OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERIALS: Place in book return to remove charge from circulation records ‘\ 'II’ PHYSICOCHEMICAL STUDIES OF POTASSIUM CATION INTERACTIONS WITH NEUTRAL LIGANDS AND WITH ANIONS By Emmanuel Schmidt A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1981 ABSTRACT PHYSICOCHEMICAL STUDIES OF POTASSIUM CATION INTERACTIONS WITH NEUTRAL LIGANDS AND WITH ANIONS By Emmanuel Schmidt 'The kinetics of complexation of the potassium cation with the ligand lB-crown-6 (1806) were studied in five sol- vents or solvent mixtures by potassium-39 nuclear magnetic resonance. The salt used was potassium hexafluoroarsenate or iodide. The solvent systems were acetone, methanol, l,3-dioxolane, acetone-l,A-dioxane (80-20% vol.) and acetone- tetrahydrofuran (80-20% vol.). In solutions containing equimolar amounts of solvated and complexed K+ ions, a kinetic process is observed in which the potassium cation exchanges between the solvated and the complexed site. The kinetic parameters were obtained from a quantitative analysis of the line width of the apparent resonance of the sol- vated cation. In 1,3-dioxolane and probably in other solvents too, the cation exchange proceeds via the bimolecular exchange Emmanuel Schmidt mechanism instead of the dissociative mechanism which other workers found to be operative in water. The activation energy for the bimolecular exchange strongly depends on the solvent; it is 9.2 kcal mol-1 in acetone and methanol, 1 13.8 kcal mol- in the acetone-dioxane mixture and about 16 kcal mold1 in dioxolane. The differences in activation entropies, which vary between -A cal mol-l deg.l in acetone and +15 cal moi"l deg-l in dioxolane, largely compensate the differences in activation energies so that the ex— change rates at 25°C vary by a factor of less than 50. The results point to a "loose" transition state in which bond breaking would occur prior to bond forming. Carbon-l3 NMR studies were performed on the free cryptand 221 (C221) in twenty solvents over a large tem- perature range. It was shown that C221 interacts with acceptor solvents such as water, formamide, nitromethane, acetonitrile, methanol and dichloromethane. In these sol- vents, as the temperature decreases, the conformation of the cryptand molecule tends towards that found in the C221-K+ cryptate whereas in other solvents such as acetone and di- methylformamide (DMF) the C221 conformation does not change with temperature. Hydrogen-bonding is likely to be res- ponsible for the cryptand-solvent interaction. The temperature independence of the potassium-39 chemical shift of the C221-K+ cryptate in several solvents indicates that, in solution, there is only one kind of Emmanuel Schmidt complex formed. The relatively large paramagnetic shift, which varies between 10 and 15 ppm, suggests that the K+ ion is tightly embedded in the cryptand cavity. Potassium-39 and carbon-13 NMR studies were also per- formed on the potassium ion complexes with dibenzo-2l- crown-7 (DB21C7), dibenzo-2A-crown-8 (DB2A08) and dibenzo- 27-crown—9 (DB2709). Only the presence of 1:1 complexes was detected by both techniques. With DB2ACB, the log- arithm of the formation constants (measured by 13C) are l.AAi0.08, 3.A6:0.l7, 3.70:0.16 and 3A in DMF, pyridine, acetonitrile and nitromethane respectively. In nitro- methane, the three aliphatic carbon peaks coalesce at K+/DB2A08 mole ratio of 1, indicating the formation of a tridimensional "wrap-around" complex. In the other sol- vents, where the complex is less stable, the coalescence is not observed, indicating that a strong K+—solvent interaction might prevent a complete stripping of the cation solvation shell by the ligand. The DB2A08 ligand has the minimum size required to wrap around the K+ ion. The 23Na, 39K, 133Cs and 205cm chemical shifts of the crown complexes with the corresponding monovalent cations in various solvents were collected from the literature. An analysis of these data in terms of the repulsive over- lap effect provides detailed-information about the ion size-cavity size relationship, the conformational properties of crown molecules and the cation-solvent interactions. Emmanuel Schmidt Potassium-39 chemical shifts in molten salts correlate those in aqueous solutions of these salts but cover a much larger range. ACKNOWLEDGMENTS The author wishes to express his deepest appreciation to Professor Alexander I. Popov for his guidance and encourage- ment throughout the course of this work. Professor POpov will be particularly remembered as a master in the dif- ficult titration of necessary guidance with necessary freedom. Professor Stanley R. Crouch is acknowledged for his helpful suggestions as second reader. Thanks are also extended to Professors James L. Dye and Michael J. Weaver for interesting and helpful discussions concerning the kinetics study and to Dr. Jean-Pierre Kintzinger (Université Lbuis Pasteur, Strasbourg, France) for his precious last- minute help during his visit at MSU. The financial assistance of the Department of Chemistry, Michigan State University, and the National Science Founda- tion is gratefully acknowledged. ‘ Many thanks go to Wayne Burkhardt, Tom Clarke and Alan Ronemus for their efforts in keeping the NMR spectrometers in operating condition and for allocating me expanded time slots on these instruments. Deep appreciation is given to Professor Bernard Tremillon (Université Pierre et Marie Curie, Paris, France) 11 for "sending" me to America to make sure that Professor Popov practices his French on a regular basis and for "sending" his son Jean-Michel to check that I did not forget mine. Special thanks go to all the past and present members of the research group (alias the U.N. lab) and in particu- lar to Dale, Davette, Elisabeth, Lee, Richard, Sadegh, and Zhi-fen for their friendship and stimulation during these intense years. Finally, I wish to thank Catherine for her love, patience and unending encouragement throughout this study. Her everyday letter considerably shrank the ocean. iii Chapter LIST OF LIST OF CHAPTER A. B. E. F. CHAPTER A. B. CHAPTER TABLE OF CONTENTS Page TABLES. . . . . . . . . . . . . . . . . . . vii FIGURES . . . . . . . . . . . . . . . . . . xiii I. HISTORICAL REVIEW . . . . . . . . . . . 1 Introduction. . . . . . . . . . . . . . . . 2 Kinetic Properties of Alkali Cation Complexes. . . . . . . . . . . . . . A Conformation and Solvation of Cryptands and Cryptates . . . . . . . . . . 33 Interaction of Conventional Ligands with Alkali Cations . . . . . . . . 36 Nuclear Magnetic Resonance in Molten Salts. . . . . . . . . . . . . . . . 37 Potassium-39 NMR. . . . . . . . . . . . . . A0 II. EXPERIMENTAL PART. . . . . . . . . . . A5 Salt and Ligand Purification. . . . . . . . A6 Solvent Purification and Sample Preparation . . . . . . . . . . . . . . . A7 Instrumental Measurements and Data Handling . . . . . . '.° 3 . . . . . . A9 1. Potassium-39 NMR. . . . . . . . . . . . A9 2. Carbon-l3 NMR . . . . . . . . . . . . . 57 3. Data Handling . . . . . . . . . . . . . 58 III. KINETICS OF COMPLEXATION OF POTASSIUM CATIONS WITH 18- Crown-6 IN NONAQUEOUS SOLVENTS BY POTASSIUM-39 NUCLEAR MAGNETIC RESONANCE. . . . . . . . . . 60 iv Chapter Page A. Introduction. . . . . . . . . . . . . . . . 61 B. Choice of Solvents and Salts. . . . . . . . 62 C. Results and Discussion. . . . . . . . . . . 65 l. Potassium-39 NMR of the Solvated and Complexed Potassium in the Absence of Chemical Exchange. . . . . . 66 2. KinetiasStudy in Pure and Mixed Solvents. . . . . . . . . . . . . . . . 93 D. Conclusion. . . . . . . . . . . . . . . . . 133 CHAPTER IV. MULTINUCLEAR NMR STUDY OF THE FREE CRYPTAND 221 and OF THE C222°K+ CRYPTATE. . . . . . . . . . . . . . . . . . . . 13A A. Carbon-l3 NMR Study of the Solvation and the Conformation of Cryptand 221. . . . 135 1. Introduction. . . . . . . . . . . . . . 135 2. Results and Discussion. . . . . . . . . 136 3. Conclusion. . . . . . . . . . . . . . . 155 B. Potassium-39 NMR Study of the C221-K+ Cryptate. . . . . . . . . . . . . 156 1. Introduction. . . . . . . . . . . . . . 156 2. Results and Discussion. . . . . . . . . 157 3.. Conclusion. . . . . . . . . . . . . . . 162 CHAPTER V. MULTINUCLEAR NMR STUDY OF THE COMPLEXATION OF POTASSIUM IONS WITH CROWN ETHERS AND WITH "CONVENTIONAL" LIGANDS . . . . . . . . . . . . . . . . . . . . 16A A. Potassium Cation Interaction with Crown Ethers. . . . . . . . . . . . . . . . 165 1. Introduction. . . . . -.- . . . . . . . 165 2. Results and Discussion. . . . . . . . . 165 Chapter B. Potassium Cation Interaction with the "Conventional" Ligand Bipyridine. 1. Introduction. . . . . . . . . . 2. Results and Discussion. . . . . . CHAPTER VI. NUCLEAR MAGNETIC RESONANCE STUDIES OF MOLTEN SALTS . . APPENDICES. . . . . . . . . . . . . . APPENDIX 1. APPLICATION OF COMPUTER PROGRAM KINFIT TO THE CALCULATION OF COMPLEX FORMATION CONSTANTS FROM NMR DATA APPENDIX 2. SUBROUTINE EQUATION. APPENDIX 3. DERIVATION OF THE EXPRESSION OF THE RECIPROCAL LIFETIME OF THE SOLVATED SPECIES APPENDIX A. CALCULATION OF THE STANDARD DEVIA- TION ON 1/IA VALUES . . . . . . . . . . . . REFERENCES. vi Page 203 203 20A 209 221 221 223 22A 226 228 LIST OF TABLES Table “rt . Page 1 Kinetic Parameters for M+-Anti- biotic Complexes in Methanol at 25°C . . . . . . . . . . . . . . . . . . . 8 2 Rate Constants and Activation Parameters for the Formation of Some Cryptates at 25°C . . . . . . . . . . 13 3 Dissociation Rates of Some Cryptates in Several Solvents at 25°C. . . . . . . . 15 A Activation Parameters for the Dis- sociation of Some Cryptates at 25°C. . . . l7 5 Rate Constants and Activation Param- ' eters for Some Crown Complexes at 25°C . . . . . . . . . . . . . . . . . . . 21 6 Key Solvent Properties and Correc- tion for Magnetic Susceptibility on WH-180. . . . . . . . . . . . . . . . . 52 7 Diamagnetic Susceptibility Correc- tions with Respect to Acetone d6 . . . . . 59 8 Key Solvent (and Solvent Mixtures) Properties and Corrections for Magnetic Vii Table 10 11 12 Susceptibility on WH-180 and WM-250 Spectrometers. . . . . . . . . . . . . . Potassium-39 NMR Chemical Shifts and Reciprocal Transverse Relaxation Times for Solutions Containing KAsF6 and 18C6 at lacs/K+ Mole Ratio (MR) of 0, 0.5 and 1.0A in Acetone and at Various Tempera— tures. . . . . . . . . . . . . . . . Potassium-39 NMR Chemical Shifts and Reciprocal Transverse Relaxation Times for Solutions Containing KAsE6 and 1806 at 1806/K+ Mole Ratio (MR) of 0, 0.5 and 1.0A in Methanol and at Various Tempera— tures. . . Potassium-39 NMR Chemical Shifts and Reciprocal Transverse Relaxation Times for Solutions Containing KASF6 and 1806 at 1806/K+ Mole Ratio (MR) of 0, 0.5 and 1.0A in 1,3-dioxolane and at Various Temperatures . . Potassium-39 Chemical Shift and Recipro- cal Transverse Relaxation Times for Solutions Containing KAsF6 and 1806 at 1806/K+ Mole Ratio (MR) of 0, 0.5 and 1.0 in Acetone-1,A-dioxane and at - Various Temperatures viii Page 6A 71 72 7A 77 Table 13 1A 15 16 17 18 Potassium-39 Chemical Shift and Reciprocal Transverse Relaxation Times for Solutions Containing KAsF6 and 1806 at 1806/K+ mole Ratio (MR) of 0, and at Various Temperatures. Potassium-39 Chemical Shifts and Transverse Relaxation Rates for the Solvated and the Complexed K+ ions in Various Solvents at 25°C Potassium-39 NMR Chemical Shifts and Line Widths for Acetone-1,3-dioxolane Mixtures Containing 0.05 M KAsF6 and 0.05.M 1806 at 23°C. Potassium-39 Chemical Shifts and Transverse Relaxation Rates for the Solvated and the Complexed K+ Ions in Various Solvents at the Tempera- tures Corresponding to the Middle of the Intermediate Region in Each Solvent. Values of 1/1A and l/TAX[K+'18C6] in Acetone. Values of l/TA and l/tAXEK+-18C6] in Methanol ix 0.5 and 1.0 in Acetone-THE Page 79 86 89 98 .112 113 Table Page 19 Values of l/rA and l/TAXEK+-18C6] in 1,3-Dioxolane. . . . . . . . . . . . . . . 11A 20 Activation Parameters and Exchange Rates for K+°18C6 Complexes in Various Solvents . . . . . . . . . . . . . . . . . 115 21 . Values of l/TA and l/TAX[K+'18C6] in ACetone-Dioxane Solutions. . . . . . . . . 127 22 Carbon—13 Chemical Shifts of the Sodium and Potassium C221 Cryptates at 25°C. .3. . . . . . . . . . . . . . . . 1A2 23 Carbon-l3 Chemical Shifts of the Free C221 Cryptand in Various Solvents and at Several Temperatures. . . . . . . . . . 1A3 ' 2A Potassium-39 Chemical Shifts and Line Widths of the C221'K+ Cryptate in Several Solvents and at Various Temperatures . . . . . . . . . . . . . . . -158 25 Potassium-39 Chemical Shifts and Line Widths for Acetone Solutions Containing KAsF6 and 1,10-dithia-18-crown-6 at Various Mole Ratios and at 25°C. . . . . . 168 26 Potassium-39 Chemical Shifts and Line Widths for Acetonitrile and Pyridine Solutions Containing KAsF6 and Dibenzo- 27—Crown—9 at Various Mole Ratios and at 2A°C. . . . . . . . . . . . . . . . . . 172 Table 27 28 29 3O 31 32 33 Page Potassium-39 Chemical Shifts of Potassium Ion Complexes with Various Crown Ethers in Several Solvents and at 25°C. . . . . . . . . . . . . . .7. . . 17A Sodium-23 Chemical Shifts of Various Na+-Crown Complexes in Several Solvents at 30°C. . . . . . . . . . . . . . . . . . 177 Cesium-133 Chemical Shifts of Various Cs+-Crown Complexes in Several Solvents at 30°C. . . . . . . . . . . . . . . . . . 178 Thallium-205 Chemical Shifts of Various Tl+°Crown Complexes in Several Solvents at 30°C . . . . . . . . . 179 Formation Constants of M+°DB2AC8 + + + + Complexes with M = Na , K , Cs and T1+ in Various Solvents at 25°C. . . . . . 19“ Carbon-13 Chemical Shift - Mole Ratio Data for the K+°DB2AC8 Complex in Various Solvents at 25°C . . . . . . . . . 197 Potassium-39 Chemical Shifts and Line Widths for Nitromethane Solutions Containing KAsF6 (0.075 M) and 2,2'- Bipyridine at Various BP/K+ Mole Ratios (MR) and at 25°C. . . . . . . . . . 205 xi Table 3A 35 36 Physical Properties of Some Alkali Metal Salts and Eutectic Mixtures. Potassium-39 Chemical Shifts and Line Widths for Some Molten Salts and Molten SaltsMixtures . . . Potassium-39 Chemical Shifts for Aqueous Solutions of KSCN at Various Concentrations . . . . . xii Page 211 21A 215 Figure LIST OF FIGURES Structures of some naturally occurring and some synthetic macrocyclic com- ‘pounds. . . . . . . . . . Three possible forms of cryptand C222. . . . . . . . . . . . . . . . . Semilog plots of potassium-39 transverse relaxation rates for the solvated and the complexed K+ ion vs l/T in various solvents. Potassium-39 chemical shifts of the solvated and the complexed K+ ion vs l/T in various solvents. . . . . Potassium-39 relaxation rates for the solvated K+ ion in methanol Potassium-39 chemical shifts and line widths for the K+-1806 complex in acetone-1,3-dioxolane mixtures . Concentration dependence of the potassium-39 chemical shift and line width for KAsF6 solutions in tetra- hydrofuran and 1,3-dioxolane. . . ‘41,), /XiV Page 3A 68 69 70 88 91 Figure 10 11 12 13 1A Page Potassium-39 NMR spectra for solu- tions containing KAsF6 and 18-crown—6 at various 18-crown—6/K+ mole ratios (MR) in acetone-l,A-dioxane (80-20% vol.) at various temperatures . . . . . . 95 Semilog plots of l/T2 vs l/T for acetone solutions containing KAsF6 and 1806 at ligand/K+ mole ratio of 0 and 1.0 . . . . . . . . . . . . . . . . 96 Semilog plots of l/T2 y§_1/T for methanol solutions containing KI and l8-crown—6 at ligand/K+ mole ratio of 0, 0.5 and 1.02 . . . . . . . . . . . . . 105 Semilog plots of potassium—39 relaxa- tion rates vs l/T in 1,3-dioxolane solutions 106 Temperature dependence of the 39K chemical shift in methanol solutions with 18c6/K+ ratios of 0, 0.5 and 1.02. . 107 Semilog plots of l/TA vs l/T in various solvents. . . . . . . . . . . . . lll Plot of l/TAx[K+°l8C6] Ea 1/[K+] in 1,3-dioxolane at various temperatures. . . . . . . . . . . . . . . 118 XV Figure 15 16 17 18 Potassium-39 NMR spectra for solu- tions containing 0.10 M KAsF6 and 0.05 M 1806 in acetone-THE ' (80-20% vol.) at various tempera- tures Semilog plots of 1/T2 Kg 1/T for acetone and acetone-dioxane (80-20% vol.) solutions containing 0.1 M KAsF6 (bottom curves), 0.10 M KASF6 and 0.05 M l8-crown-6 (middle curves), 0.10 M KAsF6 and 0.10 M 18-crown-6 (top curves) Semilog plots of l/T2 YE 1/T for acetone and acetone-THE (80-20% vol.) solutions containing 0.10M KASF6 (bottom curves), 0.10 M KASF6 and 0.05 M l8-crown-6 (middle curves), 0.10 M KAsF6 and 0.10 M 18—crown-6 (top curves) Temperature dependence of the 39K chemical shifts in acetone- dioxane (80-20% vol.) solutions xvi Page 123 12A 125 126 Figure Page 19 Carbon-13 spectra of the C221-K+ cryptate in methanol and of the C221-Na+ cryptate in pyridine at 25°C. . . . . . . . . . . . . . . . . . . 138 20 A comparison of the patterns observed in the carbon-13 spectra of C221 and of its cryptates with the Na+ and the K+ ion. . . . . . . . . . . . . . . . . . 139 21 Carbon-l3 spectra of the free C221 cryptand in various solvents and of the C221-K+ cryptate in water . . . . . . 1A7 22 Carbon-13 spectra of the free C221 cryptand in acetonitrile and DMF at various temperatures . . . . . . . . . 1A8 23 Carbon-l3 spectra of the free C221 cryptand in nitromethane, nitroethane and 1-nitropropane at various tem- peratures . . . . . . . . . . . . . . . . 151 2A A plot of the difference in chemical shift between the two N_C_H2 resonances of the free C221 cryptand vs tempera- ture in various solvents. . . . . . . . . 152 25 Potassium-39 chemical shifts as a function of the DT18C6/K+ ratio in acetone solutions . . . . . . . . . . . . 169 xvii Figure Page 26 Potassium-39 chemical shift YE DE27C9/K+ mole ratio in acetonitrile and pyridine solutions at 25°C. . . . . . 173 27 Sodium-23 chemical shifts of various Na+°crown complexes in nitromethane, acetonitrile and pyridine at 30°C . . . . 180 28 Potassium-39 chemical shifts of various K+-crown complexes in acetonitrile. . . . . . . . . . . . . . . 181 29 Cesium-133 chemical shifts of various Cs+-crown complexes in acetonitrile, pyridine, acetone, methanol and nitromethane . . . . . . . . 182 30 Potassium ion-crown ether complexes of various stoichiometries. . . . . . . . 187 31 Carbon-13 spectra for pyridine solu- tions containing KAsF6 and dibenzo- 2A-crown-8 at various mole ratios and at 25°C . . . . . . . . . . . . . . . 196 32 Carbon-13 chemical shifts (vs acetone d6) as a function of the K+/DB2AC8 mole ratio in aceto- nitrile and nitromethane at 25°C. . . . . 199 33 Carbon-l3 chemical shifts (vs acetone d6) as a function of the xviii Figure 3A 35 + K /DB2AC8 mole ratio in DMF and pyridine at 25°C. Variation of the potassium-39 chemical shift and line width as a function of the 2,2'-bipyridine/K+ ratio in various solvents at 25°C. . Variation of'thepotassium-39 chemical shift as a function of the KSCN con- centration in water at 23°C xix Page 201 206 216 DNMR C222B DMF PC THF THP DMSO DB PY EEUX LIST OF ABBREVIATIONS Dynamic Nuclear Magnetic Resonance Benzo cryptand 222 Dimethylformamide Propylene carbonate Tetrahydrofuran Tetrahydropyran Dimethylsulfoxide Dibenzo Pyridine Ethylenediamine XX CHAPTER I HISTORICAL REVIEW 1. Introduction The coordination chemistry of alkali metal cations (M+), which for many years was thought to be non-existent, has matured in the last fifteen years to become a coherent discipline. Two major diScoveries prompted chemists to develop the field very rapidly. The recognition of the biological role of Li+, Na+ and K+ cations (1-3) was soon followed by the synthesis of macrocyclic polyethers, called crowns (A), and of macrobicyclic ligands, called cryptands (5-7). Crowns and cryptands display strong and selective The ccnnplexive abilities towards alkali metal cations. Iwecent and fascinating chemistry of these ions has evolved from this property . Syntheses of new multidentate macrocycles, some of WTlich exhibit complicated topologies and fluctuating ring Etizes (8), are reported almost weekly. The equilibrium Ixrbperties associated with the interactions of alkali ‘3Eitions with these macrocycles have been investigated by a. number of physicochemical techniques (9) and collected 3111 several reviews (10-13). Some of the most useful tech— rlilques are potentiometry, calorimetry and conductometry as well as electronic, vibrational and nuclear magnetic reson- aJace (NMR) spectroscopy. In this laboratory we probe directly the immediate chemical environment of the alkali ions in solution by using alkali metal NMR. This sensitive technique has proved valuable for studying ionic solvation and complexation re- actions, particularly in nonaqueous solvents (9,1A). Po- tassium-39 NMR, however, has been used only sparingly in the past due to the very low sensitivity of this nucleus, as compared to those of the other alkali elements. However, with the advent of superconducting magnets and the develop- ment of Fourier transform NMR, high quality 39K NMR spectra may be obtained even in dilute solutions. In this dissertation we describe various aspects of tune kinetics and thermodynamics of complexation of the betassium cation with some crown ethers and cryptands in no naqueous solvent 3 . In the first part of this thesis we investigated by 35%K NMR the solvent dependence of the kinetics of complexa- trlon of the potassium cation with a crown ether molecule. The second part is devoted to a 130 NMR study of the S<31vent and temperature dependences of the conformation C>:f'a.cryptand. Cryptands and their complexes, cryptates, ekist as equilibrium mixtures of three conformations but IICDthing is known about the conformational equilibria in S<31ution. Likewise the ligand-solvent interactions have Ilot been explored, except in water and methanol solu- t ions . The third part of this work describes the extension of the thermodynamic studies of potassium cation-crown ether complexes initiated by Shih (15) in this group. In par- ticular, we examined the possibility of measuring the stability constants of complexes by 13C NMR. Some ex- ploratory studies of alkali complexes with the so-called "classical" or "conventional" ligands are also reported. We selected 2,2'-bipyridine for this investigation. The last part is devoted to some 39K NMR measurements in low melting salts of eutectics. The object of this ex- ploratory study was to investigate the influence of a 39 direct K+-anion interaction on the K resonance in the absence of a solvent. We will review separately each of the four topics OLrtlined above except the third one. The literature up tc> early 1978 on potassium complexes has been covered in tile Ph.D. thesis of J. S. Shih (15). 23. Kinetic Properties of Alkali Cation Complexes Thus far, the kinetics of alkali cation complexes have r‘eceived rather limited attention despite their importance fIor the understanding of ion transport processes in organic er biological membranes (1) and of catalytic phenomena (l8). Naturally occurring antibiotic macrocycles such as valinomycin, monactin (Figure l) and enniatin, have been ‘3‘ z . {'7 Ohm". x _ 0 c c’ 1 0 3w Cum ““0 0...“... b 5“. o 1 s a “I : 2: c 0 ‘ 1:- ” 0 I L“““ I #0 . = - .‘ > in , ~‘ ', ‘ n.:9_,=a)=a~ :CH.’ Nonqclin , 0 0 | fl.:°2;=3 :CH) 9‘: ::F“ Vcflocfi" : 9 ' ~: 9.;9 9,33 :chtm o D 9.,3 '-'I"OU a fl- 2 9.. ’mana:t.n . “000mm Coo, 0cm, . . CL) .> C. .3 . - ° \J k/Sx) ' 18-‘cnowN-6 1,10-DITHlA-18C6 15-cnoww-5 (1806) (m 13cc) (15cs) ©{3P DIBENZO-18-CROWN-6 BENZO-15-CROWN-5 (DB18C6) (315cs) (IO/‘0 ' flohoh O u’\, u . @OWOQJ;3© V2333 n:1 0821C? 8:0 CRYPTAND 221 C221 "=2 032403 m CRYPTAND 222 c222 ":3 0327C9 Figure 1. StruCtures of some naturally occurring and of some synthetic macrocyclic compounds. better characterized in terms of their complexation kin- etics than their synthetic counterparts (10,19) although most studies were performed only in methanol solutions be- cause of problems of low so1ubility and low complex stability in water (16). “r" Grell t 1. (17) investigated the formation of the M+-valinomycin complexes in methanol, using an ultrasonic absorption method. The data indicated that the uncomplexed valinomycin undergoes some rapid conformational equilib- ria, and the mechanism could be simply described as shown below. + kl2 + k23 + M (solvated) + L(so1vated) I M (solvent)L I ML k21 k32 (l) A. diffusion controlled bimolecular collision, between an Cmben form of the ligand molecule and the solvated cation, 218 followed by the rate-determining conformational change C>f the ligand leading to the compact final structure of che M+-valinomycin complex. Chock 33 al. (16) reported Eéxtensive relaxation studies on complex formation of tnacrocyClic (215;: nonactin, monactin) and open chain ( e.g., monensin) antibiotics with monovalent cations. UDheir data indicated the existence of a ligand conforma- 1:ional change prior to complexation: k 12 L1 k1 L2 (2) 21 . k23 + L2 + M : ML2 (3) k 32 The formation rates R23 approach the limits imposed for difquion controlled processes. Degani (20) used 23Na NMR to monitor the kinetics of complexation of Na+ by the open chain antibiotic, monensin, in methanol. The dominant sodium exchange pathway was found to be the first order (dissociative mechanism k + 33 + , Na + monensin + Na —monensin (A) k32 Tnne small activation entropy for complexation (A833 = “3-9 cal. mOl-ldeg-l) indicates a small conformational re- Elrrangement of the monensin anion prior to complexation. The kinetic parameters for the formation of some M+- ELntibiotic complexes in MeOH are collected in Table 1. qjhe formation rates (k23) approach the limits imposed for Cliffusion controlled processes. It is clear from Table l tfihat alkali metal complexes show wide variations in the (lissociation rates (k32). Indeed, for the cryptates, these IFates may vary by as much as ten orders of magnitude (21). loo haw .mQOHpmuoc pom Amv coauMSUm com Imsom comm am .mppososmo ommnd poom> monocmo om> .COdepomnm oficommhpas mopocmo m: mm m< .sfiDQMCHDu one capomcfio Dom mafia: 020 x .om monopomom Eopm damn . IaoE.Hmox m.ouu x\masumax .AmeIHm> new +mzuam> Dom pdooxo mummoanopoo EsficoEEmHmusnmpump a H.o mo Coco .NH oocopomom Eopm mama .mCOHumpo: pom Amv COfiumDUm mom x H h .Hm> Loo moan: ozm .Hm> pom .cfipowcfio pom .COHummeop dean modummeEop mono: .mH no Acfipomcoev mm mocopomom 80pm mama .mCOfipmpoc pom AHV soap o: .:.o m.mln m< x m ohm mpHCDo m ohm mpacon Imopd one CHV mCOHumLqu capp050pondoppoodm an omcfieaopoo ucmpmcoo mufiHHnmpm ucopmdd¢m he moaxw.m :oaxm.= moaxw.a m.H moax>.H +mz ncfipomcfime m: :oaxm.a :oaxa.m moaxm.» m.o woaxH.H N.oagmé moaxm.: +mo ncfipomcfio he moaxm :oax:.m moaxm.a +mz ccfipomsfio m: moaxm :oaxm.: moaxm.a v.0 moaxa.a N.oaxmo moaxH.H +mz ncfipomch .o.m.> moaxm.m +x ncfipowcoz x.o.m.> monH.H mwz zcfipomcoz mo w moaXm.m moaxm m.m mOme.H monH F: +32 me> be moaxm +mo mam> he soaxm.e +em wao> a o o o woe m: moaxw n moaxm H hoaxa o : moaxa moax: :oaxm +m me> m x x . x x . m m mfimo moa m woa : m m N.oa m NOH s N z + 2 Ma > mzz 2mm moaxa mm Noaxm m +mz mcfimcocoz o o m m m I o so 2 mu: am an an: an: H- Hun an: an: coarse commas m m mm NH Hm NH m mm 3 63 0M 3 2x m3 .oomm om moo: ca moxofiasoo OHpOHpszapo< one monopmcoo opmm .m canoe .omm ca as Rose 1A ooumHSOHmoo .mm monopomomo .wm ocm mm moocmpomom Eogm oopmfisoamoo .hopwz Ca cm>Hw mam mpmuoewpmd cofium>fipom ones .oopmoHUCH omfizhmcuo mmmac: om oocopmmmm Echo woman moaxme moaxme mmmo Noaxm.me moaxme ammo +mo moaxw.a woaxw.a moaxm.m moaxfimm mmmo Noaxo.w noaxm.m moaxa.: ammo +nm woaxo.: Noaxm.m Noaxm.m moaxm.a moaxn.: moaxo.m mmmo x . . x . . . moa w m moaxm He NOH m : woaxw m Noaxw H ammo +x moaxH.H moax>.m moaxm.fi mmmo woaxw.a moaxm.m Noame monN.H moaxm.m ammo aoaxa.m moaxm moaxw.w moaxfi.m mofixm.m flame +mz moaxme Noaxw.a ammo Noaxmv moax:.H :oaxm.a moaxw.a moaxm.: moaxo.w Hamo +Hq om mzo omza m0pm moo: 0mm caduozpo coapmo m Aaun Huzv x .ooschcoo .m canoe mmmmo owoaxo.m zoaxze mmmo moaxme :oaxm.me ammo +mo mloaxm.m HIoaxod moax:.a mmmo OHxH.H oaxm.> ammo +Qm mcoaxa.: muoaxm.a m.~ mmmo m.me HIonm.H mo.H moaxo.m ammo +x 5 72055 EN omoafieg ammo Huoaxm.m anoaxm.> mnoaxm.m mnoaxmm.m oaxmz.a ammo HIOHxH.n m.m moax:.a Hamo +mz Hoaxm.a Hoaxm.n m.H ammo nmloaxm.a nmuoaxm.m :Ioaxo.m mloax:.: mloaxm.m Hamo +fiq EB omzo zoom moo: omm oceanic soaemo haumv ox .oomm um muco>fiom Hmpo>om CH monopdmpo oEom mo moumm coaumfioommfim .m canoe .m -u...~:~u.~.~...e o» .s~.-\.~. .um mocmpmmomo l6 .zm wocopomomo .mm monopomomn .ooHMHoodm omHzponpo mmch: om mocmpomom Eoom madam mOHx:.m mmmmo e NOmee mmmo m NOsze Hmmo + o HIOwa.H mmmo m.» Hmmo +nm mmmo NIOHx>.m Hmmo +x oH.H oo.w ommH Hv mmmo NIOHv Hmmo . HHmo +mz Hmmo x . . nmloH : m . n:|0me H HHmo +HH ooaenEAom Hm Ema «mm om cemeaaso coaumo Aaunv ox .omschcoo .m mHDwB l7 .om mocopmmmm Eopm ompmHonmon .hmpoeempHo> OHHomo monocoo .>.om .oonuoe OHLpoEOQozocooo em mzz mOMMH .mH+ m.eH m.m m.eH om mmmmo em mzz mommH m.eH+ m.mH o.w m.mH mzo mmmo +uo . . m om o posocoo om 2H 0 m mmmo +em mm d.>.o m.mH+ m.om m.eH m.om 0mm mmmo +x : : m.N I m.NH 3.3H m.NH .o A.HH: m.mH m.mH m.mH o m Hmmo +mz e e a.eHu m.mH m.sH m.mH am Hmmo e e w.mmu m.MH m.om H.=H moasmeaom HHmo e e m.mHu e.mH 0.0m o.mH mza HHmo e e . m.mHn m.mH s.mH H.oH omza HHmo e e m.mHu o.mH s.mm e.mH Mm HHmo mm mzz Hue e.o + e.om m.om m.Hm o m HHmo +aq .oom oonuoz HnHoE.Hmox HIHoE HIHoE HIHoE ucm>Hom ocmpdmmo COHpmo .Hmox .Hmox .Hmox o o o m mme xme use m .Oomm pm moumudmho oEom mo coHumHoommHQ 02p pom mpouoewpmm COHpm>Hpo< .3 dance l8 - The transition state for complex formation lies very close to the reactants. Apparently only a small amount of desolvation of the cation has occurred on forma- tion of the transition state. - The dissociation rates increase sharply with the donor ability of the solvent (this rule has a few excep- tions (3A))- - The dissociation of the cryptates is acid catalyzed (26,29). These general features were mostly found from rate data. Very few activation parameters have been reported, particularly in nonaqueous solvents (see Table A and Reference 57). Kinetic studies of crown ether complexes have been rather Sparse. They require a great deal of effort since none of the classical Optical techniques are applicable. Alkali metal NMR seems to offer a more general approach than 1H or 13C NMR to the investigation of the complexation kinetics (A6). It can yield exchange rates in a large temperature range so that the activation energies may be 23Na and obtained. When a quadrupolar nucleus, 21%;, 39K, is placed in an environment which does not have cubic symmetry, the line width of the NMR signal increases due to the asymmetry of the electric field at the nucleus (3A). In general the lines are broader for the crown complexes than for the cryptates, sometimes by one order 19 of magnitude or more (A7,58). This indicates that the \electric field produced by the planar structure of the former is less symmetrical than that created by the tri— dimensional cavity of the latter. A noticeable exception to this behavior was found for the 2:1 (ligand:metal) "sandwich" crown complexes, 345;, Na(15C5)2 (59), which give quite narrow signals because the cation is surrounded by a large number of binding sites not Coplanar with it. Both the number and the position of these sites are es- sential in determining the alkali metal NMR line width. For example Kintzinger and Lehn (60), studying the sodium cryptates, found that the 23Na quadrupole coupling constant (which determines the line width, see Chapter III) de- creases as more and more oxygen atoms are disposed around the cation. When the signals are relatively narrow, 142;, Avl/2 s 100 Hz, and if the catiOn exchange is slow, two separate alkali resonances are observed in solutions containing an excess of the salt. Only in this case can a complete line shape analysis be performed to obtain the kinetic param- eters. This method was used by Ceraso e3 al. (33,3A) for the Na+-C222 cryptates and by Cahen 22 al. (35) for the L1+-C2ll and Li-C221 cryptates in various solvents (Table A). Mei _e_t_ _e_l. (36,37) also applied this technique to the Cs+C222 and Cs+C222B cryptates in DMF and PC, respectively. Unfortunately this technique fails for crown complexes 20 13305 which have a small quadrupole except for nuclei like moment (61) and give narrow signals. Mei §£.§l- (38) found by 133Cs NMR small activation enthalpies (m8 kcal. mol-l), but large negative activation entropies (N—lA e.u.) for the release of Cs+ ion from the crowns l8C6 and DC1806 in PY and PC respectively (Table 5). For the nuclei with large quadrupole moments, the line width of the bound site is usually broad and sometimes undetectable. In the last case only one apparent resonance is observed (A6), but in some cases the exchange kinetics may still be deduced from the line shape analysis (62,63) with an ap- proximate treatment (see details in Chapter III). Ten years ago, Shchori gt a1. (A6,A7) studied by the above technique the kinetics of complexation of the Na+ ion with DC18C6, DB1806 and its derivatives in DME, DMF and MeOH solutions using 23Na NMR (Table 5). 'They ob- served that the cation exchange proceeds via the dissocia- tive mechanism, the same as that which is favored for the cryptates. Table 5 shows that the apparent activation energy Ea for the release of the Na+ ion from DBl8C6 and its derivatives is nearly independent of the solvent._ The much smaller value of Ea’ 142;, 8.3 kca1.mol'l, found for DC1806, compares well with the result of Mei 33 Q1. (38) for the Cs+-DC18C6 complex in PC. Shchori 2£.§l- (A7) suggested that the energy barrier for the release of the Na+ ion is determined by the energy 21 H.mH 222 +62 emumHmoee m.mH 229 +62 mmomHmoz 2.6H AooemnVOHm 2062 +2 m.mH eOHRm.H ROHRm.6 222 +62 6.6H mOHxH 20H26 222 +62 2.HH sone.H 66me.m 2062 +62 emomHma m.2 2062 +62 222HV662H62 .2H. m.» m.m om +60 oAaHomomHoo 6.22- 6.2 e.6 66226.6 622 +66 20H22.2 wOHRm.e ow2 +66 soaxm.H moa22.2 o 2 +62 H.m 6.6H 6.0H moaxs.m monm.e 062 +2 N.3226 monm.m 062 +62 soaxe e soHame om2 +H2 eeowH onxN.H woaxe.e o 2 +62 AOme.s woaxm.e . 062 +2 20H22.2 onxe.m 062 +62 omomH .s.6..HuHoe .H6ox _ .aHta HuaHrs. H-6.Hu2 666>Hom soaumo ozone mMme mm2< 62 mm2 mmx mmx m.oomm um momedEoo czomo oEom pom mpouoempmm COHpm>Hpo< ocw mucmpmcoo muwm .m oHnme . o :: mocopmm mg .002H20 H22662H0 u 602H2020 .26266 666v 6 2 2H 6m2 .66 6626662626 H: H: . . .006H20 locHs6H6 u 602H2022 .006H20-066H2H0 u 602H202 .2 66206H 602 602666 2H .2: 662626262.H .< HmEomH How mocmum Hom COHpmo 23o20 m m 6 m mm mm .Umscfipcoo .m oHDMB 23 required for a conformational rearrangement of the crown (A7). In this case, the greater flexibility of DCl8C6 as compared to DBl8C6 could account for the lower Ea as- sociated with this crown (A7). This hypothesis was sup- ported by other studies.rrIn 1969, Wong gt a;. (55) reported a 1H NMR kinetic study of the complexation of dimethyl- DB18C6 with the fluorenyl sodium ion pair (Na+Fl’) in THF-d8 solutions. This is a special case in which the fluorenyl ring currents cause large chemical shifts of the ligand protons, thus making possible the measurement of the activation energy Ea for the exchange process, which the authors assumed to be the bimolecular process shown below - + + - + 6 Fl C*Na + C + F1 CNa + 0* (5) The value 0f Ea, 12.5 kca1.mol'l, closely matches that measured by Shchori for DB18C6 (Table 5). Similarly Shporer and Luz (A8) measured by 39K NMR an activation energy of 12.6 kCal.mol-l for the release of the K+ ion from DB18C6 in MeOH solution. However, they could not detect by 87Rb NMR any kinetic process for the Rb+—DB18C6 complex, the decomplexation rates being much faster. The results of Eyring g3 a1. (A2) were along the same lines. Using an ultrasonic absorption method these authors ob- served that in water the enthalpy of activation for a con- formational rearrangement of 1806 (vide infra) closely 2A resembles that required for the dissociation of the K+°18C6 complex (see Table 5). According to these authors the simplest conceivable mechanism consistent with the relaxa- tion data is the two-step process represented by the equa- tions kl2 + k21 k + 23 + M + CR2 : MCR2 (7) k32 where CR1 and CR2 denote the unreactive and the reactive form of the crown respectively. This mechanism involves a fast ligand conformational change followed by a step- wise substitution of the coordinated solvent molecules by the CR2 conformer of the crown. It is identical to the mechanism originally proposed by Chock for the complexa- tion of various monovalent cations with DB3OClO (AA). We have already encountered this two—step process in the case of antibiotic ionophores (16). The above series of studies, except Wong's in THF-d8, are restricted to four solvents with high and similar donicities (3A) so that the observed apparent constancy of Ea for DB18C6 and its derivatives may be coincidental, as Shchori g3 QM. themselves (A6) pointed out in their original paper. The same comment applies to l8C6 and to 25 D01806 complexes. In order to ascertain if the energy barrier to exchange is determined by the barrier for a con- formational twist of a crown molecule, it would be necessary to have kinetic data for a single crown ether reacting with more than two metal ions, in several solvents over a sub- stantial temperature rangE (A2). The various mechanisms proposed for the formation of crown complexes are worth being examined. Eyring gt 3;. (39-A2) carried out ultrasonic absorption studies of the conformational equilibrium of 1806 and its complexation with Li+, Na+, K+, Rb+, 03+, T1+, Ag+, NH: and Ca2+ ions as well as of 1505 and its complexation with Na+, K+, Rb+, Tl+ and Ag+ ions in aqueous solutions (Table 5). The equilibrium constants for the conformational change (Equa- _ - -2 tion 5) are K21 - k21/k12 — (2 i 2) 10 for 1806 and K21 i 0.1 for 1505. This means that most of the free crowns are in the "reactive" CR2 form and that the observed forma- 23 ones k23. For a given cation, the values of k23 are about the same for both crowns (Table 5) and also for the much tion rates k' = k23/1 + K21 are very close to the actual larger and more flexible antibiotic macrocycles (Table 1). In contrast R32 increases for all the cations in going from 1806 to 1505, due presumably to the smaller ring size and the increased rigidity of 1505 (A1). As seen for the cryp- tates the selectivity arises mainly from differences in the decomplexation rates. 26 Turning back to the formation rates, Liesegang gt 1. (A0) observed that for the crown complexes with the K+, + Rb+ and Cs ions, the k 8 23 M-1 8-1 whereas for the EDTA and nitrilotriacetic acid values level off at A.3 x 10 complexes they are in the order K+ < Rb+ < 05+. In addition, for the crowns the k23 values are all about one order of magnitude smaller than the corresponding specific solvent exchange rate constants kex (22). The first point sug- gests a ligand dependence of the complexation kinetics due, for example, to the presence of a second conformational change. Since typically the rate determining step in ligand substitution of an aquo ion is the rate of loss of primary coordinated water from the ion, the second point may indicate that binding the ligand to the ion will require the loss of more than one inner layer water molecule. Thus the kinetic evidence appears to suggest a mechanism more complex than the elementary two-step process pre- viously considered. Several conformational processes (in small crowns) with free energy barriers ranging from 6 to 10 kcal.mol_l were reported by Krane (A9,50,66,67) who used low tempera- ture (W—lOO°C + -l60°C) DNMR techniques in CHClF2/CHF012 mixtures and in some other low melting solvents. The blocking of the inversion of the ether linkages either in the free ligand (50) or in the complex (A9) provokes at low temperature a split of the peaks corresponding 27 to different sets of ligand nuclei. The NMR relaxation times Tl also yield conformational information. Fedarko (68) detected by this method segmental motion in the free D01806 and DB18C6 molecules. Many factors other than the donor ability of the ~“~‘ solvent and the rigidity of the crown may affect the kine- tic parameters for crown complexes. Among these factors are the dielectric constant of the solvent, the structure of the anion and that of the cation, the solvation state of the ion pair, the ligand solvation (H-bonding, etc.). 'Since all the alkali cations have the same "noble gas" electronic structure, we must include in this discussion some studies concerned with organic cation - crown ether complexes (M6rg 0). De Jong et a1. (51,69,70), using a DNMR method with the cation protons, found the activation energy Ea for the release of t-BuNH;(PFg) from 1806 in CDCl3 to be 18 i l kcal.mol-l. This high value is un- expected since most reported values of Ba for alkali crown complexes (M+C) fall in the range of lO-lA kcal.mol-1 (A6-A8,55). Indeed it resembles the Ea values measured for the cryptates (Table A). No explanation was proposed by the authors. Instead of the dissociative mechanism which applies to most M+C complexes, the bimolecular pro- cess (Equation A) was found to be predominant. Both the structure of the cation and the low polarity of the sol- vent may be responsible for these facts. Recently 28 Krane e£_a;. (52) studied by 13C DNMR the ring size effect on the decomplexation barrier of crown complexes with aryl diazonium tetrafluoroborate salt in CHCl F. The preferred 2 host for this salt is the crown ether 2107 (AGS‘: 11.1 1 1 kcal.mol' at TC = -52°C). For 1806, A07 = 10.1 kcal.mol- d at T0 = -75°C. A comparison of these values with other data obtained at 25°C is impossible since as: is unknown. Once again the chemical shift difference between the free and the complexed form of the ligands is too small to permit the measurement of the activation energy Ea’ In general the organic cations (51,71) as well as Ag+ and Tl+ (A0,A1) complex much faster with 1806 than do the alkali cations (A0,Al). For these cations the association rates in CD013 or CD2012 (109 = 1010 MI1 8-1) are characteristic of a diffusion controlled process (51,71). Thus the or- bitals of the metal ions appear to play as decisive a role in the complexation kinetics as do dissimilarities in the ligands (19)- In low dielectric media, the cation-anion interactions may also affect the exchange rates. If the forward and reverse reaction rates vary in a significantly different extent, the resulting thermodynamic parameters will also change. Consequently a different selectivity order can be observed in a polar solvent and in a low polarity medium such as THF or THP. For example Wong et 3;. (55) found the stability sequence with dimethyl-DB18C6 in THF to be 29 Na+ > K+ > Cs+ > Li+ whereas both Pedersen (A) and Izatt 33 21° (10) found an affinity order in water of K+ > Cs+ > Na+ > Li+ for D01806. The polarity of the sol- vent, rather than the difference in the ligand molecules was found to be responsible for this effect (55,72). Ionic association can give either contact or solvent-sep- arated ion pairs (73), and contact ion pairs can be ex- ternally solvated or complexed (7A). Furthermore, com- plexation by the crowns induces the formation of another kind of species, namely the ligand-separated ion pairs (7A). Smid 3£.§l° (72,7A) extensively studied the com- plexation of alkali metal fluorenyl salts (Fl‘M+) with various small crowns (C) in THF and THP by electronic spectroscopy. They obtained the equilibrium constants for the formation of externally complexed ion pairs (Fl-M+C) and ligand- separated ion pairs (Fl-CM+). In this particular case where only contact ion pairs exist in the absence of crown at 25°C, the following equilibria were considered -+ .— Fl M + C 2 Fl M+C K1 (8) -M+ + C + Pl’CM+ K — K K ( ) -+ + - + F1 M C + F1 CM K3 (10) _+ — F1 M C + C 2 Fl CM+C Ku (11) 30 The first type of ion pair prevails over the second one, 1222, K1 > K2, in the less polar solvent THP, as it does for potassium salts in both solvents due probably to the weaker specific solvation of K+ as compared to Na+ or Li+. In general the formation of an Fl'CM+ complex from Fl-M+C is exothermic and so is the formation of Fl-SM+ from Fl-M+ where S is a cyclic ether solvent molecule (72). Therefore a temperature decrease results in an increase in the con- centration of both the solvent-separated and the ligand- separated ion pairs. This also has a profound effect on the selectivity order (72). The kinetic origin of the varia- tions seen in the thermodynamic parameters is unknown. Considering the large number of species present in such systems, 222;: contact, solvent and ligand-separated ion pairs, 1:1 and 2:1 (C:M+) complexes, solvated and com- plexed ion pairs, ion triplets . . ., and the necessity of taking into account the temperature dependence of the con- centration of each species, the treatment of data obtained by any technique may be extremely complicated, as shown by the work of Khazaeli (75). Usually a single technique cannot "see" all the species present. Electronic spectros- copy, whenever possible, and NMR spectroscopy of several nuclei, as well as electrical conductance and electrochemical techniques must all be used to assemble the puzzle. The solvation state of the ion pairs and the kinetic parameters may be altered by changing the solvent, the 31 temperature and, last but not least, the anion. A strik- ing example of the anion effect was recently reported by Lin and Popov (59). They studied by 23Na NMR the exchange of the Na+ ion between its solvated site and its 1806 complex in'THF and 1,3-dioxolane, using the salts NaBPhu, NaClOu and NaI. With NaBPhu the exchange is slow at 25°C in both ethereal solvents, whereas it is fast with the other salts. The coalescence temperature decreases by about 60°C when BPh; is replaced by C10; or I‘. This effect has not been fully rationalized yet, but it is most likely related to the fact that, in THF at 25°C, NaBPhu is a solvent-separated ion pair (76) whereas NaClOu is a contact ion pair (76,77). Cation-anion interactions can drastically affect the rates of the decomplexation step even for cryptates. For example Lehn E£.§£~ (32) observed that the coalescence temperature of the 1H NMR peaks of 0222 and T10222+ de- creases from +39°C to -6°C upon replacement of T101 by T1N03. Yee 33 al. (25) found a strong association between the tripositive Eu3+ and Yb3+ cryptates and small anions F- and OH'. They could not decide if the cation-anion interaction is direct or not, but the presence of signif— icant concentrations of the OH- and F- ions had a marked accelerating effect on the dissociation. In their detailed IR study of the nitrate ion structure in the system + - Na ~C221 NO3 (in DMSO-d6 solutions), Edgell gg al. (72) 32 detected two NO3 sites. One of them was attributed to the solvent-separated ion pair. The second site was tenta- 3 would interact with the sodium cation through the strands tively assigned to a "close ion pair" in which the N0 of the cryptand molecule having replaced one or more sol- vent molecules. Crystallographic studies of solid macrocyclic complexes have shown that the cation is either coordinated to one macrocyclic ligand only (79,80) or to the ligand plus (i) one anion (81,82) or (ii) two anions (83) or (iii) a water molecule which in turn is H-bonded to an anion (8A,85) or (iv) a second ligand molecule (86) or (v) two solvent mole- cules and two anions alternatively (87). This list is not exhaustive. X-ray crystallographic data, although very useful, indicate rather than demonstrate the geometrical structure of the ion pairs in Solution (88).‘ In summary, the kinetics are better understood for cryptates than for crown complexes. Fast rates and ion pairing effects for the latter drastically increase the complexity of the data collection and the data interpreta- tion, respectively. For both systems the actual mechanisms might be much more involved than the apparent ones. A clear picture of the mechanisms will not emerge before the conformational equilibria are fully understood. 33 C. Conformation and Solvation of Cryptands and Cryptates A recent compilation of thermodynamic data on macro- cyclic complexes (10) reveals that most cryptates and crown complexes are enthalpy stabilized and entropy destabilized. Only a few cryptates in water do not follow this rule (56). In nonaqueous media the release of solvent molecules from the cation and the free ligand upon complexation introduces a favorable entropy term. On the other hand, the conforma- tional contribution to the entropy of complexation is negative since the number of conformations available to the ligand decreases upon complex formation (89). This effect is important enough to overcome the entropy increase as- sociated with desolvation of the ion and the free ligand. Cryptands exhibit interesting conformational properties. Like the macrobicyclic diamines of Simmons and Park (90, 91), cryptands exist as equilibrium mixtures of three con- formational isomers (6). These forms, which are denoted as in-in, in-out, and out-out, depending on the respective positions of the bridgehead nitrogen lone pairs inside or outside the cryptand cavity (Figure 2), may interconvert rapidly via'nitrogen inversion. It has been shown (6,92) that in the solid state the cryptand 0222 is in the in-in form whether it is free or complexed. In solution the situation is far from clear for the free ligand but the in-in form is strongly favored 3A out-out in-in out-in Figure 2. Three possible forms of Cryptand 222. in the complex since it allows both nitrogen to contribute to the complex stability. The formation of cryptates brings about an increase in the rigidity of the ligand which results in the non-equivalence of the two protons of certain CH2 groups of the non symmetrical ligands, QLgL, C221 (93). Crystal structures of C222.M+ with cations of increas— ing sizes (M+ = Na+ (9A), K+ (95) and Cs+ (96)) show a progressive opening up of the 0222 cavity with torsion of the ligand around the N...N axis. For a given molecular cavity the larger the cation size, the larger is the extent of the overlap between the outer orbitals of the cation and those of the ligand binding sites, and the larger the down- field shift observed by alkali metal NMR. For the tight fitting inclusive C222-Cs+ complex, Kauffmann g3 a1. (97) reported a 13303 NMR chemical shift of +2AA ppm (lg aqueous 03+). Conformational information may be obtained 35 by 1H NMR; however, except for the highly symmetrical 0222, the complexity of the patterns and superimposition of the signals render the analysis difficult (93). Kndchel eg‘a1. (98) found a relationship between the donicity of the solvent and the chemical shift of the N-CM2 protons of 0222. According to them the position of the signal of the N-CH groups allows an estimation of the 2 conformation of the cryptand in solution. However their data are restricted to only three solvents, CD013, CD3OD and D20, and this precludes a completely general interpre- tation. Ligand solvation and ligand conformation are intimately related. For example, strong ligand-solvent interactions may limit the number of conformations available to the . ligand. It has been stated that the solvation of cryptands and cryptates is small or at least does not change very much from solvent to solvent when compared with the ionic solvation (99). Indeed several authors even proposed that cryptates might provide a new method for obtaining single- ion enthalpies (AHt) and free energies (AGt) of transfer (99-101). However in 1977 Abraham g3 a1. (102) found by direct calorimetric measurements that the enthalpy of transfer of C222 from water to methanol, MAH(C222), is +13.9 kcal.mol'1. This represents about three times (in absolute values) the corresponding AH for Na+ and K+ t ions (-A.9 and -A.5 kcal.mol-l, respectively). The same 36 authors showed that the extrathermodynamic assumption [AHt (C222.M+) = 0] does not hold for enthalpies of trans- fer in the HZO-MeOH system (103-105). The very large HAHt (C222) is almost totally compensated by the entropy of transfer [fiASt (0222) = A3 cal.mol-1.deg-l] to give M - WAGt (0222) = 1.1 kca1.mol 1. The large increase in entropy indicates desolvation of 0222 and/or an increase in the number of conformations available to the ligand. Besides, the substantial variation in WAGt (C222.M+) with the central ion M+ indicates that the central ion is not shielded from the environment (103-105). Water and methanol form strong hydrogen bonds with the two amine nitrogens and, to a lesser extent, with the ether oxygens of an uncomplexed cryptand. The hydrogen bonding is, at best, considerably weaker in a cryptate in which the ether oxygens and the two nitrogens are bonded to the metal ion (106). Between dipolar aprotic solvents S A S S transfer activity coefficients ly 2 of C222-M+ with M+ = Na+, K+, T1+ and Ag+, are almost equal to that of the 1 and 82, the cryptand, but due to H—bonding this equality does not hold if S1 is methanol (105). D. Interaction of Conventional Ligands with the Alkali Cations In view of the rapid expansion of the coordination chemistry of alkali and alkaline earth cations, recent 37 studies of the alkali complexes with the non macrocyclic ligands are rather sparse, although as a review article by Poonia and Bajaj (13) shows, some attempts to study such complexes have been made even in the pre-macrocycle days (107). In most cases, however, such work was directed towards the isolation of new adducts of the alkali salts with a variety of chelating or non chelating ligands and anions (108). For example the first solid alkali complex of 2,2'-bipyridine (BP) seems to be (K-BP)+PhuB-, recently isolated by Grillone and Nocilla (109). However, few attempts have been made to study the existence and the stability of such adducts in solutions, and particularly in nonaqueous solutions. In general alkali cations show a greater tendency to complex with O-donor than with N-donor ligands (110); alkali complexes of 1,10-phenanthroline have~received the most attention (107-112). By contrast, 2,2'-bipyridine does not seem to exhibit much complexing ability towards these cations (111,113). E. Nuclear Magnetic Resonance in Molten Salts Molten salts differ from conventional nonaqueous sol- vents in two major respects; high melting point and strong ionic character (11A). The latter confers high electrical and thermal conductivity and promotes the solubilization of ionic solutes, e.g., metal oxides. In general molten 38 salts are better and more versatile solvents than water. Although molten salts have been studied mostly by electrochemical methods (115-120) it was felt as early as 1958 that NMR techniques might offer valuable information with regard to identifying the species present and determin- ing the small structural changes or the changes in the chemical bonding which accompany a variation in the composi- tion of the melt. Due to the favorable properties of the 205T1 nucleus (spin 1/2, high sensitivity, large chemical shift range), Rowland and Bromberg (121) followed by Hafner and Nachtrieb (122,123) studied solid and molten thallium salts. Except for T1(C10u)3, the 205T1 NMR shift on fusion is in the paramagnetic direction, and the magnitude of the shift is less than the difference between that of the melt and that of an aqueous thallous solution. The direction of the shift was accounted for by overlap effects due to a decrease in the cation-anion distance (122). The authors (122,123) showed that NMR provides a valuable means of exploring deviations from purely Coulombic interactions between ions. The degree of covalency in the cation- anion bond was found to increase with temperature. In Tl+x‘-M+x' (M+ = alkali cation) mixtures, the 205T1 shift varies linearly with the mole fraction of M+X- added, and the direction of the shift depends upon the radius of the alkali cation (123). In 1973 Harold-Smith studied the structure of the alkali nitrates by measuring the 39 spin-lattice relaxation time T 23 7 of Na in NaNO3 (12A) 1 and of Li in LiNO3 and Li/KNO3 mixtures of various com- positions (125). The quasi-lattice, random-flight model (126,127), which assumes the existence of a pseudo lattice for the liquid, was found to apply to molten NaN03, distance de- 3 creases whereas the electric field gradient at the lithium Concerning the mixtures, the average Li+-NO nucleus increases with increasing mole fraction of KNO 3 (125). Alkali halides and nitrates are typical ionic salts (128). At the other extreme are compounds such as aluminum chloride which undergo only slight dissociation into ions and in which more complex bonding forces oper- ate (115,116). The first attempt to apply NMR methods to the study of the chloroaluminates (A1013 - alkali halide mixtures) seems to have been that of Anders and Plambeck (129). These authors measured the 23Na, 27Al and 3501 line widths in the low melting ternary A101 -NaCl-K01 3 system. The 27A1 line width increases with added A1013 but the 23Na line remains sharp (9 Hz). In addition there was no detectable 23Na or 27A1 chemical shift with changing stoichiometry. The authors concluded that fused alkali 7 ionic, the entities in the melt being primarily M+ and chloroaluminates of A1201 stoichiometry are essentially A12C17. (01‘, A1013, A1201", A130110, A12Cl6,...) depend on the The species present in chloroaluminate melts A0 composition and are still under debate at the present time (130). Recent developments in molten salt research include 1H and 13C NMR studies of AlCl3-based fused salts that are molten at or near room temperature (131,132). sgr. ..._ F. Potassium 39_Nuclear Magnetic Resonance A 39K, 0K and “1K, possess a Three potassium nuclei, magnetic moment (133). Potassium—39 is the most sensi- tive as well as the most abundant in nature (93.1%). How- ever, the sensitivity of this nucleus is still about 200 23 1 times lower than that of the A 2 Na nucleus, i.e., 5 x 10- lg 9.3 x 10' with respect to H at constant field (61). Consequently 39K NMR studies have been rather sparse even after the development of the Fourier Transform NMR tech- nique and the availability of high field superconducting magnets. Apart from the sensitivity, the general features of 39K NMR are quite comparable to those of 23 Na NMR which have been described by several authors (61,3A) and which were recently reviewed (57). Several theories have been 23 proposed to calculate the Na chemical shifts and quad- rupolar coupling constants and to relate these two quan- tities (61,13A). Given the similarities of the two nuclei, only the main features of 39K NMR are mentioned here. For the 39K nucleus: A1 - The resonance frequency is low (9 = 2.8 MHz in a field of 1.A Tesla). - 39K has a spin of 3/2 and it possesses a quadrupole moment of 0.07b (61), smaller than that of 23Na (0.10b). This is a sizeable difference when considering that the linewidths vary as the square of the quadrupole moment. - The dominant relaxation mechanism is quadrupolar and occurs through molecular reorientation of the solvent. — The natural line width (narrowest one observed thus far in aqueous solutions) is about 6 Hz (133). - In solution the chemical shifts range from about -20 ppm to +20 ppm with respect to K+ at infinite dilution in water (15,135). - Changes in the paramagnetic screening constant, 0p, are much larger than changes in the diamagnetic screen- ing constant, 0d, and thus dominate the chemical shift (61). - According to Kondo and Yamashita's theory (136), the most important contribution to the chemical shift arises from the overlap repulsive forces between the 3p orbitals of K+ and the outer s and p orbitals of the neighboring anion(s), ligand(s) and solvent molecule(s). - The extreme narrowing approximation (137), i.e., 39K wt << 1, where w is the resonance frequency of the C nucleus and TC is the correlation time for solvent re- orientation, is valid down to the melting point of common A2 solvents since w is very small. This implies T1 = T2. - The resonance line shape is Lorentzian (FT of an exponential function) and the line width Av 2 (full width 1/ at half height) equals l/nTé. - Since the magnetogyric ratio 7 is small, the con- tribution from the magnetic field inhomogeneities to the observed line width, yAH/2n, is small and may often be neglected (see details in the experimental part of Chapter I), 39K NMR studies reported An extensive review of the up to early 1978 can be found in the Ph.D. thesis of J. S. Shih (15). Very few papers have appeared since then (138-1A0). A brief review of the 39 K NMR work is given below. I The 39K chemical shifts of several potassium salts in aqueous solutions were measured by Deverell and Richards (1A1) and by Bloor and Kidd (1A2) who used relatively low fields (.0 9A0.mmv 0.mH ww.H m.mw AQHMB 666.0: 6H0.0 H.mm m6.6 2.6H 62H0Hs22 s... .662. .62.. .6666. mmwwwmwmmw wmmewm 0262600 6HoaH0. .06Hu22 so 26HHH6H6066606 oHpmcwmz pom COHpooHHoo 0:6 moHpHodopm uco>Hom 202 .0 oHnme 53 gas and a spin rate of 10-12 Hz were maintained in order to minimize the temperature gradient across the sample. (c2) Other Studies - Samples (5 or 10 ml) were con- tained in 15 mm or 20 mm tubes, depending upon whether D20 or another deuterated solvent was used as look. In the latter case the chemical shifts may not be directly compared since the magnetic field has to be adjusted for lock purposes in going from one solvent to another. This arises from the fact that the 2H resonance frequency varies with the solvent (and also with the temperature). For example at 25°C we found = 6D 0 + 2.2 (3) 5 Acetone-d6 2 where the subscripts denote the lock solvents. The cor— rections are about 2 ppm for CQ3OD and 2.8 ppm for CD3CN. These corrections were not systematically applied in the studies at variable temperature. d. Line Width Measurements - Line width calibrations were made with a 0.25 M solution of KI in a 3:1 (vol.) H20-D20 mixture. The half height line width (Av1/2) of the 39K NMR signal (25 scans, SW = 5000 Hz, 32 K of memory) of this solution was 7.5 i 0.3 Hz (corresponding to T2 = l/wAv1/2 = 02 i 2 msec.) before and after each set of 54 experiments. A comparison of this value with the smallest one reported in the literature (133) for potassium salts at infinite dilution in H20, ;;2L, 5,7 0,6 Hz (T2 = 56 i 6 msec.), indicates that the line width contributions from field inhomogeneities range from 1 to 2 Hz at the most despite the large volumes of sample used in this study. This contribution is smaller than, or comparable to, the estimated precision of our measurements which is about 2% for successive runs with the same or different samples. In our kinetics study, rate data were obtained from differences in line widths so that small systematic errors.would essentially cancel out. For these reasons, the measured line widths were not corrected for inhomogeneous line broadening. The relaxation rates l/T2 were calculated from the line widths (l/T2 = wAv ) of the absorption 1/2 spectra. e. Data Acquisition and Signal Processing - The high quality spectra required for kinetics measurements are particularly difficult to obtain in 39K NMR due to the very low sensitivity of this nucleus. To overcome this problem we used a high magnetic field and a large sample' volume, but it was still necessary to carefully optimize every step of the data acquisition in order to obtain spectra with large signal to noise ratio in a reasonable amount of time (<2 h routinely). 55 Signal Averaging + In dilute solutions (<0.l H in K ) 39K signals cannot be detected in l scan and extensive signal averaging is necessary. At 25°C the typical numbers of scans (NS) to obtain spectra with a S/N ratio of about 20 for a solution 0.1 M in K+ (20 mm O.D. tube) are NS 20 Hz 2000 if Avl/2 = 50 Hz and LB NS 50000 if 300 Hz and LB 80 Hz Av1/2 where LB is the artificial line broadening (see below). Fortunately 39 K relaxation times are short so that many signals can be accumulated in a short time. However care must always be taken to avoid both truncation and satura- tion which distort the line shape and cause erroneous line width measurements (176). All measurements were done in such a manner that the NMR signal decayed completely during the time interval (>5 Tl) between the rf-pulses. Even when this condition applies, a very fast acquisition often leads to some baseline distortion; 1 K was the minimum memory size used in this work. Zero Filling Technique The theoretical and practical aspects of this tech- nique were recently reviewed by Lindon and Ferridge (177). 56 In short, one can increase the point to point resolution and even improve peak lineshapes and positions by adding more than N zeroes to an N-point free induction decay (FID) prior to Fourier transformation. We used this technique for all the kinetics studies except those in acetone and 1,3-dioxolane. Typically the spectra accumulated using 102” or 20MB points of memory were zero filled to 8192 or 16384 points (177,178). With a spectrum width of 5000 Hz, the chemical shifts were accurate to $0.3 ppm and the line widths to 12%. For the low sensitivity nuclei and the broad signals, the zero filling technique brings about 'a considerable time saving. Furthermore, without it, the acquisition of good quality spectra in unstable super- c001ed solutions, E;§;, K+°18C6 complex in Acetone-THF (80-20% vol.) at -u0°C, would have been impossible. Sensitivity Enhancement The FID's were multiplied by the universally used negative exponential weighting function, so that the line shape remains Lorentzian after transformation. This method improves the sensitivity and does not introduce any lineshape distortion (177). For each spectrum a compromise must be found between sensitivity enhancement and undesirable line broadening. Typical artificial line broadenings (LB) applied to our spectra were 57 LB = 5 - 10 Hz for AVl/Z = 50 Hz LB = 15 - 50 Hz for AVl/2 = 100 Hz LB = 50 - 120 Hz for AVl/2 = 300 Hz. For our systems the useaof a matched filter (176), i.e., LB = Avl/2’ results in a very broad line especially for + K -crown complexes and therefore is not recommended. _2. Carbon 13 NMR 13 a. Instruments - Most C NMR measurements were performed on a Varian OFT-20 spectrometer operating at a field of 1.868 Tesla and a frequency of 20.0 MHz. A Bruker WM-250 spectrometer operating at a field of 5.87 Tesla and a frequency of 62.9 MHz was also used. .The latter was Coupled to an Aspect 2000 computer with 32 K of memory. A large number of measurements were made on both spectrometers. b. Locking Procedure, Referencing and Corrections - The sample solution was usually contained in an 8 mm o.d. NMR tube (Wilmad) which was coaxially centered by means of teflon spacers in a 10 mm o.d. NMR tube containing dry acetone-d6 as the lock. The methyl carbon peak of acetone- d6 was used as the external reference. Carbon-13 chemical shifts were corrected for the differences in bulk diamag— netic susceptibilities between the sample solvent and 58 acetone according to Equation (2) (WM250) or to Equation (u) (CFT20) ' = 21 ref , sample acorr aobs + :3 (Kv ' Kv‘ ) (u) sample v When values of K were not available from reference, approximate corrections were obtained by running the same sample on the two spectrometers. The correction in Equation (4) is a third of the difference in the observed chemical shifts between the two spectra. Table 7 gives the corrections for each spectrometer. c. Temperature - The temperature was measured before and after each experiment with a calibrated Doric digital thermocouple inserted in an NMR tube containing acetone (CFT 20). For temperature measurements on the WM 250, see Paragraph IIIA3. 3. Data Handling The kinetic parameters and the stability constants of complexes were calculated by fitting the rate-temperature data and the chemical shift-mole ratio data respectively to appropriate equations using a weighted non-linear least- squares program KINFIT (179) on a CDC 7501 computer. Details on the use of this program are given in the Ap- pendices l and 2. 59 Table 7. Diamagnetic Susceptibility Correctionsa with Respect to Acetone—d6. Correction for Correction for Varian OFT-20 - Bruker WH-l80 Solvent and Varian DA-60 and Bruker WM—250 Acetone 0 0 MeCN -O.155 +0.310 CHCl3 -O.587 +1.17“ CH2012 -O.518 +1.036 Formamide -0.l91 +0.382 DMF —O.237 +O.U7u DMSO -O.3ON +0.508 MeOH -O.ll6 +0.232 EtOH -O.2u2 +0.”8u Nitromethane +0.1uu -0.288 Nitroethane -0.078 +0.156 l—Nitropropane -0.l8b +0.36b Pyridine -0.319 +0.638 PC -O.365 +0.73O Tetramethylguanidine -0.273 +0.5A6 H2O- -O.545 +1.09 THF -O.321 +0.6U2 1,3-Dioxolane -0.I40b . +0.81b Ethylene Glycol -0.508 +1.016 Toluene -0.331 +0.662 Anisole —0.uuu +0.888 Methyl Acetate -0.162 +0.32“ ain ppm at 25:10°C. bThis work, approximate values (see text). CHAPTER III KINETICS OF COMPLEXATION OF POTASSIUM CATIONS WITH l8-CROWN—6 IN NONAQUEOUS SOLVENTS BY POTASSIUM 39 NUCLEAR MAGNETIC RESONANCE 60 A. Introduction It is surprising to note that while hundreds of forma- tion constants of macrocyclic complexes are known, the kinetic properties for the complexation of alkali ions with 18C6, one of the most common and early synthesized crown ethers, have been investigated only in water. Shporer and Luz (A8) showed that the exchange of the po- tassium cation between its solvated site and the K+°DB18C6 39K NMR. From complex is slow enough to be observed by these results it was clear that the same technique could be used to study the complexation kinetics of K+ol8C6 in ' a variety of solvents because the stabilities of the K+°l8C6 and K+-DB18C6 complexes are comparable (log Kf 2.03 and 6.05 for K+ol8C6 in H 0 and MeOH respectively vs 2 1.67 and 5.00 for K+°DB18C6 (10)) and probably so are the decomplexation rates. The goal of the work presented in this chapter was} to determine if the activation energy for the release of an alkali cation from a small crown molecule (18C6) depends on the solvent or n0t, since the previous studies concerned with this problem were not conclusive. Spurred by the results of Lin in 1,3-dioxolane and THF (59,1A8), we were particularly interested in studying the ethereal solvents. 61 62 B. Choice of Solvents and Salts The relatively high concentrations required for detec- tion by 39K NMR (>0.05 m) and the low solubilities of potassium salts and K+ol8C6 complexes place some restriction on the choice of solvents and salts. Many other factors, which further limited our choice, had to be considered. The donor abilities of the various solvents used in this study had to be comparable since we wanted to con- centrate on the effect of changing the dielectric constant and the solvent structure upon the kinetic parameters. For this reason acetone was chosen as the primary solvent since it has a donor number (1A9,150) of 17 close to those of THF and 1,3-dioxolane (Table 8). It is a polar solvent in which all the species of interest are sufficiently soluble (1 0.1 M) for 39K measurements even at low tem- peratures. It is miscible in any proportion with the three cyclic ethers used in this study. It has a low melting point as well as a low viscosity so that 39K NMR signals are not too broad in it. Methanol was selected because the only reported kinetics study by 39K NMR (A8) was done in this solvent, thus allowing some comparisons to be made. In addition to THF and 1,3-dioxolane, which we used in an attempt to observe with K+ the slow cation exchange seen ~with Na+ (59,1A8), we chose l,A-dioxane in connection with its interesting behavior in the propagation of anionic polymerization (76). This reaction involves free ions 63 such as styrene—(8‘) and ion pairs such as S-Na+. The concentration of free ions increases rapidly in THF but slowly in dioxane. The S-Li+ ion pair is the most re- active and the S-Cs+ the least reactive in THF but the pattern is reversed in dioxane (76). We also resorted to acetone-dioxane and acetone-THF mixtures due to solubility problems. The key properties of the solvents are given in Table 8. Potassium hexafluoroarsenate is soluble in many or- ganic solvents (151,152), including THF (153) and 1,3- dioxolane. Approximate solubility measurements in these two solvents gave values of 0.18 M and 0.20 M, respectively. Potassium halides are insoluble (KCl m 10'“ m) in THF even in the presence of DCl8C6 (”,15A). Even potassium salts with soft anions such as ¢COO-, CH3C02’ picrate or B0; (153) are insoluble in THF and 1,3-dioxolane.' We used KAsF6 for all our measurements except those in MeOH where K1 was used. In most organic solvents, the solubility of 18C6 is greater than 0.1 M (ELgL, >0.22 M in MeOH (155)), but the K+-18C6 complex is only sparingly soluble in ethereal sol- vents. Using KAsF6, we found that, at 25°C, the solubility of K+-18C6 is less than 0.01 M in THF and is about 0.035 M in 1,3-dioxolane, whereas both the free salt and the ligand are soluble (>0.1 M) in both solvents. This is in contrast with the general enhancement of the solubility of inorganic 6A mzz 020 69 0060600600 .6006 00206000m 0060 0060650600 mno ppoo m mo0hpoo + 0 u 0 06356om 0:6 :6 00050066:6 00 OH 6066.6 .6 m 6 66. u 26 66 page 6666.6 +|||u 66666.6 006 60 00:6000 66 60055: pocoo acmeusw 0:8 .006 00:060m0m0 .666 00:0600000 6000 + m CO660006 0:0 po0 6060:»:0 0>66000c .66660 .co66 .6666.m66v .0060060c6 066360£uo 6606:: 000 06 000 a 060061066.6 06oomuvm.66 000.0: 060.0 00 00: 06.06 660.0 00.6 6.60 0002 QAOONIV0.HH 066.0: 060.0 00 006: 0.06 006.0 06.6 0.6 . 006 66 66. 66.66 66.6 0:66oxo6aum.6 066.0: 000.0 606 0.66 0.66 Aoomv600.6 06.0 6.6 0:06o6ml6.6 .6 .6o> 606606 . 0cmxo60 060.08 0068 I6.6|0COp0o< 060666066.o 06oomuvm.m6 00.6: 006.0 00 00: 0.66 600.0 00.6 6.06 0cou0o< 65000 06x6: AOoV 6000 6200 6000 A06n000 pampmcoo 600>6om gmpmznfixm . 000630 .Q.n .Q.E 9609532 6066oom6> pcmsoz 0660006060 0.0066o0 .6o> pocom 06oa60 .660605oppo0am 006:23 0:0 006:03 606560 so 6060666966000630 0660:002 pom 6co600066o0 0:0 06066600060 uc0>6om 60x .0 06909 60 65 salts in organic solvents by crowns and cryptands (156). A decrease of solubility upon complexation has already been observed by Wong 23 El- (55) for several crown complexes of fluorenyl alkali salts in ethereal solvents and by Cambillau 33 al. (157) for the 1806 complex of potassium enolate in THF. The poor anion solvating capacity of ethers (150,158) might be responsible for this effect. Complexation of K+ by 1806 breaks a large proportion of contact ion pairs (89,159), as shown below. This in turn leads to anion activation (160) and in our case to pre- cipitation of the complex. In MeOH, the K+-18C6 complex precipitates if PF; or Ang is used as the anion but with I’ its solubility is greater than 0.2 M, due probably to the possibility of weak H-bonding with MeOH. It is knOwn that PFE has a poor coordinating ability (161,162); the ASFE anion probably shares this property (151). These results show that any complexation data based on increased solubility of inorganic salts in low polarity media (and also in MeOH) may not yield a clear picture of how ef- fective the various salts complex with crown ethers. C. Results and Discussion The kinetics of complexation of 18C6 with potassium cation were studied in five pure or mixed solvents by 39 using K NMR line shape analysis. The five solvent systems were acetone, acetone—l,A-dioxane mixture 66 (80-20% vol.), acetone-THF mixture (80-20% vol.), methanol and 1,3-dioxolane. The structures of the three cyclic ethers are shown below. ' O THF 1,3-dioxolane l,A—dioxane As pointed out in the Introduction, the treatment of kin- etics data obtained by alkali metal NMR depends on the differences in chemical shift and line width between the solvated cation and the complexed catiOn, in the absence of chemical exchange. Therefore it is necessary to characterize the two sites before discussing the kinetics data. In this discussion, site A refers to the solvated site and site B refers to the complexed site. 1. Potassium-39 NMR of the Solvated and Complexed Potassium in the Absence of Chemical Exchange To obtain the transverse relaxation rates, l/T2A and l/TZB’ and the chemical shifts, 6A and 6B, in the ab- sence of exchange, two solutions in each solvent were prepared. The first contained the potassium salt only 67 and the second contained equimolar amounts of the potassium salt and of l8C6. Since the K+°18C6 complex is quite stable, l;2;: KA > 10Ll (10), in all solvents studied and throughout the temperature range considered, the second solution contained only complexed potassium cations. In both solutions chemical exchange is absent. The temperature dependence of l/T2 and l/T2B is given A in Tables 9-13. Figure 3 shows semilog plots of these ob- served relaxation rates vs reciprocal absolute tempera- ture for all the solvents studied. Figure A shows the temperature dependence of the chemical shifts for sites A and B. 622.6. For K+ ion in solution, we expect to find T1 to be equal to T2. This is indeed the case as is shown in Figure 5 where the reported longitudinal relaxation rates l/Tl for a solution of 0.5 M KI in methanol (“8), are plotted together with our values of l/T2 for a solution of 0.2 M KI in MeOH. The two solutions differ only by a small change in viscosity and the two sets of data are in good agreement. However, while Shporer and Luz (A8) drew a straight line through their data points, our values which are much less scattered clearly show a curvature. Figure 3 also shows some curvature for solvents other than methanol. Therefore the curvature must be real. 68 60.2 12.5 -23.2 -510 —73.g -91.4 °c ' I I f 2000 '- d Figure 3. Semilog plots of potassium-39 transverse relaxa- tion rates for the solvated (open symbols) and the complexed (closed symbols) in various solvents. Cl Dioxolane; O MeOH; v Ac-dioxane; 0 Ac-THF; A Acetone. 69 60.2 12.5 -23.: 51.0 -73‘._2 -9L4 °c +2L ppm Figure A. Potassium-39 chemical shifts for the solvated (open symbols) and the complexed (closed sym- bols) K ion in several solvents.A 1,3-dioxolane; O acetone; 0 acetone-dioxane (80-20% vol.); V acetone-THF (80-20% vol.), 0 methanol. 800 600 2° I n l_____|____l_6 3.5 4.0 4.5 5.0 5.5 103/1" K" Figure 5. Potassium-39 relaxation rates for solvated K+ ion in methanol. .5 0.5 M_KI (Reference “8); O 0.2 M KI (this work). .0600006 0609p 600 x 0 so 0 6 0666 zpos0z0 .00-0:op00< 0:0 .mpoE0E no .66666 a 6.66 060 :003009 >0:0500:0 00:0:060: m :6 00:0606660 0:» non 00po0apoo 602 6 0 x 06 :o x 0 0:66: 00006553000 0603 06900000 .00-0:ou00¢ ”9:0>6om xooq 71 9 0.01 600 060.0 0.00- 0.0I 606 666.0 0.60- 6.0- 606 000.6 0.66- m.0l N06 mm0.6 m.00l 0.6 I How m:m.m 0.60I 6.6- 606 006.6 6.66- 6.0 - 666 666.0 0.66- 0.6I 66m 060.6 6.mml 0.0 I M66 000.6 0.66I . 0.6- 006 000.6 6.66- 0.0 I 60 606.6 6.60- W 0.6- 000 006.6 0.00- 0.6 I 00 060.6 6.60- 6006 060.6 6.60- 0.0- 066 600.0 0.06- 0.6 I 06 000.6 0.66- 0006 006.6 0.mml N.6I 066 600.0 0.M6I 6.0 I 00 N06.6 0.6m- 066 600.0 0.06- 0.0- 660 600.0 0.0 6.0 I 00 600.0 0.06- 666 666.6 6.66 6.6- 666 666.6 6.66 6.6 - 66 666.6 6.66- 666 666.6 6.66 6.6- 666 666.6 6.66 6.66- 66 666.6 6.6 6.6- 000 066.0 0.66 0.0- 666 000.0 0.00 0.66- 66 600.0 0.66 E00 6.6-6 6Ix 0o E00 Mum 6Iv6 0pc .500 mum 6Ix ouo 00 B\6 B\006 u 00 B\6 B\006 00 B\6 B\006 060.6 u m: 00.0 n m: 00 u m: 9.606360600509 mSO6p0> 00 0:0 60.6 0:0 060 .0 go 606000 0602 +x\0006 00 0006 0:0 000600 0:6:60p:oo 6:06us6om wmmmmmfl you 60569 :06p0x06mm 0660>6:0LB 6000606000 0:0 600690 60065090 022 001556660000 .0 06906 72 Table 10. Potassium-39 NMR Chemical Shifts and Reciprocal Transverse Relaxation Times for MeOH Solutions Containing Kla and 1806 at 18c6/K*‘Moie Ratio (MR) of O, 0.5 and 1.02 and at Various Tempera- tures. MR = 0 MR = 0.5 56 lOEiT 1(3? 5 J: lofiT 1(3? 5c s ppm C K s ppm 24.7 3.357 56 —7.H 2u.7 3.357 217 —5.0 - 0.9 3.672 75 -6.6 1H.7 3.H73 236 -H.6 - 5.2 3.731 81 -6.H 8.2 3.55H 251 -H.5 -1o.u 3.805 85 -6.2 3.0 3.621 26u' —u.3 -1H.9 3.872 90 -5.9 -2.1 3.689 273 -H.1 -19.7 3.9H5 95 -5.8 -6.5 3.750 283 —u.o -25.3 H.03H 107 -5.0 -11.6 3.823 305 -H.O -29.5 H.103 107 -5.H -l6.6 3.897 333 -3.7 -3H.3 H.186 12H -5.1 -21.6 3.975 355 -3.H -38.9 H.268 129 -H.9 -26.2 H.0H9 . 377 -3.5 -H3.8 H.359 1HH -H.8 -31.H H.136 . H05 -3.5 -H8.6 H.H52 160 -H.6 -HO.7 H.301 H18 -3.9 -53,9 H.560 176 -H.3 -H5.7 H.396 H12 —u.1 -58.l H.6H9 195 -H.2 -50.7 H.H9H H08 -H.3 -63.5 H.769 229 -3.9 -56.2 H.608 377 -H.H -67.9 H.871 261 -3.7 -60.6 H.7OH 352 -H.5 -73.5 5.008 31H -3.5 -65.9 H.82H 336 —H.O -79.5 5.163 H02 -3.3 -68.6 H.888 3H2 -H.O -83.9 5.283 H87 -3.1 -70.l H.92H ‘ 339 -H.O -89.H 5.HHl 628 -2.7 -75.9 5.068 H12 -3.3 -9H.0 5.580 785 -3.1 -81.5 5.216 H87 -3.2 73 Table 10 - Continued. MR = 1.02d t 10312 l/Té 50 °C K- "" 5.1 ppm 2h.7 3.357 386 -2.5 17.1 3.uu5 377 -1.9 2.9 3.622 u62 -2.o -11.8 3.826 540 . -2.9 -25.8 u.ou2 666 —1.0 -38.3 H.257 832 -1.3 —53.2 n.545 1056 -o.2 -65.2 H.808 1366 -0.2 a0.2 m KI. Lock solventzMethanol-dU. bAll spectra were accumulated using 1 K or 2 K data points and zero filled to 16 K. cUncorrected for the differences in 2H resonance frequencies ‘ between D20 and Methanol-d“ (N2 ppm). dNumber of scans = 6000 (high temp.) to 15000 (low temp.). 7 u Table 11. Potassium-39 NMR Chemical Shiftsa and Reciprocal Transverse Relaxation Times for Solutions Con- taining KAsF6 and 1806 at 1806/K+ Mole Ratio (MR) of 0, 0.25, 0.50 and 1.02 in liggDioxolane and at Various Temperatures. MR = ob’g MR = 0.25b’f t lofiT 1(3? 5 t 10311 173? 8 °C K S ppm °C K 8 ppm 66.5 2.9u6 47 -17.0 61.2 2.990 239 -l6.3 56.5 3.035 50 -16.8 55.2 3.0u5 239 —1u.7 u6.2 3.133 50 -16-2 51.0 3.089 261 -1u.u 35.8 3.238 63 -15.8 u5.9 3.139 280 -1u.9 21.2 3.399 - 66 -1u.9 no.6 3.187 283 -1u.6 9.6 3.539 88 -14.5 35.u 3.290 283 —1U.6 0.0 3.663 9n —1u.2 29.9 3.299 26a -1u.8 - 9.8 3.796 122 -13.6 2u.6 3.358 229 -15.2 -19.2 3.937 152 -13.1 18.5 3.u28 21M -15.0 -38.2 u.255 258 —12.0 19.2 3.480 192 -1u.7 8.6 3.599 160 -1u.9 u.3 3.602 1H9 -1u.7 - 0.9 3.672 138 -1u.6 - 5.7 3.738 137 —1u.2 -10.3 3.8OM 135 ~13.7 -1u.9 3.872 1M8 -13.6 -19.5 3.992 163 -13.2 -2u.2 u.016 188 -13.1 -29.2 9.098 212 -12.9 Table 11 - Continued. 75 MR = 0.5b’g MR = 1.020,f t 103{T 1(3? 5 t 103{T 1/T2 5 °C K- 8 ppm °C K- s'1 ppm 96.2 3.133 977 -10.1 69.2 2.969 759 -9.8:1 35.5 3.291 528 -12.2 56.9d 3.036 817 -10.1:1 29.0 3.367 993 -l3.8 93.8 3.155 869 -0.3il 15.0 3.970 320 -19.8 33.9d 3.269 992 +1.3:1 9.9 3.591 251 -19.5 23.2e 3.379 1005 +0.9:0.5 5.1 3.593 219 -19.6 0.0 3.660 179 -19.5 - 9.2 3.788 170 -19.1 -l3.7 3.854 179 -13.8 -18.7 3.929 190 -13.5 —23.3 9.002 --- -13.1 aUncorrected for diamagnetic susceptibilities. d6 (see Table 9 footnote d). b0.1 M KAsF5. 00.05 M KAsF6, unless otherwise indicated. d0.09 M KAsF6. Lock Acetone— 8Obtained by extrapolation from data in Acetone-1,3-Dioxo- lane mixtures (see Table 15). fMemory size 2 K or U K. gMemory size 8 K or 16 K. Zero filling to 16 K. No zero filling. 76 Tablella. Potassium-39 Chemical Shifts and Line Widths for KA8F6 at Various Concentrations in 1,3-Dioxolane and in THF at 25°C. 1,3-Dioxolane THF [KAsF6] 6a Avl/2b [KAsF6] 83 Avl/25 M ppm HZ M ppm HZ 0.0077 -15.99 19.70 0.0070 -16.69 15.00 0.0115 -16.27 15.3 0.0120 -16.68 15.0 0.0195 —16.27 13.7 0.0229 -17.06 21.9 0.0176 -16.98 19.5 0.0386 -17.32 19.8 0.0213 -16.60 15.3 0.0505 -17.50 23.0 0.0293 -16.75 15.1 0.0882 -17.68 23.9 0.0995 -17.10 17.8 0.1186 -17.89 26.8 0.0710 -17.19 19.7 0.158 -17.95 30.0 0.115 -17.99 29.0 ai0.07 ppm. bil Hz. 0 For the low concentrations, the reported AVl/2 values probably exceed by l or 2 Hz the real values due to a small instability of the field over the long period of time necessary to obtain these spectra. 77 m.5 ::H mam.: o.mmu 0.5: :NH :m:.: a.» sma om:.: m.omu H.mu mma amm.= 6.» mad Ho:.= o.mau :.mu mmfi mom.: a.» moa mom.a H.H:- m.mu med mmm.a m.m :oa mmm.z 3.6m- H.mn omm smfi.= m.w OOH H:H.= s.am- :.mu mum mmo.: o.m Hm sno.a H.6m- m.au wsafi 530.: H.0m- o.mu mmm osm.m m.m mm msm.m m.Hms m.mu :mw sme.m o.sn m.wu Ham oom.m m.mn mew mom.m o.mH m.mu ma: omm.m 0.0 .m5 :mm.m :.NH- H.mu mam osm.m m.mm mmo.o m.su om: ems.m m.mu :mm osm.m m.mm mo.o H.mu Hm: amo.m mm mmo.m m.m . m.mn smm omm.m 3.3m oa.o $.57 mam :mw.m o.mn mmo mam.m s.wm m.su mam mom.m mm zsm.m 0.0 m.:u mom oom.m m.mm m.mu owm mm:.m s.mu mm: smo.m s.ma mmo.o H.wu mmm mmz.m mm mm:.m m.mH m.:. as: smo.m s.m: mo.o m.mn mam mmm.m mm mam.m m.mm w.:n pm: smo.m s.m: oa.o s.mu mom m:m.m m: mma.m :.mm w.:u em: ozo.m s.mm o.ma ,msm amfi.m m: moo.m m.ma I a Edd manm Hum Do 2 Ed mHIm aux main HIM 090 6.60 E\H B\moa p 6:60 6.66 B\H B\mOH B\H a mod m:o.a m2 m.o m2 m2 0.9.moL3uMLoQEoB m30fi9m> pm cam A.Ho> Romlomv ocmxoaQ|:.HI020poo< CH :o.H 6cm m.o .o no Amzv moapmm 6H6: +x\momH pm momfi cam ammmmCMpB HmOOLanom pew mpmanm HmOfiEmno mzz mmlesfimmmuomgwnmanme 78 .pmxoaa 1500 ohm mcofipmo EsfimmmpOQ one Ham pmnu beamco on pocpm mm: coma mo mmmoxo pawfiam aom xooa onen .oopQOfipcfi omfizmmSuo mmoacs mmm methanol > acetone-dioxane > acetone, for site A whereas the order is dioxolane > acetone-dioxane > acetone > methanol for site B. This last order is also the order of solu- bilities of the complexes. The change seen for methanol is surprisihg.’ We observed a large increase in the relaxation 85 rate of site A in acetone upon addition of 18C6 (see below) or 1,9-dioxane (Figure 16). Thus the presence of these ethers markedly increases the viscosity of acetone solu— tions. This is not so for MeOH. The K+-18C6 complex is probably more solvated in methanol than in other solvents and thereis some experimental evidence that a stronger solvation results in a smaller NQCC (172). However, we can neither measure nor calculate the NQCC or Tc to evaluate their relative contributions to l/T2. 23 Kintzinger and Lehn (60) obtained the Na quadrupolar coupling constants of four sodium cryptates by calculating 13 TC from the C relaxation times T of the CH2 carbons of l the cryptates and introducing T in Equation (1). They C made the assumption that T (from 13C data) also repre- c -sents the reorientational motions which modulate the 23Na quadrupole interaction. This assumption, pr0bab1y valid for the rigid cryptates, may not be reliable for crown ether complexes due to the fluctuation of the coordinating sphere. Thus we cannot obtain the NQCC. The weak solvation of the complex is also indicated by the small solvent dependence of the resonance fre- quencies of the complexed potassium cation. The chemical shifts measured by Shih (15) are given in Table 19 along with our results. While the chemical shifts of the sol- vated K+ ion vary over a range of more than 22 ppm, the signals of the bound site are found between -3.8 ppm and .ma mocopmmom Eopm «ammo .Apxou momv modam> NB\H opsumthEop 30H 0:» Song pmpmaoomcuxmm 86 we H.m w.Hu m.o+ 20mm :02 0.52 m.o- 5+2 nomzm come ma? m.me m.ml ml? . nmzo 5.5 . 6mm mm H: 5.: m.m- 9.5 - mom: m.ma coca ow HMH m.mH :.o+ m.mHI , mamaoxOHolm.H 5.m 5m: om om m.m m.m| w.HHI R >\> om\ow mmauchpmo< m.ma 5mm mm H5 m.m m.ml o.mHI & >\> om\ow o:mxoaalz.almcopmo¢ H.@ 5m: cm mm H.@ . m.m| m.HH| ozoumo< mm9\¢m9 . m m um Ema Eon Eda pcm>aom mm- «m. mpuaa manas me am B\H B\H .oomm pm mucm>aom mSOHLm> CH mCOH +x Am opfimv onoHQEoo on» cam A< oufimv cmpm> IHom on» now mmpmm coaummemm ompm>mcmpe pew numenm HmofiEono mmIESHmmMpom .zH manna 87 +0.9 ppm. It is noticeable that the extent of variation of 6 with the solvent is about the same for K+-1806 and K+-c221 (138). The potassium cation fits nicely into the 18C6 cavity (82), so that only a small portion of its sur- face is exposed to the solvent (189). The X-ray crystallo- graphic data show that in the K+-c221 complex, K+ also lies in the cavity of the l8-membered ring (92). Therefore, K+ may also interact with the solvent, at least in one direction. Further details are given in Chapter IV. For site B as for site A, the signal is shifted in the para- magnetic direction in all solvents and at a rate of 3-5 ‘ ppm/100°C. DioXolane exhibits a behavior different from the other solvents. First, the relaxation rate of site B is large, i.e., l/T = 1000 at 25°C. Second, the chemical shift 2B of site B is the largest downfield shift of the K+cl8C6 complexes at 25°C whereas the signal of K+ solvated (or ion paired) in dioxolane is the most upfield one (Table 9). Third, between 99°C and 56°C, the chemical shift of site B drops by about 10 ppm while, in other solvents, 6B de- creases steadily with increasing temperature (Figure 9). The low solubility of the K+ol8C6 complex in dioxolane (0.035 M at 25°C) precluded a direct measurement of 5 and Avl/Z' As shown in Figure 6,we obtained the values of these parameters at 23°C (Table 19) by extrapolating to pure dioxolane the values measured in acetone-dioxolane 88 340- '-4 1 I 0.2 0.4 0.6 0.8 Volume fraction 1,3 -dioxolane Figure 6. Potassium;39 chemical shifts and line widths for the K °18C6 complex in acetone-1,3-dioxolane mixtures. - 89 Table 15. Potassium—39 NMR Chemical Shifts and Line Widths for Acetone-1,3-Dioxolane Mixturesa Containing 0.05 M KASF6 and 0.05 M 18C6 at 23 1°C. Acetone 1,3-Dioxolane 5b AV1/2 l/T2 V01 % vol % ppm Hz S-l lOO - O -3.5 199 968 80 - 20 (-9.5) 162 509 60 - 90 -l.3 202 635 “0 - 60 -0.8 232 729 20 - 80 -0.2 ' 281 883 0 - 100 +0.9c 320C 1005 aLock Acetone d6. See Table 9 footnote d. bUncorrected for diamagnetic susceptibilities. CExtrapolated. The K+-l8C6 complex is soluble less than 0.05 E in 1,3-Dioxolane. See Figure 6. 90 mixtures (Table 15). Both 6 and Avl/2 vary continuously from pure acetone to acetone-dioxolane (80-20% vol.) and there is no appreciable preferential solvation of the complex (185-187,153), as far as the concept of preferen- tial solvation mayebe applicable to complexes. The ex- trapolated values of 6 and AV1/2 in dioxolane are in ac— cordance with the values measured directly at slightly higher temperatures where the K+-18C6 complex is more soluble (see the points corresponding to l/T2 and 5 for K+-18C6 in dioxolane at the highest 103/T in Figures 3 and 9 respectivelyl The simplest way to determine if complexed ion pairs are responsible for the broad line and the large shift observed is to study the concentration dependence of 6 and l/T2 of site B. Unfortunately the present instrumen- tation does not allow us to use concentrations of less than 0.05 M_when the signal is broad. Ion pairing may, however, be studied for the pure salt at low concentration since the signal of the solvated site is narrow (Table 11a). Figure 7 shows the variation of 6 and AVl/2 of K+ as a func- tion of the KAsF6 concentration in THF and dioxolane solu- tions. As the concentration increases, the signal broadens and shifts upfield, thus indicating increasing contact ion pairing. In solutions m0.05 - 0.1 fl’in both solvents, a large fraction of the salt is associated, as shown by the leveling off of the chemical shift. The very small 91 25 0A.)... 1 Hz 15 l l. on ""—F o q) 1 l 0.05 0.10 0.115 0.20 [KAsF6] M Figure 7. Concentration dependence of the potassium-39 chemical shift and line width for KAsF in THF (circles) and 1,3-dioxolane (trianglesg. 92 difference in 6 and Au between THF and dioxolane shows 1/2 that these solvents provide similar environments for the potassium cation. Judging from the large upfield shifts in the concentra- tion range 0.007 M'- 0.15 E, it can be said that the donor ability of these solvents with respect to K+ is weak. To be more precise, the signal of K+ at the lowest concentra- tion studied (0.007 M) in THF is only 9.5 ppm downfield (-l6.6 ppm vs -2l.l ppm) from the infinite dilution chemical shift of K+ in the very weak donor nitromethane (DN = 2.7). The Gutmann donor number of THF is 20.0 (199,150). That of dioxolane has not been measured but a value of 19.7 has been calculated to Griffiths and Pugh (165) from the fre- quency shift of the 0-D vibrational band of methanol—d (188). A lower D.N. for dioxolane than for THF is con- sistent with the decreaSe of the basicity of the coordinated oxygen atom due to the presence of a second oxygen in the ring. However’the extent of the decrease seems to depend on the cation considered because steric factors are impor- tant for small cyclic ethers (99,55). Chan et a1. (99) found the ratio of the solvent- separated to contact fluorenyl lithium ion pairs to be about 50 times larger in THF than in dioxolane. The authors attributed this behavior to the lower basicity of dioxolane although they pointed out that other factors, such as the stability of the contact ion pair, steric hindrance, and 93 repulsion between the two oxygen atoms, might be important. Shatenstein E£.§l° (190) measured the concentrations of radical anions, ELEL’ lithium and sodium biphenyl and sodium naphthalene, formed in a series of ethers. They observed that in going from Li+ to Na+ the solvating power of dioxolane increases with respect to THF and 1,9-dioxane. For the sodium salt, 1,3-dioxolane and 1,3-dioxane have the same solvating power which is intermediate between that of THF and that of TRP (190). Our results indicate that the donating abilities of THF and dioxolane for the K+ ion are comparable and relatively weak. Steric hindrance may indeed explain the low sol- vating power of dioxolane for the Li+ ion since it appears that the solvating power of dioxolane increases with respect to THF as the cation size increases. The chemical shifts of K+ at infinite dilution, 611m, in THF and 1,3-dioxolane cannot be obtained from Figure 7, being given the steepness of the curves at low concentra- tion. Several potassium salts soluble up to 0.1 M minimum are needed to calculate 611m (75). If, as we expect, 611m is several ppm downfield from -16 ppm (see Figure 7 and Chapter V), the breaking of contact ion pairs should result in a large downfield shift. In this case breaking of contact ion pairs upon complexation by 1806 could explain the large downfield shift found for the K+'18C6 complex in dioxolane at room temperature (Figure 9). 99 It would also explain the apparent abnormality of the temperature dependence of 6B in this solvent (Figure 9). As the temperature increases, contact ion pairing in the complex causes a large upfield shift as is found for the uncomplexed salt. This happens only at high tempera- ture probably because the Ang anion is bulky (rAsF6' =3 9 (172)) and cannot easily come into contact with K+ complexed by 1806. 2.. Kinetics Study in Pure and Mixed Solvents Kinetics data for the exchange of potassium ions be- tween the solvated site and the K+-18C6 complex were ob- tained from the temperature dependence of the 39K transverse relaxation rates of solutions containing both species. In each solvent a solution containing KAsF6(or KI) and 18C6 with a l8C6/K+ mole ratio (MR) of 0.5 was prepared. Due to the large formation constant of the complex in all the solvents studied, the concentration of the complex was equal to that of the l8C6 added to the salt. 39 Under our experimental conditions, only one K reson- ance was observed throughout the temperature range covered. This is seen in the middle part of Figure 8 where typical spectra obtained in acetone-1,9-dioxane mixture of the solvated form, the complexed form and the mixture are shown on the left,on the right and in the middle, respec- tively. Figure 9 shows a typical semilog plot of l/T2 95 .mopsumngEop msoapm> um A.Ho> momlowv ocmxoapls.almcouoow CH Amzv mofipmn mHoE +x\oI:30LOImH msofinm> um moma new mmmfipooomop .o>L:o mB\H 2 m 50.0 m.ma mma o2 0.3H: m.m m . mm+ osoaoxoaoum.a mo.o . 0.0 em m.o- m.=- 0.: 50. on- moo: 00.0 0H2 H0 H72 m.mu m.: 5 5H: 5 >\> omlow amenocooooa 00.0 ‘ o.mH 00 m.Hu 0.0- H.s H0: 5 u 5 >\> omuom ocmxofiml:.filocouoo< 00.0 0.5 om m+2 0.5: 3.: 00- mm- osoeooa mmeAm>u<>0 mme\aom < m < H: s z 9: 0 0 B\moa w p m p ”oso>aom zoom :5 coamom mumapoEpoch on» no mappfiz on» on wcfipcoomoppoo megapmpoosoe on» as muco>aom mSOfimm> CH mGOH +x Am opfimv poonQEoo on» pew A< opfimv pmum> IHom on» non wopmm coaumxmaom ompo>mCMLB pew mamasm HooHEoco mmlszfimmwuom .ma canoe 99 data can be treated by use of relations which omit the chemical shift, which simplifies greatly the data analysis. If the product (VA — 0B) T2B were large, both the nuclear transfer and the chemical shift would affect the relations describing the transverse relaxation rate 1/T2, and make the data treatment more complex. In fact it would have been necessary to measure by a pulse sequence the longi- tudinal relaxation times which are not affected by the chemical shift. This would have required an enormous amount of time given the low concentrations of solutions used in this study. A theoretical description of the effects of nuclear transfer on apparent relaxation times has been given by Woessner (62). The actual derivation of the kinetic para- meters, which has been done by Shchori gt a1. (96) for our relatively simple case, is outlined below. In the absence of a chemical shift between the two magnetic environments the T2 relaxation curve F(t) is a sum of two exponential functions F(t) = P' exp (- —;—) + P' eXp(- —$—) (3) A T2A B T2B where P', Pé and T2A’ T2B are the apparent population fractions and transverse relaxation times, respectively. 100 The Fourier transform of F(t) is a superposition of two Lorentzian functions B (9) 1 - w T where w is the Larmor frequency of the two sites. The apparent intensities (P’, Pg) and widths (l/TéA’ l/TéB) of the two Lorentzian curves are functions of the trans- verse relaxation rates in the absence of chemical exchange (l/T2A’ l/T2B) and the mean lifetimes of the two species (TA, TB). '— - PA - l Pé (5) 1 l l l 1 1 (- - -)(T----) + - + - P, = l _ 1 TB TA 2A T2B TA' TB (6) B 2 2 2 1/2 E + ”1 2A 2B A B TATB - 2 l 5%.. 2[T___1 +____T1 +_1_+._1__[(T_l__T_l_+ 1 __1_ 2A 2A 2B TA TB 2A 2B TA TB 1 1 (7) 101 l l 1 l l l 1 l l 1 ‘T-v—=2[5—+5'—+?‘+?‘+E(T—“5—+T—-7) 2B 2A 2B A B 29 2B A B n 1/2 + . TATE] l (8) The fractional populations of each site, PA and PB, are related to T and TB by the equations A TA TB ' P = ——-———-—+ and P = —————- (9) A IA TB B TA + TB (10) For our system T2A is much larger than T2B and PA is equal to or larger than PB. It follows from Equations (5—8) that PAT2A >> PéTéB. Hence the observed signal I(w) given by Equation (9) is a single Lorentzian curve (see Figure 8) with a line width l/T2 = l/TéA. It is the ap- parent resonance of the solvated site. In our experi- ments, the free induction decay following the pulse was recorded after a time delay of 1200 to 1500 us in order -to allow for a nearly complete decay of the intensity of the broad line due to the complexed potassium. Rearrangement 102 of Equation (7) (see the derivation in Appendix 3) based on the equilibrium condition (Equation (10)) leads to the following expression for l/TA in the intermediate region TA l/T — 1/T2 2av where l/T = PA/TZA + PB/T2B . (12) 2av In the fast exchange region, l/T2 is given by (96), _ _ 2 l/T2 - PA/TzA + PB/T2B + PAPB(wA wB) T (13) where T is the mean lifetime of the interaCtion; that is T = TATE/TA + TB' In our case the difference (TA - TB) is small so that l/T2 = l/T at high temperature as 2av seen in Figure 9. The parameters necessary to compute 1/1A can all be read off plots of the type shown in Figure 9. The values of l/T2A in the intermediate region were obtained by the extrapolation of the low temperature values of l/T2 assum- ing that l/T2A and l/T2Aref have the same temperature de- pendence. A direct verification of the validity of this assumption was possible at high temperatures where the 103 values of l/T2B were obtained directly from the spectra. The values of l/T2A calculated from Equation (12) were the same as the extrapolated ones, thus validating our assumption. In general, we did not rely on any other extrapolation than that mentioned above. Unlike Shchori g2 al. (96,97) and Shporer and Luz (98), we were able to detect the signal of the bound site, due to the following reasons . (l) The signal of the K+-18C6 complex is narrower than that of the Na+ or the K+.DB18C6 complex, which might be due to the absence of benzene rings that increase the asymmetry of the electric field at the metal nucleus. (2) We accumulated a very large number of scans - up to 100000 - to obtain certain spectra for the complex. The direct measurement of l/T2B increases the overall reliability of the results since the extrapolation of the high tempera- ture values of l/T2 in order to obtain l/T in the inter- 2av mediate region is no more necessary. In fact this extrapo- lation is subject to large errors in those solvents where the low and high temperature values of l/T2 do not fall on parallel straight lines due to the curvature of the l/T2A curve (an example is shown in Figure 10). The procedure used in this kinetics study consists of four steps: - measurement of the line widths in a large tempera- ture range; 109 - calculation of 1/TA from the relaxation rates with Equation (11); — computer fitting of the Arrhenius or the Eyring plots to obtain the activation parameters; and - determination of the exchange mechanism. We will examine successively the results for the pure solvents and for the mixed solvents. Pure Solvents Four pure solvents, acetone, methanol, 1,3-dioxolane and DMF, were investigated. In DMF the ratio T2A/T2B is not large enough to allow quantitative rate data to be obtained. The values of l/T2 measured in the three other solvents are presented in Tables 9-11. Figures 9-11 il- lustrate the temperature variation of l/T2 for acetone, methanol, and 1,3-dioxolane, respectively. For each sol- vent the l/T2B and l/T2Aref curves were reproduced from Figure 3. In addition, the temperature dependence of the chemical shift for the samples with a ligand/K+ ratio of 0.0, 0.5 and 1.0 in methanol is shown in Figure 12. The pattern observed for the chemical shifts closely resembles that for the relaxation rates. In particular, at high tem- perature the signal for the solution with a mole ratio of 0.5 is half way between the two signals of the solvated and the complexed potassium. This clearly confirms that the 8—1 1/T2 105 2000 I- 10009 800- ‘ 600i- 400 - 00'- 100- 80-- 60- Figure 10. 103/ 1' °K"' Semilog plots of l/T2 XE l/T for methanol solu- tions containing KI and 18C6 at ligand/K+ mole ratio of 0 (l/TzAref)’ 0.5 (l/T2)and 1.02 (l/T2B)' The extrapolations of l/T2A and l/T 2av are described in the text. - 106 126.8 60.2 12.5 _23.2 -51.0 °c r I 1 1r l 2000+ . 1000 "' d ,7 800 q "’ 600 . .. 13‘ 400 - \ F 200% - 100 - 80 - 60 4 40 -' 1 1 l . 1 11 2.5 3.0. 3.5 4.0 4.5 103/ T °K Figure 11. Semilog plots of potassium-39 relaxation rates 1g l/T for 1,3-dioxolane solutions. —Top curve: K 1806 complex (0.05 M); -Middle full curves: (0)18C6/K+ . 0.5 (I)-18C6/K+ = 0.25; -Lower curve: KAsF (0.1 M); -dashed curves: extrapolations of 1/T2av (details in the text). 107 l I I I T +2- ‘ 0" .. s - PPM _4_ .1 -6- d -8- .. 1 1 1 1 1 3.5 4.0 4.5 5.0 5.5 103/ 1' K“ Figure 12. Temperature dependence of the 39K chemical shift in methanol solutions with 18C6/K+ ratios of 0(0), 0.5 (I) and 1.02 (0). 108 observed kinetic process involves potassium cations ex- changing between sites A and B. The temperatures corresponding to the maximum and the minimum of the l/T2 curve respectively appear in Table 16. Going from acetone or methanol to 1,3-dioxolane the active temperature range is shifted to higher temperatures by about 60°C. With Ang as the anion, the exchange of K+ between sites A and B is observable at room temperature in 1,3-dioxolane. This result allows us to propose an explanation for the anion effect reported by Lin and Popov (59). As mentioned in the historical part, it is observed that in 1,3-dioxolane solutions of NaBPhu at 25°C, the exchange of Na+ ions between sites A and B is slow (i.e., the 23Na NMR is not affected by the exchange). It is fast however with perchlorate or the iodide as the anion (59, 198). The formation constant of the K+ol8C6 complex is between one and two orders of magnitude larger than that of the Na+-18C6 complex in H 0 and MeOH (10) and most likely 2 in other solvents too. Bearing in mind that the formation rates of crown ether complexes do not vary to a large extent in going from Na+ to K+ (90) we would expect to find the decomplexation rates to be smaller for potassium than for sodium complexes. This is indeed observed with NaClOu and NaI, but the reverse is found with NaBPhu. The above behavior is probably due to the difference 109 in the types of ion pairs formed by sodium salts. We have shown previously that, in 1,3-dioxolane and in THF, KAsF6 forms a tight ion pair. The same observation probably holds for NaClOu (77) and NaI salts in these solvents. In these cases the anions compete with the ligand 1806 for the cation, and therefore the complexes are destabilized. This destabilization apparently manifests itself by a marked increase in the decomplexation rates. Conversely, in THF solutions, NaBPhu is believed to form solvent-I separated ion pairs (76,193) which would result in a more stable Na+°l8C6 complex and in smaller decomplexation rates. The increase of 60°C in the coalescence temperature, Tc’ when C10; or I- is replaced by BPh; (198) is consistent with this model. The fact that Tc increases by the same amount in THF and in 1,3-dioxolane probably indicates that in both solvents the solvation state of the ion pair for each salt is the same. However it should be noted that differences in specific solvation between Na+ and K+ ions - may account for some variation in the exchange rates. The THF binds more strongly to Na+ than to K+ (55,199). It is noticeable that in 1,3-dioxolane the 23Na NMR signal of the solvated Na+ ion is several ppm downfield from that of the complex (whatever the salt used), whereas the correspond- ing 39K NMR peaks are in the reverse order (Table 19). The change in the chemical shift of the complex rather than of the solvated species is responsible for this. The 110 smaller extent of overlap between the outer orbitals of the Na+ ion and those of the oxygen atoms of the crown probably accounts for the large upfield shift of the 23Na resonance in the Na+°l8C6 complex (See Chapter V). Due to the poor solubility of the K+'18C6 complex in 1,3-dioxolane the values of l/T2B were measured in solutions which were 0.05 or 0.09 M whereas the other parameters were measured with solutions containing 0.1 M KAsF6 and various amounts of 1806. Hence the l/T2B curve, calculated from Equation (12), lies slightly above the experimental points (Figure 11). The values of l/TA for the three solvents are collected in Tables 17-19. Semilog plots of l/TA XE l/T, shown in Figure 13, give straight lines. Activation energies, Ea’ are obtained from the slope of these plots (Table 20, first column). The data of Liesegang pg al. (92) for the same system in water are inserted in Table 20 for comparison. The important result is the very strong solvent de- pendence of the activation energy Ea which varies between 9.2 kcal mol-1 and 16.8 kcal mol-l. The lower value found in acetone and methanol solutions slightly exceeds that reported (38) for the CS+-l8C6 complex in pyridine (Table 5). It also compares well with the activation energies (Table 5) found for the release of Cs+ ion from D018C6 in PC (38) and for the release of Na+ ion from DCl8C6 in methanol (97). The energy barrier to decomplexation is 111 20.9 - 2.9 -23.2 -40.6 - 55.8 °C I I I 1000 am) 600 400 5-1 Ilt, 100 so 60 4o- ] l l l l 3.4 3.7 4.0 4.3 . 4.6 1 03/ T K“ Figure 13. Semilog plots of l/TA vs l/T in various solvents. 1,3-dioxolane (sample with 1806/K+ 0.25); 1, 3- dioxolane (sample with l8C6/K+ 0.50); acetone-1,9-dioxane (80- 20% vol. ); acetone; methanol. Ibooo 112 Table 17. Values of 1/1A and l/TAX[K+'18C6] in Acetone in the Intermediate Temperature Region.a 103/T l/T2Bb l/TZAb 1/T2 i/tA° l/TAX[K+°1806] Srd K-l S-l S-1 S-l S-1 M'-l S-l T 3.997 850 100 915 1192 22890 13 9.163 990 110 393 538 10760 7 9.333 1150 122 298 222 9990 5 9.529 1385 i 190 292 112 2290 9 9.730 1660 169 198 35 700 22 a[K+'18C6] = 0.05 M. bThese values were interpolated or extrapolated from the values measured at other temperatures (see Table 9 and Figure 9). CCalculated with Equation (11). dCalculated relative standard deviation on l/TA or l/TAX [K+°l8C6] (see Appendix 9). Table 18. Values of l/TA and l/TAXEK+°18C6] in Methanol in the Intermediate Temperature Region. 113 103/T l/T2B l/T2A 1/T2 1/tAC 1/1Ax[K+-18c6] Sr0 -1 s-1 S—1 S-l S-1 -1 S-1 % 9.301 890 150 918 739 7390 15 9.396 925 165 912 976 9760 10 9.999 1010 189 908 357 3570 7 9.608 1125 209 377 217 2170 7 9.709 1230 238 352 131 1310 8 9.829 1380 280 336 59 590 15 a[K+-18C6] = 0.1 M. These values were interpolated or extrapolated from data given in Table 10. CSee Table 17 for details. See extrapolations in Figure 10. Table 19. Values of 1/1A and l/TAX[K+°18C6] in 1,3-Di- 119 UOUULUUOUOUOLA) oxolane in the Intermediate Temperature Region.a 18C6/[K+] = 0.5 3 , _ +. c 10 /T l/T2B _ 1/T2A l/T2 l/TA 1/TAXEK 1806] Sr K-l s—l s-l S-l s—l S-l % 3.367 1190 82 993 799 19980 12 3.970 1250 93 320 300 66000 10 3.591 1335 103 251 171 3920 13 3.593 1910 110 219 119 2280 18 3.660 1515 122 179 60 1200 29 1806/[K+] = 0.25 .299 1070 69 269 782 31280 .358 1130 68 229 397 13880 .928 1200 79 219 299 9760 .980 1260 79 192 166 6690 .599 1350 87 160 89 3560 .602 1920 93 199 69 2560 .672 1530 103 138 38 1516 a [KAsF6] = 0.1 M. b Table 11. cSee Table 17 for details. Values interpolated or extrapolated from data given in See extrapolations in Figure 11. .m: monopomom an ooDMEapmo coapwa>op pnmocmpm . 0 aces: . HI. .Azm monopomom oomv _meop pampempm 029w .eHmZHx .HH pew H Ho coauoapomop Lon axon comm .mwcmp 115 m cadumpmoEop o>apom map :a oonzmwoe monam> 0:» Bone popmaoomhuxo Lo nonmaSOHmo mapoopap pozpao who: mozam> one .mmoma.+xux<5\a xv mapooamop .mqoppcm cam moamnuco coapw>auom one mam m< cam .mwpoco coaum>apom msacocnp ca moxoaoEoo mowa +M you mopmm owcmnoxm .mpco>aom pew mpouosmamm soapm>apo< .om canoe 116 somewhat larger in water (10.8 kcal mol’l) although this solvent exhibits a stronger donor character (195). The result obtained in 1,3-dioxolane is the most striking. In this solvent the activation energy is almost twice that found in acetone and methanol solutions. Two different samples (MR = 0.25 and MR = 0.50) gave the two values, 15.5 and 16.8 kcal mol-l, which are in fair agreement consider- ing the narrow temperature range accessible to study in this solvent. The activation energy in 1,3-dioxolane is comparable to the Ea values reported for some very stable cryptates such as Li+-C2ll (35) and Na+-C222 (39) in various solvents (Table 9). It also approaches the high Ea value of 18 i l kcal molfll found for the decomplexation of + 3 The exchange of K+ ion between sites A and B may proceed tBuNH from 1806 in 00013 (51). via two mechanisms: the bimolecular exchange process (I) and the dissociative mechanism (11). k 1 *K+ + K+ol8C6 : *K+-l8C6 + K+ - I k-l k + i3 + K + 18C6 + K '18C6 II k —2 Since these two mechanisms may contribute to the over- £111 potassium ion exchange, the general expression for 117 the reciprocal mean lifetime of the solvated species (96) is: l/TA = kl[K+-l8C6] + k_2[K+fl8C6]/[K+] (19) The contributions of mechanisms I and II may be determined by plotting l/TA[K+-l8C6] XE l/[K+]. This was done for 1,3-dioxolane. The values of l/TA were measured for two samples with 1806/K+ = 0.25 and 0.5 respectively (see Figure 11 and Table 11). The ionic strength was kept constant (0.1 M), since rates of reactions between ionic species may be strongly affected by the ionic strength of the solution; there is also a small ionic strength effect on the rates of reactions between ions and neutral mole- cules (196a). The values of 1/rA were not obtained exactly at the same temperatures in the two samples; therefore they were read from the computer fitted Arrhenius plots shown in Figure 13, at several temperatures. The plot of l/TA[K+'18C6] 1s 1/[K+] is shown in Figure 19. The values of l/TA[K+-18C6] are constant within experimental error. This indicates that the contribution of mechanism II to the exchange is negligible (any contribution of mechanism II would show as a positive slope in the plot). Hence the .Arrhenius plots of Figure 13 yield the activation energy :For'the exchange and not for the release of K+ ion from 18C6. 118 30 103/1' ‘ 1 '1“, 25+ ——t 4— 33 _ '7 2 75 20- ' 0 2?. ‘3: 15+ ' p< \\ > 10 3.4 . 5,- . ‘0‘ 3.5 "' —H 4— 3.6 1 1 1 1 1 5 1O 15 20 25 1/[K‘1 M" Figure 19. Plot of 1/rA x [1051806] 1; l/[K+] in 1,3- dioxolane at various temperatures. 119 This result was unexpected because for all alkali crown ether complexes for which it has been tested the cation exchange proceeds via the dissociative mechanism II (89). For example Liesegang 2: il- (92) reported that for the K+-18C6 complex in water, where the decomplexation rates are much faster than in 1,3-dioxolane (Table 20), mechanism II - coupled with a conformational rearrangement of the crown - best fits their ultrasonic relaxation data. Thus the solvent effect is two fold. In comparison with water, 1,3-dioxolane considerably reduces the dissociation rates and changes the mechanism. In this respect it is interesting to note that for - + the system tBuNH3-18C6/CDC13 already mentioned (51) mechan- iSms I and II both contribute to the exchange but at 20°C 1 the decomplexation rate, k_2, is very small (60:10 5. ) while the exchange rate k1 is large (1.5 x106 M.1 s'l). Also, in the case of the system Fl-Na+-dimethyl-DB18C6/ THF-d8, Wong _3 _l. (55) postulated that the mechanism I is operative and calculated the exchange rate to be 3200 M71 3'1 at 2°C. Other workers (95) observed that the ' exchange rates are small in ethereal solvents and in CDC13. The two aspects of the solvent effect on the complexa- tion kinetics are probably related in the following way; ‘when the energy barrier to decomplexation becomes very large it may eventually reach a point where lower energy jpathways become available. In our case the lower energy 120 route is the bimolecular exchange with a symmetrical energy profile. This last mechanism was not encountered for cryptates even when the activation energy for the release of the metal ion from the ligand molecule is very large. However, the mechanism does not seem to have been tested in THF and pyridine in which solvents the decomplexation rates are small (39) and mechanism I is the most likely to par- ticipate to the cation exchange. It could be argued that i even in these solvents the contribution of mechanism I is probably small because one cation must leave the cryptand cavity before another one can move in. The case of 18C6 is different. The molecule is sym- metrical and the probabilities of access of the cation to the 18C6 binding sites from the "top" and from the "bottom" of the molecule are identical. This might favor mechanism I. However, this mechanism is not favored on electrostatic grounds since it requires a collision between like charges. The mechanism was tested in acetone by the same method. The measurements at a mole ratio of 0.25 were not as ac— curate as those at a mole ratio (MR) of 0.50. A rough calculation for the former sample yielded an activation energy of 10.6 kcal mol-l. This value is larger by 1.9 kcal mol’1 than the value obtained with MR = 0.5. We feel that the value of E3 reported in Table 20, 9.2 kcal Inol-l, is the more reliable. As in l,3-dioxolane,our data 121 strongly favor mechanism I. The mechanism was not tested in methanol. It would be rather difficult to do so because the ratio T2A/T2B is not very large in this solvent (Figure 10 and Table 10). _For a sample at MR < 0.5 the differences of relaxation rates involved in Equation (11) would be small and therefore subject to a large relative error. Besides, the active temperature range is quite narrow. We can, however, speculate on the mechanism. Assuming that the measured activation energy, 9.2 kcal mol-l, is for the decomplexa— tion step (mechanism II), we can calculate the rate of release of K+ from 1806 at 25°C by extrapolation of the low temperature values. Using k_2 = 131 s‘1 at -60.6°c (Table 18) with Ea = 9.2 kcal mol-l, we obtain k_2 = 6.8 x 10Ll s-1 at 25°C. This value combined with Lamb's (197) 6.05 10 complexation constant Kf = 10 M."1 s.1 which is about one order of magnitude larger than yields k2 = 7.6 x 10 the theoretical value for a diffusion-controlled reaction in methanol (22,39). There are two plausible explanations. As noted by Liesegang g; ai. (39) in a similar case, the k_2 and Ea values may not lend themselves to an extended extrapolation (-61 + +25°C in our case). In other words Ea may vary with temperature. Another possibility is that mechanism I contributes to the exchange. Both factors may act in the same direction. It seems reasonable to say that in general the slower 122 the dissociation of the complex, the more likely the con- tribution of mechanism I to the cation exchange. The results of the kineticsstudy in mixed solvents will shed some light on the behavior of ethereal solvents. Mixed Solvents The spectra obtained in acetone-dioxane (80/20 v/v%) and acetone-THF (80/20 v/v %) are shown in Figures 8 and 15, respectively. For the latter mixture the signal-to- noise ratios are relatively low, since only a few minutes were available for the measurement before the complex precipitates (see Table 13, Footnote d). The relaxation rates and chemical shifts appear in Tables 12-13. Semi- log plots of l/T2 1g l/T for each mixture are shown in Figures 16 and 17 where the results in pure acetone are reproduced for comparison. In both cases the presence of ethereal solvent in the [mixture shifts the kinetic process to higher temperatures and increases the activation energy, which is related to the absolute slope of the l/T2 curve in the intermediate region. These effects are more pronounced with 1,9-dioxane than with THF. Due to the precipitation of the complex, the low tem- perature portion of the l/T2 curve for the THF containing mixture could not be obtained. As a result the l/T2A values are missing. This precludes the calculation of 123 fix 53'“ 1.334! 2545 .QJB (13 - 508 1 145.5 -2545 -36&)’ -43£> 1,;4H50 300 Hz Figure 15. Potassium-39 NMR spectra for solutions con— taining 0.10 M KAsF6 and 0.05 M 18C6 in acetone- THF (SO-20% vol.) at various temperatures. 2000 b 1000? 600 b 600 .— 8-1 400 '- 11": 200 - 129 60.2 12.5 -23.2 - -73.2 -91.4 °c Figure 16. Semilog plots of l/T2 gs l/T for (I) acetone and (O) acetone-dioxane (BO-20% Vol.) solu- tions containing 0.1 M KAsF5 (bottom curves), 0.1 M KAsF5 + 0.05 M 18C6 (middle curves), 0,1 M KASF6 + 0.1 M 18C6 (top curves). 60.2 12.5 -23.2 -51.0 -73.2 -91.4 °C r , l l I T l 1000 a w 600 . 600 ' .- 400 I a U) 13' '. \. 20°F I. ' I I a 100- . - BOP 60*- 40b 1 L 1 1 1 1 3.0 3.5 4.0 4.5 5.0 5.5 103/ T °K“ Figure 17. Semilog plots of l/T2 vs l/T for (I) acetone and (O) acetone-THF (80/20% vol.) solutions containing 0.1 M KAsF5 (bottom curves), 0.1 M + 0.05 M 1806 (middle curves), 0.1 M KAsF5 + 0.1 Mfl8C6 (top curves). 126 PPNI -8 - - .A -10 - ‘ -4:!P “ l 1 1 1 3.0 3.5 4.0 4.5 -1 103/ T K Figure 18. Temperature dependence of the 39K chemical shifts in acetone-dioxane (80/20% vol.) solutions. 0. o 18C6/KI A 18C6/K+ o 1806/K+ 1. 3 0.5; 0. 127 Table 21. Values of 1/7A and l/TAXEK+'1806] in Acetone- l,9—Dioxane (80/20% vol.) in the Intermediate Temperature Region.a 103/T l/T2Bb 1/T2Ab l/T2 l/TAC 1/TAx[K+°18C6] Sr,d s'1 s’1 3"1 s"1 5'1 M51 s-l % 3.757 890 71 930 1635 32700 23 3.830 955 75 912 888 17760 11 3.900 1005 79 391 639 12680 7 3.976 1075 85 339 388 7760 5 9.055 1150 91 276 235 9700 5 9.137 1230 98 220 139 2780 6 9.225 1330 109 166 65 1300 10 9.303' 1920 108 138 31 620 16 a[K+°l8C6] = 0.05 M. bThese values were interpolated or extrapolated from the values measured at other temperatures (see Figure 16). CCalculated with Equation (11). dCalculated relative standard deviation on l/TA or l/TAX [K+-18c6]. (See Appendix 9.) 128 the activation energy for this solvent system. The dioxane-containing mixture is more amenable to a quantitative study since the values of l/T2 could be measured down to the minimum of the l/T2 curve. It is noticeable that the two curves 1/T2 and l/T2Aref meet at this minimum. This is not found in acetone (Figure 16) which is the only solvent studied in which the addition of 0.05 M 1806 markedly increases the relaxation rates of the solvated K+ cations. The temperature dependence of the chemical shifts in the mixture is illustrated in Figure 18. The variation of 5 is steeper for site A than for site B (m6 ppm 12 m9 ppm/100°C); for each site it is steeper than the corresponding variation in methanol (Figure 12). The values of l/TA and l/TA[K+'18C6] are given in Table 21. The semilog plot of l/TA 1g 1/T, shown in Figure 13, yields an activation energy Ea of 13.8 kcal mol-l. Thus Ea in- creases by 9.6 kcal mol-1 between acetone and the acetone- dioxane mixture. This is a very large increase if we consider that the mole fraction of dioxane is only 0.18. It confirms the results obtained in dioxolane that the presence of ethereal solvents drastically increases the activation energy. The dielectric constant of the mixture is about 16 instead of 20.7 for pure acetone. Besides, 1,9-dioxane is a poor donor solvent so that we can expect a larger population of contact ion pairs (for the uncom- .plexed potassium) in the mixture than in acetone. The 129 ion pairing would tend to destabilize the complex and, as seen earlier, to increase the decomplexation rates. 0n the other hand, the lower donicity of dioxane as compared to acetone would tend to stabilize the complex and perhaps to reduce the decomplexation rates. Thus, the two factors are competing. However, what is observed by NMR is the exchange process not the decomplexation step so that we cannot use these arguments to interpret the kinetic param- eters. General Discussion The treatment of the exchange rates with the absolute rate theory of Eyring (198, 69) yields the activation parameters given in Table . The expression used for the exchange rate is k = ‘—’ e (15) where kB is the Boltzmann constant, h.is the Planck constant 7 = AH¢ - TAS7. The other symbols have their usual and AG meaning. It was assumed that mechanism I is operative in all solvents except methanol. In this solvent the activation parameters calculated assuming mechanism I or mechanism II are given in Table 20. Inspection of Table 20 reveals that the large 130 differences in activation energy of exchange between ace— tone-dioxane and 1,3-dioxolane on one hand, acetone and methanol on the other hand, are not reflected in the free activation energies which are all close to 10 kcal mol-1.. Hence the exchange rates vary only by a factor 50 at room temperature. Differences in activation entropies account for this fact. While the activation entropy is —9 cal -1 1 mol deg"1 in acetone and -3 cal mol- deg.l in methanol, it is +12 cal mol"l deg.l in acetone—dioxane and +15 cal mol-1 deg.l in 1,3-dioxolane. Thus the effect on A37 of adding 20% dioxane to acetone is almost as large as that of replacing acetone by 1,3-dioxolane. Between acetone and acetone-dioxane, the compensation of the difference in AH¢ # by the difference in AS is total. Such a compensa- tion effect is not uncommon (65). For example, it had been observed (35) in the case of the Li+dC2ll cryptates in various solvents. The activation entropy for the release of K+ ion from 1806 (mechanism II) is positive in water (3.1 cal mol"l deg-l). A similar value was found in the case of Li+ and Na+ cryptates; it probably reflects a participation of the solvent in the transition state (39,35). More interesting are the large positive activation entropies found in 1,3-dioxolane and in acetone-dioxane. For macro— cyclic complexes in nonaqueous solvents, such large values have only been encountered for the release of Cs+ from 131 C222 in DMF and C222B in PC (37). Since mechanism I is applicable to our system, the overall entropy change is 2 zero, so that we cannot relate the AS values to the cor- responding thermodynamic parameters. The activation enthalpies are a little more tractable, since they are related to the interactions between the species present. In reactions between ions, the electro- static interactions predominate. In our system, changes in the enthalpies of solvation of the reactant cations from one solvent to another should not affect the reaction rates significantly, because the solvent systems used exhibit relatively poor donor character as seen from the upfield shifts of the solvated K+ ions (Figure 9).. We have seen earlier that the complexation of K+ with 18C6 in dioxolane involves the breaking of contact ion pairs (below m90°C). The most likely transition state for mechanism I is the sym- metrical "complex" represented by K+ol8C6oK+. If the ion pairs are broken in the transition state, a large amount of Coulombic interaction has to be spent in going from the reactants to this transition state. This would account for a the large AH in 1,3-dioxolane. The same explanation probably holds for the acetone-dioxane mixture. In addition, in these two cases where the uncomplexed potassium salt exists as a contact ion pair, the solvent does not assist g the exchange. The increase in AH due to this second factor may even be predominant. The large AH7 values point to a 132 "loose" transition state, as pictured below. Since the exchange mechanism resembles the SN2 type mechanism, this "looseness" implies that bond-breaking must occur prior to bond-forming (199). K" K+ K” K+ 18C6 ' 18C6 "loose" transition state "tight" transition state In the low polarity solvents, strong Coulombic forces such as cation-cation repulsion and cation-anion attraction would tend to favor such a "loose" transition state. How- ever, the large repulsive forces between the two like charges in the transition state are likely not to change in a large extent from one solvent to another, since the local dielectric constant is expected to vary much less than the bulk dielectric constant. The large A87 value associated with a "loose" transition state might arise partly from the ligand conformational entropy. If the cations are not tightly bound to the ligand in the transi- tion state, the faster segmental motion of the ether 133 fragments as compared to that in the complex (95) should increase the conformational entrOpy. D. Conclusion The kinetic parameters for the K+°l8C6 complex are very sensitive to the solvent medium. In ethereal solvents and probably also in acetone and methanol, the cation exchange between the solvated and the complexed forms proceeds via a bimolecular process. The very large activation energies found in the former solvents are compensated by differences in activation entropies. As a result the exchange rates at 25°C do not vary in a large extent. Changes of solvent 9 exert their influence on AH in a rather complex manner although ion pairing effects seem to be predominant. The 2 variations in AS are difficult to interpret and quite a generally follow the variations in AH so as to minimize AG7. Kineticsstudies in solvent mixtures offer an interest- ing perspective for studying the various factors respon- sible for the solvent effect. CHAPTER IV MULTINUCLEAR NMR STUDY OF THE FREE CRYPTAND C221 AND OF THE C221-x+ CRYPTATE 139 A. Carbon-l3 NMR Study of the Solvation and the Conforma- tion of Cryptand 221 I. Introduction The solvation of ligands is generally believed to be much less important than ionic solvation (222). As a result, the former has received less attention than the latter. However, in the case of tetraaza ligands, Hinz and Margerum (223,229) proposed that the "macrocyclic effect" (225) is due to the weaker solvation of the macro- cyclic ligands as compared to open-chain ligands with the same donor groups. In protic solvents, hydrogen bonding is likely to play an important role in the solvation of basic ligands such as tetraaza ligands and cryptands (106,229). The situation is far from clear in the di- polar aprotic solvents. Conformational entropy as well as solvent ordering differ from solvent to solvent and can alter the "enthalpic" selectivity. The negative entropy associated with the formation of most cryptates in nonaqueous solvents is composed of several terms, such as the release of solvent mole- cules from the cation and the ligand solvation shells (entropy increase), changes in the ligand conformational 135 136 entropy (entropy decrease). Therefore it was of in- terest to us to study these two somewhat related factors. Several techniques must be used to collect the confor- mational information and the thermodynamic parameters. The purpose of this study was to investigate by 13C NMR the interaction of the free and the complexed cryptand 221 with a large number of solvents. 2. Results and Discussion The cryptand C221 has three sets of'OgH2 carbons in the 38-92 ppm region and two sets of NQH2 carbons in the 29—29 ppm region (with respect to the methyl carbons of acetone-d6). In this discussion we will be mostly interested in the NQH2 carbons (C(9) and C(5)) since the assignment of the 09H2 peaks is not definitive yet. A The spectrum of the C221°K+ cryptate (Figure 19), which is virtually identical in all solvents studied, shows the NQH2 carbons of the short bridge, C(5), about 9.7 ppm upfield Of the NQH2 carbons of the long bridges, C(9). In contrast, in the C221-Na+ cryptate, the C(5) resonance is d0wnfield of the C(9) resonance by about 0.5 ppm (Figure 19,20). When it is complexed, the ligand is assumed to be in the in-in form (Chapter I, Part C) to allow the two nitrogen atoms to contribute to the complex stability. Since the in—in form should dominate in both cryptates, 137 Figure 19. Carbon-13 Spectra of (a) the C221°K+ cryptate in methanol (S = solvent peak) and (b) the C221 Na+ cryptate in pyridine (small peaks). The large peaks in spectrum b are those of the uncomplexed cryptand. 139 4 II II 45 40 35 30 25 20 S; pp"! Figure 20. A comparison of the patterns observed in the Carbon-l3 spectra of 0221 and of its cryptates with the Na and the K+ ion. 1 C221 in MeOH. 2 C221 Na+ in pyridine. 3 0221 K+ in methanol. 9 C221 in nitromethane (+25°C). 5 0221 in nitromethane (-20°C). 190 the patterns observed in the 13C spectra are not simply related to the in-in form of the ligand. The differences in the patterns reflect changes in the conformation of the bridges of the macrobicyclic ligand. The C221-Cs+ cryptate gives the same pattern as the C221-K+ cryptate (97). The change in pattern observed when Na+ ions are re- placed by K+ or Cs+ ions may be attributed to two factors: First, the larger sizes of the latter ions might force the cryptand to increase its cavity size by taking a different conformation than the one it has in the C221-Na+ cryptate; second, the electric field around the Na+ ion is larger than that around the larger K+ and Cs+ ions. The first factor should affect each resonance in a dif- ferent extent,while the second one should shift all reson- ances by approximately the same amount. Crystallographic data (92) indicate that several torsional angles about the N-C segments and the corresponding conformations (anti or gauche) are changed when the Na+ ion is replaced by the K+ ion in the C221 cavity. In solution, what is observed by NMR is an average of various cOnformations and torsional angles. Although the effect of the two factors mentioned above cannot be isolated at the present time, it can be said that the 130 spectra indicate large changes in tor- sional angles. In going from C221-K+ to C221-Na+, the chemical shift of the C(5) carbons is virtually unchanged while the C(9) 191 resonance shifts upfield by about 5 ppm (Table 22). If the y effect (226) is operative here, the shielding of C(9) would indicate that the C(9)-N bond changes from the anti to the gauche conformation with respect to the C(5)- C(3) bond. Since the y effect is a mutual effect, the C(3) carbons should also be shielded. Indeed the C(3) resonance shifts upfield by 2.7 ppm, which is more than the upfield shift observed for the corresponding OCH2 carbons of the long bridges. An even larger upfield shift could be expected for C(3) since there are one short bridge and two long ones and, on the average, the C(5)-C(3) bond should be in the "gauche" conformation longer than each of the two C(9)—N bonds. The above discussion may not be carried very far until the field effect is precisely evaluated. This effect, which consists of shifting upfield the 13C resonances of the ligand molecule when a cation is introduced in the cavity, is quite small when the K+ (or Cs+) ion is involved. In this case, the conformational effects dominate the changes in chemical shifts between the free and the complexed ligand. For example, in methanol, the C(5) and the C(2) resonances shift upfield by 3.01 and 2.22 ppm, respectively, (Tables 22 and 23, Figures 19 and 20) upon formation of the complex C221-K+. This strongly indicates that the C(5)-N and the C(2)-C(9) segments switch from the anti to the gauche conformation with respect to each other. The 192 Table 22. Carbon-l3 Chemical Shifts of the Sodium and Potassium Cryptates. Ga (ppm) c221~Na+ C221-K+ (pyridine)b ~(methanol)b Diff. (Na+-K+) 1C 39.13 39.95 —0.82 2C 37.07 38.51 -1.99 3 37.78 90.95 -2.67 9 23.58 28.59 -9.96 5 29.02 23.86 +0.16 aReference: Methyl carbons of acetone-d6. bThe chemical shifts are virtually solvent-independent. CThe assignment of these peaks is tentative and the respec— tive position of peaks 1 and 2 is assumed to be identical in the two cryptates. 4 .‘nm 193 5m.O 50.0w 5H.5m m5.0s 50.0= 00.0: 00 0m.O 0H.5m H0.5m 00.0: 00.00 0:.H0 00 A0510 mm.O 00.50 50.50 0O.H: NO.H0 m0.H: :0 mm Hososooz :0.a m5.:m 5m.0m za.00 ma.00 05.00 0:1 50.5 05.00 00.00 00.50 00.50 00.50 00 100-0 0m.O 00.00 50.00 50.50. 50.50 0H.O5 00 0.50 mouoaasoasooooa 50.0 50.00 00.50 05.0e 00.00 00.00 00 5.50 Hoosao osoflmmmm 05.0 O0.0m 00.00 O0.0z 00.50 O0.O0 O01 Om.O. 00.50 00.50 H5.0: m0.0: 00.0: ON: HsOHIO O O0.5m O0.5m NO.H: NO.H: 50.H: mm mm osmoosaopoazua 50.H 50.00 00.50 O5- 00.0 O0.0m 00.50 501 H5010 50.0 05.50 00.50 0m 0m oscsooosoaz 05.0 00.00 05.50 05.00 00.50 00.0: 00. 550-0 00.0 00.00 O0.5m 00.O: 0=.O0 50.0: 0m 5.0m ososoosospaz 0a.m m0.zm m0.5m 00.05 00.00 00.05 0m 05 A00 topaz 00.0 H0.0m ma.0m H5WO: 50.50 H5.O0 am 5OH 500 coasospom Hsoov 0 a m m H 50.00 0 506.6.50 00120 58000 0 pco>aom .mwpzpmpoasoe Hatm>om 00 0:0 mpcm>aom mSOHLm> :H unauazpo ammo 0050 on» no mpmanm awanono malcoopwo .mm mant luu 05- 0 00.05 00.05 00.05 50.50 55.05 0 05.05 05.05 05.05 00.05 55.05 00: 50.0 50.55 00.55 05.05 55.05 05.55 55 505-0 0e 55.55 55.55 50.05 50.55 50.55 50 5.05 00-0000000 0e 05.55 05.55 55.05 55.05 05.55 05: 500uev 0:505 00.0 55.55 55.55 05.05 50.55 00.55 55 0.55 -050055000505055 0e 00.05 00.05 00.05 00.05 50.05 50: 555-0 50.0 50.55 50.55 00.05 00.05 00.55 55 5.55 00505555 5.5a 0.05 5.05 55.05 55.50 00.05 05w 00.5 55.05 55.55 50.05 00.05 50.05 55- 555-0 0e 00.55 00.55 00.55 00.55 00.55 05 0.0 005000505050050 05.5 50.05 55.05 05.05 50.00 50.05 50: 55.0 50.05 05.05 55.05 05.05 50.55 00- 5555-0 0e 05.55 05.55 00.05 00.05 55.55 05 0.55 5000000 05.5 00.55 00.05 05.05 05.50 .x05.05 00- 505-0 50.0 55.05 00.05 05.05 05.00 05.05 05: 50:52002 50000 0 5 0 5 5 50000 5 50° 000 00:00 ucm>50m AEQQV 0 .000050000 .05 05005 lUS 500:0 0 50.05 50.05 05.05 05.55 50.55 50 5.0 0000000 55000: 0 00.55 00.55 05.05 05.05 55.05 50: 50.50n0 0 50.55 50.55 55.05 55.05 05.55 00 0.5 0500500 0 55.05 55.05 05.50 05.50 00.05 00: 0 00.55 00.55 50.05 50.05 05.05 00: 505.0 0 05.55 05.55 50.05 55.05 55.55 05 5.5 0000500 0 55.05 55.05 05.05 50.50 05.05 05: 505-0 00.0 00.55 55.55 00.05 00.55 00.55 55 00050505010.5 0 50.55 50.55 05.05 05.05 00.05 00: 0 55.55 55.55 00.05 00.05 00.55 00- 5505-0 50.0 05.55 55.55 50.55 50.55 00.55 55 0.5 005 0 05.55- 05.55: 55.5- 55.0- 00.0- 00- 0 55.55- 55.55: 05.5: 05.5: 00.0- 05- 550.0 0 50.05- 50.05- 50.5: 05.5- 00.0- 55 5.5 0-0000000500 0 50.55 50.55 50.05 55.05 05.05 00- . 00.0 55.55 50.55 50.05 50.55 00.55 55 550-0 0 50.05 50.05 00.05 05.55 00.55 50 5.00 020 50000 0 5 0 5 5 50000 0 50° 000 00:50 pcm>50m AEQQV 0 .000055000 .05 05005 1&6 same behavior is observed in other solvents (see below). In the case of sodium, the field effect is much larger, and the whole spectrum of C221°Na+ complex is shifted up- field with respect to that or C221-K+ (Figure 20); the lchemical shifts are difficult to rationalize without an estimation of the field effect. The solvent dependence and, whenever possible, the 13 temperature dependence of the C spectrum of the free cryptand 221 were studied in 20 solvents. The chemical shifts are listed in Table 23. No change in chemical shift was observed when the concentration was varied between 0.05 and 0.25 M, indicating that, in this concentration range, intermolecular interactions are negligible. The 130 spectra of the free C221 in various solvents and of the C221-K+ cryptate are shown in Figure 21. Table 23 gives the difference in chemical shift between the C(h) and the C(5) resonances in all solvents studied. At room temperature, this difference varies between -O.l (CDCl3) and 2.3 ppm (Formamide). When the difference (an-55) is large, the spectrum shows a pattern similar to that ob- served for the C221-K+ cryptate (Figures 20,21); we will say that the spectrum is of the "K+ltype", On lowering the temperature, the difference (5M-55) in- creases in solvents such as nitromethane and acetonitrile (Table 22 and Figure 22) whereas in other solvents peaks H and 5 remain superimposed down to the melting point. The two 1147‘ M J -51.... FORMAMIDE MOCN 0.00, bL WUWL H20 COMPLEX c221 - K" _ "2° Figure 21. Carbon-l3 spectra of the free C221 cryptand in various solvents and of the C221-K+ cryptate in water. 1&8 . 55 M n 24 -42 MeCN -53 Figure 22. Carbon-13 spectra of the free cryptand 221 in acetonitrile and DMF at various tempera- tures. 1&9 types of behavior are illustrated in Figure 22 for aceto- nitrile and DMF, respectively. It is clear that certain solvents interact with the C221 molecule while others do not. This interaction may be simply represented by C221 + n8 2 C221-Sn AS° < 0 where S stands for a solvent molecule. Due to the negative entropy associated with this "complexation" reaction, the equilibrium is shifted to the right at low temperature (AH° is independent of temperature), which explains the evolution of the spectra seen for example in acetonitrile (Figure 22). In this solvent and in all the solvents which give a separation of the peaks H and 5, which we will now call A-type solvents, the low temperature spectra are of the K+-type. A similarity in pattern does not necessarily imply a similarity in conformation. However, in the case of potas- sium, the field effect is small. Also the data given in Table 23 indicate that in A-type solvents and with decreas-' ing temperature the C(H) and C(5) resonance signals tend toward the corresponding signals in the 022l-K+ cryptate (Figure 20). This behavior suggests that indeed A-type solvents fknmma the C221 cryptand to take a conformation similar to that observed in the C221-K+ cryptate. 150 It is not surprising that water, methanol and formamide interact with C221 since these solvents can form hydrogen bonds with the bridgehead nitrogen atoms. The oxygen atoms may also participate in H-bonding with the solvent molecules; however, according to Lord and Siamwiza (227), they probably interact, at best, to a very limited extent. These authors studied by IR the interaction between C222 and CDCl3 and concluded that the six oxygen atoms appear to be inaccessible to the chloroform-d. Their conclusions should be valid for the case of C221 in various solvents. An exception could be water. Water molecules are only slightly larger than K+ ions (2.9 vs. 2.66 fi) so that they can probably penetrate into the C221 cavity as do K+ ions. In this case they may have access to the oxygen atoms. It is noticeable that A-type solvents are all acidic even if some of them, e.g., CH2012, are very weakly acidic. Methanol probably interacts via the O-H group and formamide via the NH2 group. Both acetonitrile and nitromethane have an electron withdrawing group and most likely interact via the positively charged methyl group. In order to verify I this hypothesis, we studied the spectral changes along the series nitromethane, nitroethane and l-nitropropane, in a large temperature range (Figure 23). The differences (GU-65) are plotted as a function of the temperature in Figure 2H. It is seen that, at a given temperature, the quantity (Sn-65) decreases as the methyl group is removed further from the 151 . mma cm> 00 m 05 3009 ©0059 05w Euppomam 3000 :H .mmpsumpmasmp 030550> 5M mCQWOLQWLuficmw 000 0:05500555: 02055050505: :5 555 acupquo moan 055 no 0550000 mHIconpmo .mm mpsw5m N N N n no N On 0 Z :0 :0 IO 2 1 I N n O 02 IO 5< j 15 a5 152 l 1 1 l L 1 1 -80 -60 -40 -20 0 26? 4O 6O Tennperature °C Figure 2“. A plot of the difference in chemical shift between the two NgHg resonances of the free cryptand 221 as a function of temperature in several solvents. (O) nitromethane; (0) nitro- ethane; (A) l—nitropropane; (O) acetonitrile; (I) methanol; (V) ethanol. 153 electron withdrawing group; the largest decrease occurs, as expected, between nitromethane and nitroethane. Consequently the methyl group of these solvents ap- pears to be responsible for the interaction with the C221 ligand. A decrease Of (GM-55) was also observed in going from methanol to ethanol (Figure 23). This result does not indicate that the methyl group is responsible for the interaction; it simply shows that the interaction is weaker in the case of ethanol. It should be noted that, within a series of solvents, the differences (GM-55) are about the same when calculated at the same T/Tm.p. ratio. Ethylene glycol, which is an H-bonding solvent, gives a difference (5u-55) of 0.37 ppm at 24°C. The high melt- ing point of this solvent precludes a low temperature study. Toluene and anisole gave about the same chemical shifts, indicating that the methyl group of anisole is not acidic enough to interact noticeably with C221 even at low tempera— ture. Perhaps the best evidence that the solvent acidity (defined in terms of acceptor ability) is responsible for the solvent-C22l interaction is provided by the drastic spectral change observed between C221 in DMF and C221 in formamide (Figure 21). At 3M°C, peaks A and 5 are superimposed in DMF whereas they are separated by 2.32 ppm in formamide. The substitution of two methyl groups by two hydrogen atoms increases the acidity. Formamide is 15H capable of H-bonding and its acidity is comparable to that of the lower aliphatic alcohols (150). The spectrum of C221 in formamide is about the same as that observed in water, which indicates a relatively strong C221-formamide interaction. The only acidic solvent, or acceptor solvent in the Gutmann sense (150), which does not seem to interact with C221 is chloroform-d. Dichloromethane, which is much less acidic than chloroform, gives a 4-5 Split of about 1 ppm at ~M2°C while chloroform-d does not split the two peaks at all, even at -60°C. The latter solvent has been shown by IR (227) to form hydrogen bonds with the nitrogen atoms of C222, the ligand being presumably in the in-out or the out-out form. Chloroform—d probably behaves in the same way with C221. However, the interaction is not detected by 13C NMR, which indicates that either the interaction is too weak to affect the spectrum or the conformation of C221 hydrogen-bonded to chloroform-d is the same as that of C221 in non acidic solvents such as ethers. The C221-di- chloromethane interaction might be favored by the presence of two acidic hydrogen atoms and by the larger dipole moment of dichloromethane as compared to chloroform-d (1.55 KE- 1.o Debye (165)). Ethers such as THF and 1,3-dioxolane, which are poor acceptor solvents (150), appear not to interact with C221, even at -90°C (Table 22). Tetramethylguanidine is capable 155 of hydrogen-bonding (228); however no interaction can be detected. Steric hindrance due to the four methyl groups of the TMG molecule may prevent the acidic hydrogen from gaining access to the nitrogen atoms of C221. It is noticeable that solvents possessing two or three acidic hydrogen atoms appear to interact with C221 stronger than solvents having only one acidic hydrogen atom. For example, formamide gives a much larger 4-5 Split than meth- anol and ethanol although, as previously mentioned, these solvents have similar acidities. The same holds for di- chloromethane and chloroform-d. A solvent molecule with two hydrogen atoms may be able to bind the two nitrogen ~ atoms of C221 at the same time if the cryptand is in the in-in conformation and if the solvent molecule is small enough to penetrate into the ligand cavity. Water is probably the only solvent which meets the last condition. It is unlikely that other solvents form 1:1 "complexes" with C221 because the entropy associated with the complexa- tion reaction would be very small and spectral changes with decreasing temperature would also be very small. 3. Conclusion In solvents which exhibit well developed acceptor properties, the free cryptand 221 appears to have the same - conformation as the one it has in the C221°K+ and C221-Cs+ 156 cryptates. This conformation is favored at low tempera- ture. Hydrogen-bonding between the acidic hydrogen atoms of the solvent molecules and the nitrogen atoms of the cryptand is probably responsible for the solvent—C221 interaction which forces the cryptand to take the con- formation of the C221-K+ cryptate. In a series of solvents with the same functional group, the strength of the C221-solvent interaction follows the order of the acidities. However if solvents with different functional groups are considered, not only the acidity but also the number of acidic hydrogen atoms, the size of the 'solvent molecule and the steric hindrance around the acidic group seem to play a determining role in the C221-solvent interaction. Dr. J. P. Kintzinger (Université’Louis Pasteur, Strasbourg, France) is now completing a study of the same 13 system by using 1H NMR, C relaxation measurements and calorimetric techniques. B. Potassium-39 NMR Study of the C221-K+ Cryptate 1. Introduction The size of the K+ ion is a little smaller than the cavity size of 0221, 2.2 K (229) Kg. 2.66 X (202). In the similar case of Cs+ ion and C222, Kauffmann 32 a1. (97) found that the Cs+ ion can form two kinds of complexes with 157 C222, an inclusive and an exclusive complex. Therefore it was of interest to continue the studies of Shih and Popov (138) in order to determine if one or two modes of complexa- tion exist for C221-K+. 2. Results and Discussion The 39K chemical shifts and line widths of the C221-K+ cryptate in various solvents are given in Table 23. Cor- rections due to the use of different lock solvents were not applied in this table. The corrected values are given in the footnotes. The chemical shifts of c22l.K+ range from 10.2 to 15.2 ppm. The variation of 5 ppm, which roughly corresponds to that found in the case of 18C6'K+ (Chapter III), shows that the K+ ion is not completely isolated from the surround- ing solvent molecules. A chemical shift range of about 50 ppm was reported by Gudlin and Schneider (230) for the 205T1 chemical shift of the C22loTl+ cryptate. Consider- ing that the chemica1_shift ranges for the free ion via common solvents are about 40 and 500 ppm for the K+ and Tl+ ion, respectively, the variations seen for the cryptates are comparable. This is expected since the K+ and the Tl+ ion have similar sizes and have the same area of their surface exposed to the solvent molecules. Although the chemical shift range is quite small, it can.be said that the C221-K+ cryptate is an exclusive complex. 158 05: 005 0.055.55 55 00 0.000.05 00 55 0.05 .05 50 00000 05- 005 55 55 00: 055 0.050.05 05: 555 0.05 .05 . 55 00 00.05 .05 2000 00000 50000002 05: 055 0.050.55 55: 005 0.050.05 55 55 50.055.05 05000 20000 0055005000000 50: 500 55 05 50: 055 0.055.05 05: 50 0.050.55 0 00 0.055.05 55 00 50.050.05 2000 00:00 55 50 0.055.05 whmM Oma msoumo< 500005 Aumv om\59< AEQQ00 . 55mm uco>5om pcm>5om 0005 .mmpsummeEoe 030550> 50 0:0 muco>5om 5mpm>mm :5 005059050 5mmo.+x no 055053 0:55 0:0 0m555£m 50058050 msz 0MIES5mmmuom .55 05209 n 159 mwl 00>? 00: 000 55 55 05- 055 55 55 05- 005 50.055.00 0 055 0.055.55 05 505 0.055.55 50 00 0.050.05 om mm m.05m.OH 00 50 0.055 05 00000 00-005 005 50 005 55 05 0000 050 00 mml mmm 0.05m.m5 55 555 0.050.55 0000 55 505 0.055.05 2000 050 00505000 55 005 0.055.05 00 005 0.050.55 0000 050 0000 Acovcp Ammv 05\5>< Aeaavm p50m uC0>5om uC0>5om 0005 .000055000 u 55 05000 160 . .050>5500Qm0n .00 m0 0:0 20000 500:0:0500< 05 0005 0:5 20:3 Ema 0.5 0:0 5500 0.5 «Egg 5.5 50005050 .0505Q0 5005E0£0 005009000 025 £50590 090 .0 050.0 u 55550.+000 OOH+U. 5\5>0 55 505-05 .00 005v 5\5>0 05 0050 .Nm oomA 2 mo 0 u m5mmo.+mmn .555 0050030 0000 0000 0:5 050 05C0>500 0005 0S5 05 0005 0Q >500 000 0500>500 500500050 25 00050550 050530 50055020 00 :005L0QEoo 00 .003Q55coo I :N 05909 161 The large paramagnetic shift (the central value is about 13 ppm) is similar to that found in the case of the DBl8C6°K+ complex (Chapter V). It indicates a strong over- lap of the orbitals of the donor atoms of the ligand with the outer p orbitals of the cation. Such an overlap would not be possible if the K+ ion were displaced from the center of the lB-membered ring of C221 towards the outside of the cavity as the crystallographic data indicate (92). Also a larger solvent dependence would be observed in this case. Therefore the K+ ion appears to lie inside the ligand cavity although it is able to come into contact with the solvent molecules. When KPF6 is replaced by KSCN, the cryptate signal shifts downfield by 1.5 ppm in acetone and 1.2 ppm in pyridine (Table 23). The direction of the shift is that found in the absence of C221 (135), which could indicate some extent of ion pairing in the cryptate. However, the magnitude of the shift is almost within experimental error and does not give a conclusive evidence for ion pairing. In methanol the results obtained for KSCN and KI are about the same. The ratio of the K+ ion size over the C221 cavity size, 2.66/2.2 = 1.21, is virtually identical to the correspond- ing ratio for the Cs+ ion and c222, 3.3u/2.8 = 1.19. Two kinds of complexes, one inclusive and one exclusive, were found in the case of C222-Cs+ (97). However, in the 162 39 case of C221-K+, the quasi invariance of the K resonance frequency in a large temperature range (Table 23) shows that there is only one mode of complexation for the K+ l3C spectrum of C221-K+ with ion. The invariance of the the solvent and with the temperature (Part A) supports the above conclusion. 39K shift with increasing tempera- The invariance of the ture is in contrast with the large upfield shift observed with 1806-K+ (Chapter III) in all solvents studied. It probably indicates a weaker K+-solvent interaction in the cryptate as compared to the crown complex. Such a weak interaction in turn suggests that the K+ ion is closer to the center of the C221 cavity than crystallographic data indicate. However, the invariance of the shift could also arise from the compensation of two effects, e.g., weaker solvation (+ upfield shift) and stronger cation-ligand overlap (+ downfield shift). At low temperature the 39K chemical shifts do not converge towards a solvent independent .Shift, indicating that the K+ ion remains exposed to the sol- vent molecules. 3. Conclusion The large paramagnetic shift of the C221.K+ cryptate shows that the K+ ion is tightly embedded in the cryptand cavity. The relatively small solvent dependence of the 163 shift and its quasi invariance with temperature clearly indicate that in solution there is only one kind of C22l°K+ cryptate in which the K+ ion is closer to the center of the C221 cavity than it is in the solid state. CHAPTER V MULTINUCLEAR NMR STUDY OF THE COMPLEXATION OF POTASSIUM IONS WITH CROWN ETHERS AND WITH "CONVENTIONAL" LIGANDS 1611 A. Potassium Cation Interaction with Crown Ethers Introduction The complexation reactions of the K+ ion with the crown ethers 120“, lSCS, BlSCS, 18C6 and DB18C6 were investigated in several rmmmqueous solvents by Shih (15) using 39K and 13C NMR. A number of formation constants were reported for the smaller crowns and the presence in solution of "sandwich" 2:1 (ligand :K+) complexes with 1505 and MBlSCS was detected. Broad signals precluded a quantitative study for 18C6 and DB18C6. In this part, the above study was extended to other ligands such as dithia-1806 and to large members of the crown family such as DB2lC7, DBZHCB and DB27C9 (Figure 1). Results and Discussion Potassium-32 NMR Studies The 39K chemical shifts were measured as a function of the ligand/K+ mole ratio. The concentration of the salt, KAsF6 or KI, was held constant at 0.05 or 0.075 M and the ligand concentration was varied. In all cases the exchange rate of the K+ ion between the solvated and the complexed site was fast on the 39K NMR time scale and only one 165 166 population-averaged line was observed. If only a 1:1 complex is present and if ion pairing is negligible, the observed chemical shift, 6 is then given by obs’ 5 (l) 5 =X 6 +X M+L M M+ M+L where 6 and 6 are the chemical shifts of the solvated M+ M+L and X are the respective and complexed cation and X M+ M+L mole fractions of the two species, respectively. The con- centration equilibrium constant for the formation of the complex is, K = _M:L_ (2) where CM+L’ CM+ and CL denote the equilibrium molar con- centrations of the complex, the cation and the ligand, respectively. By combining Equations (1) and (2) with the mass bal- ance equations it can be shown that - t t 2 t 2 t 2 t t = — - + - Sobs KfCM KfCL l) KfCL + KfCM 2KfCLCM + (2K ct + 2K Ct+l)l/2 EM:EEE + 5 (3) f L r M 2K t ML fCM In Equation (3) the total concentrations of the cation and 167 the ligand (0; and CE, respectively) are known and 5M is determined by measuring the cation chemical shift in the absence of the ligand. The two unknown quantities, Kr and 6 L’ can be evaluated by a non-linear-least-squares procedure M starting with reasonable estimates of Kf and 5ML' The pro- gram KINFIT (179) was used to perform the iterations and to obtain statistical information regarding the unknowns. De- tails about the use oftfifltsprogram can be found in the Ap- pendix l. t The systems investigated were (i) dithia-lBC6 (DTl8C6) + KASF6 in acetone, (11) DB2107 + KASF6 in acetonitrile, (iii) DBZHCB + KASF6 in nitromethane, acetonitrile, pyri- dine and methanol (KSCN was used in the last solvent), (iv) DB27C9 + KAsF6 in acetonitrile and pyridine. The chemical shifts and the line widths as a function of the ligand/K+ molar ratio are given in Table 25 and plotted in Figure 25 for the system involving DT18C6. In this case, the quite narrow signal gradually shifts in the paramagnetic direction with increasing ligand/K+ mole ratio: The chemical shift does not reach a limiting value (GML).uP to a mole ratio of 2.17, the highest one which could be obtained due to the low solubility of the ligand. This behavior is indicative of a rather weak K+oDTl8C6 interaction. A data analysis using KINFIT yielded a value of 2.8:O.8 (log Kf = o.u500.12) for the association constant Kf. It 168 P Table 25. Potassium-39 Chemical Shifts and Line Widths in Acetone Solutions Containing KAsF6a and Dithia-18C6 at Various Mole Ratios and at 25°C.b L2%i%%él 50(ppm) Avl/Zd (Hz) o -13.1 12 0 M2 -12 18 o u2e — 8 7 ul 0 65 -11 o 20 o 90 -10 2 22 1.00 —lO.2 2n 1 10 - 9.8 25 1.25 - 9.6 25 1.72 .- 8.5 27 2.17 - 7.5 33 aKAsF6 = 0.075 m. bUnless otherwise indicated. ci0.2 ppm. dil-Z Hz depending on the line width. et = -u2°c. 169 l I l l l I 0.4 0.3 1.2 1.6 2.0 2.4 DT1806/ K” Figure 25. Potassium-39 chemical shifts as a function of the DT1806/K+ ratio in acetone solutions. 170 should be noted that this value roughly corresponds to the lower limit of the range of reliable values measurable by 39K NMR with our method. Since ion pairing was not con- sidered in the calculation of 6 the actual value of Kf obs’ is probably larger than 2.8. A Kf value of 1” (log Kf = 1.15) was reported (200) for the K+oDT18C6 complex in methanol. Considering that the formation constants of M+ocrown complexes are of the same order of magnitude in methanol and in acetone (15), our result is in the ex- pected range. Reported (201) log Kf values for M+oDT1806 complexes in acetone include 0.61:0.09 (M+ = Cs+) and 2.98: 0.01 (M+ = Tl+). Although the sizes of Tl+ and K+ are comparable (l.u0 and 1.33 A, respectively (202)), the former cation gives a much more stable complex than the latter. This is expected since there must be some covalent inter- action between the "soft" acid (203) Tl+ ion and the "soft" sulfur atom. In the case of K+ and Cs+ ions, the replace- ment of two 0 atoms by two S atoms in the ligand ring de- creases the formation constant by several orders of magni- tude (data for CS+ ion in Reference 201). The limiting chemical shift of the K+oDTl8C6 complex, of 6.613.? ppm, is rather imprecise because the chemical shifts could not be measured at high mole ratios. However, it resembles the limiting shifts found for K+-DBlBC6 in various solvents (15). In this last complex, the K+ ion is "squeezed" in the ligand cavity (see below). The same 171 reason probably holds for DT18C6 which has a smaller cavity than 18C6 since S atoms are larger than 0 atoms. However this analogy must be regarded with caution because the electron donor atoms are not the same. The signals of the K+ ion complexes with the large dibenzo crowns are rather broad, i;§;, 100-220 Hz (Table 27), which increases the experimental error in the chemical shift to such an extent that in most cases a quantitative analysis of the data could not be done. Moreover, without the zero-filling technique (see Experimental part), the time required to perform a complete mole ratio study, 3:523 at least 9 spectra, becomes prohibitive when the lines are broad. For these reasons, only the data relative to DB27C9 are presented here (Table 26). As seen in Figure 26 in pyridine solutions, the experi- mental points are quite scattered. However, in acetonitrile, the plot of 6 z§_mole ratio (Figure 26) clearly shows a break at a mole ratio of about 1. This behavior indicates the formation of a rather stable (log K 3 u) K+-DB27C9 f complex. It should be emphasized that in the case of 39K NMR, the upper limit of the reliable Kf values measur- able with our method is lower than that attainable by 133Cs NMR ( 105) since the experimental error is larger and higher concentrations have to be used. We can estimate this upper limit to be between 5 x 103 and 10“. For example, a quantitative analysis of the data, shown in 172 Table 26. Potassium-39 Chemical Shifts and Line Widths for Acetonitrile and Pyridine Solutions Containing KAsF6a and DB27C9 at Various Mole Ratios and at 2H°C. Acetonitrile - _. Pyridine [Dfiilig] 87ppm) Avl/2(Hz) [D::Z:9J 5(ppm) Avl/2(Hz) 0 0 - 2.4 12:1 0 e.ut0.2 39:2 0.28 - 5.6 “4:3 0.16 - 1.5i0.3 82:3 0-“5 - 7.3 52:3 o.uu - 7.0:0.6 105:5 0.59 - 8.5 78:3 0.63 - 9.6;o.6 130010 0.71 - 9.4 95:5 0.91 -11.3:0.6 170:10 0.91 -11.7 100:5 1.00 -12.2:O.6 188310 1.00 -1l.7 110:5 1.30 -ll.0:0.6 240320 1.09 -12.6 107:5 1.86 -12.u¢0.6 220020. 1.21 -12.0 102:5 1.29 -l2.6 97:5 1.57 -12.2 107:5 2.22 -12.2 130:5 l3 a[KAsF6] = 0.075 b$0.3 ppm. l l l l l 0.5 1.0 1.5 2.0 2.5 Dance/K“ Figure 26. Potassium-39 chemical shift vs. DB27cg/K+ mole ratio in acetonitrile (O) and pyridine (V) solutions at 25°C. .Nm omHIOOH u N\H>| m.m| Hocmnpoz :.mH- mHH- Am.mflv N.NH- o.m u mcfiefimsm mp m.HHu H.0H- m.p- mcoumoa 1 :.ma. NH- ems- m.a o.MH- m.: - m.m- H.s - a.m- mmwmwwm oAmHIV m.HHI m.mal m.m| m.mat 04ml ocmnuoe . nonpfiz aoammo mozmma Noamma mama sowfima mfimomav momfi mAmomHmv momfim mfiaomflv scm>fiom E 3.3 MAEQch Acv Ao.nv Any Any ADV .oomm um cam mucm>aom Hmpm>om CH mLmSpm esopo mSOHLm> Sufi; momeQEoo COH Esammmpom mo mumacm awofismco mmIESHmmMQom .Nm mfinme 175 Figure 26, for the K+-DB27C9 complex in acetonitrile yielded a formation constant of (8.5:lu.6) x l03! Although the central value seems reasonable on the basis of that measured for the Cs+oDB27C9 complex (7.8:O.5) x 103 (205), there is ample room for various interpretations! The association constants of the K+oDB2hC8 complexes were measured by 13C NMR (see below). Unlike the formation constants, the chemical shifts of the K+olarge dibenzo-crowns complexes could be reliably evaluated. Table 27 gives the results together with the chemical shifts of various K+ ion complexes found in this work or in the literature (15). It is seen in Table 27 that the chemical shift does not vary along the series DB2lC7, DB2NC8 and DB27C9 and that within experimental error, (0.5-l ppm for broad.lines), it is independent of the solvent for the last two crowns; only one solvent was studied in the case of DB2lC7. Thus in the large crowns, the potassium cation seems to be efficiently shielded from the surrounding solvent molecules, as found for example for the K+-C222 cryptates (138). The invariance of the shift with the crown as well as its value, -l2:l ppm, call for some comment. A solvent- independent shift of similar magnitude was found by Shih (15) for the "sandwich" K+.(15C5)2 complex (Table 27). It is the largest 39K NMR diamagnetic shift (with respect to K+ ion at infinite dilution in water) observed for the 176 crown complexes (and the cryptates) studied thus far (Table 27). In order to rationalize this behavior, it was necessary to analyze the 39K NMR data in conjunction with other alkali metal NMR data for similar complexes. The~23Na, 39K, 133Cs and 205 T1 NMR chemical shifts of the crown complexes with the corresponding monovalent cat- ions were gathered from the literature (59,148,15,38,2OM- 206,168,169) and collected in Tables 27-30. Figures 27, 28 and 29 illustrate the variation in several solvents of the chemical shifts of the M+-(crown)n complexes (n = 1 or 2) with M+ = Na+, K+ and Cs+, respectively. This variation can be fully understood in terms of the repulsive overlap effect. On theoretical grounds, the extent of the para- magnetic shift is expected to depend on the short-range re- pulsive overlap of the electron donor with the cation (136). This was found to be borne out by experiment in a number of cases such as the "inclusive" Cs+-C222 complex (see Chapter I, Part C). Indeed the application of this model to the data illustrated in Figures 27-29 allows a detailed analysis of the ion size-cavity size relationship, of the conformational properties of the crown ether molecules and of the cation- solvent interactions. Due to its intermediate size, the K+ ion forms rather stable complexes with each crown in the series 120“ DB27C9 (see Table 27). However, the 39K NMR data were the most recently obtained because the measurements are the most difficult. The body of data is now large enough to 177 .mmH monopommmo .wzH mocmpmmomm m.mI 0mm H.@HI m.mI om 5H. 3.2: m.mn are m.mI m.mI m.mI omzm m.®l 0.0! .mzm H.» I . m.ml m.m I N..N.I m.ml :.ml mcfivfihzm m.m I :.oHI w.wl mcouoo< o.w I m.wI m.OHI m.mHI m.:| H.2I mafipuficouoo< m.OHI m.pl :.m I N..mHI NHmI H.HHI m.mI m.:l . mcmSHmEthHz coaoommo cwozmma csoammo «coma momfiao mAmomHv cmomfi cmomflm pccsfiom . Assays muco>aom Hmpm>om CH moonQEoo csoso.+mz msoapw> mo womanm HwOHEono mmwmwmowm .mm magma 178 .wmfi mofimfimhmmw .wm QOCmL®.H®m U ooCmCommmU .zom moCmCmmomo .me ooCoCmmomC .coCHHCmCCC ohm mmCHm> copoHomCmm wwI m.mmI mmI wmI m.::I H.0HI H.smI m.wm om m I . m I w I ww m.mm w I omzo HI m.mHI WWI w + m.m:I :.m WHI can H.5HI smI s.mmI m.mHI wI «WI ww Hosanna: :.HHI . m.mHI m.omI o.m + meI m.OH m.MHI mm cchHasa m.mHI m.oml o.mml H.@ I pal :.m I H.mml a: mCoumo< CHI w.mHI m.:HI :.m mmI mwaH wwI m.mm+ cHHCpHC IOpoo<. m.mHI m.mmI :.mmI 0.:NI WWI HWI H.mm+ ascnpce IOCsz DOHoommm cmosmmo nmozmma QNOHmma mflmomHv cmowH cmomHmm cmomHam psc>Hcm CAEQavo IHom HCCo>om CH momeCEoo Czopo.+mo mCOHCm> mo .ooom pm mpcc> mpCHsm HccHscno mmHIesHmco .mm cHnme 179 Table 30. Thallium-205 Chemical Shifts of Some T1+°Crown Complexes in Various Solvents at 30°C. (ppm)a’b Solvent DA18C6 DB21C7 DB2u08 Nitromethane -252 -308 -247 Acetonitrile - 12.” -27U -250 Acetone - 32.6 -278 -260 Methanol -209 (-215) DMF - 72.7 (-336) DMSO >275 (+150) aReference 169. bThe brackets indicate that the experimental error is large, due in general to the low stability of the complex. 180 1 l l l l I 1 I 81505 15cs uses)2 18C6 D8 D8 I DB DB 21C? 24ca 27cs 30C1O Figure 27. Sodium-23 chemical shifts of various_Na+.cpown complexes in several solvents at 30°C. ( V) nitromethane; ( O) acetonitrile; ( I) pYri- dine. Figure 28. 181 L l l l l l l l l J. 81505 1505 DB DB DB 1806 I 2107 DB 27C9 (12C42 (315(5)2 (Iscs), 1806 24C8 Potassium-39 chemical shifts of various K+.crown complexes in acetonitrile. The vertical bar for 18C6 represents the range of chemical shifts observed in various solvents (Table 1“). .182 /‘\ . ‘ \ -4OP § 1 DB 180608qu 1866 21C7 24C8 27C9 30CB10 Figure 29. Cesium-133 chemical shifts of various Cs+°crown complexes in several solvents. (O) aceto- nitrile; (I) pyridine; (A) acetone; (0) methanol; (V) nitromethane. 183 permit a general interpretation. The cavity sizes of the large crowns, DB2lC7 and above, are larger than the K+ ion size. As a result, there is not much overlap between the ether oxygens and the K+ ion; a large upfield shift results (Figure 28). The small solvent dependence of the shifts is consistent with the possibility for the large crowns to "wrap around" the cation (207). Conversely, the DB18C6 cavity size is smaller than the K+ ion, and the signal shifts in the paramagnetic direction by about 20 ppm with respect to DB2lC7. This is due to the increase in the repulsive overlap between the ether oxygens and the K+ ion. The chemical shift of K+cl8C6 occupies an intermediate position corresponding to the intermediate size of 1806 between those of 031806 and 082107. Since for the K+-1806 complex the shift is not known in acetonitrile, the range of shifts in various solvents is shown in Figure 28. As seen in Chapter III, the solvent dependence of the shift is rather limited for this complex. The large upfield shift observed in going from DB18C6 to 1505 (Figure 28) indicates that the K+ ion is displaced from the center of the cavity. If this were not so, a very large paramagnetic shift would have resulted. Thus the K+ ion lies above the plane of the 15C5 ring and forms an "exclusive" type of complex, as does Cs+ with DB18C6 (38). This is supported by the very large solvent dependence of the shift (Table 27) and by the formation of a quite stable 184 2:1 (15C5:K+) complex (log Kf : 2 in MeCN (15)). It would be interesting to determine the minimum cavity size of a crown ether which could accommodate the K+ ion in the center of its cavity. This could be done for ex— ample by adding a (third) benzene ring to DB18C6 to reduce the cavity size, and by measuring the chemical shift of the resulting K+-"tribenzo-18C6" complex. If the signal is at lower field than that of K+-D818C6, then the K+ ion should be at the center of the cavity. The opposite result would indicate that the cavity is too small to accommodate the K+ ion. The difference between the ionic radii and the COordination radii (208) of this and other ions could be estimated by this method. The 39K chemical shifts of the 2:1 (15C5:K+)comp1ex is about 8 ppm upfield from that of the 1:1 complex. In a recent review (89) Dye pointed out that the solvent-in- dependent shifts of -A8 ppm for Cs+-(18C6)2 is about that expected for an ether type solvent, indicating that the Cs+-O interactions are essentially relaxed in this complex (indeed 5 for a THF solution of Cs+ octanoate 0.01 g is -Mh ppm (209)). The same explanation probably holds for K+°(1505)2. This complex gives a shift of -12 ppm which, as previously mentioned, closely resembles the shifts seen for the large crowns. In both kinds of complexes, the binding sites of the ligand possess a large motional free- dom due to the looseness of the fit and behave as 185 individual ether-type solvent molecules with respect to the cation. It was assumed in Chapter III that the chemical shift of the potassium ion solvated by THF or 1,3-dioxolane is several ppm downfield from -16 ppm. From the results obtained With K+°(15C5)2, we anticipate a value close to -12 ppm, which would validate our assumption. The number of binding sites does not seem to be im- portant. In the K+ ion complexes with DB2lC7, 032008, DB27C9 and 2 x 15C5, the shift is Constant while the number of ether oxygens varies from 7 to 10. This is because the K+-O interactions are relaxed in the A complexes, and the overlap is minimum. The K+°BlSC5 and K+-(l2CA)2 complexes give similar shifts in acetonitrile (Figure 28). What was originally considered (15) as a 1:1 must actually be a 2:1 (12CM:K+) complex since the shift is virtually solvent independent (Table 27). Both the 1:1 and 2:1 complexes are detectable in the case of B15C5. As found for 15C5, the 2:1 complex gives a signal at higher field than the 1:1 complex in acetonitrile and at lower field in nitromethane. It is interesting to note that the signal of the 2:1 complex is gradually shifted upfield when 12C“ is replaced. by B15C5 and by 1505 (Figure 28). This might indicate that as the ring size increases, the repulsion between the two rings also increases due probably to the smaller distance of approach of these two rings. This tends to 186 loosen the fit and to reduce the overlap between the rings and the cation. This explanation seems to be borne out by the higher stability of the 185031505)2 complex as compared to K+-(15C5)2. In acetonitrile log Kf values of 2.7 and 2.0 respectively were reported (15). The data can also be rationalized by considering that the K+ ion can or cannot accommodate 10 oxygen atoms around it depending on the spatial arrangement of these atoms. Carbonél3 NMR studies (207) and X-ray diffraction studies (210) have shown that 10 oxygen atoms can be evenly dis- posed around the K+ ion when it is complexed by DB3OClO both in solution (207) and in the solid state (210, Figure 30). In this case the 10 oxygen atoms are "wrapped around" the cation like the seam of a tennis ball (same description as for the K+-nonactin complex (211)). When these atoms are located on the two B15C5 rings, they can probably still come into contact with the K+ ion since the shift of —9.5 ppm for the K+-(B15C5)2 complex indicates that the K+—O interactions are not completely relaxed. The benzene rings must not induce steric hindrance since they can easily avoid each other as they do in the solid state (110, Figure 30). In order for the 10 oxygen atoms of the two 15C5 molecules to come into contact with the cation, the two rings should be very close to each other since 15C5 is larger than BlSC5. This is not possible due to the repulsive forces between the rings. The potassium cation is then left in a cage 187 «”0 © (K*)20324ca Figure 30. Potassium ion-crown ether complexes of various stoichiometries (from Reference 110 except K '1806 Reference 82). 188 slightly larger than itself and the overlap is minimum. In the case of K+°(12C") little repulsion between the two 2, distant rings allows a tighter fit and results in a para- magnetic shift despite the decrease in the number of bind- ing sites. The 39K NMR data in Figure 28 demonstrate that the steric factors rather than the electron donor abilities of each crown or the number of binding sites determine the chemical shifts of K+ccrown complexes. For example the downfield shift observed upon replacement of 18C6 by DBl8C6 clearly arises from steric changes. It has been proposed (10) that the addition of benzene rings reduces the cavity size and induces a decrease in the basicity of the ether oxygens. If these oxygens were free to move like solvent molecules, the decrease in basicity should decrease the para- magnetic shift since the solvent paramagnetic shift origin- ates from electron-pair donation from the solvent to the outer p orbitals of the cations (6l,lUl). Contrary to this, however, an increase in the paramagnetic shift is observed due to the larger "squeezing" of K+ in the smaller DBlBC6 cavity. The notion of "inclusive" and "exclusive" complexes which is well accepted for cryptates (97) may now be ex- tended to the crown complexes. The K+ ion complexes with DBlBC6 (which corresponds to the maximum of the curve) and with all the larger crowns (at the right of this maximum) 189 are inclusive, as well as the "sandwich" complexes ob- tained with the small crowns. In an inclusive complex, the cation is located at the center of the crown cavity, or of the cavity formed by two crown molecules. On the other hand the 1:1 complexes with the smaller crowns than DB18C6 are "exclusive", i434, the cation is displaced from the center of the cavity. In the case of potassium complexes, both the chemical shift of the complex and the extent of the solvent de- pendence of this shift may be used to determine the "in- clusive" or the "exclusive" character of the complex. The inclusive character is associated with a small or negli- gible solvent dependence, ELEL’ less than 5 ppm for K+-18C6, whereas the exclusive character is reflected by a large solvent dependence of the shift, ELEL’ 20 ppm for x+o1505 (Table 27). The patterns seen in Figures 27 and 29 for the Na+ and the Cs+-crown complexes, respectively, may be ration- alized in the same manner. Only the distinctive features of the plots will be discussed. In the case of sodium, there is no sharp maximum of the chemical shift in the series B15C5 . . . DBBOClO (Table 28). However, the very small variation of 5 between 15C5 and B15C5 suggests that the maximum is in between these two (closer to BlSCS than to 15C5). This means that the cavities of all the crowns shown except B1505 are equal to or larger than the Na+ 190 ion. The chemical shifts exhibit a limited solvent de- pendence (3 ppm or less). It can be said that the complexes are "inclusive". Another interesting feature is the very large solvent- independent upfield shift of the Na+-18C6 complex. The 1806 cavity is larger than the Na+ ion but it is too rigid to twist around this cation so that the Na+-O interactions are even more relaxed than in the large crowns. The larger DB2lC7 molecule is able to depart from the planar structure or to contract itself in order to bring the oxygen atoms into contact with the cation. A substantial paramagnetic shift results (Figure 27). The DB2HC8 molecule is even more deformable upon complex formation; a "wrap around" complex is formed in this case, as shown by the relative narrowness of the 23Na lines (169). The DB3OC10 molecule seems to be too large to be twisted enough to hold the Na+ ion tightly, as postulated by Live and Chan (A5). The chemical shifts of the Na+oDB3OClO complexes are somewhat dependent on the solvent (Table 27) and follow the same sequence as the chemical shifts of the solvated Na+ ions in the various solvents (see for example 16“). This behavior strongly indicates that in the complex the Na+ ion is not completely stripped of solvent molecules. The opposite conclusion was arrived at by Live and Chan (”5) who used 1H NMR. However it is to be expected that the 23Na chemical shift is a much more sensitive probe of 191 the sodium ion interaction than the 1H chemical shift. Once again, the shifts seen for the large crown com- plexes with Na+ ions resemble those seen for the Na+ ion in ethereal solvents and for the "sandwich" Na+°(15C5)2 complex. This last complex, which is unstable due to the matching of the 15C5 cavity with the Na+ ion, could only be detected in the weakly donating nitromethane (148). The chemical shifts of the Cs+ocrown complexes do not reach a distinct maximum for any particular crown (Figure 29). However, the pattern observed suggests that the maximum "squeezing" of the Cs+ ion would occur for a crown molecule with a size intermediate between those of 18C6 and DB2lC7. As seen before, the Cs+-O interactions are relaxed in the Cs+-(1806)2 complex (89) even more so than in the large crowns. Only the DB30ClO molecule is large enough to "wrap around" the Cs+ ion, as indicated by the very limited solvent dependence of the shift. Besides, 133 the downfield shift of the Cs resonance upon replacement of DB2AC8 or 032709 by 0330010 suggests that the Cs+ ion is in direct contact with the ether oxygen atoms of the last ligand whereas the Cs+-O interactions are weaker with the other two ligands. As previously noted, a change in the number of binding sites does not seem to affect the chemical shift. However, the nature of the electron donor is a predominant factor. It is seen in Table 29, by comparing DAl8C6 with 18C6, 192 that the substitution of two oxygen atoms by two N-H groups in 18C6 shifts the resonance signal of the complex by about 40 ppm in the paramagnetic direction. A large para- magnetic shift was also observed in the case of Na+ ions (206, see Table 28). The direction of the shift is con- sistent with the fact that alkali metal NMR resonances are at lower fields in N-donor than in O-donor solvents (135, 195). The solvent dependences of the chemical shifts of Cs+ocrown complexes are relatively large and follow the sequence 0330010 < 031806 2 032008 ~ 032709 2 032107 5 1806. It is noticeable that, in contrast with the Na+ocrown com; plexes, the Cs+ocrown complexes are of the "exclusive" type with the exception of the Cs+oDB3OClO and, of course, the "sandwich" Cs+-(l8C6)2 complexes. This "exclusive" charac- ter stems directly from the relatively large size of the + ion. Even when this ion is located at the center of a Cs crown cavity, such as DB2lC7, a large area of the surface of the ion remains exposed to the surrounding solvent mole- cules or anions. This does not apply to smaller cations. Two important consequences may be drawn. First the "inclusive" or "exclusive" character of the complex should be defined from the solvent dependence of the chemical shift rather than fromtfluaposition of the cation in the cavity. Second, the contact and the solvent-separated ion pairs are much more efficiently broken upon complexation 193 of small cations than large cations by crown ethers. The parallelism observed in various solvents for the variation of the 133Cs chemical shift as a function of the crown (Figure 29) is interesting in that it allows a reliable prediction of the chemical shifts of a series of Cs+'crown complexes in a given solvent, if the chemical shift of only one Cs+-crown complex in the same solvent is is known. Some predicted values are underlined in Table 29. Carbon-l3 NMR Studies The formation constants of some alkali cation crown complexes may be evaluated from the variation of the salt/ ligand ratio. The procedure is nearly the same as that previously described for alkali metal NMR. Wilson (212) reported that with a 100 MHz spectrometer the upper limit of reliable values of formation constants obtainable by 13C NMR is about 100. This observation was confirmed by the results of Shih (15) and of Lin and Popov (59) who used an 80 MHz spectrometer. The purpose oftflflmsstudy was to measure the stability constants of the K+oDB2HC8 complex in various solvents. Since this complex was found to be quite stable in methanol (log K > 3, see Table 31), f the upper limit of our technique had to be increased by about two orders of magnitude. This was achieved by 13 running the C spectra on a 250 MHz spectrometer at a 194 .mHm moCoComomo .sz ooCoComomo .mHm moCmLommmo .poumOHCCH omHspoCuo mmmHC: xpoz mHCBn .pouonoCH omthono wmoHC: mmH moCoCmmomm so.owem.H mo.owoo.s . HH.owo=.m oH.0hom.m osHlosm o.Hv so.ome.H . oe . omzo AoH.HV so.OHOH.m mo.oses.H . oe mzo om=.m oma.m osm.H 0mm somImooz sos . oom.m ooe.m omo.owms.m ode.m so.oan.m mo.owmo.m oom.m Honored: mo.ome.= mo.ost.m ocooooe oom.= oos.m mo.osHm.: so.oh:m.m pH.oHos.m so.owmm.m oHHhoHsooooa mx mo.owHH.z :M. NH.osas.m osmCooEossz m+HB m+mo n+x m+mz pCm>Hom Cm mom .oomHmm pm mpCm>Hom mCOHCm> CH +2 Csz mmonQEoo muzmmm.+z mo mpCMumCoo COHmeCom .Hm oHCmB + H E-i 'U C: (U (I) C.) Q 5‘? C0 2 ll 195 point-to-point resolution of 0.015 ppm. 130 chemical shift with the K+/ The variation of the DB2AC8 molar ratio was studied in nitromethane, aceto- nitrile, pyridine and DMF. Some spectra obtained in pyri— dine are shown in Figure 31. The chemical shifts are given in Table 32 and plotted in Figures 32 and 33 as'a function of the mole ratio. The DB2MC8 molecule has three sets of four aliphatic carbons. The three signals shift upfield upon complexation and the signal at the lowest field shifts the most. In nitromethane solutions (Figure 32) the addition of potassium hexafluoroarsenate to a DB2M08 solution results in a gradual coalescence of the three signals into a single peak at equimolar concentrations of K+ ions and the ligand. This behavior, which was previously observed for the K+o DB3OClO complex (207) in the same solvent,.seems to indi- cate an essentially equal interaction between the 8 oxygen atoms of the polyether ring and the K+ ion. Such equal interaction is only possible if in solutions the ligand is "wrapped around" the cation in the same manner as found (207) for the DB30ClO complex (Figure 30). The 1:2 (DB2AC8:K+) complex, which exists in the crystalline state (Figure 30), could not be detected in nitromethane solutions. As expected, the log Kf value for the 1:1 complex is larger than A in this solvent. In acetonitrile, pyridine and DMF solutions, the three 196 0.93 1.15 0.54 M 1.52 0.80 2.28 K7D82408 Figure 31. Carbon-l3 spectra for pyridine solutions con— taining KAsF5 and dibenzo-2U-crown-8 at various mole ratios and at 25°C. 197 mom.mm HHH.mm mmm.wm mo.: owe.om mso.mm on.mm mm.H mmm.mm oom.mm ooo.om mm.m sow.mm smo.mm mmm.mm :m.H mmm.mm Hmm.mm mHo.mm mm.m som.om smo.mm mmm.mm Hm.H mmm.mm :mm.mm mHo.mm mm.H Hom.mm smo.om mmm.mm mH.H mmm.mm Hmm.mm mHo.mm mm.H m:w.om .ommnmm mmm.mm mo.H mHs.mm Hmm.mm mso.om mH.H mmm.om mHs.om omm.mm so.H os:.mm mom.mm moH.om HHo.H moo.os Hom.mm mom.mm mHm.o Hem.mm mmm.mm MOH.mm 3mm.o Hom.o: mom.mm mso.mm mom.o mHm.mm mmm.mm mom.mm ome.o mom.os Hmm.mm www.mm HHs.o smm.oe mms.mm om:.mm mmo.o meo.oz mHH.o: oHs.mm mom.o smm.os 330.0: mms.mm mom.o ssm.H: mes.os mom.mm o mmm.HH mmm.o: mmm.o: o m m H Hmonmmou m m H Hmoemmmg o c ©.om m m+xg_ o m 6.0m m m+mu moHthHsooooa.h. occHlosa .OOHHmN pm mu£m>fiom mSOfihw> cH ondsoo mosmmo.+x one hoe some oHosm oHoZIoCHnm HmoHsono mzz mHIcoohmo .mm cHome 198 .Hm so.ov ocHoHsHa sH odooxo a mo.o u Hmosmmom .oouomppoo poC mCm AEQQ mo.oHV mCMHCm HCoHEoCo one .ARHo> om\omv oCOCoomICoumz CH mCOuoom no mConCmo mmm u ooCoComomo m .A m mHntv moHpHHHCHquomCm oHuonmE Com pmuoophoo who mumHCm HacHEoCo 0C9 .AEQQ mHo.ov CoHpCHommC pCHoa on uCHoa on no Cooho on» no mH Coemo HmuCoEHCono mCBC .AHH Cmpdmno oomv mo oCouoow mung mo mConCmo QM u ooCoCommmo .omlemo CmHCm> 0 Co puma who: mpCoEoCCmmoE 0C9 o .ommlzz CoxCCm 0 Co moms who: mpCoEoCCmmoE oCBm omm.o: omo.oe mme.om Hm.m \ I s:s.o: msH.o: Hoo.mm om.H @. O Mu My 0 @ ommoa iHoe omoém HmH mmm.o: mam.o: mHe.mm mm.H 0. Hog. ‘0 mHoHH mama: Sham oHH n w _ omH.H: omm.o: Hmm.mm mmm.o mHm.HH omm.oz mom.mm How.o so.sm so.sm so.sm om.H ssm.Hs mmm.o: Hmm.mm mss.o om.mm mH.mm ow.sm Hm.o «Hm.HH mmm.os owo.o= oom.o ew.om ss.mm oo.wm o smo.m: sss.os mmm.o: o m m H Hmoemmou m m H H oemmou o o to CIIHHMHII o o o.oo hrdemHII nmCmeoEOCsz moUHEmECOMHmeoEHQ .ooadHosoo I mm cHhoe 199 4 1,2,3 l l l 0.5 1.0 1.5 2.0 167032403 Figure 32. Carbon-l3 chemical shifts (vs acetone d5) as a function of the K+/DB2HC8 mole ratio in (O) acetonitrile and (A) nitromethane at 25°C. 200 signals tend to approach each other as the mole ratio in- creases, but they do not coalesce even at high mole ratio (Figure 33). This behavior is particularly clear in DMF where the spread of the three signals remains quite large (ml.l ppm is m0.3vppm in the other two solvents). The formation constants, which were measured from the variation of the chemical shift of carbon 3 of the DB2UC8 molecule (see Figure 31), are given in Table 31 together with some literature values. In acetonitrile we found the log Kf value to be equal to 3.70:0.16 (Kf = (5i1.8) 103), which corresponds to the value obtained by polarography (213). Thus it seems to be possible to measure Kf values 13C NMR technique provided a much larger than 100 by the high field spectrometer is used. However the large stan- dard deviation clearly indicates that the upper limit of the technique is reached. In pyridine, the signals shift by a large amount upon complexation (Figure 33), which made possible the evaluation of log Kf with each of the three carbons. The three values obtained were identical within experimental error.- However, carbons l and 2 gave large standard deviations indicating once again the limit of the method. As in other solvents, the log Kf value measured with the carbon 3 was kept (log Kf = 3.A6i0.l7). It is interesting to note that the DB2HC8 ligand selectively complexes K+ over Tl+ ions in pyridine while the order is reversed in the other solvents 201 42 "' 8 41k- 3 ppm \ 404- ' +2 _ +1 . 2 39— ‘ ‘ 1 1 1 l L 0.5 1.0 1.5 2.0 2.5 ' K7 032408 Figure 33. Carbon-13 chemical shifts (vs acetone d ) as a function of the K+/DB2AC8_Eole ratio n (0) DMF and (I) pYridine at 25°C. 202 studied (Table 31). This reversal most likely reflects the competition be- tween solvation and complex formation. There is probably some extent of covalency in the Tl+epyridine interaction due to the "softness" of both the Tl+ ion and the nitrogen donor atom of the pyridine molecule (203). The stronger solvation of Tl+ ions by pyridine results in a weaker com- plexation. Although pyridine exhibits a strong donor character (150), it does not solvate strongly the hard alkali cations, and the complexes of these cations with crown ethers are usually quite stable (205). In DMF, which is a good donor, the K+-032uc8 complex is much less stable than in the other solvents studied (log Kf = 1.00:0.08). As found in the other solvents, the stability is lower than that found in the case of Cs+ ions. There is a definite relationship between the stability of the K+'DB2MC8 complex and the spreading of the three signals in a given solvent. In the weakly donating nitro- methane solvent, the complex is stable and the three peaks coalesce at high mole ratio; a "wrap-around" complex is formed. In acetonitrile and pyridine where the complex is less stable, three signals are distinguishable even at high mole ratio, but the separation of the two extreme sig- nals does not exceed 0.3 ppm; the "wrap-around" complex is almost formed. In-DMF, the resonance signals are widely separated (61—63 = 1.1 ppm, see Figure 33) even at a mole ratio of 2.5 where about 85% of the DB2AC8 are complexed 203 (with [DB2AC8] = 0.075 M). This behavior suggests total that the ligand does not "wrap around" the K+ ion. In this last case, it is very likely that one or several solvent molecules remain attached to the K+ ion, as seen for example intflmal:l complex of DB3OC10 with rubidium thiocyanate (216). It was noted earlier that Cs+ ions form "wrap-around" complexes with DB3OC1O only (in the series of ligands studied), whereas Na+ ions do so with DB2lC7 and DB2AC8 (DB27C9 was not investigated). It can be inferred from these results that the DB2HC8 molecule must have the mini- mum size required to surround the K+ ion. Consequently a strong K+-solvent interaction might prevent a complete stripping of the solvation shell by the ligand. The stability of the complex and its conformation are inti- mately related in this case. B. Potassium Cations Interaction with the Conventional Ligand 2,2'-Bipyridine Introduction In general interactions of "conventional", liéia non macrocyclic, ligands with the alkali ions are quite weak, and often they are beyond the sensitivities of most physicochemical techniques. Therefore, in order to detect the formation of such complexes, it is necessary to use 20“ very sensitive probes of either the ligand or the metal cation. Small changes in the immediate environment of the alkali ions in solutions may be detected by the alkali metal NMR technique. Hourdakis (218) studied the interactions of alkali metal ions with 2,2'-bipyridine (BP) in several nonaqueous solvents, using 7Li, 23Na, 133Cs and 13C NMR. Recently 23Na NMR for Vogtle gt 31. (113) referred to the use of a qualitative differentiation of the stability of Na+ ion complexes with ggphenanthroline, 2,2',2"—terpyridine and 2,2'—bipyridine and concluded that the last ligand gives the least stable complex. Hourdakis found the stabilities of the BP complexes to be in the order Li+ > Na+. Complex formation was not detected for the Cs+—BP system. It was of interest to us to complete the study by the "borderline" case which is the K+ ion. Results and Discussion The variations of the 39K chemical shifts and of the linewidths as a function of the BP/K+ mole ratio are given in Table 33 and shown in Figure 3“. Since the complexation was expected to be weak, only solvents with intermediate and weak solvating ability were investigated. Both 6 and AVl/2 values show much less change upon addition of BP than was observed with Li+ and Na+ (100,218), although the range of the chemical shift is larger for 39K than for 205 Table 33. Potassium-39 Chemical Shifts and Line Widths for Nitromethane Solutions Containing KAsF6 (0.075 M) and 2,2'-Bipyridine at Various BP/K+ Mole Ratios (MR) and at 25°C. Nitromethane THF MR 6(ppm) Avl/2 (Hz) MR 5(ppm) 301/2 (Hz) 0 -22.8 .9 0 -17.8 24 0.75 -21.9 17 1.00 -l7.6 26 1.21 —21.5 24 1.98 -17.7 26 1.99 -20.9 30 3.25 -l7.4 29 2.50 -20.3 35 4.37 -17.4 30 3-08 -l9-9 39 7.75 -l7.1 36 3.45 -19.8 41 9.91 -16.6 36 4.25 119,o 54 11.07 —16.6 38 Acetonitrile PC 0 - 2.2 10 O -15.1 62 1.07 - 2.0 13 0.91 -14.6 80 2.18 - 1.9 14 1.76 -l4.7 78 2.96 - 1.8 16 2.93 -14.5 89 3.94 - 1.6 17 3.96 -14.1 95 7.45 - 1.4 21 7.00 -13.8 115 10.63 - 1.2 28 9.86 - -13.4 145 206. I 1 J, 1 2 4 6 8 IO '2 [BP]/[K’] I L l l I Figure 34.' Variation of the 39K chemical shift and line width as a function of the 2,2'-bipyridine (BP)/ K+ ratio in various solvents. 207 7Li or 23Na. It is obvious that the K+-BP interaction is very weak even in nitromethane solutions. Except in this last solvent, the increase in the line width reflects that in the viscosity. It should be noted that while Grillone and Nocilla (109) succeeded in isolating solid (K°BP)+PhuB- complex, in order to do so they had to have a 6-7 fold excess of the ligand. Hourdakis (218) found that the limiting chemical shift in nitromethane is 4.0 ppm (vs LiClOu 4 M in H2O). This value seems to be the largest downfield 7Li chemical shift observed thus far for solutions of lithium salts, and it is considerably further downfield than that observed for dilute lithium salt solutions in pyridine (+2.5 ppm (219)). On the other hand, the highest upfield shift seems to be that of fluorenllithium diethyl ether complex in ether sol- vent of -6.2 ppm (220). Although two different references were used for these measurements, 4.0 M LiClOu and 20% solution of LiCl (both aqueous), the difference between the two is smaller than the experimental error. These results seem to support the explanation proposed t a_. (221) for the very high and low shieldings by Maciel of 7Li in acetonitrile and pyridine solutions, respectively. The neighbor anisotropy effect is even stronger for the Li(BP): complex than for Li+ ion in pyridine solutions. In the Li(BP): complex, Li+ ion is coordinated to the nitro- gen in the plane of the ring, while in the Li+Fl’-Et20 208 complex it is located directly above the plane of the aro- 7 matic carbanion (220). The resulting Li shifts are 10 ppm 7Li shifts apart, about twice as much as the total range of in dilute solutions of lithium salts in common solvents (219,221). It is clearly seen from the above results and from the Hourdakis'results that 2,2'-bipyridine does complex Li+, Na+ and K+ ions in nonaqueous solvents, and that the strength of the interaction varies in the order Li+ > Na+ > K+; it also varies inversely with the solvating ability of the solvents. CHAPTER VI NUCLEAR MAGNETIC RESONANCE IN MOLTEN SALTS 209 A. Introduction In the presence of a solvent, the anion competes with the solvent molecules for the cation and such factors as the dielectric constant and the donor character of the solvent, the sizes and the charge densities of the species in solution determine the result of this competition. In a molten salt only cation and anions are present; however, the nature and the extent of the cation-anion interaction may vary in a large extent, depending on the species present and on the temperature. Nuclear magnetic resonance has proven to be a sensitive technique for the study of cation-anion interactions. Besides, it has been shown (121—125) that NMR can provide valuable information about the structure of molten salts and the identification of the species present (Chapter I, 205 7 Part C). The reported studies were made by Li 23 T1, and Na NMR. The relatively low melting points of some potassium salts or their mixtures with other alkali metal salts (Table 34) led us to extend our 39K NMR studies to molten salts. B. Experimental Part Most measurements were carried out on a highly modi- fied single-coil, multinuclear Varian DA-6O spectrometer 210 .AxoUCH 0>Huomamon u av me n u Csz CoCMHCOHmon .AwmmHv mmopm oHEoUwo< .xoonUCmm pHmm CopHoz qumh .h .0 Sosa spmmm 211 smH soImm mHoHeIHosz m smmv momImomz 1 mmH mmImHIsm moszmoszImozHH moH . oeIzm mozoZImozHH om.m NMH smIma moonmozHH ommms.m. mH.H ommHom.H wmm m0on onoo.m o.H ommomm.H OHm mozmz oommo.m m.m osmHss.H mmm mozHH oaH m.moIm.om moszzomsz mmH osIom ZomHIzomoz oomsm.s oa.m oomImsHom.H msH . zone nw.m mmm Zommz do C o mIEo.w .mm a Hoe Empmzm .CEop .QEoumpHmCoa E CoHpHmOQEoo m.mopsprz 0Hpoouzm 0C0 mpHmm Hmpoz HwaH< oEom Co moHpCoqopm HCOHmmCm .zm oHCmB 212 equipped With an external lH lock system. Special probe inserts were made by W.Burkhardt for various frequency ranges. A high melting glue was used for the attachment of the radio frequency coil to the glass part of the insert. The designs of the inserts and of the home built probe allowed measurements to be made at temperatures up to ”250°C. Preliminary measurements were made on a Bruker WH-l80 spectrometer described in Chapter 11. With this instru- ment, measurements are not possible at temperatures over 150°C or 200°C depending on the core diameter. Samples (W5 ml) were prepared by directly weighing appropriate amounts of purified salts (purification methods were described in Chapter II) in the NMR sample tubes (Wilmad, 15 mm O.D.). In all cases studied, pressure caps were sufficient to resist to the pressure resulting from the heating of the samples. For the measurements, the samples were heated in a sand bath at the temperature of the measurement and rapidly transferred to the probe which was preheated at the same temperature. A few minutes were allowed in order to reach equilibrium. Depending on the composition of the sample and on the line width, the number of scans varied between 1 and 100. The highly moisture sensitive chloroaluminate mixtures were directly inserted in NMR sample tubes under nitrogen atmosphere. 213 C. Results and Discussion The 39K chemical shifts and line widths of the systems investigated are given in Table 35. At 200°C, the signals are quite narrow despite the relative viscosity of molten salts (Table 34). The line widths are comparable to those observed in concentrated aqueous solutions of potassium salts at 23°C (Table 36). At a given temperature, the line widths were nearly the same for all systems investi- gated. The chemical shifts vary in a large range. The shift 0f -17 ppm observed for the LiNO3-KNO3 eutectic mixture at 200°C approaches that observed for the K+ ion solvated in nitromethane at 25°C (—2l.l ppm (15)). Bloor and Kidd (142) studied the anion and the concentration dependence of the 39K chemical shift in aqueous potassium salt solutions. These authors found that the paramagnetic Shielding of the cation induced by the anion is related to the ability of the anion to overlap with the cation. In the molten salt as in aqueous solution, the N03 ion gives an upfield shift 39 of the K resonance with respect to the solvated K+ ion, indicating that the K+-NO3 overlap is smaller than the K+-H2O overlap. In KSCN and in NaSCN—KSCN mixtures, a shift of 13 ppm is observed. The chemical shifts obtained in aqueous potassium thiocyanate solutions are given in Table 36 and plotted in Figure 35. In this figure it is seen that 214 Table 35. Potassium-39 Chemical Shifts and Line Widths for Some Molten Salts and Molten Salt Mixtures. Composition t 6 Av1/2 System mol % °C 20.5 ppm Hz KSCN 200 13.1 23 NaSCN-KSCN 10-90 200 12.8 26 20-80 200 12.6 27 30-70 200 11.4 28 190 10.9 35 170 10.9 52 LiNO3—KNO3 43-57 200 -l7.0 24 (LiNO3-KNO3)-KSCN 90-10 200 -l4.1 31 130 -14.8 104 LiN03-NaN03-KNO3 27-18-55 200 -17.4 23 190 -l6.8 27 180 -16.9 30 170 -l7.2 34 160 -17.3 42 150 -l7.2 52 140 -17.8 63 NaOH-KOH . 50-50 240 43 33 215 Table 36. Potassium-39 Chemical Shifts for Aqueous Solu- tions of KSCN at Various Concentrations.a Concengration Si; . AVE/2 - Hz 0.097 0.06 8 0.200 0.06 8 0.296 0.06 10 0.403 0.15 10 0.502 0.15 10 0.593 0.21 11 0.807 0.30 11 0.985 0.35 11 2.010 0.79 11 3.986 1.95 13 9.028 5.58 20 9.028 (58°C) 4.71 14 9.028 (75°C) 4.42 12 9.028 (87°C) 3.11 11 8At 23°C unless otherwise indicated. b£0.05 ppm. °il Hz. 216 I F r r f“ l I l 16- - A 12- - S ' /’ / ppm 8 ' /’// .- 4 — _ 0 _ I l I l I l 1 l 2 4 6 8 1o 12 14 16 [KSCN] M Figure 35. Variation of the potassium-39 chemical shift as a function of the KSCN concentration in water at 23°C. The dashed line is an extrapolation. The triangle is for the molten KSCN at 200°C. 217 an extrapolation of the chemical shifts observed in aqueous KSCN solutions to the concentration of the pure salt (W16 M) gives a value of about 11 ppm in fair agreement with the value of 13 ppm measured in the molten salt. As more and more NCS- ions are disposed around the potassium ion, the environment of this ion resembles more and more that in molten KSCN; the extent of the electron donation of NCS- to K+ seems to be comparable in both media since the chemi- cal shifts are nearly the same. It should be noted that the strong temperature de- pendence of the chemical shift in aqueous solution (Table 36) is not observed in molten salts (at least in the few cases studied). The upfield shift with increasing tempera- ture is a quite general phenomenon in solutions (Chapter III). It is probably related to the loosening of the cation solvation shell. Sahm and Schwenk (133) obtained values of 23.0, 49.6, '57.6 and 62.5 ppm for the 39K chemical shift of crystalline powders of KF, KCl, KBr and K1 respectively. The order of the shifts is the same as that observed for aqueous solutions of these salts (15,142). Therefore it appears that there is a correlation between the chemical shifts in aqueous solutions and those in molten salts or in crystals A decrease in temperature does not significantly affect the chemical shift, nor does a change in the cation composi- tion of the mixture as long as the anion remains the same. 218 This behavior, which was observed for the nitrates as well as the thiocyanates mixtures (Table 35), is consistent with the fact that the nearest neighbors of the cations remain the same. Nevertheless, it.should be noted that a change in the composition of a mixture results in a slight Change in the average K+—anion distance, due to the dif- ferences in electrostatic attractions of different cations for the common anion (125). The change in distance should, in turn, affect the chemical shift. In fact the chemical shift variation is too small to be detected in the mixtures investigated here. Harold-Smith (125) could not detect any 7Li chemical shift upon changing the cation composition in LiNO3-KNO3 mixtures. Likewise, we could not detect any 39K chemical shift upon replacing in part LiNO3 by NaNO3 in the same mixtures (Table 35). Relaxation times are more sensitive than chemical shifts to changes in the composi- tion of the mixtures (125). The upfield shift observed in going from the pure KSCN to the NaSCN-KSCN (30-70 mol %) mixture is only three times larger than the experimental error, which precludes any interpretation. [However, it is noticeable that the direc- tion of the shift is that expected from an increase in the K+-NCS- distance due to the stronger electrostatic attrac— tion of the sodium ion for the nitrate ion. When the anion cOmposition is changed, the chemical shift should change according to the respective populations 219 F of the various anions in the mixture. Indeed when 10% Of KSCN replace 10% of LiNOB-KNO3 mixture, the 39K resonance signal shifts downfield by 3 ppm, which is 10% of the dif- ference in chemical shift between KSCN and the nitrates mixture (Table 35). The large paramagnetic shift of 43 ppm observed in the molten NaOH—KOH (50-50 mol%) mixture at 240°C is nearly the same as that observed for the crystalline powder of KCl, 49.6 ppm (133). In molten hydroxides the shift is large considering that, in an aqueous solution of NaOH and KOH (5 M each), the chemical shift is only 6.8 ppm. A few 23Na and 27 Al NMR measurements were made in chloroaluminate melts. At 193°C, in the AlCl3-NaC1—K01 (66-20-14 mol %) eutectic mixture (mp = 89°C) which cor- responds to the (Na,K)Al2C17 stoichiometry, the 27A1 chemi— cal shift is 105 ppm, which is in the range of shifts found for the species A101; and A12C16 in various solvents (85 ). In the same mixture the line width broadens from 350 to 700 Hz in going from 193 to 144°C, which makes the chemical shifts measurements difficult at the lower temperatures. The 27A1 signals are much narrower, iLQL, Avl/2 = 100 Hz at 160°C, in the A101 -NaCl (50-50 mol %) mixture 3 (mp 154°C (115)), due to the tetrahedral environment around the Al nucleus in the A101; species. Within experimental error, the chemical shift is identical to that measured for the previous ternary system, so that chemical shifts 220 cannot be used to identify the various species present in chloroaluminate melts. Conclusion The 39K chemical shift range, which is about 40 ppm in dilute solutions, extends to about 100 ppm in molten salts and in crystals due probably to the closer distance of approach of cations and anions in these systems. In molten salts, potassium-39 lines are relatively narrow. The chemical shifts observed in molten salts correlate those found in aqueous solutions of the corresponding salts and do not vary significantly with the temperature or the cation composition of the mixtures. APPENDIX 1 APPLICATION OF THE COMPUTER PROGRAM KINFIT TO THE CALCULATION OF COMPLEX FORMATION CONSTANTS FROM NMR DATA The KINFIT computer program was used to fit the potas- sium-39 and carbon-13 NMR chemical shift X§_mole ratio data to Equation (3) of Chapter V which was used as the SUB- ROUTINE EQN. t t = [(KfCM"KfCL - 1) + (K20t2 + K20t2 - 5 f L f M obs °M ' 6ML )+ 5 2K 0 )1/2] ( t f M + 2K Ct + 2K Ct + 1 t C M f L f M C 2K (3) 2 t f L ML Equation (3) has two unknown quantities, and K °ML f’ designated as U(1) and U(2) respectively in the FORTRAN code. The two input variables are the analytical concentra- tion of the ligand (Ct, M) and the observed chemical shift (60b8, ppm) which are denoted as XX(l) and XX(2) respectively in the FORTRAN code. Starting with a reason- able estimate for the value oflgsand 6 the program ML’ 221 222 fits the calculated chemical shifts (the right hand side of equation (3) to the observed ones by iteration method. The first control card contains the number of data points (columns 1-5 (Format I5)), the maximum number of iterations allowed (columns 11-15 (F 15)), the number of constants (columns 36-40 (F 15)) and the convergence tol- ! erance (0.0001 works well) in columns 41-50 (F 10.6). The second control card contains any title the user desires. The third control card contains the values of CONST(1) (0;, M) in columns 1-10 (F 10.6) and CONST(2) (5M, ppm) in columns 11-20 (F 10.6); other constants can be listed in columns 21-30, 31-40, etc. The fourth and final control card contains the initial estimates of the unknowns U(1) = 5ML and U(2) = Kf,in columns 1-10 and 11-20 (F 10.6) respectively. The fifth through N cards are the data cards which contain XX(1) = C1: L variance on XX(1) in columns 11-20, XX(2) = the chemical in columns 1—10 (F 10.6), the shift at XX(1) in columns 21-30 (F 10.6) and the variance on XX(2) in columns 31-40 (F 10.6) followed by the same parameters for the next data point. Each card may contain two data points. The SUBROUTINE EQN is given below. APPENDIX 2 SUBROUTINE EQN 1Q nu<75 “New/VQEDT/I'ET“ ~qu/onxvf/«OOT-JOD sgtcw xlb.1OODoUI 7 lo ‘ orccoI.LnOt :0 1,039.95 1 9(Incu I ' C (fur?u.NC.-l' GO to 3‘ YHDN i . - CCIouIlI a Yuan x ((1%? I‘d”! IfI;.:r~.~:,-gv oo 70 70 offIJON CON! [Nut affnou cQ-Itzmnf at anq CO"!I~"( anHON CO"! t~ll€ ntfunn [~O 2223 onloq ‘OIJ~I.IYLD£.JY‘9‘:O {'7IL.’|I'Vcoounp56U3VI90NnhN‘010”. I YI‘II x.It