View Article Online / Journal Homepage / Table of Contents for this issue

Interpretation and use of Mott-Schottky plots at the semiconductor/ electrolyte interface

W. John Albery,* Gerald J. O'Shea and Alec L. Smith Physical Chemistry Laboratory, South Parks Road, Oxford, U K O X 1 3QZ

Downloaded by Pondicherry University on 02 March 2013 Published on 01 January 1996 on http://pubs.rsc.org | doi:10.1039/FT9969204083

The Poisson Boltzmann equation is solved for a semiconductor without making the usual assumption of no majority carriers in the depletion layer. The capacative behaviour of the semiconductor is combined with that of the Helmholtz layer to provide a complete description of the potential distribution in the electrolyte and in the semiconductor for all applied potentials including accumulation, the flat band transition and depletion. Calculations show that good linear Mott-Schottky plots are obtained in the depletion region even when 90% of the potential is across the Helmholtz layer. A procedure is described which uses the curvature of the Mott-Schottky plot in the region of the flat band potential to determine the ratio of the semiconductor capacitance to that of the Helmholtz layer. This allows one to determine experimentally the amount of band bending in any system. The procedure is illustrated with data for tin-doped indium oxide. Reasonable values are obtained for the doping density, the Helmholtz capacitance and the true flat band potential.

In the study of the semiconductor electrolyte interface, MottSchottky plots'V2 are used to find the flat band potential and the doping density by plotting C-2 against the applied voltage ED, where C is the observed capacitance. Straight lines are usually found and it is customary to assume that the capacitance of the interface is dominated by that of the semiconductor. The Helmholtz capacitance on the electrolyte side of the interface is assumed to be so large that it is negligible since the observed capacitance, C, is given by: 1 / c = 1/cH

the Debye length, xDL, is

(3) describes the permittivity and ND is the doping density. We have assumed that P is zero in the bulk of the semiconductor and in depletion rises to some positive value at the electrolyte interface. Multiplication of eqn. (2) by dP/ax and integration gives:
ICE,

XDL = (ICE0

kl/ND e2)'12

+ 1/cSC

(1)

-=

ap

ax

f2[P

+ exp(-P)

(4)

where C, and C,, are the capacitances of the Helmholtz layer and the semiconductor, respectively. De Gryse et aL3 have pointed out that linear plots can be obtained even when C, is comparable to Csc. In their treatment they assume that the applied potential is sufficiently removed from the flat band potential that there are no majority carriers in the depletion layer. In this paper we do not make this assumption but solve the full Poisson Boltzmann equation. This allows us to explore the behaviour of the capacitance for all values of the applied potential including accumulation of majority carriers and the transition through the flat band potential. We show in agreement with De Gryse et aL3 that linear Mott-Schottky plots can be obtained when C, is significant. We quantify the error in the flat band potential obtained from extrapolation of such plots. We show that by analysing the curvature of the full plot we can obtain the true flat band potential, the doping density and the value of the Helmholtz capacitance. We exemplify our new procedure with an analysis of data for tin-doped indium oxide.

where we have used the fact that both P and aP/aX tend to zero at large values of x. For a p-type semiconductor the plus sign is to be taken when P is negative and majority carriers are accumulating while the minus sign is to be taken when P is positive and there is a depletion layer. If q is the charge in the semiconductor of area A we then have C C = 4/Eo S where C, is
= (CO/P,>

c1 - exp(-P)I dx

(5)

(6) The expression to be integrated in eqn. ( 5 ) is the same as that on the right-hand side of eqn. (2) and hence is equal to (aP/d&, which can be found from eqn. (4), giving:
11)1'2I (7) For small values of Po expansion of the exponential shows that C,, M C, and C, is therefore the capacitance of the semiconductor at the flat band potential. The capacitance is determined by a small perturbation to the potential using ac impedance. From the definition in eqn. (5), Csc is an integral capacitance, but ac impedance measures the differential capacitance Csc where
-

c, = e2NDAxDL/kT

csc = I ( c o I p o ) ~ 2 ~ p oexP(-Po) +

The semiconductor
The Poisson Boltzmann equation for a p-type semiconductor is:
-= 1 - exp(-P)
a2p

ax2

Ckc = dq/dE, By considering dq, the following general relation may be deduced between the two capacitances: C$&, = 1 + d ln(C,,)/d ln(P,) From eqn. (7) and (8) we find
cgc = I COC1 - exP(--o)1/{2CP,

where the dimensionless potential, P, measured from the bulk semiconductor, is

P = EJkl

(8)
-

the dimensionless distance from the electrolyte interface, X, is

x = x/xDL

+ exp(-PP,)

11}1/2 I

(9)
4083

J . Chern. SOC., Faraday Trans., 1996,92(20), 4083-4085

View Article Online

When Po is small C$, becomes equal to C, and C,,; the differential term in eqn. (8) is negligible. On the other hand when Po is large the differential term tends to - 1/2 and Csc = ( 1/2)Cs,. Under these conditions the exponential terms in eqn. (9) are negligible and it is satisfactory that eqn. (9) reduces to the Mott-Schottky equation',2 where we have rearranged it in the form that is usually plotted:

The Helmholtz layer

The charge, q, in the semiconductor plus any charged surface groups, q o , must be balanced by an opposite charge in the Helmholtz layer. Hence the potential, P,, , at the Helmholtz plane in the electrolyte is given by:
Downloaded by Pondicherry University on 02 March 2013 Published on 01 January 1996 on http://pubs.rsc.org | doi:10.1039/FT9969204083
D 'C

pDC

=0 '

+ (4+ qO)/cH

Fig. 1 Mott Schottky plots of y2 from eqn. (15) vs. PDC from eqn. (1 1) for the given ratios of CdC,

We have labelled this DC because if the charge in the diffuse layer is negligible then this will describe the dc potential difference between the bulk of the semiconductor and the bulk of the electrolyte. This potential difference will be controlled by integral capacitances. Hence from q = PoCsc and eqn. (7) we find
pDC

potential causes more change across the Helmholtz layer and less in the semiconductor. Hence one may be exploring a small section of a curved Mott-Schottky plot because there is little change in Po compared with P,, .

k (cO/cH)(2[p0

+ exp(-PO)

11}1'2 qO/cH +
(11)

New procedure for analysis of data

We now propose a procedure to analyse such data to find the true flat band potential, and the Helmholtz capacitance. We use the curved part of the Mott-Schottky plot where the potential is close to flat band. Given a set of data a classical Mott-Schottky plot is made and the linear portion extrapolated back to find a pseudo flat band potential, E? . From eqn. (14) this will be given by:
p? = 1 - (1/2)(CO/cH)'

where + refers to depletion and - to accumulation. To describe the dc potential distribution we now definef,,, the fraction of the potential difference that is dropped across the Helmholtz layer:

Neglecting the qo term,f,, can be calculated from eqn. (11) and (12) and is a function only of Po and Co/CH.

(16)

Mott-Schot tk y plots
We assume that the Helmholtz capacitance does not vary with potential so that from eqn. (8) the differential and integral capacitances are the same. Then the capacitance measured by ac impedance, Cobs, given by is
l/cobs

We now construct a theoretical potential scale with respect to this potential: AP

= PDC P? -

(17)

From experimental data A P can be calculated by Ap

= (F/RT)(E,c - E?)

(18)

+ l/cH

(13)

Following De Gryse et ~ l . one can show that in the Mott, ~ Schottky region from eqn. (9), (11) and (13),

Fig. 3 shows a set of curves of ln(y2) plotted against ln(AP), calculated from eqn. (17) for different values of CJCo . In the Mott-Schottky region all the plots have unit slope. However when C,/Co is smaller than one there is a significant curved portion before the linear region is reached. The values of A P

Hence one will obtain linear Mott-Schottky plots with the correct gradient to determine the doping density. However the first term on the right-hand side may cause significant errors in the flat band potential determined by extrapolating the linear plot to cut the potential axis. The linearity of the plot is no guarantee of the correctness of the procedure. To illustrate this point we now define y where
y
= (cH/cobs)/(l

+ cH/cO)

(15)

To investigate the character of Mott-Schottky plots we investigate plots of y 2 us. P, for different values of CJC,. The , definition of y has been chosen so that for each case y = 1 at the flat band potential, Po = 0. For each value of CH/Co values of Po are chosen and then the corresponding values of y , P,, andf,, are calculated. A set of Mott-Schottky plots is shown in Fig. 1 and the corresponding variation offm in Fig. 2. It is clear that good linear plots are indeed found for all the values of CHICo.A contributory factor that leads to good linearity is that when CHis smaller than Co changing the applied
4084

00 . 0

20
D 'C

40

60

Fig. 2 Plots of fDc which describes the fraction of the potential dropped in the Helmholtz layer for the given ratios of CHIC,

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

View Article Online

-0.5' 0
Downloaded by Pondicherry University on 02 March 2013 Published on 01 January 1996 on http://pubs.rsc.org | doi:10.1039/FT9969204083

2 3 In(AP)

Fig. 3 Theoretical working curves of ln(y2) us. ln(AP) where A P measures the potential from the pseudo-flat band potential

NaClO,. We have presented the theory for a p-type semiconductor, since it is easier to have P increasing from zero in the bulk of the semiconductor. Tin-doped indium oxide is an n-type semiconductor. However the same theory will apply except that in depletion P will now be negative. Since we plot In(AP), A P must be positive. Hence for an n-type semiconductor we measure the potential of the electrode with respect to the solution. The fit of the data to the theoretical curve is illustrated in Fig. 4. Good agreement is found. From the analysis of the linear Mott-Schottky plot we found Co = 130 pF cm-2. This gave a high doping density of 2.4 x 1021cmP3. The fit to the working curve gave CHICo= 0.2 which is a very reasonable value for a highly doped semiconductor. The capacitance ratio resulted in a value for C, of 26 pF crnp2. This again is a typical value. Hence the analysis of the curved part of a Mott-Schottky plot can be used to determine the voltage distribution at the semiconductor electrolyte interface4 and to settle the interminable debate about the degree of band bending that occurs by an appeal to experiment. We thank SERC and Unilever for financial support.

at which there is the transition to the linear region are diagnostic of the value of the capacitance ratio. Note that where a curve crosses the line ln(y2)= 0 is the flat band potential. Below the line the semiconductor is in accumulation. A working curve of the experimental data plotted as ln(l/Czbs) vs. ln(AP) calculated from eqn. (18) is constructed. The curve is matched to the theoretical curves by adjusting the fit along the y axis but not along the x axis. When the fit has been found the value of C , can be found from the capacitance ratio and the value of C , found from the gradient in the linear Mott-Schottky region and eqn. (14). Knowing the capacitance ratio one can calculate P? from eqn. (16) and hence the correction required to turn E? into the true flat band potential. In eqn. (11) we included a term in qo which described surface charge. An advantage of the procedure is that if qo remains unchanged as a function of potential, then in the calculation of AP, qo cancels out and therefore the procedure is still valid even if there is a substantial surface charge. We now illustrate the procedure with a typical set of data for a tin-doped indium oxide electrode in 0.10 mol d m P 3
23.0

Appendix
List of symbols A C C H
Cobs

csc csc

C O
E EO
DC

.f"?
ND P
P O
D 'C

P? 22.5

0.5
h

4
X XDL

2 Y
v w

01

A v
t -

c -

area of interface integral capacitance of interface capacitance of Helmholtz layer capacitance of interface as measured by ac impedance integral capacitance of semiconductor differential capacitance of semiconductor integral capacitance of semiconductor at flat band potential potential value of E at the semiconductor/electrolyte interface value of E at the Helmholtz plane pseudo flat band potential fraction of potential that is dropped across the Helmholtz layer doping density dimensionless potential defined in eqn. (2) value of P at the semiconductor/electrolyte interface value of P at the Helmholtz plane dimensionless pseudo flat band potential charge in the semiconductor distance Debye length of the semiconductor dimensionless variable defined in eqn. ( 1 5) = x/xDL ; dimensionless distance in the semiconductor

22.0

0.0

References
1 N. F. Mott, Proc. R. SOC. (London) A, 1939,171,27. 2 W. Schottky, 2. Phys., 1939,113,367. 3 R. De Gryse, W. P. Gomes, F. Cardon and J. Vennik, J . Electrochem. SOC., 1975,122,711. 4 W. J. Albery, R. A. Fredlein, G. J. O'Shea and A. L. Smith, Faraday Discuss. Chem. SOC., 1990,90,223.

21.5 0

0.5
1 2 3 In(A P ) 4

Fig. 4 Fit of data for tin-doped indium oxide to the theoretical curve for CJC, = 0.20. marks the flat band potential.

Paper 6/027 16D; Received 18th April, 1996

J . Chem. SOC.,Faraday Trans., 1996, VoZ. 92

4085