Studies of Oriented Aromatic Hydrocarbons in the Phos Phorescent State at Low Magnetic Fields

WINER, Arthur Melvyn, 1942- ELECTRON PARAMAGNETIC RESONANCE STUDIES OF ORIENTED AROMATIC HYDROCARBONS IN THE PHOS PHORESCENT STATE AT LOW MAGNETIC FIELDS.

The Ohio State University, Ph.D., 1969 Chemistry, physical

University Microfilms, Inc., Ann Arbor, Michigan

THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED ELECTRON PARAMAGNETIC RESONANCE STUDIES OF ORIENTED AROMATIC HYDROCARBONS IN THE PHOSPHORESCENT STATE AT LOW MAGNETIC FIELDS

DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By Arthur Melvyn Winer, B.S, ***** The Ohio State University 1969

Approved by

Adviser Department of Chemistry To My Father

ii ACKNOWLEDGMENTS

It is a pleasure to acknowledge the contribution to this work of Professor Roger E. Gerkin, who not only pro vided direction and encouragement throughout the research, for which I am grateful, but who also established an atmos phere of conviviality which made my tenure as his student most enjoyable. Valuable discussions with Mr. William J. Rogers, Mr. Peter Szerenyi and Mr. David L. Thorsell of this labora tory and with Professor Clyde A. Hutchison of the Univer sity of Chicago are gratefully acknowledged. I wish to thank Mr. Thorsell also for taking the photographs of research equipment which appear in the dissertation. The work of Mr. Terence Ireland in maintaining and improving the electronic equipment was essential to the completion of this research and was greatly appreciated. I wish to thank Professor Jack G. Calvert, Dr. Susan Collier and Dr. T. Navaneeth Rao for use of the Turner spectrophotometer, and Professor Peter W. R. Corfield for

iii iv use of an x-ray diffractometer and for his consultation concerning the crystal structure analysis of p-terphenyl. Fellowships from the Gulf Oil Corporation and the Continental Oil Company provided helpful financial assis tance during the final two years of research. Computer facilities were provided by The Ohio State University Computer Center. I am deeply grateful to ray parents for their con tinuous support and encouragement throughout the years of ray education. Finally, I would like to express my sincere gratitude to ray wife, Jane, who stood by me through the many and sometimes difficult years of study and research, and who therefore deserves to share fully in this accomplishment. ip

VITA

1942, Hay 5 Born - New York City, New York 1962-1964 Laboratory Assistant, Korad Corporation, Santa Monica, California 1964 National Science Foundation Undergraduate Research Participant, Department of Chemistry, University of California at Los Angeles, Los Angeles, California 1964 B.S. Chemistry, University of California at Los Angeles, Los Angeles, California 1964-1966 Teaching Assistant, Department of Chemistry, The Ohio State University, Columbus, Ohio 1966-1967 Research Assistant, Department of Chemistry, The Ohio State University, Columbus, Ohio 1967-1968 Gulf Fellow, Department of Chemistry, The Ohio State University, Columbus, Ohio 1968-1969 Continental Oil Fellow, Department of Chemistry, The Ohio State University, Columbus, Ohio

PUBLICATIONS "The Decay of 0£ (a ) in Flow Experiments^' J. Phys. Chem. 70, 302 (1966). Co-authored with Dr. Kyle D. Bayes, Department of Chemistry, University of California at Los Angeles, Los Angeles, California

v vi

"Observation of Multiplet Structure in Low-Field Magnetic Resonance Spectra of Triplet States of Oriented Aromatic Hydrocarbons." J. Chera. Phys. 47, 2504 (1967). Co-authored with Dr. Roger E. GerkinTHDepartment of Chemistry, The Ohio State University, Columbus, Ohio "Deuterium Isotope Effect in Zero-Field Splittings of Phosphorescent Fhenanthrene Oriented in Biphenyl." J. Chem. Phys. 50, 3114 (1969). Co-authored with Dr. Roger E. Gerkin7"~bepartment of Chemistry, The Ohio State University, Columbus, Ohio "Proton Hyperfine Structure at Magnetic Fields Below — 100 G in Electron Magnetic Resonance Absorptions by Triplet States of Aromatic Molecules." J. Chera. Phys. 51, 1664 (1969). Co-authored with Dr. Roger E. Gerkin7*~bepartraent of Chemistry, The Ohio State University, Columbus, Ohio

FIELDS OF STUDY Major Field: Physical Chemistry Studies in Kinetics. Professor Frank Verhoek Studies in Thermodynamics. Professors George E. MacWood and David White Studies in Quantum Mechanics. Professor William J. Taylor TABLE OF CONTENTS

DEDICATION...... ii ACKNOWLEDGMENTS ...... iii VITA ...... v LIST OF TABLES ...... xi LIST OF FIGURES ...... xiv

Chapter I. INTRODUCTION AND HISTORICAL BACKGROUND .. 1 Introduction Historical Background II. THEORY...... 9 Introduction The Spin Hamiltonian Solution of the Fine-Structure Spin Hamiltonian The Low-Field Eigenvalues III. EXPERIMENTAL APPARATUS AND METHODS ..... 18 The Low Field Spectrometer Static Magnetic Fields The Helmholtz Magnet Static-Field Measurements Cancellation of the Laboratory Field Calibration of the Helmholtz Magnet Homogeneity of the Resultant Field Microwave Fields Frequency Counting Signal Detection Optical System Crystal Temperature The High Field Spectrometer

vii viii Chapter Page IV. PREPARATION OF SAMPLES...... 42 Purification of Materials Guest Materials Chrysene-h^ 2 2,3-Benzocarbazole Chrysene-di2 1,2; 5,6-Dibenzanthracene-h^4 1,2; 5,6-Dibenzanthracene-d^4 Naphthalene-hg Naphthalene-dg Phenanthrene-nxo Phenanthrene-d^Q Dibenzothiophene 1,2,3,4-Tetrahydroanthracene Picene Host Materials p-Terphenyl-hi4 p-Terphenyl-dj* Symmetric Octanydroanthracene Biphenyl Durene 2,2'-Binaphthyl p-Quaterphenyl 2,6-Dimethylnaphthalene Growth of Single Crystals Host Crystal Structures p-Terphenyl Symmetric Octahydroanthracene 2,2*-Binaphthyl EXPERIMENTAL RESULTS AND DISCUSSION FOR ORIENTED CHRYSENE IN ITS PHOSPHORESCENT S T A T E ...... 68 Introduction Multiple Resonance Absorption in Low-Field Fine-Structure Spectra of Triplet Chrysene Oriented in p-Terphenyl Zero-Field Splittings and Spin Hamiltonian Parameters for Chrysene Basic Relations Triplet Chrysene-d^ in Single Crystals of p-Terphenyl-h^ 4 Triplet Chrysene-d^ 2 in Single Crystals of p-Terphenyl-di4 Triplet Chrysene-d^ in Single Crystals of s-Octahydroanthracene Triplet Chrysene-dio in Single Crystals of Biphenyl ix

Chapter Page

Multiple Orientation of Chrysene Triplets in Single-Crystal Hosts Lifetime of the Triplet State of Chrysene-d^ Chrysene-d^ 2 1° Single-Crystal Hosts p-Terphenyl-hi4 Host s-Octahydroanthracene Host Chrysene-di2 1° EPA Glasses Temperature Dependence of the Zero-Field Splittings of Chrysene-di2 in p- Terphenyl-hi4 and s-Octanydroanthracene Deuteration Effects Deuteration of Chrysene Deuteration of p-Terphenyl Conclusion VI. EXPERIMENTAL RESULTS AND DISCUSSION FOR OTHER MIXED CRYSTAL SYSTEMS...... 160 1,2; 5,6-Dibenzanthracene Oriented in p- Terphenyl and in 2,2*-Binaphthyl 2,3-Benzocarbazole Oriented in p-Terphenyl and in Chrysene Impurity in s-Octahydroanthracene VII. DEUTERIUM ISOTOPE EFFECTS IN ZERO-FIELD SPLITTINGS OF PHOSPHORESCENT PHENANTHRENE ORIENTED IN BIPHENYL...... 170 Introduction Results for Phenanthrene Oriented in Biphenyl Discussion VIII. PROTON HYPERFINE STRUCTURE AT MAGNETIC FIELDS BELOW -100 GAUSS IN MAGNETIC RESONANCE AB SORPTIONS BY TRIPLET STATES OF AROMATIC HYDROCARBONS ...... 192 Introduction Results and Discussion Reasons for Previous Failures to Observe Hyperfine Structure at Low Fields Comparisons of High-Field and Low-Field Hyperfine 'Patterns Low Fields Between — 75 and — 100 G Zero Field and Fields Below — 50 G Effects of Resolved Proton Hyperfine Structure on Precision Determinations of Zero- Field Splittings at Low Fields Chapter Page

IX. SUGGESTIONS FOR FURTHER RESEARCH ...... 213 Introduction Mixed Crystal Systems to be Investigated Picene in p-Terphenyl p-Terphenyl-h^ and p-Terphenyl-d^ In 4,> Dimethylbiphenyl p-Quaterphenyl Investigation of Multiplet Structure in Optical Spectra of Organic Mixed Crystals APPENDIX A. Formulas for g-Tensor Diagonal Elements ... . 220 B. Computer Program Listings ...... 222 Calculation of g-Tensor Diagonal Elements Least-Squares Calculations Calculation of Limiting Phosphorescent Lifetime REFERENCES CITED ...... 230 LIST OF TABLES

Table Page 1. Emission spectrum of chrysene-d^2 EPA glass at 77K...... 46 2. Magnetic resonance data for chrysene-d^2 oriented in p-terphenyl: Transitions near - 0.116 cm"1...... 89 3. Magnetic resonance data for chrysene-di2 oriented in p-terphenyl: Transitions near - 0.063 cra~l...... 90 4. Magnetic resonance data for chrysene-d^2 oriented in p-terphenyl: Transitions near - 0.053 cm"1...... 91 5. Triples of best values of zero-field splittings for chrysene"di2 in p-terphenyl 97 6. Predicted and experimental values of lD + E[/hc for chrysene-d^2 p-terphenyl..... 99 7. Magnetic resonance data for chrysene-d^o oriented in s-octahydroanthracene and in biphenyl...... 102 8. Zero-field splittings and spin Hamiltonian parameters of chrysene-d^2 *n s-octahydroanthracene...... 104 9. Angles between the projection of H in the BC plane and B]_ and B2 for chrysene'triplet states in p-terphenyl...... 117 10. Angles between H and the B fine structure axes of the two translationally inequivalent triplets corresponding to each of Signals 14-18 in Run 13 a,b...... 121

xi Table Page

11. Magnet base angles corresponding to closest approach of H to B fine-structure axes of component triplet states in the -'0.053 cra-1 multiplet.»».»•*».•»•*»•»•••••••»»»»•»»»••»»•••» ,124 12. Comparison of differences among angles be tween H and B fine-structure axes given in Tables~9, 10 and 11 for the various multiplet signals.»••»•••«***••••••••••••»••••»• 129 13. Lifetime of the lowest triplet state of chrysene-di2 In p-terphenyl-hi4 from - 64K to — 90K...... 133 14. Temperature coefficients of zero-field splittings and spin Hamiltonian parameters for chrysene-di2 1 ° p-terphenyl and s- octahydroan thracene...... 149 15. Selected D - E zero-field splittings of chrysene-hi2 and chrysene-dj^ oriented in p-terphenyl at — 77K...... 152 16. D - E and 2E zero-field splittings of chrysene-di2 In p-terphenyl-hj. 4 and p- terphenyl-di4 ...... 154 17. Zero-field splittings of 1,2; 5,6-dibenzanthracene oriented in p-terphenyl-hi4 ...... 162 18. Nominal concentrations of phenanthrene in biphenyl mixed crystals used in isotope- effect study...... 175 19. Resonance absorption linewidths and signal- to-noise ratios for phenanthrene-hô and phenanthrene-dô in biphenyl...... 181 20. Magnetic resonance data for phenanthrene- dô In biphenyl at 77K...... 182 xiii

Table Page

21. Zero-field splittings and spin Hamiltonian parameters of phenanthrene-hio and phenanthrene-dxQ in biphenyl at 77K...... 184 22. Zero-field splittings and spin Hamiltonian parameters of naphthalene-hg and naphthalene-dg in biphenyl at 77K...... 189 LIST OF FIGURES

Figure Page 1. Behavior of the triplet energy levels at low fields ...... 17 2. The low-field Helmholtz magnet...... 23 3. The low-field header...... 31 4. Block diagram of the low-field spectrometer...... 36 5. Emission spectrum of chrysene-d-t o in EPA glass at 77K...... 43 6. Disconnectable vacuum manifold, powder trap and crystal tube used in preparation of mixed crystals...... 53 7. Mixed single-crystal samples in various stages of preparation...... 55 8. Bridgman furnace...... 57 9. p-Terphenyl crystal structure...... 63 10. p-Terphenyl axis system...... 66 11. Chrysene axis system...... 70 12. Incompletely resolved resonance absorptions from chrysene-dio ln p-terphenyl at — 0.126 cm"1 and ' 275 G...... 73 13. Resolved multiple resonance absorptions from chrysene-d^2 ip p-terphenyl below — 120 G, at — 0.117 cm"1...... 75

xiv XV

Figure

14. Multi-absorption pattern observed at low fields for the chrysene-d^2 (in p-terphenyl) transi tions near — 0.063 cra~l...... 77 15. Multi-absorption pattern observed at low field for the chrysene-d^ 2 (in p-terphenyl) transi tions near — 0.054 cra**l...... 79 16. Resonance absorption from chrysene-d^o observed at — 2200 G for H very nearly parallel to the A axis of one translationally inequivalent set of chrysene triplets...... 82 17. Plot of data obtained in precision determination of chrysene-d^2 (in p-terphenyl-hi4 ) zero- field splittings near — 0.116 cm"!...... 93 18. Magnetic resonance decays for chrysene-d^2 in p-terphenyl-hi4...... 132 19. Triplet lifetime as a function of temperature for chrysene-d^ 2 in single-crystal hosts...... 134 20. Zero-field splittings of chrysene-di2 in s- octahydroanthracene as a function of temperature...... 146 21. Multi-absorption pattern observed at low fields for the 1,2; 5,6-dibenzanthracene-di4 in p-terphenyl transition near — 0.115 cm"l...... 163 22. Resolved electron magnetic resonance absorptions at — 0.147 cra"l by triplet phenanthrene-d^o and triplet phenanthrene-hig present together in a biphenyl single crystal at 77 K...... 177 23. Proton hyperfine structure in electron magnetic resonance absorptions by triplet naphthalene oriented in biphenyl at low and high fields 198 24. Proton hyperfine structure in electron magnetic resonance absorptions by triplet phenanthrene, and by triplet dibenzothiopliene at lowfields. •• 200 xvi

Figure 25. Proton hyperfine structure in electron mag netic resonance absorptions by triplet naphthalene at low fields...... 204 26. Plot of data obtained in precision determina tion of phenanthrene-hio (*-n biphenyl) zero- field splitting near — 0.093 cm~l...... 210 CHAPTER I INTRODUCTION AND HISTORICAL BACKGROUND

A. INTRODUCTION During the past decade there has been considerable interest in the experimental determination of the energy differences between the components of the lowest triplet state of aromatic molecules at zero magnetic field (so- called zero-field splittings). Zero-field splittings arise from the magnetic dipole-dipole interaction of two unpaired pi electrons and are one of the fundamental properties of aromatic molecules, related to the electron spin distribu tion and to the molecular symmetry (of the lowest triplet state) in such molecules. The precise values of zero-field splittings are also sensitive to environmental effects such as molecular crystal fields, as well as to intramolecular isotope, atom, and functional group substitution. Additio nally, the goal of calculating splittings which are in good agreement with precise experimental values continues to challenge the models and methods of theoretical chemists. Electron paramagnetic resonance absorption spectroscopy

1 2 is well suited to investigation of zero-field splittings, as well as other properties of the phosphorescent state of aromatic molecules such as triplet lifetimes and proton hyperfine interactions. Three principal kinds of electron magnetic resonance studies of triplet states have been car ried out, by various workers cited below. Of the three, the "low-field method" affords the greatest precision in the determination of zero-field splittings. Despite the high precision permitted by the low-field method, the only molecules to which it had been applied prior to initiation of this work were naphthalene and phe nanthrene (and an as yet unidentified impurity in fluorene). Accordingly, the original goal of this research was to undertake a low-field investigation of more- extensively-conjugated aromatic molecules beginning with chrysene and 1,2;5,6 dibenzanthracene. Experiments were carried out using the spectrometer and experimental methods discussed in detail in Chapter III, on samples prepared as described in Chapter XV. At an early stage in the research multiple resonance absorptions were found for chrysene oriented in single crystals of p-terphenyl. The observation of a number of sets (or triples) of zero-field splittings led to a greatly expanded investigation of the chrysene in p- terphenyl system, since only a single triple of splittings was expected as in the case of previous low-field experi ments for naphthalene and phenanthrene.^-*^ The results of studies of the "multiplet structure" in the low-field mag netic resonance spectra of chrysene-hj^ an(* chrysene-di2 oriented in biphenyl, in p-terphenyl-h^ and in sym- octahydroanthracene single crystals are given in Chapter V. Data from studies of dibenzanthracene for which multi plet structure was also observed (for p-terphenyl host), and from investigations- of the impurities detected by mag netic resonance absorption during the chrysene research, are presented in Chapter VI. Two additional novel effects were established in the course of the chrysene research. First, deuterium iso tope shifts in the zero-field splittings of phenanthrene (and subsequently of naphthalene) were established unambig uously for the first time. Then, resolved proton hyper fine structure was observed in low-field resonance absorp tions from naphthalene, phenanthrene and dibenzothiophene in contrast to failures to observe such structure in the previous low-field studies (for naphthalene and phenan threne) . The results of investigations of the deuterium isotope effect in zero-field splittings, and of resolved proton hyperfine structure at fields below 100 gauss, are presented in Chapters VII'and VIII, respectively. 4

B. HISTORICAL BACKGROUND Electron paramagnetic resonance absorption was first o observed in 1945 by Zavoisky. At this time also, Lewis and Kasha demonstrated that the origin of the long-lived phosphorescence emission by aromatic molecules (in rigid solvents and at low temperatures) is the lowest triplet state.They pointed out that the triplet-state mecha nism for phosphorescence requires the occurrence of photo- magnetic effects, and subsequent static magnetic suscepti bility experiments on phosphorescing systems did establish such photomagnetisra.a~fi—8 However, attempts to observe para magnetic resonance absorption in systems for which static field susceptibility experiments had been carried out were unsuccessful.^ Weissraan pointed out that this failure could be due in part to anisotropy broadening of absorp tions (from randomly oriented samples) due to a non degeneracy of the three magnetic sublevels of the triplet state in zero magnetic field.^ This explanation for the previous failures was subsequently shown to be correct. Upon studying naphthalene oriented in a durene single crystal and thereby eliminating anisotropy broadening, Hutchison and Mangum in 1958 reported electron paramagnetic resonance absorption by an aromatic molecule in its phos phorescent state for the first time.^ In their experi ments, Hutchison and Mangura confirmed the triplet character 5 of the phosphorescent state of naphthalene, obtained values for the zero-field splittings and spin-Hamiltonian parameters (designated D and E), and observed resolved proton hyperfine structure in the triplet resonance absorptions. 12 These experiments were carried out at X-band and K-band frequencies, for which resonance absorptions were observed at fields between approximately 2,000 and 10,000 gauss ("high-field"). The absorptions corresponded to transitions (between the triplet levels) satisfying the selection rule A M g = i 1. Electron magnetic resonance absorption was subsequent ly observed for triplets randomly oriented in glasses by van der Waals and de Groot,*3,14 who observed the so- called AMs= i 2 transitions for which the resonant fields are relatively independent of molecular orientation and are of intermediate magnitude (— 1500 gauss for X-band radiation). Yager, Wasserman and Cramer later showed that AMs= i l transitions could be observed from the relatively few triplet molecules in a randomly oriented sample for which one of the three principal magnetic axes is approxi mately aligned with the external field.15,16 Because of the greater convenience of experiments on randomly oriented samples, zero-field splittings and values of D* (D*= (d 2 + 3 e ^)^) have been measured for a very large number of molecules in glasses and plastics by many workers,13-27 wh£ie high-field determinations of zero-field splittings for triplets oriented in single crystals have been reported for only relatively few mole cules: benzene,naphthalene, 12,29,30 qUinoxaline,^ and isoquinoline,^ phenanthrene,^ pyrene,^3j34 tetra- methylpyrazine,36 ancj mesitylene. ^ Although less dif ficult experimentally, glass studies in general yield less precise values of zero-field splittings, do not readily give the relative signs of the spin Hamiltonian parameters and do not provide as much information about hyperfine interactions as experiments on triplets oriented in single crystals. The disparity between single-crystal and glass studies, with respect to the precision attainable in de terminations of zero-field splittings, increased with the development by R.E. Gerkin, in C.A. Hutchison's Laboratory (1958-1960), of a method for measuring splittings at near zero magnetic fields (below— 75 gauss). This method has the advantage that the energy of interaction of the trip let spins with the magnetic field constitutes at most a very small perturbation on the energy differences of the i triplet levels at zero field, whereas in high-f^.eld studies the Zeeman contributions to the observed

t splittings are usually at least twice the magnitude of the zero-field splittings themselves. Results reported by 7

Brandon, Gerkin and Hutchison in 1962 and 1964 for naphtha lene oriented in durene and biphenyl and for phenanthrene oriented in biphenyl, established that the precision at tainable in zero-field splitting determinations by the low- field method was superior to that reported for any half field or high-field studies for either randomly oriented or mixed single-crystal samples.(A fourth method of obtaining zero-field splittings is zero-field spectroscopy in which the frequency is scanned and either microwave or optical detection is used. These techniques are capable, in principle, of precision comparable to that of the low- field method. However, zero-field spectroscopy has been qg 9 employed in only a few cases, and the uncertainties reported to the present have been approximately an order of magnitude greater than those found in low-field studies.) Gerkin and Szerenyi have recently described a low- field investigation of phosphorescent triphenylene ori- A O ented in single crystals of dodecahydrotriphenylene. Perhaps the most important result of this work was the un ambiguous demonstration of a non-zero value for the spin Hamiltonian parameter E, indicating less than three-fold symmetry for the lowest triplet state of triphenylene (oriented in a host which does not, necessarily, impose less than three-fold symmetry on the guest). In contrast, all previous magnetic resonance studies for triphenylene found E equal to zero within the (larger) uncertainties of those measurements. In addition to the low-field studies cited above, the only other published reports describing investigations of oriented triplets by the low-field method have been those of Gerkin and Winer based on the researches described in this thesis.43-45 CHAPTER II THEORY

A. INTRODUCTION With a few exceptions, the zero-field splitting data obtained in this research could be analyzed in terras of well-established theory. Since theoretical treatments of electron magnetic resonance absorption by aromatic mole cules in the triplet state at both low and high fields can be found in a number of references,12,42,46-48 oniy a condensed version of this theory will be presented here. Those cases for which the existing theory apparently does not account for the experimental splitting data are dis cussed individually in appropriate Sections of Chapters V-VIII. To our knowledge there has been no report of a theo retical treatment concerned explicitly with the phenomenon of raultiplet structure in low-field magnetic resonance spectra, or of deuterium isotope effects in zero-field splittings. Theory relevant to an interpretation of these phenomena is presented with the discussions of them

9 10

(Chapters V and VII, respectively). Similarly, theoreti cal considerations which bear on the observation of re solved proton hyperfine structure at zero and near-zero fields are presented in Chapter XIII.

B. THE SPIN HAMILTONIAN Four principal magnetic interactions must be con sidered in formulating a general Hamiltonian operator for an isolated aromatic molecule in the triplet state and in a static magnetic field. These are: (1) the electronic Zeeman interactions; (2) the dipolar interaction of the two unpaired electron spins; (3) nuclear-electron hyper fine Interactions; and (4) nuclear Zeeman interactions. Experimental results from previous studies^**2 as Well as the present studies have established that differences in molecular-crystal-field effects for different organic host structures can be as large as several percent of the magni tudes of the zero-field splittings. Thus, a general Hamiltonian for the triplet state oriented in a single- crystal host would require, in addition, a term which ac counts for the substantial effect of the molecular-crystal field. For completeness, an explicit term accounting for the effect of orbital angular momentum contributions to the triplet state magnetic moment would be included but for the following reasons need not be. First, theoretical 49 calculations by Hameka and Oosterhoff for the triplet state of benzene established the magnitude of the spin- orbit contribution to the zero-field splitting parameters to be approximately 10"^ cm"*, a value only slightly larger than typical uncertainties in zero-field splittings determined in this research. Evidences that spin-orbit effects are comparably small in the molecules studied here are that their phosphorescent lifetimes are of the same magnitude as that of benzene and that very nearly free- spin g-values have been found experimentally for these molecules. Finally, to first order, spin-orbit effects merely shift the three components of the triplet state equally and do not affect the energy differences between components (zero-field splittings). It is most practical to consider explicitly only the magnetic interactions cited above in formulating an appropriate Hamiltonian and to treat the derived zero-field splitting parameters as implicit functions of the magnitudes of the molecular crystal-field and the spin-orbit coupling. Since the bulk of the zero-field splitting data was obtained for perdeuterated molecules, for which resolved hyperfine structure is not observed, we will consider here only the electronic Zeeman and the electron fine-structure interactions. In this approximation, we may write the Hamiltonian of two electrons in a triplet state, and 12

static magnetic field H as

1-1= m • 8 • (S1+S2)4-sV [ ^i g--(g--‘£^ g2‘~) j » (1)

where the first term represents the electronic Zeeman interaction ( /3 is the Bohr magneton, and g is the field- spin coupling tensor), and the second terra is the operator form of the classical interaction energy between two elec tron magnetic moments (r being the vector joining the two electrons). Upon expanding the dipolar terms in the com ponents of the spin vectors, reducing the resulting expres

sion in terms of the total spin (S= S^+ §2 )» and averaging over all positions r^ and of the two electrons, an ef fective spin Hamiltonian is obtained,

J-|[= |0|h • g • s - a ' s a 2 - b ' s b 2 - c' Sc 2, (2)

where A, B and G are the principal magnetic fine structure axes and A* , and & are the zero-field eigenvalues. (The zero-field Hamiltonian is diagonalized by eigenfunc tions which transform as Cartesian coordinates and which are simple linear combinations of the infinite-field triplet spin functions.) Relation (2) is called the spin Hamiltonian, since it operates on wavefunctions expressed in terms of spin variables only. Since the trace of the 13 zero-field splitting tensor is zero (i.e., A/ 4- B/ +

Q = 0 ), the zero-field splittings can be rewritten in terms of just two independent parameters, which are com monly called D and E:

M = |/S|S - S • g + »CSc2 - 1/3S2) + E(Sa 2- Sb2). (3)

Since the 1/3 terra in Relation (3) does not affect the energy differences between components of the triplet state, and if we assume that the magnetic fine structure axes A, B and C coincide with the principal molecular axes x, y and z (as defined for each molecule studied), then Relation (3) becomes

I H t = |/s]H-g.S+ D Sz2 + E(Sx2 - Sy2) (4) and D + E, D — E and 2E are the zero-field splittings. c* SOLUTION OF THE FINE STRUCTURE SPIN HAMILTONIAN The eigenvalues and eigenvectors of the spin Hamil tonian (4) for H along each of the three principal axes are (from Stevens, reference 46): 14

' H || z;

wl = D + SZZ |/*l lsl+E tanflz> cosSg |l)+ sin^ 2 |l) ,

WI = D - g 2 2 |/3| j H]— E tane2, sin0 z |l}— cos^ll), where tan 20 ^ = E/gzz lollsl !

5 11 x: Wx= ^(D + E) + gxx |/3| |h|-*e(D + K) tan 9^, c°s0x WQ= D - E, (5)

WT= ^(D + E) - gxl£ \(B\\n\ + ^(D+E)tan0x , sin0 x Jl) — cos0 x |^,

where tan 20X= — (D 4- E)/2g J/3] |h | ; X}C

H II y:

W 1 = % ( D - E ) + g |/3||h|+ %(D - E) tan9y, cos , W0 = D + E,

WX=%(D + E) - Syy|i8 ||H| - %0> - E)tan0y, Sin<}y| 1>- cos0y |X>, where tan 20y= (D — E)/2g |/3||h|.

The diagonal elements of the g tensor (gv„,XX g„„ yy and g__) can be calculated from data obtained in high-field experi ments according to the relations given in Appendix A.

D - IfDi LOW-FIELD EIGENVALUES Although at intermediate and high values of the static magnetic field exact solution of the spin 15

Hamiltonian (4) is straightforward only for the case of H parallel to the principal axes, a general solution is readily found at low fields. In particular, from zero and up to fields for which G2 Cs I 11H |)^ Is much less than the square of the smallest (or in some cases the next smallest) zero-field splitting, the eigenvalues obtained are valid for any orientation of the molecular axes with respect to the static field. These inequalities,

G2 << (D + E) 2 or G 2 << (2 e)2, provide perhaps the most useful working definition of low field, with respect to the Zeeman and electron spin-spin interactions. In the usual case, say for phenanthrene, chrysene and dibenzanthracene, low fields according to this criterion would be from 0 to approximately 75 G. If we (justifiably) neglect any anisotropy in g be cause of the small contribution of the Zeeman terra to the total splitting (in any case, for the molecules studied the anisotropy itself is very small, e.g., for phenan threne gxx = 2.0021, gyy = 2.0028 and g2z = 2.0041), then in the field interval specified above the eigenvalues be have as follows:(from Brandon, Gerkin and Hutchison, reference 2 )

D + E-> (D+E) + (G2 /2E) 1 - [C/(D+E)] J (6)

D-E-> (D-E) - (G2 /2E) 1- [C/(D-E)]| (7)

0 -> — £ G2 /(D2 - E2), (8) >2 o where C = Jt (D - E) -f- ro^D + E), / = cosine of angle be tween H and the x axis, and ra = cosine of angle between H

and the y axis. Relations (6 )-(8 ) provide the basis for analysis of the low-field precision data obtained in this research. A qualitative plot of the triplet energy levels as a function of field (at low fields) is shown in Figure 1 for H parallel to the x, y and z axes, respectively. In all low-field experiments carried out in this study, the static magnetic field was always oriented approximately normal to the microwave magnetic field. The zero-field selection rules are that the D + E, D - E and 2E "tran sitions" are x, y and z polarized, respectively, and the transitions which were observed in these "perpendicular" experiments are indicated in Figure 1. w A Hlix

D-E

D+E

Figure 1. Zero-field and Zeeman splittings of a triplet state for low magnetic field parallel to the principal axes. The "perpendicular" transitions observed in this research are indicated.

* CHAPTER III EXPERIMENTAL APPARATUS AND METHODS

A. THE LOW-FIELD SPECTROMETER 1. Static Magnetic Fields a. The Helmholtz Magnet The magnetic field experienced by a sample in this study was the vector sum of the residual laboratory field, and the field produced by a pair of water-cooled, air-core, circular coils set parallel a distance apart equal to their radius. This "Helmholtz” configuration, and the specific cross-section dimensions of the coils, pro duced an especially uniform magnetic field at a region halfway between the coils along their axis (hereafter called the center of the magnet or sample site), and fur ther insured that the field produced was to a good approxi mation linear in the current flowing through the coils (see below)• Each coil was formed from round No. 10 solid copper wire wound in twenty-six layers containing 24 turns per layer, the radius of the first layer of conductor being

18 7 7/8 ± 1/8". Cooling was provided by means of two platens for each coil, each platen carrying two (3/8" O.D. by 1/32" wall thickness) copper tubes taped integrally with the coils. The cooling platens arid conductor layers were vacuum impregnated with epoxy resin and then wrapped (half-lap) with glass tape to produce finished coils of approximately 20" diameter with approximately a 4" x 4" cross-section. (The coils were constructed locally by the National Electric Coil Co.). The coils were mounted in the Helmholtz configuration on a circular disk of 1%" thick aluminum which in turn was mounted, via a brass bearing at its center, on a massive aluminum plate supported off the floor by four levelling bolts at its corners. Degree markings were inscribed on the rotatable aluminum disk on which the magnet was mounted, and could be referenced to a fiducial mark and vernier settings on a block fixed to the aluminum base. The vernier scale permitted angular resolu

tion of 0 . 1 degree and reproducibility in angle determina tions of ~0.1 degree. Inlet and outlet water hoses con nected to the cooling manifolds were of ample length to permit free rotation of the magnet. The Helmholtz coils were connected in parallel and ener gized by a Harvey-Wells HS-1050A current-regulated preci sion power supply. The regulation was determined and found adequate to restrict field variations to less than 20

0.001 G over the field range for precision measurements. Static fields of up to 500 G could be produced. The fields employed in precision measurements (— 3 to "-30 G) were well below those (above — 200 G) at which heating of the coils could be detected during continuous operation. The static magnetic field could be swept linearly in time, at varying rates (from 14 G sec"^ to 4 G min”^) over any arbitrary range of field (from — -0 to --500 G) by means of an electronic field-sweep programmer constructed by A. B. Ledwith in this department. The field-sweep pro grammer was used during searches for resonance absorptions and to study the resolved proton hyperfine structure pat terns observed at low field (see Chapter VIII). In pre cision measurements of zero-field splittings, the field was swept by hand because the field corresponding to the absorp tion maximum could be established more accurately in this way.

k* Static-Field Measurements The current In the Helmholtz coils was deter mined by measuring the voltage drop across a standard re sistor in series with the magnet. The standard resistor consisted of a pair of Leeds and Northrup Model 4360, 0.1 ohm standard resistors (Serial Nos. 1681298 and 1636607) connected in parallel. For approximately the first third of the precision data obtained in this research the 21

voltage drop across the standard resistor was measured using a Leeds and Northrup Model 8687 potentiometer. The potentials of two Eppley standard cells (Nos. 811602 and 811607) measured by this potentiometer differed from the National Bureau of Standards calibrated potentials for the cells by -0.08 and -0.07 mV, respectively. For the re maining two-thirds of the precision determinations of zero-field splittings made in this study, voltages were digitalized via a Hewlett-Packard 2212A voltage-to-frequency converter (which produced an output pulse frequency di rectly proportional to the amplitude of the dc input vol tage) and a Hewlett-Packard 5245 L electronic frequency counter (described below). The 0.1 volt scale of the voltage-to-frequency converter was calibrated against the Eppley standard cell voltages to better than 0.025 mV. As a check on the reliability of the voltage-to-frequency converter, a given voltage drop was measured from time to time with each of the independently calibrated instru ments. The values so obtained were shown to differ by negligible amounts (less than 0.1 roV). The difference between two measurements of the (central) field for resonance at a fixed frequency in experiments to determine zero-field splittings was calcu lated for 282 pairs of field determinations, and the mean deviation of the pairs from their respective average 22 values was found to be 0.09 G. Thus, the most important source of error in determination of resonance fields was the uncertainty in determining where the absorption reached a maximum rather than any uncertainty in the cali bration of the Helmholtz magnet (discussed below) or the uncertainties in the values of the voltages (fields) measured with the potentiometer or voltage-to-frequency converter.

c. Cancellation of the Laboratory Field Two additional, smaller sets of Helmholtz coils were wound on Plexiglas formers using #28 copper wire. The coils of one magnet each had 100 turns of wire, were 35 cm in diameter, and were connected in series. The other magnet consisted of 28 cm diameter coils with 85 turns on each, again connected in series. Each set of coils was energized by a constant current power supply built by T. Ireland in this department. These magnets (hereafter called cancelling coils) were both mounted on the Dewar support in the gap of the main Helmholtz magnet, &s shown in Figure 2, — one with the plane of the coils vertical, the other with the plane of the coils horizontal-- so that their "center fields" were at the sample site. The optimal orientations and currents required to minimize the laboratory field at the sample site (which was — 0.6 G) by cancellation of fields, were determined with a 23

Figure 2. The low-field Helmholtz magnet (with cooling- water manifolds on left). Shown also are the horizontal and vertical cancelling coils at tached to the dewar support (the smaller inner coils were for 60 Hz modulation and were not used in this study). Shown in the foreground

are the A-H6 lamp and focusing lens (note win dow in dewar support and dewar to permit exci tation of samples)• 25 rotating-coil gaussmeter. The gaussmeter was powered by a Bodine KYC-26 electric motor, which rotated at 3600 rpm and produced a 0.28 rooz torque. The gaussmeter coil con sisted of 270 turns of #36 insulated copper wire wound on a bakelite coil former (1.2 cm O.D.). The output of the gaussmeter could be conveniently displayed on either an oscilloscope or a voltmeter.

d. Calibration of the Helmholtz Magnet The field-vs-current function for the main magnet was determined by: (a) employing the rotating-coil gaussmeter to obtain 60 measurements at approximately 0 . 5 G intervals over the range 0.43 to 27.3 G, and (b) ob serving electron magnetic resonance absorption by diphenyl- picrylhydrazyl (DPPH) to obtain 16 measurements over the range 3.0 to 17.8 G. The rotating-coil voltage was deter mined with a Fluke 883 AB differential voltmeter. The voltages produced by the rotating-coil gaussmeter and the corresponding potential drops across the standard resistor were least-squares fitted to give a best linear function. The root-raean-square deviation of the indivi dual points from the best-fit line was 0.029 G. The fields calculated from the DPPH resonance frequencies were least-squares fitted vs the corresponding voltages across the standard resistor to give a best linear function. The rms deviation of the individual points from this best-fit 26 line was 0.032 G. Thus, the field-vs-current function was taken to be linear between 0.5 and 27 G within the accuracy required for these experiments.

Five additional DPPH resonance calibrations of the Helm holtz magnet were carried out during this research. From the best-fit functions for the six calibrations, the weighted mean value of the conversion factor between the potential drop across the standard resistor and the Helm holtz field at the sample site was calculated to be 0.2342 ± 0.0008 G mV"1. From the intercepts of the best-fit functions, the total uncancelled or residual laboratory field at the sample site was calculated to be less than 0.03 i 0.02 G during all precision determinations of zero-field split tings. Experiments to determine the magnitude of the re sidual field by measuring the ratio of the rotating-coil gaussmeter voltages for the cancelled and uncancelled laboratory field, yielded somewhat higher values for the residual field prior to the final optimization of the can celling coils' positions and currents on 1968 July 29. Specifically, from 1967 September 26 to 1968 April 24 the total uncancelled field was, according to this method, 0.07 G; from 1968 April 24 to 1968 July 24 it was 0.10 G; after 1968 July 29 the total residual field was measured as 0.03 G (the estimated uncertainty for these fields was 27

+ 0.01 G). The source of the discrepancies between the values of the residual field determined in the DPPH cali brations and those found in the rotating-coil gaussmeter experiments (prior to 1968 July 29) is not known. However, it must be emphasized that even the larger values of the residual field were insufficient to necessitate any cor rection to the voltage measurements for the molecules studied, to within the indicated uncertainties in the zero- field splittings.

e. Homogeneity of the Resultant Field Using the rotating-coil gaussmeter the magnetic-field homogeneity at the sample site was examined, and a variation in field of less than 4 parts in 1000 was

found within a sphere of 2 cm diameter centered on the sample site. Within the volume of the sample ( 0.5 x 3 0.5 x 0.5 cm ) no variation in the coil voltage could be detected within the uncertainty of the measurements ( 2 parts in 1000). The experimentally measured homogeneity was substantially less than that calculated for the main magnet. This was due to the fact that in the present ap paratus the overall field-horaogeneity was limited by the (smaller) cancelling coils and the magnitude of the laboratory field, and not by the properties of the main magnet. Reproducible placement of crystal samples at the center of the Helmholtz field from experiment to experiment was 28 assured by a fiducial designation of this position.

2. Microwave Fields The microwave radiation required for these experi ments was produced by three oscillators which together covered the range from 0.9 to 4.5 GHz. The oscillators and their respective ranges of operation were: (l)General Radio, Type 1218-A (0.9-2.0 GHz); (2) General Radio, Type 1360-A (1.7-4.1 GHz); and (3) Polarad, Model 1207M1 (3.8- 8.2 GHz). Frequency drift-rates are approximately 0.1% during warm-up for these instruments and they were always turned on from several hours to tens of hours prior to making precision measurements in any experiment. The after-warm-up frequency stability of the GR 1360-A oscil- £ lator is specified to be approximately 5 parts in 10 .

The average difference between two values of the microwave frequency produced by this oscillator several minutes apart (one measured preceding and the other following the determination of the static magnetic field required for resonance) was calculated for 157 pairs of precision £ measurements and found to be 2.3 parts in 10 for fre quencies near 2.7 GHz, and 1.7 parts in 10^ for fre quencies near 3.5 GHz. The average difference between pairs of frequency determinations for microwave radiation produced by the GR 1218-A and the Polarad 1207M1 in 29

experimental runs were calculated and found to be 5.2 parts 6 6 in 1 0 (for 120 pairs of frequencies) and 1.3 parts in 10 (for 34 pairs), respectively.

3. Frequency Counting Microwave frequencies were measured with a Hewlett- Packard (H-P) Model 5245L electronic counter with an H-P 5253B frequency converter plug-in unit. The counter was used in conjunction with an H-P 540B transfer oscillator which extended its frequency measurement range into the microwave region. The microwave frequency was determined by zero-beating (by means of an oscilloscope in the 540B) the input signal with a known harmonic of an extremely stable signal genera ted by the 540B transfer oscillator and by then measuring the 540B fundamental frequency on the counter. Multiplying the counter readout by the harmonic number gave the micro wave frequency. The harmonic number and corresponding frequency were established unambiguously in every experi ment by comparison of frequencies calculated for several adjacent-harmonic zero-beat readings. The electronic counter (whose specified aging rate was 9 less than + 3 parts in 10 per 24 hours) was calibrated seven times during this research (at approximately 4 month intervals) by intercomparison with the 60 kHz signal 30

transmitted by the National Bureau of Standards at Fort Collins, Colorado (WVB), The signal was received by an antenna on the roof of Evans Laboratory and the intercora- parison was made using an H-F 117A VLF phase comparator. The stability of the frequency counter was such that, during the intervals between calibrations, the accumulated deviations of its time base from the USFS input frequency 9 were not more than 8 parts in 10 . Recalling from above that the frequency reproducibility in precision measure- g ments ranged from 1.3 to 5.2 parts in 10 , it is clear that for the measurement of frequencies (as in the measure ment of fields) the calibration error was.negligible com pared with the uncertainties in the experiments themselves. It is also clear from these data that the uncertainty in frequency was determined primarily by the stability of the microwave oscillators.

4. Signal Detection The output from the microwave oscillator selected for a given experiment was coupled by coaxial cable to the header, shown in Figure 3, via a Narda variable probe coupler (Model No. 3080) and two PRD Type 327 matjching ele ments. The output from the variable probe coupler was fed to the 540B transfer oscillator. Probe-coupling of the microwave energy to the resonant cavity at the bottom of 31

Figure 3. The low-field header with microwave cavity (right). Note (left) BNC connectors for coupling of microwave and modulation lines and knurled knob for adjustment of dielectric tuning plunger.

the header (Fig. 3) was made by insulated phosphor-bronze wires inside cupro-nickel tubing. The resonant cavities were of a cylindrical re-entrant type described pre-

viously 2 * 12 and were operated as transmission cavities. Cavities of 0.64, 1.00 and 1,53 inch depths were required for the frequencies used in this study. The use of adjus table caps on the cavity center-posts and polyethylene "loading” plugs of various sizes permitted each cavity to be employed over a relatively wide range of frequencies ( — 1GHz). In precision measurements, fine-tuning of the cavity to the selected microwave frequency was accomplished by means of a dielectric plunger whose insertion into the capacitive gap at the top of the cavity could be adjusted continuously from the top of the header. Cavities were pressurized slightly with hydrogen gas to prevent filling by liquid nitrogen or condensable gases. The cavities used differed significantly from pre- 2 vious ones only in that modulation of the static field was accomplished via a single loop (of insulated copper wire) mounted inside the cavity through epoxy seals in the cavity floor. In previous studies, multiple-turn external modulation coils were used. It was found necessary to em ploy internal modulation in order to remove a prominent recorder baseline dependence on the amplitude of the static magnetic field which was observed when external 34

static-field modulation was used for cavities built in this department. (This phenomenon was not observed for cavities constructed for Gerkin at the University of Chi cago) . The cause of this baseline shift (which unfortu nately was most prominent in the range of field for which precision measurements were made) has not been definitely established. However, it may have resulted from the win dow shrouds being driven by the external modulation coils which were wound around the shrouds in the normal operating configuration, since both shortening the shrouds and re placing the cupro-nickel shrouds by copper shrouds very substantially reduced the baseline shift. The magnitudes of the modulation fields at the sample site produced by the two methods of modulation were calculated and shown to be typically less than one gauss. Either lOOKHz or 125KHz modulation was employed. The "output" side of the header was coupled by coaxial line to an FXR Model N200A crystal detector and matching element. Sylvania N21F crystal diodes were used in the de tector. The rectified output of the diode was fed to a narrow-band pre-amplifier and then to a phase-sensitive detector. Both instruments were built by T. Ireland in this department. The pre-amp gain and relative phase (between the reference signal from the modulation oscil lator, and the 100 or 125KHz rectified crystal output) 35 could be varied to optimize signal-to-noise ratios. The diode output was displayed on a Simpson raicroamroeter for convenience in setting the microwave power level, and to permit tuning of the cavity and the microwave line to the resonance frequency. An RC circuit in the phase-sensitive detector furnished effective spectrometer time constants of 0.04, 0.5, 2.3 and 9.0 seconds. The output of the phase-sensitive detector was dis played on a Leeds and Northrup Speedomax G stripchart re corder, which had an 0.25 second response time. The chart recorder displayed first derivative traces of resonance ab sorptions. The point at which a first derivative trace crossed the interpolated baseline (from the above- and below-signal baselines) was taken as the center (or maxi mum) of the absorption. Normally a chart speed of 2” per minute was used, but to obtain triplet decay data a chart speed of 1 " per second was employed. A block diagram of the low-field spectrometer is shown in Figure 4.

5. Optical System The light source for the continuous photo-excitation of samples was a water-cooled, General Electric A-H6 high- pressure mercury arc. During precision measurements the output from this lamp was filtered either by an aqueous Microwave Oscillator Chart 0 Antenna Recorder

Attenuator comparator 100 KHz 100 KHz Preamplifier phase sensitive modulator detector

Matching Transfer Frequency Crystal Matching elements oscillator counter diode elements Dewar Voltage Standard resistor to frequency converter

A-H6 Magnet Field Filter power sweep o supply Lens

power Helmholtz coil supply Earths Resonant field cavity cancel ing coils Lo O' Figure 4. Block diagram of the low-field spectrometer, CuSO^* 5H^O solution (100 g/~*) in a 5 cro-path quartz cell,

or by an aqueous solution, bandpass filter of CoSO^*7 H2 O (45 g

6. Crystal Temperature Measurements were made principally with the reso nant cavity immersed in liquid nitrogen boiling at atmos pheric pressure, but measurements were also made at lower temperatures by reducing the pressure above the bath, and at higher temperatures by admitting cold nitrogen gas to the experimental dewar at controlled flow rates or by al lowing a massive lead heat-sink in contact with the cavity to warm slowly from 77 K. In each of the latter cases, a copper-constantan thermocouple junction was soldered to the bottom of the cavity and the EMF relative to a refe rence junction was recorded before and after each resonance measurement to permit temperature determination by inter polation. In a similar series of experiments, the variable- temperature junction of a second calibrated thermocouple was embedded in a crystal mounted in the cavity as for a resonance experiment; by determining the "crystal EMF" as 39 a function of the "cavity EMF", a temperature could be as signed to the crystal for each magnetic resonance observa tion to within an estimated uncertainty of approximately one Kelvin degree. To minimize the difference between the crystal temperature and the bath or ambient temperature, the ultraviolet-transmission solution-filters described above were used in the excitation path. Thermocouple ex periments similar to those just described showed that for the CuSO^ solution-filter the average temperature difference between the illuminated sample and the liquid nitro gen bath was (+) 2.9 i 0.4 K, while for the three compo nent bandpass-filter this difference was at most (+) 0.3 K. One experiment was carried out in which the reso nant cavity was immersed in liquid oxygen boiling at atmos pheric pressure and at reduced pressures. In this and other experiments in which the dewar support was partially evacuated, the pressure was regulated by a Cartesian manostat (Manostat Corporation Style No. 8 ) and read on an open-tube manometer. Bath temperatures were then calcu lated from the measured vapor pressures of the liquid oxy gen or nitrogen.

B. THE HIGH-FIELD SPECTROMETER The static magnetic field was produced by a Varian 12" low-impedance magnet and a Varian Fieldial-regulated power 40

supply, and its value was determined by proton resonance magnetometry. The microwave radiation source was a Varian

4 217C klystron energized by a Hewlett-Packard 716B power supply. A Hewlett-Packard 5245L frequency counter and 2590B frequency converter were used to measure the micro wave frequency and the proton resonance frequency. The counter time-base was periodically calibrated as described

above and was found to have deviated by not more than 2 8 parts in 1 0 from the standard during any interval between calibrations. The spectrometer was modulated at 100kHz, and the . phase-detected signal was displayed by a Leeds and Northrup Speedomax W/L stripchart recorder with a one-second re sponse time. The cavities were similar to those previously described, 2 ’ 12 and the samples were continuously excited

by an A-H6 high-pressure mercury arc lamp except during lifetime measurements. The experimental procedures used for the high-field measurements have been described previously.^’^ All these experiments were done with the cavity immersed in liquid nitrogen boiling under atmospheric pressure. The description of the high-field spectrometer is limited to the above account due to the availability of many detailed experimental descriptions of electron mag netic resonance studies of oriented optical triplets at high fields, 2,12,30-37 an(j because only a very limited number of (largely qualitative) high-field experiments were undertaken in this research. CHAPTER IV PREPARATION OF SAMPLES

A * PURIFICATION OF MATERIALS 1. Guest Materials

a. Chrysene-hi9 Eastman chrysene No. 4217 (violet fluorescence grade) was variously subjected to vacuum sublimation (for example 4.1: 56, 58, 62), zone-refining (4.1: 185, 237, 239, 245) column chromatography on Woelm neutral alumina (4.1: 248), an acetylation reaction (4.1: 268) and recrys tallization (for example 4.1: 224) in order to enhance its purity. (The parenthetical numbers found throughout this Chapter specify the research notebook number (i.e., 4.1) and the page on which the indicated procedure is described in detail.) Of these procedures, only column chromato graphy (with benzene eluant) yielded a product from which most 2,3 benzocarbazole, the tenacious and major 51-52 con taminant of commercial high-purity chrysene, had been re moved. The removal was demonstrable by the absence (visually determined) of the characteristic orange

42 43 phosphorescence of 2,3-benzocarbazole in single crystals of chrysene photoexcited at 77K by the A-H 6 lamp. Moreover, four additional resolved impurity bands developed on the column and were also separated from the chrysene. Although these additional impurities were not studied further, the recorded retention characteristics and UV and visible ap pearances of their respective bands on the column should aid any future investigation of their identity (4.1: 248- 253, 256). It should also be noted that a custom sample of chrysene (supposedly of higher purity) prepared for us by Eastman (chrysene 4217X) showed little or no reduc tion in benzocarbazole phosphorescence. Subsequently, a CO synthesis of chrysene was attempted, but only the inter mediate chrysene-3,6-quinone was recovered (4.2: 31-41). b. 2 j 3-Benzocarbazole As discussed fully in Chapter VI, magnetic resonance absorptions from phosphorescent 2 ,3-benzocarba zole oriented in p-terphenyl and chrysene hosts were de tected and investigated. Due to differences in zero-field splittings, signals from 2 ,3-benzocarbazole in p-terphenyl were shown to be readily distinguishable from those of chrysene in p-terphenyl, thus removing any remaining uncer tainty about possible complications arising from the presence of traces of 2 ,3-benzocarbazole in the chromato graphed chrysene samples. The 2,3-benzocarbazole used in 44 these studies was obtained from Chemical Procurement Labo ratories Inc. and was purified by vacuum sublimation (4.3: 33, 44).-

✓ c. Chrysene-diy Chrysene-d-^ (nominal isotopic purity 98%) was obtained from Merck, Sharpe and Dohme, Ltd. (M.S.D., Lot No. AD-024) and was purified by column chromatography (4.2: 155-158) as described for chrysene-h^ above. Since no previous report of the total emission spectrum of chromato graphed chrysene-d- ^ 2 was found, such a spectrum was re corded for a sample suspended in Hartman-Leddon spectroscopic-purity EPA glass at 77K, at an excitation wave- o length of 3130A, using a Turner 210 spectrophotometer (4,4: 50-58). The spectrum obtained was very similar to that published for chrysene-h-^* 54 except for an approxi mately threefold increase in the phosphorescence to fluorescence quantum-yield ratio. The spectrum is shown in Figure 5 and the positions of the maxima are tabulated in Table 1. The emission and absorption monochromators of the Turner instrument had been calibrated against mer cury lines to within i;lA by Dr. N. T. Rao. The average deviation of wavelengths from the mean, for the more intense emissions, was z 2A and z 5A for the fluorescence and phosphorescence emissions, respectively. 3500 REL. INTENSITY Figure 5. Emission spectrum of chrysene-d-^ in EPA glass at 77K. at glass EPA in chrysene-d-^ of spectrum Emission 5. Figure 4000 X X 10 WAVELENGTH {£) 5000

5500 60004500 ■is 46

Table 1. Emission Spectrum of Chrysene-d10 in EPA Glass at 77K

Fluorescence Phosphorescence (A) (A) 3599 4957 (max) 3634 5031 3700 5097 (sh) 3789 (max) 5164 (sh) 3823 (sh) 5331 3903 5387 3999 5462 4040 (sh) 5548 (sh) 4129 5753 4232 5820 4267 (sh) 5898 4381 4497 47

Some chrysene-d^2 P “terphenyl mixed crystals were prepared using unpurified chrysene-d-^. For such crystals magnetic resonance absorptions from 2 ,3-benzocarbazole could also be observed as discussed in Chapter VI. d. 1,2; 5,6-Dibenzanthracene-hj/,

Rutgers 1,2; 5,6 -dibenzanthracene showed no (impurity) phosphorescence and was not purified further,

Eastman No, 3272 1,2; 5,6 -dibenzanthracene was chromato graphed on Woelra alumina (4.2: 239-242) as in the case of chrysene. Four distinct impurity bands were observed during the purification. A portion of the purified ma terial was sent to M.S.D. to be perdeuterated. However, they did not find it possible to deuterate this dibenzanthracene sample to an acceptable isotopic purity by the same procedures formerly found successful (for presumably less pure starting materials). This suggests that one or more of the impurities in commercial dibenzanthracene cata lyzes the deuteration reaction. e. 1,2; 5,6-Dibenzanthracene-djy. M.S.D. did supply 1,2; 5,6-dibenzanthracene-d^ of 98% nominal isotopic purity (lot No. B-127), prepared from their own starting material. This sample was used without purification. f. Naphthalene-h^ The naphthalene used in this research was 48

Eastman No. 168 (Lot No. 36A) which had been zone-refined by D. Burley for approximately ten passes (see Burley note book, page 9). g. Naphthalene-dg Naphthalene-dg was obtained from M.S.D. (Lot No. B-312; 98% nominal isotopic purity) and was not puri fied further. h. Phenanthrene-h^Q Eastman phenanthrene No. 599 was used without purification. i. Phenanthrene-d-^Q M.S.D. furnished phenanthrene-d^Q (Lot No. 1403; 98% nominal isotopic purity), which was not purified further. j. Dibenzothiophene Eastman No. 4594 dibenzothiophene was used without further purification. k. 1, 2,3,4“-Tetrahydroanthracene 1,2,3,4-Tetrahydroanthracene was synthesized from 9,10-dihydroanthracene produced according to the pro- 55 cedure of Orchin (4.4: 20-25). The tetrahydroanthracene produced was chromatographed on Woelro neutral alumina using benzene eluant and recovered from ethanol (4.4: 28-29). 1. Picene Rutgers picene (nominal purity 99.9%) was used* 49

without purification.

2, Host Materials a. p-Terphenyl-—14 Scintillation grade p-terphenyl-h^ was obtained from Arapahoe Chemicals Inc. and was used without further purification, since no (impurity) phosphorescence was de tected visually from single-crystals of it photoexcited at 77K, nor was magnetic resonance observed from such crys tals. p-Terphenyl from K and K Laboratories Inc. and from Eastman was also used without further purification for the preparation of a small number of mixed crystals. b. p-Terphenyl- ii4 p-Terphenyl- dl4 (nominal isotopic purity 98/0 was obtained from M.S.D. (Lot No. 1116) and was used with out further purification. c. Symmetric Octahydroanthracene Symmetric octahydroanthracene (OHA) obtained from Chemical Procurement Laboratories Inc. was zone- refined 50 passes (4.2: 205, 211), chromatographed (4.3: 20) or recrystallized twice from ethanol (4.3: 30, 34). Although these procedures removed a yellow impurity from the sample, they failed to remove completely a specie giving rise to a yellow-green phosphorescence emission from single crystals of OHA photoexcited at 77K or at 273K. 50

However, magnetic resonance absorptions were detected from the triplet state of a molecule believed to be the source of this emission, and the zero-field splittings and phos phorescent lifetime of this triplet state were determined (both at 77K and at 273K) and shown to be much different from those of chrysene-d^ in OHA. Results from this investigation are presented in Sections IV. B. 2 and Chapter VI. Weak blue phosphorescence was observed from unpurified or zone-refined samples of OHA photoexcited at 77K, but not at 273K. Blue phosphorescence was not observed, even at 77K, from photoexcited samples of OHA which had been recrystallized twice from ethanol. A number of mixed crystals were prepared using unpuri fied OHA. d. Biphenyl Eastman No. 721 biphenyl (Lot No. 53A) was used without purification. e. Durene Eastman No. 2295 durene was used without puri fication. -Binaphthyl - Aldrich B3440 2,2 -binaphthyl was zone-refined 100 passes (4.3: 266). A number of boules were prepared using unpurified 2 ,2 * -binaphthyl (which showed no impurity 51 phosphorescence). g. p-Quaterphenyl

Eastman No. 6 8 6 6 p-quaterphenyl, which was an opaque, orange powder, was chromatographed on Woelm neutral alumina using benzene eluant and colorless, transparent crystals of p-quaterphenyl were recovered (4.2: 285-287).

h. 2,6 -DimethyInaphthalene

Aldrich D17-100 2,6 -dimethyInaphthalene was zone-refined 43 passes (4.2: 227).

B. GROWTH OF SINGLE CRYSTALS The canonical way of preparing "crystal tubes" for ap proximately the last 110 of the 134 boules grown in the course of this research was as follows: A 10" length of 12 ram. O.D. Pyrex tubing was glassblown to a male 12/30 standard taper joint, and the other end was pulled off into a gradually tapered point. This piece was then thoroughly flamed in air (to a temperature near the softe ning temperature of Pyrex) in order to oxidize completely any possible organic contaminants. After the tube was tested for leaks, approximately half of the host material was transferred to the tube, then the guest sample and finally the remaining host material was transferred. The crystal tube and a powder trap were placed on a vacuum « line via a manifold which also contained a powder trap. 52

A crystal tube, powder trap and manifold are shown in

Figure 6 , The use of double powder-traps and individual manifolds (each member of the research group having his own) minimized the risk of contamination of one sample by another through contamination of the vacuum line. (Discon- nectable vacuum-line couplings manufactured by Fisher- Porter Co. made the use of individual manifolds practical.) A sample was generally evacuated (using only a mechanical

pump) for approximately 4 to 6 hours. At the end of this period, the pressure in the line was measured using a McCleod gauge. Then the crystal tube was pulled off under vacuum and a glass-rod hook was glassblown to the top of the tube (see Figure 7). In the case of approximately the

first 2 0 boules grown, materials were sublimed under vacuum (or in a partial atmosphere of nitrogen) into the crystal tube which in addition to being flamed had been washed with a hot Alconox solution. Also, a mercury dif fusion pump was employed in addition to the forepump. No apparent differences in magnetic resonance observations or crystal quality resulted after the less involved prepara tive method described above was adopted. Boules were grown from a mixed melt of the pirified reagents by the Bridgman method in an air-core furnace

(depicted in Figure 8 ) similar to that described by Sher- wood and Thomson. 56 By employing chart-drive motors of 53

Figure 6 . Disconnectable vacuum manifold, powder trap and crystal tube used for preparation of mixed crys tals. (Note second powder trap built into vacuum manifold.) During degassing of sample material, crystal tube would be inserted in standard taper joint at bottom of powder trap. ^crwrJtYi 55

Figure 7. Mixed single-crystal samples in various stages of preparation: (a) crystal tube (sealed under vacuum) containing unmelted starting materials; (b) crystal tube and boule (phenanthrene-h^ in biphenyl) after removal from Bridgman Furnace; (c) boule (chrysene-d-^ in p-terphenyl) after re moval from crystal tube; (d) cleaved boule (naphthalene-dg in biphenyl) from which single crystal samples will be cut, (note reflection of light from the primary cleavage planes). w \ vtafe*# ^n-y'-f 4 3 w b # •>, . ,r> S:vfc$V .'■ r ~ i s i £f/«?t.i'L >* '5';’•.•I**"’1 i * ,.“' t ? ;

S&SSiSl^t^vA ‘ t - -. ISIIS*; i:*S3

«&!$$$

^laSAi&w * *-« ■-. .

VMMMtwUr; ^ >Si.-; t-‘A --:-.1;- ■• ■ : !:\ ■ §^£V0^-r-^\ »i**.''V;:vU'?j - -_—■». - li' hr-' N i t * ;

ISMl-tti

S*«r5« iii &>&'•&* •?“>•*'i ^ . i i .• •;*.; -■!*;ifetr-C.V* "*:-i '*•••• fJSSS^*r.K 1 ’

« £ ? | | * - 1-

V - * ' i m g m fi r j

$ 57

Figure 8 . BRIDGMAN FURNACE

A. Haydon Chart Drive Motor; 1/24, 1/96 or 1/192 rph. B. Line to 7.1 amp, 120 volt Variac. C. Aluminum plate, 1/2". D. Maranite (Johns-Mansville), 1"; insets cut for Pyrex tubing. E. Stainless-steel, worm-drive hose clamp, 2%". F. No. 18 Nichrome wire insulated with fishspine; length

chosen so that resistance is approximately 2 1 ohms. G. Aluminum rod with threaded ends, 1/2". H. Pyrex tubing. 58

rTJr-iiriln

T O P VIEW

SIDE VIEW 59 either 1/24, 1/96, or 1/192 rph speeds, growth rates of from 3 to 0.3 mm hr"^ could be obtained. In general, boules grown at rates of less than 1 mm hr"^ were of best quality. Specific descriptions of the properties of boules grown from various mixed samples (other than the ones discussed below), are given in the appropriate Sec tions of Chapters V-VIII. A unique identification of the boules grown was made by recording the materials (and their weights) used in the preparation of each boule on a separate page in a Crystal Notebook. A boule was then assigned the number of that page in the Crystal Notebook on which its preparation was described. This procedure was strictly followed for the 134 boules prepared for this study, with the result that no ambiguity ever arose about the nominal composition of any crystal used in the research. Boules of p-terphenyl were grown with nominal concen trations of chrysene-d^ 2 ranging from 0 . 0 1 to 0 . 0 0 0 0 1 mole fraction and often were of excellent quality (showing clear striations on well-defined cleavage planes). Spec- trophotometric (UV) analysis of crystal pieces used in mag netic resonance experiments (4.2: 267, 281; 4.3: 40-43) gave the corresponding actual concentration range as ~ 0.004 to ~ 0.00001 mole fraction. Nominal concentrations for boules of other mixed crystal systems ranged between 60

"■'0.005 and 0.0003 mole fraction.

Mixed crystals of chrysene- d 1 2 in unpurified OHA grown by the Bridgman method were generally not of good quality, being glassy with no clearly defined cleavage planes within a boule, nor possessing reproducible cleavage planes from boule to boule. In contrast, chrysene-d^ mixed crystals prepared from OHA recrystallized twice from ethanol were ice-clear and showed definite cleavage planes as they first emerged from the growth zone. Only later did these boules become translucent and the cleavage planes less well-defined as they cooled toward room temperature. A third type of behavior was observed for mixed crystals of recrystallized OHA containing tetrahydroanthracene (the suspected impurity-source of the phosphorescence emission and magnetic resonance absorptions in unpurified OHA). These boules consistently shattered violently, with suf ficient force to break the crystal tube some hours after emerging from the heated region of the air-core furnace. In contrast, un-doped boules of recrystallized OHA re mained ice-clear and retained definite cleavage planes months after being grown. Two groups of investigators have described phase transformations in OHA similar to those observed here, but neither group discussed the role 57-58 of impurities in these processes. A thorough inves tigation of this problem lay outside the scope of this 61

research, but such a study seems worthwhile. A brief at tempt was made to obviate possible undesirable effects of a solid-solid phase transformation by growing crystals of chrysene-d-^ in OHA from benzene solutions at room tem perature (4.3: 107). Little phosphorescence was observed from the crystals produced, and apparently too little guest was incorporated to be useful. (However, it should not be concluded on the basis of these limited results that suitably doped crystals could not be grown from solution if more extensive efforts were made toward that end.) Single-crystal samples were cleaved from boules using new razor blades from which all grease had been removed by washing with acetone. In most experiments crystal pieces were oriented approximately (visually), and were mounted in the cavities on a bed of Apiezon N grease (for good thermal contact). In a few experiments, identified below, single-crystal pieces were oriented on polystyrene wedges machined to known angles to improve the alignment of the static magnetic field with a magnetic fine-structure axis. Typical dimensions of single-crystal pieces used in mag netic resonance experiments were 7 mm x 6 ram x 3 mm. 62

C. HOST CRYSTAL STRUCTURES !• P-Terphenyl The crystal structure of p-terphenyl has been de termined by Pickett,^ Hertel and Romer,^^ and more re- 61 cently by Rietveld and Maslen. p-Terphenyl crystallizes in the monoclinic prismatic class with two molecules per unit cell, and the space group is P£ (l)/c. If the unit cell is taken to have = 92° l1, then the unit cell edge lengths are a = 13.613 A, b = 5.613 A, and c = 8.106 A. The structure of p-terphenyl is shown schematically in Figure 9. Rietveld and Maslen, and Pickett, have shown that: (a) the molecule in the crystal is essentially planar; (b) the dihedral angle between p-terphenyl molecu- Q lar planes is 69 + 1 ; (c) the two molecules in the unit cell have their long in-plane axes parallel, and these axes are almost exactly in the plane normal to the sym metry axis (b axis) of the crystal; and (d) the angle between the a axis and the projection of the x axes in the crystal symmetry plane is 15.3°. These data were sup plemented by data from an X-ray diffraction study of limi ted scope, which enabled us to relate recurring external morphological features (primary and secondary cleavage planes and striations thereon) to the known structure. Specifically, it was shown that the primary natural clea vage plane (which invariably was parallel to the boule Figure 9. p-Terphenyl crystal structure. 64

2^ 65

axis) was the be plane, and that the intersection of the most prominent secondary cleavage plane with the primary plane was parallel to the b-axis or dihedral-angle bisec tor. The axis system chosen for p-terphenyl is shown in Figure 10 and is analogous to that previously chosen for biphenyl.^ 2. Symmetric Octahydroanthracene A detailed crystal structure determination has ap parently not been reported for OHA, although a limited num- 61 ber of structural data are available. However, since a comprehensive knowledge of the chrysene-d^ splittings had been obtained from the prior experiments employing p- terphenyl as the host, it was not necessary to have com plete knowledge of the OHA structure. Further, a quali tative understanding of the chrysene-d^ orientation in a given experiment in OHA host was obtainable by observing which zero-field selection rules were satisfied. 3 • 2_'-Binaphthyl No crystal structure data could be found for 2,2*- binaphthyl. Interpretation of experiments using this host material more extensive than were carried out in this research may require at least a limited structure determi nation. Figure 10. p-Terphenyl axis system. 67 CHAPTER V RESULTS AND DISCUSSION FOR ORIENTED CHRYSENE IN ITS PHOSPHORESCENT STATE

A. INTRODUCTION Electron magnetic resonance absorption from proto- chrysene oriented in p-terphenyl was first observed in this laboratory on 1965 August 16 when two closely-spaced but distinct resonance absorptions were detected at low fields near — 0.116 cm In succeeding experiments ad ditional resonance absorptions in the energy range from — 0.1156 to 0.1170 cm ^ were at least partially resolved, and it was established that at least 3 of the signals were associated with similar but clearly different zero-field splitting values. A plausible explanation of the multiple absorptions (and zero-field splittings) was that only one of the signals arose from oriented phosphorescent chrysene, and that the remainder originated from impurities in the guest or host materials, or both. However, extensive puri fication of materials (Sec. IV.A) failed to affect the num ber of signals observed. Multi-signal patterns were also

68 69 observed from single crystals of chrysene-dj^ in p-terphenyl at — 0.116, 0.063 and 0.054 cm"*, and each signal was found to be associated with a unique zero-field splitting. These observations of multiple magnetic resonance absorptions cor responding to different zero-field splittings for one che mical species were novel and unexpected. This chapter is devoted to a description of the ex tensive study made of "multiplet structure" in low-field magnetic resonance spectra cf chrysene oriented in p- terphenyl (and other host structures). Results from in vestigations of the temperature dependences of the zero- field splittings and triplet lifetime of oriented chrysene are also presented.

B. MULTIPLE RESONANCE ABSORPTION IN LOW-FIELD FINE- STRUCTURE SPECTRA OF TRIPLET CHRYSENE ORIENTED IN p_- TERPHENYL As discussed above, the x axes of the two molecules in the p-terphenyl unit cell are virtually parallel* Thus, if chrysene molecules replace p-terphenyl molecules substitu- tionally in such a manner that corresponding axes (the axis system chosen for chrysene is shown in Figure 11) are parallel, only a single resonance absorption by triplet chrysene should be observed for a field scan (from — 400 to 0 G) for H directed approximately parallel to the Figure 1 1 . Chrysene axis system. p-terphenyl x axes. On this basis, as well as on the basis of previous low-field studies, 1 * 2 it was expected that only one signal would be observed for microwave frequencies cor responding to — 0,116 cm"*' in experiments in which the p- terphenyl mixed crystal pieces were oriented with the pri mary cleavage plane (0 0 1 ) approximately normal to H (in this orientation the angle between the p-terphenyl x axes and H is approximately 16°). What was in fact observed in such experiments on chrysene-hj^ 1 ° p-terphenyl crystals (at fields below -—100 G) was an overlapped pattern consis ting of at least four distinguishable resonance absorptions. Three of these signals could be resolved at fields below ■—25 G to the extent that their zero-field splittings were separately determinable. The three values found, namely 0.11704, 0.11615, and 0.11573 cm \ although similar in magnitude differed from each other by approximately twenty standard deviations or more. Since intensive effort to purify chrysene-h^ did not result either in any characteristic alteration of the ab sorptions or in reduction of the number observed, single crystals of purified chrysene-d^ p-terphenyl were pre pared. As anticipated, deuteration of the guest reduced its resonance absorption linewidth from — 20 G to ~ 6 G, resulting in the following observations. 72

When the microwave frequency was chosen so that resonance absorption occurred at *—400 G, a single severely distor ted signal was observed. At a frequency such that reso nance occurred at — 275 G, the pattern shown in Figure 12 was observed, and as the field for resonance was lowered further, this pattern was further resolved. Finally, when the resonance absorption which occurs at the lowest field (for P = "'-0.117 cm ) was centered at '—10 G, the striking pattern (multiplet) shown in Figure 13 was obtained. Cor responding multi-absorption patterns observed for the transitions near — 0.063 cro“^ and — 0.054 cm~^ and at com parably low fields are shown in Figures 14 and 15, respec tively. Systematic studies of crystal pieces cleaved from ap propriately prepared boules showed that neither the num ber of signals within a pattern nor their relative ampli tudes depended significantly upon (a) guest concentration over three orders of magnitude, down to only ten parts per million (a limit set by practical convenience in signal de tection); (b) the rate of growth of boules from 0.36 mm hr"^ to 3.5 ram hr"-*-; (c) the particular chrysene or p- terphenyl samples used; or (d) the particular boule chosen as the source of experimental samples. Upon setting the microwave frequency so that patterns of the types shown in Figure 12 were obtained, then changing 73

Figure 12. Incompletely resolved resonance absorptions

from chrysene-d^ 2 in p-terphenyl at ~0.126 cra"^- and "~275 G. t \ t 74 Figure 13. Resolved multiple resonance absorptions ob served at low fields from chrysene-dj^ *n p-terphenyl at ^0.117 cm”*. ~ 120 G

" v l 77

Figure 14. Multi-absorption pattern observed at low fields

for the chrysene - d 1 2 (in p-terphenyl) transi tions near ^0.063 cm“^.

Figure 15. Multi-absorption pattern observed at low fields for the chrysene-d^ (in p-terphenyl) transitions near *~0.054 cm"*-. ao

1 0 0 G 81 the frequency by small increments, the centers of the indi vidual signals were moved successively to zero magnetic field, permitting precision determinations of the zero- field splittings for each resonance absorption. The en hanced field-separation of one of a set of multiple sig- 4 nals as the zero-field frequency associated with it is approached quite closely by the experimental frequency is due to the substantial change in the value of the first derivative of the differences between energies of the triplet levels with respect to the magnitude of the static field, even over the small range between 100 and 0 G. (The value of d £/dH ranges between 0 and 1 as H varies from 0 to oo ; at 100 gauss its value is 0.285.) Moreover, at fields above ~1000 G, the value of this first deriva tive is sufficiently large (0.95) that the field- differences corresponding to the zero-field energy dif ferences among components of a multiplet are less than the resonance absorption linewidths for the individual transitions. Thus, only a single (composite) resonance absorption should be observed from one of the translationally inequivalent sets of molecules at such fields, and this indeed was found to be the case in high-field experiments for H parallel to the x axis of one translationally inequivalent set of p-terphenyl molecules. Fi gure 16 presents a resonance absorption observed at 82

Figure 16. Resonance absorption from chrysene-d-^ observed at ~2200 G for H very nearly parallel to the A axis of one translationally inequivalent set of chrysene triplets. 83

C0 84

~-2200 G in such an experiment, which corresponds to the low- field transition for which the multi-absorption pattern shown in Figure 13 was obtained; the width of the high- field absorption was 5.9 G 1 1 G, whereas the average width of the component signals at low field (Figure 13) was 5,6 G i 1 G. Thus, in the case of this transition, the seven resonance absorptions observed at low fields are very nearly superimposed at fields above approximately 2000 G. In high-field experiments in which the static field was rotated in the yz plane of p-terphenyl, on the other hand, multi-absorption patterns were observed, and the resolution of the patterns and the number of distinct absorptions observed in them was found to depend upon sta tic field orientation. These observations had important implications with respect to the possibility of inequivalently oriented sets of chrysene guest molecules in the p-terphenyl host structure and will be discussed further in Section V.D.

G. ZERO-FIELD SPLITTINGS AND SPIN HAMILTONIAN PARAMETERS

FOR CHRYSENE-d 1 2 1. Basic Relations For description of magnetic resonance absorption in triplet chrysene we have used (for reasons discussed in Chapter II) the spin Hamiltonian 85

]}-{[ = s|/3|h.S + D Sz 2 -f E(SX 2 - Sy2), (9) where x, y and z designate the principal axes of chrysene (Figure 11), which are assumed (at least provisionally) to be parallel to the principal axes of the magnetic fine structure tensor. The three eigenvalues of the spin Hamil tonian (9) when [h| ss 0 are D + E, D — E and 0, and the three values for P = frequency/c for the zero-field transi tions (splittings) are |d + Ej/hc, |d — e|/he, and |2e| /he. As H is increased from 0 by amounts such that, for the case of chrysene, G2 << (2E) 2 where G = g [/f] H, we recall from Chapter II that the eigenvalues behave as follows:

0 + E-> (D+E) + (G2 /2E)[1- C/(D+E)J , (6 )

D - E-> (D-E) - (G2 /2E)[1- C/(D-E)] , (7)

0 - G 2 £ / (D2 -E2) , (8 ) where C = 2 2(D — E) + m2 (D + E) , f = cosine of angle between H and the axis, and rn = cosine of angle between H and the y axis. Since for the precision data presented here the largest Zeeman contribution to any splitting corresponded to

G2 /(hc) 2 ~ 7 x 10 ^ cm" 2 whereas (2E)2 /(hc) 2 was determined _2 to be — 3 x 1 0 J cm , the inequality stated above as the condition of validity of Relations (6 ), (7), and (8 ) was in deed very well satisfied. Accordingly, all precision data 86 were least-squares fitted to expressions of the form:

Pi = a + ylgj 2 , (1 0 )

where is c " 1 tiroes the i measured frequency for resonance, is the measured field for resonance at the i n measured frequency, and a and Y are parameters. Mean values of a from least-squares fits of the data for the several determinations of each splitting were taken to be the best values of the triplet state zero-field energy- level separations divided by he. The values of Y depend not only upon the splitting parameters of a particular triplet species and the value of g (assumed isotropic, as noted above), but also upon the orientation of the fine- structure axes (A, B and C) with respect to H, and thus, upon C as defined above. Hence the values of y may vary substantially from one experimental series to another. However, for a given experimental series in which the relative orientation of crystal and static field is fixed, differences (adjusted for the small variations of the splitting parameters for the various components of the raultiplet) larger than the combined uncertainties in the best-fit values of Y for the various component signals within the raultiplet are indicative of differences in the orientations of sets of guest triplets giving rise to the component signals. This will be elaborated upon in Section V.D. 87

Data obtained in a high-field experiment in which the p-terphenyl mixed crystal was oriented so that H could be brought very nearly parallel to the A axis of one translationally inequivalent set of chrysene triplets were used to calculate a value for gxx. The iterative calculation was made using Relation (1) in Appendix A (a listing of the computer program employed is given in Appendix B) and a value of 2.0027 was obtained for gv„. Observation of multi-absorption patterns in y, z experiments at high fields prevented unique determinations of gyy and gzz. Thus, for the analysis of all low-field data g was taken to be isotropic and of value 2.003. 2. Triplet Chrysene-d-jo in Single Crystals of p-Terpheny1-h a. Determination of Zero-Field Splittings Ninety-five precision determinations, compri sing approximately 500 measurements of pairs of values

ISil for resonance were made for 2 0 different zero-field splittings studied for chrysene-d^ oriented in p-terphenyl- hj^. Each accepted value of [H^j for resonance was ob tained by averaging at least two measurements of the reso nant field, once sweeping upfield and once sweeping down- field, for the given value of Briefly, but more spe cifically, the procedure used in a precision determination of a splitting was: (a) match the cavity and line to the 83 selected microwave frequency; (b) record a harmonic of this frequency; (c) sweep the field up from 0 to the maxi mum in the resonance absorption ("baseline crossover" for the first derivative of the absorption); (d) record the potential drop corresponding to the crossover field and then sweep to the high field side of the absorption; (e) retune the cavity;(f) sweep down to crossover and record the voltage, and (g) record the (zero-beat) fre quency harmonic again. This procedure was repeated for each v^t JH^| pair in the set of ^ ' s and |h^| *s compri sing a single precision determination. The primary data obtained in this manner for the 2 0 zero-field splittings for chrysene-d^ 2 p-terphenyl are too numerous to be presented in raw form here, but the raw numbers are tabu lated in Notebook 4.C.1, pages 77 through 238. These pri mary data and selected derived data are summarized in Tables 2, 3 and 4 (for the transitions observed near — 0.116, 0.063 and 0.053 cm"*1, respectively). The data for each experimental series were least- squares fitted according to Equation (10), weighting all measurements equally. The best-fit values of the slopes and intercepts are given in Columns 7 and 10, respectively, of Tables 2, 3 and 4. The standard deviations of the slopes and intercepts were calculated by the usual least- squares formulae (a listing of the least-squares computer Table 2. Magnetic Reaanaaee Data fa r OiTT*rnr-d Oriented la p-Terphenyl. T n n jltio o i ae«r ** 0.116 cm'1 .

Fun t u b e r Cry* U l Ro. o f d iffe r* Benge o f Beg* Apprnxlaate Sign Beet nluee Standard Calculated Seat nluee Standard Mean n lu e e o f end *1(0*1 tnqjeratur* en t v j fo r netlc field orientation of • of of J for deviation Zeexan of tero-fleld d ev iatio n te ro -f le ld o u fe tr (01) ufalcb I g i l o n r which £ end J , , r i t o f Iq . o f r oootrlbutloa apllttlng* o f beat epllttlnga, me aeeeortd rteonencearteptctirely dwl (10) to ta o b e e ra d f o r f i t o r n lu e e o f «1U) atandard were deter* (p-terphenyl d lU * 10* e p llttln g a derlatlona, at < % ! Eq. (10) to z.r.s. alned (C) axle eyeten) at eaalleat data (® *l > e o .i 0 K * fig. 9 end 10 x l e r (®*M (c»*r ) (e**l C‘a ) ( a 1 G*») * 10* x 10*

1- I 60.1 5 6.1-19.9 *; n e 1.286 3.2 6 0.116998 1 0.117006 2 - 1 80.1 1 6 .3- 13.* *; t* 4 1.372 5-7 5 0.117013 1 0.000006 1- 1 7T-2 3 6 .9 - H .5 *; r t 4 1.571 1.6 8 0.117022 < 1 5 - 1 90.5 3 3 . I . 10.2 x; y* 4 1.113 3-0 2 0.116967 < 1 15- 1 60.1 1 2 .5 -1 3 .2 *; »x 4 1.285 3-1 1 0.117060 < 1 0.117060 "

1- 2 B0.1 3 6 .6- 10.1 *i 7* * 2.059 17.1 8 0.116967 1 0.116977 2 * 2 80-1 1 1.9-12.5 * i 7* 4 1.570 7-7 3 0.116905 1 0.000007 1 - 2 77 .2 2 H . l - 13.5 *! 7* 4 1.676 - 22 0.116986 _ 5 - 2 90.5 3 I . 3-10.5 *i 7* 4 1.365 8 .B 3 0.116939 1 15- 2 80.1 1 1 .6 - 12.1 *1 71 4 1.368 12.8 2 0 . 1170;* < 1 0 . 11702*

1 - 5 80.1 5 6 . 3 ^ o .9 *1 7* 4 1.110 1 .3 6 0.116883 < 1 0.116891 2 - 3 80.1 6 6 .9 -22.2 *S 7* 4 1.383 0.5 7 O .H 6892 < 1 0.000005 3 - 3 80.1 6 5-7-22.1 XI n 4 l . U l 2.9 5 O .II6897 1 1 * 3 77-2 1 1.1-15.2 * t 7* 4 1.523 9-* 1 0.116902 1 5 - 3 90.5 1 5.0-27.2 *1 7* 4 1.J77 0.5 1 0.116852 < 1 1. 1 5 - 3 e o .i 5 1.0-20.7 * ; 7* 4 1.383 0 .8 2 0.116962 < 1 0.1169*2

1 - 1 80.1 1 5.7-23-6 XI 7* 4 1.105 1.1 1 0.116633 < 1 0.116639 2- 1 80.1 6 6 .2-23.3 XI 7* 4 1.399 0 .3 6 O .U 663S < 1 0.000005 3 - 1 80.1 5 5 .1-21.3 *1 7X 4 1.152 1.6 1 0 . 11661.1 < 1 1* 1 77.2 6 9-1-25.5 *1 yx 4 1.128 2.1 12 0 . 1166*5 1 5* 1 90.5 1 5.0-21.7 x ; yx 4 1.353 1.2 1 0.116611 < 1 w 1 5 - 1 80.1 5 2.5-21.1 x; yx 4 1.392 1.6 1 0.116720 < I 0.116720

1 - 5 80.1 1 5.8-25.0 * ; yx 4 1.159 1.1 5 0.116121 < 1 0.116133 3.1-22.8 2 - 5 80.1 6 *i yx 4 1.127 0 .2 1 O.II6156 < 1 0.000008 3 - 5 80.1 6 8.3-21.2 xj yt 4 1.133 0 .2 10 0.116138 < 1 1 - 5 77.2 5 9.1-22.1 x i yx 4 1.397 0.5 12 0.116153 < 1 5 - 5 90.5 5 5.6*25.1 xj yx 1.391 0 .6 5 0.116090 < 1 b 5 .5-21.8 1 5 -5 80.1 5 x ; 71 4 1.116 1.6 5 0.116117 < 1 0.116U 7

1- 6 80.1 1 5.9-25.6 x i yx 4 1. I 1I 1.9 6 0.115687 1 0.115697 2 - 6 00.1 6 5 . 1-22.2 xj 71 4 1.105 0.6 3 0.115697 < 1 0.000006 3 - 6 8» . l 7 6.6-25-6 x; yx 4 1.106 1-5 7 0.115697 < 1 1 - 6 77.2 1 10.9-28.0 x> yx 4 1.131 0 .3 17 0.115708 < 1 5 - 6 90.5 1 5-1-23-5 x; T* 4 1.188 1.1 5 O .II5669 1 b 1 5 -6 00.1 1 7.5-19.5 xi yx 4 1.309 1.5 7 0.1157*1 < 1 0.1157*1

0.115612 2 - T 80.1 6 1 . 2- 16.0 xi yt 4 1.375 0 .8 2 O .II56U < 1 0.000002 . 5 - 7 90.5 3 1 .6 - 16.2 x; yi 4 1 .3 U 2.1 3 0.115578 < 1 1 5 -7 80.1 1 6 .1-11.8 x j yt 4 1.JT5 3-7 5 O .II5678 1 O.U56T0

a Mm b n in e * m » obtained mllnalng cocmraioQ of data for Rucx * and 5 to 8o.l° K oiioc the teaperatur* coefficient dctarmlsad la ibla reeearcb. b Run 1J data n r* obtained fn * dtrytene-d^ oriented la p-teTpbeayl-d^ • All other data nr* obtained uiinf p-tarpbeuyl-h^ . Table 3- Magnetic Resonance Data for CTiryaene-d Oriented In p-Terphenyl Transitions near ~ 0.06}' cm-1. 12

Run nuc&er Crystal Ho. of differ Range of mag Approximate Sign Beat values Standard Calculated Best values Standard Mean values of and signal temperature ent «i for netic field orientation of of of y for deviation Zeeman of xero-field deviation xero-field number ( ° 1C) which (jM was over which JJandRj , fit or Eq. of y a contribution splittings of best splittings, measured resonances respectively dv*t (10) to data x 10* to observed for fit of values of with standard were deter (p-texphenyl dJHjJ X 10* splittings Eq. (10)to Z.F.S. deviations, at mined (o) axis system) at smallest data (cm-1 ) 8 0 .1 0 K * (cm-1 c‘s ) (cm"1

6 - 8 77.2 6 9 .0-2 2 .1 x ; x 0.918 1 .1 8 0.062136 < 1 0.062157 7 - 8 80.1 3 12.0 -2 2 .1 y ; x + 0 .6 1 3 1.2 9 0.062182 < 1 0.000016 8a- 8 80.1 5 1 3 .8-2 1 .2 * i x - 0.117 2.9 • 9 0.062150 1 8b- 8 80.1 6 12.8-21.7 y . * + 0.750 1-3 15 0.062165 1

6- 9 77*2 l* 8.5-19-5 = . * 0.961 1.9 7 0 .06270 ! < 1 0.062700 7- 9 80.1 1 6.5-25.1 y . x + 0.162 3.0 1 0.062700 1 0.000006 8a-9 6o.l 5 8.3-25.1 * t x - 0 .1 7 9 1.5 1 0.062692 1 8b- 9 Bo.l 1 13.8-25.0 y ; * + 0 .5 6 3 1.8 12 0.062710 1

6- 10 77.2 1 6 .0 -1 3 .0 x ; x 1.005 1.8 1 0.062825 < 1 0.062836 7- 10 80.1 1 7.0-21.9 y i x + 0.168 2.1 3 0.062851 < 1 0.000011 8a- 10 8o.i 5 11.1-21.3 x ; x - 0.103 1.0 5 0.062830 < 1 8b- 10 B0.1 5 8.8-15.9 y ; x + 0.181 1.3 1 0.062811 1

6- 11 77.2 6 6.3-16.6 t ; x l.oll 2.1 1 0.062893 < 1 0.062901 7* 11 8 0 .1 3 9.3-22.2 y v x + 0.60a 0.7 5 0,062911 < 1 0.000008 8b- 11 60.1 6 11.1-22.1 y ! x 0.765 1.1 16 0.062907 < 1

6- 12 77.2 l» 8.8-16.9 X J X . 1 .1 5 2 1.8 9 0.063088 1 0.063086 7- 12 60.1 3 10.1-20.1 y ; x + 0.573 3 .8 6 0.063090 1 0.000006 8a- 12 8 0 .1 5 11.3-23.3 X J X — 0.112 1.8 8 0.063073 1 8b- 12 80.1 6 15.7-26.0 y 1 x + 0.719 1.1 18 0.063093 1

9- 13 77-2 1 9.3-19*1 s i x - 1.021 1.8 9 0.063302 < 1 0.063298

a Kean values were obtained following conversion of data for Run 6 to BO.l ° K using tbs temperature coefficient determined In thla reaeareb.

to o M lt t, ttogaetlc Retooanc* Eata for Otryieee«d Oriented la p-Terpbecyl. Transition* near ** 0-053 ™’A *

Bm m rfcer d j i t i l Bo. of differ Bines of d i Appreildte Sign Bait value* S tu d ird O eleulxted Beit nine) S tm d ird Hein nluei of u l a lg o il t o f c n t m ent Vj far nette field orientation of o f c f y fo r d e d itlo n ZeeDM of lero-flsld d e riitio o x e m -fie ld m a te r CD ntdebl^jluM d i r vhlcb fl,»nd R, , fit of Eq. contribution ■pllttlngi o f b e it ■jdlttlngl, w r n i l rtio m n e ti rttpectlsely (10) to data “' x f c to obi erred for fit of n l o e i o f irllh itindud vers deter- (p-terphenyl d ^ l x 10T ■ p u ttin g s »1. (10) to z . r . s . derlitloni, it Blued CO) u l s e y itm ) ■t toe lic it ( a - 1) 80.1 0 It • rt|. 9 ud 10 (e* fl 0*a ) (CB-I O '*) Hj ( o f 1 ) n » - > ( = ’1 ) 1 10* x 10®

11- lli 8 o .i T 5.1-22.5 > 1 t 1.216 1.1 3 0.052308 < 1 0.052312 181- n TT-5 5 2B.6-J6.7 y ) i 4 0.712 9.5 59 0 .05232B 10 0.000007 10b- Ik TT.5 5 16-9-21.0 r ; 1 4 2.k33 10.5 71 0.052316 k 16- lk e o .i k 11.1-22.1 s t * - 1.325 k.6 16 0.05233T 1 0.052337 b

U - 15 80.1 8 9-1-25-5 1 i 1 1.157 3-1 10 0.052620 1 0.05262k 10- 1J B o.l 6 9 .0-20.8 1 ; . 1.216 5-6 10 0.052606 1 0.000010 I k . 15 80.1 6 9.8-21.7 * 1 * 1.313 l.k 13 0.052621 < 1 1 7 i- 15 77.5 5 20.3-31.6 y ; i ♦ 0.710 1.6 29 0.052612 2 1Tb- 15 77.5 5 12.1-19-6 r ; t 4 2.326 7 .1 33 0 .05263a 2 18* - 15 77-5 k 20.9-52.J y • * ♦ 0.716 5-2 . 33 0.052611 k 18b- 15 77-5 k 12-3-18.3 y i < a 2.193 11.7 ’ JT 0.05263T 3 80.1 16- 15 6 8 .0 -23.3 * • * 1.363 1.2 9 0.052599 < 1 0.052599 b

11- 16 80.1 7 7.k-23.T x ; * l.U T 1.7 6 0 .0528J1 1 0 .05237k 13- 16 7T.2 5 6 . 1-17.8 x ! * 1.159 3-8 5 0.052877 1 0 .00000k lk - 16 80.1 6 7.6 -2 1 .5 x ; x 1.316 0.7 T 0.052872 < 1 1B«- 16 TT.5 k l 6.k -28.6 y i * + 0.T2T 3-9 19 0.052887 2 18b- 16 TT.5 k 9-T -l6.k y t * 4 2.260 k .6 21 0.052885 1 16- 16 80.1 5 9 -2-22.0 * i * 1.313 0 .6 u 0.052875 < 1 0.052075 b

U - IT 80.1 6 6-8-23.9 X J I 1.088 1.1 5 0.053111 < 1 0.053113 1 ?- IT 77.2 6 T.5-1T.8 x ; x l . l k l 3.2 6 0 , 05311k 1 0.000003 lk - IT 80.1 5 7*6-20.7 * 1 * 1-319 0.9 T 0.053H5 1 I 6 i - 17 TT.5 5 22.3-35.0 y 1 * ♦ 0.739 2.1 37 0.053152 2 18b - 17 77.5 5 12.5-19-3 y i * « 2.531 5-9 10 O.O53I I 9 2 18c- 17 T7.5 k 22.1-33.1 y i * 4 0.71k 2-9 36 0.053153 2 lBd- IT TT.5 k 12-1-17.8 y t ♦ 2-592 8.8 39 0.053150 2 16- IT 80.1 k 9-3-25.2 X 1 t 1.352 1.2 13 0.053168 < 1 0.053168 b

11- 18 80.1 6 7-3-22.6 X i X 1.062 l.B 6 0.05392k 1 0.053926 l j - 18 77.2 5 7-1-25.2 x i x • 1.018 . 3.5 6 0.053932 1 0.000002 18a - 18 77.5 5 19.2-53-1 y t * 4 0.619 2 .0 25 0.05393k 2 18 b - 18 TT-5 5 I0.k-1T.B y * * 4 2.267 7 .8 25 0.05393k 2 1 6 - 18 80.1 5 5 * l-2k.O x ; x - 1.17k 2 .0 lk 0.053ST3 1 0.0539T3 b

U l - 19 8 0.1 9 5 .1 -2 0 .k X i I 1.003 1.6 3 0.053998 < 1 0.053999 U b - 19 8 0.1 5 9-2-19.9 1 ; * _ 0.998 k.k a 0.053998 1 0.000002 15- 19 TT.2 k 7 .0-16.9 x ; x * 1.001 21 .0 7 O.05I 009 1 16- 19 60 .1 k U . 2-20.0 X t x * 1.166 0.5 15 0.051011 < 1 O.CSlOkk b

1 1 - 20 B o.l 9 6.0-23.5 -X 1 I l.o k o i . e k 0.051195 1 0 . 05 U 95 12 - 20 TT.2 k 8 .0 -2 2 .6 X t * • 0 .96k 2.k 6 0.051202 1 0.000000 16- 20 80 .1 5 6 .1 -1 8 .2 X J X ■ 1.191 0 .9 5 0.051196 1 0.051196 b

o >te*a valuta vert obtained foUmrioc coorvraloQ of data for ftaa12, 1J, IT, aid 10 to 00.1 0 K ualng.tha tecpartton caaffleltfit daUrsiotd la thi* ntw rch. ■ _ Em 16 data w ri obtained fro* ehrjam t-d^ orlenttd In p*luptiinirl^4 - All other data irrra obtained uiln* p-t*rpbatqrl*h^ * 92 program is given in Appendix B) and are given in Columns

8 and 11, respectively. The mean values of the intercepts (Column 12) were calculated weighting each individual a by the number of different for which |h J was measured in its determination (Column 3). Figure 17 is a plot of data from .Series 2 for the ~ 0.116 cm - 1 transition of chrysene-d^ 2 oriented in p-terphenyl-h^ together with the best-fit linear functions. Extrapolation of these functions to zero field, as shown, yields seven discrete intercepts (zero-field splittings). The values of the intercept, a, found for twenty dis tinguishable resonance absorptions for chrysene-d-^ In p- terphenyl-h^ matched closely one of the three values

'-'0.116 cra”^ (7 values), ~ 0.063 cm"*' ( 6 values), or ~ 0.053 cm"^ (7 values). The largest of these values is the sum of the other two, and there are twelve possible different assignments of the three zero-field eigenvalues, 0, D + E, and D - E, of the spin Hamiltonian (1) to the three approximate experimental energies. The number of allowed assignments was less, since the observed signs of diVd|H] for various orientations of H with respect to the p-terphenyl axes (given in Columns 6 and 5, respectively of Tables 2, 3 and 4) ruled out certain of the possible assignments. These signs can be deduced for each assign ment from Relations (6 ), (7) and (8 ) upon assuming that the iue 17. figure ENERGY/he (cm' 0.116600- 0.115800 — 0.1 17000 s 0.115400 0.116200 - 0.117400 Plot of data from from data of Plot transitions of chrysene-d-^ oriented in p- in oriented chrysene-d-^ of transitions terphenyl-h^. 3 5 4 3 2 I0~2x 2 H Run 2Run (gauss2) o te ~ the for 0.116 cm

93 94

x, y and z axes chosen for chrysene-d^ (F*-Sure 1 1 ) are approximately parallel, respectively, to the x, y and z axes of p-terphenyl molecules as given in Figures 9 and 10. Subject to this assumption, the only assignments al lowed by the data given in Tables 2, 3 and 4 were found to be:

(D - E)/hc=i 0.116 cm"-*- (D - E)/hc=i 0.053 cm" 1

1 1 o r 1 - 1 (D + E)/hc;= Z 0.063 cm" 1 (D 4* E)/hc=Z0.063 cm 1 and thus the only allowed approximate values of D/hc and E/hc were found to be: (A) (B)

D/hc = ± 0.090 cm" 1 D/hc ± 0.005 era" 1 1 or -1 E/hcn Ijp 0.027,0m"1 E/hc — 310.058 cm" , only the upper or lower signs to be taken in each case. Specific experiments which would permit completely unam biguous assignment of D/hc and E/hc to either values (A) or values (B) were not feasible in this study due to ex perimental restrictions imposed by the present apparatus. Those experimental results which had bearing on this ques tion could be interpreted in favor of assignment (A) and not in favor of assignment (B) but such results were not completely conclusive. Although the assignment of D and E could not be made unambiguously on the basis of any low- field or high-field single-crystal experiment, there are other bases on which this choice can be made for chrysene. 95

In particular, values of the spin Hamiltonian parameters

n i g for chrysene calculated theoretically * and measured 0*1 experimentally (in glass studies ) are in good agreement with assignment (A) but in complete disagreement with as signment (B). Moreover, for the series of polyconjugated aromatic hydrocarbons to which chrysene belongs and in fact for every molecule for which |D|/hc and [E|/hc have been reported, the ratio |d /e | has always been shown to be greater than unity. Thus, while independent assignment of D and E on the basis of data obtained in this study would be preferable, there is ample justification for choosing assignment (A) and for rejecting assignment (B). b. Correlation Among Zero-Field Splittings To' permit interpretation of the observed multiplet structure, correlation was sought among the indivi dual zero-field splittings of the three sets of splittings. Accordingly, a matrix was constructed whose row-headings were the experimental values of Jd - E|/hc, whose coluran- headings were the experimental values of [2 e| /he, and whose elements were the differences of the row and column headings. These elements were thus all possible "predicted" values of iDH-Ef/hc consistent with the observed values for the other two splittings. By comparing these predicted values of |D+ E| /he with the observed values, it was pos sible to select unambiguously three "triples" of zero-field 96

splittings, since the magnitude of the uncertainties in the mean values of the splittings were well established and small (the average of the standard deviations of the •*6 —1 mean values of the zero-field splittings was 6 * 1 0 cm” : see Tables 2, 3 and 4, Column 12). Following the selection of the first three triples, assignment of the three addi tional sets was then unambiguous. (We have described a choice as unambiguous if no other possible choice lay nearer than 6 o to it.) . The resulting sets of triples are displayed in Table 3. No resonance absorption has yet been observed near the frequency corresponding to the difference of the remaining pair of |d - e| /he and ]2 e] /he values. It is difficult to explain this failure since other resonance absorptions have been found by this postu- lated-triples method. (An alternative interpretation of the data is to take only triples 3, 4, 5 and 7 in Table 5 as unique, and to assume double degeneracy for the

D - E value of Set 6 and the 2E value of Set 1, and triple degeneracy for the D +■ E value of Set 1. In this way nine triples are obtained with reasonable agreement between predicted and observed values. However, since signals —6 -1 separated by only 25 • 1 0 era have been resolved, the degeneracies would all have to be nearly exact. Further more, neither the observed relative signal amplitudes nor the measured linewidths of the resonances associated with 97

Table 5. Triples of Zero-Field Splittings for Chrysene Oriented in p-Terphenyl.

Signal |d -e | /he |D4-e | /he |2 e|/ he set (era-1) (cm"*) (era"1)

1 0.117006 0.063086 0.053926

2 0.116977 - 0.052624 3 0.116891 0.062901 0.053999 4 0.116639 0.062457 0.054195 5 0.116133 0.062700 0.053443

6 0.115697 0.062836 0.052874 7 0.115612 0.063298 0.052312 98

the indicated splittings support the idea that these splittings are doubly and triply degenerate, respectively.) Additional support for the assignment of triples made in Table 5 is found in the fact that the average agreement between predicted and experimental values of|D-f- E[/hc —6 1 (see Table 6 ) within triples, namely 9-10 era"1, is in excellent agreement with the mean uncertainty of the zero- —6 —1 field splittings themselves (as given above, 6 • 1 0 cm ). The overall average values of the three sets of -1 -1 splittings are, from Table 5, 0.1164 cm , 0.0631 cm ,

and 0.0533 cm"3, for |d - E|/hc, |d + E|/hc and [2e |/he, respectively. These splittings lead to average values of |D]/hc and ]E|/hc for chrysene-d-^ in p-terphenyl-h^ at 80.1 K of 0.0897 cm ^ and 0.0267 cm"’*', respectively. These values may be compared to those measured by Brinen 21 and Orloff for chrysene-h^ in 2-MTHF glasses (at an un specified temperature): these authors report 0.095 era"*- and 0.025 cm”** for |D|/hc and |E|/hc, respectively (no uncertainties are given for these values). In the present studies the mean linewidths and the mean magnitude of the deviations from the mean linewidths found for the component absorptions for each transition for chrysene-d]^ in p-terphenyl-h^ were 5.6 i 1.0 G for the transitions near 0.116 cm"**, 10.5 i 3.0 G for those near 0.063 cm"*', and 6.1 t 1.1 G for those near 0.053 cm"*'. 99

Table 6 . Predicted and Experimental Values of |d 4- e|/he for Chrysene-d^ 2 Oriented in p-Terphenyl.

Signal Predicted Observed set |D+E] / he |DfE| / he (cm***') (cm"*-)

1 0.063080 0.063086

2 0.064353 - 3 0.062892 0.062901 4 0.062444 0.062457 5 0.062690 0.062700

6 0.062823 0.062836 7 0.063300 0.063298 100

(No explanation of the substantially larger mean linewidth and standard deviation for the D4-E absorptions has been developed, and this result may be significant since in the case of OHA host the mean D 4- E linewidth agrees very well with those for D - E and 2 E absorptions indicating that the linewidths of D + E resonance absorptions for chrysene- ^12 are not intrinsically greater.) Average signal-to- noise ratios for a 2.3 second time constant were 150, 55 and 60 for the D - E, D + E and 2E resonance absorptions, respectively.

Results of triplet lifetime studies of chrysene-d^ 2 in p-terphenyl-h^ 4 are presented in Section V.E below.

3. Triplet Chrysene-di9 /in Single-Crystal p-Terphenyl-

ii4 14 precision determinations, comprising 60 measure ments of pairs of values , [hJ for resonance were made for 14 different zero-field splittings studied for chrysene-d^ 2 oriented in p-terphenyl-d^. These primary data and selected derived data are summarized in the entries denoted by (b) in Tables 2 and 4. It is readily apparent that only small intrinsic dif ferences exist between this case and the preceding case. Section V.E. below contains a discussion of these differences. 101

4. Triplet Chrysene-di9 in Single Crystals of.s- Octahydroanthracene Chrysene-d^ from the sample used to prepare the p-terphenyl-host mixed crystals, from which as discussed above a total of twenty resonance absorptions were ob served, manifested in mixed single-crystals with s-octahydroanthracene (OHA) as host only six resonance absorp tions, two resonance absorptions for each of the three low-field transitions. Data from precision determinations of the six corresponding zero-field splittings were ob tained and analyzed as described in the preceding sections. The results of the analyses are presented in Table 7. The assignment of the observed splittings to the zero- field energies D - E, D + E and 2E was made by analogy to the results in p-terphenyl, since the crystal structure of OHA is not available. (It is necessary to assume that the crystal field shift on going from p-terphenyl to OHA re- . suits in a slight increase in the splittings associated with D 4- E, and a slight decrease in the splittings associ ated with 2E and not in an interchange of the magnitudes of these two splittings. The crystal field shift neces sary to produce such an interchange of splitting values would be almost a factor of four larger than any crystal field effect previously observed, and in addition such a nearly exact reversal of values seems extremely unlikely.) Table 7* Magnetic Resonance Data for Chrysene-d^ Oriented in s-Octahydroanthrncene and In Biphenyl.

Run number Boat Crystal Designation No. of differ Range of mag Sign Beat values Standard Calculated Standard Beat values and signal teapernture of chrysene ent wj for netic field of of y for deviation Zeeman deviation of zero-field t*V) transition nuaber vbich jji over Which dvj_ fit of Eq. of y contribution of beat splittings was measured resonances (10) to to observed values of for fit of ~ , cm"1 were deter- data . a i k l x 10° splittings Z.F.S. Eq. (10) to stlned (0) * 107 at smallest x 10® data (as-i G~a) (ca*1 Q"2 ) Hi (as*1) («-M I * x 10® fan-i)

19a- 22 OHA 77.5 ~ O.II76 6 6.4-20.4 + 1.455 ' 1.7 5 < 1 0.117615 19b- 22 OHA 64.9 " O . U 76 5 4.5-19-9 ♦ 1.517 4.3 5 1 0.117691

19a- 23 OHA 77-5 ~ 0.1151 6 8.6-22.4 ■f 1.416 1.5 11 < 1 0.115075

20-24 OHA 77-5 ~ 0.0687 6 3 .4-23.8 - I .503 1.0 2 < 1 0.068695

20-25 OHA 77-5 - 0 .06&4 6 8.9-25-3 * 1.545 1.6 13 1 0.068381

+ < 1 21a- 26 OHA ~ 77 “ 0.0409 7 4-7-19.5 2.911 2*1 7 0.048917 21b- 26 aiA 66.5 “ 0.0409 4 7.6-20.4 + 3-193 8.8 19 5 0.040938

22a- 27 CHA ~ 87 - 0.0467 6 4.9-18.8 + 2.912 3-3 7 1 0.046706 22b- 27 CHA **77 *- 0.0467 4 12.4-22.6 + 1.629 2.4 25 l 0.046681

23 - 28 BIPHENTL 77-5 ~ 0.1182 6 7-1-22.0 + 1.343 1.0 7 < l O.UB 199

23-29 BIFHSOCL ~ 87 *» 0.1181 5 7.9-22.1 4- 1-391 3.5 9 1 0.118152

O (O 103

This assignment having been made, the association of split tings into triples was straightforward. Average values of the zero-field splittings and the spin Hamiltonian parame ters obtained for chrysene-d]_2 In OHA at ~77K are given in

Table 8 . (No temperature corrections have been applied to these data, which were obtained under several different "crystal-temperature" conditions.) The average linewidths found for the two absorptions observed for each transition were: for the 0.116 cra"^- transition, 5.4 G; for the 0.068 cra"^ transition, 5.4 G; and for the 0.048 cra~^ transition, 5.9 G. In experiments near 0.116 cm"^ and near 0.068 cm"-*- in which both resonances observed for each transition were studied under the same conditions, the signal-to-noise ratio for one signal in each transition was approximately five times as great as that for the other signal. (No data of this type were obtained for the third transition.) Upon combining the signals into triples as described above (i.e., on the basis of their zero-field splittings), it was found that the signals of larger magnitude occurred in one triple and the signals of smaller magnitude occurred in the other. Thus the zero-field splittings forming the first triple in Table 8 appear to be associated with a set of guest molecules occupying OHA sites which are approxi mately five times as densely populated as the sites Table 8 . Zero-Field Splittings and Spin Hamiltonian Parameters of Chrysene-d^ Oriented in s-Octahydroanthracene at~77K.

Signal Observed Observed Predicted Observed Set M l /he |D+E| /he |2E| /he N /he |d |/he 1E|/he (cm-1 ) 1 (cm-1 ) (cnrl) (cnrl) (cm“l) (cm-1 )

1 0.117615 0.068712 0.048903 0.048917 0.093163 0.024458

2 0.115074 0.068382 0.046692 0.046706 0.091778 0.023353 105 occupied by the set of guest molecules associated with the second triple of zero-field splittings.

5. Chrysene-di 2 in Single-Crystal Biphenyl Weak phosphorescence observed from mixed single crystals of chrysene-d^ 2 in biphenyl indicated that rela tively little chrysene-d^ 2 was incorporated in boules grown by the Bridgman method. Due in part to the resul tant low signal-to-noise ratios, this system has not been studied extensively. A single prominent resonance absorption has been ob served near 0.118 cra”i, and has been associated with the quantity |d - E|/hc. Precision determination of this splitting at 77K (the bandpass filter was employed), gave a value for |d - E)/hc of 0.118199 cm”*-. The width of this resonance absorption was observed to be 6.2 G, a value which is consistent with occupation of a definite host site. No resonance absorptions corresponding to |D -f*E|/hc or |2E|/hc were observed, presumably because of insufficient signal intensities. The substantial crystal-field shift in the value of |D - E| /he for chrysene-d^2 in biphenyl host from the aver age values of this quantity found for p-terphenyl-hi4 and OHA hosts is attributed to severe crowding of chrysene in the biphenyl sites. 106

D . MULTIPLE ORIENTATION OF CHRYSENE GUEST TRIPLETS IN P-TERPHENYL HOST A considerable number of experimental results can be cited in support of the conclusion that impurities are not responsible for the observation, at low magnetic fields, of multiple signals from mixed crystals of chrysene-d^ 2 P" terphenyl. Among these are the following: 1.) All resonance absorptions observed from such mixed crystals had linewidths typical of a perdeuterated species ("5 G). If impurities in the host material were responsible for any of the observed signals, those reso nance absorptions would be expected to have linewidths typical of a proto species (~ 20 G). 2.) Purification of the chrysene-dj^ sample, which resulted in the removal of at least four distinct impuri ties, failed to alter either the number of absorptions ob served in the raultiplet patterns or their relative ampli tudes. 3.) The zero-field splittings of the known major im purity in commercial chrysene, 2 ,3-benzocarbazole, were determined for p-terphenyl host and shown to be different

from all of the 2 0 zero-field splittings observed from chrysene-d^2 *-n p-terphenyl (see Section VI.B). 4.) The relative amplitudes of signals within a 107 raultiplet did not depend on nominal guest concentration over three orders of magnitude. Since it was shown by UV analysis that actual concentrations of chrysene in p-terphenyl mixed crystals were substantially less than nominal concentrations except at the lowest concentrations em ployed, the observed lack of dependence of relative signal amplitude on nominal concentration would require that all supposed impurities have solubility properties essentially identical to those of chrysene in p-terphenyl. 5.) Within a raultiplet pattern the magnetic resonance lifetimes were all comparable and the linewidths of the component absorptions were the same within uncertainties.

6 .) Results from experiments in which various UV transmission filters were employed showed that the rela tive amplitude of signals within a raultiplet was not sig nificantly affected by variation in the wavelength of the exciting light. In particular, for filters whose maxi- o muro transmission-wavelength was below ~3500A and which transmitted little light between ~3500A and ~3600A (the important absorption region for chrysene in p-terphenyl crystals), only weak signals were observed. Nonetheless all of the resonance absorptions observed in the D - E multiplet pattern for unfiltered excitation were present for filtered excitation and had the usual relative ampli tudes. For a bandpass filter (transmission peak at 108

o ~2800A) which transmitted essentially no light in the 3500 to 3600A region no resonance absorptions were observed. Thus all species giving rise to raultiplet signals must have had very similar if not identical singlet absorption bands. 7.) Twenty different resonance absorptions were ob served in the chrysene-dj^ raultiplet patterns for p-ter phenyl host whereas only six absorptions were observed for OHA host, despite the fact that both types of mixed crystal systems were prepared frora the same chrysene-d^ sample.

8 .) Since a typical impurity concentration in the (un purified) chrysene-d^ 2 sample would be ~1% (with the pos sible exception of 2 ,3-benzocarbazole discussed separately in item 3), it would be expected that for impurity life times and linewidths comparable to those of chrysene-d^ 2 the signal-to-noise ratio of a resonance absorption arising from an impurity would be approximately 1 / 1 0 0 of that for chrysene-d^2 * fact, the absolute magni tudes of signals observed in the multiplet patterns were all comparable to each other (within factors of 2 to 4) as shown in Figures 13 - 15. Furthermore, for the measured lifetimes and linewidths of the component signals, signal- to-noise ratios were consistent in each experiment with the nominal chrysene concentration of the crystal piece 109 employed in that experiment and not with a concentration

1 / 1 0 0 as great. 9.) Differences between zero-field splittings corres ponding to the various component signals within raultiplets were very much smaller than typical differences between zero-field splittings obtained for different molecules* Moreover, while it might be expected that one or even pos sibly two of the zero-field splittings of an impurity were almost exactly the same as the corresponding splittings for chrysene-dj^* the probability of all three splittings being in near-perfect agreement with those of chrysene-d^ 2 for six different impurities is (on the basis of compari sons of zero-field splittings obtained for all molecules studied to date) very small. On the basis of these nine factors it seems securely established that all of the twenty resonance absorptions observed in the multiplet patterns shown in Figures 13, 14 and 15 arose from a single chemical species, chrysene- d^2' This being so, consideration must next be given to the question of whether pairing (or higher-order associa tions) of chrysene molecules is responsible for multiplet structure. Again there are a number of evidences which contradict this type of explanation of the origin of multiple absorptions. One, cited above, is the lack of dependence of the patterns on guest dilution down to concentrations*as low at 10"^ mole fraction. The occur rence of significant associations of guest molecules during slow crystal growth in thermally equilibrated systems in which there are 1 0 0 , 0 0 0 times as many host molecules seems very unlikely. Moreover, for any substantial aggregation of chrysenes resulting in the formation of guest micro crystals, guest-guest triplet annihilation processes would surely be significant, and a reduction in apparent magnetic resonance lifetime would be expected even at very low temperatures but none is observed. (It should be noted in this connection that aggregations of chrysenes in crystal defects will not be observed because of the reduc tion in lifetime resulting from triplet-triplet annihila tion processes.) Finally, in experiments on similar mixed crystal systems with substantial concentrations of two kinds of guests [for example, naphthalene and phenanthrene in biphenyl;^- phenanthrene-h^Q and phenanthrene-diQ in biphenyl (see Chapter VII)], zero-field splittings for the two guests were shown not to differ from the splittings found for each guest alone in the same host structure. This result by extension supports a presumed absence of any substantial specific interaction or association of guest molecules in these dilute chrysene-d^ 2 *n p-terphenyl mixed crystals. Inasmuch as strong cases could be made against both Ill impurities and association of guest molecules as the ori gin of raultiplet structure in the lour-field magnetic reso nance spectrum of chrysene-d^ in p-terphenyl, alternative explanations were sought. Two aspects of the experimental data presented in the preceding section seemed particularly significant in this regard. First, the range of zero-field splittings found within multiplets was of the same magni tude as the shift in zero-field splittings observed for a given guest in different host structures. For example, for chrysene-d^ 2 *-n p-terphenyl the largest differences between zero-field splittings within multiplets were

14 ■ 10"^, 8 * 1 0 **^ and 19 * 10“^cm"^ for the patterns near ~0.116, 0.063 and 0.053 cra”^, respectively, while the largest and smallest shifts of zero-field splittings of triplet naphthalene upon going from durene host to bi phenyl host are ~30 • 10"^ and ~ 6 •* 10“^cm”^.^ The dif ferences observed in the zero-field splittings of naph thalene are attributed to the different molecular-crystal fields experienced by the guest triplet in the two host structures. Clearly, separations within a raultiplet can be of the same order of magnitude as the "structure shift" observed upon going from one host to another. A second important feature of the precision data pre sented in Tables 2-4 was that values of Yt calculated for

* best fits of Equation (10) to data obtained for the 112

components of a mu1 tiplet in a single experiment in which the static field orientation relative to the crystal piece was not changed (a "fixed” experiment), differed in some cases by many standard deviations of Y . Consideration of Equation (10) and of Relations (6 )-(8 ) from which it is formulated shows that the value of Y (the slope in a fit of precision data) depends not only upon the value of the zero-field splittings of the triplet state being studied but also upon the orientation of the principal fine-structure axes of the triplet state with respect to the direction of the static magnetic field. Thus, dif ferences in values of Y (determined for the components within a raultiplet in a "fixed" experiment) which are larger than can be accounted for by uncertainties or by the various slight differences in zero-field splittings for the components must be due to differences in orienta tion of the fine-structure axes of the respective triplet states with respect to the static field. For a "fixed" experiment, different orientations of the guest triplets with respect to the static field necessarily imply dif ferent orientations of the triplets relative to each other. Thus, a plausible interpretation of raultiplet structure based on the two experimental factors cited is that the component signals in the raultiplet patterns arise from sets of differently oriented chrysene guest triplets 113 which therefore occupy several kinds of sites of substitu tion at which there are significantly different molecular- crystal fields corresponding to the observed differences in the zero-field splittings of the component signals. For the sake of clarity in the discussion which fol lows, the concept of sets of guest triplets which differ in orientation from each other by small rotations about one or more principal fine structure axes will be described by the term "multiple orientation" (of guest triplets). Although it is recognized that such differently oriented triplets would be translationally inequivalent with re spect to each other, the latter term will be reserved to describe the fact that the p-terphenyl crystal structure requires in general that there be two translationally in equivalent sets of any guest triplet per unit cell, whose mean molecular planes are separated by a substantial di hedral angle and whose long in-plane axes are essentially parallel. Thus, if there are sets of multiply oriented guest triplets in p-terphenyl there must be two transla tionally inequivalent sets of such sets per unit cell. Four experimental and calculational approaches were used in attempting to obtain a quantitative as well as qualitative understanding of the orientations of chrysene triplet states in the p-terphenyl host structure; 114

1.) Expressions for the slopes ( y) of fits of data sets to Equation (10) were obtained by taking appropriate differences of Relations (6 )-(8 ) and differentiating the resulting equations for the energies corresponding to reso nance absorption, with respect to |H]^ ( G 2 /g2 0 2). Thus we have (still in terms of energies rather than reciprocal centimeters):

, l D -l- 3E 02 D -4-3E 9 y°-E “ W + 2E(D+E) 2E(D-E) m *

y' _ 1 D-3E p2 D-3E 2 D+E 2E “ 2E(D+E) ” 2E(D-E) m *

1 D p D 2 2E E + E(D+E) E(D-E) “ ’ where Y* ~ Y he, / = cosine of angle between H and the A fine-structure axis, m rs cosine of angle between H and the B fine-stimeture axis and D/E < 0 and E is taken to be negative.

The data cited for Run 8 a,b in Table 3 were obtained in an experiment in which the static magnetic field direc tion was approximately aligned with the yz plane of p-terphenyl. In this "fixed" experiment neither the crystal position nor the static field orientation was changed during the time precision data were being recorded. Data were obtained both above and below the zero-field frequency 115 for the two absorptions arising from the two translationally inequivalent sets of guest triplets for each of Signals 8 , 9, 10 and 12 in the D -4- E multiplet pattern. It was calcu lated from Equation (12) that in a D 4 E experiment for chrysene in which H is exactly in the BC plane (HRp || A; h -La, hence ? = 0 ), resonance absorption will be observed for microwave frequencies above the zero-field frequency (positive y') if the angle between H and B is less than 34 and below the zero-field frequency (negative Y ) if the angle between H and C is less than 56°. Thus for the crystal orientation in Run 8 a,b, one translationally in equivalent set of chrysene triplets must have had B ap proximately aligned with H while for the other set C was more nearly aligned with H. Rewriting Equation (12) for the two translationally inequivalent sets of guests, we have for the slopes obtained in such an experiment:

y (+) = a + b + cmj^ , (14)

y'(-) = a + b / 2 2 + c m 2 2 , (15) where a = 1/2E, b = - ( 2E (D4-E) and ra-^ and ra2 are t^le direction cosines for the translationally inequivalent sets of triplets. We further have relations 1X6

^ 2 4- “f- n i^ = ^

/ 2 -h m22 + n22 = 1, (17) where n = cosine of angle between H and the C fine-structure axis, and it is assumed that the A fine-structure axes for the two translationally inequivalent sets are parallel (hence J?2 = ii)* Since the values of y*(-)> b and c were determined experimentally for each signal inves tigated in Run 8 a,b, there were five unknown values in the four Equations (14) - (17). A fifth equation in these unknowns was obtained in terras of the dihedral angle $ between the B (or C) fine-structure axes of the two trans lationally inequivalent sets, namely, ra2 ■= m^ cos

B^ and B 2 for each triplet state corresponding to these signals* These angles are shown in Table 9 for a number of dihedral angles between 50 and 75*. Theuncertainties In the angles shown range between~ 0.3 and ~1.0 for a 117

Table 9. Angles Between the Projection of H in the BC Plane and Bj. and B 2 for Triplet States Corres ponding to Signals 8, 9, 10 and 12 in Run 8 a,b.

Dihedral Angle Signal Number H proj- B^ H proj- B2 (degrees) (degrees) (degrees)

50 8 34.2 84.2 9 39.1 89.1 10 44.2 94.2 12 34.8 84.8 55 8 34.6 90.4 9 38.9 94.9 10 43.5 98.5 12 35.1 90.1 60 8 34.4 94.6 9 38.3 98.7 10 42.5 102.6 12 34.8 94.8 68 8 33.1 101.1 9 36.6 104.6 10 40.3 108.3 12 33.5 101.5 75 8 31.4 106.4 9 34.6 109.6 10 38.0 113.0 12 31.8 106.8 118 one sigma variation in opposite directions (worst case) of the slope values obtained from Table 3. Thus, the data given in Table 9 indicate significant differences (rela tive to uncertainties) in the orientation of the B (or C) axes of at least three of the sets of triplet states giving rise to component signals in the D + E multiplet. Specifi cally, these differences take the form of small rotations relative to one another about the long in-plane fine-struc ture axes (A axes) of the chrysene triplet states. No sig nificant differences were found in the calculated orienta tions (relative to H) of the A fine-structure axes of the sets of triplet states corresponding to Signals 8, 9, 10 and 12. For example, for a dihedral angle of 55* the cal culated angles between the static field and the A axes corresponding to these four signals were 58.6, 59.4, 58.5 and 58.8 degrees, respectively. Further discussion of these results is given below, but it is of interest to note at this point that the differences in angles for the four signals are quite insensitive to the dihedral angle chosen. For example, the maximum difference (Signal (10) - O O Signal (8) in or B 2 angles ranges only from 6.6 to 9.9 for dihedral angles between 75* and 50*, respectively. 2.) In certain 2E experiments (Runs 17a,b; 18a,b), chrysene-d^ 2 p-terphenyl single-crystals were oriented © with the primary cleavage plane on the face of a 16 wedge 119 and the b-axis striations parallel to the front edge of the wedge. For this orientation the plane of rotation of the static magnetic field was expected to be essentially coincident with the yz plane of p-terphenyl. In experi ments of this kind, in general, two resonance absorptions were observed for frequencies above the zero-field fre quency and at fields above ~10 G. The two signals, desig nated "down-field" and "up-field" absorptions, arose from the two translationally inequivalent sets of chrysene triplets and were resolved when H — |Jb of one translatio nally inequivalent set while H — f|c for the other, since in this case the values of Y for the two signals were quite different (see Table 4). (Since they,corresponded to the same zero-field splitting the two absorptions merged at fields below ~10 G. The signals necessarily merged also when H was approximately parallel to the bisector of the dihedral angle between B^ and B2 . This fact was utilized to establish that, as expected, the bisector of the dihedral angle between B^ and B2 guest axes corresponded, at least to within a few degrees (and probably exactly), to the bi sector of the dihedral angle between y^ and y 2 host axes.) For H very nearly aligned with the B axes of one trans lationally inequivalent set (the corresponding angle reading on the magnet base was recorded), precision data were obtained for both resonance absorptions for each of 120

Signals 14-18. From the resulting values of Y* for the "up-field" absorption it was possible to calculate self- consistent values of I ^ and m^ using Equation (13) and by assuming that any misalignment of H from was in the AiB^ plane. The angles which H made with the three fine- structure axes were then obtained directly from ra-^ and n^. Again assuming that H t*le direction cosine m2 was obtained from the value of y' for the "down-field" absorption using Equation (13). The angles which H made with B-^ and B2 for the sets of two translationally inequi valent triplet chrysenes corresponding to Signals 14-18 in Run 18a,b are given in Table 10. The angles between H and

B 2 were very nearly equal to the dihedral angle between

B]_ and B2 , since the (average) misalignment of the BC 1 planes for chrysene triplets from the rotation plane of the magnet was indeed quite small (~6°) in this experi ment according to the values of angles between H and B^ in Table 10. It should be noted that since values of the angles between H and B^ were calculated on the assumption that H was in the A^B-^ plane, *r/2 minus these angles are then the angles between H and A^. Thus, subject to the assumption cited, significantly different orientations of the A fine-structure axes are indicated for triplets -cor responding to Signals 15 and 18 by the data of Table 10. The results obtained from Run 18a,b suggest that the 121

Table 10. Angles Between H and the B Fine Structure Axes of the Two TranSlationally Inequivalent Triplets Corresponding to Each of Signals 14 - 18 in Run 18 a,b.

Signal Magnet Number Base Angles between H and Angle B1 b2 (degrees) (degrees)

14 -27 8.0 55.8 15 -27 0.8 55.5 16 -27 5.7 51.2 17 —28 4.8 58.0 17 +28.5 61.1 7.8 18 -27 16.0 60.0 122 dihedral angle between the mean molecular planes of translationally inequivalent triplets varies over ~10 with a mean value of ~57°. Since the dihedral angle between the mean molecular planes of translationally inequivalent p- terphenyl molecules is 69*, these data indicate a significant average misalignment (~6 ) of the chrysene B,C fine-structure axes from the p-terphenyl y,z molecular axes in addition to an approximately five degree spread in the orientation of the B (or C) fine-structure axes for the component triplets relative to each other within each of the two translationally inequivalent sets* 3.) A "direct" method of demonstrating multiple orientations of chrysene triplet states was based on es tablishing that the static field orientation corresponding to closest approach of H to B (in a 2E experiment similar to that described for method 2) differed for various com ponent signals in the 2E multiplet pattern. With crys- 0 tal pieces mounted on a 16 wedge so that H rotated very nearly in the yz plane of *p-terphenyl, the static field was swept until the chart recorder pen was near the center of the "up-field" resonance absorption. For a fixed fre quency, as the static, field was rotated by one degree inter vals through the angle for closest approach of H to a B axis, the absorption moved "up and down field" as the field for resonance increased to a maximum (corresponding to 123 closest approach) and then decreased. When the first deri vative of the absorption is displayed, an absorption moving up or down in field corresponds to a left-right movement of the recorder pen and by allowing the chart recorder to run continuously, a peak was traced as H was rotated. The magnet base angle was marked on the peak trace at one de gree intervals, each angle being held for ~ 30 seconds to establish clearly the mean pen position. The angle cor responding to the maximum in such peaks for each signal was associated with closest alignment of H to the B fine- structure axes of chrysene triplets giving rise to that signal. Several H-rotation peaks (for alternately in creasing and decreasing angles) were taken for each of Signals 15-20, so that any systematic error due to base line drift or hysteresis in pen position with angle would be minimized. Average values and standard deviations of the magnet base angles corresponding to closest approach of H with B obtained (for triplet states in one transla tionally inequivalent set) in two separate experiments of this kind are presented in Table 11. Again, significant differences (with respect to uncertainties) between B fine-structure axis orientation are found for several of the six sets of triplet states giving rise to the compo nent signals in the 2S.multiplet pattern. The agreement between differences in magnet base angles found for the 124

Table 11. Magnet Base Angles Corresponding to Closest Approach of H to B Fine-structure Axes of Component Triplet States in the ~ 0.053 era"*- Mu1tiplet.

Angles (degrees) Signal Run I. Run II. Number

15 16.0 0.8 154.8 1.0 16 17.3 3.1 153.0 0.4 17 15.0 0.4 156.8 0.8 18 19.8 1.0 151.5 0.4 19 18.8 0.6 151.0 0.0 20 17.8 0.6 152.8 0.3 125 different signals for Run I and Run II was quite good within the uncertainties given. 4.) As discussed in Section V.5, resonance absorptions which arise from triplet states whose zero-field splittings differ only very slightly from each other will be most re solved at the lowest fields and will merge at high fields. However, if two such triplet states are translationally inequivalent from each other then in general at sufficient ly high fields their corresponding resonance absorptions must once again separate and continue to separate with Increasing field. This is a direct consequence of the fact that the slopes of the 1/ vs H (high-field) functions for the two translationally inequivalent triplets will be different. Thus, results from high-field experiments for chrysene-d-^2 p-terphenyl, especially with respect to the number of signals observed for H along the three principal fine-structure axes, are also relevant (at least qualitatively) in a discussion of possible multiple orien tations of guest triplets. As shown in Figure 16 only a single resonance absorp tion was observed at ~2200 G for H very nearly aligned with the A fine-structure axes of one translationally in equivalent set of chrysene triplets. In contrast, in ex periments at ~3100 G in which H rotated very nearly In the yz plane of p-terphenyl and hence very nearly In the BC 126 plane of triplet chrysenes, up to six signals were observed in raultiplet patterns from one translationally inequivalent set of (multiply oriented?) chrysenes. The number of re solved signals observed and their separations depended sensitively on the static field orientation. For example, for one static field orientation four resolved absorp tions of comparable amplitude and widths of 5 to 6 G were observed at an average separation o f — 60 G, while for another orientation two pairs of merged signals were ob served — 125 G apart. The observation at high fields of a pattern of absorp tions for H in the BC plane is consistent with a model of sets of chrysene guest triplets whose mean molecular planes are not parallel and which in particular differ in orientation by small rotations about the A axes. The ob servation of only a single resonance absorption for a D-E transition at — 2200 G for H very nearly parallel to A axes of chrysene triplets suggests that the A axes of the dif ferent sets of chrysenes corresponding to component sig nals cannot be greatly misaligned from each other. How ever, due to the comparative insensitivity of Y to varia tion in direction cosine values in the case of D-E experi ments for chrysene, small misalignments of A axes cannot be ruled out on the basis of this observation. In each case, results obtained by the four methods 127 described above indicate that the sets of chrysene guest triplets giving rise to the different component signals in multiplet patterns differ in orientation from each other by at least small rotations about the principal fine struc ture axes. (Magnetic resonance experiments give no quanti tative information regarding translational differences which may occur between different sets of chrysene trip lets in the p-terphenyl host structure, although they can in principle indicate that such differences exist.) All of the angular data obtained by these methods re fer of course to the principal fine-structure axes of the chrysene triplet state. However, there is certainly no basis on which to believe that the guest-host intermolecular interactions result in multiple orientations of the fine-struoture axes for a unique molecular orientation. On the contrary it is much more reasonable to presume that the indicated differences in triplet state orienta tions are a manifestation of differently oriented chrysene molecules in this host structure. Thus, differences measured in fixed experiments between angles between H and fine-structure axes are taken to correspond to dif ferences in molecular orientation of the various sets of chrysene guests which give rise to raultiplet structure in low-fleld magnetic resonance spectra. A comparison of differences in B-axis angles of 128

orientation determined by Methods 1, 2 and 3 and tabulated in Tables 9, 10 and 11, respectively, is made in Table 12. Comparison of results obtained in D + E and 2E experiments was based on the association of zero-field splittings (or Signals corresponding to them) into triples as discussed in Section V.C. Reasonably good agreement (relative to estimated combined uncertainties of — 2* to ~ 3 #) is found between angle differences obtained in the several experi ments, except in Comparisons 3 and 5. No satisfactory ex planation of the discrepancies in these comparisons has been found apart from the possibility that one or more assumptions made in the analyses of the experimental data are invalid. A similar case of overall consistency but with discrepancy of a few results occurs with respect to A axis orientation data. The data obtained indicate preponderantly that the long in-plane axes of sets of chrysene triplets are essentially parallel for all sets and that the dominant character of multiple orientation is rotation by various amounts about the long in-plane axes. Notable exceptions to this, however, are the angles found between H and A for Signals 14 and 18 (Table 10) by Method 2. Due to the magnitudes of uncertainties relative to the small angular differences being investigated, and the dis agreement in a few cases between angle values found by 129

Table 12. Comparison of (Absolute Values of) Differences Among Angles Between H and B Fine-structure Axes Given in Tables 9, 10“and 11 for the Various Multiplet Signals.

Comparison Signals for Number which angle Angle Differences differences Table 9 Table 10 Table llb are compared3 (degrees)

1 2 0 (8)-18 (1 2 ) 0.5 ----- 2.0,1.3

2 20 (8)-17 (9) 4.3 2 .8 ,4.0

3 2 0 (8)-16 (1 0 ) 8.9 0.3,0 . 2

4 18(12)-17(9) 3.8 1.5C 4.8 ,5.3 5 18(12)-16(10) 8.4 4.4 2.3,3.8

6 18 - 15 ---- 2.3 3.8 ,3.3

7 17(9)-16(10) 4.6 2.9C 2 .5,3. 8

8 1 6 - 1 5 ---- 1.5 1.5,1.8 aD + E signal numbers corresponding to the 2E signal num bers given are shown in parentheses to facilitate reference to the original data in Tables 9-11. ^Values given in this column were obtained from data of Runs I and II, respectively. cThe angle between H and B 2 for Signal 17 in Table 10 was corrected by one degree (to 57.0°) to account for the one degree difference in magnet base angle for this measure ment. 130

different methods, the results presented here are consi dered tp be only semi-quantitative. It presently appears that only by a major research effort, probably involving the development of new microwave cavity and header designs (to permit rotation of a single-crystal sample about three normal coordinate axes from outside the dewar with

~ 0 . 1 degree precision), will it be possible to obtain a substantially more precise and complete determination of the small angular differences indicated between sets of chrysene guests* Although it is by no means certain that significantly more quantitative information can be gained in this manner by low-field electron paramagnetic resonance spectroscopy, there does not appear to be any other method which can furnish even as much detailed information as is presented here,

E. LIFETIME OF THE LOWEST TRIPLET STATE OF CHRYSENE-Pi? 1. Single-Crystal Hosts a * P-Terphenyl-h^/ The lifetime of the lowest triplet state of

chrysene-d^ 2 in p-terphenyl-h^ has been studied by mag netic resonance methods as a function of temperature from ~ 64 to ~90 K by varying the pressure above liquid oxygen and liquid nitrogen baths in which the resonant cavity was

immersed. A bandpass filter (Sec. III.A.6 ) was used for 131 all lifetime'measurements so that the crystal temperature was at most 0.3 K above the bath temperature. The bulk of the decay data were taken for the four well-resolved sig nals in the ~ 0.116 cm”** raultiplet (Numbers 3-6, Table 2). The data were taken from high-speed stripchart recordings of an extremum in the first derivative of the resonance absorptions obtained before and following shuttering of the A-H6 source. Triplet resonance decays for chrysene-d^ in p-terphenyl-h-j^ for several different temperatures are shown in Figure 18. Except at the two highest temperatures all decays were exponential for points between 0 . 9 and 0.25 of the initial signal amplitude. From these data triplet lifetimes ( Te) were calculated by a least-squares method (see computer program listing in Appendix B). (For decays obtained at 90.4 and 85.4 K, fits were made of data taken between 0.6 and 0.2 of the initial signal amplitude.) The mean lifetimes obtained (from several determinations) for each of the four signals at each temperature are shown in Table 13 and are plotted in Figure 19. The average uncer tainty in the 32 mean values of T q was 0.3 seconds. The temperature dependence of (guest) triplet life times in organic mixed crystals has been extensively studied for a variety of systems.^4-67 ^ag ^,een pro. posed that the dominant cause of the temperature dependence Figure 18. Magnetic resonance decays for chrysene-d]^ chrysene-d]^ for decays resonance Magnetic 18. Figure

Log (I/Iq ) 0 90.3 K 90.3 in p-terphenyl-h^. in 2 85.2 K 85.2 Time 4 2.6K 6 . 82 (sec.) 6 • 8 K 64.1 n 132 10 133

Table 13. Lifetime of the Lowest Triplet State of Chrysene- d12 ln p-Terphenyl-hy,. from ~64 K to ~ 90K. Data were obtained for the four well-resolved signals (Numbers 3-6, Table 2) in the 0.116 cm" 1 multiplet.

Triplet Lifetime ( Te) (seconds) Signal No. 3 4 5 6 Temperature(K) "

90.4 0.9 0.9 1 . 1 1 . 0

85.4 2.7 2 . 6 2.4 2.5 82.8 3.8 3.4 3.6 3.1 80.0 4.2 4.9 5.3 4.7

77.4 6 . 0 5.7 6 . 8 6.7

75.0 6.3 6 . 8 8.4 7.6

69.7 7.2 8 . 8 10.5 9.8

64.2 8 . 1 9.2 1 1 . 0 1 0 . 0 134

Figure 19. Triplet lifetime as a function of temperature

for chrysene-d^ 2 1 ° slngle-rcrystal hosts: closed circles are for p-terphenyl-h^ host open circles are for s-octahydroanthracene host.

% P H O S P H O R E S C E N C E LIFETIME (sec) 0 40 80 (K) E R U T A R E P M E T 120

160

200

240

230

320 135 136- in dilute mixed crystals for which Ae

AET/k . 6 5 , 6 7 Since the limiting lifetimes for Signals 3-6 could not be measured experimentally (due to the unavailability of a helium-temperature cryostat), predicted values of T were calculated from the data in hand by least-squares fitting ln(l/r - 1 /T°) vs 1/T, treating f ° as an adjus table parameter. (A listing of the limiting-lifetime cal- O culation program is given in Appendix B.) 7 was stepped by small increments from a value equal to the longest ex perimentally determined lifetime (in each Set, 3-6) to sufficiently large values such that the standard devia tion of the experimental values from the best fits was minimized (the linear correlation coefficient was maximized

n simultaneously). The value of T corresponding to this * minimization was taken to be the best estimate of the limiting lifetime for each signal and was 8 .2 , 9.3, 11.1 and 10.1 seconds for Signals 3 through 6 , respectively. (In a repeat calculation in which lifetimes for the two lowest temperatures were deleted, values of T 137

corresponding to minimization of standard deviations were found to be 7.5, 8.5, 11.5 and 10.4 seconds for Signals 3-6, respectively. Comparison of these values with those obtained in the first calculation indicates that the limi ting lifetimes calculated for the complete data are some what too small but that the differences between them are significant.) If the predicted differences in limiting lifetimes for the four signals are substantiated by ex periment, they can be attributed to the different guest- host interactions experienced by the four sets of chrysene- d.12 molecules giving rise to Signals 3-6. This would constitute an interesting example of the documented sensi- fi S Afi tivity of triplet lifetimes to host environment. J>DO From least-squares fits of the data in Table 13, and using the limiting lifetimes given above, values of AE^, were calculated to be 910, 1021, 1073 and 1095 era"* for Signals 3-6, respectively; the average uncertainty in AE^ was 40 cra“l. These values may be compared with values for the difference in the spectroscopically determined S ’triplet energies AE^t of chrysene-d^ and p-terphenyl-h^

which range from 600 cra"l to 800 cra“l .69 T ^ e differen ces between values of AE,p found for the four sets of chrysene-di2 molecules corresponding to Signals 3-6 are ucomparable to shifts observed in the triplet energy of a given guest in different host structures. (For example, 138

the triplet energy of naphthalene-dg guest goes from 21,180 cra~l to 21,380 cm"! from durene host to benzophe- none host. ) Thus, the calculated differences (larger than uncertainties) in AE^ can be interpreted in terms of differences in the triplet energies of the sets of chrysene-

d -^2 guests giving rise to Signals 3-6.

* Clearly the data obtained from the magnetic resonance lifetime studies of chrysene-d^ i-n p-terphenyl are con sistent with the phenomenology of orientational inequiva lences and corresponding differences in molecular crystal fields for the sets of chrysene-d^ guests which give rise to multiplet structure in low-field spectra, b. s-Oc tahydroanthrac ene The temperature dependence of the magnetic resonance lifetime of (the larger resonance absorption by)

chrysene-d^ 2 oriented in OHA was studied in the tempera ture range from ~77 to ~ 250 K, by methods described in

Section III.A.6 . The mean values found for e“l times

from exponential decays at 1 2 temperatures are plotted in Figure 19. The average uncertainty in the mean lifetimes was 1.1 seconds. From Figure 19 the limiting low-temperature lifetime for triplet chrysene-d^ 2 ln OHA is approxi mately 11.7 seconds. The values plotted in Figure 19 show a scatter which appears to be characteristic of data c from mixed crystal systems for which AE-j. (spectroscopic) 139

> 5000 cra**-*-,^"^ of which chrysene-d^ in OHA is an example. The principal cause of the temperature dependence of the triplet lifetime in such systems has not yet been es tablished unambiguously. Temperature dependence of the radiationless transition rate^^ and thermally activated processes involving impurities and crystal defects 67 * 6ft have been invoked as causes. Some previous workers despite admitting that thermal activation from the guest triplet to the host triplet is unlikely to be an important process in accounting for the observed temperature depen dences of triplet lifetimes in most systems for which AE^ S > 5000 cm -1 , have nevertheless evaluated AE^ for such systems in the manner described above for systems in which A E ^ « 5000 cm"-*-. For the most part the values of AE^ so obtained (which are always much smaller c than AE^ ) appear to be erratic and largely meaningless; The inconsistency in these values (in so far as correspon dence of with a "mean activation temperature" is con cerned) stems at least in part from the (varying) gradual rates of decrease of lifetime with increasing temperature found for many of the Ae^, S > 5000 cra“A 1 systems, including chrysene-d^ 2 OHA. These contrast with the steep lifetime-temperature functions observed for systems in which AE^ S « 5000 cm —1A and which can successfully be described 140 by a two energy level (E^H* E^q ) Boltzmann scheme. (For example, compare the temperature-lifetime functions in Figure 19.) The form of the lifetirae-temperature functions c obtained for many mixed crystal systems for which AE,j, > 5000 cm"^ suggests that many energy levels intermediate to the lowest guest and host triplet states may partici pate in the thermal depopulation (by annihilative processes following triplet energy transfer) of the guest triplet state. These multiple levels might include, in addition to impurity and defect bands, weak states which arise from mixing of the guest singlet and triplet states, c. Biphenyl Decays of the prominent resonance absorption observed for chrysene-d^ in biphenyl near ~ 0.118 cm"** were recorded at the boiling point of N 2 using a bandpass filter. The decays were found to be exponential with a mean lifetime of 11,4+ 1 seconds.

2. Chrysene-d^o in EPA Glasses Decays of the phosphorescent emission of chrysene- d ^ 2 in Hartmann-Leddon high-purity EPA glass at 77 K were obtained at three wavelengths (4964, 5400 and 5840A), using the Turner Spectro 210 with a high-speed stripchart recorder. Decays obtained at 4964 and 5400A were exponen tial and had mean lifetimes of 14.0 sec. (with an average 141

standard deviation of 0 . 1 sec.) and 14.3 sec. (with an average standard deviation of 0 . 2 sec.), respectively. O Decays recorded at 5840A were exponential during the first e"^ time and gave a mean lifetime of 17.1 sec. with an average standard deviation of 0.9 sec. The substantially o longer lifetime measured at 5840A can be attributed to the fact that the triplet-triplet absorption band maximum o 7Q yi for chrysene occurs at 5835A. * (In studies of several other molecules in glasses of approximately the same con centration as used in these experiments, Brinen and Hodgson showed that longer lifetimes are obtained in phosphorescence measurements at wavelengths for which the triplet emission and absorption bands overlap strongly and explained this on the basis of triplet-triplet reabsorp tion. ^) In the only report we have found of the phosphorescent lifetime of chrysene-d^2 > Kellogg and Wyeth give the value 73 12.3 sec. for chrysene-di2 *-n polymethylmethacrylate.

F. TEMPERATURE DEPENDENCE OF THE ZFS PARAMETERS OF CHRY-

SENE-D1 2 IN P-TERPHENYL-H|/t AND IN S-OCTAHYDROANTHRACENB Due to the decrease in the triplet lifetime of chrysene- d ^ 2 ^ p-terphenyl to less than ~0.5 sec. at ~95K, low- field magnetic resonance studies above this temperature were not feasible, and for the present study facilities 142 for experiments substantially below the boiling point of liquid nitrogen were not available. While studies of the temperature dependence of the zero-field splittings of chrysene-d^ 2 p-terphenyl over a broad range were thus not possible, experimentation was carried out in the acces sible temperature range.

The seven values of |d - e|/he for chrysene-d^ 2 p-terphenyl-^ were measured at the boiling point of oxy gen, for which (since a bandpass filter was used) the crystal temperature was taken to be 90.5 K. The values so obtained were subtracted from the mean values of the splittings determined at an average crystal temperature of 80.1 K (CuSO^ filter). The mean difference in the splittings at the two temperatures was 36 • 10"^ dl 7 • lO'^cm"^ K“^. Assuming a linear temperature dependence over this ten degree interval, these values gave as an average temperature coefficient:

d lp " ,EI ^ = -(3.5±0.7) * lO^cra" 1 K*1. (19) dT In early experiments conducted at the boiling point of liquid nitrogen, zero-field splittings were determined with the image of the arc of the A-H 6 lamp focused below the crystal (off-axis focus), for which the crystal tem perature was taken to be that of the bath, 77.2K. When the |D — e|/he values obtained in this manner were 143 compared to the mean values measured at 80.1 K, an average temperature coefficient of (-4. lit 2 .1 ) • 1 0 "^cm"'^ K"^ was found for |d - e] /he at ~77K. Thus the values of the [D - e|/he coefficient obtained by the two methods are in good agreement.

By comparing |d e | /he and |2e | /he splitting values determined for off-axis focus (77.2K) with mean values of these splittings measured at 80.IK, mean temperature coef ficients for these splittings were found to be

d jp + E| /he _ -1 .4 . l O ^ c m " 1 K " 1 (20) dT and

d l2E^ /hc s -2.4 • l O ^cm - 1 K-l. (21) dT Note that the sum of Coefficients (20) and (21) should equal the value for Coefficient (19) and does so within the experimental uncertainty. Coefficients (19) - (21) were used to apply tempera ture corrections to appropriate sets of the data in Tables 2, 3 and 4. It should be emphasized that the corrections for the zero-field splitting data taken at 7/K and 80K are all approximately equal to, or smaller than, the average standard deviation in the mean values of the splittings.

In the case of the substantial correction for the |d - e | / he data taken at 90.5K, the temperature coefficient is 144 well determined and the average uncertainty which can be introduced by the correction is 7 • 10 -6 cm "1 . Clearly these corrections can in no way affect either the numeri cal correlations or the interpretations in Section V.C. Finally, from the mean temperature coefficients for the zero-field splittings (19) - (21), we have as the mean coefficients for the spin-Hamiltonian parameters for chrysene-d^ 2 in p-terphenyl at 77K,

6 d|D |/hc as -2.3 O o ,• in"10 em-l cm" 1 K-lK“l (22) dT and

field splittings (due primarily to temperature variation

in the course of a splitting determination) is 2 • 1 0 cm L (which, however, is less than %% of the total change in the splittings with temperature in the range of interest). The values of |d - E[/he and |d 4- E|/hc for chrysene- d ^ 2 In OHA are plotted as a function of temperature in Figure 20. The data were fit to polynomial expansions in T to give, for temperatures between ~ 6 5 K and ~ 250K,

|D - e| /he 0.118082 - 5.5772 -10 " 6 T - 4.4631 • 10~9 T 2 - 3.3740 • lO^VcnT1, (24)

|D 4 e|/he = 0.069011 - 3.9374 * 10"6T - 1.8414 * 10~9 T 2

- 3.5154 * 10”1 1 T 3 cm"^, (25)

from which we have

|2E|/hc = 0.049071 - 1.6398 • 10"6T - 2.6217 • 10"9 T 2

41.414 * 1 0 "^2 T 3 cm~l. (26)

The standard deviations of the points from the best- fit functions were 6 . 6 • 10"^ cm"*- for |d - E|/hc and 5.7 • 10“^ cm"* for |d 4 Ej/hc. The expansions were trun cated at the T^ term since T^ terms contributed less than one average mean standard deviation to the zero-field splittings at the highest temperature. From the zero- field expansions we obtain for the temperature dependence 146

Figure 20. Zero-field splittings of chrysene-d^ in s-octahydroanthracene as a function of temperature. 147

0.069000 0.117500 .

0.068500 0.117000

D + E I / h c I D*-E 1/ he

(cm"1) (cm*1)

0.1165000.068000

0.067500 0.116000

0.067200 0.115700 50100 150 200 250 300 TEMPERATURE (*K) 148 of the spin Hamiltonian parameters for chrysene-d^ in OHA from ~ 65K to ~ 250K,

|d|/he = 0.093547 - 4.7573 • 10"6T - 3.1523 • 10”9 T 2

- 3.4447 • 10**1 1 T3, (27)

J e| /he = 0.024535 - 0.8199 • 1 0 "6T - 1.3109 • 10“9 T 2

+ 0.707 * 10"1 2 T3. (28)

The temperature coefficients were evaluated at 77.2K from Equations (24) - (28) and are tabulated with values of the coefficients for chrysene-d^ in p-terphenyl-h^ in Table 14. The substantial differences (compared with uncertainties) of the values for the two hosts indicate that the temperature coefficients are not determined by intramolecular properties alone, which is quite plausible but had not been shown previously. The temperature coeffi cients for chrysene-di2 at ~77K are also of interest in that they are considerably smaller (for both hosts) than the temperature coefficients measured for other mole cules. For example, values reported for —*^ ■" at '-77K are for naphthalene in biphenyl,'*' -13 * 10"**, for triphenylene in dodecahydrotriphenylene, ^ 2 -19 • 1 0 "**, for 1,3,5-triphenylbenzene in trimethylborazole, 317 ~ + 90 • 10"**, and for coronene in Lucite,2** ~20 • 10"**cm"^ Table 14. Temperature Coefficients of Zero-Field Splittings and Spin Hamiltonian Parameters for Chrysene-di2 in p-Terphenyl and s-Octahydroanthracene.

d(|D-E|/hc)/dT d( |D+E| /he)/dT d( |2E|/hc)/dT d( |D| /he)/dT d( 1e | /he)/dT Host

p-Terphenyl 3.5 1.4 2.4 2.3 1 . 2

OHA 6.9 4.9 2 . 0 5.9 1 . 0 149 150 *

The data presented here for the variation with tempe rature of the zero-field splitting parameters of chrysene- d ^ 2 oriented in OHA result from the first precision study of these functions over an extended temperature range re ported for any molecule* While some progress has been made in theoretical treatments of the temperature depen dence of the splitting parameters of (nearly) three-fold symmetric molecules,^4,75 theoretical treatments for less symmetric molecules are not available. The chrysene-d]^"* OHA data should provide a good test of any forthcoming theoretical predictions of the temperature dependence of D and E for oriented chrysene-d^*

G. DEUTERATION EFFECTS 1. Deuteration of Chrysene As discussed above, the reduction of absorption linewidths (by approximately a factor of four) by deutera tion of chrysene-h^ 2 was essential for quantitative study of multiplet structure in the low-field magnetic resonance spectra of chrysene-d^ In p-terphenyl. Conversely, com paratively few data have been obtained for chrysene-h^ 2 in p-terphenyl because of the overlap of signals within a multiplet. However, three distinct zero-field splittings have been determined for resonance absorptions at ^0.116 cm“^ at ~ 80K (no temperature corrections have been 151 applied to the chrysene-h^ data but either "off-axis focus" or a CuSO^ filter was used in all cases). The three values are shown in Table 15 with what are taken to be the corresponding splittings for perdeuterochrysene in p-terphenyl at 80.IK (taken from Table 5). A compari son of these splittings must be made with reservation since at least two of the chrysene-h^ resonance absorp tions for which |D - E|/hc was determined (0.117036, 0.115733) may be superpositions of more than one indivi dual resonance. Nevertheless, in view of the apparent correspondence in the three pairs of values, and the fact that in the case of chrysene-hj^ the largest and smallest splittings shown bracket the multiplet (i.e., no other splittings are obtained outside these values), it appears justifiable to make a qualified comparison of these data. Although for each pair of values in Table 15 the chrysene-hj^ splittings are larger, the mean difference is only 29 £ 6 • 10" cm" while the estimated combined —6 uncertainty for the two sets of splittings is 2 0 • 1 0 cm"*. Thus, if the assignment of the six splittings into the given pairs is correct, no significant deuterium i isotope shift can be established for the (D - E) zero- field splittings of chrysene in p-terphenyl.at — 80K. Further evidence for the absence of a substantial isotope shift is provided by the observation that, in a D - E 152

Table 15. Selected D - E Zero-Field Splittings of Chrysene hi 2 and Chrysene-di2 p-Terphenyl at ~77K.

Guest Observed [d - E|/he ( cm’^)

Chrysene-h1 2 0.117036 0.116155 0.115733

Chrysene-d1 2 0.117006 0.116133 0.115697 153

experiment at ~80K for a single crystal of OHA containing both chrysene~h^ 2 ehrysene-d-j^* resonance absorp tions from the two guests could not be completely resolved even at fields as low as ~ 25 G. This can be contrasted with the case for single crystals of biphenyl containing both phenanthrene-h-^Q and phenanthrene-d^Q at ~ 77K (see Chapter VII), where the two guest resonance absorptions could be almost completely resolved at ~ 25 G (for the transitions near ~ 0.147 and ~ 0.093 cra”^). Moreover, the mean difference in the [d - E|/hc values for proto- and perdeuterophenanthrene was 206 • 10"^ it 10 • lO^cra"^. Thus, if deuteration of chrysene produces a shift in the D - E zero-field splitting at '•77, the magnitude of the effect is at most approximately one-tenth the shift in the (D - E) splitting of phenanthrene upon deuteration.

2* Deuteration of p-Terphenyl An effect in the zero-field splitting of a guest due to deuteration of the host (in organic mixed crystals) has not been previously reported. However, we have ob served such an effect for the splittings of chrysene-d^ in p-terphenyl ~hjyf and - d ^ mixed crystals. Table 16 shows the values of |d - E|/hc and |2E|/hc determined for chrysene-d^2 p-terphenyl-h^ and - d ^ at '-'80.1K. Six of the (7) D - E and five of the (7) 2E splittings Table 16. D - E and 2E Zero-Field Splittings of Chrysene-di2 in p-Terphenyl-h^ (C1 8 H3 4 ) and p-Terphenyl-d1 4 (C18D14)’

• SIGNAL SET

1 2 3 4 5 6 7

0.117060 0.117024 0.116942 0.116720 0.116117 0.115741 0.115678 C18D14 |d -e | /he (cm*l) 0.117006 0.116977 0.116891 0.116639 0.116133 0.115697 0.115612 c 18h14

0.053973 0.052599 0.054044 0.054196 0.053468 0.052871 0.052337 1C18D14 |2E /he (cm"i) 0.053926 0.052624 0.053999 0.054195 0.053443 0.052874 0.052312 i C18H14 154 155 were larger for p-terphenyl-d^ host while the remaining ones were smaller. (Why three splittings decreased while all others increased is not known at this time but these three values do establish that a systematic experimental error was not responsible for the observed shifts.)

The mean shifts in the values of |d - E|/hc and [2E|/hc which increased were (57 it 13) • 10 ^ and (29 it 17) •' 10“^cm"l, respectively. The corresponding mean increases in the spin-Hamiltonian parameters of —6 chrysene-d^ 2 upon deuteration of p-terphenyl were 44 * 10 and 15 * 10"^ era”* for |d] /he and |E|/hc, respectively. In the case of host deuteration there is no experi ment analogous to the '‘unambiguous” , two-guest experiments described above (and in Chapter VII) for the study of guest-deuteration effects in chrysene and phenanthrene. Nevertheless, since the average uncertainty in the chrysene-d^2 zero-field splittings measured in this study was 6 * 10"^ cm"*', the occurrence of an isotope shift in

(at'least) the |d - E]/hc and |d |/he values for chrysene- p-terphenyl-h^ and - d ^ seems clearly established by the data presented. Further discussion of these effects is given in Section VII.C. The mean resonance linewidths for chrysene-dj^ *-n P“ terphenyl-d]_4 were 4.8 it 0.4G and 5.3 i 0.5G for D - E and 2E resonance absorptions, respectively. These are both 156

0.8G ( ^ 15%) less than the mean linewidths found for

chrysene-d^ 4 in p-terphenyl-h^ (Section V.C.). Although the mean linewidths for chrysene-d^ in p-terphenyl-h^

and -d^ 4 agree within their combined uncertainties, the probability (calculated by Fishers "t" test, for unpaired variates) that the "true" linewidths for p-terphenyl-h^ and - d ^ hosts are the same, was only 15% for the D - E data and 39% for the 2E data. Thus, the decrease in linewidth for p-terphenyl-d^ may be significant, indicat

ing that chrysene-d^ 2 resonance absorptions are perhaps broadened to a small extent by protons on nearest- neighbor p-terphenyls.

H. CONCLUSION At about the same time (1965) that multiplet struc ture in low-field magnetic resonance spectra was first ob served in this laboratory, Hochstrasser and Small repor ted observation of multiplet structure in the electronic spectra of phenanthrene in biphenyl at ^ 4 K, and sug gested the possibility of inequivalent guest orientations in this system.^6,77 Prior to this time, Brandon, Gerkin o and Hutchison had reported misalignment of phenanthrene in biphenyl on the basis of high-field magnetic resonance experiments at 77 K, but no multiplet structure was 157

observed in the corresponding low-field experiments at this temperature. Also, Charles, Fisher and McDowell had 33 reported misalignment of pyrene-d^Q in fluorene. In a note published in 1967, Cleghorn suggested that there are "probably two types of 2-methylnaphthalene triplet state 78 molecules substituted in the durene lattice." However, this interpretation is not required by any of the data presented by Cleghorn and in fact is in conflict with some of the data. For example the number of absorptions shown in a figure included in the note is not consistent with this interpretation. Moreover, Cleghorn failed to consider the possibility of rotation of the fine structure axes (as a result of the methyl substitution) for a well- oriented molecular framework. Thus, the validity and usefulness of this note must be seriously questioned. Recently Hochstrasser and Small reported multiplet structure in the optical spectra of several additional mixed crystal systems (pyrene in biphenyl, anthracene in biphenyl and phenanthrene In naphthalene) at temperatures 79 below ''■'20 K. On the.basis of results from polariza tion measurements and semi-empirical calculations for the intermolecular guest-host interactions, they have in terpreted the observed structure as arising from inequivalently oriented sets of guests, although they have also stated that distortions of the host structure in the 158 neighborhood of fixed inequivalent guests cannot be de finitively rejected as a possible source of the roultiplets. Several interesting differences can be noted between the systems studied by Hochstrasser and Small (HS) and those studied here. First, in all of the systems inves tigated by HS, interconversion of states (indicated by collapse of the multiplet to a single line) was observed above ~10 K. In contrast, for the chrysene-d^ systems multiplet structure persisted to the highest temperatures for which study was permitted (by the triplet lifetime temperature dependences), namely -~90 K for p-terphenyl host and ''■'250 K for OHA host. This implies much higher barriers to thermal interconversion between inequivalent guests in the latter systems, and in fact may indicate that the inequivalent sets of chrysene-d-^ molecules are "locked in" as these mixed crystals grow. Secondly, in the HS systems multiplet structure typically took the form of doublet or triplet components whereas for chrysene- d^ 2 p-terphenyl and other mixed crystal systems studied here (see following Chapter), between five and seven components were observed in multiplet patterns. Thus, although there is a strong likelihood that the same phenom enon is responsible for multiplet structure in both opti cal and magnetic resonance spectra, this remains to be shown by comparing results from the two types of studies 159 of the same mixed crystal systems. (See Chapter IX for

suggested researches of this kind.) Although each of the

two experimental methods offers its own advantages with regard to this type of investigation, it appears at this

time that low-field magnetic resonance spectroscopy af fords the most direct and quantitative evidence for multiple * orientations of guest molecules in host structures. Cer tainly, on the basis of the results presented in this and the following Chapter, it seems clear that the (formerly prevalent) presumption of unequivocal orientation of aro matic guests in aromatic host structures can no longer be considered generally valid. CHAPTER VI EXPERIMENTAL RESULTS AND DISCUSSION FOR OTHER MIXED CRYSTAL SYSTEMS

Several limited studies were undertaken in connection with the major researches described in Chapters V, VII and VIII. Results from these studies are discussed briefly here.

A. 1,2; 5,6-DIBENZANTHRACENE ORIENTED IN p-TERPHENYL AND IN 2,2'-BINAPHTHYL Multiplet structure was observed in the low-field. mag netic resonance spectrum of 1,2; 5,6-dibenzanthracene-h^ (DBA-h]^) oriented in single crystals of p-terphenyl-h^ at '■'■'77K. A total of eight resonance absorptions corres ponding to different zero-field splittings were detected: four at ~ 0.113 cm”*’; three at ^0.062 cm”*- and one broad absorption at ^0.051 cm”*. The precise values of the zero-field splittings determined for DBA-h^ in p-terphenyl-h^ at approximately 80 K (CuSO^ filter) are

160 161 given in Table 17. (The raw data from which these values were calculated are too numerous to report here but are tabulated in Notebook 4.C.1, pages 80-91, 116-118, 120 and 121.) Approximate values of resonance linewidths for the individual components of a multiplet were determined and found to be 15 gauss for absorptions near 0.113 cm"*- and 18 gauss for absorptions near 0.062 cm *". The magnetic resonance lifetime of DBA-h^ in p-terphenyl was measured at ^77 K and determined to be 1.2 seconds, which agrees well with the value of 1.5 seconds reported for DBA-h^ in EPA glass. -1 In a single experiment at ''■'0.113 cm for DBA-d^ in p-terphenyl-h^, at least seven distinct resonance absorp tions were resolved due to narrower linewidths (4 G) for the perdeutero-dibenzanthracene absorptions. The multiplet pattern observed at '■'■'0.115 cm"*' from DBA-d^ in p-ter phenyl is shown in Figure 21. From a single magnetic resonance decay, the triplet lifetime of DBA-d^ in p- terphenyl at ^77 K was found to be 4.7 seconds. For the resonance absorption associated with the largest zero-field splitting of DBA- d14 in p-terphenyl, the transition energy corresponding to absorption at ^ 24 G was 0.11485 cm"*', while for the equivalent signal from DBA-h^ in p-terphenyl the transition energy corres ponding to absorption at ^24 G was 0.11458 era"*'. These 162

Table 17. Zero-field Splittings of 1,2; 5,6-Dibenzanthra- cene-h^ 4 Oriented in p-Terphenyl-h^.

Transition (cra“^)

0.113 0.062 0.051

0.114465 0.063864 (0.05074 - 0.05090)a 0.113765 0.062901 0.113606 0.062336 0.113417

a Broad absorption observed near zero field for this range of energy/hc. /V 97 G

Figure 21. Multi-absorption pattern observed at low fields for the 1,2; 5,6-di benzanthracene-di4 in p-terphenyl transition near *-0.115 cm"l. 163 164 results indicate a very substantial deuterium isotope shift in the (D - E) zero-field splittings of dibenzanthracene (in p-terphenyl). In contrast to the multiple resonance absorptions ob served from dibenzanthracene oriented in p-terphenyl, only a single resonance absorption was observed for each of the three possible triplet transitions for DBA-h^ oriented in single crystals of 2,2*-binaphthyl. The zero-field splittings (at '-'77 K) associated with the three resonance absorptions were determined to be 0.113765, 0.062981 and 0.050768 cm The linewidths of these absorptions were 15.4, 18.8 and 19.1 G, respectively. Assignment of |D + E|/hc and |2E|/hc to the 0,062 cra"^ and ~ 0.051 cra"*^ experimental values determined in this study cannot be made unambiguously on the basis of o the data obtained thus far. (The signs of di'/dH ob tained in experiments on DBA in 2,2*-binaphthyl are of little help in this regard since no report of a crystal structure determination for 2,2*-binaphthyl has been found.) A complete Investigation of 1,2; 5,6-dibenzanthracene triplet resonance absorptions will require a study com parable in scope to that described in the preceding chap ter for chrysene since a comparable number of signals are observed in the multiplet patterns for DBA oriented 165 in p-terphenyl. However, the results already obtained are sufficient to establish, for dibenzanthracene guest, dra matically different substitutional behavior in p-terphenyl and in binaphthyl host structures.

B. 2,3-BENZ0CARBAZ0LE ORIENTED IN p-TERPHENYL AND CHRYSENE Commercial chrysenes, even after purification by re crystallization, have been shown to contain as much as 10% 51 52 of the heterocyclic impurity, 2,3-benzocarbazole. * In order to remove any doubt as to whether this tenacious impurity was the origin of one or more of the resonance absorptions occurring in multiplet structure patterns from mixed crystals of chrysene in p-terphenyl, the low- field magnetic resonance spectrum of phosphorescent 2,3- benzocarbazole was investigated. From mixed crystals of unpurified chrysene in p-ter phenyl, in addition to the multiplet patterns observed at **0.116, '"'0.062 and ''■*0.053 cm ^ (which were assigned to chrysene in the analysis in the preceding chapter), two other sets of absorptions were observed. For one set the corresponding zero-field splittings were shown to range between 0,0847 and 0.0858 era"*', while zero-field splittings corresponding to absorptions in the other set ranged in value from 0.0636 to 0.0644 cm"**. At least five resonance absorptions were observed in each of these 166 patterns. If 2,3-benzocarbazole occupies substitutional sites in the p-terphenyl host structure so that its fine structure axes are approximately parallel to those of p- terphenyl, then provisionally (on the basis of satisfac tion of the zero-field selection rules) the zero-field ••1 splittings at ^0.085 cm" may be associated with |D - E| /he and those at ''-0.064 cm ^ may be associated with |D4*E|/hc. When an authentic sample of 2,3-benzocarbazole was used as the sole guest in p-terphenyl mixed crystals, multiplet structure patterns were observed at ~ 0.085 cm ^ and at ~ 0.064 cm from such crystals, and component sig nals in these patterns were shown to correspond one-to- one with components of the multiplet patterns observed -1 (at 0.085 and 0.064 cm ) from impure chrysene in p- terphenyl crystals. However, no patterns were observed at the three energies corresponding to the chrysene zero- field splittings. Conversely, when p-terphenyl mixed crystals prepared with chromatographed chrysene were studied, the three absorptions attributed to chrysene were observed but the two patterns attributed to 2,3-

* benzocarbazole were not. These results clearly establish that 2,3-benzocarbazole indeed was not the source of any of the resonance absorptions attributed to chrysene. Single crystals of unpurified chrysene grown by the Bridgman method were of very poor quality. Nonetheless, 167 when excited by the A-H6 lamp at 77 K such crystals mani fested intense orange phosphorescence (as did 2,3-benzocar bazole in p-terphenyl). Two weak resonance signals were observed from the (2,3-benzocarbazole in) chrysene crys tals. The zero-field splittings associated with the ab sorptions were determined (for unfiltered excitation), and found to be 0.08594 era"*' and 0.06373 cm"1 (£ 0.00005 cm *"), respectively. Thus, the values found for the zero-field splittings of 2,3-benzocarbazole in p-terphenyl host (average values) and in chrysene host are very simi

lar. If the assignment of (|d | -f- |e ) )/hc and ([d | — |e | )/hc to experimental values at -'0.086 and <-'0.064 cm"*' is cor rect, then for 2,3-benzocarbazole in chrysene at -'77 K,

|d | /he = 0.07483 cm*1 and |E|/hc = 0.01110 cm"1 (£o.00005 cm"1).

C. IMPURITY IN S-OCTAHYDROANTHRACENE As discussed in Section IV.A.2, standard purification methods failed to remove from s-octahydroanthracene (OHA) an impurity which gave rise to a yellow-green phosphor escence emission in single crystals of OHA excited by the A-H6 lamp, both at 77 K and at 273 K. Two magnetic reso nance absorptions were detected from such single crystals of OHA and the zero-field splittings associated with the 168

absorptions were determined (for unfiltered excitation; liquid nitrogen bath) to be 0.111420 cm"^ and 0.083237 cm*"^ (estimated uncertainties for replicate determinations were -5 -1 ''-'l *10 cm ). The magnetic resonance lifetime was measured in an experiment at ~ 0.083 cm ^ and found to be 1.0 sec. The absorption linewidth for the signal near 0.1114 cm ^ was approximately 15 G. In an experiment in which the resonant cavity was im mersed in a distilled water-ice slush and the bandpass filter was employed, the resonance absorption associated with the zero-field splitting value 0.11142 cm~* at ~ 7 7 K was determined (semi-quantitatively) to have a correspond- -1 ing zero-field splitting at 273 K of 0.1097 cm On the basis of the similarity of the magnitudes of the triplet energy, zero-field splittings, magnetic reso nance lifetime and temperature coefficient of the D - E zero-field splitting of the impurity to the values of these properties for naphthalene, it is suggested that the impurity may be 1,2,3,4-tetrahydroanthracene. (An additional basis for this suggestion is that in the cata lytic hydrogenation of anthracene both tetrahydroanthracene and OHA are produced.) However, it has not been pos sible to establish whether this is the impurity molecule being studied, since (as discussed in Section IV.B) all mixed crystals of tetrahydroanthracene in recrystallized octahydroanthracene grown to date have shattered after emerging from the hot zone of the Bridgman furnace. CHAPTER VII

DEUTERIUM ISOTOPE EFFECT IN ZERO-FIELD SPLITTINGS OF PHOSPHORESCENT PHENANTHRENE ORIENTED IN BIPHENTt

A. INTRODUCTION On 1968 August 30, while examining a chart recording from a low-field experiment for phenanthrene-h^Q carried out by him on 1964 July 1, Dr. Gerkin noted instances of multiple resonance absorptions in several field scans and brought these to my attention. These data raised a ques tion as to whether the (multiple) signals observed were due to multiply-oriented sets of phenantlirene molecules and hence whether "multiplet structure" had been recorded but not noted prior to the observation of this phenomenon for the system chrysene in p-terphenyl. Immediately after this question was raised a biphenyl crystal containing phenanthrene-d^Q was prepared, since from the chrysene studies it was clear that if multiple signals were not ob-

A served for deutero-phenanthrene then multiplet structure could not be expected for proto-phenanthrene in biphenyl due to greater linewidths of proto-signals.

170 171

Only a single resonance absorption was observed near

0.147 cm~Vfrora the phenanthrene-d^Q-biphenyl mixed crystal at 77K. Thus, multiplet structure is not found in the low-

0 field magnetic resonance spectrum of phenanthrene in bi phenyl at 77K. Consequently, the detection of multiplet structure for chrysene in p-terphenyl in 1966 still appears to be the first observation of this effect by magnetic reso nance methods. (More detailed consideration subsequently, of the 1964 July 1 experiment led to the realization that the multiple signals recorded arose from the two translationally inequivalent sets of phenanthrene guests in bi phenyl and were to be expected for the transition, crystal orientation and fields chosen in this experiment.) In the same experiment which established that only a single signal is observed for phenanthrene-d^Q in biphenyl at '-■'0.147 cm-^, the zero-field splitting corresponding to this resonance absorption was measured at approximately 87K. The value of the splitting disagreed with the value reported by Brandon, Gerkin and Hutchison for the corres ponding splitting for phenanthrene-h^Q in biphenyl at ap proximately 84K. (The temperatures given here are the approximate crystal temperatures measured in the respective researches for unfiltered excitation by the A-H6 lamp.)

The difference in the values determined for the 0.147 cm"* splitting was significant since its magnitude was 172 approximately ten times the combined estimated uncertain ties of the measurements and further seemed too large to be explained on the basis of differences in the crystal temperatures for the two experiments assuming a reasonable value for the temperature coefficient of the splitting. Although magnetic resonance studies of the proto- and deutero-species of several molecules (naphthalene, phenan threne and pyrene, for example) had been made at either high or low fields, no measurements of zero-field split tings at one temperature and for one host structure for both isotopic species of a particular molecule had been reported. Interestingly however, Hirota, in describing a high-field experiment for a biphenyl crystal containing both phenanthrene and phenanthrene-d^Q, had stated: "Since the spin Hamiltonian parameters D and E for the protonated and deuterated phenanthrene are almost the same, the sig nal due to one species cannot be distinguished from that due to the other because the signals overlap."®® Since Hirota offered no experimental evidence to support his statement that D and E for. the two isotopic species are "almost the same" (which implies that, in fact, he be lieved the values to be slightly different) it must be con cluded that this assertion was conjectural. Thus, it was clearly of considerable interest to establish whether a deuterium isotope effect in the zero-field splittings of 173 phenanthrene could be demonstrated and if so to determine to its magnitude* As discussed in Chapter V, the field resolution of resonance absorptions associated with slightly different zero-field splittings becomes progressively greater as • zero field is approached. Thus there was reason to believe that in contrast to the high-field results discussed by Hirota, resolution of absorptions from phenanthrene-h^Q and phenanthrene-d^Q oriented in the same biphenyl crystal . would be possible at very low fields. In any case, because the discrepancy between the |D - E|/he values of the proto- and deutero-species of phenanthrene was of the same magnitude as typical uncertainties in high-field measure ments of zero-field splittings and also because of the relative lack of resolution at high fields (see calcula tion, page 176), it appeared probable that only the low- field method (or zero-field spectroscopy) would permit study of an isotope effect of the indicated magnitude. Accordingly, the low-field investigation described below was carried out.

B. ■ RESULTS FOR PHENANTHRENE ORIENTED IN BIPHENYL Experiments were performed at the boiling point of liquid nitrogen on biphenyl host crystals containing either phenanthrene-h^Q or phenanthrene-d^Q or both. The 174

_nominal concentrations of the four boules from which crys tal pieces were cleaved for use in these experiments are shown in Table 18. In the two-guest boules, the ratios of concentrations were chosen so that signal-to-noise ratios for phenanthrene-h^Q and phenanthrene-d^Q would be approxi mately equal (by accounting for differences in lifetimes and linewidths). Crystal pieces used in these experiments were on the average 7 mm on a side and 3 mm thick. In every precision experiment (excepting those to determine temperature coefficients of splittings) the crystals were excited by an A-H6 lamp whose output was filtered by the three-component ultraviolet-transraission solution described in Section III.A.5. Thus, in no precision experiment (with the exception noted) did the crystal temperature exceed that of the bath by more than 0.3 K (III.A.6). Data pre sented below show that the restriction in the variation of crystal temperature realized by use of this solution filter led to approximately an order of magnitude reduction in the uncertainties of the phenanthrene-d^Q zero-field splittings over the uncertainties obtained in any other low-fieId study. In experiments on two-guest crystals at ^0.147 cm"* and *''0.054 cm , it proved possible to resolve the phenanthrene-h^Q and phenanthrene-d^Q resonance absorptions at fields below approximately 50 G. The maximum separation of 175

Table 18. Nominal Concentrations of Phenanthrene in Biphenyl Mixed Crystals Used in Isotope Effect Study

Boule No. Guest Concentrations (mole %) Phenanthrene-h-^Q Phenanthrene -d^Q

273 0.161 274 0.199 275 0.141 0.031 276 0.124 0.011

I 176 the centers of the absorptions (for which the signal "nearest zero field" could still be completely observed) was approximately 25 G and 15 G for the 0.147 cm***- and 0.054 cm ^ transitions, respectively. Since half the sum of the linewidths of the proto- and deutero-absorptions was 10 G, essentially complete resolution of the two sig nals was achieved in experiments at 0.147 cm"^ at the lowest fields (for which both signals could be observed completely). Shown in Figure 22 are resolved magnetic resonance absorptions from (two-guest) Boule #276 in a linear field scan for which u ss0.147268 cra"^. The phenanthrene-h^Q and phenanthrene-d^Q absorptions are centered at 40.9 and 15.7 G, respectively, and are dis tinguishable by their linewidths (and by their decay - times as well). (The separation of resonance absorption centers for an analogous high-field experiment (H j| x) was calculated for an (X-band) frequency of 9.60 GHz and using the zero-field splitting values determined in this study. The separation was calculated to be 2.0 G at 2120 G, or an order of magnitude smaller than that found in the low field experiments just described. Thus, Hi rota must have observed essentially complete overlap of the phenanthrene-h-^Q and phenanthrene-d^Q resonance ab sorptions in his two-guest, high-field experiment.) Resolution of the resonance absorptions from the two 177

Figure 22. Resolved electron magnetic resonance absorp

tions at ~0.147 cm*’*' by triplet phenanthrene-

d^Q and triplet phenanthrene-h-^Q present to

gether in a biphenyl single-crystal at 77 K.

The absorptions are centered at 15.7 G and

40.9 G, respectively (the field scan is

linear). These absorptions are distinguish

able by their linewidths and decay times as

well as by the magnetic fields at which they

occurred. 178

~ 7 5 G

';>\A aW V

I____I L 179

Isotopic species in the same crystal piece at low fields permitted the determination of their respective 0,147 cm"*’ zero-field splittings under identical experimental condi tions. Thus, the difference found in the two splittings was shown unambiguously not to arise from any systematic experimental error. Moreover, the values of the zero- field splittings determined in the two-guest experiment were in excellent agreement with the values determined for the same splittings for each isotopic species alone in biphenyl. This agreement indicates an absence of any significant interaction between the two kinds of guests in the quite dilute three-component crystals. In a two-guest experiment at 0.093 cm"* some overlap of signals occurred (due to the smaller isotope shift for this transition) but this was not sufficient to com promise the measurement of the centers of the resonance absorptions in precision determinations of the zero-field splittings. The 0.147 cm"* and 0.093 cm"* splittings having been determined, their differences gave expected values for the 0.054 cm"* splittings which were then determined experimentally for each isotopic species alone in biphenyl. Both the predicted and experimental values (which were in good agreement) showed that the isotope shift in the 0.054 era"* splittings was even smaller than that for the 0.093 cm"* splittings. In addition, for 180 the usual orientation of a biphenyl crystal piece in a D + E experiment (the 0.054 cm”*- transition for phenan threne) there is a high probability of observing resonance absorptions both above and below the zero-field frequency from the two translationally inequivalent sets of guest molecules. Thus, because of this possibly complicating factor (in the case of a two-guest experiment), and the expected poor resolution of signals (due to the smallness of the isotope shift) no two-guest experiment was carried -1 out for the 0,054 cm transition.

The precision determinations of the zero-field splittings discussed above (as well as the measurements of triplet lifetimes and resonance linewidths) were car ried out in the same manner as described in Chapter V for the chrysene studies. The best signal-to noise ratios

(for a 2.3 second time constant) and the mean resonance lindwidths found for absorptions from phenanthrene-h^Q and phenanthrene-d^Q for each of the three transitions are given in Table 19. The magnetic resonance data from the precision zero- field splitting determinations are summarized in Table 20. (The raw data which are too numerous to present here, are recorded in Notebook 4.C.1, pages 172 to 197). The assign ment of the experimental energies near 0.147, 0.093 and

0.054 cm"** to the |d - E|/hc, |2E|/hc and |D -f- E|/he 181

Table 19. Resonance Absorption Linewidths and Signal-to- noise Ratios for Phenanthrene-tuQ and Phenan- threne-d-^Q in Biphenyl.

Species Transition (era"*') 0.147 0.093 0.054

Resonance Linewidths 14.5 14.8 a 22.4 a G14H10 0.3 0.8 0.8 5.29 5.06 5.95 C14D 10 0.09 0.39

Best signal-to-noise ratios observed for two second time constant

C14H10 80 45 40

c 14d10 100 100 65

a Determined by interpolation from linear field scans. Table 20 . Magnetic Resonance Data for Fhenenthrene-h and Phenanthrene-d Oriented in Biphenyl at 77.5 K . 10 so

Run nisber I s o to p ic One-guest or Designation Ho. of differ Range of mag S ig n Beet values S ta n d a rd Calculated S ta n d a rd Best values and signal species of tw o -c u e at o f ent wj for netic field o f o f y f o r d e v ia tio n Zeeman d e v ia tio n of tero-field n tn b e r phenanthrene c r y s t a l b phenanthrene which Ijji) over Which dw^ fit of Eq. o f y contribution o f b e s t splittings s tu d ie d * transition was measured resonances (1 0 ) t o to observed v a lu e s o f for fit of v , ca *1 were deter d a ta (cm"1 G'2 ) splittings Z .7 .S . Eq. (10) to m ined (G) (eo-i O'2) at smallest (cm -1 ) d a ta H* ( cm*1 > (cm*1 ) x 10T x 109 x 10°

?L - JO d e u te ro one - 0 . 11*7 6 8 .5 -2 2 -1 + 1 .5 5 0 0 -9 12 0 .3 0.11*72271

2 5 - JO d e u te ro two - 0.1U7 5 7 -7 -1 9 -1 + 1 -5 9 3 0-5 9 0 .1 0.11i72256

26- 31 deutero one - 0 . 0 9 J 6 6 . 2 - 2 1 .1 - 1 . 1*20 0 .3 5 0 .1 0.0933303

ST- J1 d e u te ro two ~ 0 .0 9 J 7 6 . U 2 2 .7 - 1-595 1 .2 e 0 . 1* 0.0933317

2 8 - J2 d e u te ro one - 0 . 0 5 3 5 1 8 .5 -2 7 -9 - 0 .8 3 6 0 .7 29 0 . 1* 0.0538967

2 9 - 35 p ro to one - 0 . 11*7 6 7 .6 - 2 3 .8 + 1 .3 0 1 0 .5 8 < 1 0.11*7030

JO - 53 p ro to two - 0.11*7 5 1 2 . 8- 2l* .l + 1 .5 6 3 0 .8 25 2 0.11*7013

J l - J t p ro to one - 0 . 0 9 3 1* 1 2 .6 - 21* .6 + 0 .6 2 6 0 . 1* 10 < 1 O.093178

J 2 - Jl* p ro to one - 0 .0 9 3 5 9 -8 - 2 1 .6 - 1.1*96 1 .6 111 1 0 .093196

3 3 - Jl* p ro to two - 0 .0 9 3 5 1 0 .8 -1 8 .3 - 1-539 **•9 18 1 0 .093199

Jl*- 35 p ro to one “ 0 .0 5 3 5 11 . 2 - 11* .0 ♦ 2.275 30- 28 5 O.O538O6

J 5 - J5 p ro to one - 0 .0 5 3 13 1 2 .0 -2 1 .9 - 0 .6 0 1 T-ti 10 2 0.053835

a proto - phenanthrene-h10 ; deutero ■ phenanthrene-d^o

b Two-guest crystals contained both phensnthrene-hlo and phenanthrene-d^ . 182 183 splittings, respectively, follows the detailed analysis and assignment for phenanthrene-h^Q by Brandon, Gerkin and

Hutchison. Zero-field splittings for phenanthrene-h^Q were determined to within approximately jrlO *10" cm and for phenanthrene-d^Q to within approximately il 10 * 1 0 cm"'*-. These estimates are based upon the splitting values in Table 20 which are summarized in Table 21, and which show this order of agreement (a) between values measured for a par ticular zero-field splitting from different crystals and (b) between the appropriate combination of any two of these splittings and the third splitting from a given crystal. The phenanthrene-h^Q zero-field splittings and spin Hamiltonian parameters presented in Table 21 agree well with the values for phenanthrene-h^Q reported by Brandon, Gerkin and Hutchison. To be compared strictly to the present data, however, the values of BGH should be corrected to 77 K. The temperature coefficients for the zero-field splittings of phenanthrene-h-^Q in biphenyl were qualitatively determined by measuring splittings with and without the bandpass filter in the optical path and were found to be of the order -5*10”^, -5*10"^ and 0 #10"^ cm"^K"’^ for d(|D - E|/hc)/dT, d( |2E|/hc)/dT and d( D + E /hc)/dT, respectively. Since in the BGH inves tigation crystal temperatures were shown to be Table 21. Zero-field Splittings and spin Hamiltonian Parameters of Phenanthrene-h^Q and Phenanthrene-d^Q Oriented in Biphenyl at 77K.

Guest(s) Guest Observed Observed Predicted Observed |D| / he |E|/ he present studied iD-El/hc |2E| /he |D+E|/hc |D+E|/he (cm“l) (cm"i) (cm"l) (cm“i) (cm~^) (cm-1)

0.1472271 0.0933303 0.0538968 0.0538967 0.1005619 0.0466652 c 14D 10 c 14d 10

c 14h10 c 14d10 0.1472256 0.0933317 0.0538939 a 0.1005598 0.0466659 C14d10

C14h10 0.093199 0.100414 0.046599 C14H10 0.147013 0.053814 a C14D10

0.147030 0.093187 0.053843 0.053820 0.100437 0.046593 C14H10 c 14H10

a This entry was not determined for reasons discussed in the text. 185 approximately six kelvins above the bath temperature, cor rections to the [D - E|/hc, |2E| /he and |D+E|/hc values would be approximately -f3*10"-*, +3*10 ^ and 0 cra~X, respec tively. Thus, the result of this temperature correction is to bring the two sets of data into still better agreement; in particular to well within the combined uncertainties of the respective sets of splittings.

Determination of zero-field splitting values to seven significant figures has not been reported for any other study. The attainment of this precision in the case of the phenanthrene-d^Q splittings can be attributed princi pally to the very small crystal-temperature variation ob tained by using the UV bandpass filter and to the small magnitude of the temperature coefficients for the zero- field splittings of phenanthrene-d^Q. (Two other contri buting factors were large signal-to-noise ratios and small linewidths for the phenanthrene-d^Q resonance absorptions.)

The temperature coefficients were determined qualitatively in the same manner as described above for phenanthrene-h^Q and were found to be approximately -3*10 —6 , -2*10"° A and

-1»10"^cra“xk “X for d(|D - E|/hc)/dT, d( |2e ) /hc)/dT and d(|D + E|/hc) dT, respectively. Whereas in the previous studies temperature variations of several degrees were £ possible, leading to uncertainties of order i2K x 5*10“ cra“XK~^^ ±1.10“5cm"X, in the present studies the maximum 186 contribution to the splitting uncertainties (for phenanthrene-d^) due to temperature variations was of order i0.15K x. 3* 10“^cm"^K“^ = it4.5 * 10**^cra"^. (The maximum variation in the nitrogen bath temperature was calculated from barometric pressure data taken in each experiment and was found to be 0.038K. Thus, this was a negligible source of crystal temperature variation compared with crystal heating due to the incompletely filtered output of the A-H6 lamp.) Since the mean uncertainties in the previous study (for phenanthrene-h^Q) and in the present study (for phenanthrene-d^) were approximately dl2# 10“^ cm —1 and!T5.5 -I- • 10 “7 cm —l , respectively, it is clear that variations in crystal temperature can account for a major part of the uncertainty in both cases.

The larger uncertainties associated with the zero- field splittings of phenanthrene-h 10 (f*±10 • 10 ) were due in part to the lower signal-to-noise ratios for these measurements but were due principally to the presence of resolved proton hyperfine structure in phenanthrene-h^Q resonance absorptions for many of the precision determina tions. The effect of the proton-electron hyperfine inter action on the precision splitting data for phenanthrene-h^Q is discussed in the following Chapter. 187

C. DISCUSSION The deuterium isotope shifts in the zero-field split tings of phenanthrene at 77K are, from the appropriate averages of values in Table 21, 206*10"^, 117*10"^ and

77* 10"^cra"** for |d - e | /he, |2e |/he and |D+E[ /hc, res pectively. From these values the calculated shifts in the spin Hamiltonian parameters fD| /he and |e |/he are 136•10" and 69*10" , respectively, and the fractional . changes in D and E are 2(DQ - Dp)/(DD + Dp) = 1.35 • 10"^ and 2 (Ed - Ep)/(ED + Ep) = 1.48 • 10"**, thus very nearly equal. Although the shifts cited above are small in absolute magnitude they are substantial with respect to the estimated combined uncertainties in the proto and —6 _»1 deutero splittings (approximately 15*10 cm ). Since, in addition, approximately one half of the precision data was obtained in "two-guest11 experiments (described above) which effectively eliminates systematic error as a source of the shifts, a deuterium isotope effect in the zero-field splittings (and spin Hamiltonian parameters) of phenan threne is well established. This is not the case for chrysene for which (as dis cussed in Chapter V) isotope shifts in the zero-field splittings could not be established with respect to the experimental uncertainties, at 77K, although a significant shift in the chrysene splittings was observed upon 188

deuteration of the p-terphenyl host. Since the combined uncertainties in measurements for chrysene were comparable to those for phenanthrene ( ^ 15 *10”^ cm" ^) the magnitude of any triplet-state deuterium isotope effect at ^77K for chrysene is at most one tenth of that described here for phenanthrene. Results from an isotope effect study for naphthalene oriented in biphenyl are presented in Table 22. These values indicate that still a third case occurs in the iso tope shifts for naphthalene (compared with those for phenanthrene and chrysene). Namely, that at 77K, there is a substantial isotope shift for |d | /he, 141*10"^crn~^, but not for |E|/hc, -13*10"^cm“^. Not only are the magnitudes of the shifts in D and E not comparable as they were for phenanthrene in biphenyl at 77K (although the fractional changes, 1.42*10 —3 for D and -0.84*10 —3 for E are much more nearly equal), but [e |/he apparently decreases from proto- to deutero-naphthalene at 77K. (The shift in |E|/hc, cited above from the ,!predicted,, values for |2e ) /he in Table 22, has been confirmed by the experimental determina tion of |2E| /he for naphthalene-hg and naphthalene-dg in biphenyl by D. L. Thorsell in this laboratory.) Interpretation of the observed isotope effects does not seem straightforward in view of the variety of results found for the first three molecules investigated. Of 189

Table 22. Zero-field Splittings and spin Hamiltonian Para meters of Naphthalene-h8 and Naphthalene-d8 Oriented in Biphenyl at ^77K.

Guest Observed Observed Predicted |D| /he lEl /he ID-El /he lD+E| /he 12E|/he (cra-1) (cm-1) (cm“l) (cm"1) (cm X)

Naphthalene-dg 0.114929 0.083947 0.030982 0.099438 0.015491

Naphthalene-hg 0.114801 0.083793 0.031008 0.099297 0.015504 190

particular concern is the difficulty in establishing the

partition of the observed effects between intermolecular

and intramolecular causes. A plausible basis for an

intramolecular effect on triplet splittings is an altera

tion of the molecular vibrational spectrum (particularly

for those modes involving the mass of the hydrogen iso

topic species critically); this alteration in turn would

affect, via vibronic coupling, the electronic magnetic di polar interaction which leads to the triplet splittings.

The fact that (according to the preliminary values cited

above) the temperature coefficients of the phenanthrene- h^Q splittings are approximately a factor of two larger

than those for phenanthrene-d^Q in the same host structure and over the same range of temperature is consistent with

this intramolecular interpretation. (The observation of different temperature coefficients also emphasizes the need for precision measurements of the temperature dependences

of zero-field splittings over an extended range of tem perature. Facilities for carrying out such studies were not available during the present investigation.) A second, related mechanism for an intramolecular effect is the

small but possibly significant effect deuteration might have on molecular hydrogen-hydrogen ("steric") interactions.

The nature and magnitudes of such distortions have been cal culated for phenanthrene and chrysene by Coulson and 191

Haigh. 81 "Cancellation" of deuteration effects on steric distortion might be expected for chrysene due to its higher symmetry (although an additive effect cannot be ruled out) but no such "cancellation" should occur for the steric interaction between the 4, 5 protons (deuterons) in phenanthrene. The observation of a measureable isotope effect for phenanthrene but not for chrysene is consistent with this mechanism for an intramolecular effect. On the other hand, the substantial effect on the zero-field splittings of chrysene observed for deuteration of the p-terphenyl host (V.F.2), but the lack of an isotope shift for deuteration of the chrysene itself, indicates that any explanation based on purely intramolecular effects will be inadequate. Thus, although the present work has clearly established for the first time the occurrence of deuterium isotope effects in triplet state splittings, a definitive interpretation of these effects will require the accumula tion of data for more systems. Specifically, what is needed are low-field precision measurements of the tempera ture dependences of zero-field splittings of a considerable number of isotopically substituted guests in various isotopically substituted hosts. CHAPTER VIII PROTON HYPERFINE STRUCTURE AT MAGNETIC FIELDS BELOW 100 GAUSS IN MAGNETIC RESONANCE ABSORPTIONS BY ORIENTED TRIPLET STATES OF AROMATIC MOLECULES

A. INTRODUCTION

In the course of one of the first experiments concerned with establishment of the deuterium isotope effect in zero- field splittings, ill-defined but persistent structure was noted in lox*-field resonance absorptions by phenanthrene- h^Q oriented in biphenyl. In a succeeding experiment suffi cient resolution was attained (after optimization by appro priate rotation of the static magnetic field) to warrant the conclusion that proton hyperfine patterns were being detected. This conclusion was supported by comparison of these spectra with a high-field hyperfine pattern for phenanthrene in biphenyl reported by Brandon, Gerkin and

Hutchison, and was quite unexpected since intensive in vestigation of this same mixed-crystal system at low fields had been reported by these same authors but no hyperfine structure had been detected. In fact, in reporting results

192 193 from their magnetic resonance studies of phosphorescent phenanthrene, Brandon, Gerkin and Hutchison (BGH) failed to discuss the fact that high-field resonance absorptions manifested resolved hyperfine structure whereas low-field absorptions did not, from which it could be inferred that this was to be expected. On this basis, we had assumed that hyperfine structure in resonance absorptions by triplet states at low fields was not observable. Following the unexpected observations from phenan threne, hyperfine structure was also obtained at fields below <~100 G in resonance absorptions by oriented triplet naphthalene (which had also not been observed in previous studies of naphthalene at these fields by BGH^) and by oriented triplet dibenzothiophene (for which no previous studies in single crystals have been reported). Presented in this chapter are results and conclusions from a study of resolved proton hyperfine structure at low fields, which was carried out primarily to answer the fol lowing questions: First, could the previous failures to observe resolved structure at low fields be explained? Second, do hyperfine patterns obtained at high and low fields differ and if so in what ways? Third, co|uld re solved hyperfine structure be observed at zero field? This latter question was particularly timely since in re cent publications van der Waals and his co-workers^ 194 had misconstrued the low-field experiments reported by BGH as zero-field experiments, and had (incorrectly) cited the failure to observe hyperfine structure in the previous low- field studies as experimental evidence that hyperfine struc ture is not observable in zero-field transitions. The effects of resolved hyperfine structure on preci sion determination of zero-field splittings of proto- species at low fields are also discussed below.

B* RESULTS AND DISCUSSION Dilute mixed single-crystals of naphthalene or phenan threne in biphenyl and of dibenzothiophene in s-octahydroanthracene (OHA) were prepared as described in Chapter IV. Biphenyl single-crystal pieces could be well oriented with respect to the static magnetic field (H) since the bi phenyl crystal morphology and structure are known in detail. 82 Since the crystal structure of OHA has not been reported, approximate alignment of dibenzothiophene fine structure axes with H was achieved by "trial and error" orientation of the OHA single-crystal pieces (see below). Reasons For Previous Failures to Observe Hyperfine Structure at Low Fields Two factors have been observed to enhance substan tially the resolution of hyperfine patterns below ~100 G: (a) reduction of misalignment of a triplet fine-structure 195 axis from H, and (b) uniform field-scanning, a.) In the previous low-field studies of naphthalene and phenanthrene oriented in biphenyl, no special efforts were made to align the static magnetic field precisely along principal fine- structure axes, since this was not required even for pre cision determinations of the zero-field splittings of these •* molecules. In particular, cleavage planes themselves rather than the molecular axis orientation with respect to the planes were used as the basis for positioning crystal pieces relative to H. As a result, the average misalign ment of the fine structure axes from H would be expected to have been 15 to 25°. In the present investigation the average misalignment was estimated to be approximately 5 to 10°, and it was shown that misalignments of order twenty degrees were sufficient to cause substantial loss of resolution in a hyperfine pattern at a given (low) field, b.) In the previous low-field studies the magnetic field was swept in a (necessarily) highly non-uniform manner by manual setting of resistors in series with a Helmholtz mag net. As discussed in Section III.A.l, a linear, electronic fieldsweep unit was used in the present experiments and the scanning rate was decreased until no further improve ment in resolution was obtained. Two additional experimental factors can be cited which, although significant, are felt to have played a less 196

important role in the successful observation of hyperfine patterns at low fields: c.) At a given magnetic field the field stability was superior in the present apparatus since a current-regulated power supply was employed whereas in the previous investigation regulation was limited to that of the line voltage (obtained from a wall socket), d.) Some variation in pattern resolution was observed from crystal to crystal in these studies, indicating that boule quality may influence the resolution of hyperfine structure at low fields. It is believed that some improvement in single-crystal quality has resulted from modifications in the design of Bridgman furnaces used in the present studies, but this cannot be established definitely. Consideration of these four factors leads to the con clusion that while no one of them should have prevented observation of resolved hyperfine patterns in the previous studies, their combined effect did. The relative impor tance of these experimental factors in accounting for the previous failures is, however, not subject to quantitative determination. 2. Comparisons of High-Field and Low-Field Hyperfine Patterns a * Low Fields Between ~75 and -*100 G « Hyperfine patterns observed at *''100 G differed only very slightly from patterns observed in X-band experiments 197 on the same systems. For example, for naphthalene for H very nearly parallel to the long in-plane axis (x-axis), field increments between adjacent extrema in one-half of the hyperfine pattern were, respectively, 3.8 ill 0.4; 4.4i 0.3; 4.0 ± 0.2; 4.0 ± 0.2; and 4 . 3 ± 0.2 (central dif ference) gauss at 84.4 G and 3.4; 4.7; 3.6; 4.3 and 3.4 (central difference) gauss (estimated uncertainties as above) at 3939.4 G. Thus, only the central differences for these two fields differed from each other by more than the combined uncertainties; moreover, the total field-spans of these patterns between outermost extrema were 36.2 it 0.4 G and 35.1 i 0.4 G, respectively* The great similarity between resonance absorptions from triplet naphthalene obtained at *~104 G and at '''3940 G, respectively, is shown in Figure 23. The hyperfine pattern obtained for H very nearly paral lel to the long in-plane axis of phenanthrene and for the resonance centered at 72.3 G is shown in Figure 24a. This too is very similar to the equivalent pattern observed at high fields and shows excellent resolution of the seven line structure. Although the detailed crystal structure of s-octahydroanthracene is not available, the orientation of dibenzo thiophene guest molecules could be determined qualitatively 2 by utilizing the zero-field selection rules and 198

Figure 23. Proton hyperfine structure in electron magnetic resonance absorptions by triplet naphthalene oriented in biphenyl at ~77 K, for H very nearly parallel to the long in-plane axis of naphthalene: (a) pattern obtained at low field (resonance centered at ~104 G); (b) pattern obtained at high field (resonance centered at 3940 G). 199

a 200

Figure 24. Proton hyperfine structure in electron magnetic resonance absorptions by oriented triplets at 'v77 K: (a) phenanthrene pattern observed at a (central) field of 72.3 G for H approximately parallel to the long in-plane axis of phenan threne; (b) dibenzothiophene pattern observed at a (central) field of 32.4 G for H approxi mately normal to the mean molecular plane of dibenzothiophene. 20 6

(a)

10 6 202 quantitatively by analysis of precision zero-field splitting data. These data also confirmed that dibenzo thiophene (and not an impurity) gave rise to the observed spectrum, since the value of the zero-field splitting cor responding to Id 4- E| /he was found at -*77 K to be 0.111057 cm“\ in excellent agreement with the value of 0.1109 it 0.0003 cm"*- reported for dibenzothiophene in EPA , 22 glass at "'77 K (from Am = Z1 studies). The hyperfine pattern obtained for H approximately normal to the mean molecular plane of dibenzothiophene and for the resonance absorption centered at 32.7 G is shown in Figure 24b. It seems clear from these cases that if resolved hyper fine structure can be observed in a triplet resonance ab sorption at high fields, such structure should also be observed in the corresponding low-field experiment (at least at fields above ~/50 G), provided attention is given to the experimental factors cited above. b. Zero Field and Fields Below <~50 G Although hyperfine patterns observed above ~75 G were almost indistinguishable from the patterns ob served in X-band experiments on the same systems, in each case a gradual, progressive loss of resolution of the hyper fine patterns was observed as the field for resonance ap proached zero field. This behavior is illustrated by a series of naphthalene resonance absorptions, shown in 203

Figure 25, obtained during a single experimental run under essentially invariant detection conditions. For this series, H was approximately parallel to the x-axis of naphthalene and its (central) value for resonance was 51.4, 24.9, 13.8, 10.4 and ^0.1 G for patterns (a) through (e), respectively. (With regard to the lowest field ('"'0.1 G), we recall that the total residual laboratory field was less than 0.03 G at the time the experiments described here were carried out.) The associated zero- field splitting, designated |D - E|/hc, was determined in this experiment and found to be 0.114740 cra“^ (not corrected for hyperfine structure effects) pattern (e) was then obtained for a field scan at v = 0.114740 cra"^. Since the pattern did not change abruptly as v was de creased further (to 0.114726 cm“^), it was not possible to characterize a "zero (central) field experiment" of this type any more unambiguously. It is of particular interest that the "zero-field" pattern shows definite evi dence of structure and spans a range of field comparable to the range spanned by the corresponding portions of the preceding patterns. (For £centralj static fields less than half the total field-range of absorption, a complete pattern is, of course, not seen in this type of experi ment.) Similar series of traces for triplet phenanthrene were also obtained and showed the same type of behavior. 204

Figure 25. Magnetic resonance absorption by triplet naph thalene oriented in biphenyl at ^77 K, for H approximately parallel to the long in-plane axis of naphthalene. The (central) fields for resonance are 51.4, 24.9, 13,8, 10.4, and ~0.1 G for patterns (a) through (e), re spectively. Excepting small changes In the spectrometer gain, these patterns were obtained under invariant detection conditions in a single series of measurements. The field-scan rate was uniform throughout. 205 206

In recent publications van der Waals, Schmidt and deGroot have invoked the Kramers theorem (concerning the degeneracy of the electronic states of any odd-spin system at zero magnetic field) to explain their failure to ob serve hyperfine structure in zero-field absorptions from fluorene.38,41 ^ g y attempted to generalize this result by citing the experiments reported by Brandon, Gerkin and Hutchison as additional examples of zero-field experiments in which hyperfine structure was not observed. In fact, BGH reported only studies at (high and) low fields ( ~3 to ~75 G) and not at zero field. Moreover, the present work has shown that resolved proton hyperfine structure is ob servable in low-field resonance absorptions from oriented triplet states, and further, suggests that the zero-field transitions of naphthalene and phenanthrene are indeed split by proton hyperfine interactions. Contrary to the interpretation of van der Waals (as given in References 38 and 41), it does not appear to this writer that the oc currence of hyperfine structure in the zero-field transi tions of naphthalene and phenanthrene would constitute a violation of Kramers' degeneracy, since for both of these molecules even numbers of protons and electrons are in volved in the hyperfine interaction. The reason(s) for the loss of resolution In hyperfine patterns as the field for resonance approaches zero field is not known with certainty. Variation of experimental parameters such as modulation amplitude and crystal tem perature appear to have small but not decisive influences on resolution. Computer simulation has shown that just a small increase in the width of individual hyperfine com ponents can produce loss of resolution comparable to that observed experimentally. Using QCPE Program 83 (adapted for the IBM 1620-1627 Plotter) proton hyperfine patterns could be synthesized by superposition of Gaussian compo nents. Treating the width of the individual hyperfine com ponents as a variable, patterns were generated for H paral lei to the long in-plane axis (x-axis) of phenanthrene. It was found that (for a given set of splitting constants) an increase in component width of approximately one gauss (from the "high-field" width) produced a decrease in reso lution in the generated patterns very similar to that ob served in a series of experimental patterns, for H very nearly parallel to the x-axis of phenanthrene, obtained from 72 G down to 9 G. Whether or not a broadening of the individual hyperfine components below ~75 G is actually responsible for the observed behavior remains to be shown (perhaps by low temperature studies). It must be empha sized that conclusions drawn from the computer simulations cited are of limited value, since the calculations were based on a high-field model of the proton-electron 208 hyperfine interactions and clearly at the lowest fields employed this model is not suitable. In fact, any valid calculation of hyperfine patterns at fields of order 10 gauss is very problematical since perturbation treatment approximations fail completely at these fields. A second possible explanation of the observed loss of resolution below ~50 G is that each observed resonance ab sorption is actually an "average" signal (corresponding to an "average" zero-field splitting) arising from a distri bution of guest molecules which occupy many nearly iden tical but distinct sites in the host structure. (The ef fect suggested here would be much smaller in magnitude than that giving rise to multiplet structure.) For the reasons discussed above (in connection with resolution of overlapped absorptions in the studies of multiplet struc ture and isotope effects), it is precisely near and at zero field, that even a very narrow distribution in zero- field splittings (about an average value), resulting from a distribution in site crystal-fields, could cause a "smearing" of hyperfine structure. A complete investigation of the effects discussed in this and the following section, almost certainly will re quire by itself a research effort of major proportions. 209

3. Effects of Resolved Proton Hyperfine Structure on Precision Determinations of Zero-Field Splittings at Low Fields Upon analyzing precision data cited in the preceding chapter, it was found that particularly in experiments for which resolved proton hyperfine structure was observed, but to a lesser extent in other measurements for proto- 2 phenanthrene, plots of (precision values of) vs did not appear to extrapolate to the "zero-field frequency". That is, for frequencies which should have corresponded to resonance absorption centers occurring at very low but finite fields on the basis of the higher-field data, in fact, less than one half of the resonance absorption sig nal was observed in scans from zero field indicating that the central field for resonance was in effect "below zero field." An example of the behavior described is shown in Figure 26, in which data from an experiment at ^0.093 carf are plotted. Sets of ^'s and H ^ s were determined both above and below the zero-field frequency in this run and in both cases, when the field was scanned for the lowest frequency chosen, less than one half of the resonance ab sorption was observed (indicated by the two points "below zero field"). Thus, resonance absorption centers were not observed at the fields predicted by extrapolation of the behavior at higher fields, but in effect occurred at 0 210

i0.093220

0.093160

0.093100 1 2 34 5 H2 x I0~2 (G2)

Figure 26. Plot of data obtained in precision determina tion of phenanthrene-hjn (in biphenyl) zero- field splitting near ^0.093 cm***-. 211 or "negative" fields, indicating (at least on the basis of 2 a v£ vs fit) that the "true" zero-field frequency should be somewhere between the lowest and next-to-lowest frequency in each case. Similar behavior was observed in data taken at ~0.147 cm and at ''■'0.054 cm“^ with the re sult that, despite the reduction of variation in crystal temperature discussed above, no substantial improvement in the precision of the phenanthrene-h^Q splittings was obtained beyond the precision ( ** i 2 * 10“^) found in the extensive low-field investigation of protophenanthrene in o biphenyl reported by Brandon, Gerkin and Hutchison for which an unfiltered excitation source was used. A clear implication of the present result is that in order to ef fect a reduction in the uncertainties in zero-field splittings determined for a proto species (comparable to that effected for phenanthrene-d^Q here), explicit account will have to be taken of the influence of proton-electron hyperfine interactions. In particular, analysis of such data will have to be made in terms of eigenvalues obtained from solution of a more general spin Hamiltonian (than that employed here) which includes a hyperfine interaction term (see Chapter II). In this regard it should be noted that the fields at which the vs H^^ fit in Figure 26 (and in similar fits to the other data obtained) fails, are indeed those for which the electronic Zeeman 212

Interaction energy is of the order of the proton hyperfine interaction energy (i.e., at ~10 G)• CHAPTER IX SUGGESTIONS FOR FURTHER RESEARCH

A. INTRODUCTION In the course of the investigation of multiplet struc ture in low-field magnetic resonance spectra described in Chapter V, a number of additional interesting studies which might be undertaken occurred to this investigator. Results from some such studies (e.g., 2,3 benzocarbazole in p-terphenyl and chrysene, and studies of impurities in OHA)were presented in Chapter VI. Since only preliminary efforts with regard to others of these problems have been made, they are presented here as potentially worthwhile future researches. It is most convenient to discuss these problems in terms of the mixed crystal systems to be investigated.

B. MIXED CRYSTAL SYSTEMS TO BE INVESTIGATED 1. Picene in p-Terphenyl Inasmuch as picene can be viewed as a "substituted p-terphenyl" it would be of interest to show whether or

213 214 not raultiplet structure is manifested by picene in p- terphenyl and if so how many components appear in the raultiplets. This result would bear on the presumption that fewer inequivalent guest orientations are likely to occur in systems in which the guest and host mole cules have a high degree of "compatibility" from the standpoints of molecular size and symmetry and of intermolecular interactions. In addition, picene is in herently of interest since no report of the experimental determination of its zero-field splittings has been found. (A theoretical value of |d |/he has been calcula ted for picene by Brinen and Orloff.®^) Mixed single crystals of picene in p-terphenyl were prepared in the usual manner and the feasibility of in vestigating this system at 77K, from the standpoints of guest substitution and effective phosphorescent lifetime, was indicated by the observation of intense (green) phos phorescence from such crystals. (Green phosphorescence is expected from picene since the energy of its lowest 84 triplet has been determined to be 2.49 eV. ) Preliminary searches for magnetic resonance absorption from Boule #238 were made by scanning from 0 to ~50Q G at microwave frequencies ranging from ~3050 to ~3860 MHz, for H approximately parallel to the x axis of p-terphenyl. No signals were observed in these scans. These negative 215 results and the theoretical value for D (which was not found until after the experiments described were carried out) suggest that in future experiments on this system searches for (D - E) resonance absorptions should be made at microwave frequencies between ~3900 and 'v4400 MHz.

2. p-Terphenyl-h^ and p-Terphenyl-dj^. in 4,4'-

Dimethylbiphenyl Theoretical calculations of the spin Hamiltonian parameters D and E have been made for a number of linear 85 and angular polyphenyl benzenes but experimental data have been found only for biphenyl, m-terphenyl and tri- 23 phenylbenzene. Since high-purity p-terphenyl-h^ and a sample of p-terphenyl-d^ were available (from the chrysene experiments), consideration was given to studying these species as guest molecules in their own right, and 4,4*-dimethylbiphenyl was chosen as potentially the most suitable host structure from among available compounds. Specifically, 4,4*-diraethylbiphenyl appeared to best satis fy the criteria of relative molecular size, geometry, and singlet and triplet energies with respect to p-terphenyl guest. Although a sample of 4,4'-dimethylbiphenyl was ob tained the proposed investigation remains to be pursued. Determination of the p-terphenyl zero-field splittings would prov.lde a further test of the model employed in 216 theoretical calculations of spin Hamiltonian parameters for the polyphenyl benzenes* Moreover, investigation of the p-terphenyl-diraethylbiphenyl mixed crystal system should yield (especially if both proto- and deutero-p-terphenyl are studied) further information regarding intermolecular interactions in polyphenyl structures (again, can multiplet -structure be observed?) and specifically might aid in clarifying the host isotope shift observed in the zero- field splittings of chrysene-d^ 2 upon perdeuteration of p- terphenyl.

3. p-Quaterpheny1 A very impure sample of p-quaterphenyl was obtained from Eastman but was successfully purified as a potential host structure for chrysene. Although it proved unsatis factory as a host material for chrysene, p-quaterphenyl would be of interest as a guest molecule itself for the same reasons just given in the case of p-terphenyl (theore tical values of D and E have been calculated for this polyphenyl benzene also 85 ). A suitable host structure for p-quaterphenyl was not readily apparent from among available compounds.

4. Investigation of Multiplet Structure in Optical Spectra of Organic Mixed Crystals The application of optical emission and absorption 217

spectroscopy to the mixed crystal systems for which multip- ' let structure has been observed by low-field electron mag netic resonance spectroscopy (LFEPR) seems worthwhile for at least two reasons. First, to attempt to confirm that multiplet structure in electronic spectra is a manifesta tion of the same phenomenon which gives rise to multiplet structure in low-field magnetic resonance spectra. Secondly, data obtained from optical studies of systems previously investigated by low-field magnetic resonance spectroscopy will tend to be primarily supplementary and not redundant, since EPR in contrast to optical spectro scopy yields no information about the multiplet character of the singlet manifold and little information for non- perdeuterated molecules, while on the other hand optical (polarization) results cannot give so quantitative or di rect an evaluation of the magnitudes of orientational in equivalencies as LFEPR. Some specific experiments which could be undertaken, the data desired from each experiment and the kinds of in formation expected from analysis of these data are briefly as follows! a.) Measure the absorption, fluorescence and phos phorescence multiplet component splittings at ^77 K for several prominent bands in each of the spectra of the systems of interest. Confirm that the multiplet splittings 218 are a constant of vibronic band. Compare the number of • components, their relative intensities and splitting-to- transition energy ratios to the analogous values obtained in LFEPR studies at ~77 K. b.) Determine from fluorescence or absorption (or both) spectra the number of components, their relative intensities and splittings as a function of temperature from ^4 K to ~,300 K. These measurements should yield directly information about the relative populations of the inequivalent orientations. Carry out a similar tem perature dependence study for the phosphorescence multip let spectra. These data (referenced to the lowest energy component of the multiplet) can provide estimates of barrier heights and rate constants for site interchanging among the inequivalently oriented sets. c.) Using the known crystal structure and external morphology of p-terphenyl, carry out polarized absorption experiments to obtain polarization ratios for each of the components of the multiplets. The polarization ratios for the propagation direction of the incident light paral lel to the long in-plane axes should exhibit a range of values at least in qualitative agreement with the relative orientation angles observed directly in LFEPR studies. d.) If the multiplet separation is sufficient in any (or all) of the systems proposed for study, experiments in 2X9 which the individual multiplet components are selectively excited,should be attempted. Information obtained by monitoring the relative intensities in the fluorescence and phosphorescence multiplets upon selective excitation combined with the component splitting determinations can lead to an understanding of the relative energies of the , three states being studied (SQ, S-^, T p for the inequivalently oriented molecules. APPENDIX A

* Formulas for g-Tensor Diagonal Elements

220 S x x = 2hv - %[

_ 2hv - %[(D - E)/2] [(D - E)/2gyyj3H10y] + [(D - E)/2gyy0 H ^ ] J + . . . . Syy (H10y 4- H0Iy)

_ 2hv - [

222 223

C C THIS PROGRAM CALCULATES BY ITERATION THE DIAGONAL C ECEMENT S“ 0F THE-_SPTN::TTA'MTXTON1ATT~G“TENSOR USTNU ------C T H E FORMULAS OF HUTCHISON AND MANGUM (J.CHEM. PHYS. C 377” 908-(T9^r) ) ^ INPUT VATUES~AKE THE 1 F S - E S P U m - " C I N ERGS, THE UPPER AND LOWER FIELDS {HMAX,HMIN) FOR C RE'SONA'N'C E“ AT‘ ~ FR EQU EN C Y- GNU ~T E R G i n 7 ATT rNTTYAT ------C APPROXIMATE VALUE FOR G IGAPP), THE VALUE OF THE C BUFTR- MAGNETON {BETATY-AND THE”D 1F F EH E N C F ' H E T W E E N ------C GAPP AND G FOR WHICH THE CALCULATION IS TERMINATED c e e t s t l n f ; ------:------c C TF IT YP E = T j C A T C U CA T E S- G XX, SP CI T = D+l C IF ITYPE=2, CALCULATES GYY, SPLIT=D-E C rF"TTYPE=3, caccu CATES- GZZ, S P L I T = 2 E c C------P R QTG RATCTTR ITT E N~ F 0 R— 3'6T0 "BY- A R T H U R - M7— W T N El C c ------DIMENSION TITLEII 6 ) “TT"FORMAT ( I ZTT5AAT3Er58/TFTam ------1 0 FORMAT (1X,16A4//) Y*TF ORM A T T T H o n 9H G T E N S 0 R ” CA ECU LA T T 0 N 7 7 T 16 FORMAT (1H0,5HGXX =,2XF10-5////» I T T ' O R MAT (YK0T6HB ET A_—=T2 XTE 2 5T677TH0 5HGNU"^7EX7 ------1E15.6//1H07HSPLIT =,2X,E15.6//IH0,6HHMIN =,2X,F8.3// 2TH0 * 6HHMAX- YT2XTF3T377TH0T6RGATP“n"ZX7 F10 . 577YH --- 38HEPSILN =,2X,F10.5////) T 8' F0R M A T (1H075 HGYY- =7 2X*, F8T5777T ------19 FORMAT ( IHO, 5HGZZ =,2 X ,F 8 .5///) READ(5~r8)"TTYPFr(TITLFrT)7”J"=m 6m BErAnGND_;------ISPLIT, HMIN, HMAX, GAPP, EPSILN WRIIE 16,14)------WRITE 16,10) (TITLEtJ), J = 1,16) --— IFtlTYP E7NET0T GO- T 0— 45 ------WRITE(6,40) "4B- FQ R"MA1 ri'H 07 T 6 N W* * T A L ' C U LA T10 N- TYPE- NT3 T~SP EC T F T E D ------1-EXECUTION TERMINATED**#*) C A c m E x n ------:------45 IF (ITYPE.LT.04) GO TO 60 W R I T E r 6 " 7 5 0 1 ------50 FORMAT (1H0,86 H****CALCULAT I ON TYPE INCORRECTLY “ T” SPECTFTECr-EXEUUTTON- rERMrNATELr*W*-) ------C A L L E X I T 60 GO TU (101,102,103), IIYPE------101 WRITE (6,17) BETA, GNU, SPLIT, HMIN,HMAX, GAPP, EPSILN T2TTT) E N 0M ~ = — G T O * B ET A * TH M IN + H M A X 7 ------_____ ------GNUM =* 16.0*GNU-(SPLIT**2/(GAPP*BETA*HMIN) ) 224

1 - ( S P L I T * *2 / ( G A P P * B E T A * H M A X) ) GXX = GNUM/DENOM WRITE (6,T6) GXX ' " ------IF (ABS(GXX-GAPP).LT.EPSILN) GO TO 180 G A P P = G X X G O TO 120 T 0 2 WRITE-' C6Vr7) BETA’; GNU7~SPLTTT H'MTNTTTMAX, GAPP,' EPSILN 130 DENOM = 8.0*BETA*(HMIN+HMAX) G N U M — 16> U^ G N U - t SPL I r * ’(t2 y (G A P P * B E rA * R M IN I I • l-{SPLIT**2/(GAPP*BETA*HMAX)) GVY = GNUM/UfcNOM WRITE (6,18) GYY I F I A B S ( G Y Y - G A P P J T T T VE P S IL N ) G O- T O" 18 0 G A P P = GY Y G U IU 130 103 WRITE (6,17) BETA, GNU, SPLIT, HMIN,HMAX, GAPP, EPSILN 140 D E N G M - BETTA* (HMAX+HMIN ) GNUM = 2.0*GNU-(SPLIT**2/(2.0*GAPP*BETA*HMAX)) 1— (SPLI1**2/(2.0*GAPP*BET A*HMIN)) GZZ = GNUM/DENOM w k i re 16 ,iv) g z z IF (ABS(GZZ-GAPP).LT.EPSILN) GO TO 180 G A P P = G Z Z G O TO 140 1"8“0 S T O P END 225

THIS PROGRAM CALCULATES A LEAST SQUARES FIT FOR N PAIRS OF VALUES X AND Y, PUNCHEO WITH DECIMAL POINTS AT 9 AND 27. BEGIN EACH SET OF DATA WITH A CARD SHOWING THE VALUE OF I TYPE IN COLUMN 2, AND N IN COLUMN !>. THE PARAMETER I TYPE DETERMINES THE FORM TAKEN BY THE LINEAR FIT. IF ITYPE=1,THE INPUT FORMS OF Y AND X ARE USED IF ITYPE=2»LOG Y IS FORMED, AND X IS USED FROM THE INPUT IF ITYPE=3* LOG Y AND LOG X ARE USED IF ITYPE=4,LOG Y AND 1/X ARE USED IF ITYPE=5,LN Y AND 1/X ARE USED IF ITYPE=6,Y IS USED FROM THE INPUT AND X=l/X IF ITYPE=7, LN Y IS FORMED ANO X IS USED FROM THE INPUT • I TYPE MUST BE SET TO ONE OF THE ABOVE VALUES, OR PROGRAM EXECUTION WILL BE TERMINATED. ‘T D I T G M n S 6-72 MAY "BE USED'■FORnrHE''TTrLE"Or”E^CH CALCULATION THE STANDARD DEVIATION AND CORRELATION COEFFICIENT CALCULATIONS HERE ADAPTED FROM HUNTSBEftGER» ELEMENTS OF STATISTICAL INFERENCE. DIMENSION X(200), Y1200), TITLE 1161 DATA BLANK/1H / 10 FORMAT!12,13,16A4,A3/(2F18.8)) 20 FORMAT 11H1»23HMETH0D OF LEAST SQUARES//1X,16A4// 117X, lHX,2"3X,lHY//> 30 FORMAT!1H0,7X,F18.3,5X,F18.8) 40 FORMAT (IX//1X,7HSL0PE = ,2X,F18.8//) bQ FORMAT tlH HHINTERCEPT =,2X,F18.8/J R E A L M U READ!5,10)I TYPE,N»(TITLE(J),J=1»16)»END,(X!I),Y(I) ,I=1,N) WRITE 16,20) (TITLE IX), J - 1,12) WRITE (6,30) (X {I), Y (I), I = 1, N) X 5 U M = 0. Y S U M = 0. S t G X Y - 0. S X S Q = 0. X Y D E V = 0 . 0 X 2 D E V = 0 . 0 Y2D£V=0.O "" ...... IF! ITYPE.NE.OJGO T 0 7 7 7 WRITE(6,102) 102 FORMAT (1H0,76H****CALCULATI0N TYPE NOT SPECIFIED l-EXECUTTON’ "TE71MINATED***’*") ...... "" C A L L E X I T 777^^nWP"F.XTT0QTCa^0^TCf ■'f'YVI'rtT ' '"r’Y A\ A A “ “ “ ' “ ■ . 1 ■ ■ !■.■■■! WRITE(6»103)

T O 3 F"0 R~M A T (T h O f~85H****CALCULAnuN TYPE INCORRECTLV" ISPECLFIED-EXECUTIQN TERMINATED****J 226

CALLEXIT------200 GO TO (201,202,203,204,205,206,207,200,209,210),ITYPE 201 ‘WRTTE'CBtffOri' ..... 601 FORMAT(1HI,10HX AND Y ARE LINEAR) GO TO 300 202 WRITE(6,602) 6 02' ''F'GRTIA_TT11TL7TGHY^C0iry7^— rS~T DO 402 1=1,N 402 rTrF^"A'nJGYOTY ( I)') ...... GO TO 300 203 WRI 1 E16,603J 603 FORMAT(1H1,13HY=L0GY,X=LQGX) D 0 4'0 3 I = 1, N Y (IJ =AL0G10IY( I) ) 403 X {I)=ALUG10(X(I) 1 GOT 0300 204 WRITE(6,604) 6 04 FORMAT(1HI,i4HY=LOGY,X=1-0/X) DO 404 r=l,N Y (I) = AL0G10(Y( I) ) 404 X tI) = 1•0/X( I ) GOT0300 "205 WRTT£T6,6'03> 605 FORMATt1H1»34HARRHENIUS CALCULATION,Y=LN Y,X=1/X) DU 403T = r 7 N YCI )=ALQG(Y(I)) 403 X I 17=1. 07X11) --- GO TO 300 206 WRITE(6,606) ...... 606 FORMATtIH1,17HY IS LI NEAR,X=l/X) DO 406 I = T,N 406 X (I>=l.O/X(I)

"207 W1TET6,60/) "" * 607 FORMAT(1H1,35HPHOSPHORESCENT LIFETIME CALCULATION) ------d0-^07- 1^17^

407 Y (I)=ALOG(Y (I)) 20B~CONTl“NUE ' : 209 CONTINUE Z1D~C'0NTTNUE------G O TO 3 0 0 ~3oo~vfRiTrr6"T4T)— rx (t t t y rrrri= i t n j------44 FORMAT (IH0,27HC0NVERTE0 VALUES OF X AND Y/.1H0, ‘"1 16X Tl HXT2TX7TFi Y7TF107 CSX 7F13T875XTFT0T 8T)------D O 12 I = 1, N 5T5UM = X'SuM + x( n : YSUM = YSUM + YII) 'STGXY~^~STGXY + ” x m W ( T ) — ------sxso = sxsq + xm*x( i) 227

12 C O N T I N U E R = F L O A T IN) M - V R ^ S T G X Y - XSUM*Y S UM77 TR * S XS“U - XSUM*XSTJMl B = (SXSQ*YSUM - XSUM*SIGXY)/(R*SXSQ - XSUM*XSUM) “ S T A NO A RD 0 E VI AT f O N C A L C U L A TVON X 8 A R = X S U M / R Y 6 A R = Y “S U M 7 R DO 14 1=1,N X*YDT£V=~XYIT (£VT~CX~Cl ) -XBA'RT^rYTIT^tLATH X2DEV=X2DEV+(X( I )-XBAR)**2 Y 2 0 EV=^Y2 D E V + ( Y I T ) - Y BA R ) **~2 14 CONTINUE S 2 YX=TY2D X YD £ V ) / T R - 2 7 1 S2B=S2YX» I X2PEV»R»XBAR*XBAR)/{R*X2DEV) ______

SDM=SQRT{S2M) S 0 B= S QRTTS 2 8 ) IF t ITYPE.NE.05)G0 TO 15 r£'M'P=M """ " ” TEMP=-l.987*TEMP • WRITEt6,104)TEMP 104 FORMAT!1H0,19HACTI VAT I ON ENERGY= ,F12.4,15H CALORIES/MOLE) 15 IF! IT Y P E . N E . 07) G O TO 13 T E M P = M r E M P = - l . / M WRITE!6,107)TEMP 107 FORMAT!1H0,10HLIFETIME =,F12.4,5H SEC) T E M P = M T E M P = M * M TEMP=SDM/(M*M) WRI TE !6 , " 3 0 7 ) TE M P 307 FORMAT!IHO,25HUNCERTAINTY IN LIFETIME =,2X,F6.3//> 13 CONTINUE WRITE 16,40) M WRITE !6,50) B WRITE(6,90)SDM,SDB 90 FORMAT I I HO , 2 9 F I S T ANDARD DEVIATION OF SLOPE =,2X, IF18.8/1H0,33HSTANDARD DEVIATION OF INTERCEPT =,2X, 2 F 1 8 . 8 ) LINEAR CORRELATION COEFFICIENT CALCULATION V A RY 2 = Y 2 0 £W ( 1 . 0 7 T R - 1 T 0 D VARX2=X2DEV*(1.0/IR-1.0)) SX^S'QRn VARK 2T"----- SY=SQRT(VARY2) ------

WRITE(6,91)CC ~9 1~F OR MAT (IVi0 V32HLTNEAR C 0 R T < E C A T IG N ~ C O E f F ICTEM f'^7 2XV F3T4T IF(END.NE.BLANK) CALL EXIT G O“ TO 11 “ END 228

LIMITING LIFETIME 360 PROGRAM

TH I S" P R O G R A M " L E A S T S QU AR E5~ FIT S“ PK0"SP RORESCE fTl------LIFETIME (TAU) VS TEMPERATURE (TEE) WITH FUNCTIONAL D'E PE N D ENGE“ ACC0ROTNG-T0~KTNTE TTC“ SCTTEME TN~WHTCH LIFETIME IS GOVERNED PRINCIPALLY BY THERMALLY ACTTVATED rRTPLET- ENERGTTPTAArSTER, "T ITEATTfiGTrHE ------LOW-TEMPERATURE LIMITING LIFETIME (TAO) AS A “VAR'I A BL“E “ 0 V ER” T H E- 1NTERVAt T AO L N T n T A T r T O ------TO TAOFIN(AL) BY INCREMENT DELTAO• PROGRATtfTR I TT E N “ B Y"AR T H U RT ' M T T TNER7 I 95T"'A0GCrST V

DIMENSION X1200), YI 2 0 0 ), TEEI200), TAU1200), TITLE (16) “D AIA-BXANK7XIT-7 ------10 FORMAT!15,16A4,A3/(2F18.8)) “ 2 0 FORMATTVHTT 2T5fi M E THO0— OF"XE'AST_ S WA R T S 7 7 T X , 16"A477 117X,1HX,23X,1HY/) “ 3 0~FlJRMmTTOT7XrFT8Tff75XrFTST81 ------4 0 FORMAT (1X//1X,7HSLOPE =,2X,E10. 8 //) "3 0 F 0R MAT f 1H~ 11 H INTERCEPT” ,"27VFr8T07) ----- “ ------11 READ15,10)N,(TITLE (J >,J=1,16),END,tTEE(I), TAUlI),1 = 1,N) W R I T E ( 6T2 0 ) { TIT LE T J ) , ~ ~ J “^ ~ 1 7 T 2 ) ------WRITE (6,30) (TEE(I), TAUtI), I = 1,N) ro i ' W R T T F r 6r r o ^ ) ------102 FORMAT (1H1, 29HLIMITING LIFETIME CALCULATION/) 112 F O R M A T - ( 5X7" 3FT0T3T READ (5,112) TAOIN, DELTAO, TAOFIN TAT)=-TAOTN 110 DO 51 1=1,N XTrr-^— TTZTEEllT ------Y (I) = ALOG((1«/TAU(I)) - (I./TAO)) "5 r “C O N T T N U E ------! 211 FORMAT (IHO, 19HLIMITING LIFETIME =,2X,F10.2) 210“ H RTTE r5T2Tn T AO 2 Q 0 W R I T E (6 ,44) (X(I),Y(I),1=1,N) ~4'4~FUR'MATTTH0T27'hGOWV'E R TED” AUDITS V "OF X- AND"Y7, IHO------1,16X,1HX,23X,1HY/1H0/(8X,F18.8,5X,F18.8)) XSUIT"=~"0^ ------Y S U M = 0 . STG7Y~="07------:------S X S Q = 0. S r m E V ^ D T Q ------:------X 2 D E V = 0 . 0 Y2DEV“=0T0------;------:------DO 12 I = 1» N 229

XSUM = XSUM + X(I) YSUM = YSUM + Y11) SIGXY = SIGXY + X(I)*Y(I) SXSQ = SXSQ + X(I)*X(I) 12 CONTINUE R = FLOAT (N) SM= (R*SIGXY - XSUM*YSUM)/(R*SXSQ - XSUM*XSUM> B = (SXSQ*YSUM - XSUM*SIGXY)/(R*SXSQ - XSUM*XSUM) STANDARD DEVIATION CALCULATION XBAR=XSUM/R YBAR=Y'SUM7R 0 0 1 4 I = LiN XYDEV=XYDEV+(X tI)-XBARf*TYtI)-YBAR) X2DEV=X2DEV+(X(I)-XBAR)**2 Y 2D EV= Y 2D E V + ( Y ( I )-Y BARJ **2 14 CONTINUE S2Y X=(Y2DEV—SM*XYDEV)/(R-2•) S2B=S2YX*(X2DEV+R*XBAR*XBAR) /(R*X2DEV) S2M=S2YX7X2DEV SDM=SQRT(S2M) SD B - S W T f S T B l 13 CONTINUE WRITE ( 6 , 40THSM ' WRITE (6,50) 8 WRITE(6,90)SDM,SDB 90 FORMAT(IHO,29HSTANDARD DEVIATION OF SLOPE =,2X, IF 18 . ByiHO,33HST ANDARD- DEV I AT I ON OF INTERCEPT =,2X, 2F18.8) ITTN^E'AirTO^R^rA^TTOlOrOEl^ETirrETTTnCATrClJrA'TTOTsi VARY2=Y2DEV*(1.0/(R-1.0)) VAR X2 = X’2 0‘EV*T1.0 / TR-F70T) SX=SQRT(VARX2) r^^TTTTVAiTO") :------CC=SM*(SX/SY) write f6 7 9 i rcc 91 FORMAT(IHO,32HLINEAR CORRELATION COEFFICIENT =,2X,F8.4//) 126 TAO = TAO + DELTAO IF ( E N D . N E . B L A N K ) C A L L EXIT rF TTTiTrrirrTTAijFi n j go ro~n G O TO 110 END «

REFERENCES CITED

1. R. W. Brandon, R. E. Gerkin, and C. A. Hutchison, Jr., J. Chera. Phys. 37, 447 (1962)

2. R. W. Brandon, R. E. Gerkin, and C. A. Hutchison, Jr., J. Chera. Phys. 4l, 3713 (1964) 3. E. Zavoisky, J. Phys. U.S.S.R. 9, 211, 245, 447 (1945)

4. G. N. Lewis and M. Kasha, J. Ara. Chera. Soc., 66, 2100 (1944) — 5. G. N. Lewis and M. Kasha, J. Am. Chem. Soc. 67, 994 (1945)

6. G. N. Lewis and M. Calvin, J. Ara. Chera. Soc., 67, 1232 (1945) — 7. P. Frohlich, L. Szalay, and P. Szor, Acta. Chera. Phys. Univ. Szeged (Hungary) 2^, 96 (1948) 8. G. H. Lewis, M. Calvin and M. Kasha, J. Chera. . Phys. 17., 804 (1949) 9. J. A. Weil, "Search for the paramagnetic resonance absorption of an organic dye in its phosphorescent state," thesis, University of Chicago, 1955, p.2. 10. S. I. Weissraan, private communication to C. A. Hutchison, Jr.

11. C. A. Hutchison, Jr. and B. W. Mangum, J. Chem. Phys. 29., 952 (1958) 12. C. A. Hutchison, Jr. and B. U. Mangura, J. Chem. Phys. 34, 908 (1961)

230 231

13. J. H. van der Waals and M. S. de Groot, Mol. Phys. 2, 333 (1959) 14. J. H. van der Waals and M. S. de Groot, Mol. Phys. 3, 190 (1960) 15. W. A. Yager, E. Wasserman, and R. M. R. Cramer, J. Chem. Phys. 37., 1148 (1962) 16. E. Wasserman, L. C. Snyder, and W. A. Yager, J. Chem. Phys. 41, 1763 (1964) 17. B. Smaller, J. Chem. Phys. 37, 1578 (1962) 18. G. V. Foerster, Z. Naturforsch. 18A, 620 (1963) 19. M. S. de Groot and J. H. van der Waals, Mol. Phys. 6, 545 (1963) 20. C. Thomson, J. Chem, Phys. 4 1 , 1 (1964) 21. J. S. Brinen and M. K. Orloff, J. Chem. Phys. 45, 4747 (1966) — 22. S. Siegel and H. S. Judeikis, J. Phys. Chem. 70, 2201 (1966) — 23. J. S. Brinen, J. G. Koren, and W. G. Hodgson, J. Chera. Phys. 44, 3095 (1966) 24. M. S. de Groot and J. H. van der Waals, Physica 29, 1128 (1963) 25. M. S. de Groot and J. H. van der Waals, Mol Phys. 6, 545 (1963) 26. R. E. Jesse, P. Biloen, R. Prins, J. D. W. van Voorst and G. J. Hoijtink, Mol. Phys. 6_, 633 (1963) 27. J. H. van der Waals and G. ter Maten, Mol. Phys. 8, 301 (1964) 28. M. S. de Groot, I. A. M. Hesselman, and J. H. van der Waals, Mol. Phys. 13, 583 (1967) 29. A. W. Hornig and J. S. Hyde, Mol. Phys. 6_, 33 (1963) 232

30. A. Schraillen and G. V. Foerster, A. Naturforsch, 16A, 320 (1961) ~ ~

31. J. S. Vincent and A. H. Maki, J. Chera. Phys. 39, 3088 (1963) —

32. J. S. Vincent and A. H. Maki, J. Chera. Phys. 42, 865 (1965) — 33. S. W. Charles, P. H. H. Fischer and C. A. McDowell, Mol. Phys., 9, 517 (1965) 34. 0. H. Griffith, J. Phys. Chera. 69,, 1429 (1965)

35. J. S. Vincent, J. Chem. Phys. 47, 1830 (1967)

36. M. S. de Groot, private communication to Y. Gondo and A. H. Maki, J. Chem. Phys. 50, 3270 (1969)

37. M. S. de Groot, I. A. M. Hesselman, and J. H. van der Waals, Mol. Phys., ID, 91 (1965)

38. J. Schmidt and J. H. van der Waals, Chera, Phys. Letters 2^, 640 (1968)

39. E. T. Harrigan and N. Hirota, J. Chem. Phys. 49, 2301 (1968) — 40. D. S. Tinti, M. A. El-Sayed, A. H. Maki and C. B. Harris, Chem. Phys. Letters 3^ 343 (1969)

41. The Triplet State, A. B. Zahlan et al. Eds. (Cara- ' bridge University Press, Cambridge, England 1967) p. 63

42. R. E. Gerkin and P. Szerenyi, J. Chera. Phys. 50, 4095 (1969) 43. R. E. Gerkin and A. M. Winer, J. Chera. Phys. , 47, 2504 (1967) 44. R. E. Gerkin and A. M. Winer, J. Chera. Phys. > 50, 3114 (1969)

45. R. E. Gerkin and A. M. Winer, J. Chem. Phys. » 51» 1664 (1969)

46. K. W. H. Stevens, Proc . Roy. Soc . (London) A 214, 237 (1952) 233

47. A. Carrington and A, D. McLachlan, Introduction to Magnetic Resonance (Harper and Row, Publishers,-Tnc., New York,~i>l. Y., 1967)

48. S. P. McGIynn, T. Azumi and M. Kinoshita, The Triplet State (Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1969)

49. H. F. Hameka and L. J. Oosterhoff, Mol. Phys., 1, 358 (1958) 50. M. Kasha, J. Chem. Phys. 3£, 929 (1948)

51. P. Mabille and N. P. Buu-Hoi, J. Org. Chera., 25, 1937 (1960) — 52. D. F. Bender, E. Sawicki and R. M. Wilson, Anal. Chem. 36, 1011 (1964)

53. W. Davies and G, N. Porter, J. Am. Chem, Soc., 1957, 4967 54. M. Kasha, Radiation Res. Suppl., Z, 243 (1960)

55. M. Orchin, J. Am. Chera. Soc. 66, 535 (1944)

56. J. N. Sherwood and S. J. Thomson, J. Sci. Instr., 37, 242 (1960) 57. V. N. Vatulev and A, F. Prikhot*ko, Fiz. Twerd. Tela., 7, 42 (1965) * 58. Yu. V. Mylch and M. A. Isneva, Dokl. Akad. Nauk. S.S.S.R., 162, 326 (1965) 59. L. W. Pickett, Proc. Roy. Soc. 142, 333 (1933)

60. E. Hertel and G. H, Romer, Z. Physik Chem. B21, 292-6 (1933) --- 61. H. M. Rietveld, Ph.D. Thesis, University of Western Australia (1962). Data from this thesis were graciously communicated to us by Dr. E. N.| Maslen, Department of Physics, University of Western Australia, Nedlands, Western Australia. i 62. Barker Index of Crystals, Vol. 12 (1956) !

i 234

63. J. S. Brinen and M. K. Orloff, Chera. Phys. Letters, 1, 276 (1967) 64. S. G. Hadley, H. E. Rast and R. A. Keller, J. Chera. Phys., 39, 705 (1964) 65. E. T. Harrigan and N. Hirota, J. Chera. Phys., 49, 2301 (1968) 66. See reference cited in footnotes 1-3 in the preceding paper* 67. M. Kinoshita, K. Stolze and S. P. McGlynn, J. Chera. Phys., 48^, 1191 (1968) 68. W. Siebrand, J, Chem. Phys. 47, 2411 (1967) 69. G. Porter and M. Windsor, Proc. Roy. Soc., 245A, 238 (1958) 70. D. S. McClure, J. Chera. Phys. 19, 670 (1951) 71. D. P. Craig and 1. G. Ross, J. Chera. Phys. 47 2946 (1967) ~~ 72. J. S. Brinen and W. G. Hodgson, J, Chera. Phys . 47, 2946 (1967) 73. R. E. Kellogg and N. C. Wyeth, J. Chera. Phys. , 45, 3156 (1966) 74. Y. Marechal, J. Chera. Phys., 44, 1908 (1966)

75. M. H. Perrin and M. Gouterman, J. Chera. Phys. * ££» 1019 (1967) 76. R. M. Hochstrasser and G. J. Small, Chera. Cororaun. 5, 87 (1965) 77. R. M. Hochstrasser and G. J. Small, J. Chem. Phys., 45, 2270 (1966) 00

. H. P. Cleghorn, Mol. Phys. 13, 93 (1967) 79. R. M. Hochstrasser and G. J. Small, J. Chem. Phys. 48, 3612 (1968) 235

80. N. Hirota, J. Chem. Phys. 43, 3354 (1965)

81. C. A. Coulson and C. W. Haigh, Tetrahedron, 19, 527 (1963) —

82. A. Hargreaves and S. H. Rizvi, Acta. Cryst. 15, 365 (1962) —

83. J. S. Brinen and M. K. Orloff, Chera, Phys. Letters 1, 276 (1967)

84. R. N. Nurmukhametov, Russian Chem. Rev., 35, 469 (1966) —

85. M. K. Orloff and J. S. Brinen, J. Chem. Phys., 47, 3999 (1967) —

Studies of Oriented Aromatic Hydrocarbons in the Phos­ Phorescent State at Low Magnetic Fields

Studies of Oriented Aromatic Hydrocarbons in the Phos Phorescent State at Low Magnetic Fields