ION PAIRING Al.IiL H'DF.,:9EIf B':lLiDING IN THE EXCITED STATE
OF ALKALI CARBANION SALTS
M. JOHN PLODINEC
A DISiEF.TATIlCl PRESENTED TO THE GRADUATE COUNCIL OF
THE Ii'. -F'PSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIF.EllIEfJl
FOR THE :,ELF EL OF DOC T,.i:F OF PHILOSOPHY
UtiIV'EFSITY OF FLORIDA
For those who wanted to come home again. .
Digitized by the Internet Archive
in 2011 with funding from
University of Florida, George A. Smathers Libraries with support from LYRASIS and the Sloan Foundation
The author would like to take this opportunity to thank all the
members of his Supervisory Committee, Dr. Wallace Brey, Dr. Gardiner
Myers, Dr. George Butler, and Dr. Stephen Schulman, for their aid,
encouragement, and counsel. Special thanks must go to Dr. Schulman,
both for allowing his equipment to be used, and for his many helpful
Thanks are due Jimmie McLeod and Lynn Williamson for their heroic
attempts to read the turgid style and illiterate scrawl in which this
dissertation was written.
Thanks to the Boss, for putting up with the gripes and the grop-
ings, clumsiness and, often, ignorance, of this theoretician turned
Finally, thanks is due to the author's wife, Louise; she made each
day a little better.
TABLE OF CONTENTS
Acknowledgements . .... .. iii
Abstract . . .. .. .. v
I. iiJT DUCTION . . 1
II. EXPERIMENTAL PROCEDURES . .. 12
Preparation and Purification of Sample Systems .. 12
Spectral Measurements . . 14
III. CATION AND SOL\ '!TI EFFECTS . 22
Fluorenyl Systems: Experimental Results and Discussion 22
Fluoradenyl Systems: Experimental Results and Discussion 42
The Radical Anion of Anthracene: Results and Discussion 55
IV. ..i.'ii lF TO GENERATE CARBANIONS FROM EXCITED HYDROCARU'ij3 58
V. AGGREGATION EFFECTS ON (.- r J ICil FLUOFE' Wi E ... 66
VI. GENERAL DISCUSSION AND SUMMARY . .. 86
Cation and Solvent Effects . .. 86
Radical Anions . . 95
Aggregation Effects .. . 97
Ionization . . 98
Appendix 1: INNER FILTER EFFECTS . .. 100
Appendix 2: POPULATION ANALYSIS OF THE FLUORENYL ANION BASED ON
F'i itL CALCULATION . .. 103
Appendix 3: EVIDENCE FOR THE AGGREGATION OF FL'.':: D.ENE IN PROTIC
SOLVENTS . . 105
References and Notes . . 107
Biographical Sketch . . 112
Abstract of a Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment
of the '-'equ;rements for the Degree of Doctor of Philosophy
ION PAIRING Ai HYDROGEN FiH.iDING IN THE EXCITED STATE
OF ALKALI CARBANION SALTS
M. John Plodinec
Chairman: Thieo E. Hogen Esch
Major Department: Chemistry
The fluorescence and excitation spectra of the alkali metal salts
of the anions of fluorene and fluoradene, and the radical anion of
anthracene, were studied at room temperature in protic and aprotic
solvents. As expected, the excitation spectra were usually identical
to the absorption spectra of these salts, and displayed the same be-
havior with changing cation and solvent.
The shifts in the fluorescence maxima of the salts in aprotic
solvents are explained in terms of an equilibrium between contact
and solvent-separated ion pairs, the proportion of the latter increa-
sing as the cation is changed from a larger to a smaller, as the sol-
vent is changed from a poorer-to a better solvator of cations, or as
a cation completing agent, such as a crown ether, is added.
At smaller salt concentrations in ether solvents of low dielec-
tric constant, free ions were observed. For one such system, fluora-
denyl sodium in tetrahydropyran (THP), a dissociation constant was
calculated from the excitation spectra which agreed reasonably well
with the value obtained from conductance measurements.
Significant effects due to cation-solvent interactions were also
observed in the lifetimes and relative intensities of these salts.
fi L:e are explained in terms of a "normal" heavy atom quenching effect,
which should decrease in importance from cesium to sodium, and another
effect increasing in importance from cesium to sodium. Several differ-
ent detailed mechanisms for this second effect are discussed.
Excited fluoradenyl sodium was investigated in protic solvents,
and red shifts (higher wavelengths) in the fluorescence maximum seen,
compared to the free anion in acetonitrile. This is explained in terms
of hydrogen bonding to the free anion. In mixed ether (THP)-alcohol
(n-propanol) solvents, a similar red shift was seen. However, upon
addition of a cation completing agent, the maximum shifted back to the
position of a normal separated ion pair. This is interpreted in terms
of the cation assisting in hydrogen bond formation.
Unsuccessful attempts were made to generate carbanions from excited
hydrocarbons. The reasons for these failures are discussed and used to
explain the solvent dependence of the acid dissociation constant of
fluoradene in terms of aggregation of the hydrocarbon in protic solvents.
Finally, the effects of aggregation on the absorption and fluor-
escence spectra of carbanion salts were examined, by applying simple
exciton theory to bisfluorenyl barium in THP and tetrahydrofuran (THF).
Detailed structures are derived for the anion dimer which are reason-
able in view of the greater cation solvating ability of THF. Qualita-
tive statements, based on exciton theory, are made about structure of
fluorenyl-alkali metal salt aggregates, and certain unusual spectral
I!; l i'. AUCTION
:'there'.e-r a salt is dissolved in a solvent, dissociation of
the salt into its free ions may not go to completion. Depending
on such factors as the charge of the ions, their size, the dielectric
constant of the solution, the ability of the solvent to solvate
any or all the individual ions, and the concentration and ionic
sti r.rth of the electrolytic solution, the degree of ioni.in may
be nearly unity or almost nil.
However, in order to fully characterize electrolytic solutions,
it may be necessary to invoke the presence of other species. The
non-dissociated ion pairs may associate with themselves to form
neutral a''reates such as dimers, trimers, or, in general, n-mers.
At the same time, the non-dissociated ion pair may associate with
free ions to form charged ac'r--gtes such as triple ions. The
chemical behavior of such species should be highly dependent upon
their structure, but, except for certain dye molecules at high
concentrations, the structure of such associated species has
not been extensively examined.90 Also, the free ions, or the
ion pairs, may interact with the solvent to form charge-transfer
species, or, in protic media, hydrogen bonded species, either of
which may also affect the chemical behavior of the electrolyte.11
To further complicate this picture, the non-dissociated ion
pair may exist in two forms, contact and solvent separated ion pairs.
The latter species, first invoked by Winstein to explain solvolysis
phenomena,15'16 may be thought of as the result of the diffusion of a
single layer of solvent molecules between the anion and the cation of
a contact ion pair. This species still travels through the medium
as a single entity, as would a contact ion pair, but also may exhibit
some of the drastically different behavior expected of free ions.
This concept of a solvent-separated ion pair has been of great import-
ance in explaining such diverse phenomena as the mechanism and stereo-
chemistry of organic reactions, the rates of initiation and
propagation of ionic polymerizations,119 the electronic and vibra-
tional absorption spectra of organic and inorganic salts, the
electron spin resonance spectra of radical ion salts, and the
nuclear magnetic resonance spectra of certain salts.2
In Figure 1 is a pictorial presentation of the different possible
forms of the ion pair, and a plot of potential energy vs. inter-
ionic distance for a simple 1:1 electrolyte in a medium of dielectric
constant 20, originally due to Grunwald. The -l;::ical basis of
the Grunwald scheme is as follows. Assume two free ions in solution,
infinitely separated. As they approach, the potential energy of
the system decreases. However, each ion may have a solvation shell
which will be compressed as the two ions approach, this compression
requiring energy. At some point, the energy necessary to compress
the solvation shell further will be greater than the stabilization
of the system due to the closer approach of the ions, thus causing
an increase in the potential energy. As the two ions continue to
approach, the energy of compression of the solvation shells will,
at some distance, be the same as the energy of formation of this
shell, and the ions will collapse into the contact pair, i.e. the
solvation shell will be squeezed out. Thus, one may visualize at
least two other distinct chemical entities, as well as free ions:
one, corresp: nl1r-,nl to the complete collapse of the free ions, the
contact ion pair; the other, corresponding to partial collapse of
the individual free ions but with the maintenance of a layer of
solvent molecules between them, the solvent-separated ion pair.
It must be noted that while the difference between contact
and separated ion pairs has been presented as between two species,
there is evidence for two families of ion pairs,22 since both the
contact and solvent-separated species may exhibit varying amounts
of peripheral solvation. A compilation of the various possible
equilibria is given in Figure 2.
The foregoing has dealt with well-known ground state phenomena;
there is no reason to assume a priori that these same considerations
will not hold true in the electronically excited state of an ion
pair as well. Indeed, recently there have been several attempts to
explain data on excited molecules in terms of dissociation of
ionized excited species into free, or hydrogen bonded, ions. 37
However, while the presence of ion pairs has been postulated, there
has been no systematic investigation to determine the validity of
this postulate; and, thus, there has been some skepticism shown.38
The presence or absence of ion pairing phenomena in such excited
state processes as electrochemiluminescence could play a critical
role in both the qualitative and quantitative understanding of these
Further, by studying ion pairing in the excited state, one
n(solvent molecules) M IA
ion pair solvent-separated ion pair
ion pair t< solvent-separated ion pair
<- free ions
M+I IA- M + A
vet- te M1 + A
on pair free ions
an pair *- free ions
2 -A + + + +A +
2M A (M-A ),2 M A- + (M A )2 (MA )3, etc.
aggregation to form n-mers
S -. +- + + -
2M A- M+AM + A or M + A-MA
triple ion formation
Figure 2. Equilibria possible in ionic solutions.
could use this information to elucidate the nature of the other,
more specific, phenomena of aggregation and hydrogen bonding
in the ground state, referred to above. Intimately bound with these
aims would be the effort to determine similarities and differences
between ground and excited states, the effect of electronic excitation
on their ion-pairing properties, and to examine at least some of
the pathways available to the excited state to allow it to return
to the ground state.
Thus, the goals of the present work, broadly stated, are the
(1) The determination of how far the validity of the concept
of ion pairing extends for the excited state.
(2) An investigation of the usefulness of information about
the excited state of ion pairs for the determination of specific
ground state phenomena, such as dissociation, aggregation, and
(3) An examination of the differences between the ground and
excited states of ion pairs and the role of cation-solvent relaxation
processes, in these differences.
(4) An examination of the "deexcitation reaction," i.e. attempting
to show what factors determine how fast, and in what manner, the
excited state ion pair returns to the ground state.
Some of the most extensively investigated systems exhibiting
ion pairing in the ground state are the alkali metal salts of fluorene.
As shown by Hogen Esch and Smid,39 in low dielectric constant media,
with decr-: -inri temperature or changing from a poorer to a better
cation solvating medium, a second peak appears in the absorption
spectra, due to the separated ion pair. Tihu:, the absorption spectra
of these salts are sensitive indicators of cation and solvent effects
in the ground state. Consequently, it was thought that their fluores-
cence spectra would give the same sort of information about the
excited state in such media as tetrahydrofuran (THF), tetrahydro-
pr'ran (THP), 1,2-dimethoxyethane (DME), dioxane, and toluene. Also,
macrocyclic polyethers such as dicyclohexyl-18-crown-6 (2,5,8,15,18,
21-hexaoxatricyclo[18.104.22.168]hexacosane), a crown ether, were used to
obtain loose ion pairs, especially under conditions where they would
not otherwise be formed.220 Thus, these systems should be useful
in determnirinc the validity of ion-pairing for the excited state,
looking at cation-solvent relaxation processes, and examining the
Further, the bisfluorenyl barium salt should be a good model
system for a triple ion or ion pair dimer, since: (1) conduc-
tance studies indicate that one is dealing with essentially only one
species (there is no evidence for higher aggregates and the first
dissociation constant is low, Kd = 3 x 10 1/mole, in THF), and (2)
some data are already available about its structure in solution. This
could be applied to the lithium and sodium fluorenyl salts in dioxane,
and lithium fluorenyl in toluene, which are all believed to be
aggregated on kinetic grounds.39
In order to more meaningfully discuss radical ion processes,
the sodium and cesium salts of anthracene were investigated in THF,
THP, and THP-glyme mixutres. These systems are known to exhibit
ion-pairing in the ground state, and are well characterized.44-48
ion-pairing in the ground state, and are well characterized.
In order to examine hydrogen bonding to excited state carbanion
salts, the alkali salts of fluoradene were investigated. Because of
the relatively great acidity of the hydrocarbon, this anion can
exist in a much greater variety of solvents than the fluorenyl anion,
and has been shown to hydrogen bond in the ground state, the
cation playing a significant role in the hydrogen bonding of the
non-dissociated salt. Thus, the fluoradenyl salts were investigated
in THF, THP, DME, dioxane, acetonitrile, methanol, ethanol, n-propanol
(n-PrOH), n-propylamine (n-PrNH2) and t-butanol (t-BuOH). See Table 2
and Figure 3.
Figure 3. Chemicals
Table 1. Summary of ion-pairing in the ground state of alkali
fluorenyl salts at room temperature
Cation Solvent Type ion pair Principal absorption maximum
20% C :
95% C :
20% C :
Na (CE)c'd THF
348, 371 (shoulder)
346, 371 (shoulder)
50% C : 50% S
50% C : 50% S
SC = contact, S = separated, F = free.
Data from reference 39, supplemented by author.
d CE = slight excess of dicyclohexyl-18-crown-6 present.
Data from reference 40, supplemented by author.
Data from references 41 and 43.
Table 2. Summary of ion pairing in the ground state of alkali
fluoradenyl salts at room temperature
Cation Solvent Form of the ion pair Absorption maxima
=50% C : 50%
- :'.. C : 20%
388, 530, 570
361, 374, 512, 547
367, 378, 518, 553
361, 376, 525-40c
350, 360, 388, 495, 520,
340, 350, 495, 523
a Data from reference 13, supplemented by
C = contact, S = separated, F = free, H
= hydrogen bonded.
Preparation and Purification of Sample Systems
Tetrahydrofuran (THF), tetrahydropyran (THP), and 1,2-dimethoxy-
ethane (DME) were purified by refluxing over sodium-potassium alloy
for about 12 hours, then distilled onto fresh alloy. A small amount
of benzophenone was added, and the resultant purple dianion solution
degassed on a vacuum line. The benzophenone anion acted as an
indicator of the presence of water or oxygen.3
Dioxane was refluxed over CaH2 for approximately 12 hours, then
fractionally distilled and sodium-potassium alloy added. A small
amount of fluorenone was added, and the resultant green solution
degassed on the vacuum line.39
Methanol (MeOH), ethanol (EtOH), and n-propanol (n-PrOH) were
refluxed over magnesium filings activated by iodine for approximately
three hours, then distilled under vacuum, and degassed.51
Toluene', pyridine, n-propylamine (n-PrNH2), hexane, and t-butanol
(t-BuOH) were stirred over CaH2, for 12 hours, distilled under vacuum
onto fresh CaH2, stirred, degassed, then distilled under vacuum and
Acetonitrile was stirred over CaH2 for 12 hours, distilled
under vacuum onto P205, stirred for 12 hours, and distilled again
under vacuum into an ampoule of lithium fluorenyl.52
Deionized water was degassed by distilling under vacuum and
freezing the distillate, pumping on the resultant solid, then allowing
the solid ice to melt. This was repeated three times.
Fluorene was recrystallized from absolute ethanol; fluoradene
from hexane. Purity was checked by melting point, and ultraviolet
Fluorenyl and fluoradenyl salts were prepared from the corres-
ponding salts of the 1,1,4,4-tetraphenylbutane dianion (TPB ), usually
in THF, which were available in the laboratory. Transfer of the salt
to other solvents was achieved by evacuating the THF solution to
ultimate vacuum (about 107 torr), distilling the desired solvent
onto the salt under vacuum, mixing, reevacuating, then adding more
of the solvent desired. As an extra precaution, solvents purified by
the various means above were usually added to a dry salt sample. If
there was any decoloration of the salt, the solvent was repurified.
If not, the solvent was distilled from the solution to the salt
sample of interest, under vacuum.3
Anthracene was recrystallized from n-propanol, then dried in
vacuo. Sodium radical anion salts of the hydrocarbon were formed
by reacting a solution of the hydrocarbon with a sodium mirror
under vacuum. The cesium salt was formed by reacting the hydrocarbon
in THF with the metal, under vacuum.
All solutions were stored under vacuum in ampoules equipped
with break-seals. When not in use, all samples were kept in a
freezer at -200 C, where they usually were stable.
The crown ether used was dicyclohexyl-18-crown-6, obtained
from Dr. H.K. Frensdorff of E.I. du Pont de Nemours Elastomers
Department, and recrystallized from petroleum ether.53 Later
samples were recrystallized from acetonitrile and stored under vacuum.
Samples of crown ether were added to salt solutions by means of
evacuated break-seals; if any decoloration or significant loss in
optical density occurred, the samples were not used. Due to their
low solubiliby, especially in THP, the crown ether-salt samples
were usually filtered before use. (See Figure 3.)
Reagent grade sodium tetraphenylborate was purified according
to a modification of the method of Skinner and Fuoss, as follows.
The salt was partially dissolved in an eight-to-one mixture of
methylene dichlbride and acetone. The solution was filtered,
and toluene added until a white precipitate started to appear.
The mixture was then immersed in a dry ice-isopropanol bath, and
the white precipitate collected on filter paper. The solid was
placed in an ampoule and dried on the vacuum line for approximately
two hours. This procedure was necessitated by the destruction of
fluorenyl samples by the reagent grade salt, which smelled like
phenol. After purification in the above manner, the sodium tetra-
phenylborate did not destroy anion solutions, even when added in
excess by a hundred-fold, to determine common ion effects.
Salt samples were usually formed in an apparatus similar to that
of Figure 4, at a concentration of about 10 M, in the following
manner. After the apparatus was built (all glass except for the
quartz optical cells and the spacer), it was attached to the vacuum
line and tested for pin-holes with a Tesla coil (and repaired, if
necessary). It was then flamed out, evacuated to about 10-7 Torr,
and sealed from the line at constriction a. The hammer and an
external magnet were used to break the break-seals of the fluorene
and TPB solutions' ampoules and the two were mixed. The absorption
spectrum of the resulting solution was taken with a Beckman Acta V,
in the range 325-600 nm, in the 2 mm cell with either a 1.8 mm or
1.9 mm spacer, to determine the concentration. A typical absorp-
tion spectrum (of sodium fluorenyl in THP) is reproduced in
The solution was then poured through the constriction b, and
the walls of the apparatus "washed" with solvent, by application
of a dauber, dipped in liquid nitrogen, to the outside. After the
walls were clean, the receiver was sealed away from the rest of the
apparatus at b.
Dilutions of the sample were accomplished by pouring most of the
solution into the sidebulb, through constriction c, and distilling
solvent back into the cell by application of a cold dauber. Concen-
trations less than 10- M could be calculated from the visible and
near ultraviolet spectrum, and known extinction coefficients.39
Fluorescence emission and excitation spectra were taken on
a Perkin-Elmer MPF-2A spectrofluorimeter in the ratio record
mode, courtesy of Dr. Stephen G. Schulman of the College of Pharmacy,
in the following manner. One of the principal absorption maxima
was chosen as the exciting w3%elenrith, and the emission spectrum
manually scanned to find the maximum. Then, holding the wave-
length of emission fixed, the excitation spectrum was scanned manually
to find an optimum excitation wavelength. At this point, exciting
with light of the optimum wavelength, the emission spectrum was
V) r ,
+, 4- A
0 U u0
(Q (1) -1
*H 0 +-i 0
0-i HO 0
F 0 0 0
H O 0F
0 0C < < r-
) Q C [D
Absorption spectrum of fluorenyl sodium in THP.
scanned and recorded. Then, selecting an emission wavelength of
significant intensity, the excitation spectrum was scanned and
Lifetimes were measured by Mr. Anthony Capomacchia of Dr.
Schulman's group at the College of Pharmacy, on a TRW nanosecond
decay time fluorometer, using a pulsed nitrogen lamp and a Tektronix
556 dual-beam oscilloscope with two IAI plug-in dual-channel ampli-
The values given here represent the lifetimes obtained from
at least two different concentrations of the same salt (except for
cesium fluorenyl, which was anomalous). The accuracy of the life-
times of the fluorenyl salts is probably much less than that of the
fluoradenyl salts for the following reasons. In the systems studied,
there was always residual hydrocarbon present, either fluorene or
fluoradene. However, there was never any evidence of the formation
of an excimer of fluorene, meaning that the output signal of the
irradiated solution always contained a component attributable to
the hydrocarbon. For the salts of fluorene with lower lifetimes,
this was a major source of error. Thus, the data are considered
to be no better than 10% and probably no worse than 25%, with the
longer lifetime salts being most accurate. For the fluoradene
salts, however, the accuracy was probably nearer 10%, since, in the
solvents examined, there was very little hydrocarbon monomer emission;
the fluoradene hydrocarbon mainly emitting through an excimer state
of much lower intensity, relative to the fluoradenyl salt emission,
then the intensity of the fluorene monomer relative to its salts.
This difference could be easily distinguished by visual comparison
of the oscilloscope signals of the fluorenyl and fluoradenyl
salt systems. Little use is made of the absolute numbers, in any
event, and the general trends noted are of greater importance.
Relative intensities were obtained either by comparison of
peak height to an internal standard (the free ion for the fluoradenyl
salts; the crown ether-separated, or solvent-separated ion pair for
the fluorenyl salts), or by comparison of peak heights between two
different salt solutions at known concentrations. This is a less
accurate procedure than the former, since different samples might
have different concentrations of quenching impurities. However,
results obtained in this manner were reproducible to within 20%.
Implicit in the above work for the fluorenyl ion pairs was the
assumption that there was no difference between a solvent-separated
ion pair and a crown-ether-separated ion pair. To check this, a
solution of fluorenyl sodium in DME (20% contact, 'i^, solvent-
separated ion pairs in the ground state) was prepared, and its
fluorescence spectrum compared to that of the same solution to
which a slight excess of crown ether had been added. There
was no difference in terms of peak positions (528, 568 nm for both)
or peak heights, which justified the assumption.
After a series of spectra had been obtained for a particular
salt, using the salt in the receiver part of the apparatus in
Figure 4, weighed amounts of reagents such as crown ether or common
ion could be added by using the hammer and an external magnet to
open the appropriate break-seal, mix the salt solution with the
.re-w.it~hed solid contained in vacuum, filter the solution through d,
and repeat the series of spectra. After the completion of an
experiment, the apparatus could be turned on its side so that the
side-bulb was down, and the solution poured into the side-bulb, the
tubing "washed" around constriction c, and the solution sealed away
from the rest of the receiver apparatus and stored in the freezer.
CATION AND SOLVENT EFFECTS
Fluorenyl Systems: Experimental Results and Discussion
The first systems investigated were the alkali metal salts of
fluorene. Typical emission spectra are shown in Figure 6, those
of fluorenyl sodium (NaFl) in THP, at different concentrations.
All the emission spectra of the fluorenyl systems displayed two
peaks as shown, so that it is highly unlikely that they represent
two different species. Further, their relative intensities, at
a given concentration, were unaffected by the addition of common
ion or mode of preparation, and they persisted, with about the
same relative intensities, from 10 M. down to the lowest concentra-
tion studied (109 M). For these reasons, it was concluded that the
doublet arose from emission from the lowest vibrational state of
the first excited state (S ) into two vibrational states of the
electronic ground state. Additional evidence for this lies in
the fact that, according to Berlman,5 the parent hydrocarbon,
fluorene, also has two peaks in its fluorescence spectrum. Also,
the separation of the two peaks (at least below 103 M) is constant
at 1240 + 10 cm near where the hydrocarbon, and fluorenyl-
calcium chloride57 have both been reported to have a vibration
of appropriate symmetry to couple with the electronic transition
(1277 and 1219 cm respectively).
At higher concentrations (102 M to 10 M), both the position and
relative intensity of the two peaks are dependent upon concentration.
Emission wavelength, nm
Effect of concentration on the emission spectrum of fluo enyl
sodium in THP. A [NaFl] = 2x10 M; B5 [NaFl] = 6.5x10 M;
C. [NaFl] = 2x10 M; D. [NaFl] = 4xl0 M.
As the concentration decreases toward 104 M, both peaks shift to
lower wavelengths, and the lower wavelength peak gains in relative
intensity. At concentrations below 10 M1, while some shifts in
the position of peaks are still observed at lower concentrations,
the relative intensities of the peaks are now independent of concen-
tration. This is shown graphically for the fluorenyl salts in
Figure 7, where the ratio of the lower to the higher wavelength
peak heights is plotted as a function of concentration, for several
of the salts.
If one examines the excitation spectra of these salts (see
Figure 8), as a function of concentration, one finds that in the
high concentration region, above 10- M, anomalous spectra are
obtained. However, at concentrations below 10 H, the spectra
are nearly identical to the absorption spectra, as expected589
(although there are significant differences in relative intensities,
which will be discussed later).
There are, basically, two important causes of the above phenomena.
First, reabsorption processes must be expected to play a significant
role. For example, for sodium fluorenyl in THF, while the first
absorption maximum occurs at 486 nm, there is significant absorption
e-en at 530 nm (c530 150). Under the conditions of the emission
experiments, there should be a great deal of reabsorption of emitted
light at the lower wavelengths. Assuming the average path of an
emitted photon to be 0.5 cm, for sodium fluorenyl in THF, 95%
transmittance of the fluorescent beam would not be achieved until
concentrations below 3 x 10 M. Thus, as concentration is decreased
there should be an increase of intensity at lower wavelengths as
-1 -2 -3 -4 -5 -6 -7
-1 -2 -3 -4 -5 -6 -7
I0 0 1 1 I
-2 -3 -4 -5 -6 -7
Peak height ratio as a function of concentration for several
fluorenyl salts in THP. A. Fluorenyl sodium with an excess
of crown ether; B. Fluorenyl sodium; C. Fluorenyl potas-
sium; D. Fluorenyl cesium.
Effect of concentration on the excitation spectrum of
fluorenyl sodium in THP.
A. [NaFl] = 2x10 M;
B. [NaFl] = lxl- M;
C. [NaFl] = 2xl0 5M;
D. [NaF1] = 4x10 M
(identical spectra for still lower concentrations).
Excitation Wavelength (in nm)
more of the lower wavelength fluorescence passes through the solution
without reabsorption, which is observed.
The effect of concentration on excitation spectra is less well
defined, but, as shown by McDonald and Selinger,6 for high absorbance
solutions there should be peaks in the excitation spectrum corres-
ponding to troughs in the absorption spectrum, and the results
should be dependent upon the geometry of the sampling system.
Thus, for high absorbance solutions, if the incident beam must
pass through the solution, it will be attenuated so that most of
it will be absorbed near the front of the fluorescence cell; i.e.
the solution will act as a filter, and most of the emission
produced will be near the front of the cell, and out of view of the
detection photomultiplier of a conventional spectrofluorimeter
employing right angle geometry.
In appendix 1, it is shown that, given the right-angle geometry
of the spectrofluorimeter, the change in the excitation spectrum
with concentration is that expected for the salt, assuming this
"inner filter" effect.
Another factor in the behavior of the salts at high concentration
is the formation of triple ions, and higher aori-ctes. Since this
will introduce a much greater degree of complexity, discussion of
the effect of -'r-aotion will be postponed until the behavior of
bisfluorenyl barium is examined.
Another complication is the possibility of excited complex
formation. 1-6 By addition of a ten-fold excess of fluorene, it
was shown that if an excited complex was formed, fluorene was not
involved, since there was no change in the fluorescent behavior
of a solution of sodium fluorenyl, in THF.
The Effect of Cation
In order to determine the effect of cation on the fluorenyl
emission, the fluorescence spectra of the alkali metal salts in
THP were taken at concentrations below 10-4 M. A..-'.r~ld to Table
1, the sodium, potassium, rubidium and cesium salts are entirely
contact ion pairs in the ground state. The emission results are
contained in Table 3.
Hogen Esch and Smid explained the shifts seen in the absorption
spectra of these salts in the following manner. The anion, in the
ground state of a contact pair, is stabilized by the cation, which
occupies its equilibrium position with respect to the ground state
charge distribution. Upon absorption of light, the new electronic
configuration of the anion is rapidly attained (=0-1 sec), but,
in accordance with the Born-Oppenheimer approximation, the cation
does not have time to move to its new equilibrium position with
respect to the excited anion, which, therefore, is not as stabilized
by the cation as is the ground state. Thus, the energy difference
between the ground and excited states is increased relative to the
free ion, and this increase should be greater the greater the cationic
field, i.e. the smaller the radius, for alkali cations; thus, the
absorption spectra should be blue-shifted (shifted to lower wavelengths)
for contact pairs going from cesium to lithium (See Figure 9).
The above assumes that there is a sufficient difference in the
charge distribution of the ground and excited states to cause cation-
anion reorientation. As .pointed out by Birks and Dyson,65 the lack
of mirror symmetry between the absorption and fluorescence spectrum
i Free Anion
-- Contact Ion Pair
-- --- Free Anion
Contact Ion Pair
AE1 > AE AE2 < AE
1 o 2 o
Figure 9. The effect of cation.on the spectra of contact ion pairs.
1) C) Cn cl) u u
** u 0 0 u u u
II O *,-
0 o w
nUE (4 -
0 C0 U
tO X3 U '
of a compound is a sensitive indicator of changes in the electronic
distribution in that compound between the ground and excited states.
A comparison of Figures 5 and 6 would indicate such a lack. Further,
as shown in Appendix 2, simple Huckel calculations for the fluorenyl
anion also indicate major changes in the electronic distribution
of the anion in going from the ground to the first excited state.
Analogous reasoning should explain the shifts in the fluorescence
spectra, if one assumes that the lifetime of the excited state is
long enough to permit the cation to reach its equilibrium position
with respect to the excited anion (see Figure 9). During emission,
the cation does not have time to reach its ground state equilibrium
position, and the excited anion may be more stabilized than it is
in its ground state just after emission. This means that the energy
difference is now decreased relative to the free ions, this difference
being greater the greater the cationic field, i.e. the smaller the
cationic radius. Thus, a red shift (shift to higher wavelengths)
would be expected for a series of contact ion pairs as the cationic
radius is decreased, i.e. going from cesium to sodium fluorenyl in
THP, with lithium open to question in THP, due to the significant
amount of solvent-separated ion pairs present in the ground state
(see Table 1).
From Table 3, it is obvious that the expected shifts do occur
from cesium to sodium, but that lithium fluorenyl emits at 528 nm.
The position of this peak was unaffected by the addition of lithium
tetraphenylborate, a source of lithium ions, so that the possibility
of dissociation of the contact pair into free ions in the excited
state seems unlikely. To further identify the emitting species in
this case, the fluorescence spectrum of lithium fluorenyl in THP
was compared to those of both the sodium and potassium salts to which
had been added a slight excess of dicyclohexyl-18-crown-6. Since all
three have the same emission maximum, 528 nm, it seems safe to identify
the emitting species in the lithium fluorenyl case as the solvent-
separated ion pair.
This significant finding justifies the assumption that the
lifetime of the excited state is long enough to permit the cation
to attain its equilibrium position with respect to the new charge
distribution of the excited anion, before it emits. Not only is
there enough time for the cation to move to its new position, but
there is enough time for a layer of solvent molecules to diffuse
between cation and excited anion. As will be seen later, the measured
-7 -8 66
lifetimes of the excited state (10 -10 sec) are orders of magnitude6
longer than solvent relaxation times (10 -10 sec).6
From Huckel calculations (Appendix 2), it is to be expected
that the excited fluorenyl ion pair should be somewhat looser than
the ground state one. Assuming the cation to lie above the cyclo-
pentadienyl ring in solution,3 the ground state anion has almost
two-thirds of the negative charge on those five atoms, while the
excited anion has less than one-third there.
Although the constant for the equilibrium between the excited
contact and the excited separated ion pair cannot be measured in this
case, its value can be estimated from the spectroscopic data, by use
of the so-called Forster cycle,61'66 with the known value of the
equilibrium constant in the ground state.
In Figure 10, this cycle is shown as it specifically pertains
M I Fl
AG* = AG + AG AG
0 S c
Figure 10. Forster cycle and ion pairing in the excited state.
to the equilibrium between contact and separated ion pairs in the
excited state. Denoting the difference in free energy between the
ground and excited state of a contact and separated ion pair by
AGC and AGS, respectively, the free energy difference for the excited
state process, AG*, is related to the free energy difference for the
ground state process, AGo, by:
AG* = AG + AG AG
0 S C
If the entropy difference for the process is about the same in both
the ground and excited states, then AGS AGC can be approximated by
the enthalpy differences: AGS AGC z AHS AHC. Since the enthalpy
difference between the ground and excited states of the contact or
separated pair in solution is virtually identical to the internal
energy, AE, which can be approximated by averaging the 0-0 lines of
the absorption and emission spectra, it can be shown that:41
pK = pK -
where pK is the negative common logarithm of the equilibrium constant,
h is Planck's constant, c is the speed of light, k is the Boltzmann
constant, T the temperature (in K), and VC, vS the average of the
0-0 lines of the absorption and emission spectra for the contact
and separated ion pairs, respectively. If the value of the ground
state equilibrium constant is 3/7, then pK* = log (7/3)-2.35 = -2.08,
or, K* = 120, which is in striking accord with the fluorescence spectrum.
The Effect of Solvent
A compilation of the behavior of the alkali metal salts of
fluorene in different solvents is given in Table 4, as well as
assignments of the type of ion-pairing in the ground and excited
Hogen Esch and Smid39 explained the effect of solvent on the
absorption spectra of these salts in the following manner. For a
contact ion pair, as the solvent is changed to one better able to
solvate cations, it decreases the amount of perturbation of the anion
by the cation, and the absorption spectra will shift toward that of
the free ion. Thus, the lithium fluorenyl contact ion pair absorbs
at 343 nm in toluene, 346 nm in dioxane, and 349 nm in THF.
This greater cation solvating ability of one solvent over another
may also manifest itself as an increase in the amount of separated
ion pairs present. Thus, sodium fli--r.-:'l absorbs at 355 nm in THP,
absorbs at 356 nm with a shoulder at 372 nm in THF, and at 373 nm
with a small peak at 358 nm in DME, reflecting an irneia-in, amount
of separated ion pairs, and hence, a greater cation solvating
ability of these solvents.
This same rationale, as can be seen in Table 4, seems to hold
true equally well for the excited state. Indeed, the same order of
cation coordinating power can be obtained from the table as was
found by Hogen Esch and Smid:39
DME > THF > THP > Dioxane > Toluene.
However, this is not the only explanation possible, and other
explanations will be examined in the General Discussion.
As noted in Table 4, the position of the sodium salt in THF
seems somewhat anomalously shifted, relative to the same salt in
THP. Since both contact and solvent-separated ion pairs are present
Table 4. Effect of solvent on the fluorescence of alkali metal
salts of fluorene, at concentrations below 10-4 M.
Emission Type of ion paira
Cation Solvent Maximum(nm) Groundb Excited
l, l -
70% C : 30% S
,.,. C : 80% S
95% C :
20% C :
5% S 25%
a C = contact, S = separated, F = free.
b See Table 1.
c System is j-prea3te'd, see text.
d From F6rster cycle calculations, see text.
e Seen in solutions of sodium fluorenyl in THF, at concentrations
below =10-6 M.
C : 75% S
in the ground state, the possibility of an excited state equilibrium
is indicated. Addition of sodium tetraphenylborate, a source of
common ion, had no effect on the emission maximum, which indicates
free ions are not involved. Further, a combination of a contact
ion pair spectrum (such as sodium fluorenyl in THP) with a separated
ion pair spectrum (such as the crown etherate of sodium fluorenyl)
in a ratio of 1:2 yields a spectrum nearly identical to that of
sodium fluorenyl in THF.
Since the ground state equilibrium constant is known (0.064),
a F6rster cycle calculation could be performed, giving pK* = -0.538
or K* = 3.4. Thus, it seems likely that the fluorescence spectrum
of sodium fluorenyl in TH[ is composed of the emission from both
types of ion pairs.
Lifetimes and Relative Intensities
Lifetimes and relative intensities for several of the alkali
fluorenyl systems at the same concentrations (1.10-5 M) are listed
in Table 5. For all the salts examined, except that of cesium,
the lifetime at concentrations above 1.104 M was considerably
lower than the listed value. For example, the lifetime of sodium
fluorenyl in THP at 2.104 M is 30 ns, and at 6.104 M is 24 ns.
However, at concentrations below-10- M, further dilution left the
lifetime of the salt unchanged.
The cesium salt, on the other hand, showed a continued decrease
of lifetime with concentration throughout the concentration range
studied. However, in light of the excess of fluorene present in
all systems, it is possible that it interfered with the cesium
results, since the lifetime of fluorene is comparable.55
The general behavior of the salts, in terms of relative intensities
at the emission maximum, is the same. As Table 5 indicates, the
free ion has the longest lifetime and emits most intensely; the solvent-
separated, or crown ether-separated, ion pair emits nearly as intensely
and has nearly the same lifetime; the sodium, potassium, and rubidium
salts all have nearly the same intensity and lifetime; while the
cesium salt, and the lithium salt in dioxane are of low intensity
emitters, with the shortest lifetimes.
The general behavior can be explained as a combination of three
effects. The low lifetime and intensity of the lithium salt in
dioxane, a system which is probably aggregated (on the basis of
kinetic data39), will be considered in greater detail later.
The anomalously low emission intensity and lifetime of the
cesium salt is probably due to the so-called heavy atom effect,
whereby atoms of high atomic number cause a breakdown of the spin-
selection rules, and thus enhance intersystem crossing from the
first excited state to the lowest triplet state of the chromophore.
However, if this were the only effect operative, one would expect
to see a significant increase in lifetime and intensity as the
cationic atomic number decreased from 55 (cesium) to 37 (rubidium)
to 17 (potassium) to 11 (sodium). The invariance of lifetime and
relative emission intensity to changes in atomic number for the last
three leads to one of two conclusions: (1) there is no heavy atom
effect operative for these nuclei, or (2) there is some other effect
operating in the opposite direction to the heavy atom effect, thus
tending to counterbalance it.
The first possibility, that there is no heavy atom effect for
o Co -H CO CO CO Lt c
H 2- 2-t Hl CO cnO
23- o H o ui> if) (N (0
(N 2- 2- 2- H H CO cO
LO O l C( CO CO CO
D co ct o co n n cO co
in un LI) )O in u) uOn o
0 E- E- E- [i E-[ E-4
4- + + + +- 0)
*H C + ) ) O U ?
i.i 0u Z ( 01
H 0 -
these nuclei, seems highly unlikely, since the rubidium cation is
isoelectronic to the bromide anion, which has been shown to quench
the fluorescence of several compounds more effectively than the
chloride ion, which is isoelectronic to the sodium cation.
(Indeed, a careful reading of reference 70 would indicate a general
cation quenching effect.) Thus, it seems likely that the heavy
atom effect is operative for these nuclei, but is opposed by another
quenching mechanism. '.,ile there is no unequivocal evidence in the
present work for any specific mechanism, several possibilities will
be examined in the general discussion.
Fluoradenyl Systems: Experimental Results and Discussion
As with the alkali fluorenyl salts, the alkali metal salts of
fluoradene were affected, at higher concentrations, by inner filter
and reabsorption effects. However, the problem was somewhat more
serious for the fluoradenyl systems, since the molar extinction
coefficients were considerably higher.
This was especially serious for the separated ion pairs. The
Stokes shifts (difference between highest wavelength absorption and
lowest wavelength emission) for both the fluorenyl and fluoradenyl
separated ion were comparable (8 nm for lithium fluorenyl in DME
vs. 10 nm for lithium fluoradenyl in L'1.), but the molar extinction
coefficient for the fluoradenyl system was almost ten times higher
(for lithium fluoradenyl in L!1E:, E(570) = 7800, compared with lithium
fluorenyl in DME, E(520) = 800. Thus, the emission spectrum of the
separated pair was both red-shifted, and the intensity considerably
decreased, just as for the fluorenyl system, and these effects
persisted down to concentrations about ten times lower than they
had in the separated fluorenyl ion pairs, i.e. about 10 M.
The problem was also more serious for the contact ion pairs of
fluoradene than for the contact ion pairs of fluorene. However, the
inner filter and reabsorption effects were less severe than for
the separated fluoradenyl ion pairs, due to two factors. First,
the Stokes shifts of the contact ion pairs of fluoradene are much
larger than those of the separated pair (in fact, they are somewhat
larger than for the fluorenyl ion pairs). This means that there
is less interference by the visible absorption band on the emission
band. Second, the extinction coefficient of the visible band is
somewhat less for the contact ion pairs than for the separated
ion pairs of fluoradene; e.g. for sodium fluoradenyl in THP,
E(540 nm) = 5300, for the crown ether complex, E(570 nm) = 7800.
Thus, for the contact ion pairs, these inner filter and reabsorption
effects persisted down to about 104 M.
Effect of Cation
As noted by Hogen Esch,3 the fluoradenyl anion is sensitive to
the same parameters of cation, solvent, and temperature that the
fluorenyl anion is. However, in the ground state, fluoradenyl ion
pairs tend to be somewhat looser than their fluorenyl counterparts.
For example, a lithium fluoradenyl solution in THF contains virtually
all solvent-separated ion pairs, while a lithium fluorenyl solution
in THF has 25% contact ion pairs. Further, the greater acidity of
the parent hydrocarbon, fluoradene, allows one to study the anion in
a greater range of solvents.4
As can be seen from the data in Table 6, the fluoradenyl salts
in THP display much the same behavior as the fluorenyl salts, with.
two exceptions. First, for all the fluoradenyl salts, the dissociation
of the ion pairs into free ions could be detected directly at low
concentrations in THP.
Secondly, the sodium salt shows this behavior even at relatively
high concentrations (10-4 M), so that one finds a dependence of the
position of the emission maximum upon the excitation wavelength.
If excited at wavelengths corresrporling to the contact ion pairs'
Table 6. Effect of cation on the ion pairing of excited alkali
fluoradenyl salts, at room temperature in THP.
SFluorescence Type Ion Pair
maximum (nm) Groundb Excited
Na+ 585-600,c 600d C/F C
Nat e 580 F/C F
Na(CE) f 581 S S
Kt 594 C C
K CE) f 580 S S
Cst 590 C C
Free 580 F F
a C = contact, S = separated, F = free.
b For reference, see Table.2.
c At concentrations from 8x10-5 M to 5x10-6 M, excited at 359 nm,
or 540 nm.
d Independent of excitation wavelength, in the presence of hundred-
fold excess of sodium tetraphenylborate.
e Excited at 389, under same conditions as c.
f Slight excess of dicyclohexyl-18-crown-6 added.
g Seen in all the above at low concentrations.
absorption maxima (550, 371, 359 nm), the sodium salt has a broad
emission band, 580-600 nm, der'-ndirn on concentration; as the
concentration increases, the peak shifts toward 600 nm. If excited
at 388 nm, where the solvent-separated ion pairs, or free ions, absorb,
sodium fluoradenyl emits at 580 nm. Upon addition of a hundred-fold
excess of sodium tetraphenylborate, the emission maximum shifts to
600 nm and becomes independent of excitation wavelength. This
indicates that the species emitting at 580 nm is not a solvent-
separated ion pair, but corresponds to the free ion.
There are two possible paths for the creation of excited free
ions in this system. In the first, the contact pair, after excitation,
dissociates into free ions:
Na Flad- + hv -- (Na Flad )* -- Na + (Flad-)*
--- Nat + Flad- + h' .
The second is simply excitation of the free ion, i.e.
Na Flad Na+ + Flad-
Flad + hv --- (Flad )*
(Flad ) --- Flad + hV'
Since addition of sodium ion causes not only changes in the emission
spectrum, but also causes corresponding changes in the excitation
spectrum, it must be concluded that no pathway which depends upon
excitation of a single species can explain the behavior, which means
that the first alternative must be discarded. See Figures 11 and 12
(Figure 12 is an absorption spectrum included for comparison).
Using the room temperature dissociation constant of the salt
Effect of common ion on excitation and emission spectrum of
fluoradenyl sodium in THP; [NaFlad] = 5x10-5.
A. X(emission) = 600 nm;
B. X(excitation) = 371 nm
C. A(emission) = 580 nm;
D. X(excitation) = 388 nm;
E,F after addition of 1 equivalent NaBph4 independent of
emission or excitation wavelength, respectively.
550 600 6
Emission wavelength, nm.
300 350 400
Excitation wavelength, nm.
350 450 550
Figure 12. Absorption spectrum of fluoradenyl sodium in THP;
[NaFlad] = lxl0-4M.
0 5 5 5
*Hue1.Asrto pcrm ffurdnlsdu nTP
in the ground state (obtained from preliminary conductance measurements
in this laboratory, in which a value of 48 for the limiting conductance
of fluoradenyl sodium in THP was used) of 1.1xl0 mole/l, it was
thought desirable to try to calculate a dissociation constant from the
excitation spectrum of the salt at a known concentration to compare
with the number obtained from conductance. Using the excitation spec-
trum of the salt at 6.25x10 M, comparing peak heights at 359 nm and
389 nm, subtracting the contributions of one species to the other's
excitation maximum, and taking into account the differences in quantum
yield (see below) a value, Kd = 2x10 M was obtained in quite reason-
able agreement with the value obtained from conductance. This method,
admittedly used here very crudely, gives promise of being quite useful
for salts with very low dissociation constants.
As in the fluorenyl systems, the cation has a large effect on
the intensity and the lifetime of the emission of the fluoradenyl
anion. As the data in Table 7 indicate, again cesium acts to quench
the fluorescence more than does sodium, while the free ion emits most
intensely and has the longest lifetime. Although the lifetime of
the crown ether-separated pair was not obtained, its intensity is
Table 7. Effect of cation on the lifetime and intensity
of fluorescence of the fluoradenyl anion.
Cation Solvent Lifetime (ns) Intensitya
Cs+ THP 4.2 8
Na+ b THP 11.8 25
Free c Acetonitrile 47.8 100
In relative units.
In the presence of a slight excess of sodium tetraphenylborate.
Lithium as counterion.
roughly the same as that of the free ion. Again, as in the fluorenyl
salts, the addition of crown ether has a striking effect, not only
on the position of the emission maximum, but on its intensity.
Effect of Solvent and Hydrogen Bonding
In Table 8 are listed the salts and their emission maxima in
different solvents. As opposed to fluorenyl systems, there is no
evidence for charge transfer-type interactions in any of the systems
A comparison of Tables 4 and 8 shows that, for the aprotic
solvents, the same order of cation coordinating ability is obtained
for the fluoradenyl salts as was found to hold for the fluorenyl salts.
Also, as in the fluorenyl systems, there is virtually no difference
in position of the emission of the separated ion pair and that of the
free ion. More remarkable, in view of the differences in cliri'e
distribution between the ground and excited states, there is virtually
no effect of solvent polarity on the position of the emission maximum
of the separated ion pair, or free ion, from THP dielectricc constant
= 5.61) to acetonitrile dielectricc constant = 37.5).71 (The same
lack of a solvent effect is seen in the absorption spectrum of these
salts.) This indicates either that both the ground and excited
states of the anion are solvated to the same extent, or that neither
is specifically solvated at all. Alth..uph this point will be more
fully examined in the General Discussion, the redistribution of
charge, indicated by HUckel calculations and invoked to help explain
the cation dependence of both the absorption and the fluorescence
Table 8. Effect of solvent upon the ion pairing of excited
alkali fluoradenyl salts, at 1x10-5M.
Emission Type of Ion F ira
Maximum (nm) Groundb ited
Li+ Dioxanec,e 582, 595 C C
r; 581 S S
Acetonitrile 580 F F
THF 580 S S
Na' THPd,e 600 C C
THF 580 50% C : 50% S S
n-PrNH 583 80% C :20% S S
3n-PrOH:7THPf 583 C, S-H C,S
t-BuOH 588 C C-F
n-PrHf 587 F-H F-H
EtOH 586 F-H F-H
MeOH 585 F-H F-H
K+ THP 594 C C
Cs+ THP 590 C C
n-PrNH + r.-PIrJH 582 S S
Na (CE)g THP 581 S S
THF 580 S S
3n-PrOH:7THP 581 S S
t-BuOH 580 S S
n-PrOH 587 F-H F-H
SC = contact, S = separated, F = free, H = hydrogen bonded.
See Table 2.
c Anomalous system, see text section on aggregation.
In the presence of a large excess of sodium tetraphenylborate.
SDependent on excitation wavelength.
Broad peak, centered at position indicated.
g Slight excess of crown ether added.
spectra, is inconsistent with any model invoking specific solvation
of the anion, barring an accidental cancellation of effects.
In the protic solvents examined, there is a small red shift of
the emission maximum of the free ion compared to the free ion in THP.
That there is hydrogen bonding to the excited anion is indicated by
the increase in peak width at half height (1170 cm- for the free
ion in EtOH, 650 cm1 for the free ion in THP), the decrease in
intensity (the free ion in THP emits approximately nine times more
intensely that it does in the protic solvents), and the slight red
shift in the position of the maximum.252737
Also, the results in the THP-n-propanol system suggest that,
as in the ground state,3 the carbanion-alcohol hydrogen bond can be
facilitated by the presence of the cation. In a 1 x 10 M sodium
fluoradenyl solution in 30 per cent n-propanol, 70 per cent THP,
the carbanion emits at 587 nm. Addition of crown ether shifts the
emission maximum to 581 nm. Dilution to about 1 x 10 M causes
the emittion maximum to shift back to 586 nm.
The shift of the emission spectrum relative to the aprotic
solvents can be explained by an argument anilo:u: to that used to
explain the effect of cation. The hydrogen bond formed to the
excited anion is not the same as that to the ground state anion.
Assuming that the solvent has time to rearrange and reach its
equilibrium position to the excited anion within the lifetime of
the excited state, the hydrogen bond formed should stabilize the
excited anion more than the ground state anion which it will
become immediately following emission (the Franck-Condon ground
state anion), i.e. the energy difference between the excited and
ground state free ion in a protic solvent will be less than that
for the free ion in an aprotic solvent. See Figure 13.
AE > AE H
Hydrogen bonding causes a red shift,
in the fluorescence spectrum.
Effect of hydrogen bonding on the fluorescence
of the free fluoradenyl anion.
The Radical Anion of Anthracene: Results and Discussion
To determine how applicable the concept of ion-pairing was to
the excited state of radical systems, the sodium and cesium salts of
the anthracenide radical anion were prepared as previously described in
A typical absorption spectrum of the sodium salt in THP is given
in Figure 14. This agrees well with other published spectra of these
salts. As in the other systems investigated, these salts are
known to exhibit ion-pairing in the ground state. The rationale of
the position of the absorption peaks exactly parallels that of the
Since the MPF-2A allows excitation only at wavelengths below
700 nm, in looking at the excitation spectra of these salts, it was
found that the peaks corresponded to those of anthracene. Since there
was always unreacted hydrocarbon in the solution, its presence is
not surprising. However, that the excitation spectra of the salt
correspond to those of the hydrocarbon is not a trivial result because
it shows a significant avenue of energy transfer in these systems.
A typical fluorescence spectrum contained a single peak near the
end of the instrument's wavelength range for emission. The results
for all the salts studied are compiled in Table 9. The fluorescence
spectrum of the sodium salt in THF at 1 x 10- M has a peak at
773 nm which shifts to 760 nm on dilution to 1 x 10- M while
increasing in intensity. Further dilution leaves this peak position
unchanged. The spectrum at 1 x 104 M could either be due to the
equilibrium between tight and loose pairs (as in fluorenyl sodium
in THF) or be due to ionization, since the dissociation constant
is fairly high (4 x 10-6 M).73 Since the absorption band is extremely
broad, the extinction coefficient, even at 760 nm, is quite high
(E760 = 3000). This would indicate that the peak at 773 nm was due
to either separated ion pairs or free ions, but shifted by reabsorption
effects. This is even more likely since the sodium salt, in THP,
where neither free ions nor separated ion pairs would be expected,
has no observable emission, until very low concentrations (<8 x 10-6 M).
Table 9. Ion-pairing in the ground and excited states of alkali
Cation/solvent Absorption Fluorescence Ion Paia
maximum (nm) maximum (nm)
Na /THP or THF 707 >770 C
Na /THP + glyme-5b 750 760 S
Cs /THF 725 768 C
Free ionC/THF 750 759 F
C = contact, S = separated, F = free ion.
Glyme-5 25 per cent by volume, completing agent for cations; see text.
c Seen in solutions of Na+ or Cs Anth7.
Crown ethers were not used as completing agents because it was
feared that they might react with the radical anions.7 Instead, glyme-
5 (CH 0[CH CH 01 CH ), a straight chain analog of 18-crown-6, was used
to complex the cation. As can be seen from the table, as for the other
systems studied, the separated and free ion have the same fluorescence
maximum. Also, again the free ion emits approximately an order of
magnitude more intensely than does the cesium contact ion pair.
j.,TTi;F'PT TO GENERATE CARBANIONS FROM EXCITED H'i e.:.-,. b:IJ
In the ground state,75 fluorene has an acid dissociation constant
of about 1023. However, based on Forster cycle calculations,676
its first excited state is estimated to be about 29 orders of magnitude
more acidic than in the ground state.
Fluoradene, in the ground state, shows a rather striking dependence
of its pKa on the solvent. In methanol,7 the pKa is 18.2; in dimethyl-
sulfoxide,78 the pKa is 10.5. The Forster cycle method indicates the
excited hydrocarbon to be about 27 orders of a i-iritude more acidic
than in the ground state.
In view of the great acidity of the hydrocarbons in the excited
state, as indicated by Forster cycle calculations, several attempts
were made to generate the excited state carbanion, especially in
protic media. These attempts were unsuccessful, but some of the
factors involved may help elucidate some of the data for the
Solutions of fluorene and fluoradene were prepared under vacuum,
with purified, degassed, solvents. To these solutions were added
known amounts of base via evacuated ampoules. Absorption and
fluorescence spectra were taken as before.
In Figures 15 and 16 are absorption spectra of fluorene and
fluoradene in various media. In Figures 17 and 18 are emission
spectra of fluorene and fluoradene in various media.
As can be seen from the figures, fluorene is surprisingly
200 250 300
spectra of fluorene in various solvents;
B. Methanol; C. Water.
250 300 350
Figure 16. Absorption spectra of fluoradene in various
solvents; A. Hexane; B. Ethanol; C. Water.
250 300 350 400
Figure 17. Fluorescence spectra of fluorene in
various solvents; A. Water; B. Methanol;
350 400 450 500 550
Emission Wavelength, nm.
Emission spectra of fluoradene in various
media. A. Hexane, lx10- ;; B. Hexane,
lx10-5M; C. Methanol, lxl0-5M; D. Hexane,
5x10-2M; E. Ethanol, 4x10-6M (not to scale).
soluble in the protic solvents, roughly 10 M. At the same time,
fluoradene is almost completely insoluble in water, though not in
other protic solvents.
Figure 18 shows that, in the protic solvents, fluoradene emits
from an excimer state exclusively, while in hexane this excimer
emission is seen only at higher concentrations. As is shown in
Appendix 3, this is a good indication of aggregation in the fluoradene-
protic solvents systems.
Addition of base to fluorene in the protic solvents and irradiation
gave no sign of fluorescence from the anion. Since the emission of
the anion could still be seen at 10 M, if present, it must be con-
cluded that the concentration of the excited anion is less than this.
Additions of base to fluoradene solutions were somewhat more
successful in producing anion. The anion did not appear at all in
water, or in a mixture of water and 5% ethanol, when base was added,
but titrations in methanol and ethanol did produce anion, but not
until H values of about 16; yielding a pKa of 18, consistent
with literature values.77
From the above, one must conclude that while, thermodynamically,
the equilibrium between excited hydrocarbon and excited anion lies
far on the side of the excited anion, there are other factors which
make attainment of this equilibrium nearly impossible.
(1) As pointed out by Mason and Smith,79-81 the rate of ionization
of the excited state carbon acids in protic solvents is probably
limited by the amount of reorganization required by the solvent.
The extensive network of hydrogen bonds in a solvent such as water,
around a hydrophobic species, requilrL-Vn a considerable expenditure
of energy in order to reorient itself to accommodate a proton and
(2) In fluoradene, the possibility of excimer formation would
significantly decrease the amount of "free" excited monomer available
to react with base.
(3) Ground state aggregation of the hydrocarbons, especially in
protic solvents, could hamper diffusion of base to the active site
of the carbon acid.
(4) The intensity of the exciting light would determine the
concentration of the excited hydrocarbon, and, hence, of the excited
carbanion. The relatively weak source of a commercial instrument
would not produce too high concentration of excited carbanion.
(5) Lastly, the breaking of a carbon-h-,!-Ir -n bond is involved,
which would probably require a large energy of activation.
It was not unexpected that the attempts to generate carbanions
from their excited 1T'.'i:.c cirbons failed; however, the information
obtained from these experiments points to a previously ignored factor
which might account for the tremendous difference in the pKa of
fluoradene in methanol and DMSO: aggregation of the hydrocarbon in
If one considers only a hydrocarbon dimerization reaction, in
addition to the carbon acid dissociation in alcohol, then:
RH + OR- I a R + ROH
where RH is the hydrocarbon acid, R its conjugate base, and ROH/OR
are the alcohol and alkoxide, respectively. Then, assuming that
most of the hydrocarbon is a2r rated, i.e. [(RH)2] >>[RH] + [R ],
it follows that [(Rh)2] = Co/2, Co the initial amount of hydrocarbon
(1) K = ; K -
a [RH][OR ] [RH]2
(2) K R- 2KD 1/2.
a [OR ] C
Substituting (2) into (1) yields KD = C /(2[RH]2). Since, in methanol,
no monomeric fluoradene could be detected, one must conclude that
[RH] < 1 x 10 M (a conservative estimate for the least amount of
monomer detectable). Thus with C = 1 x 10 M, it follows that
KD > 5 x 10, indicating that virtually all the fluoradene is
aggregated in methanol. So far, no assumptions are made about the
Now, suppose there is no difference in the pKa of the hydrocarbon
monomer in DMSO or methanol, but that the aggregated form is virtually
inert to base. Thus, the pKa of fluoradene in methanol is actually
app app _[R-1 a
an apparent value, pKaapp. From (2), K app [- RaI
[OR ]Co (2K C)1/2
or, pKaapp = pKa + 1/2 log 2 + 1/2 log (KDCo).
If KD > 5 x 1018, C = 1 x 10-5 M, then pKaapp > pKa + 7, or, since
pKa in L'[Ml0 is 10.5, pKaapp in methanol > 17.5.
Thus, apgretation of the hydrocarbon in protic solvents may be
quite a significant factor in the apparent solvent dependence of the
acid dissociation constant.
AGGREGATION EFFECTS ON CARBANION FLUORESCENCE
Earlier, it was proposed that in order to understand aggregated
systems, the barium salts would serve as good models. This is due
to several factors. First, in the fluorenyl systems which are
actually aggregated in the sense of forming n-mers (such as lithium
fluorenyl in toluene or dioxane), the only information available is
the average value of n in solution, obtained from kinetic experi-
ments.3982 In the barium fluorenyl systems, one can focus on the
anion dimer (with respect to the anion). Also, the barium fluorenyl
system has the significant advantage of having an absorption band
reasonably isolated from others, which is not true of the barium
However, there are certain anomalies to the barium fluorenyl
salt which must be borne in mind in applying results from this system
to others. The size of the barium cation is roughly that of the
potassium ion (1.35 A for Ba 1.33 A for K ), but the charge/radius
ratio is nearly that of lithium (1.48 for Ba 1.67 for Li ).
Thus, while certain anomalies of lithium fluorenyl which have been
ascribed solely to its small size884 may not be elucidated by data
for the barium system, the large electrostatic field of the barium
cation, or more particularly of fluorenyl Ba compared to sodium
fluorenyl, as felt by another fluorenyl anion, may cause "collapse"
of the aggregate which would not occur for other systems.
A more significant problem is the temperature dependence of
the absorption spectrum of bisfluorenyl barium in THF (the room
temperature spectrum is shown in Figure 19). There is little
qualitative change in the spectrum from 250 to -700 C; even at the
lower temperature, there are only about 20% separated ion pairs.
As the temperature is decreased still further, there is a dramatic
increase in the peak at 372 nm, due to the shift of equilibrium
(1) to the right as the temperature is lowered. At -1000 C, the
salt is virtually all in the separated form.85
BaFl2 + nTHF F FIBat+ IFl (1)
This is in striking contrast to the behavior of sodium fluorenyl
in THF, which has a similar absorption spectrum at room temperature,
but which shows a regular increase in the 372 nm peak as the temp-
erature is decreased, indicating a regular increase in the amount of
separated pairs present. This contrast calls into question the
nature of the 372 nm peak in the absorption spectrum of bisfluorenyl
barium in THF at room temperature.
Thermodynamic data on the bisfluorenyl strontium salt,42 a
similar system, show AH and AS for (1) to be -12.3 + 2 kcal/mole
and -47 + 7 entropy units, respectively. Assuming that AH for the
barium salt is not too different from that of the strontium salt
(AH for lithium fluorenyl is about the same as AH of sodium
fluorenyl, in THF39), one finds that:
log K300 AH 1 1
( ) (2)
K 4.58 200 300
If AH = -12.3 kcal/mole, K200 4, then (2) implies that K300
1.10-5, i.e. there are only about 0.001% separated ion pairs in the
400 450 500
Figure 19. Absorption spectrum of bisfluorenylbarium in THF.
bisfluorenyl barium in THF solution at room temperature. Thus,
on thermodynamic grounds, one is led to doubt that the absorption
peak at 372 nm, for this system, is due to separated pairs.
The absorption spectrum of the salt in THP at room temperature,
which also has a shoulder at 372 nm, gives further evidence that this
peak is not due to separated ion pairs. Given the much greater
cation solvating ability of THF compared to THP, this makes it very
likely that the 372 nm peak in both solvents is primarily due to
some other effect than equilibrium (1).
Another effect one would hope to be able to explain is the
severe hypochromism of this system. As noted by Smid,43 the linear
extinction coefficient of the 347 nm band is 7300, compared to a
value of 11,000 to 12,000 for contact alkali metal salts in THF.
The fluorescence and excitation spectra of the salt in THF and
THP are very instructive (see Figures 20 22 and Table 10)., The
Table 10. Fluorescence of barium fluorenyl in THF and THP, at lx10-5M.
Solvent w wavelength Type ion pair
THP All absorbing 568 nm contact-"aggregate"
THF 373 nm 528 nm separated
347 nm =530, =570 nm separated; contact-
THF + 20% CEa 347 nm 533 nm mainly separated
373 nm 528 nm separated
a [dicyclohexyl-18-crown-6] = 0.20 [Ba++].
550 600 500
Emission spectra of bisfluorenylbarium in THF, as a
function of exciting wavelength. A,B [Fl-] = lxl0-5M;
C,D [Fl-] = 3xlO-6M; A,C excited at 373 nm; B,D
excited at 347 nm.
Emission at 580 nm.
300 350 40
Emission at 530 nm.
Figure 21. Excitation Spectrum of bisfluorenyl barium in
THF as a function of emitting wavelength; total
fluorenyl concentration = 3xl0-6M.
Excitation wavelength, nm.
500 550 600
Emission wavelength, nm.
Excitation and emission spectrum of bisfluorenyl barium in
THP; total fluorenyl concentration = lxl0-4M.
spectrum in THF has peaks at 528 and 568, if excited at 373 nm,
identical to those of the separated ion pair, or free ion. However,
if excited at 347 nm, the intensity of the peak at 568 nm increases
relative to the lower wavelength peak, indicating that there are two
As has already been shown, it is highly unlikely that there is
any significant amount of separated ion pairs in bisfluorenyl barium
in THF. This implies that the species excited at 373 nm is the free
ion. As a check of this, the emission of fluorenyl sodium in THF
was compared to that of the barium salt, when both were at the same
anion concentration (3 x 10-6 M), and excited at the same wavelength.
Under these conditions, the barium salt should have approximately
3% free ions (Kd = 3 x 10-9 R/mol), while the sodium salt should have
33% free ions (K = 6 x 10 7/mol). The relative intensities at
528 nm are 11:1, in striking agreement with the assumption that the
species excited at 373 nm is the free ion.
The fluorescence of the other emitting species, excited at
347 nm, is better seen in THP, where this other species, the contact-
"aggregate" is the only species present. This is to be expected,
since the dissociation constant of the barium salt in THP should be
significantly lower than in THF. The intensity of the emission from
this contact-"aggregate" is extremely low; in fact, about thirty times
lower than that of the free ion at the same wavelength (568 nm), and
75 times less than the intensity at the free ion maximum, indicating
a great deal of self-quenching by the contact-"aggregate".
The excitation spectrum of the emitting species in THP is rather
interesting, since it contains a peak around 355 nm. This peak has
no counterpart in the absorption spectrum, and its nature is unclear.
It will be discussed in somewhat greater detail in the General
Addition of about 25% crown ether to bisfluorenyl barium in
THF had two effects. First, it increased the intensity of the peak
at 528 nm (excited at 373 nm) by a factor of about ten (at 4 x 10-5 M).
Secondly, there is an increase in the intensity of the emission from
the other species, but it is much more modest, and mostly obscured
by the free ion or separated ion pair spectrum. However, if the
salt is excited at 347 nm, emission occurs at 533 and 568 nm, with a
larger peak at 568 nm than for a pure separated ion pair. If one
subtracts the contribution of the separated ion pair from this
spectrum, one obtains the spectrum of a species emitting around 540 nm,
presumably the contact cation, Fl Ba which would emit about 30%
as intensely as the separated pair. This is not seen in the uncomplexed
case, as will be discussed in the General Discussion.
The low intensity of the salt's emission is probably due to
two effects. First, the barium cation, being isoelectronic to the
cesium ion, should cause reduced intensity, due to a heavy atom
effect. The greater charge-to-radius ratio of barium would be
expected to cause an accentuated effect, however, by forming a
tighter ion pair, increasing the interaction between the cation and
anion, thus causing greater quenching.
A second mechanism of quenching is specifically due to
aggregation, the so-called exciton interaction, which Simpson, and
co-workers, and Kasha, and coworkers, have applied to dyes and
In the following discussion, the basic relations of the theory
of molecular excitons will be set down as they apply specifically
to the dimer case, in the manner of Kasha. It is assumed that
intermolecular overlap between the two species is small, but finite,
so that the monomer units preserve their individuality and the
aggregate wave-functions and energies may be obtained by applying
perturbation theory to the monomer. Denoting the two molecules in
the dimer (in this case, fluorenyl anions) by A and B, the splitting
of the monomer band due to exchange of excitation energy between A
and B, AE, is given by:
2MA'M 6(M A'R)(MB*R)
AE B A B
where MA and MB are the vector transition dipoles (such that
A 12 = i 2 = 2, M the transition moment for the monomer),
and R is a position vector from the center of MA to the center of
MB, i.e. R is the distance between the transition moment vectors of
the two monomer units. This simplifies to E = 21M- G, where G
is a factor depending on the geometry of the aggregate. Further,
the intensity of the transition from the ground state to the exciton
states is proportional to the vector sum of MA and M Thus, while
the exciton splitting will always occur, only one transition need
be seen because the vector summation constitutes a sort of selection
The mechanism of quenching is thus due to a lowering of the
energy difference between the excited singlet and its associated
triplet state, enhancing the rate of intersystem crossing, since
the rate of intersystem crossing is proportional to the reciprocal
of this energy difference. Hence, the enhancement of phosphorescence
usually observed in such systems, and the accompanying quenching of
fluorescence come from the same cause.
If one assumes the geometry of the barium fluorenyl salt to be
that of Smid and Hogen Esch, but allows the two essentially planar
anions to tilt toward one another (x-ray patterns of similar fluorenyl
salts assume this pattern),83'84 then the exciton model predicts two
bands, such that the oscillator strength of the first band, divided
S R / (ignoring the position of the cation)
by the oscillator strength of the second is equal to the square of
the tri~=>nt of the angle between MA (or MB) and the position
vector R, i.e. = tan a, where fL fH are the oscillator strengths
H L H
for the low and high wavelength exciton bands, respectively, and
a is indicated in the figure.
In order to determine the oscillator strengths of the separate
bands, plots of e(v), the decadic molar extinction coefficient (in
-1 -1 -1
lmole cm as a function of v, wave number (in cm ), had to
be made by converting the absorption spectrum of the dimer from
wavelength to wave number. The areas of these plots were measured
by a planimeter, and the ratio fL/fH derived from the ratio of the
two areas. For bisfluorenylbarium in THP this ratio was 18.67,
making angle a = 770; for THF, a =610.
This result is quite in accord with expectations. THF is a
much better cation solvating agent than THP, so that one would
expect more specific peripheral solvation of the barium cation
by THF, which would cause the two anions to open farther, as is
It should be noted that the geometry assumed here for the salt
is mathematically equivalent to one where the two anions are in
parallel planes directly above each other, but twisted about R,
the line joining the two centers. In this model, a would be the
nirile of twist of one ring relative to the other. That geometry,
while formally equivalent, provides no rationale for the different
values of a in THF and THP, and so has been disregarded.
As a check on the accuracy of the theory, the distance R
was determined by transforming the monomer spectrum (assuming
sodium fluorenyl with a slight excess of crown ether to be a very
good approximation of the unperturbed monomer in THP) as above, and
the monomer transition moment evaluated by:
2 3he f -30 f
M = 2 r = 2.126 x 10 -
8lTmc < v>
8 me < v >
where f is the monomer oscillator strength, and < v > is the average
wave number of the absorption band, determined by:
f = 4.319 x 10-9 n fe(v)d
< v > = fvE(v)dv / fe(V)dv
where the term involving n, the refractive index (= 1.4200 for THP),
is a correction for medium effects. Thus, M was finally evaluated
12 = 1.304 x 10 (fe(v)dv)
This value for the monomer, 15.95 x 10-36
e.s.u., was then used to
evaluate R3 according to the basic equation for the exciton splitting
energy, which for the assumed geometry in THP is:
3 2(1 + cos2 a) M 2
AE(1.9863 x 10 )
where AE is measured in cm and is equal to 1940 cm
1 + cos a = 1.051, and the numerical factor converts cm to
ergs. This yields a value of R = 4.43 A in THP, in fair agreement
with expectations, considering the gross nature of the theory.
For the salt in THF, AE = 1720 cm 1 + cos a = 1.235, which
yields a value of R = 4.87 A, quite in line with qualitative
A similar theoretical treatment is applicable to the hypochromism
of these systems,87 which predicts a dependence of the amount of
hypochromism on the geometry of the aggregate. The theoretical
treatment also requires knowledge about the oscillator strengths
of other transitions, which are not known, so that it will not be
considered further here.
Thus, the simple exciton interaction model gives good semi-
quantitative results for this system. Attempts were made to apply
the knowledge gained from the barium fluorenyl system to other systems
thought to be aggregated: sodium fluorenyl in dioxane, and lithium
fluorenyl in dioxane and toluene.
As would be expected for an aggregate, the fluorescence spectrum
of all three of these salts is considerably quenched compared to a
normal contact ion pair, good evidence in itself that all three are
aggregated. However, unlike barium fluorenyl, the absorption spectra
show no exciton splittiri so that it is difficult to make any
quantitative statements about the structure of the aggregates.
But one can use the fluorescence spectra to attempt to make some
qualitative statements about the structure of the aggregates.
See Figures 23-25.
For sodium fluorenyl in dioxane, the principal absorption
maximum is at 354 nm, compared to 356 nm for the same salt in THP,
a slight blue shift. Although an accidental cancelling of the geo-
metrical factor cannot be ruled out, the most likely explanation
is that the transition moment vectors are parallel and stacked.
The fluorescence spectrum of sodium fluorenyl in dioxane (see Figure
36), beyond its low intensity, is quite similar to that of the same
salt in THP, although very slightly red-shifted (maximum of 538 nm
in THP, and 540 nm in dioxane), and contains little more helpful
information. Indeed, the possibility that it is non-aggregated
sodium fluorenyl that is emitting is not inconsistent with the
experimental data, especially since the excitation and absorption
For lithium fluorenyl in dioxane, the fluorescence spectra
are more interesting. As noted above, the intensity is low relative
to a "normal" contact pair, and the system has a shorter lifetime
than would be expected. If one looks at the fluorescent maximum
as a function of concentration one finds that it decreases as
the concentration goes down, from 545 nm at 1 x 10-4 M to 540 nm
at c < 106 M. Even more interesting are the excitation spectra,
which have peaks at 345 nm and a shoulder at 360 nm, then at the
lowest concentration show only a peak at 360 nm.
Excitation wavelength, nm.
Emission wavelength, nm.
Figure 23. Excitation and emission spectrum of fluorenyl sodium in
dioxane, [NaFl] = 8x10 M.
I I I
500 550 600
Excitation wavelength, nm.
Figure 24. E
Emission wavelength, nm.
xcitation and emission of fluorenyl lithium
n dioxane (not to scale). A. 2x10 M;
. 2.3xl0-5M: C. lxl0-6M.
300 350 500 550 600
Excitation wavelength, nm. Emission wavelength, nm.
Excitation and emission spectrum of fluorenyl
lithium, in toluene; [LiFl] = .lxl0-5M.
There are two possibilities. Either one is seeing a change in
the form of the oligomer to, presumably, a lower aggregation state,
or one is seeing dissociation of the ion pair into free ions. To
test for this, the spectra (both excitation and fluorescence) of
fluorenyl cesium in dioxane were examined. This salt is known to
be non-aggregated in dioxane,82 so that it could provide a good test.
If the anomalous peak appeared, it would be due to dissociation of
the ion pair. If it did not appear, this would indicate that the
effect was due to aggregation of the lithium salt in dioxane.
Over the concentration range 5 x 105 to 1 x 10 M, the emission
and excitation spectra remained unchanged, indicating that the
changes noted above, for fluorenyl lithium in dioxane, are probably
due to changes in the state of aggregation, rather than dissociation
into free ions.
Lithium fluorenyl in toluene should form even tighter aggregates
than in dioxane, so that the above transition should be less likely
to occur at a concentration that would allow it to be observed.
As the fluorescence spectra show, this is the case. Throughout
the concentration range, emission occurs at 552 nm, and the excitation
maximum is at 344 nm. This is in qualitative agreement with the
exciton splitting picture, since in dioxane, the distance between
anions would be somewhat larger due to peripheral solvation thus
decreasing the exciton splitting term relative to toluene. This
would send the upper state higher and the lower state lower, causing
absorption at a lower wavelength 343 nm vs. 346 nm) and emission
at a higher wavelength (552 nm vs. 545 nm) in toluene relative to
dioxane. This is depicted in Figure 26.
Dimer in dioxane
Dimer in toluene
For absorption, AE(toluene) > AE(dioxane), thus:
X(toluene) = 343 nm < X(dioxane)
= 346 nm.
For emission, AE(toluene) < AE(dioxane), thus:
X(toluene) = 552 nm > A(dioxane) = 545 nm.
Figure 26. The effect of solvent on the fluorenyl lithium
a. i r-e- ate.
For the fluoradenyl salts which are apgre-.ated, the situation is
much the same, although complicated by the greater number of bands, so
that it is difficult to separate exciton splitting bands.
Barium fluoradenyl in THF has very clearly defined absorption
bands due to a separated pair, as well as some ill-defined bands due to
the anion in the aggregate. In THP, no contribution from the separated
ion pair is apparent in the absorption spectrum.
The fluorescence spectrum of both these systems corresponds to the
separated, or free, anion. While this is not too surprising for the
salt in THF, it is not clear why this is true in THP as well. There is
a considerable hypochromic effect on the absorption bands of the salt,
larger than for the fluorenyl systems, so that fluorescence from the
anion within the "aggregate" may be more effectively quenched than in
the fluorenyl systems. The intensities of all the emission spectra were
Lithium fluoradenyl in dioxane shows anomalies both in its absorp-
tion bands and in its emission spectra. There are ill-defined absorp-
tion bands at 354, 366, and 382 nm, compared to the other fluoradenyl
salts (except barium) which have only two bands in this region. The
probable explanation is exciton splitting of the normal band.
The fluorescence spectrum is very interesting since it is both
excitation wavelength and concentration dependent. If excited at 355 nm
at =8x10 M (a saturated solution), emission occurs at 591 nm; at
1.3x10 M, excitation at the same wavelength causes emission at 582 nm.
Also, at 8.10 M, excitation at 382 nm causes emission at 582 nm, while
at 1.3x10 M, this band, which is present in the absorption spectrum
throughout, has disappeared from the excitation and emission pattern.
GENERAL DISCUSSION AND ii1U.-.RY
As has been seen, the concept of ion-pairing is just as valid in
the excited state as in the ground state. In both states, there is an
equilibrium between contact and separated pairs, which lies, for the
excited state, farther toward the loose pair than in the ground state.
This is a direct result of the different charge distribution in the
excited state; for other systems, in which the charge becomes more
localized at some atom, upon excitation the ion pairs might be tighter
in the excited state.
Cation and Solvent Effects
As in the ground state, it is possible to distinguish spectroscop-
ically between contact and separated ion pairs, or between contact ion
pairs and free ions, but it is not possible to distinguish separated
ion pairs and free ions. Further, for the contact ion pairs the spectral
shifts caused by different cations are smaller in the fluorescence
spectra than in the absorption spectra. For example, fluorenyl sodium
in THP absorbs at 355 nm compared to 373 nm for the separated ion pairs,
a shift equivalent to 1359 cm- (about 3.9 kcal/mole); the excited state
system, (fluorenyl sodium)* in THP, emits at 538 nm, compared to 528 nm
for the separated pair, a shift of only 352 cm- (about 1 kcal/mole).
The linear relationship between 1/r (r the cationic radius) and
v max, the wave number at the maximum, observed by Hocen Esch and Smid
for fluorenyl absorption,39 also holds for emission (see Figures 27 and
28). The plot for the fluorenyl system yields quite a reasonable value
( (10 cm )
S (lo cm )
Plots of emission maximum vs. functions of the
cationic radius, for the fluorenyl salts.
A. 1/r vs. V; B. Warhurst plot, 1/r +2 vs. V.
S (10 cm )
Plots of emission maximum as functions of cationic
radius for the fluoradenyl salts. A. 1/r vs max
B. i/r +2 vs V (Warhurst plot).
for X max of the emission of the free ion, 527 nm; a Warhurst-type plot88
of l/(r + 2)vs. V max, while it gives a reasonable straight line, yields
a much poorer value for X max of the free ion, 517 nm. For the fluor-
adenyl systems, a value of 576 nm for the free ion's wavelength of max-
imum emission was extrapolated from the plot, while the Warhurst-type
plot gave a value of 562 nm. (These plots were constructed assuming that
the cationicc radius" of a separated ion pair was equal to the length of
one molecule of THP and the radius of the sodium cation, 5.75A.)
While there is little reason to expect one scheme to be better at
predictinr the spectral maximum than the other, it should be noted that
the Warhurst model is much worse at describing the behavior of these
The nature of the counterion affects not only the position of the
emission maximum, but also the intensity and lifetime of that emission.
From the data presented, one must conclude that the cation quenches the
fluorescence of the excited anion in at least two ways. First, it can
quench through a "normal" heavy atom effect, which is the predominant
effect in cesium salts, presumably by increasing spin-orbit coupling
from the excited singlet to the triplet state. This effect should de-
crease as the atomic number of the cation decreases.
The cation may also quench the excited anion through a mechanism
involving some perturbation of the rigid, planar anion, which depends
on the size of the cationic field for its effectiveness, increasing as
the cationic radius decreases. Although no firm conclusion about the
nature of this other effect can be reached on the basis of the present
work, some of the factors involved can be mentioned. (For convenience,
the anion discussed will be the fluorenyl anion.)
In general, the rate constant for non-radiative deactivation of the
excited state is proportional to:
i I > Fk)2
where P1 are the ground and excited state wave functions, respect-
ively, Ok is the k'th normal vibration mode of the molecule, and Fk is
a vibrational term involving the Franck-Condon coupling factor.
One would expect the energy of the cation-anion vibration to
increase as the radius of the cation decreased, thus requiring fewer
vibrational quanta to deactivate the excited state. Thus, the effect of
the cation on the purely vibrational part of the above expression, Fk,
would be similar in nature to the effects seen in substituting deuterium
for hydrogen in aromatic hydrocarbons90 91 (Deuterated forms have longer
lifetimes and higher quantum yields.), with this effect greater for
sodium than potassium, etc.
Perhaps more significant would be the effect of the cation on the
electronic factor, <(1 /3Qkl >. As indicated in Appendix 2, charge
density is more dispersed into the benzene rings for the excited state
free anion, while it is concentrated in the cyclopentadienyl ring in
the ground state of the free anion. The cation may reasonably be
expected to polarize the r-electron system and draw charge density
toward itself. No matter what position the cation occupies relative to
the excited anion, this effect should alter the excited state wave
function, and hence the amount of coupling between it and the ground
state wave function via any of the vibrational modes. Particularly
affected should be skeletal vibrational modes of the conjugated system.
Since there apparently is such a vibration coupled to the electronic
transition (the vibration responsible for the second peak in the
fluorescence spectra), and (from Figure 7) there is some variation of
relative peak heights with cation, this could be an important factor in
deactivating the excited state, which increase in importance as the
cationic radius decreases, i.e. sodium should polarize the anion more
Another possible mechanism is one involving electron transfer
from the anion to the metal cation (similar to that observed for the
juen.hin,<; of anthracene fluorescence by inorganic anions ). This would
be expected to increase in importance as the cationic radius decreased,
or as the electron affinity of the cation (=-ionization potential of the
metal) increased. Recent work on the quenching of carbazole (a system
isoelectronic to the fluorenyl anion)37'93 indicates that quenching by
proton donors is less important than quenching by electron acceptors for
carbazole. For the fluorenyl or fluoradenyl salts, the formation of an
ion pair would be a necessary prerequisite for such a mechanism to hold
true. Recently, such a mechanism was invoked to explain non-Stern-
Volmes behavior in the quenching of the short-lived phosphorescence of
ruthenium (II) complexes by anionic coordination complexes.92 Assuming
that the additional quenching was kinetically controlled by the ion pair
association-dissociation equilibrium, the authors were able to derive
reasonable dissociation constants.
As is readily apparent, little has been done to quantify the effects
of ion-pairing on lifetime, quantum yield, or other properties of the
excited state. Such a study, coupled with data on the phosphorescence
of these compounds, could go far to help explain the storage and transfer
of electronic energy in solution, especially since the effect of chemical
parameters on the ion-pairing has been so extensively studied.
Intimately connected to the quenching mechanisms is their virtual
elimination upon the addition of crown ether. This effect is especially
dramatic for the bisfluorenyl barium salt. In THF, in the absence of
crown ether, emission is due to the free anion and the contact-"aggre-
gate," with no evidence for emission from the free Ba Fl species.
Upon addition of crown ether, emission occurs not only from the separated
pair, but also from the species (CE)Ba Fl One is forced to ask why
the emission intensity of these two species is so different.
Suppose the barium ion rapidly resonates through the crown ether
cavity,41 between the two anions. The presence of the crown ether will
decrease the energy of the barium cation-fluorenyl anion vibration (i.e.
the "!*nr1" between the two will not be as tight), decrease its ability
to act as an electron acceptor, decrease its ability to polarize the
anion, and decrease the overlap between the lowest vacant orbitals on
the barium cation and the highest occupied orbitals of the fluorenyl
anion (since the crown ether will be putting charge density onto the
cation). Thus, no matter what the mechanism of cation quenching, the
barium-crown ether complex should be much less effective as a quencher
than the uncomplexed barium ion. The free Ba 2+Fl ion probably has the
barium cation embedded in the anion, thus increasing its effectiveness
as a quencher.
To this point, no effect of solvent has been considered. It has
been seen in previous chapters that the same solvent effects observed in
the ground state of these salts are observed in the excited state. In
order to more firmly establish the explanation given as the proper one
for these systems, attempts were made to correlate the spectral behavior
of these salts with some of the most widely used schemes in the litera-
ture for non-specific solvent effects.
As shown by Hogen Esch and Smid,13 39 there is no correlation
with Kosower's Z-value of solvent polarity,94 or dielectric constant.
Another scheme, due to Lippert, attempts to correlate the Stokes shift
(difference between the 0,0 absorption and fluorescence bands) in wave
numbers of the chromophore with:
2 2 "
2(e g)2 [ ~-1 n2-1
e g 11
hca3 2E+l 2n2+1
where pe', g are the dipole moments of the excited and ground state
species, E is the static dielectric constant, n the index of refraction
of the solvent, and a is the Onsager radius. Assuming the Stokes shift
can be approximated by the difference between the longest wavelength
absorption and shortest wavelength fluorescence maxima, a plot of AX vs.
the quantity in brackets should be linear with a slope proportional to
(j -1 ) As Figure 29 shows, there is no such linearity.
Thus the experimental results obtained can not be explained on
the basis of any non-specific solvent effect, but rather can only be
explained in terms of the specific interactions of ionic species with
solvent molecules. For cations, this manifests itself in the increasing
proportion of separated ion pairs as the solvent is changed from one
that solvates cations poorly to one with greater cation solvating abil-
ity, or as the cationic radius is decreased. For anions, while there is
apparently little interaction with aprotic solvents, with protic sol-
vents hydrogen bonding to the anion is seen. In systems such as mixtures
of n-propanol and THP, one sees solvent-shared species, since complex-
ation of the cation with crown ether disrupts the hydrogen bonding.
Plot of Stokes shift of fluoradenyl sodium as a function
of Lippert's measure of solvent polarity.