electrodes - American Chemical Society

Department of Chemistry, Mount Holyoke College, South Hadley, Massachusetts 01075. (Received: October 29, 1984). The energy threshold for electron ...
0 downloads 0 Views 628KB Size
J . Phys. Chem. 1985,89, 1453-1457 Acknowledgment. For support of this work we are indebted to the National Science Foundation 69A Program for an instrumentation grant and to the Department of Energy, Office of Basic Energy Sciences. We are also grateful for the electronic work of A. Chace and the laboratory work of E. Fitzgerald. This work

1453

would not have been possible without the donation of the excellent samples of ZnO crystals by Prof. Helbig. Registry No. 9E-DiSC2(3),35077-88-4; DiQC,(l), 20766-49-8; DiBAC1(3),32690-13-4; ZnO, 1314-13-2;KCl, 7447-40-7.

Examination of the Energetic Threshold for Dye-Sensitized Photocurrent at SrTiOs Electrodes L. P. Sonntag and M. T. Spitler*+ Department of Chemistry, Mount Holyoke College, South Hadley, Massachusetts 01 075 (Received: October 29, 1984)

The energy threshold for electron transfer has been determined for the oxidation of an excited dye molecule over the conduction band of SrTi03singlecrystal electrodes. When the pH of the electrolyte is varied from 4 to 11, the conduction band energy of SrTi03 was shifted from below to above the energy of the electron in the excited state of two donor dyes, 2,2’-diethylthiacarbocyanine and 2,2’-diethyloxadicarbocyanine. The resulting threshold has been analyzed with the aid of several models for electron transfer at semiconductor electrodes.

The relationship between the rate of an electron-transfer reaction and the free energy has been well-defined for solution reactants. In most of these studies the transfer rate between an acceptor and a variety of electron donors was measured as a function of the energy change of the reaction;’-3 in other cases4 the reduction rate of a series of electron acceptors was determined by using a common electron donor. As the energy change of these reactions ranged from endothermic to exothermic the reaction rates increased exponentially until a diffusion-limited rate constant was observed. These characteristics have been analyzed in terms of current theories of electron transfer and have been used to suggest improvements and modifications in Similar correlations between heterogeneous rate constants and the reaction energies have been made in electrochemical systems. At the surface of organic molecular crystals the rate constants for hole injection into the electrode by oxidants follows theory very we11.8 At photoexcited semiconductor electrodes the oxidation of reductants as a function of their redox potentials yields relative rate constants which also rise exponentially to a diffusion-limited val~e.~J* At semiconductor electrodes it is also possible to determine the relation between the electron-transfer rates and free energy for just one electron donor or This can be done at the surface of oxide electrodes such as ZnO, SrTi03, and TiOz because of a pH-dependent double layer through which the energy levels of the solid can be shifted with respect to solution levels by 60 mV per unit change in pH. In this type of experiment, however, more refined information about electron transfer can be obtained than is possible in the experiments involving series of electron donors and acceptors. In particular the distribution function can be determined which describes the energy of the donor or acceptor species as a function of its particular solvation state. Such a distribution constitutes the instantaneous density of states for the electron donor or acceptor in solution. This point has been discussed in detail by Vanden Berghe et ai.’* in a study of the one-electron reduction of Fe(CN)63- at ZnO single-crystal electrodes in which the magnitude of the current density a t two p H values was used to calculate the rearrangement energy of the Fe(cN)63 ion. Further use of this feature of semiconductor electrochemistry has been made by Clark and SutinI3 in the estimation of the rearrangement ‘Presently on leave at the Solar Energy Research Institute, Golden, CO.

0022-3654/85/2089-1453$01.50/0

energies of several ruthenium complexes photooxidized from solution a t TiOa electrodes. However, there has been no comparable study of rearrangement energies and thresholds for electron transfer involving reactants adsorbed at the semiconductor surface at the interface of two differing dielectric media. In this work this method will be employed to determine the distribution function for the instantaneous electron energies of adsorted reductants and to assess the accuracy of the Gaussian function which has been proposed to describe it.I4J5 The photoelectrochemical system selected for this experimentation consists of a SrTi03singlecrystal electrode as the electron acceptor and two adsorbed cyanine dyes which serve as electron donors in their excited states. These dyes have been selected so that the average energy of the electron in the excited state equals that of the conduction band edge of SrTi03 a t about pH 4. Thus a threshold for electron transfer should be crossed as the electrolyte pH is increased because the energy of the SrTi03 conduction band moves above that of electron in the excited dye. The ability of these excited dyes to inject electrons into the conduction band of SrTi03 will be measured through a laser attenuated total reflection (ATR) technique which will provide QP, the quantum yield for current production at each pH. (1) Rehm, D.; Weller, A. Ber. Bunsenges. Phys. Chem. 1969, 73, 834. (2) Vogelmann, E.; Scbreiner, S.;Rauscher, W.; Kramer, H. Z . Phys. Chem. (Wiesbadenl 1976. 101. 321. (3) Bakrdini, R’.; Van& G:; Indelli, M.; Scandola, F.; Balzani, V. J. Am. Chem. SOC.1978, 100, 7219. (4) Nagle, J. K.; Dressick, M. J. Meyer, T. J. J. Am. Chem. Soc. 1979, 101,3993; ’ ( 5 ) Indelli, M.; Scandola, F. J. Am. Chem. SOC.1978, 100, 7733. (6) Efrina, S.; Bixon, M. Chem. Phys. 1976, 13, 447. (7) Van Duyne, R.; Fischer, S . Chem. Phys. 1974, 5, 183. (8) Willig, F. Adu. Electrochem. Electrochem. Eng. 1981, 12, 1. ( 9 ) Tamura, H.; Yoneyama, H.; Kobayashi, T. In “Photoeffects at Semiconductor-Electrolyte Interfaces; Nozik, A., Ed.; American Chemical Society: Washington, DC, 1981; p 131. (IO) Inone, F.; Fujishima, A.; Honda, K. Bull. Chem. Soc. Jpn. 1979,52, 721 7.

(11) Schumacher, R.; Wilson, R. H. Harris, L. A. J . Electrochem. SOC. 1980, 127, 96. (12) Vanden Berghe, R. A,; Cardon, F.; Gomes, W. P. Surf. Sci. 1973, 39, 368. (13) Clark, W. D. K.; Sutin, N. J. Am. Chem. Soc. 1977, 99, 4676. (14) Gerischer, H.Adu. Electrochem. Electrochem. Eng. 1961, 1 , 139. (15) Gerischer, H. Photochem. Photobiol. 1972, 16, 243.

0 1985 American Chemical Society

1454 The Journal of Physical Chemistry, Vol. 89, No. 8, 1985 Experimental Section The SrTiO, single crystals were obtained from Atomergic Chemetals Corp. in the form of a boule which was cut in slices to expose the (1 11) plane for use as the working surface. A portion of the samples were polished to a 0.25 pm finish and cut to an area of 10 mm2 for use as electrodes in the standard photoelectrochemical setup. Other crystals were cut, ground, and polished in a similar manner to the shape of prisms for use as internal reflection elements in the laser ATR system. All electrode material was doped and annealed by heating in wet H2 at 900 OC for 1 h. The quality of the electrode surfaces was determined through photocurrent-voltage curves by using band-gap radiation and impedance measurements. Only those crystals exhibiting ideal photocurrent-voltage relationships and linear Mott-Schottky plots were used in these studies. The two dyes selected for study were 2,2’-diethylthiacarbocyanine (DiSC2(3)) and 2,2/-diethyloxadicarbocyanine(DiOC2(5)). They were obtained from Accurate Chemical Co. and examined for purity by using thin-layer chromatography. Fluorescence measurements of these dyes were made with a Perkin-Elmer MPF44B spectrofluorimeter equipped with a DCSU-2 differential corrected spectrum unit. The electrochemical characteristics of thee dyes were examined by using cyclic voltammetry performed at a glassy carbon electrode with an IBM EC225A potentiostat after the manner of Loutfy and Sharp.I6 The standard photoelectrochemical apparatus was employed to obtain photocurrent-voltage curves and photocurrent action spectra for the dyes. This equipment consisted of a 150-W Xe lamp, a monochromator, and a chopper which provided modulated, monochromatic light to illuminate the SrTiO, electrode. All potentials in this work are referenced to the SCE. The modulated currents were amplified in a home-built potentiostat and detected by an EG&G Model 5204 lock-in amplifier. Impedance measurements were made with this equipment using the lock-in technique. The laser ATR method for determining 9,has been described in detail in previous work.I7-l9 Its basic features can be summarized briefly. Light from an Ar+ laser or dye laser is internally reflected within a semiconductor electrode shaped as an internal reflection element; as the beam exits the electrode, its intensity is measured by a photodiode. Absorption of laser light by dye adsorbed on the electrode surface results in a decrease in the photodiode signal and an increase in the photocurrent measured by the potentiostat. The incident laser light is modulated by an electrooptic shutter to allow pulses of light ranging from microseconds to milliseconds to be employed. In this way a simultaneous measure of the optical density of the adsorbed dye and the photocurrent produced by this layer may be made; with this information, a, the number of electrons measured as current per unit photon absorbed, can be determined. Results Fluorescence spectra were taken for DiOC2(5) and DiSC2(3) to determine the influence of pH on the photophysics of these dyes. Solutions prepared in the same manner as those in the photoelectrochemical experiments were used, and the wavelengths of maximum excitation and emission were found to remain constant to within 2 nm over the pH range of 4.0-1 1.0. The quantum efficiencies of fluorescence for these dyes also remained constant with pH to within 5%. Over this same pH range the oxidation potentials for the dyes were determined by using cyclic voltammetry in a methanolic solution which was 0.1 M in LiCl. The current-voltage curves for all dye solutions, although irreversible, remained identical in form, showing potential shifts of less than 20 mV. The oxidation (16) Loutfy, R.;Sharp, J. Photogr. Sci. Eng. 1976, 20, 165. (17) Kavassalis, C.; Spitler, M. T. J. Phys. Chem. 1983, 87, 3166. (18) Spitler, M. T. In “Photoelectrochemistry: Fundamental Processes and

Measurement Techniques”; Wallace, W., Nozik, A., Deb, S.,Eds.; Electrochemical Society: Princeton, NJ, 1982; p 282. (19) Spitler, M.; Lubke, M.; Gerischer, H. Ber. Bunsenges. Phys. Chem. 1979, 83, 663.

Sonntag and Spitler potentials of these dyes were found to be +0.82 V vs. SCE for DiSC2(3) and +0.62 V for DiOC2(5). With excitation energies E, of 2.17 and 2.07 eV the excited-state oxidation potentials O E D . p + of these dyes are -1.35 and -1.45 V whereZo oED*ID+ = OE,/Dt - E , It must be noted that these values for the energy of the donor levels of the dyes are approximations that have served as guides in the selection of the dyes to be used in these experiments. Their accuracy as guides is limited by the use of an ‘E,, value estimated from an irreversible oxidation wave in the cyclic voltammetry. Through the very thorough work of Tani2I comparing the potentials of reversible and irreversible oxidation waves of over 50 cyanine dyes (including these two) with M O estimates of their highest occupied energy levels, the error can be assessed reliably at no more than 100 mV. At the surface of ZnO single crystals, the situation is different. For dyes which are irreversibly oxidized at metal electrodes, the lifetime of the oxidized radical at ZnO surfaces has been measured to be as long as 30 ms.22 This is certainly much longer than the microseconds required for the photocurrent measurement and the hundreds of picoseconds involved in the initial electron transfer. Therefore, the reaction can be considered to be effectively reversible within the context of these experiments. Given that these essential characteristics of the dyes are pH independent, it was next confirmed that the SrTi03 electrodes exhibited a pH-dependent flat-band potential. When the flat-band potential that has been derived from both the photocurrent and impedance methods was used, the band edges of SrTiO, were observed to shift by 58 f 3 mV per pH unit. At pH 7.0 and a frequency of 100 Hz, the photocurrent-voltage curves and Mott-Schottky plots gave a flat-band potential, Efb,of -1.1 V and a donor density of 1 X 1019/cm3. This value for Eb agrees well with other literature report^.^,-^^ Following this preliminary work, the pH dependence of the photocurrent action spectrum of DiSC2(3) was determined in an electrolyte 0.1 M in KCl and 5 X in dye. The electrodes were biased at 0.0 V with respect to SCE so that a band bending of 700-1 100 mV was maintained throughout these experiments. The results are shown in Figure l a where the photocurrent is plotted as a function of the wavelength of incident light. A plot of the photocurrent maximum of the dye as a function of pH reveals a definite threshold for sensitized photocurrent which is shown in Figure 2a. Depicted in Figure 1b and 2b are data for DiOC2(5) which were gathered in a similar manner (the data of Figure 2b for DiOC2(S) are taken from ref 27). The adsorption of organic dyes on semiconductor surfaces is known to be pH dependent.26 Therefore, the laser ATR apparatus must be employed in studies of this system since it accounts for the pH-dependent dye adsorption in the calculation of 3,. 9,was determined at 575 nm for DiSC2(3) on SrTi03 as a function of pH; the results are given in Figure 3a. A threshold for 9,for DiOC2(5) at 585 nm was also observed at about pH 7 and is shown in Figure 3b. The experimental errors in these measurements were smallest for DiSC2(3) and grew progressively larger for DiOC2(5) from higher to lower pH. The 9, values of Figure 3 represent the average for surface coverages where the dye molecules are calculated to be separated by an average distance of 35 A or greater, that is, an absorption of less than 0.002 of the incident light. The @p for surface coverages greater than this do, however, reflect (20) Gerischer, H.; Willig, F.In “Topics in Current Chemistry”; Davison, A., Ed.; Springer-Verlag: New York, 1976; Vol. 61, p 31. (21) Tani, T. Photogr. Sci. Eng. 1972, 16, 258. Tani, T.; Kikuchi, S . Photogr. Sci. Eng. 1967, 1 I, 129. (22) Matsumura, M.; Mitsuda, K.; Tsubomura, H. J. Phys. Chem. 1983, 87, 5248. (23) Vanden Kerchove, F.; Vandermolen, J.; Gomes, W. Ber. Bunsenges. Phys. Chem. 1979,83, 230. (24) Bolts, J.; Wrighton, M. J . Phys. Chem. 1976, 80, 2641. (25) Watanabe, T.; Fujishima, A,; Honda, K. Bull. Chem. SOC.Jpn. 1976, 49, 355. (26) Spitler, M.; Calvin, M. J. Chem. Phys. 1977, 66, 4294.

The Journal of Physical Chemistry, Vol. 89, No. 8, 1985 1455

Dye-Sensitized Photocurrent at SrTiO, Electrodes

(0)

I

I

I I

I

,

450

I

,

Disc,@)

t

,

550

500

I

600

I

,

650 4

WAVELENGTH (nrn)

6

5

7

9

8

IO

II

PH (b)

( b)

El A

T

c Y I c

._ c

.-

?

DiOC,(5)

A

A--\

e> - --

;\

0

z

w L.' LL

w LL

t

-

I

I

450

I

I

'

500

I

1

550

'

600

8

'

1

\

\

-

\

\

1

650

WAVELENGTH (nm)

Figure 1. Photocurrent action spectra for DiSC2(3)at SrTi03at several values of the electrolyte pH. The dye concentration was 2 X lo-' M in a 0.1 M KC1 electrolyte. The peak at 470 nm is the increasing extinsic absorption tail of the crystal truncated by a 450-nm cutoff filter. (b) Photocurrent action spectra for DiOC2(5)are also seen to be a function

of pH.

4

5

6

8

7

9

IO

II

PH Figure 3. (a) aPdetermined for DiSC2(3)at 575 nm as a function of pH by using the laser ATR apparatus. The best fit of the data to eq 3 and 4 is given as a solid line for W = 3.0. The dashed lines show how the calculated threshold varies as values for Wof 8.0 and 1.0 are used. (b) @p measured at 585 nm for DiOC,(S) as a function of pH. The best fit for the data is given by the solid curve for W = 2.0. The dashed lines show the calculated threshold for rearrangement energies of X = 250 mV

and X = 300 mV. dependent energy difference between Ep/D+and the conduction band edge. However, it is possible that the electron transfer from the dye can be followed by a surface trapping and a subsequent recombination with the parent dye. If the rate of this recombination were dependent on the ED./D+-band edge difference, then the measured photocurrent would reflect the kinetics of this process as well as that of the initial electron transfer. This recombination model has been discussed in detai117*27-30 with the conclusion that 9, may be written

9, =

RH

Figure 2. Maximum current at 570 nm of each action spectrum for

DiSC2(3)in Figure la plotted to reveal a threshold for photooxidation as the pH is increased from 4 to 11. (b) Similar plot of the photocurrent maximum at 595 nm for DiOC2(5)as a function of pH. thresholds similar to those seen in Figure 3 although with larger error. 9,for those dyes at lowest surface coverage was estimated to be 2 X for Di0,(5) and 1 X for DiSC,(3). Discussion Before analysis of the data, it must be determined whether or not the photocurrents measured in these experiments are an accurate reflection of the ability of the dye to inject an electron in the semiconductor. This ability should be a function of the pH-

@et@,

where @'et represents the quantum efficiency of electron transfer at the surface, and 9 , is the efficiency of escape from recombination. Recent work17 has shown that , 9 can remain constant over a large energy range so that it is likely that aPdoes reflect 9 , . The good agreement of theory with results in the experiments of this work supports this assumption. It is also possible for a fast energy transfer between the adsorbed dye molecules to be responsible for the low @p.32*33 With this (27) Spitler, M. T. J. Chem. Educ. 1983, 60, 330. (28) Gerischer, H.; Spitler, M.; Willig, F. In 'Electrode Processes 1979"; Bruckenstein, S.,Ed.; Electrochemical Society: Princeton, NJ, 1980; p 115. (29) Gerischer, H. Surf. Sei. 1980, 101, 518. (30) Arden, W.; Fromherz, P.J . Electrochem. SOC.1980, 127, 370. (31) Bressel, B.; Gerischer, H. Ber. Bumenges. Phys. Chem. 1983,87,398. ( 3 2 ) Liang, Y.; Ponte Goncalves, A.; Negus, D. J . Phys. Chem. 1983,87, 1.

(33) Miyasaka, T.; Honda, K. Surf.Sci. 1980, 101, 541.

Sonntag and Spitler

1456 The Journal of Physical Chemistry, Vol. 89, No. 8, 1985

In light of the comments about trapping of the injected electron, it is not clear that the semiconductor acceptor states will have the distribution given by the conduction band density of states, D(E). The acceptor states may well be surface trapping states with a fairly constant distribution with respect to energy. In the evaluation of eq 1, however, it makes very little difference which one is used because the exponential function of eq 2 dominates the integral of eq 1. We will use a constant density of states because it greatly eases the data analysis. With a bias potential on SrTi03 of 0.0 V,f(E) can be taken to be zero; if it is assumed that the variation in v ( E ) with E is small, then the expression for j may be simplified to j

a

Lm [D*] exp[-(E

- 0E)2/4XkT] d E

(3)

Given the efficiency of current production of about low2and a low light intensity, it can be assumed that [D*] does not change with time. In this case [D*] can be brought out of the integral. With a change of variables eq 3 may then be expressed as j

semi conduct or dye Figure 4. Energy-level diagram in which the variables of eq 1 and 2 are pictorially defined. Electron transfer occurs from a distributionof states D*(E) about a mean value OE to the conduction band of the semiconductor characterized by its band edge E,.

mechanism the excitation quickly finds its way to a site where it is quenched without electron transfer to the electrode. The measured current would directly reflect an electron-injection process which is merely a relatively slow and minor side reaction for the dye in the excited state. In this case the relationship between the energetics and kinetics of the electron-transfer process would also be preserved in the current drawn from the electrode. The analysis of the results of Figure 3 will proceed in a manner similar to Vanden Berghe et al.'* with the difference that this work concerns the oxidation of an excited dye molecule instead of a dark reduction reaction. An expression for the current flow resulting from oxidation over the conduction band of semiconductor electrodes has been derived by Gerischer.14 For the oxidation of an excited dye D', the result takes the form E,

(1)

where u(E) represents a transfer frequency for the electron, D(E) is the density of states of the conduction band, D*(E) is the number of excited dye molecules at the surface, and f(E) is the fermi function. All of these factors are a function of the energy of the electron, E . In subsequent work15 Gerischer defined D*(E) more explicitly for the donor dye level as D'(E) = [ D * ] ( ~ T X ~ T )exp[-(E '/~ - 0E)2/4Xkr]

[D*]Jmexp[-(x

+ p + X)2/4XkT]

dx

where X = ( E - E,) and p = E, - 'ED./D+. As E, shifts with pH for SrTi03, p changes and as a result j changes also. Thus the pH dependence of j can be calculated if an appropriate value for X is used. However, in order to compare such a theoretical prediction with results of Figure 3, it must be recast in the form of a quantum efficiency for electron transfer

Ev

j = q I m u ( E )D ( E ) [ l - f(E)]D*(E) d E

0:

(2)

where X is the reorganization energy for the excited dye, [D*] is the concentration of excited dye molecules, and 'E represents the most probable energy for the electron in the excited state of the dye. These terms are depicted in the energy-level diagram of Figure 4. Equation 2 represents a probability distribution function for the energy of the excited electron and is determined by the fluctuations in its energy of interaction with the surrounding dielectric. In this case the environment consists of half electrolyte and half semiconductor as the dye is adsorbed at the interface of the two.

*p

= [k(ku,X)/(k(P,h)+ kq)I@esc

where k(p,X) = j/([D*]@.,) and kq represents the sum of all other rate constants for quenching the excited state. We take it to be possible for k(p,X) to attain a maximum at any energy in the range where O E is below E,. This entire range of possible maxima was covered in determining the best theoretical fit to the data. To evaluate 9, as a function of pH, the ratio W = k,/k(p,X) must be estimated. At monolayer concentrations of the dye k, is determined by energy-transfer quenching and has been measurdg2to be no faster than 1.6 X 1O'O s-'. Under conditions which maximize &,A), k,, can be e ~ t i m a t e d ~ ~to* be ~ ~in- 'the ~ range of 0.3-1 X 10'O s-l. Therefore Wcan be as large as 7, but as small as 2. Given the low surface concentrations used in these experiments which reduce energy transfer between the dye molecules, the lower values of W would be more accurate. With these considerations the value of aPhas been calculated for each dye as a function of pH for W = 3. When experimental measures of E, and ED*/D+were used, the solid curves of Figure 4a, were found to fit the data best. No shift of this curve along the horizontal axis was necessary even with the error in the E, and ED.lD+values, and the implicit assumption that X for the ground and excited states is the same. This best fit was found when k(p,X) was taken to be a maximum at 'E = E,. For DiSC2(3) X was found to be 375 (f25) mV while for DiOC2(5) it was 275 (f25) mV. This fit also depends on W, in Figure 3a the theoretical prediction is also shown for W = 8 and 1. This interpretation of the results is subject to the qualification that the dye molecule is assumed to be adsorbed outside and onto the pH-dependent sheath on the electrode surface so that E, is shifted the full 59 mV per pH unit relative to 'E. If a shift of less than 59 mV were effective, the potential scale in Figure 3 would then be compressed and the value of X would change. Given these results and those of Figures 2 and 3, it can be concluded that the Gerischer picture of electron transfer formulated in eq 1 and 2 constitutes an accurate description of the energetics of oxidation of reactants adsorbed on the electrode surface. In particular the Gaussian distribution of eq 2 used to (34) Spitler, M.; Lubke, M.; Gerischer. H. Chem. Phys. Lett. 1978, 56, 571. (35) Nakashima, N.; Yoshihara, K.; Willig, F. J . Chem. Phys. 1980, 73,

3553. (36) Goncalves, A., private communication

J . Phys. Chem. 1985,89, 1457-1460 describe the fluctuating energy levels of the electron predicts 9, well, especially in regions where the reaction is only mildly exothermic. Qualitatively, the shape of the threshold agrees very well with the solution experiments of Clark and Sutin.I3 It is clear, however, that uncertainties in kq and the error in the data do not permit a very precise evaluation of the rearrangement energy to be made. Nevertheless, the values of X between 0.3 and 0.4 eV for these dyes are comparable to the 0.3 eV evaluated for the like-sized dye rhodamine B in studies of simple outer-sphere electron transfers at the surface of organic molecular crystals.s We conclude that the substance of the Gerischer formulation of electron transfer may be applied with confidence to reactions of species adsorbed at the semiconductor-electrolyte interface. These dye molecules are, however, only weakly adsorbed to the surface of the electrode with heats of adsorption on the order of 6-10 kcal/mol. Therefore, it is still unknown as to whether

Phosphate-H,O

1457

oxidants or reductants deposited or chemically bound on the surface would correlate well with solution models or obey the simple distribution of levels described in eq

Acknowledgment. For support of this work we are indebted to the National Science Foundation 69A Program for an instrumentation grant and the Department of Energy, Office of Basic Energy Sciences. We also thank Dr. Ronald H. Wilson of the General Electric Corp. for the doping and annealing of the SrTi03 crystals and Artner Chace for his invaluable help with electronics. Registry No. DiSC2(3), 18403-49-1; DiOC2(5), 37069-75-3; SrTi03, 12060-59-2; KCI, 7447-40-7. (37) Gerischer, H. In 'Electrocatalysis on Non-Metallic Surfaces"; Franklin, A. D., Ed.; National Bureau of Standards: Washington, D.C., 1976; P 1.

Interactions in Concentrated Aqueous H3P04Solutions

R. Camidti,* P. Cucca, and D. Atzei Istituto di Chimica Generale, Inorganica e Analitica, Universitd di Cagliari, 091 00 Cagliari, Italy (Received: March 16, 1984; In Final Form: November 16, 1984)

X-ray diffraction data from two concentrated aqueous H3P04solutions were examined. A peak in the range 3.6-3.9 8, in the correlation functions reveals the presence of phosphate-H,O interactions. Each H,PO, molecule interacts with about four water molecules. A detailed analysis was made to analyze the influence of the H3P04hydration phenomena.

Introduction

Numerous diffractometric studies on solutions of sulfate in the presence of different have showed that this oxyanion has a hydration shell but it is less geometrically determinate than that found for halogen ions.6-8 Also a study on a 2 M H2S04 aqueous solution has been ~ t a r t e d . In ~ this case the absence of cations and therefore of the possible maskings due to them permitted us to study better the problem of sulfate anion hydration. Only one study has been madelo of the H3P04 hydration phenomena but the main object was the study of complex formation between metals (cadmium and nickel) and the phosphate group. Some indications about the presence of H3P04-H20contacts have been obtained. The number of contacts was about four, relatively small in comparison to the number of about eight found for the sulfate group. At the same time the existence of a coordination shell of about 12 for the Po43-has been shown in different studies on solid corn pound^,^^-'^ and also in a recent study of the molecular dy(1) Caminiti, R.; Paschina, G.; Pinna, G.; Magini, M.Chem. Phys. Lett. 1979, 64, 391.

(2) Caminiti, R. 2.Naturforsch., A 1981, 39A, 1062. (3) Caminiti, R. Chem. Phys. Lett. 1982, 86, 214. (4) Caminiti, R. Chem. Phys. Lett. 1982, 88, 103. (5) Radnai, T.; Pilinkis, G.; Caminiti, R. 2.Naturforsch., A 1982, 37A, 1247. (6) Caminiti, R.; Licheri, G.; Piccaluga, G.; Pinna, G.; Magini, M. Reu. Inorg. Chem. 1979, 1 , 333. (7) Caminiti, R.: Cucca, P. Chem. Phys. Lett. 1982, 89, 110. (8) Caminiti, R.; Musinu, A.; Paschina, G.; Pinna, G. J. Appl. Cvstallogr. 1982, 15, 482. (9) Caminiti, R. Chem. Phys. Lett. 1983, 96, 390. (10) Caminiti, R. J . Chem. Phys. 1982, 77, 5682. (11) Schrocder, L. W.; Mathew, M.;Brown, W. E. J . Phys. Chem. 1978, 82, 2335.

TABLE I: Composition of the Solutioos solution [H,PO,] [H,O] d,C g cm-) 1 2

2.00' 0.0384b 5.38' 0.1147b

50.067' 0.9616b 41.506' 0.8857b

: 1

cm-*

1.098

1.7465

1.275

2.6738

'In units of mol/L. 'x in the structural unit [H3P04]x[H20]1-x. 'd is the density. d p is the linear absorption coefficient calculated for Mo Ka radiation.

namicsI4 PO4%was found to be widely hydrated. On the contrary, in the crystal structure of H3P04-0.5H2015the number of H3P04-0contacts (with 0 from other phosphate molecules or from an H 2 0 molecule) was found equal to five. Regarding the parameters of the ionic hydration (particularly the average number of water molecules around the phosphate group) we think that depends on the prevailing species present in the studied solutions. To clarify this aspect of the structural problem we started this first study on two H3P04 concentrated aqueous solutions as part of a comprehensive study on H3P04 solutions in which, with increasing pH, other species like H2PO4-, HP042-, and Po43should be present. In such a way we may also check if in aqueous solutions the number of contacts of these species with water molecules increases passing from H3P04 to HZPO,, HP042-,and PO-: and reaches the number of 12 for the last species, as found in solid phase. This (12) Mathew, M.; Kingsbury, P.; Takagi, S.; Brown, W. E. Acta Crystallogr., Sect. B 1982, 838, 40. (13) Takagi, S.; Mathew, M.; Brown, W. E. Acta Crystallogr., Sect. B 1982, Sect. B38, 44. (14) Lee, W. K.; Prohofsky, E. W. J . Chem. Phys. 1981, 75, 3040. (15) Mighell, A. D.; Smith, J. P.; Brown, W. E. Acta Crystallogr., Secr. B 1969, B25, 776.

0022-3654/85/2089-1457$01.50/00 1985 American Chemical Society