J. Phys. Chem.

1983, 87, 5405-5411 5405

Transport and Kinetics in Photoelectrolysis by Semiconductor Particles in Suspension

J. S. Curran’ and D. Lamouche

Laboratolre de Physlcochlmle des Interfaces, Ecole Centrale de Lyon, B.P. 163, 69131 Ecul/y Cedex, France
(Received: February 22, 1983)

Assuming that small semiconducting particles retain the same optical and electronic properties as macroscopic
samples, we obtain an expression for the quantum yield of photoelectrolysis. It is shown that small unmodified
particles are in principle capable of efficient photoelectrolysis if surface recombination and the back-reaction
of surface-adsorbed intermediates is suppressed. The case of metal-loaded particles is discussed in this light.

Introduction sample. The discussion is restricted to “small” particles

Recent interest in the behavior of illuminated semi- in the first instance, since they may be considered to be
conductor particles in suspension in an electrolyte is partly uniformly illuminated. The illumination is diffuse under
due to the possible use of this sample system as a converter most experimental conditions, so uniform illumination can
of solar energy to fuel (e.g., hydrogen). Some progress has be assumed for ro < a-l, where ro is the particle radius and
been made in this domain, as recent results on Ti02,1-4 a the optical absorption coefficient. For submicron par-
SrTi03,4+ and CdS7-9 show. In addition, some general ticles this condition is met unless very short wavelength
features of such systems are beginning to emerge. Un- light is used. The grains are not, however, considered to
treated semiconductor particles are generally ineffective be so small that band theory does not apply. To take an
or totally inactive, whereas when modified by metallic example from the work of Pichat,14 ro N 0.01 pm, the
deposits they are more efficient a t photoelectrolysis.lJwlz illumination condition is met down to X = 300 nm and the
The quantity of metal deposited has an important influ- particles contain N lo5 atoms, quite enough to establish
ence and it has recently been shown that a maximum in a band structure.
the efficiency-metal quantity curve exists.13J4 It has also Between the initial creation of a hole-electron pair and
been shown that small-sized particles are more efficient,13 the final appearance of the products of the photoelectro-
and that the electrolyte composition has an important lysis reaction in solution there exist several obstacles. In
influence on the photoelectrolysis yield?J3 However, there chronological order these are (a) bulk recombination in the
has been little attempt to interpret these results quanti- semiconductor particle, (b) surface recombination a t the
tatively, except by Pichat et al., whose model concentrates electrolyte-semiconductor interface, (c) back-reaction of
on surface catalysis mechanisms.14 This article attempts surface intermediates, and (d) back-reaction of products
to show that semiconductor powder behavior can be in- in solution. For each of these distinct processes a yield
terpreted by using the conventional concepts of photoe- can be defined; thus the overall problem can be decom-
lectrochemistry provided some important differences be- posed into four independent parts. The overall quantum
tween macroscopic “contacted” and microscopic yield of the system qtot is the product
“uncontacted” electrodes are borne in mind. Vtot = ~iqSRqabs?lpr
T h e Model where (a) qi is the fraction of photoexcited carriers which
The behavior of untreated powders is considered first. escape bulk recombination and reach the interface, (b) qSR
The grains are assumed to be microcrystals of spherical is the fraction of carriers arriving a t the interface which
shape which retain all the equilibrium, transport, and escape surface recombination, (c) qabs is the fraction of
optical properties attributed to a macroscopic crystalline carriers escaping the processes in (a) and (b) which react
to form product molecules in solution rather than per-
(1)S.Sat0 and J. M. White, J.Phys. Chem., 85,592 (1981). forming back-reactions with adsorbed intermediates (if
(2)T.Kawai and T. Sakata, Chem. Phys. Lett., 72, 87 (1980).
(3)E. Borgarello, J. Kiwi, E. Pellizetti, M. Visca, and M. Gratzel, any), and (d) qpr is the fraction of product molecules
Nature (London),281, 158 (1981). formed in solution which escape back-reactions a t the
(4)D.H. M. W. Thewissen, M. Euwhorst-Reinten, K. Timmer, A. H. interface and diffuse away into the bulk of the solution.
A. Tinnemans, and A. Mackor, “Proceedings of the International Con-
ference on the Photochemical Conversion and Storage of Solar Energy”, Each of these factors is dealt with in turn below, beginning
Jerusalem, 1982,p 261. with qi.
(5)J. M. Lehn, J. P. Sauvage, R. Ziessel, and L. Hilaire, Isr. J.Chem.,
22, 168 (1982). T r a n s p o r t Equation for Electrons a n d Holes
(6)J. M. Lehn in “Photochemical Conversion and Storage of Solar
Energy”, Academic Press, New York, 1981,p 161. under Illumination
(7)J. F. Reber, K. Meier, and N. Buhler, “Proceedings of the Inter- Four simplifications are made, discussed in parts a-d,
national Conference on the Photochemical Conversion and Storage of
Solar Energy”, Jerusalem, 1982,p 252. which enable us to obtain an analytically soluble differ-
(8)D.Duonghong, J. Ramsden, and M. Gratzel, “Proceedings of the ential equation for hole and electron transport, and which
International Conference on the Photochemical Conversion and Storage lead to a simple expression for the quantum yield.
of Solar Energy”, Jerusalem, 1982,p 251.
(9)T.Kawai, “Proceedings of the International Conference on the ( a ) Negligible Electrical Field within Small Particles.
Photochemical Conversion and Storage of Solar Energy”, Jerusalem, 1982, In general, when a semiconductor is in contact with an
p 284. electrolyte in the dark, its Fermi level must equalize with
(10)A. R. Ellis, J. Solid State Chem., 22, 17 (1977). that of the solution. For macroscopic electrodes this leads
(11)B. Krauetler and A. Bard, J. Am. Chem. SOC.,100,4317 (1978).
(12)Y. Kakato, K. Abe, and H. Tsubomura, Ber. Bunsenges. Phys. to the formation of a space charge layer a t the interface,
Chem., 80,1002 (1976). and a considerable potential drop in the semiconductor.
(13)T. Sakata, T.Kawai, and K. Hashimoto, Chem. Phys. Lett., 88, The situation is quite different for a small semiconductor
50 (1982).
(14)P. Pichat, M. N. Mozzanega, J. Disdier, and J. M. Hermann, particle in suspension. The space charge layer must always
Nouu. J. Chim., 6,559 (1982). form, but in the absence of an external contact this same
0022-365418312087-5405~01.5010 0 1983 American Chemical Society
 Curran and Lamouche

V (NHE)

I
0.2 0.4 0.6 0.8 -8

Flgure 1. A plot which shows the criterion for the flat-bands as-
10
:r
\E
L
0.01 0.02 0.03 0.04
- x
Flgure 2. The energy diagram for a small TiO, particle in suspension
sumption, where the potential drop between the center and the surface in H 2 0 of pH 7.
of a particle radius r o is 2 k B T l q . Any given powder can be placed
on this diagram, and will either be in the “field-free’’ zone under the zone of Figure 1 the total absence of electrons in the
curve, where the electric field is too small to affect carrier transport, conduction band and the small value of potential difference
or above it, where the flat-band assumption must be otherwise justified that is possible inside the semiconductor has the important
or abandoned. consequences that under most practical circumstances the
Fermi level is no longer associated with the conduction
depletion layer exists under the entire surface. More im- band (n-type case) but is placed between the bands in line
portantly if the radius of the particle is smaller than the with the solution redox (Fermi) level. Figure 2 shows such
thickness of the space charge layer formed in a macroscopic a diagram for T i 0 2 in H 2 0 a t pH 7.
electrode of the same material the potential drop in the Exactly the same is true of a p-type particle. The initial
semiconductor becomes limited by the size of the particle. doping type for small particles on the lower left of Figure
Under these conditions all the donors (for the n-type case) 1is immaterial to the form of the energy diagram of Figure
are ionized and there are no electrons left in the conduction 2.
band. Under total depletion the potential drop A V be- ( b ) Lack of Effect of Illumination on the Electrical
tween the center and the surface of a spherical particle can Field. It is assumed that illumination of the particle does
be easily calculated by applying Gauss’ theorem.15 If we not affect the “flat-band’’condition discussed above. Since
assume a dielectric constant t and a donor density N there is no reason for either carrier type to accumulate at
the center, only the Dember potential, normally neglected,
can cause a change in the internal field, even if the total
charge on the particle is changed by illumination, as ob-
Hence served by Bard et
( c ) Pseudo-First-Order Bulk Recombination. The
A V = V(r=O) - V(r=ro) = s, E ( r ) GL
r0 qN
dr =
ro
r dr
Shockley-Read-Hall treatment of bulk recombination via
trapping levels gives a general expression for the recom-
bination current per unit volume RE:
N
AV = -ro2q
6tto
The potential difference A V is the largest that may exist
inside the particle, but its value is small for small ro. In where c, and ch are the electron and hole concentrations,
particular if AV < 2 k B T / q , its influence on carrier trans- ci is the intrinsic carrier concentration, and 7, and are
port may be neglected. This inequality is satisfied for the electron and hole lifetimes. For wide bandgap n-type
semiconductors under low injection conditions the ex-
ro( iv)1/2 ~ (12kBT~o)1/2 pression simplifies to
q2 RE = C h / T h = kCh
< 407 where k is a first-order recombination rate constant, since
Figure 1illustrates this criterion graphically, in such a way c, >> c h and c, >> ci.
that a given particle system may be located either in the This situation also prevails in the case of the small
field-free zone under the curve, or in the field-present zone particle under illumination, but for an entirely different
above it. In many cases in the literature the size of the reason. In the dark as described above the particle is
particle is not reported, but Pichat et al.14 state that ro totally depleted, but under illumination the holes and
varies from 0.01 to 0.015 pm. The donor density for TiOz electrons created react at the semiconductor-electrolyte
ranges from about 1015 for undoped material to 10l8 interface a t different rates in the normal case. Thus, for
for material heat treated in hydrogen.16 Metal-loaded example, ce >> ch when k h >> k,, and first-order recombi-
particles are likely to have the latter doping density, unless nation kinetics prevails. We adopt this plausible as-
no heat treatment has been used. With respect to the sumption for which some experimental evidence in the case
energy diagram for a particle which lies in the field-free of CdS particles has recently been reported18 by Rosetti
and Brus, who state that luminescence from radiative re-
(15) W. J. Albery, Keynote Lecture of the 33rd Meeting of the Inter-
national Society of Electrochemistry, Lyon, France, Sept, 1982. (17) M. D.Ward, J. R. White, and A. J. Bard, J.Am. Chem. Soc., 105,
(16) R. H. Wilson, L. A. Harris, and M. E. Gerstner, J.Electrochem. 27 (1983).
SOC.,126, 844 (1979). (18) R. Rosetti and L. Brus, J.Phys. Chem., 88, 4470 (1982).
Photoelectrolysis by Semiconductor Particles in Suspension The Journal of Physical Chemistry, Vol. 87, No. 26, 1983 5407

combination has a linear and not a quadratic dependence tlli

on light intensity.
( d ) First-Order Interfacial Kinetics. The interfacial
reaction kinetics of both electrons and holes is assumed
to be first order in electron and in hole concentrations.
The respective rate constants kh and k , represent the sum
of surface recombination plus forward and backward
Faradaic charge transfer processes. We show below that
the back-reaction of products in solution is negligible.
However, in principle surface recombination and back-
reactions of surface intermediates may introduce a sec-
ond-order term into the interfacial kinetics, with the
consequence that kh (and k,) increase with illumination
intensity. This would be observable as negative deviation , IO~L,
from linearity in an experiment of the type reported by 1 2 3 4 5 6 1 8 9 1 0 .

Rosetti and Brus for CdSs as we show in Appendix 11, and Flgure 3. The internal quantum yield 7, of a small semiconductor
the same negative deviation in a plot of photoelectrolysis particle of radius r o = 0.03 p m as a function of the diffusion length
L D (cm). The rate-limiting interfacial rate constant k, is given a value
yield vs. illumination intensity, neither of which have been of 600 cm s-l and two values of the hole diffusion constant are used,
reported. 2.6 X lo-* and 0.39 cm2 s-', corresponding approximateb to TiOz and
Linearity up to 1015 photons cm-2 s-l has recently been CdS.
confirmed for hydrogen evolution from the system Pt/
Ti02/H 2 0/ propanol. l9
( e ) Solution of the Transport Equation. Using the
simplifications explained in parts a-d, we can write an
analytically soluble pair of differential equations for the
steady state on for each carrier type, using the normal
formula for spherical diffusion:

where I is the illumination intensity. Equation 2 can be .04 .08 .12

solved directly for ch. The first boundary condition is a t
the center Flgure 4. The internal quantum yield 7, of a semiconductor particle
as a function of it's radius r o for various values of the interfacial rate
dch/dr = 0 a t r = 0 (spherical symmetry) constant k,. The dotted lines show the results of applying an ap-
proximation (discussedin the text) which neglects diffusion. L , is taken
and the second a t the interface where the current density as 1 0 - ~cm.
is attributed a value, J , eventually experimentally acces-
sible, which must be a t steady state be the same for both Figure 4. From a physical point of view, it is worth noticing
electron and hole currents: that applying the simplified eq 4 for vi is equivalent to
Jh = J , = khch a t r = ro neglecting the diffusion (e.g., neglecting the diffusion terms
in the general eq 2) and thus to considering c, and c h as
The mathematical details are given in Appendix I. The constant within the particle. From Figure 4,it is clear that
essential result is that this approximation appears to be very useful as long as kh
3(R coth R - 1 ) < lo3and for small particles say r < 0.1 km for the specific
'li = (3) case considered here, or in general for R < 1.0.
kr0
+
R2 -(R coth R - 1)
kh
( f ) Surface Recombination Factor vSR. By defining a
first-order rate constant for surface recombination kSR,for
the same reasons as discussed in parts c and d and for the
where R = ro/LD,L D being the diffusion length defined by forward Faradaic process kh,f,we can write the fraction of
L D = (D/k)lI2 photogenerated carriers escaping surface recombination
as
Values of vi as a function of LD for fixed values of ro and
kh are shown in Figure 3. In Figure 4 we have calculated kh,f
vi as a function of ro for a fixed LDand various values of VSR =
+ kh,fkSR
kh.
Using a series expansion and retaining only the first- (g) Intermediate Back-Reaction Factor 'labs. It is im-
order terms gives a simplified expression of eq 3 for vi possible to write an expression for this term without
adopting a specific mechanism and making some rather
1 arbitrary assumptions about the rates of various processes.
Vi = L (4)
We make no attempt to estimate this parameter but sim-
1 + -ro
rc

3kh ply point out that it may play a crucial role.

(h)Product Back-Reaction Factor vpr., We are able to
This expression was used to calculate the dotted lines in show that for small particles this factor is close to unity,
which eliminates one complication. The physical reason
(19) P. Pichat et al., unpublished results. is that enhanced diffusion at small particles is able, at the
5408 The Journal of Physical Chemistry, Vol. 87, No. 26, 1983 Curran and Lamouche

1
low current densities involved, to remove products rapidly E INHE) CATHODE ANODE NEITHER
4 A T
so that their concentration a t the interface is very low.
The factor vpr can be written by inspection, taking for
example the back-reaction of H2 0.0
-
vPr - k H
k',,f[H,OI
4
1
b[ 21 + k6,f[H201 + 1.0
where kb and k i , f are second-order interfacial rate con-
stants (since it is necessary to take account of the con-
centration of the respective reactants). [H2] is estimated + 2.0
by equating the ideal current density J (vi, ~ S R vlab
, = 1.0)
to the diffusion flux in solution; it is thus an overestimate.
If we assume that diffusion takes place over a stagnant
layer of thickness S, beyond which [H,] is virtually zero,
and with a diffusion coefficient D for H2
+ 3.0 4 0.02 prn
C

DW21S
J(idea1) = Flgure 5. The energy diagram for metallized TIO, (TiO,/Pt etc.) (a) with
(S - r o h the metal acting as a cathode, (b) with the metal as an anode, and
(c) with the metal acting neither as cathode nor as anode, but as a
It is commonly assumed that S is quite large compared recombination zone. Both (a) and (b) are impossible. V , is the intrinsic
to the ro values we consider in this article. Thus a sim- barrier height between n-TiO, and Pt, estlmated as 0.5 V.,'
plification takes place.
m i c r o s ~ o p $ ~and
~ it is generally assumed that these spots
act as cathodes, the structure behaving like a Schottky
barrier solar cell with efficient hole-electron pair separa-
Iar2 tion by the internal field. However, it is easy to show that
[H21 = 30 this can only be the case under rather restricted thermo-
dynamic conditions. We show that a t least for Ti02/
Inserting appropriate values for Iar, (lo* M cm-2 s-l for P t / H 2 0 the metal spots can neither act as cathodes or as
normal illumination intensities) and ro/D (0.2 cm-l), and anodes.
using [H20] = 0.055 M ~ m - one ~ , finds that vpraturns out Figure 5a shows the energy diagram for this structure,
to be unity unless k , / k , , > los, a condition unlikely to be supposing that the metal spot is a cathode. The quantity
met in practice. Vbi is the barrier height between Ti02 and Pt, calculated
Interim Conclusion. So far we have been able to show to be 0.8 V13 and measured on a single crystal to be 0.5 V.27
that for sufficiently small particles vi and qpr are near to Clearly, the cathode hypothesis is false, since the carriers
unity in theory at least. If such small particles turn out migrate in the wrong direction. Figure 5b shows the same
to be poor at photoelectrolysis it is logical to conclude that diagram supposing the metal acts as an anode. Again it
either surface recombination of intermediate back-reac- is immediately evident that this alternative is false. The
tions are responsible, as long as it can be shown that either only possible conclusion is that no current passes through
the k h or k , reactions are sufficiently downhill to be rea- the metal, that the bands are flat, and that the metal spot
sonably rapid. acts as an inefficient recombination zone. The logical
Theory20and experiments on meta121-23organic crystal24 consequence is that the beneficial effect of the metal must
and semiconductor electrodesz5show that such rate con- be explained by another mechanism, which we propose to
stants reach values of around lo4 cm s-l for one-electron be an increase in k,,f and/or kh,f,and/or a decrease in kSR
transfers for large negative free energy changes (30.5 eV) due to submonolayer coverage of the semiconductor surface
such as could be the case for oxygen evolution steps on by metal atoms. It is known that submonolayer metal
TiOz and similar semiconductors. Oxygen evolution on deposits can lead to dramatic changes of the type pro-
oxides may be a four-electron process26but given the very po~ed.~~,~~
negative values for AGO = -2.0 eV individual steps can also The same arguments applied to the SrTi03/H20case
be expected to be rapid. For the typical case of Ti02/H20 lead to precisely the same conclusion-the bands are flat
we expect from Figure 2 that kh is in the range 102-104 cm for a small particle with a metal deposit.
s-l ( k , is much smaller) and that curves such as in Figures If, as argued above, the bands are flat and the metal spot
3 and 4 can be used. causes recombination, a very simple explanation can be
given for the maximum in the metal quantity-efficiency
Particles Modified by Metallic Deposits curve. On increasing the number of spots, recombination
It is well-known that semiconductor dispersions are must eventually overtake the beneficial electrocatalytic
frequently more effective if they have been submitted to effect, and the yield begins to drop.
treatments which leave small quantities of metals or The thermodynamic condition which must be satisfied
metallike oxides on the surface of the semiconductor if the metal is to act as a cathode for a reduction reaction
particles. Small metal spots can be observed by electron of redox potential E , is easily written:
2kBT
(20) H. Gerischer, Top. Appl. Phys., 31, 115 (1979).
(21) P. Bindra, H. Gerischer, and L. M. Peter, J.Electroanal. Chem.,
ER > Vdn-type) + v b i + -
4
57,453 (1974).
(22) C. Bernstein. A. Heindrichs. and W. Vielstich, J. Electroanal.
c h K , 87,81 (197sj. (27) N. Yamamoto, S. Tonomura, and H. Tsubomura,J.Electrochem.
(23) J. Kuta and E. Yeager, J. Electroanal. Chem., 59, 110 (1975). SOC.,129, 444 (1982).
(24) F. Willig, Adu. Electrochem. Eng., 12, 1 (1981). (28) K. W. Frese, M. J. Madou, and S. R. Morrison, J.Electrochem.
(25) R. Memming and F. Mollers, Ber. Bunsenges. Phys. Chem., 76, SOC., 128 1939 (1981).
.. . (1972).
475 \-- -, (29) A. Heller, "Photoeffectsat SemiconductorElectrolyte Interfaces",
(26) S. Trasatti and G. Lodi, in "Electrodes of Conductive Metal American Chemical Society, Washington, DC, ACS Symp. Ser. No. 146,
Oxides", Part B, Elsevier Scientific, Amsterdam, 1981. 1980, p 57.
Photoelectrolysis by Semiconductor Particles in Suspension The Journal of Physical Chemistty, Vol. 87, No. 26, 7983 5409

and for the anode case:

2kBT
E,, < V,(n-type) + E, - vhi - - 9
where V, is the flat-band potential and E, the bandgap.
The term 2kBT/q arises from the minimum value of the
potential drop across the particle capable of assisting
charge separation. At nonequilibrium the overvoltage on By subtracting (1-2) from (1-1)and rearranging one finds
the metal must be taken into account, and effective charge
separation would generally require more than a 2kT/q
drop. Thus only very positive couples seem likely to satisfy
the cathode condition. The reason why water-alcohol where c(r) = DhCh - Dece. Using
mixtures can be relatively efficiently photoelectrolyzed by
T i 0 2 / P t is less likely to be due to the fulfillment of the u(r) = rc(r) (1-4)
cathode condition or the anode condition than to an in- and substituting in (3), one obtains
crease in the surface back-reaction factor vah, for relatively
subtle kinetic reasons. (1-5)
It should be mentioned that the extension of the above
argument to particles suspended in water totally free of Hence
dissolved ions, or in very pure organic solvents, is not
necessarily straightforward, due to the presence of an ex- u=ar+b (1-6)
tended double layer in the solvent, and to difficulties in where a and b are constants. When r = 0 a may be :Found
defining flat-band and redox potentials. It would seem from the definitions of c and u:
logical to make experiments in the presence of an inert
electrolyte as is the general practice in electrochemistry. a = DhCh - DeC, (1-7)
Conclusions and thus
We have been able to calculate the internal quantum 1
ce = - a)
-((DhCh (1-8)
yield of a small uniformly illuminated semiconductor De
particle in suspension, as a function of it's radius, the This shows that the electron concentration is linearly re-
Faradaic rate constants and the diffusion length in the lated to the hole concentration. We now proceed ito the
semiconductor. solution of eq 1-1using the substitution
The assumption is made that there is a considerable
difference between the Faradaic rate constants for holes w(r) = rch (1-9)
and electrons. Under these conditions bulk recombination Rearrangement gives
becomes first order and the qusntum yield is controlled
by the fastest interfacial rate constant (probably holes for
Ti02) in competition with bulk recombination. We show (1-10)
that for particles with a radius near to the diffusion length,
and for interfacial rate constants in the range expected for which is a linear second-order equation with respect to w.
the rather downhill hole reactions, the internal quantum The general solution can be written as
yield approaches unity. w = up + wh
Since we have also been able to show that the back-re-
action of products is negligible for small enough particles, where up is a particular solution of (1-10) and wh is the
the often low photoelectrolysis efficiency of unmodified general solution of the homogeneous differential equation
particles is either due to surface recombination or to associated with (1-10):
back-reactions of surface intermediates.
It has also been suggested that unless rather restrictive (1-11)
thermodynamic conditions are met, semiconductor par-
ticles with metallic deposits (spots) cannot be considered wh can be found easily
as Schottky diodes. The metal spots act as recombination
zones, and it is suggested that the beneficial electrocatalybic a h = A eXP[-(k/Dh)1/2r] + B e~p[(k/DJ'/~r] (1-12)
effect of metallic submonolayers on the semiconductor is
responsible for increasing the quantum yield. This may where A and B are constants. If we take
be through either suppression of the intermediate back- Iar
reactions or surface recombination. wp = - (1-13)
k
An extension of these arguments to larger particles re-
quires a calculation which accounts for nonuniform illu- A e~p[-(k/Dh)'/~r]+ B e ~ p [ ( k / D h ) ~ / ~ r ]
mination, and especially an examination of the conditions Ch(d = t-
r k
under which a flat-band condition can be met dynamically ch(r) must remain finite when r approaches zero, hence
by the collection of majority carriers in the center of the
particle. We are undertaking such a study. A = -B
Acknowledgment. We thank Professor W. J. Albery for and
his inspiration, and Professor P. Clechet, Dr. Pichat, and 2A'sinh (k/Dh)1/2r la
Dr. J. M. Herrmann for their constructive comments. ChW =
r
+- 12
(1-14)
Appendix I The constant A'is determined from the surface boundary
The differential equations (see text) to be solved are condition a t r = ro:
5410 The Journal of Physical Chemistry, Vol. 87, No. 26, 1983 Curran and Lamouche
ch(r=ro) = ch,O (1-15) (ro/R2)(R coth R - 1)
(kro/R2)(R coth R - 1)
J-0 sinh (r/LD)
(11-4’)
sinh (ro/LD) r
where vi is the internal quantum yield. Equations 1-14 and
1-15 lead to If kh can be considered constant, we obtain for the re-
combination current

since the diffusion length LD is defined by and

LD = (D/k)’I2 Iem= k’RE’ =

It is now easy to calculate the total recombination current

in the particle, RE’:

RE’ = k s“
0
4rr2ch(r)dr (1-18)
From eq 11-4’ and 11-6 we can see that the integral term
is independent of I (excitation) and thus
Using eq 1-17 we obtain (2) = k’ta x

(Icu 11lol(&
kh constant
kD
’r - ch,$) x
sinh (ro/LD) k - i)H + du = constant

( r o cosh 2 - LD sinh
with k ” = kk’, le,being the detected fraction of the total
(11-7)

radiative recombination current.

By defining now Now, if we consider kh as dependent of C0,h and c0,, we
G - RE’ can reasonably assume that dkh/dc0,, > 0, dkh/dCo,h > 0,
=- Ti (1-20) and so dkh/dI > 0. If we use eq 6, differentiation with
G
respect to I leads to
where G is the total hole-electron pair generation rate in
the particle, and using eq 1-16, 1-19, and 1-20, we obtain

By rearranging, and using

R = ro/LD (1-22) The last term on the right-hand side can be simplified in
such a way that we can invert the order of differentiation
we obtain and integration because the differentiation and integration
parameters (Iand r) are independent. Thus
3(R coth R - 1)
Vi =
kr0
R2 + -(R coth R - 1)
kh

Appendix I1
We can write (11-9)
As dkh/dI > 0, F > 0, and SOP04rr2H
dr > 0, hence
(11-1)

(11-2) Equations 11-9 and 11-10reveal the nonlinearity of I,, with

respect to I and the negative deviation of the emission
and using signal in the case where k h is dependent on C?J, with respect
Iaro (R coth R - 1) to the emission signal when kh can be considered as con-
khCO,h = V i T = ICYro (11-3) stant (linearity between I,, and I).
kr0
R2 + -(R coth R - 1)
kh Glossary
r0 radius of particle (average), cm
gives a optical absorption coefficient, cm-l
c, (ch) electron (hole) concentration, ~ r n - ~
De (Dh) diffusion coefficients, cm2 s-l
k , (kh) pseudo-first-orderinterfacial rate constants, cm s-l
k pseudo-first-order bulk recombination rate con-
where F , G, and H represent respectively stant, s-l
 J. Phys. Chem. 1983, 87, 5411-5417 5411

I light intensity, photons cm-2 s-l kB Boltzmann constant

N donor density, cm-3 RE grain recombination current
4 electron charge, C vfb flat-band potential, V
t dielectric constant, dimensionless E, bandgap, eV
AV potential drop between center and radius, V
€0 permittivity of free space, F cm-' Registry No. Ti02, 13463-67-7; SrTiO,, 12060-59-2; Pt,
11 quantum yield, dimensionless 7440-06-4; hydrogen, 1333-74-0.

Electron Spectroscopy for Chemical Analysis Studies on Oxyligated Holmium

Compounds

W. 0. Mllllgan, D. F. Mulllca," H. 0. Perklns, C. K. C. Lok,

Departments of Chemistry and Physics. Baylor Un;vers;ty, Waco, Texas 76798

and V. Young
Department of Chem;stry, Texas A d M Unlverslty, College Station, Texas 77843 (Received: March 14, 1983)

ESCA studies on Hoz03,HoOOH, Ho(OH)~, and amorphous holmium hydrous oxide are reported. Experimental
results imply that small amounts of 4f covalent bonding exist in these ionic compounds and that electron transfer
from the conduction band to the valence band is possible. The key to understanding the catalytic power of
these compounds may be related to the 4f energy level. The origin of "shake down" at 4d and 3d levels is a
two-electron process which involves the creation of an n-d hole followed by the electron jump from the 5d
conduction band to the 4f energy level. The 4d-4f couplings are ligand sensitive. The el,
photoelectron signals
are identical with the nonequivalent oxygens in these compounds. Using the Doniach-SunjiE line shapes folded
with a Gaussian approximating the resolution of the instrument, one can completely separate the 4f band and
the OZpband. The photoionization energies of the 4f band approximate the theoretical values of the major
spin excitation A-(t) at 8.8 eV and the Hund's rule ground state A(J)at 5.4 eV. The peaks of the OZpband
at 4.5 and 7.6 eV indicate different oxygen sites. An electron flow from OZpto 4f orbitals and the quasi-at-
omic-model calculation indicate a covalent effect in these ionic compounds.

Introduction of the present work was to conduct systematic studies on

Investigations related to electronic structure and the energy levels of holmium trihydroxide, oxyhydroxide,
bonding in oxyligated lanthanide compounds have become sesquioxide, and amorphous hydrous oxide by employing
necessary, since their catalytic power has been reported an HP 5950A ESCA spectrometer.
in the literat~re.l-~Attempts have been made to use the Experimental Section
01,binding energy to correlate EPA (electron-pair accep-
tor) and EPD (electron-pair donor) with the catalytic Holmium sesquioxide used in the crystal growth of the
properties of heterogeneous catalyst^.^ The absence of the holmium trihydroxide, oxyhydroxide, and holmium hyd-
5d conduction band is apparent in these insulating com- rous oxide was obtained commercially (Reagent grade,
pounds. However, the 4f and the OZpbands play an im- 99.9-99.99%) and was used without any further purifica-
portant part in the valence level, and the shift of 01, tion. Crystals of holmium trihydroxide were produced by
binding energy has a more subtle meaning as this shift is placing the corresponding oxide into a Teflon bomb con-
governed by the external screening effects and the Ma- taining a 20 N NaOH solution. Hydrothermal aging at 180
delurig potentials. "C for 3-5 days was required and then the Teflon bomb
The method of using Doniach-SunjiE lines folded with was transferred to a step-down oven at 80 " C for several
experimental resolution function proposed by Wertheim4 hours before harvesting at room temperature. The crystals
and the theoretical calculation of the coefficients of frac- of oxyhydroxide were grown in a like manner except for
tional parentage on the open shell by Cox5 have gained the hydrothermal aging temperature, which was 300 "C.
success in explaining the behavior of lanthanide alloys in These crystal structures were verified by comparing ob-
solid-state physics.6-8 In light of this technique, the aim tained X-ray powder diffraction data with literature val-
U~S.~JO The amorphous material was prepared by dis-
(1) S. Bema1 and J. M. Trillo, J. Catal., 66, 184 (1980).
(2) B. H. Davis, S. N. Russell, P. J. Reucroft, and R. B. Shalvoy, J. (7) J. K. Lang, Y. Baer, and P. A. Cox, Phys. Reu. Lett., 42,74 (1979).
Chern. SOC.,Faraday Trans. 1 , 7 6 , 1917 (1980). (8) G. Kaindl, C. Laubschat, B. Reihl, R. A. Pollak, N. Mortensson,
(3) H. Vinek, H. Noller, M. Ebel, and K. Schwarz, J. Chern. Soc., F. Holtzberg, and D. E. Eastman, Phys. Reu. B , 26, 1713 (1982).
Faraday Trans. I , 73,734 (1977). (9) G. W. Beall, W. 0. Milligan, and H. A. Wolcott, J . Inorg. Nuc2.
(4) G. K. Wertheim and S. Hufner, Phys. Rev. Lett., 35, 53 (1975). Chern., 39,65 (1977).
( 5 ) P. A. Cox, Struct. Bonding (Berlin),24, 59 (1975). (10) W. 0. Milligan, G. W. Beall, and D. F. Mullica, "Crystallography
(6) M. Campagna, G. K. Wertheim, and E. Bucher, Struct. Bonding in North America", American Crystallographic Association, New York,
(Berlin),30, 99 (1976). 1983, Section F, Chapter 9, pp 362-9.

0022-3654/83/2087-5411$01.50/00 1983 American Chemical Society