Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Aqueous coordination chemistry of aluminum and gallium with 3-hydroxy-4-pyridinone ligands Nelson, William Otto 1988

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1989_A1 N44.pdf [ 9.29MB ]
Metadata
JSON: 831-1.0060259.json
JSON-LD: 831-1.0060259-ld.json
RDF/XML (Pretty): 831-1.0060259-rdf.xml
RDF/JSON: 831-1.0060259-rdf.json
Turtle: 831-1.0060259-turtle.txt
N-Triples: 831-1.0060259-rdf-ntriples.txt
Original Record: 831-1.0060259-source.json
Full Text
831-1.0060259-fulltext.txt
Citation
831-1.0060259.ris

Full Text

AQUEOUS CORDINATION CHEMISTRY OF ALUMINUM AND GALIUM WITH 3-HYDROXY-4-PYRIDINONE LIGANDS by WILIAM OTO NELSON B.Sc, University of Wisconsin-Superior, 1984 A THESIS SUBMITED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGRE  OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Chemistry) We acept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 198 © Wiliam Oto Nelson, 198 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) ii ABSTRACT Aluminum(II) and galium(II) complexes with the folowing ligands were synthesized: 3-hydroxy-2-methyl-4(lH)-pyridinone (Hmp), the 1-methyl (Hdp), 1-ethyl (Hmep), and 1-hexyl (Hmhp) derivatives of Hmp, and P-[N-(3-hydroxy-4-pyridinone)]-a-aminopropionic acid (mimosine). The 3-hydroxy-4-pyridinone ligands employed in this study (except mimosine) were prepared by the conversion of an oxygen heterocycle, 3-hydroxy-2-methyl-4-pyrone, to the coresponding nitrogen heterocycle by reaction with a primary arnine. These bidentate ligands contain an a-hydroxyketone moiety and their conjugate bases form neutral complexes with trivalent metals. The ligands and the metal complexes were fuly characterized by mas spectrometry and infrared, proton NMR, and ultraviolet spectroscopy. The structures of several ligands and metal complexes were determined by X-ray difraction. Hmp, Hdp, and Hmep  crystalize as centrosymetric O-H—0=C hydrogen bonded dimeric units. The facial geometric isomers of Al- and Ga(dpp)3 crystalize as the dodecahydrate in which the water molecules are asociated in hexagonal rings similar in structure to that of ice Ih. The oxygen atoms of the metal complexes are hydrogen bonded to bridging waters so that the water rings and metal complexes are interconected in a thre-dimensional aray. An analogous water network is found in the structures of Al- and Ga(mepp)3. The proton NMR spectra in CD3OD and D2O indicate the metal complexes are fluxional above -30 °C. Variable-temperature proton NMR experiments identified the exchange proces as facial to meridional geometric isomerization. Ligand exchange experiments using proton NMR indicated the isomerization folows an intermolecular rather than intramolecular pathway in CD3OD. Variable-pH 27Al NMR experiments show the tris-ligand aluminum complexes to be resistant to hydrolysis from pH 4-9. The formation iii constants of the metal-ligand complexes were determined by potentiometric titrations, and this study indicates the galium complexes have a similar pH region of hydrolytic stability. The overal formation constants for the tris-ligand aluminum and galium complexes were al greater than 1030, indicating that these ligands could compete for aluminum and galium in blod plasma models. Water solubilities and octanol/water partition coeficients of the ligands and metal complexes were measured and they indicate the suitability of these complexes for 67Ga animal biodistribution experiments. The results of the biodistribution study show that under conditions of ligand exces  67Ga is redirected from transferin; however, the 67Ga-ligand complexes do not localize in any organs. It apears the ligands greatly enhance the removal of the radionuclide from the body. iv TABLE OF CONTENTS Abstract ii Table of Contents iv List of Tables viii List of Figures x List of Abbreviations xiii Acknowledgements xvi Chapter I General Introduction 1 Chapter n Synthesis and Characterization 10 A. 3-Hydroxy-4-Pyridinone Ligands 10 2.1 Introduction 10 2.2 Materials and Methods 15 Method A 2.2.1 Preparation of B zmpp 15 2.2.2 Preparation of Hmpp 16 2.2.3 Preparation of Hdpp 16 2.2.4 Preparation of Hmhpp 16 2.2.5 Preparation of H2exn 16 Method B 2.2.6 Preparation of Hmpp 17 2.2.7 Preparation of Hdpp 17 2.2.8 Preparation of Hmepp 18 2.3 Discussion of the Synthetic Procedure 18 2.4 Characterization of the 3-Hydroxy-4-Pyridinone Ligands 22 2.4.1 Elemental Analysis 23 2.4.2 Infrared Spectroscopy 24 V 2.4.3 Proton NMR Spectroscopy 29 2.4.4 Electron Impact Mas  Spectrometry 3  B. Tris(3-Hydroxy-4-Pyridinonato) Metal Complexes 36 2.5 Introduction 36 2.6 Material and Methods 41 2.6.1 Preparation of Al(mpp)3 42 2.6.2 Preparation of Ga(mpp)3 42 2.6.3 Preparation of Al(dpp)3 42 2.6.4 Preparation of Ga(dpp)3 42 2.6.5 Preparation of Al(mhp>3 42 2.6.6 Preparation of Ga(mhpp)3 43 2.6.7 Preparation of Al(mepp)3 43 2.6.8 Preparation of Ga(mep)3 43 2.6.9 Preparation of Al(mimo)3 43 2.6.10 Preparation of Ga(mimo)3»H20 4  2.7 Discusion of the Synthetic Procedure 4  2.8 Characterization of the Tris(3-Hydroxy-4-Pyridinonato) Metal Complexes 48 2.8.1 Elemental Analysis 49 2.8.2 Infrared Spectroscopy 50 2.8.3 Mas  Spectrometry 57 Chapter HI Solid State Studies 58 A. 3-Hydroxy-4-Pyridinone Crystal Structures 58 3.1 Introduction 58 3.2 Results and Discusion 62 B. M(dpp)3 Crystal Structures 68 3.3 Introduction 68 vi 3.4 Results and Discusion 70 3.4.1 The/ac-M(dp)3 Unit 70 3.4.2 The Hydrogen Bonded Water Network 76 C. M(mep)3 Crystal Structures 79 3.5 Introduction 79 3.6 Results and Discusion 81 Chapter IV NMR Studies 93 A. Aluminum-27 NMR Spectroscopy 93 4.1 Introduction 93 4.2 Materials and Methods 96 4.3 Results and Discusion 97 B. Variable-Temperature Proton NMR Spectroscopy 109 4.4 Introduction 109 4.5 Materials and Methods 12 4.6 Results and Discusion 12 Chapter V Solution Studies 129 A. Aqueous Solubility 129 5.1 Introduction 129 5.2 Materials and Methods 130 5.2.1 Determination of Molar Absorptivity 131 5.2.2 Determination of Aqueous Solubility 131 5.3 Results and Discusion 13 5.3.1 Ultraviolet Spectroscopy 13 5.3.2 Aqueous Solubility at 25 and 37 °C 135 B. Octanol/Water Partition Coeficients 140 5.4 Introduction 140 5.5 Materials and Methods 142 vi  5.6 Results and Discusion 14 C. Potentiometric Equilibrium Measurements 150 5.7 Introduction 150 5.8 Results and Discusion 151 Chapter VI Galium-67 Biodistribution Study 161 6.1 Introduction 161 6.2 Materials and Methods 162 6.3 Results and Discusion 163 Chapter VII Conclusion and Sugestions for Future Work 168 References 171 Apendix 180 Procedure A.l. Potentiometric Equilibrium Measurements 185 vii LIST OF TABLES Page Table 1.1. Efective ionic radi and transferin binding constants 5 Table 2.1. Results of the free ligand elemental analyses 23 Table 2.2. Infrared bands for Hmp  and substituted analogues 26 Table 2.3. Selected infrared absorption bands in the fre ligands 27 Table 2.4. Proton NMR chemical shift data for the free ligands 31 Table 2.5. Mas  spectral data for the fre ligands 3  Table 2.6. Comparison of T| values and C/TR for selected bivalent metal ions 37 Table 2.7. Results of elemental analyses of the metal complexes 49 Table 2.8. Asignment of VM-0 m the Al and Ga complexes 51 Table 2.9. Characteristic IR absorptions of the Al and Ga complexes 51 Table 2.10. *H NMR data for mimosine and M(mimo)3 complexes 5  Table 2.1. *H NMR data for the tris-ligand metal complexes 56 Table 2.12. Data from FAB-MS spectra of the tris-ligand metal complexes 57 Table 3.1. Bond lengths and angles for the free ligands 61 Table 3.2. Comparison of bond lengths betwen 2- and 4-pyridinone 64 Table 3.3. Fre ligand H-bond parameters and IR stretching frequencies 65 Table 3.4. Bonding parameters for the M(dp)3»12H20 complexes 72 Table 3.5. Intra-anular torsion angles of the chelate rings in M(dp)3 73 Table 3.6. Bond angles for the metal-ligand interactions in M(dp)3 74 Table 3.7. Metal ionic radi and M-O bond lengths in M(dp)3 complexes 75 Table 3.8. Hydrogen bond distances and angles for M(dp)3»12H20 76 Table 3.9. Bond parameters for M(mep)3»12H20 complexes 83 Table 3.10. Unit cel dimensions for M(dp)3 and M(mep)3 complexes 84 Table 4.1. NMR properties of several quadrupolar nuclei 95 Table 4.2. 27A1 NMR data for the tris(3-hydroxy-4-pyridinonato) Al complexes 98 ix Table 4.3. Kinetic data for ris(3-hyclroxy-4-pyridinonato) metal complexes at Tc 17 Table 5.1. ^max and E for the B-band in the fre ligands 13 Table 5.2. Xmax and £ for the B-band in the metal-ligand complexes 134 Table 5.3. Water solubility of fre ligands at 25 °C 136 Table 5.4. Water solubility of the metal-ligand complexes at 25 °C 137 Table 5.5. Log P values of some comon drugs 141 Table 5.6. Log P values for the 3-hydroxy-4-pyrone complexes 147 Table 5.7. Log P values for the 3-hydroxy-4-pyridinone complexes 148 Table 5.8. Log K and log (3n for 3-hydroxy-4-pyridinone metal complexes 152 Table A. 1. Molecular weights for ligands and metal-ligand complexes 180 Table A.2. Crystalographic data for Hmp, Hdp, and Hmep 181 Table A.3. Crystalographic data for M(dp)3 complexes 182 Table A.4. Crystalographic data for M(mep)3 complexes 183 Table A.5. Hydrogen bond distances and angles for In(dp)3«12H20 184 Table A.6. Hydrogen bond distances and angles for M(mep)3»12H20 184 X LIST OF FIGURES Page Figure 1.1. Speciation diagram for ImM Ga-*+ at 25 "C 3 Figure 1.2. Catechol to o-semiquinone to o-benzoquinone redox cycle 6 Figure 1.3. Heterocyclic ligands comparable to o-semiquinone 7 Figure 2.1. Mechanism for the conversion of a 4-pyrone to a 4-pyridinone 10 Figure 2.2. Precursor 3-hydroxy-4-pyrones 11 Figure 2.3. Ligand structures and abbreviations 14 Figure 2.4. Schematic representation of Method A 18 Figure 2.5. Infrared spectrum of Hmpp from 1700 to 300 cm - 1 28 Figure 2.6. Proton NMR spectrum of Hmpp in (CD 3) 2SO at 80 MHz 32 Figure 2.7. Mass spectrum of H2exn 35 Figure 2.8. Five-membered chelate ring 36 Figure 2.9. Tris(N-substituted-3-hydroxy-4-pyridinonato) metal complexes 41 Figure 2.10. Equation for the formation of the tris-ligand metal complexes 44 Figure 2.11. IR spectra of Al(mpp)3 from 1450 to 1200 cm-1 46 Figure 2.12. IR spectra of Hmpp and M(mpp)3 complexes from 900 to 300 cm - 1 52 Figure 2.13. IR spectra of Al(dpp)3 and Hdpp from 1700 to 1400 cm'1 53 Figure 3.1. 4-Pyridinone resonance forms 58 Figure 3.2. ORTEP view of Hmpp, Hdpp, and Hmepp 60 Figure 3.3. Resonance forms for 2-pyridinone 63 Figure 3.4 Hdpp hydrogen bonded dimeric unit 65 Figure 3.5. Stereoview of the packing arrangement in Hmpp 66 Figure 3.6. Stereoview of the packing arrangement in Hdpp 67 Figure 3.7. Stereoview of the packing arrangement in Hmepp 68 Figure 3.8. ORTEP view down the c axis of the M(dpp)3 unit cell packing 71 Figure 3.9. Stereoview of the M(dpp)3 unit cell packing and stereoview of ice Ih 72 xi Figure 3.10. ORTEP view of a part of the H-bonding network of waters in M(dp)3..73 Figure the tris(ligand) portion of M(dp)3 75Figure 3.12. ORTEP view of the tris0igand) portion of M(mep)3 86 Figure 3.13. ORTEP view down the c axis of the unit cel packing of Al(mep)3 91 Figure 3.14. ORTEP view down the c axis of the unit cel packing of Ga(mep)3 92 Figure 4.1. 27A1 NMR spectra of Al(mp)3 at pH 1.8 9 Figure 4.2. Variable pH 27Al NMR spectra of Al(mp)3 102 Figure 4.3. 27A1 NMR spectra of Al(mpp)3 at pH 8.9 103 Figure 4.4. Variable-pH 27Al NMR spectra of Al(dp)3 106 Figure 4.5. Variable-pH 27Al NMR spectra of Al(mimo)3 107 Figure 4.6. Variable-pH 27A1 NMR spectra of 3:1 Hmep  to A1(N03)3 108 Figure 4.7. Isomers of trivalent metals with asymetric bidentate ligands 109 Figure 4.8. 30 MHz *H NMR spectrum of Ga(dp)3 in CD3OD at 18 °C 13 Figure 4.9. 30 MHz *H NMR spectrum of Al(dp>3 in C D 3 O D at 18 °C 14 Figure 4.10. 30 MHz lH NMR spectrum of Al(dpp)3 in (CD3)2SO at 18 °C 15 Figure 4.1. 30 MHz *H NMR spectrum of Al(dp)3 in D2O at 18 °C 16 Figure 4.12. CH3c spectra for Al(dp)3 and Ga(dp)3 in C D 3 O D at -30 °C 17 Figure 4.13. Experimental and simulated CH3C spectra in Al(dp)3 18 Figure 4.14. Variable-temperature *H NMR spectra of CH3C in Al(dp)3 121 Figure 4.15. Variable-temperature *H NMR spectra of Ga(dp)3 in C D 3 O D 12 Figure 4.16. lH NMR spectrum of 3:1 Al(ma)3 to Al(dpp)3 in C D 3 O D at 18 °C 126 Figure 4.17. *H NMR spectra of Ha and Hb in Al(dp)3, Al(ma)3, and 3:1 mixture.. 127 Figure 4.18. Variable-temperature lH NMR spectra of Hb in mixed-ligand species 128 Figure 5.1. Equilibrium equations and constants for M3+ with bidentate ligands 150 Figure 5.2. Speciation diagrams for Al-Hdp at 3 mM and 3 uM Hdp 154 Figure 5.3. Comparison of 27Al NMR spectra and Al-Hdp speciation profiles 15 xi  Figure 5.4. Speciation diagram for Ga-Hdp 156 Figure 5.5. Protonation equations for 3-hydroxy-4-pyridinone ligands and maltol 157 Figure 5.6. Plot of Al3+ complexation vs. log of total ligand concentration 158 Figure 5.7. Plot of Hmp, Hdp, and Hmep  logP3 values 160 Figure5.8. Plot of Hmp, Hdp, and Hmep pKa values 160 Figure 6.1. Liver biodistribution ploted as 67Ga uptake vs. ligand 165 Figure 6.2. Graph of %Ga complexed vs. ligand injection concentration 167 xii LIST OF ABREVIATIONS Abreviation Meaning acac acetylacetonate a-RT a-substituted tropolonate AI13 [A104Ali2(OH)24(H20)i2]7+ Al(dp)3 tris(3-hydroxy-1,2-m^ethyl-4-pyriaUnonato)aluminvim(II) Al(ma)3 tris(maltolato)alumum(IJI) Al(mep)3 tris(3-hydroxy-2-methyl-1 -ethyl-4-pyridinonato)aluminum(nT) Al(mhp)3 tris(3-hydroxy-2-methyl-1 -hexyl-4-pyrioUnonato)alurninurn(lI) Al(mimo)3 rris(nTimosinato)aluminum(III) Al(mp)3 ris(3-hyo!roxy-2-methyl-4-pyridinonato)aluminurn(i) Pn overal formation constant B-band benzenoid ultraviolet absorbance band BB blod-brain barier Bzma 3-benzyloxy-2-methyl-4-pyrone Bzmp  3-benzyloxy-2-methyl-4(lH)-pyridinone Catechol 1,2-dihydroxybenzene Chlorokojic acid 5-hydroxy-2-(chloromethyl)-4-pyrone C/IR charge to ionic radius ratio AV frequency shift in Hertz betwen peaks in absence of exchange Ave experimentaly observed frequency shift Sx.y vibrational in-plane bending mode £ molar absorptivity EDTA ethylenediaminetetracetic acid FAS flame atomic absorption spectroscopy xiv fac facial geometric isomer Ga(ma)3 tris(maltolato)galium(JTJ) Ga(dpp)3 tris(3-hydroxy-1,2-cumemyl-4-pyricUnonato)gaUijmi(II) Ga(mep)3 tris(3-hydroxy-2-methyl-1 -emyl-4-pyricUnonato)galium(in) Ga(mhp)3 tris(3-hyclroxy-2-methyl-l-hexyl-^pyricUnonato)galium(in) Ga(mimo)3 tris(mimosinato)gam'um(in) Ga(mpp)3 ms(3-hyo!roxy-2-memyl-4-pyridinonato)galium(II) rj absolute hardnes H-bond hydrogen bond Hdp 3-hydroxy-1,2-dimethyl-4-pyridinone H2exn 1,6-oU(3-hyd^oxy-2-methyl-4-pyridinone)hexane Hmep 3-hydroxy-2-methyl-1 -ethyl-4-pyridinone Hmhp 3-hydroxy-2-methyl-1 -hexyl-4-pyridinone Hmp  3-hydroxy-2-methyl-4(lH)-pyridinone IR infrared Kn stepwise formation constant Kojic acid 5-hydroxy-2-(hydroxymethyl)-4-pyrone Xmax wavelength of maximum absorbance L-dopa L-3,4-dihydroxyphenylalanine Maltol 3-hydroxy-2-methyl-4-pyrone mer meridional geometric isomer mQ milUCurie; equals 3.7 x 107 disintegrations/second Mimosine P-[N-(3-hydroxy-4-pyridinone)]-a-aminopropionic acid NMR nuclear magnetic resonance Vx.y vibrational stretching mode P H-oca/wate partition coeficient XV 7CX-y pyromeconic acid tfac T c T i T 2 TQ UV W1/2 vibrational out-of-plane bending mode 3-hydroxy-4( lH)-pyridinone trifluroacetylacetonate coalescence temperature spin-lattice (longitudinal) relaxation spin-spin (transverse) relaxation nuclear quadrupole relaxation correlation time ultraviolet peak width at half height xvi ACKNOWLEDGEMENTS I would like to thank the technical staf of the Mas  Spectroscopy lab and the Chemistry Department suport staf for al their help. I would especialy like to thank Dr. S.O. Chan and Marieta Austria of the NMR lab for their help with the 27Al NMR, and Steve the elder and Steve the younger in the glasblowing shop. Many thanks to Randy for the lengthy conversations about D N M R 3 and to Ove for listening to my organic ramblings. The financial asistance in the form of a University Graduate Felowship and a Teaching Asistantship is gratefuly acknowledged, as is a research grant received from Sigma Xi. Those whose work apears in this thesis are deserving of special mention. Therefore I would like to thank Steve Retig to whom I owe a great debt, Peter Borda for his prompt determination of the elemental analyses, Dave for al his hard work and copious quantities of speciation diagrams, Tim K. and Carey for their synthetic endeavors, Tami for having to smel as much octanol as I did, and Don, Teri and Gordon for al their work at the hot lab. This would not be complete without acknowledging the suport and encouragement of al the members of the "Orvig Team" past and present. A special thank you is extended to Alexis for changing Orvig's student into Orvig's students, and of course to Chris himself for being wiling to take on a graduate student who, acording to F. Aubke, was not only older than his research director but also had more hair! Finaly, to Marian who would make my life a misery if I didn't thank her, and to Sir Isac Newt wherever he might be, "thank's for making those lonely nights in the lab les lonely." xvi  To Marian, and to my parents, Bud and Doris,who always wanted a doctor in the family. I leave my coleagues on the Orvig team with the imortal words of Yogi Bera in the seventh game of the 1960 World Series: "It ain't over 'til it's over." 1 Chapter I General Introduction Aluminum is the most abundant metal and the third most comon element in the earth's crust. It is bound as oxides, primarily alumina, and complex aluminosilicates. Surface water concentrations have remained minimal due to the low solubility of the natural Al minerals; however, acidification due to acid precipitation has facilitated the transport of Al from soil to surface waters.1 Elevated concentrations of Al have ben reported for lakes and rivers in regions throughout the northern temperate zone that receive high inputs of acidic substances, including eastern Canada2 and the northeastern United States.3 The increasing concentrations of Al in surface waters can be considered an unforesen consequence of human activity. By design, municipal water suplies can have high levels of Al (up to 1 mg/L) because Ai2(SO,4)3 is comonly used as a floculant during water purification.4 Al is also found in relatively high concentrations in a number of products used for human consumption5 including drugs (e.g., 10-15 mg per tablet of bufered aspirin), procesed fods (chese can be up to 0.7% Al), and baking powder (about 5% Al). In spite of this exposure, Al is largely excluded from the body because of low gastrointestinal absorption and eficient renal excretion. On average, Al is ingested in the range of 20-50 mg/day and the total body burden in normal persons is about 30 mg.2 As recendy as 1974 Al was regarded as a generaly benign element,6 but over the last twenty years a large body of evidence has acumulated to link Al with several neurological dysfunctions that apear to have an environmental etiology.4'78 Dialysis encephalopathy (DE) and amyotrophic lateral sclerosis (ALS) are especialy interesting because of their asociation with Al in drinking water. Patients on long term dialysis are 2 the victims of DE and a corelation has ben established betwen the levels of Al in the dialysis solutions and this often fatal syndrome.9 Natives of the island of Guam have an inordinately high incidence of ALS (comonly refered to as Lou Gehrig's disease) and it has ben hypothesized that this could be related to the high levels of Al found in the soils and surface waters of the regions where the disease is endemic.10 The asociation of Al with Alzheimer's disease was first recognized through animal experiments where it was found that aluminum-induced encephalopathy had a clinical course and histopathology similar but not identical to that found in Alzheimer's disease.11-2 It has since ben established that victims of Alzheimer's have elevated Al levels (10 to 30 times above normal) in the nuclei of the afected brain cells.13 Although it has not ben proven that Al is the cause of Alzheimer's or any other human neurological disorder, the circumstantial evidence is compeling when coupled with the wel established neurotoxicity of Al (first documented14 in 186). The case against Al is strong enough that several studies have ben undertaken to ases chelating agents for the removal of this "benign" element from the body.4'815 It was in the context of the potentialy deleterious consequences of the increasing environmental exposure to aluminum that we first became interested in the aqueous cordination chemistry of this group 13 metal. Our interest in galium is due to its utilization as a radiopharmaceutical imaging agent. Ga has two isotopes that lend themselves to the detection methods of nuclear medicine: 67Ga (ti/2 = 78.1 h; y= 93.3,185, 30 KeV; acelerator product) and 68Ga (ti/2 = 68.3m;y=51 KeV from (3+ anihilation; generator product). The discovery in the 1960's that 67Ga (administered as the citrate) localized in soft tisue tumors16 engendered considerable clinical research and the citrate (the only comercialy available radiopharmaceutical of 67Ga) is now widely used in oncological nuclear medicine.17"9 Despite its availability and easy detection, the use of 67Ga has ben limited to the detection of tumors and inflamatory lesions.20'1 3 The aqueous cordination chemistry of Ga with multidentate ligands incorporating hydroxy, amino, carboxylate, and catechol moieties has ben explored with nuclear medicine in mind;2-24 however, there has not ben much work on the aqueous chemistry of this trivalent metal with simple bidentate ligands. One reason for this is the hydrolytic instability of the group 13 metals; Al and Ga undergo extensive and complex hydrolysis that is pH, concentration, and time dependent.7-25 Figure 1.1 is a diagram showing the pH dependent speciation of Ga3+ (the speciation diagram for Al3+ is very similar). The diagram is simplified as no polymeric species are included and it ilustrates the amphoteric nature of the Ga-hydroxide complexes. Al and Ga form insoluble neutral hydroxides in the pH 4-8 region and the logarithm of the overal formation constants for the Al- and Ga(OH)3 gelatinous precipitates are 12.7 and 23.2, respectively (based on data from ref. 25). To chelate these metals in an aqueous environment, a ligand must be able to compete with the hydroxide ion. %of total An . Ga 40 H Ga hexaaqua ion Ga(OH) Ga(OH)2 Ga(OH)3 Ga(OH)4 PH 10 Figure 1.1. Speciation diagram for ImM Ga3+ at 25 °C (based on data from ref. 25). 4 We are interested in ligands and metal-ligand complexes that are "biologicaly relevant". For the purposes of our research, biologicaly relevant compounds would have the folowing properties: molecular weight (MW) < 60, water solubility, enough lipophilicity to cros cel membranes, neutral charge at physiological pH (7.4), and (for the metal-ligand complexes) significant thermodynamic stability. In adition, the ligands ned to be reasonably nontoxic as they are to be used in animal biodistribution studies with 67Ga. This is of use both to determine the eficacy of the ligands as in vivo directors of this radionuclide and to gain some idea of the biodistribution of the Al complexes as there are no Al isotopes suitable for this type of study. In 67Ga imaging experiments, water solubility at the |J.M level (or lower) is suficient; for facile spectral characterization of the metal-ligand complexes, solubility > 1 mM is desirable. The constraints on size, lipophilicity, and charge are necesary because of our interest in ligands and metal-ligand complexes that could conceivably enter the brain. The brain is separated from the circulatory system by the most selective barier in the body, the blod-brain barier (BB). It has ben shown that the permeability of the BB is directly related to the size and lipophilicity (and therefore charge) of a substance 26 The uper limit on size has yet to be determined, but it is thought that penetration of the BB is not inhibited by size for compounds of MW < 500.27 For in vivo experiments, the ned for thermodynamic stability is dictated primarily by the afinity of the blod serum protein transferin for trivalent metal ions 28 Transferin is an iron transport protein that has two binding sites for metal ions similar in size to Fe3+. (The ionic radi and transferin binding constants (Ki and K2) for Al, Ga, In, and Fe are listed in Table 1.1.) The large binding constants and the high concentration of vacant transferin binding sites in human blod (-50 p^M29) make transferin a powerful scavenger of trivalent metal ions. The size specificity of transferin binding is shown by the large diference in the stability of the Al and In complexes. When 67Ga-citrate is used for 5 imaging, the observed biodistribution is that of 67Ga-transferin as citrate canot compete for the metal ion. A ligand must be able to maintain the radionuclide when confronted with the high levels of transferin in the circulatory system even to be considered as a 67Ga imaging agent. Table 1.1. Efective ionic radi (six cordinate)3 and transferin binding constants.1* Al Ga In Fe Ionic Radius (A) 0.535 0.620 0.80 0.645 Log Ki 12.9 20.3 30.5 2.7 Log K.2 12.3 19.3 25.5 2.1 a From ref. 30. b In from ref. 28, Al and Fe from ref. 29, and Ga from ref. 31. The group 13 metals have an afinity for oxygen containing compounds and the research in our laboratory has concentrated on bidentate ligands with dihydroxy or a-hydroxyketone binding groups. My initial project was the synthesis of Al and Ga complexes with several 1,2-dihydroxybenzene derivatives (catechols). Various catechols have ben used as ligands for a number of metals,32 including aluminum33 and gallium.34 In adition, the catecholamines (dopamine, noradrenaline, and adrenaline) are adrenal medulary hormones responsible for the cordination of the sympathetic nervous system and the adrenal medula. The catechol ligand of greatest interest to us was L-3,4-dihydroxyphenylalanine (L-dopa). L-Dopa is not a neurotransmiter, but it is a biological precursor that is converted to dopamine by L-dopa decarboxylase.35 L-Dopa can cros the BB and it is the principal therapeutic for the treatment of Parkinson's disease, another neurological disorder considered to have an environmental etiology.36 6 The ability of L-dopa to localize in the brain made it a god starting point for our research. However, it has two properties that are liabilities within the framework of this study: the tris-ligand metal complexes are trianionic and fre (and complexed) L-dopa is readily oxidized in water. The first property makes it more dificult for the metal-ligand complexes to cros  cel membranes. The second property is one shared by al catechols as they can be oxidized to the fre radical o-semiquinone and to o-benzoquinone (Fig. 1.2). The redox sensitivity of the catechol ligands complicated the synthesis of the tris(catecholato) Al and Ga complexes, and we were unable to develop this ligand system with these metals. Figure 1.2. The catechol to o-semiquinone to o-benzoquinone redox cycle. Our research group enjoyed greater suces with two clases of heterocyclic ligands containing the a-hydroxyketone moiety. The binding group is the same as that in o-semiquinone (Fig. 1.3) and a number of transition metal complexes with o-semiquinone ligands have ben isolated.32 These heterocyclic ligands do not have the facile redox properties of the catechols and their tris-ligand metal complexes are neutral at physiological pH. Contemporaneous with my catechol research, other members of our group were synthesizing Al and Ga complexes with several 3-hydroxy-4-pyrones.37 The ligand that proved the most interesting was 3-hydroxy-2-methyl-4-pyrone (maltol), a naturaly ocuring compound that is used as a fod aditive. In animal studies conducted by our colaborators, tris(maltolato)aluminum(II) (Al(ma)3) was found to be 20 times as 7 neurotoxic as Al-lactate, the agent comonly used to induce aluminum encephalopathy.38 The work in our laboratory established the afinity of the oc-hydroxyketone binding group for Al and Ga; however, maltol was not able to redirect 67Ga from transferin in preliminary imaging experiments. o-semiquinone O O" 3-hydroxy-4-pyrone O O* 3-hydroxy-4-pyridinone O O' I H Figure 1.3. Heterocyclic ligands comparable to o-semiquinone. The 3-hydroxy-4-pyridinone* nitrogen heterocycles ofered several advantages over the 3-hydroxy-4-pyrones. The primary advantage was synthetic versatility: the cyclic ether in 4-pyrone is esentialy inert whereas a variety of substituents can be atached to the ring nitrogen in 4-pyridinone. The 3-hydroxy-4-pyridinones are also stronger bases (as indicated by the pKa of the 3-hydroxyl group) and this should result in tris-ligand metal complexes that are more stable than those formed by the 3-hydroxy-4-pyrones. Therefore, my research was diverted from the catechols and was directed toward the synthesis of tris(3-hydroxy-4-pyridinonato) Al and Ga complexes. One of the 3-hydroxy-4-pyridinone ligands employed in this study has ben used in clinical trials as a therapeutic chelator for iron overload diseases.39'40 The ligand was shown to be nontoxic which indicates this clas of nitrogen heterocycles would be suitable for animal biodistribution studies. It was * These compounds are also commonly referred to as 4-pyridones or 3,4-dihydroxypyridines. 8 felt the 3-hydroxy-4-pyridinones were similar enough to maltol to retain the positive atributes (suficient water solubility and lipophilicity) of the Al- and Ga(ma)3 complexes while having the potential to form tris-ligand metal complexes of greater stability and variability. My project involved the synthesis of a homologous series of N-alkyl substituted-3hydroxy-4-pyridinones. This was done to provide bidentate ligands of increasing lipophilicity that could posibly have diferent biodistributions. A potentialy tetradentate ligand was also synthesized by linking together two 4-pyridinones via the ring nitrogens. Once isolated, the ligands were used to synthesize tris(3-hydroxy-4-pyridinonato) Al and Ga complexes that (by design) were al near the uper size limit for unhindered pasage into the brain.* The ligands and metal-ligand complexes were fuly characterized by elemental analysis, mas  spectrometry, and infrared, proton NMR, and ultraviolet spectroscopy. Several studies were then undertaken to determine how wel these compounds fit our criteria for biological relevance. The water solubility was measured and, to indicate the degre of lipophilicity, the octanol-water partition coeficients were determined. The hydrolytic stability of the Al complexes was examined by 27Al NMR and potentiometric titrations were used to determine the thermodynamic stability of the Al and Ga complexes. The proton NMR spectra of the metal complexes indicated an exchange proces was ocuring and this proces was identified by a variable-temperature NMR study. Single crystals of several ligands and metal complexes were grown and the solid state structures were determined by X-ray difraction. A concerted efort was made to corelate the solid state structures to the aqueous solution behavior of these compounds. Particular emphasis * See Table A.1 in the Appendix for a listing of molecular weights. 9 was placed on examining the hydrogen bonding interactions that were a dominant force in the solid state and were also found to persist in solution for both the fre ligands and the tris-ligand metal complexes. Based on the results of this research project, several of the ligands were used in a 67Ga biodistribution study. This study is stil underway and the preliminary findings are briefly discused in Chapter VI. Under conditions of ligand exces, the 3-hydroxy-4-pyridinones can redirect 67Ga from transferin; however, the resulting biodistribution is not significantly diferent from that of Ga-citrate and it is doubtful these ligands would have any aplicability as radiopharmaceutical imaging agents. Because the transferin binding constants for Ga are seven orders of magnitude greater than those for Al, a ligand that can compete for 67Ga has the potential to form stable Al complexes in vivo. The 67Ga biodistribution study does indicate the 3-hydroxy-4-pyridinone ligands have potential as therapeutic Al chelators and this aplication may be pursued in the future. This research project had another goal separate from the examination of the cordination chemistry of group 13 metals and the apraisal of the biological utility of a clas of ligands. This goal was the development of a methodology that would be generaly aplicable for future projects undertaken in this laboratory. We especialy wanted to identify and ases the techniques that would be the most useful in determining the biological potential of a ligand prior to the transition from in vitro to in vivo studies. 10 Chapter II Synthesis and Characterization A . 3-Hydroxy-4-Pyridinone Ligands 2.1 Introduction The first synthesis41 of a 4-pyridinone was reported in 184 and since then a considerable literature has acumulated on these nitrogen heterocycles 42-44 The methods for their synthesis can be grouped into thre categories: ring closure of acyclic compounds, conversion of other heterocyclic ring systems, and substitution and displacement on pyridine or its derivatives. A review of the literature indicated that ring conversion was the simplest method for the synthesis of 3-hydroxy-4-pyridinones. Figure 2.1. Mechanism for the conversion of a 4-pyrone to a 4-pyridinone. 1 One of the oldest ring conversions involves the amonolysis and aminolysis of the cyclic ether 4-pyrone. The acepted mechanism (Fig. 2.1) is nucleophilic atack by a primary amine, folowed by ring opening, los of water, and ring closure to give the coresponding 4-pyridinone.45 There is no direct prof ofered for this mechanism, but molecular orbital calculations predict an enhanced probability of nucleophilic atack ocuring at position 2.46 Further indirect evidence is the efect ring substituents have on the conversion reaction: by induction (when in the 2 position) and resonance interaction, electron-withdrawing groups enhance reactivity while electron-donating groups have the oposite effect44 Although the electron-donating hydroxyl substituent can reduce the efectivenes of the reaction, there are many examples of the conversion of 3-hydroxy-4-pyrones to the coresponding 3-hydroxy-4-pyridinones. The 3-hydroxy-4-pyrones that were considered as posible synthetic precursors in this study are listed below. Ri,R2, R3 = H Pyromeconic acid O R l v J L o H R! = H; R2, R3 = C02H Meconic acid l| 11 Rlf R3 = H; 2 2OH3 Kojicacid *2 ° Rs R1? R2 = H; R3 = CH3 Maltol Figure 2.2. Precursor 3-hydroxy-4-pyrones. The conversion reaction was first used for the structural determination of naturaly ocuring compounds. The structures of maltol47 (in 1906) and kojic acid48 (in 1924) were confirmed by their reactivity with primary amines; conversely, the structures of 4-pyridinones were also verified with the conversion reaction. Meconic acid and its 12 pyrolysis product pyromeconic acid were employed in the structure elucidation of 3-hydroxy-l-alanine-4-pyridinone (mimosine*); this was a dificult problem that provoked some controversy prior to its definitive resolution in 1947 49,50 The synthetic utilization of the conversion reaction folowed from its investigative role: kojic acid and maltol were used in the atempted synthesis of 4-piperidinols,51 pyromeconic acid was the starting material in the first total synthesis of mimosine,52 and a series of N-alkyl substituted-4-pyridinones was made from pyromeconic and meconic acid.53 The conversion reaction with the 3-hydroxy-4-pyrones, however, was sen to be somewhat inconsistent. For example, meconic acid gave the expected 4-pyridinone with methyl-, ethyl-, and propylamine but not with n-butylamine or a-phenylethylamine.53 The reported yields were further evidence of inconsistency: a 40% yield with metiylamine and a 15% yield with ethylamine. The enhanced reactivity imparted by carboxyl groups was shown by the even lower yield (10%) for the reaction of pyromeconic acid with methylamine. An extensive study on the conversion reaction concluded that the more basic and les hindered amines would give the greatest yields of 4-pyridinones.54 However, even the results with amonia were subject to variation: Heyns and Vogelsang found a 30% yield for the amonolysis of kojic acid but others found this reaction to be unproductive.43 It is posible to improve the reactivity of the 3-hydroxy-4-pyrones by blocking the hydroxyl group. The utility of including a blocking step has long ben recognized: in 1906 it was found that maltol and pyromeconic acid would not undergo amonolysis but the 3-methoxy derivatives would react to give the expected 4-pyridinones.47 A typical Earlier authors refer to this compound as leucaenol or leucaenine, a natural product isolated from the seeds of Leucaena glauca in 1937. It was subsequendy shown to be identical to mimosine, first isolated in 1936 from the sap of the tropical shrub Mimosa pudica; since mimosine was coined first, proof of identity made the other names redundant 13 blocking-deblocking sequence used dimethyl sulfate to form the methoxy derivative with deblocking by acid hydrolysis in HI or HBr. The blocking reaction was reasonably facile, but the rigorous conditions for deblocking could be a major experimental impediment. Spenser and Notation atributed the por yield and ireproducibility of the first mimosine synthesis to the six hours of refluxing in aqueous HI required to remove the methyl blocking group. His mimosine synthesis (from pyromeconic acid) used a benzyl blocking group because it could be removed under les strenuous conditions.55 A continued interest in mimosine* led Haris to atempt the improvement of its synthesis; ancilary to this goal, he reported a preparation for the amonolysis of maltol at ambient conditions that also employed the benzyl blocking group.56 His preparation provided the starting point for our synthesis of N-alkyl substituted-3-hydroxy-4-pyridinones. Maltol was chosen as the synthetic precursor despite the electron-donating methyl group in the 2 position. By induction the methyl substituent reduces the eficacy of the conversion reaction, but by the ortho effect57 it reduces the acidity of the hydroxyl proton (pKa of 8.36 for maltol vs. 7.69 for pyromeconic acid and 7.6 for kojic acid58). Maltol is therefore a stronger base and a beter ligand for Lewis acid metal ions. Recently it was found that maltol formed Al complexes of greater stability than did kojic acid.59 Results from our laboratory had shown maltol to be a god ligand for group 13 metals and most importantly, the metal complexes had the desired degre of water solubility and lipophilicity.37 We felt the 2-methyl substituent would play a similar positive role in the chemistry of the 3-hydroxy-4-pyridinones synthesized from maltol thereby compensating for the reduced efectivenes of the conversion reaction. Animals fed the seeds or leaves of Leucaena glauca suffered from hair loss, and mimosine was found to be the active ingredient responsible for this. This depilitory activity was the reason for the renewed interest in its synthesis as it was thought to show promise as a chemical de-fleecing agent to improve wool harvesting. 14 3-Hydroxy-2-methyl-4-pyridinone ligands were synthesized from maltol and the folowing primary amines: methyl-, ethyl- and /i-hexylamine; amonia and 1,6-diaminohexane. The diamine was used to synthesize a bispyridinone that would be a potentialy tetradentate ligand. The monopyridinone ligands were named from the first leters of the substituents, e.g., 3-Hydroxy-2-methyl-l-hexyl-4(para)-pyridinone became Hmhp. The substituents were ordered to emphasize the acidic proton so that the conjugate base of the ligand could be readily identified, e.g., mhp". The bispyridinone H2exn was named from its hexane progenitor again stresing the protons. The structures of the ligands synthesized in this study are in Figure 2.3. 3-Hydroxy-2-methyl-4-pyridinone O Hmp R = H (CH2)5CH3 CH3 C H 2 C H 3 Hdp Hmep Hmhp R l,6-Di(3-hydroxy-2-memyl-4-pyridinone)hexane H2exn Figure 2.3. Ligand structures and abreviations. 15 2.2 Materials and Methods Of the preparations in Method A, that of Hmp was acording to Harris;56 the others were reported previously by us, with ours being the first synthesis of H2exn. 6 0 Hmp  has ben synthesized by other methods.61 The synthesis of Hdp  from maltol62-63 and of Hmhp  from maltol D-glucoside64 had ben reported prior to our work. Further experimentation was undertaken to simplify Method A by eliminating the blocking step and the results are reported as Method B. The precedents and rationale for this aditional synthetic efort wil be presented in Section 2.3. All chemicals were reagent grade or beter and were used without further purification. The progres of the reactions was monitored by thin layer chromatography (TLC) on silica gel plates with 5% methanol in CH2CI2 as the eluting solvent. The starting material 3-benzyloxy-2-methyl-4-pyrone (Bzma) was made from the comercialy available maltol (Aldrich) by the method of Haris (> 95% yield).56 Bzma is a viscous yelow liguid that was used without further purification (TLC pure) The melting points were measured with a Mel-Temp aparatus and are uncorected. Unles stated otherwise, the quoted yields were for purified compound and they were calculated from maltol. The characterization data, including the elemental analyses, are found in Section 2.4. Method A 2.2.1 3-Benzvloxv-2-methvl-4(lHVpyridinone. Bzmp. This intermediate was prepared by the treatment of Bzma with a solution of amonia in ethanol acording to the method of Harris.56 Crystalization from hot ethanol gave 4.70 g (60% yield) of pale yelow crystals. Mp 16 °C. 16 2.  3-Hvdroxy-2methvl-4(lH)-pvridinone. Hmp. Bzmp was deblocked by acid hydrolysis with 40% HBr in acetic acid.56 The crude product was redisolved in water and N H 4 O H was aded to adjust to pH 8. The solution was coled overnight at 10 °C and gave 2.63 g (47% yield) of pink crystals. Mp 265 °C dec. 2.2.3 3-Hvdroxy-1.2-dimethvl-4-pvridinone. Hdp. A solution of Bzma (10.0 g, 46.0 mol) in 50 mL ethanol and 10 mL 40% methylamine in water (130 mol) was placed in a sealed flask at 20 °C for 72 hours. The exces amine was removed in vacuo and the oily residue taken up in 100 mL water. The organic phase was extracted into CH2CI2 (2 x 100 mL) and evaporation gave a yelow oil. The benzyl ether (Bzdp) was disolved in 50 mL THF and hydrogenated under ambient conditions over 5% Pd/C catalyst (1 g) until hydrogen uptake ceased. The solution was filtered and the solid extracted with boiling water (3 x 100 mL). The aqueous solution was concentrated until crystalization comenced and then chiled at 10 °C overnight. Recrystalization from hot methanol gave 3.46 g (52% yield) of white crystals. Mp 260 "C dec. 2.2.4 3-Hvdroxv-2methvl-1 -hexvl-4-pyridinone. Hmhp. Bzma (7.10 g, 32.8 mol) and hexylamine (16.6 g, 164 mol) were disolved in 10 mL methanol and placed in a sealed flask at 20 °C for 72 hours. The solution was concentrated in vacuo and 10 mL 12M HC1 was aded to the oily residue. After heating on a steambath for 1 hour, the exces acid was removed in vacuo. To the residue was aded 10 mL water and the pH adjusted to 8 with 2 M NaOH at which time a precipitate formed. Sublimation of this beige solid gave 3.40 g (48% yield) of a yelow powder. Mp 123 °C. 2.2.5 1.6-Di(3-hydroxy-2methyl-4-pyridinone)hexane. H2exn. Bzma (3.0 g, 14.0 mol) and 1,6-diaminohexane (8.0 g, 69.0 mol) were stired in 25 mL methanol for 72 hours at 20 °C. Concentration in vacuo gave an oil that was treated with 50 mL concentrated HC1 and heated for 30 minutes on a steambath. The acid was removed in 17 vacuo and the residue was disolved in 50 mL water. The pH was adjusted to 8 with 2 M KOH and the resulting precipitate was colected. The filtrate was extracted with CHCI3; evaporation left a solid residue that was aded to the earlier precipitate to give 0.43 g (18% yield) of a yelow powder. Sublimation gave analyticaly pure sample. Mp 295 °C dec. Method B The reactions were done under positive N2 presure. The flasks were fited with condensing columns but the reactions were only heated to 50 °C. Before extraction or aqueous recrystalization, the pH was adjusted to take advantage of the diference in acidity betwen the 3-hydroxy-4-pyridinones and maltol (pKa of -9.8 and 8.4, respectively). Purification at pH 8 ensured disproportionately greater ionization of the unreacted starting material thus aiding in the removal of the primary contaminant in these reactions. The solution pH was measured with a Fisher Acumet model 805 pH meter. 2.2.6 3-Hydroxv-2-methvl-4-(lH)pyridinone. Hmp. To a chiled solution of maltol (2.53 g, 20.1 mol) in 30 mL of water was aded 4 mL concentrated NH4OH (60 mol) folowed by the slow adition of 5 mL 6 N HC1. The pH was adjusted to 9.3 and the solution was heated for 36 hours. Concentration in vacuo and coling at 10 °C for 24 hours gave 1.42 g of crystaline product. Recrystalization from hot methanol gave 1.20 g (48% yield). 2.2.7 3-Hvdroxv-1.2-dimethvl-4-pvridinone. Hdp. Maltol (5.05 g, 40.3 mol) in 75 mL water was placed in a 3-necked flask to which was conected an adition funel with 15 mL 40% methylamine in water (193 mol) and 50 mL water. The solution was heated for 12 hours with the amine being slowly aded during the first 4 hours. The solution was concentrated in vacuo and coling at 10 °C overnight gave 2.80 g of a yelow microcrystaline solid. The filtrate was extracted with 150 mL CH2CI2 for 12 hours in a liquid-liquid extractor. Evaporation of the solvent left a residue that was washed with 18 acetone and aded to the filtrant. Purification by sublimation gave 3.87 g (70% yield) of a white powder. 2.2.8 3-Hvdroxv-2-methvl-1 -ethy 1-4-pvridinone. Hmep. A solution of maltol (2.52 g, 20.0 mol), 70% ethylamine in water (13.3 g, 206 mol) and 40 mL of water was coled in an ice bath. 30 mL of 6 N HC1 was slowly aded and the pH adjusted to 9.8. The solution was heated for 24 hours, transfered to a continuous liquid-liquid extractor, and the product was extracted into 100 mL CH2CI2 for 6 hours. Evaporation of the solvent left a brown solid that was washed with cold acetone. Sublimation gave 1.80 g (58% yield) of a white powder. Mp 205 °C dec. 2.3 Discusion of the Synthetic Procedure Initialy, the amonolysis of maltol was atempted under the same conditions as reported herein for its benzyl ether derivative Bzma. The resulting failure to observe any conversion to Hmp  confirmed the utility of employing a blocking group. Method A (Fig.2.4) can be divided into thre steps: blocking of the ring hydroxy! group, the ring conversion reaction, and deblocking to give the 3-hydroxy-4-pyridinone. The blocking reaction is a Wiliamson ether synthesis with maltol going to Bzma (an oil) in high yield. O O o CH3 OH R R Figure 2.4. The schematic representation of Method A. 19 The conversion reaction was done at ambient temperature and the maximum yield was obtained after thre days. A similar procedure alowed seven days but the yield for Hdp was the same as in Method A (~50%).63 Atempts were made to sped up the reaction by refluxing, but product isolation was complicated by the resulting tary byproducts and yields were significantly lower. The reaction was done in alcohol/water mixtures because of the hydrophobicity of Bzma and this may have retarded the reaction; an earlier study had concluded that the conversion reaction was more eficient in water than in alcohol.51 In the synthesis of Hmp, the 4-pyridinone benzyl ether (Bzmp) was isolated as an analyticaly pure solid when triturated by acetone and recrystalized from ethanol. The other benzyloxy intermediates only formed as oils (probably due to the lipophilic N-alkyl substituents) and the preparations proceded direcdy to the deblocking step. For Hmp, deblocking was done by catalytic hydrogenation or by acid hydrolysis in 40% HBr in acetic acid, with the later giving the greater yield. The results were reversed with Hdp, and hydrogenation gave beter results. Acid hydrolysis with 6 N HC1 was used for Hmhp, and H2exn required concentrated HC1 to remove the benzyl blocking group. Restricted aces to the hydrogenation aparatus and the ned for more rigorous deblocking left acid hydrolysis as the method most often employed. The conversion reaction was done with a five-fold exces of amine, and the unreacted amine was removed by rotary evaporation. For 1,2-diaminohexane this was insuficient and acid hydrolysis gave the dihydrogenchloride salt of the precursor amine as a byproduct. Separation of the product was dificult and undoubtedly contributed to the porer yield for H2exn compared to that for Hmhp  (18 vs. 48%). The problem of salt formation was a complicating factor in al the preparations albeit to a leser extent. The least desirable aspect of Method A was the time required as the thre steps tok up to five days to complete. A posibility for shortening the procedure would be to heat the reaction in a sealed glas tube: maltol and its 3-methoxy derivative gave 35 and 69% yields 20 respectively when heated with aniline in sealed tubes (40 hours at 150 °C).65 Our eforts with sealed vesels were unsucesful as both maltol and Bzma failed to react with either amonia or n-hexylamine. The decision to forego further experimentation with this technique was facilitated by the physical restrictions of the glas tubes; this hmitation led to recent atempts to circumvent sealed tube reactions in a study with kojic acid and arylamines.66 Another way to sped up the procedure would be the elimination of the blocking step. This step was included to avoid deprotonation of the ring hydroxyl group as the increased electron-donating capacity of the O" anion would impede formation of the 4-pyrone resonance hybrid most susceptible to nucleophilic atack. Instead of replacing the proton, the ionic state of maltol could be controled by bufering the reaction at a pH < the pKa of the hydroxyl proton. Hdp was obtained in a 55% yield from a procedure employing a methylamonium acetate bufer. The reaction was done at slightly acidic, neutral, and slightly basic pH with the later giving the best results.62 The pH was not specified but repeating the preparation gave an initial pH of 9; however, we were unable to achieve reasonable yields for Hdp  or Hmep  with this procedure. It was felt the problem lay with the acetate bufer (and with the chromatographic separation step this necesitated) and the results encouraged further experimentation with bufered systems. Recently the conversion reaction was done without blocking or buffering.67 This simple preparation gave a 50% yield for the reaction of maltol with methylamine, but the results were porer with ethyl- and n-propylamine (24% and 21% yield, respectively) showing once more the inconsistency of the conversion reaction. This limitation made it worthwhile to continue to lok for a procedure that would have wider aplicability than the above preparations and that would be faster than Method A. But "method" is the wrong word; rather what is reported herein as Method B is simply an experimental strategy that recognizes the primary importance of the ionic state of the reactants (and consequently the solution pH) to the suces of the conversion reaction. 21 The utility of this strategy can be demonstrated by examining the preparation of Hdp  without blocking or buffering.67 The ratio of reactants was 3:1 methylamine to maltol; given the concentrations used, the initial pH was 1.5 and maltol would be 9.6% ionized. At the start of the reaction there was esentialy no unionized or reactive maltol present. As the reaction was heated, the concentration of the volatile methylamine was reduced and after the prescribed 5 hours of refluxing the pH was 8. At this pH, methylamine would be 99% ionized and unreactive. Somewhere in betwen these two extremes there is a pH region where the nucleophilic atack mechanism is operable but this preparation does not control the time spent at this optimal pH. As wel, the hydroxide ion is a nucleophile able to cleave 4-pyrones68 so the avoidance of side reactions is another reason for controling the solution pH. A simple way to control the pH without bufering would be the slow adition of base and this was done in the preparation of Hdp  as reported in Method B. Methylamine was aded to an aqueous maltol solution over 4 hours and because of the volatility of the amine (Bp -6.3 °C), the reaction was heated at 50 °C overnight rather than refluxed. This simple preparation gave higher yields (70 vs. 52%) and was faster than the preparation in Method A. It was a cleaner preparation than the other syntheses of Hdp  (including the repeats of literature preparations) and gave microcrystaline precipitate directly from the reaction mixture. The reaction with amonia provided a beter trial for this strategy since every amonolysis of 3-hydroxy-4-pyrones of which we are aware employed a blocking group, a sealed reaction vesel, or both. This is probably due to the volatility of amonia (Bp -3.4 °C) and its reduced nucleophilicity compared to alkylamines. In retrospect, our choice of amonolysis rather than aminolysis to explore the necesity of employing a blocking group was an unfortunate one. Consistent with the dificulty of this reaction, atempts to prepare Hmp  simply by the slow adition of amonia ended in failure; beter 2  results were obtained with an amonium chloride bufer. Bufering at pH 7.5 and 8.5 produced no reaction (monitored by TLC) but a significant decrease in the amount of maltol ocured upon overnight refluxing at pH 9.3. These results coupled with the failure of the unbufered reaction (pH > 10) indicated a narow range of optimum reactivity. For the Hmp  preparation in Method B, the solution was bufered at pH 9.3 with initial concentrations such that the ratio of unionized reactants would be 15:1 amonia to maltol. This was a one pot synthesis that gave the same yield (48%) as in Method A but was again a simpler and faster preparation. The N-ethyl derivative Hmep was prepared in an analogous maner to Hmp. The bufered preparation gave a higher yield (58%) than either Method A (35%) or the unbufered preparation cited previously (24%).67 By controling the ionic state of the reactants it is posible to convert maltol to its 3-hydroxy-4-pyridinone analogues without using a blocking group. The yields are the same or beter and the preparations are simpler and faster for Method B when compared to Method A. The routine synthesis of 3-hydroxy-4-pyridinones in our laboratory is now done without a blocking group, as are the exploratory eforts to synthesize new pyridinone ligands. 2.4 Characterization of the 3-Hvdroxy-4-Fyridinone Ligands The ligands were characterized by elemental analysis, infrared (IR) and proton NMR spectroscopy, and electron impact mas spectrometry (EI-MS). The characterization data were completely consistent with the structures as shown in Figure 2.3. Particular atention was paid to the hydrogen bonding (H-bonding) in these compounds, the IR spectra indicated that the compounds were H-bonded polymers in the solid state and that the intermolecular H-bonds involved the hydroxyl and the carbonyl moieties. The IR 23 results for Hmp  showed the cyclic secondary amine was also a H-bond donor, a proton NMR experiment indicated the H-bonding persisted in solution. The elemental analyses (C, H, N) were performed by Mr. Peter Borda of the Microanalytical Laboratory of this Department. The IR spectra were recorded with a Perkin Elmer PE 783 in the range 400-20 cm-1. All samples were prepared as KBr disks and spectra were referenced to polystyrene film. The proton NMR spectra were recorded with a Bruker WP-80 and a WP-40  instrument. The 80 MHz instrument was run by the author, and spectra at 40 MHz were suplied by the U.B.C. NMR service. The EI-MS was performed on a Kratos MS 50 spectrometer and al mas  spectra were suplied by the U.B.C. mas spectrometry service. 2.4.1 Elemental Analysis Prior to submision for analysis, all samples were purified either by recrystalization or by sublimation in vacuo (0.03 tor) with heating. The temperatures for sublimation were relatively mild (<140 °C) except with H2exn. This compound was to insoluble for recrystalization and the analyzed sample was sublimed at > 285 "C using a Wod's metal bath. Table 2.1. Results of the ligand elemental analyses (Found/[Calculated]). %G 57.58 [57.57] 60.45 [60.41] 62.56 [62.72] 68.56 [68.87] 64.81 [65.03] 5.70 [5.65] 6.49 [6.53] 7.28 [7.25] 9.01 [9.15] 7.19 [7.29] %N 11.23 [11.24] 10.00 [10.07] 8.98 [9.14] 6.61 [6.69] 8.40 [8.43] 24 2.4.2 Infrared Spectroscopy Of the spectroscopic techniques employed in the study of these compounds, IR is the most useful as it afords a facile confirmation of the outcome of the conversion reaction. The ring skeleton vibrational modes of the pyridinones and the pyrones are related to those found in benzene.42 This typicaly results in four ring-stretching modes betwen 1650 and 140 cm"1, and the patern changes predictably when the cyclic oxygen is replaced by nitrogen. An examination of the IR spectrum readily confirms the transformation of maltol to the coresponding 3-hydroxy-4-pyridinone. In the 1960's there were a number of IR spectroscopic studies on the 4-pyridinones motivated primarily by the dificulty of asigning the carbonyl stretching mode in these compounds.69"71 Although none of these studies included any 3-hydroxy derivatives, it was posible to make tentative spectral asignments based on this work. The asignments in Tables 2.2 and 2.3 were suported by reference to Belamy's review of structural corelations in the infrared.72 Hmp  is of special interest because of the aditional H-bond donor in this compound; the asignments for the modes in Hmp  afected by H-bonding are included with those of the benzyl ether intermediate (Bzmp), and the dideuterated analogue, d2-Hmp  in Table 2.2. Deuteration was by repeated crystalization from D2O and was undertaken to determine the relative strength of the H-bonds. The spectral asignments for the other compounds are in Table 2.3. The IR spectrum of Hmp  from 170  to 30 cm-1 is reproduced as Figure 2.5. The abreviations used for the vibrational modes are: V, stretching; 8, in-plane bend; n, out-of-plane bend; as, asymetric; sy, symetric. The asignment of VOH in Hmp  is confirmed by Bzmp where the 3270 cm-1 band disapears and by d2-mp, where it shifts to 2420 cm"1. This gives a V Q H A ' O D of 25 1.35, indicative of a relatively weak H-bond. The theoretical ratio is 1.375 (from the reduced mases) and for fre OH it is 1.355.73 This ratio fals systematicaly as the strength of the H-bond increases and can reach unity for very short H-bonds.74 A combination of a lower zero point energy for H than D, and a weaker O—D—O than O—H—O interaction explains this low ratio.74 In the N-alkyl derivatives the VOH drops below 320 cm"1 which sugests stronger H-bonding than in Hmp; this is supported by the lower Vc=o m these compounds, e.g., 1645 cm-1 in Hmp  and 1630 cm"1 in Hdp. In al cases the bands are broad and within the range for normal polymeric intermolecular H-bonding.72 The asignment of V N H is confirmed by Bzmp, d2-Hmp, and Hdp: the band is at 280 cm"1 in Hmp, shifts to 2650 cm"1 on benzylation, shifts to 210-2150 cm"1 (split, mean = 2180 cm-1) on deuteration, and finaly disapears on methylation. The ratio (VNH/VND) is 1-28 and spliting of V N D is not without precedent, as it is reported in deuterio-2-pyridinone.75 The authors list a 310  cnr1 band (in KBr) for the protio compound which suports our asignment, and also state that the 285 cm-1 VNH band for 2-thiapyridinone is the strongest NH hydrogen bonding yet recorded for other than zwiterionic structures. The deuteration ratio for the thiapyridinone is 1.29, and the typical N—H—S stretch apears at 3160 cm-1. It is imposible to separate the Vc=o and the higher energy Vring stretches in any of the compounds. The bands are extensively coupled and there is no mode which is localized solely in the carbonyl bond.76 The highest wave number band has the most C=0 character and the relatively low energy of this band (below 1650 cm-1 in al spectra) indicates it is acting as a H-bond aceptor. In the SCH region, weak to medium intensity bands are observed with aditional bands for the compounds with N-substitution consistent with the presence of aditional methyl and methylene groups. There is extensive mechanical coupling of the 6OH and Vco modes in the region 1325 to 1235 cm-1 26 so that an unambiguous assignment is not possible . 7 4 ' 7 7 The 8Ri ng and T C R j n g assignments correlate well with the literature and the sharp out-of-plane bending mode (ca. 820 cm - 1) is the most distinctive feature of the lower energy region in all of the spectra. Table 2.2. Infrared bands for Hmpp and substituted analogues (cm - 1). Assignment Hmpp d2-Hmpp Bzmpp V O H (D ) 3270 b 2420 V N H ( D ) 2800 b 2180 2650 b vc=o 1645 1630 1630 sh and 1620 1550 1620 VRing 1540 1540 sh 1535 1500 1490 1500 1420 w 1420 w 1410 VcOand 1300,1270 1325 1190(Vcoc) 50H(D) 1245 900 b 8Ring 1225 1245 sp 1215 sp 1110 1150 1110 1045 1090 1050 27 Table 2.3. Selected infrared absorption bands (cm-1). All bands are strong except where noted.3 Asignment Hmp Hdp Hmep Hmhp H2exn V O H (b) 3270 3150 3180 3195 3195 VCHring (w) 310 3010 3040 3060 3060 H 3 ( 2 ) 2920 w 2940 w 2980 w 2960,2935 2960 2860 2890 vc=o 1645 1630 1630 1630 1630 and 1620 1565 1575 1580 1580 VRing 1540 1530 1530 1530 1530 150 1515 1510 1510 1505 1420 w 140 w 1405 w 1405 w 140 m 8asCH 1450 w 1460 1450 w 1465 w 1460 w 8syCH 1380 1380 m 1365 1380 w 1380 w 135 1350 1360 m Vcoand 130, 1270 1280 1260 1270 sh 130,1280 80H 1245 1250 b 1230 1240 b 1235 b SRing (w) 125 1230 1250 120 120 110 125 135 170 170 1045 105 1040 130 1040 1030 1040 TCRing 830 820 830 850 850 m, medium absorption; w, weak absorption; sp, split band; b, broad band; sh, shoulder. 28 Figure 2.5. Infrared spectrum of Hmp  from 170 to 30 cm"1 (KBr disk). 29 2.4.3 Proton NMR Spectroscopy The spectra of Hmpp, Hdpp, and Hmepp were recorded in D2O; the lipophilic Hmhpp and the insoluble H2exn required 20% CD3COOD/D2O. The acidic NH and OH protons were detectable if the samples were rigorously dried and the spectra were recorded in aprotic solvents: (CD3)2SO for Hmpp and Hdpp, and CDCI3 for Hmepp and Hmhpp. H2exn was only soluble in acidic solution so it was not possible to detect the hydroxyl proton signals. The spectra in D 20 were referenced to an external (CH3)3Si(CH2)3S03Na (DSS) signal and the spectra in aprotic solvents were internally referenced to the solvent signal. The chemical shifts listed in Table 2.4 were recorded at 40 MHz except those of the OH protons that were at 80 MHz. The spectra have AB doublets for the ring protons (Ja>b of 7 to 8 Hz), a singlet for the ring methyl group, and a series of signals for the methyl and methylene protons of the R substituent. The signal from the protons on the carbon direcdy attached to the ring nitrogen is shifted downfield from that of the ring methyl group. Deshielding by the electronegative nitrogen is the reason for this and for assigning the lowfield doublet to H a. 7 8 For Hdpp and Hmepp, the signals from the alkyl protons are easily assigned. With the exception of the CH2d,e methylene protons in Hmhpp, we did not attempt to resolve the second order spectra of the alkyl protons in Hmhpp or H2exn. In all cases, peak integrations are consistent with the assignments as given in Table 2.4. In the solid state, Hmpp forms intermolecular H-bonds and this interaction persists in solution as evidenced by the spectra recorded in (CD3)2SO. As the sample concentration is increased, the OH resonance shifts downfield from 5.64 ppm at 0.03 M to 6.90 ppm at 0.09 M. The direction of the shift is indicative of H-bonding and a dependence on concentration is typical of intermolecular H-bonding. The OH chemical shift is also 30 afected by the concentration of water in the solvent. The spectrum of an Hmp  sample (0.09 M) prepared in a drybox under N2 is reproduced as Figure 2.6 and the OH resonance is asigned to the broad peak at 6.90 pm; leaving the NMR tube under ambient atmosphere for several days caused the OH resonance to shift to 6.32 pm. Similar results were obtained with a 0.03 M solution of Hmp  and with Hdp samples. This shift is probably due to the difusion of atmospheric water into the previously dry solvent and the upfield direction of the shift sugests the intermolecular H-bonds are stronger than those betwen the 3-hydroxy-4-pyridinone and water. The NH resonance in Hmp  is asigned to the broad signal at 1.35 pm, based on a peak in Bzmp  at 1.30 pm (both in (CD3)2SO at 80 MHz). The adition of a drop of D2O to NMR samples of Hmp  results in the los of both acidic proton signals and the los of the second order coupling to Ha which apears as a broad, les wel  resolved doublet only with this compound (se Fig. 2.6). The presence of an NH chemical shift of 12.05 pm in 2-pyridinone has ben taken to sugest that the compound exists as H-bonded dimers in solution79 just as it does in the solid state.80 Our results for Hmp  also indicate the existence of intermolecular H-bonded arays in solution with both the NH and OH acting as proton donors. The IR spectra and the crystalographic results (se Chapter II) clearly show that this is the case in the solid state. 31 Table 2.4. Proton NMR chemical shift (8) data for the free ligands at 400 MHz (ppm). o R= H b H ^ J L / O H C H , X x H N " C H 3 r a | 3 C R l3d C H . C H 2d 3e C R . C H C H . C H C H U C H . 2d 2e 2f 2g 2h 3 1 H 3 C C \ C H 2 d C H 2 e C H 2 f C H 2 g C H 2 h C H 2 i - N Assignment Hmpp Hdpp Hmepp Hmhpp H 2exn** O H 6.90 6.35 6.43 5.96 Ha(d) 7.58 7.58 7.56 7.57 7.55 H b (d) 6.51 6.46 6.43 6.71 6.69 C H 3 c (s) 2.34 2.36 2.33 2.16 2.13 C H 2 d *3.73 (s) 4.00 (q) 3.87 (t) x3.87 (t) C H 2 e *1.26 (t) 1.37 (tt) yi.37 (m) CH 2 f ,g ,h 0.85 (m) 20.93 (m) C H 3 i 0.38 (m) abbreviations: s = singlet; d = doublet; q = quartet; t = triplet; tt = triplet of triplets; m = multiplet. * ReadCH3. ** For H 2exn, the two rings are apparently equivalent, e.g. read: H a > a « with integral equal to two protons. x Read C H ^ i with integral equal to four protons, y R e a d C H 2 e > h z Read C H 2 f , g " 32 9 03 O n rt Figure 2.6. Proton NMR spectrum of Hmpp in (CD3)2SO at 80 MHz. 3  2.4.4 Electron Impact Mas  Spectrometry The molecular ions, base peaks, and selected fragment ions for the compounds are listed in Table 2.5 and the mas spectrum of F^exn is reproduced as Figure 2.7. Some asignments were made by comparison with the literature on 2-pyridinones because of the scant work available on the 4-pyridinones.81 The ring fragments exclusively by the los of CO (M-28) and HCO* (M-29) and not by the los of HCN (M-27) or CH3CN (M-41), as is comon in the 2-pyridinones. 3-Hydroxy-2-pyridinone82 fragments with the los of H2O (M-18) and this did not ocur in any of the 3-hydroxy-4-pyridinones. The molecular radical cation (M*) is present in al the spectra and this is probably a result of the ability of the ring nitrogen to localize positive charge. In the spectrum of Hmp, the M* at m/z 125 is the base peak and m/z 97 and 96 are due to the los of CO and of HCO respectively. The m/z 125 ocurs in four of the five spectra and is the base peak in thre; the prominence of the ring radical cation is due to the stability of the dihydroxypyridinium moiety which is known to be favored in the gas phase.83 Los  of CO and HCO* results in the m/z 11 and 10 peaks in Hdp and the m/z 125 and 124 peaks in Hmep. Hmep  can also fragment by cleavage of the exocyclic N-C bond with hydrogen migration to give the ring molecular cation at m/z 125; los of CO and HCO* from this ion is the best explanation for the prominent peaks at m/z 97 and 96 (base peak in this spectrum). This fragmentation mode is not operable with the N-methyl group so the Hdp spectrum was without a m/z 125 peak. The los of carbonyl fragments from the M* is not observed in the spectra of either Hmhp  or H2exn. The N-hexyl group in Hmhp  fragments by sucesive los of methylene groups and again (by hydrogen migration) the ring radical cation is the base peak. The bispyridinone H2exn fragments through los of one ring to give a m/z 208 peak 34 followed by the same methylene cascade as in Hmhpp. The base peak is m/z 125 and both spectra have peaks at m/z 97 and 96. The least ambiguous spectroscopic evidence for the structure of H2exn is provided by its mass spectrum as the IR and proton N M R spectra afford little distinction between the mono- and bispyridinone. Table 2.5. Mass spectral data (m/z) with the percent relative intensity in parenthesis. Ligand M * M O O M-HCO Fragment Peaks Hmpp 125* 97 (9) 96(47) Hdpp 139 (98) 110* 111 (25) Hmepp 153 (74) 125 (91) 124 (28) 97 (58) 96* Hmhpp 209 (37) 125* 97 (14) 96 (20) H2exn 332 (15) 208 (64) 125* 97 (10) 96(31) * Base peak (relative intensity = 100%). 36 B. Tris(3-Hydroxy-4Pyridinonato) Metal Complexes 2.5 Introduction With the exception of H2exn, the conjugate bases of the 3-hydroxy-4-pyridinones are anionic bidentate ligands that can chelate a metal ion via the deprotonated 3-hydroxyl and the carbonyl oxygens. Each ligand would form a five-membered chelate ring with the metal center and the tris-ligand metal complexes would be of neutral charge. We refer to H2exn as a "potentialy" tetradentate ligand because it contains two bidentate moieties; however, the position of the oxygen atoms on the ring and the length of the hexyl bridge make it virtualy imposible for both rings to chelate one metal center. H2exn could chelate two metals to give a M2L3 dimer as was the case for a structuraly similar diprotic tetradentate ligand containing two l-hydroxy-2-pyridinone rings.84 Ligands with oxygen donor atoms are usualy hard bases. Trivalent Al and Ga easily fit the definition of hard acids as aceptor atoms of high positive charge and smal size. Therefore, the reaction of the 3-hydroxy-4-pyridinones with Al and Ga should conform to the Hard and Soft Acid Base (HSAB) principle, i.e., hard acids prefer to R CH3 Figure 2.8. Five-membered chelate ring. 37 cordinate to hard bases.85 The HSAB principle is a generalization based on experimental facts; however, a theoretical derivation has ben developed for this venerable canon of inorganic chemistry.86 The derivation is based in part on the concept of absolute hardnes (I - A) (JI) which has an operational (and aproximate) definition of T| =-^—^—- where I is ionization potential and A is electron afinity. This work has also produced a chemical definition of hardnes as "the resistance of the chemical potential to change in the number of electrons". Using the TJ values, it is posible to rank metal cations by their relative hardnes. To demonstrate the utility of this, the rj values and the charge to efective ionic radius (C/IR) ratios (based on data from ref. 30) are compared for several six cordinate trivalent metal ions in Table 2.6. (The left hand column is included for perspective: B3+ has the highest T| value, and Ca2+ is a hard acid, Ag1+ is a soft acid, and Cu2+ is clased as borderline in the earlier clasification system.85) The C/IR ratio has ben used to indicate the relative hardnes of metal ions and the T| values alow a greater discrimination of this property. The first row transition metals (al high spin) have similar C/IR ratios but Fe3+ is significantly harder than its neighbors, as expected for its d5 electron configuration and as indicated by the rj values. By either the TJ values or the charge to ionic radius ratios, the order of hardnes for the group 13 metals is Al» Ga > In. Table 2.6. Comparison of r\ values86 (eV) and C/IR ratio (A-1) for selected trivalent metal ions. C/IR Tl C/IR Tl B3+ 11 Cr3+ 4.8 9.1 Al3+ 5.61 45.8 Ca2+ 19.5 Mn 3+ 4.65 8.8 Ga 3+ 4.84 17 Cu2+ 8.27 Fe3+ 4.65 12.1 In 3+ 3.75 13 Ag 1+ 6.96 Co 3+ 4.92 8.9 3.34 10.4 38 Another measure of the relative acidity of a metal ion is the first disociation constant of the hexaquo species defined as: [M(H20)6]3+ [M(H20)5OH]2+ + H + pKa Based on pKa values of 2.6, 4.0, and 5.0,25 the order of acid strength is Ga > In > Al. The hardnes of Al is exagerated by both of the scales in Table 2.6 and the distortion is greatest in the absolute hardnes value. Aluminum's large T] value is due to its very large fourth ionization potential which requires the removal of an electron from an inert gas (Ne) core. The ionization potentials of Ga3+ and In3+ are based on the removal of a 3d and a 4d electron, respectively. Not surprisingly, the ionization potential for Al3+ is almost twice that of Ga3+ or In3+ (1,563 vs. 6150 and 571 kJ mol"1).87 The question of the chemical relevance of the fourth ionization potential of Al was raised by the developers of the absolute hardnes ranking,86 and it is clear that the T) values do not acurately predict the relative acidity of the group 13 metals. This is in contrast to the situation with the first row transition metals where the ionization potentials are due to the removal of valence electrons from the same shel. The reason for the order of the hexaquo pKa values can be deduced from the third ionization potential of these metals. Instead of decreasing on going from Al to Ga it increases from 274  to 2962 kJ mol"1, and the value for In is only slightly lower than that of Al (2704 kJ mol"1). The reason given for this is d-block contraction in atomic size and a higher efective nuclear charge for Ga and, to a leser degre, for indium.88 The electrons of the filed d shels do not completely shield the 10 aded positive charges on the nucleus. When this is contrasted to the shielding aforded by the Ne core of Al, the reduced acidity of Al3+ (or the enhanced acidity of Ga3+ and In3+) is quite understandable. 39 Regardless of the scale that is used, however, the group 13 metals are hard Lewis acids and any aqueous synthetic chemistry with A l or Ga requires the consideration of their pH dependent speciation. The complexity of their aqueous behavior can be attributed to the affinity of the hard base water (r\ = 9.5) for these hard metal ions. The following equations stress the pH dependence of hydroxide formation.8 9 Al(OH) 3 + 3 H+ " A l 3+ + 3 H 2 0 Kso, = ^ = 1 0 . 0 , [ A 1 ( H 2 0 ) 6 ] 3 + = 10 10.7 x IQ - 3pH From the final equation, 3 X 10" 1 2 M is the highest concentration of free A l allowed by amorphous Al(OH)3 at pH 7.4. The practical consequence is that the typical synthetic concentrations (mM) are supersaturated in Al(OH)3 and a similar situation pertains with Ga. Unless a ligand has a high affinity for these metals, a gelatinous precipitate will form. Work in our laboratory with 3-hydroxy-4-pyrones had indicated that the a-hydroxyketone group was a strong enough binder of A l and Gato preclude hydroxide formation at neutral pH and this was confirmed by the large overall formation constant P3 reported for Al(ma)3. 5 9 (For a definition of P3 refer to Section 5.7.) The 3-hydroxy-4-pyrones are stronger acids than their pyridinone analogues because of the heterocyclic oxygen's greater capacity (compared to that of nitrogen) for stabilizing the negative charge of the deprotonated hydroxyl anion. The 3-hydroxy-4-pyridinone ligands used in this study have a hydroxyl p K a ~1 pH unit higher than maltol. This increased basicity ensured there would be no hydroxide formation in the synthesis of the tris-ligand A l and Ga complexes. 40 There have ben a number of aqueous solution studies of both of these metals with bidentate oxygen containing ligands, e.g., Al with salicylate ions90 and with hydroxy carboxylic acids91 and Ga with hydroxyaromatic ligands.92 Recent interest in the biological and environmental roles of Al produced several studies on its aqueous speciation with the potentialy tridentate citric acid.93'94 Potentiometric studies indicate what ligands are god binders of these metals but do not aford any synthetic precedent. The field of Fe cordination chemistry is probably the best place to lok for relevant synthetic studies as high spin Fe3+ is close in size and electronic properties to Ga3+ (se Table 2.6). The most pertinent work was Raymond's synthesis and structural characterization of Fe3+ complexes with l-hydroxy-2- and 3-hydroxy-2-pyridinones.95 Raymond has also reported the synthesis of Ga complexes with catecholate and benzohydroxamate ligands.34 Cordination chemistry with the 3-hydroxy-4-pyridinone ligands provides much les fertile ground as regards synthetic antecedents. In the mid 70's, N-aryl derivatives were touted as extractants for the separation of 67Ga and ^Zn radionuclides.96 At this same time the stability constants of mimosine (and several related compounds) with a number of divalent metal ions were determined.97'98 Another study99 reported a log 03 of 29.2 for tris(mimosinato)aluminum(II), although their methodology was questioned.97 Several N-substituted-3-hydroxy-4-pyridinones have ben evaluated as chelating agents for the treatment of iron overload diseases and a large formation constant (log 03 = 34.5) was determined for Fe(dpp)3.100 This result was suported by the log P3 of 35.1 found for the Fe3+ complex with 3,4-dihydroxypyridine.95 Based on our results with the 3-hydroxy-4-pyrones and the above literature, the synthesis of Al and Ga complexes was atempted with the 3-hydroxy-4-pyridinone ligands prepared in our laboratory and with L-mimosine (abreviated as mimo). The later was included to ascertain what efect the amino acid substituent would have on the synthesis of the tris-ligand metal complexes. It was also felt the M(mimo)3 complexes could have an 41 interesting biodistribution as mimosine is the 4-pyridinone analogue of the catechol L-dopa, a compound that readily enters the brain (se Chapter I). Figure 2.9 is a diagramatic representation of the tris(3-hydroxy-4-pyridinonato) metal complexes that were sucesfuly isolated and characterized in this study. M(mp)3, M(dp)3, and M(mhp)3 have ben reported by this laboratory101 and this is the first report of the synthesis of M(mep)3 and Ga(mimo)3 of which we are aware. R i = CH, R = H M N J3 CH3 CH2CH3 (CH2)5CH3 M(mpp)3 M(dpp)3 M(mep)3 M(mhp)3 R i =H CH2CH(NH2)C02H M(mimo)3 Figure 2.9. Tris(N-substituted-3-hydroxy-4-pyridinonato) metal complexes. 2.6 Material and Methods The preparations were the same for Al and Ga with any given ligand; the details are given for the Al preparations and only the reactant concentrations and yields are given for the coresponding Ga preparations. The only diference of synthetic importance was the reduced aqueous solubility of the Ga complexes. For a given ligand, the Ga complex was ~50% les soluble than its Al analogue and this necesitated only minor adjustments in the preparations. 42 Unles stated otherwise, the reported yields were for analyticaly pure compound. Mimosine was available comercialy (Sigma, aprox. 99%) and was used without further purification. The other ligands were prepared and purified as described in Section 2.2. 2.6.1 Tris(3-hyo^oxy-2-methyl-4-pyridinonato)aluminum(in). Al(mp)^. Hmp  (1.25 g, 10.0 mol) and A1(NC«3)3»9H20 (1.24 g, 3.30 mol) were disolved in 20 mL water. The pH was raised to 8 with 2 M NaOH and the volume reduced to 10 mL by heating at 70 °C. While the suspension was stil hot, the pink solid was colected by filtration and gave 1.06 g (83% yield). Mp 310 °C dec. 2.6.2 Tris(3-hydroxv-2-methvl-4-pvridinonato)galium(II). Ga(mp)3;. To Hmp (1.09 g, 8.70 mol) in 20 mL H20 was aded 1.45 M GaCl3 (2.0 mL, 2.90 mol). Yield 1.13 g, 88%. Mp 290 °C dec. 2.6.3 Tris(3-hydroxv-1.2-dimethyl-4-pyridinonato)aluminum(II). Al(dp)3. Hdp (1.69 g, 12.2 mol) and A1(N03)3«9H20 (1.52 g, 4.05 mol) were disolved in 50 mL water. The pH was raised to 8 with 12 M N H 4 O H and the volume reduced to 20 mL by heating at 70° C. The pale pink product was colected after overnight coling at 20 °C and gave 1.39 g (79% yield). Mp 30 °C dec. 2.6.4 Tris(3-hvdroxv-1.2-dimethvl-4-pvridinonato)galium(II). Ga(dp)3. To Hdp (1.2 g, 8.80 mol) in 20 mL water was aded 1.45 M GaCl3 (2.0 mL, 2.90 mol): Yield 1.10 g, 79%. Mp 280 °C dec. 2.6.5 Tris(3-hydroxv-2-methyl-l-hexyl-4-pvridinonato)aluminum(II). Al(mhp)3. Hmhp  (0.86 g, 4.1 mol) and AlCl3»6H20 (0.3g, 1.3 mol) were disolved in 40 mL (1:1) methanol/water. Deprotonation by the adition of 12 M N H 4 O H (2 mL) was folowed by heating at 80 °C for one hour to remove exces amonia and methanol. The product 43 was extracted into CH2CI2 (2 x 50 mL) and the solution was dried over MgS04. Vacuum distillation left a brown oil that was triturated with hexanes. Drying in vacuo (18 hours) gave 590 mg of a yellow powder (67% yield). Purification was by sublimation (300 °C, IO"2 torr). Mp 305 °C dec. 2.6.6 Tris(3-hvdroxv-2-methvl-l-hexvl—4-pvridinonato)gallium(III). Ga(mhpp)3. To Hmhpp (0.63 g, 3.0 mmol) in 40 mL (1:1) methanol/water was added 1.45 M GaCl3 (0.69 mL, 1.0 mmol). Yield 0.50 g, 71% . Mp 285 °C dec. 2.6.7 Tris(3-hv(iroxv-2-memvl-l-emvl-4-pyridinonato)aluminum(III). Al(mepp)3. Hmepp (422 mg, 2.89 mmol) and Al(NO3)3«9H 20 (360 mg, 0.96 mmol) were dissolved in 25 mL of water forming a pale pink solution of pH 2.2. The pH was raised to 7 by the slow addition of 2N NaOH and then the solution was heated at 70 °C for 30 minutes. The cloudy yellow solution was transferred to a liquid-liquid extractor and the product extracted into 75 mL CH2CI2 over 12 hours. The organic layer was removed in vacuo. The pink solid was washed with diethyl ether and gave 425 mg (92% yield). Mp 250 °C dec. 2.6.8 Tris(3-hvdroxv-2-methvl-l-ethvl-4-pvridinonato)gallium(III). Ga(mepp)3. To Hmepp (669 mg, 4.37 mmol) dissolved in 15 mL of water was added 1 mL of 1.45 M Ga(Cl3) aqueous solution. Yield 652 mg, 85%. Mp 240 "C dec. 2.6.9 Tris(mimosinato)aluminum(III). Al(mimo)3. L-mimosine (591 mg, 2.98 mmol) and A1(N03)3»9H20 (371 mg, 0.99 mmol) were dissolved in 25 mL of water forming an orange colored solution of pH 1.8. Concentrated N H 4 O H was slowly added to bring to pH 6.5 and the yellow solution was left stirring at 20 °C for 1 hour. Methanol was added until cloudy and the solution was cooled at 5 °C for 16 hours. The pink precipitate was collected, the filtrate was concentrated to 15 mL in vacuo, and the precipitation procedure was repeated a second time. The two precipitates were combined giving 510 mg (83% yield). M p 2 5 0 ° C d e c . 4  2.6.10 Tris(rumosinato)galium(ID hydrate. Ga(mimo)3,H20. To L-mimosine (593 mg, 2.9 mol) disolved in 25 mL of water was aded 1.45 M GaCl3 aqueous solution (0.69 mL, 1.0 mol). Yield 571 mg, 86%. Mp 240 °C dec. 2.7 Discusion of the Synthetic Procedure The preparation of the tris(3-hydroxy-4-pyridinonato) Al and Ga complexes was relatively straightforward (Fig. 2.10). The 3:1 stoichiometry was closely maintained as the formation of the tris-ligand metal complexes was thermodynamicaly favored and it was not necesary to use an exces of ligand to push the reaction to completion. The preparations were in water except for M(mhp)3; because of the lipophilic N-hexyl substituent, a 1:1 methanol to water solution was used. To overcome purification problems with several of the preparations, other solvents (methanol, ethanol, acetone) were tried but the best results were obtained in water. The heating step was ostensibly included to ensure complete reaction but its primary utility became that of volume reduction to induce precipitation of the tris-ligand metal complexes. raise to pH 8 with BOH 3 HL + MX 3 • M L 3 + 3 B X + 3 H 2 0 A HL = 3-hydroxy-4-pyridinone ligand M = Al or Ga X = NO3 or Cl B = Na or NH4 Figure 2.10. Equation for the formation of the tris-ligand metal complexes. 45 All of the complexes were prepared in god yield (70 to 90%), and at no time were any side reactions observed. The only synthetic complication was the separation of the products from the salt formed in the neutralization proces. This separation was efected either by solvent extraction or by fractional crystalization in those cases where the solubility properties of the complexes precluded simple extraction. M(mp)3, M(dp)3 and M(mimo)3 were only soluble in protic solvents and they were isolated by fractional crystalization. The initial preparations of Al(mp)3 used AlCl3«6H20 with NaOH as the base. The reaction mixture was concentrated (at 70 °C) until Al(mp)3 precipitated thereby taking advantage of the large diference in solubility betwen the product and the byproduct salt (Al(mp)3 was predicted to be soluble at the mM level). The solution was coled and filtered; the IR and proton NMR spectra of the precipitate were consistent with the Al(mp)3 formulation and showed no proton containing impurities. However, the analytical results were indicative of a non-carbon containing contaminant (%C expected 54.18, found 46.93) and the likely cause was coprecipitation of NaCl. The preparation was repeated using A1(N03)3 since NaN03 is twice as water soluble as NaCl, and NO3" has a strong sharp IR band at 1380 cm"1 (a region fre of ligand absorbances) whereas CI" is invisible in the IR. A simple experiment showed that NO3" was present in the precipitate and ofered an easy purification method: the IR spectrum of the initial precipitate had the distinctive 1380 cnr1 band but if the solution was concentrated further and filtered while stil hot (70 °C), this band disapeared only to reapear when the solution was alowed to col to 20 °C before filtering. The hot (on the left) and the cold (on the right) Al(mp)3 precipitate IR spectra are reproduced as Figure 2.1 (the NO3" band at 1380 cm-1 is marked with an asterisk). The simple expedient of hot filtration gave analyticaly pure product in 83% yield. The Al(mp)3 preparation set the patern for the rest of the Al complexes in that 46 Al(NO>3 was used as the starting material to take advantage of the IR visibility of NO3". As the most convenient source of Ga was the chloride salt (obtained by dissolving the metal in HO), the reaction conditions were optimized for the Al preparation before being repeated with GaCl3 . » I I I I 1 1 4 0 0 1 2 0 0 1 4 0 0 1 2 0 0 Figure 2.11. The IR spectra of Al(mpp)3 from 1450 to 1200 cm'1. Al(dpp)3 did not precipitate from a hot solution (due to its high solubility temperature coefficient) but overnight cooling at 20 °C gave the pure product in good yield. To avoid problems with coprecipitation, the reaction mixture was more dilute than with Al(mpp)3 and N H 4 O H was used as the base ( N H 4 + salts are more water soluble than Na + salts). In the preparation of Al(mimo)3, fractional crystallization from hot water did not work nor was it possible to induce precipitation by cooling as the complex was too water soluble. Better results were obtained by using methanol as a second solvent; when precipitation was allowed to proceed slowly (16 hours), an analytically pure product was obtained. 47 Fractional crystalization gave low yields (50%) of Al(mepp)3. Although this complex is not soluble to a measurable extent in any aprotic solvents, it was posible to isolate pure product in yields > 85% by continuous liquid-liquid extraction with CH2CI2. Al(mhpp)3 is quite lipophilic and forms an intractable gumy solid in water. Its preparation from A1(CH3)3 was atempted using anhydrous organic solvents and a Schlenk line. The product was isolated in low yield (10%) but the results did not justify the complications inherent in a highly reactive starting material. Further experimentation using A1(N03)3, alcohol-water mixtures, and product extraction with CH2CI2 produced a reasonable preparation. The yields were somewhat low (70%) and sublimation (30 °C, 0.08 tor) was neded to give samples suitable for analysis. Despite considerable efort, we were unable to isolate any metal-ligand complexes with the H2exn ligand. Reactions using a 3:2 stoichiometry invariably gave precipitates that were so insoluble as to defy characterization or purification. The mas  spectra of a sample from an Al reaction did give an AI2L3 molecular ion but this was the sole piece of structural evidence to emerge from this work. The insolubility of the products proved to be an insurmountable problem. H.2exn itself is only soluble in acidic solution; this is rather surprising since Hmhp is readily soluble in organic solvents and the other ligands al have apreciable water solubility. The anomalous behavior of the bispyridinone may be due to the formation of H-bonded dimeric units. The hydrophilic sites in the molecule (the a-hydroxyketone moieties) would be involved in intermolecular H-bonds and the hexyl bridge is hydrophobic: the result would be water insolubility. The polar H-bonds would make the complex lipophobic: the result would be complete insolubility. A similar argument could explain the insolubility of the metal complexation reaction precipitate, as the M2L3 dimer would have hydrophobic hexyl bridges and lipophobic metal centers. 48 It is also posible for H2exn to polymerize by forming H-bonded chains rather than dimeric units. The monopyridinones would be much les likely to form chains and this may ofer a beter explanation for the higher melting point and sublimation temperature (both near 30  °C), and for the decreased solubility of H2exn when compared to the other ligands. In the metal complexation reaction, oligimers could be formed rather than dimers. Whether due to dimerization or polymerization, the established insolubility of H2exn and the conjectured insolubility of its M2L3 dimer is reasonable. 2.8 Characterization of the Tris(3-Hvdroxv-4-Pvridinonato') Metal Complexes The IR and proton NMR spectra of the tris-ligand metal complexes show diagnostic changes from those of the fre ligands. The changes are consistent with metal chelation via the deprotonated hydroxyl and the carbonyl oxygen atoms. The elemental analyses and the molecular ions in the mas spectra are consistent with an ML3 formulation. The spectra of the fre ligands were analyzed in some detail; therefore, only the diferences betwen the fre and complexed ligand ned be adresed in this section. Unlike the fre ligands, the tris-ligand metal complexes are not volatile enough for EI-MS but rather require positive ion fast atom bombardment (FAB) ionization. All the spectra were recorded with an AEI MS 9 and the samples were introduced on a coper tiped probe in either a glycerol or a thioglycerol matrix dependent on solubility. With this exception, the instrumentation and conditions are the same as those reported for the characterization of the fre ligands (se Section 2.4). 49 2.8.1 Elemental Analysis The samples were submited for analysis under N2 because the tris-ligand metal complexes were hygroscopic. Rigorous drying was required to remove water from the recrystalized products: heating in vacuo (0.05 tor) at > 65 °C for a minimum of 12 hours. Those samples that were sublimed were not subjected to further drying. Table 2.7. Results of elemental analyses (Found [Calculated]) Compound Formula %C %H %N Al(mp)3 [Ci8Hi8AlN306] 53.89 [54.13] 4.69 [4.5] 10.60 [10.52] Ga(mp)3 [Ci8Hi8GaN306] 48.65 [48.90] 4.29 [4.1] 9.53 [9.51] Al(dp)3 [C21H24AIN3O6] 57.40 [57.13] 5.54 [5.49] 9.67 [9.52] Ga(dp)3 [C2iH24GaN306] 51.98 [52.09] 5.10 [5.01] 8.54 [8.68] Al(mep)3 [C24H30AIN3O6] 59.72 [59.61] 6.25 [6.27] 8.90 [8.69] Ga(mepp)3»H20 [C24H32GaN307] 53.52 [53.85] 6.0 [5.85] 7.72 [7.85] Al(mimo)3 [C24H27AlN6Oi2] 46.61 [46.60] 4.60 [4.41] 13.52 [13.59] Ga(mimo)3»H20 [C24H29GaN6Oi3] 42.18 [42.43] 4.08 [4.31] 12.20 [12.37] Al(mhp)3 [C36H54ALN3O6] 6.50 [6.34] 8.19 [8.35] 6.34 [6.45] Ga(mhp)3 [C36H54GaN306] 62.42 [62.25] 7.94 [7.85] 5.93 [6.05] 50 2.8.2 Infrared Spectroscopy IR studies on pyridine derivatives show that the substituents vibrate largely independently of the ring.42 The fre ligand IR spectra were consistent with this conclusion and predictably, the spectra of the tris-ligand metal complexes and the coresponding fre ligand are quite similar. The bands are broadened somewhat but the general features of the spectra are the same. Also, the spectra of the Al and Ga complexes with any given ligand are virtualy identical above 80 cm-1 (al bands are within + 5 cm"1). There are, however, thre diagnostic diferences betwen the spectra of the fre and complexed ligand: the apearance of thre new bands below 80  cnr1, the los of V o H i and the bathochromic shift of Vc=0. The thre new bands (Table 2.8) are tentatively asigned as VM-0 but they are probably also coupled to the ring deformation modes.102 This is the only region of the spectrum where distinctions due to the diference in mas betwen the two metal ions are observable as at least one of the bands is consistently at lower energy in the Ga complexes. Figure 2.12 is a comparison of the spectra of Hmp  (top), Al(mpp)3 (midle), and Ga(mpp)3 (botom) from 90  to 30 cm"1 (the VM-O bands are marked with asterisks). The zwiterionic amino acid moiety obscures the JR spectrum of mimosine but the low energy region is clean enough to be of diagnostic utility in asigning the V M - O of the M(mimo)3 complexes. The los of the V O H is the most noticeable change in the spectra of M(dpp)3, M(mepp)3, and M(mhpp)3. The region above 310  cnr1 has only the ubiquitous band at 3450 cm"1 due to water. For M(mpp)3, the band at 3280 cnr1 is asigned as V N H with the hypsochromic shift of this mode (at 280 cm"1 in Hmp) ascribed to the absence of NH 51 intermolecular H-bonds. The asignment is not unusual for this mode72 and the band is much sharper in the tris-ligand metal complex than in the fre ligand. Table 2.8. Asignment of VM-O (cm-1, KBr disks) M(n Al W)3 Ga M(d Al 5P)3 Ga M(m< Al 3>P)3 Ga M(mr Al 1PP)3 Ga M(m Al imo)3 Ga 735 640 460 730 625 365 b 710 575 40 710570 370 705 580 460 705 565 410 720 580 475 b 720 570 340 b 60 620 40 65 b 60  b <30  The characteristic four band patern of mixed Vc=o and V R i n g is shifted upon formation of the tris(3-hydroxy-4-pyridinonato) metal complexes (Table 2.9), with the most pronounced bathochromic shift ocuring for the highest energy band (ca. 30 cm'1). The superimposed spectra of Al(dpp3 (on top) and Hdp  (on botom) in Figure 2.13 ilustrate the changes typicaly observed. Table 2.9. Characteristic infrared absorptions (cnr1). All are sharp and strong except as noted. Asignment M(mp)3 M(dp)3 M(mep)3 M(mhp)3 Vc=0 1620 1605 1600 1605 and 1595 1560 1555 1560 VRing 1535 sh 1515 1515 1520 1505 b 1495 1490 1490 52 54 2.8.2 Proton NMR Spectroscopy The spectral changes on formation of the tris(3-hydroxy-4-pyridinonato) metal complexes are minor, but there is a diagnostic shift in the resonances of the ring protons. The resonances are closer together and the Ha signal is shifted slightly upfield (typicaly ca. 15 Hz). The alkyl proton resonances are also shifted but in a les consistent fashion. Under the same conditions as described previously for the fre ligands, the OH resonances are absent in al the tris-ligand metal complexes. The NH signal is at 12.02 and 12.07 pm respectively for Al- and Ga(mp)3 (in (CI>3)2SO at 80 MHz). An exchange proces  is ocuring in the tris-ligand Al complexes at ambient temperature; however, the chemical shift changes are smal (from 8 to 20 Hz at 40 MHz) and are only observable on higher field strength instruments. This is another instance where a diference betwen the metals is observed as only the averaged spectrum is observed for the Ga complexes at the ambient probe temperature (typicaly 18 °C); this was not unexpected as Ga is known to be more labile than Al.103 (The temperature-dependent NMR of the tris(3-hydroxy-4-pyridinonato) Al and Ga complexes is adresed in Chapter IV.) The fluxionality of the Al complexes results in overlaping signals so the chemical shift of the central signal is reported. The spectral asignments for al but the M(mimo)3 complexes are listed in Table 2.1 and the peak integrations are consistent with the asignments. It is posible for mimosine to chelate via the amino acid nitrogen and oxygen atoms although this was considered to be unlikely given the relative softnes of nitrogen. The IR spectra are indicative of chelation by the a-hydroxyketone moiety, but the broadnes of the amino acid bands in both the fre and complexed ligand makes it imposible to rule out 5  entirely binding at this site. The compounds are not soluble in aprotic solvents so it was not posible to lok for the acidic protons by NMR; however, comparison of the chemical shifts of fre and complexed mimosine in D2O (Table 2.10) shows the same smal shifts in the Ha and Hb resonances that were sen with the other ligands. The He resonance (mimosine is the only ligand in this study without a 2-methyl substituent) is shifted upfield so that it is betwen the AB doublets, rather than downfield of them, in the M(mimo)3 complexes. The exocyclic proton signals, found as thre multiplets from 4.2 to 4.7 pm that each integrate as one proton, are broadened but they did not shift significantly on chelation. The similarity in the movement of the ring proton signals indicates that mimosine is chelating the metals in an fashion analogous to the other 3-hydroxy-4-pyridinone ligands employed in this study. Table 2.10. *H NMR chemical shifts (8) for mimosine and M(mimo)3 complexes (pm). Recorded at 40 MHz. M I Cr ^CH 2 CH(NH 2 )C0 2 H H Asignment Mimosine Al(mimo)3 Ga(mimo)3 Ha(d) 7.78 7.73 7.72 Hc(s) 7.82 7.46 7.54 Hb (d) 6.85 6.72 6.7 56 Table 2.11. *H NMR chemical shifts (8) for the tris-ligand metal complexes (ppm). Recorded at 400 MHz. n R = H C H 3 d C H 2 <j C H 3 e C H 2 d C H 2 e C H 2 f C H 2 g C H 2 h C H 3 i M(mpp)3 M(dpp)3 M(mepp)3 M(mhpp)3 M(mpp)3 Al Ga M(dpp)3 Al Ga M(mepp)3 Al Ga M(mhpp)3 Al Ga Ha(d) Hb(d) C H 3 c (s C H 2 d C H 2 e CH2f,g,|» C H 3 i 7.50 6.56 2.31 7.48 6.61 2.34 7.56 6.50 2.35 3.81* 7.49 6.47 2.43 3.80 s* 7.53 6.48 2.32 4.08 1.30* 7.52 6.54 2.36 4.08 q 1.30* 7.02 6.48 2.38 3.82 1.66 1.27 0.85 7.24 6.70 2.56 3.96 t 1.73 m 1.30 m 0.87 m * Read C H 3 . All recorded in D2O except: @ = CD3OD; # = CDCI3. D2O spectra externally referenced to DSS; spectra in other solvents are internally referenced. Abbreviations: s = singlet; d = doublet; q = quartet; t = triplet; m = multiplet. 57 2.8.3 Mass Spectrometry The FAB-MS spectral data are listed in Table 2.12. In every case the molecular ion is found as the HML3 + (HM+) peak, and the ML2 + base peak is formed by the loss of one ligand. For the Ga complexes, the peaks are in the natural isotopic ratio of 3:2 6 9Ga to 71Ga. The presence of peaks due to the matrix make the assignment of the spectra beyond this point profidess. Table 2.12. Data from FAB-MS spectra of the tris-ligand metal complexes (m/z). M(rr Al PP)3 Ga M(di Al >P)3 Ga M(n* Al 3>P)3 Ga M(m Al W)3 Ga M(m Al mo)3 Ga HML3 4 ML 2 + 400 275 444 442 319 317 442 303 486 484 347 345 484 331 528 526 375 373 652 443 695 693 487 485 619 421 662 660 465 463 58 Chapter III Solid State Studies A. 3-Hydroxy-4-Pyridinone Crystal Structures 3.1 Introduction The structures of Hmpp, Hdpp, and Hmepp (Fig. 3.2) were established by single crystal X-ray diffraction. Apart from the obvious motive of structure confirmation, the crystallographic study was undertaken to establish the extent of double bond delocalization in these compounds, to determine if N-alkyl substitution increased the delocalization, and to discover what effect intermolecular H-bonding had on the crystal packing arrangement. o o ' o" o" o R R R R Figure 3.1. 4-pyridinone resonance forms. There are several possible resonance forms for 4-pyridinone 1 0 4 (Fig.3.1) and this is of some importance when considering the potential stability of the tris(4-pyridinonato) metal complexes. A significant contribution from the resonance forms with a partial negative charge on the carbonyl oxygen might lead to complexes of increased stability as the bonding interactions with the hard acids A l and Ga are primarily electrostatic in nature. 1 0 5 Delocalization of the ring double bonds and a lengthening of the carbonyl bond 59 would be evidence of resonance hybridization in the 4-pyridinone ligands. It was thought that alkyl substitution at the ring nitrogen might enhance its ability to accommodate a positive charge thereby increasing the contribution from the pyridinium resonance form; the veracity of that supposition was examined by comparing the structures of Hmpp and its N-alkylated analogues. Intermolecular H-bonding had been indicated by the IR and proton N M R spectra and not surprisingly, the 3-hydroxy-4-pyridinones crystallized as dimeric H-bonded units. Our interest in this interaction is due to the potential connection between H-bonding and the solubility properties of the ligands and the metal-ligand complexes. In addition, our research group would eventually like to enter the field of inclusion chemistry and extensively H-bonded structures have potential for the synthesis of clathrate compounds. Knowledge of the packing arrangement could be useful in determining the suitability of Hmpp, Hdpp, and Hmepp as host molecules. The structures of Hmpp and Hdpp have been reported previously by this laboratory 6 0 and this is the first report of the structure of Hmepp. The single crystals of Hmpp were grown from a supersaturated water solution: a twice recrystallized sample of the ligand was dissolved in dilute acetic acid (pH 2), the pH was adjusted to 8 with 2 M NaOH, and overnight cooling at 10 °C gave crystals suitable for X-ray diffraction. Crystals of Hdpp could not be grown from water and were obtained by liquid-liquid diffusion from methanol with diethyl ether as the second solvent. Hmepp crystals were grown by slow evaporation from a dilute water solution. A l l of the crystal structures in this thesis were determined by Dr. Steven J. Rettig of the U.B.C. structural chemistry laboratory. In all cases, the crystals were stable under ambient atmosphere and were mounted on glass fibers for X-ray diffraction. Figure 3.2. ORTEP view of Hmpp, Hdpp, and Hmepp. 61 Table 3.1. Bond lengths (A) and angles (deg) for the free ligands (bond lengths in the upper and bond angles in the lower portion of the table). In this and all subsequent listings of bond parameters, the estimated standard deviation (esd) is in parenthesis after the data entry. Atoms Compound Hmpp Hdpp Hmepp 0 ( 1 ) - C ( 3 ) 1.358 (2) 1.360 (3) 1.357 (3) 0 ( 2 ) - C ( 4 ) 1.280 (2) 1.272 (3) 1.258 (3) N - C(2) 1.356 (2) 1.369 (4) 1.378 (3) C(2) - C(3) 1.371 (2) 1.376 (4) 1.370 (3) C(3) - C(4) 1.431 (2) 1.430 (4) 1.433 (3) C(4)-C(5) 1.411 (2) 1.407 (4) 1.410(3) C(5) - C(6) 1.365 (2) 1.364 (4) 1.350 (4) C(6) - N 1.345 (2) 1.352 (4) 1.351 (3) C(2) - C(l) 1.493 (2) 1.489 (4) 1.490 (4) N - C(7) 1.482 (4) 1.485 (3) C(7) - C(8) 1.484 (4) 0 ( 1 ) - C ( 3 ) - C ( 2 ) 117.8 (2) 118.2 (3) 118.3 (2) O(l) - C(3) - C(4) 120.4 (2) 119.2(2) 118.8 (2) 0 ( 2 ) - C ( 4 ) - C ( 3 ) 120.0 (2) 120.5 (3) 121.3 (2) 0(2) - C(4) - C(5) 124,4 (2) 124.3 (3) 124.3 (2) C(6) - N - C(2) 121.8 (2) 120.8 (2) 120.5 (2) N - C(2) - C(3) 119.0(2) 118.6 (3) 118.6 (2) C(2) - C(3) - C(4) 121.8 (2) 122.5 (3) 122.9 (2) C(3) - C(4) - C(5) 115.5 (2) 115.2 (3) 114.3 (2) C(4) - C(5) - C(6) 120.7 (2) 121.1 (3) 121.9 (2) C(5) - C(6) - N 121.1 (2) 121.7 (3) 121.8 (2) N - C(2) - C( l ) 118.3 (2) 119.0 (3) 120.0 (2) C( l ) - C(2) - C(3) 122.7 (2) 122.3 (3) 121.5 (2) C(6) - N - C(7) 118.4(3) 117.3(2) C(2) - N - C(7) 120.7 (3) 122.0 (2) N - C(7) - C(8) 111.8 (2) 62 3.2 Results and Discussion The six-membered pyridinone rings are slightly non-planar and the maximum deviations from the mean planes are 0.019(2) A in Hmpp, 0.009(3) A in Hdpp, and 0.013(3) A in Hmepp; the distortions from planarity are toward a C(4) envelope, a C(2)-C(5) boat, and a N-C(4) boat respectively. A n examination of the bond angles (Table 3.1) shows only minor differences between the three structures. N-alkylation results in a smaller intra-annular bond angle at nitrogen (C(2)-N-C(6)), and the C(3)-C(4)-C(5) angle is compressed in all three compounds to a minimum of 114.3° in Hmepp. This compression is indicative of the strength of the C=0 bond (shortest in Hmepp) and is due in part to the lone pairs of electrons on the oxygen. The bulk of the N-alkyl groups predictably affects the position of the ring methyl group (C(l)) and this increases the N-C(2)-C(l) angle from 118.3° in Hmpp to 120° in Hmepp. The small changes in the bond angles involving the oxygen atoms are due to the packing of the H-bonded dimeric units. The nitrogen coordination is planar within experimental error in all three compounds and the bond lengths indicate partial delocalization of the formal double bonds. By comparing the observed bond lengths with the values calculated (using the empirical relationship between bond order and bond length) for the double bond character consonant with the probable resonance forms, it is possible to relate double bond delocalization (as determined crystallographically) to the approximate contributions from each resonance h y b r i d . 1 0 6 Calculations of this nature were done for 2-pyridinone, a structural isomer of 4-pyridinone that is capable of similar resonance interactions involving the nitrogen lone pair of electrons.8 0 The formula used was R =•—^1 f ) ^ * ^* w n e r e ^ 1 S t n e observed bond length, R i and R 2 are the standard carbon single- and double bond lengths 63 and x is the bond order expressed as the fraction of double bond character. The combination of 2-pyridinone resonance forms that gave the best agreement with the observed bond lengths are shown in Figure 3.3. O + N o 0 0 • A o 20 10 5 50 15 Figure 3.3. Resonance forms for 2-pyridinone (percent contribution is in bold type). Analogous calculations were not done for the 3-hydroxy-4-pyridinones as a comparison of the bond lengths listed in Table 3.2 clearly indicates the similarity in the degree of double bond delocalization between Hmpp and 2-pyridinone. (Compound C was included to show that N-alkyl and hydroxy substitution did not appreciably alter bond delocalization in the 2-pyridinone ring.) The carbonyl bond is significantly longer in Hmpp and the similarity in the ring bond lengths is readily apparent; additionally, the mean of the six ring bonds is the same and Hmpp even has a slightly smaller standard deviation (a). This indicates there is as much delocalization in Hmpp and suggests there may be as significant a contribution from the pyridinium resonance forms for Hmpp as was seen in 2-pyridinone (35%). Contrary to our expectations, N-alkylation did not increase the amount of double bond delocahzation in the 4-pyridinones. Compared to Hmpp, there was no significant change in Hdpp bond lengths and the changes in Hmepp were indicative of a decrease rather than an increase as the C(5)=C(6) and C=0 bonds were shorter and the C - N bonds were longer than in Hmpp (see Table 3.1). 64 Table 3.2. Comparison of bond lengths (A). (The ring carbons are numbered as in Hmpp.) O . O H 3 ^OH 2 A " C H 3 Hmpp 2-Pyridinone J x N ^ O I n- Bu 3-Hydroxy-1 -butyl-2-pyridinone Bond A Compound B# c@ c=o • 1.28 1.24 1.24 C(2)-C(3) 1.37* 1.44 1.43 C(3)-C(4) 1.43 1.33* 1.35* C(4)-C(5) 1.41 1.42 1.41 C(5)-C(6) 1.37* 1.37* 1.33* N-C(2) 1.36 1.40 1.37 N-C(6) 1.34 1.34 1.38 Mean (a) 1.38 (0.033) 1.38 (0.044) 1.38 (0.038) # Data from ref. 80. @ Data from ref. 107. The 3-hydroxy-4-pyridinones crystallized as centrosymmetric 0 - H - 0 = C hydrogen bonded dimeric units. (The dimeric unit from Hdpp is included as Figure 3.4 and the packing arrangements of Hmpp, Hdpp, and Hmepp are given as Figures 3.5, 3.6, and 3.7 respectively.) The dimeric units of Hdpp and Hmepp are separated from one another by normal van der Waals distances but in Hmpp each dimer is linked to four others by N-H—0=C hydrogen bonds to form a three-dimensional network. The strength of the 65 H-bonds can best be judged by examining both the H-bond parameters and the IR stretching frequencies (Table 3.3). In all three structures, the O - H - O bond lengths are intermediate (<2.70 A 1 0 8) and the angles are within the range typical for H-bonds (140-180*109). in Hmpp, the N - H - O distance and VNH indicate relatively strong nitrogen H-bonds; 1 0 8 the constraints imposed by the N H hydrogen bonding causes a weakening of the O - H - O bonds as indicated by the bond angle of 144* and by the VOH of 3270 cm- 1 (VOH > 3200 c m - 1 is classified as weak H-bonding 1 0 8). A comparison of VOH in the three ligands supports the proposition that the weakening of the bonds between the dimeric units in Hmpp is due to the formation of the three-dimensional network. Figure 3.4 The Hdpp hydrogen bonded dimeric unit. Table 3.3. A comparison of the free ligand H-bond parameters and the IR stretching frequencies. Values involving the nitrogen in Hmpp are in italics. Atoms Hmpp Compound Hdpp Hmepp H • O (A) 1.90(3) 1.90(2) 1.94 (5) 1.83(4) (N)O-O (A) 2.670 (2) 2.796(2) 2.692 (3) 2.659 (2) ( J V j O - H - O (deg) 144(3) 166(2) 154 (3) 150 (2) Vw)OH (cm-1) 3270 2800 3150 3180 66 In the packing of the dimeric units, the alkyl groups are staggered to allow the closest spacing of the pyridinone ring;, and this is most readily seen in the unit cell of Hdpp (see Fig. 3.6). Comparing Hdpp to Hmepp shows that the N-ethyl group disrupts the stacking of the dimeric units; this is reflected in an increase in the volume of the unit cell (both compounds crystallized in the orthorhombic space group* Pbca ) from 1312 A 3 in Hdpp to 1628 A 3 in Hmepp. The bulkier ethyl group also results in an appreciably lower calculated density for Hmepp (1.14 g/cm3) than either Hmpp or Hdpp (1.42 and 1.41 g/cm3 respectively). Despite the similarities in the strength of the H-bonds, the Hmepp crystal lattice appears to have been weakened due to the steric requirements of the N-ethyl group. This conclusion is supported by the lower melting point (205 vs. 260 *C) and the greater water solubility (five-fold increase at 25 *C) of Hmepp when compared to Hdpp. Figure 3.5. Stereo view of the packing arrangement in Hmpp. * The crystallographic data (including the unit cell dimensions) for Hmpp, Hdpp, and Hmepp are in the Appendix as Table A.2. 67 Figure 3.6. Stereoview of the packing arrangement in Hdpp. 68 Figure 3.7. Stereoview of the packing arrangement in Hmepp. 69 B. M(dpp)3 Crystal Structures 3.3 Introduction The structures of A1-, Ga-, and In(dpp)3 were determined; the facial (fac) isomers crystallize in the trigonal space group* P3 and the empirical formula is M(dpp)3«12H20. The threefold symmetry of the fac isomers results in an asymmetric unit consisting of 1/3 of a metal ion, one ligand, and four water molecules. The water molecules form hexagonal rings that are H-bonded to the chelating oxygen atoms of the complexes by bridging waters. The water molecules do not coordinate the metal; rather they are structural w a t e r s 1 1 0 which form a three dimensional framework that is an important factor in determining the structure of the inorganic complex. The unit cell diagram (Figs. 3.8 and 3.9) accentuates the spatial relationship of the water channels to the tris(pyridinonato)metal portion of the structure (/ac-M(dpp)3 unit). The view perpendicular to the symmetry axes of the water channels and the/ac-M(dpp)3 units in Figure 3.10 affords the best perspective of the entire water network. Because of the uniqueness and the structural importance of the water network, the Results and Discussion wil l be divided into two parts: the /ac-M(dpp)3 unit in Section 3.4.1 and the hexagonal water network in Section 3.4.2. The A l and Ga complexes are isostructural but due to a slight variation in the packing arrangement, the In complex is not crystallographically equivalent; however, the/ac-M(dpp)3 units of the three complexes are isostructural (Fig. 3.11) and the bonding parameters will be examined together. Despite * The crystallographic data for the M(dpp)3 complexes are in the Appendix as Table A.3. 70 the differences in packing, the water network is the same in all three compounds and will be addressed as such. The single crystals were grown from supersaturated water solutions: an aqueous suspension was heated to 80 °C, filtered, and crystals formed after several days at 20 °C. The crystals were stable under ambient atmosphere and the integrity of the water network was exemplified by Al(dpp) 3 which successfully analyzed for 12 waters. The crystals were collected by filtration, dried overnight in a desiccator, and submitted for analysis under nitrogen: expected (found) C 38.35 (38.15); H 7.37 (7.39); N 6.39 (6.31). The strength of crystal lattice accounted for the rigorous conditions necessary to isolate the anhydrous compounds (see Section 2.8.1). To determine the importance of the lattice waters to the geometry of the/ac-M(dpp)3 units, numerous attempts were made to grow crystals from other solvents, either by liquid-liquid diffusion or by slow evaporation. The latter technique gave In(dpp)3 crystals from 95% ethanol, but the complex again formed as the dodecahydrate. This was the only system that produced crystals suitable for X-ray diffraction. The A l - and Ga(dpp) 3 structures were reported previously by this laboratory. 1 0 1 - 1 1 1 The synthesis of In(dpp)3 was analogous to that of its congeners and the structure was reported in a study of In complexes. 1 1 2 Although my doctoral research did not involve the synthesis of indium complexes, the synthetic procedures were based on my work with the tris(3-hydroxy-4-pyridinone) A l and Ga complexes. I was directly involved in the growing of single crystals of In(dpp)3 and the similarity of the three structures motivated the inclusion of the In complex in this thesis. A number of In complexes were also part of the octanol/water partition coefficient study that is included in Chapter V . 71 Figure 3.8. O R T E P view down the c axis of the unit cell packing of the M(dpp)3 complexes (M = A l and Ga). 72 73 i * >4 Figure 3.10. ORTEP view of a part of the H-bonding network of waters inM(dpp)3. This view down the a axis shows all of the independent O atoms; all of the atoms in the M(dpp)3 units are omitted except for the M 0 6 octahedral coordination sphere. 74 3.4.1 The fac-M(dpph Unit The ligand rings are planar within experimental error. There are no significant differences in the ring bond lengths between the A l and Ga complexes and only minor changes are seen in comparison to the In complex (Table 3.4). For In(dpp)3, one C-N bond is longer but the other is shorter and the same relationship holds for the formal double bonds (C(2)-C(3) and C(5)-C(6)); this produces no net change in double bond delocalization. Comparison to the free ligand reveals only minor changes and the extent of ring double bond delocalization is essentially unaffected by metal chelation. The delocalization in the C-0 bonds is greater in the metal complexes than in the free ligand: the difference between 0(1)-C(3) and 0(2)-C(4) decreases from 0.088 A in Hdpp to 0.028 A, 0.038 A, and 0.054 A in A1-, Ga-, and In(dpp)3 respectively. The extent of delocalization is clearly seen by comparing bond lengths with the related dihydroxy ligands, the catechols. The average C-O distance (calculated from a number of different metal complexes) is 1.36(1) A for the catecholate anion and 1.29(1) A for the delocalized semiquinone radical anion. 3 2 For the dpp anions in our metal complexes, the 0(1)-C(3) bonds are signifcantly shorter (mean = 1.337(8) A) than the catecholate bond and the 0(2)-C(4) bonds are the same (mean = 1.297(8) A) as the semiquinone bond (within the esd). The only significant variations in ligand bond angles occur in the carbon atoms (C(3) and C(4)) that are part of the chelate ring. Chelation causes a compression in the interior angles of the chelate ring: compared to the free ligand values, the 0(1)-C(3)-C(4) and 0(2)-C(4)-C(3) bond angles are decreased in the A l and Ga complexes and are the same in the In complex. The smaller the metal the greater the compression. Further evidence of this is seen in (C6Hs)2B(dpp) where these angles for the dpp" anion are further compressed to 111.8(4)° and 113.6(3)° respectively. 1 1 3 Figure 3.11. ORTEP view of the tris(ligand) portion of the M(dpp>3 complexes ( M = Al, Ga, and In). 76 Table 3.4. Bonding parameters for the M(dpp)3#12H20 complexes. The bond lengths (A) are in the upper portion and the bond angles (deg) are in the lower portion of the table. For comparison, the parameters for the Hdpp ligand have been included in the last column. Atoms M(dpp)3 M= Al Ga In Hdpp 0(1)-C(3) 1.327(3) 1.342 (5) 1.343 (3) 1.360 (3) CK2)-C(4) 1299 (3) 1.304 (5) 1.289 (3) 1.272 (3) N - C(2) 1.369 (3) 1.372 (5) 1.355 (3) 1.369 (4) C(2) - C(3) 1.385 (3) 1.382 (6) 1.393 (3) 1.376 (4) C(3)-C(4) 1.423 (3) 1.409 (6) 1.403 (3) 1.430 (4) C(4)-C(5) 1.398 (3) 1.403(6) 1.410 (3) 1.407 (4) C(5) ~ C(6) 1.360 (3) 1.359 (6) 1.344 (4) 1.364 (4) C(6) - N 1.360 (3) 1.349 (5) 1.371 (3) 1.352 (4) C(2)-C(l) 1.493 (3) 1.486 (7) 1.487 (4) 1.489 (4) N - C(7) 1.476 (3) 1.470 (6) 1.463 (3) 1.482 (4) 0(1)-C(3)-C(2) 124.5 (2) 122.4 (4) 120.1 (2) 118.2 (3) CXI) - C(3) - C(4) 115.3 (2) 116.9(4) 119.1 (2) 119.2 (2) 0(2)-C(4)-C(3) 116.0(2) 117.9(4) 120.0 (2) 120.5 (3) 0(2)-C(4)-C(5) 125.7 (2) 124.2 (4) 123.0 (2) 124.3 (3) C(6) - N - C(2) 121.2 (2) 121.1 (4) 120.2(2) 120.8 (2) N - C(2) - C(3) 119.0(2) 118.9(4) 119.7(2) 118.6 (3) C(2)-C(3)-C(4) 120.2 (2) 120.7 (4) 120.8(2) 122.5 (3) C(3)-C(4)-C(5) 118.4(2) 117.9 (4) 116.9(2) 115.2(3) C(4)-C(5)-C(6) 119.5(2) 119.7(4) 120.8 (2) 121.1 (3) C(5) - C(6) - N 121.7 (2) 121.7 (4) 121.6 (2) 121.7 (3) C(6)-N-C(7) 117.7(2) 117.7(4) 117.0(2) 118.4 (3) C(2) - N - C(7) 121.1 (2) 121.1 (4) 122.8 (2) 120.7 (3) N - C(2) - C(l) 119.2 (2) 119.4(4) 118.8 (2) 119.0(3) C(l) - C(2) - C(3) 121.9 (2) 121.7 (4) 121.4 (2) 122.3 (3) 77 Table 3.5. Intra-annular torsion angles of the chelate rings in the M(dpp)3 complexes (deg). Atoms M = A l Ga In 0(2)-M-0(l)-C(3) - 5.05 (14) - 5.4 (3) 7.47 (14) M-0(l)-C(3)-C(4) 4.7 (2) 5.0 (4) - 7.2 (2) 0(l)-C(3)-C(4)-0(2) - 1.0 (3) - 0.8 (6) 0.8 (3) M-0(2)-C(4)-C(3) - 3.1 (2) - 3.8 (5) 6.0 (3) 0(l)-M-0(2)-C(4) 4.50 (15) 5.0 (3) - 7.21 (15) The increasing size of the metal ion is accompanied by a reduction in ligand strain as evinced by compression of the O-C-C bond angles. By this criterion, In best fits the dpp" anion since it caused the least deviation from the bond angles in the free ligand. This is a tentative conclusion as the changes are small and crystal packing forces could be a factor in the observed variations. The planarity of the chelate rings (Table 3.5) varies in the opposite direction. The metal ion is shifted out of the plane and the deviation from planarity is the greatest for In(dpp)3. There is a compression of the M(dpp)3 units from ideality along the C3 axis leading to a decrease in the 0(l)-M-0(2) angles and an increase in the exocyclic 0(l)-M-0(2') angles in all the structures (Table 3.6). The exocyclic 0(1)-M-0(1') and 0(2)-M-0(2') angles are 90°(±1°) in the A l and Ga complexes, but they increase to 93.58 and 92.40 A respectively in the In complex The buckling of the chelate ring and the compression along the C 3 axis increase directly with the ionic radius of the metal center. A n examination of the packing arrangement (see Figs. 3.8 and 3.10) shows that along the c axis (parallel to C3) the 78 distance between fac-M(dpp)3 units is governed largely by a single bridging 0(3) water. The length of the H-bonds from 0(3) to the ligand oxygens does not vary significantly among the three structures. If the rigid water network will not expand, an increase in the size of the metal center must result in compression along the C3 axis and a twisting of the chelate plane. Table 3.6. Bond angles (deg) for the metal-ligand interactions in M(dpp) 3 complexes. Atoms M = A l Ga In O(l) - M - 0(2) 84.23 (6) 83.22 (12) 77.87 (6) 0 ( 1 ) - M - 0 ( 1 ) ' 90.81 (8) 90.90 (12) 93.58 (6) 0 ( l ) - M - 0 ( 2 ) ' 95.71 (7) 96.65 (12) 97.74 (7) 0 ( 2 ) - M - 0 ( l ) ' t r a n S 171.85 (7) 170.48 (12) 166.18 (6) 0(2) - M - 0(2)' 89.83 (8) 90.01 (12) 92.40 (6) M - O(l) - C(3) 112.20 (13) 110.7 (2) 110.82 (13) M - 0(2) - C(4) 112.00 (14) 110.9 (3) 111.39 (14) The M-O bond lengths are listed in Table 3.7; the differences in the M - O bond lengths and the differences in metal ionic radii are included in italics. Comparison of the differences shows that the M-O bond lengths are reasonably close to the values predicted by the differences in the size of the metal. The Al-0(2) bond shows the most deviation and this slight increase from the predicted length suggests the metal-oxygen bonding is weakest for Al(dpp) 3. The strength of the Ga-0 bonds is seen by comparison to the 1.986(6) A average Ga-O distance in K3[tris(catecholato)Ga(JJi)].34 The dative bond (Ga-0(2)) is the same length as the catecholate bond (within esds) and Ga-O(l) is significantly shorter. The 79 relative shortness of Ga-0(2) is further evidence of a significant contribution from the pyridinium resonance forms with a partial negative charge on 0(2). The length of the Ga-0 bonds also compares well to the bonding in tris(3-hydroxy-l-butyl-2-pyridinonato)iron ( I H ) ; 1 0 7 when adjusted for the differences in metal ionic radii (a difference of 0.025 A), the M - 0 distances are 2.024 and 1.955 A. Table 3.7. Metal ionic radi i 3 0 and M-O bond lengths3 in M(dpp)3 complexes (A). M = A l Ga In Ionic Radius 0.535 0.085 0.620 0.180 0.800 M - 0(1) 1.893 (2) 0.074 1.967 (3) 0.167 2.134 (2) M - 0(2) 1.923 (2) 0.067 1.990 (3) 0.175 2.165 (2) a Going along a row, the differences between entries in adjoining columns are in italics. The structural differences between the free and the complexed ligand are minor and the changes are indicative of increased double bond delocalization in the metal-ligand complexes. The/ac-M(dpp)3 units are isostructural and the variations between the structures are readily rationalized by consideration of the metal ionic radii and the rigidity of the packing arrangement. The metal-oxygen interactions are strong and compare favorably to those of other oxygen containing bidentate ligands known to be good chelators of trivalent metals. 80 3.4.2 The Hydrogen Bonded Water Network The water molecules form a three dimensional array: half (H20(3) and H2CX6)) form a bridge from the/ac-M(dpp)3 units to the hexagonal channels formed by the other half of the water molecules (H20(4) and H20(5)) in the corners of the unit cell (see Figure 3.10 for numbering). According to the classification of Falk and K n o p , 1 1 4 in which waters are designated by the number and type of hydrogen bonding water neighbors (a is a proton acceptor and d is a proton donor), H20(3) is a, H2CX6) dda, and the waters in the hexagonal channels (H20(4) and H20(5)) are ddaa. Table 3.8. H-bond distances (A) and angles for M(dpp)3»12H20. Interaction H- O o- •O O - H • O (deg) M = A l Ga A l Ga A l Ga 0(3)-H(a> 0(1) 1.95(6) 1.98(9) 2.861(3) 2.859(5) 161(4) 165(8) 0(3)-H(b)-0(2) 2.08(4) 2.19(6) 2.849(3) 2.842(6) 164(4) 159(7) 0(4)-H(a)-0(6) 2.12(5) 2.19(6) 2.772(3) 2.765(6) 159(5) 164(10) 0(4)-H(b)-0(4) 1.62(8) 1.58(12) 2.747(3) 2.746(5) 168(5) 154(8) 0(5)-H(a)-0(4) 2.06(10) 2.44(12) 2.802(4) 2.807(7) 150(9) 145(22) 0(5)-H(c)-0(5) a 1.50(13) 1.53(24) 2.793(4) 2.779(7) 174(6) 160(10) 0(5)-H(d)-0(5) 2.17(13) 2.793(4) 149(12) 0(6)-H(a)-0(3) 1.72(4) 1.99(8) 2.729(4) 2.734(7) 174(3) 167(8) 0(6)-H(b)-0(5) 2.00(4) 1.91(7) 2.791(4) 2.778(7) 174(3) 171(6) a This interaction involves H(05b) for the Ga compound. 81 The H-bonds* (Table 3.8) between the ligand O atoms (O(l) and 0(2)) and H20(3) are relatively strong considering these oxygen atoms are chelating the metal atoms. A l l six of the chelating O atoms are hydrogen bonded to 0(3) water molecules; the latter form an infinite chain down the c axis bridging from the hydroxy 0(1) in one ligand to the keto 0(2) of a ligand rotated by 120° and translated by one unit cell. (This is why Figure 3.8 shows the contents of two unit cells). The O - H and O - O distances for the chelating oxygens vary from 1.95(6) to 2.19(6) A and from 2.842(6) to 2.861(3) A respectively. Not surprisingly, the chelating hydroxy O forms shorter O - H bonds than the chelating keto O. The hexagonal channels of water molecules in the corners of the unit cell are the most unique feature in these structures. (The water 0(5) was found to be twofold disordered in the A l complex and is shown with four half protons bound to it in Figure 3.10.) Each of the water rings has crystallographically imposed 3 or S6 symmetry and the rings essentially adopt the structure of i c e 1 1 5 in its stable low pressure form, ice-I n . (Refer to Fig.3.9 for a comparison of the water rings and the structure ice.) Every water molecule in the ring is hydrogen bonded to four nearest neighbors with the added distinction that the overall structure is predominantly proton-ordered, unlike ice-I n which is completely disordered. The O - O distances in the channels (those involving 0(4), 0(5), and 0(6)) vary from 2.75 to 2.81 A compared with the average value of 2.75 A for O - O in ice-I n at 100K .U6 Within each water ring the hydrogen bonding is h o m o d r o m i c 1 1 7 ' 1 1 9 because of the unidirectional circular bonding pattern. A l l the O - H - O bonds run in a counterclockwise direction when viewed down the hexagonal axis (see Figure 3.9). The arrangement is crystallographically imposed by the space group (P3), however, and does not occur * There were no significant differences in the H-bond parameters for In(dpp)3 so they have been placed in the Appendix as Table A.5 82 independent of symmetry constraints as do the water networks in the structures of some nucleosides 1 2 0 and P-cyclodextrins. 1 2 1 ' 1 2 2 Homodromic hydrogen bonding arrangements are favored (and more frequently observed despite symmetry constraints) over heterodromic or antidromic arrays because of an inherent lower dipole moment. 1 1 8 There is a considerable cooperative effect which results in increased hydrogen bonding activity for a hydroxyl group when it is already the donor or acceptor in a hydrogen bond. Quantum mechanical calculations have confirmed that chain structures (particularly cyclic) of hydrogen bonds are energetically favored over individual interactions. 1 2 3 This is the first example where this arrangement of water rings occurs in hydrates containing (relatively) large metal complexes; there has recently been reported a tris(2-pyridinonato) iron complex that contains hexagonal water rings, but the rings are discrete and there is no three-dimensional water network. 1 0 7 Water rings are, however, well known in the crystal structures of i c e 1 1 5 - 1 1 6 and the clathrate hydrates 1 2 4 ' 1 2 5 The clathrate hydrates are crystalline compounds which consist of a hydrogen bonded water host network (often a H40O20 pentagonal dodecahedron) within which a guest is held by an interaction which varies from weakly hydrogen bonding, to ionically bonding where one or more ion is associated with, or incorporated in, the water framework. 1 2 6 We note similarities here with hexamethylenetetramine hexahydrate ((CH2)6N4*6H20), an unusual hydrate in which the host lattice is not based on a regular polyhedron, so there are no well defined polyhedral cavities. 1 2 7 It shares with the structures reported herein the hexagonal water rings; however, the water rings are staggered around the cage-like amine molecules in a spiral (instead of a linear chain) to which the latter are bound. The above observations suggest that the M(dpp) 3 complexes present (when crystallized from water) appropriate conditions for the formation of the water channels in what is a previously unobserved, but energetically favored, hydrogen bonding arrangement. This probably results by virtue of both the complex size and the hydrophobic 83 core of the unit cell formed by the pairs of methyl groups on the ligands from the two /ac-M(dpp)3 units. There is a clear alternation of hydrophilic and hydrophobic regions along the ab diagonals of the unit cell; no doubt this feature also contributes to the unique hydrogen bonding arrangement. It may be that a driving force for the formation of the dodecahydrates is a variation on hydrophobic c lathrat ion 1 2 4 » 1 2 5 as well as the cooperativity 1 0 9 of H-bonding networks. C. M(mepp)3 Crystal Structures 3.5 Introduction The importance of the H-bonded water network to the M(dpp) 3 structure was stressed in the preceding sections; because of the constraints inherent in such a rigid structure, it was felt that even a small variation in the ligand might be sufficient to alter the nature of the water network or to preclude its formation altogether. In the packing arrangement of tris(3-hydroxy-l-butyl-2-pyridinonato)iron(III) there are hexagonal water rings H-bonded to the chelating oxygens. 1 0 7 There are no bridging waters and no water channels: one water ring is sandwiched between two metal complexes (forming discrete units) and alternate waters in the ring are H-bonded to the chelating hydroxyl oxygens (the carbonyl oxygens do not act as H-bond acceptors) of one of the metal complexes. As soon as the first M(dpp)3 structure was determined, we wondered if the water network was 84 unique to the Hdp  ligand; this very interesting Fe structure piqued our curiosity as to the water arangements that might be posible with the other 3-hydroxy-4pyridinone ligands. Eforts to grow more crystals culminated in the determination of the structures of Al- and Ga(mep)3. The crystal growing atempts with the other metal complexes had met with litle suces and the M(mep)3 complexes were synthesized specificaly for this structural study. The crystals were grown by slow evaporation (over a period of weks) from dilute water solutions. Crystal growth was much slower than with the M(dp)3 complexes and the single crystals were considerably smaler. The replacing of a methyl with an ethyl group is, of course, a very minor structural change; however, this smal change was suficient to alter significantly the packing arangement of the fre ligands and it was thought a similar outcome might be posible with the tris-ligand metal complexes. When the Al(mep)3 crystals analyzed as a decahydrate (the analytical sample was prepared analogously to the previously mentioned Al(dp)3 sample), we suspected we were dealing with a similar structure. The detenination of the crystal structure proved that the water network was able to acomodate the N-ethyl group: the empirical formula was Al(mep)3«12H20, the space group* was trigonal P3, and the water network was unchanged.* The Ga(mep)3 crystal structure showed that a change in metal ionic radi in conceit with N-ethyl substitution was also insuficient to alter the water network. The Al and Ga complexes were esentialy isostructural; the modifier "esentialy" was necesitated by the same crystalographic inequality that distinguished the In(dp)3 complex from its Al and Ga analogues. For this reason, the In(dp)3 packing arangement was not adresed previously and the nature of the inequality wil be examined by a comparison of the unit cel packing in the Al- and Ga(mep)3 complexes. * The crystallographic data for the M(mepp)3 complexes are in the Appendix as Table A.4. ** Because there was no significant differences from the water network in the M(dpp)3 structures, the M(mepp)3 H-bond parameters are in the Appendix as Table A.6. 85 3.6 Results and Discussion The/ac-M(mepp)3 units of the A l and Ga complexes are isostructural (Fig. 3.12), as is readily apparent by a comparison of the bond lengths and angles in Table 3.9. The C(5)-C(6) bond is significandy longer in the A l complex (by 0.02 A) and that is the only difference in ring bond lengths. There appears to be an increase in delocalization compared to the free ligand but this is primarily due to change in only one bond, C(5)-C(6), that was the shortest for any of the structures solved (free ligands and metal complexes) in Hmepp and was the longest in Al(mepp)3~ 1.350 vs. 1.380 A. There is more delocalization in the C - 0 bonds than in either Hmepp or the M(dpp)3 complexes. The difference between the two C-O bonds is 0.099 A in Hmepp, and 0.020 and 0.024 A in A l - and Ga(mepp)3 respectively. This leads to the same conclusion as was reached with the M(dpp)3 complexes; i.e., an increase in double bond delocalization occurs upon chelation. Aside from the obvious difference in the N-alkyl groups, the /<ac-M(dpp)3 and /ac-M(mepp)3 units are isostructural. The ligand bond lengths and angles are virtually the same (compare Tables 3.4 and 3.9). The chelate ring angles show a similar compression along the C3 axis and the M-O bond lengths are comparable. In the water network, some of the H-bonds are slightly longer but, as was the case with the increased size of the metal ion, there is no evidence from the H-bond parameters that the water network is significantly affected by the additional methylene group in the M(mepp)3 complexes. 86 Figure 3.12. ORTEP view of the tris(ligand) portion of the M(mepp)3 complexes. 87 Table 3.9. Bond parameters for M(mepp)3«12H20 complexes. The bond lengths (A) are in the upper portion and bond angles (deg) are in the lower portion. For comparison, the parameters for the Hmepp ligand are in the last column. Atoms M = A l Ga Hmepp M - O ( l ) 1.894 (1) 1.962 (1) M - 0 ( 2 ) 1.930 (1) 2.00 (1) 0(1) - C(3) 1.317 (2) 1.327 (2) 1.357 (3) 0(2) -C(4) 1.297 (2) 1.303 (2) 1.258 (3) N - C(2) 1.373 (2) 1.376 (3) 1.378 (3) C(2)-C(3) 1.388 (2) 1.387 (3) 1.370 (3) C(3)-C(4) 1.424 (2) 1.422 (3) 1.433 (3) C(4)-C(5) 1.399 (2) 1.396 (3) 1.410 (3) C(5)-C(6) 1.380 (3) 1.360 (3) 1.350 (4) C(6) - N 1.342 (3) 1.349 (3) 1.351 (3) C(2 ) -C( l) 1.494 (2) 1.485 (3) 1.490 (4) N - C(7) 1.485 (2) 1.488 (3) 1.485 (3) C(7) - C(8) 1.504 (3) 1.500 (4) 1.484 (4) O(l) - M - 0(2) 84.23 (5) 83.03 (6) o(i) - M - o(iy 90.46 (6) 90. 52 (6) O(l) - M - 0(2)' 94.87 (5) 95.29 (6) O O - M - O d W , 172.50 (5) 171.34 (5) 0(2) - M - 0(2)' 90.93 (6) 91.79 (6) M - O(l) - C(3) 112.1 (1) 111.2 (1) M - 0(2) - C(4) 111.3 (1) 110.7 (1) 0(1)-C(3)-C(2) 124.0 (2) 122.4 (2) 118.3 (2) 0(1)-C(3)-C(4) 115.5 (1) 117.0 (2) 118.8 (2) 0(2)-C(4)-C(3) 116.2 (1) 117.5 (2) 121.3 (2) 0(2)-C(4)-C(5) 125.6 (1) 124.9 (2) 124.3 (2) C(6) - N - C(2) 121.6 (1) 120.9 (2) 120.5 (2) N - C(2) - C(3) 118.8 (2) 118.8 (2) 118.6 (2) C(2)-C(3)-C(4) 120.5 (1) 120.6 (2) 122.9 (2) C(3)-C(4)-C(5) 118.1 (1) 117.6 (2) 114.3 (2) C(4)-C(5)-C(6) 119.2 (2) 120.1 (2) 121.9 (2) C(5) - C(6) - N 121.8 (2) 122.0 (2) 121.8 (2) N - C(2) - C(l) 120.5 (2) 120.2 (2) 120.0 (2) C ( l ) - C ( 2 ) - C ( 3 ) 120.7 (2) 121.0 (2) 121.5 (2) C(6) - N - C(7) 117.4 (2) 118.0 (2) 117.3 (2) C(2) - N - C(7) 121.0 (2) 121.1 (2) 122.0 (2) N - C(7) - C(8) 112.6(2) 112.5 (2) 111.8 (2) 88 Table 3.10. Unit cell dimensions for M(dpp) 3 and M(mepp)3 complexes. Dimension M(dpp)3 M(mepp)3 M = A l Ga In A l Ga a (A) 16.600 (2) 16.6549 (6) 16.842 (1) 17.1734 (8) 17.247 (1) c(A) 6.877 (1) 6.8691 (4) 6.8078 (7) 6.827 (1) 6.830 (2) Volume (A3) 1641.3 (3) 1650.1 (1) 1672.3 (2) 1743.7 (3) 1759.4 (1) D c @ (g/cm3) 1.33 1.47 1.48 1.33 1.40 @ D c = calculated density. There are some differences in the unit cell dimensions (Table 3.10) of the structures that were determined. (In the trigonal space group P3, a =b * c, a = P = 90° and y= 120°.) The rigidity of the water network that was responsible for the compression of the M-O-C bond angles along the C3 axis (parallel to the c- axis) can be seen in the length of c. For the M(dpp)3 complexes, this dimension decreases (slighdy as the metal radius increases) to a minimum for In; the increase in the size of the N-alkyl substituent likewise causes a decrease in this parameter, e.g. 0.050 A between Al(dpp)3 and Al(mepp)3- It is along the ab diagonal that the changes in the metal and ligand are accommodated by the water network. The result is an increase in a:, e.g., a 0.242 A increase for In- vs. Al(dpp) 3, and a 0.573 A increase for Al(mepp)3 vs. Al(dpp)3. The packing diagram of Al(mepp) 3 (Fig. 3.13) shows how the N-ethyl group fits into the hydrophobic core that is made up of one ligand from each of the two/ac-M(mepp)3 units in the cell. The flexibility of the structure in the ab diagonal is due to the large ring that encircles the core and consists of bridging waters, one side of the hexagonal water 89 channels, and the M-O bonds in the /ac-M(mepp)3 units (the oxygen atoms have been darkened to highlight this ring). The ligands in the core are separated by 3.5±1 A; i.e., normal van der Waal's contacts. The points where the ligand carbon atoms approach the water rings, C(8)-0(5) and C(6)-0(6), are likewise separated by ~3.5 A. To maintain this distance from the water network, the N-ethyl group must twist out of the ligand plane. The length of c (> 6.8 A) is considerably greater than van der Waal's contact distances so there is room to accommodate the N-ethyl group without an increase in c and, therefore, without an increase in the length of the H-bonds that determine this dimension. Focussing on this inner core and its encircling ring, it is possible to see why the increase in metal radius did not disrupt the water network. Increasing the length of the O-M-0 portion of the inner ring pushes the water channels away from each other (increases a) but no strain is put on the water channels because they are connected via the /ac-M(mepp)3 unit rather than directly connected by H-bonds. The structure can be thought of as four rigid water columns held together by the more flexible /ac-M(mepp)3 units. A n increase in the size of the metal simply pushes the water columns apart in the a and b directions, and in the one direction where direct strain could be placed on water H -bonds, the octahedral metal complexes are compressed before the H-bonds are stretched. The crystallographic inequality that distinguishes In(dpp)3 from its A l and Ga analogues also separates the M(mepp)3 complexes from each other. A comparison of the packing diagrams of A l - and Ga(mepp)3 clearly indicates that the unit cells are not equivalent. There is a rotation of the/ac-M(mepp)3 units by ca. 60° about the C 3 axis with respect to the water network. Each unit cell contains one A and one A stereoisomer; this was necessary for the ligands in the hydrophobic core to lie parallel to each other and, therefore, to define the dimensions of the core. In order for the ligands to be as close as possible, the rings are staggered so the C(l) methyl groups point away from each other. In the unit cell of Al(mepp)3 (see Fig. 3.13), the A isomer is in the upper left of the cell and 90 the methyl groups project into the plane of the paper, the A isomer is in the lower right and the methyl groups are toward the viewer. Figure 3.14 shows the packing in the unit cell of Ga(mepp)3 and here the position of the stereoisomers is reversed. This was referred to as a crystallographic inequality simply because it only involves the unit cells and does not alter the/ac-M(mepp)3 units or the interaction of the water rings with these units. It is possible this alteration of the packing arrangement is caused by the size of the metal as for both ligands, the change occurs in the tris-ligand complex with the largest metal; i.e., In(dpp)3 and Ga(mepp>3. However, it is not readily apparent why this rotation of the ligand orientation occurs and it must be considered simply a crystallographic oddity. 91 Figure 3.13. ORTEP view down the c axis of the unit cell packing of Al(mepp)3. 92 93 Chapter IV N M R Studies A. Aluminum-27 NMR Spectroscopy 4.1 Introduction The 2 7 A l nucleus has a nuclear spin of 5/2 and, therefore, it has a nuclear quadrupolar moment (Q). When a quadrupolar nucleus is placed in a magnetic field, the N M R energy levels are perturbed by quadrupolar effects. This quadrupolar interaction is affected by the direction of the electric field gradient that is fixed by the molecular framework. Thus the quadrupolar energy can be modulated by the Brownian motion of the molecule and if this occurs at the proper rate, spin-lattice relaxation (Ti) wil l be induced. Because the molecular motion is random, it has random phase and this leads to the loss of phase coherence between nuclei, i.e., spin-spin relaxation (T2). The result is an efficient magnetic relaxation mechanism dominated by nuclear quadrupole relaxation. 128,129 The electric field gradients are generated by ligand field asymmetry so there is a direct connection between the line width of the 2 7 A 1 N M R signal and the geometry of the coordinated ligands or coordinated solvent molecules. Octahedrally solvated A l 3 + , tetrahedral AIX4", and AI2X6 dimers exhibit high symmetry and have relatively narrow line widths. For trigonal species, the line widths (measured as peak widths at half height, W1/2) become much larger. The overall range of line widths is from 3 Hz (for [A1(H20)6] 3 +) to over 6000 Hz .1 3 0 There have been a number of studies using 2 7 A l N M R line widths as a probe to determine the coordination number of organoaluminum complexes 94 in organic solvents. 1 3 1 A far lesser amount of data has been collected in aqueous solution for A l complexes with biologically active l igands. 1 3 2 " 1 3 4 The 2 7 A l isotope is 100% naturally abundant and has a relative receptivity one-fifth that of the proton. Comparison with the 1.11% natural abundance and 0.016 relative receptivity of 1 3 C indicates why 2 7 A 1 N M R does not require the instrument time typical of 1 3 C N M R . The practicality of 2 7 A l N M R can best be illustrated by comparing its nuclear properties to those of its congener Ga and of two other quadrupolar nuclei, 1 7 0 and 1 5 N , that are used as N M R nuclei. The data in the last two columns of Table 4.1 were calculated from the following equation for nuclear spin quadrupolar relaxation (TQ) that is valid in the Umit of fast motion. 1 2 8 1 _ 1 _ 1 _ 3 21 + 3 /eQ\2 ' a 7 ~ - 2 TQ " T I " T 2 " 40 r 2 (21 - 1) I R J ( d z 2 ) 1 I is the nuclear spin quantum number, eQ is the electric quadrupole moment, h is Planck's d 2 v constant divided by 2%, -j-^ is the maximum electric field gradient at the nucleus, and T c is the correlation time for Brownian motion. The line width factor, defined as (21 + 3) Q 2 I 2 (21-1) can be used to compare the line widths of quadrupolar nuclei. Normalized to the value for 2 7 A l , the relative line widths of the Ga isotopes are the largest and the other two quadrupolar nuclei have peaks < 10% that of 2 7 A l . However, when the differences in abundance and receptivity are used to determine the relative peak heights, defined as abundance * receptivity line width factor 2 7 A l N M R has a hypothetical sensitivity an order of magnitude (or more) greater than the other quadrupolar nuclei in Table 4.1. As a result of this, high resolution 2 7 A l N M R has been used to quantify A l 3 + to 1 ppm (37 p M ) . 1 3 5 95 Table 4.1. The N M R properties of several quadrupolar nuclei (refs. 128 and 129-Appendix 2) Isotope I Isotopic abundance (%) N M R frequency (MHz) Relative1 receptivity Quadrupole Relative2 moment line (10- 2 8m 2) width Relative3 peak height 27A1 5/2 100 26.08 0.206 0.149 1.00 54.6 69Ga 3/2 60.4 24.04 0.042 0.19 6.7 1.00 7 1 G a 3/2 39.6 30.55 0.057 0.12 2.6 2.24 1 7 0 5/2 0.037 13.56 0.00011 -0.026 0.03 0.003 14N 1 99.63 7.23 0.001 0.01 0.07 3.8 l H 1/2 99.98 100.0 1 — — — 1 Relative (to lH) receptivity at a constant field with equal numbers of nuclei. 2 Relative (to 2 7 A l ) line width. 3 Relative (to 6 9 Ga) peak height. We did attempt some studies with 7 1 G a N M R , but its lack of sensitivity (see Table 4.1) coupled with the low solubility of our Ga complexes made it impossible to detect any signals in a practical time span. 2 7 A l N M R served as a characterization technique and it was also used to readily determine the success of our tris-ligand A l complex preparations. The constraints on instrument access unfortunately restricted this latter application. Our primary utilization of this N M R technique was the determination of the hydrolytic stability of the tris(3-hydroxy-4-pyridinonato) A l complexes. 2 7 A 1 N M R has been used extensively to examine the pH dependent speciation of A l , 1 3 6 and the chemical shifts and narrow line widths of the predominant A l species at pH < 4, [A1(H20)6] 3 + , and at pH > 9, [Al(OH)4]", are quite different from those of the tris-ligand complexes in this study. When a solution of an A l complex was sampled at a 96 series of p H values from < 2 to > 10, the 2 7 A l N M R spectra readily indicated the pH region where the hydrolysis products first appeared. The broad peaks of our complexes made it difficult to quantify by this technique, but it did paint a clear picture of what we referred to as the "window" of hydrolytic stability for Al(ma)3. 3 7 Similar A l hydrolysis experiments using 2 7 A 1 N M R have been carried out with the tris-ligand A l complexes of acetylhydroxamate, 1 3 7 oxalate, 1 3 8 lactate, 1 3 9 and a number of hydroxycarboxylate ligands. 1 3 4 The closest comparison to the spectra of the tris(3-hydroxy-4-pyridinonato) A l complexes were the spectra of Al(ma)3. 4.2 Materials and Methods The 2 7 A l N M R spectra were recorded at ambient probe temperature (ca. 18 °C) on a Varian XL-300 N M R spectrometer at 78.16 M H z and a pulse width of 15 p.s. The first experiments used a sweep width of 20 K H z and an acquisition time (T a q ) of 0.20 s. 6500 Transients were needed to achieve a reasonable signal to noise ratio (S/N) and the recording of one spectrum required 22 minutes. Because of the broad lines, we felt the resolution could be reduced in order to complete one variable-pH experiment (involving ten or more samples) in the two hour time blocks available on this instrument. Combinations of sweep width (20 to 60 KHz) and T a q (0.05 to 0.40 s) were tried before settling on 50 K H z and 0.12 s. These acquisition parameters allowed the resolution of both the narrow signals from the hydrolysis products and the broad signals from the Al-ligand complexes. If the sample concentrations were > 10 m M , 3500 transients gave a good S/N and the 7 minute run time per sample was compatible with the limitations on instrument availability. The spectra were referenced to the [A1(H20)6]3 + signal (set as zero) from 0.20M A1(C1C>4)3 in 97 0.10M HCIO4. D2O was added for locking and downfield chemical shifts were positive. The spectra were recorded by the author. Purified compounds, distilled and deionized water, and a Fisher Accumet model 805 pH meter (calibrated with pH 4 and 10 buffer solutions) were used in the variable-pH experiments. Solutions of the A l complexes were made in 15 mL H2O and 2 mL D2O (for a lock signal) and the initial pH was adjusted to < 2 by the addition of 8 M HC1. The pH was raised by the addition of 8 M NaOH and the solution was equilibrated between pH changes for > 10 minutes. Aliquots were withdrawn and filtered through glass fibers into the N M R tube. The sampling was initiated at the lowest p H and was repeated at approximately one pH unit intervals up to pH > 10. The hydrolysis was completely reversible within the 2-3 hour time frame of the experiment. 4.3 Results and Discussion The chemical shift range of the 2 7 A l nucleus is approximately 450 ppm and, like the line widths, the shift is characteristic of the ligand symmetry about the A l atom. The signals for hexacoordinate A l nuclei are quoted as occurring from 20 to -46 ppm and A1C>6 species usually appear close to 0 p p m . 1 3 0 There are exceptions to this rule: the tris(hydroxamato) A l complexes and the alumichrome trihydroxamato peptides resonate at 36-42 p p m 1 3 7 and a dimeric acetate species gives a signal at 38 p p m . 1 3 6 These exceptions have led to the proposal that downfield shifts near 38 ppm could be characteristic of chelation by small-ring-forming bidentate ligands. 1 3 6 The tris(3-hydroxy-4-pyronato) A l complexes have chemical shifts of ca. 39 ppm and the tris(3-hydroxy-4-pyridinonato) A l complexes resonate at essentially the same frequency. The downfield shifts are due to the 98 inequivalence of the chelating hydroxy and carbonyl oxygen atoms. The A l nuclei are not subject to a rigorously octahedral field and this is also reflected in the relatively broad peaks of these tris-ligand A l complexes. The chemical shifts and the W1/2 values of the tris(3-hydroxy-4-pyridinonato) A l complexes are listed in Table 4.2; the data are from the variable-pH experiment spectra of the p H 7-8 samples. Al(mhpp)3 is not sufficiently water soluble for the variable-pH experiment and its spectrum was recorded in CD3OD. The values for Al(ma)3 are included for comparison and to put these data in perspective, Al(acac)3 (D3 point group) resonates at 0 ppm with W1/2 = 100 H z . 1 3 0 Table 4.2. 2 7 A l N M R data for the tris(3-hydroxy-4-pyridinonato) A l complexes. > Al(ma)3 Al(mpp)3 Al(dpp)3 Al(mepp)3 Al(mimo)3 Al(mhpp)3 ppm 38 38 37 37 37 38 W i / 2 ( H z ) 900 600 700 780 1600 1400 Our other data (spectroscopic and crystallographic) indicate the A106 coordination sphere is very similar for all the 3-hydroxy-4-pyridinone ligands; it is not likely that differences in ligand symmetry would account for the variations in line widths that were observed. N M R line widths are also affected by temperature, solvent viscosity, exchange processes, and the mass of the solute. 1 2 9 The proton N M R indicate an exchange process is occurring at room temperature in the A l complexes and this could contribute to the variations in line widths. In comparison to Al(mpp) 3, the larger R groups in Al(mimo)3 and Al(mhpp)3 may be decreasing the molecular tumbling rate; this would increase the T c for Brownian motion and, as indicated in the above equation, the relaxation rate (and therefore the W1/2) for the larger tris-ligand A l complexes would increase. 99 After Subtraction Before Subtraction Solvent Blank I i i I i I i i I i I i I I I I i M I I I i N I M i i | I I i I | i I I I | '200 150 100 50 0 -50 -100 -150 PPM -200 Figure 4.1. 2 7 A l N M R spectra of Al(mpp)3 at pH 1.8 showing the background correction. 100 The minimum concentration for the variable-pH studies was 10 m M ; this cutoff point is not due to the sensitivity of the technique but rather is due to the background signal from the A l in the ceramics of the N M R probe. This is a well documented phenomenon 1 4 0 - 1 4 1 and the probe A l has a broad (~6000 Hz) signal centered at ca. 60 ppm. The background signal from the probe overlaps the signals from the tris-ligand A l complexes (near 40 ppm). The probe signal is also out of phase with the signals from the solution A l nuclei which makes it difficult to properly phase the spectrum. It is possible to do a background correction by subtracting the free induction decay (FID) of a solvent blank from the sample FID and Fourier transforming the resulting signal. Figure 4.1 depicts the before and after spectra of a p H 1.8 Al(mpp)3 sample with an initial A l concentration of 15 m M . (The sharp signal at 0 is from [A1(H20)6]3+ and the smaller broad peak at 17 ppm is from [Al(dpp)(H20)4] 2 +.) This background correction works well but the experimental procedure presented an additional complication. As the pH was raised, the composition of the solution changed due to the addition of base and this was hard to duplicate in the solvent blank. If the conductivity of the solvent blank was sufficiently different from that of the sample, it effectively detuned the probe so the phase shifts were altered and the subtraction did not work. This problem was handled by making several solvent blanks and using the one that gave the best results for a particular sample. The interference from the background signal can be avoided by using A l concentrations > 20 m M . For Al(mepp)3 and Al(mimo)3,40 m M solutions were used; Al(mpp)3 is soluble to 16 m M and background subtraction was necessary in a number of the spectra. Al(dpp)3 is only soluble to 1.2 m M at neutral pH, but.we were able to make a supersaturated solution with an initial A l concentration of 30 m M at pH 1.8. The insolubility of the tris-ligand A l complex is due to the incorporation of water molecules (see Chapter HI) and this is not an instantaneous process. There was no appreciable precipitation in the base solution over the two to three hours required for the variable-pH experiment. The N M R spectra were recorded immediately after sampling to avoid precipitation in the N M R tube. (Crystals did form in the tubes 101 overnight and it was this experiment that first indicated X-ray diffraction grade crystals of this compound could be grown from water.) The aqueous solubility of Al(mhpp)3 is similar to that of Al(dpp)3 and a concentrated solution can be made in acidic solution. However, its low solubility is due to the lipophilic N-hexyl groups and the variable-pH experiment was not feasible because a pervasive precipitate formed at ca. p H 2.5. (A detailed summary of the solubility properties of these complexes is in Chapter V.) The variable-pH 2 7 A l N M R spectra for Al(mpp)3 are shown as Figure 4.2. The acidic hydrolysis is evinced by a shoulder at pH 3.4 on the upfield side of the Al(mpp)3 peak at 37 ppm. This results from the partial protonation of ligands in the coordination sphere of the A l and their replacement with waters. At p H 3.1 this shoulder is resolved into two shoulders resulting from the [Al(mpp)2(H20)2]+ (28 ppm) and [Al(mpp)(H20)4] 2 + (17 ppm) species. When the solution is acidified further, the completely hydrated species [A1(H20)6] 3 + is observed at 0 ppm. As the pH drops the signals from the hydrolysed species gain in intensity at the expense of the Al(mpp)3 peak. Even at the lowest pH of 1.8, there is still a signal from the monoligand species although most of the A l is in [A1(H20)6] 3 +. When the p H is raised, the ligands are replaced by the hydroxide ion to form ultimately [Al(OH)4]" at pH 11 (80 ppm, W1/2 = 60 Hz). The basic hydrolysis of A l 3 + is time dependent and this explains why no intermediate species analogous to those formed at acidic pH were found in the time frame of the experiment. The Al(mpp)3 pH 8.9 sample was left (in an N M R tube) at room temperature for two weeks and the spectrum recorded at that time is contrasted to the initial spectrum in Figure 4.3. After two weeks, the intensity of the Al(mpp)3 signal is reduced by one-third and the broad peak at 61 ppm is probably due to four coordinate mixed aquo/hydroxo/mpp A l complexes. In the basic hydrolysis of tris(lactato)aluminum(III), mixed hydroxo/lactato complexes are ascribed to a broad peak at 60 p p m . 1 3 9 102 i i i | i i i i | i i i I | i i i i | i i i i | i i i i | i i i i | i i i i | i 2 0 0 1 0 0 0 - 1 0 0 p p m Figure 4.2. The variable-pH 2 7 A 1 N M R spectra of Al(mpp)3. 103 —I I . | I J I I j I I i ! | 1 i I I | 1 M I | I . I I j i I I I | i I I I | 1 200 100 0 .100 ppm Figure 4.3. 2 7 A l N M R spectra ofAl(mpp)3 at pH 8.9. The bottom spectrum is 2 hours after sampling and the top spectrum is the same N M R sample 2 weeks later. A comparison of the variable-pH spectra of Al(mpp)3 with those for Al(ma)3 indicates that the tris(3-hydroxy-4-pyridinone) complex is more stable to both acidic and basic hydrolysis. In the spectra of Al(ma)3, the signal from [A1(H20)6] 3 + first appears at pH 3.2 and Al(OH)4" is evident at pH 9; the spectra of Al(mpp)3 show no signals from either of these species at similar pH values. The difference is particularly evident at higher pH where the [Al(OH)4]" signal does not occur until pH 10.8 in Al(mpp)3. The Al(dpp)3 spectra (Fig. 4.4) are similar to those of Al(mpp)3: the intermediate species occur at 26 ppm ([Al(dpp) 2(H 20)2j +) and at 14 ppm ([Al(dpp)(H20)4]2+). At pH 2.3 there appears to be significantly more of the bisligand species as the two peaks at 26 and 14 ppm are resolvable, unlike in the Al(mpp) 3 spectra. The only significant change in 104 the Al(mimo)3 spectra (aside from broader peaks) is a shift in the window of hydrolytic stability toward lower p H (Fig. 4.5). It must be stated that it is somewhat unfair to compare the spectra of the A l complexes in this study. We are observing a competition for the A l ion among water, hydroxide ions, and the 3-hydroxy-4-pyridinone ligands; the concentrations of the first two are fixed at any given pH but the ligand concentrations were different for each complex studied. It is possible to conclude that the tris(3-hydroxy-4-pyridinonato) A l complexes are more stable to hydrolysis than Al(ma)3 on the basis of this experiment, however, since the initial Al(ma)3 concentration was larger (50 mM) than any A l concentration used in this study. The first variable-pH experiment with Al(mepp)3 (Fig. 4.6) gave spectra that showed a second hydrolysis product at pH > 9.0 (62 ppm, W1/2 = 50 Hz). This narrow peak was likely due to the polymeric [A104Ali2(OH)24(H20)i2] 7 + (commonly referred to as " A I 1 3 " ) species whose symmetric tetrahedral AIO4 core resonates at 62.5 p p m . 1 3 1 A I 1 3 is the principal 2 7 Al N M R detectable base hydrolysis product that is formed when solutions containing only [A1(H20)6] 3 + are neutralized by the addition of 2.5 equivalents of base. The polymer is unstable at the acidic p H where it forms and it can undergo slow transformations to give other less symmetric species that have much broader signals. The appearance of the spectrum is very different depending on the speed of the hydrolysis and this added to the difficulty in determining the precise nature of the A l hydrolysis reactions. 1 3 6 Given the similarity of the 3-hydroxy-4-pyridinone ligands, it seemed unlikely that Al(mepp)3 was undergoing hydrolysis in a significantly different fashion from the other complexes where no A I 1 3 signals were observed. In the other variable-pH experiments, the isolated tris-ligand A l complexes were used as the starting material, but in this experiment A1(NC>3)3 and the free ligand (in stoichiometric ratio) were used. It appears there was a slight excess of A l that was hydrolysed to the A l 13 polymer, as the length of the 105 experiment was short enough to ensure this would be the preferred hydrolysis product. This was confirmed by repeating the experiment with purified Al(mepp)3 as the starting material: there was no AI13 signal in the spectra at pH 9.6 and the addition of A1(NC»3)3 resulted in the appearance of the 62 ppm peak that subsequently disappeared when an excess of the free ligand was introduced. The pH region where the AI13 signal was found is not as readily explained. The AI13 species forms at ca. pH 4 and at pH > 8 it should disappear as, in the absence of other ligands, [Al(OH)4]" is the dominant A l hydrolysis product at higher p H . 1 3 6 The lack of a signal at lower pH could be ascribed to an exchange with the Al(mepp)3 complex, but the persistence of the signal (there was no significant change in the spectrum after one hour) at higher p H is something of a mystery. The AI13 species was seen in hydrolysis experiments with tris(lactato)aluminum; there was a weak signal at 62 ppm that appeared at pH 7 and was completely gone by pH 10 where only [Al(OH)4]" was present. 1 3 9 No signal due to AI13 was found in a pH controlled study of a series of hydroxy carboxylic acids even at 1:1 ligand to A l ratios. 1 3 4 The robustness of these A l complexes is evident from their wide window of hydrolytie stability, from pH < 4 to > 9. They appear to be more stable to hydrolysis than Al(ma)3 and this is corroborated by the formation constants (see Chapter V) as the log P3 for the tris(3-hydroxy-4-pyridinonato) A l complexes is ~32 compared to a log P3 of 22.5 for Al(ma)3. 5 9 The variable-pH experiment with Al(mepp)3 directly illustrates the ability of the 3-hydroxy-4-pyridinones to compete with the hydroxide ion. The results of this variable-pH 2 7 A 1 N M R study indicate the tris(3-hydroxy-4-pyridinonato) A l complexes should resist in vivo hydrolysis except in the highly acidic conditions of the stomach. 106 Figure 4.4. The variable-pH 2 7 A 1 N M R spectra of Al(dpp>3. 107 M I j I I I I ) I I I I J I I I I ) I I | | I | | | i I | 1 1 | • | | , i . , wo ido i -lboppm Figure 4.5. The variable-pH 2?A1 N M R spectra of Al(rrumo)3. 108 1 I 1 1 I I 1 I | 1 I I M I I 1 I I I I I 1 I I I I I I I 1 1 I I I I I M I 2<Jo ldo o -iboppm Figure 4.6. Variable-pH 2 7 A 1 N M R spectra of a 3:1 Hmepp to A1(N03)3 rnixture. 109 B. Variable-Temperature Proton NMR Spectroscopy 4.4 Introduction In tris-ligand metal complexes, asymmetric bidentate ligands may assume a facial (fac) or a meridional (mer) geometry. The fac and mer geometric isomers* are each enantiomeric pairs of A and A stereoisomers. (Figure 4.7 is a generalized diagram illustrating the four isomers.) The rearrangement reactions are geometric isomerization and racemization; these reactions can occur separately or simultaneously depending on the mechanism of the rearrangement. Three types of mechanisms have been proposed: (1) the complete dissociation of one ligand to give a four-coordinate intermediate, (2) the rupture of one metal-ligand bond to give a five-coordinate intermediate, and (3) twisting processes (Bailar or RSy-Dutt twists) that do not require the cleavage of any metal-ligand bonds. 1 4 2 The majority of research on the rearrangement reactions of tris-ligand metal complexes has been directed towards differentiating between the two mtramolecular mechanisms. fac-A fac-A mer-A mer-A Figure 4.7. Isomers of tris-ligand metal complexes with asymmetric bidentate ligands. *They are frequently referred to as cis and trans isomers respectively. 110 Gordon and H o l m 1 4 3 define two limiting types of tris-ligand systems based on the rates of the rearrangement reactions. The kinetically "slow" systems are those in which the geometric isomers can be completely separated and partially resolved. The reaction rates 1 4 4 are ? 10"2 s"1 and complexes containing inert metal ions such as C r 3 + and C o 3 + are in this category. The second type is designated "fast"; this means that the intramolecular rearrangements are rapid enough to prevent separation or resolution of the isomers, but they are not so rapid as to disallow isomer detection by N M R spectroscopy or other techniques such as low temperature H P L C . 1 4 5 Tris-ligand A l and Ga complexes fall into the fast category. N M R is particularily suited to the examination of the rearrangement reactions because the fac isomer has a threefold symmetry axis and the mer isomer is asymmetric; the three ligands of xhefac isomer are magnetically equivalent and the chemical shifts of the nuclei on these ligands will be different from that of their inequivalent counterparts in the mer isomer. If the chemical shift differences are large enough, the isomers may be identified and it is possible to measure the rates of isomerization and racemization. In the early 1960s, Fay and Piper used variable temperature 1 9 F N M R to examine the fluxional behavior of tris(trifluroacetylacetonate) (tfac) complexes with a variety of metals including A l , Ga, and I n . 1 4 6 This classic work established the utility of variable-temperature N M R for the study of stereochemically nonrigid inorganic complexes. Proton N M R has been used extensively to study the rearrangement reactions of tris(P-diketonato) metal complexes and the group 13 metals have figured prominently in this w o r k . 1 4 4 - 1 4 7 The fluxionality of tris-ligand A l and Ga complexes with other ligands such as E D T A 1 4 8 and a-substituted tropolonates ( a - R T ) 1 4 9 - 1 5 0 has also been examined by N M R techniques. The tropolone complexes are of particular interest as these bidentate ligands chelate the metal center with the same binding group (the a-hydroxyketone moiety) as the 3-hydroxy-4-pyridinones. Ill The tris(3-hydroxy-4-pyridinonato) A l complexes exhibit complex proton N M R spectra at the ambient probe temperature of the 300 M H z N M R instrument. Under the same conditions, the Ga analogues give the expected first order spectra. The spectra of the A l complexes can be rationalized on the basis of fac-mer isomerization that is slow enough to result in overlapped signals from the geometric isomers. A variable-temperature N M R study of the M(dpp) 3 complexes was conducted to verify that ligand rearrangement is the source of the difference between spectra of the analogous tris-ligand A l and Ga complexes. The results of this study indicate that two exchange processes are occurring and the coalescence temperature (T c) for the higher temperature process is near room temperature for Al(dpp) 3 and is at -9 °C for Ga(dpp)3. The solubility properties of the M(dpp) 3 complexes necessitate the use of protic solvents. The majority of this work was done in C D 3 O D due to the solubility of the M(dpp) 3 complexes and the large temperature range available with this solvent. Ga(dpp)3 is too insoluble in D 2 0 to allow the acquisition of a spectrum in a reasonable length of time. Ga(dpp)3 is quite soluble in (CD 3 ) 2 SO but it does not reach the region of slow exchange above the freezing point (18.5 °C) of this solvent. Therefore, the difference in lability between the two metals was examined by variable-temperature N M R in C D 3 O D . Because of the importance of water interactions to the solid state structure of the M(dpp) 3 complexes, we were also interested in determining if the rate of the rearrangement reaction for Al(dpp) 3 is significantly different in water. The exchange process for Al(dpp) 3 was examined in D 2 0 , ( C D 3 ) 2 S O , and C D 3 O D ; because of the low water solubility of Al(dpp) 3 , the fluxionality of the more water soluble Al(mpp) 3 was also investigated in D 2 Q . 112 4.5 Materials and Methods The spectra were recorded on a Varian XL-300 N M R spectrometer equipped with a variable temperature probe. The thermocouple was calibrated using a methanol calibration standard and was accurate within ± 1 ° over a range of -70 to 60 °C. The spectra in CD3OD and (CD3)2SO were referenced to the solvent peak arid those in D 2 0 were referenced internally to (CD3)2CO. The spectra were recorded by the author. A l - and Ga(dpp)3 and Al(mpp)3 were prepared and purified as reported herein (Section 2.6). In(dpp)3 and Al(ma)3 were synthesized by others in this laboratory (C. Matsuba and M . Finnegan) and were purified by recrystallization. The equilibrium distribution of the Al(dpp)3 geometric isomers in the absence of exchange was determined by computer simulation using the N M R line-shape program D N M R 3 . 1 5 1 4.6 Results and Discussion The spectra of Ga- and Al(dpp)3 in C D 3 O D at 18 °C clearly show the differences that prompted this variable-temperature N M R study (Figs. 4.8 and 4.9~the spectral regions without signals have been omitted for clarity). The singlet for the CH3 C group has a W1/2 of 1.6 Hz in Ga(dpp)3 compared to an exchange broadened 7.0 Hz in Al(dpp)3. The downfield doublets in the Al(dpp)3 spectrum are also broadened and a second signal is just starting to appear as shoulders on the Hb doublet. The Al(dpp)3 spectra in (CD3)2SO (Fig. 4.10) and D 2 O (Fig. 4.11), also at 18 °C, afford better resolution of the signals from the fac and mer isomers due to the slightly larger chemical shift difference and the higher coalescence temperature in these solvents. In both spectra, there are four distinct signals from the three inequivalent CH3 C groups of the mer isomer and the one unique methyl of the fac isomer. 113 Hdd s:2 B;Z O;E 2;E r e g;e B;E o;t-e a f B 1 9 8 B a I B 8 • »t > «a a I I t I 1 1 t t I I I i • • »I » t I I I I l f • I l t 11 I I I t t I I • 11 I I I I I I I I 11 I I I I 11 I I 11 I t I I 11 I I 11 I I 1 I 11 1 I o 3 H3 Y CIO(H*CD) £HDN LO o o oo o o Ndd 2-9 fr'9 9*9 B-9 0 7 2 7 f 7 9 7 9 7 I I t I 11 I 1 I I I I I I I I I I I 11 I I I I I I I I 1 I I I I 11 I M I I I I I 11 I t i III I 1111 i t 1 I I I I 11 I 1111 i N 11 1 1 1 1 I I 1 1 1 1 I 1 1 1 I I 1 1 1 1 115 Figure 4.10. 300 M H z proton N M R spectrum of Al(dpp)3 in (CD3) 2SO at 18 °C. Figure 4.11. 300 M H z proton N M R spectrum of Al(dpp)3 in D 2 0 at 18 e C . 117 The chemical shift differences in the downfield doublets are quite small so the H a and H5 ring protons are of limited utility for identifying the geometric isomers in any of the solvents. The same is true of the N -CH3 signal. For this reason, the kinetic parameters were determined from the temperature-dependent C E ^ spectra. Among the four signals, the one that was different in intensity was assigned to the /ac-isomer, this assignment rationale was used by Piper and Fay for the fluorine resonances in the M(tfac)3 complexes. 1 4 6 Figure 4.12 shows the C H ^ signals for A l - and Ga(dpp)3 in the absence of exchange (-30 °C) and the signal assigned to the fac isomer is marked with an asterisk. (The scale in Hertz indicates the magnitude of the peak separation and not the chemical shift) Two of the mer isomer signals in Ga(dpp)3 appear as shoulders, but the assignment of the peak with the odd intensity to the fac isomer has been made with other tris-ligand metal complexes where only two of the four peaks are resolvable. 1 4 9 ji 1111111 iji 111 j 111 \j\ 11111111 y 11111111j JM11111111) 111 ^111111111 y 111111 n Figure 4.12. CH3c spectra for Al(dpp)3 (left) and Ga(dpp)3 (right) in CD3OD at -30 °C.. A statistical distribution of the geometric isomers would give four peaks of equal intensity, i.e., the fac isomer would be 25% of the total concentration. It has been stated that the mer isomer is the more stable isomer due to its lower dipole moment and this was 118 used to explain the smaller than statistical (18%) equilibrium distribution for the kinetically fast Al(tfac)3 complex. 1 4 6 Gordon and Holm maintained that unless the complexes were sterically constrained, a statistical or nearly statistical distribution of isomers would usually be formed in solution. 1 4 3 The A l - and Ga(cc-RT)3 complexes had a slight excess of the ./ac-isomer at equil ibrium 1 5 0 and this was also the situation for Al(dpp) 3. D N M R 3 was used to simulate the equilibrium spectrum of Al(dpp)3 in the absence of exchange and the best fit (Fig. 4.13) was obtained with a 32% distribution for the /ac-isomer. The enhanced stability of the fac isomer was interesting when considered in concert with the solid state structures (see Chapter III). It was thought that the water networks imposed a facial geometry on the M(dpp)3»12H20 complexes, but these N M R results in methanol suggest that the fac isomer enjoys at least a small thermodynamic advantage independent of the H-bonded water network. The close match of the computer simulation also supports the assignment of the larger signal to the/ac-isomer. Figure 4.13. Experimental (left) and simulated (right) Al(dpp)3 CH3c spectra in CD3OD (-30 *C). For the exchange of nuclei between two inequivalent sites, the rate of exchange at Tt Av the temperature of coalescence (kj c) is given by: kx c =—p— where Av is the frequency V2 separation (in Hertz) between the resonance components in the absence of exchange. 1 5 2 119 For the CH3 C spectra, A v e (the experimentally observed frequency separation) was taken as the difference between the signal for the/ac-isomer and the furthest downfield signal from the mer-isomer. A source of error in using this equation is the temperature dependence of the chemical shifts that is presumably due to solvent-solute interactions. 1 5 3 To correct for this temperature dependence, the A v e was plotted against temperature at several points in the region of slow exchange, and the line was extrapolated to the region of fast exchange. Then Av was read from the extrapolated line at the T c . In our systems, the values of A v e were small (5-8 Hz) and the variations due to temperature dependence were barely within the resolution error of the instrument; therefore, the required adjustments in Av were minor. The value of the free energy of activation (AG+Tc) was calculated from the Eyring equation 1 5 4 (assuming the transmission coefficient to be unity): k r c = - j p T c e -AGf/RTc where h is Planck's constant, K B is Boltzman's constant, and R is the gas constant. The simplified Gutowsky-Holm equation 1 5 2 is used to calculate the rate constants for an exchange process from the peak separations in the slow exchange region; the rate constants can then be used to determine the E a , AH^, and AS^ for the exchange reaction. The size of the error in using this approximation is related to the length of T 2 and the magnitude of Av. When Av is small and T 2 is short, the errors become comparable to the calculated rate constants and the peak separation method should not be used. 1 5 5 A value of less than 0.33 for the ratio of T 2 (expressed as W1/2) to Av was cited as the minimum requirement for using the simplified equation. 1 5 6 For the CH3 C spectrum in Al(dpp)3, the T 2 determined by computer simulation is 0.135 s (W1/2 = 2.36 Hz) and the Av is 5.1 Hz. The ratio of W1/2 to Av is 0.46 and the simplified equation could not be used for this complex, or for Ga(dpp)3 either as the CH3 C signal from the Ga complex is even more poorly resolved. 120 The preferred method for estimating rate constants is line-shape simulation. The D N M R 3 computer program allows mutual and non-mutual exchange (referred to in the program documentation as "otherwise"). The non-mutual exchange routine has multiple chemical configurations that can have unequal populations and it is possible to simulate the equilibrium spectrum of Al(dpp) 3 in the absence of exchange with this routine; however, the non-mutual routine cannot directly accommodate exchange between chemical configurations of different symmetry. With the mutual exchange routine, the population difference between the fac- and mer-isomers cannot be incorporated so the larger peak due to the fac isomer is not modelled. Because of these limitations, the D N M R 3 simulated spectra did not correspond well enough to the experimental spectra to merit inclusion in this thesis. Qualitatively, the M(dpp) 3 complexes appear to undergo two exchange processes in a fashion similar to that of the M(cc-RT)3 complexes. 1 4 9 - 1 5 0 This is illustrated by the C H 3 c spectra for Al(dpp) 3 over a temperature range of -90 to 38 °C which are included as Figure 4.14. The spectra for Ga(dpp)3 are similar and Figure 4.15 is a view of the exchange process for all of the protons in the complex (the spectral regions without signals have been omitted in this reproduction). In both M(dpp) 3 complexes there is a low temperature exchange process (LTP) with a T c near the -98 °C low temperature limit of our C D 3 O D study. There is a second exchange process with a T c of 21 and -9 °C in A l - and Ga(dpp)3, respectively. The kinetic parameters at the T c for this higher temperature process (HTP) are listed in Table 4.3. The L T P for the M(oc-RT)3 complexes was identified as racemization by means of a trigonal twist and the HTP was fac-mer isomerization. By comparison to the spectra of the M(cc-RT)3 complexes, the variable-temperature N M R spectra of the M(dpp) 3 complexes support the conclusion that the exchange process near ambient temperature is due to fac-mer isomerization. 122 Figure 4.15. Variable-temperature proton N M R spectra of Ga(dpp)3 in CD3OD. 123 Table 4.3. Kinetic data for tris(3-hydroxy-4-pyridinonato) metal complexes at the T c . Complex Solvent [ec] Av (Hz) T c (°C) a k r ^ s ' 1 ) AG^Tc (kcal/mol)b Ga(dpp)3 C D 3 O D 4.8 -9 10.7 14.2 ± 1.2 Al(dpp)3 C D 3 O D [32.7] 5.1 21 11.3 15.7 ± 1.3 (CD 3 ) 2 SO [40.7] 7.2 38 16.0 16.5 ± 1.4 D 2 O [78.5] 8.0 34 17.8 16.0 ± 1.4 Al(mpp) 3 D 2 O 6.0 36 13.3 16.5 + 1.4 a Errors were estimated to be ± 2 °. b Errors were estimated assuming an order of magnitude error in the rate constants. c Dielectric constants at 25 °C. The M(dpp)3 complexes are more labile than A l - and Ga(tfac)3 that have a AG+TC of 21.4 and 18.7 kcal/mol respectively (in C D C I 3 ) . 1 5 7 Since the rate-accelerating influence of the CF3 groups is well documented, 1 5 8 it would appear that the 3-hydroxy-4-pyridinones constitute a relatively labile tris-ligand metal system. This is only speculation, however, as the contribution from solvent effects has not been taken into account. The rearrangement rates for d° and d 1 0 metals are dependent on the ionic radius of the m e t a l 1 4 4 and the M(dpp)3 complexes exhibit the predicted kinetic order (Al < Ga < In). Ga(dpp)3 has a lower AG+Tc than Al(dpp)3 and the In analogue gives only an averaged spectrum down to -80 "C in C D 3 O D . For a series of aprotic solvents, it was found that the T c and AG^Tc of Al(tfac)3 decreased as the dielectric constant of the solvent increased. 1 5 9 There was no indication of a similar trend in this study and the kinetic parameters for the Al(dpp)3 exchange process were essentially the same in the three solvents that were used (see Table 4.3). The kinetic 124 data for Ni(phenanthroline)3 showed no simple correlation between the racemization rate and the dielectric constant of the solvent (this study included water and methanol). 1 5 8 The results for Al(dpp) 3 are in agreement with this conclusion. In order to obtain mechanistic information on the rearrangement reaction, it is necessary to ascertain that one is dealing with an intramolecular process. This can be accomplished by examining the N M R spectrum of a mixture of the tris-ligand metal complex of interest and another tris-ligand metal complex. If it takes an appreciable amount of time to form the mixed-ligand species, it can be concluded that ligand exchange is slower than the rearrangement reaction, e.g., no N M R signals attributable to mixed-ligand species were detected in one hour at room temperature for a mixture of Al(acac)3 and tris(l-phenyl-5-methylhexane-2,4-dionato)aluminum(III).160 When ligand exchange is completed in a shorter period of time (in minutes at room temperature), the N M R spectrum at a temperature above the T c for the rearrangement reaction can be used to determine the relative rates of the two processes. If multiple signals assignable to the mixed-ligand species are observed at a temperature where the rearrangement process gives an averaged spectrum, it can again be concluded that the intermolecular process of ligand exchange occurs at a slower rate. This is considered sufficient proof that the rearrangement reaction is an intramolecular process. Our interest in the possibility of solvent effects on the exchange process made it worthwhile to qualitatively compare the rates of ligand exchange and ligand rearrangement for Al(dpp) 3 in C D 3 O D . The other tris-ligand aluminum complex used in the ligand exchange experiment was Al(ma)3. The Al-maltol complex was chosen because, unlike the tris(3-hydroxy-4-pyridinonato) A l complexes used in this study, the chemical shift of the ring protons was sufficiently removed from that of Al(dpp) 3 to permit easy differentiation of the signals from the two ligands. 125 The ratio of the N M R sample was 3:1 Al(ma)3 to Al(dpp)3 and Figure 4.16 shows the proton N M R spectrum fifteen minutes after mixing (at a probe temperature of 18 °C). It is evident that mixed-ligand species have already formed and the 2:1 ratio of the signals from the N-methyl group of the dpp" ligand (at 3.82 ppm) suggests the formation of Al(dpp) 2ma and Al(ma)2dpp species. Figure 4.17 highlights the spectra of the H a and Hb ring protons: in the 3:1 mixture, the signals from the maltol in the mixed-ligand species are shifted slightly upfield from those of Al(ma)3 and the signals at the chemical shift of Al(dpp)3 indicate that all of the dpp* ligands are on mixed-ligand species. The T c for Al(dpp)3 in C D 3 O D is 21 °C and as the top spectrum in Figure 4.17 illustrates, the ligand rearrangement rate is slow at 18 °C. However, the signals from the mixed-ligand species are sharp and well resolved at this temperature. This is evidence that the ligand exchange rate is comparable to the rate of ligand rearrangement. To verify this, the 3:1 mixture was cooled (in 10° increments) to -50 °C and the variable-temperature N M R spectra of the Hb proton on the dpp' ligand are reproduced as Figure 4.18. The sharp overlapped doublets of the 18 °C spectrum are significantly broadened at -10 °C (a similar effect is observed for the other signals in the spectrum). At -20 °C the ligand exchange rate is slow enough to permit the observation of signals from the inequivalent dpp ligands on the Al(dpp)3.n(ma)n species. The ligand exchange rate is slower at -30 °C and this can be considered the spectrum in the absence of ligand exchange as it is unchanged at -40 and -50 °C. It is apparent that ligand exchange is occurring at a rate somewhat faster than that of the rearrangement reaction. It is necessary to cool the sample to a temperature comparable to the T c of the more labile Ga(dpp)3 (-9 °C) in order to slow the ligand exchange process enough to observe the signals from all of the mixed-ligand species. 126 Figure 4.16. Proton N M R spectrum of 3:1 Al(ma)3 to Al(dpp)3 in CD3OD at 18 "C. Figure 4.17. Proton N M R spectra of Ha and Hb in Al(dpp)3 (top), Al(ma)3 (middle) and the 3:1 Al(ma) 3 to Al(dpp)3 mixture (bottom). A l l spectra in CD3OD at 18 *C. 128 Figure 4.18. Variable-temperature proton N M R spectra of H5 in the dpp" ligand on the mixed-ligand species (in CD3OD and temperatures are °C). The ligand exchange experiment indicates the rearrangement reaction for Al(dpp)3 in CD3OD is an intermolecular process. The racemization of Ni(phenanthroline)32+ in water proceeds through a dissociative mechanism and the similarity of the activation energies for the process in methanol (and a number of other solvents) is used to support the proposition that the same mechanism is occurring in all of the solvents studied. 1 5 8 By analogy, the similarity of AG+Tc in D 2 0 and (CD3)2SO to the value in CD3OD suggests there is an intermolecular exchange occurring in all of these solvents. It can be concluded on the basis of the variable-temperature N M R experiments that fac-mer isomerization is the reason for the fluxional N M R spectra of the tris(3-hydroxy-4-pyridinonato) A l and Ga complexes . It appears the exchange process is intermolecular and, therefore, the isomerization mechanism likely involves ligand dissociation. In addition, the solubility properties of these A l and Ga complexes make it very unlikely that the mechanistic details of their ligand rearrangement processes could be determined. 129 Chapter V Solution Studies A. Aqueous Solubility 5.1 Introduction During the synthesis of the ligands and the tris-ligand metal complexes, their interesting and at times unexpected aqueous solubilities were duly noted. This led to solid state studies on single crystals of several compounds grown from, and in the case of the metal complexes grown with, water. The crystal packing arrangements were dictated by H-bonding and we wanted to discover if this form of molecular association also affected the solution behavior of these compounds. The accurate measurement of aqueous solubility was one method used to establish the correlation between the solid state and solution properties. Accurate knowledge of the aqueous solubility was also of practical value as regards synthetic and biological applications. The unusual aqueous solubility properties of nitrogen heterocycles were noted as early as 1899 when it was found that the insertion of hydroxyl groups progressively decreased the water solubility of purine. This unexpected result was also observed for 7C-deficient N-heterocycles as exemplified by 4-hydroxy-2-pyridinone which is 160 times less water soluble than either 2- or 4-pyridinone. To explain this, Albert suggested that intermolecular H-bonding from the ring nitrogen to the exocyclic oxygen atom was preferred over H-bonding to water molecules. 1 0 4 The reduced aqueous solubility of the 3-hydroxy-4-pyridinones can be attributed to strong intermolecular H-bonds, but with the exception of Hmpp, the ring nitrogen was not directly involved. In comparison to the free 130 ligands, the interesting feature of the metal-ligand complexes was that their solubilities were governed by solvent rather than intermolecular H-bonding and still the result was a lowered aqueous solubility. The solubility was determined at 25 °C and in an attempt to gauge the relative strength of the H-bonds, the change in solubility upon heating to 37 °C was also measured for several compounds. In keeping with our interest in biological applicability, the measurements were done in isotonic pH 7.4 buffer. The 3-hydroxy-4-pyridinones have a TC->7t* transition (~280 nm) and the solute concentrations were measured by monitoring this band. Since it was employed primarily as an analytical technique (it was also used in the determination of the «-octanol/water partition coefficients), ultraviolet (UV) spectroscopy was included in this chapter rather than with the other spectroscopic techniques in Chapter II. 5.2 Materials and Methods Water was deionized with Barnstead D8902 and D8904 cartridges and distilled in a Corning MP-1 Megapure still. Trizma-7.4 HC1 (tris(hydroxymethyl)aminomethane) is available from Sigma. The isotonic buffer solution was 0.15 M NaCl and 0.05 M Trizma which gives pH 7.42 (±0.05) at 25°C. Solution temperatures (+0.1 °C) were maintained with water-jacketed beakers and a Julabo circulating waterbath. Samples were centrifuged at 3300 rpm in a Centrific Centrifuge model 228. A l l aliquots were withdrawn with Eppendorf digital pipettes, either 10-100 U.L or 100-1000 p L (1-1.5% error). The electronic spectra were measured from 370 to 210 nm with a Perkin Elmer Coleman 124 U V - V i s spectrophotometer (1 cm cell). Trizma absorbs in the U V near the 210 nm instrument cut-off point so it was necessary to use buffer as the reference solution. 131 L-Mimosine was recrystallized from water; the other ligands and metal complexes were prepared and purified as previously described in Chapter JJ. A l l samples were rigorously dried prior to weighing. 5.2.1 Procedure for the Determination of Molar Absorptivity (£) By comparison to other 3-hydroxy-4-pyridinones,161 a reasonable estimate of £ for the free ligands was 12,000 L moHcnr 1 for the selected band (A,max)- Based on this, 1 m M stock solutions (in 100 mL volumetric flasks) were made from accurately weighed compound. Five 10 mL standard solutions were made (5 to 80 p.M for the free ligands, 3 to 40 p M for the metal complexes) that covered an absorbance range of 0.070 to 1.200. Plots of concentration (c) versus absorbance (A) were made and all the compounds obeyed Beer's Law (A = £bc with b = 1 cm) over these concentration ranges. Each plot had at least five points and in some cases a second set of measurements was made and all the points were included in the graph. The values of £ were calculated by linear regression analysis (correlation coefficients > 0.999). The £ values in Tables 5.1 and 5.2 have an estimated error of ± 500 L moHcm _ 1; this is consistent with the propagation of dilution and instrument errors and is < the standard deviations (a) calculated from several of the Beer's Law plots. The instrument error in Xmax is ± 1 nm. 5.2.2 Procedure for the Determination of Aqueous Solubility at 25 and 37 °C Suspensions of each compound in buffer were placed in the 25 °C waterbath, stirred for 30 minutes, and equilibrated for an additional hour to allow the finely suspended 132 solids to settle. Aliquots were withdrawn from the saturated solutions, filtered through glass wool into centrifuge tubes, and centrifuged for five rninutes. This step was necessary because of the large dilutions required to obtain concentrations suitable for spectroscopy; any solids in the final aliquot were redissolved upon dilution and resulted in errors in the calculated solubility. Accurate aliquots were withdrawn from the centrifuged samples and diluted with buffer to the concentrations necessary to give absorbances between 0.500 and 1.500 (the dilution factors ranged from 10 for Ga(mhpp)3 to 16,667 for Hmepp). Four measurements were made for each solution and the average absorbance (A25) and the previously determined £ values were used to calculate the concentrations. Propagation of the weighing, dilution, and instrument errors would allow the reporting of concentrations to three significant figures. The additional systematic errors inherent in solubility studies (such as the effects of small amounts of impurities) and in this procedure (such as the lack of temperature control once the samples were removed from the waterbath) limited our confidence so the concentrations were reported with two significant figures. In Tables 5.3 and 5.4, the concentrations have an estimated error of ±5 in the second digit. The concentrations of M(mpp)3, M(dpp)3, and M(mepp)3 were measured on three separate occasions and the reported values for a (r=3) supported this estimated error. For the absorbance measurements at 37 °C, the suspensions were equilibrated at the higher temperature for 30 minutes and treated as above. The absorbances (A37) were corrected for changes in the dilution factor and the percent change on heating was calculated 133 5.3 Results and Discussion 5.3.1 Ultraviolet Spectroscopy The ultraviolet spectra of pyridine and its derivatives are similar to that of benzene; they have an E (ethylinic) and a B (benzenoid) band above 210 nm that originate from Tt—>TC* transitions. 1 6 2 Due to the lone pair of electrons on the nitrogen, there is a weak n->7t* transition (R-band) that is generally only observable in the vapor phase. The B-band in pyridine is at 257 nm (in water with £ = 2750 L moHcnr1); conjugated and/or electron-donating substituents cause this band to shift to lower energy. 1 6 2 - 1 6 3 O H and CH3 groups cause bathochromic shifts of 18 and 5 nm respectively. 1 0 4 In good agreement with this, the B-band in the N-substituted ligands is at 280 ± 2 nm and is at a slightly shorter wavelength in Hmpp (Table 5.1). Compared to their respective ligands, the B-band in the metal-ligand complexes show a small bathochromic shift (9-16 nm) and the Ga complexes are consistendy at the longest wavelength (Table 5.2). Table 5.1. Xmax (nm) and £ x 10 - 3 (L moHcnr1) for the B-band in the free ligands. Hmpp Hdpp Hmepp Hmhpp Mimosine H2exn A-max 273 278 279 279 282 278 £ 12.4 13.6 14.2 12.4 15.0 26.1 134 Table 5.2. Xmax (nm) and £ x IO"3 (L moHcnr1) for the B-band in the metal-ligand complexes. M(m A l ?P)3 Ga M(d A l 'P)3 Ga M(mepp)3 A l Ga M(m A l 1PP>3 Ga M(mimo)3 A l Ga X £ 285 286 26.4 27.1 290 291 28.4 29.4 292 293 30.7 30.8 293 294 28.8 30.0 291 294 30.2 30.5 For pyridine derivatives, the intensity of the B-band can be greatly enhanced by auxochromic substituents (saturated groups with nonbonded electrons). 1 6 3 This enhancement accounts for the large increase in the £ between the 3-hydroxy-4-pyridinones and pyridine. Increasing solvent polarity also has a marked hyperchromic effect on the B-band attributable to solvent H-bonding with the nitrogen lone pair of electrons and this increase in intensity is at a maximum for pyridinium salts. 1 6 2 The smaller £ values for Hmpp and the M(mpp)3 complexes may be due to a decreased electron density on the ring nitrogen when compared to the N-substituted compounds. This could result in somewhat weaker solvent H-bonds and, therefore, a reduced hyperchromic effect. The differences are small (from 1200 to 4300 L moHcm1) but consistent; the one exception is Hmhpp and it is possible that the hydrophobic N-hexyl group could disrupt the H-bonding sufficiently to account for the lower £ of this ligand. 135 5.3.2 Aqueous Solubility at 25 and 37 °C At 25 °C, Hmpp is less than half as water soluble as Hdpp (Table 5.3). The replacement of a hydrophilic amino with a hydrophobic methyl group increases water solubility and intermolecular H-bonding is the likely explanation for this unusual behavior. The evidence from IR spectroscopy and the crystal structure indicated that the O - H - O hydrogen bonds in Hmpp were weaker than in Hdpp so the decrease in water solubility must be due to the N-H—0=C hydrogen bonds. The strength of the Hmpp H-bonds in solution was shown by the proton N M R experiment in which the intermolecular interactions persisted in (CD3)2SO, a good H-bond acceptor solvent. It is interesting to compare these compounds to 4-pyridinone and 4-hydroxy-2-pyridinone whose water solubility (at 20 °C) is 105 and 0.65 m M respectively. 1 0 4 Despite the presence of two methyl groups, Hdpp has a solubility comparable to 4-pyridinone. This could be due to the relative position of the H-bonding sites and, therefore, the extent of the H-bonded network. When the H-bond sites are ortho, as in Hdpp, it is possible to form dimeric units. When they are para, as in 4-pyridinone, the most likely interaction would be the formation of H-bonded chains that could reduce the relative water solubility. The addition of a second H-bond donor in 4-hydroxy-2-pyridinone presents the possibility of a three-dimensional H-bonded structure, and as with Hmpp, the result is a further reduction in water solubility. The situation is not so straightforward when Hmepp is compared to Hdpp; the substitution of an ethyl for a methyl group increases the water solubility almost sixfold. The increased volume of the unit cell and the appreciably lower melting point for Hmepp compared to that of Hdpp led us to expect the water solubility to be greater, but the magnitude of the difference is surprising. The compounds both form dimeric units and on 136 the basis of IR stretching frequencies, the H-bonds are of similar strength. The large increase in solubility must be due to a weaker crystal lattice in Hmepp; although intermolecular H-bonds are important, they are by no means the only factor in determining aqueous solubility. Table 5.3. Water solubility (mM) at 25 °C. Hmpp Hdpp Hmepp Hmhpp Mimosine H2exn Solubility 38 95 570 5.6 19 0.56 Hmhpp is minimally water soluble and, unlike the other 3-hydroxy-4-pyridinones studied, Hmhpp is soluble in a number of aprotic solvents (including diethyl ether, CH2CI2, and acetonitrile). It is apparent that the N-hexyl group, and not the potential for H-bonding, is the deciding factor in the solubility properties of Hmhpp. The results for H2exn indicate why this compound was difficult to characterize and support the conclusion that the lower water solubility is due to H-bonded polymerization. The low water solubility of L-mimosine, a compound with four H-bonding sites, strongly suggests that intermolecular H-bonding is forming a crystal lattice capable of resisting solvation by water molecules (or by any other solvents short of dilute acids and bases). The increased size of the tris-ligand metal complexes should result in a decrease in water solubility compared to that of the free ligand. However, the water solubility of the free ligands is strongly influenced by their ability to form intermolecular H-bonds via the cc-hydroxy ketone moiety and formation of the metal complexes removes this H-bonding site. It is possible that diminished intermolecular association could offset the increase in 137 size and result in greater not lesser aqueous solubility for the metal complexes. This appears to be the case with mimosine as Al(mimo)3 is five times as soluble as the free ligand (Table 5.4). The substituent zwitterion becomes the dominant factor in determining the water solubility and, given the acidity of the ammonio proton (pKa of 7 in the free ligand97), one would predict appreciable water solubility. Ga(mimo)3 is approximately one-half as soluble as its Al analogue and the other ligands show a similar contrast between their Al and Ga complexes. The reduction in solubility is consistent with the larger size of the metal center, but the magnitude of the difference is not easily rationalized considering that in no instance does the metal account for more than 16% of the mass of the tris-ligand metal complex. Table 5.4. Water solubility (mM) at 25 *C and % change f ^ m ^ S f 1 " ^ x 100 M(m Al 'P)3 Ga M(d Al 'P)3 Ga M(mepp)3 Al Ga M(m Al 1PP)3 Ga M(mimo)3 Al Ga Sol. % 16(1) 7.5(5) 42 20 1.2(1) 0.70(2) 1.2 0.74 19(1) 11(1) 3.3 1.9 0.69 0.20 12 3.6 95 55 500 289 The M(mimo)3 complexes are the exception rather than the rule and the rest of the tris-ligand metal complexes have a much lower water solubility than that of the free ligand. The M(dpp)3 complexes have the maximum change as the solubility of Ga(dpp)3 is < 1% that of Hdpp (see bottom row in Table 5.4). The decrease is not as large in the M(mpp)3 complexes (e.g., 42% for Al(mpp)3); this is probably due to the lower solubility of the free ligand and the availability of the NH group for solvent H-bonding in the metal-ligand complexes. That the M(mhpp)3 complexes are less soluble than Hmhpp is not surprising 138 since there are three lipophilic hexyl groups per molecule instead of one. The large decrease in solubility for the M(dpp)3 complexes results in solubilities on the order of those for the hydrophobic M(mhpp)3 complexes and this is rather surprising. The solid state packing arrangement presents a possible explanation for the low water solubility of the M(dpp)3 complexes. Solid water consists of hexagonal water rings interconnected in a honeycomb arrangement to give a rigid quasi-infinite network (refer to Fig. 3.9). On melting, some of the H-bonds are disrupted and the network of water molecules becomes irregular and less open than in the ice structure. It is generally accepted that there would still be an infinite network linking together many finite discrete networks of varying sizes but the exact structure of the finite units is still very much open to debate. 1 6 4 In the solid state, the M(dpp)3 complexes are able to fit into and in a sense hold together separate water structures. A similar situation could be occurring in aqueous solution; rather than being pulled apart by the solvent molecules, the metal-ligand complexes could be incorporated into the disordered fabric of liquid water to form insoluble polymers of the inorganic complexes and the water columns. The M(dpp)3 and M(mepp)3 complexes have the same solid state packing arrangement and both show similar large decreases in water solubility when compared to the free ligands. But the M(mepp)3 complexes are ca.16 times as soluble as the M(dpp)3 complexes and they are even more water soluble than the M(mpp) 3 complexes despite the obvious difference in hydrophilicity between the N-ethyl and N - H groups. Hmepp is substantially more water soluble than Hdpp for reasons that are not readily apparent and the M(mepp)3 complexes could be forming water polymers that are simply more soluble than their M(dpp)3 analogues. 139 The difference between the M(dpp) 3 and M(mepp)3 complexes is further shown by the change in absorbance found on heating to 37 °C. The experiment was motivated by the large aqueous solubility temperature coefficient that was observed in the synthesis of the M(dpp)3 complexes and was exploited in crystal growing to make the supersaturated solutions from which the single crystals were obtained. Neither the M(mpp)3 nor the M(mepp)3 complexes appear to exhibit this property and the results of the quantitative study confirm the experimental observations. On heating to 37 °C, the absorbance of the M(dpp)3 solutions increases by nearly 200% while the M(mpp)3 and M(mepp)3 solutions show no significant change (A37 is within ± 10% of A25). (For perspective, the absorbance of all the free ligands increased by ca. 30%.) This simple experiment serves to illustrate the unique aqueous solution behavior of the M(dpp)3 complexes. The 3-hydroxy-4-pyridinone ligands synthesized in this study varied comparatively little in size or electronic configuration yet they exhibited a considerable range of water solubilities. The order of water solubility expected for the free ligands was Hmhpp < Hmepp < Hdpp < Hmpp and the order found was Hmhpp < Hmpp < Hdpp « Hmepp. This illustrates the interesting, albeit somewhat unexpected, results that can be obtained by simple N-alkyl substitution. The determination of the aqueous solubilities also made it possible to correlate the spectroscopic and crystallographic studies directly to a property of concern to our goal of developing biologically relevant ligands. 140 B. Octanol/Water Partition Coefficients 5.4 Introduction The distribution of a solute between two immiscible phases in which it is soluble has long been a subject of investigation. The early workers established that the concentrations of the solute in the two phases was constant and independent of the relative solution v o l u m e s . 1 6 5 While the physical chemists strove to develop a theoretical description of the partition phenomena, applications for this physicochemical property developed from the discovery (near the turn of the century) that the relative narcotic activities of drugs was often direcdy related to their oil/water partition coefficient. 1 6 6 In the late 1960's, Hansen and co-workers proposed that the rc-octanol/water system was a good model for the lipoidal biophase in living organisms and, therefore, could be useful in studying the distribution of solutes between blood and lipid in living organisms. 1 6 7 In pharmacological research, the n-octanol/water partition coefficient (P)* has been accepted as the operational definition of lipophilicity and is now widely employed in the design and development of new bioactive compounds. The measurement of log P for the Ga and In complexes was useful in establishing their potential as radiopharmaceutical imaging agents (In has two suitable radioisotopes). The increasing interest in A l chelators for the treatment of various neurological disorders This may be represented as Koctanol/water (ow)> Pow or D o w - Log P is used in the computational methods for the estimation of P and in the numerous quantitative structure-activity relationship (QSAR) studies involving this property; therefore, log P is the most frequently (but by no means exclusivley) encountered symbol for this property. 141 linked to a l u m i n u m 1 5 ' 1 6 8 made it worthwhile to determine the log P values for the free ligands and the tris-ligand A l complexes. A log P of 2 has been proposed as ideal for the design of barbituates, but the complexity of brain uptake makes it impossible to set a lower log P limit in neurological drug design. 2 7 Levin reported a structure-activity relationship between the permeability of the blood-brain barrier, and the log P and molecular weight of a substance.26 He found that drugs with log P values as low as -3.0 could penetrate the brain in detectable amounts. The following table is intended merely to give an idea of the range in lipophilicity found for small molecules that readily enter the brain. Table 5.5. The log P values of some common drugs. 2 7 ethanol -0.31 morphine 0.07 nicotine 0.45 cocaine 1.05 caffeine 0.08 phenobarbital 1.42 codeine 0.23 amphetamine -0.84 The log P can be measured by shaking a solute with the immiscible solvents and measuring the solute concentration in one or both of the phases. Conceptually simple, the consensus opinion is that the classic "shake-flask" method is experimentally difficult and very time consuming. 1 6 9 ' 1 7 0 The difficulties in measuring the partition coefficient have at times caused wide variations in the reported log P values; in a recent study, an author went so far as to eliminate from consideration all log P values that were in conflict with those reported by Hansch and his co-workers. 1 7 1 (This reflects both the high regard that is afforded Hansch and the difficulties of measuring P.) This has led to the development of numerous alternative methods for determining log P v a l u e s . 1 7 0 ' 1 7 2 ' 1 7 3 The most widely used alternative is reverse phase high performance liquid chromatography (HPLC) with 142 methanol-water as the eluent. 1 7 0 * 1 7 4 The H P L C retention times are converted to log P by means of a standard set of solutes for which the shake-flask log P values have been measured; however, this method has been criticized as being unreliable, especially for hydrophobic compounds that require high proportions of methanol (> 50%) in the eluent. 1 6 9 The log P is an additive-constitutive property of a substance and as such it can be estimated using a substituent constant, JC , defined in an analogous manner to the Hammet a c o n s t a n t . 1 6 5 A manual algorithm was developed that has since given way to a computerized method of estimating log P: for a homologous series, the log P values are measured for the compounds with key structural features capable of mutual interaction and then estimated for the other members of the series. The computations are non-trivial (and expensive) and the estimated log P values are often quite different from the experimentally derived results. Despite the problems in its execution, the shake-flask method is considered the most accurate and reliable method for measuring log P values. 1 6 9 5.5 Materials and Methods Reagent grade rc-octanol was distilled and the first and last quarters were discarded. The pH 7.4 isotonic buffer was prepared as described in Section 4.2. The two solvents were mutually saturated by stirring a 1:1 mixture overnight and the saturated solvents were used for all measurements. The ligands and metal-ligand complexes were purified by recrystallization or sublimation. The solute concentration in the aqueous phase was determined by monitoring the absorbance of the B-band; the spectra were recorded on the U V spectrophotometer used for the aqueous solubility study. Trial solutions made with 143 n-octanol saturated buffer indicated the small amount of n-octanol was not affecting the position or intensity of the B-band. The reference solution was n-octanol saturated buffer that had been centrifuged for the same length of time as the sample solutions. This was necessary because /i-octanol absorbs near 210 nm and a small difference in its concentration between the reference and sample solutions was found to cause serious baseline drift. The initial absorbance was kept between 0.5 and 1.0 which required concentrations of -30 p M for the metal complexes and ~60 p M for the free ligands. 1 m M stock solutions were diluted (using graduated cylinders) to make 25 mL solutions at the desired initial concentrations. A 10 mL aliquot was withdrawn and placed in a 15 mL centrifuge tube labelled as the initial solution. Two 6 mL aliquots were withdrawn and placed in centrifuge tubes labelled as extraction tubes. 6 mL of /i-octanol was added to the extraction tubes and they were inverted 100 times (>2 minutes contact time). After >15 minutes equilibration, the tubes were centrifuged until the two layers were visibly clear (typically at least 15 minutes). The n-octanol layer was removed with a pasteur pipette, and the aqueous layer was used to rinse the cuvettes and to make one absorbance reading per tube. The initial solution was also centrifuged and two absorbance readings were made. This was repeated three times per compound thus producing six measurements of the initial absorbance and six of the post-extraction value. The values of P were calculated from the absorbance of the aqueous phase as fol lows 1 7 5 and the mean log P (r > 6) and a were reported. (initial absorbance)-(post-extraction absorbance) volume of buffer post-extraction absorbance x volume of w-octanol From 0 to 25 °C, the log P can vary from 0.005 to 0.01 log units/degree;165 this is within the error for the shake-flask method and no attempt was made to control the 144 partitioning temperature. Ideally, samples of both phases should be analyzed to check for material balance as a guard against unforeseen losses. This requires an analytical procedure for both phases which further increases the time required for the measurements. It has been shown that if care is taken to ensure that no special solute interactions are occurring, reliable results can be obtained by analyzing only one phase. 1 7 6 5.6 Results and Discussion Once we decided on using the shake-flask method, it remained to choose an analytical technique. One option was flame atomic absorption spectroscopy (FAAS) and this technique has been used to determine the log P for a variety of Hg complexes. 1 7 7 The viscosity of n-octanol did present problems for sample aspiration, but a procedure was developed and several A l and Ga complexes were studied (analyses done by T. Karpishin). The compound that was used to verify our F A A S procedure was Al(ma)3. Shake-flask extraction with U V spectroscopy as the analytical technique gave a log P of -0.17 for 57 p M Al(ma)3 3 7 (this value was reproduced by our collaborators at another laboratory 1 7 8). Using the F A A S procedure, the Al(ma)3 log P was -1.22 ± 0.10 for a 2.5 m M solute concentration and a similar large deviation from the value determined with U V spectroscopy was seen for Ga(ma)3. The log P was also found to be concentration dependent as the Al(ma)3 log P values ranged from -1.70 at 15 m M to -1.10 at 1.5 m M . The partitioning should be done at the lowest solute concentrations possible since P is concentration dependent and only theoretically valid at infinite dilution; however, concentrations of 10_ 1 M are considered sufficiendy dilute for neutral molecules that have little tendency to associate in solut ion. 1 7 6 The Al(ma)3 log P is not constant at m M 145 concentrations and this suggests intermolecular association. The fact that the log P is an order of magnitude higher at 57 | i M than at 2.5 m M certainly supports this conclusion. To avoid errors from solute association, the partition concentrations can be decreased until no further change in log P is f o u n d . 1 7 6 This is not possible with F A AS because of the detection limits for A l and Ga (the bottom end of the working range is 1.5 and 0.72 m M , respectively 1 7 9). We experimented with F A A S because of problems with the procedure we used to measure the log P values of tris(3-hydroxy-4-pyronato) A l and Ga complexes (analyses done by T.Lutz). 3 7 This procedure was prone to give large fluctuations in log P values and changing the analytical technique offered a possible improvement. F A A S did solve the problem of precision, but the large discrepancy from the "true" log P for Al(ma) 3 made the accuracy of the results suspect. The experiments with FAAS did alert us to the potential complication that intermolecular association presented. The large molar absorptivities of the 3-hydroxy-4-pyridinones made U V spectroscopy the analytical technique of choice for this study. (pM detection limits). The problems with our first procedure for determining log P were not rectified by changing the analytical technique so adjustments were made to the extraction steps and the revised procedure is reported herein. Ga(ma)3 was used to test the revised procedure and the log P (in brackets in Table 5.6) was in agreement with our previously published value. The error of ±0.03 log units was calculated on the basis of two separate experiments (r=12) and it was the same as the error considered acceptable for the shake-flask method. 1 7 6 The lower limit that can be measured with acceptable precision by this procedure is a log p of -1.75. For log P values lower than this, the difference between the initial and post-extraction absorbances (initial absorbance of 1 and 1:1 extraction) is < three times the instrument error (± 0.005 absorbance units). 146 The accepted way to extend the log P range is to increase the ratio of n-octanol to buffer, thereby increasing the difference between the two absorbance readings (this also decreases the error in log P). However, extractions at these ratios formed emulsions that were very difficult to remove and the failure to get completely clear solutions is known to produce large errors in log P values. 1 7 6 We found that letting the solutions stand for > 24 hours did not remove the emulsions and even centrifuging for several hours did not always give clear solutions. Based on the log P values of the 3-hydroxy-4-pyrone metal complexes (Table 5.6), the lower limit on log P imposed by 1:1 extractions was acceptable as it was thought the 3-hydroxy-4-pyridinone metal complexes would have log P values within an order of magnitude of these results. We also felt that a log P of -1.75 made a reasonable cut-off point as compounds with values below this are quite hydrophilic and this would significantly limit their ability to cross cell membranes. The log P values for a number of In complexes were determined using the revised procedure. 1 1 2 When the results for the In complexes of maltol and kojic acid are included with the log P values determined previously for the A l and Ga analogues (Table 5.6), a reasonable pattern emerges (for the structures of the 3-hydroxy-4-pyrones see Fig.2.2). With its additional hydroxyl group, kojic acid is considerably more hydrophilic than maltol. For both ligands, the metal complexes have smaller log P values than the free ligands, and the relative order of the lipophilicity of the metal-ligand complexes is the same; i.e. In > A l > Ga. The results for the tris-ligand metal complexes with pyromeconic acid and cholorokojic acid (5-hydroxy-2-(chloromethyl)-4-pyrone) were predictable: removing the ring methyl group from maltol (to give pyromeconic acid) produced a decrease in lipophilicity and replacing the methyl O H group in kojic acid with a C l atom (to give chlorokojic acid) produced an increase in lipophilicity. 147 Table 5.6. Log P values for the 3-hydroxy-4-pyrone complexes (a values are in parenthesis). M L 3 Free Ligand 3 A l Ga In Maltol 0.090 - 0.17 - 0.29 [-0.22(3)] -0.009(8) Pyromeconic acid -1.12 -- -0.62(6) Kojic acid - 0.64 - 1.06 - 1.10 -0.82(6) Chlorokojic acid -- -- -0.31(4) a These values are from ref. 180 and were determined by the shake-flask method. The log P values for the 3-hydroxy-4-pyrones are presented as an example of a well behaved metal-ligand system. The log P values are reasonable for metal complexes and changes in structure produce predictable changes in the log P. By comparison, the 3-hydroxy-4-pyridinones could be called a "truculent" metal-ligand system based on their log P values (Table 5.7). With the exception of A l - and Ga(mhpp)3, the metal complexes have much lower than expected log P values. Instead of being similar to the values of the 3-hydroxy-4-pyrone metal complexes, they are at or near the lower log P limit for the shake-flask procedure employed in this study. Professor Hansch graciously agreed to measure the log P of a Ga(dpp)3 sample and the result reported to us (italicized entry in Table 5.7) is in agreement with the value that was determined in our laboratory . The error in log P (±0.10 log units) is significantly larger than the ± 0.02 log units typical for what Hansch referred to as "well behaved compounds." 1 8 1 The tris(3-hydroxy-4-pyridinonato) metal complexes have log P values that are lower and more difficult to measure than was predicted on the basis of the results for the 3-hydroxy-4-pyrone system. 148 Table 5.7. Log P values for the 3-hydroxy-4-pyridinone complexes. M L 3 Free Ligand A l Ga In Hmpp - 0.52 (8) < - 1.75 - 1.51 (6) < - 1.75 Hdpp - 0.74 (8) < - 1.75 - 1.55 (20) -1.59 (10) < - 1.75 Hmepp - 0.37 (3) - 1.68 (6) - 1.64 (20) Hmhpp 0.95 (10) 1.32 (20) 1.38 (12) It is possible that the interaction of water molecules with the tris(3-hydroxy-4-pyridinonato) metal complexes is affecting the partitioning process. The partitioning in this laboratory was done at 30 p M solute. concentrations, and Hansch used Ga(dpp) 3 concentrations from 27 to 3 uM without observing any concentration dependence. This indicates that intermolecular association is not responsible for the low log P values. The solid state and the water solubility studies demonstrated the strength of the interaction between water molecules and the metal complexes. If the metal-ligand complexes are associating with liquid water in a fashion comparable to the M(dpp) 3 and M(mepp)3 solid state structures, the reduced log P values would be reasonable as the most lipophilic portion of the complex would be shielded from the n-octanol molecules (see Chapter III). It is difficult to account for the effects of solvent-solute association on partitioning, 1 6 5 but water association comparable to that found in the solid state structures could explain both the lower water solubility and the lower log P values of these metai-ligand complexes. The free ligands were certainly better behaved than the metal complexes, but the order of lipophilicity was unexpected; instead of Hmpp < Hdpp < Hmepp < Hmhpp as predicted on the basis of the N-substituents, Hmpp was more lipophilic than Hdpp. If one 149 equates lipophilicity to a lower water solubility, Hmpp was significantly more lipophilic than Hdpp and this was thought to be the result of intermolecular H-bonding (see Table 5.3). The log P of Hmpp was measured at 52 p M and H-bonded dimerization is not considered to be a significant factor in log P measurements at concentrations below the m M leve l . 1 6 5 Also, if there is intermolecular association, the log P should be smaller not larger than expected. Once again the physical properties of the homologous series of ligands did not follow the predicted order. The M(mhpp)3 complexes have log P values larger than that of Hmhpp; obviously the lipophilicity imparted by three hexyl groups dominates the partitioning process just as it governed the aqueous solubility. The M(mpp)3, M(dpp)3, and M(mepp)3 complexes have smaller log P values than their respective ligands and this indicates they would have potential as therapeutic chelating agents. The ideal chelating agent would be a ligand that is lipophilic enough to be distributed to the sites of metal accumulation and that forms metal-ligand complexes more hydrophilic than the free l i g a n d . 1 6 8 This would diminish redistribution of the metal to other tissues and would expedite the elimination of the metal from the body. Taking the removal of A l from the brain as a hypothetical clinical situation, the ligands—Hmpp, Hdpp, and Hmepp—are small enough and are sufficiently lipophilic to cross the blood-brain barrier (based on Levin's structure-activity relationship2 6). The tris-ligand A l complexes are at least an order of magnitude more hydrophilic than the free ligands and this fits the above criteria for a therapeutic chelating agent. 150 C. Potentiometric Equilibrium Measurements 5.7 Introduction In the reaction of a trivalent metal with a bidentate ligand, a series of equilibria are established involving the metal, the ligand, and metal-ligand complexes. These equilibria are illustrated in Figure 5.1 along with the definitions of the stepwise (K n ) and overall (Pn) formation constants. The generalized formula for the overall formation constants is pn = [MLn] [M] [ L P * [ML] M + L ^2 ML K i = jjfffo p! = K , ML + L ^ M L 2 K 2 = pSSfip,- P2 = K i K 2 M L 2 + L -p: M L 3 K 3 = " M ^ L T & = K-K2K3 Figure 5.1. Equilibrium equations and constants for the reaction of a trivalent metal with bidentate ligands. The above equations describe the equilibria for A l and Ga in aqueous solution (neglecting mixed hydroxo complexes). For the 3-hydroxy-4-pyridinone ligands, L is the conjugate base formed by the dissociation of the hydroxyl proton, and the neutral ML3 complex was predicted to be the dominant metal-ligand species at physiological pH. Variable-pH 2 7 A l N M R experiments provided a qualitative picture of the pH-dependent 151 speciation of the A l complexes that clearly supported this prediction. To obtain a quantitative evaluation of the A l speciation, and to confirm our assumption of an analogous speciation for the Ga complexes, potentiometric titrations were performed to determine the formation constants. The preparation and standardization for all the solutions and the potentiometric titration procedure are included in the Appendix as Procedure A . l . This procedure was developed by Dr. David Clevette of this laboratory and it has been reported previously. 1 8 2 The computational methods used to calculate the formation constants were developed elsewhere and are referenced as such in the procedure. The data for Hmpp, Hdpp, Hmepp, and Hmhpp (Table 5.8) were reported to the author by Dr. Clevette. 5.8 Results and Discussion The 3-hydroxy-4-pyridinone ligands form A l and Ga complexes of great stability (Table 5.8); the overall stability constants P3 are all > 10 3 0 at 25 °C and an ionic strength of 0.15 M NaCl (isotonic). In the equilibrium calculations a hydrolysis model consisting of the species [M(OH) n ] ( 3 _ n ) + (n = 1, 2, 3, 4) with formation constants according to ref. 25 was applied. The polynuclear species [Al2(OH)2] 4 + and [Al3(OH)4] 5 + were also included in the calculations. In conjunction with the studies in this laboratory, Professor Staffan Sjoberg examined the equilibrium reactions of A l with Hdpp and Hmpp. He found that effects due to possible mixed A l 3 + - OH" - L" species were negligible at total ligand to aluminum ratios of 1, 2, 3, and 5 ; 1 8 3 therefore, a total ligand to metal ratio of just greater than three was used in all the potentiometric titrations performed in our laboratory. 152 Table 5.8. Log protonation constants (K) and log p n for the equilibrium reactions of A l , Ga, and In with the 3-hydroxy-4-pyridinone ligands synthesized in this study (25 °C, 0.15 M NaCl). Constant Metal Hmpp Hdpp Hmepp Hmhpp 0 l o g K i 9.80 (1) 9.86 (3) 9.81 (2) 9.92 (2) logK2 3.65 (1) 3.70 (1) 3.64 (2) 3.59 (1) log Pi A l 11.87 (3) 11.91 (2) 11.75 (4) 11.51 (1) Ga 13.34 (3) 13.17 (15) 13.15 (9) ~ a In 13.51 (1) 13.60 (2) 13.53 (1) logp2 A l 22.54 (3) 22.83 (2) 22.52 (5) 22.49 (1) Ga 24.41 (1) 25.16 (15) 24.98 (11) — a In 23.70 (1) 23.93 (3) 23.78 (1) log P3 A l 32.05 (3) 32.25 (5) 32.17 (6) 31.71 (3) Ga 32.85 (6) 34.32 (14) 33.88 (13) — a In 32.76 (2) 32.93 (3) 32.80 (1) a The In values were included in reference to Figure 5.7. b Data for the Ga and In complexes were not available due to their low water solubility. 153 The affinity of the hydroxide ion for A l and Ga has been alluded to at several points in this thesis and one of the primary objectives of this research project was the synthesis of ligands that would prevent the formation of metal hydroxides at neutral pH. The competition between the ligands and the hydroxide ion can be represented graphically by using the data in Table 5.8 to produce plots of metal speciation as a function of solution pH. Since the variation in log p n values among the four ligands is minor, the Al-Hdpp speciation diagrams are representative of the ligands studied. In Figure 5.2, the speciation diagram for 3 m M Hdpp (top figure) shows no hydrolysis at neutral pH and [Al(OH)4]" does not occur in significant amounts (> 1% of the total Al) below p H 8.5. Even with a thousand-fold dilution in ligand concentration (bottom figure), the metal-ligand species still predominate from pH 4 to 8 and 60% of the initial A l is in Al(dpp) 3 at physiological pH. The qualitative observations of an absence of hydrolysis products in the synthesis of the tris(3-hydroxy-4-pyridinonato) A l complexes are verified by these quantitative results. The pH 4 to 9 window of hydrolytic stability for the tris-ligand A l complexes that was determined by the variable-pH 2 7 A l N M R study (see Section 4.3) is confirmed by the potentiometric titration data. Using the same concentrations of A l and Hdpp as in the N M R experiment ([Al] = 35 m M and 3:1 A l to Hdpp) it is possible to "simulate" the 2 7 A l N M R spectral results in a qualitative fashion. In Figure 5.3, several spectra from the variable-pH 2 7 A l N M R study are shown next to bar graphs of the equilibrium speciation profile calculated at the same pH. The agreement between the two methods is very good and this comparison also serves as a verification of the assignment of the upfield 2 7 A l N M R signals to the mono- and bisligand A l species. Figure 5.2. Speciation diagrams for 3:1 Hdpp to A l at 3 m M (top) and 3 p M (bottom) ligand concentrations (25 °C, 0.15 M NaCl). 155 Figure 5.3. Left: variable-pH 2 7 A 1 N M R spectra for 0.035 M Al(dpp)3. Chemical shifts (ppm) are: [Al(OH)4r (80), [Al(dpp) 3] (38), [Al(dpp)2(H 2 0)2] + (26), [Al(dpp)(H 2 0)4] 2 + (14), and [ A 1 ( H 2 0 ) 6 ] 3 + (0). Right: A l speciation profiles calculated for 0.035 M A l using data at 25 °C and 0.15 M NaCl (L = dpp). 156 It was not possible to perform variable-pH 7 1 G a N M R experiments because of the low sensitivity of this quadrupolar nucleus and the low solubilities of the Ga complexes. Their hydrolytic stability could be inferred from the similarity of their synthesis, characterization, and physical studies data to that of the corresponding A l complexes. The determination of the formation constants establishes that the Ga-ligand complexes exhibit the same resistance to hydrolysis as their A l analogues and this is readily apparent from the speciation diagram for 3 m M Hdpp and 1 m M Ga (Fig. 5.4). The greater stability of the Ga-hydroxides and the larger log p*n values of the Ga-Hdpp complexes results in a slight shift in the region of hydrolytic stability towards lower pH. A comparison of the A l and Ga speciation diagrams shows that the cross-over point at which the [M(H20)6] 3 + species is reduced to 50% of the total metal concentration is shifted from a pH of 2.2 to a p H of 1.4 and the [M(OH)4]" species appears at a lower p H and in greater concentration in the Ga diagram. Figure 5.4. Speciation diagram for 3 m M Hdpp and 1 m M Ga at 25 °C and 0.15 M NaCl. 157 The 3-hydroxy-4-pyridinone ligands and their precursor maltol are amphoteric (Figure 5.5). The two stepwise protonation constants for the ligands (Ki and K 2 ) are given in Table 5.8. At the conception of this research project, it was thought that the N-alkyl groups would enhance the ability of the ring nitrogen to stabilize a positive charge. This should have resulted in more ring double bond delocalization and larger K 2 values by increasing the stability of the pyridinium resonance hybrid and the pyridinium cation. The data from the solid state studies (refer to Section 3.2) and these protonation constants both show that this anticipated effect of N-alkylation did not occur. The log (J3 values for the metal complexes also indicate the metal binding efficacy of the 3-hydroxy-4-pyridinone ligands was not significantly affected by N-alkylation. O O O H Figure 5.5. Protonation equations for 4-pyridinone ligands (X = N-R) and maltol (X = O) Hmpp is the 4-pyridinone ligand closest in structure to maltol and comparison of the protonation constants shows that the 4-pyrones are stronger acids. (The data for maltol 5 9 were collected in 0.6 M NaCl; because the magnitude of the formation constants is dependent on ionic strength, the following values for H m p p 1 8 3 are also at 0.6 M NaCl and are slightly different from the values at 0.15 M NaCl reported in Table 5.8.) Maltol has a hydroxyl log K i of 8.38 while that for Hmpp is 9.58. In both heterocyclic rings there is an 158 additional protonation constant with a log K 2 of -0.71 in maltol (p. = 0.5) 5 8 and 3.74 in Hmpp. The 4.5 log unit difference in K2 is an effect of the ring nitrogen atom, which is better able to delocalize a positive charge than a ring oxygen. This would stabilize the dihydroxypyridinium cation in acidic solution and increase K2 for Hmpp. The difference in K i is due to the greater ability of the ring oxygen to delocalize the negative charge of the deprotonated hydroxyl oxygen. This would stabilize the conjugate base and decrease K i for maltol. The net effect is an increase in the basicity of the chelating oxygen atoms in the 3-hydroxy-4-pyridinone ligands. This results in a large enhancement of the formation constant for the metal-ligand complexes: log P3 for Al(ma)3 is 22.48 compared to a value of 30.41 for Al(mpp) 3. 100 x ^ 8 50 0 m A A" • • • • A -7 -6 -5 -4 log [total ligand] -3 -2 • Hdpp A edta • maltol A catechol l p M A l 50 u M Tf sites 0.1 m M citrate pH7.4; 25°C Figure 5.6. Plot of Al 3 " 1 " complexation (%) vs. log of total ligand concentration (conditions and ligands are indicated in the legend). Figure 5.6 is a plot that compares the metal binding affinities of several ligands regardless of denticities. Using formation constants taken from the literature for A l complexes with citrate 1 8 4 (100 pM) and transferrin29 (50 p M vacant sites), a simple model of the metal binding capacity of human blood can be made (pH 7.4, 25 "C, 0.15 M NaCl). 159 For 1 | i M A l , the plot shows that Hdpp is more efficient (lower ligand concentration required) at complexing 100% of the A l than is E D T A 1 8 5 (which is hexadentate and tetraprotic), pyrocatechol 1 8 6 (bidentate and diprotic), and malto l 5 9 (bidentate and monoprotic). It must be emphasized that this model is limited in an absolute sense, but it is valid in a comparative sense and it certainly illustrates the enhanced metal binding capacity of the 3-hydroxy-4-pyridinone ligands. The solid state structures of A1-, Ga-, and In(dpp)3 showed little variation in bonding parameters, although the M-O bond lengths did indicate the bonds were somewhat weaker in Al(dpp) 3 (see Section 3.4.1). However, the Ga and In complexes have significantly larger log {3n values than the A l analogues and this is illustrated by a plot of the log p3 values for Hdpp, Hmpp, and Hmhpp (Figure 5.7). This graph shows that the order of stability for the tris-ligand metal complexes is the same as the order of the first dissociation constants for the hexxaquo ions; i.e., Ga > In > A l (see Section 2.5). Using the p K a of the 3-hydroxyl proton (log K i ) as a measure of basicity, a plot of p K a vs. ligand (Figure 5.8) correlates exactly with the relative stabilities of the metal complexes. Hdpp is the strongest base and it forms the most stable metal complexes. Taken together, these two graphs nicely illustrate that this ligand-metal system does comply to the H S A B principle paraphrased as: "the harder acid prefers the harder base." The formation constants for the metal-ligand complexes show that the 3-hydroxy-4-pyridinones are very good bidentate chelators for the group 13 metals. These quantitative results are in complete agreement with our experimental observations and they also verify the hydrolytic stability of the tris-ligand complexes as established by 2 7 A 1 N M R . The predictive capability that these data affords is particularly useful in the context of our stated goal of using in vitro techniques to assess the suitability of a ligand for in vivo studies. 160 3 4 34.0-33.5-33.0 3 2 . 5 1 32.0 31.5 Hmpp Hdpp Hmepp Figure 5.7. Plot of log P3 for the Ga, In, and A l complexes (as indicated in the legend) vs. ligand. i 9.90 T 9 . 8 5 1 9 . 8 0 1 9.75 Hmpp Hdpp Hmepp Figure 5.8. Plot of pKa of the hydroxyl proton vs. ligand (25 *C, 0.15 M NaCl). 161 Chapter VI GaIIium-67 Biodistribution Study 6.1 Introduction Our goal of developing radiopharmaceutical imaging agents and our interest in aluminum, which has no suitable isotopes for radiolabelling studies, explain why 6 7 G a is the radionuclide that was chosen for these imaging experiments. The favorable nuclear decay properties and the relative paucity of work with this radionuclide (see Chapter I) were also factors that prompted our interest in 6 7 G a . Despite the differences in electronic configuration and ionic radius, the aqueous coordination chemistry of A l and Ga is very similar. They are only found in the +3 oxidation state in water and their aqueous chemistry is dominated by their shared property of strong Lewis acidity. These 6 7 G a biodistribution experiments with the 3-hydroxy-4-pyridinones can be considered first order approximations of the biological fate of the A l analogues. The similarity of in vitro aqueous behavior makes 6 7 G a the best model available for A l . The ligands—Hmpp, Hdpp, Hmhpp, and L-mimosine—form tris-ligand Ga complexes that are sufficiently water soluble for a biodistribution study; the solubilities range from 0.74 m M for Ga(dpp)3 to 55 m M for the mimosine complex. Ga(mhpp)3 has a log P of 1.4 and the other complexes have log P values near -1.6. None of these complexes has the the same combination of these properties seen in Ga-maltol (water solubility of 31 m M and log P of -0.22), but they do provide a range of lipophilicity that could result in variations in the biodistribution of 6 7 G a . The formation constants for the Ga-ligand complexes are large enough to ensure hydrolytic stability in vivo. 162 Once the ligands and tris-ligand Ga complexes were completely characterized, a preliminary screening experiment with 6 7 G a was performed to determine if the 3-hydroxy-4-pyridinones showed promise as imaging agents. It was at this point that the developmental work with maltol ceased as the image obtained for 6 7Ga-maltol was identical to that of 6 7Ga-citrate, that is to say, the same as o 7Ga-transferrin. 1 8 7 However, the results with Hdpp indicated that this ligand was affecting the 6 7 G a biodistribution. Based on this screening and the in vitro studies, a biodistribution study using Hdpp, Hmpp, Hmhpp, and L-mimosine was initiated. This study is still in progress; therefore, only a brief description of the methodology and a summary of results to date wil l be presented herein. This work was done in collaboration with Dr. Donald Lyster and a complete report of the results is forthcoming. 1 8 8 6.2 Materials and Methods The biodistribution study was done in the Radiopharmacy at Vancouver General Hospital. The animal work-up and the data analysis were performed by Dr. D. Lyster, T. Rihela, and G. Webb. The ligands were synthesized and purified in our laboratory. Solutions at the proper concentrations were prepared in isotonic Trizma pH 7.4 buffer and delivered to the Radiopharmacy for radiolabelling and animal injection. The 6 7Ga-citrate starting material was available by purchase. For the preliminary screening experiment, anesthetized rabbits were injected with 0.43 m M Hdpp solution containing 0.5 mCi of 6 7Ga-citrate (33 nM) and the imaging was done with a Siemans Large Field of View Gamma Camera. For a control, the same dosage 163 of 6 7Ga-citrate without Hdpp was injected in a second rabbit and the images were compared at various time intervals over a period of hours. The biodistribution experiments were performed in BALB/c mice which were sacrificed at varying time intervals up to 2 days post-injection. The percent injected dose per organ (i.d./organ) values were obtained for blood, liver, spleen, stomach, kidneys, lungs, heart, and brain. The organs were placed in vials and the radiation was measured in a well gamma counter. In all the experiments, the ligand to Ga molar ratio was kept constant (~104 to 1) and each mouse was given a 0.1 mL injection containing 1 p C i of 6 7 G a . For example, the injection concentrations were 3.4 p M Hdpp and 0.25 n M 6 7 G a . The study was then repeated using the same amount of 6 7 G a and concentrated (near saturation) ligand solutions. Again using Hdpp as the example, the injection concentrations were 80 m M Hdpp and 0.25 n M 6 7 G a for a molar ratio of 10 8 to 1. The injection concentrations of the other ligands were 1, 20, and 40 m M for Hmhpp, L-mimosine, and Hmpp respectively. As a control in both the dilute and concentrated ligand studies, 6 7Ga-citrate solutions without added ligands were also administered to a separate test population. 6.3 Results and Discussion The 6 7 G a biodistribution experiments with dilute ligand solutions were done to assess the potential of these ligands as 6 7 G a imaging agents. The injection concentration of 6 7 G a necessary to produce the optimal radiation is on the n M level; therefore, formation of the tris-ligand 6 7 G a complex is ensured by using p M ligand solutions. To assess the 164 efficacy of an imaging agent, it is desirable to keep the ligand concentration at a minimum because of the influence that a large excess of ligand can have on the biodistribution. Metal ions are thought to cross membranes by one of two generalized processes: active transport requiring energy dependent ion pumps or passive transport; i.e., adsorption onto the cell membrane followed by diffusion into the c e l l . 1 8 9 The latter process may be facilitated by ligands that could aid the transport of the hydrophilic metal ion across the hydrophobic membrane. This can be envisaged as passage from the aqueous medium to the interior of the cell along a ligand cascade of increasing binding strength. 1 9 0 Therefore, the abundance and strength of extracellular metal-ligand binding affects metal uptake which is clearly demonstrated by the effect organic ligands can have on metal toxicity. 1 8 9 The objective of cellular accumulation could be adversely affected by a large excess of the imaging agent itself. This was of particular importance with these ligands since, with the exception of Hmhpp, the tris-ligand Ga complexes were significantly less lipophilic than was considered ideal. The preliminary screening had indicated that the ligands were affecting the 6 7 G a biodistribution with p M ligand injection concentrations. To minimize potential interference with cellular uptake, the same ligand to metal ratio was maintained in the dilute ligand experiments. The assessment of these ligands as chelating agents was done with concentrated ligand solutions. The objective of a chelating agent is the removal of a metal from the body and high concentrations of ligand in the blood can only facilitate this goal. In long term experiments, high levels can be maintained by repeatedly administering the ligand. We were conducting experiments of a shorter duration and were interested in determining if injection of saturated ligand solutions would be sufficient to produce a change in biodistribution. If this were to occur, it would be good evidence of the potential of these ligands as chelating agents. 165 The rjreUminary results of this study can be summarized be examining the uptake of 6 7 G a (as percent of total 6 7 G a ) in the liver at 24 hours post-injection. The bar graph (Fig. 6.1) contains the data for the four ligands, both the dilute and concentrated experiments, and for the 6 7Ga-citrate control. The liver is the principle organ where transferrin accumulates; therefore, any alteration from the biodistribution of 6 7Ga-transferrin can be most readily observed in this organ. The uptake of 6 7Ga-citrate (taken to be that of Ga-transferrin) is shown in the last column. It is obvious that these 3-hydroxy-4-pyridinones significantly reduce the uptake of 6 7 G a in both the dilute and concentrated ligand experiments. This confirms in vivo what the in vitro experiments had shown—the 3-hydroxy-4-pyridinones are very good bidentate chelators of Ga.. 10 LIGAND • I Dilute ES3 Concentrated Figure 6.1. Liver biodistribution plotted as 6 7 G a uptake vs. ligand for the dilute and concentrated ligand experiments. These results indicate that these 3-hydroxy-4-pyridiones wil l not be useful as imaging agents; although they redirect 6 7 G a from transferrin at concentrations that would 166 be viable for further development, they are not localizing the radionuclide in any particular organ. The 6 7 G a uptake in the other organs examined is similar to that of the liver, and the levels of 6 7 G a are uniformly reduced from that of 6 7Ga-citrate. It appears that the ligands enhance the removal of the metal from the body, presumably via the urine; but this cannot be definitely proven until this study is completed. It was thought that the lipophilic Hmhpp complexes might have a different biodistribution. This did not occur, although it is possible that 6 7Ga-Hmhpp may be removed through the hepatobiliary tract. This also has yet to be determined. We intend to exploit the ability of this class of ligands to redirect 6 7 G a from transferrin. This wi l l be done by synthesizing additional 3-hydroxy-4-pyridinone ligands with other substituents that could alter their biodistribution. The further decrease in 6 7 G a uptake seen for the concentrated ligand experiments is especially encouraging when viewed in concert with the positive results from the extensive studies of Hdpp as an iron chelating agent.3 9 The fact that the two trivalent metal ions closest in size to A l can both be mobilized by Hdpp definitely indicates the potential of the 3-hydroxy-4-pyridinones as A l chelators. The results are even more impressive when the differences in transferrin binding constants are considered since the affinity of transferrin for A l is ca. eight and ten orders of magnitude less than for Ga and Fe, respectively. This in vivo study also allows us to assess the relative merits of the methodology that we have been employing in our laboratory. The log P values for these ligands were significantly lower than ideal and this is undoubtedly a factor in the rapid elimination of the metal-ligand complexes from the body. Knowledge of the difference in log P values between the ligand and the metal complexes is very useful since the biodistribution results further indicate their potential as chelating agents. The aqueous solubility data were used in the design of this biodistribution experiment and this simple determination merits inclusion in future studies. 167 The formation constants can be employed in a manner complementary to experiments such as these biodistribution studies. Figure 6.2 is a plot that accounts for the dilution of the ligand and the 6 7 G a in the circulatory system. It indicates what ligand injection concentration (in this case, Hdpp) is necessary to compete effectively with transferrin, using the transferrin binding constants and concentrations of human blood serum. This graph demonstrates both the utility of this model and the difficulties encountered in applying any simple model to the complex interactions in vivo. The model predicts that 80 m M Hdpp used in the concentrated ligand experiment is sufficient to complex almost all of the 6 7 G a and this is supported by the results of the liver uptake of 6 7 G a . The model also predicts that p.M Hdpp should not alter the biodistribution, but the animal study indicates some redirection from transferrin. It is necessary to improve the model by including other factors such as Fe concentrations and concentrations of metal binding proteins such as albumin. This is not an easy task but the predictive capacity shown by the simple model definitely indicates this is a goal worth pursuing. 100 -i % o f Ga complexed with Hdpp 80 -60 -40 20 H • • • • —r-20 1 picoM Ga 0.1 m M citrate 50 p;M Tf sites pH7.4; 25°C —r~ 40 60 80 100 120 m M Hdpp injection concentration Figure 6.2. Graph of %Ga complexed as a function of Hdpp injection concentration (conditions as indicated in legend). 168 Chapter VII Conclusion and Suggestions for Future Work The primary objective of our research is the investigation of the aqueous coordination chemistry of A l and Ga. We are also interested in biological applications of this chemistry with the dual goals of developing 6 7 G a radiopharmaceuticals and using 6 7 G a as a model for the biodistribution of A l . The preceding chapter addresses the biological aspects of our work and it serves as both the conclusion for that part of our research and as an indicator of the direction our future research in that area will take. The core of my research was the synthesis and characterization of several 3-hydroxy-4-pyridinone ligands and their A l and Ga complexes. The ligand synthesis initially involved the use of a benzyl blocking group, but because of the length of the procedure, a preparation was developed using buffered solutions to control the ionic state of the starting materials, maltol and the primary amines. The buffered preparations pointed us towards the next logical step in this synthesis and this work is now being carried out in our laboratory. The formation of Al(ma)3 enhances the reactivity of the complexed maltol ligands to nucleophilic attack. The electropositive metal acts to remove electron density from the 4-pyrone ring and it is well established that electron-withdrawing groups facilitate the conversion of 4-pyrones to 4-pyridinones. By the reaction of Al(ma)3 with an excess of methylamine, Al(dpp)3 can be produced almost quantitatively. This metal template reaction is now being used in the synthesis of other 3-hydroxy-4-pyridinone ligands. We attempted the synthesis of multidentate ligands and the bispyridinone H2exn was the initial product of this work. However the solubility properties of this compound restricted its use as a ligand. Multidentate ligands with the cc-hydroxyketone moiety could 169 be made by joining two 4-pyridinone rings via an amide linkage and this should afford increased aqueous solubility. 3-Hydroxy-2-methyl-l-(P-ethylamino)-4-pyridinone, formed by the conversion of maltol with ethylenediamine, was prepared and isolated in our laboratory. Using a dicarboxylic acid (such as butanedioic or pentanedioic acids), it would be possible to link two of these ligands together via their N-ethylamine substituents. The coupling agent dicyclohexylcarbodiimide (DCC) could be used to form the peptide bonds and the length of the bridge between the rings could easily be varied as aliphatic dicarboxylic acids and D C C are readily available commercially. One of the most interesting aspects of our work was the unusual solid state structures that were determined for the M(dpp)3»12H20 complexes. Because of the extensive water network in these structures, attempts were made to grow crystals from water of the other metal complexes. This culminated in the determination of the M(mepp)3»12H20 structures. The stability of the water network was somewhat unexpected and a close analysis of the structures suggests that a slighdy larger N-alkyl group would disrupt its formation. The synthesis of 3-hydroxy-2-methyl-l-propyl-4-pyridinone is now being carried out; i f the metal complexes of this ligand are sufficiently water soluble, attempts will be made to grow crystals to determine whether the water networks can still be formed with this larger N-substituent. Also, variable temperature potentiometric titrations will be performed to establish the enthalpic and entropic contributions of this water network to the overall thermodynamic stability of the M(dpp)3 complexes. The variable-temperature N M R experiments with A l - and Ga(dpp)3 established that these metal complexes undergo fac-mer isomerization. Ligand exchange experiments with Al(dpp)3 and Al(ma)3 indicated that the ligand rearrangement process likely involved a dissociative mechanism. Further ligand exchange studies are planned to determine the lability of several of the tris-ligand metal complexes in aqueous solution. If the exchange 170 rates permit N M R detection, the ligand exchange between the tris(3-hydroxy-4-pyridinato) A l complexes and Al-transferrin is the ultimate goal of this research. The affinity of the 3-hydroxy-4-pyridinone ligands for A l and Ga made the synthesis of the metal complexes relatively straightforward. The stability of the metal complexes is due to the basicity of both the oxygen atoms of the bidentate ligand. The ring nitrogen is primarily responsible for this base strength because of the stability of the pyridinium resonance hybrid and pyridinium cation. The hydrolytic stability of the A l complexes was established by variable-pH 2 7 A 1 N M R experiments, and potentiometric equilibrium measurements showed the thermodynamic stability of the tris ligand A l and Ga complexes. The favorable properties of these ligands reported in this thesis ensure this ligand system will be the basis of further research in our group. 171 References 1. Driscoll, C. T. Environ. Health Perspect. 1985, 63, 93. 2. Aluminum in the Canadian Environment; Havas, M . , Jaworski, J . F., Eds.; National Research Council of Canada, Assoc. Committee Sci. Criteria Environ. Quality, Rep. No. N R C C 24759,1986; p 331. 3. Cronan, C.S.; Schofield, C. L. Science 1979,204, 304. 4. Alfrey, A . C. Adv. Cline. Chem. 1983,23, 69. 5. Krishnan, S. Can. Res. 1988, 32. 6. Sorenson, J. R. J . ; Campbell, I. R.; Tepper, L. B.; Lingg, L. B. Environ. Health Perspect. 1974, 8, 3. 7. Macdonald, T. L.; Martin, R. B. TIBS 1988,13, 15. 8. Krueger, G. L.; Morris, T. K.; Suskind, R. R.; Widnek, E. M . CRC Crit. Rev. Toxicol. 1984,13, 1. 9. Alfrey, A . C ; Legendre, G. R.; Kaehny, W. D. N. Eng. J. Med. 1976,294, 184. 10. Perl, D. P.; Gajdusek, D. C ; Garutto, R. M . ; Yanagihara, R. T.; Gibbs, C. J . Science 1982, 217, 1053. 11. Crapper, D. R.; Krishnan, S. S.; Quittkat, S. Brain 1976, 99, 67. 12. Crapper, D. R.; Krishnan, S. S.; Dalton, A . J . Science 1973,180, 511. 13. Crapper, D. R.; DeBoni, U . in Aluminum Neurotoxicity; Liss, L., Ed.; Pathotox Publishers: Illinois, 1980; pp 3-15. 14. Dolken, A . Arch. Exp. Pathol. Pharmakol. 1897,40, 58. 15. Llobet, J. M . ; Domingo, J. L.; Gdmez, M . ; Tomds, J . M . ; Corbella, Pharmacol. Toxicol. 1987, 60, 280. 16. Edwards, C. L.; Hayes, R. L. J. Nucl Med. 1969,10, 103. 17. Hoffer, P. /. Nucl. Med. 1980,21, 394. 18. Hayes, R. L.; Huber, K. F. Metal Ions Biol. Syst. 1983,16, 279. 19. Tsan, M . F.; Scheffel, U . /. Nucl. Med. 1986, 27, 1215. 20. Hoffer, P. /. Nucl. Med. 1980,21, 484; Int. J. Nucl. Med. Biol. 1981, 5, 243. 21. Tsan, M . B. /. Nucl. Med. 1985,26, 88. 172 22. Green, M . A. ; Welch, M . J. ; Mathias, C. J . ; Fox, K. A . A. ; Knabb, R. M . ; Huffman, J . C. J. Nucl Med. 1985, 26, 170. 23. Taliaferro, C. H . ; Martell, A . E . Inorg. Chem. 1985,24, 2408 and references therein. 24. Taliaferro, C. H . ; Martell, A . E . Inorg. Chim. Acta 1985,85, 9 and references therein. 25. Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley: New York, 1976; pp 112-123, 313-327. 26. Levin, V . A . /. Med. Chem. 1980, 23, 682. 27. Hansch, C ; Bjorkroth, J . P.; Leo, A . /. Pharm. Sci. 1987, 76, 663. 28. Kulprathipanja, S.; Hnatowich, D. J . ; Beh, R.; Elmaleh, D. Int. J. Nucl. Med. Biol. 1979, 6, 138. 29. Martin, R. B.; Savory, J . ; Brown, S; Bertholf, R.L.; Wills, M . R. Clin. Chem. 1987, 33, 405. 30. Shannon, R.D. Acta. Crystallogr. 1976, A32, 75. 31. Harris, W. R.; Pecoraro, V . L. Biochemistry 1983, 22, 292. 32. Pierpont, C. G.; Buchanan, R. M . Coord. Chem. Rev. 1981,38, 45. 33. Hancock, R. A. ; Orszulik, S. T. Polyhedron 1982,1, 313. 34. Borgias, B. A. ; Barclay, S. J . ; Raymond, K. N . /. Coord. Chem . 1986,15, 109. 35. Veca, A . ; Dreisbach, J . H . /. Chem. Ed. 1988, 65, 108. 36. Rajan, K. S.; Mainer, S.; Davis, J . M . Bioinorg. Chem. 1978, 9, 187. 37. Finnegan, M . M . ; Lutz, T. G.; Nelson, W. O.; Smith, A. ; Orvig, C. Inorg. Chem. 1987,26,2171. 38. McLachlan, D. R. Neurobiol. Aging 1986, 7, 525. 39. Kontoghiorghes, G. J . ; Aldouri, M . A. ; Hoffbrand, A . V.; Barr, J . ; Wonke, B.; Kourouclaris, T.; Sheppard, L. Br. Med. J. 1987,295, 1509. 40. Albert, A . Selective Toxicity, 7th. ed.; Chapman and Hall: New York, 1985; pp 430-489. 41. Lerch, J . U . Monatshefte 1884, J , 407. 42. Boulton, A.J . ; McKillop, A . in Comprehensive Heterocyclic Chemistry Volume 2; Katritzky, A . R., Rees, C.W., Eds.; Pergamon Press: New York, 1984; pp 1-27. 173 43. Brody, F.; Ruby, P. R. in The Chemistry of Heterocyclic Compounds Volume 14: Pyridine and its Derivatives Part One; Klingsberg, E.,Ed.; Interscience: New York, 1960; pp 99-589. 44. Meislich, H . in The Chemistry of Heterocyclic Compounds Volume 14: Pyridine and its Derivatives Part Three; Klingsberg, E., Ed.; Interscience: New York, 1962; pp 509-890. 45. Elkaschef, M . A-F. ; Nosseir, M . H . /. Am. Soc. Chem. 1960, 82, 4344. 46. Brown, R. D. /. Chem. Soc. 1951, 2670. 47. Peratoner, A . ; Tamburello, A . Gazz. Chim. Ital. 1906,36, 33. 48. Yabuta, J . /. Chem. Soc. 924, 575. 49. Bickel, A . F. J. Am. Chem. Soc. 1947, 69, 1801. 50. Adams, R.; Jones, V . V . /. Am. Chem Soc. 1947, 69, 1803. 51. Campbell, K. N . ; Ackerman, J . F.; Campbell, B. K. /. Org. Chem. 1950,15, 337. 52. Adams, R.; Johnson, J . L. /. Am. Chem. Soc. 1949, 71, 705. 53. Kleipool, R. J . C ; Wibaut, J . P. Rec. Trav. Chim. Pays Bos 1950, 69, 1041. 54. Heyns, K.; Vogelsang, G. Chem. Ber. 1954, 87, 1377. 55. Spenser, I. D.; Notation, A . D. Can. J. Chem. 1962,40, 1374. 56. Harris, R. L . N . Aust. J. Chem. 1976,29, 1329. 57. March, J. Advanced Organic Chemistry; McGraw H i l l : New York, 1977; p 258. 58. Chous, G.; Benoit, R. L. /. Org. Chem. 1967, 32, 3974. 59. Hedlund, T.; Ohman, L.-O. Acta Chem. Scand. in press. 60. Nelson, W. O.; Karpishin, T. B.; Rettig, S. J . ; Orvig, C. Can. J. Chem. 1988, 66, 123. 61. Chem. Abstr. 1985,102, 113315Z. (patent) 62. Severin, Th.; Loidl, A . Z. Lebensm. Unter.-Forsch, 1976,161, 119. 63. Kontoghiorghes, G.J. Ph.D. Thesis, University of Essex, U . K. , 1982. 64. Yasue, M . ; Kawamura, N . ; Sakakibara, J . Yakugaku Zasshi, 1970, 90, 1222. 65. Jakopcic, K.; Tamhina, B.; Zorko, F.; Herak, M.J . /. Inorg. Nucl. Chem. 1977, 39, 1201. 66. Looker, J . H. ; Cliffton, M . D. /. Heterocycl. Chem. 1986,23, 5. 174 67. Kontoghiorghes, G. J . ; Sheppard, L . Inorg. Chim. Acta 1987,136, L l l . 68. Cavalier, L. F. Chem Rev. 1947, 41, 525. 69. Katritzky, A . R.; Jones, R. A . /. Chem Soc. 1960, 2947. 70. Cook, D. Can. J. Chem. 1963, 41, 515,.2575. 71. Batts, B. D.; Spinner, E. Aust J. Chem. 1969,22, 2581. 72. Bellamy, L. J. The Infrared Spectra of Complex Molecules Volume 1,3rd. ed.; Chapman and Hall: New York, 1975; p 433. 73. Odinokov, S. E.; Nabiullin, A . A. ; Mashkovsky, A . A . ; Glazunov, V . P. Spectrochim. Acta 1983, 39A, 1055. 74. Bellamy, L. J. The Infrared Spectra of Complex Molecules Volume 2, 2nd. ed.; Chapman and Hall: New York, 1980; pp 240-280. 75. Bellamy, L. J . ; Rogasch, P. E. Proc. Roy. Soc. (A) 1960,257, 98. 76. Katritzkky, A . R.; Taylor, P. J. in Physical Methods in Heterocyclic Chemistry Volume TV', Katritzky, A . R., Ed.; Academic Press: New York, 1971; pp 266-432. 77. Hadzi, D.; Sheppard, N . Proc. Roy. Soc. (A) 1953, 216, 247. 78. White, R. F. M . ; Williams, H . in Physical Methods in Heterocyclic Chemistry Volume TV; Katritzky, A . R., Ed.; Academic Press: New York, 1971; pp 177-216. 79. Cox, R. H. ; Bothner-By, A . A . /. Phys. Chem. 1969, 73, 2465. 80. Penfold, B. R. Acta Crystallogra. 1953,6, 591. 81. Porter, Q. N . Mass Spectrometry of Heterocyclic Compounds, 2nd ed.; Wiley: New York, 1985; pp 603-663.. 82. Undheim, K.; Hurum, T. Acta Chem.Scand. 1972, 26, 2075. 83. Maquestiau, A. ; van Haverbeke, Y.; de Meyer, C ; Katritzky, A . R.; Cook, M . J . ; Page, A . D. Can. J. Chem. 1975, 53, 490. 84. Scarrow, R. C ; White, D. L.; Raymond, K. N . /. Am. Chem. Soc. 1985,107, 6540. 85. Pearson, R. G. /. Am. Chem. Soc. 1963, 85, 3533. 86. Pearson, R. G. /. Chem. Ed. 1987, 7, 561.; Pearson, R. G. Inorg. Chem. 1988, 27, 734. 87. Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry, 5th ed.; Wiley: New York, 1988; pp 209. 88. Greenwood, N . N . ; Earnshaw, A . Chemistry of the Elements; Pergamon Press: Oxford, 1984; pp 250-256. 175 89. Martin, R. B. Clin. Chem. 1986,32, 1797. 90. Secco, R ; Venturini, M . Inorg. Chem. 1975,14, 1978. 91. Motekaitis, R. J . ; Martell, A . E. Inorg. Chem. 1984,23, 18. 92. Letkeman, P.; Martell, A . E.; Motekaitis, R. J. /. Coord. Chem. 1980,10, 47. 93. Martin, R. B. J. Inorg. Biochem. 1986,28, 181. 94. Gregor, J . E.; Powell, H . K. J . Aust. J. Chem. 1986,39, 1851. 95. Scarrow, R. C.; Riley, P. E.; Abu-Dari, K.; White, D.L.; Raymond, K. N . Inorg. Chem. 1985,24, 954. 96. Tamhina, B.; Herak, M . J . ; Jakopcic, K. /. Less-Common Met. 1973,33, 289.; Herak, M . J . ; Tamhina, B.; Jakopcic, K. /. Inorg. Nucl. Chem. 1973,35, 1665. 97. Stunzi, H . ; Perrin, D. D.; Teitei, T.; Harris, R. L . N . Aust. J. Chem. 1979,32, 21. 98. Stunzi, H . ; Harris, R. L . N . ; Perrin, D. D.; Teitei, T. Aust. J. Chem. 1980,33, 2207. 99. Tsai, W. C.; Ling, K . - H . /. Chin. Biochem. Soc. 1973,2, 70. 100. Kontoghiorghes, G. J. Inorg. Chim. Acta 1987,135, 145 and references contained therein. 101. Nelson, W. O.; Karpishin, T. B.; Rettig, S. J . ; Orvig, C. Inorg. Chem. 1988,27, 1045. 102. Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, 3rd. ed.; Wiley-Interscience: New Y o r k , 1978; pp 249-311. 103. Kido, H . ; Saito, K. /. Am. Chem. Soc. 1988,110, 3187. 104. Albert, A . Heterocyclic Chemistry an Introduction; Athlone Press: London, 1959; p 424. 105. Klopman, G. Chemical Reactivity and Reaction Paths; Klopman, G., Ed.; Wiley: New York, 1974; pp 55-166. 106. Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell University Press: Ithaca, N.Y., 1960; pp 221-264. ' 107. Scarrow, R. C ; Raymond, K. N . Manuscript submitted to Inorg. Chem. 108. Novak, A . Struct. Bond. (Berlin) 1974,18, 111. 109. Jeffrey, G. A . ; Takagi, S. Acc. Chem. Res. 1978,11, 264. 110. Evans, R. C. An Introduction to Crystal Chemistry, 2nd ed.; Cambridge Univ. Press: Cambridge, 1964; pp 283-301. 111. Nelson, W. O.; Rettig, S. J . ; Orvig, C. /. Am. Chem. Soc. 1987,109, 4121. 176 112. Matsuba, C. A. ; Nelson, W. O.; Rettig, S. J . ; Orvig, C. Inorg. Chem. 1987,27, 3935. 113.. Nelson, W. O.; Orvig, C ; Rettig, S. J . ; Trotter, J . Can. J. Chem. 1988, 66, 132. 114. Falk, M . ; Knop, O. in Water: A Comprehensive Treatise Volume 2; Franks, F., Ed.; Plenum Press: New York - London, 1973; pp 55-113. 115. Peterson, D. W.; Levy, H . A . Acta Crystallogr. 1957,10, 70. 116. Kamb, B. in Structural Chemistry and Molecular Biology, Rich, A . , Davidson, N . , Eds.; W. H . Freeman: San Francisco, 1968; pp 507-542. 117. Saenger, W. Nature (London) 1979,279, 343. 118. Saenger, W. Nature (London) 1979,280, 848. 119. Saenger, W.; Lindner, K. Angew. Chem. Int. Ed. Engl. 1980,19, 398. 120. Neidle, S.; Berman, H . M . ; Shieh, H . S. Nature (London) 1980, 288, 129. 121. Betzel, C ; Saenger, W.; Hingerty, B.; Brown, G. M . /. Am. Chem. Soc. 1984, 106, 7545. 122. Zabel, V . ; Saenger, W.; Mason, S.A. /. Am. Chem. Soc. 1986,108, 3664. 123. Del Bene, J . ; Pople, J . A . /. Chem. Phys. 1970,52, 4858; 1973,58, 3605. 124. Jeffrey, G. A. ; McMullan, R. K. Prog. Inorg. Chem. 1967, 8, 43. 125. Jeffrey, G. A . Acc. Chem. Res. 1969, 2, 344. 126. Wells, A . F. Structural Inorganic Chemistry, 5th ed.; Clarendon Press: Oxford, 1984; p 660. 127. Mak, T. C. W. /. Chem. Phys. 1965,43, 2799. 128. Akitt, J . W. Ann. Rep. NMR Spectrosc. 1972,5A, 465. 129. Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Pitman: London , 1983; pp 66-94, 118-142, Appendix 2. 130. Hinton, J. F.; Briggs, R. W. in NMR and the Periodic Table; Harris, R. K., Mann, B.E., Eds.; Academic Press: London, 1978; pp 279-308. 131. Delpuech, J. J. in NMR of Newly Accessible Nuclei Volume 2; Laszlo, P., Ed.; Academic Press: London, 1983; pp 153-195. 132. Karlik, S. J . ; Elgavish, G. A . ; Pillai, R. P.; Eichhorn, G. L. /. Magn. Reson. 1982, 49, 164. 133. Karlik, S. J . ; Elgavish, G. A. ; Eichhorn, G. L. /. Am. Chem. Soc. 1983,105, 602. 134. Greenaway, F. T. Inorg. Chim. Acta 1986, 116, L21. 177 135. Bertsch, P. M . ; Barnhisel, R. I.; Thomas, G.W.; Layton, W. J . ; Smith, S. L. Anal. Chem. 1986,58, 2583. 136. Akitt, J . W.; Mil ic , N . B. J. Chem. Soc, Dalton Trans. 1984, 981 and references therein. 137. Llinas, M . ; De Marco, A . /. Am. Chem. Soc, 1980,102, 2226. 138. Jaber, M . ; Bertin, R ; Thomas-David, G. Can. J. Chem. 1977, 55, 3689. 139. Karlik, S. J . ; Elgavish, G. A. ; Eichhorn, G. L. Inorg. Chem. 1983,22, 525. 140. Akitt, J . W.; Mann, B. E. /. Magn. Reson. 1981,44, 584. 141. Benn, R.; Ruflinska, A. ; Janssen, E.; Lehmkuhl, H . Organometallics 1986,5, 825. 142. Girgis, A . Y. ; Fay, R. C. /. Am. Chem. Soc. 1979, 92, 7061. 143. Gordon, J . G.; Holm, R. H . /. Am. Chem. Soc. 1970, 92, 5319. 144. Pignolet, L . H . Top. Curr. Chem. 1975,56, 91. 145. Henderson, D. E.; Saltzman, J . J . ; Uden, P. C ; Cheng, Z. Polyhedron 1988, 7, 369. 146. Fay, R. C ; Piper, T. S. /. Am. Chem. Soc. 1963, 85, 500. 147. Hatakeyama, Y. ; Kido, H. ; Harada, M . ; Tomiyasu, H. ; Fukutomi, H . Inorg. Chem. 1988, 27, 992. 148. Gennaro, M . C ; Mirti , P.; Casalino, C. Polyhedron 1983,2, 13. 149. Eaton, S. S.; Hutchinson, J . R.; Holm, R. H. ; Muetterties, E. L. /. Am. Chem. Soc. 1972, 94, 6411. 150. Eaton, S. S.; Eaton, G. R.; Holm, R. H . ; Muetterties, E. L. /. Am. Chem. Soc. 1973,95, 1116. 151. Binsch, B. G.; Kleier, D. A . , Quantum Chemistry Program Exchange QPCE, Program No. 165; DNMR3, Chemistry Department, Indiana University. 152. Gutowsky, H . S.; Holm, C. H . /. Chem Phys. 1956,25, 1228. 153. Fay, R. C ; Piper, T. S. Inorg. Chem. 1964,3, 348. 154. Glasstone, S.; Laidler, K.; Eyring, H . The Theory of Rate Processes; McGraw-H i l l : New York, 1941; p 190. 155. Allerhand, A . ; Gutowsky, H . S.; Jonas, J . ; Meinzer, R. A . /. Am. Chem. Soc. 1966, 88, 3185. 156. Pinnavaia, T. J . ; Sebeson, J . M . ; Case, D. A . Inorg. Chem. 1969,8, 644. 157. Grossman, D. L.; Haworth, D. T. Inorg. Chim. Acta 1984, 84, L17. 178 158. Basolo, F.; Pearson, R. G. Mechanisms of Inorganic Reactions, 2nd ed.; Wiley: New York, 1967; pp 300-334. 159. Fay, R. C.; Piper, T. S. /. Am. Chem.Soc. 1962, 84, 2303. 160. Hutchinson, J. R.; Gordon, J . G.; Holm, R. H . Inorg. Chem. 1971,10, 1004. 161. Imafuku, K.; Takahashi, K.; Matsumura, H . Bull. Chem. Soc. Jpn. 1979,52, 111. 162. Silverstein, R. M . ; Bassler, G. C.; Morrill, T. C. Spectrometric Identification of Organic Compounds, 4th ed.; Wiley: New York, 1981; pp 305-331. 163. Mason, S. F. in Physical Methods in Heterocyclic Chemistry, Volume II; Katritsky, A . R.; Ed., Academic Press: New York & London, 1963; pp 1-84. 164. Geiger, A . ; Mausbach, P.; Schnitker, J. in Water and Aqueous Solutions; Neilson, G.W., Enderby, J. E.; Eds.; Adam Hilgef: Bristol, 1986; pp 15-30. 165. Leo, A . ; Hansch, C.; Elkins, D. Chem. Rev. 1971, 71, 525. 166. Meyer, H . Arch. Exptl. Pathol. Pharmakol. 1899,42, 110. 167. Helmer, F.; Hansch, C.; Kiehs, K. Biochemistry 1968, 7, 2858. 168. Yokel, R. A. ; Kostenbauder, H . B. Toxicol. Appl. Pharmacol. 1987, 91, 281. 169. Leo, A . J . /. Pharm. Sci. 1987, 76, 166. 170. Partition Coefficient Determination and Estimation; Dunn IV, W. J . , Block, J. H . , Pearlman, R.S., Eds.; Pergamon Press: Oxford, 1986; p 154. 171. Dunn III, W. J. ; Koehler, M . G.; Grigoras, S. /. Med. Chem. 1987, 30, 1121. 172. Clarke, F. H. ; Cahoon, N . M . ; /. Pharm. Sci. 1987, 76, 611. 173. Gobas, F. A . P. C ; Lahittete, J. M . ; Garofalo, G.; Shiu, W. Y.; Mackay, D. /. Pharm. Sci. 1988, 77, 265. 174. De Kock, A . C ; Lord, D. A . Chemosphere 1987,16, 133. 175. Herz, W. "Der Verteilungssatz", Ferdinand Enke, Stuttgart 1909, p 5. 176. Purcell, W. P.; Bass, G. G.; Clayton, J . M.Strategy of Drug Design: a guide to biological activity; Wiley: New York; Appendix I pp 126-143. 177. Arnold, A . P.; Canty, A . J . ; Moors, D. W.; Deacon, G. B. /. Inorg. Biochem. 1983,19, 319. 178. McLachlan, D. R. Personal communication. 179. Bassett, J . ; Denney, R. C ; Jeffery, G. H . ; Mendham, J . Vogel's Textbook of Quantitative Inorganic Analysis, 4th ed.; Longman: London and New York, 1985; pp 834-835. 179 180. Kontoghiorghes, G. Inorg. Chim. Acta 1988, 257, 101. 181. Hansch, C. Personal communication. 182. Lutz, T. G.; Clevette, D. J . ; Rettig, S. J . ; Orvig, C. Submitted for publication. 183. Clevette, D. J . ; Nelson, W. O.; Nordin, A. ; Orvig, C ; Sjoberg, S. Submitted for publication. 184. Ohman, L. -O. Inorg. Chem. 1988,27, 2565. 185. Martell, A . E.; Smith, R. M . Critical Stability Constants; Plenum: New York, 1974-1982; Vols. 1-5. 186. Ohman, L. -O.; Sjoberg, S. Polyhedron 1983,2, 1329. 187. Larson, S. M . Seminars in Nuclear Medicine 1978, 8, 193. 188. Lyster, D. M . ; Clevette, D. J . ; Nelson, W. O.; Rihela, T.; Webb, G. A. ; Orvig, C. Manuscript in preparation. 189. Langston, W. J. ; Bryan, G. W. in Complexation of Trace Metals in Natural Waters; Kramer, C. J. M . , Duinker, J. C , Eds.; Martinus Nijhoff/Dr W. Junk: The Hague, 1984; pp 375-392. 190. Williams, R. J. P. Proc. R. Soc. (B) 1981,213, 361. 191. Gran, G. The Analyst 1952, 77, 661; Anal. Chim. Acta 1988,206, 111. 192. Motekaitis, R. J . ; Martell, A . E. Can. J. Chem. 1982,60, 168. 193. Motekaitis, R. J . ; Martell, A . E. Can. J. Chem. 1982,60, 2403. 180 Appendix Table A . l . Molecular weights (MW) of the 3-hydroxy-4-pyridinone ligands and their tris-ligand metal complexes. (The entries in italics are for the 3-hydroxy-4-pyrone ligands) Ligand M W M W of Tris-Ligand Metal Complexes AIL3 GaL3 InL 3 Hmpp 125.1 399.4 442.1 487.2 Hdpp 139.1 441.2 484.0 529.1 Hmepp 153.1 483.2 526.0 571.1 Mimosine 198.2 618.6 661.3 706.4 Hmhpp 209.3 651.8 694.6 739.7 H2exn* 332.4 1048 1134 1224 Maltol 126.2 402.3 445.0 490.1 Kojic acid 142.1 450.2 493.0 538.1 •Molecular weights are for the M2L3 dimer. 181 Table A.2. Crystallographic data for the 3-hydroxy-4-pyridinone ligands. compound Hmpp Hdpp Hmepp formula C6H7NO2 C7H9NO2 C 8 H n N 0 2 formula weight 125.1 139.2 153.2 crystal system monoclinic orthorhombic othorhombic space group Pl\ln Pbca Pbca a (A) 6.8351 (4) 7.3036 (4) 12.5907 (8) b(A) 10.2249 (4) 13.0490 (6) 11.7477 (6) c(A) 8.6525 (4) 13.7681 (7) 11.0040(6) P(deg) 105.215 (4) V(A3) 583.51 (6) 1312.2 (1) 1627.6 (3) Z 4 8 8 D c (g/cm3) 1.424 1.409 1.14 F(000) 264 592 592 diffractometer Enraf-Nonius CAD4-F Enraf-Nonius Rigaku AFC6 u (Cu-K a ) (cm-l) 8.65 8.21 6.62 radiation Cu-Ka C u - K a C u - K a W A ) 1.540562 1.540562 1.54178 1.544390 1.544390 N i filter N i filter graphite-monochromated temperature 23°C 23°C 23°C Mmax (deg) 150 150 150.1 reflections with I > 3a (I) 914 857 1228 number of variables 111 128 145 R; Rw 0.037; 0.046 0.044; 0.046 0.053; 0.085 max A/a (final cycle) 0.05 0.05 0.02 goodness of fit indicator 2.349 1.014 3.50 residual density (e/A3) 0.20 0.32 0.45 182 Table A.3. Crystallographic data for the M(dpp) 3 complexes (recorded with a Enraf-Nonius CAD4-F diffractometer). compound Al(dpp) 3 -12H 2 0 Ga(dpp)3-12H 20 In(dpp) 3-12H 20 formula C21H48AIN3O18 C 2 iH48GaN30i8 C2iH48lnN 3 Oi 8 formula weight 657.6 700.3 745.44 crystal system Trigonal Trigonal Trigonal space group P3 P3 P3 a (A) 16.600 (2) 16.6549 (6) 16.842 (1) c(A) 6.877 (1) 6.8691 (4) 6.8078 (7) V(A3) 1641.3 (3) 1650.1 (1) 1672.3 (2) Z 2 2 2 D c (g/cm.3) 1.331 1.470 1.480 F(000) 704 740 776 radiation C u - K a Cu-K« M o - K a W(A) 1.540562 1.540562 0.70930 1.54439 1.54439 0.71359 nickel filter nickel filter graphite monochromator p (cm - 1) 11.97 17.89 7.67 temperature 22°C 22°C 21°C 2Qmax (deg) 150 150 60 reflections with I > 3a (I) 1662 1653 2496 number of variables 202 195 190 R; Rw 0.045; 0.051 0.047; 0.055 0.033; 0.037 max A/a (final cycle) 0.17 0.023 0.027 goodness of fit indicator 1.020 1.023 1.492 residual density (e/A3) 0.23 0.48 -0.55 to +0.75 (near In) 183 Table A.4. Crystallographic data for the M(mepp)3 complexes (recorded with a Rigaku AFC6). compound Al(mepp)3-12H20 Ga(mepp)3«12H20 formula C24H54AIN3O18 C24H54GaN3Oi8 formula weight 699.7 742.4 crystal system Trigonal Trigonal space group P3 P3 a (A) 17.1734 (8) 17.247 (1) c (A) 6.827 (1) 6.830 (2) V(A3) 1743.7 (3) 1759.4 (1) z 2 2 D c (g/cm3) 1.33 1.40 F(000) 752 788 radiation C u - K a M o - K a tea (A) 1.54178 0.71069 graphite-monochromated graphite-monochromated p (cm-1) 11.56 8.50 temperature 21°C 21°C 2Qmax (deg) 150.3 55.0 reflections with I > 3a (I) 1157 1918 number of variables 207 215 R; Rw 0.032; 0.038 0.029; 0.036 max A/a (final cycle) 0.23 0.02 goodness of fit indicator 1.63 1.46 residual density (e/A 3) 0.10 0.28 184 Table A.5. Hydrogen bond distances (A) and angles for In(dpp)3«12H20. Atoms Interaction O-H (A) H O (A) o-o (A) O - H - O (deg) 0(3)-H(03a)-0(l) 1.00 1.92 2.900(3) 165 0(3)-H(03b)-0(2) 0.87(3) 1.99(3) 2.839(3) 168(3) 0(4)-H(04a)-0(6) 0.70(4) 2.07(5) 2.744(4) 160(5) 0(4)-H(04b)-0(4) 0.87(9) 1.90(9) 2.755(3) 169(6) 0(4)-H(04c)-0(4) 0.72(8) 2.04(8) 2.755(3) 170(6) »0(5)-H(05a)-O(4) 0.82 1.99 2.789(3) 166 0(5)-H(05b)-0(5) 0.77(5) 2.01(5) 2.782(3) 179(5) 0(6)-H(06a)~0(3) 0.66(4) 2.10(4) 2.759(4) 177(5) 0(6)-H(06b)-0(5) 0.74(4) 2.06(4) 2.780(4) 164(4) Table A.6. Hydrogen bond distances (A) and angles for the M(mepp)3«12 H2O complexes. Interaction O - H H- 0 0 - •O O - H - 0 (deg) A l Ga A l Ga A l Ga A l Ga 0(3)-H(l)-0(l) 0.81(5) 0.75(4) 2.10(5) 2.14(4) 2.877(2) 2.883(3) 161(4) 173(4) 0(3)-H(2) 0(2) 0.92(7) 0.77(3) 1.95(7) 2.08(3) 2.838(2) 2.834(3) 163(5) 168(3) 0(4)-H(3)-0(6) 0.83(7) 0.85(4) 1.99(7) 1.95(7) 2.795(3) 2.784(3) 166(4) 167(4) 0(4)-H(4)-0(4) 0.85(8) 0.71(7) 1.97(7) 2.11(7) 2.811(2) 2.810(3) 170(6) 172(5) 0(5)-H(6)-0(4) 0.89(1) 0.81(3) 2.04(6) 2.03(4) 2.828(3) 2.821(3) 145.83 164(3) 0(5)-H(5)-0(5) 0.82(6) 0.76(3) 2.02(6) 2.08(4) 2.833(2) 2.835(3) 171(4) 175(4) 0(6)-H(7)-0(3) 0.91(6) 0.79(4) 1.86(6) 2.00(4) 2.771(3) 2.763(3) 173(4) 165(4) 0(6)-H(8)-0(5) 0.76(8) 0.77(4) 2.11(8) 2.12(4) 2.859(3) 2.849(4) 168(5) 159(4) 185 Appendix Procedure A . l Potentiometric Equilibrium Measurements Potentiometric measurements of the ligands in the absence, and presence, of metal ions were performed with an Orion Research E A 920 pH meter equipped with Orion Ross research grade glass and reference electrodes. A Metrohm automatic buret (Dosimat 665) was used to add the standard NaOH. The temperature was maintained at 25.0 ±0.1 °C throughout with water-jacketed beakers and a Julabo circulating bath, and the ionic strength was adjusted to 0.15 M (isotonic) by the addition of NaCl. A l l solutions were continuously degassed with prepurified A r during the course of a titration. The ligands were twice recrystallized or sublimed; concentrations were obtained by weighing. A l l metal-containing solutions were obtained from appropriate dilution of atomic absorption standard solutions of A l and Ga (Sigma or Aldrich). The exact amount of excess acid present in the metal ion solutions was determined by a Gran's p l o t 1 9 1 of (V G + V t ) x 10 _P H vs. V t , where V G = the initial volume of 1:1 metal-Na2EDTA solution, and V t is the volume of added standard NaOH. The base consumed is equal to the excess acid plus the N a 2 E D T A protons. The metal-ligand titrations were performed at a total ligand (C) to total metal ion (B) concentration ratio of just greater than three. NaOH solutions (0.1 M) were prepared from dilutions of 50% NaOH (less than 0.1% N a 2 C 0 3 ) with freshly boiled, distilled, deionized water and standardized potentiometrically against potassium hydrogen phthalate (KHP). The electrodes were calibrated with standard aqueous HC1 and NaOH solutions to read - log[H + ] directly. The range of -log[H + ] available was limited from 1.5 to 12 in which the electrode behavior was reversible and linear. Protonation and deprotonation reactions of the ligands were studied within the range 2 < -log[H +] < 10 and the metal-ligand titrations were studied within the range 1.5 < -log[H+] < 4. A l l titrations were performed in sets of 3 or 4 runs. 186 The proton dissociation constant of the ligands were determined by using the Fortran computer program P K A S . 1 9 2 In the M(IU) systems, the computations allowed for the presence of M(OH)2+, M ( O H ) 2 + , M ( O H ) 3 and M ( O H ) 4 \ In addition, A l 2 ( O H ) 2 4 + and A l 3 ( O H ) 4 5 + were included. Formation constants for these various metal species were taken from ref. 25. The stability constants for the main species M L 2 + , M L 2 + , and ML3 were determined by using the Fortran computer program B E S T . 1 9 3 This program sets up simultaneous mass-balance equations for all the components present at each addition of base, and calculates the pH at each data point according to the current set of stability constants and total concentrations of each component. Stability constants judiciously chosen by the user are automatically adjusted in order to minimize the sum of squares of differences between the calculated and observed values of - log[H + j . Adjustment is continued until there is no further improvement in the fit. The constants are reported to the second decimal place, which is representative of the reproducibility of the potentiometric equipment employed. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0060259/manifest

Comment

Related Items