Ion pairing and hydrogen bonding in the excited state of alkali carbanion salts


Material Information

Ion pairing and hydrogen bonding in the excited state of alkali carbanion salts
Physical Description:
vi, 112 leaves. : illus. ; 28 cm.
Plodinec, Matthew John, 1946-
Publication Date:


Subjects / Keywords:
Carbanions   ( lcsh )
Hydrogen bonding   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis -- University of Florida.
Bibliography: leaves 107-111.
Statement of Responsibility:
M. John Plodinec.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000582640
oclc - 14152322
notis - ADB1017
sobekcm - AA00004928_00001
System ID:

Full Text







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.



Acknowledgements . .... .. iii

Abstract . . .. .. .. v


I. iiJT DUCTION . . 1


Preparation and Purification of Sample Systems .. 12
Spectral Measurements . . 14


Fluorenyl Systems: Experimental Results and Discussion 22
Fluoradenyl Systems: Experimental Results and Discussion 42
The Radical Anion of Anthracene: Results and Discussion 55




Cation and Solvent Effects . .. 86
Radical Anions . . 95
Aggregation Effects .. . 97
Ionization . . 98

Appendix 1: INNER FILTER EFFECTS . .. 100

F'i itL CALCULATION . .. 103

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



M. John Plodinec

December, 1974

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

results explained.

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 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
9 10
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



(l> It



0 4

H 0

0 *H




U o

4-k 0

0 4)


+-1 0
0 -

O -
* '
U *H


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

+ A


free ions
<- free ions

M+I IA- M + A
vet- te M1 + A

solvent-separated i

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.

M+A- +
MA +


1-- -


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

hydrogen bonding.

(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[]hexacosane), a crown ether, were used to

obtain loose ion pairs, especially under conditions where they would
22 40
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

deexcitation process.

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



70% C
20% C :

95% C :
20% C :

349; 373
349; 373

30% S
80% S

5% S
80% S


372 (shoulder)

Li+ b


K+ b

Rbt b

Cs+ b


Na (CE)c'd THF

348, 371 (shoulder)
346, 371 (shoulder)

Ba+2(CE)c THF

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.


Ba+2 e


349, 373
349, 373

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

Li+ Dioxane




=50% C : 50%
- :'.. C : 20%



529 ,


510, 540
388, 530, 570
525, 555
505, 535

361, 374, 512, 547

367, 378, 518, 553

361, 376, 525-40c

n-PrNH3' n-PrNH2




529, 570
529, 570
524, 564
525, 567

350, 360, 388, 495, 520,
340, 350, 495, 523

a Data from reference 13, supplemented by
C = contact, S = separated, F = free, H
cBroad maximum.
Broad maximum.

= hydrogen bonded.

Na (CE)




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

degassed again.

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.

Spectral Measurements

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

Figure 5.

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





< <



00 0



V) r ,

+, 4- A
0* 0

0 U u0
0 *-

(Q (1) -1
*H 0 +-i 0

0-i HO 0

F 0 0 0
H O 0F

0 0C < < r-

) Q C [D



i Q
c^ )










Absorption spectrum of fluorenyl sodium in THP.

Figure 5.

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

 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.


Fluorenyl Systems: Experimental Results and Discussion

General Considerations

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).
-2 -4L
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.

Figure 6.

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

Figure 7.

0 0

-1 -2 -3 -4 -5 -6 -7
log [salt]


-1 -2 -3 -4 -5 -6 -7
log [salt]

I0 0 1 1 I

log [salt]

o0 00

-2 -3 -4 -5 -6 -7

log [salt]

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).

Figure 8.


400 450

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


Excited State

i Free Anion
-- Contact Ion Pair


-- --- Free Anion

Contact Ion Pair

Ground State
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

o \

** u 0 0 u u u

'- +







( U


o a

0) 0

II O *,-
0 o w
nUE (4 -

0 C0 U
U z'.
*H h

tO X3 U '



+ +
cc, m

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

hc(vC )
pK = pK -
2.303 kT

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

Li+ Dioxane

l, l -



70% C : 30% S
,.,. C : 80% S

95% C :
20% C :

5% S 25%
80% S




Na (CE)



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

> ,

r-C )




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

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 i

0) 0

0 o



*p 0


O- E

,H o


H 0 -
n 0






0 l



C 0



rl bO

*V *H

4J C4

H 42
'O- 0
0 >q

*H H

(tf A

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

General Considerations

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

Figure 11.

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
Cation Solvent
Maximum (nm) Groundb ited
Ground Ex>: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
3 2

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.

Solvent Protic

o H

Hydrogen bonding causes a red shift,
in the fluorescence spectrum.

Figure 13.

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

Chapter II.

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

other systems.

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



0 (



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
metal-anthracenide salts.

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.


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

fluoradene-fluoradenyl system.

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

Figure 15.

A. Hexane;

spectra of fluorene in various solvents;
B. Methanol; C. Water.



SB \



S\ \
0.5 -


250 300 350

Figure 16. Absorption spectra of fluoradene in various
solvents; A. Hexane; B. Ethanol; C. Water.


I i


250 300 350 400

Figure 17. Fluorescence spectra of fluorene in
various solvents; A. Water; B. Methanol;
C. Hexane.

350 400 450 500 550

Emission Wavelength, nm.

Figure 18.

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

an anion.

(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:
2RH~-- (RH)
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

present. Thus,
[R-] [(RH)2]
(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

pKa values.

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.


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

fluoradenyl salt.

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

Wavelength, nm.

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.

Excitation Eii
Solvent w wavelength Type ion pair
wavelength maximum

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

Wavelength, nm.

Figure 20.

Wavelength, nm.

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.

Wavelength, nm.

Emission at 580 nm.

300 350 40

Wavelength, nm.

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.

Figure 22.

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

species present.

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)
3 R5

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
L 2
vector R, i.e. = tan a, where fL fH are the oscillator strengths
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

the case.

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),
86 2
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 )
-1 -i
where AE is measured in cm and is equal to 1940 cm
2 -1
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.
-1 2
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

spectra coincide.

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.

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.

Figure 25.

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

very low.

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.


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
c c
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

(A) 1.2




(B) 0.4-



r +2





( (10 cm )

1.88 1.86

S (lo cm )

Figure 27.

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.
C c


(A) 1.2



0 _


4 -1)
(10 cm-

1.71 1.67

S (10 cm )

Figure 28.

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).
c max:



r +2

Na (CE)



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
k o

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

than cesium.

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
2+ -
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.
e g
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.





Figure 29.

Plot of Stokes shift of fluoradenyl sodium as a function
of Lippert's measure of solvent polarity.









- .DME