cation effect on the cdse-liquid...

J. Electrochem. Soc., Vol. 140, No. 9, September 1993 �9 The Electrochemical Society, Inc. 2581

29. L. Hollan, J. C. Tranchart, and R. Memming, This Jour- nal, 126, 855 (1979).

30. H. Gerischer, Ber. Bunsenges. Phys. Chem., 69, 578 (1965).

31. W. W. Harvey, This Journal, 114,472 (1967). 32. P. Allongue and S. Blonkowski, Y. Electroanal. Chem.,

316, 57 (1991); 317, 77 (1991). 33. S. Adachi, J. AppI. Phys., 58, R1 (1985). 34. T. L. Tansley, in Semiconductors and Semimetals, Vol.

7: Applications and Devices Part A, R. K. Willardson and A. C. Beer, Editors, p. 293, Academic Press, Inc.,

New York (1971). 35. H. Kroemer, in VLSI Electronics Microstructure Sci-

ence, Vol. 10: Surface and Interface Effects in VLSI, N.G. Einspruch and R. S. Bauer, Editors, p. 121, Aca- demic Press, inc., Orlando, FL (1985).

36. M.O. Watanabe, J. Yoshida, M. Mashita, T. Nakanisi, and A. Hojo, J. Appl. Phys., 57, 5340 (1985).

37. D. E. Aspnes, S. M. Kelso, R. A. Logan and R. Bhat, ibid., 60, 754 (1986).

38. J. van de Ven, J. L. Weyher, J. E. A. M. van den Meer- akker, and J. J. Kelly, This Journal, 133, 799 (1986).

Cation Effect on the CdSe-Liquid Junction Derek Harris, *'~ Jack Winnick* and Paul A. Kohl*

Georgia Institute of Technology, Atlanta, Georgia 30332-0100

ABSTRACT

The effect of cation on the CdSe/polysulfide photoelectrochemical (PEC) solar cell has been investigated. The current- potential response of the polysulfide electrolyte on a plat inum electrode was measured as a function of the alkali metal cation in solution. There was a -33 mV shift in the redox potential and an increase in the current when the cation was changed from Na § to Cs § The performance of the CdSe/polysulfide PEC cell with both polycrystalline and single-crystal semiconductor electrodes increased in the open-circuit potential, the short-circuit current, the fill factor, and the energy conversion efficiency when Cs polysulfide was used as compared with K or Na polysulfide. Impedance measurements were made on the electrodes in polysulfide solutions and hydroxide solutions with the different cations. The results for both electrolytes showed a negative shift in the flatband potential, an increase in the apparent charge-carrier concentration, and an increase in the frequency dispersion of the measurements with the addition of Cs +. The impedance measurements also showed a dependence on the orientation of the CdSe crystal. The effects above are related to a combination of a change in the solution and the surface of the semiconductor. The stoichiometric distribution of the polysulfide species changes with the addition of Cs + and the surface of the semiconductor changes, facilitating charge transfer across the semiconductor/ electrolyte interface.

Initial work on the cadmium chalcogenide/polysulfide PEC solar cell was published by Gerischer ~ and Ellis et aI.2 A great deal of work has been performed since those origi- nal studies on improvements to the semiconductor, the surface condition of the semiconductor electrode, and the electrolyte. CdS, CdSe, CdTe and mixed CdSe~T%x 3-~ have been investigated to increase the energy-conversion efficiency and stability of the PEC cells. Several researchers have investigated cadmium chalcogenide PEC solar cells using polycrystalline electrodes. Long-term stabilities have been demonstrated with conversion efficiencies up to 5% for polycrystalline CdSe. 7-1~ The semiconductor/liquid junction solar cell is particularly well suited to low cost polycrystalline electrodes.

Changes in the make-up of the polysulfide electrolyte have a large effect on the performance of the solar cell; the effect of hydroxide ion concentration and the sulfide- to-sulfur ratio have been investigated by some researchers.~l~ Both the hydroxide ion and the sulfur-to-sulfide ratio affect the equilibrium concentration of the electroactive species, the stability of the solution, and the light absorption by the solution. Changing the cation in the solution leads to changes in the performance of the cell. ~6-~9 The cation is not involved directly in the redox reactions at the semiconductor electrode but the performance of the cell increases significantly when the cation in solution is changed from Na § or K § to Cs § The improvement is in the stability of the semiconductor, the open-circuit photovoltage (Voc) and the photocurrent response. The flatband potential (V~) also shifts negative with the addition of Cs*; yet the only difference between the cations is the ionic ra- dius and thus the charge density. The change in cation can affect the surface of the semiconductor by altering the de- gree of adsorption, or by incorporation into the semicon-

* Electrochemical Society Active Member. a Present address: IBM Corporation, Hopewell Junction, New

York 12533-6531.

ductor itself; or by changing the activity of the species in solution, or the distribution of the polysulfide species.

The CdSe/polysulfide system is examined here. First, the current-potential response of polysulfide on a plat inum electrode was measured. Impedance measurements then were made to determine how the change in cation affects the band energies and charge-carrier concentration of the semiconductor. The measurements were made in a variety of systems to separate the different effects: CdSe single- crystal electrodes in hydroxide electrolytes with the different cations, CdSe single-crystal electrodes in polysulfide with the different cations, and with CdSe polycrystalline electrodes in polysulfide. The impedance measurements indicate changes in the performance of both single-crystal and polycrystalline CdSe/polysulfide PEC cells with the different cations.

Polysulfide Electrolyte Polysulfide solutions have been studied extensively in

the pulp and paper industry. The equilibrium relationships of the solution species are presented in Table I. ~~ Lessner et aI. 21 demonstrated that active species participating in the reduction at the cathode is supersulfide, [$2] �9 The concentration of [$2]- is maintained in the solution by the dissociation of [$4] 2 by reaction 7. In those studies on platinum, the rate of reduction was limited by both the charge-transfer reaction and the homogeneous chemical reaction maintaining the supersulfide concentration. These findings were confirmed in studies of the influence of aprotic solvents on the current-potential response. Giggen- bach and others have shown that with the addition of dimethyl formamide (DMF), the equilibrium of reaction 7 was shifted in favor of [$2]-. 22~23 Lessner showed that the addition of DMF in the polysulfide solutions increased the current density, again indicating that it was tied directly to the concentration of [$2]- at the cathode. 24

The electroactive species and rate-limiting step at the photoanode have been the subject of several papers. 14' 24-2~ It

2582 J. Electrochem. Soc., Vol. 140, No. 9, September 1993 �9 The Electrochemical Society, Inc.

Table I. Equilibria in aqueous polysulfide solutions. A log K= ~ + B T + C + Dq-T

Range Ki Reaction A B C D (~ Eq.

K~ H2S ++ H § + HS- -1158 -3.10 15-35 1 K,2 HS- +~ H + + S 2- -1426 -12.0 15-270 2 K,/2 S~- + HS- + OH- +-~ 2S~- + H20 -4.2 0.35 20-140 3 K m 2S~- + HS- + OH <--+ 3S~- + H20 -1.75 0.35 20-140 4 K3/4 3S5- + HS- + OH- +-~ 4S~- + H20 5.6 0.35 20-140 5 Kt/3 4S 2- + 8 OH- + H20 <--+ 2S20~- + 10 HS- -48,168 -0.2855 261.2 0.90 150-210 6 Kd S~- <---> 2S~ -4533 6.42 100-250 7

was thought that the rate of sulfur dissolution was the rate- limiting step. 27 Winnick e t aI 1 in a survey of measurements of the peak current decay with varying sulfur-to-sulfide ratios in the electrolyte, showed that the performance de- pends on the [$2] 2- concentration. 14 Ardoin, in another study on the temperature dependence of the short-circuit current, demonstrated that the activation energy was approximately 71 kJ/mol while that for sulfur dissolution was 10.5 kJ/mol. 25 It was shown that the higher activation energy was tied to the equilibrium reaction between [$4] 2- and [S~] 2-, a combination of reactions 3 and 4 in Table I

2S~- 4 2 2- 4 4 + ~ H 2 0 - - + ~ $ 4 + ~ H S - + ~ O H - [8]

At the photoanode, disulfide is oxidized to supersulfide, with the disulfide concentration maintained by the homogeneous reaction of [$4] 2- to [S~] 2-. At the cathode, the reduction of [$2]- to [$2] 2- takes place, with the [$2]- maintained by the homogeneous dissociation of [$4] 2- to [$2]-.

The polycrystalline surface is a nonideal semiconductor- l iquid interface because of the grain boundaries, disloca- tions, and inclusions present in the material. To understand better the performance improvements with Cs +, the semiconductor/electrolyte interface was simplified. Experi- ments were conducted on n-CdSe single crystals with different crystal faces exposed to the electrolyte. To separate the different effects, impedance experiments were performed in both polysulfide and hydroxide electrolytes with the different cations.

Experimental The solutions were 3M in the hydroxide, 1M in H2S, and

1M in sulfur. They were prepared by bubbling H2S through the appropriate hydroxide solutions. NaOH, KOH, CsOH, and tetraethylammonium hydroxide were used. The mass of H2S added to the hydroxide solution and the pH change was monitored. Elemental sulfur then was added to the sulfide solution to make the polysulfide. The approximate distribution of the polysulfide chains in the solution was determined using the equilibrium relations listed in Eq. 1-7.

A standard, three-electrode cell was used for all experiments; all were carried out at ambient temperature. The electrodes comprised a platinum wire reference, a 2 cm 2 platinum foil counter, and the working, either Pt or CdSe, under an argon atmosphere. The solution was stirred at all times, but no effect of the stirring on the currents or potentials was seen.

Initial experiments were made using a Pt wire as the working electrode, sealed in a glass rod, and then polished flat on the end; the surface area of the resulting tip was 0.018 cm 2.

The general method for making the polycrystalline CdSe films used here was developed by Hodes e t aI.6 CdSe pow- der was combined in deionized water with ZnC12 and Tri- ton-X 100 to make a slurry. This slurry was painted on oxidized Ti substrates and annealed in air at 600~ for 15 min.

After cooling, the electrodes were photoetched in 0.1M HC1 by short-circuiting the electrode to a stainless steel wire mesh and illuminating for 5 s. This photoetching

helped in removing some of the surface defects of the film and reducing the number of surface states. The t i tanium rod then was inserted into an acrylic insulating jacket and cemented in place with the CdSe surface flush with the jacket face.

The CdSe was illuminated through a quartz window with a 650 W tungsten-halogen lamp. The distance from the electrode to the light was calibrated with a pyrometer to give 100 mW/cm 2 of incident energy. The current-potential measurements were made with a PAR 362 (Princeton Ap- plied Research, Princeton., N J) potentiostat.

Impedance measurements were made by superimposing a 5 mV sinusoidal signal on the dc bias. The current response from the cell was separated into the in-phase and out-of-phase elements with an Ithaco 391A lock-in amplifier. The phase shift and magnitude of the response were used to determine the impedance of the cell.

CdSe crystals were obtained from Cleveland Crystals (Cleveland, OH); they were oriented on the (0001) and (1120) planes. The crystals were mounted so that the Cd face (0001), the Se face (0001), or the mixed face (1120) would be exposed to the solution.

Electrical contact was made on the back of the crystal with a gall ium-indium alloy. The electrodes were etched in a concentrated solution of HC1/HNO3 to remove defects from the surface. The crystals then were dipped in a hot 10% KCN solution to dissolve any elemental selenium formed on the surface of the semiconductor in the previous step. The CdSe and CdS crystals were then photoetched in 6M HC1 by short-circuiting the electrodes with a platinum foil and illuminating with i00 mW/cm 2.

The current-potential measurements were made with a PAR Model 273 potentiostat under the control of an IBM PS/2 70 computer. A three-electrode configuration was used: the counterelectrode was a 2 cm 2 platinum foil, and the reference was a saturated calomel electrode (SCE) or a platinum wire. The cell was illuminated with the same 650 W tungsten-halogen lamp used in the polycrystalline experiments. The impedance measurements were made by using a PAR 5301 lock-in amplifier in conjunction with the 273 potentiostat.

Results The current-potential behavior of solutions containing

Na § K +, and Cs § were measured at a platinum electrode (Fig. 1). The oxidation and reduction currents were higher for the polysulfide solution containing the Cs +. The oxidation current, from 50 to i00 mV overpotential, for Cs + was double that of K + which was double that of Na § The reduction current with Cs + was larger than that of K + and Na +, which were approximately equal.

The redox potential, Eredox, was measured on platinum with respect to a SCE reference. The Er~dox in Cs § polysulfide was the most negative, at -0.786 V vs. SCE; in K § polysulfide it was -0.771, and in Na § polysulfide it was -0.753 V vs. SCE. This result agrees closely with the measurements made by Licht: 18 -0.775, -0.768, and -0.748 V vs. SCE for Cs +, K § and Na +, respectively.

Solar cell performance measurements were made in three different polysulfide solutions with Na *, K +, and Cs § as the cation. The results of these measurements are shown in Fig. 2 with the important differences enumerated in

2583

f J

E

m Cz

- - - - K

3 Na

2

1

0

-150

J. Electrochem. Soc., Vol. 140, No. 9, September 1993 �9 The Electrochemical Society, Inc.

-100 -50 0 50 100

O v e r p o t e n t l a l , m V

Fig. 1. Polysuifide current response on platinum with alkai metal- polysulfide solutions: Cs* (solidi; K § (dashed); Na § (dotted).

Table II where Voc is the open-circuit potential, I= is the short-circuit current, FF is the fill factor, and ~ is the energy-conversion efficiency. There is an increase in both the open-circuit potential and the short-circuit current with Cs § as the cation in the polysulfide electrolyte. The trend in I= is Cs ~ >> K § >> Na *. The Vo~ for Na + and K § however, within the limits of error are equal and are about 30 to 40 mV less than that with Cs § ~ for the three different polysulfide electrolytes shows the same trend with Cs § >> K § >> Na § The fill factors, however, for the three different electrolytes are approximately equal.

The performance was measured for the cell with single- crystal CdSe in either K § polysulfide or Cs § po]ysulfide for the t_hree different orientations of the CdSe crystal, (0001), (0001), and (1130). The performance curve for the (0001) face is shown in Fig. 3, and the characteristics for each system are given in Table III. The addition of Cs § gave the highest performance for each of the different crystal faces. The improvement was achieved with an increase in the fill factor for each of the different faces. Only for the Se face, (0001), was there an improvement in Voc. I~r was approximately equal for the different polysulfides with each of the different crystal faces. Licht 18 reported similar results at CdSe crystals where the short-circuit current changed only raarginatly, from 20.8 mA/cm 2 for K ~ polysulfide to 21.3 mA/cm 2 for Cs * polysulfide. In that study, the fill factor showed an increase to 0.647 with Cs* from 0.624 for K § This is different from the polycrystalline experiments where the fill factor remained approximately the same; however, Voo and I~ increased when Cs + was the cation in the polysulfide.

The (0001) Se face with Cs § was the most effective orientation for the solar cell; with the efficiency close to 12 %. Voo for the (0001) face was slightly greater than with the (0001)

Table II. Polycrystalline CdSe performance.

(mV) (rnA/crn 2) FF ~1

Na 0.632 11.9 0.365 2.74 K 0.639 13.9 0.356 3.16 Cs 0.672 16.9 0.356 4.04

face, but the short-circuit current was smaller. The (1120) face was the least effective face. Overall,_the largest improvements with Cs + were with the Se (0001) face. The efficiency increased from 8% with K + to almost 12% with Cs +. With the (0001) face, the improvement was from 9.6 to 10.6%, and with the (1130) face it was from 5.1 to 6.6%.

Impedance measurements were made with the polycrystalline electrodes in K+-polysulfide and Cs§ at frequencies ranging from 500 to 1000 Hz. The capacitance was used to determine the flatband potential from the Mott-Schottky (M-S) equation

C= is the semiconductor capacitance, N= is the doping density, A is the area, e is the dielectric constant, and the other symbols have their usual meaning. Plots of 1/Csc2 v s . Vwere linear with some dispersion of lines between the frequencies, but the separate lines converged to the same x- in ter - cept. Figure 4 shows a comparison of the M-S plots for CdSe in K+-polysulfide and the Cs+-polysulfide as the electrolyte. The same CdSe electrode was used in this measurement to compare the slopes. The Cs+-polysulfide gave a smaller slope, 5.1 x 101~ cm4/(F2V) compared to 9.6 x 10 TM cm4/(F2V) for the K+-polysulfide. This difference corre- sponds to a greater apparent concentration of charge carriers when Cs § was the cation. The absolute value of the charge-carrier concentration was not determined because of the surface roughness of the polycrystalline electrode.

When the term k T / q is negligible, the x- intercept of the M-S plots gives the flatband potential, V~. The M-S plots were reproducible for the different CdSe electrodes. Change in electrode area affects the slope of the M-S plots but not the intercept. The VFB in Cs § polysulfide was -1.533 _+ 0.01 V vs. SCE and in the K+-polysulfide is - 1.441 • 0.02 V v s . SCE. The negative shift in V~ increases the maximum photovoltage attainable by the cell.

To understand the nature of the semiconductor-electrolyte interface in the absence of faradaic charge transfer, impedance measurements were made with the single-crystal semiconductors in four hydroxide solutions, NaOH, KOH, CsOH, and (C2Hs)4NOH (Fig. 5) for the CdSe (0001) face in 1M NaOH. A linear regression of the data is shown. The resu]ts for the CdSe (0001) face in KOH were similar, but in CsOH an increase in slope with increase in frequency was observed. The flatband potential for the (0001) CdSe

15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

e a 10

5

0 ~ ~ ' ' f ' N ~ - ~ lO0 200 300 400 500 600 70(}

Photovol tage , ( - m V )

Fig. 2. Polycrystailine CdSe performance curves with Na, K, and Cs pelysulfides. Illuminated at simulated AM1 spectrum with 100 mW/cm 2.

40

35 .............. -'~'"~ ~ ........ " ......................................................

,~ 3O E >T, 25 ............................. ... u) X a

~ 15 t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \ \ \ ......

0

. . . . . . N

O " t i i T

0 1 O0 200 300 400 500

B i a s v . S C E , ( - m V )

\ \

~, t

600 700

Fig. 3. CdSe (0001) performance with K and Cs polysulflde. Illumi- nated at simulated AM1 spectrum with 100 mW/cm 7.


Table III. Single-crystal CdSe performance.

( m v ) (mA/cm 2) FF "q

(0001) K 580 37.0 0.37 8.25 Cs 670 37.5 0.51 11.86

(0001) K 680 34.0 0.43 9.6 Cs 660 33.5 0.51 10.6

(1120) K 610 28.0 0.30 5.1 Cs 610 28.5 0.38 6.57

face in CsOH, though not a function of frequency, shifted approximately 120 mV negative compared to NaOH and KOH as shown in Fig. 6. The flatband potential measurements for the (0001) and the (0001) faces in the different hydroxide solutions are summarized in Table IV.

The change in V~ in the CsOH is coupled with a change in the M-S slope, or charge-carrier concentration, similar to that found with the polycrystalline electrodes in polysulfide electrolytes. At 1.6 kHz the difference in the slopes for the M-S plots between CsOH and NaOH or KOH was sig- nificant; the slope for CsOH was 1.5 x 1015 cm~/F2V and for NaOH and KOH was 2.1 • 1015 cm4/F2V. The difference in the M-S slope decreased as the test frequency increased; at 63 kHz the slope for CsOH was approximately 2.0 • 1015 cm4/F2V, and for NaOH and KOH it was again 2.1 x 1015 cm4/F2V. The magnitude of the shift in Vr~ remained the same throughout the frequency range.

VFB and the slope of the M-S plots did not change for the CdSe (0001) face in NaOH, KOH, or CsOH solutions. As indicated in the table, the (Jnly crystal face that shows a flatband shift when the cation is changed to Cs is the CdSe (0001) face. (C2Hs)4NOH also shows a negative flatband shift, but the M-S plots were much less ideal with more frequency dispersion and a smaller linear range. Others have measured values for Vr~ in NaOH ranging from -700 to -1200 mV v s . SCE. 11'13'29'3~

Single-Crystal Impedance Experiments in Polysulfide The same series of impedance experiments for CdSe

(0001), (0001), (I120) were performed for the four different cations, Na § K § Cs § and (C2Hs)4N § in polysulfide solutions to investigate any effects due to the presence of the [Sz] / [$2] 2- couple and the other polysulfide species. The impedance measurements in (C2Hs)4N§ were not reproducible. The results for CdSe (0001) at 25 kHz with the different polysu]fides are shown in Fig. 7. There was a large negative shift in V~B for this face as described above. The decrease in the slope of the M-S plots with Cs § as the cation parallels the results from the hydroxide experiments. The difference in the slopes between the polysulfide with Cs § as the cation and that with K§ Na § as the cation is most evident at the lowest frequencies. At 63 kHz the slope was similar to that of the the other two cations.

61117

500

4OO

~ 3OO

~ 2OO

700 [] Cs F K ~//_

IOO O - I .6 -1.5 -1.4 -I.3 - I .2 -I.1 -1 -0.9 -0.8 -03 -0.6 -0.5

Applied Bias v. SeE, V

Fig. 4. Molt-Schottky plot for polycrystalline CdSe in K and Cs ~lysulfide (800 Hz).

Table V is a summary of the V~ measurements compared with the measurements of Voc for the single-crystal CdSe electrodes in the different polysulfide solutions. The measurements made for the polycrystalline electrodes and values from the literature are included.

The etched CdSe (0001) face, listed on the third line of Table V, showed no negative shift with Cs + polysulfide as compared with Na § or K + polysulfide in either Voc .or in Vn. The slopes of the M-S plots between the different cations for the (0001) face were not significantly different. This finding parallels the results with the CdSe (0001) face in the different hydroxide solutions. Consistent through all the other data presented in Table V is a large increase in both Voc and VFB when the cation is changed to Cs § from Na + or K +. Voc and VF~ for Na + and K +, however, showed no difference between the two cations.

The CdSe (0001) and (0001) faces were tested after operating in K+-polysu]fide and Cs+-polysulfide solar cells for 20 min. The electrodes were biased at -0 .4 V v s . the counterelectrode while il luminated at 100 mW/cm 2. That potential was close to the optimum as measured in the performance curves shown earlier. The impedance was then measured as before without illumination. For all the crystal orientations the frequency dispersion increased once the CdSe electrode was exposed to il lumination and current was passed. The slopes for the M-S plots were smaller than before running in the cell.

After operating in the cell, the flatband potential for CdSe (0001) in both the K + and Cs+-polysulfide shifted to more negative potentials. V~ for the CdSe (0001) in Cs + polysulfide shifted 200 mV more negative from -1300 to -1500 mV. V~ for the CdSe (0001) in K § polysulfide shifted only approximately 60 mV negative from - 1350 to - 1410 inV. These values are much closer to those for the CdSe (0001). Thus, after a short period of operation in the cell, the CdSe (0001) also demonstrated a negative shift in V~ with change to Cs +.

Discussion Several changes in the electrode-electrolyte interface for

hydroxide and polysulfide solutions with Na + or K § to Cs § cations have been observed: the oxidation current and, to a lesser extent the reduction current on platinum, increased when the potential was scanned. The redox potential shifted to more negative values with Cs + cations, the open- circuit potential increased, and the flatband potential shifted to more negative values. The slope of the M-S plots decreased and showed more frequency dispersion with Cs + than with K + or with Na +.

The cation is not involved directly in the electrode reactions; however, its charge density affects several solution and surface equilibria. The smaller cations have a larger hydration shell, with Li, K, Na, and Cs having primary hydration numbers of 5 _+ 1, 4 _+ 1, 3 --- 2, and 1 ___ 1 respectively? 2 In the experiments with the different polysulfide solutions on platinum, Eredo x changed a total of -30 mV when the cation was changed from Na § to Cs § due to a change in the equilibrium concentrations of the redox species. The electroactive species in the charge-transfer reactions are disulfide, [$2] 2 , and supersulfide, [$2]- which are maintained by the reactions listed in Table I. The basic- ity of the solution affects the distribution of polysulfide species in solution.

An increase in the activity of the hydroxide solutions when the cation is changed to larger alkali metal cations has been reported previously. 33 For 3 molal solutions, the hydroxide ion activity was 1.48, 2.35, and 3.24 for LiOH, NaOH, and KOH, respectively; data are not available for the CsOH solutions at these concentrations. Assuming that the change in activity in the polysulfide solutions behaves the same, there would be an approximately 40% increase in the activity of OH from Na + to K +.

The change in the distribution of species with a change in the OH activity can be estimated using the equilibrium relationships listed in Table I. An increase in the activity of OH increases the concentration of the smaller chain poly-

30

0 I

-1200 -1000

"7 25

o~ 20 "O

,, 15

E O

10

O

" 5

-800 -600 400 -200 0 200 400 600

J. Electrochem. Soc., Vol. 140, No. 9, Sep tember 1993 �9 The Electrochemical Society, Inc. 2585

Applied Bias v. SCE, mV

Fig. 5. Molt-Scholtky plot for CdSe (000i) in 1M NaOH (1.6 to 63 kHz).

sulfide species, [$3] 2- and [$2] 2-, and decreases the concentration of the longer chain [S~] 2- and [$5] 2-. The concentration of supersulfide, [$2] , is maintained by the dissociation of [$4] 2- and decreases with an increase in the OH~ activity. Thus, the E~ao~, as given by the Nernst equation shifts approximately - 7 to -10 mV when sodium is replaced by potassium, due to the changes in reactant and product activities. The measured change in E~eao~ is -18 mV, which is within the error in the measurements and equilibrium data.

Also evident from the different polysulfide solutions on platinum is an increase in the current density with a change to Cs § The oXidation current for Cs+-polysulfide at an overpotential of 80 mV is more than 4 times that of the Na +- polysulfide. There was also a slight increase in the reduction current with a change to Cs § as the cation.

The current-potential measurements in this study agree with those of Lessner e t al. on platinum electrodes. 24 In that study, it was determined that the rate expression for the charge transfer reaction coupled With the homogeneous chemical reactions could be writ ten as in Eq. 10

i ~-F = k,[$24 ]2/31OH-]4/3[HS-]4/3e (F*/2~T) - ko[Si]e ( z ~ r / [10]

The change in the current is caused by either an increase in the charge-transfer rate constants, k~ and kc, or a change in the activities of [S~] ~-, [OH]-, [HS] , and [S~]-.

The increase in oxidation current also can be related to the increase in [OH]- activity discussed above, with the

change in Ereaox. At high anodic overpotentials (e.g., 80 mY), the cathodic back reaction can be neglected. The increase in current with change in cation can be estimated using the equilibrium relationships in Table I. The change in polysulfide species concentrations assuming a 40% increase in the OH- activity with change from sodium to potassium leads to an increase in the current by a factor of 1.6 to 1.8. This is close to the measured difference between the current with Na § and K*, an increase by a factor of 2.

The reduction current, however, also shows a small increase in magnitude. The shift in the distribution of the polysulfide species described above does not predict this increase but rather a small decrease in the magnitude of the reduction current by 2 to 3 %. Thus, we must conclude that in addition to the change in activity of the polysulfide species, there is a change in the rate constants. Others working with the ferricyanide-ferrocyanide redox couple on platinum have found that the current increases with a change in cation in the order of Li + << Na + << K § << Rb ~ << Cs+. ~4'a~ The rate constants that Kuta and Yeager calculated for the ferricyanide-ferrocyanide couple in the different alkali metal chloride electrolytes on a gold electrode are 0.014, 0.02, 0.03, and 0.05 cm/s for NaCI, KCI, RbCI, and CsCI, respectively. The increase in the rate constant appears related to the cation size and is sufficient to explain the increase in reduction current density.

Licht et al. ~8 also measured the I-V response of the different polysulfide solutions of Cu2S and found that the current increased in the order of Li + << Na § << K + << Cs +. They

30

25 ,qe

O

'~ 20 r "0

,- 15 I1.

E 10

0 5

0 -1200 -1000

I I I ( L

-800 -600 400 -200 0

i !!iiiii;1 I I

200 400 600

Fig. 6. Mott-Scholtky plot for CdSe (000]) in NaOH, KOH, and CsOH at 25 kHz. The two series with CsOH show the reproducibil- ity with different CdSe (0001) electrodes.

Potential v. SCE, mV


Table IV. VFe in hydroxide solutions.

(mY vs. SCE)

NaOH KOH CsOH (C2H~)4NOH

(ooo~) -842 -854 -971

-1328

Table V. Comparison of V~ and VFB results with literature.

(OOOl) -720 -742 -732 -885

re la ted the increase in the magni tude of the current to an increase in the bulk conduct iv i ty in the polysulf ide solu- t ions wi th the larger cat ions which they measured as 0.23, 0.265, 0.365, and 0.372 S /cm for Li § Na § K § and Cs*, respectively. However, here the cell was conf igured so as to reduce the effect of the solut ion resis tance and the bulk resis tance of the solut ion is small, less than 5 ~. For the currents measured on p la t inum in this exper iment (<<50 ~A) this would be an IR drop of less than 0.3 mV at anodic overpotent ia ls of 80 mV. The solut ion IR drop does not contr ibute s ignif icant ly to the overal l po lar iza t ion and it is unl ikely that the increase in bulk conduct iv i ty can expla in the current differences on p la t inum wi th the different cations.

The results f rom this s tudy indicate that the two mecha- nisms, a change in the act ivi t ies of the polysulf ide species and an increase in the charge- t ransfer ra te constant wi th Cs § affect the cur ren t -po ten t ia l response on p l a t inum in an addi t ive way. Increas ing the ac t iv i ty of the hydroxide in the polysulf ide solut ion also changed the d is t r ibut ion of the polysulf ide species. This change in the solut ion shifts Er~dox slightly negatively in increments from Na § to K + to Cs +. In addition, there is a surface effect which acts to increase the rate of the charge transfer accounting for the reduction current increase and part of the oxidation current increase which occurs in a jump from Na § or K + to Cs § The solution effects show stepwise changes from Na § to K § to Cs § whereas those effects due to a change in the surface show a large jump with change to Cs +. The effects of changing the cation to Cs § shown in the performance and impedance measurements also can be attributed to a combination of solution and surface effects.

Changes in the short-circuit current and the fill factor of the phot0voltaic cell can be attributed primarily to solution effects. I~, the maximum net current, measured on the polycrystalline electrodes showed a steady increase for Na, K and Cs of 11.9, 13.9, and 16.9 mA/cm 2, respectively. Licht ~8 measured a smaller but steadily increasing trend with single-crystal CdSe from 19.7 to 20.8 to 21.3 mA/cm 2. This result follows the same pattern as the increase in the oxidation current on platinum. The increase in rate constant with Cs § also contributes to the increase in current, but as in the measurements on platinum, to a lesser extent. The short-circuit current for the single-crystal electrodes here does not show improvement with the addition of Cs § cations as does the short-circuit current with the polycrystalline electrodes. This result probably is due to limitations in charge transport within the semiconductor at the higher currents.

2 . 3

+ +

�9 . + . 4 2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i ~ . . . . . I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . + . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . . . . . . . . . . : ~ . . . . . . . . . ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . : ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

~ . 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

�9 N a O H + K O H ~ C s O H

1.5 I i i L I I I I I I I I I I I I I

10 1 0 0

F r e q u e n c y , k H z

Fig. 7. Molt-Schotlky slope vs. frequency for NaOH, KOH, and CsOH.

x

o

Na K Cs

This study: polycrystalline c -1.31 - - -1.32 -1.44 -1.37 -1.55 ( 0 0 0 ~ [ ) a - - -1.41 -1.35 -1.42 -1.44 -1.57 ( 0 0 0 1 ) a - - -1.30 -1.45 -1.35 -1.43 -1.29 (0001) b . . . . 1.41 - - -1.50 (1120) c - - -1.36 -1.38 -1.37 -1.40 -1.51

Licht et al. TM -1.38 -1.53 -1.38 -1.54 -1.48 -1.64

Allongue et al. 13 - - - 1.55

a Fresh etched crystal. b Aged. ~ pH 12.

Small fill factor is a consequence of low current densities at potentials close to VFB where the cathodic back reaction, the reduction of [$2]- to [$2] 2- limits the net oxidation. The shift in the distribution of the species in the polysulfide solut ion favors an increase of [$2] 2- and a decrease in the concent ra t ion of [$2]- w i th Cs §

The negat ive shif t in VFB and the subsequent increase in Voc are due to a change in the surface of the semiconductor. The values for both paramete rs are approx imate ly equal for Na § and K § whi le there is a consis tent improvement wi th Cs +. This pa t t e rn is observed also for the polycrys- ta l l ine photoanodes , and the (0001) and (1170) s ingle-crystal photoanodes in the polysulf ide and hydroxide electrolytes.

In addit ion, the observat ion that these effects are dependent on the crystal or ienta t ion strongly indicates that this is a sur face- re la ted effect. The change in the surface is fa- c i l i ta ted by the presence of chalcogenide (e .g. , S or Se) a toms on the surface of the crystal . The CdSe (0001) face when c leaved is cadmium; the (0001) face is selenium. I t is acknowledged tha t the surfaces may change wi th the e tch- ing and cleaning of the surfaces pr ior to testing. The difference in the measurements remain however wi th all but the or ienta t ion of the face held constant. The difference in the f indings for the two faces is t ied to the behav ior of surface atoms on a compound semiconductor. The different a toms in a mixed semiconductor behave as Lewis acid and base sites. 36 A surface cadmium atom is more l ikely to act as a Lewis acid site and be more react ive wi th an anion in solution. A surface selenium atom is more l ikely to act as a Lewis base site and thus be more prone to adsorp t ion by a pro ton or cat ion such as Cs § The (0001) face, p redomi- nan t ly Se, shows a larger change wi th Cs § while, the (0001) face, in i t ia l ly Cd, is less sensi t ive to the cat ion in the solution.

Af te r running the CdSe (0001) crystal in the solar cell under i l luminat ion, a negat ive shift in the V~ wi th Cs + was observed. One effect of i a rada ic current wi th the CdSe- polysulf ide cell is the fo rmat ion of a CdS f i lm by exchange with the solution ions 3~-39

CdSe+S~ +HS-+OH--->CdS+SeS~-+H20 [ii]

The ion exchange occurs during the flow of charge when illuminated. The negative shift in the VFB with Cs + is related to the availability of chalcogenide sites.

The decrease in slope of the Mott-Schottky plots with the addition of Cs + is also due to a change in the near-surface region of the semiconductor. This effect is most evident for the (0O01) face. The change in M-S slope with Cs + is larger than that with Na + and K +, which are approximately equal. As shown in Fig. 7, the slopes of the M-S plots for NaOH and KOH are approximately equal throughout the range of frequencies while the slope for CsOH is significantly lower, particularly at the lower frequencies.

A frequency dispersion in the slope of the M-S plots was found with the addition of Cs + to the hydroxide electrolytes. Frequency dispersion in the slope most often has been a t t r ibu ted to surface states. 36 The surface states have

J. Electrochem. Soc., Vol. 140, No. 9, September 1993 �9 The Electrochemical Society, Inc. 2587

a range of time constants, RC, for charging and discharg- ing. The apparent concentration of charge carriers changes for different frequencies as the number of surface states able to respond at a particular frequency changes. That the addition of Cs § causes a frequency-dependent slope gives further evidence that the surface is undergoing a change.

The effect of cation in the solution has been explained as a shift in the distribution of the active species due to the increased hydroxide ion activity with the addition of Cs § The changes in the CdSe surface with the addition of Cs § are unexpected, particularly the negative shift in the VFB with the addition of Cs +. The addition of the sulfide anion and its subsequent adsorption on the surface shifts the flatband potential negative; V~ shifted negatively by about 600 mV with the sulfide compared to the hydroxide electrolyte. The adsorption of a cation on the surface of the semiconductor is expected to shift V~ and the band energy levels in the positive direction. If the Cs + were adsorbed more easily than the other cations, the excess positive charges on the surface of the semiconductor would have shifted the bandedges and the V~ positive.

The adsorption of alkali metal atoms on the surfaces of metals and semiconductors has been shown to cause this effect; however, the work function of the surfaces of semiconductors and metals is reduced with the adsorption of alkali metals. ~~ A reduction in work function results in a negative shift in the bandedges, and V~. The reduction in work function is dependent on the extent of coverage of the alkali metal on the surface and which alkali metal cation. It has been demonstrated on several different metallic surfaces, Ta W, Fe, Ag, Ni, and on Si that there is a larger reduction in the work function with Cs § as opposed to Na + or K+. 41"44 There is an approximately I eV greater reduction of the work function with Cs § adsorption as opposed to Na +.

The mechanism for this change in the work function and band energies has been related to the chemisorption of the alkali metal with the material surface. 42'43,4~ The chemisorption leads to charge rearrangement and induces surface states in the semiconductor which reduces the work function This result correlates well with the data presented here. In the M-S plots there was an increase in the charge- carrier concentration which could be an increase in the surface states, additionally there was increased frequency dispersion in the slope with Cs + as the cation indicating the possible increased presence of surface states. Goldstein and Van De Mark also theorized that the Cs § cations acted as more efficient charge transfer centers in their experiments with the ferrieyanide couple.

Summary The changes in the semiconductor-electrolyte interface

with the nature of the alkali-metal cation in polysulfide have been identified. The improvement in the performance of the CdSe/polysulfide PEC solar cell with the addition of Cs § reported by other researchers has been confirmed. The performance improvement occurs through both an increase in the current and in the Voc with Cs + polysulfide as compared to the other polysulfides.

The effect with Cs * is a combination of solution effects and semiconductor surface effects. There was a shift in the E~dox of - 33 mV with a change in cation from K* to Cs § as measured on platinum. This change is accounted for by a shift in th~ distribution of the polysulfide redox species with a change.in charge density and activity of the cation. The major component of the large increase in the anodic oxidation current on plat inum has been attributed to the same shift in the polysulfide species. The i-V behavior of the CdSe/polysulfide cell under i l lumination shows an increase in current with the addition of Cs + also due to the shift in the polysulfide species.

The surface of the semiconductor undergoes a change when exposed to Cs § Along with the large increase in the oxidation current on plat inum there was a small increase in the reduction current with the addition of Cs + caused by an increase in the charge-transfer rate constant. The increase in Voc and negative shift in V~B with Cs § were dependent on

the orientation of the crystal. The slopes of the M-S plots with Cs § in both polysulfide and hydroxide solutions were lower, indicting an increase in charge-carrier concentration in the semiconductor. There is larger frequency dispersion of the M-S plots with Cs § as the cation, indicative of a surface effect.

Manuscript submitted Feb. 8, 1993; revised manuscript received May 20, 1993.

REFERENCES 1. H. Gerischer, J. Electroanal. Chem. Interracial Elec-

trochem., 58, 263 (1975). 2. A. B. Ellis, S. W. Kaiser, and M. S. Wrighton, J. Am.

Chem. Soc., 98, 1635 (1976). 3. J. M. Bolts, A. B. Ellis, K. D. Legg, and M. S. Wrighton,

ibid., 99, 4826 (1977). 4. A. Heller, K. C. Chang, and B. Miller, ibid., 100, 684

(1978). 5. G. Hodes, Nature, 285, 29 (1980). 6. G. Hodes, D. Cahen, J. Manassen, and M. David, This

Journal, 127, 2252 (1980). 7. B. Miller, A. Heller, and M. Robbins, ibid., 124, 1019

(1977). 8. R.A. Boudreau and R. D. Rauh, Solar Energy Mater., 7,

385 (1982). 9. C.J. Liu and J. H. Wang, This Journal, 129, 719 (1982).

10. G. Hodes, A. Albu-Yaron, E Decker, and P. Motisuke, Phys. Rev. B, 36, 4215 (1987).

11. K. W. Frese, Jr., and D. G. Canfield, This Journal, 131, 2616 (1984).

12. S. Licht and J. Manassen, ibid., 132, 1076 (1985). 13. E Allongue, H. Cachet, M. Froment, and R. Tenne, J.

Electroanat. Chem., 269, 295 (1989). 14. J. Winnick, E M. Lessner, E McLarnon, and E. J.

Cairns, Appl. Phys. Lett., 53, 1985, (1988). 15. E M. Lessner, Ph.D. Thesis, University of California,

Berkeley, CA, pp. 6 -18, (1986). 16. S. Licht, R. Tenne, H. Flaisher, and J. Manassen, This

Journal, 131, 950 (1984). 17. S. Licht, R. Tenne, G. Dagan, G. Hodes, J. Manassen, D.

Cahen, R. Triboulet, J. Rioux, and C. Levy-Clement, Appl. Phys. Lett., 46, 608 (1985).

18. S. Licht, R. Tenne, H. Flaisher, and J. Manassen, This Journal, 133, 52 (1986).

19. R. Tenne, C. Levey-Clement, and G. Sawatzky, J. Vac. Sci. Technol. A, 4, 2471 (1986).

20. E M. Lessner, Ph.D. Thesis, University of California, Berkeley, CA, p. 7 (1986).

21. R M. Lessner, J. Winnick, R R. McLarnon, and E. J. Cairns, This Journal, 133, 2517 (1986).

22. K.M. Abraham, R. D. Rauh, and S. B. Brummer., Elec- trochim. Acta, 23, 501 (1978).

23. W. Giggenbach, Inorg. Chem., 11, 1201 (1972). 24. R M. Lessner, F. R. McLarnon, J. Winnick, and E. J.

Cairns, This Journal, 134, 2669 (1987). 25. N. Ardoin and J. Winnick, ibid., 135, 1719 (1988). 26. A. Heller, G. P. Schwartz, R. G. Vadimsky, S. Menezes,

and B. Miller, ibid., 125, 1157 (1978). 27. D. Lando, J. Manasseh, G. Hodes, and D. Cahen, J. Am.

Chem. Soc., 101, 3969 (1979). 28. S. Licht, Nature, 33O, 148 (1987). 29. A.B. Ellis, S. W. Kaiser, J. M. Bolts, andM. S. Wrighton,

J. Am. Chem. Soc., 99, 2839 (1977). 30. K.W. Frese, Jr., JAppl. Physics, 53, 1571 (1982). 31. R H. Allongue and R. Tenne, This Journal, 138, 261

(1991). 32. J. O'M. Bockris and A. Reddy, Modern Electrochem-

istry, Vol. 1, p. 117, Plenum Press, New York (1970). 33. B. E. Conway, Electrochemical Data, p. 75, Elsevier

Publishing Company, New York (1952). 34. J. Kuta and E. Yeager, Y. Electroanal. Chem. Interracial

Electrochem., 59, 110 (1975). 35. E. L. Goldstein and M. R. Van De Mark, Electrochim.

Acta, 27, 1079 (1982). 36. S. R. Morrison, Electrochemistry at Semiconductor

and Oxidized Metal Electrodes, p. 14-15, Plenum Press, New York (1980).

37. R. N. Noufi, R A. Kohl, J. W. Rogers, Jr., J. M. White, and A. J. Bard, This Journal, 126, 949 (1979).

38. M. Peisach, C. A. Rabe, C. A. Pineda, D. Mahalu, and R. Tenne, J Vac. Sci. Technol. B, 6, 1506 (1988).

39. A. Silberstein, J. AppI. Phys., 54, 5428 (1983).


40. G. E. Rhead, Surf. Sci., 203, L663 (1988). 41. P A. Serena, J. M. Soler, N. Garcia, and I. P. Batra,

Phys. Rev. B, 36, 3452 (1987). 42. R. M. Nieminen and K. O. 5ensen, ibid., 38, 5764 (1987). 43. Y. Enta, T. Kinoshita, S. Suzuki, and S. Kono, ibid., 39,

1125 (1988). 44. D. Heskett, T. M. Wong, A. J. Smith, W. R. Graham, N. J.

DiNardo, and E. W. Plummer, J. Vac. Sci. Technol. B, 7, 915 (1989).

45. J. D. Levine, Surf. Sci., 34, 90 (1973).

H/D Isotope Effect on Electrochemical Pumps of Hydrogen and Water Vapor Using a Proton-Conductive Solid Electrolyte

T. Hibino, K. Mizutani, and H. Iwahara Synthetic Crystal Research Laboratory, School of Engineering, Nagoya University, Chilcusa-ku, Nagoya 464-01, Japan

ABSTRACT

From the standpoint of an H/D isotope effect, the electrochemical pumping of hydrogen or water vapor has been studied using a high temperature proton conductor as a solid electrolyte. When a potential was applied to the cell at 1173 K, the pumping rate of deuterium was remarkably smaller than that of hydrogen. During hydrogen pumping, the total current was depressed by replacing H § and D +. In addition, an increase in cathodic overpotential was observed by the current-interruption method. Similar behavior was recognized in the pumping of water vapor, except that the transport number of the conducting ion decreased with replacement of D § by H §

High temperature proton-conductive solids are of indus- trial interest because of their application as solid electrolytes for a solid-oxide fuel cell, for a steam electrolyzer to produce hydrogen, or for a hydrogen gas sensor. 1 Fur- thermore, an electrochemical reactor for organic com- pounds is possible using the proton conductor as a di- aphragm. 2'3 High temperature proton-conductive ceramics based on SrCeO3, BaCeO3, CaZrO3, and SrZrO3 were first found by us ~-7 and have been investigated by various au- thors from different view points. 8-16 Particularly, the studies regarding the conduction mechanism are being made dy- namically and can be summarized as follows: (i) temperature programmed desorption (TPD) spectra of the above oxides indicate that the proton concentration becomes larger as the electronegativities of constituent atoms in oxides decrease, namely, their basicities increase, 15 (if) in- frared spectroscopy indicates that a proton is bound to an oxide ion in the lattice, 8'1~ (iii) decrease in ac conductivity by replacing H + by D + suggests that the conducting ion is not a hydroxide ion but rather a proton and that it mi- grates via the dissociation of the O-H bond, s'12'13'16 (iv) when hydrogen and argon are supplied to the anode and the cathode, respectively, and a direct current is sent to the electrolyte, hydrogen evolves at the cathode. This fact demonstrates that the charge carrier is the proton. ~

This mechanism indicates a possibility for the separation of tritium-, or deuterium-, species from hydrogen or water vapor as a new application of the proton conductor. Using the proton conductive ceramic as a solid electrolyte, one can construct an electrochemical pump of hydrogen or water vapor, the working concept of which is described in Fig. 1. If a mixture of H-, D-, and T-species is introduced into the anode and a potential is applied to the electrolyte, the order of the amount evolved at the cathode will be H- > D- > T-species as a result of the isotope effect. It is thought that the isotope effect results from the difference in rate of the migration in the bulk and/or of the electrode reaction among isotopes. Thus, as the basic research for this separation, we examine here the H/D isotope effect on the pumping of hydrogen or water vapor. Furthermore, the isotope effects on the migration of conducting ions and on their electrode reaction are investigated independently by current interruption.

Experimental Preparation of SrCeo.95Ybo.osO3-a ceramic.--The ceramic

was prepared by a solid-state reaction, followed by sinter- ing. The desired amounts of raw materials (SrCO3, CeO2,

and Yb203) were mixed with ethanol and calcined in air at 1673 K for 10 h. The oxide was ground in a ball mill and pressed into the cylindrical pellet at 2 ton cm -2. The pellet was sintered in air at 1803 K. The density of the ceramic was more than 95% of its theoretical value, and the crystal phasewas confirmed to be a single perovskite-type phase by x-ray diffraction.

Measurement of ac conductivity.--The electrical conductivity was measured at 973 to 1173 K by a complex impedance method. The ceramic was cut into a square pil- lar-shaped specimen to eliminate the contribution of the electrode resistance. Porous Pt metal was attached to the top and bottom faces of the specimen as an electrode material. The measurement was carried out in hydrogen, deuterium, or wet air. Wet air was prepared by passing air through a water or heavy water bubbler at 293 K.

Electrochemical pumping of hydrogen or water vapor.- The schematic illustrations of the apparatus are shown in Fig. 2 a and b. The ceramic was sliced into a thin disk (thickness: 0.5 mm and diam: 13 mm), as shown in Fig. 2a. Platinum paste was baked on both faces of the disk at 1173 K as an electrode material. Each electrode was attached with two Pt lead wires, one of which was connected to a potentiostat and the other of which was connected to an electrometer to eliminate the ohmic resistance of Pt lead wire. Platinum paste also was coated on the side face of the disk as a reference electrode, and was exposed to the atmosphere with a constant partial pressure of water vapor. On the reference electrode, three equilibria, Eq. 1, 2, and 3, were established simultaneously between the ceramic oxide and the atmosphere, regardless of the anodic and the cathodic reactions 17

_1 O2 + V~ = Ox + 2h" [1] 2

H20 + 175 = Ox + 2H + [2]

1 H20 + 2h" = 2H + + ~ O2 [3]

where H § h', V~, and Ox denote the proton, hole, oxygen vacancy, oxide ion on a normal lattice site in the oxide, respectively.

In the experiment of the electrochemical hydrogen pump, 1 atm hydrogen or deuterium and argon were introduced into the anode and the cathode, respectively, at 1173 K. The voltage was applied potentiostatically, and hydrogen or deuterium evolved at the cathode was detected by a gas

cation effect on the cdse-liquid...

Documents