J O U R N A L O F C H E M I S T R Y Materials Feature Article Developing an understanding of the processes controlling the chemical bath deposition of ZnS and CdS Paul O’Brien* and John McAleese Department of Chemistry, Imperial College of Science Technology and Medicine, South Kensington, London, UK SW7 2AY. E-mail: p.obrien@ic.ac.uk Received 24th June 1998, Accepted 3rd August 1998 The deposition of cadmium sulfide by chemical bath area which is well documented.10 Applications in optoelecmethods is straightforward and involves an alkaline solu- tronic or photovoltaic devices is another area receiving attention of a cadmium salt, a complexant and a chalcogen tion, see for example Squeiros et al.,11 but has yet to be fully source, often thiourea or thioacetamide.Supersaturation developed.In CdS based solar cells, the use of wider bandgap of the bath with respect to ‘Cd(OH)2’ is necessary for the materials such as ZnS or CdxZn1-xS could lead to decreases deposition of good quality films under a wide range of in window absorption losses and improvements in the short conditions. In contrast zinc sulfide is more difficult to circuit current of the cells. Saitoh et al.have discussed the use deposit. In this paper we discuss the literature concerning of ZnS in passive structures such as optical waveguides,12 the deposition of these chalcogenides and use equilibrium while Landis et al.13 have described the index-matching of models to rationalise many of the observations found in wide-bandgap epitaxial heterojunction windows including ZnS the literature.Strategies for the deposition of high quality for silicon based solar cells. On mercury cadmium telluride films of ZnS by CBD are discussed. (HgCdTe), ZnS films can be used as gate dielectric layers in infrared photodetectors for CCDs14 and hence have a Introduction potentially wide range of applications. There is considerable interest at present in the soft processing of materials.1 Zinc and cadmium sulfide are compound The CBD process semiconductors with a wide range of potential applications.Film deposition These materials have many similarities, both exist in cubic or hexagonal forms and are wide-, direct-bandgap In chemical bath deposition experiments, solid material is semiconductors. formed which means that the bath must be thermodynamically The chemical bath deposition (CBD) process uses a unstable with respect to precipitation of the solid phase controlled chemical reaction to eVect the deposition of a thin formed, i.e.supersaturated. There are two parent possible film by precipitation. In a typical experiment substrates are reactions leading to solid material, notably: (i) within the bulk immersed in an alkaline solution containing the chalcogenide of the solution (homogeneous precipitation); (ii) at a surface, source, the metal ion and added base.A chelating agent is the substrate or adventitious reaction on the reaction vessel also added to control the hydrolysis of the metal ion. The surface (heterogeneous precipitation). process relies on the slow release of S2- ions into an alkaline It is the second of these routes which leads to film formation.solution in which the free metal ion is buVered at a low There is a tendency in the literature to discuss the deposition concentration. The speciation of metal ions, in particular the of thin films in terms of two distinct mechanisms or models. free metal ion concentration, is controlled by the formation The models form end points in a complex series of potential of complex species, e.g.[Zn(NH3)6]2+. The supply of sulfide processes for the deposition of adherent films encompassing ions is controlled by the decomposition of an organic sulfur many diVerent possibilities (Fig. 1). The first of these is the containing compound, usually thiourea or thioacetamide. The solubility product of the compound in question helps maintain the stoichiometry of the deposited material, homogeneous compounds can be obtained as a result.A large number of physicochemical factors control the growth of the deposit under a specific set of reaction conditions. The supersaturation with respect to an individual phase is important as well as the kinetics of the growth processes.The technique has been used extensively to grow CdS and this is reflected in the numerous papers on the subject.2–8 CBD is the production method for the CdS component of the ‘Apollo’ (CdS/CdTe) series solar cells manufactured by BP Solar. The deposition of ZnS by CBD is a more diYcult proposition than that of CdS. In particular it is evident that there is a much wider range of conditions in which the concurrent deposition of zinc sulfide and oxide can occur.It would be useful to be able to deposit ZnS by CBD. There is a diverse range of applications for thin films of this semiconductor including as waveguides, heterojunction devices and in thin-film electroluminescent displays in which it is the most commonly used host material.9 The potential of ZnS layers in Fig. 1 Schematic representations of the processes which could lead to a thin film: ion-by-ion, cluster-by-cluster and mixed. blue light-emitting diodes (LEDs) and laser diodes is also an J. Mater. Chem., 1998, 8(11), 2309–2314 2309so-called ion-by-ion process in which the ions condense at the reacting surface to form the film. The second is termed the cluster-by-cluster process in which agglomeration of colloidal particles pre-formed in solution, by the homogeneous reaction, leads by absorption at the surface to particulate films.In practice both processes may occur and or interact, leading to films in which colloidal material is included in the growing films. The predominance of one mechanism over another is governed by the extent of heterogeneous and homogeneous nucleation.Key factors include the degree of supersaturation of the solution and the catalytic activity of the surface (substrate). 15 As indicated in (Fig. 1) there are several chemical reactions which can potentially be involved in each of these Fig. 2 Graph of growth rate vs. [ligand ]. processes. The mechanisms of CBD processes are really quite poorly The situation is diVerent for ZnS and Dona and Herrero21 understood and this is reflected in the literature.The actual have suggested that as the growth rate for ZnS appeared be processes leading to the formation of good quality adherent independent of stirring this allows the possibility of diVusion films are complex. The CdS system illustrates the point. Good control to be discarded.They found that graphs of growth quality (adherent) films are only obtained from baths which rate against [NH3] or [N2 H4] had similar profiles, see Fig. 2. are supersaturated with respect to the precipitation of cadmium There is an acceleration of growth rate with concentration up hydroxy species, irrespective of the substrate used. As the to an optimum concentration of the ligand (at molar concensupersaturation of Cd(OH)n species is a very important factor trations, 1.5–2.0 nm min-1) and then the rate rapidly declines.in the formation of adherent films, the processes that might Similar relationships are observed for both complexing agents. be important include: (i) the formation of clusters ‘CdxOHy’; Decreases in ligand concentration resulted in increases in the (ii) the absorption of these ‘nuclei’ at the surface; (iii) the free zinc metal ion concentration and therefore an enhancemetathetical reactions of surface bound ‘nuclei’ with sulfide ment of homogeneous precipitation.Increases in concentration ions or thiourea in a heterogeneous process to form good of either complexing ligand must reduce the amount of free quality CdS; and (iv) the formation and reactivity of pendant zinc ion in solution and therefore restrict the growth rates of OH groups on the surface.both homogeneous and heterogeneous precipitation. In early work, Kitaev et al. postulated16 that the presence Mokili et al.22 have hence proposed that the formation of of hydroxide particles in solution was necessary for the growth Zn(OH)2 must be minimised in order to obtain ZnS due to of good quality films, the decomposition of thiourea being possible competition between the formation of sulfide and stimulated by a solid phase such as cadmium hydroxide.This hydroxide in basic solutions [Zn(OH)2, Ksp=10-15.3 and ZnS, proposal was supported by the observations of Kaur et al.17 Ksp=10-23.8]. in that adherent films were only prepared in the presence of Froment and Lincot18 found ZnS films to be diVerent from ‘Cd(OH)2’ in solution.The importance of ‘Cd(OH)2’ species those of CdS and to be an aggregation of spherical particles, in the growth mechanism of CdS even in conditions where no in a more or less close packed structure. The distance between observable macroscopic precipitate is present is clear.Froment reticular planes revealed that the films were composed mainly and Lincot18 suggested that the mechanism of film formation, of cubic ZnS. Electron diVraction patterns are similar to those similar to that proposed by Parfitt,19 could be represented by for CdS colloids. Evidence like this indicates that the growth the following series of consecutive surface adsorption/reaction mechanism for ZnS is probably a colloid aggregation process.steps: Similar findings have been reported by Mokili et al.,22 TEM results revealing ZnS films composed of aggregated grains in Cd2+aq+site+2 OH-�Cd(OH)2,ads (1) an amorphous matrix. (NH2)2CS+Cd(OH)2,ads�C* (2) In contrast, Dona and Herrero21 suggest that the deposition process has its basis in the slow release of Zn2+ and S2- ions C*�CdS+site (3) in solution.These ions, it is suggested, condense on the where C* is a reaction intermediate. substrates. This idea suggests an ion-by-ion process, opposing The supersaturation limit for the formation of ‘Cd(OH)2’ the statement of others. Contradictory statements such as may be only coincidentaly signigficant as the pKa of surface these have prompted us to form the opinion that no proper, bound Cd2+ could be similar to that of the ion in solution.fundamental understanding of the process exists to date. The formation of surface bound hydroxy species could be the Table 1 summarises some typical chemical bath conditions important step which could serendipitously occur at a similar used in ZnS deposition. value of pH to the formation of a precipitate of the hydroxide.Various other sequences of these reactions could lead to Reaction kinetics adherent CdS. It is interesting to note that freshly precipitated Cd(OH)2 is readily metathesised to CdS in the presence of The kinetics of typical CBD processes (Fig. 3) appear to follow a sigmoidal profile similar to those observed for thiourea; this reaction appears not to happen with the zinc analogue.The solution species formed in the case of zinc are autocatalytic reactions.23,24 In solid state nucleation and growth processes, such evidently diVerent as is the degree of supersaturation. CdS has been reported to grow epitaxially on single crystals reactions are often described by a formal kinetic expression such as the Avrami equation: by CBD such as InP.20 This observation supports an ion-byion mechanism because of the register required between the a=1-e(-kt) (4) substrate and growing layers at the atomic level.Further evidence comes from an HRTEM18 study of material deposited where a is the fractional decomposition, t the time and k a rate constant. This is an expression of the rate of formation before coalescence of the CdS film. Monocrystalline nuclei (30 nm) are formed at the substrate surface with the c-axis of nuclei in the special case when there is completely random nucleation.25 If we consider such processes in a qualitative perpendicular to the surface. These crystallites coalesced to form a continuous film composed of microcrystallites (hexag- way the kinetics are dominated by three phases: (i) initiation, often nucleation, the initial step usually requiring a high onal, 20–80 nm). 2310 J. Mater. Chem., 1998, 8(11), 2309–2314Table 1 Some typical chemical bath conditions reported to have deposited ZnS Counter ion, Ligand 1, Ligand 2, Added base, Sulfiding agent, T / °C, Band gap/ Phase Ref. conc./M conc./M conc./M conc./M conc./M growth eV reported rate/A° min-1 sulfate, 0.14 NH3, 3.75 HZ, 0.508 Na2B4O7, 0.03 (NH2)2CS, 0.50 80, 10 34 90, 17 chloride, 0.20 NH4+/NH3, 2.00 TAA, 0.04 3.60 cubic 15 n-type chloride, 0.02 TAA, 0.02 4.10 cubic 15 >3.7 sulfate, 0.025 NH3, 1.00 HZ, 3.00 (NH2)2CS, 0.035 3.76 cubic 16 sulfate, 0.05 TEA, 0.20 2.2 ml pH 10 TAA, 0.04 17 NH3/NH4Cl sulfate, 0.05 TEA, 0.225 TAA, 0.02 17 19 sulfate, 0.025 NH3, 1.00 HZ, 3.00 (NH2)2CS, 0.035 3.76 cubic 21 acetate, HZ, 0.50–3.00 NH4+, 0.02 TAA, 0.001–0.1 <16.7 3.60–4.00 18 0.001–0.01 chloride, HZ, 1.10 (NH2)2CS, 25, 1.5 3.85 22 0.075–0.190 0.100–0.150 chloride, 0.20 HZ, 4.00 (NH2)2CS, 0.20 16.7 3.70–3.80 42 acetate/sulfate, HZ, 1.1–1.2 HZ, 3.00; (NH2)2CS, 0.035–0.1 25, 0.4 34 0.025–0.050 en, 0.2; TEA, 0.07 HZ, hydrazine (N2H4); TEA, triethanolamine; en, ethylenediamine; TAA, thioacetamide (CH3CSNH2).Glass substrates were used in all cases except ref. 17 where CuInSe2 substrates were also used. cadmium is extremely important in controlling the nature of the films deposited. There appears to be no such simple relationship for zinc sulfide deposition. The degrees of supersaturation in any system will depend on various factors including pH, ligand concentration and temperature.It is interesting to compare the speciation in similar chemical baths for cadmium and zinc. If we take the ethylenediamine system as an example, a bath modestly supersaturated with respect to cadmium hydroxide is readily obtained at a 251 ligand to metal ratio ([Cd]=180 mmol dm-3, [en]= 360 mmol dm-3), Fig. 4(a) pH> ca. 9.5. In contrast a similar Fig. 3 Kinetic profile for a generalized autocatalytic process, fraction of reaction (a) vs. time. activation energy in which reactive centres which catalyse the reaction are formed; (ii) the main phase of the reaction, sometimes zero order in which a number of reactive sites gives rise to a lower activation energy pathway, autocatalysis in a general sense, but in a heterogeneous reaction the growth of nuclei; and (iii) a termination step in which the reagent becomes depleted and the reaction begins to slow and eventually stops.Intuitively this kind of general model is attractive for CBD processes and one thing which is interesting is that many CBD processes show a substantial linear portion in their kinetic profile (see for example ref. 18). This observation may suggest that in this phase of the reaction the process is controlled by a constant number of saturated reaction sites. There is an analogy here with the growth of thin films by MOCVD which is often arranged to be in a diVusion limited regime. Equilibrium considerations The metal ions zinc and cadmium are both labile, so it is reasonable to assume that in stirred solutions equilibria will be rapidly established, hence equilibrium models are useful in assessing the starting points of these chemically reactive baths.In any reaction which involves precipitation, the state of supersaturation of the system will be crucial and equilibrium Fig. 4 Speciation diagram to illustrate the fundamental diVerences calculations help to assess its extent.It is known in the between cadmium and zinc complexation: (a) cadmium, (b) zinc. In deposition of CdS that the supersaturation of the reacting each case [M]=180 mmol dm-3 and [en]=100 mmol dm-3 (equilibrium constant values are at 50 °C and from references 8 and 26). chemical bath with respect to hydroxy/oxy complexes of J. Mater. Chem., 1998, 8(11), 2309–2314 2311released. In a typical deposition experiment (90 °C, 1 h), a conformal film ca. 0.6 mm thick was deposited. It would be interesting to look at the possibility for ternary material deposition from such baths which operate at much lower pH values than are typical for CBD. The films deposited by this method had bandgaps close to that expected for ZnS. Nature of the films deposited in ZnS experiments Electronic properties.Another important issue concerns ndgaps of films deposited from ammonia free baths discussed above27 which are close to that expected for bulk ZnS (3.6 eV). Films deposited from ammonia containing baths have been reported to have bandgaps (4.1–3.9 eV) consistently higher than the bulk value,28 Table 1. There has been speculation that this eVect is due to quantum confinement in the very small (grain size 3–6 nm) crystallites of which these films Fig. 5 Limiting solubility lines for ‘M(OH)2’ {[M]/mol dm-3 (for were composed.29 In essence, a 3-D quantum size eVect is supersaturation) vs. pH, M=Cd or Zn} and corresponding supersaturation line ([S-]/mol dm-3 vs. pH) [for illustration pKsp values used suggested to be occurring in which electrons are localised in at 20 °C: ‘Zn(OH)2’, 15.3; ZnS, 23.8; ‘Cd(OH)2’, 14; CdS, 28].individual crystallites of polycrystalline thin films.30 CdSe films Double-headed arrow shows typical region used for CdS deposition, have been grown where bandgap values were apparently up the path A to B to C is explained in the text. to 0.5 eV higher than in single-crystal samples.31 Similar observations relate to PbSe films32 which were found to contain bath containing zinc is unstable with respect to precipitation crystals of varying size depending on the deposition paramof the hydroxide at much lower values of pH>ca. 7.5.In the eters, in particular the nature of the complexing agent, the presence of modest amounts of a sulfiding agent, in the alkaline film thickness and the deposition temperature, and ZnS parregion, the cadmium bath deposits CdS while in the ticulate films (particle size 30–60 A° ).33 Although such eVects corresponding zinc bath ZnO/hydroxide is more easily formed.have been explained by quantum confinement, it is entirely The above calulations draw attention to a fact that we and possible that some of the films are chemically inhomogeneous others have commented on in several papers, i.e.the impor- with a significant quantity of oxide/hydroxide incorporation. tance of the supersaturation of cadmium solutions with respect to hydroxy species as a necessary condition for the growth of good quality films of CdS. It is possible to further develop Film composition. Many of the films reported as ZnS which have not been rigorously characterised are probably at best this approach in a comparative way by considering the solubility limits for metal hydroxy species as a function of pH.At heavily contaminated with zinc oxide or hydroxide. In an RBS study Mokili et al.22 showed that CBD deposits contained any value of pH there is a limiting value of pM (defined by KspOH2 /[OH]2) below which supersaturation will occur.These significant amounts of oxygen. EXAFS confirmed the presence of oxygen in the form of hydroxide and a high oxygen to limits are approximately linear on plots of pM vs. pH, Fig. 5. At pH 10–11, we know good quality CdS would be deposited sulfur ratio was found. The compositions of some films were even reported to be close to Zn(OH)2.In more basic solutions from such a solution. This condition is suYcient to define a minimum sulfide ion concentration at which metathesis to the containing hydrazine, deposits of material close in stoichiometry to ZnS were obtained, but even such samples were not sulfide becomes a preferred reaction (KspS/[M]). It is striking that CdS is much more stable than ZnS and this is again of the pure sulfide and were contaminated with either oxide (ZnO0.5S0.5) or hydroxide [Zn(OH)S0.5].graphically illustrated (Fig. 5). The actual supersaturation in a reactive chemical bath will depend on the sulfide ion concen- On the basis of these studies they proposed a deposition mechanism involving hydroxide intermediates and suggested tration, however the relative positions of the lines for cadmium and zinc will be unchanged. that if the transformation of hydroxide into sulfide is too slow, significant amounts of hydroxide would be present in the final If we assume that as CdS and ZnS are similar materials (structurally), a similar path is in principle possible for their film.A general conclusion is that the presence of a significant quantity of oxide or hydroxide in the ZnS films may be deposition as thin films. We can investigate this idea by defining equivalent conditions for ZnS and CdS in terms of explained by the proximity of the solubility lines of the sulfide and the hydroxide.supersaturation for the sulfide, which are found by following the line B–C on Fig. 5. Later EXAFS studies by Mokili et al.34 showed that thin films prepared by CBD in alkaline ammonia solutions contain- There are several important conclusions that we can draw from this simple calculation: (i) equivalent levels of supersat- ing a zinc salt, thiourea and diVerent amine additives were zinc hydroxosulfide, the oxygen content of which ranged uration for the sulfide are found at pH values ca. 2.5 lower for ZnS than CdS; (ii) if reactions are carried out in a high between 39 and 82%.FTIR studies supported the evidence suggesting that the films consisted mainly of Zn(OH)2 with pH regime more sulfiding agent (by ca. 4 orders of magnitude!) will be needed for the zinc system, even so the degree of little or no ZnO present. A clear absorption at 648 cm-1 was assigned to a ZnMOH stretching mode by KauVman et al.35 supersaturation for zinc with respect to the hydroxide will always be greater than for cadmium with respect to its The ZnMO bond peak position (at 430 cm-1 from a ZnO powder sample) was very weak in most spectra thus rejecting hydroxide (values of pM and pH being equal ); and (iii) at lower pH the rate of hydrolysis of the sulfide source is likely the hypothesis that significant amounts of ZnO were formed in the films.Such compositions were present even without to be lower so again even if a lower pH is used more sulfide source will be needed. post-annealing. A point worth stressing is the similarity of the conditions It is interesting to note that an intuitive attempt to develop this kind of approach has been reported.27 The chemical bath used in the Mokili study34 to those used in the majority of the other studies of ZnS by CBD, Table 1. All of the reported used contained zinc ions and urea, at modestly acid values of pH.The urea slowly hydrolysed in the bath to provide conditions for the deposition of ZnS as detailed in Table 1 are remarkably similar and suggest that there is the distinct hydroxide and thioacetamide hydrolysed as the hydroxide was 2312 J.Mater. Chem., 1998, 8(11), 2309–2314possibility of heavy oxide contamination or incorporation in many of the films reported as ZnS. The chemistry of sulfide delivery In essence, the chemistry involved in the CBD of ZnS is straightforward. Thiourea decomposition will occur in CBD baths essentially as described in the kinetics studies carried out by Marcotrigiano et al.36 into the desulfuration of thiourea in sodium hydroxide.They postulated that in alkaline solutions, thiourea first gives sodium sulfide and cyanamide which is then transformed into cyanamide, amidinourea, guanidine or, at pH 12, almost quantitatively into urea which isomerizes to ammonium cyanate and is finally hydrolysed to ammonium carbonate: (NH2)2CS+OH-ANCNH2+SH-+H2O (5) Fig. 6 Equivalent solutions. (A) An equivalent solution contour plot NCNH2+H2OAONC(NH2)2 (6) for pM=9 ([M]total vs. [en]total) for the cadmium ethylenediamine system at 50 °C (see ref. 8 for relevant constants); (B) [en]total/[M]total ONC(NH2)2ANH4CNO (7) for pM=9, (C) [en]/[M]total=3. (valid for pH&pKa2 of ethylenediamine). OCN-+2 H2OAHCO3-+NH3 (8) The cyanate and carbonate ion are thought to form very to predict for the CdS system the composition of baths which early in the reaction and may be responsible for the hydroxide will produce high quality films at diVerent concentrations or consumption being greater than that corresponding to reaction how the substitution of one ligand with another, e.g. ethylene- (2).It is assumed that the hydroxide consumption equals the diamine for ammonia, can be made in such a way as to SH- formation.It is known that thiourea behaves as a maintain good quality film deposition. A typical plot is shown zwitterion37 which may play a role in the thiourea desulfurin Fig. 6 for the ethylenediamine cadmium system. The contour ation. Recent observations in our laboratory may suggest that defining constant pM is a curve and much higher ratios of under typical CBD conditions, decomposition of thiourea ligand to metal are needed as the overall concentration proceeds only as far as urea. decreases.This idea is consistent with the tendency of complexes to dissociate on dilution. The third line shows the Strategies for growth contour for a constant ligand to metal ratio of 351 which The fact that the CBD of CdS produces highly crystalline, emphasises that this line is diVerent from that defining a conformal, adherent films relatively easily suggests that one constant pM.There will of course be a critical ratio if one of strategy for growing good films of ZnS would be to alter the the variables is held constant as can be seen from Fig. 6 and bath conditions to favour the same mechanism that governs the concept is useful at high concentration, but this idea does CdS deposition. Froment and Lincot18 suggested that by not allow for the method to be generalised. changing the composition of the reacting solutions competition In related work on PbSe Gorer et al.41 appear to support between the processes of homogeneous and heterogeneous the possibility of a mechanism crossover.Experimental connucleation could be altered to favour thin film growth. ditions were demonstrated where either one or both growth Subsequent experimental work by Mokili in conjunction with mechanisms could lead to fine control of the film properties. Froment and Lincot,22 however, has shown that no marked changes with regards to the growth curves occurred for ZnS Conclusions understanding ZnS deposition in when the conditions approached and/or crossed the precipirelation to physico-chemistry tation lines.This behaviour is quite unlike that of the cadmium sulfide systems for which a clear change in the growth habitat The use of a second ligand is observed when supersaturation is achieved.38 They have also suggested that there is little or no induction time associated Many workers report the use of a second ligand in the chemical bath when attempting the deposition of ZnS.Hydrazine is a with ZnS growth which may suggest that growth depends on colloidal material in this case. popular choice in the CBD of ZnS. A typical observation is that of Ortega-Borges et al.27 who reported that the use of Gorer and Hodes39 suggested for CdSe that the presence or absence of hydroxide particulate material in solution may ammonia and thiourea without hydrazine resulted in films which were not homogeneous or adherent.govern the transition between atom-by-atom and colloidal growth. The change in mechanism was apparently indicated In a related observation Dona and Herrero21 have suggested that the addition of hydrazine, ‘although not essential, by a sharp change in crystal size (observed by shifts from blue to red in optical spectra).They suggested a ‘critical ratio’ R(c) improves homogeneity, specularity and growth rate’. Ndukwe also used hydrazine.42 Dona and Herrero have suggested that (concentration of complexing agent, nitrilotriacetate, to cadmium salt) below which Cd(OH)2 was present and above the rate-determining step for the heterogeneous process may involve the dissociation of a Zn2+ML bond.This suggestion which it was not, despite a visible Cd(OH)2 suspension under any condition. This led them to propose that below R(c) the is unlikely as zinc is classically a labile metal ion. Another possibility is that the hydrazine complex ions have a lower co- mechanism is initiated on the Cd(OH)2 colloidal particles adsorbed on the substrate and above R(c), deposition occurs ordination number (and therefore less steric impediment for the approach of the sulfide ion).Hydrazine could potentially directly on the substrate by initial film formation without Cd(OH)2. act as a bridging ligand and perhaps facilitate surface binding.One thing which must be noted is that at a constant pH the However, it is hard to justify the importance of a critical ratio of metal to ligand. We have defined a useful concept of addition of a second ligand can only lead to a lowering of the free metal ion concentration. In the case of zinc this will lower equivalence in terms of a contour defining a constant value of pM on a plot of total metal ion concentration against total supersaturation with respect to either oxide/hydroxide phases or the sulfide.ligand concentration.40 The use of such plots has enabled us J. Mater. Chem., 1998, 8(11), 2309–2314 2313The most detailed investigation into the addition of amines Amersham International Research Fellow and the Sumitomo/STS Professor of Materials Chemistry.has been carried out by Mokili et al.22 They found that, in all cases, the addition of either hydrazine, triethanolamine or ethanolamine to ammonia increased thin film growth rates. References Using hydrazine, a maximum growth rate (increased by a 1 M. Yoshimura, J. Mater. Res., 1998, 13, 795. factor of four) could be achieved at 3 mol dm-3.However, 2 J. M. Dona and J. Herrero, 12th E.C. Photovoltaic Solar Energy they suggest that hydrazine accelerates the hydrolysis of thio- Conf., Harwood Academic Publishers, Amsterdam, 1994, 597. urea. Other amines tended to increase the growth rate but did 3 M. T. S. Nair, P. K. Nair and J. Campos, Thin Solid Films, 1988, not change significantly the speciation in the bath (precipi- 161, 21.tation line, pH, etc.). Hence for hydrazine and triethanolamine 4 D. Lincot and R. Ortega Borges, J. Electrochem. Soc., 1992, 139, 1880. they proposed that these molecules participated in the 5 P. C. Rieke and S. B. Bentjen, Chem. Mater., 1993, 5, 43. decomposition of thiourea. 6 H. Hu and P. K. Nair, J. Cryst. Growth, 1995, 152, 57. Dona and Herrero reached similar conclusions21 and have 7 Y.Hashimoto, N. Kohara, T. Negami, N. Nishitani and T.Wada, speculated that the relationship between complexing agents Sol. Energy Mater. Sol. Cells, 1998, 50, 71. and growth rates are typical of a system which has ‘two 8 P. O’Brien and T. Saeed, J. Cryst. Growth, 1996, 158, 497. competing processes’ namely heterogeneous and homogeneous 9 R. Mach and G. O.Muller, Phys. Status Solidi A, 1982, 69, 11. 10 S. Yamaga, Jpn. J. Appl. Phys., 1991, 30, 437. precipitation. They also consider the heterogeneous process to 11 J. M. Squeiros, R. Machorro and L. E. Regalado, Appl. Opt., be ‘limited by complex ion adsorbability’ and by ‘the adsorp- 1988, 27, 2549. tion points on the substrate’. In considering the possibility 12 T. Saitoh, T.Yokogawa and T. Narusawa, Jpn. J. Appl. Phys., that the eVect of additional amines is primarily on the sulfiding 1991, 30, 667. agent Ortega-Borges et al.27 have developed an essentially 13 G. A. Landis, J. J. Loferski, R. Beaulieu, P. A. Sekula-Moise, S. M. Vernon, B. Spitzer and C. J. Keavney, IEEE Trans. Electron. ligand free system, vide supra. Devices, 1990, 37, 372. 14 M.A. Kinch, Semicond. Semimet., 1981, 18, 312. 15 A. E. Nielsen, Kinetics of Precipitation, Pergamon Press, Oxford, Important factors 1964. CBD is potentially a method for the deposition of thin films 16 G. A. Kitaev, A. A. Uritskaya and S. G. Moksushin, Russ. J. Phys. Chem., 1965, 39, 1101. of ZnS. It is also a method free of the many inherent problems 17 I. Kaur, D. K. Pandya and K.L. Chopra, J. Electrochem. Soc., associated with high temperature techniques such as MOCVD 1980, 127, 943. including increased point defect concentration, evaporation of 18 M. Froment and D. Lincot, Electrochim. Acta, 1995, 40, 1293. ZnS leading to polysulfide and a limited choice of substrate. 19 G. D. Parfitt, Pure Appl. Chem., 1976, 48, 415. However to date only a small number of papers have reported 20 D.Lincot, R. Ortega-Borges and M. Froment, Appl. Phys. Lett., 1994, 64, 569. on the deposition of ZnS, by the CBD method, in any detail. 21 J. M. Dona and J. Herrero, J. Electrochem. Soc., 1994, 141, 205. At present progress may be limited by a poor appreciation of 22 B. Mokili, M. Froment and D. Lincot, J. Phys. III, 1995, 5, C3- the complexity of the system and a more thorough understand- 261.ing of the underlying mechanism involved is required for the 23 P. W. M. Jacobs and F. C. Tompkins, in Chemistry of The Solid design of better deposition systems. State, ed. W. Garner, Butterworth, London, 1955, p. 184. In conclusion there are a number of fundamental diVerences 24 A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd edn., John Wiley & Sons, Chichester, 1965, p. 19. between the CdS and ZnS systems which need to be appreci- 25 M. Avrami, J. Chem. Phys., 1941, 9, 177. ated: (i) CdS must, in some cases, be deposited by a mechanism 26 A. E. Martell and R. D. Smith, Critical Stability Constants, 1972, operating at close to the atomic level; epitaxy can be observed. vol. 2: L. D. Petit, IUPAC Stability Constants Data Base, The deposition of ZnS is often diVerent and involves clusters Academic Software, 1997.of ZnS. Part of the challenge in developing the deposition of 27 R. Ortega-Borges, D. Lincot and J. Vedel, in Proc. 11th E.C. Photovoltaic Solar Energy Conf., Harwood Academic Publishers, ZnS and related ternaries is to drive the process toward a Switzerland, 1992, p. 862. surface controlled ion-by-ion process. (ii) The eVect of pH in 28 O. Madelung, Semiconductors other than Group IV Elements and the CdS system is to form Cdx(OH)y species; those crucial for III-V Compounds, Springer, Berlin, 1992, pp. 26–27. film growth are bound to the surface of the substrate. These 29 M. Moskovitz, Chemical Physics of Atomic and Molecular metathesise to the sulfide.In the ZnS system high pH tends Clusters, ed. G. Soles, North Holland, Amsterdam, 1990, p. 397. 30 R. Rossetti, R. Hull, J. M. Gibson and L. E. Brus, J. Chem. Phys., to lead to the formation of Zn(OH)2/ZnO either on the 1985, 83, 1406. surface or as bulk material. At high values of pH metathesis 31 G. Hodes, A. Albu-Yaron, F. Decker and P. Motisuke, Phys. of these compounds to the sulfide is much less favourable than Rev., 1987, 36, 4215. for the cadmium species which can be rationalised by the 32 S. Gorer, A. Albu-Yaron and G. Hodes, J. Phys. Chem., 1995, relatively small diVerences in the solubility products for ZnS 44, 16442. (pKsp=24 at 25 °C) and ZnO/Zn(OH)2 (pKsp=17 at 25 °C). 33 X. Kang Zhao and J. H. Fendler, J. Phys. Chem., 1991, 95, 3716. 34 B. Mokili, Y. Charreire, R. Cortes and D. Lincot, Thin Solid (iii) The deposition of good quality ZnS appears to be optimal Films, 1996, 288, 21. in the presence of two complexants, typically ammonia and 35 J. W. KauVman, R. H. Hauge and J. L. Margrave, J. Phys. Chem., hydrazine. The dominant advantageous eVect of the second 1985, 89, 3541. ligand may be to enhance the rate of decomposition of thiourea 36 G. Marcotrigiano, G. Peyronel and R. Battistuzzi, J. Chem. Soc., increasing the rate of sulfidization of the growing film. (iv) In Perkin Trans. 2, 1972, 1539. 37 J. L. Walter, J. A. Ryan and T. J. Lane, J. Am. Chem. Soc., 1956, many cases there is the possibility that the literature does not 78, 5560. strictly report ZnS growth but more accurately a zinc 38 R. Ortega-Borges and D. Lincot, J. Electrochem. Soc., 1993, 140, hydroxysulfide material. 3464. We consider that a more closely controlled growth 39 S. Gorer and G. Hodes, J. Phys. Chem., 1994, 98, 5338. environment is a necessity if substantial advances in this field 40 P.O’Brien, unpublished results, 1996. are to be made. 41 S. Gorer, A. Albuyaron and G. Hodes, Chem. Mater., 1995, 7, 1243. 42 I. C. Ndukwe, Sol. Energy Mater. Sol. Cells, 1996, 40, 123. P.O’B. thanks the EPSRC for financial support and BP Solar Ltd. for gifts of substrates. P.O’B. is the Royal Society Paper 8/04692A 2314 J. Mater. Chem., 1998, 8(11), 2309–2314