The electrochemical reduction processes of solid compounds in high temperature molten salts.

Chem Soc Rev Published on 18 February 2014. Downloaded by Lomonosov Moscow State University on 18/02/2014 11:34:28.

TUTORIAL REVIEW

Cite this: DOI: 10.1039/c3cs60327j

View Article Online View Journal

The electrochemical reduction processes of solid compounds in high temperature molten salts Wei Xiao and Dihua Wang* Solid electrode processes fall in the central focus of electrochemistry due to their broad-based applications in electrochemical energy storage/conversion devices, sensors and electrochemical preparation. The electrolytic production of metals, alloys, semiconductors and oxides via the electrochemical reduction of solid compounds (especially solid oxides) in high temperature molten salts has been well demonstrated to be an effective and environmentally friendly process for refractory metal extraction, functional materials preparation as well as spent fuel reprocessing. The (electro)chemical reduction of solid compounds under cathodic polarizations generally accompanies a variety of changes at the cathode/melt electrochemical interface which result in diverse electrolytic products with different compositions, morphologies and microstructures. This report summarizes various (electro)chemical reactions taking place at the compound cathode/melt interface during the electrochemical reduction of solid compounds in molten salts, which mainly include: (1) the direct electro-deoxidation of solid oxides; (2) the deposition of the active metal together with the electrochemical reduction of solid oxides; (3) the electro-inclusion of cations from molten salts; (4) the dissolution–electrodeposition process, and (5) the electron hopping process and carbon deposition with the utilization of carbon-based anodes. The implications of the forenamed cathodic reactions on the energy efficiency, chemical compositions and

Received 16th September 2013

microstructures of the electrolytic products are also discussed. We hope that a comprehensive

DOI: 10.1039/c3cs60327j

understanding of the cathodic processes during the electrochemical reduction of solid compounds in

www.rsc.org/csr

molten salts could form a basis for developing a clean, energy efficient and affordable production process for advanced/engineering materials.

Key learning points (1) The molten salt electrolysis of solid compounds is a clean, energy efficient and affordable process for refractory metal/alloy extraction, the synthesis of functional oxide composites, spent nuclear fuel reprocessing and the controllable preparation of nanostructured semiconductors. (2) The electrochemical reduction mechanisms of the solid compounds in molten salts show a significant influence on the energy efficiency, chemical compositions and microstructures of the electrolytic products, which therefore determines the economic viability and environmental benignity of the process. (3) The electrochemical reduction mechanisms of the solid compounds in molten salts show a dependence on the solubility of the solid compounds in molten salts. (4) All reaction mechanisms summarized here are generic and can be extended to other solid electrode processes.

1. Introduction Metals, alloys and semiconductors are building blocks of the modern society and thus are always needed and hence are produced on a large scale. The vast majority of these materials are being industrially produced by various pyrometallurgical methods such as carbothermic and/or metallothermic reductions. School of Resource and Environmental Sciences, Wuhan University, Wuhan 430072, PR China. E-mail: [email protected]; Fax: +86 27 68775799; Tel: +86 27 68774216

This journal is © The Royal Society of Chemistry 2014

Upon utilizing carbon or metals as energy carriers, the present carbothermic or metallothermic processes generally register an unsatisfactory energy efficiency and a considerable carbon footprint. Alternatively, electrons with adjustable activities and potentially green origins are ideal reductants for metallurgical processes. The advantage of the electrons as energy carriers has provoked enormous motivation on the development of electrometallurgy and electro-synthesis of materials, especially using high-temperature molten salts as electrolytes due to their low cost, low toxicity, abundant availability, high heat capacity, low vapour pressure, large electrochemical window, high ionic

Chem. Soc. Rev.

View Article Online

Published on 18 February 2014. Downloaded by Lomonosov Moscow State University on 18/02/2014 11:34:28.

Tutorial Review

conductivity and their ability to dissolve species. The economic viability and environmental benignity of molten salt electrolysis can be also justified by the following facts: (1) the wellestablished infrastructure, huge output and technological maturation of the electricity industry and (2) the increasing installed capacities of electricity from renewable energies in grids. Electricity from renewable energies can be used for heating molten salts to the desired reaction temperatures and driving chemical transformations; while the inevitable heat loss of molten salts at high temperatures can be (at least partially) compensated by joule heat. Although electrolysis has achieved ´roult a great success in industrial aluminium (namely Hall–He process) and active metal extraction in which an electrodeposition process occurs, it still remains a challenge to extend the electro-deposition process to refractory metal production since it requires dissolved feedstock and liquid deposits. It is well known that solid ores are the most straightforward natural materials and hence should be the most favourable feedstock for metallurgical processes. The direct electrochemical reduction of solid compounds to solid products in molten salts represents a breakthrough of refractory metal metallurgy and has received intensive attentions across academic and industry in recent decades.1–3 In addition to metallurgical applications, the molten salt electrolysis of solid compounds (e.g. oxides) is applicable to some more elaborate and sophisticated processes such as the reprocessing of spent oxide nuclear fuel4 and the preparation of nanostructured semiconductors,5,6 in which the economic viability and environmental benignity of the molten salt electrolysis can be further enhanced. As energy and resources become more valuable, technologies that can utilize energy and resources more efficiently and in a sustainable way are urgently needed. In this context, the electrochemical reduction of solid compounds has been demonstrated as an affordable, green and versatile technology. A comprehensive and

Dr Wei Xiao studied Chemistry in Wuhan University and received his BS in 2002 and PhD (Physical Chemistry) in 2007. He then performed research on electrochemical energy storage/ conversion devices through a postdoctoral research fellowship at the National University of Singapore and Nanyang Technological University. After a two and half years stay in Singapore, he performed Wei Xiao postdoctoral research on electrolytic silicon extraction with the University of Nottingham (UK). He joined Wuhan University as an associate professor in July, 2011, and his current research focuses on novel electrochemical technologies towards energy, resources and environmental sustainability.

Chem. Soc. Rev.

Chem Soc Rev

deep understanding of the (electro)chemical reaction mechanisms of solid compounds in molten salts is the prerequisite for the optimization and application of the forenamed process. In the present manuscript, various cathodic reduction mechanisms of solid compounds, especially solid oxides, upon molten salt electrolysis are summarized and discussed, with the purpose of providing a brief overview of the involved (electro)chemical reactions and their influences on the energy efficiency, chemical compositions or microstructures of the electrolytic products. The involved cathodic reduction mechanisms and the corresponding electrolytic products at the cathode are summarized in Scheme 1. It should be noted that the cathodic reduction mechanisms governing the molten salt electrolysis of the solid compounds are determined by both thermodynamics and kinetics. And the dimension/microstructure of the feedstock and electrolysis parameters exhibit significant impacts on the actual reaction mechanisms, incurring somewhat controversial reports on the cathodic reduction mechanisms of solid oxides in molten salts. In addition, several mechanisms will occur in parallel or in series during the molten salt electrolysis of bulk solid compounds. Furthermore, the cathodic mechanisms are interconnected with anodic reactions. Therefore, the specification of the practical cathodic reduction mechanisms of individual or mixed oxides falls out from the scope of the manuscript, but has been well documented in previous reports.1,3

2. Direct electro-deoxidation of solid oxides This mechanism was accepted as the cathodic mechanism in the FFC-Cambridge Process which involved the direct electrochemical reduction of solid oxides to metal/alloys in molten chlorides.2 An oxygen ionization mechanism has been introduced to describe the electrochemical reduction of solid oxides

Prof. Dr Dihua Wang received his BS (1991), MS (1994) degree in Environmental Chemistry and PhD (1998) in Electrochemistry at Wuhan University, China. He joined the faculty of Wuhan University in 1994 and was promoted to full professor in 2007. His research focuses on the green electrochemical technologies for sustainable resources and energy. He spent half year at the University of Dihua Wang Nottingham as a visiting scholar in 2004. From 2006 to 2008, he worked with Prof. Donald R Sadoway at MIT as a visiting professor. He is author/co-author of more than 100 peer-reviewed papers and holds 12 patents.


View Article Online


Chem Soc Rev

Tutorial Review

Scheme 1 A summary of the main reaction mechanisms and cathodic electrolytic products of the molten salt electrochemical reduction of solid compounds.

in molten salts, such as molten CaCl2.2 The FFC-Cambridge Process is consequently represented by the following reactions. Overall reaction: Using inert anode: MOx(s) - M(s) + x/2O2

(1a)

Using a carbon-based anode: nMOx(s) + xC(s) - nM(s) + xCOn

(1b)

Cathode reaction: MOx(s) + 2xe - M(s) + xO2

(2)

xO2 - x/2O2 + 2xe

(3a)

Inert anode:

Carbon anode, here n = 1 or 2: nO2 + C - COn + 2ne

(3b)

where, M represents metals (e.g. Ti, Cr, Nb and Ta) or semimetals (e.g. Si). The key innovation of the FFC-Cambridge process is based on the thermodynamic fact that the decomposition voltages of many solid oxides at high temperatures, ranging from 873 to 1173 K, are lower than the decomposition voltages of the inorganic molten salts (see Fig. 1).2,3 All of the thermodynamic data were derived from HSC Chemistry version 6. During the electrolysis of a solid oxide feed in a molten salt (e.g. molten CaCl2) with the ability to dissolve and transport O2 ions, the direct electrochemical reduction (Reaction (2)) of the oxide


Fig. 1 The decomposition voltages of typical solid oxides, calculated based on the corresponding Gibbs free energy changes derived from HSC Chemistry version 6.

feeds can occur at a potential more positive than the equilibrium potential of Ca2+/Ca in molten CaCl2. It should be noted that the ionized oxygen (O2) transports and exists in the melt during the electro-deoxidation. It was reported that the discharge of

Chem. Soc. Rev.

View Article Online


Tutorial Review

O2 on the graphite anode is thermodynamically more favourable than that of Cl.6 The decomposition of the molten chlorides (e.g. CaCl2 or LiCl) during the electrochemical reduction of solid oxides is determined by the decomposition voltage of CaO or Li2O. However, it is acknowledged that the different concentrations (activities) of O2 and Cl in the molten CaCl2 and the electrochemical interface should also be considered. Upon applying a suitable cathodic overpotential to the oxide cathodes, solid metals can be generated at the cathode via the direct electrochemical reduction or electro-deoxidation of oxides (as shown by Reaction (2)), given the fact that both the oxide precursors and the corresponding metals do not dissolve significantly in the molten salts. Simultaneously, the oxygen in the oxides is ionized at the cathode. The ionized oxygen diffuses from the cathodes to the melt and is eventually electro-oxidized to oxygen at an inert anode surface (Reaction (3a)) or CO/CO2 at a carbon anode (Reaction (3b)). The direct electro-deoxidation of solid oxides can not only be driven in a molten salt electrolyzer (e.g. in the FFC-Cambridge process), but also can be triggered in a molten salt primary battery configuration. The latter was well demonstrated during the reduction of solid Nb2O5 to Nb powder in molten CaCl2 by an electronically mediated reaction (EMR) using liquid Ca as the reductant.7 The overall reaction in the EMR is a calciothermic reduction, without a direct physical connection between the feed oxide and reductant (Ca). Liquid Ca is oxidized to Ca2+, with the emitted electrons being introduced into solid Nb2O5 in different locations through molten CaCl2. Then solid Nb2O5 is reduced to Nb also via Reaction (2). The direct electro-deoxidation mechanism follows a similar reduction pathway to the so-called second-type electrochemical process. Based on the previous research on the electrolysis of solid silica, the initiation–expansion–shrinking–disappearing (IESD) process of the solid (conductive product)/solid (oxides)/ liquid (molten salt) three-phase reaction boundary (TPB) was proposed as the mechanism describing the direct electrodeoxidation of the solid oxides.8–11 Given that the metal oxide (MO) is an insulator, a good initial contact between the current collector and the oxide are essential for the onset of oxygen ionization at the cathode. The direct electrochemical reduction of the insulating MO starts at the initial current collector/solid MO/molten CaCl2 TPB.3,8,12 The reduced metal (M), which has a reasonable electrical conductivity, is porous due to the decrease in the molar volume. Simultaneously, molten CaCl2 enters the pores, leading to the formation of a new M/MO/CaCl2 TPB which propagates along the surface of, and also penetrates into, the solid MO until complete reduction. When the MO is reduced, the O2 ions move into the neighbouring electrolyte, surrounding the reduced and porous metal layer, and are then transported to the bulk electrolyte before being discharged at the anode. The propagation of the three-phase reaction boundary both along the surface and towards the bulk leads to the complete electrochemical reduction of oxides.3,8,12 The kinetics of the electrochemical reduction of an oxide cathode can then be described based on the dynamic (IESD) three-phase (solid conductive product/solid oxides/liquid molten salt) reaction boundary models.8,12 It should

Chem. Soc. Rev.

Chem Soc Rev

Fig. 2 A typical cyclic voltammetric (CV) plot of the NiO powder loaded metallic (Mo) cavity electrode (MCE) in molten CaCl2 at 1123 K recorded at 100 mV s1.

be noted that the established direct electrochemical reduction mechanism is general and can be extended to all systems related to the electrochemical reduction of solid insulators. 2.1

The one-step electro-deoxidation of solid oxides

The electrochemical reduction of solid oxides that have a single thermodynamically stable M–O phase or are in low-valence states can be fulfilled via a one-step electro-deoxidation of the solid oxides (e.g. NiO). Also, the phase diagram of the NiO–CaO binary systems shows the absence of any thermodynamically stable calcium nickelates. Therefore, the electrochemical reduction of solid NiO in molten CaCl2 should be a one-step electro-deoxidation of solid NiO to metallic Ni, as evidenced by the cyclic voltammetry of solid NiO in molten CaCl2 at 1123 K shown in Fig. 2. (Counter electrode: graphite rod; reference electrode: mullite-sheathed Ag/Ag+,13 denoted as Ag/Ag+ in the figures.) The NiO powder loaded metallic (Mo) cavity electrode (MCE) was employed as the working electrode.14 It is clearly shown that the electro-deoxidation of solid NiO (resulting a cathodic peak of C1) occurs at the potential more positive than that of the decomposition of the melt (resulting a cathodic peak of C2), with the generation of sintered Ni nodules (see Fig. 3a) on the cathode upon the cathodic polarizations. The direct electrodeoxidation of solid NiO can be expressed as: NiO(s) + 2e - Ni(s) + O2

(2a)

2.2 The continuous or stepwise electro-deoxidation of solid oxides Many metal–oxygen binary systems exhibit various thermodynamically stable M–O phases, suggesting that the precursory oxides can possibly be electro-reduced stepwise (if the M–O binary systems have multiple-phase oxides) or continuously (if the M–O binary systems can form solid solutions). A series of oxygen ionization steps may lead to the formation of partially reduced oxides before the complete metallization of the cathode. The generation of the partially reduced oxides during the direct electrochemical reduction of the oxide at the cathode (Reaction


View Article Online


Chem Soc Rev

Tutorial Review

Fig. 3 Typical SEM images of (a) the electrolytic Ni obtained from the electrolysis of NiO in molten CaCl2 at 1173 K (adapted with permission from The Electrochemical Society); (b) the electrolytic TiNi for the electrolysis of mixed oxides between TiO2 and NiO in molten CaCl2 at 1173 K (adapted with permission from Springer); (c) the electrolytic Ti from the electrolysis of CaTiO3 in molten CaCl2 at 1123 K (adapted with permission from John Wiley) and (d) the electrolytic W from the electrolysis of CaWO4 in molten CaCl2–NaCl at 1023 K (adapted with permission from The Electrochemical Society).

(4)) has been confirmed in the electrochemical reduction of solid Nb2O5 and TiO2 in molten CaCl2. The stepwise routes following the transformations of Nb2O5–NbO2–NbO15 and TiO2–Ti3O5– Ti2O3–TiO16,17 have been demonstrated, respectively. In addition, TaO was observed as the intermediate oxide during the electrolysis of solid Ta2O5.18 The formation of the intermediate highly conducting Magnelli phases (TiO2x) during the electrochemical reduction of TiO2 in molten CaCl2 was proposed and confirmed.2 The formed Magnelli phases allow the passage of electrons and therefore assist in improving the energy efficiency of the electrolysis. MOx(s) + ae - MOxa/2(s) + a/2O

2

MOxa/2(s) + (2x a)e - M(s) + (x a/2)O2

(5)

As exhibited in Fig. 4, the voltammetric (CV) plot of the TiO2-MCE in molten CaCl2 (counter electrode: graphite rod; reference electrode: mullite-sheathed Ag/Ag+ (ref. 13)) shows four well-defined reduction peaks more positive than the calcium metal deposition (indicated as C5), which indicate the four electrochemical reduction processes in the reduction of titanium oxide. Together with the analysis of the intermediate products, the cathodic behavior of solid titania can specified. In particular, it was demonstrated that the reduction peaks C1 and C4 should be attributed to the formation of low-valence Ti–O (Reaction (4a)) and the further reduction of low-valence Ti–O to metallic Ti (Reaction (5a)).20 TiO2(s) + ae - TiO2a/2(s) + a/2O2

(4a)

TiO2a/2(s) + (4 a)e - Ti + (2 a/2)O2

(5a)

(4)

The porosity of the solid oxide cathode during the electrochemical reduction should be considered since larger electrode porosities allow a faster O2 diffusion and lower solution resistance. The metal-to-oxide molar volume ratio should be considered to optimize the porosity of the feedstock. If the metal-to-oxide molar volume ratio is close to unity, the porosity of the generated metal layer is too small to provide an efficient diffusion path for O2. Actually, the metal-to-oxide molar volume ratio of Ti/TiO, is very close to unity (0.91) and this high ratio was considered as an intrinsic barrier to the solidstate reduction of TiO2 to Ti.19 This kinetic barrier then can be overcome or bypassed by utilizing a highly porous precursory pellet. It was recently suggested that when using a TiO2 pellet with porosities larger than 50%, the electro-deoxidation rate was drastically accelerated in comparison with parallel experimental runs carried out using a less porous TiO2 pellet.19 It is worth noting that the Ti is highly-sintered at high temperatures.19,20 The sintering of metals causes the shrinkage and the decreased porosity of the pellets, which should be considered when designing


the cathode. It was also proposed that the electro-deoxidation dense solid TiO2 in molten CaCl2 might follow the less thermodynamically favorable but effectively more direct path of TiO2– Ti2O3–TiO.17 The incorporation of the calcium species into solid TiO2 tends to block the O2 diffusion channels. However, the calcium-incorporation reactions in the less-porous solid TiO2 cathode can be alleviated.17 The intermediates might be short-lived during electrolysis. Although some intermediates are stable if the applied potential is removed, the extraction of high-purity intermediate oxides upon the molten salt electrolysis of bulk oxides remains a challenge yet. This is because the electro-deoxidation process is generally accompanied by other reactions (e.g. the electroinclusion of cations from the melts) due to the overlapping of oxygen ionization and the cation incorporation in thermodynamics and kinetics.1 The intermediate oxides generated can be totally electro-reduced to metals by prolonging the duration of reduction or by carrying out electro-reductions at larger overpotentials via Reaction (5).

Fig. 4 A typical CV plot of the TiO2 powder loaded metallic (Mo) cavity electrode (MCE) in molten CaCl2 at 1123 K recorded at 50 mV s1.

Chem. Soc. Rev.

View Article Online


Tutorial Review

Chem Soc Rev

The continuous or stepwise electro-deoxidation process also occurs in the molten salt electrolysis of mixed solid oxides. The inevitable variation of the decomposition voltages among the different oxide components makes it preferential for the initial reduction of the less thermodynamically stable oxides. The initially generated metal dispersed through the whole cathode not only improves the electrical conductivity of the whole cathode, but also facilitates the formation of more electrochemical active sites (TPB). This suggests that the initially generated metal can function as a depolarizer to enhance the fast electrochemical reduction of other oxides. The merit of more TPB in the cathode was demonstrated by an investigation on the extraction of a TiNi alloy from the electrolysis of mixed oxides in molten CaCl2.21 It was observed that Ni was first generated at the initial stage of electrolysis and the full electrometallization or the formation of the TiNi alloy was achieved within 4 h, whilst the TiO2 pellet of the same size and structure required around 10 h. The optimal current efficiency is thus higher than 80%, with an energy consumption as low as 6 kW h (kg-TiNi)1.21 As shown in Fig. 3b, the interconnected TiNi nodules were electro-generated and such a morphology is similar with that of the electrolytic Ni (Fig. 3a). More TPB in the oxide cathodes can also be achieved by mixing conducting agents with oxides to fabricate cathode preforms before electrolysis. In this case, the penetrating depth for complete electrolysis can be defined as the diameter of the oxide grains. Such a short oxygen diffusive path remarkably speeds up the reduction and therefore decreases the energy consumption of electrolysis. It should be noted that the added conducting agents should be prudently chosen in case of the formation of undesired products. Nohira et al. investigated the effect of Si addition on the electrochemical reduction of a porous SiO2 pellet in molten CaCl2.22 A remarkable acceleration in terms of the reduction rate was achieved by the addition of Si powder to the porous SiO2 pellet as a conductive additive. In this case, the Si powder, which possesses a merely moderate electrical conductivity, was selected to function as a conducting agent because the added Si introduces no impurities to the desired products (Si).22 2.3

Simultaneous electro-deoxidation and de-calcification

If the calcium-enriched oxides are thermodynamically more stable than the corresponding metal oxides, they can also be generated via the chemical combination reaction between CaO in molten CaCl2 and metal oxides, as shown by Reaction (6). The formation of calcium-enriched oxides via a combination reaction has been confirmed in the cases of the electrochemical reduction of solid oxides such as TiO2,20 Nb2O515 and Ta2O518 in molten CaCl2. MOx + dCaO - CadMOx+d

(6)

The in situ generated calcium-enriched oxides then can be further electro-reduced to metals via either a combination of Reaction (7) and Reaction (8) or via Reaction (9), if CaO is

Chem. Soc. Rev.

thermodynamically more stable than the calcium-enriched oxides. CadMOy + 2( y z d)e - MOz + dCaO + ( y z d)O2 (7) MOz + 2ze - M + zO2

(8)

CadMOy + 2(y d)e - M + dCa2+ + yO2

(9)

It should be noted that calcium incorporation can also be achieved via an electrochemical route in addition to the above combination reaction route. The electro-inclusion route will be discussed later. The in situ generated calcium-enriched oxides produced generally exhibit larger molar volumes than oxides.20 This means that the generated calcium-enriched oxides tend to block the diffusion path of O2, causing higher diffusion polarizations and an inferior energy efficiency. Also, the perovskite phases (CadTiOx, d r 1, x r 3, typically CaTiO3 and CaTi2O4) are thermodynamically more stable than the simple oxide (TiOy) phases and hence require a more negative potential to become reduced. However, the in situ generated perovskites upon the electrolysis of the bulk solid TiO2 is inevitable, which accounts for the low efficiency of Ti extraction from TiO2. Alternatively, ex situ ‘‘perovskitisation’’ can be used to address this negative effect by using porous CaTiO3 instead of TiO2 as the feedstock.20 Ex situ ‘‘perovskitisation’’ is a process for converting TiO2 to CaTiO3 by a reaction with CaO at elevated temperatures. The decreased porosity and then restricted O2 diffusion in the electrolysis of TiO2 stemming from the in situ generated perovskites is expected to be avoided in the electrolysis of porous CaTiO3. It was found that the electrolysis of porous CaTiO3 for the preparation of a Ti sponge (see Fig. 3c) took a much shorter time, leading to an increase in the overall current efficiency that was almost double the original value (about 28%), without compromising the product quality.20 As shown in Fig. 4, the reduction peaks C3 can be attributed to electrochemical reduction of perovskite to Ti–O (Reaction (7a)).20 CadTiO2a/2 + 2(2 a/2 z d)e - TiOz + dCaO + (2 a/2 z d)O2

(7a)

The Reactions (7), (7a) and (9) suggest a simultaneous electro-deoxidation and de-calcification of the calciumenriched oxides in molten CaCl2. Such a mechanism was also confirmed in the direct electrochemical reduction of CaWO4 (scheelite) into fine tungsten powder in molten chlorides. For the electrolytic tungsten from solid precursors, scheelite is a more proper feedstock than solid WO3 due to the significant loss of WO3 from the cathode due to the formation of highly volatile WO2Cl2 in molten CaCl2.23,24 The simultaneous electrodeoxidation and de-calcification of scheellite in molten CaCl2 for the electrolytic preparation of tungsten can be expressed as Reaction (9a). As displayed in Fig. 3d, the electrolytic W is of nanoparticles, indicating that the molten salt electrolysis can be a refining process. Such a phenomenon shall be discussed later. The successful preparation of the Ti–W alloy from mixed TiO2 and WO3 was recently reported.25 It was reported that the


View Article Online

Chem Soc Rev

Tutorial Review


reduction of the mixed oxides initiated with the rapid reduction of WO3 to a fine W–Ti particulate. Therefore, the loss of WO3 in molten CaCl2 can be alleviated.25 This indicates that alloying process can be an effective approach to address the loss of the feedstock. This function can be extended to immobilize liquid metals, as discussed in Section 2.4. CaWO4 + 6e - W + Ca2+ + 4O2

(9a)

2.4 The immobilization of active liquid metal with the formation of alloys upon the electro-deoxidation of mixed oxides The extraction of low-melting-point metals such as Ga (m.p. 303 K), Sn (m.p. 505 K) and Al (m.p. 933 K) via the molten salt electrolysis of the individual solid oxides might be problematic due to the consistent loss of metals from the cathode since the produced metal is liquid in molten salt electrolytes. The immobilization of the liquid metals can be enhanced by the utilization of mixed electrolytes with lower operation temperatures, and also by the electro-deoxidation of mixed oxides for the direct extractions of alloys with desired stoichiometries. NiMnGa,26 Nb3Sn,27 and Ti–6Al–4V28 alloys were successfully electrochemically produced via the molten salt electrolysis of oxide mixtures. The formation of alloys can immobilize liquid metal elements with low melting points, leading to the generation of solid alloy products with desired stoichiometries. The formation of Al-alloys occurs at potentials more positive than that of individual Al.28 The exothermic nature of the alloy process provides additional advantages on enhancing the kinetics of electro-deoxidation. 2.5

Cathodic passivation

The applied electrode potential (or cell voltage) has a direct influence on the reduction rate and the energy efficiency of the electrolysis process. Using electrolytes with a high solubility of oxide ions is imperative for the onset of the oxygen ionization mechanism. Molten CaCl2 is a suitable state-of-the-art electrolyte for the FFC-Cambridge Process, due to its relatively low cost, low toxicity and high solubility for O2. The saturation of CaO in CaCl2 at 1173 K occurs at about 19 mol%.8 Large cathodic overpotentials during electrolysis are favourable for a fast oxygen ionization rate. However, an excessively large overpotential may lead to O2 saturation, which causes the precipitation of CaO on the porous cathode surface.8 In such cases, the reduction mechanism is a combination between an oxygen ionization (Reaction (2)) and CaO precipitation mechanism (as shown by Reaction (10)). The precipitation of CaO in the cathode incurs the deteriorated purity of the final product, the decreased conductivity of the cathode and the blocked diffusion path of O2 in the cathode. Thus, the electrolysis potential, porosity and thickness of the cathode should be prudently controlled to ensure the absence of CaO precipitation. It was observed that the reduction current decreased sharply during the electrolysis of quartz in molten CaCl2 at potentials more negative than 0.90 V (vs. quartz sealed Ag/AgCl).29 This anomalous phenomenon, called cathodic passivation, is ascribed to the precipitation of CaO.8


MOx + xCa2+ + 2xe - xCaOk + M

(10)

It should be noted that cathodic passivation originates from O2 diffusion barriers in reaction zones. It is believed that such a scenario could become absent during the electrolysis of highly porous feedstock even at large overpotentials. The influence of the applied potentials on the current efficiency and energy consumption (electrolysis only) of the electrolysis of quartz in molten CaCl2 was highlighted by the derived volcanic-profile correlation based on the penetrating TPB model.8 The combined effects of the reduction kinetics and the residual current of the molten salts account for such a volcanic profile. The enhanced reduction kinetics occur upon electrolysis at higher overpotentials, rendering an increased reduction rate of electrolysis at higher overpotentials. However, the residual current of the molten salts, which is discussed in detail in Section 6, increases with the increased overpotentials. Therefore, the decreased energy efficiency appears in the electrolysis at excessively high overpotentials. It is clear that increasing the thickness of the cathode causes a decreased current efficiency and an increased energy consumption. The derived volcanic-profile variation of both the current efficiency and the energy consumption against the overpotentials indicates that there might be an ‘‘optimal potential’’ at which the reaction rate, current efficiency and energy consumption all reach their optimal values. For instance, for the electrolysis of a quartz pellet with a depth of 1.0 mm, the optimal potential appears to be between 0.80 and 0.90 V (vs. quartz-sealed Ag/AgCl),29 which is most likely the potential at which the reduction generated O2 ion accumulates to saturation in molten CaCl2. Correspondingly, the optimal current efficiency is higher than 85%, with a value of the energy consumption as low as 10 kW h (kg-Si)1.8 2.6 The direct electrochemical reduction of solid compounds other than oxides In addition to solid oxides, the direct electrochemical reduction mechanism which is a typical second-type electrode process can be extended to the electrochemical reduction of solid compounds other than oxides. The direct electrochemical reduction of solid metal sulphides (e.g. MoS2) in molten chlorides was confirmed,30 in which the direct electrodesulfidation occurring in the solid-metal-sulphide cathode facilitates the generation of metals. Interestingly, the carbonbased anode was found to be an effective inert anode in the molten salt electrolysis of solid metal sulphides, rendering the occurrence of electro-splitting the solid metal sulphides into metals at the cathode and elemental sulphur at the anode.

3. The deposition of the active metal together with electrochemical reduction of solid oxides In practice, molten CaCl2 inevitably contains some calcium oxide (CaO) since CaCl2 has a hygroscopic nature and thus tends to absorb moisture. The presence of moisture incurs the

Chem. Soc. Rev.

View Article Online


Tutorial Review

Chem Soc Rev

hydrolysis of CaCl2 leading to the formation of CaO in the melt. Even in prudently pre-treated CaCl2 (e.g. treated with hydrochloric gas), ionized oxygen (O2) that is generated from the direct electro-deoxidation of solid oxide transports in molten CaCl2. Hence the existence of dissolved CaO in molten CaCl2 is inevitable. The decomposition voltage of CaO is comparable to those of many oxides such as Tb2O3 and La2O3 (Fig. 1). This indicates that a cell voltage higher than the decomposition voltage of CaO is required, in order to bring about the direct electrochemical reduction of the above oxides. On the other hand, a higher cell voltage than the decomposition voltage of CaO should be employed to overcome the kinetic barriers arising from electrochemical, ohmic and concentration polarizations, even for the effective electrochemical reduction of oxides whose decomposition voltages are much lower than that of CaO. In both cases, the deposition of the active metals (either from molten salts or oxide cathodes) is present during the molten salt electrolysis of solid oxides. 3.1 The deposition of the active metal followed by metallothermic reduction The electro-metallothermic reduction, namely the Ono–Suzuki (OS) process is also a possible mechanism under cathodic polarizations to achieve the reduction of precursor oxides in molten salts.31–34 Using Ca as an example, the electro-calciothermic reduction mechanism can be represented by the initial electrodeposition of metallic Ca (Reaction (11)) and the following calciothermic reduction of oxides (Reaction (12)). xCa2+ + 2xe - xCa(l)

(11)

MOx + xCa(l) - M + xCaO

(12)

There have been disagreements over the kind of mechanism that is dominant while electrolyzing solid oxides at a voltage higher than the decomposition voltage of CaO, especially for the reduction mechanisms of solid TiO2 and rare earth metal oxides.35 The different reduction kinetics between direct electro-deoxidation and electro-metallothermic reduction could provide some indication of the real reduction mechanism, although both mechanisms are thermodynamically possible under cathodic polarizations. Assuming Reaction (11) to be the faster step of the two, the Ca activity in the molten salt near the cathode would follow the cathode potential according to the Nernst equation. This, in turn, indicates that the current flow on the cathode would be only affected by the cathode potential, disregarding the nature of the metal oxide on the cathode. This is contrary to the findings observed in the cyclic voltammograms of various metal oxides, which were found to be very different.14,20,35 Also note that in this case, the Ca activity corresponding to the electrode potential via the Nernst equation is forced to unity considering the deposition of a distinct liquid Ca phase in the melts. Consequently, when the potential changes to a more negative potential, Reaction (12) should proceed at a constant rate. This contradicts the reported results on the reduction of solid Tb4O7 in molten CaCl2 suggesting thereby a higher

Chem. Soc. Rev.

reduction rate with an increased voltage, ranging from 3.1 to 3.5 V.35 Assuming that Reaction (12) is the faster step or at least that it occurs at the same rate as Reaction (11), the Ca would be consumed immediately by Reaction (12) upon production from Reaction (11). Hence, Ca is effectively absent at the cathode, and the combination of Reaction (11) and Reaction (12) leads to Reaction (13). xCa2+ + MOx + 2xe - M + xCaO

(13)

Considering that CaO will dissociate in the molten salt according to the following reaction. xCaO - xCa2+ + xO2

(14)

Introducing Reaction (14) to Reaction (13) results in the simple Reaction (2), which indicates the occurrence of the oxygen ionization mechanism in that case. It should be noted that although both Reaction (13) and Reaction (10) have very similar forms, the physicochemical implications of the two reactions are dissimilar. Only at potentials more negative than the decomposition potential of CaO can Reaction (13) occur, whilst a potential more negative than the decomposition potential of CaO is not essential for the occurrence of Reaction (10). The occurrence of Reaction (10) is ascribed to the sluggish kinetics of the O2 diffusion through the porous bulk cathode.8 Both direct electro-deoxidation and electro-metallothermic reduction mechanisms are thermodynamically possible during the electrochemical reduction of solid oxide in molten salts under large cathodic polarizations. For bulk electrolysis, the former requires a feedstock in well assembled and integrated forms (e.g. oxide pellets sandwiched by current collectors) due to the compulsory physical connection between the feed and reductant (electrons). While in situ electro-generated reductants (liquid metals) can be well dispersed throughout melts in the latter, rendering that the surfaces of every grain of feed oxide can be effective reduction sites. Correspondingly, the propagation paths to a complete reduction in the latter are determined by the grain sizes of individual oxide particles and are significantly shorter than that in the former determined by the depth of the oxide pellets. In this sense, the electro-metallothermic reduction is more promising in terms of space-time yield, given the comparable thermodynamic barriers on the electrogeneration of liquid metals and direct electro-deoxidation. The electro-metallothermic reduction was reported to be predominant during the removal of oxygen from rare-earth metals containing ca. 2000 mass ppm oxygen due to the higher thermodynamic barriers of the direct electro-deoxidation of rare metal–oxygen solid solutions.36 Calciothermic reduction and direct electro-deoxidation were elaborately connected in the reduction of solid Nb2O5 to Nb powder in molten CaCl2 by an electronically mediated reaction (EMR) using liquid Ca as the reductant.7 The EMR can also be arranged in a two-compartment configuration,37 in which the electrodeposition of the liquid Ca metal occurs in a molten salt


View Article Online

Chem Soc Rev

Tutorial Review

electrolysis (MSE) chamber and solid oxide is reduced to metal (Reaction (2)) via an EMR process appears in an EMR compartment.


3.2

Electro-deoxidation under active metal deposition

The above discussions indicate that the electro-calciothermic reduction mechanism is not mandatory to the reduction of active oxides. Therefore, oxides such as TiO2, ZrO2 and rare earth metal oxides can be completely reduced via the oxygen ionization mechanism.35 Also note that the in situ generated Ca does not necessarily ensure the occurrence of the calciothermic reduction of solid oxides due to thermodynamic or kinetic barriers.35 In such a scenario, the formation of calcium is a side-reaction for the reduction of solid oxides. The oxygen ionization path should be one of the possible mechanisms that govern the reduction of solid active oxides in molten salts even at the potentials more negative than that for the decomposition of CaO. The decomposition voltages of the active solid oxide such as lanthanide- and actinide-oxides are generally comparable to those of the decomposition of CaO. Therefore, both the electrometallothermic reduction and direct electro-deoxidation mechanisms can be present processes governing the reduction of such solid oxides upon molten salt electrolysis. Such molten salt electrolysis is applicable and hence important for not only the mass production of active metals (e.g. Tb)35 but also for the recovery of spent oxide nuclear fuels (U38 and Ce4). Upon electrolysis in molten Li2O–LiCl, the reduction of solid U3O8 involves both the above mechanisms.38 While the direct electrodeoxidation mechanism is believed to be the predominant process governing the reduction of CeO2 in molten LiCl.4 Again, it is acknowledged that there have been disagreements over the reaction mechanism at large overpotentials. The reaction mechanism is dependent on both thermodynamics and kinetics, which should be considered case-by-case and deserve further investigation. 3.3

Underpotential deposition of active metals

For those metal–calcium binary systems with thermodynamically stable intermetallics, intermetallic compounds between metals and calcium (from molten salt) can possibly be generated during the electrochemical reduction of solid oxides in molten CaCl2 via Reaction (15). Such underpotential deposition of active metal upon molten salt electrolysis is promising for the preparation of alloys/intermetallic compounds. M + xCa2+ + 2xe - CaxM

(15)

Fig. 5 A typical CV plot of the W–SiO2 electrode in molten LiCl at 1123 K recorded at 100 mV s1.

positive than the equilibrium potential of the Ca/Ca2+ couple, but more negative than 0.95 V (vs. quartz-sealed Ag/AgCl, denoted as Ag/AgCl).29 The generated intermetallic Ca–Si compounds possessing poor electrical conductivities and larger molar volumes (densities of CaSi, Si and SiO2 are 2.17, 2.33 and 2.20 g cm3, respectively) not only cause increased ohmic polarization and also block transfer channels for O2 in the cathode, incurring a retarded reduction rate, decreasing the current efficiency and increasing the energy consumption during electrolysis.12 The generated calcium silicides can react with water during a post-rinse process (resulting in the formation of Ca(OH)2, SiO2 and H2) and eventually deteriorate the purity of the desired Si product, which should be avoided for silicon extraction. The underpotential deposition of the Li metal on uranium oxides was observed upon the electrolysis of solid uranium oxides in molten LiCl–Li2O.39 Utilizing a homemade W–SiO2 electrode,12 the cathodic behaviour of solid SiO2 in molten chlorides was investigated. In the CV curve of the W–SiO2 electrode in molten LiCl at 1123 K (counter electrode: graphite rod; reference electrode: quartz-sealed Ag/AgCl electrodes29) shown in Fig. 5, the direct electro-deoxidation of solid silica to silicon renders the shoulder C1 (depicted as Reaction (2b)). With the prolonging of the cathodic scan, electrochemical reduction of Li+ from the melt occurs at the in situ generated Si (C2), resulting in the formation of Li–Si alloys (Reaction (15a)). In the following cathodic scan, the electrodeposition of the Li metal incurs the formation of C3. SiO2(s) + 4e - Si(s) + 2O2

(2b)

Si + xLi + xe - LixSi

(15a)

+

Thermodynamic calculations demonstrate that the Ca–Si can form at potentials more negative than Ca/Ca2+.12 This thermodynamic finding indicates the importance of controlling the electrode potential for the production of high purity silicon. It was suggested by the cyclic voltammograms of a W–SiO2 electrode and potentiostatic electrolysis of porous silica pellets in molten CaCl2 that the reduction of Ca2+ over the reductiongenerated Si (Reaction (15)) can occur at potentials more


4. The electro-inclusion of cations from the molten salt During the electrochemical reduction of solid oxides, electrons are introduced at the cathode. Because of the electric neutrality, the injection of electrons into a solid oxide, at the cathode, can

Chem. Soc. Rev.

View Article Online


Tutorial Review

Chem Soc Rev

lead to either the removal of oxygen anions or the uptake of metal cations from the electrolyte into the oxide phase. The latter reaction involving the insertion of cations (using Li+ as an example) into solid phases is similar to the intercalation mechanism occurring in lithium ion batteries and can thus be presented as Reaction (16).40 The successful preparation of stoichiometric LiTi2O4 and LiTiO2 utilizing the electrochemical reduction solid TiO2 in molten LiCl has been demonstrated recently, which opens a new avenue for synthesizing inorganic functional materials.40 MOx + dLi+ + de - LidMOx

(16)

The electro-inclusion of cations from molten salts was also observed during the electrolysis of solid TiO2 in molten KCl. Fig. 6 shows the CV of MCE-TiO2 in molten KCl at 1173 K recorded at 100 mV s1. (Counter electrode: graphite rod; reference electrode: quartz-sealed Ag/AgCl electrodes.29) The peak pair of a 0 and a can be assigned to the electrochemically intercalation and de-intercalation of K+ from the melt into/out TiO2. Upon the constant-voltage electrolysis of solid TiO2 for 900 min in molten KCl with a graphite anode, one-dimensional K3.2Ti8O16 (inset of Fig. 6) was successfully prepared, as represented by Reaction (16a). 8TiO2 + 3.2K+ + 3.2e - K3.2Ti8O16

(16a)

As exhibited in Fig. 4, the reduction peak C2 should be attributed to the formation of perovskite via the electroinclusion mechanism, as depicted as Reaction (16b).20 TiO2a/2 + dCa2+ + 2de - CadTiO2a/2

(16b)

The formation and reaction mechanism of different perovskites during the electrolysis of solid TiO2 in CaCl2 is quite complicated. Correspondingly, in situ characterization methods, e.g. synchrotron

X-ray diffraction technique were applied to specify the chemistry of perovskites.41 It was demonstrated that different perovskites such as CaTiO3 and CaTi2O4 appear due to the simultaneous electrodeoxidation and electro-inclusion of Ca2+. Furthermore the generated perovskites can disproportionate into low-valence TiOx and other perovskites. The removal of CaO from present perovskites (simultaneous de-oxidation and de-calcification) was finally achieved before the completion of the reduction.41 The electro-deoxidation process generally accompanies the electro-inclusion of cations from the melts due to the overlapping of oxygen ionization and the cation incorporation in thermodynamics and kinetics.1 The electro-inclusion of the cations becomes predominant when the transportation of O2 appears kinetically sluggish, as evidenced by the presence of less perovskites in the more porous TiO2 cathode upon electrolysis in molten CaCl219 and the predominant formation of Li–Ti–O or K–Ti–O upon the electrolysis of TiO2 in molten LiCl40 and KCl (as shown in Fig. 6) that have an inferior ability to CaCl2 on transporting O2. In addition, the host oxide with lattice voids for the accommodation of cations might be favourable for the kinetics of electro-inclusion of cations, similar to the reversible Li+ intercalation–deintercalation in/ out layered LiCoO2 in lithium ion batteries. In particular, the partially reduced oxides are supposed to feature different or anomalous properties in comparison with their stable counterparts. It was reported that nanosized TiO2x–CadTiOx (TCT) composites prepared via the electrochemical reduction of solid TiO2 in molten CaCl2 are excellent catalyst supports for Pt and the corresponding Pt/TCT showed promising electro-catalytic activities on the complete 8-electron oxidation of borohydride without hydrogen evolution in a potential region negative to the reversible hydrogen electrode42 or for the enhanced catalytic oxidation of CO and CH3OH.43

5. Dissolution–electrodeposition process 5.1

Fig. 6 A typical CV plot of the TiO2 powder loaded metallic (Mo) cavity electrode (MCE) in molten KCl at 1173 K recorded at 100 mV s1. The inset shows an SEM image of K3.2Ti8O16 prepared via electrolysis in molten KCl at 1173 K for 900 min at a constant voltage of 2.4 V between a TiO2 pellet and a graphite anode.

Chem. Soc. Rev.

The solubility of the oxides in molten CaCl2

The electrochemical redox mechanisms of solid compounds are highly relevant to their solubility in electrolytes. The basic electrode process of solid PbSO4 with a solubility product (Ksp) of 1.6 108 in aqueous H2SO4, which mediates the reversible charge–discharge process of lead-acid batteries, involves a dissolution–electrodeposition mechanism. The solubility of several oxides (NiO, Fe2O3, Cr2O3, Co3O4, CaWO4 and NiFe2O4) in molten CaCl2 was measured as the saturated concentrations of the corresponding metal cations in the melts with saturated oxides. The saturated concentration of metal cations can be translated to the solubility of oxides by counting the mass of oxygen.13,44 It was determined that the saturated concentrations of the cations from such oxides in molten CaCl2 fall in a range between 102 and 104 M, which are much higher than that of the lead cations from PbSO4 in aqueous H2SO4 (roughly 105 M). A sharp increase in the solubility of NiO and Fe2O3 can be observed with increasing temperatures, as shown in Fig. 7a.


View Article Online


Chem Soc Rev

Tutorial Review

Therefore, the effect of the CaO concentrations in molten CaCl2 on the solubility of the oxides should be considered. As shown in Fig. 7b, the solubility of NiO decreases when the concentration of CaO is increased, while the solubility of Fe2O3 increased sharply when the concentration of CaO was increased. The phase diagrams suggest that Fe2O3 can combine with CaO to form CaFe2O4; whilst such a combination can hardly occur between NiO and CaO. The generated CaFe2O4 tends to dissolve in the melts and thus increase the solubility of Fe2O3.44 It is believed that the acidity/alkalinity of the oxide is relevant to the tendency of solubility against the CaO concentration, in which the solubility of the acidic oxides, e.g. Fe2O3 increases with CaO concentration whereas the solubility of the alkaline oxides, e.g. NiO decreases. Such an assumption agrees well with our recent results showing that the solubility of acidic silica in molten CaCl2 increased with the CaO concentration.45 5.3 Verification and the implications of the dissolution– electrodeposition process

Fig. 7 Saturated concentrations of Fe cations (notated as black squares) and Ni cations (represented as red triangles) in molten CaCl2 at different temperatures (a) and in 1123 K molten CaCl2 with the addition of different amounts of CaO (b).

Such a positive variation of solubility against temperature was also reported in solid CaWO4.13 The observed considerable solubility of the oxide suggests that dissolution–electrodeposition probably exists in parallel with the well-documented direct electrodeoxidation mechanism during the molten salt electrolysis of solid oxides. The reaction between oxides and molten CaCl2 should also be considered. It was reported that WO2Cl2 can be generated in molten CaCl2 via the reaction between solid WO3 and molten CaCl2.23 5.2 The effect of the CaO content on the solubility of the oxides in molten CaCl2 The all oxide feeds discussed in the manuscript are those do not significantly react with CaCl2 (with the formation of metal oxychloride/chloride). It is believed that the molten salt electrolysis of oxides capable of reacting with molten salts (e.g. extraction of W from WO3) is of very limited practical meaning due to the consistent loss of oxide feed and the consumption of molten salt. It should be noted that the formation of metal oxychloride/chloride tends to be less favourable when the concentration of CaO is increased. As discussed before, ionized oxygen appears at electrochemical interfaces and transports in molten CaCl2, suggesting that the presence of CaO in TPB and melts is inevitable.


A solid-to-solid direct electrochemical reduction mechanism was proposed to address the electrochemical reduction of solid silica in molten CaCl2 as expressed in Reaction (2b) and in Fig. 5 (C1), in which solid SiO2 is directly electro-reduced to solid silicon.8–10 It was estimated that the concentration of in situ generated O2 at the electrochemical interfaces during the electrochemical reduction of solid silica in molten CaCl2 could be higher than 1 mol L1.8 Such a high concentration is believed to be adequate to trigger the formation of silicates in the silica cathode since the Gibbs free energy change of Reaction (6a) at 1123 K is 139.76 kJ mol1.45 SiO2(s) + CaO - CaSiO3(s)

(6a)

Therefore, the presence of CaSiO3 at the cathode is inevitable during the electrochemical reduction of solid silica in molten CaCl2, regardless of the addition of CaO in the melt. The presence of silicates in the cathode during the course of the electrochemical reduction of solid silica in molten chlorides was recently proposed and confirmed.6,45–48 If the solubility of CaSiO3 in the melt is remarkably high, silicon could also be produced through the electrodeposition of the dissolved silicate as described below. CaSiO3(s) - Ca2+ + SiO32

(17)

SiO32 + 4e - Si(s) + 3O2

(18)

The solubility of CaSiO3 in molten CaCl2 was recently measured to be 1.56 wt% at 1123 K and the occurrence of the above dissolution–electrodeposition mechanism has been proved to be present in the reduction of solid silica in molten chlorides.45 The CVs of the Mo electrode in molten CaCl2 with the addition of SiO2, CaO and CaSiO3 are displayed in Fig. 8 (counter electrode: graphite rod; reference electrode:13 mullitesheathed Ag/Ag+). The potential was reported with respect to the equilibrium potential of the Ca/Ca2+.45 As can be seen, the CV of the Mo electrode (plot a) in molten CaCl2 with the

Chem. Soc. Rev.

View Article Online


Tutorial Review

Chem Soc Rev

Fig. 9 Typical SEM images of the electrolytic products generated by the potentiostatic electrolysis of (a) a silica pellet for 10 h at 0.60 V (vs. quartz sealed Ag/AgCl) in molten CaCl2 at 1173 K (adapted with permission from The Royal Society of Chemistry) and (b) a GeO2 pellet for 12 h at 1.35 V (vs. mullite-sheathed Ag/AgCl) in molten CaCl2–NaCl at 873 K (adapted with permission from Elsevier).

Fig. 8 The CVs of the Mo electrode in (a) molten CaCl2 (434 g) + 0.1 mol SiO2, (b) CaCl2 + 0.1 mol SiO2 + 0.1 mol CaO and (c) CaCl2 with the addition of 0.051 wt% CaSiO3 at 1123 K recorded at 500 mV s1. Adapted by the permission of The Royal Society of Chemistry.

addition of 0.1 mol silica only shows the peaks ascribed to the formation and re-dissolution of Ca. While another cathodic peak (onset at 0.75 V vs. Ca/Ca2+) appears in plot b with the consecutive addition of 0.1 mol CaO, ascribed to the redox reactions of the dissolved silicates. Such a cathodic peak reoccurs in the CV (plot c) of the Mo electrode in molten CaCl2 with 0.051 wt% dissolved CaSiO3, suggesting the existence of the dissolution–electrodeposition mechanism during the electrochemical reduction of solid silica in molten CaCl2. It is also shown that the electrodeposition of silicon from dissolved silicates started at a potential of 0.75 V (vs. Ca/Ca2+), which is far more positive than the formation potential of Si–Ca intermetallics. The above results confirm that both direct electrodeoxidation and dissolution–electrodeposition mechanisms are possible during the molten salt electrolysis of solid silica in molten CaCl2. The two mechanisms will simultaneously occur during electrolysis, in which dissolution–electrodeposition tends to be more evident when concentrations of O2 in the vicinity of the reaction zones become higher. Intriguingly, it was found that the electrolytic Si and Ge obtained upon the molten salt electrolysis of solid SiO2 and GeO2 were of nanowires or nanoparticles (as exhibited in Fig. 9).5,6 The formation of nanostructured Si or Ge upon the electrolysis of solid oxides in molten chlorides indicates that the molten salt electrolysis of solid oxides is a generic, versatile and template-free approach for the production of nanostructured semiconductors. It was believed that the preparation of nanostructures via molten salt synthesis may be challenging since the present high temperature facilitates overgrowth and the violent aggregation of nuclei or building blocks. Such a situation was verified by the formation of over-aggregated even sintered solid metallic products such as sponge-like Ti,20 Ni14 and TiNi21 generated from the molten salt electrolysis of corresponding solid oxides (see Fig. 3a–c). Metal-based materials contain large amounts of delocalized electrons, which can drive thermal-induced strain. And the thermal-induced strain governed

Chem. Soc. Rev.

by the mobility of the delocalized electron tends to facilitate the agglomeration of generated metallic species. For those with relatively low melting points such as Ti (m.p. 1941 K) and Ni (m.p. 1728 K), the present highly mobile delocalized electrons facilitate the coalescence of the metal species, allowing the formation of interconnected even sintered solid products. For refractory metals, e.g. W (m.p. 3695 K), the mobility of the inside delocalized electrons is highly restrained due to their high melting points, resulting in the significantly retarded agglomeration of the electro-generated metal species and the formation of lessinterconnected nanopowders (Fig. 3d). When it comes to electrolytic Si and Ge, they are merely semi-metals. In short of sufficient delocalized electrons to prompt thermally-induced strain, the aggregation of generated Si/Ge seeds is not accelerated, facilitating thereby the formation of ultrafine products. Such an ultra-refining mechanism was verified and addressed based on the in situ dissolution–electrodeposition mechanism.5,45 The dissolved silicates can be electro-reduced into silicon which is capable of functioning as nucleation centres for the growth of nanostructured silicon. Fewer nuclei are generated upon the electrolysis of porous silica at low overpotentials, facilitating the formation of high-aspect-ratio silicon nanowires. Silicon nanoparticles are produced upon the electrolysis of porous silica at high overpotentials, due to the presence of more nuclei.6,45,49

6. Residual current during molten salt electrolysis Except for the potential range related to the formation and redissolution of Ca, the CV in Fig. 8a exhibits small and sloped currents. Such a sloped and linear (against potential) current can be defined as the residual current of molten salts. After normalization in terms of the apparent electrode area, the background current is suggested to be very small in comparison with the reduction current of SiO2.12 These sloped and linear variations in the residual current are ascribed to the direct conduction of electricity through the molten salt.12 This is in conjunction with the electron hopping mechanism, as described by Reaction (19), in which M can be Ca and/or other multivalent metals and the front subscripts represent the


View Article Online

Chem Soc Rev

Tutorial Review

relative locations of the redox active species between which electron hopping occurs in the electrolyte.12 Because of its electrical nature, the residual current increases correlate with the increasing overpotential.8


AM(I)

+ BM(II) 2 AM(II) + BM(I)

(19)

The residual current becomes more influential when the electrolysis potential is fixed to be more negative than that of the calcium deposition. The deposited Ca contributes to electron transport through the molten salt and then increases the residual current. The increased residual currents incur lower current efficiencies in the electrolysis of oxides such as TiO2, at voltages higher than the decomposition potential of CaO in comparison with those of the electrochemical reduction of oxides such as SiO2 at lower voltages.8,12 It should be noted that when carbon-based anodes are employed in electrolysis, the CO2 generated at the anode can react with O2 to form CO32 ions dissolved in the molten CaCl2. The CO32 ions can diffuse to the cathode, where discharge and carbon deposition occur via Reaction (20).18,50 Such an electrochemical process might account for the negative (cathodic) currents in both the forward and backward reaction in the CV.12 CO32

+ 4e - C + 3O

2

(20)

Then the carbon deposited at the cathode shall contaminate the product and reduce the current efficiency. A recent study on electrolytic Ta showed the existence of carbon among the product, which caused a large leakage current of the Ta capacitor fabricated from electrolytic Ta powder.18 The present carbon oxides originate from both an air leak and generated gas at carbon anodes, and other impurities such as metal species in the molten salts also supply to the residual currents.50 It is acknowledged that the residual current is an inherent property of the molten salts and its elimination remains highly challenging. Decreasing the impurity levels of the molten salt and the atmosphere above the salt by the utilization of pre-electrolysis, well sealing the electrolyzer and supplying a positive inert gas pressure to decrease the impurity levels of the electrochemical bath should be helpful to diminish the residual current and thus decrease the electrolytic energy consumption.

7. Perspective In the present manuscript, the laboratory-scale fundamental research on the cathodic process in the molten salt electrolysis of solid compounds is summarized and the factors influencing the energy efficiency, purity and microstructures of the electrolytic products are discussed. Although a great success on the laboratory-scale has been fulfilled, the molten salt electrolysis of solid compounds is still in its infancy and its commercialization in the short-term still remains invisible. Innovations on materials, electrochemistry and engineering are highly desired. The marriage between the electrochemist and materials and


the collaboration between academic and industrial communities are the prerequisites for fully exploiting the merits. Also, applications for the molten salt electrolysis of solid compounds have been extended from the production of structural materials to functional materials to spent nuclear fuel reprocessing and also from the preparation of bulk materials to the controllable construction of nanostructured materials. Therefore, its value and far-reaching impact are sprawling from the original metallurgical community to the electrochemistry, energy, nuclear, nano and material communities. In particular, the molten salt electrolysis of solid compounds provides unprecedented opportunities on massive and controllable preparation functional materials such as nanostructured semiconductors, hydrogen-storage alloys and functional oxides that cannot be facilely prepared by traditional processes or on the preparation of existing materials with different or anomalous structures. The physicochemical, mechanical and biological properties of the electrolytic products are not fully investigated, which deserves further study.

Acknowledgements The financial support from the NSFC (20873093, 50934001, 51071112, 21203141 and 51325102), MOST (2009DFA62190) and Wuhan University (121075) and the essential contribution from our colleagues and groupmates are acknowledged.

Notes and references 1 A. M. Abdelkader, K. T. Kilby, A. Cox and D. J. Fray, Chem. Rev., 2013, 113, 2863–2886. 2 G. Z. Chen, D. J. Fray and T. W. Farthing, Nature, 2000, 407, 361–364. 3 D. H. Wang, X. B. Jin and G. Z. Chen, Annu. Rep. Prog. Chem., Sect. C, 2008, 104, 189–234. 4 Y. Zhang, H. Y. Yin, S. D. Zhang, D. Y. Tang, Z. W. Yuan, T. H. Yan, W. F. Zheng and D. H. Wang, J. Rare Earths, 2012, 30, 923–927. 5 H. Y. Yin, W. Xiao, X. H. Mao, W. F. Wei, H. Zhu and D. H. Wang, Electrochim. Acta, 2013, 102, 369–374. 6 W. Xiao, X. B. Jin and G. Z. Chen, J. Mater. Chem. A, 2013, 1, 10243–10250. 7 T. H. Okabe, I. Park, K. T. Jacob and Y. Waseda, J. Alloys Compd., 1999, 288, 200–210. 8 W. Xiao, X. B. Jin, Y. Deng, D. H. Wang, X. H. Hu and G. Z. Chen, ChemPhysChem, 2006, 7, 1750–1758. 9 X. B. Jin, P. Gao, D. H. Wang, X. H. Hu and G. Z. Chen, Angew. Chem., Int. Ed., 2004, 43, 733–736. 10 T. Nohira, K. Yasuda and Y. Ito, Nat. Mater., 2003, 2, 397–401. 11 Y. Deng, D. H. Wang, W. Xiao, X. B. Jin, X. H. Hu and G. Z. Chen, J. Phys. Chem. B, 2005, 109, 14043–14051. 12 W. Xiao, X. B. Jin, Y. Deng, D. H. Wang and G. Z. Chen, J. Electroanal. Chem., 2010, 639, 130–140.

Chem. Soc. Rev.

View Article Online


Tutorial Review

13 D. D. Tang, W. Xiao, H. Y. Yin, L. F. Tian and D. H. Wang, J. Electrochem. Soc., 2012, 159, E139–E143. 14 G. H. Qiu, M. Ma, D. H. Wang, X. B. Jin, X. H. Hu and G. Z. Chen, J. Electrochem. Soc., 2005, 152, E328–E336. 15 X. Y. Yan and D. J. Fray, J. Electrochem. Soc., 2005, 152, E308–E318. 16 K. Dring, R. Dashwood and D. Inman, J. Electrochem. Soc., 2005, 152, D184–D190. 17 D. T. L. Alexander, C. Schwandt and D. J. Fray, Electrochim. Acta, 2011, 56, 3286–3295. 18 T. Wu, X. B. Jin, W. Xiao, X. H. Hu, D. H. Wang and G. Z. Chen, Chem. Mater., 2007, 19, 153–160. 19 W. Li, X. B. Jin, F. L. Huang and G. Z. Chen, Angew. Chem., Int. Ed., 2010, 49, 3203–3206. 20 K. Jiang, X. H. Hu, M. Ma, D. H. Wang, G. H. Qiu, X. B. Jin and G. Z. Chen, Angew. Chem., Int. Ed., 2006, 45, 428–432. 21 Y. Zhu, M. Ma, D. H. Wang, K. Jiang, X. H. Hu, X. B. Jin and G. Z. Chen, Chin. Sci. Bull., 2006, 51, 2535–2540. 22 K. Yasuda, T. Nohira, K. Takahashi, R. Hagiwara and Y. H. Ogata, J. Electrochem. Soc., 2005, 152, D232–D237. 23 M. Erdogan and I. Karakaya, Metall. Mater. Trans. B, 2010, 41, 798–804. 24 M. Erdogan and I. Karakaya, Metall. Mater. Trans. B, 2012, 43, 667–670. 25 R. Bhagat, M. Jackson, D. Inman and R. Dashwood, J. Electrochem. Soc., 2009, 156, E1–E7. 26 A. J. Muir Wood, R. C Copcutt, G. Z. Chen and D. J. Fray, Adv. Eng. Mater., 2003, 5, 650–653. 27 X. Y. Yan and D. J. Fray, Adv. Funct. Mater., 2005, 15, 1757–1761. 28 D. Hu, W. Xiao and G. Chen, Metall. Mater. Trans. B, 2013, 44, 272–282. 29 P. Gao, X. B. Jin, D. H. Wang, X. H. Hu and G. Z. Chen, J. Electroanal. Chem., 2005, 579, 321–328. 30 G. M. Li, D. H. Wang, X. B. Jin and G. Z. Chen, Electrochem. Commun., 2007, 9, 1951–1957. 31 R. O. Suzuki, M. Aizawa and K. Ono, J. Alloys Compd., 1999, 288, 173–182.

Chem. Soc. Rev.

Chem Soc Rev

32 R. O. Suzuki, K. Teranuma and K. Ono, Metall. Mater. Trans. B, 2003, 34, 287–295. 33 R. O. Suzuki, J. Phys. Chem. Solids, 2005, 66, 461–465. 34 K. T. Jacob and S. Gupta, JOM, 2009, 61, 56–59. 35 D. H. Wang, G. H. Qiu, X. B. Jin, X. H. Hu and G. Z. Chen, Angew. Chem., Int. Ed., 2006, 45, 2384–2388. 36 K. Hirota, T. H. Okabe, F. Saito, Y. Waseda and K. T. Jacob, J. Alloys Compd., 1999, 282, 101–108. 37 www.okabe.iis.u-tokyo.ac.jp/docs/020822THOFinalversion. pdf. 38 S. M. Jeong, H. S. Shin, S. S. Hong, J. M. Hur, J. B. Do and H. S. Lee, Electrochim. Acta, 2010, 55, 1749–1755. 39 J. M. Hur, S. M. Jeong and H. Lee, Electrochem. Commun., 2010, 12, 706–709. 40 K. Jiang, X. H. Hu, H. J. Sun, D. H. Wang, X. B. Jin, Y. Y. Ren and G. Z. Chen, Chem. Mater., 2004, 16, 4324–4329. 41 R. Bhagat, D. Dye, S. L. Raghunathan, R. J. Talling, D. Inman, B. K. Jackson, K. K. Rao and R. J. Dashwood, Acta Mater., 2010, 58, 5057–5062. 42 K. L. Wang, K. Jiang, J. T. Lu, L. Zhuang, C. S. Cha, X. H. Hu and G. Z. Chen, J. Power Sources, 2008, 185, 892–894. 43 X. X. Shi, X. B. Jin, W. Xiao, X. Hou, H. L. Chen and G. Z. Chen, Chem.–Eur. J., 2011, 17, 8562–8567. 44 H. Y. Yin, L. L. Gao, H. Zhu, X. H. Mao, F. X. Gan and D. H. Wang, Electrochim. Acta, 2011, 56, 3296–3302. 45 W. Xiao, X. Wang, H. Yin, H. Zhu, X. Mao and D. Wang, RSC Adv., 2012, 2, 7588–7593. 46 S. K. Cho, F. R. F. Fan and A. J. Bard, Electrochim. Acta, 2012, 65, 57–63. 47 J. Y. Yang, S. G. Lu, S. R. Kan, X. J. Zhang and J. Du, Chem. Commun., 2009, 3273–3275. 48 E. Juzeliunas, A. Cox and D. J. Fray, Electrochim. Acta, 2012, 68, 123–127. 49 J. Zhao, J. Li, P. Ying, W. Zhang, L. Meng and C. Li, Chem. Commun., 2013, 49, 4477–4479. 50 G. H. Qiu, K. Jiang, M. Ma, D. H. Wang, X. B. Jin and G. Z. Chen, Z. Naturforsch., A: Phys. Sci., 2007, 62, 292–302.


Electrochemical Synthesis of Magnesium Hexaboride by Molten Salt Technique.

The selection of buffer salts for the high-performance liquid chromatography of basic compounds.

Quinoid compounds: high-performance liquid chromatography with electrochemical detection.

The smallest quaternary ammonium salts with ether groups for high-performance electrochemical double layer capacitors.

Rapid mass production of two-dimensional metal oxides and hydroxides via the molten salts method.

Absolute thermodynamic properties of molten salts using the two-phase thermodynamic (2PT) superpositioning method.

Theoretical studies of structure and dynamics of molten salts: the LiF-ThF4 system.

The room temperature electrochemical synthesis of N-doped graphene and its electrocatalytic activity for oxygen reduction.

High-performance Ag-Co alloy catalysts for electrochemical oxygen reduction.

Electro-deposition and re-oxidation of carbon in carbonate-containing molten salts.

Harvesting capacitive carbon by carbonization of waste biomass in molten salts.

Facile and scalable synthesis of Ti5Si3 nanoparticles in molten salts for metal-matrix nanocomposites.

Reactor modelling for electrochemical processes.

Engineering the metathesis and oxidation-reduction reaction in solid state at room temperature for nanosynthesis.

Comparative Toxicities of Salts on Microbial Processes in Soil.

The preparative electrochemical reduction of pteroylpentaglutamate.

Recycling of salts in partition protein extraction processes.

High-pressure injection injury with molten aluminum.

High temperature electrolysis in alkaline cells, solid proton conducting cells, and solid oxide cells.

A novel molten-salt electrochemical cell for investigating the reduction of uranium dioxide to uranium metal by lithium using in situ synchrotron radiation.

Graphite Carbon-Supported Mo2C Nanocomposites by a Single-Step Solid State Reaction for Electrochemical Oxygen Reduction.

Development of double-perovskite compounds as cathode materials for low-temperature solid oxide fuel cells.

Non-enzymatic reduction of organomercurial salts in biological systems.

The phenomenon of formation of prebiological compounds in volcanic processes.