SPECTROSCOPIC INVESTIGATIONS OF ELECTRONIC TRANSITIONS IN
CERTAIN ACENE QUINONES
RODGER N. CAPPS
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
"Training is everything. The peach was once a
bitter almond; cauliflower is nothing but cabbage
with a college education."
-- Pudd'nhead Wilson's Calendar
The author is indebted to Professor Martin T. Vala for his initiation
and support of this work. His many suggestions and comments, as well as
his never-failing optimism that things would eventually work out, were
of incalculable value in the course of this research.
The expertise contributed by the electronics shop personnel was a
help in carrying out these experiments, as was the assistance in construc-
tion of equipment contributed by Mr. E.C. Whitehead.
Dr. Glenn Boutilier and Mr. Gary Walden are gratefully acknowledged
for the performance of lifetime measurements included in these experi-
ments. The author would also like to thank Professor J.D. Winefordner
for the loan of the Eimac lamp, and Dr. E.J. Gabbay for the daunorubicin.
Numerous people must be acknowledged for moral support. The author
would like to thank his fellow graduate students, Ed. Voigtman, Joe
Baiardo, Dave Powell, and Bob Brittain for their camaraderie. He would
also like to acknowledge his parents for support and encouragement.
J.M. Nicovich, Justine Strand, Bette Ackerman, and Dr. S.O. Colgate are
also gratefully acknowledged for their moral support.
Finally, the author would like to thank Adele Koehler for her able
assistance in preparing this manuscript in a finished form.
TABLE OF CONTENTS
. . . iii
LIST OF TABLES . . .
LIST OF FIGURES . . .
ABSTRACT . . .
I INTRODUCTION . . .
II EXPERIMENTAL DETAILS . .
Materials . . .
Monochromators and Light Sources . .
Detection and Recording System . .
Shpolskii Matrix Experiments . .
Lifetime Measurements . .
Polarized Excitation and Emission Spectra .
III THE ANTHRAQUINONE SYSTEM . .
Introduction . ..
Shpolskii Systems and Impurity-Lattice Interactions.
Experimental .. . .
Results and Discussion . .
Analysis of Phosphorescence Spectra .
Temperature-Dependent Emission .
IV QUINIZARIN AND DAUNORUBICIN . .
Introduction . . .
Theory of Photoselection . .
Experimental . . .. .
Results and Discussion . .
Infrared Spectra . .
Lifetimes . . .
Absorption and Luminescence Spectra .
Photoselection and Assignment of Transitions .
. . 97
V SUMMARY AND CONCLUSIONS . . 115
REFERENCES.................. .. ....... .117
APPENDIX FREQUENCIES OF NORMAL MODES OF VIBRATION OF
9,10-ANTHRAQUINONE . . .. 122
BIOGRAPHICAL SKETCH ....................... 124
LIST OF TABLES
I Experimental Equipment and Manufacturers (Lifetimes) 11
II Phosphorescence Vibrational Analysis for Anthraquinone
in Hexane at 10 K . . 32
III Phosphorescence Vibrational Analysis for Anthraquinone
in Heptane at 10 K . . 35
IV Representations of the Normal Modes of Vibration of
Quinizarin . . 91
V I.R.-Active Fundamental Modes of Vibration of Quinizarin 93
VI Fluorescence Vibrational Analysis for Quinizarin in
Octane at 14K . ..102
LIST OF FIGURES
1. Block Diagram for Shpolskii Matrix Experiments .
2. Block Diagram for Lifetime Measurements .
3. Block Diagram of Experimental Set-Up for Polarization
Experiments . . .
4. Mechanism of the 3Blg 1A Phosphorescence in
Anthraquinone . . .
5. Phosphorescence of Anthraquinone in Hexane at 140K in
the Origin Region . .
6. Phosphorescence of Anthraquinone in Hexane at 540K in
the Origin Region . . .
7. Phosphorescence of Anthraquinone in Hexane at 740K in
the Origin Region . . .
8. Phosphorescence of Anthraquinone in Hexane at 940K in
the Origin Region . . .
9. Phosphorescence of Anthraquinone in Hexane at 1200K in
the Origin Region . . .
10. Phosphorescence of Anthraquinone in Hexane at 1590K in
the Origin Region . . .
11. Phosphorescence of Anthraquinone in Heptane at 14K
in the Origin Region . . .
12. Phosphorescence of Anthraquinone in Heptane at 74K
in the Origin Region . . .
13. Phosphorescence of Anthraquinone in Heptane at 1670K
in the Origin Region . . .
14. Phosphorescence of Anthraquinone in Octane at 14K in
the Origin Region . . .
15. Phosphorescence of Anthraquinone in Octane at 74K in
the Origin Region . . .
16. Phosphorescence of Anthraquinone in Pentane at 14K
in the Origin Region . . 55
17. Phosphorescence of Anthraquinone in Pentane at 74K
in the Origin Region . .... 67
18. Plot of Activation Energies of Temperature-Dependent
Emission of Anthraquinone in Hexane. . ... 69
19. Phosphorescence of Anthraquinone in Hexane at 140K .. 75
20. Phosphorescence of Anthraquinone in Hexane at 360K 76
21. Phosphorescence of Anthraquinone in Hexane at 760K 77
22. Structures and Axis System of Anthraquinone, Quinizarin,
and Daunorubicin . . 82
23. Illustration of Relative Polarization for an Electronic
Transition . . .. 87
24. Plot of A vs. C for Quinizarin in Hexane .... .. 95
25. Plot of A vs. C for Quinizarin in Methanol ... 96
26. Fluorescence of Quinizarin in Hexane at 140K ...... 98
27. Fluorescence of Quinizarin in Octane at 14K ...... 99
28. Fluorescence of Quinizarin in EPA at 770K. .. 106
29. Plots of Absorbance and Relative Polarization of
Quinizarin in EPA. .. . . 108
30. Plot of Polarization vs. Wavelength for Quinizarin
Fluorescence. . .. 110
31. Plot of Excitation Polarization of Anthraquinone as
Compared to Quinizarin . . 111
32. Plots of Absorbance and Relative Polarization of
Daunorubicin in Alcohol-Water-Glycerin ... 114
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
SPECTROSCOPIC INVESTIGATIONS OF ELECTRONIC TRANSITIONS IN
CERTAIN ACENE QUIRONES
Rodger N. Capps
Chairman: Martin T. Vala
Major Department: Chemistry
The positions and natures of the lowest excited emitting electronic
levels in 9,10-anthraquinone and two of its derivatives, quinizarin and
daunorubicin, have been characterized. In the first case, that of
anthraquinone, highly resolved emission spectra at low temperatures were
obtained in Shpolskii matrices of n-pentane, n-hexane, n-heptane, and
n-octane. In the latter three matrices, the origin of the purely
electronic transition is observed very weakly or not at all. In the
n-pentane matrix, a variety of both non-totally symmetric and totally
symmetric modes are present. In hexane, heptane, and octane, the phos-
phorescence of anthraquinone is shown to contain progressions built upon
vibrations of types blu, b2u, and b3u, with the most intense peaks being
those of type blu. Analyses of the phosphorescence emission of anthra-
quinone are presented which are shown to be consistent with the assign-
ment of the lowest triplet level of anthraquinone as 3Blg. The results
obtained in this work tend to corroborate those of Khalil and Goodman,
but include several additional spectral features not reported by them.
A temperature-dependent emission at energies higher than the origin of
the purely electronic transition is also obtained in n-hexane, heptane,
and octane, but not in n-pentane. Based upon the behavior of this
emission as a function of temperature and solvent, it is suggested here
that the emission is a manifestation of the anthraquinone-lattice inter-
action which allows the spin and parity forbidden electronic transition
to gain radiative properties.
The nature of the lowest excited emitting electronic level in quin-
izarin is characterized by lifetime measurements, photoselection measure-
ments, and low temperature Shpolskii matrix emission spectra. The
species involved in the absorption and emission is shown to be the
quinizarin molecule, rather than an aggregated form, by concentration
studies. An analysis of the absorption spectrum of quinizarin in the
infrared and far infrared regions is also presented.
Photoselection measurements were made upon both quinizarin and
anthraquinone. It was observed that the relative polarizations of
transitions in the two molecules were the same. Based upon the axis
system chosen for quinizarin, electronic transitions occurring in the
ultraviolet and visible absorption spectra of quinizarin have been
Highly resolved, low temperature emission spectra were recorded for
quinizarin in n-hexane, n-heptane, and n-octane. It is suggested that
the emission in n-hexane and n-heptane arises from two sites, at least
one of which is severely distorted. An analysis of the vibrational pro-
gressions for quinizarin in n-octane is presented which is consistent
with the observed ground state I.R. frequencies.
Based upon the measured emission lifetime of 6.5 1 nsec, photo-
selection measurements, and analyses of low temperature Shpolskii matrix
spectra, the lowest excited emitting electronic level in quinizarin is
assigned as a ir,i* singlet arising from a charge-transfer interaction of
the hydroxy substituents with the benzene ring of the anthraquinone
Photoselection measurements upon daunorubicin revealed that the
polarization of major electronic transitions in this molecule are the
same as those of quinizarin. Based upon these results, it is concluded
that the polarizations of transitions in this molecule, as assigned
byGabbayand co-workers, are incorrect.
The 9,10-anthraquinone molecule and its related compounds are of
chemical importance from both an applied and a theoretical point of view.
It is an excellent subject for studies of heteroatom effects upon ex-
tended r-electronic systems, of spin-orbit coupling mechanisms, and of
possible intramolecular exciton interactions. Its high degree of sym-
metry allows an excellent correlation of theory and experiment through
the methods of group theory. Anthraquinone is also the parent molecule
of a large number of compounds which find use in the dye industry.
Studies of its electronic structure and photochemical reactivity might
find use in understanding the photodegradation of dyes. It has also
been demonstrated that this compound and certain of its derivatives dis-
play a mutagenic activity upon intestinal bacteria found in humans (1),
while other compounds having the 9,10-anthraquinone molecule incorporated
into their skeletal structures show promise as anti-leukemia drugs
The earlier investigations of the nature of electronic transitions
in 9,10-anthraquinone were concerned with visible and near ultraviolet
absorption spectra (4, 5). Labhart carried out investigations of the
excited states of substituted anthraquinones using polarized absorption
spectra (6, 7).. Using polarized single crystal spectra, Sidman established
the polarizations of the electronic transitions in anthraquinone (8).
Photoselection experiments (9, 10) and a determination of the T1 < S
absroption moment (11) showed that the phosphorescence of anthraquinone
was in-plane polarized, n-n* type.
In reference (11) it was concluded that the lowest triplet level of
anthraquinone was A Later studies (12, 13) indicated that this
assignment was incorrect. The earliest reported Shpolskii matrix study
(14) was inconclusive, since no ground state vibrational frequencies were
available at that time. A later study (15) at 77K indicated that the
lowest triplet was 3Blg. However, there is normally a great difference
in resolution of spectra run at 770K and those obtained at temperatures
approaching O'K. One of the objectives of this study was to try to
determine the nature of the lowest triplet state of anthraquinone from an
analysis of the vibrational structure of the phosphorescence spectrum.
Quinizarin, which is a 1,4-disubstituted hydroxyanthraquinone, has
not been as extensively studied. Absorption and fluorescence spectra
of quinizarin in the vapor phase and in ethanol solutions were recorded
by Borisevich and Gouzinskii (16). A later work (17) reported visible
absorption and fluorescence spectra in the vapor phase, along with a
partial analysis of the I.R. spectrum.
Shcheglova, Shigorin, and Dokunikhin (18) recorded the luminescence
spectrum photographically at 770K in n-octane, but no attempt at analysis
was made. An early work (5) on substituent effects upon the visible and
ultraviolet absorption spectrum of anthraquinone included a study of
quinizarin. ElEzaby et al. (19) studied the absorption spectra of
anthracene, anthraquinone, and several hydroxy-anthraquinones in dif-
ferent solvents in order to determine the nature of the electronic
transitions in these compounds. Platonova, Popov, and Smirnov (20)
concluded from dichroic ratios of absorption spectra of quinizarin in
polyvinyl alcohol films that the relative polarizations of electronic
transitions in quinizarin were the same as those of anthraquinone.
The above works indicated that the behavior of quinizarin was
peculiar in several respects. A discussion of the effect of substituting
OH groups upon anthraquinone by Abrahamson and Panik (21) indicated that
the n-electronic levels of the molecule should be shifted relative to
anthraquinone. Experimentally (17), a very large wavelength shift in
emission was observed. Also, the emission from the molecule was appar-
ently fluorescence, rather than phosphorescence (17). Normally, carbonyl
containing compounds will exhibit phosphorescence (22).
The present work was undertaken to resolve some of the ambiguities.
The daunorubicin was also included in this study to determine whether
the assignments of electronic transitions in an earlier work were cor-
rect (23). The results listed in this work indicated a polarization of
electronic transitions contrary to results obtained here for quinizarin,
suggesting that chromophores present in daunorubicin had a very large
effect upon the electronic energy levels of the anthraquinone skeleton.
This work, then, presents a study of electronic transitions in
anthraquinone and two of its derivatives, quinizarin and daunorubicin,
using room temperature luminescence and absorption spectra, Shpolskii
matrix emission spectra, I.R. spectra, and polarized excitation spectra.
The anthraquinone used in these experiments was synthesized by ring
closure of ortho-benzoyl benzoic acid with concentrated H2SO4 (24).
Technical grade ortho-benzoyl benzoic acid was extracted from an
ether solution with 3 M NaOH solution. The sodium salt solution of the
acid was then filtered, and the ortho-benzoyl benzoic acid was regenerated
with 3 M HC1 solution. This was followed by suction filtering and
washing with distilled water. The ortho-benzoyl benzoic acid was then
recrystallized twice from ethanol and water. The melting point was
127-1280C, in good agreement with the literature value of 1280C (25).
The ortho-benzoyl benzoic acid was placed in a large excess of 96%
H2SO4. The mixture was then heated to 150-1600C, and maintained in this
temperature range with vigorous stirring for approximately two hours.
The reaction mixture was then poured over ice and allowed to stand
overnight. It was then mixed thoroughly with saturated NaHCO3 solution,
suction filtered, and thoroughly washed with more NaHCO3 solution and
distilled water. Two recrystallizations from an acetone-chloroform
mixture, followed by four vacuum sublimations, were deemed adequate to
insure the purity of the anthraquinone. An I.R. spectrum was run to
insure that the compound formed was indeed anthraquinone. Since the
compound sublimes, rather than melts, purity was checked by thin-layer
chromatography. Only one band was observed.
The quinizarin used was received from Eastman Kodak, and was vacuum
sublimed twice before use. The melting point was 200-202C. The litera-
ture value is also 200-202C (25).
The daunorubicin was a gift from Dr. E.J. Gabbay, and was used as
Aldrich Gold Label (purity > 99%) n-hexane, n-octane, and 2-methyl
butane (isopentane) were used as received. Fisher analytical grade
n-pentane and Mallincrockt analytical grade n-heptane were used as re-
ceived, along with absolute ethanol and anhydrous ether. All solvents
were checked for absorption and emission in the spectral regions of
interest, and found to be satisfactory.
Monochromators and Light Sources
Heath EU-700 monochromators were used in this research. These are
0.35 meter, single pass, modified Czerny-Turner type mount monochromators
with folding mirrors which provide a common optical axis for entrance
and exit beams. The excitation monochromator was supplied with a plane
grating ruled with 1180 lines/mm and blazed at 3500 A, while the
emission monochromator grating was blazed at 5000 A.
The aperture ratio of the monochromator is f/6.8 at 2000 A. Wave-
length accuracy, relative to a fixed reference line, is 1.0 A throughout
the wavelength range. Resettability of the monochromator is 0.1 A.
Slits are continuously manually variable from closed up to 2000 microns.
Reciprocal linear dispersion is approximately 20 A/mm at the exit slit.
An external control unit provides scanning rates from 0.05 to 20 A/sec,
and can be synchronized with the chart rate drive of the Heath 700-C
chart recorder module used.
An Eimac 300 watt xenon lamp (Eimac is a division of Varian Associ-
ates) with a regulated power supply was used as a continuum excitation
source for the polarized excitation and emission spectra. A Hanovia
12 watt low pressure mercury discharge lamp was ordinarily used as an
excitation source for the Shpolskii matrix experiments on anthraquinone.
Detection and Recording System
The photodetector used in the Shpolskii matrix experiments and the
photoselection experiments was an RCA 1P-28A photomultiplier tube with
an S-5 spectral response. It was enclosed behind a shuttered quartz
window in the.light-tight compartment of a Heath photomultiplier module.
The supply voltage normally used was -900 volts.
The current output of the photomultiplier tube was fed into a
Keithley 416-S picoammeter. The picoammeter amplified the signal, con-
verted it from a current to a voltage signal, and filtered it. The
output of the picoammeter was then fed into the potentiometric amplifier
section of the Heath 700-C chart recorder module.
The potentiometric amplifier will accept any signal from 1 millivolt
up to 50 volts, and convert it to 1 volt full scale output to the chart
recorder. The recorder module was also equipped with a D.C. offset
module which allowed the baseline to be shifted to compensate for any
D.C. component of the signal.
The recorder output was plotted as a record of signal intensity
versus wavelength. The recorder drive could be synchronized with the
scan drives of the monochromators, so that a predetermined scale of
A/inch of chart paper could be recorded.
Shpolskii Matrix Systems
A block diagram of the experimental arrangement is shown in Figure 1.
An Air Products Model CSW202 Displex helium closed cycle refrigerator was
used as the cooling source in these experiments. Samples were contained
in a copper cell sealed with an 0-ring and a quartz Suprasil window.
The cell was mounted on the cold tip of the Displex at a 135 degree
angle to the face of the mercury lamp. The temperature of the sample
was measured by two gold-Chromel thermocouples, the upper of which was
fixed to the cold tip of the Displex, and the other mounted directly to
the cell. Liquid nitrogen was used as a reference junction. The tem-
perature of the upper thermocouple could be read directly from the
temperature controller module, while that of the lower one was indicated
by the thermocouple emf as registered on a Keithley digital voltmeter.
In general, the two temperatures agreed to within 1K. Due to its
proximity to the sample, that of the lower thermocouple was taken as
being a more accurate indication of the sample temperature. If desired,
the sample temperature could be increased by using a resistance heater
on the cold tip of the Displex. Temperature control was normally con-
stant to within 0.50K after the system had equilibrated.
Samples of the compounds to be run in n-paraffin matrices were pre-
pared by the quick-freeze method. The cell was filled and mounted on the
cold tip of the Displex. Good thermal contact was insured by the use of
an indium gasket between the cell and the cold tip. The heat shield of
the Displex was placed around the cell, and the cold tip was immersed
in liquid nitrogen. When the readout of the Displex temperature con-
troller indicated that the cell was at 770K, the outer shroud of the
Displex was put into place and the system put under vacuum by a two-stage
mechanical pump with a liquid nitrogen cold trap. After approximately
fifteen minutes, the compressor of Displex was started. Cool-down time
to 100K was approximately thirty mintues.
Light from the mercury discharge lamp was passed through a Corning
CS 7-54 filter and focused onto the sample cell by quartz optics.
Emission from the sample cell was viewed at right angles to the direction
of excitation. It was focused by quartz optics onto the entrance slit
of the monochromator. Radiation passed by the monochromator was detected
by the RCA 1P-28A photomultiplier tube. The photomultiplier current was
processed as described in the previous section.
Slit widths of the monochromator were in the range from 18 to 120
microns, for a spectral bandpass of approximately 0.4 to 2.4 A. Scan
speed was normally 0.5 A/sec, with 20 A/inch recorded on the chart re-
A block diagram of the apparatus used to measure lifetimes of
various peaks in the phosphorescence spectrum of anthraquinone in hexane
at 770K is shown in Figure 2. The same apparatus was used in an effort
to determine whether or not any phosphorescence was present in the
emission spectrum of quinizarin. Table I lists the model numbers and
- I (-
Table I. Experimental Equipment and Manufacturers (Lifetimes)
Itm Model Number
High Voltage Power
Avco Everett Research Labora-
tories, Everett, Mass.
American ISA, Inc.,
Metuchen, N.J. 08840
RCA Corp., Lancaster, Pa.
PAR Corp., Princeton,
Heath Co., Benton Harbor,
Wavetek, Inc., San Diego,
Hewlett-Packard, Palo Alto,
PAR Corp., Princeton, N.J.
Data Laboratories Ltd.,
Mitcham, Surrey, U.K.
Friden, Inc., Rochester, N.Y.
DEC, Maynard, Mass. 01754
manufacturers of the equipment. Both Figure 2 and Table I are reproduced
with permission from Dr. Glenn Boutilier (26). A complete discussion of
the design and operation of the system may be found in reference (26).
Samples were placed in a 30 cm length Suprasil tube of 2 mm I.D. and
4 mm O.D. This was fitted into a Teflon cylinder using Teflon tape and
slip-fitted into a NMR spinner assembly. The spinner assembly fitted into
the top of the cover of a sample compartment containing a nitrogen
immersion Dewar and was used to position the sample cell. Nitrogen was
flushed through the lower section of the compartment to prevent conden-
sation of moisture on the Dewar.
A nitrogen discharge laser with a pulse width of 7.7 nsec was used
as the excitation source. The 3371 A line was used. Emission lines from
the laser were focused into one plane by the beam steering mirrors. The
laser beam was passed through an interference filter with a peak trans-
mittance of 42% at 3400 A and a bandwidth of 100 A to block non-lasing
nitrogen emission lines at wavelengths longer than 3600 A. The beam was
then focused onto the sample tube by a 3 inch diameter quartz lens of
8 inch focal length.
Light from the sample was focused onto the entrance slit of a
monochromator by a 15 mm diameter 25 mm focal length quartz lens. The
monochromator was a 0.1 meter focal length monochromator with an aperture
ratio of f/3.5 and a reciprocal linear dispersion of 80 A/mm. Detection
of phosphorescence selected by the monochromator was by an RCA 4857
photomultiplier tube enclosed in a light-tight compartment.
The photomultiplier socket was especially wired so that voltage
supplied to the photomultiplier dynodes could supply sufficient current
to maintain a linear response for the large pulses encountered using a
pulsed nitrogen laser. A wiring diagram of the P.M. socket may be found
in reference (26).
Both the signal average and the boxcar integrator used required a
voltage input. The photomultiplier output current was passed through a
current-to-voltage amplifier which could be gated to avoid saturation
effects due to stray light or fluorescence. A discussion of the design
and operation of the amplifier is included in reference (26). The laser
pulse repetition rate was controlled by the Wavetek signal generator
shown in Figure 2.
In the case of lifetime measurements of anthraquinone, the slit
width of the emission monitored at 4974 A was 0.1 mm (giving a spectral
bandwidth of 8 A). In the case of the peaks at approximately 4503 and
4530 A, the slit width was increased to 0.5 mm for a spectral bandwidth
of 40 A. The same setting was also used in attempts to detect phos-
phorescence in the emission spectrum of quinizarin. Supply voltage to
the photomultiplier tube was -800 volts. The amplifier was gated off
for 200 usec in all phosphorescence experiments.
In measuring lifetimes, the filtered output of the amplifier was
connected to the signal average. Sweep time of the signal average
could be varied by factors of 2 from 5 msec to 81.92 sec. Delay before
the start of sweep could either be set to zero or varied in factors of
2 from 0.32 msec to 5.12 sec. The signal average acquired 1000 points
per sweep and summed the value at each point into memory. The proper
repetition rate, sweep time, and amplifier gain were selected from an
oscilloscope display of the signal. After the requisite number of sweeps
had been averaged, the contents of the signal average were output to
the paper tape punch in 16 bit words. A PDP 11/20 microcomputer was
later used to process the tapes. The programs used are also listed in
the appendices of reference (26).
The arrangement used to measure fluorescence lifetimes of quinizarin
was similar. A 1.5 nsec laser was used as the excitation source, and a
fast boxcar integrator was used to process the signal. Instead of being
digitized, the data were recorded on a chart recorder. The limiting
factor here was the response time of the photomultiplier (1.5 nsec).
What was actually measured was the fluorescence lifetime convoluted with
the rise and decay time of the photomultiplier. Lifetimes were estimated
by extrapolating the time for a decay curve of known width of time to
decay to l/e of its original value.
Polarized'Excitation and Emission Spectra
When measuring a polarized spectrum, the substance in question is
normally imbedded in a clear, strain-free glass, dissolved in a dilute,
highly viscous solution, or grown in a single crystal. A glass is
generally much more convenient to prepare than a single crystal and is
preferable to a solution, since rotational depolarization will be
In the case of quinizarin, EPA was found to be the most satisfactory
substance for forming a glass (EPA is a 5:5:2 vol.-vol.-vol. mixture of
ether-isopentane-ethanol). Attempts were made to use 3-methylpentane
and isopentane-methyl cyclohexane mixtures, but solubility problems pre-
vented the use of these compounds. Daunorubicin was not soluble to any
appreciable extent in EPA or other organic solvents commonly used to form
glasses. A 55-25-20 vol.-vol.-vol. mixture of ethanol-methanol-water
was used in an attempt to form a glass, but was found to be unstable when
the daunorubicin was dissolved and the solution frozen. Alternatively,
the daunorubicin solution was added in a volume ratio of 1:6 to glycerin,
and the mixture cooled by the simple expedient of lowering the room
temperature with air conditioning to approximately 190C. This arrange-
ment gave satisfactory results for obtaining the polarized excitation
spectrum of daunorubicin.
A number of experimental methods for obtaining polarization ratios
are discussed by Parker (27). The method used was that of Azumi and
McGlynn (28). This procedure corrects for such factors as selective
transmission of the emission monochromator, selective reflection of the
cylindrical sample tube and cylindrical Dewar, and spectral shifts due
to rotation of the Glan-Thompson polarizers. It makes use of the fact
that emission viewed in a direction at a right angle to an incident
horizontally-polarized beam must be unpolarized. If the intensities of
the vertical and horizontal components of fluorescence with horizontally
polarized exciting light are denoted as IBE and IBB, then the ratio
IBE/IBB is a correction factor for the effects mentioned above, since
any difference between IBE and IBB will be due to instrumental factors.
If the corresponding intensities of fluorescence with vertically
polarized light are denoted by IEE and IEB, then the corrected polariza-
tion of emission is calculated from
EE E B (IBE/IBB)
SI + (I BEBI
EE EB BE BB
Two different experimental methods were used which gave essentially
the same results. Polarized excitation and emission spectra were first
run on an Aminco-Bowman spectrofluorimeter equipped with a liquid
nitrogen immersion Dewar and Glan-Thompson polarizing accessory. How-
ever, in obtaining the polarized excitation spectra there was some
second-order scattering of the exciting light at short wavelengths.
Spectra were recorded on a relatively small area of chart paper, which
made an accurate interpretation of intensity versus wavelength somewhat
difficult. In order to obtain more highly resolved spectra, the experi-
mental arrangement shown in Figure 3 was employed.
Quinizarin, dissolved in EPA, was placed in a 2 mm I.D. quartz
Suprasil tube. The tube was then placed in a quartz immersion Dewar,
which was clamped into place on a ring stand mounted on the optical table.
A stream of air was blown across the Dewar to prevent condensation.
Light from the Eimac lamp was passed through the excitation mono-
chromator. The exciting light was collimated and focused on the sample
by quartz Suprasil optics. Polarization of the excitation beam was
achieved by passing the focused light through a Glan-Thompson polarizer
placed directly in front of the immersion Dewar.
Emission was collected at right angles to the direction of excita-
tion. Polarization of the emitted light was obtained by using a
polaroid sheet mounted in front of the entrance slits of the emission
In obtaining the polarized excitation spectra, the excitation
monochromator had to be scanned from 2000 A up to past 5300 A. This,
coupled with two other factors, led to the necessity of scanning the
excitation spectra in two segments. The Dewar used held enough liquid
<<.^_ r, 'I
nitrogen for approximately 36 minutes. In order for a set of spectra to
be internally self-consistent, a scan of all four possible orientations
of the polarizers had to be taken on a single glass sample without re-
filling the Dewar. Also, since the output of the Eimac lamp was much
greater in the visible than in the ultraviolet region, a readjustment of
monochromator slit widths had to be made before scanning into the visible
region to keep the chart recorder from going off-scale. Therefore, the
excitation spectra for quinizarin were usually scanned from 2000 A up
to 3800 A. A scan of the visible region was made on a fresh glass sample
from 3600 A up to past 5400 A. An average of polarization values from
the two different runs was taken in the area of overlap. In general,
the agreement was good between the two sets of data.
When a set of polarized excitation spectra was obtained, the emis-
sion monochromator was set on a wavelength corresponding to the maximum
of a vibronic band in the emission spectrum. The chart recorder was
synchronized to the scan drive of the excitation monochromator, and the
wavelength region of interest was scanned. Slits of the excitation
monochromator were maintained at 1800 microns or less, while those of
the emission monochromator were maintained at 1000 microns or less.
In the emission mode, the strip chart recorder was synchronized with
the scanning drive of the emission monochromator. The excitation mono-
chromator was maintained at a fixed wavelength, and the fluorescence
emission spectrum was obtained. Spectra were also obtained here for all
four possible orientations of the two polarizers.
The experimental arrangement for obtaining the polarized excitation
spectra of daunorubicin was similar, with the distinction that the quartz
immersion Dewar was not used. As mentioned before, it was necessary to
use the highly viscous glycerin-alcohol-water mixture to obtain polarized
excitation spectra, since no suitable glass could be found for daunoru-
bicin. The mixture was placed in a 1/2" O.D. quartz Suprasil tube and
mounted in the same position as the immersion Dewar had been, and spectra
were obtained as described in the case of quinizarin. Since the
daunorubicin exhibited broad, structureless fluorescence, no polarized
fluorescence emission spectra were recorded.
THE ANTHRAQUINONE SYSTEM
The molecular structure of anthraquinone and its crystalline mor-
phology are well known (29, 30, 31). The 9,10-anthraquinone molecule is
planar and belongs to the D2h point group, while its crystalline space
group is C2h.
Group theoretical considerations of the electronic configurations of
anthraquinone show that emission from the triplet level could occur from
either the Au, B1g, B2g, or B3u states. It is generally accepted (13, 15)
that the lowest triplet energy levels of the 9,10-anthraquinone molecule
are the 3B and 3A states. These two states are degenerate if inter-
actions between the two carbonyl groups are neglected. On the basis
of polarized single crystal absorption measurements at 770K, Dearman
and co-workers (11) have assigned the lowest triplet state of 9,10-
anthraquinone as A This assignment was refuted in later polarized
Zeeman (13) and Stark-Zeeman (12) absorption measurements, where it was
concluded that the lowest triplet state of 9,10-anthraquinone is 3Bg
and that it is separated from the A state by 410 cm- in the crystal
The transition Big A is forbidden on the grounds of spin and
parity conservation, while the transition A A is only spin for-
It has been shown (9, 10) that the phosphorescence of anthraquinone
is n-n* type, in-plane polarized. Measurements of the polarized S-T
absorption spectra (11) showed that the dipole moment of the transition
lies in the plane of the molecule parallel to the C-O axis (the Z axis,
in Mulliken's notation). Therefore, the vibrational bands present in the
emission spectrum of anthraquinone may be viewed as primarily being
polarized along the Z axis. If the conclusions in (12, 13) are correct
3 3 1
and the 3Bg is the lower triplet level, then the transition 3B A
will occur primarily as a result of borrowing intensity from the nearby
lu (TT*) state.
It is well known (32) that, in the case of forbidden electronic
transitions where vibronic coupling plays an important part in allowing
the transition to occur, only odd quanta of the coupling vibrations will
appear. These vibrations will be antisymmetric with respect to any
symmetry operation to which the transition moment integral is antisym-
metric. In the case of 9,10-anthraquinone, this will be vibrations of
the type b If the carbonyl stretching frequency and vibrational pro-
gressions built upon it are present, as is normally the case for n-n*
phosphorescence from carbonyl-containing compounds (22), then the low
temperature Shpolskii matrix phosphorescence should reveal them. If the
lowest triplet state were indeed the 3B1g, then a close analysis of
vibrational progressions present in the phosphorescence spectrum should
Analyses of the spectra were handicapped by a lack of reliable Raman
data. As an adjunct to this portion of the research, it was decided to
obtain polarized single crystal Raman spectra of anthraquinone. However,
for reasons which are discussed in a later section of this paper, these
attempts were unsuccessful. Also, a highly resolved low temperature
Shpolskii matrix emission spectrum of anthraquinone was published (33)
before the analysis presented here was complete. Therefore, the present
results will only be reported to the extent that they extend those findings.
In comparing spectra obtained at 100K and 770K, it was noticed that
several small high energy peaks had appeared as the temperature was in-
creased. The separation between the location chosen as the 0-0 position
of the anthraquinone phosphorescence and the highest-energy temperature
dependent band was approximately 424 cm-1. This was very close to the
value of 410 cm-1 tentatively suggested as the triplet level separation
in the crystal (13). It was thought that perhaps the upper triplet level
might be thermally populated by the lower one. Such behavior has been
observed in other molecules (34, 35, 36). Therefore, a series of experi-
ments were undertaken in an effort to determine whether the higher-
energy peaks were emission from a thermally populated triplet level of
anthraquinone, or whether they had some other origin.
In summary, this portion of the research had three major objec-
tives: (1) to obtain and analyze highly resolved phosphorescence spectra
of anthraquinone in n-paraffin matrices at low temperatures; (2) to
obtain polarized single crystal Raman spectra in order to make definitive
assignments of Raman-active normal vibrations; (3) to determine the
exact nature of the temperature dependent higher energy emission bands
in the phosphorescence of anthraquinone.
Shpolskii Systems and Impurity-Lattice Interactions
The low temperature phosphorescence spectrum of anthraquinone in
n-hexane is an example of what is commonly referred to as the "Shpolskii
effect." The Shpolskii effect was discovered by E.V. Shpolskii (37)
as part of a systematic investigation of luminescence and absorption
spectra of large organic molecules in frozen crystalline n-paraffin
matrices. As in the case of substituted molecular crystals, i.e.,
naphthalene in durene, the effect is generally considered to arise from
substitution of "guest" molecules into the crystalline lattice of the
host. The guest molecules exist as discrete molecular substitutional
sites, rather than as crystalline aggregates (33, 38, 39). In inves-
tigations of electronic transitions, the Shpolskii matrix method
possesses the inherent advantage that the electronic energy levels of
the host are usually much greater than those of the guest, so that the
only electronic transitions which occur upon excitation are those of
the guest molecule.
In an experiment involving the Shpolskii method the substance
under investigation is dissolved in a suitable solvent, usually an
n-alkane, and the solution is then frozen. At very low temperatures the
guest molecules undergo electronic transitions in both absorption and
emission with a very high probability that either no change or small
changes in the phonon energy of the host will take place. This will
depend upon the nature of the electronic transition and upon the extent
of coupling of inpurity molecules to the host lattice.
Spectra produced by these transitions are quasi-linear in nature,
having bands which are very sharp and well-defined, with residual half-
widths of approximately 1-10 cm- at very low temperatures. They are
sometimes considered to be the optical analogue of the Mossbauer effect
(40). Since emission in a condensed medium normally occurs from the
ground state vibrational level of the electronic excited state, there is
usually a very good correlation between the frequencies of such peaks in
the emission spectrum and the ground state vibrational frequencies of
the guest molecules. These sharp peaks are often accompanied by more
diffuse, less intense bands which are usually assigned as "lattice" bands
in the analysis of a Shpolskii matrix spectrum. More will be said about
For molecules in which the electronic transition is an allowed one,
there is an overlap of the 0-0 transition in absorption and emission. A
temperature dependence upon the intensity of the spectrum is also ob-
served, with the intensity increasing with decreasing temperature. It is
also observed that considerable broadening and loss of fine structure
occurs as the temperature is raised. This is often accompanied by
thermal line shifts.
There also appears to be a critical dependence upon the size of the
solvent molecule. This is manifested experimentally in several different
ways. One of these is the fact that a spectrum can often be interpreted
in terms of several different spectra of varying intensities which are
displaced from each other. This effect is explained in terms of dif-
ferent sites in the host crystal in which the guest molecules reside. It
is also found that the solvent molecule which has a long axis dimension
comparable to that of the guest molecule gives the sharpest, most well-
As mentioned earlier, a common feature of electronic spectra in
Shpolskii matrix systems is the appearance of sidebands which are
associated with vibrational-electronic transitions. These sidebands are
usually taken -to be evidence of the electron-phonon interaction. The
exact nature of the impurity-host relationship is not well understood.
The first attempts at formulating a theory to describe the effect ob-
served in Shpolskii systems were made by Rebane (41, 42). The quasi-
linear bands were associated with a phononless optical transition and
the diffuse background was associated with phonon bands. Interest in the
exact nature of the impurity-lattice interaction has led to attempts to
refine the theory of interactions (43, 44, 45) and also to account for
thermal broadening and shifts of the so-called "zero-phonon" lines in
Shpolskii systems. Recently (46-53), experiments have been performed,
utilizing both absorption and emission spectra, upon various systems in
attempts to clarify the nature of the impurity-lattice interaction.
In any treatment of the theory of optical spectra of impurities in
solids (43-45, 54, 55), several assumptions are normally made. Both the
Born-Oppenheimer approximation and the Franck-Condon principle are taken
to be valid. It is generally assumed that the impurity concentration is
low, so that impurity-impurity interactions are taken to be negligible.
It is also generally assumed that electronic eigenstates of the impurity
are at different energies than the electronic bands or states of the
host crystal. The presence of the impurity will serve to destroy the
translational symmetry of the host crystal, so that the relevant sym-
metry will be the site symmetry of the impurity. It will also perturb
the lattice vibrations of the crystal by giving rise to localized
vibrations which do not exist in the vibrational spectrum of the unsub-
stituted host crystal. In general, the impurity will interact with both
band and localized vibrations (55).
Detailed.treatments of the theory of lattice-impurity interactions
in the optical spectra of impurities in solids are available (42, 43, 54,
55). In the general case, the optical spectrum of the impurity crystal
will consist of a narrow zero-phonon line corresponding to the purely
electronic transition in the impurity site, accompanied by broad phonon
sidebands caused by transitions with simultaneous excitation of phonons
(55). In the case where localized modes are present in the crystal,
vibrational replicas of the zero-phonon line will be produced at
frequencies + nw from the zero-phonon line, where w is the frequency of
the localized vibration, n is an integer, the plus sign corresponds to
absorption, and the minus sign to emission (55). Physically, the inter-
action of the electronic transition in the impurity site with the thermal
vibrations of the crystal will manifest itself in its influence upon the
intensity, linewidth, and position of the zero-phonon line.
The theory of the electron-phonon coupling predicts an exponential
dependence of the relative integrated intensity of the zero-phonon line
on temperature. The ratio of the intensity of the zero-phonon line to
the total spectrum intensity may be expressed as (56):
Io/I = exp[-. pi(2ni + 1)] = exp[-S(T)] (1)
In this expression, i is an index for acoustic and local vibrations,
pi is the dimensionless "Stokes loss" per crystalline oscillator i and
is equal to (mi mw Ag/2h), w. is the frequency of the ith oscillator,
mi is the effective mass, Ai is the shift in the equilibrium position
of the ith oscillator during an optical transition in the impurity site,
ni = [exp(fihi/kT) 1]-1 is the thermal average of the occupation number
of the ith oscillator, and S(T) is the Huang-Rhys factor which denotes
the strength df impurity-lattice coupling (56).
- L. I -
An examination of the relative integrated intensity of the zero-
phonon line as a function of temperature would serve to give information
about the strength of the lattice-impurity coupling, since the overall
integrated intensity of the zero-phonon line and its accompanying phonon
bands is independent of temperature in the Franck-Condon approximation
(55). Intensity lost by the zero-phonon line should be gained by the
phonon side band, since increased electron-phonon interaction would occur
as the temperature increased. The extent of the impurity-lattice
coupling would be characterized by the resulting value of S, since the
larger the value of S, the greater the impurity-lattice coupling is
taken to be.
The nature of the broadening of the zero-phonon line will depend
upon the structure of the electronic energy levels of the impurity (56).
For the case where the maximum phonon energy is greater or equal to the
difference in the energies of the electronic levels, the broadening may
be mainly determined by nonradiative thermal processes when absorption
or emission of a single resonance phonon occurs at the same time as the
electronic transition. In the opposite case, that where the difference
in the energies of the electronic levels is greater than the maximum
phonon energy, Raman scattering of phonons by the impurity center may
play an important part in the thermal broadening (53).
If the electronic transition which takes place is a phonon-assisted
one, certain additional features will manifest themselves. The vibronic
bands present due to a phonon-assisted transition will involve the
emission of photons and the creation or annihilation of phonons. De-
pending upon the energy levels involved, either or both of Stokes and
anti-Stokes emission of phonons will occur. Di Bartolo (57) has derived
the relevant matrix elements for such processes.
Since thermal vibrations are involved, the intensities of the
phonon bands on either side of a zero-phonon line will be temperature-
dependent. In the simplest case, that of a one-phonon transition, the
process will depend upon the number of phonons present. The temperature
dependence is contained in the expression for the number of phonons
np= [exp(h, p/kT) 1]-1 (2)
The value of n becomes very small as T is reduced. As shown in
(57), this will result in the almost total disappearance of emission of
anti-Stokes phonons at low temperatures, while the Stokes processes will
persist. It is also expected that as the temperature increases, an
increase of the multi-phonon background will occur. These processes will
tend to smooth out the peaks so that an almost continuous background
Anthraquinone and n-paraffin solvents were obtained as described
in Chapter II. The techniques described there for obtaining Shpolskii
matrix spectra and measurements of phosphorescence lifetimes were em-
ployed. The I.R. spectra were obtained in KBr disks on a Perkin-Elmer
As was mentioned in the introductory section of this chapter, un-
successful attempts were made to obtain polarized single crystal Raman
spectra of anthraquinone. There were several reasons for attempting
these experiments. There is some disagreement concerning the presence or
absence of certain lines in the powder spectrum of anthraquinone, as well
as the assignment of certain experimentally observed frequencies to
normal modes (58, 59). The most complete experimental study upon powder
samples is that of Rasanen and Stenman (59), with frequencies being
assigned on the basis of normal coordinate calculations of Strokach,
Gastilovich, and Shigorin (60, 61). In cases where the crystal struc-
ture of a compound is known, the method of polarized single crystal Raman
spectroscopy has proven to be very useful in the unambiguous assignment
of Raman active normal modes of vibration (62, 63).
Single crystals of anthraquinone were grown by the Bridgman tech-
nique. The design of the furnace was based upon those described by
Lipsett (64). Two weeks were normally required to pass an anthraquinone
ingot through the furnace.
The Raman spectrometer used was a Spex Ramalog 5, with a Spex 1401
double monochromator and an RCA 3140 cooled photomultiplier tube which
could be operated in the photon-counting mode. The excitation source
normally employed with this system was a Coherent Radiation Model CR-5
argon ion laser. Malfunctions of the argon ion laser and the photo-
multiplier tube resulted in a lengthy period of inactivity in this portion
of the research.
It was found that anthraquinone fluoresced strongly under excitation
by the argon ion laser lines, so that no Raman spectra could be obtained.
The exact nature of the fluorescence which often occurs in Raman spectra
is not clear. In this case, it was decided to employ the accessory
Coherent Radiation Model 490 dye laser as an excitation source to see if
going to longer wavelengths of excitation would decrease the fluorescence.
Attempts to use the dye laser proved to be only a little more
successful. The power output of the dye laser was very unstable. Since
the Raman effect is ordinarily on the order of 10-5 times the accompany-
ing Rayleigh scattering, a very poor signal to noise ratio was obtained.
It was also discovered that no signal below 1000 cm-l of the exciting
line of the dye laser could be detected due to background in the anthra-
quinone powder spectrum. The reason for this was not immediately clear.
It was later discovered that the birefringent tuning element of this
particular model of dye laser transmits a structured fluorescence
emanating from the dye itself (65). In addition, background fluorescence
was observed in portions of the anthraquinone spectrum which could be
It is possible that the fluorescence emanating from the dye itself
could have been removed by either an appropriate wavelength notch optical
filter or a tunable grating filter. It was also planned to use a KIM-1
microcomputer to employ the technique of digitized frequency-modulated
spectroscopy (66) to remove any fluorescence intrinsic to the anthra-
quinone itself. This technique requires an excitation source which is
constant in time. All attempts to correct the problem of the dye laser
power fluctuations, including those of a Coherent Radiation field service
representative, were unsuccessful.
The only spectra obtained were those of powder samples. The
accessory for mounting single crystals on the spectrometer was not
available. Attempts to fabricate holders resulted in unsuccessful
attempts to obtain single crystal spectra, so it is probable that the
problem was one of alignment. In any case, in view of all the diffi-
culties encountered, it was deemed expedient to terminate this phase
of the research and rely upon the frequencies and assignments of Raisnen
and Stenman (59).
Results and Discussion
Analysis of Phosphorescence Spectra
Tabulations of the analyses of the phosphorescence spectra of
anthraquinone in Shpolskii matrices of n-hexane and n-heptane are shown
in Tables II and III, respectively. The analysis is complete in the case
of n-hexane, while only the major spectral features of the emission in
n-heptane are presented. The analysis is much more straightforward in
the case of n-hexane, since anthraquinone exhibits only one-site emission
in this solvent. Interpretation is much more complicated in the heptane
and octane matrices, particularly in the case of the weaker combination
bands, since two sites appear to be present. Two sites also appear in
the case of anthraquinone in n-pentane. In n-hexane, heptane, and
octane, the origin of the electronic transition is absent, while it is
observed in n-pentane. As was noted by Khalil and Goodman (33), this
may be attributed to the fact that at least the site symmetry of the
molecule is conserved in the fonner three matrices, while distortion
and subsequent lowering of symmetry occur in the n-pentane case.
Literature values (67) of the I.R. active fundamentals were checked
for accuracy. As explained earlier, attempts to obtain Raman spectra
were unsuccessful. The frequencies and assignments of Rasanen and
Stenman (59) were used instead. A tabulation of the fundamental modes
of vibration of 9,10-anthraquinone used in this work can be found in the
Table II. Phosphorescence Vibrational Analysis for Anthraquinone in
lexane at 10"K
+ 789 Raman
+ 789 Raman + lattice
+ 1146 Raman
+ 1146 Raman
+ 1146 Raman + lattice
+ 365 Raman
0-v32 + la
0-3 x v66
0-v31 + la
0-3 x V32
0-v62 + la
0-v31 + 25
0-v31 + 36!
0-v28 + la
Table II. (Continued)
Wavelength Energy AE Assignment
(A) (an1, vac) (cm-, vac)
5031 19875 1908 MW 0-v24 + 239 Ranan
5039 19841 1942 W 0-v24 + 258 Raman
5044 19819 1964 W 0-v62 + 1146 Raman
5051 19972 1992 VW 0-v62 + 1146 Raman + lattice
5065 19737 2046 MS 0-v24 + 365 Raman
5071 19714 2069 MW 0-v24 + 365 Raman + lattice
5104 19589 2194 W 0-v46 + 1146 Raman
5121 19521 2262 VW
5124 19511 2272 VW
5130 19487 2296 MS 0-v31 + 1667 Raman
5136 19467 2316 VW 0-v31 + 1667 Raman + lattice
5148 19419 2364 VW 0-v24 + 688 Raman
5152 19404 2379 VVW 0-v24 + 688 Raman + lattice
5160 19373 2410 VVW
5176 19314 2469 MS 0-v24 + 789 Raman
5183 19290 2493 W 0-v24 + 789 Raman + lattice
5199 19228 2555 VVW 0-v61 + 1597 Raman
5228 19124 2659 VVW O-v24 + 978 Raman
5248 19051 2732 VW
5259 19009 2774 VVW
5276 18948 2835 M 0-v24 + 1146 Raman
5283 18922 2861 W 0-v24 + 1146 Raman + lattice
5306.5 18839 2994 VVW
5313 18816 2967 W 0-v27 + 1667 Raman
5329 18761 3022 VW 0-v25 + 1440 Raman
5360 18653 3130 VVW
5370 18616 3167 W 0-v24 + 1480 Raman
5379 18585 3198 VVW
5391 18544 3239 MS 0-v25 + 1667 Raman
5399 18518 3265 MW 0-v25 + 1667 Raman + lattice
5407 18489 3294 W 0-v24 + 1597 Raman
5411 18476 3307 W 0-43 + 789 Raman +1146 Raman
5422 18439 3344 VS 0-v2 + 1667 Raman
5428 18419 3364 M 0-v24 + 1667 Raman + lattice
5489 18213 3570 VW
5498 18183 3600 VW
5530 18087 3705 M 0-v24 + 1667 Raman+365 Raman
5537 18055 3728 W 0-v24+1667 Raman+365 Raman+lattice
5576 17928 3855 W
5608 17826 3957 W 0-v31 + 2 x 1667 Raman
5643 17715 4068 W 0-v25 + 1667 Raman+789 Ranan
5663 17654 4129 W
5748 17392 4391 VW O- 25+1597 Raman+1146 Raman
5756 17366 4417 VW 0-v24+1597 Raman+1146 Raman
5782 .17289 4494 W 0-v24+1667 Raman+1146 Raman
5791 17280 4509 W 0-v24+1597 Raman+1243 Raman
5820 17177 4606 W
Table II. (Continued)
Wavelength Energy AE Assignment
(A) (cm-1, vac) (cm-1, vac)
5893 16965 4818 VW 0-v24+1667 Raman+1480 Raman
5917 16896 4887 VW 0-v24 + 2 x 1597 Raman
5928 16864 4919 W 0-v25 + 2 x 1667 Raman
5946 16813 4970 W 0-v24 + 2 x 1667 Raman
5954 16791 4992 W 0-3 x v24
Phosphorescence Vibrational Analysis for Anthraquinone in
Heptane at 10K
Site Site AE I Assignment
(cm-1) (cm-1) (cn-l)
VVVW 0-0, H.E. site
VVW 0-0, L.E. site
W 0- 49
'W 0-3 x v66
'W 0-v32 + 365
W 0-3 x v66
W 0-v32 + 365
W O-v31 + 365
W 0-v31 + 365
S 0-v2 + lat'
'S 0-v24 + latl
M 0-24 + 365
W 0-24 + 265
S 0-24 + 365
1W O-v24 + 365
WI 0-v2 + 166
'W 0-v31 + 166
M 0-v31 + 166
W 0-v31 + 166
W 0-v24 + 789
'W 0-v24 + 789
W O-v24 + 789
1 0-v24 + 789
W 0-v24 + 114(
1W 0-v24 + 1141
1W 0-v27 + 166
W 0-v27 + 166
Raman + lattice
Raman + lattice
7 Raman + lattice
7 Raman + lattic
Raman + lattice
Raman + lattice
Table III. (Continued)
Site Site AE I Assignment
(an-1) (cm-1) (cm-1)
18524 3342 S 0-v + 1667 Raman
18508 3358 W 0-v24 + 1667 Ranan + lattice
18418 3343 VW 0-v24 + 1667 Raman
18401 3360 MW 0-v24 + 1667 Raman + lattice
18164 3702 W 0-v24 + 1667 Ranan + 365 Raman
18058 3703 MW 0-v24 + 1667 Raman + 365 Raman
17910 3956 VW 0-v31 + 2 x 1667 Raman
17803 3958 W 0-v31 + 2 x 1667 Raman
17787 4079 VW 0-v25 + 1667 Raman + 789 Raman
17700 4061 VW 0-v25 + 1667 Raman + 789 Raman
17372 4492 VW 0-v2 + 1667 Raman + 1146 Raman
17350 4516 VW -v24 + 1597 Raman + 1243 Raman
17269 4492 W 0-v24 + 1667 Raman + 1146 Raman
17245 4516 VW 0-v24 + 1597 Raman + 1243 Raman
16947 4919 VW 0-v + 2 x 1667 Raman
16874 4992 VW 0-3 x v24
16844 4917 VW 0-v25 + 2 x 1667 Raman
16771 4990 W 0-3 x v24
We find 7 out of 11 possible blu vibrations to be present in the
phosphorescence spectrum of 9,10-anthraquinone in hexane. These, in
the notation of reference (67), are v 32' v31' V28' v27' V26' v25' and v24
This is in agreement with the results of Khalil and Goodman. One fea-
ture which was not included in reference (33)was the presence of a peak
at 1661 cm-1 from the origin. This peak is clearly visible, and is
assigned in the present study as a combination of v27 and the 365 cm-1
Raman line of anthraquinone.
The most intense peaks in the spectrum were those based upon the
non-totally symmetric C = 0 stretch, v24, as was expected. Both one and
three quanta of this vibration are reported in this work, although only
one quantum was reported in reference (33). Assignments of multiple
quanta of vibrations will be discussed later.
Khalil and Goodman reported combinations of v24 with Raman lines of
frequencies 240, 360, 790, 1149, 1595, 1666, 1666 + 360, and 1666 + 1149
cm In the present study, peaks were found which were assigned as
combinations of v24 with Raman lines of frequencies 239, 258, 365, 688,
789, 978, 1146, 1480, 1597, 1667, 1667 + 365, 1597 + 1146, 1597 + 1243,
1667 + 1480, 2 x 1597, and 2 x 1667 cm1. The difference is probably
attributable to the much greater sensitivity in this region of the spec-
trum of the photomultiplier used here, as compared to that used by
Khalil and Goodman (33).
Khalil and Goodman (33) assigned lines in the phosphorescence
spectrum of anthraquinone of type b2u corresponding to v46' U44' and
v43. In addition, another b2u vibration, V47' appeared in combination
with the Raman 360 cm- line. In this study, 5 out of 11 possible
b2u vibrations were observed. These were v43' v44' v45' v47, and v49'
Combinations of v49 + 1146 cm-1 Raman, v46 + 1146 cm-1 Raman, and
v43 + 789 cm- + 1146 cm- Raman were observed.
At this point, some discussion concerning peaks corresponding to
"46 is necessary. There is some disagreement (67, 58, 68) concerning
the frequency of this fundamental mode of vibration, since it did not
appear in the single crystal spectra of anthraquinone and anthraquinone
d-8 (67). Normal coordinate calculations (60, 61) favor the assignment
of a frequency of 1034 cm- (68). Khalil and Goodman used a frequency
of 1155 cm as the basis for an assignment of v46' In the present
work, no peak is observed at this position. Instead, a peak is observed
at 1067 cm1 from the origin which might be assigned to either v46 or
v29. Since other combination bands occur in the spectrum which agree
very well with a frequency of 1034 cm-1 for v46, this value is chosen
and assignments are made accordingly.
Out of six possible b3u fundamentals, three are observed in this
work and three are also reported by Khalil and Goodman. These are v66,
v63' and v62. Combinations of v64 with the 1146 cm-l Raman line are
found in the present work and in reference (33). Khalil and Goodman
assign a combination of v61 with a 360 cm- Raman line, while a com-
bination of v61 with the 1597 cm-l Raman line is found in this study.
A combination of v62 with the 1146 cml Raman line and of v63 with the
789 cm-l Raman line is found in both results.
As regards the search for odd quanta in the vibrational progressions
of the phosphorescence of anthraquinone, there is evidence for three
such peaks. The band located at 521 cm-1 from the origin may be assigned
to three quant-a of v66' a skeletal deformation, or a combination of v32
with the 305 cm- Raman line. This was assigned by Khalil and Goodman (33)
as 3 x v32 on the basis of deuteration experiments. The line which
occurs at 692 anc from the origin might either be assigned to the Raman
685 cm line, to 3 x v32' or to a combination of the b2u mode vu4
(387 cm-l) with the Raman 305 cm-l line. On the basis of deuteration
experiments, Khalil and Goodman assigned this line to 3 x v32, so it has
likewise been assigned here.
The third peak in the spectrum of anthraquinone in hexane which
may be assigned as an odd quantum progression lies at 5954 A. This line
is close to the position of several mercury lines. However, it appeared
in spectra excited by the mercury lamp using a Corning CS 7-54 filter and
a WG-9 emission filter. The Corning 7-54 filter served to prevent
transmission of visible lines from the lamp, while the WG-9 emission
filter served to prevent the occurrence of second-order mercury lines
in the spectrum. The band in question, along with several other weak
peaks in the area, was observed in spectra excited by the Eimac xenon
lamp, so the conclusion that it belongs to the anthraquinone phosphor-
escence seems a valid one. This line, along with several others reported
in this work, was not reported by Khalil and Goodman, since the spec-
tral response of the photomultiplier to be used by them was very weak
in this region.
Results of analyses of Shpolskii matrix emission spectra of anthra-
quinone obtained in this study tend, in general, to confirm the results
of Khalil and Goodman. Crystal field effects of the host lattice do not
affect the location of the origin to a great extent, since the origin of
the electronic transition lies at similar positions in pentane, hexane,
heptane, and octane. In pentane, the origin is observed, with two site
emissions present. In hexane, heptane, and octane, it occurs weakly or
not at all. This indicates that at least the site symmetry (Ci) of the
anthraquinone molecule is preserved in the hexane, heptane, and octane
matrices. Non-totally symmetric I.R. active vibrations of types blu,
b2u, and b3u occur in the phosphorescence spectrum, indicating that the
electronic state from which the phosphorescence takes place is B g.
The BIg A transition is both spin and symmetry forbidden.
Apparently, the state from which it borrows intensity is a Blu
(IT-Tr* state). As discussed by Khalil and Goodman (33) and Strokach and Shi-
gorin (69) mixing of states to allow the transition to occur may happen
in three different ways. The 3Blg state may be viewed as mixing directly
in first order perturbation theory by means of the spin-electronic-
vibrational interaction operation H vso. The two states may also be
mixed in second order perturbation by intermediate states. These may
be either or both of a A 7,7,* state or a A n,,r* state. The possi-
abilities are shown in Figure 4. Here, Hvso = Hso /Q where Hso is the
spin-orbit interaction operator with a component of Blg symmetry and Q
corresponds to a blu type normal mode of vibration. H is the
operator for electron-vibrational interaction, aH o/Qa, where Ho repre-
sents the Hamiltonian of the unperturbed system.
The question of the temperature-dependent emission in anthraquinone
is now considered. Figures 5-17 illustrate the behavior of emission in
the region of the origin of the electronic transition of anthraquinone
in octane, heptane, hexane, and pentane as the temperature is increased.
The phenomenon of temperature-dependent emission is observed in octane,
kAb sua ui
-'I \ -
heptane, and hexane matrices, where the origin of the purely electronic
transition is absent, but not in the pentane case where the origin is
The possibility that the peaks are an impurity emission can be
largely discounted by consideration of the purity of the chemicals used.
That the peaks might be due to a photoproduct being formed was eliminated
by demonstrating a reversible dependence of the emission upon temperature
and a lack of time dependence of intensity upon photopumping.
In cases where thermally activated emission processes between two
adjacent electronic levels occur, an Arrhenius plot of In I/I1 versus
T 1,will normally give a straight line whose slope agrees well with the
energy separation between the levels if no additional depletion processes
are present (34, 35). The results of such a plot for anthraquinone in
hexane are shown in Figure 18. There is an apparent exponential tem-
perature dependence, but the activation energies do not agree with the
energy level separations. If the highest-energy peak (- 4503 A) is taken
as the origin of a thermally populated 3A state, then it is separated
from the origin of the 3Bg state by approximately 420 cm-1. The value
of the activation energy obtained from the Arrhennius plot for this peak
is 220 cm- and the correlation coefficient of the plot is 0.981. There
is no obvious reason why the discrepancy of a factor of approximately
2 should occur if this peak is indeed the origin of emission from the
thermally populated 3A state. Likewise, the peak at approximately
4530 A gives an activation energy of 176 cm-. If it were a vibrational
progression in the emission from the 3A state, a separation of approxi-
mately 68 cm n- should be present. A careful study of the emission of
the origin region in hexane, as a function of temperature, as shown in
E = 176
E. = 220 cm-1
0.8 1.0 1.2 1.4 1.6 1.8
(1/kT x 103)
Plot of Activation Energies
of Anthraquinone in Hexane
of Temperature-Dependent Emission
Figures 5-10, revealed that the peaks appeared sequentially as the
temperature was increased. The lifetime of the emission as monitored
at 4974 A (0-v24) was 3.21 + 0.32 msec, while those of the 4503 and
4530 A peaks were 3.77 + 0.38 and 3.71 + 0.33 msec, respectively. The
value of the lifetime of the emission monitored at 4974 A agrees well
with the literature values (70, 71). The close correlation of the life-
times of the three peaks indicates that they all originate from the same
electronic level. In view of all these factors, it was concluded that
the peaks were not emission from the second triplet level, but were
either vibrational hot bands or bands due to electron-phonon interaction.
This conclusion (that the peaks were not emission from the second triplet
level) was supported by a later paper (72) on the temperature dependence
of anthraquinone. The authors observed a strange effect upon adding
hexyl iodide to anthraquinone in heptane. No effect upon the emission
from the 31g level was observed, but a new set of weak bands in the
origin region were present which were assigned to the 3A level. These
peaks were separated from the 3B1g emission by approximately 60 cm-1
In this same paper, it was tentatively suggested that the temperature-
dependent emission was due to vibrational hot bands from the distorted
This does not seem to be a reasonable conclusion, since, in condensed
media, vibrational relaxation times are normally much faster (10-11 to
10-14 sec) than radiative lifetimes. Hot bands are therefore normally
found (73, 74) only in gaseous systems. In the rare cases (75, 76),
where it has been suggested that they were present in quasi-linear
systems, the electronic emission is spontaneous fluorescence where the
radiative lifetime is of a similar order of magnitude to the vibrational
The fact that the temperature dependent emission is found in the
hexane, heptane, and octane matrices, but not in the pentane matrices,
lends support to the idea that the peaks are due to impurity-lattice
interactions rather than vibrational hot bands. That is, the transition
appears to be a phonon-assisted one in the cases of anthraquinone in
hexane, heptane, and octane, where the origin of the electron transition
is absent. Before proceeding further in this direction, a number of
points concerning the crystal structures and phonon spectra of the
n-paraffins C5-C7 will be discussed.
The crystalline structures of n-pentane, hexane, heptane, and octane
have been well characterized by X-ray methods (77-80). The n-pentane
structure is orthorhombic, with 4 molecules per unit cell and a volume
of approximately 543 A Both n-hexane and n-octane are triclinic with
one molecule per unit cell and volumes of 165 and 208 A3, respectively.
N-heptane is triclinic, with 2 molecules per unit cell and a volume of
The phonon spectra of these crystals have been investigated using
Raman light scattering and neutron scattering techniques (81-83). A
remarkable similarity in the phonon spectra of n-hexane, heptane, and
octane is found, as might be expected in view of their similar crystal-
The origin of the electronic transition in the phosphorescence of
anthraquinone occurs at similar locations in n-hexane, n-heptane, and
n-octane. Examination of Figures 5-15 shows that similar behavior
occurs in all three cases, with the temperature dependent emission
appearing at approximately the same position in all three matrices.
The discussion which follows is restricted to the case of anthraquinone
in n-hexane, since the situation is uncomplicated by multiple site
The temperature dependent emission is first noticeable at approxi-
mately 500K. Figures 5-10 illustrate the behavior of the origin region
of the phosphorescence of anthraquinone in n-hexane over the temperature
range 140K-159K. It can be seen that, as the temperature is increased,
the origin region begins to resemble a continuum, with maxima appearing
at positions separated by approximately 420, 290, and 180 cm-1 from the
origin. The word approximately is used here because the peaks are
slightly skewed and overlapped so that it is difficult to determine
their exact position. Also, an examination of the position of the most
intense component of the phosphorescence at 4974 A from 14-820K revealed
a shift on the order of 8-10 cm-1 toward the blue as the temperature
was increased. As Figure 10 shows, increasing the temperature to a
point just below the melting point of the n-hexane matrix results in an
almost complete loss of spectral definition in the origin region and an
extension of the emission to even shorter wavelengths. Although not
shown, similar behavior was observed in the heptane and octane cases.
In the case where a molecular impurity with intramolecular frequen-
cies of vibration is substituted into a crystal, two different situations
may arise (55). The molecular frequency wm may fall into a region
corresponding to frequencies also found in the perfect host crystal.
The resonance between the molecular and crystal frequencies will allow
energy to be transferred from the molecular impurity to the crystal.
The second case is that where the molecular frequency bm occurs in a
region which does not correspond to vibrations found in the perfect
crystal. If the perturbation produced by the impurity is strong enough,
an anharmonic vibration may set in which does not correspond to a peak
in the phonon spectrum of the crystal (55). The relatively intense
phonon sidebands which accompany all the major vibrational progressions
in the phosphorescence of anthraquinone in hexane appear to be of this
type. For instance, in the emission corresponding to the vibrational
progression 0-v24, the accompanying phonon sideband occurs at a separa-
tion of 22 cm No such vibrational frequency occurs in the phonon
structure of n-hexane, while anthraquinone has a Raman active lattice
mode of 28 cm-1 (59). It is known (84) that such "lattice" modes will
appear in electronic spectra for either one of two reasons. The first
is that the excited molecule has an excited state equilibrium disposition
in the crystal lattice which differs from its normal ground state con-
formation. The second reason is that the lattice mode perturbs the
electronic transition in a way so as to assist the electronic motion to
gain radiative properties.
The temperature-dependent emission which occurs in the emission of
anthraquinone in hexane matrices has peaks separated from the origin of
the 3Bg level by approximately 180, 290, and 420 cm-l. Peaks in the
phonon structure of n-hexane (83) occur at 176 cm- 296 cm-, 372 cm-l
and 460 cm There is a relatively close correspondence between the
separation of the peaks occurring in the temperature dependent emission
and the peaks occurring in the phonon structure of n-hexane. Anthraquinone
itself has Raman active vibrations at frequencies of 301 cml, 419 cm-1
and 150 cm so that it is more probable that the peaks correspond to
resonance modes set up between hexane vibrations and anthraquinone
As to the temperature-dependent emission, it is assigned here as
arising from a temperature-dependent anti-Stokes scattering of phonons
occurring simultaneously with the electronic transition. There are
several reasons for this conclusion. Figures 5-17 clearly reveal the
temperature dependent emission is present in the n-hexane, heptane, and
octane cases, but not in the n-pentane case. For each of the former
three cases, the origin of the purely electronic transition occurs with
negligible intensity, while in the latter case it is present. This in-
dicates that the transition is a phonon assisted one in the former three
cases. The occurrence of phonons in assisting transitions to take place,
while relatively rare, is not unknown. A most thoroughly documented
case (84) is that of the phosphorescence of molecular crystalline
pyrazine. Raman scattering of phonons has also been shown (50) to play
an important part in the thermal broadening observed in Shpolskii sys-
tems. Finally, the behavior of anthraquinone in n-paraffins C6-C8
resembles closely the temperature dependence of the emission of the
system V2+ in MgO, where anti-Stokes scattering of phonons was also
ascribed as the reason for the temperature-dependent emission observed
in this system (85).
A study of the impurity-lattice coupling in the anthraquinone-hexane
system might be helpful in corroborating the above conclusion. Judging
from the quasi-linear appearance of the spectra, it is probable that the
coupling is weak to intermediate. No attempt was made to evaluate the
coupling. This can be best understood with reference to Figures 19-21.
These indicate the decrease in intensity and broadening of the zero-
phonon line at 4974 A over the temperature range 14-780K. Studies of
impurity-lattice interactions (50) are usually performed on zero-phonon
4925 4950 4975 5000 5025
Figure 19. Phosphorescence of Anthraquinone in Hexane at 14K
4925 4950 4975 5000 5025
Figure 20. Phosphorescence of Anthraquinone in Hexane at 36K
I L I I I
4925 4950 4975 5000 5025
Figure 21. Phosphorescence of Anthraquinone in Hexane at 760K
lines which can be accurately separated from their phonon sidebands over
a wide range of temperature. The broadening in the anthraquinone case
makes this extremely difficult, and would introduce a large error into
calculations of the relative integrated intensity of the zero-phonon
line. Since the relative integrated intensity of the zero-phonon line
as a function of temperature characterizes the strength of the coupling,
this would cause a large error in the estimation of the strength of the
QUINIZARIN AND DAUNORUBICIN
The geometrical structure of quinizarin is not known with certainty.
In all that follows, it will be assumed that the molecule is planar and
belongs to the point group C2v. There are several bases for this
assumption. Quinizarin is 1,4-dihydroxy anthraquinone. Free rotation
about the 0-H bonds could be expected to occur, so that on the average
the 0-H groups might lie either in or out of the plane of the molecule.
Experimentally (17, 86), it is found that the C=0 stretching frequency
in the infrared spectrum of this compound is 1631 cm-1, as opposed to
1676 cm in anthraquinone. This indicates that quite strong hydrogen
bonding is present. Such a situation would occur if the 0-H groups lie
in the plane of the molecule. Quinizarin also displays a limited solu-
bility in polar solvents, which might be interpreted as additional
evidence that the 0-H groups are highly coordinated with the carbonyl
groups. Therefore, for purposes of analysis, it is assumed that 1,4-
dihydroxy anthraquinone (quinizarin) is a planar molecule in the ground
state and belongs to the point group C2v. The axis system in quinizarin
is chosen here to be the same as that used for anthraquinone in
The visible and ultra-violet spectra of quinizarin show several
quite strong absorption bands (5, 19). The longest wavelength absorption
band lies in the visible region, extending from approximately 4000 to
5300 A. The absorption appears to be due to one electronic transition
with vibrational fine structure. All of the UV and visible bands in the
quinizarin absorption spectrum have E values greater than 103, in-
dicating that the electronic transitions responsible for their occurrence
take place with a relatively high degree of probability. Under C2v
symmetry, elements of the electric dipole moment operator transform as
al, bl, and b2. Quinizarin has a closed shell electronic configuration,
so that its ground electronic state is A1 Transitions of the type
11 Al 1A1 B1 A1 B2 would be expected to occur with a high
degree of probability in absorption, while transitions of the type
1A1 1A2 would be forbidden by selection rules and would occur only
by "stealing" intensity through vibronic interactions with allowed
Abrahamson and Panik (21) have discussed the effect of introducing
a basic group such as Cl, OH, NH2, etc.,into the rings of the anthra-
quinone system. If resonance interaction between orbitals on the basic
group and the 7 electronic system of anthraquinone occurs, then the
highest filled i orbital of the ground state will be raised in energy
and that of the basic group will be lowered. For highly electronegative
substituents, an inductive effect may be present which will interact
with n orbitals on the anthraquinone system to depress their energy
levels. Both of these effects will tend to decrease the energy level
separation of the n and n orbitals of the anthraquinone system and in
the extreme case might cause a reversal of the energy levels.
A study of the emission characteristics of the quinizarin molecule
can give information concerning the disposition of singlet energy levels.
Compounds which have an n,w* singlet state as the lowest excited singlet
state do not normally exhibit fluorescence, since almost complete inter-
system crossing takes place from the lowest excited n,w* state to a
triplet state, from which phosphorescence occurs (22). Solvent effects
upon absorption spectra may also be useful, since n,r* transitions will
usually display a shift towards shorter wavelengths in going from a
non-polar to a polar solvent, while r,7* transitions display the opposite
The structures of anthraquinone, quinizarin, and daunorubicin are
shown in Figure 22. It can be seen that the only symmetry element which
daunorubicin possesses is the identity element. It therefore belongs to
the point group C1. All electronic transitions will be of type A 1A,
with subsequent easing of symmetry constraints concerning the "allowed-
ness" or "forbiddenness" of transitions.
This portion of the research deals with investigations of electronic
transitions in quinizarin utilizing visible and ultraviolet absorption,
room temperature luminescence, lifetime measurements, photoselection,
and Shpolskii matrix emission. In addition, I.R. and far I.R. spectra
of quinizarin are reported, with an analysis of I.R. active normal modes
of vibration of quinizarin. Photoselection measurements upon daunoru-
bicin are also included, in an effort to determine whether the assign-
ments of electronic transitions of daunorubicin in reference (23) are
Theory of Photoselection
Numerous examples of the use of photoselection in the investigation
of electronic transitions may be found in the literature (28, 87-91). An
cr O = =
o ao c- \
--1 ~- C
\r /- = f
\\, ,/ >0
\ '*- ^ / LTI
Uj --^- -- 10
extensive exposition of the theory of photoselection and discussions of
its application may be found in the article by Albrecht (92). The
treatment which follows is similar to Albrecht's.
In order to understand how photoselection might be used in the
assignment of electronic transitions, it is first appropriate to review
the concept of the polarization of an electronic transition. Since
electric dipole transitions are generally orders of magnitude stronger
than magnetic dipole or electric quadrupole transitions and are found
much more frequently in organic compounds, the example given is restricted
to the case of electric dipole transitions.
Consider, as the most general case, a transition between two
stationary electronic states qi and if brought about by incident radia-
tion of frequency v = (Ef Ei)/h. For simplicity, the states are
assumed to be non-degenerate and their eigenfunctions are assumed to be
real. The probability that the transition will occur is proportional
to the square of the transition moment integral
Mif = J i R 4 f dr f xi xf do (1)
where M = e ri', i is the spatial eigenfunction of the ith state and
xi is the corresponding spin function. For simplicity, it is assumed
that AS = 0. The electronic transition will be an allowed one if and
only if Mif is not equal to zero.
In the case where one or both states are degenerate, the transition
will be an allowed one provided that the direct product of the species
(r) of 0i with that of 1f belongs to the same species as one of the
The selection rule as discussed above applies only for fixed nuclei.
The nuclei are not actually fixed, and the dependence of the total eigen-
function upon nuclear motion must be considered. Neglecting rotational
motion (justified physically by the fact that during an actual experiment
the molecules undergoing a transition are normally fixed in a rigid
medium) the total eigenfunction may be expressed in the Born-Oppenheimer
iik = ei(q,Q) ok (Q) (2)
where ,'ik is the vibronic state corresponding to the kth vibrational
level of the ith electronic state, q indicates all the electronic coor-
dinates, and Q stands for all the nuclear coordinates.
If the electric dipole is resolved into two parts, one due to
electrons and the other due to nuclei,
M = M + M (3)
then the electric dipole transition moment becomes
Mik,fl f ik h1 dQ f e M of dq + I k mn Q1 dQ f o of dq (4)
Since the electronic wavefunctions for a given position of the
nuclei are muturally orthogonal, the second term in the above expression
If the dependence of 0. upon nuclear coordinates is taken into
account, we obtain
Mif = ei (q,Qo) M e f (q,Q ) dq (5)
Mik,fl = Mif (Qo) f ak 1i dQ
where Q is the configuration of the nuclei near the equilibrium
Thus, the transition probability, which is proportional to M2, has
been resolved into two components, one of which depends upon nuclear
motion, and the other which depends upon electronic motion.
Following Albrecht, this may be expressed in a slightly different
form by using the Q dependence of the electronic wave functions.
of (q,Q) = e + ~ fr (Q) oe (6)
The electronic wavefunction for state f at Q is expressed in terms
of the complete set of electronic wavefunctions evaluated at Q = 0,
corresponding to the ground electronic state. The sum in Eq. (6) goes
over all states but f, where the rth member of the set is The coef-
ficient Afr serves to mix the zeroth order wavefunction e6 with e through
a vibrational motion of the proper symmetry. Such mixing will be neg-
ligible for the ground state, so that 0. (q,Q) e?. The transition
moment may then be expressed as
Mikfl f = k 1 dQ + r M r fr dQ (7)
The physical interpretation of Mik,fl is that it represents the
polarization of the vibronic transition ik fl. Equation (7) may be
viewed as being made up of two parts, an "allowed" and a "forbidden"
part. It is the allowed part of the transition which will be the most
important part in determining the nature of the electronic transition.
This is the first term in Eq. (7), and represents the modification of
combining vibrational wave functions (Franck-Condon factor) upon an
electronic transition. The second term in Eq. (7) represents the forbidden
part of the transition. A given vibronic transition will be polarized
according to the vector sense of Mf if it is electronically allowed,
or according to the vector sense of the appropriate M? if the transition
involves appropriate non-totally symmetric vibrations. This follows
from the fact that integrands in the integrals over nuclear space will
not contain totally symmetric components for both the allowed and for-
bidden components simultaneously for a non-totally symmetric Afr. There-
fore, vector addition of the two terms will not occur except when Xfr
contains a totally symmetric vibration.
In an actual experiment involving photoselection, there are essen-
tially two different methods which may be employed. In one case, there
is an active attempt to cause molecules to become oriented (as in the
case of a single crystal). The second method does not depend upon the
active orientation of molecules, but relies upon the anisotropic nature
of the exciting light. Ordinary light is isotropic. In order to render
it anisotropic, it is necessary to pass the beam through one of several
different types of polarizing media, so that the light which emerges is
a beam of plane polarized light.
Consider a rigid solution in which the fluorescent solute molecules
are randomly oriented. Assume, for the sake of simplicity, that the
transition moment for absorption coincides with that for emission. In
the absence of depolarizing factors this will correspond to an excitation
from the ground state to the first excited singlet state, and emission
from the first excited state to the ground state.
Following Parker (27), we may consider such a situation with
reference to Figure 23. The incident light is vertically polarized and
Illustration of Relative Polarization for an Electronic
- Wu -
propagating along the X axis. Consider a molecule with a transition
moment directed along OR, inclined at an angle a with OZ and an azimuthal
angle y with OY. The R' is the projection of R on the plane XOY.
Emission is viewed along OY. The probability that such a molecule will
be excited by the vertically polarized light will vary as cos2a. For a
given number of excited molecules, the intensity of the vertical com-
ponent of light viewed along OY will also vary as cos a, while that of
the horizontal component will vary as sin asin y.
The degree of polarization is defined as
P = (I, I )/(lI + Ii) (8)
where In and LL are the intensities of the observed parallel and per-
pendicular components of the emission. In the case where the absorption
and emission moments coincide exactly, P is equal to 1/2 for vertically
polarized exciting light. Perrin (93) and Jablonski (94) have shown
that if the angle between absorption and emission oscillators is B, for
vertically polarized light P will be given by
P = (3cos28 l)/(cos2B + 3) (9)
For a = 0, P takes on the value of 1/2. For B = 90, when the
absorption and emission oscillators are at right angles, P will have the
value -1/3. Differences found in practice from these values are normally
attributed to such factors as overlap of two closely spaced transitions
or external factors such as strains in the rigid glass matrix.
The polarization of the emission as a function of excitation wave-
length can be.used to establish the relative orientation of an emission
transition dipole to a particular absorption band in the absorption
spectrum of the molecule in question. This is generally done by com-
paring the polarization curve to bands in the absorption spectrum. In
the ideal case, there will be a region of approximately constant polari-
zation for a given absorption band, followed by a region of changing
polarization where two absorption bands overlap, and then another region
of approximately constant polarization over most of the second absorption
band. The polarization may change sign or not depending upon whether
the two absorption bands have mutually perpendicular transition moments.
The experimental procedures used to obtain Shpolskii matrix spectra,
photoselection measurements of quinizarin and daunorubicin, and lifetime
measurements for quinizarin are described in Chapter 2. Reagents and
equipment utilized in this set of experiments are also described in this
Infrared absorption spectra of quinizarin in the range 4000-250 cm-
were obtained in KBr disks on a Perkin-Elmer 621 spectrometer. Spectra
in Nujol mulls over the region 1800-400 cm-l and in the far I.R. were
obtained on a Digilab FTS-20C Fourier Transform I.R. spectrometer.
It has been suggested that dimerization takes place in the con-
densed phase in hydroxy anthraquinones (95). In order to determine
whether the emission and absorption spectra obtained were from quinizarin
molecules or from dimers, room temperature luminescence and absorption
spectra were run as a function of concentration in n-hexane and methanol
solvents. A Cary 17 spectrophotometer with a pair of matched 1 cm
path-length cells was used for absorption measurements, while a