common third nucleus, coupling is proportional to the gyromagnetic ratios of the isotopes so will be larger for the nucleus with the larger gyromagnetic ratio,.g.g.. 3(1H—15N) > J(1H—14N) and J(l9F-11B) > J (NF—1013). 85 Fourth, the magnitude of the coupling constant generally falls off with distance, that is, as the number of bonds intervening between two coupled nuclei increases, the coupling constant decreases. There are exceptions to this as will be pointed out later. There may also be an angular dependence of the coupling. Fifth, the coupling constant is independent of the applied magnetic field and often also of temperature, whereas chemical shifts vary directly with the applied field and are often quite temperature dependent. This provides an experimental method for simplifying the inter- pretation of a spectrum. By obtaining spectra at two or more field strengths (preferably each 20—30% larger than the one previous), centers of multiplets which are chemi- cally shifted will be spread further apart, whereas the spacings within the multiplets will be the same regard- less of the increased field value, as these spacings are due to spin-spin coupling. This is one stimulus for achieving very high magnetic fields since most spectra will approach first-order structure and analysis will be simpler. Finally, theoretical calculations (130) and ex— perimental measurements agree that the coupling constant has a sign associated with it and signs for most J'values 86 are now known. Mechanism 9f SpineSpin ngpligg The mechanism involved in spin coupling should now be examined, to better understand these values. The ori— ginal explanation proposed was interaction of electronic orbital currents with the nuclear spins: namely, one nucleus induces currents in the electron cloud of the bond. which then pass the information to the other nucleus. This model failed, however, because calculations based on it failed to agree with observed values by an order of mag- nitude or more (126 and 127). A mechanism was proposed later in which the nuclear spins interact with the elec- tron spins, following the Hund rule that all spins remain antiparallel. Calculations based on thisxnodel yield values which have the correct order of magnitude. Theoretical derivations based on the use of a Hamiltonian for the motion of the electrons in the fields produced by nuclei having magnetic moments yield a coup- ling constant which is the sum of three major terms. The first is from interaction between electron orbital and nuclear magnetic moments, the second arises from magnetic dipole interaction between electronic and nuclear spins and the third is the Fermi-contact term. The most sig- nificant term and the one which has the biggest effect is 87 the last one, the cxantact term, which is expressed (130): OCCo un s J(AB)Fermi - '“YAYs/AE) :3 fl (“‘1de M15913 1 l (34) Where K is a constant, YA and 73 are the gyromagnetic ratios of the two nuclei A and B,respectively, and AB is the mean value of the excitation energies. The summations are over all occupied and unoccupied orbitals. Similarly, the other two terms involve the gyromagnetic ratios of the two coupled nuclei and an approximate excitation energy. The standard quantum mechanical approaches of molecular orbital theory or valence bond theory, combined with the variational method, have been employed for cal- culations of J values but a complete discussion of the theoretical calculation of coupling constants is beyond the scope of this reviewy further details may be found in discussions given by Pople, Schneider and Bernstein (128) and Emsley, Feeney and Sutcliffe (129). For protons the main contribution to the coupling comes from the Fermi-contact term, Equation (34% with the other two terms making little or no contribution. The chief reason for this is that the hydrogen atoms use a ls-type atomic orbital in bond formation: the other coup— ling terms drop out unless other orbitals such as p, d and 88 f are involved. For nuclei other than protons the other two terms, the electron-orbital and the electron-nuclear dipole terms, also enter into the calculation as p- and d— type atomic orbitals now must be considered. Because of the inclusion of these terms, and because more orbitals are contributing, it is easy to see why coupling involving nuclei other than protons if; in general larger than proton-proton coupling. Since more orbitals must now be evaluated, it is understandable that the task is also considerably more difficult. S ' ° ' N' n C m A brief review of spin coupling in alkyl amines and ammonium salts will now be given. Alkyl ammonium salts contain the elements carbon, hydrogen, and nitrogen and there are many different pairs of nuclei which are capa- ble of coupling but this discussion will be limited to H-H and N-H couplings as these are readily observable and the only ones of major concern to us here. Proton—proton couplings will be discussed first as these are the most obvious in the spectra since they always occur, whereas the proton-nitrogen couplings may not always be observed as discussed above (Section II B). 89 Proton-Proton CouplingConstantg_ The first type of proton-proton coupling to be con— sidered is that between an NH proton and the alkyl s—hydrogens,.;.g. proton-proton coupling across three bonds and through nitrogen. (Incidentally, all couplings are classified first as to the number of intervening bonds between the two spin-coupled nuclei, as we have mentioned above that the coupling decreases with distance: couplings with two and three intervening bonds are designated 2J and 3J , etc. A subscript is then added to identify the particular nuclei involved.) This particular coupling, 3J(§-N-C§), is present in all members of the series of alkylamines and alkylammonium salts, the simplest ones being methylamine and ammonium ion. The most extensive list of these is that given by Freifelder, Mattoon and Kriese (131), who reported values for N-monosubstituted methylamin—es. Their results, together with a few other similar couplings, are given in Table 2. From the values listed it can be seen that the NH proton coupling with the Nqalkyl proton is about 5 Hz. .Also, there is a measurable difference in 3J(§NC§) depending on the R group on the nitrogen. Clearly we are seeing the effects of more than just the s electrons and contributions tO.J 90 Table 2. Three bond proton-proton coupling through nitrogen of the type 3J(R-N§:Q§3). R-Group 3J (ENC?) Reference Amides (BCO--)* 4.5-5.0 Hz 131 Formamides (HCO--) 4.9-5.0 132 Ureas, Biurets ) )(BNHCO--)4.0-5.0 131 and Guanidines ) Ethyl Carbamyl (EtOCO--) 4.9 97 Sulfonamides (B-SOZ--) 5.0 l31 Nitramines (N02--) 4.0 86 Dinitro- trinitroPhenyls 5.3-5.7 133 *B represents a hydrocarbon group. from terms other than the Fermi contact term. Unfortunately, Freifelder §t_§;, (131) and Hedberg g; _l. (133) did not speculate in this regard only commenting that in some compounds no coupling was observed, probably because these were weak bases and proton exchange was occurring (sf, Section II A). Quite interestingly, if the three- bond proton-proton coupling through nitrogen involves something other than methyl or simple alkyl substituents 91 on nitrogen a wider range of values are found. For example, Freifelder gt_§;, (134) report a value of 6 Hz for an acetylated amine of the structure -CH2—NngOCH3 (protons coupling are underlined). For substituted form- amides, coupling of the NH proton with the formyl proton over three bonds exhibits a much wider range than the simple 5.0.: 1.0 Hz of Table 2. Sunners, Piette, and Schneider (135) report a range in various solvents of 1.7 to 2.1 Hz for formamide and Randall and Baldeschwieler (132) a range of 1.8-2.3 Hz for N-methylformamidey the pro- 0 \\C-N/ H/’ \\H . tons being cis in both cases, i,§fl The values are much larger, 12.9-13.6 Hz, for the trans coupling, i.e. C-N , of formamide (135). Thus, we can now see how difficult it is to rationalize coupling constants as not only the number of intervening bonds but also the bond types and angles quite naturally causes differences. In addition, another complicating factor is the medium ("solvent effect") which is the reason for the range of values for the formamides mentioned (132 and 135). 92 Values of 3J’(§m0§) for quarternary ammonium ions are summarized in Table 3, along with the concentrations and solvents, both of which may possibly be important From the table we see that for the same concentration in water, there is a decrease in the magnitude of the coupling constant from monomethyl- to trimethylammonium ion- Although these values come from and were used in the kinetic studies (9;, Section II A), the authors of those articles never commented on the J values. One value in Table 3 that appears to be very much out of line with the others in its group is the 6.5 Hz reported by Emerson etual. (20) for trimethylammonium ion in sulfuric acid solution. Fraenkel and Asahi (38) noted that this coup- ling constant is a function of the sulfuric acid concen- tration but never found values as large as 6.5 Hz. 93 mm mm fin ON mm mm mm ON hm III I. [j :0 memM 0.5 Hum .HH@ .2 a wagoaaouuxa vanmé 483 x»: 8.0 .2 m. [moz+mzfmmuv gim neon vaumum .2 0.9.6.0 IHU+mZMAmmUV m 35 omm .2 0.04.0 uHo+mz Ammuv . v N . . m m m o om m «33 31mm .2 m oum o 18.32 A moo ~.m omm .2 m.~..m.o uHo+mzmxmmuv Wm omm .2 564.0 Lu+~mz~xmmuv :6 E nao+mmzmmu e v N I. ‘ I e m m Hmo ommauzmmvm resume 16+sz 3.0 n omé cum .2 This; Lu+mm2mmu um mo.o w 36 of .z méumé Lu+mmzmmu mam .uuw iucm>aom 20600 ocsomfiaw. . A+leofivhm wahu 93 mo ammouuac :msounu mafiamsou cauoumlcououm unanimounn. .m wanna. 94 pave uficowasmocusaoulm I moan: aw o.m Hum .HH@ .2.“ maauaummaa uaanuoann.z.z we H.o onus uaumon an smoam z ¢.o 9: m.o mafiaacnasnuaeHanz.z emmuuemex 1 A+mwam¢bm .Jumw .uewwaom ..ucou ocsomamw )l Anosaaucoov m manna 95 Other types of compounds have been included in Table 3 for general interest. The value for the N,N' —dimethylpiperazinium ion, a cyclic tertiary ammonium ion, is about the same as those for the acyclic tertiary tri- methylammonium ion. The value for the other tertiary ammonium ion, N,N-dimethylanilinium ion, may or may not be significant. It would first have to be measured in dilute aqueous HCl for comparison with the others. Simi- larly, based on the values of J for dimethylpiperazinium and trimethylammonium ions one would expect the J value for the pyrrolidinium ion to be near that of the dimethyl- ammonium ion: here the difference may be due to strain in the five-membered pyrrolidinium system. The second type of proton-proton coupling to be reviewed is the coupling within the alkyl group itself and the only one which we are concerned with is the three bond coupling 3J(§CQ§) within the ethyl group, that is, the coupling between the methyl and the methylene protons. For convenience, a few of these values have been listed in Table 4 in order of increasing coupling constant. Examining Table 4 it can be seen that the values depend on the atom to which the ethyl group is attached. This seems to have escaped the attention of earlier workers, 96 om umu vs Ao~m\m0mumzoaammuvooom m.n _ Hm +mmzmum v.n r we o~m\aomumz~muu m.h m mma mmum Nv.n a we o~m\Hozzmum m.~ “ mma mmum ~.n _ we o~o\szum mv.h and monouum ~.n om o~m\umzvum «.5 sma .uma Nunez o.» as o~mxumzvum m.» om mama o.» v» onxumzvum um ~.a om vummnz um m.o oocouowem uce>aom .pcsomfiou Amuum¢bmu oucouowmm pcsomsoo Amuumvhm .msoum Hague on» how nucnuncou madamsoo cauoumlcou0hm .4 «Hana 97 N ovH “.ch o . m ova maoamum o.m ova Naoammum Ao.mv mma «mvum ~.o w m.n ovH cucum m.n oucuuuwuw acm>aom . pcsomfiou Afloumv hm vocmuummm ocsommmw Among hm lql Cfucouv v manna. 98 probably because they were studying either the effect of the central element on internal chemical shift between methyl and methylene (80, 81 and 140) or its effect on the relative couplings between the central element and methylene and methyl hydrogens (80 and 81). The value of 3J(§CC§) for ethyl substituted quaternary ammonium ions seems to be affected little,if at all,by other substitu- ents on nitrogen. Massey.§t,al. (80) detected no differ- ence on quaternization, reporting the same value, 7.4 Hz, for triethylamine and tetraethylammonium ion. Also, Freireider,_e_t_ 1;. (134) report 3a =- 7.0 Hz for several diethylamines. 14N-Pzgton Coupling angggggg Now we come to the third major type of spin-spin coupling, the proton-nitrogen coupling. Coupling between protons and nitrogen will not always be observed depending on the electric field gradient at nitrogen (Section II B) . Nevertheless, there have been a number of reports of 1H-14N coupling constants so it is worthwhile reviewing them. Surprisingly enough, there are a sufficient number of wealthy chemists around so that a considerable number of 15N enriched compounds have been synthesized. The im- portance of this is that 15N (I - 1/2) has no quadrupole 99 moment so will always exhibit coupling with protons. From the lH-lsN couplings one can calculate the corresponding 1H-14N values. We shall review both,considering the 14N values first. Here again they are grouped according to number of intervening bonds, those for coupling across one bond being listed in Table 5. Some of the values are clearly inaccurate, the older ones being particularly suspect. Some general trends can be seen, however: first, protonation definitely increases the value as can be seen by comparing lJ(1H-14N) for liquid ammonia and the ammonium ion, where an increase of 8 to 9 Hz is observed. Second, replacing ammonium hydro- gens by alkyl groups causes a change in lJ(lH-14N). Third, if the nitrogen is part of a double bond system or a hetero— aromatic ring there is an extremely large increase. Because the precision of these values is rather poor, further com- ment is reserved until the lH-lsN values are considered since there is no doubt about the accuracy of these. In addition, a number of workers who have reported 15N data have made suggestions to explain the various differences in the values. Muira and Saika (148) recalculated the coupling constant for liquid ammonia and showed that the Fermi contact term accounts for about 70% of the value: they agree with Pople and Santry (130) that this value is lOO ma as v mac cmm Hovmz n.4n mva vommm wen umvmz m.o H m.mm «ea mozm rue: coauacaua aware, mozvmz H.o w m.~m mm .mm 000 .oaom oauuom Hmaunamo uavmz o.H “ o.~n med Hon 2 m :a naumusumm Hovmz o.~ w o.Hm was How 2 m mozvmz o.H M o.om ma AH v rev oar +vmz o.om mm weaved ao mmz «.0 M m.mv ma canvas mun mmz m.o u m.m¢ «ea shaved and mmz m.m¢ e canvaa sun mmz o.~ u o.uv v carved sun mmz o.uv HvH weaved sun mmz um o.H w o.ov museummmm uno«M4psow, pcsomeou ,l. Azealmavha ll .Ucon mco nachos nucnuncoo madamzoo zvalma .m wanna lOl D: Eon 03333333 5 8202 n 2:53 or: 3; Hon 358m 5 $222 m sausage n.o H 9?. no e0m «m 5 $3 merges o. 8 m 0.08 :82 «382:8 0.”. w ode m 0093 :32 028033350214 o.m H... can m 032 :82 02.562 9m w 5.3 m 003 :82 29:3 9m H o.nm m 002 SE “V3332” 5 eon How . masseuse oé H 0.3 av usau> pounaquno .ONE HUmZMANMUmmUV o.nn m cum ouauaoaum :« eon HommzmmUmmo o.~ w o.om 3 008 an 35:32 85% Ana“: 9: m o~m @3333 5 mom 8323mm“: o.m w 98 em emu coauaraum ca 2 o.mue.a Hommzmmu m.H w o.¢m m of 333.63» 5 mom Hommzmmo o.~ « adv moseuowam escauacsoo ,, .mrsomeou szalmavha AU.uaouv m Canoe 102 positive. They point out one of the major pitfalls of all theoretical caluclations of coupling constants, namely, the difficulty of estimating a value for the mean excita- tion energy [gf. Equation (34)]. The two-bond nitrogen-proton couplings are listed in Table 6. Some generalizations can be made. First, the two-bond coupling 23(CH3-N) is consistently greater for methyl than is 2J(CH2-N) for the methylenes of ethyl and other derivatives. This can be seen not only in the values for the quarternary alkylammonium ions, but also in those of the isonitriles and was pointed out by Massey, £3,3l, (81) who noted that it is true also of alkyl groups on ele- ments other than nitrogen. They gave no theory to explain this fact although it seems rather obvious that replacing a methyl hydrogen with a methylene (i.e. a carbon-hydrogen bond with a carbon-carbon bond) will cause a change in the hybridization and excitation energies, and hence, in the coupling constant. The problem is to come up with a theory to explain the direction of the change. The answer to this perhaps comes from the third point which isn‘t empha- sized in Table 6,namely, the sign. McFarlane (149) has shown the two-bond NH coupling to be negative, as had been proposed by Bourn,.g;,al. (154), in ethyl isonitrile where the nitrogen couples through an alkyl carbon. Bourn 103 Table 6. 1H-l‘N coupling constants across two bonds. 2J(1H-14N) Compound a References 0.2 Hz (CH3CH2)4N+ 149 < 0.5 (CH3C§2)4N+ 71; 72 N 0 (m3q1_2)4n+ 74, 150 - 0.4 (Q§3)4N+ 149: 151 0.5 (Ca3)3NCH - CHR+ 76: 152 0.54 i 0.05 (qg3)4N+ 25 0.55 t 0.05 (qg3)4N+ 81: 32 ~1.0 (@3)4N+ 70 1.33 t 0.06 0&3-NC0 90 1.30 t 0.2 C‘Hs-cgch 85 1.80 I 0.2 (CH3)20§NC 85 - 1.9 CH3c§2Nc 149 - 2.0 CH3c§2Nc 150 2.0 CH3(CH2)20§2NC 85 - 2.3 c§3Nc 149 2.35 Q§3NC 153 2.7 Cfl3Nc 85 2.5 (CH3)3NC§ - CHCOOH+ 76 + 2.9 (CH3)3N0§ - CH-c6H5+ 152 + 3.5 (CH3)3Nc§ - CH2+ 152 3.6 (CH3)3Ncg - CH2+ 76 aProtons coupling with the nitrogen—are underlined. 104 and coeworkers (154) showed that this is also true for N-methyl amides, the coupling between the amide nitrogen and the methyl proton being negative, However, Ohtsuru and Tori (152) have shown that the two-bond coupling of nitrogen with the geminal proton of the vinyl group is positive, so that comparison of values in the table without knowledge of the signs is dangerous. They attribute this sign change to different hybridization of either the nitrogen or the carbon which sounds quite reason- able as Bourn.et,al. (154) show that for N-methylform- amides coupling between nitrogen and the formyl proton is positive whereas coupling between nitrogen and the methyl protons is negative. The number of compounds for which three-bond lH—14N has been reported, is so large that a complete listing is impractical,so only a few are given in Table 7 for illustrative purposes. Some of the more interesting studies are now described in detailo The nitramines re- ported by Lamberton.g£‘gl, (86) constitute a new class of compounds for which 1H-14N coupling can be directly ob- served. The somewhat unusual feature is that the coupling is with the nitro—group nitrogen as was shown by hetero- nuclear decoupling experiments (87) The work of Lehn and Seher (76), as well as that of Ohtsuru and Tori (152), 105 Table 7. 1H-14N coupling constants across three bonds. 3J(1H—14N) ‘_Compound References 1.38 t 0.08 Hz CH3CN 89: 90 1.45 CH3CH2Nc§3N02 86 1.5 (Qfl3)2NN02 86 1.65 CH3c§2NCH3NOZ as 1.5 (CH3CH2)4N+ 70 1.65 (CH3CHZ)4N+ 72 1.7 (CHBCH2)4N+ 79v 82 + 1.8 (QE3CH2)4N+ 71: 74v 149 1.8 1' 0.03 (CH_3CI-12)4N+ 75 1.77 I 0.03 2,N,N-Trimethylpiperidinium 79 1.9 Several alkyl CH and CH2 74 quaternary N +2.0 (CH3)3CNC 149 2.4 t 0.1 R-—<§:EN—CH2C§3 78 +2.5 C§3CHzNC 1499150 2.6 (CH3)2CHNC 85 3.5 (CH3)3CNC 85 1.8 (CH3)3NCH - CHCOOH+ (21;)3 76 2.2 (CH3)3NCH - QEC6H5+ (915)“ 152 2.6 (CH3)3NCH - c§2+ 76: 152 3.0 t 0.2 R-— 1 (38) .Application of H2 power at this level causes a redistribu- tion of the energy level populations which results in intensity changes (either increase or decrease) of certain lines in the A multiplet, or if Condition (37) holds, of the collapsed multiplet (174, 178 and 179). If this Condition (38) is met while a double resonance experiment of Type (b) is performed, then this is termed a generalized nuclear Overhauser effect (NOE) (174, 175, and 178 - 181). Before proceding with more detailed descriptions of tickling, INDOR, TSI and NOE, the aspects of spin decoupling should be pointed out. It may be recalled that proton exchange (Section IIA) and quadrupolar re- laxation effects (Section IIB) both cause collapse of spin multiplets which appears to be analogous to spin 121 decoupling, Condition (37). In actuality, these three phenomena have quite different origins. A clue to this difference is the rather complex spectra obtained from intermediate values of H2 and w: in decoupling experiments, whereas intermediate exchange rates and relaxation rates produce very simple partially coalesced spectra (174 and 179). Another point to be emphasized is that spin de- coupling does not arise from saturation of the X group resonance as can be seen by Conditions (37) and (38). The decoupling process involves the orientation of the mag- netization vectors and saturation of the X group is only a side effect (179 and 182). At this point, we shall quickly look at the important features of the NMDR experiments mentioned above. The common feature of them all is that they may be applied to either homonuclear or heteronuclear spin systems. Spin tickling, Case (1), may be conducted in either field (a) or frequency sweep (b) modes, the latter being the best because the point of irradiation in the X multi— plet may be controlled. The result of this, is that lines sharing a common energy level will be split into doublets. The amount of splitting depends on whether or not the transitions between energy levels giving rise to the line are progressive (in series) or regressive (in parallel). 122 There are a number of limitations to this technique. First, it works very well for first-order spectra and when m2 is applied to well separated non-degenerate lines. For spectra with many lines, it is a small task to see which are affected and which aren't: in addition, overlapping lines increase the difficulties and effects on lines broadened by a quadrupolar nucleus such as 14N may be undetected (174 and 175). The TSI experiment is performed in field sweep mode (a), and at low power such as Case (1). This experi- ment is executed by applying H2 to a line in the X group after passing a line in the A multiplet and turning off H2 before reaching the next adjacent line in the A multiplet. Thus, for changes in the.A spectrum to be observed, the sweep time from line An to Ann must be less than T1(X). The advantage of this over tickling NMDR is the ease of identifying the progressive and re- gressive transitions of non-degenerate lines. Progressive transitions show an intensity increase compared to the unperturbed spectrum, and a deflection above the baseline, whereas regressive transitions show an intensity decrease and a deflection below the baseline (174). The difference between TSI and NOE is that for TSI H2 is given by Case (1): also TSI is usually conducted in field sweep (a), 123 and H2 is only transitory (174 and 178). The difficulty is that each non-degenerate line in the A group must be observed and interpretation for closely lying lines or second-order spectra will not be simple. NOE experiments are performed in the frequency- sweep mode (b) and with H2 amplitude sufficient for saturation, Case (3). This is different from TSI both is in the irradiating amplitude H2 and the fact that H2 applied continuously to a line in the X multiplet (178). The major limitation in the use of NOE is that T1 (A) must be greater than T1(X) (174, 178-181). The general character- istics of INDOR have been given, namely frequency sweep of a special kind (c) and the relatively low H2 power of Case (1). A: for TSI, the time required to sweep from one line, Xn' to another, Xn+l’ is shorter than the relaxation time T1(A)(174L Again. as for TSI, regressive transitions show a decrease (deflection below the baseline) while pregressive transitions show an increase (deflection above the baseline)- This is an advantage over the tickling method. A major advantage over both spin tickling and TSI is that only those lines in the X group having a common energy level with the A line being ob- served are recorded (recall that in INDOR, Type (c). ml is set on one line in the A multiplet). Thus, not only are mul- tiplets not coupled with A undetected,_e_ 3 L multiplets for which 124 J(AL) = 0, but also lines in the X multiplet not connected 'With the Particular A line being observed at ml are also not detected (175 and 183) . An additional application of INDOR is in the heteronuclear mode. Here one can make use of the strong inherent sensitivity of one nucleus while recording the spectrum of another of lower sensitivity. This was shown by Baker e; 11., who observed either 1H or 19F at m1 while sweeping $2 through spectrum of the desired nucleus,.g.g. 13C, 14N, 29si or 31P (176 and 177). This then concludes the basic NMDR methods, techniques, and descriptions which will be necessary to discuss the applications of NMDR to nitrogen containing compounds. There are a number of other methods and techniques, such as pulsing (184 and 185) and modulating (186) the frequency, which were not discussed because their applications are quite limited. The theoretical background has also been neglected because the emphasis in the following will be on experimental results. Details of the theory can be found in the early and extensive re- view of Baldeschwieler and Randall (179) or the more up-to- date work of Hoffman and Forsen (174). Throughout the following review, the notation of Baldeschwieler and Randall (179) will be used. This notation designates the observed nucleus first with the 125 irradiated nucleus second and in brackets, A-{X}v,g.g. lH-{14N] means that hydrogen was observed while 14N was irradiated. The first lH-{14N} NMDR experiment performed was that by Piette, Ray and Ogg (187). It was of Type (a) Case (2) 14N coupling and the sole purpose was to remove partial which was broadening the hydrogen resonances as a result of an intermediate rate of 14N quadrupolar relaxation (cf. Section IIB). Thus, they were then able to measure all of the proton-proton coupling constants for pure formamide, and for various partially deuterated formamides in aqueous solution. As a result, they demonstrated the existence of a solvent effect on both chemical shifts and coupling constants and in addition proved the thesis of Roberts (8) that the broadening of NH proton peaks in pure amides arises from partially collapsed 1H—14N coupling rather than from hydrogen—bonding, viscosity or inter- mediate proton exchange rates. In a similar manner, Randall and Balderschwieler (132) used 14N decoupling to simplify the spectrum of N—methylformamide for complete analysis of proton chemical shifts and coupling constants. They obtained quantitative values for the solvent effects on the coupling constants and suggested an explanation of these effects. In addition 126 they performed lH-{14N} decoupling on N,N-dimethylformamide, acetamide and N-methylacetamide, but were unable to resolve all the proton-proton couplings. .More recently Kamei (188) was able to completely decouple 14N and so measure the proton chemical shifts and coupling constants for formamide, acetamide and N-methylacetamide. Another example is the 1H—{14N} decoupling of pyridine and pyridinium ion, in order to obtain better resolved proton spectra so that computer fitting of the spectra would be easier. This was done by Merry and Goldstein (189), who were measuring changes in the proton coupling constants with protonation, and by Castellano,g; ,al., (190) who wanted a well resolved spectrum in order to obtain very accurate proton chemical shifts and coupling constants. A more sophisticated use of decoupling is to prove that coupling exists between two sets of nuclei. Hetero- nuclear decoupling, 1H-{14N}, was used by Anderson, Baldschwieler,,g§”§1. (71) to prove that the triplet fine structure of the methyl group of aqueous tetraethylammonium (4Et) bromide arises from large long range 1H—14N coupling and was confirmed by Milner and Turner (185) and by McFarlane (149). Whereas Anderson and Baldeschwieler (71) could detect no effect of 14N decoupling on the methylene of 127 the tetraethylammonium ion, McFarlane (149) reported a 4% 14N. McFarlane thus intensity increase on decoupling proved that the slight broadening of the methylene peak in the single-resonance spectrum is due to unresolvable lH-14N coupling. In addition, by decoupling 14N, McFarlane (149) definitely proved that the triplet structure of the proton peaks of tetramethylammonium ion (25) and of isonitriles (85) is due to 1H-14N coupling. In a similar manner, Lehn and Seher (76) both simplified the spectra and proved the presence of 1H—14N coupling in five tri— methyl (substituted vinyl) ammonium salts by lH-{14N}. Most recently, Hampson and Mathias (87) used lH—{14N} de- coupling to prove definitely that the fine structure of certain nitramines (86) arises from proton-nitrogen coupling. The last example leads to a third use of NMDR. Since ml and a: are known very accurately and since wz must be at the center of the X multiplet for optimum decoupling, the difference, “1"2' gives a very accurate value for the chemical shift between the A and X multiplets. This technique is extremely useful for two major reasons, first, the accuracy just mentioned and, second, it permits location of peaks which may be hidden, as in homonuclear decoupling, or unobserved, as in the case of heteronuclear decoupling. An example of homonuclear decoupling is given 128 by Kowalewski,gtwa1. (175), who performed 1H-{lH} decoupling on 3-acetylpyridine. By means of homonuclear INDOR they were able to determine accurately all the frequencies of the doublets of the hydrogens at the 2 and 6 positions even though these were so broadened by the pyridine nitrogen as to be unresolvable in the single-resonance spec- trum. Other examples are given in the reviews (174 and 179). For heteronuclear NMDR chemical shift determination, the A multiplet is observed while the X nuclei are not observed, wz being at a very different radio frequency. Baldeschwieler and Randall (191) were the first to demon— strate this technique for obtaining nitrogen chemical shifts by lH-{14N}. They did field-sweep decoupling, NMDR Type (a) Case (2), on ammonia, ammonium ion, pyridine and pyridinium ion (incidentally, a printing error has been pointed out (182) in (191), the figures are mis- labeled and incorrect values for the chemical shifts are given in (191) but appear correctly in the review (192).) Briefly, the method is to obtain the value of oz for optimum 14N decoupling of some reference compound, such as NHZ, while observing the hydrogen spectrum, and then to scan the other nitrogen compounds again noting the value of oz for optimum decoupling. The difference between 129 these latter .2 values and the m2 of the reference NH: are the 14N chemical shifts with respect to the ammonium nitrogen. The chief advantage of this method is that it makes use of the inherently greater sensitivity of hydrogen NMR while obtaining the 14 N "spectrum," hence gives a better signal-to-noise ratio. At about the same time as the work of Baldeschwieler and Randall (191), this method was independently proposed by Glasel, Jackman and Turner (193). This technique was used by Hampson and Mathias for obtaining the 14N chemical shifts of twenty-seven primary and secondary amides (194) and six thioamides (109), using a modified INDOR method, Type (c) Case (1). Thus, in concluding we come to the example which brought us on the subject of chemical shifts by NMDR, the 14N decoupling of nitramines by Hampson and Mathias (87). Nitramines have two nitrogen atoms in different chemical environments, 14 RzNNOZ' The a frequency necessary to decouple N from 2 the R group fell in range of nitro group chemical shifts and was over 100 ppm different from amine chemical shift values. The NMDR experiment clearly shows that the alkyl group proton coupling is to the nitro group nitrogen rather than the amino nitrogen. The most effective way of obtaining chemical shifts from NMDR, particularly of the heteronuclear variety, is by 4 (I f) () tm 4 () IQ 130 INDOR. Baker et al. (176 and 177) have shown how the complete spectrum of the unobserved nucleus can be re- corded by setting al on a line in the A spectrum which is spin coupled to X. For example, they obtained the 13C spectrum of trifluoroacetic acid by observing 19F, the 2981 spectrum of tetramethylsilane by lH-[29$i} de- coupling, the 31P spectrum of trimethylphosphate by lH-{31P} decoupling and,of interest here, the 14 1 N spectrum of the ammonium ion by 1H-{ 4N} decoupling. The nature of the INDOR method permits coupling constants to be very accurately obtained. Baldeschwieler and Anderson (88 and 144) showed that field-sweep double resonance, Type (a) Case (1), was a sensitive method of accurately determining the nitrogen resonance frequency provided that a theoretical fit of the observed partially decoupled spectrum could be made. Baldeschwieler (144) confirmed this by performing 1H-{14N} decoupling on ammonium—14N ion and lH—{15N] de- coupling on ammonium-lsN ion. Actually the first NMDR experiments on 14N and 15N ammonium ion were done by Anderson, Pipkin and Baird (143). Later, Anderson and Baldeschwieler (88) applied the method to 14N ammonia but found that, since they did not take into account the broadening due to nitrogen relaxation, their accuracy was 131 limited to :1 Hz and that experimentally H; was limited to low values, Case (1). The basic method has been out- lined by Anderson (195), who pointed out that it only works for the field-sweep type of NMDR because values of oz above and below optimum decoupling yield mirror— image spectra (thus “point“ to the correct w ), whereas 2 the frequency-sweep type of NMDR yields symmetric identical spectra for values of .2 above and below optimum decoupling. As a consequence of being able to determine the optimum decoupling frequency in heteronuclear NMDR, one can very accurately measure the frequencies of both nuclei. Dividing one by the other, one is thus able to obtain a ratio of the gyromagnetic ratios with an accuracy of better than 1 part in 108. This is a minor applica— tion of NMDR and the values of 7(14N) / 7(1H) and 7(15N) / 7(1H) obtained from the ammonium ion were reported by Anderson at al. (143). Baldeschwieler (144) later reported slightly more accurate values for these and Baker (176) reported a value for y(l4N) / y(lH) in a— greement with that of Baldeschwieler. Now we come to the most sophisticated application of NMDR, the determination of relative signs of coupling constants and, very closely related to it, the determina— tion of the energy level arrangement of the spins. 9‘). {l (I'D h :1- 132 A positive coupling constant is defined as re- sulting from an interaction which minimizes the spin coup— ling energy when the two nuclear spins are antiparallel and, conversely, a negative coupling constant results from an interaction minimizing the energy when the spins are parallel (174). First—order spectra remain invariant to sign reversal of coupling constants and, as a consequence, spectra calculated theoretically with various sets of relative signs will all have the same appearance. Second- order spectra, on the other hand, have a quite different appearance for each set of relative signs and so theoreti- cally calculated spectra can be used to obtain relative signs. Now we see the real power and utility of NMDR, as it can be applied to obtain relative signs of both first- and second-order spectra and, thus, is the only method of obtaining relative signs for first—order spectra. .Absolute signs are usually obtained from NMDR by relating all signs of the coupling constants to one which is known absolutely and the reference is 1{NIH-13¢) which has been calculated theoretically by molecular orbital theory (130) to be positive. The technique may be explained very briefly with an example. Consider the three—spin system ALX in which all three types of nuclei are spin coupled with each other 133 and all couplings are resolvable, also the chemical shifts are large so that no multiplets overlap. Irradiation with H2 (Case (2)) of the low field line of A can have two effects on the L multiplet. If the low field lines of LX multiplets change in intensity, then J(AX) and J(LX) have the same sign, as double irradiation affects those A and L nuclei connected with the same X spin state. On the other hand, if irradiation of the low field line of A had produced perturbations in the high field LX multiplets, then 3(AX) and J(LX) are of opposite sign. Because lines appearing in the resonance spectrum are due to transitions between energy levels of the spin states, and as we have just seen here, resonance lines arising from spin states in common can be detected, the application of NMDR to the construction of an energy level diagram is straightforward. Recall also, that use of low H2 power, Case (1) with either spin tickling or INDOR, can dis— tinguish whether the energy levels in common are progressive or regressive. More details on the methods of relative sign determination and energy level diagram construction can be found in the two reviews (174 and 179) and in a short note by Friedman and Gutowsky (196). Kowalewski, de Kowalewski and Ferra (175) used 134 homonuclear INDOR, 1H—{1H}, to construct the energy level diagram and prove the relative signs of the coupling con- stants in a substituted pyridine. Using the same tech- nique, Baker (183) performed 19F-{19F} decoupling experi- ments on the compound Igggg-CClFBrCFBrCF-CFCl for the same two purposes. McFarlane (160) determined the absolute sign of 3J(1H-15N) as negative by relating all signs to 13(1H-13C), obtaining the relative signs by lH—{13C} and 1H—{15N} NMDR experiments on acetonitrile with 96% 15N en— richment. In a similar manner, from lH-{13C} and lH-{14N] experiments, he (149) obtained the absolute signs of 23(1H-14N) and 33(1H—14N) in methyl-, sthy1-, and t-bu- tylisocyanides and in tetraethylammonium hydroxide. By means of 1H-{lsN} decoupling, Bourn, Gillies and Randall (154) were able to obtain the relative signs of 2J(1H-15N) and 4J(1H-1H) for N,N-dimethylformamide-lsN. From the relative signs of 3(1H—15N) and J(1H-1H) in other form— amides,and from the knowledge that 1J(1H-15N) is absolutely negative, they were able to assign the absolute signs of all the other coupling constants. The relative signs of several proton-nitrogen coupling constants were determined by lH—{lHJ NMDR. Bourn and Randall obtained the relative signs of all the 135 3(13-15N) values for formamide-15N, N-methylformamide-15N (162) and N,N-dimetrylformamide-lSN (151). Ohtsuru and Tori (152) found all the 14N~-vinyl proton couplings to be of the same sign for trimethylvinylammonium bromide. 2J(1H-14N) was demonstrated to be opposite in sign to 3J(1H-14N) for ethylisonitrile and tetraethylammonium iodide by the frequency sweep NMDR, Type (b) Case (1), ex— periment of Maher (150). Recently Tori, Ohtsuru and colleagues (170) reported the absolute signs for 2J(1H—15N) and 3J(lH-15N) for quinoline-15N, its ethiodide and its N-oxide showing that complexing the nitrogen and solvents not only change the magnitude but also the sign of 1H—lsN couplings. A summary of the applications of NMDR which were discussed above will now be given. First, a spectrum can be simplified by removing couplings obscuring lines which it is desired to measure. Second, definite proof of which nuclei are coupling and, hence, definite proof of struc— ture can be obtained. Third, by completely decoupling and collapsing a multiplet, chemical shifts may be determined, since this is the irradiation frequency necessary for optimum decoupling. Fourth, the frequencies from hetero- nuclear decoupling provide very precise ratios of 136 gyromagnetic ratios, because electronic counters are available which can measure frequency to ten or twelve significant figures. The fifth and most sophisticated application is the determination of relative signs of coupling constants and construction of the energy level diagram. III. EXPERIMENTAL A. Purification of Compounds For the 14N spin-lattice relaxation time experi- ments alkylammonium chloride salts (lMe3H, 2Me2H, 3MelH, EtMeZH, lEt3H, 2Et2H, and 3EtlH where lMe3H = CH3NH3+ etc.) were obtained as reagent grade chemicals from the Eastman Kodak Company. Although Heilbron (197) recommends re- crystallizing these salts from ethanol or ethanol-ether mixtures, absolute ethanol-isopropyl alcohol was found to be a better solvent system. In most cases one recrystal- lization was found to be adequate but for a few compounds two or three recrystallizations were necessary. In one case recrystallization proved to be detrimental and at- tempts to purify the ethylmethylammonium salt (EtMeZH) seemed to hasten decomposition since the material became very dark and an oil formed. Other methods of purifying this salt also proved futile and, because its purity was in doubt, a complete relaxation time study was not made on this substance, The recrystallized products were then ”file on“ my- 9.- 0 o. r IE. l'flr ‘Hh Rev- D o. ‘u. 138 stored in an evacuated desiccator over P205 for several weeks prior to making up the aqueous solutions and obtain- ing the spectra. Solutions were acidified with conc. HCl prior to final dilution and pH measurement. pH values were measured with a Sargent Model LS pH Meter after cal- ibrating with a standard pH = l 0 solution. The final con- centrations and pH values are given in Table 9. Table 9. Concentrations and pH values for solutions of substituted ammonium salts used in this work. Compound Conc. _1pH, Compound Conc..._pr 4H 5.04 M 0.2 EtMeZH 5.00 M < 0.5 1Me3H 5.11 < 0.5 lEt3H 5.00 < 0.5 2Me2H 5.00 < 0.5 2Et2H 4.80 0.4 3MelH 4.80 0.4 3EtlH 4.07 0.8 4Me 3.23 -- 4Et IVB.OO ___ 2,2,2-Trifluoroethylamine hydrochloride (TFEA-HCl) was prepared essentially by the method of McKay and Vava- sour (198) and Bissell and Finger (199). 59 of trifluoro- emhylamine (Pierce Chemical Co.) was dissolved in 30 ml of ether. Dry HCl gas was slowly bubbled through the solution ina 2 X 50 cm "test tube," which was cooled by an ice bath. 139 After an hour bubbling was stopped and the ether decanted from the white precipitate. The product was washed with fresh ether, recrystallized from ablolute ethanol-ether and stored ig‘yggug over P205 for several days. A 4.36 M aqueous solution was made up and acidified (pH < 0.0): however, the TFEA'HCl is such a strong acid, pKa = 5.6 (200), that rapid proton exchange is still occurring and no sepa- rate NMR signal for the NH+ group can be detected. At— tempts to reduce proton exchange by dissolving TFEA-HCl in conc. HCl proved unsuccessful as the TFEA°HC1 is much less soluble in conc. HCl. Benzylamine hydrochloride was prepared from benzyl- amine (Eastman Kodak Co.) by the method outlined above. It also was recrystallized from absolute ethanol-ether and stored under vacuum over P205. Here again the solubility in water was so poor that the signal from the aromatic hy- drogens could barely be seen. The ammonium salts (chloride, iodide, nitrate, sul- fate, acetate and formate) were high purity reagent grade chemicals and were used without further purification. As a precaution these materials were stored for a period of several days in a vacuum desiccator over P205, prior to use. 140 Ammonium trifluoroacetate was prepared by slowly bubbling gaseous ammonia through a 50% aqueous solution of trifluoroacetic acid in a test tube which was cooled by an ice bath. After several hours, the ammonia was stopped and the excess ammonia and water removed 13 gaggg. The salt was recrystallized from absolute ethanol-isopropyl alcohol, then stored over P205 in a vacuum desiccator for a few days until ready to use. Most solvents were used without special preparation as the reagent grade materials were pure enough for NMR purposes. Special care was taken in a few cases as listed below. Glacial acetic acid and trifluoroacetic acid were dried by adding a few drops of the corresponding anhy- drides just before preparing solutions in these solvents. Dimethylsulfoxide (DMSO) was stored over Linde Molecular Sieve 5A for several days. It was shown to be quite dry by scanning the NMR spectrum at high gain when no proton signal from water protons could be seen. On adding one drop of water per 5 ml. of DMSO the OH reso— nance could just be detected. Commercial nitromethane was purified by the pro- cedure of Clarke and Sadler (201)followed by refluxing 141 for 24 hours over BaO under a nitrogen atmosphere. The nitromethane was then distilled from the reflux flask and the fraction boiling at 97-980C/750 mm. was used. All the solutions of the ammonium salts studied were saturated, as solubility was usually poor. Excepuons were 100% sulfuric acid, which was made 2.5 M with ammo- nium sulfate, and aqueous ammonium chloride which was made 5.0 M. A few drops of conc. HCl was added to the latter solution before dilution to the mark in a volumetric flask and the pH was measured and found to be 0.2 whereas an unacidified 4.99 M solution of ammonium chloride had a pH of 4.22. B. NMR Spectrometers The two NMR spectrometers used in the course of these investigations were the DA-60-IL and the HA-100, both manufactured by varian Associates (202). Both of these instruments had the V-4354-Internal Reference Field- Frequency Stabilization Unit, which permits field or fre— quency sweep. In addition, homonuclear double resonance may be performed with this unit provided that an audio- frequency (for wz) is supplied from an external oscillator (either a Hewlett—Packard Model 20000 or Model 200ABR 142 Audio Oscillator was used for this purpose). Details of the operation of this unit were given by Freeman (203a) and systems similar to it have been described in the literature (203b—f). Both instruments were located in an air—condi- tioned room and the 110 VAC to each console was main- tained constant with a Stabiline Voltage Regulator. The HA-lOO was a standard instrument. This parti- cular one had an electromagnet with 12" diameter polepieces, the tapered pole caps allowing a magnet gap of only 0.75": a Varian V-2100 Power Supply provided regulated current and a V-3506 flux stabilizer gave further stabilization of the magnetic field. An audio frequency counter (202) was sup- plied with the instrument. A V-4343 Temperature Control Unit was used with the V—4333 NMR Probe for variable tem- perature experiments. The DA-60-IL instrument had been converted from a DP-60, which in turn had been converted from an HR-40 spectrometer, and both energizing coils on the V-4012A Electromagnet had been replaced. This system contained a V-3521 Integrator and Baseline Stabilizer Unit and as a result required an attenuator when using the V-4333 NMR Probe. For homonuclear decoupling and variable temperature work, the V-4333 NMR Probe (tuned for 60 MHz). which was designed to be used with the V—4354 "lock box," 143 the V-4343 Temperature Controller, an external audio oscillator (either HP Model 20000 or 200ABR), and an exter- nal audio counter (either HP-Model 521A or 5245L electronic frequency counter) were used. For heteronuclear decoupling work, a specially modified V-4331A NMR Probe (containing two transmitter coils), an NMR Specialties SD-60/EC-60 unit, an audio oscillator (HP-200CD or ZOOABR), an audio counter (HP-521A or 5245L), and an additional cathode ray oscilloscope (CRO) were used. Details on the operation of these follow (III C and III D), A varian C—1024 time averaging computer, was used to smooth out the noise on samples where the NH protons gave a broad signal which could be seen as a gradual and reproducible drift above the baseline (e.g. 3MelH, 2Et2H and 3EtlH). For this'situation, only 10 to 20 scans of the C-1024 were necessary. C. Homonuclear Decoupling- Variable Temperature Experiments The internal reference system operates by locking the magnetic field and the radio frequency with a signal obtained from some peak in the spectrum of the sample. For this purpose hexamethyldisiloxane was used. Since it is immiscible with the aqueous solutions under 144 invesflgafion, it was contained within a sealed capillary tube which was held concentric to the NMR sample tube by means of two Teflon plugs. The capillary of disiloxane was sufficient to generate a strong lock signal: in addi— tion, disiloxane had a boiling point well above the high- est temperature used. It should be mentioned that other methods of generating a look signal were attempted. For example, the water soluble sodium 3-(trimethylsilyl)- propyl sulfonate was tried but the lock signal was so weak that the system kept jumping lock to the alkyl or water peak in the spectrum. Locking on the solvent water peak was also tried, but its lock signal was so intense that noticeable sidebands fell on the peaks of interest. Homonuclear decoupling was carried out in the fre- quency-sweep mode for the reason given in Section II E. Since the varian system has some convenient features about it which make decoupling particularly easy, these will be pointed out. The first step in the homonuclear decoupling-variable temperature experiment was setting the V-4343 at the approximate temperature and making a rough check of tmnnperature with the appropriate calibration standard (ethylene glycol at high temperatures, methanol at low temperatures) 145 and the chart supplied with the V-4343. Next, the sample of interest was placed in the probe and allowed to reach thermal equilibrium for about 20 min.; since the acidified aqueous alkylammonium salts are conductors, leakage had to be adjusted with the probe paddles at this point. The third step was to scan the complete spectrum (sweep range 500 Hz or 10 Hz/cm,.gg;-Figure 5, Section IV.A). The recorder arm was mechanically linked to the sweeping frequency (wl), hence the recorder was positioned on one side of the alkyl multiplet and the selector set to D1, or difference one, position. This is the difference between sweeping frequency (all) and locking frequency (.33), ,i.g., Dl - wl-w3, and its value was read on the audio fre- quency counter, and noted on the spectrum. The recorder was then moved to the other side of the multiplet and D1 again read, noting its value on the spectrum. Having done this, the spectrum was then spread out (;,g., a narrower sweep range, 250 Hz or 5 Hz/cmpwas selected) and the NH+ peaks moved onto the chart paper LgfL-Figure 6, Section IV A). Since the alkyl multiplet could no longer be seen. the importance of the D1 values read previously is now apparent. The fourth step was the actual decoupling to remove the fine structure resulting from coupling with 146 alkyl protons which was present in the NH peaks. This was accomplished in the following manner. An audio frequency from an external oscillator (m2) was introduced by switch- ing to SPIN DECOUPLE mode while simultaneously adding about lOdb more RF power on the 60 MHz RF Unit in order to restore power to all sidebands now that an additional one had been added. Now the selector was set to D2, or dif- ference two, position, which gave the difference between the external (or decoupling) frequency and the locking fre- quency,.i.g., D2 - m2—w3. The external oscillator was varied to obtain a value of D2 corresponding to one of the values of D1 read before and the NH spectrum scanned. Next the external oscillator was changed, so that D2 approached the other value of D1, and the NH spectrum again scanned. Again the external oscillator was changed with D2 approach— ing the other Dl value, the NH spectrum scanned, and the whole process repeated until D2 reached the second Dl value. Thus, the irradiating oscillator (an read by D2) slowly marched through the alkyl multiplet and we detected this by observing the NH multiplets ngé Figure 10. Section IV A). The NH multiplet patterns show sharpening as m2 approaches the center of the alkyl group, should be sharpest when m2 is at the center, and broaden out more as 147 wz approaches the other side of the alkyl multiplet. By looking for the sharpest NH group, the optimum frequency for decoupling could be determined. The external oscillat— or was then set to this value by reading D2. Now the am- plitude of the external oscillator was varied, each time scanning the NH group. This was done until no further sharpening in the NH group could be detected on increasing the amplitude. The correct decoupling amplitude, H2, had then been achieved, and the spectrum contained only the nitrogen single-bond proton couplings. The fifth and final step was to get an accurate measure of the temperature and this was done by putting the appropriate temperature cali— bration standard in the probe. After thermal equilibrium was reached, the peak was recorded and D1 values on each side measured. At a later time the position of this peak was calculated. The temperature was then calculated from the appropriate linear equation listed below since the calibration curves supplied with the V-4343 were found to be in error (204): TEMP(°C) - 191.7—l.076 (AMEFRQ) TEMP(°C) - 191.9—1.017 (ETGFRQ) where AMEFRQ is the separation between the two methanol peaks in Hz and ETGFRQ is the separation between the two 148 ethylene glycol peaks in Hz. As these two equations were derived for a spectrometer frequency of 100 MHz, the cor— responding values of peak separations obtained from the DA-‘O-IL had to be divided by 0.6 before using them in the appropriate equation to obtain the temperature. Correc- tions similar to this have been noted by others (205a) and other NMR I'thermometers" have been used (205b and c). A 36:1 gear reducer was installed on the external audio frequency oscillator, which was used to generate the irradiating frequency m2. With this attachment m2 could be changed in le increments: thus, “marching" across the alkyl group was under much better control than using an unmodified oscillator. From the foregoing description, it can be seen that a number of hours were consumed in obtaining one good de— coupled spectrum. The spectrum then had to be analyzed or decomposed to yield the usable parameters, coupling con— stants, chemical shifts, and relaxation times. Proton— proton coupling constants were obtained directly from the spectra, utilizing calibration markers obtained from Dl readings. These coupling constants were obtained from spectra in which the peaks for the alkyl groups were spread out, §.g., a 50 Hz sweep width or 1 Hz/cm (cf. 149 Figure 13, Section IV A). Some proton—nitrogen coupling constants were read directly from the spectra whereas others were obtained from the computer output. This pro- cedure together with that used for determination of the relaxation times, which also came from the computer fit is described separately (Section IV B). In cases where the 14N relaxation time was short, .i.g., the NH triplet in the proton spectrum was coalescing or was a single broad peak, decoupling of the alkyl pro- tons had no observable effect. This is reasonable, be— cause no multiplet from coupling of the alkyl group with the NH protons could be detected throughout the temperature range. This was found to be true for 3MelH, 2Et2H and BEtlH. However, what was saved in decoupling effort had to be spent in time averaging. Since there are fewer hydrogens on the nitrogen, and also the concentrations are lower, the signal/noise ratio for the NH+ group peak was poor. As a consequence, the intensities from spectra of this type would be of dubious value for comparison with computed spectra and it was necessary to use the Varian C- 1024 to obtain smoother line shapes. The C-1024 was used for 3MelH and BEtlH when scanning on the DA-60—IL spectro- meter and for 3MelH, 2Et2H, and BEtlH when scanning on the 150 HA—lOO spectrometer, since the signal/noise ratio was poor- er. Generally 15-20 accumulations in the C-1024 were sufficient to smooth the curves. D. Heteronuclear Decoupling Experiments 1H- For the purposes of heteronuclear decoupling, {14N}, a Spin Decoupler Unit, Model SD-60 (206) was em- ployed. This unit operated at a nominal frequency of 4.33 MHz. Initial decoupling experiments were made by double tuning the V-4331A NMR Probe transmitter coil to oscillate both at 60 and 4.33 MHz. High 4.33 MHz power levels were required to decouple the ammonium protons, as can be seen from the magnitude of the coupling, lJ(1H-14N)ce 50Hz, and Equation (37). Consequently a large amount of heat was generated in the probe and this caused the leakage to vary,resulting in poor spectra. Another experiment with a second (4.33 MHz) transmitter coil and proper insulation, although not eliminating the problem completely, greatly improved the situation and showed the direction to move. Thus, it was decided to mount the second coil on a Dewar vessel concentric with,and in between,both the 60 MHz trans- mitter and receiver coils. Since care must be taken to insulate the system properly, it was decided to construct a complete variable temperature system and thereby be able 151 to carry out variable temperature-heteronuclear decoupling experiments. A number of variable temperature systems have been described (207a—e). Two of these designs (207C and d) were abandoned because the fabrication was too involved and they were for wide-line rather than high—resolution work. Two others (207a and b) were rejected because the leads from the receiver coil are twisted when it is ro- tated and rotation is necessary to obtain minimum coupling to the surrounding transmitter coils. The design of Shafer (207e) was selected as most suitable because the receiver coil, leads, and female BNC connector are all attached to a tube and rotate together. The seal between this receiver coil insert and the Dewar vessel is made by a closely fitting Nylon bottom piece and gaskets (cf. detailed drawing in the upper right hand corner of Figure l). The 60 MHz receiver coil insert was made by mounting the coil on selected 8 mm O.D. quartz tubing cut 11 cm long. The tube was scratched on one end while three 2 mm dia. holes were burned through about 8 mm from the other end. The tube was then cleaned by washing successively with hot trisodium phosphate, water, aqua regia, water, concentrated aqueous ammonia, water and then oven dried. It was placed in a wooden jig which held it upright and 152 wrapped with uninsulated #30 copper wire (oxygen free, high conductivity) (three turns for a 60 MHz coil). One end of the wire was tied to a rubber band, which was then stretched around the back of the jig and tied with the other end of the wire so the wire was held under tension. Using Nylon thread as a spacer, the turns were moved close together and Craftsman epoxy cement (Sears, Roebuck & Co.) was applied and allowed to set for 4 hours, until the resin was tacky. The spacer thread was then removed and a very light coat of additional epoxy applied which was allowed to set overnight. The two ends of the coil were measured for proper length, cut, bent down, and one soldered to the outside "dimple" of a machined female BNC (UG-290/U). The tube and BNC were removed from the jig, carefully bending the lead over, and the other lead soldered to the center pin of the BNC. A Teflon "cup" bearing was inserted in the bottom end of the tube and epoxy smeared over the bottom of the ENC. The tube was put upright on the BNC and held in this position in the jig while the epoxy was allowed to cure overnight. It was then removed from the jig and placed in an oven at 70°C for an hour, after which the temperature was slowly (10°C per hour) raised to 120°C, in order to cure the epoxy cement. 153 The spacing between the windings is quite important. If the coil is wound tight without spacing, one can find two nulls on rotation but the paddles will have little or no effect on leakage adjustment. If the spacing is too wide, rotation nulling and paddle leakage control will be normal but such a coil has a poor Q and consequently the instrument will have poor signal-to—noise. After several failures, the best spacing was found to be the same as the diameter of the wire. The 4.33 MHz transmitter coil attached to the out— side of the Dewar vessel proved more of a challenge. After a few attempts to wind it with a continuous wire (insulated, #30 gauge copper) in the curved configuration, first on the Dewar vessel itself and afterward on a card— board chart-paper center piece, a very simple method was used. First, the transmitter coil was wound, one-half section at a time. A jig, consisting of two pieces of plexiglass with a 2.0 cm by 3.0 cm retangular brass spacer the thickness of the wire, was used for this purpose. One piece of plexiglass was notched so that epoxy cement could be applied in four places to the wound coil. The jig was held together by two bolts for easy dismantling after wind- ing the coil, which had eight turns (le). After both half sections were made, they were bent (on the 2 cm sides) 154 to fit the curved contour of the Dewar jacket. The half sections were positioned on the Dewar jacket held in place by rubber bands and the epoxy cement applied. After the epoxy cement set, one lead from each half section was soldered together on the side while the other leads were brought around the back underneath the Dewar jacket sideneck. Details about loop antennae (receiver coil), transmitter antennae, and characteristics of LC circuits are found in Terman's book (208). Another reference source for material of this general nature, as well as hints on construction techniques, is the "Radio Amateur‘s Handbook" (209). How- ever, for details about coil construction and characteristics specific to NMR the best source is Sontif and Gabillard (210). The Dewar jacket - 4.33 MHz transmitter coil was placed in the probe and one lead each from the 4.33 MHz and the 60 MHz transmitter coils were soldered to a ground lug inside the back of the probe. The other lead of the 4.33 MHz coil was soldered to one pin of the female'Twinax connector at the rear of the probe body. To the second pin of this Twinax connector was soldered the other lead of the 60 MHz transmitter coil. The Dewar jacket was sealed to the top and bottom of the 60 MHz receiver coil by means of gaskets cut from Silastic 55 (Dow Corning, Midland, Michigan). The Dewar jacket and receiver coil were held in 155 place by a threaded piece at the top, which had an ”O“ ring to hold the receiver coil and provided pressure to lock the Dewar vessel in place. Details of this can be seen in Figure 1, left hand drawing. A vacuum jacketed tube (Pyrex) was fabricated to hold the standard varian heater sensor element- The V—4343 Temper- ature Controller Unit was used to regulate temperature. The superstructure for holding the Dewar jacket, receiver coil and spinner housing (see Figure l) and the rig to hold the vacuum jacket and heater sensor (Figure 2) were machined according to blueprints kindly provided by Dro Paul Shafer (ZOTeL The complete assembly can be seen in Figures 1, 2 and 3, however certain details should be pointed out. The rear connection of Tygon tubing to the heater—sensor jacket, as well as the ball-joint connection between the front of this and the receiver coil Dewar jadket, were insulated by wrapping with foam rubber (see Figure 3). The inside of the probe was cooled by passing air through a polyethylene tube in the re- ceiver trimmer hole in the bottom of the probe (see Figure 3). This has a channel which goes into the receiver coil area (see detail, left-hand side of Figure l) ., Tygon tubing was attached to the exit ports mounted above the receiver coil top and be- low the spinner housing bottom (see left—hand side of Figure 1). One of these can be seen in Figure 3. The purpose of this is to lead exit gas from the probe away from the magnet. The lH—{14N} double resonance system coasisted of an 156 sou: (CM) _ A # # v 012.343 semen t§ ”W E\ \‘ . ‘vtzt a ~ ~23 6.1% , 4\ \ AIR FLOW ;/ ”our 81.48770 55 64W: I II” QUARTZ VACUUM JCKET \\ ‘S‘x>‘ g;f\>\\ \ i ./// I / ' " ‘// l', I 4..” ”NZ xm C OIL ,/ / / 60 ”HI REC COIL SLASTK' 55 6448““ Lil “ . \ . \ \\\ m— ALWUM | 00.1an TUBE ‘ 60 M92 IRECEWER xenon IE.” #854le / '1. Erma: g- A m \ \' \Q FIGURE 1. DETAIL 0F DENAR JACKET, 60 MHZ RECEIVER COIL INSERT. AND 4.33 MHZ TRANSMITTER COIL FOR 1H-14N DECOUPLING EXPERIMENTS. =3.— ‘vc-e __ 157 .3533: szmon—mzm .8 22:8; “2:32 wz_:o_._m mhzuxEumxu wz_._n_=ouuo $325.“.sz 53.5; mo... uncan— nmIon do .:<._.mn meuix qQ§onmz Baum omumaouamu manager IIoII me P e b D b sdnozs TKXTV go JeqmnN 180 '43 1Me3H .muaom EstOEEmatho on» now mumfirm Houefimnu cauoum m2 lEt3H lemme omega Hooaeoru oououo m2 2Me2H o.HI m.HI ‘ ¢ 3EtlH . o N 3MelH . I . .1 . EtMGZH Number of Alkyl Groups 2Et2H P .ha ousmfim 181 I F1 scale IlOppm R3N Relative Nitrogen Chemical Shifts O I'-' O N O U C b C Figure 18. Nitrogen chemical shifts 32, op values (Grim) for amines and substituted ammonium ions. The chemical shifts are relative within each series. Relative Nitrogen Chemical Shifts 182 -L RCN I scale 10 ppm. A l A j 0 10 20 30 40 a P Figure 19. Relative nitrogen chemical shifts and up values for nitriles and nitro-compounds. Relative Nitrogen Chemical Shifts 183 values for amides. T T T j ' I __l I l , I I l RCONHZ scale 110 ppm q I CH3CONHR I HCONHR 4% 10 15 20 25 a P Figure 20. Relative nitrogen chemical shifts and up 184 .4443E7V7' 677? W W .417WHLASZ2IM/CIJNKTfESZ2 .44fl3fi7V7‘ .551: M W 8. Cf; COONhII/N my CECOUH 5¢7H2' .AAABWWV7' CMEKRQCVWHIAN’AKR9CVT xWMEflEVV7' (MBZT 111 M £2C7§CZNQA#£ UNN£M4529 AME/ENT AME/EN? ENhICL /N DMSO F C/‘SCOO/W W Chi/VG. FIGURE 21. PROTON NMR LINESHAPES FOR AMMONIUM SALTS IN VARIOUS SOLVENTS AT SELECTED TEMPERATURES. 185 Y I I I T b - oosOoooooooooooOoooooooo.oo.oooo.oooooooo o........N.I:I4:I..Fvooc. (HTFAC) -ooooooooooooooooooooooooooo o co. (N114) 2804 (H2804) .n.unnunuunuu.uu NH4I (DMSO) + NHQCI (DMSO) ---- “----—..-_ ___ NHATFAC (DMSO) -_ -.-_ NH4NO3 (DMSO) 5H2 ,,,,,,, .”"",Known Relaxation ‘3 33 k p o ___‘_ Un nown rocess U) s ,2 ---°--- Known Exchange 0 4.) '44 " ~ - .H "- ~~ r: ~~ ‘~ .- U) ‘~- ~ F‘I -.°‘~~-~‘ NH4C1 (H20,pH=-0.2) 8 b s ‘ ~~ ~~ s g T T ‘ ~.o T T T ‘ " o ~-o_ -- . .s ‘ ~ ~~~ 9 U .‘ ‘~~..-~ NH4C1 (H20,pH=0.48) “ ~~‘.~“- ‘ ‘ S ‘. 3~~ ‘~--- mum (H20. Int-4.22) “.o~‘ -~~ o--- ‘ ‘ ‘ ~..NH4 Form (H Form) A L A l A 0 10 20 3O 4O 50 Temperature Change (0C) 3Figure 22. Change of NH proton chemical shift with tempera- ture for ammonium salts in several different solvents. 186 B. Data Analysis by Digital Computer In order to obtain a least-squares fit of the cal- culated spectrum with the experimental spectrum, estimated values for the nitrogen spin-lattice relaxation time (TWON) and the proton-nitrogen coupling constant (CUPJ) had to be supplied as input data. This was done by making a visual match of the experimental ammonium-proton spectrum (for optimum alkyl-group proton decoupling) with a set of theoretically calculated spectra. Program QUADRELX (APPENDIX A) generates a set of spectra for NH protons according to Pople's equation (9)_ for assigned values of?2 (ETASQ), where‘? = lOnTlJ. The areas from a set of such curves (a set of ETASQ values) are all normalized to the largest area and are plotted out for the same ordinate units and abscissa units. A set of such curves are shown in Figure 7, and the corresponding values of relaxation times for the two nitrogen-proton coupling constants are given in Table 10. 187 Table 10. 14N relaxation times for various J values. 72 =(10TTT1J)2 T1(14N) = mm T1(14N) = TW¢N ___:£EEEEL. {QIIH-14HIegg2g=so.0 lJ(lH-14N)=CUP.T=55.0 513.34 14.42 x 10-3 13.11 x 10'3 269.76 10.46 9.506 122.43 7.044 6.404 62.00 5.013 4.557 13.03 2.298 2.089 4.68 1.377 1.252 An extensive set of curves such as these was then used to visually match with the experimental alkyl-proton de- coupled ammonium-proton spectrum. From this visual fit, ETASQ was obtained and a relaxation time (TW¢N) calculated from it and the experimentally found (or estimated) lJ(lH-14N), called CUPJ. TW¢N and CUPJ obtained in this manner were used as initial values for the least-squares fitting of the experimental spectrum using prOgram NMR FIT8(APPENDIX B). The experimental spectrum is fed in as a set of frequency, intensity (X, Y) pairs. These values ‘Were obtained from the experimental alkyl-ammonium-proton Spectrum by the following procedure. First, the spectrum 188 with optimum alkyl proton decoupling was chosen; thus, it represented only the NH line shape for which POple's equa— tion was written (9). Second, a baseline was drawn be— tween the two flat portions on each side of the spectrum and, in cases where the right-hand side started to drift upward (due to the broad ”wing“ from the solvent water at 60 MHz), the baseline was made to follow this curvature. Next, the center of the lineshape was located and a verti- cal line drawn to intersect both baseline and curve. Using this centerline as a base, vertical lines were drawn at every 2 Hz interval for 80 Hz above and below the center- line giving a total of 81 frequency values (X) running from 0 to 160 Hz with 80 Hz as the center. Initially, only forty-three values were used but the fits were not as good. Now the intensity, intersection of vertical line with base- line and curve, was measured with a ruler to the nearest l/4mm. In order to insure a symmetrical shape and to re- duce measuring errors, corresponding intensities on each side of the center were averaged, e.g., Y at O was aver- aged with Y at 160, Y at 50 was averaged with Y at 110, etc. The spectrum thus digitized was then punched on IBM data cards. In addition to TWON and CUPJ mentioned above, the baseline (BSLN) and center (CEN) were input values. 189 The program then incremented TWON, CUPJ and BSLN to give the best fit of experimental and theoretical spectra. Normally, 26 loops of the set of three parameters was sufficient to give a good fit. Further loops through NMR.FIT8 gave changes in the fourth and fifth digits of the incremented parameters and little or no decrease in the sum of the squares of the differences between experi- mental and calculated values ($0). The output parameters sought were TW¢N and CUPJ. The values of these for the best fits are listed in Table 11 (Section C). Programs QUADRELX and NMR FITS were run on a Control Data Corpor- ation, Model 3600, Digital Computer. In addition to the above computer programs, the IBM QUIKTRAN library program LINFIT was used to get linear least-square fits to viscosity data (Table 15). From the slope and intercept of the plot of log against 1/‘1‘ for each compound viscosities Were calculated at each temperature for which the relaxa- tion times were known. These results are presented in Table 16. C . Data 190 am o.H mu,mm u 0>M mh¢.m Hw.mm on Hm mom 0 mm om cm 00 awe v no em on ma ovm.m oa.om oo-~m mmm m ma mm em 0H mma.m on em mo.m¢ omo.m ma.mm mo.m Amamsmv soucoeemasEUSEAue mo o.H oo.om “96 can m mm.mm mm.om own 0 mm mm ma.mm who.» we mm mmeo mmoam no em ma.m~ 5mm.o mm.mm mm.mm mom m we em mm ea omo o om.Vm on.am mam m CH mm we w Ammos~v souoosemasrooaflo me o + on em "mom oo.o4 me.¢m ma.o~ no 4H m0.¢m mo hm mm.oa eh.¢m ma.m~ on ma eh.¢m be me Noo.m mm.¢m mn.NH pom om Ha Nm me om vomiov .uva 94.5.0 um h0.¢m uommo AmmMZAV ancoeewamcuoaocoz .oommoaxae NEAzeaImihH Oodfioa . 0mm moaxas mm Azoalmiha Do .959 .mwmmLmOCHH CODOMQ BoaGOEEm mo mcauuaw woumowmlommma amusmeou mo mudsmwm .HH manna 191 .8. + . «were 88 88 88 988.8 88.88 88.88 888.8 88.88 88.88 888.8 88.88 88.88 888.8 88.88 88.88 888.8 88.88 89.8 889.8 88.98 88.88 888.8 88.88 98.8 Am8ummv 8888088888888888 88.8.H 88.88 “moo 888.8 88.88 88.88 888.8 88.88 89.88 888.8 88.88 88.88 888.8 98.88 88.8 888.8 88.88 88.88 988.8 88.98 88.8 1888888 888868888988888 . H. . not/m 88 8 88 88 889.8 88.88 88.88 88.88 88.88 88.88 888.8 89.88 88.88 98.88 88.88 88.88 898.9 88.88 88.8 88888.88 88 89.88 8088.88 888 888.8 88 98.88 8889.8 “888888 888co8888988888os 088 H N I . . 088 .8. I . 888x .8. 8:288 8888 Do 888.8. 888x .8. 83288 8888.8 Do 888.8. ~.8.o:ooe 88 88888 192 888.8 88888.8: 888.8 88.888 88.88 899.9 98888.8I 888.8 88.888 88.88 988.8 88888.8I 888.8 88.888 88.88 888.8 88898.8: 888.8 88.888 89.88 888.8 98888.8: 988.8 88.888 88.88 888.8 88888.8: 898.8 88.888 88.88 898.8 89998.8I 898.8 89.988 88.88 888.8 88888.8: 898.8 88.898 88.8 88828 98.88 88888.8I 888.8 88.888 88.98 88.88 88988.8: 888.8 98.888 98.88 888.88 88888.8I 888.8 88.888 88.88 888.88 88888.8- 988.8 88.888 88.88 888.88 88888.8I 898.8 88.888 88.88 888.8 88888.8: 988.8 88.888 89.88 .088 888.8 89888.8I 8Ixo888.8 8088.888 .8688.8 88828 Omm Az88v8ex8888 1288888 868 8Ieos\8888 soar 0o» .mCOH 8:8COEEM owusuflumnsm new wasumuwmewu £u83 OSeu scaumxmamu ZVH mo c08DM8um> .NH OHQMB 193 88.88 88888.8I 888.8 88.888 88.88 98.88 88888.8I 898.8 88.888 88.88 88.88 89888.8: 888.8 88.888 88.88 889.8 89888.8I 888.8 88.888 88.88 888.8 89888.8: 888.8 88.888 88.88 898.9 88888.8I 888.8 88.888 88.8 888.8 89888.8I 988.8 88.898 89.8 88888 898.8 88888.8: 888.8 88.888 88.88 888.8 88888.8: 898.8 88.888 88.88 888.8 88888.8: 888.8 88.888 88.88 898.8 88888.8I 888.8 88.888 89.88 888.8 88888.8I 888.8 89.888 88.88 888.8 89888.8: 888.8 88.888 88.88 .688 888.8 98888.8: 8:86888.8 so88.898 .8688.8 88828 00m 128888exooe8 8288888 868 8-8oe\8888 woe 0o» «.heucooV NM 88885 194 00N.N 00mmm.ml H00.m 0N.mmm 00.00 888.8 88889.8: 888.8 88.888 88.88 mmh.a Mbvmh.ml mmH.m mfi.vam mm.0¢ 500.a momhh.ml HmN.m mh.¢0m mm.Hm 888.8 88888.8I 888.8 88.888 88.88 888.8 89888.8: 888.8 98.888 89.8 0MN.H 00500.NI m00.m hm.th hm.¢ maumm mmm.m 0N¢H¢.NI mmm.N mm.vmm om.H0 wwm.m 00mmv.mi V50.m vm.mmm GH.Nm 888.8 88888.8I 888.8 88.988 88.88 mom.m m0¢0m.ml 00m.m 00.n0m Ah.0m 888.8 88988.8- 988.8 89.888 88.8 .Omm hmm.a N0000.NI HIMOOGm.m Moqm.mhm .UOGm.m Emumm Umm A2888888888 1288888 868 88.6.8.8888 goo coo 8.8.8oooe 88 88889 195 omo.a am.mwm| mmom.al OmmmJOI maumm mom.a mo.omvl nooa.al mnnm.OI mmumN HmH.N mo.onvl «88¢.OI mvmm.OI mmuma boo.m 88.8mwl Noon.0I v~mm.ou mamzm voa.m mm.mm¢l vmnm.0I mmmm.OI mmmzN .Hmux Hm¢.H mm.onI mmmm.OI mvmm.0I mmmza ~00x acumm macaw AmmouwucH .wwmoo .uuou vasomeoo .mumv mudumuwmemulweflu cofiumxmawu mo mafiuuflm mmumsvmlummmd mo muasmmm .MH magma 196 005.0 N.mm¢H omoo.¢l mmmm.o maumM 00¢.m m.mmaa thm.mi mwmm.o mmumm wmm.m N.mmm Namm.ml momm.o mmumH 8v8.8 0.888 8888.8: oooo.8 mamz8 fiwv.m 8.8mm mmmm.ml momm.o mmeN .Hmux v0m.m m.mmh Homm.ml oooo.H mmmza A.uwm>0 m: wQOHm AmmuumucH .Huou ocsomeou .mumo musumummemulmuflmoowfl> mo mCHuuflm mmumsvmlummwa mo muasmmm .va magma 197 8889.8 89989.8 888.8 8.888 8.88 8888.88 88888.8 898.8 88.888 88.88 8898.88 88888.8 898.8 8.988 9.88 88888 8889.8 88898.8 888.8 8.888 8.88 8888.8 88889.8 898.8 9.888 8.88 8888.8 88889.8 898.8 9.888 8.88 8988.9 99888.8 898.8 8.988 9.88 88888 8888.8 98888.8 888.8 8.888 8.88 8888.8 98888.8 888.8 8.888 9.88 9889.8 88888.8 888.8 8.888 8.88 88888 8888.8 88888.8 888.8 8.888 8.88 8888.8 88888.8 898.8 9.888 8.88 8888.8 88988.8 898.8 8.988 9.88 88828 8888.8 88888.8 888.8 8.888 8.88 8888.8 88988.8 898.8 88.888 88.88 8888.8 98888.8 898.8 8.988 9.88 88828 8888.8 88988.8: 888.8 8.888 8.88 9888.8 88888.8 988.8 8.988 8.88 808888.8 88988.8 898.8 808.988 .809.88 88828 no ® N 8808 moaxEK: Moe Oou 6:509:00 .mm8u8m00m8> c089580m umMSmmmz .mH manna 198 th.® 0H.Nma haom. vmowfid mom.N ovm.® hm.mva vwmfl. mmH¢.H v50.m MMH.© m0.m®H moam. mvmo.H NVH.M 05v.m Nm.NwH 0H05. mhma.m 00m.m Hm©.v m®.MHN 00mm. 0000.N 0H¢.m wm®.m ¢m.mmm mONN.H m¢®¢.m mmm.m ®N®.m mh.m5N Hoom.d hOHm.¢ mm®.m mHmEm 0HN.® 00.HNH mm0m. 0N00.H 0N0.m Nhh.h h©.mNH Gohm. NHON.H MMH.m hm0.0 mH.¢¢H Namfl. hh¢m.a 00H.m vm®.® mm.0mH 000m. HNmm.H omN.m ¢©m.© ma.hmH 00mm. m®N®.H 50m.m mmm.m mmohma Hmaw. hHmw.H mhm.m th.m hm.mmH Nmmh. ¢¢mH.N 05¢.m mmo.v 50.0HN mmmm. ¢00©.N Ohm.m ENOSN h0.HH m®.mm Noon. 0Nw®.0 ¢N0.m 0m.NH Hm.00 omfim. vmmh.0 NHH.M @00.HH H0.¢m thN. m000.0 HmH.m 0mm.0H hN.Hm damn. 0H00.H 50m.m mmm.0H 0m.®m omafi. mNNN.H mhm.m Nwm.m 0m.m0H hmmm. ¢Nm¢.H hm¢.m 888888.8 8:08888.888 xo\888888. 88 8988.8 8:80888.8 88828 0mm moaxfia 8Iomm 8.8.\H Mo\ao B\& 008 no & 8:80 B\ooo.n 6:859:00 .m08u8moomw> COHHDHOm Umumasoamo 6cm mmumu c08umwamH 008uumalcwmm .88 manna 199 888.8 88.888 8888.8 8888.8 888.8 888.8 98.888 8888.8 8888.8 888.8 889.8 88.888 9888.8 8888.8 888.8 988.8 88.888 8988.8 8888.9 888.8 888.8 88.888 8888.8 888.88 888.8 888.8 88.889 8888.8 888.98 888.8 888.8 88.888 8888.8 888.88 888.8 88888 888.8 88.888 8888.8 8988.8 888.8 888.8 88.888 8888.8 8888.8 898.8 888.8 88.888 8988.8 8888.8 888.8 888.8 88.888 8888.8 9889.8 888.8 888.8 88.888 8888.8 8888.8 988.8 988.8 89.888 8888.8 888.88 888.8 88888 888.88 88.88 8888.8 8888.8 888.8 898.88 88.88 8898.8 8888.8 898.8 888.88 88.88 8888.8 8888.8 888.8 889.8 88.888 8988.8 8888.8 888.8 888.8 88.888 8888.8 8898.8 888.8 898.9 88.888 8888.8 8888.8 888.8 888888.8 8|88888.888 xo\mo8888.8 88 8889.8 8uxo988.8 88888 8888888889 8.888898 8.888 9% 888 no& 8-8.8 .8.\8888 8568860 8.8.uzouv 88 88889 200 Table 17. Results of leaiX-squares fitting of data to the N).g§8‘W/T. equation 1fl1( Compound Corr. Coeff. Intercept SIOpe 1Me3H 0.9968 59.306 8758.57 2Me2H 0.9978 75.018 15127.91 3MelH 0.9902 104.278 11386.93 lEt3H 0.9947 53.636 7922.13 ZEt2H 0.9561 260.478 6381.02 3EtlH 0.9738 470.831 4286.60 Table 18. Relaxation rates calculated for three selected values of viscosity/temperature. Relaxation Rates l/Tl sec"1 (fiVT) .002 cp/bx .040 cp/ox .080 cp/oK 1Me3H 76.8 sec-1 409.6 sec.-1 760.0 sec-1 ZHeZH 105.3 680.1 1285.0 3Me1H 127.0 559.8 1015.0 lEt3H 69.5 370.5 687.4 2Et2H 273.2 515.7 771.0 3EtlH 479.4 642.3 813.8 Table 201 Proton coupling constants and chemical shifts for alkylammonium ions. Temp °C 3 J(D-L)a 3J(A-D) 6(L) 6(0) 6(A) 1216.33! 6.00 6.26 Hz -- 466.7 Hz 188.88 Hz .._ 30.35 6.17 -- -- 183.94 -- b 30.35 6.16 --- -—— —-— .._(4M) 40.50 6.14 -- -—— __- ___ 51.70 6.15 --- 465.48 186.19 -- ZMgZH 9.90 5.71 -- 507.3 189.92 ——— 34.00 5.68 —-- 505.7 191.40 -- 51.70 5.72 -- 505.6 193.51 --- 31‘1ng 4.25 5.23 --- —-- 205.82 -—- 10.00 5.14 --- 576.3 206.00 -__ 31.00 5.18 -- 575.1 208.20 -- «44825 5.20 --— 575.75 209.87 --- 51.50 5.12 --- 574.7 210.92 -—- .58.50 5.01 --- 576.6 212.16 —-- £329 30.35 —- —~- --- 218.90 --- aWhere L is the ammonium proton (N§+), D is the alkyl Proton a. to the nitrogen (+NCE) and A is the alkyl proton /3 to the nitrogen (+NCQE). bThis conc 4M all others 5M. Table 19 (cont'd) 202 Temp 3J(D—L) 3J(A8n) €(L) C(D) 5(A) EgMgZH 30.35 5.76 Hzc 7.31 Hz 509.7 Hz 188.40 --- --- --- --- --- 212.06 103.84 Hz JEEQH 0.60 5.93 7.33 476.3 209.2 102.42 9.50 5.92 7.27 475.6 210.2 103.16 20.30 5.91 7.30 475.1 211.3 104.20 31.00 5.98 7.39 475.5 212.2 105.35 34.00 5.90 7.26 474.0 212.6 105.65 41.00 5.88 7.34 474.9 213.8 106.50 51.30 --- --- 474.4 --- --- 60.00 5.91 7.34 474.3 215.4 107.90 2E§2H 5.00 --- --- 516.9 -—- --- 9.90 --- 7.32 -—- 210.37 103.49 10.00 --- -- 516.3 -—- --- 19.80 —-- --- 516.2 —-- -- 30.35 6.23 7.28 -—- 211.96 104.93 443.70 -- -—- 513.7 -—- --- 51.60 6.19 7.30 514.0 213.61 106.73 60.80 6.14 7.31 514.4 214.47 107.42 3Et1H 4.00 --- --- 561.17 --- --- 9.90 4.99 7.26 --- 219.48 107.78 10.00 --- --- 561.05 --- --- cWhere D is the Me. 203 Table 19 (cont'd) Temp °C 3J(D-L) 3J(A-tn 0(L) 6(D) 6(A) 19-75 --- --- 560.34 Hz -—- --- 31.00 4.92 Hz 7.32 Hz --- 220.44 106.2 34.00 4.97 7.29 560.16 220.51 106.6 38.75 -—— -- 558.30 --- --— 47.00 --- --- 560.90 --- -—— 57.70 4.92 7.45 560.20 222.39 108.37 48; 30.35 --— 7.27 --- 218.40 98.09 204 Table 20. Measured nitrogen and proton NH chemical shifts. Compound Abbr. Exper. 6(14N) Calc. ((1433) M1353) NH4+ 4H 0:0.23 ppm 0.00 ppm CH3NH3+ 1Me3H +1.6210.81 -5.2 ppm -0.246 (CH3)2NH2+ 2M62H -0.3510.81 -10.5 -O.898 (CH3)3NH+ 3MelH -5.7710.69 -15.75 -2,059 (CH3)4N+ 4M8 -21.0210.23 CH3CH2NH3+ lEt3H --14,7810.23 -10.75 -0.391 (CH3CH2)2NH2+ zstzn -31.18¢1.15 ~21.5 -1.060 (CH3CH2)3NH+ 3EtlH -35.10¢2.31 -32.25 -l.809 (CH3CH2)4N+ 4Et -42.9610.23 (CH3CH2)CH3NHZ+ EtMeZH -12.70¢1.15 -15 95 -0.978 205 Table 21. Nitrogen chemical shifts for substituted amines- Compound State 6(14N) Op Reference lhujuuwuuatlumuxzi CH3NH2 Aq NaOH 0.011 ppm 0 (118) CH3CH2CH2CH2NH2 7 -39.031 10 (118) CH3CH2NH2 Aq NaOH -15.0¢1 14 (118) csnsnnz EtZO -24.013 18 (118) IJq. -35.0 18 (119) NH3 IJq. +23.o:1 —- (118) +21.o -- (117) n 8,. Secondary Amines 20": (CH3)2NH I1q. +5.1 0 (117) Ag NaOH +2.0:2 0 (118) CGHSCH3NH EtZO -25.0¢4 18 (118) (CH3CH2)2NH Aq NaOH -3o.0:2 28 (118) Q" ngtgggx Amineg (CH3)3N IJq. +16.0 0 (117) Etzo +3.013 o (118) (CH3CH2CH2CH2)3N Et20 -l6.0115 30 (118) C6H5(CH3)2N Et20 -16.0¢4 18 (118) (CH3CH2)3N 11g. -24.0120 42 (118) MeNo2 -28.013 42 (112) 206 Table 22. Nitrogen chemical shifts for substituted ammonium ions. *— 4‘; ——v v—v— wwv—f Compound State 6(14N) Op Reference W + NH4 Aq 0.0011.0 ppm -- (117 118, _ and 119) Ag (C1 ) 0.0010.23 -- This WOrk CH3NH2+ Ag (01") +1.621-0.81 0 This Work CH3CH2NH2+ Aq (Cl-) -14.78:0.23 14 This Work + - C6H5NHZ+ A. (c1‘) 47.00310. 18 (118) Ag (c1’) -32.00 18 (119) W (CH3)2NH2+ Ag (of) -0.35:tO.81 0 This Work (CH3CH2)CH3NH2+ Aq (c1‘) -12.70¢1.15 14 This Work (c2113c112)21~1112+ Aq (Cl‘) -31.181-1.15 28 This Work 98W (CH3) 31m" Ag (01") -5.77:.tO.69 0 This Work (CH3CH2)3NH+ Aq (c1‘) -3s.10¢2.31 42 This Work 207 Table 23. Nitroqen chemical shifts for nitriles and isonitriles. W :8: Compound 6(14N)a UP A N' ile CGHs-CN -230-¢2.0 ppm 18 C6H5CH2-CN -221-j3. l7 CH3CH2-CN -2l7.1l- l4 CH3—CN —217.il. 0 (CH3)3CCN -215.11. 44 B. Iggnitgileg CH3-NC -136-10.5 0 CH3CH2-NC -151.3o.5 14 aReference 114. 208 Table 24. Nitrogen chemical shifts for amides. Compound 84fl54N1a 9’ AL Formamides HCONH-CH3 -89.482.0 ppm 0 HCONH-CHZCH3 -104.811.5 14 HCONH—C6H5 —118.512.5 18 B. Acetamides CH3CONH-CH3 —84.812.0 0 CH3CONH-CH2CH3 -100.2i3.5 l4 CH3CONH—(CH2)3CH3 -104.912.0 10 CH3CONH-(CH2)2CH3 -106.9r3.5 10 CH3CONH—C6H5 -110.282.0 18 CL_Jflxafléflflgflaiéfluégi CH3-CONH2 ~83.982.0 0 CH3CH2-CONH2 -80.782.5 l4 CH3(CH2)2-CONH2 -81.112 5 10 (CH3)2CH-CONH2 -80.7i3.5 27 C6H5CH2-CONH2 ~79.012.5 17 C6H5-CONH2 —71.412.0 18 aReference 194. 209 'Table 25, NitrOgen chemical shifts for nitro compounds. Compound 6(14N1a u? CH3-N02 -354.sio.5 ppm 0 CH3(CH2)2-N02 -364.5 10 CH3(CH2)3-N02 -364.5 10 CH3(CH2)4-N02 -364.5 10 CH3(CH2)5-N02 -364.5 10 C6H11-N02 -374.5 23 (CH3)2CH-N02 -378.5 27 (CH3)3C-N02 -384.5 44 C6H5CH2-N02 -362. 17 C6HS-N02 -346.5 18 alReference 125. Table 26. Measured viscosities for ammonium salts in various solvents. W “V“ Sglution Kinematic Viscosities 5.04 M Aqueous NH4C1 at pH = 0.2 0.948 881 (22°C) Satd. NH4OOCCF soln in dry CF3COOH 7.74 cSt (22°C) 3 2.5 M (NH4)ZSO4 in conc. 8250 44.94 cSt (22°C) 4 210 .88: 88 us 88 :8 808888 88088686 8888 vooaeasOU 888um usn .AEs can» uwmuscm m.m08 mm 8688868 860868-00 8.888 88 88840 882808880 .ousm 808888xum 08608 8688880 888 xumm :o8uaswneoo pecan 0:0 ouc8 cauoun Hxxonumu c883 maommanoo uwaa8uu Nmz no 83 ma.~nl8288lm8068 umHQ8uu oocmpsoum 0.888 mm AQE an on. mucw>808 nso8ue> s8 muHMm 6:8:0558 How 8885808 :22 no quEEsm .88 08889 211 88.88 Ihzwaumavha no.8 I 8\8 >8 .umamaua Nv‘mmm o.mm 08.88 IAzwaumHqu n.8 n 8\H >8 .8888889 m8.888 m.m8 Ana¢v 8:88.88 IAzwaumH.uH 8mm‘8 n 8\8 >8 .uaam8us 8a 88.888 o.o A8.oumavxmnqmz z wo‘m 580888 uxmvo 088 A88 8:8 m8 mummy 8m H M 8mlcf:wmavn8 hon AEOOUIV xlwm Gamcwm audnm muunm accoluwmz xuom vacunflsbc mumuuauu xuva uouuoum Amoco + wmzv x-om co8uucflneou 88088 Mu—CQEEOU “QHDHUDHUM Xflmm 088 8880888 88:8” 8.88m 824 08008 man Amy Uém Ho> Hdsuw + c808 m>on¢ m.oon gem AHV .caom .umm o.mmv Q58 vmzomUmmu .8888 ”888808 no mauve moooamu Hufiuuaw uaumnocoo 888:8 Ewan .mmm Auov mama an.ucoov 5N mHQMB 212 omU co Haemam o>umnno no: @8500 ¢mz0008mu omo co cumm uoncqm guy: mozwmz Hucmwm 0: 30am Btu acaum ON .omu co sown umamcflm xdflk 80¢mz 0:88 cocouuonn 30cm ecu madam om “Guam Hdcmflm uOHmCHm vaOOUmmU 8vIEuowouoazu Hacmam O>ummno no: 08500 mozvmz flat Hucmam o>uonno uoc 08:00 Huvmz £51. 888888 o>ummno 80: 88:00 888208886 854 8088888 880888800 888888888 8882088.80 8m 8.888 8888 828 880285 c fiddduuaduuflm. mucmEEoo housuusuum xnwm qumlucoo unanm Emnu .mmm Auov mama AU.uCOUV hm mHQIB 213 .88cm8w u>uumno was @8500 VOmNAvmzv £88 0888 8883 880888 “888cm88 Hmuw>08 a>uwmno 68:00 8888200088 36¢ 8888888 800808800 8.888 88 88 88.0mu8288-88088 0888888 888888088 888000880 8.888 888 88.8 n 8.888 88 88.88a 8\8 >8 8 8888888 88088 808882 8.888 888 88.8 n 8.888 88 88.8 n 8\8 >8 8 8888888 88088 8882 8.8.8 888 88.8 I 8.888 88 88 88.8 n 8\8 >8 8 8088888 88088 80888 8.888 888 88088 888> 8000880 8.888 888 A 0888 Momunnm uam 80080 0828 8883 880888 A m + «>088 8.888 858 8 8088888 888000880 88 8.888 888 80828 80 88088 888808888 8 888888 mucmEEOU swusuusuum xmwm uHMMIocou uwficm E050 «wmm 800v mews L 8%.»:OUV hm mHDwB 214 880 800 0 808.0 88. .0080 080.0 880 8008 0 8.88 088880808882 80880 08.8 88. 80080 888.8 880 .008 0 0.88 0808 088808 ADV AOON v m.m.v any Acnmv 88.8 Any Aoov v n.m¢ mnaxowaam Hanuofiaa 880 800~u0 0.8 880 80080 800.0 880 8008 0 8.8 .80 .0080 808.0 880 800080 8.8 8808080880 880 80080 888.8 880 8008 V 08.0 80080 08.8 880 80080 0880.8 800 8008 0 «8.0 888808880 0808 088008 axm Amov >uan00ua> 08fiacma unauucoo uauuuoaoan ucu>80m [ .mucm>AOm mo mmfiuuwmoum HflUHuhnm .mN OHQnB 215 880 08.0 80880 00.8 I 8&0 .mmm oucouomuu on human Afiv smacks» any uuouumH use Amy Aoomv Noo.H va AOmNV momm.o 880 80080 088.0 AHV AOOMV omh.o 880 80880 88.08 880 80880 88.88 888 880 808 888 880 808 Aoomv hm.om 80880 80.88 800 0 88.8 Aoomv m.mm AOOHV 0.0NH AOMNV 0.00H HounS o8u¢ owumUMOMOSHM8uB 080$ oausuazm Amuv huauouua> 08Enc>a 8:888:00 U8uuum8man qu>HOm Afl.ucouv mm manna V. DISCUSSION A. N-H Lineshapes in the Proton Spectra Of Substituted Ammonium Ions Contributions to Liggngdtns from Proton Exchange It is first necessary to show that proton exchange has no significant effect on the lineshapes in the proton spectra of the substituted ammonium ions in strong acid solution. Once this has been established the interpreta- tion of the observed lineshapes in terms of 14N spin-lat- tice relaxation will be possible. A very extensive review of proton exchange kinetics between alkylammonium ions,alkylamines and solvents, was presented in Section IIAL The purpose of that review was to examine the conditions under which proton exchange made a significant contribution to the observed lineshapes in the proton spectra of alkylammonium (or alkylamine) salts. It was pointed out that at low pH values (pH < 1.0) ex— change by any mechanism was described as "slow,' resulting in little or no observable effect on the NMR spectra of the alkylammonium ions. Another way of viewing this is that the residence time of the proton on the amine nitrogen is long on the NMR time scale for pH < 1.0. Indeed, the ex— change process is so slow that the technique employed to 216 217 study it was to start with deuterated alkylammonium salts (e g. R“199340 in acidified water, or alkylammonium salts in acidified deuterium oxide, and observe the change in either the NH+ or alkyl multiplet on slowly replacing the ammon- ium protons with deuterons or gigggygggg. This will now be demonstrated in more quantitative terms with two calculations relevant to the systems stud- ied in this thesis- Using the data and information of Grunwald, Loewenstein, and Meiboom (2), we begin by com— bining their Equations 15 and 16 to yield: Lvex = k16/'[H+] (39) where hvex is the contribution to the half-height width of the NH; peak from proton exchange, and kl6 is a rate con- stant defined by their Equation (17). Using their values of k16 at several listed pH values, the contribution of exchange to the width of the central NH+ line of aqueous 1Me3H can be calculated. Subtracting these results from their values of observed total linewidths (their Table V), the half—height width of the central NH+ line without ex- change is obtained and is given in the last column of Table 29. 218 Table 29. Calculated linewidths* for central NH proton of aqueous monomethylammonium ion PH LV 1/2 total Lv ex Av 1/2 without exchange 3.12 19.9 Hz 1.3 Hz 18.6 Hz 3.41 21.2 2.6 18 6 4.01 28.5 9.9 18.6 *Reference 2 Thus, the calculated half—height width of the NH+ central line without exchange is 18.6 Hz for 4.47 M 1Me3H at 19t2°C while we found a value of 18.0 Hz for the corresponding line in the spectrum of a 5.11 M solution of 1Me3H at pH 4 0.5 and 30°C. Our line width is approximately the same as that calculated for no exchange, so no appreciable exchange can be occurring: the small difference of 0.6 Hz may be due to either the difference in temperature,since the lineshapes are a function of temperature,or sample spinning, as Grun- wald g$.§1‘.did not spin their samples. The exchange broadening may also be calculated from the results of Grunwald, g§.g;‘. Substituting their values for kll = 22.6 x 10'3 sec-1M (their Table II) and p = 0.6 (their Table IV) for 4.00 M 1Me3H at 25tl°C, into their 3. For a solution Equation (17) we obtain k1 = 5.88 x 10- 6 in which [n+1 = 0.108 M (pH = 0.97) this yields Avex = 0.02 219 Hz using Equation (39). Therefore, for our 5.11 M 1Me3H at pH = 0.5 the calculated contribution to the NH+ line- shape from exchange is negligible at ambient temperature. Loewenstein and Meiboom (33) point out that the NH+ lineshapes for 2Me2H and 3Me1H are dependent on the 14N relaxation rates and that quantitative evaluation of the exchange broadening of these lines is impracticable. Since all of the exchange rate constants decrease for the methyl series,and 3Et1H values are smaller than 3Me1H (36 and 41) it is reasonable to assume that exchange broadening of the NH+ lines for the ethyl series at pH < 1.0 is also negligible. Thus, exchange kinetics has a negligible effect on NH+ lineshapes for solutions with pH < 1.0 at ambient temperatures with nothing said about other temperatures. Conner and Loewenstein, (21) made a variable-temperature NMR study of 1Me3H and 4H in order to obtain the activation energies for the kinetic processes, and showed that there was very little effect on these for 1Me3H but a larger effect for 4H. The significant thing is that they attri- bute the temperature effect on the NH+ lineshape to changes in 14N quadrupole relaxation rates rather than to proton exchange. For the specific alkylammonium salts studied in this 220 thesis we feel confident that, at the pH values used. the line shapes are primarily determined by the 14N spin- 1attice relaxation rates. In addition to the calculations given above, our experimental results are in agreement with this assumption. As can be seen in Figure 8, increasing the temperature for solutions of 1Me3H and lJ(lH-14N)) in the lEt3H not only causes the triplets ( proton spectra to sharpen, evidence of a 14N spin—lattice relaxation rate decrease (cf. Section IIB.), it also leads to resolution of the spin coupling with the alkyl protons (3J(lH—1H)) which appears as fine structure on each of the broad triplet components, evidence that pro~ ton exchange in these systems is negligible. Additional evidence for low proton exchange rates can be found in the fact that we observe the multiplet structure of the alkyl hydrOgens a to the ammonium group; these show multiplet structure from coupling with ammonium.nm1protons. It may be recalled that in all the kinetic studies reviewed in Section IIA exchange causes a collapse of these multiplets. The value of the ammonium proton-e alkyl proton coupling (3J(§NC§)) for each compound shows a decrease (analogous to collapse) with temperature (Table 9) which could be interpreted as resulting from an increased exchange rate at higher temperatures. However, one must exercise extreme 221 caution when considering changes in coupling constants as was pointed out in Section IIDy there are many things which contribute to spin—spin coupling and thus we may be observing a real change in 301§yc§) due to changes in the solute-solvent structure. In the various studies of proton exchange, changes in the lineshapes rather than only multiplet collapse were considered and all the a alkyl proton peaks show slight broadening with increasing temperature. Here again, one should not be too ready to attribute this to increased proton exchange, since it may be due to unresolved coupling between nitrogen and the protons a‘to it (23(1H—l4 N) < 0.5 Hz) as was mentioned in Sections IIB and D. As an exam- ple, spectra of the methylene hydrogens of the ethyl groups are presented in Figure 13 where it can be seen that there is broadening for 4Et where no exchange occurs but where coupling with 14N is possible. various authors have sug- gested that the broadening of the a hydrogen peaks is due 14N and this does appear like- to unresolved coupling with 1y; it also probably contributes to the broadening of the methyl group resonance (cf. Figure 5) in the Et series and experiments showing this are presented later, Section VB. As yet unexplained is the exceptional broadening for the methylene multiplet of ZEtZH. 222 Contributions from 14N Spin-Lattige Relaxation and We Having confidence new that the lineshapes for the ammonium ion protons (Figure 6) are governed mostly by 14N relaxation rates, the relaxation rate data obtained from the computer least-squares fitting of the observed spectra may be discussed. These are shown in Table 1,, Section IVC. Considering the values in Table 11, the most striking thing is the variation in 1J(1H-14N) with temperature. Since the majority of the values (2Me2H, 3MelH 2Et2H and 3EtlH) were not observed directly but obtained from the computer analysis we cannot really say whether or not they do vary with temperature. A complete discussion of this point is deferred until after discussion of the relaxation times. In Figure 6 the gradual collapse of the ammonium triplet may be seen. This results from the changes in the 1‘4N spin- lattice relaxation times caused by variations in electric field gradient from compound to compound: the changing sym- metry around nitrogen gives rise to these different field gradients, The values reported in Table 11 may now be compared with those reviewed in Section IIB, keeping in mind that most of those reported by others were estimates. Our value for 1Me3H is about one-half the 20 msec 223 estimated by Grunwald gtflal, (25). The value of Grunwald .g§.al. (25) seems a little high and ours about what it should be, as it falls in line with those reported by Anderson and Baldeschwieler (88) and by Swift.ggnal. (13) about 15.9 msec for liquid ammonia. As the ammonium ion multiplets for 1Me3H and lEt3H (Figures 5, 6, 8, and 9) are slightly more collapsed than the proton lineshapes for liquid ammonia of Swift gt,§l, (13) our T1 (14N) should be (and 18) less than the T1 (14N) BEtlH > ZEtZH > lEt3H. Considering only the effect of alkyl groups on the 14N chemical shift, it would seem reasonable that the methyl series should follow the same order, and indeed this is what we find, the order being 4Me > 3MelH > 2Me2H > 1Me3H again arranged according to low field shift. That the effect of the alkyl groups on the 14N chemical shift is mainly additive is experimentally substantiated as can be shown by arranging the three dialkyl ammonium ions in 243 order of decreasing low field shift 2Et2H > EtNeZH > 2Me2H. All of these 14N chemical shifts (relative to ammonium ion) are clearly illustrated in Figure 16. Operating on the assumption that the 14N chemical shifts are largely additive for the alkyl groups, one can use the equation of weighted averages of Maslov (120): :1 '1" " bi s E mjcj j where: 61 = chemical shift of the unknown molecule i tj = chemical shift of the known molecule j mj = number of bonds found in the molecule of type j which are also present in molecule i s = 1;}.mj= total number of bonds in molecule i along with the values 6(4H) = O, 6(4Me) s -21 and 6(4Et) = -43, which are chosen to give exact agreement,to make an empirical calculation of the chemical shifts. These calculated values are tabulated in column 4 of Table 20 and plotted as circles and dotted lines in Figure 16. Quite obviously, the experimental and calculated values shown in the figure and table do not agree too well, Although the effect of the alkyl group is additive, it does not fully account for the observed 14N chemical 244 shifts. Thus we must look further in order to explain the differences between the observed values and those given by the additive approximation. Earlier (Section II C) a number of theories for nitrogen chemical shifts were examined and their limita— tions in explaining or predicting nitrOgen shifts dis— cussed Here we will apply an empirical method which allows a fairly close approximation to any l4N shift of covalent compounds, once the chemical shifts for three or four compounds are known, In general there is a shift to low field on pro- tonation of the ammonia or amine nitrogen because the shielding effect of the lone pair electrons is removed by involving these electrons in bonding. This is shown by the large number of examples in Tables 21 and 22- Perhaps this is a proper place to point out the "bowing" of Figure 16 and offer an explanation. It can be noted that for the methyl series there is an upfield "bow" deviation from the dashed Maslov line, whereas for the ethyl series the opposite is observed, namely, a downfield “bow” deviation from the dashed Maslov line. The explanation for these lies in the observed amine shifts listed in Table 21 All of the methyl amines are at higher field than the 245 protonated forms,whereas for the ethyl series the nitro- gen resonances for free amines are at,or very close to, the values for the protonated forms. Thus, the methyl group exerts a stronger influence on the 14N shift than does the ethyl group, and the effect is to cause a slight shift to higher field for the protonated form- Further evidence for the greater influence of methyl relative to ethyl is noted in the dialkyl series where EtMeZH lies closer to 2Me2H than to 2Et2H (Figure 16 and Table 22)- Unlike Hampson and Mathias (194), who reported no correlation of 14N and 1H chemical shifts of amides, we do find a very obvious and significant correlation in the case of alkylammonium ions (cg; Table 20, Figures 16 and 17). This is due to the fact that both the 1H and 14N chemical shifts of the alkylammonium ions are governed by 5D with up as a slight perturbation. Comparing the 1H shifts with the 14N shifts measured for the same sample preparations there are some obvious differences. The 1H shifts, unlike the 14N shifts, show a distinct move- ment to low field with increasing alkyl group substitu- tion, in the order + NH4 < mono-alkyl < dialkyl < trialkyl Although the NH prOton is in between the nitrOgen and 246 solvent water, its resonance value is determined mainly by the nitrogen to which it is bonded (and indirectly by the alkyl groups acting on nitrogen) and only slightly by the much weaker hydrogen bonding to water. Hydrogen bonding shifts the NH proton resonances to low field while decreasing hydrogen bonding results in shifts to higher fields (94, 104 and 128). Accordingly, one would expect the ammonium ion and the monoalkylammonium ions to hydro- gen-bond most, the trialkylammonium ions least, so that if hydrogen bonding were the main factor determining the NH proton shifts, the result would be the direct opposite to that observed. Another possible explanation of the observed NH proton shifts could be proton exchange be- tween amine and water. variable—temperature 1H resonance studies described before (Section V‘A), however, clearly show that proton exchange is relatively slow and thus unimportant in this regard- The full explanation of the observed l4N shifts with alkyl substitution was first put forth by Witanowski .gtwal; (111, 113, 114 and 125), who suggested that in— ductive effects, either electron-donating (+1) or elec— trondwithdrawing (—I), were important. Here we are mainly concerned with alkyl groups which, as a class, are 247 electron releasing relative to hydrOgen,whatever the cause, The particular order of effectiveness of the various alkyl groups among themselves depends upon whether the dominant factor is inductometric or hyperconjugative (218)- we will now show that hyperconjugation explains the 14N shifts we report for the alkyl ammonium ions as well as a large number of 14N shifts previously reported by others. In addition, thiS‘Will allow the estimation of many nitrOgen shifts which have not been measured. Since there is a much larger wealth of shift data for 31F than for nitrogen, and since a larger number of chemical shift theories have been investigated, it was thought that a study of phosphorus chemical shifts could possibly be an aid in not only explaining the nitrogen shifts, but perhaps in allowing prediction of shifts as yet unmeasured. Of particular interest is the work of Maier (219) on primary phosphines and Grim and co—workers on secondary (220) and tertiary (220 and 221) phosphines and phosphonium salts (222). They obtain a linear re- lationship between the observed 31F chemical shifts and the sums of empirically determined paramagnetic shielding values, J a (see Table II,ref. 221). From these plots which are relative to the methyl compound as zero, 248 intercepts and slopes are obtained. The differences be- tween calculated and observed values are on the average + 1 ppm. or better and, in the worst cases, i 2.13 ppm). This is within the experimental errors in 31 P NMR spectroscopy, As the order of 14N shifts for various alkyl nitro- gen compounds seemed to follow the effect of hyperconju— gation, we thought there might be a correlation between the 14N shifts observed by us,as well as others,and the a P values of Grim (221). The results of the correlation of observed 14N shift with GP org»): are listed in Tables 21, 22, 23, 24 and 25 and plotted in Figures 18, 19 and 20. Although the data are sparse, the correlations appear to hold. In addition to explaining the nitrogen chemical shifts in the amines and ammonium ions, the chemical shifts in the alkyl nitro compounds of'Witanowski gtualfi_(125), and the N-substituted formamides and aceta— mides of Hampson and Mathias (194), are also accounted for Comparing the results of the correlations for the 14N chemical shifts with those for 31p (219-222), it can be seen that the accuracy and precision is much better for the latter. One reason for this is that 31F chemical 249 shifts are two to three times larger than nitrogen shifts thus subtler effects such as hyperconjugation will be more apparent in phosphorus resonancesthan in nitrogen reso- nances. A second reason, which may be the explanation for the first, is that phosphorus has the possibility of using d orbitals in bonding which nitrogen does not have. A third reason arises from the nuclear properties. 31F has a natural abundance of 100%»and this isotope has a spin of 1/2, so will always yield signals which are quite sharp and lead to only small errors in measurement (about i 1 ppm): 14N, on the other hand, has a natural abundance of 99 6%_ a spin of l, and a quadrupole moment associated with it. As a result, the linewidths of the nitrogen resonances are a function of the electronic structure and will yield a very broad signal when the electric field gradient at nitorgen is very asymmetric. As a result the nitrogen chemical shifts may have a large error associated with them , as is the case for a large number of compounds listed in Tables 21-24. Enrichment with the isotope 15N I = 1/2, is very expensive and so far very few 15N shifts have been reported. Unfortunately not enough data for 15N enriched compounds have been reported to complete the alkyl series discussed above; such data are needed to 250 provide another test of the extent to which hyperconju- gation correlates with the nitrogen chemical shifts. Considering the nitrogen chemical shift data in Tables 21—24, and the graphs of the 14N shifts fl.i:\6:, it can be seen that the best correlations, yielding the largest lepes, are obtained when the electron-releasing group is attached directly to nitrogen (e.g. amines, ammonium salts, nitro compounds, N-substituted amides, and possibly the isonitriles). Furthermore, their slopes are always negative. However, when the electron-releas— ing group acts through an intervening atom or group, as in the nitriles and the carbonyl carbon-substituted amides, the slopes are very definitely smaller and appear to be positive. Even though the lepes are small, it appears that the effect is present and measurable. Finally, it can be noted that, in almost all compounds, the aromatic ring causes a marked deviation, and this deviation is largest when the ring is attached directly to nitrogen, and somewhat less when acting through methylene as hithe benzyl group. This may be an anisotrOpic shielding effect caused by the w—cloud current,or some resonance form causing perturbations of the nitrogen p—orbitals. 251 In conclusion, we are suggesting again, as did Bose.gp.gli (119), that nitrogen resonances can be used to demonstrate electron-release through hyperconjugation- However, we feel that the effect is best shown if the functional group is attached directly to nitrogen rather than operating through an aromatic ring as in the com— pounds they studied. In addition, nitrogen resonance studies appear to afford a more reliable method of ob- serving electron—releasing or electron-withdrawing effects than many other techniques. Thus, measurements of basi— city constants,pr, for amines are often complicated by other effects like steric strain, as elucidated by Brown (42): C. Solvent Effects on Proton NMR Spectra of Ammonium Salts _C..h__s____fan es of lng-“m NW The main stimulus for the work reported in this part was to answer the question as to whether the varia- tions of lJ(1H-14N) with temperature, which we found from computer analysis of the complex lineshapes in the NMR spectra of alkylammonium salts, were really significant. The present study of the directly bonded lJ(lH-14N) 252 coupling itself constitutes a separate test from the study of the other coupling constant, 3J(§NC§) presented before (Section V'A). It was reasoned that the best species for this study would be the ammonium ion (4H), since it is com- pletely symmetrical and thus there would be no compli- cating effects on the spectra from 14N relaxation, pro- vided that proton exchange was minimized. Consulting the literature (Tables 5 and 8 of Section II D), it was learned that there was no agreement on the value of lJ(lH-14N) for the ammonium ion. The disagreement could be due to experimental error or it could be real, result- ing from the variety of anions and solvents used in the various investigations. Since none of these authors had noted this difference, a re-examination of the spectra of various ammonium salts in a variety of solvents became necessary. From Table 27 (Section IV C) we see that there is a measurable variation in 1J(1H-14N) with anion, solvent and temperature. For convenience, we list the values at ambient temperature in Table 30. 253 Table 30. lJ(lH-14N) for ammonium salts in various solvents.* M lJLlH-lflN) Salt W 51.0 Hz CF3COONH4 Saturated in dry DMSO 52.4 I 0.3 NH4C1 5.04 x in H20; pH = 0.2 52.9 HCOONH4 Saturated in 99+% HCOOH 53.9 (NH4)2SO4 2.5 M in 100% sto4 *Data taken from Table 27. These values may be compared with those reported in Table 5 (Section II D). The first one to consider is the value for the aqueous ammonium ion in highly acidified solution. The values of Meiboom.g§”a;& (18) and.Anderson ggggly.(l43) are lower than that reported in this work while the value of Emerson §£_§;;.(].9) is higher. These values are rather old and, as can be seen, our value agrees very well with the more recent measurement of Baldeschwieler (144)- The second value for consideration is that for ammonium ion in sulfuric acid. Our value agrees with that of Fraenkel gr Ela.(145) and both differ from the value b - for the aqueous ammonium ion. We also measured 254 1J(1H-14N) in 100% sulfuric acid solution to see if there was any change with acid concentration, and there does appear to be a change. The surprising thing we find is that all these values fall within such a narrow range (’V3 Hz) for such vastly different solvents. The point is, though, that there is very definitely a measurable difference in lJ(lH-14N) with variation of solvent and, for the same salt-solvent system, with temperature. To explain these variations would involve a more detailed study and is beyond the scope of this investigation. In the course of this search for variations in the lJ(lH-14N) coupling constant of the ammonium ion in various solvents, a number of other interesting effects were notedo A few of these are shown in Figure 21. In order to explain these lineshapes, variations of anion, solvent and temperature were made in a fashion similar to that of Randall g§“§l&,(80-82) and of Kintzinger and Lehn (65)- Thus, we found that a very simple nitrOgen compound, the ammonium salts could illustrate all the effects described in Sections II A and B_ whereas other investigators (65, 69, 71, 73, 78 and 79) had employed much more complex nitrogen compounds- 255 gegergl Effects on NH Lineshgpes The chief feature used in determining the ammonium salt/solvent interaction is the characteristic shape of the ammonium ion proton signal and the change in that lineshape with temperature. As was pointed out earlier, Roberts (8) was the first to suggest that variation in the NH proton lineshape with temperature could be used as a diagnostic method to distinguish between the effects of proton—nitrogen exchange and 14N relaxation- Pie noted that increasing temperature causes faster exchange rates and consequent collapse or coalescence of multiplets and, generally, sharpening of broad singlets (gfh Figure 7, going from F to A) when the NH lineshape is governed by exchange. The opposite behavior is observed when the NH lineshape is governed by 14N relaxation: increasing temperature then causes sharpening of triplets_ the appear- ance of a triplet from a broad peak or the broadening of a sharp singlet (g§‘.Figure 7, going from A to F). In considering a triplet or partially coalesced triplet, the diagnostic temperature test to distinguish between exchange and relaxation is quite clear cut- Howe ever, for a single line, observation of changes in line 256 shape with temperature does not lead to an unequivocal answer. One must therefore probe deeper into such things as the mechanism giving rise to the single line and con- sider whether the peak is a "pure" NH line or a I"composite" NH/solvent line. Experiments relative to this problem will be discussed below. Am oniu Ion in Ac'd'c oti Solvent At the outset we will consider acidic solvents since these were chosen first in order to reduce the effects from proton exchange. A, B and C of Figure 21 show the ammonium ion in three different acids at ambient temperature (’V30°C)- The spectra are dramatically different- If we consider viscosity changes as a possible explanation for the fact that the formic acid (HForm) ammonium ion resonance (C) is broader than the sulfuric acid ammonium ion resonance (A)_ or why the trifluoro- acetic acid (HTFAc) ammonium peak (B) is coalesced we see that the data do not support the hypothesis. From Tables 26 and 28, 7(HZSO4) > 77(HForm) 2 ’I/(HTFAc) and on this basis,with Equations (20) and (24), we would predict that 257 l/T1(HZSO4) > 1/T1(HForm) 2 l/T1(HTFAc) or T1(sto4) < T1(HForm) 3 T1(HTFAc). This would mean that the formic and trifluoroacetic acids would give approximately the same NH lineshapes and a sharper triplet than observed while the sulfuric acid would lead to a coalesced NH peak. Obviously this is an incorrect interpretation since the observed spectra do not agree with this order. Considering Equation (24) we see that the other term to consider is the field gradient. Thus, the ammonium ion in sulfuric acid must have a completely symmetrical electric field, which is true for a completely ionized salt; this would be expected for a solvent with this large a value of the dielectric con- stant (e = 100; 25°). Also judging from the well—re- solved triplet observed for ammonium formate, and the partially coalesced triplet for ammonium trifluoro— acetate, one would predict that free ions are present in formic acid but that some ion pairing is occurring in trifluoroacetic acid; these predictions are in agreement with the moderately high dielectric constant of formic acid (a = 58.5; 16°) and the much lower one for tri- fluoroacetic acid (6 = 39.5: 20°). The origin of the 258 slight residual broadening in the lines of each spectrum can be clarified by a temperature study. At ambient tem- peratures, the multiplet in sulfuric acid is noticeably broader that that in aqueous ammonium salts (pH = 0.2) and the viscosity data of Table 26, as well as the sharp- ening of the multiplet in sulfuric acid at 670 (lower viscosity) prove that the broadening is a viscosity effect on the proton lines alone. That the coalescence of the lines in trifluoroacetic acid at ambient tempera- tures is quadrupolar in origin (;.§. from the 14N) is shown by the sharpening of the lines with increased tem- perature (55°), as was discussed above. This change in relaxation time with increased temperature results largely from change of correlation time (’Q: ) with temperature. As Equation (20) shows, the correlation time is inversely proportional to the temperature and directly proportional to the viscosity. As a result of these proportionalities. increasing temperature itself and the resulting lower viscosity both operate in conjunction to cause a lower correlation time, as was pointed out by Kintzinger and Lehn (65). According to Equation (24), the lower cor- relation time means a slower relaxation rate and thus longer relaxation time (65, 92 and 93)- 259 Considering now the ammonium formate spectra variable temperature studies show that the broadened peaks at ambient temperature (rVBOO) result from exchange, At 67°C the ammonium ion triplet is completely merged with the carboxy proton peak to give the usual broad com- bination peak, whereas at 15° the triplet is much sharper than at ambient temperature. The remaining acidic solvent studied was glacial acetic acid and here the dominant process appears to be exchange, as has been shown by the extensive studies of Grunwald g£.§14_(23 and 24). Comparing the ammonium ion spectra in acetic and formic acids the major difference represents a difference in exchange rates. Since acetic acid is a weaker acid than formic acid exchange will be a dominant factor over a vast range of temperatures and concentrations for acetic acid solutions. Another factor which helps determine whether the ammonium ion proton peak will be a singlet or triplet in these two solvents is the major difference in dielectric constant. The di- electric constant will also influence the degree of ionization and thus the symmetry of the resulting species. The small dielectric constant for acetic acid promotes 260 ion association, as has been shown by Grunwald _e_§ a1. (23 and 29), and the ion pairs produce large field gradients at nitrogen leading to collapse of the ammonium triplet. A oni Ion ' Non—Acidic A otic Solv nts In interpreting the lineshape of the ammonium ion in other solvents, it is more often than not important to also consider the gegen ion. In this regard we differ 'with the viewpoint of Kawazoe‘g§,§lr (79) who consider the role of the gegen ion as being of minor importance- 14N relaxation However, they were considering changes in times of substituted quaternary piperidine salts and there the effect of the gegen ion may be reduced, Randall and co~workers (80—82) report no effect of solvent or gegen ion on lineshapes, and hence on 14 N relaxation times, in a series of quaternary alkylamine halides, which in turn was the one exception reported by Kawazoeugp.gl (79) as showing the effects! Our findings are substantially the same as those of Cocivera (31) who notes that the shape of the NH resonance for methylammonium salts is a function of the solvent and gegen ion. More apropos here, Grunwald and Price (29) report a great difference in lineshape for the ammonium ion in glacial acetic acid depending on whether the anion is acetate or chloride. 261 The answer as to why some workers report such solvent— counter ion effects whereas other workers debate the exist- ence of such effects, can be given by examination of Equa- tion (24). For all the cases mentioned above, there should be little difference in the correlation times, Tc, and thus we must focus on the quadrupole coupling constant, (eQ/h)(82V/822). Since we are talking about 14N in all these compounds, (GO/h) is constant and thus it is the field gradient which must be examined. As was pointed out above (Section IIB), the effect of the electronic environ— ment on the field gradient is short ranged, being a func- tion of (ri)'3, the distance between the i'th electron and the nucleus (9 and 57)- Thus_ as this distance grows large there will be a smaller influence on the 14N nucleus and hence a smaller influence on the relaxation time. For this reason, then, Kawazoe (79) and Randall (80-82) saw little or no effect from solvent and anion on lineshape, since their molecules were too large, the distances ri large, and consequently the influence on T1(14N) small- Also, for this reason, the effects were more obvious to Cocivera (31), Grunwald and Price (29) and ourselves- This is the main point here, and provides the main reason why the ammonium ion is such a good probe of solvent-solute 262 interactions . Another factor worth considering is the 1H—14N coup- ling constant since, for coupling over a single bond the values range from 44 to 70 Hz (Table 5), whereas the coupling across two and three bonds is only 0.2 to 5.0 Hz (Tables 4 and 7). Therefore, observation of changes for a proton one bond away give an advantage of from 8 to 120 over changes observable for protons two and three bonds distant. Bearing these things in mind, we now consider the ammonium salts in other specific solvents. In chloroform the very low dielectric constant dictates that any salt will be associated into ion pairs or higher aggregates. It is not surprising, therefore, that the trifluoroacetic acid salt (the only salt soluble enough for us to observe an ammonium proton signal) gives a rather sharp singlet be- cause of the field gradient produced by the associated tri- fluoroacetate ion. Nitromethane has about the same dielectric constant as trifluoroacetic acid and, as ion-pairing was given as the explanation for the coalesced ammonium ion peaks of ammonium trifluoroacetate in trifluoroacetic acid, it is quite rea— sonable to expect that ion—pairing will also occur in 263 ‘nitromethane. Quite unexpectedly, ammonium trifluoro— acetate in nitromethane (Figure 21F) shows a triplet less collapsed at ambient temperatures than that shown by the same salt in trifluoroacetic acid (Figure 213)‘ From this, one can infer that there is only a slight field gradient around the ammonium ion and most likely the structure con- sists of the polar nitro groups pointing toward the ammonium ion, the two oxygens of a nitro group appearing to the ammonium ion as being very similar to the two oxygens of the trifluoroacetate. Dimethylsulfoxide (DMSO) has a dielectric constant slightly higher than nitromethane and trifluoroacetic acid and slightly lower than formic acid. Indeed, conduc- tivity measurements show (101) that in concentrations greater than lO-ZM ion pairing predominates. From the concentra- tions we used, we would expect ion-pairing to be important and this is supported by the fact that the chemical shift of the ammonium ion is a function of its anion (Table 27). The shifting of the NH resonances for methylammonium salts seas noted by Cocivera (31) to be dependent on the anion in tflae solvent t—butyl alcohol (¢=12.471 25°) where ion pairs exist. The easiest ammonium salt-DHSO interaction to explain 264 is that with ammonium trifluoroacetate, which shows a clear well-defined triplet at ambient temperature. On raising the temperature about 38°, this triplet coalesces (Figure 21D), thus indicating that for the trifluoroacetate salt proton exchange between the ammonium ion and solvent is of great importance. As to why a triplet is observed for the trifluoroacetate and not for the other salts the same explanation given for trifluoroacetate in nitromethane may be used, namely, the solvent plus the trifluorocetate ion form an aggregate with the ammonium ion which has high symmetry. The nitrogen nucleus then sees a highly symmet— rical structure, the ammonium hydrogens interacting in the same manner with all oxygens whether they belong to the trifluoroacetate or DMSO. Comparing the triplet structure for the same salt, ammonium trifluoroacetate, in the three solvents-trifluoroacetic acid, nitromethane, and dimethyl- sulfoxide (Figure 213, D, and F)—one can easily see that the 14N relaxation time is longest for the DMSO. As point- ed out above, the structure for DMSO-ammonium ion is very similar to that of trifluoroacetate-ammonium lon(in the :Lon pair) and, since DHSO has only one oxygen, one might speculate that the ammonium ion comes into contact with jLLSt one oxygen of the trifluoroacetate- 265 The two halide and nitrate salts in DHSO present an example of the "single—line" problem. Indeed, here it is a dilemma, as the ammonium halide peaks broaden while the ammonium nitrate peaks sharpen with increasing temperature The fact that each of these single lines have different half-height widths and different chemical shifts means that the gegen ion has an influence on their shapes. Therefore, this in part explains the singlet rather than triplet lineshape; the gegen ion being very different from the DMSO of the solvent shell gives rise to large field gradients causing short relaxation times- Since the line width changes were perplexing, it was thought that a consideration of the chemical shifts with temperature for these salts in DMSO compared with the known changes of relaxation times or exchange rates with temperature could clarify the situation. From the kinetic work (Section IIA) there is ample evidence that there is a large chemical shift change of the NH resonance, or of the combination band, with change in exchange rates and/or ‘Cemperature- The changes in temperature with magnitude of cflnange in chemical shift (irrespective of the direccion of sflnift) are listed in Table 31 A plot of these values :is shown in Figure 22, the first point for each 266 'Tabde 31. Change of NH proton chemical shift ‘with temperature.a *' gggggm AtfiTECLigyl gjwstem AtFTbC) 15:7 (NH4)ZSO (32304) 38 1.8 Hz 5-04m aq NH4C1 23.5 4.0Hz NH4Form HPorm) 14 4.9 (pH = 0.2) 52.0 10 35 NH4TFAc (HTFAC) 21 0.4 4.99M aq NH4C1 11.5 1.9 38 0.72 (pH - Ou48) 15.0 2.5 40.2 8.7 NH4C1 (DMSO) 38 1.2 N841 (0x50) 38 1.0 4.99M aq NH Cl 11.5 6.8 NH4NO3 (muse) 38 4.7 (pH - 4.3?) 15-0 11.2 NH4TFAc (DMSO) 38 3.1 40.2 19.0 aData taken from Table 27. The change in saiftlivL is observed for the change in temperature LT. salt-solvent combination being chosen as a zero and all other points for the same system being relative to it. In essence, we have a comparison of the slopes of the plots of chemical shift y§,temperature. From Figure 22, there is a very obvious correlation Iwetween relative chemical shift and temperature, a much .larger change being noted for an exchange process than for a. quadrupole relaxation process. Incidentally, it should hue pointed out that from the slapes of the plots of Figure 267 22, one might expect that the activation energy for ex- change occurring in formic acid should be larger than for comparable processes in water (pH 5 0.48) while it should be smaller for ammonium trifluoroacetate in DMSO. It can now be said with more certainty that the change in the line shape of the peaks for ammonium nitrate in DMSO is due to an appreciably rapid exchange process because, first, the line sharpens up with increased temperature which is characteristic of a kinetic process and. second, the change in chemical shift with temperature falls in between values for known exchange processes. The same two considerations clearly identify the changes in the spectra of the ammonium halides in DMSO with temperature as being governed by a quadrupolar relaxation process. First, the broadening with increased temperature and second_ the rela- tive change of chemical shift with increased temperature falls in the range of values for known relaxation processes. The other two ammonium salts investigated in DMSO the formate and acetate reacted with DMSO at ambient tempera- ‘ture expelling gaseous ammonia- Considering the behavior of all the ammonium salts in DMSO and the reasons for some being mainly relaxation iccnutrolled and others being mainly controlled by exchange 268 processes, and still others reacting chemically. we can only be intrigued and leave this for further study. One thing is certain, however, and that is that the particular process occurring in DMSO is controlled by the gegen ion. VI. SUMMARY The NMR spectra of the mono-, di-, and trimethyl- and mono-, di-, and triethyl-substituted ammonium chlorides have been studied in aqueous solution at low pH ( = *EIS.8) 15 ' CONTINUE CALL AREAS BIGAREA = 00.000 IF(SUH(JETA).GT.BIGAREA) BIGAREA - SUH(JETA) i9 CONTINUE 00 24 JETA = 1.NUHB SKALFACT a BIGAREA/SUMIJETA) DO 24 l = 19" _ , STGIJETA.I,2) : (STG‘JETA:I;2))*SKALFACT 24 CONTINUE. B = 00.00 3 S = 00.00 00 29 JETA = 1, Nuns 00291314.“ .. IF (STGIJEIA.I.2).GT.8) B = STGIJETA,I.2) 29 __ CONTINUE STEP:(B-S)*.01 DO 3* JETA 3 11NUHB CALL GRAPH(JETA) _ HRITE(61.33) JETA 33 FORHAT(*-*,12x.*GRAPH NUMBERtl4) 34 . CONTINUE END 295 12/07/66 SUBRCUTINE AREAS THIS suaRouTINE CALCULATES AREA UNDER CURVES 87 THE TRAPAZOID APPRoxIHAIIou 115305“. COHHUN SIGIID.200.2>.~KP(101). N. NUHB. SJEPaSUN(20):XINCR 00 304 JETA s 1.NUH8 _ _ - SUHIJETA) I 0000.000“ CALC 3 (STGIJETAa1.2) + SIG‘JET‘IN12))/200 INL s N - 1 DO 301 I s 2.1NL CALC 3 C‘LC 4 STG(JET‘O'OZ) 301 CONTINUE SUH(JETA) s CALCOXINCR PRINT 303.JEYA.SUH¢JETA)oXlNCB 303 FORMAT (2X.*aREA 0F CURVE*13.* 15 *EiS.Oo5x.tXINCR IN SUBROUTINE 1 CS itE15.” 304 CONTINUE END .30 can<1rr¢ran ‘ 102 200 103 104 201 105 202 ':U.SC"O 296 12/(I7/C-h SUILnuxzfly Cih”h(d‘}fi) PI‘A“ISLLE 000% IV! HTC amp I. g _ culuu“ on" C: sis ILLILimfi IHI x-vnnltuLt. (”LUH” 1w“ h' 37% cUNIJIH: {(X), g L ]K 7!? EU '-L U5 L02H7S II LL PLU}IFU. L I? 10 H“ 30*“ U“ 1095. CTHVON STGII‘IZUfip2)u KP(101}: N: NUH89_51EP;SUH(20)'XINCR DATA(KM21HO)J(Kl=1HI)o(KH31H') IC=1 DO 102 1:2.103 KP(I\=KH KP¢1>=KI s KPIIC):KI I KPI101)=KI PRINT 200.!KP(!)-l=1o101) 7 roRnATIolt.6x,txs.13!.tF.YFII).YDIII. I = 1.») 409 FOFVLT (14.2%.r10.4.?X.FJ0.4:2?:‘10I4v2XIF10I4IZXIF10.4) END N.1.100I IES MICHIGAN STATE umvsasnv LIBRAR \lIWIHIlllHllH‘llllllllHlWIIIWHHIIIIIHHIIIEl I 3 1293 03143 1780