J. Chem. SOC., Faraday Trans. 1, 1982, 78, 1011-1019 Influence of Some Solvents and Solutes on Illuminated Red Mercury(@ Sulphide Electrodes BY R. STEPHEN DAVIDSON,* CHARLES J. WILLSHER AND (IN PART) COLIN L. MORRISON Department of Chemistry, The City University, Northampton Square, London EClV OHB Receiued 20th February, 198 1 A study has been made of the irradiation of red mercury(@ sulphide electrodes in water, acetonitrile, pyridine and methanol containing dissolved tetra-alkylammonium salts. Photocurrents were found to be due to solvent oxidation, iodide oxidation and lattice decomposition, but which of these processes actually operates is determined by the individual solvent, solvation of the iodide ion and adsorption of the tetra-alkylammonium cation on the semiconductor surface.The length of the alkyl chain of the substituted ammonium ion was critical. Instability of the sulphide resulted from adsorption of tetra-alkylammonium cations in methanol and acetonitrile. In the case of pyridine, cation adsorption was not significant, but the presence of iodide induced a negative shift of the sulphide band edges. Exploration of semiconductor-electrolyte systems to harness solar energy is well d0cumented.l Much attention has been paid to the performance of electrodes in aqueous electrolyte solutions since water is a cheap, abundant solvent. A further attractive feature of water is that, if photoelectrolysis occurs, hydrogen (an ideal fuel) is produced, and this is of great economic interest.2 If progress is to be made in developing a successful cell then an understanding of semiconductor-electrolyte interactions is necessary.The choice of electrolyte can be important in determining the stability of a system, e.g. cadmium sulphide undergoes photocorrosion in aqueous solutions of many electrolytes but is stable in sulphide + polysulphide solution^.^ We have previously shown that the performance of mercury@) sulphide electrodes is highly dependent upon the type of electr~lyte.~ Use of solvents other than water has also been investigated. Thus both n-type molybdenum selenide5 and sulphide6 are photostable in non-aqueous systems such as acetonitrile and ethanol. Cadmium sulphide is also photostable in acetonitrile solutions containing sodium iodide' and will also reduce heptyl viologen in acetonitrile.* Valuable information concerning semiconductor-band edges, flatband potentials, surface states, stability and redox reactions of solution species has been accumulated in many of the reported studies.Semiconductor-solution ' energetics' (and some kinetics) have been evaluated, and the solvent is considered an inert medium (with the exception of alcohols, where solvent oxidation can occurQ~ lo) for dispersing interesting electroactive species, and a supporting electrolyte is included to assist conductivity. This electrolyte is deemed inert and is usually either a lithium or tetra-n-alkylammonium perchlorate, nitrate or other non-oxidisable anion. This paper describes an investigation of the effect of solvent and the structure of tetra-n-alkylammonium ions as supporting electrolyte upon the performance of mercury(r1) sulphide electrodes.In earlier work we found1' that irradiation of mercury(I1) sulphide electrodes in aqueous potassium nitrate solutions containing methyl viologen led to reduction of methyl viologen at the counter-electrode. The 101 11012 I L L u MI N A TED RED ME R c u R Y(II) s u L PHI D E E L E c T R o D E s characteristic blue colour of the viologen was clearly observed. However, the efficiency of the reaction (as measured by the photocurrent and the extent of reduced material) decreased as the irradiation was continued. When the mercury(I1) sulphide electrode was removed from solution, washed and then re-irradiated in a cell containing aqueous potassium nitrate (but no viologen), very little photoreactivity was observed.There were similar results when non-aqueous solvents, e.g. acetonitrile, were employed. We concluded that the viologen had poisoned the surface of the sulphide. Since the viologen is an ammonium cation it seemed worthwhile examining whether other systems containing ammonium cations would poison the material. EXPERIMENTAL ELECTRODE PREPARATION A platinum-mesh electrodeL0 was coated with mercury(1r) sulphide in the following way. Water (25 cm3, de-ionised) was de-oxygenated by purging with nitrogen for 30 min. The sulphide (2 g) was added and the mixture boiled and then sonicated (Dawe Instruments ultrasonic probe type 7530A). The electrode was dipped into the suspension produced by this procedure, taken out and then dried in a stream of hot air.The ‘dipping and drying’ procedure was repeated three to four times to obtain an even red coating (average weight 0.025 g) on the mesh. The electrode was stored in a solution of the electrolyte for 12 h to effect equilibration. IRRADIATION PROCEDURE The lamp source was a 1.6 kW xenon lamp, the output of which was focused onto the mesh by means of a concave mirror and appropriate lenses. The light was filtered through an aqueous copper(n) chloride (0.3 mol dm-3) solution which has a maximal transparency at 515 nm with a half-bandwidth of 90nm. The light intensity at the electrode surface was measured as 2.5+ W cm-2 (Macom light meter). MEASUREMENTS The cell was of conventional design accommodating the mercury sulphide electrode, a platinum counter-electrode and a reference electrode (sealed saturated calomel electrode, Electronic Instruments Ltd).The cell held 100 cm3 of electrolyte. Potentials were measured with a Phillips high-impedance voltmeter P.M. 2431 1 (internal resistance = lo6 - lo7 a). Potentials were applied to electrodes and currents measured using a Wenking LB 75L potentiostat. MATERIALS Water was distilled prior to use. Methanol, acetonitrile and pyridine were purchased with the lowest water assay (0.05 %) and used as received. All salts were of the highest possible purity, recrystallised from the appropriate solvent if necessary and stored under vacuum. All potentials were measured against a saturated calomel electrode (SCE) in both aqueous and non-aqueous solvents, and a platinum mesh, identical to the sulphide support mesh, formed the counter- electrode.Purging with dry nitrogen was employed to free the electrolyte of oxygen and was bubbled, very slowly, through the electrolytes during the experiments. Electrolyte agitation was effected by increasing the bubbling rate. Routine atomic absorption techniques were used to analyse the electrolytes for solubilised mercury. RESULTS AND DISCUSSION We have previously reported that irradiation of mercury(I1) sulphide in aqueous iodide solutions leads to dissolution and a colour change from red to l2 The darkened material has been identified as meta-cinnabar.13 We have now examined samples, darkened under various conditions by EDAX spectroscopy* and found that * We are grateful to Dr M. Phillips of the Physics Department, The City University, for carrying out the measurements.R.S. DAVIDSON, C . J. WILLSHER A N D C. L. MORRISON 1013 the colour change is not attended by a change in stoichiometry. Furthermore, as the amount of meta-cinnabar in the cinnabar increases so the conductance increases until that of meta-cinnabar is attained. This explains why the photocurrents of mercury(1r) sulphide electrodes rise on blackening.12 Since meta-cinnabar gives a very small photoeffect, there is an optimum level for the meta-cinnabar which produces the best electrodes. The extent of mercury solubilisation and colour change of the electrode is determined by the degree of interaction of the iodide with the sulphide.* The effect of added alkylammonium iodides upon the performance of mercury(I1) sulphide electrodes was therefore studied since the solubilisation and colour-change processes should give an indication of the effectiveness of the interaction of anions with the sulphide surface under these conditions.TABLE 1 .-PHOTOEFFECTS GENERATED BY RED MERCURY(II) SULPHIDE IN SOLUTIONS OF TETRA-ALKYLAMMONIUM SALTS IN WATER relative relative photocurrent/pA at indicated photo- applied potential/V us. SCEb dissolved potential/ salta mV - 0.6 -0.4 -0.2 0.0 +0.2 remarks ~~ ~~ ~~ NH,I - 30 -3.0 -2.3 +3.2 +6.8 +31.0 “C,Hd,I - 60 d C d - 100 -5.0 -3.0 + 11.0 +20.0 e C +3.0 +17.0 “CH3141 -0.1 +O.l +0.2 +2.0 N(C,H,),I - 75 - 1.1 “C4H,),I - 75 0.0 +0.1 f0.2 +0.5 +0.4 “C6 H 13141 - 50 -0.1 0.0 0.0 +0.1 + 0.4 N(C,Hd,NO, - 90 0.0 +0.4 +0.7 +2.4 +4.O C e f g f B a Salt concentration is 0.05 mol dm-3 and the substituent alkyl chains are primary.Potentiostatically controlled. Large (> 50pA) dark current. HgS blackened and HgS blackened, no solubilisation. f HgS remained red, underwent slight solubilisation. no solubilisation. HgS blackened very slightly, no solubilisation. The photoelectrochemical behaviour of mercury(I1) sulphide electrodes in water containing various tetra-n-alkylammonium salts is shown in table 1 . The results clearly show that, as the length of the alkyl chain on the ammonium cation is increased, the photocurrents decrease. Furthermore, the extent of photochemical and/or electro- chemical reaction decreases since mercury solubilisation decreases with increase in chain length and, in particular, the efficiency of blackening.We attribute the dependence of the efficiency of the photoinduced processes upon chain length as being due to the cations being adsorbed on the surface of the sulphide. As the chain length of the cation is increased, the more the approach of the counterion (e.g. -1 or OH) will be hindered, with a consequent decrease in photoreactivity. The rather low photopotentials observed in these systems (a value of ca. - 200 mV is observed in aqueous potassium nitrate) indicate that the surface of the sulphide (and probably the Helmholtz layer) * A referee has suggested that the formation of meta-cinnabar could be accounted for by the iodide solubilising the mercury as HgIz-, so allowing the concentration of S2- ions to build up in the semiconductor.Reaction of the S2- with solubilised mercury would result in the precipitation of the black form of mercury(r1) sulphide.1014 I L L UM I N A TE D R ED M E R c u R Y(II) s u L PHI D E E L E c T R ODE s TABLE 2.-hOTOEFFECTS GENERATED BY RED MERCURY(I1) SULPHIDE IN SOLUTIONS OF TETRA-ALKYLAMMONIUM SALTS IN ACETONITRILE relative relative photocurrent/pA at indicated photo- applied potential/V us. SCEb dissolved potential/ salta mV -0.6 -0.4 - 0 . 2 0.0 + 0 . 2 remarks NH,I -140 -7.0 - 3 . 2 -2.9 +3.0 -210 -2.0 0.0 + 3 . 2 +6.0 -140 -3.6 -0.4 +2.4 +8.0 +l.O +1.2 +7.5 - 170 +4.0 +9.5 - 220 0.0 + 1.6 +6.2 C d c e C f C 9 C C g N(CH,),I N(C,H,),I N(C3H441 H(C,H,),I C c C N(C6H13)41 - 145 C 9 N(C,H,),NO, -100 -0.3 0.0 +0.1 +0.4 +0.5 g a-c As table 1. HgS remained red, extensive mercury solubilisation.HgS blackened and underwent extensive solubilisation. f HgS blackened slightly and underwent extensive solubilisation. g No blackening took place, but extensive mercury solubilisation was noted. have been modified by the cations so as to either shift the sulphide band edges to more positive potentials or alternatively decrease the extent of band-bending. The results for acetonitrile as solvent are shown in table 2. With this solvent mercury solubilisation is extensive and increases as the chain length increases (fig. 1). Furthermore, the shapes of the photocurrent against time plots vary with the length of the alkyl chain (fig. 2). Very small photocurrents were observed with tetra- n-butylammonium nitrate.This probably reflects the fact that there is not a suitable redox couple present in solution. As a consequence of this the sulphide undergoes decomposition by reaction of the positive holes with the sulphide-producing Hg2+ ions. Decomposition via this route is thermodynamically feasible.4* l4 The decomposition that is observed with the long-chain alkylammonium cations is interpreted as being caused by the adsorption of the cations on the surface hindering reaction with the iodide ions, thereby favouring the decomposition reactions. The shorter-chain compounds are less well adsorbed, thereby allowing the interaction of the iodide ions with the sulphide. This interaction will account for some of the solubilisation, and apparently reprecipitation of the mercury as meta-cinnabar occurs since blackening is observed.The results for pyridine as solvent are shown in table 3. All the iodides behave in a similar manner. The small photocurrents for the nitrate are probably due to the lack of a suitable redox couple. Even under these conditions mercury solubilisation does not occur, whereas in acetonitrite extensive mercury solubilisation was observed. The stabilising influence of pyridine has been previously noted.15 The finding that most of the iodides behave in a similar way suggests that cation adsorption on the surface is not particularly important. This is probably due to the fact that pyridine is strongly adsorbed on the surface. This apparently still allows the iodide ions to react with the sulphide, giving rise to photocurrents as a result of iodide ion oxidation and sulphide solubilisation.The results for methanol are shown in table 4. The surprising feature of using this solvent is that positive photopotentials are observed. With the tetra-n-butylammonium nitrate a positive photopotential is recorded. It appears that the methanol isR. S. DAVIDSON, C. J . WILLSHER A N D C. L. MORRISON light off / t X 0 0 X X X 0 1015 number of carbon atoms in alkyl chain FIG. 1 .-Solubilised mercury as a function of chain length in tetra-alkylammonium iodides dissolved in acetonitrile ( x ) and methanol (0). w 5 min ---- 2 2 p A L -- -9.OpA I -- -30.OpA (i) (ii) (iii) FIG. 2.--Current against time plots for mercury(I1) sulphide held at -0.4 V us. SCE in acetonitrile containing (i) ammonium iodide, (ii) tetrapropylammonium iodide and (iii) tetra-n-hexylammonium iodide.1016 ILLUMINATED RED MERCURY(@ SULPHIDE ELECTRODES TABLE 3 .-PHOTOEFFECTS GENERATED BY RED MERCURY (11) SULPHIDE IN SOLUTIONS OF TETRA-ALKYLAMMONIUM SALTS IN PYRIDINE relative photo- applied potential/V us.SCEb relative p ho tocurren t /PA at indicated dissolved potential/ sal ta mV -0.6 - 0 . 4 -0.2 0.0 +0.2 remarks - 1.6 +3.8 +4.2 NHJ - 250 ? C d N(CH3)J N(C2H5)J e e C d C d - - - - - - - - - - - - N(C,H,)*I -275 +0.8 +1.9 +3.5 +7.0 N(C4He)J -230 + 1.3 +2.2 +3.6 +6.0 N(C6H 1314’ - 265 c d N(C,H,),NO3 -155 -0.8 -0.2 +0.2 +1.4 +1.9 0.0 +0.1 +0.6 +6.0 f a-c As table 1. Solubilisation occurred. Salts were sparingly soluble. f No blackening or solubilisation occurred. TABLE 4.-PHOTOEFFECTS GENERATED BY RED MERCURY(I1) SULPHIDE IN SOLUTIONS OF TETRA-ALKYLAMMONIUM SALTS IN METHANOL relative relative photocurrent/pA at indicated photo- applied potential/V us.SCEb dissolved potential/ salta mV -0.6 - 0 . 4 -0.2 0.0 +0.2 remarks C C NH,I + 180 -38.0 -5.0 -1.0 C d N(CH3)J + 60 -54.0 - 1.4 +3.0 C d N(C2H5)J +60 -27.0 -6.5 -14.0 -11.5 C d N(C,H,)J +90 -22.0 -5.5 -5.5 -5.0 C d + 70 -5.0 -1.5 0.0 + 4 . 0 C e + 30 -8.0 -1.6 +2.0 0.0 C f f N(C4He)J N(CI3H 13141 N(CIH9)4NO, +40 -5.0 -4.5 -2.6 +4.0 +3.5 a-c As table 1. Extensive blackening and mercury solubilisation. Partial blackening and considerable solubilisation of mercury. f Very slight blackening and considerable solubilisation of mercury. reacting with the sulphide to produce a species which is part of a redox couple that can undergo electrochemical reaction at the uncovered platinum surface of the electrode.This type of behaviour has been observed for electrodes fabricated from platinum mesh and powdered titanium dioxide, in the presence of reducing ions.16 That this typk of process is occurring is shown by the photoeffect against time plots depicted in fig. 3. Part (i) shows the photovoltage plot. At point A, the sulphide is fresh and red, and initial illumination commences. The shape of the response A-B is typical of the n-type behaviour of the sulphide,12 i.e. the voltage shifts negatively on illumination. Switching off the light at point B (at this stage the electrode is now black in colour) does not cause the e.m.f. to return to the initial dark reading, but it adopts a more negative value and point C is reached.Re-illumination produces a positive photovoltage shift; to attain the voltage observed at B, point D is reachedR. S. DAVIDSON, C. J. WILLSHER AND C. L. MORRISON 1017 (i) - - -220mV - light on light off light off - - -340 mV -- - 400mV E - 20 min Smin -- 16.0 PA --• 9.5pA --• 7 . O V A t++---+ Smin 10min FIG. 3.-Plots in methanol+ tetraethylammonium iodide for (i) voltage against time and (ii) current at 0.0 V us. SCE against time. on the voltage against time plot. The decline D-E on ending irradiation is analogous to B-C. This type of behaviour is typical of a blackened sulphide electrode in the presence of a reducing ion.* Part (ii) of fig. 3 shows a typical photocurrent against time plot.The most outstanding feature is the large anodic overshoot seen on ending the irradiation. The time taken for the decline of the overshoot is quite long, but it can be accelerated by vigorous stirring of the electrolyte. Since a similar effect is observed with tetra-n-butylammonium nitrate as electrolyte it is likely that the anodic currents and positive photo-e.m.f. are due to material produced by oxidation of the solvent. When iodides are used, iodine may also contribute to these effects. However, the reaction with methanol appears to dominate the reactions in all the alkylammonium iodides, with perhaps the tetra-n-hexylammonium salt being an exception. The solubilisation of the mercury may be due to the inherent instability of the1018 I L L U MI N A TE D R E D ME R C U R Y(I1) S U L P H I D E EL E C T R 0 D ES TABLE SUMMARY OF PROPOSED ELECTROLYTE EFFECTS AND REACTIONS ~ ~ ~~ reactions probably occurring at the illuminated sulphide solvent electrodea effects of the solute and solvent water 2H,O + 4h+ -+ 0, + 4H+b 21- + 2h+ + IZc HgS + 41- + 2h+ -+ HgI,,- + S HgS + 21-+ I, -+ HgI,,-+ S (i) Cation adsorption on the (ii) Water/iodide repelled by cations (iii) Water prevents anodic (iv) Solvation of iodide by water sulp hide .8 with long alkyl chains.8 decomposition." necessary for b1ackening.f acetonitrile 21- + 2h+ + (i) Cation adsorption on the sulphide! (ii) Acetonitrile/iodide hindered by cations with long alkyl chains.g (iii) Acetonitrile fails to prevent anodic decomposition.e (iv) Acetonitrile solvation necessary for b1ackeni'ng.f (v) Mercury solubilisation enhanced with longer alkyl chains.8 (i) Cation adsorption is unimportant.(ii) Pyridine solvation of iodide (iii) Pyridine prevents anodic decomposition.e (iv) Interaction of pyridine-solvated iodide with the sulphide causes a negative shift in the band edges. of the sulphide.8 HgS + 41- + 2h+ + HgI,,- + S 1 HgS + 21-+ I, -+ HgI,,-+ S J HgS + 2h+ + Hg2+ + Se pyridine 21- + 2h+ -+ I,c HgS + 41- + 2h+ -+ HgI,,- + S HgS + 21- + I, -+ HgI,,- + S 1' causes no blackening! methanol 21-+ 2h+ + I, (i) Cation adsorption on the surface (ii) The solvent is a blackening agent.f (iii) Methanol solvation of iodide necessary for b1ackening.f (iv) Mercury solubilisation enhanced with longer alkyl chains. (v) Methanol/iodide hindered by cations with long alkyl chains.g (vi) Other, undetermined effects.HgS + 41- + 2h+ -, HgI,,- + S HgS + 21- + I, + HgI,,- + S J other unknown decomposition reactions including HgS + 2hf + Hg2+ + S" a h+ is a light-generated positive hole in the semiconductor; water oxidation; iodide oxidation; iodide-induced solubilisation; anodic decomposition (in and the fate of elemental sulphur is unknown: it may be oxidised in water and solubilised in the organic solvents); f the blackening is discussed elsewhere;l5 ~7 the longer the alkyl chain, the greater the effect.R. S. D A V I D S O N , C. J. WILLSHER A N D C. L. MORRISON 1019 sulphide in this solvent or due to its reaction with the solvent. At present we cannot ascertain the extent to which each participates.However, the extensive solubilisation observed with the tetra-n-hexylammonium compound suggests that mercury sulphide is inherently unstable in this solvent. The various chemical reactions that can occur in the different solvents are summarised in table 5. Presumably the extents of mercury solubilisation and blackening are intimately connected since the reprecipitation of the mercury has to compete with the solubilised mercury escaping from matrix of the sulphide, and the efficiency of this process will depend upon the nature of the sulphide surface and the solvent. In our earlier work4 we have attempted to determine the flatband potential of the sulphide in aqueous solution. Inspection of tables 1-3 shows that the flatband potential apparently varies with solvent and with the alkylammonium cation.This affects the photostability of the electrode, e.g. no mercury solubilisation in water and extensive solubilisation in acetonitrile. Examples are known where a change of solvent has enabled the construction of stable photo electrode^.^? However, the results described show the importance of the supporting electrolyte. In the case of iodide ions in pyridine it is shifted to negative values. Furthermore, the chemical modification of semiconductors, e.g. by attaching silanes covalently,17 may lead to impairment of the reaction due to hindering the combination of photogenerated holes with reducing ions. We thank the N.R.D.C. for a fellowship to C . J. W. and the S.R.C. for a research assistantship to C. L. M. A. J. Nozic, Annu.Rev. Phys. Chem., 1978, 29, 189; K. Rajeshwar, P. Singh and J. Du Bow, Electrochim. Acta, 1978,23,1117; M. Tomkiewicz and H. Fay, Appl. Phys., 1979,18,1; M. A. Butler and D. S. Ginley, J. Muter. Sci., 1980, 15, 1 ; R. Memming, Electrochim. Acta, 1980,25,77; A. Heller and B. Miller, Electrochim. Acta, 1980, 25, 29. H. P. Maruska and A. K. Ghosh, Solar Energy, 1978, 20, 443. A. B. Ellis, S. W. Kaiser and M. S. Wrighton, J. Am. Chem. Soc., 1976, 98, 1635. R. S. Davidson and C. J. Willsher, J. Chem. Soc., Faraday Trans. I , 1980, 76, 2587. L. F. Scheenmeyer and M. S. Wrighton, J. Am. Chem. Soc., 1980, 102, 6964. L. F. Scheenmeyer and M. S. Wrighton, J. Am. Chem. Soc., 1980, 101, 6496. K. Nakatani, S. Matsudaira and H. Tsubomura, J. Electrochem. Soc., 1978, 125, 406. F. D. Saeva, G. R. Olin and J. R. Harbour, J. Chem. Soc., Chem. Commun., 1980,401. M. Miyake, H. Yoneyama and H. Tamura, Electrochim. Acta, 1976, 21, 1065. M. Miyake, H. Yoneyama and H. Tamura, Chem Lett., 1976,633. lo R. S. Davidson, R. R. Meek and R. M. Slater, J. Chem. Soc., Faraday Trans. I , 1979, 75, 2526. l1 R. S. Davidson and C. J. Willsher, unpublished results. l2 R. S. Davidson and C. J. Willsher, Nature (London), 1979, 278, 238. l3 R. S. Davidson and C. J. Willsher, J. Chem. Soc., Dalton Trans., 1981, 833. We apologise for the fact that we failed to reference the important paper by H.A. Kagi, Y. Fujita and E. Takabatabe (Photochem. Photobiol., 1977, 26, 373) in which some photochemistry of mercury(r1) sulphide is described. R. S. Davidson and C. J. Willsher, Faraday Discuss. Chem. Soc., 1980, 70, 177. 1979, 75, 2517. M. Tomkiewicz, J. Electrochem. Soc., 1980, 127, 1518. l4 C. J. Willsher, Ph.D. Thesis (University of Leicester, 1980). l6 H. H. Chambers, R. S. Davidson, R. R. Meek and R. M. Slater, J. Chem. SOC., Faraday Trans. I , (PAPER 1/296)