The relative efficiency of electrowinning indium from chloride electrolytes

The main characteristics of electrolytic reduction applied to indium recovery from aqueous chloride solutions were considered theoretically and examined experimentally. Also other media were considered for comparison. Potentiodynamic examinations pointed out that chloride electrolytes are superior to sulphuric and nitric acid-based ones for efficient indium deposition. The optimum conditions, allowing higher than 90% cathodic current efficiencies, were determined by galvanostatic experiments. In contrast, similar sulphate electrolytes could only yield lower than 50% efficiencies because of a stronger hydrogen evolution. The galvanostatic results confirmed the findings of the potentiodynamic study. The cathodic polarisation curves revealed the major characteristics of the deposition process, implying a reduction of the electroactive In3+ ions. Coupled processes are the dynamic transformations of complex species and various rates of simultaneous hydrogen evolution. The structure and the elemental composition of the—relatively dense—cathodic deposits were examined by scanning electron microscopy and energy-dispersive X-ray analysis. Various levelling agents were also tested, and gelatine (up to a concentration of 1 g/dm3 in the chloride electrolyte) was found efficient in grain refining.


Introduction
In view of the limited resources and the fundamental role in modern electronic applications generating a steadily growing demand, indium has been declared a critical element by the European Union [1]. The primary resources are linked to the production of other metals, mostly to zinc metallurgy [2,3]. A variety of novel applications include the thermal interface material (TIM) layers to protect and connect integrated circuits and active devices [4], semiconductor compounds for solar cells, [5,6] special lead-free soldering materials [7] or the additive in alkaline battery anode alloys to substitute Cd and Hg [8,9]. Indium containing alloys are also used as control rods in nuclear reactors [10]. However, the major and the most dynamically increasing application-currently using more than half of the produced indium-is the transparent electrode material of indium-tin-oxide (ITO) in liquid crystal display (LCD) panels, [11] which really makes this element indispensable for modern technology. The recovery of indium from the residues of the ITO sputtering targets has already been commonly practised, [12] but the growing masses of the scrap LCD screens may become another important source in the future [13,14]. Secondary indium production can be mainly based on hydrometallurgy [13,15,16], and it may provide the larger part of the required supply.
As indium oxide is readily soluble in common mineral acids, leaching of the available secondary raw materials (mostly ITO targets, ITO etching cake, and LCD panels) is a relatively straightforward technology, amply covered in literature [12,13,[15][16][17]. Sulphuric and hydrochloric acid media have been found as the best-suited leaching agents [18]. There have been a number of solvent extraction methods and reagents proposed to purify and enrich the solutions for indium [18,19]. Efficient purification can be carried out also by solid-state anion-exchange [20]. The method considered so far for recovering the metal has been the straightforward cementation with zinc or aluminium as reducing agents producing an indium sponge [21]. However, the purity of the sponge is only 99-99.7% pure, which still requires further refining [22]. Recovery of a purer product or the refining of the raw sponge could be effectively made by electrowinning and electrorefining. The obtained cathodes can be melted, producing marketable indium ingots. No other processing of the obtained indium metal for structural applications are characteristic. Production of cast ingots however require cathode deposits of a suitable purity and a relatively compact state. It is known that In 3+ is the only stable oxidation state-in common aqueous media-of indium, characterised by the In 3+ /In 0 standard potential of − 0.34 V [23]. The standard redox potentials of the In 3+ /In 2+ , In 2+ /In + and In + /In 0 couples of − 0.49 V, − 0.40 V, and − 0.14 V, respectively, are increasing in this order, thereby stabilising the highest oxidation state [24]. The existence of the In 2+ and In + ions has no firm evidence in aqueous media [25]. The general electrochemical behaviour of indium at a cathode has already been studied fundamentally [26,27]. The electron transfer, the nucleation and the growth of the in crystals at the cathode surface have also been the subjects of previous investigations [28]. However, the understanding of the possible mechanisms and the major equilibrium parameters-although satisfying a theoretical interest-are not enough for the establishment of applicable recovery techniques. Indium electrodeposition has been practically used only for plating purposes required by applications in electronics [26]. Alkaline baths (cyanide, citric, tartaric, ethylene diamine tetraacetic, etc.) keep indium in a complexed form in solution and offer superior throwing power in electroplating. Acid baths (sulphate, sulphamate, fluoroborate) are simpler to prepare and stable, and they can offer higher current efficiencies [26]. Sulphate solutions are the simplest and most often used, although the relatively low (60-80%) cathodic current efficiencies are usual to be accepted. The addition of organic levelling agents has become widely applied in electroplating indium. The plating process has been commercialised with higher than 20 g/ dm 3 indium concentrations in ~ 30 g/dm 3 H 2 SO 4 baths. In order to achieve higher current efficiencies while obtaining smooth deposits, sulphamate and fluoroborate baths have been developed more recently, but chloride solutions have not been used commercially [26]. A practical interest is generated, though, by the relatively low cost of the InCl 3 salt to prepare plating baths. The choice of chloride electrolytes is naturally justified, however, if such solutions are used in the hydrometallurgical recovery of the metal. In view of the special possibility of applying efficient solution purification techniques implying the controlled formation of complex species, [29] chloride media may be thoroughly purified, thereby offering high-purity products [20]. However, the use of chloride media for indium electrodeposition has not been examined as yet. Therefore, we have investigated the practical aspects of indium electrowinning from chloride solutions. To understand the differences, mainly sulphate media were also studied.

Experimental procedure
The first series of experiments were carried out by the potentiodynamic technique aimed at revealing the effects of In concentration, pH, stirring, Cl − and SO 4 2− ion concentrations on the cathodic process. The digital potentiostatdeveloped in our laboratory-was capable of continuous polarisation sweeps even if it required a fast increase of the current. It was controlled by our specific driver and data processing software constructed on the Labview (National Instruments) platform. The working electrode of 1.5 cm 2 active surface area was supported by a copper sheet which was pre-plated with a consistent indium layer. The reverse sides and the edges were masked by acid-resistant lacquer. The anode was made of 99.99% pure graphite or a mixed metal oxide (MMO) mesh, commonly referred to as a dimensionally stable anode (DSA). The starting cathodes-of the galvanostatic experiment-had a 2.25 cm 2 active central area, which was cut out from Cu sheets of 99.95% purity and polished mechanically. The latter was mechanically polished before each run to ensure the same starting surface quality. In the second series of experiments, we examined the galvanostatic current efficiencies of indium deposition as a function of the cathodic current density from sulphate and chloride media. For this purpose, a regulated analogue DC current supply (Statron 3255.1) was used. The current was checked with a digital instrument. The electrolyte solutions were prepared from 99.999% pure indium metal and the required amount of hydrochloric and-for comparisonsulphuric acid. The compositions of the stock solutions are shown in Table 1.
The experiments were carried out in open glass cells of 100 cm 3 useful volume. The pH was set by controlled addition of NaOH, a common practice in electrowinning practice. The Cl − and SO 4 2− ion concentration was controlled by NaCl and Na 2 SO 4 addition. The reverse sides and the edges were masked by acid-resistant lacquer. The experimental setup is shown in Fig. 1.
The masses of the electrodes were measured before and after the experiments after rinsing with water and drying. The differences between the final and the starting massesalong with the recorded current and time-were used for the determination of the current efficiencies. In specific experimental runs, the pH was measured with a double junction electrode by periodically stopping the current and introducing a 350 r.p.m. stirring for 30 s. All the experiments of electrodeposition were carried out at room temperature. The aqueous speciation of indium ions was calculated using the computer program ROCC developed at our institute [30].

3
The obtained cathode surfaces were examined and characterised by scanning electron microscopy (SEM) combined with energy dispersive X-ray (EDS) analysis.

Experimental results and discussion
Efficient cathodic reduction requires the optimisation of the electrolyte solution and the current density principally. The combined goals include achieving high current efficiencies and avoiding the growth of rough or dendritic crystals at the cathode surface. The morphology of the deposit plays an important role not only in the physical purity of the product but also in the efficiency of further processing it.

Potentiodynamic investigation
As a fundamental approach, a series of preliminary experiments of potentiodynamic electrodeposition was conducted by applying hydrochloric, sulphuric and nitric media. In each solution, the indium concentration was 50 g/dm 3 , while the pH 1 value was set by preliminary controlled neutralisation. The polarisation characteristics of the examined media are shown in Fig. 1. Due to the continually changing surface roughness, the vertical axis is scaled to the directly measured current instead of the usual current densities. According to the geometric area (1.5 cm 2 ) of the active electrode surface, the corresponding "apparent" current densities-practically corresponding to the initial values-can be calculated (Fig. 2).
In the case of the sulphate solutions, the current emerges only at higher cathodic polarisation, beyond − 0.5 V, which can, however, be attributed mostly to H 2 evolution. In contrast, current appears immediately at low cathodic overpotentials in both chloride and nitrate electrolytes, which is observed to be related to indium deposition. The limiting current appeared around − 0.5 V overpotential in the stationary electrolyte. The corresponding section is shorter in the nitrate solution because the deposited indium gets redissolved by the NO 3 − ions, yielding a constant supply of In 3+ ions near the cathode surface. Increasing the cathodic polarisation beyond − 0.75 V, hydrogen evolution contributes to the generated current in all the examined electrolytes of pH 1, which was confirmed by visual observation. In the case of deposition from the chloride solution, indium showed no tendency of re-dissolution, and the evolution of hydrogen was not intensive even at higher cathodic polarisations. However, the limiting current seen in the stationary solution was reduced by the accumulation of hydrogen bubbles at the electrode surface. The reduction of H + ions may increase the total current initially, but as the active surface gets more blocked by adhering bubbles, the effective current density increases, and the current must drop if the overpotential is not increased at that rate. Due to the advantageous features, further potentiodynamic experiments were focused on the chloride electrolyte solutions.
The effect of varied indium concentration on the cathodic process in 1 M Cl − and pH 1 solutions and agitated with different stirring intensities can be seen in Fig. 3. As the indium concentration is increased, the steadily rising initial sections of the polarisation curves become longer and higher limiting currents are reached. The second, increasing section, caused by the continuous release of hydrogen bubbles, appears as the cathodic overpotential surpasses ~ − 0.75 V. However, with 25 g/dm 3 indium concentration and intensive (500 r.p.m.) stirring, the initial increasing tendency of the observed current was not broken because the limiting conditions of indium deposition virtually coincided with the onset of the intensive hydrogen evolution. If a limiting current for indium is reached, it is followed by a drop in the current if no stirring is applied. It can be attributed to the accumulation of adhering hydrogen bubbles. As expected by the Buttler-Volmer relationship, the initial slopes of the current plots are virtually similar, and the limiting currents of indium deposition are increasing as the indium concentration is increased in the stationary solutions. Intensive hydrogen evolution appears only at high (< − 0.75) cathodic overvoltages. The increase of the current in the later sections of the polarisation plots is, however, also enhanced by the roughening electrode surfaces of the gradually increased actual area.
The practical limiting current densities can be calculated from the plotted currents and the known geometric cathode surface; however, the rougher surface arising in electrolytes of higher indium concentration may significantly distort the practically expressed values. For example, with stationary solutions of 2.5 g/dm 3 indium concentration at pH 1, an apparent current density of just ~ 100 A/m 2 can be derived from the observed current and the geometric surface of the cathode, which can be used efficiently. However, 50 g/dm 3 indium in this solution could allow about ten times higher useful current densities.
Chloride electrolytes with indium concentrations below 25 g/dm 3 could not provide high currents related to indium deposition. At 50 g/dm 3 , however, the regular section of the current plot was longer, allowing ~ 1200 A/m 2 current density for the major reaction of indium deposition before the onset of limiting conditions. Thus, based on the potentiodynamic results, 50 g/dm 3 indium in the chloride solutions may offer practically advantageous conditions. As the possible side reaction of H 2 evolution may become important, the effect of pH at different stirring speeds has also been examined with 50 g/dm 3 indium concentration in the solution. Results referring to pH 3 and pH 0 are compared in Fig. 4, while the diagram for pH 1 can be seen in Fig. 3d.
The cathodic current in a wide range of overpotentials at pH 3 may arise virtually only from indium deposition. At low levels of electrolyte agitation, clear limiting currents are established, and only slight hydrogen bubble formation can be seen in the stationary solution. At pH 0, however, the initial current plots are steeper and shorter, indicating that indium deposition was accompanied by the reduction of H + ions. The rising tendency is interrupted by the limiting conditions for the main reaction, and the subsequent current drop indicates the hydrogen bubble accumulation mechanism.
Less acid would be advantageous in depressing hydrogen evolution. However, according to the equilibrium activities ( a Me + ) of the aquo-ions calculated from the hydroxide stability constants [31], much indium may precipitate from the solution if the pH rises above ~ 3 near the cathode. Some impurities-such as Fe(III), Sn(IV) and Sn(II)-may precipitate as hydroxides at even lower pH, and thus contaminate the solution and the cathode.
The stability of In 3+ ions-in the In H 2 O 6 3+ hexaaquo form-is limited to a relatively low pH range of < 2.5. [32] We have confirmed this range by determining the concentration ratio of indium remaining in the chloride solution during a controlled increase of the pH at room temperature, as shown in Fig. 5.
Although not involved directly in the cathodic process, the effects of the Cl − ion concentration were also examined with and without stirring in solutions of pH 1 containing 50 g/dm 3 indium. As seen in Fig. 6, increasing Cl − ion concentration results in steeper current development initially, but also the earlier appearance of hydrogen bubble formation. It is caused by the rougher crystal structure of the deposited metal.
As also seen in Fig. 3, remarkably higher cathodic currents are produced by applying vigorous stirring at higher cathodic overpotentials in the 50 g/dm 3 In and pH 1 electrolyte. Figure 5b shows that if the Cl − ion concentration is increased from 1 to 3 mol/dm 3 in this electrolyte, the cathodic current can further increase. The curves of 3 M and higher Cl − concentrations coincide. It is related to enhanced hydrogen evolution, also indicated by the fluctuations occurring beyond approximately − 0.5 V cathodic overpotential caused by the uneven release of the H 2 bubbles at the changing surface. With increasing Cl − ion concentrations, indium forms more stable complex ions. This is demonstrated by the relative distribution of the species shown in Fig. 7a, constructed from the computed results based on the average stability constants [31,33,34]. In a series of auxiliary experiments, we have also tested the equilibrium sorption of indium on a strongly basic anion exchange resin to find that the maximum coordination number of the tetrahedral complex is reached at about 4 mol/dm 3 free Cl − ion concentration, while anionic species of indium have not been detected by this technique in sulphuric acid. We have seen that the addition of chloride ions to the indium sulphate solutions results in significant changes in the light absorption spectra, as shown in Fig. 7b. With increasing Cl − concentration, the resulting absorption peaks shift to higher wavelengths, indicating the stronger complexing effect of the Cl − ions and the increasing coordination degrees of the formed complex species [35].
The formation and decomposition of the complex species of various coordination degrees is an equilibrium process producing the electroactive In 3+ ions. However, the increased anion concentration shifts the equilibrium towards the complex formation, thereby hindering the cathodic deposition of the metal, while the available H + ions can be reduced. The complex formation also causes a slower transport of the indium species in the solution. As a result, the surface becomes rougher, which may decrease the actual current density, and it can also increase the observed current.
Based on the potentiodynamic results, the best electrolyte solution for indium deposition in the examined range can be 1 mol/dm 3 Cl − and 50 g/dm 3 In concentrations at pH 1. However, the levelling effect of some organic additions may also be considered. As shown in Fig. 8, the initial indium deposition is not changed with the addition of all the tested organic agents to the stationary chloride solutions.
Depending on the efficiency of grain refinement, there is a potential limit where the process changes and H + ions are suddenly reduced. The generated micro-scale bubbles quickly cover the surface, blocking the further deposition of indium. As the cathode is further polarised, the hydrogen bubbles are released, and an intensive gas evolution starts. The increased addition of gelatine (2 g/dm 3 ) was evidently over the optimum rate, resulting in a rougher indium deposit-as also observed visually-and higher currents could be reached before the formation of H 2 bubbles, causing less perfect surface coverage.
The combination of 0.2 g/dm 3 naphtol to 1 g/dm 3 gelatine addition resulted in a shorter initial section of the polarisation curve when a fine-grained indium layer was deposited. It was soon followed by the onset of intensive hydrogen bubble formation, abruptly blocking the active surface until the onset of continuous hydrogen evolution. The addition of thiourea, however, remarkably changes the main cathodic process, enhancing a dendritic electro-crystallisation, and producing long outgrowths, which in turn induced a strong increase in the current.

Galvanostatic investigations
The effects of the major parameters of indium electrodeposition were determined by galvanostatic experiments simulating the practical application of the method in stationary electrolytes. The examined ranges were selected according to the fundamental findings of the potentiodynamic results in Chapter 3.1. In order to prove the features of the chloride electrolyte and to support a wider interpretation, comparisons were also made with the more common sulphate electrolytes. The effects of the most important practical factors, i.e., current density and electrolyte composition, are shown in Fig. 9. Approximately similar masses of the deposited metal were assured by the electrolysis time set inversely proportional to the current density.
The efficiencies in Fig. 9a just slightly increased with the current density in the lower range but dropped significantly beyond ~ 100 mA/cm 2 . The current densities beyond 100-120 mA/cm 2 proved to be inefficient in the stationary electrolyte because of the lagging metal ion transport, accompanied by the strengthening of H 2 evolution at the cathode. Approximately the same critical current density was found as the maximum to support fundamentally undisturbed indium deposition by the potentiodynamic results presented in Fig. 3d with the same type of stationary electrolyte solution. The remarkably lower efficiencies obtained with the sulphate electrolyte were justified by the observations of definitely stronger H 2 evolution, corresponding also to the relatively low currents observed in the comparative polarisation tests (Fig. 1). The combined effects of the indium concentration and pH can be seen in Fig. 9b. With low indium concentrations, lower pH results in stronger hydrogen evolution and correspondingly lower current efficiencies. It is also in good agreement with the potentiodynamic results discussed in Chapter 3.1. The conditions are extremely unfavourable at lower pH (< 1) values.
The effect of the Cl − ion concentration on the cathodic current efficiency is demonstrated by Fig. 10a. For comparison, the effect of SO 4 2− ions was also checked and presented. Dissolved indium is supposed to bind anions according to the equilibrium distribution illustrated in Figs. 7a and 10b. Anions were further added to test their effect on the cathodic efficiency in the form of neutral sodium salts, as indicated in Fig. 10a. This is a common practice (providing at least 1-2 mol/dm 3 addition) in the technological application to improve the conductivity of the electrolyte if acid addition is excluded because of the danger of excessive hydrogen evolution at the cathode.
The cathodic current efficiencies were generally high with the given settings in the chloride electrolyte in the whole examined range, but a sharp decrease was found with increasing anion concentration in the sulphate system. It is also in line with the potentiodynamic results. As shown in Fig. 10b, constructed from the computed results based on the averages of published complex stability constants [31,33,34], the formation of [In(SO) 4 ] 2− becomes dominant above 1 M SO 4 2− ion concentration, which suppresses the electroactive In 3+ species available for electrodeposition. With Cl − ions added into a sulphate electrolyte, however, the current efficiencies can recover. With increasing addition, the efficiency approaches that in the pure chloride solution. It shows that the formation of the complex species is more favoured in the chloride system, as also indicated by the equilibrium constants noted in Figs. 7a and 10b. Indium is mostly present in the anionic form at relatively low concentrations of the free sulphate ions. The [In(SO 4 ) 2 ] − species may be stable above 0.5 M SO 4 2− , and the [In(SO 4 ) 3 ] 3− ions may become dominant above 1 M SO 4 2− concentration.  Fig. 7b, may explain the more efficient cathodic process in chloride solutions. The electroactive In 3+ ions can be less available at the cathode surface in the sulphate system, especially if the SO 4 2− concentration is increased. For a stable electrowinning operation, the reactions at the inert anode are also important. The equilibrium redox potentials of O 2 and Cl 2 are close (1.18 and 1.35 V, respectively) to each other in the recommended (1 M Cl − and pH 1) electrolyte. Although the anodic oxygen evolution has the thermodynamic possibility, the pH of the electrolyte did not change during a galvanostatic experiment of high (> 93%) cathodic current efficiency and practically no H 2 evolution. Hence, it is understood that the evolution of Cl 2 could dominate the anodic reaction. In the sulphate system, however, the anodic reaction was O 2 evolution, generating more H + ions, than those lost at the cathode by H 2 evolution. This is confirmed by the detected decreasing pH value of the electrolyte.

The morphology of the cathodic deposits
The main information about the overall morphology of the galvanostatic deposits were first examined under the optical microscope, followed by scanning electron microscopy (SEM)-coupled with energy dispersive X-ray (EDS) analysis. The cathodes obtained with different current densities from the specified chloride media are presented in a matrix of macro-and microphotographs in Fig. 11.
Comparing the three columns of the microphotographs shows that the deposition morphology gets remarkably rougher with increasing current densities. However, even at 600 A/m 2 -which is considered fairly high in conventional metal electrowinning technology-the indium deposit is dense. At as high as 1200 A/m 2 , a constant H 2 evolution was observed in relation to the small holes found all over the cathode surface. At such high current densities, the electro-crystallisation even tends to develop loose dendrites. Apart from the holes, the deposit is smooth but more matt in visual appearance. Although the crystals deposited at the central area may look finer, the overall structure is spongier. In additional experiments, we have noticed that the deposits obtained with soluble indium anodes get rougher, caused by the lack of extra turbulence without the evolution of the anode gas. The average elemental composition of the surface was analysed by EDS. The results are in Fig. 12.
Most noticeably, the chlorine, sodium and oxygen contents increase with current density. This is caused by the increased surface area of the rougher deposits at higher current densities. During crystallisation, impurities from the electrolyte can be trapped within the growing crystals, but the rougher deposit is also harder to wash properly, and residual electrolyte can remain entrapped. While the chlorine contamination is present mainly on the surface of the crystals, but due to the rough morphology, the cathode can entrap some Cl also in the bulk. However, the minor chlorine impurity arising from the HCl-based solutions is usually eliminated during the melting of the obtained cathode, which is less efficient in the case of NaCl containing electrolytes.
Also, a higher surface area will lead to excess oxidation during washing and drying, thus causing higher oxygen content. For the purpose of comparison, also deposits from sulphate media were analysed. The results are shown in Fig. 13.   Fig. 10 The effect of Cl − and SO 4 2− ions on the cathodic current efficiencies in different (60 mA/cm 2 , pH 1, 50 g/dm 3 In) solutions (a), and the ionic species of indium in sulphate media (b) The deposit obtained from the 1 M SO 4 2− medium ( Fig. 13a and c) is built up of larger crystals; however, at 2 M SO 4 2− , as seen in Fig.13b and d, smaller crystals arise because of the turbulence resulting from the strong H 2 evolution. Figure 13e shows that at the examined higher sulphate ion concentration, the cathode was covered with a thin layer of especially high sulphur content. As a characteristic difference, more homogenous, although rougher cathodic deposits can be obtained from the pure chloride media. The sulphur contamination (5-15 ppm) is common-in the industrial practice-in deposits obtained from sulphate media [36,37]. This is mostly due to the physical entrapment of the electrolyte within the structure of the cathode. Grain refining and levelling result in denser deposits, thus lessening this effect.
It is generally known that finer electrocrystallisation and smoother cathode surfaces can be obtained by the application of organic additives [38]. Therefore, we have examined the effect of common grain refining additives in chloride electrolytes also by the galvanostatic method. However, besides the potential benefits in terms of morphology, the effects on the current efficiency need also to be considered. While the current efficiencies were found practically the same with or without the addition of 1 g/dm 3 gelatine or thiourea, the addition of 0.2 g/dm 3 naphthol significantly reduced the efficiency in electrolytes containing 1 M free Cl − , 50 g/dm 3 In, at pH 1 and run with 600 A/m 2 for 1 h. Due to the detrimental effect of 2-naphthol on the cathodic current efficiency, it is not considered further. Gelatine may, however, be applied to advantage as it has no harmful effect on the efficiency even at the relatively high (1 g/dm 3 ) rate of addition, whereas the grain refining effect can be considerable, as shown in Fig. 14.
While the bulk of the deposit in the central area appeared quite smooth (Fig. 10) without any additive at lower than Fig. 11 The cathode morphologies obtained at different current densities (pH 1; 1 M free Cl − , 50 g/dm 3 In) Fig. 12 The average concentration of elements detected with EDS at the surface of the electrodes: a − 30 mA/cm 2 ; b − 60 mA/cm 2 ; c 120 mA/ cm 2 (pH 1; 1 M Cl − , 50 g/dm 3 In) 600 A/m 2 current densities, large globular or-at the highest examined current density-even dendritic crystals developed at the edges. As suggested by Fig. 14, adding 1 g/dm 3 gelatine to the electrolyte results in a fine crystal structure over the whole cathode surface. The edge roughening is relatively slight. Increasing the added amount of gelatine, however, may have no further benefits or even may have some contrary effect on the morphology. As a general effect of grain refinement, the actual current densities on the cathode are relatively higher, and the correspondingly higher polarisation at the set current may result in more intensive H 2 evolution and lower current efficiency. Further, thiourea proved to be even detrimental to the deposit as it strongly increased the formation of long and loose dendrites at the edges. Finally, the addition of 1 g/dm 3 gelatine is considered safe and sufficient to assure dense and relatively smooth indium deposits while maintaining high cathodic current efficiencies. As a further unwanted effect, though, when grain refiners were added to the solution, dark spots of 10-80 µm appeared on the SEM images of the cathodes. According to the focused EDS results, these small areas are rich in sodium chloride salt. The sodium ions originated from the NaOH used for setting the pH of the electrolyte. However, when thiourea was added, the entrapped salt contained sulphur as well. The EDS analysis of the average cathode surfaces-with carbon below the detection limit-is shown in Fig. 15.
The entrapped NaCl can be removed from the surface either by more thorough washing or during the melting of the cathode, where it would float to the top of the melt, forming a dross phase. A further practical finding concerns the removal of the indium deposit from the mother plate. As a result of strong adherence, the soft indium deposit is rather stuck to the surface of the tested Cu, Fe, Zn or Al substrates. However, we have found that it can be readily stripped from titanium mother plates.

Conclusion
The electrochemical processes and the main practical characteristics of cathodic reduction to produce pure indium from chloride and other aqueous solutions were examined  15 The EDS analysis of cathodes obtained with 1 g/dm 3 (a) and 3 g/dm 3 (b) gelatine and with 1 g/dm 3 thiourea (c). (pH 1, 1 M Cl − , 50 g/ dm 3 In, 600 A/m. 2 , 1 h) experimentally. The effects of the major parameters of indium concentration, pH, solution agitation, and the type and concentration of the anions were pointed out by the polarisation plots obtained by potentiodynamic technique. According to these results, the chloride solution of 1 M Cl − and 50 g/dm 3 In composition at pH 1 is the best choice for indium electrowinning. Hydrogen evolution is the main interfering reaction at low indium concentrations, especially in stationary solutions. The irregularities of the polarisation curves were interpreted in terms of bubble formation and intensive gas evolution. High current efficiencies (over 90%) could be reached in galvanostatic experiments by applying as high as 600 A/m 2 apparent cathodic current density. The morphology of the obtained deposit can satisfy practical requirements. The addition of up to 1 g/dm 3 gelatine has an appreciable grain refining effect, while the current efficiency is not affected adversely, reaching 90% on average in chloride solutions, while it was only around 50% in sulphate electrolytes under similar conditions. The great difference in behaviour was interpreted in terms of the different stabilities and possibly different dynamic properties of the formed complex species. Hydrogen evolution contributes to the generated current in all the examined electrolytes, therefore, the control of pH is indispensable, keeping it as high as it is allowed by hydrolytic stability. Chloride solutions are performing best also in this respect.