714 R. C.

NELSON

Minority Carrier Trapping and Dye Sensitization1"sb

by R. C. Nelson

Department of Physics, The Ohw State University, Columbus 10, Ohio (Received October 6 . 1964)

The customary criteria for judging the plausibility of the electron-transfer mechanism of
dye sensitization in a given case are the production of charge carriers by the sensitizer when
illuminated and the sign of the contact potential difference between sensitizer and substrate.
These data do not distinguish adequately between sensitizers and nonsensitizers in certain
experimental situations, in which there is substantial evidence for electron-transfer sensitiza-
tion on other grounds. A search for an additional criterion suggests an adverse effect of
hole-trapping'in the dye phase when cationic sensitizers are used. The evidence for hole-
trapping is discussed, as well as its significance for both photochemical and photoconductive
dye-sensitized solid state processes. I t is suggested that the effect in the former case may
be on the energy of a hole-electron pair in the dye phase, whereas, in the latter, it is involved
primarily in the slowing of the relaxation process tending to restore the original conditions
after the observation of a sensitized response. This dual expression of its influence would
account for the similarities observed between photographic sensitization and sensitization
of photoconductivity in cadmium sulfide, in spite of the essential differences between the
two experimental situations.

Introduction driven by light absorbed by the dye.2R The second is

the fact that a sort of "sensitized photoconductivity"
For more than 25 years, there have been two principal
can be observed in glass in contact with a layer of
types of mechanisms contesting the field of dye-sensi-
sensitizing dye, in spite of the fact that there is no in-
tized, solid state processes. During most of this time,
trinsic process to which energy could be transferred by
the resonant energy-transfer type of process has had
resonance.2b
fairly general acceptance while the electron-transfer
However, a difficulty remains; both of these experi-
type was in eclipse. The latter type of mechanism ap-
ments discriminate strongly between good and poor
peals to experiment since its plausibility depends on
sensitizers, but it is not known upon what basis the dis-
measurable effects and relationships, and in recent
crimination rests since the result cannot be predicted
years it has been shown that, for certain sensitizer-
solely upon grounds of being able to produce charge
substrate combinations, the requirements imposed by
carriers on illumination or upon the sign of the contact
these relationships are fulfilled. It is known that, in
potential difference. Clearly, if the electron-transfer
general, sensitizing dyes in the solid form produce hole-
mechanism is valid, some criterion must exist according
electron pairs on illumination and that, in inany cases,
to which the results of these experiments can be antici-
the contact potential difference between sensitizer and
pated. Thus, a search for such a criterion imposes a
substrate is of a sign and magnitude such as to permit
test on the electron-transfer mechanism.
spontaneous transfer of electrons from the dye phase to
It is characteristic of this mechanism that it attempts
the substrate.
There are two experimental results which strongly
suggest the validity of the electron-transfer niechanis" (1) (a) This investigation was supported in part by a Public Health
Service research career program award No. GF-560-64 from the
for the cases with which they deal. One is the fact that Institute of General Medical Sciences and in part by t h e C. F.
a dye-cadmium sulfide interface shows a photovoltaic Kettering Foundation; (b) presented to the International Conference
on Photosensitization in Solids, Chicago, Ill., June 22-24, 1964.
effect, and, when this junction is made part of an elec-
(2) (a) R. C. Nelson, J . Opt. SOC.A m . , 46, 13 (1956); (b) 50, 1029
trical circuit, a continuous current will flow across it, (1960).

The Journal of P h y s i c d Chemistry

MINORITY TRAPPING
CARRIER AND DYESENSITIZATION 715

to explain the peculiarities of sensitized systems, not gested an energy-doubling process which will work only
only by consideration of transport phenomena a t the if the energy of a hole-electron pair in the chlorophyll
interface but also by application of our knowledge of phase is small.6 More traditional types of meGhanisms
the processes of charge carrier separation and diffusion resemble those of Katz, in requiring that the pair carry
in the dye phase. Thus, a possible and hopeful cri- a large fraction of the energy of the photon.6
terion for predicting the efficacy of a sensitizer may be While the significance of the energy of the pair is ob-
found in considering the energy carried by a hole- vious in a system in which chemical work is done, it is
electron pair in the dye solid since there are evidently difficult to see why it should be of importance in one in
cases of sensitization in which this must assume impor- which the only effect expected is an enhancement of
tance. This paper deals with the applicability of such photoconductivity because in this case it should suffice
a criterion to sensitized processes in general and, in that charge carriers of either sign are donated to the
particular, with its relevance to the two special cases substrate under illumination. However, there is a
just mentioned. marked parallelism of behavior of sensitizers for the
two systems, photographic sensitization and sensitiza-
The Energy Requirement of a tion of photoconductivity in cadmium sulfide, and it is
Dye-Sensitized Process of interest to attempt to uncover the reasons for this,
The resonance mechanism foresees no difficulty with in hope of coming upon a unifying idea.
energy; it is an assertion that the energy of the excited
state of the molecule is sufficient to do what needs to be
The Energy of a Hole-Electron Pair
done. The form given by Franck and Teller3 is basi- We define the energy of a hole-electron pair to be the
cally an application of the Franck-Condon principle ; separation on an energy level diagram of the final
they note that the lifetime of the excited state of the levels occupied by hole and electron when a pair is
sensitizer is of the order of lo4lattice vibration periods created. This is the difference in energy between
of the substrate, so that one can hope to carry out the ground and conductive excited states diminished by
creation of a hole-electron pair with the energy avail- any losses due to trapping of either carrier. The
able by transferring it to a favorable lattice configura- energy difference between ground and conductive
tion. This is assumed to be likely to occur a t least states can be determined by a two-step process. First,
once during tht: lifetime of the excited state. we measure, by means of the external photoelectric
Although the electron-transfer type of mechanism effect, the energy of an electron in the ground state.
rests on a knowledge of the energy level structure of We measure the energy of an electron in the conductive
the system, which can often be worked out, it still is not state by the electron-beam-retardation method. If the
a simple matter to say how much energy must be carried difference of these two be compared with the energy of a
across the interface by the hole-electron pair. The photon a t the photoconductive threshold, the two will
primary photographic process is of interest because one be found to be equal, within experimental error, for good
can set a lower limit to the energy requirement from sensitizers. For certain other dyes, more complicated
the free energy of the products. If these are silver relationships are found, but in all cases it is possible to
metal and gaseous bromine, then we must transfer a t construct an over-determined, consistent energy level
least 0.8 e.v. of energy to the silver bromide c r y ~ t a l . ~ diagram in which the energy of the threshold photon
However, the actual energy required will be determined appears as a difference between two levels, the positions
by the pathway over which the system moves in energy- of which can be determined without reference to it.’
configuration space and will, thus, depend on rates and In view of the fact that the threshold energy for
mechanisms. If we suppose that only the transfer of photoconduction is somewhat less than that correspond-
an electron from sensitizer to substrate need take place ing to the lowest allowed electronic transition of the
promptly, that is, within the lifetime of the excited molecule in solution, we must recognize this to be a
state of the dye molecule, and that all other processes remarkable result which we do not yet understand. Its
may proceed slowly enough so that thermal energy may
be used to overcome activation barriers, then the total (3) J. Franck and E. Teller, J . Chem. Phys., 6 , 861 (1938).
energy requirement need not be much greater than that (4) W. F.Berg and J. H. Webb in “The Theory of the Photographic
Process,” C. E . K. Mees, E d . , The Macmillan Co., New York, N . Y.,
stored in the products. 1954,p. 145.
I n photosynt)hesis, the luxuriance of hypotheses is ( 5 ) M. Calvin and G. M. Androes, Science, 138, 867 (1962).
such that, even for the restricted case of electron-trans- (6) E. Katz in “Photosynthesis in Plants,” W. E. Loomis and J.
Franck, E d . , Iowa State College Press, Ames, Iowa, 1949,p. 291.
fer mechanisms, the energy of a hole-electron pair may
(7) R. C. Nelson, J. O p t . Soc. A m . , 51, 1186 (1961);J . Mol. Spectry.,
appear in different roles. Calvin and Androes have sug- 7, 439 (1961).

Volunze 69. Number 8 March 1966

716
importance for the electron-transfer hypothesis of
sensitization is obvious; it also suggests a t least as a
possibility that even single dye molecules on the sur-
face of a substrate possessing a photoconductivity of
its own might be able to participate in an electron-
transfer process.
Thus, in the absence of trapping, the energy of B
hole-electron pair is nearly that of the least energetic
photon which will produce it. We now consider the
problem of trapping, confining ourselves to the dis-
cussion of cationic dyes, since our knowledge of energy
levels in these is rather more extensive than in anionic
or nonionic dyes.
Many cat,ionic dyes have large densities of electron
traps, which lie within a rather narrow range of energy
levels, so that they can be considered to be mono-
energetic for many purposes. Usually, these traps are
about 0.35 to 0.40 e.v. deep so that they represent only
about a 25% diminution of the energy of the pair.
However, they may affect sensitization by electron
transfer by bringing the effective energy of an electron
in the dye phase very near the Fermi level in the sub-
strate. These trapping levels are easily accessible to
study either by kinetic methods or by an external
photoelectric effect arising from electrons occupying
them.8
The question of hole traps is a much more difficult
one. The writer has suggested that since cationic dyes
are usually n-type conductors, the anion may act as a
hole trap.9 In the case of the halide ions, their energies
as hole traps may be estimated from the electron af-
finities of the corresponding halogen atoms. Since
these lie in the neighborhood of 3.5 e.v., to be compared
with energies of the ground states of photoconductive
dyes in the range 4.3 to 5.5 e.v., they would act as
rather deep hole traps, deep enough so that the energy
of the pair would be greatly reduced. Where the hole
is a minority carrier, it is very difficult to find out any-
thing about hole traps by ordinary methods. We shall,
instead, go at the question indirectly by considering a
classical paradox of the field of organic photoconductors
concerned with the so-called “intrinsic semiconduc-
tivity” of these substances.
Many measurements attest to the fact that, if one
plots the log of dark conductivity of a dye against the
reciprocal of the absolute temperature, the data fall on
a straight line, the slope of which is taken to be a meas-
ure of the intrinsic forbidden gap between ground and
conductive excited states, so that u = uo exp(-E/lcT),
where E is the energy of a hole-electron pair. E is
usually -1 e.v. and ordinarily not greatly different
from one-half the energy of a photon at the photocon-
ductive threshold. If we now estimate the density of
The Journal of Physical Chemistry
R. C . NELSON
electrons in the ground state, using the number of
molecules, the number of a-electrons, or a density of
states in k-space and multiply this by the Boltzmann
factor, usually the product is rather a small number,
often less than unity. If we attempt to account for the
current using this density of charge carriers, we are
forced to attribute a very large mobility to the carriers,
often -lo4 to lo6 cm.2/v. sec. On the other hand,
values of the mobility arrived a t in more direct ways are
invariably -1 cm.2/v. sec. or less. Since the number
of carriers corresponding to the large mobility is such
that shot noise effects should be prominent and these
are not seen, it seems more satisfactory to accept the
lower range of values. By the same token, we must
give up the idea that E can be the energy of a hole-
electron pair since, if we take the mobility to be low,
we must have a density of carriers several orders of
magnitude greater than is consistent with the measured
value of E. Because of this state of affairs, we shall
refer to this dark conduction as anomalous.
This paradox seems to be proof against resolution in
any simple way. It cannot be resolved using donor
states or any reasonable density of states in the lower
level. It does not seem to be due to electrode effects,
and there is evidence that it is not ascribable to the
presence of the applied field used to measure the con-
ductivity. The writer has investigated the dark con-
ductivities of a number of triphenylmethane dye
halides having known energy level structures and has
proposed a scheme in which the halide ions appear as
deep hole traps. E now has something of the character
of the height of an activation barrier over which the dis-
tribution of electrons between hole traps and the con-
ductive level is changed as a function of temperature,
and the dark conductivity data can be accounted for
reasonably well in terms of small mobilities.’O It is
not necessary to accept this scheme in order to pursue
the argument; it is sufficient to say that, if we take the
best values for the mobility, for an anomalous dark
conductor, E cannot be the energy of a hole-electron
pair, and this energy must indeed be considerably less
than E.
We now note that there is a strong negative cor-
relation between anomalous dark conductivity and
ability to sensitize photoconductivity in cadmium sul-
fide when the sensitizer is in the form of a “thick” film.
The same correlation applies to the ability to sensitize
photoconductivity in glass and to the photovoltaic
(8) B.-Y. Cho, R..C. Nelson, and L. C. Brown, J . Chem. Phys., 39,
499 (1963).
(9) R. C. Nelson, ibid., 22, 890 (1954).
(10) R. C. Nelson, ibid.,39, 859 (1963).
MIXORITY TRAPPING
CARRIER A N D DYE SENSITIZATION
effect a t a dye-cadmium sulfide interface, the character
of which is markedly different for a dye with anonialous
dark conductivity. We thus find in the anomalous
dark conductivity a characteristic which has diagnostic
value for effectiveness of sensitizers in interesting situa-
tions. There appears to be a connection between it and
the energy of a hole-electron pair. If this should be so,
it might be a link between the performance of sensitizers
in systems i n which chemical work is done and the less
interesting, but more accessible, situations in which only
the enhancement of photoconductivity is found.
Sensitization by Thick Dye Films
A “thick” dye film in the context of sensitization is
one which is much thicker than a monolayer and com-
parable to the films used to study the photoconductivity
of the dye itself. Such systems represent an extreme of
the sensitization effect. The simplest sensitized system
is one in which the sensitizer is so sparsely adsorbed on
the substrate that each molecule can be considered to be
isolated. A great many dyes produce an observable
effect in this regime. The most useful state, for most
purposes, is that in which the dye is adsorbed as a close-
packed monolayer, and, for this regime, the number of
effective sensitizers is notably less than in the first.
I n passing to multilayers, there is a further diminution of
the number of effective sensitizers, and such systemsare
highly specific and discriminatory. Within the range of
sensitizers with whirh the writer has worked, those
which perform well in multilayers are also the most
satisfactory in monolayers, so that the phenomena are
probably fundamentally similar. R!Iultilayers are also
of interest as being closely related to the dye-substrate
junction cells in which the photovoltaic effect is ob-
served.
We now list the most prominent differences in be-
havior between films which show anomalous dark
conduction (HDC) and films which do not (LDC).
(1) For HDC films the mobility inferred from the
intrinsic semiconduction model is high; for LDC films
it is low. (2) LDC films sensitize photoconductivity
in cadmium sulfide; HDC films do not. (3) The same
is true of the “sensitized photoconductivity” of glass.
(4)The photovoltaic effect a t the LDC-cadmium sul-
fide junction saturates a t low illuminance a t a value
numerically equal to the measured contact potential
difference; that for the HDC junction is proportional to
the log of illuminance over a wide range and tends to-
ward the contact potential difference as a limit. ( 5 )
the dark ( T -
LDC films show a fast decay of photoconductivity in
-
sec.); HDC films always have a
slow component in the decay curve (T lo2sec.). (6)
If a cadmium sulfide cell is sensitized with a thick layer
717
of LDC film and both are illuminated, the decay of
photoconductivity in this system is faster than in the
cadmium sulfide cell by itself and is, conversely, slower
if the film is HDC. (7) Although the photoconduc-
tivity of the HDC film persists in the dark, the photo-
voltaic effect disappears very quickly from both types
when illumination is discontinued. (8) For the HDC
film, the photovoltaic effect is very small for a very
thick film and can approach the magnitude of the con-
tact potential difference only for a thin film; for a thick
film the photo-e.m.f. is of the order of a few millivolts.
This effect is not seen with LDC films. (9) HDC films
have electron traps 0.3 to 0.4 e.v. below the conductive
level; these are inconspicuous or absent in LDC films.
Since both dye and cadmium sulfide films may be n-
type conductors, we do not consider n-p junction
effects. A fairly straightforward account of the
phenomena in HDC-cadmium sulfide junction cells can
be given on the basis of the observations listed above.
The Fermi level in the cadmium sulfide film, as meas-
ured by the electron-beam-retardation method, lies in
the neighborhood of -3.5 e.v., measured from the
vacuum level. The conductive state in many dyes,
measured in the same way, is found from -3.0 to -3.2
e.v., so that trapped electrons lie near or a little above
the Fermi level of the substrate. We suppose that,
before the measurement is made, the Fermi level in the
cadmium sulfide has been raised almost to the electron
trap energy of the dye phase, possibly by carriers pres-
ent in the dark or, more likely, excited by a small
amount of illumination incidental to the preparation of
the cell and pumping it down for measurement.
We now illuminate. In a thick dye film, only a small
fraction of the dye molecules is illuminated, whereas
the excited electrons are trapped throughout the film so
that the photovoltage is very small. For thin films and
very bright light, we are able to drive the effective
energy of excited electrons upward so that the photo-
voltage may approach the energy difference between
the conductive state of the dye and the Fermi level in
the cadmium sulfide. Since this effect is due to elec-
trons which are not in thermal equilibrium with traps,
it disappears immediately upon cessation of illuniina-
tion because there is no longer an energy gradient at the
interface. According to this point of view the sensiti-
zation effect is not entirely absent but is very small; it
could escape observation by the techniques used to de-
tect it. Since the conductivity of HDC filiiis is persis-
tent and the decay is especially slow in the sensitizer-
substrate system, only after some time and with great
(11) R. C . Nelson, J . O p t . SOC.
Am., 51, 1182 (1961)
Volume 69. Ahmber d March 1965
718
care, would it be possible to observe an energy gradient
a t the interface.
The absence of electron traps in the LDC films in-
creases the magnitude of the photo-e.m.f. , but the most
important factor in establishing the characteristic
differences is the fast recovery time of the LDC-cad-
mium sulfide system. When the system is illuminated,
there is a short transient peak in the photo-e.m.f. which
apparently is due to the rise in the Fermi level in the
cadmium sulfide due to the electrons donated by the
dye. The system very quickly comes to a steady state,
and the current remains constant. When the photo-
e.ni.f. is measured, by applying a countervoltage just
sufficient to make the current across the junction zero,
the recovery is so fast that the Fermi level in the cad-
mium sulfide is almost independent of the illuminance
under the conditions of measurement. In the case of
pinacyanole, a particularly effective sensitizer in thick
films, the photo-e.m.f. is substantially independent of
the illuminance over a range of 100 to 1. For HDC
films, not only is the photo-e.m.f. strongly dependent on
illuminance, but there are also long-term drifts and
fatigue phenomena which are probably to be associated
with changes in the Fermi level in the cadmium sulfide
due to the process of electron transfer and the slow re-
covery from it.
Presumably, considerations of a similar sort apply to
the sensitization of photoconductivity in glass. Al-
though glass is hardly a semiconductor in any ordinary
sense, there appears to be a limited number of surface
states through which electrons can enter the glass. If
one assumes these to be distributed over a range of
energies and to be accessible only to electrons more
energetic than the highest filled state, they might play
a role which simulates that of the Fermi level in
cadmium sulfide, leading to a small sensitization effect
in the case of HDC films and a large one for LDC ma-
terial. l 2
Hole-Trapping and Sensitization
We have attempted to account for the difference in
behavior between the two kinds of film as sensitizers in
terms of three characteristics, electron-trapping, pres-
ence of carriers in the dark, and relaxation time for
photoconductivity. The principal effect is concerned
with the last two since the effect of electron-trapping is
on the magnitude of the photovoltage. We have al-
ready suggested that the anomalous dark conductivity
is associated with the presence of deep traps for holes.
We now wish to suggest further that the long relaxation
time is also ascribable to hole-trapping.
In the cationic dyes of which we are speaking, there
Seems to be no explicit role for the anion in the processes
The Journal of Physical Chemistry
R. C. NELSON
of charge carrier separation and transport. The
energy levels connected with these processes are de-
termined by the colored cation and are nearly independ-
ent of the nature of the associated anion. The effect
of the anion shows in only two ways, being associated
with the dark conductivity and with the relaxation
time for photoconductivity. For crystal violet , an
HDC type material, one can decrease the dark conduc-
tivity and the relaxation time for photoconductivity,
each by two orders of magnitude, by substituting i o -
dide for chloride as the anion, without greatly changing
the eIectron trap depth, the work function, the electron
affinity, or the photoconductive threshold.
Considerations of this sort, which emphasize the dif-
ference in the roles played by anion and cation, lead to a
point of view which encourages one to speak of the do-
main of anions and that of cations and to visualize
them as interacting only weakly insofar as the phe-
nomena of photoconductivity are concerned. Similarly,
one is led to associate the role of the anion with the
minority carrier since the processes involving transport
of electrons can be associated explicitly with the cation.
Bearing in mind the negative charge of the anion, its
identification as a hole trap is a short step, and the
association of the slow decay of photoconductivity with
the weak interaction between the two domains is
natural.
There is also fairly direct evidence of deep trapping of
holes in some triphenylmethane dyes, where Petruzzella
and Nelson found a rather small p-type photoconduc-
tivity in the presence of oxygen, which quenches the n-
type conduction seen in vacuo. Since the p-type photo-
conductivity was very much smaller than the n-type,
deep trapping can be i11ferred.l~
If we then suppose that the difference in activity as
sensitizers of photoconductivity between the HDC and
LDC films is ascribable to trapping of holes in the do-
main of the anions in the former whereas in the latter
they remain untrapped in the domain of the cations, we
see that these conditions are also vitally involved in the
question of the energy of the hole-electron pair although
this energy does not enter directly into sensitized photo-
conductivity. Thus it seems likely that there will be a
good correlation between ability to convey energy to a
solid state photochemical system by charge-carrier
transfer and the ability to sensitize photoconductivity
of cadmium sulfide. Although the immediately operat-
ing causes differ, the basic cause, hole-trapping, is the
same in both cases.
(12) R. C. Nelson, J . A p p l . P h y s . . 34, 629 (1963).
(13) N. Petruzsellaand R. C. Nelson, J . Chem. P h y s . . 37,3010 (1962).