Recent Advances in Electrochemical-Based Silicon Production Technologies with Reduced Carbon Emission

Sustainable and low-carbon-emission silicon production is currently one of the main focuses for the metallurgical and materials science communities. Electrochemistry, considered a promising strategy, has been explored to produce silicon due to prominent advantages: (a) high electricity utilization efficiency; (b) low-cost silica as a raw material; and (c) tunable morphologies and structures, including films, nanowires, and nanotubes. This review begins with a summary of early research on the extraction of silicon by electrochemistry. Emphasis has been placed on the electro-deoxidation and dissolution–electrodeposition of silica in chloride molten salts since the 21st century, including the basic reaction mechanisms, the fabrication of photoactive Si films for solar cells, the design and production of nano-Si and various silicon components for energy conversion, as well as storage applications. Besides, the feasibility of silicon electrodeposition in room-temperature ionic liquids and its unique opportunities are evaluated. On this basis, the challenges and future research directions for silicon electrochemical production strategies are proposed and discussed, which are essential to achieve large-scale sustainable production of silicon by electrochemistry.


Introduction
The growing demand for low-carbon technologies has sparked a important change in global energy consumption, prompting an imminent transition from fossil fuels to renewable and sustainable energy sources [1][2][3][4][5]. In this regard, silicon is one of the most important and sustainable chemical elements due to its abundance in the earth's crust [6,7]. Silicon-based technologies are essential to harvesting and utilizing sustainable energy sources, such as wind, solar, tidal, and geothermal energy [8]. Various silicon-based energy technologies have been developed and used for efficient energy production and fast storage/release. For example, in the photovoltaic (PV) industry, 90% of the global solar cell modules are Si-based [9][10][11]. Additionally, silicon-based nanostructures are considered critical anodic materials for achieving high-energy, long-endurance Li-ion batteries and are also being established in currently niche markets, such as electric cars [12][13][14][15]. There is no doubt that silicon-based technologies have always been and will continue to be the frontier in future development.
Silicon commonly exists in oxide (silica/SiO 2 ) or silicate minerals; thus, sophisticated reduction and purification steps are unavoidable [16]. In this regard, the carbothermal reduction of silica was a chemical innovation of the 19th century and remained the dominant process as the first step in the current silicon production industry. The purity of the silicon obtained by this process is generally between 98% and 99%, called metallurgical grade silicon (MG-Si), which is not pure enough for PV or other semiconductive applications. To prepare higher-purity solar-grade (SoG-Si, 99.9999%, 6 N) and electronic-grade silicon (EG-Si, ≥99.9999999%, 9 N), metallurgical silicon is transformed into silanes that can be easily distilled, separated, and purified. The silanes are chemically reduced and purified via Siemens, Bayer, or fluidized bed reactor processes to ultra-high-purity silicon [17][18][19]. Unfortunately, the carbothermic reduction and chemical refining processes are polluted heavily, and the total heating electricity consumption is enormous. Consequently, developing alternative, environmentally friendly, low-cost, and efficient technologies remains important for silicon extractionrelated research areas. This awareness has led to an interest in molten salt silicon electrochemistry as a promising alternative to conventional processes. The silicon electrochemical extraction strategy has great potential to outperform conventional processes in terms of (a) the utilization of electrons as reducing agents rather than hazardous chemicals, (b) being straightforward as far as infrastructure is concerned, (c) the advantages of designing various silicon components, and (d) the potential to reduce the cost of producing SoG-Si [20][21][22].
Silicon electrochemistry has received tremendous attention in the last 2 decades because of its universal significance and current policy relevance in a landscape where countries around the world are advocating a low-carbon economy (Fig. 1A). Depending on the reaction mechanism, the electrochemical extraction procedure can be classified into electrodeposition and electro-deoxidation ( Fig. 1B and C). Inspired by the Hall-Heroult process for aluminum electrowinning [23], early works focused on the electrolysis of sand in molten oxides and molten fluorides (Fig. 1D). This provided the opportunity to produce pure silicon at a cost comparable to that of aluminum. In 2003, a groundbreaking work demonstrated the direct reduction of solid silica by electrochemical "deoxidation" in molten chloride (CaCl 2 ), launching a new era of silicon production by the electrolytic reduction of sand [24]. Accordingly, chloride salts (including CaCl 2 ) are extensively investigated because of their eco-friendliness, water solubility, and low cost (Fig. 1E). Recently, silicon electrochemistry has shifted its focus toward the controlled fabrication of nanostructured silicon, silicon films, high-purity silicon, and Si components (Fig. 1F). These products offer a potentially cost-effective solution for industrial-scale energy applications because of the various advantages of the synthesis process: catalystfree, template-free fabrication, and inexpensive feedstock. Additionally, ionic liquids offer new opportunities for the extraction of silicon by electrodeposition at low temperatures [25]. By now, several classic and outstanding reviews have been published to summarize significant advances in silicon electrolysis technology from various perspectives, including extraction of SoG-Si [26], fabrication of silicon-based anodes for energy storage [22,27], design of silicon surface structures [20,28], and development and recommendations for practical silicon production [21,29]. However, there is still a lack of a detailed review article with an extensive landscape, including the historical development, the current challenges, and the future pathways.
Here, we systematically review the historical progress and recent advances in the extraction of silicon by electrochemistry. The early silicon electro-reduction processes are briefly reviewed to capture the state of knowledge about half a century ago. The reaction system and mechanism, as well as the achievement of silicon extraction involved in these evolved silicon electrochemical strategies, are further discussed. Considering that CaCl 2based molten salt electrolysis offers a green and cost-effective way to synthesize new materials with various applications of silicon components, we emphatically summarize the electroreduction and electrodeposition of silicon materials in molten chlorides. Among them, the electroreduction strategy relying on solid-solid (SS) phase reaction and silicon electrodeposition process relying on solid-liquid-solid (SLS) reaction can be used to achieve electrochemical extraction and preparation of nanostructured silicon, high-purity silicon, and photo-responsive doped silicon. The electrochemical synthesis of silicides, silicon carbides (SiC), and silicon/carbon (Si/C) composites has been systematically reviewed. Furthermore, the silicon electrodeposition in room-temperature ionic liquids as a promising silicon extraction strategy has also been discussed and evaluated. More importantly, the current challenges and future directions of silicon electrochemistry are proposed and discussed in depth based on the existing process advances, with the expectation of contributing to the sustainable development of the silicon industry.

Input from the electrolytic science and inspired by the Hall-Heroult process
Electrolysis has been applied to obtain various metals since the invention of the first galvanic cells. Interestingly, the first extraction of silicon by electrochemical means dates back to the mid-19th century and has long been under investigation (Fig. 2), according to Elwell and colleagues [30,31]. The first recorded attempt was by Sainte-Claire Deville in 1854, who claimed that silicon was produced by electrolysis of impure NaAlCl 4 . The first silicon produced directly by electrolysis may be attributed to Ullik in 1865. In that experiment, silicon monomers were obtained from the electrolysis of K 2 SiF 6 in molten KF. However, systematic research on the electrosynthesis of silicon was started in the 1930s. It is documented that researchers have carried out electrowinning of silicon in various molten silicates at 800 to 1,250 °C [31]. The maximum silicon content was ~72% due to the high applied potential (which also reduced alkali or alkaline earth metals).
Following these early explorations, in the early 1950s, there was a growing interest in exploring the direct electrolysis of sand oxide to extract high-purity silicon. The extraction of pure silicon products that can be applied owes much to the knowledge gained on the Hall-Heroult process for aluminum electrolysis developed in 1886, and many of the techniques and concepts established for it are still widely utilized to date. The primary feature of this process is the electrolysis of Al 2 O 3 at a temperature above the melting point of aluminum. The used electrodes were made from glassy carbon, while cryolite was the solvent. The obtained aluminum was of high purity (99.999%). According to Faraday's law (Eq. 1), the energy consumption required for 1 kg of aluminum produced can be calculated, where F is Faraday's constant, V is the cell voltage, z is the valence of the metal, A is the atomic weight, and η is the cell efficiency. Substituting z = 3 and A = 27, we can obtain Eq. 2, and a typical cell voltage is about 4 V and η = 90%, so the power required to produce 1 kg of Al by electrolysis of Al 2 O 3 is about 47 MJ kg −1 (or expressed as 13.06 kWh kg −1 ). This value is considered to be widely accepted. Correspondingly, in the case of silicon electrolysis, z = 4 and A = 28 in Eq. 1, so Eq. 3 is obtained, and assuming that the values of V and η for silicon electrolysis are approximately the same as for aluminum electrolysis, 28.5% more electrical energy is required compared to the aluminum electrolysis process, which is 16.7 kWh kg −1 .
SiO 2 used in silicon electrolysis is cheaper and more accessible than the starting material for aluminum electrolysis. This cost reduction via the raw material equals the pretreatment consumption of raw materials for aluminum electrolysis, resulting in a slight difference in cost between silicon electrolysis and aluminum electrolysis. Thus, inspired by the Hall-Heroult process, researchers tried to apply the "sand-to-silicon" electrolysis process to industrial production in 1957 [32]. Cryolite (Na 3 AlF 6 ) was chosen as the molten salt because of its accessibility and success in aluminum electrolysis [33][34][35]. The SiO 2 /Na 3 AlF 6 system was studied in some detail by Monnier's group. Their experimental data showed that 99.9% to 99.99% purity silicon could be obtained at laboratory and pilot scales [36]. However, the project failed to achieve the goal of commercial development due to the precipitation of poorly conductive silicon in solid form on the electrode surface. The electrolytic deposition rate slows as the electrolysis advances, while the tank voltage increases, hindering continuous silicon production [37]. Huggins and Elwell [38] analyzed the limitations of the electrolytic silicon deposition process. They derived a mathematical approximation for the limited deposition rate [38], meaning that the maximum electrodeposition rate is predictable. Almost all systems used for solid electrodeposition had difficulties performing in a stable way and impurity-free at deposition rates above a few tens of micrometers per hour. However, this limitation does not affect liquid-phase electrodeposition, so the Hall-Heroult process for the commercial production of aluminum is feasible precisely because electrolysis is performed at temperatures above the melting point of aluminum. To solve the problem of the solid-phase cathode product after electrolysis, researchers at Stanford University developed (1978 to 1981) a feasible electrolytic "Hall-Heroult" process for the electrodeposition of silicon above its melting point [29]. Mattei et al. [39] conducted a detailed investigation. Silicon close to 4 N purity was obtained by electrolysis of barium silicate melt. The impurities in the product, including Ti, Al, and Fe, mainly originated from the SiO 2 source. However, the relatively high melting point of silicon (1,414 °C) inevitably raised the quality requirement of the equipment and the energy consumption.
Although the "Hall-Heroult" process for silicon did not achieve large-scale application and production, we benefited from gaining knowledge of these studies. Many technologies and concepts developed for aluminum electrolysis are still widely employed today. For example, molten salt electrolytes containing calcium oxide and calcium chloride have been shown to facilitate silica dissolution at temperatures (850 °C) far below the Si melting point, with electrodeposited polysilicon films of sufficient purity for applications in functional PV devices, as discussed in detail in the "Design of electrodeposited crystalline Si films: Photoactive layers and p-n junction".

Electrodeposition of silicon films from fluorosilicate/fluoride-based melts
Besides the electrowinning of liquid silicon at ultra-high temperatures, as discussed above, another pioneering technique proposed by Stanford University's Materials Research Center in the early 1970s was the direct deposition of silicon films in fluorosilicate/fluoride solutes [40], which simplified the tedious process of silicon wafer fabrication. As a by-product of the fertilizer industry, fluorosilicate is a rich and inexpensive silicon precursor for the electrochemical extraction of silicon. There are many types of electrolytes to choose from in the fluorosilicate/fluoride system, among which the most extensively studied are eutectic molten salt mixtures such as LiF-KF, NaF-KF, and LiF-NaF-KF, as they can obtain high-quality silicon films at temperatures well below the melting point of silicon [41,42]. Boen and colleagues [43,44] systematically investigated the effect of cathode substrates on silicon films prepared by electrodeposition in fluorosilicate/fluoride systems. The results show that silver was a suitable substrate for depositing continuous and dense silicon layers due to the ease of silicon nucleation and growing crystalline silicon on its surface. However, considering the high price of silver, the research focus was shifted toward inexpensive substrates such as graphite.
Several mechanisms and theories were suggested regarding silicon electrodeposition in fluorosilicate/fluoride systems. It was proposed that the reduction process is a quasi-reversible reduction of Si 4+ to Si. The charge transfer process is the rate-controlling step during the deposition [45,46]. However, Boen and Bouteillon [43] proposed a 2-step electroreduction mechanism (Si 4+ →Si 2+ →Si) for SiF 6 2− in the ternary eutectic melt Li-NaF-KF. Bieber et al. [47] suggested that the electrodeposition process of K 2 SiF 6 in the NaF-KF molten salt involves transient nucleation and diffusion-controlled growth. Cai et al. [48] investigated the electrochemical reduction and nucleation of SiF 6 2− in LiF-NaF-KF molten salt and concluded a one-step 4-electron mechanism. As shown above, the mechanism concerning the electrodeposition of silicon in fluorosilicate/ fluoride-based systems is still controversial, and further investigation is necessary.
The silicon films obtained in all the early related works failed to exhibit PV effects due to the high corrosiveness of fluoride-based molten salts, which led to difficulties in impurity control. Electrodeposition of K 2 SiF 6 dissolved in chloride-based molten salts (e.g., NaCl-KCl and LiCl-KCl) has also been attempted. However, the solubility of fluorosilicate is very low throughout the electrodeposition process, and the deposited silicon films were discontinuous and poorly smooth [26]. In the last decade, many studies have confirmed that water-soluble KCl-KF as a molten electrolyte can be used to deposit higher-quality silicon films [49][50][51][52]. The advantages are that KF-KCl salts can dissolve K 2 SiF 6 well at high temperatures, and the cooled and cured KF-KCl salts have high solubility in water, making it easy to remove solid impurities adhering to Si deposits by aqueous washing.
Nohira's group systematically investigated and verified that high-quality silicon films could be obtained by electrodeposition from K 2 SiF 6 in KF-KCl molten at 650 °C, revealing that the optimal electrolytic conditions were as follows: K 2 SiF 6 (2 to 3.5 mol%) and current density (50 to 200 mA cm −2 ) [49]. The authors also pointed out the relationship between the conditions of the electrolysis and the morphology of deposited silicon, i.e., the morphology of the silicon films changed from compact and smooth to nodular or coral-like with increasing current density and K 2 SiF 6 concentration. Subsequently, the effect of temperature and current density was further investigated and discussed [51]. Figure 3A depicts the characteristics of the silicon deposits at 650 and 800 °C. Since the crystallization rate at 650 °C is slower than the deposition rate, only fine silicon grains are deposited. In contrast, larger columnar silicon crystals can be grown at a higher temperature (800 °C). The size of the silicon grains deposited at 100 and 300 mA cm −1 are 15 × 30 μm and 5 × 15 μm, respectively (Fig. 3B). This indicates that the deposited silicon grains will become smaller as the current density increases. Recent studies have shown that silicon films prepared in the KF-KCl-K 2 SiF 6 system reach a 4 N purity level and exhibit n-type semiconductor characteristics [52][53][54][55]. Interestingly, high-purity SiCl 4 can also be used as a silicon source to produce the same quality silicon films, which exhibit typical p-type semiconductor characteristics. Since gaseous SiCl 4 dissolved in KF-KCl forms SiF 6 2− (or SiF x Cl y (x+y−4)− ) anions, the system is thus identical to the silicon electrodeposition in the molten KF-KCl-K 2 SiF 6 system, as shown in Eqs. 4 and 5. However, the formation and tuning of the p-type and n-type silicon films still need further investigation.
Given the advantages of the KCl-KF-K 2 SiF 6 system, Peng et al. [53] conducted a study on the direct electrodeposition of photoactive silicon films; the products were obtained in the form of silicon films and silicon nanowires, which were doped or undoped with Sn 4+ , due to the presence of liquid tin (Fig. 3C). Liquid tin has a crucial role as a medium to promote all the growth stages of dense continuous silicon films: island formation/aggregation and film formation, as shown in Fig. 3D. In addition, the tin-doped silicon film exhibited n-type semiconductor behavior, generating photocurrents. The results were acceptable, achieving ~38 to 44% of commercial n-type silicon wafers (Fig. 3E). However, the authors observed dark currents, implying the presence of undesired pinholes and cracks in the silicon film, which may be due to the aggressiveness of KF in the solute. Although the quality of the deposited silicon in KCl-KF has been improved considerably, the corrosive effects of fluoride-based molten salt remain challenging to overcome. Therefore, Laptev et al. [54] tried to replace part of the KF and KCl with a high concentration of KI to reduce the solute's aggressiveness further. Glassy carbon and tungsten substrates were used to obtain silicon films with good adhesion. The experimental parameters were as follows: KF-KCl (2:1) −75 mol% KI containing 0.075 or 0.5 mol% K 2 SiF 6 at a temperature of 725 °C. In addition, the authors also confirmed that the solute could be used to obtain n-type silicon films by adding AlF 3 as a dopant [55], but further studies on the purity and optical properties of the films are necessary.
As discussed above, dense and smooth silicon films with p-type or n-type semiconductor properties were obtained in KF-KCl or KF-KCl-KI molten salts using fluorosilicates or SiCl 4 as the silicon source. Although the control of p/n junction semiconductors was not achieved then, the possibility of applying the proposed strategy to produce PV silicon films has been elucidated.

Electrodeposition of amorphous Si from organics
The electrodeposition of amorphous silicon also makes excellent sense for solar cells. Since amorphous silicon can be deposited at temperatures close to ambient, the energy cost of electrodepositing amorphous silicon in organic solvents (electrolytes) is extremely low compared to high-temperature molten salts [56,57]. The electrodeposition process should be performed under an inert atmosphere because the precursors [e.g., SiX 4 or SiHX 3 (X = Cl, Br)] tend to react with moisture in the air to form oxides through reaction (6). The first report on the electrodeposition of amorphous silicon was conducted by Agrawal and Austin [58], who prepared 1-to 3-μm-thick Si films from organic solvents on various substrates (such as platinum, titanium, and silicon wafers) at temperatures ranging from 35 to 145 °C using SiX 4 and SiHX 3 (X = Cl, Br) as precursors. However, the obtained silicon films contained ~3% SiO 2 , traces of metallic impurities, and some Si-H and Si-H 2 entities. The ethyl orthosilicate electrodeposition in acetic acid, propylene carbonate, and (4) 1-chloropropane solvents has also been investigated. Although electrodeposition under these conditions appears promising, these deposits produce large amounts of SiO 2 upon exposure of the inclusions to air.
Gobet and Tannenberger [59] utilized SiHCl 3 , SiCl 4 , or SiBr 4 as precursors and tetrahydrofuran (THF) as the solvent to deposit silicon films on Pt, Au, Ni, Cu, glassy carbon, and In-Sn oxide (ITO) substrates. The resulting thickness of the films was only 0.25 μm. However, the film also contained impurities such as carbon (~8 at.%), oxygen (~8 at.%), and chlorine (1.5 at.%). Nicholson [60] reported that silicon was electrodeposited on n-type silicon and titanium substrates using SiCl 4 and SiBr 4 as precursors in propylene carbonate and THF as solvents. The deposited silicon showed honeycombed morphology and contained some impurities (similar to the previous case): carbon, hydrogen, oxygen, and chlorine. Interestingly, the obtained Si showed photo-responsiveness, which may be generated by pure and non-oxidized deposits close to the substrate. In addition, adding a low concentration of AlCl 3 in this system as a p-type dopant induces the formation of p-n junctions on silicon wafers. Nishimura and Fukunaka [61] electrodeposited silicon film up to a 50-μm thickness from SiCl 4 in propylene carbonate, but with very high levels of other species and immediate oxidation upon contact with air. Munisamy and Bard [62] reported the electrodeposition of silicon on Ni, Ag, and glassy carbon from acetonitrile and THF containing the SiCl 4 and SiHCl 3 precursors. Similarly, the deposits were immediately oxidized as soon as they were exposed to air, and even the least exposed fraction contained significant amounts of carbon and oxygen. The authors pointed out that annealing the deposited silicon at 350 to 850 °C effectively reduces the C, N, and H concentrations, thus raising the quality of silicon. Besides, the annealing process induces silicon crystallization and results in weak p-type photoactivity. In principle, amorphous silicon electrodeposition from organic solvents is an attractive strategy, which is very economical and convenient compared to a high-temperature deposition. However, the poor quality of the silicon films due to their extreme susceptibility to oxidation and absorption of other species from organic solvents resulted in partially abandoning this approach. No solution to the problem of silicon oxidation has been found, at least so far. Therefore, further work should focus on the appropriate antioxidant strategies and control the content of other species in silicon deposits.

State-of-the-Art Silicon Electrochemical Extraction Strategies and Current Advances
Electrochemical methods for "SiO 2 to Si" in molten chloride salt The cathodic electrochemical deoxidation (reduction) of metal oxides, called the FFC Cambridge process, has attracted increasing attention from scholars in various countries since its discovery by Chen et al. [63]. The principle of this method is to apply a voltage to the solid oxide as a cathode at a temperature below the melting point of the metal and the decomposition voltage of the molten salt electrolyte (e.g., CaCl 2 -based molten salt), whereby the oxygen is removed by ionization, resulting in a metallic material. Initially, this process was only used to purify metals such as nickel and titanium [64]. Still, as the process was refined and extended, it was applied to silicon extraction.
In 2003, Nohira et al. [24] achieved a partial reduction of SiO 2 by electrolysis of solid quartz in CaCl 2 and LiCl-KCl-CaCl 2 molten salts by the FFC Cambridge process. The photographs and SEM micrographs show that the reduced SiO 2 is mainly in contact with the Mo wire ( Fig. 4A and B). Immediately after, Jin et al. [65] obtained pure silicon powder by direct electrolytic reduction of SiO 2 porous cathode pellets in molten CaCl 2 at 850 °C. SiO 2 particles with diameters of 2 to 7 μm are converted into pure silicon of 1 to 3 μm after electro-deoxidation ( Fig. 4C and D). The porous pellets (0.5 mm thickness) required less than 4 h to achieve complete reduction with an energy of 13 kWh kg Si −1 . Due to the chemical inertness of silicon, the electrolysis product can be washed after removal from the salt, and the residual impurity metals and oxides therein can be removed using dilute acids. This discovery provided a new definition of silicon production, attracting strong interest from the international academic community and the energy industry.

The SS mechanism and its process
This mechanism is assumed to be identical to the FFC Cambridge process. This involves directly reducing solid oxides to metals/ alloys by electrochemical deoxidation in molten chlorides [66]. However, unlike electrolytic TiO 2 , SiO 2 is an insulator. It has no low-valent oxide phase, which is directly reduced to crystalline silicon by gaining 4 electrons. Solid silica insulators' direct electrochemical reduction mechanism starts when the reaction occurs in the conductive collector/silica/molten salt contact region. When a sufficient cathodic overpotential is applied to the silica precursor, electrons are transferred from the conducting collector through the collector/oxide interface into the oxide phase. Simultaneously, oxide anions in the silica migrate through the oxide/electrolyte interface into the molten salt electrolyte. Solid silica is reduced to conductive monolithic silicon, causing an increase in molar volume. The product forms a porous layer, and the molten salt enters through the pores, forming a new Si/silica/molten salt 3-phase interline (3PI) reaction area. The continuous expansion of the 3PIs at the surface and in the bulk phase results in the complete reduction of silica. Chen and colleagues [67,68] modeled the corresponding 3PI boundary model by the SiO 2 -sheathed electrode method, as shown in Fig. 4E. Thus, this is also called the 3PI reaction mechanism.
The morphology change and crystallization for Si occurred during the electrolytic reduction of SiO 2 [69]. The formed Si exhibited a typical hexagonal prismatic structure with numerous stacking faults along the <111> direction (Fig. 4F). During the electrolysis, SiO 2 is first reduced to amorphous silicon with a fluffy structure, which immediately transformed into crystalline silicon due to the formation of a new SiO 2 /Si/CaCl 2 3PI (Fig. 4G). Notably, the generation of amorphous silicon from SiO 2 is the key step of the process. The O 2− diffusion in the obtained crystalline silicon vacancies controls the electrochemical reduction rate.

From SS to SLS: The crucial role of CaO and O 2−
Initially, the SLS mechanism (i.e., the dissolution-electrodeposition mechanism of silicon, which we replace with "SLS" in this paper) was not favored because the solubility of SiO 2 in molten CaCl 2 was too low. However, in the actual electrolytic reduction of SiO 2 , silicates (denoted as Ca y Si x O 2x+y , e.g., CaSiO 3 , Ca 2 SiO 4 , and Ca 3 Si 2 O 7 ) as intermediate products changed this situation [70][71][72][73]. CaCl 2 will always undergo a hydrolysis reaction, and its solute will contain CaO independent from any pretreatment processes applied. As shown in Eqs. 8  ) because the Gibbs free energy change for this reaction at 850 °C is −139.76 kJ mol −1 . The addition of 4.8 mol% CaO to CaCl 2 by Kongstein et al. [74] was found to be beneficial for the dissolution of SiO 2 . Until 2012, Xiao et al. [71] systematically investigated the mass loss of SiO 2 (quartz rods) immersed in molten CaCl 2 containing different amounts of CaO (0, 2, 3, and 5 mol% corresponding to 1#, 2#, 3#, and 4# in Fig. 5A) for 5 h. The solubility of SiO 2 in molten CaCl 2 increased with the concentration of CaO (Fig. 5A). Moreover, the authors observed a layer of silicon deposited on the outer surface of the nickel foam substrate during electrolysis of SiO 2 in CaCl 2 (containing 2 mol% CaO). Interestingly, the nickel foam substrate was not in direct contact with SiO 2 (Fig. 5B). Pure silicon can be obtained on the surface of the nickel substrate when long electrolysis is carried out (Fig. 5C). This anomalous deposition of silicon provides evidence for the existence of an SLS mechanism.
The dissolution of SiO 2 with the assistance of CaO was also confirmed by cyclic voltammetry (CV). As shown in Fig. 5D, with the addition of only 0.1 mol of SiO 2 , there are no prominent redox peaks in the CV curve (dashed line), except for Ca formation (C1/A1). With the addition of 0.1 mol of CaO, 2 pairs of redox peaks (C2/A2 and C3/A3) appear in the corresponding CV (solid line), which are attributed to the redox reaction of the dissolved silicate. The CV of the Mo electrode in molten CaCl 2 (with different CaSiO 3 contents) is shown in Fig. 5E. The C3 peak with an initial potential of 0.75 V indicates the formation of electrodeposited silicon, which is the same as the result of the CaCl 2 -CaO-SiO 2 system (Fig. 5D). Furthermore, the C3 peak intensifies in the CV at faster sweep rates, demonstrating the mechanism of electrodeposition of silicon in the melt. The current increases with increasing sweep rate (Fig. 5F), which is characteristic of the electrochemistry of dissolved species.
Furthermore, during the molten salt electrolysis of solid SiO 2 , the slow diffusion kinetics of O 2− results in a high concentration at the electrochemical interface. This induces the rapid combination of O 2− with solid SiO 2 near the cathode to form silicates. When the soluble silicate concentration reaches a specific value, it is sufficient to trigger the silica deposition Eqs. 10 and 11. Thus, silicate is bound to be present on the cathode regardless  of whether CaO is added to the CaCl 2 electrolyte. The SLS mechanism depends on the concentration of O 2− near the solid SiO 2 cathode, which is also an important factor controlling the process. The O 2− concentration can be adjusted by various parameters, including electrolytic potential, temperature, and CaO concentration. The SLS mechanism in the electrolytic reduction of SiO 2 drives the completion of electrolysis together with the mechanism of SS deoxygenation reaction.

Controllable design of nanostructured silicon: Nanowire and nanotube
Nanostructured silicon materials provide unprecedented opportunities for a wide range of applications in sensors [75], optics [76], nanoelectronics [77], biocatalysis [78], and energy storage [27,79]. Nanostructured silicon materials (e.g., nanowires and nanotubes) exhibit significantly enhanced performance as Li-ion battery anodes due to the efficient mitigation of the volume expansion of silicon, as well as shorter diffusion channels for Li + [12].
Generally, the employment of nano-silica as starting material is considered a prerequisite for the electrolytic synthesis of nanostructured Si. Yang et al. [80] reported the formation of Si nanowires (Si-NWs) by electrolysis in a molten salt using porous nano-SiO 2 powder as raw cathodic material (Fig. 6A). The authors proposed a corresponding growth mechanism ( Fig. 6B and C). The silica near the metal electrode first undergoes deoxygenation and reduction to silicon. Since the latter is conductive at high temperatures, it will act as a nucleus to form 3PIs with unreduced silica nanoparticles and molten salt. As the reaction continues, the silicon nuclei grow in the direction of the 3PIs, resulting in the formation of nanowires. In the last decade, silicon nanowires with various characteristics have been developed by molten salt electrolysis. Bent and entangled high-purity silicon nanowires were produced by cathodic electro-deoxidation in molten CaCl 2 at 850 °C using porous silicon particles as raw material in the potential range of 0.65 to 0.95 V (versus Ag/AgCl) [81]. Free-standing bilayer silicon nanowire arrays with diameters of 50 to 200 nm and thicknesses up to 100 μm can be prepared (Fig. 6D to F) by 2 nickel grids sandwiched by a quartz cathode in molten CaCl 2 [82]. Alternatively, high-purity straight silicon nanowires can be obtained by adding catalysts to the precursors. For example, porous particles composed of metallic Ni (0.8 wt.%) and SiO 2 can be electrochemically converted into silicon nanowires [83].
However, the electro-reduction of silica is mainly driven through the 3PI reaction mechanism, which results in very low silicon yields. To address this limitation, Dong et al. [84] used soluble CaSiO 3 as the precursor to continuously produce silicon nanowires by electrodeposition ( Fig. 6G and H). Unlike direct electrolysis of SiO 2 , CaSiO 3 can be dissolved directly in CaCl 2based molten salts to generate Ca 2+ and SiO 3 2− , while SiO 3 2− is reduced to Si (8) at a certain potential. The rate-controlling step is the diffusion of O 2− , which can be accelerated by adding CaO. Furthermore, the electrolysis temperature can be lowered to 650 °C using a ternary eutectic melt of CaCl 2 -MgCl 2 -NaCl ( Fig. 6I and J). More significantly, it is demonstrated that glass waste or coal ash composed of SiO 2 , CaO, and Na 2 O can also be used as starting materials for this process, opening a new path for the sustainable production of silicon nanowires. It should be mentioned that a ton-scale pilot plant producing silicon nanowires by the electrolysis of silica in molten salt was put into function in China [85].
Except for Si-NWs, silicon nanotube (Si-NT) synthesis by molten salt electrolysis has also been investigated and reported in recent years [86][87][88][89][90]. Weng et al. [86] achieved the electrochemical synthesis of Si-NTs and Si-NT@Ag on Ni substrates by co-electrolysis of SiO 2 and AgCl in molten NaCl-CaCl 2 at 850 °C. As shown in Fig. 7A, the whole Si-NT and Si-NT@Ag growth process can be divided into 4 steps. First, Ag and Si were sequentially deposited on the Ni substrate. When their concentration reaches a certain level, liquid Ag-Si alloy is formed at 850 °C. Subsequently, the continuous electroreduction and deposition resulted in the supersaturation of Si, which induced the formation of Si-NT via a liquid-solid mechanism (steps 1 and 2 in Fig. 7A). Interestingly, Si-NTs were covered with Ag (Si-NT@Ag, verified by energy dispersive spectrometer (EDS) after adding saturated silver chloride (step 3 in Fig. 7A). The Ag deposited appeared as a furry coating on Si-NTs ( Fig.  7B to D). In addition, the authors pointed out that Si-NTs and Si-NT@Ag were automatically separated from Ag and impurities during the annealing process, thus ensuring the purity of the product (step 4 in Fig. 7A). Most importantly, this strategy ranks among the best in both current efficiency and energy consumption (Fig. 7E). This process has even higher efficiency than that of industrial production of metallurgical silicon, which is very promising. Wang and colleagues [87,88] directly selected layered CaSiO 3 as the starting material, which accelerated the formation of Si-NTs (Fig. 7F). CaO is exfoliated from CaSiO 3 during electrolysis to form SiO x (0 < x < 2) sheets. It should be noted that, unlike the typical silicon SLS mechanism proposed in previous reports, this Si-NT formation is a typical SS reduction process. Furthermore, the authors found that SiO 2 particles can also be used as a starting material for Si-NT production, where SiO 2 in the molten salt reacted with CaO to form CaSiO 3 during the initial stage of electrolysis, followed by the same reduction mechanism as described above.
The design of Si surface nanostructures by electrochemical deoxidation in molten salts is also relevant. Of particular interest at present is the surface-engineered modification of silicon layers. For example, black silicon is a new electronic material that acts like a light-absorbing sponge that captures almost all visible and infrared light, so it can significantly improve PV conversion efficiency [91][92][93][94].
The team of Fray reported a detailed investigation of the electrochemical production of black silicon in molten salt. In 2010 [95], fine nanoscale silicon with specific surface texture thin layers was obtained after electro-deoxidation in molten CaCl 2 . The starting material was p-type silicon wafers with a (8) thermally oxidized surface (Fig. 8A). The authors pointed out that even silicon oxide layers with a thickness of only a few tens of nanometers can be converted to submicrometer-sized spherical particles under prolonged electrolytic conditions. Thus, shortening the electrolysis time is beneficial for generating superfine nanostructures on the surface. Subsequently, 2-μm-thick SiO 2 film on the surface of p-type silicon underwent further electro-deoxidation in molten CaCl 2 at 850 °C [96]. The results show that the surface of the electrolyzed SiO 2 film is a porous lattice structure composed of nano-nuclei and nanofibers, as shown in Fig. 8B. The reduction of the oxide layer on the wafer surface propagates along the 3PIs, and the reduced region (black) is the obtained black silicon (Fig. 8C).
Notably, the product shows significant visible light absorption, with a measured reflectance of ~8 to 11%. To further optimize the light absorption efficiency of the porous black silicon layer, TiO 2 was deposited on the black silicon surface prepared by electroreduction in molten salt to obtain "an extremely black" surface with reflectivity as low as 0.1% (Fig. 8D) [97]. There is no doubt that these findings demonstrate the potential of electrochemical deoxidation for the industrial production of black silicon. However, an industrial-scale process is still awaited.

Design of electrodeposited crystalline Si films: Photoactive layers and p-n junction
As discussed in the "Electrodeposition of silicon films from fluorosilicate/fluoride-based melts" section, molten salt electrodeposition of silicon films provides a new opportunity to replace conventional silicon wafer production. Unlike the intensely aggressive fluoride, CaCl 2 , a low-cost, water-soluble, and ecofriendly salt, has been suggested as a promising electrolyte for depositing silicon film. Furthermore, it is known that silicates are generated in molten CaCl 2 , which can be reduced and deposited in the form of solid silicon at the cathode. Hence, producing high-quality Si films via electrodeposition in the CaCl 2 -SiO 2 system is feasible. In recent years, Bard's group has conducted intensive and systematic research in this area. Cho et al. [98] attempted the direct electrodeposition of Si films in the CaCl 2 -SiO 2 system. A relatively pure discontinuous crystalline Si film was obtained on Mo foil at 850 °C by applying a constant current. The raw material was 5-to 15-nm-sized SiO 2 nanoparticles. However, no obvious photocurrent was observed. In general, the formation of silicon films is intimately related to the selected cathode substrate because they form by depositing nuclei on the substrate surface, forming continuous silicon particles and growing consistently. The surface of Mo is unfavorable for the deposition and growth of silicon nuclei. Therefore, the silver foil was further utilized to deposit photoactive silicon films from the CaCl 2 -SiO 2 system [99]. The deposition of silicon on silver substrates is believed to occur via the formation of liquid Ag-Si alloy, followed by the precipitation of silicon after supersaturation in the liquid alloy and continued growth. The obtained silicon demonstrates high purity (>99.9%) as the impurities can be spontaneously separated during the precipitation of silicon from the supersaturated liquid. The obtained silicon exhibited a clear photoresponse, approaching half the maximum photocurrent of pure silicon wafers. However, the high price makes it difficult for a silver substrate to become more competitive in the market. The relatively inexpensive graphite has also been shown to be a suitable substrate for silicon as early as studies of fluoride electrodeposition. Zhao et al. [100] first reported that dense, continuous p-type silicon films were deposited on a graphite substrate in molten CaCl 2 containing 0.3 mol% nano-SiO 2 at a current density of 6 mA cm −2 for 1 h. The electrodeposited thin silicon films on graphite showed higher photocurrent than those obtained on silver.
It is well known that the low solubility of SiO 2 nanoparticles in molten CaCl 2 limits the net mass transport. Aiming to break this limitation, Bard and colleagues [72,73,101] demonstrated the study of continuous electrodeposition of photoactive Si films in the CaCl 2 -CaO-SiO 2 system. As shown in Fig. 9A and B, during the electrodeposition process, O 2− ionized from CaO reacts with nano-SiO 2 to form silicate ions. CaO, as an intermediate medium, facilitates the continuous ionization of SiO 2 to form silicate ions, which are reduced to silicon crystallites on the graphite surface. To achieve high-purity silicon, periodic preelectrolysis of molten CaCl 2 -CaO-SiO 2 is necessary. The impurities in the deposited silicon film should be strictly controlled to be lower than the allowable values. Since the electrodeposition process is performed at a low current and the reactor is strictly protected by high-purity argon gas, any possible CO or CO 2 emissions from the graphite anode are immediately removed. It should be pointed out that electrodeposition parameters (current density, time, etc.) are essential to improve the efficiency of the electrolytic process and to fabricate high-quality silicon films. Generally, the type and concentration of dopants can be directly controlled by adjusting the sources in the molten salt. P-type, n-type, and p-n junction silicon films were electrodeposited on a graphite substrate ( Fig. 9C to F). Silicon p-n junction films were produced directly from SiO 2 -CaO-CaCl 2 systems by a 2-step electrodeposition process, i.e., the p-type Si film was first electrodeposited on a graphite substrate with the aluminum dopant originating from the used quartz crucible or from the addition of additional Al 2 O 3 . Subsequently, the prepared p-type silicon film was polished and used as a substrate for the electrodeposition of n-type silicon film using Sb 2 O 3 or Ca 3 (PO 4 ) 2 as the dopant sources. The fabricated dense p-n junction silicon film shows typical hexagonal crystalline particles ( Fig. 9G and H) and a purity of 99.9998% (close to 6 N, solar grade), which is the highest purity ever reported in silicon electrodeposition research. The photocurrent density was approximately 40% to 50% of commercial silicon wafers, with a maximum power conversion of 3.1% as a solar cell (Fig. 9I to K).
As discussed above, significant advances have been made in the last decade in the fabrication of p-type, n-type, and p-n junction silicon films for PV applications by electrodeposition of SiO 2 in CaCl 2 molten salts. This trend will possibly continue in the future and eventually replace the production of conventional silicon wafers.

Electrochemical synthesis of various silicon-based materials
Besides the photoactive silicon films and nanostructured silicon discussed above, molten salt electrolysis has also been used as a template-free and simple green synthetic route to prepare some silicon components, such as silicide and SiC, which received considerable attention in recent years.

Silicides
Metal silicides have gained widespread applications due to their excellent high-temperature oxidation resistance and outstanding electrical/heat transfer properties [102,103]. As a potential new semiconductor material, Ca-Si alloy exhibits superconducting characteristics at high pressures with a critical temperature of 14 K [104][105][106]. Nevertheless, the high vapor pressure of Ca makes it extremely problematic to synthesize Ca-Si alloys by conventional chemical and vapor deposition methods. In contrast, preparing calcium silicide in CaCl 2 -based molten salts is simple. Ca-Si alloys can be produced through reaction (12) when the potential is more positive than the equilibrium potential E Ca/ Ca2+ [24,[65][66][67][68]. Sakanaka et al. [106] prepared Ca-Si films in molten CaCl 2 -KCl at 650 °C using a constant potential of −0.1 V. Subsequently, the multiphase Ca-Si films were converted to Si, CaSi, or CaSi 2 phases by tuning the anodic potential. In addition, the authors elucidated various conversion reactions and the corresponding equilibrium potentials, which laid the foundation for the controlled preparation of calcium silicide.
Recently, there has been an ever-growing interest in using intermetallic silicides to replace conventional graphite anodes in Li-ion batteries. These alloys used as anodes for Li-ion cells generally have lower capacities than pure silicon. Still, they show less volume change during lithiation/delithiation. Also, they maintain good structural stability and thus achieve better cycling. Various excellent results have been reported for the synthesis of silicon alloy anodes by molten salt electrolysis, such as Si-Fe [107], Si-Mn [108], Si-Cu [109], Si-Ti [110][111][112], and Si-Ge [113,114]. Zhou et al. [109] utilized SiO 2 and Cu (molar ratio 1:1) as precursors to synthesize Cu-Si nanoalloys with a constant voltage of 2.4 V in molten CaCl 2 at 700 °C. The main composition was Cu 9 Si and Si (denoted as Cu 9 Si/Si). During the electrolysis, SiO 2 is continuously deoxygenated and reduced to Si. At the same time, Cu acts as a catalyst to promote the growth of Cu-Si nanowires along the Si axis, which exhibited good cycling stability and rate performance for lithium batteries. Ti-Si alloys were also prepared by direct electrodeoxidation of SiO 2 /TiO 2 pellets (molar ratio 1:1) in molten CaCl 2 -NaCl at 700 °C, and the main product was nanoscale Si and Ti 5 Si 3 (labeled as Ti 5 Si 3 /Si). Furthermore, Ti 5 Si 3 /Si demonstrated a good specific capacity of 638 mAh g −1 after 50 cycles at a charge/discharge rate of 200 mA g −1 [110]. However, the excess of Ti 5 Si 3 in the product led to a decrease in lithium storage capacity. Hence, a superior anode is the TiSi 2 /Si nanoalloy, which can be prepared by adjusting the molar ratio of raw SiO 2 /TiO 2 (20:1) [111]. Xiao et al. [113] prepared Ge-Si alloy nanotubes and hollow particles by direct electrolysis of GeO 2 and SiO 2 nanoparticles. The cavities formed benefited from the continuous solid diffusion controlled by the Kirkendall effect. The enhanced lithium storage performance of Ge-Si nanotubes fully illustrates the potential of alloy nanotubes.

SiC and Si/C composites
SiC nanomaterials have been recognized as a new rising star [115]. Especially, SiC nanowires have attracted much attention due to their outstanding performance in many applications such as power [116] and harsh environment electronics [117], light detection devices [118], photocatalytic hydrogen production [119], and biochemical sensors [27]. Numerous works have reported that SiC can be extracted from molten chloride salts by direct electrolysis of SiO 2 /C or rice husk (RH). Yang et al. [72] obtained homogeneous SiC nanowires (SiC-NWs) by the electrolysis of SiO 2 /C (molar ratio 1:1) mixed pellets in molten CaCl 2 at 900 °C with a constant voltage of 3.1 V. They proposed the corresponding reaction mechanism as well ( Fig. 10A to C). Interestingly, SiC nanowires can also be produced by electrodeposition in molten CaCl 2 -SiO 2 /C. The reaction mechanism involves several processes: (a) composite formation, (b) dissolution, (c) electrodeposition, and (d) carbonization [Si + C → SiC, ∆G 0 = −63.94 kJ mol −1 (900 °C)] (Fig. 10B). However, since the Si nuclei cannot be surrounded by the carbon powder dispersed in molten CaCl 2 , the products are usually a mixture of Si-NWs and SiC-NWs (Fig. 10D). Obviously, these results provide an essential guideline for the controllable preparation of SiC nanostructures or Si/C composites by molten salt electrochemical processes.
Silicon and carbon composites are known to be ideal anode materials for Li-ion batteries. Introducing highly conductive carbon not only reduces the polarization of silicon particles but also increases the material's electrical conductivity, thus improving the cycling and rate performance of the battery. Such advantages of Si/C composites motivated the direct preparation of Si/C composites by molten salt electrolysis. However, Si produced by reducing SiO 2 at high temperatures would react spontaneously with C to form SiC. Thus, the preparation of Si/C composites based on molten salt electrochemistry has been deemed to be a challenging task.
The strategy of preparing Si/C composites by simultaneous pyrolysis and electrolysis of polydopamine (PDA) encapsulated SiO 2 nanoparticles in molten salts was proposed and validated [120]. During the reaction, SiO 2 is reduced to Si via electrodeoxidation. Simultaneously, PDA is pyrolyzed to obtain N-doped carbon. Notably, the gases (CO, H 2 ) generated during the pyrolysis will act as a physical barrier at the interface between Si and C, effectively inhibiting the formation of SiC. In addition, the ionized oxygen tends to react with the neighboring SiO 2 to form SiO 3 2− , which then produces new Si on the outer surface of carbon via an SLS mechanism, leading to the formation of Si@C@Si microstructures with enhanced lithium storage capacity (Fig. 10E). Alternatively, the use of MgCl 2 -based molten salts is also considered to be efficient in preventing the formation of SiC. Recently, it has been shown that core-shell Si/C composites can be prepared directly by electrochemical reduction of C@SiO 2 in molten NaCl-KCl-MgCl 2 at 650 °C [121,122]. The O 2− produced during electrolysis reacts with Mg 2+ to form MgO, which has low solubility in molten NaCl-KCl-MgCl 2 . MgO can prevent the spontaneous reaction of Si with C at high temperatures. Besides, removing MgO by water following the reaction would form voids within the carbon shell. The space formed can act as a buffer for the volume expansion of the Si/C anode material during the lithiation/delithiation process. If more cavities are needed within the carbon shell, carbon-coated magnesium silicate (C@MgSiO 3 , C@Mg 2 SiO 4 ) can be employed to generate more MgO in situ during electrolysis (Fig. 10F).
RHs are typical agricultural waste with an annual production of up to 100 million tons. It contains 15 to 20 wt.% SiO 2 , 75 to 85 wt.% organic matter, and other trace inorganic elements (such as Mg, Zn, K, S, and P) [123][124][125]. Therefore, converting RHs into SiC or Si/C composites is an attractive option, i.e., it achieves the utilization of both components and avoids the waste of resources [126,127]. Zhao et al. [128] investigated the electrochemical conversion of RHs to SiC and SiC composites in CaCl 2 or NaCl-KCl-MgCl 2 molten salts. Before electrolysis, the RHs were first converted by pyrolysis from amorphous hydrated silica and organic matter to SiO 2 and hard carbon, respectively. The SiO 2 /C composites were then electrochemically converted at a constant voltage of 2.4 to 2.8 V to obtain different products (Fig. 11A). At 2.4 V, randomly distributed SiC nanowires were obtained on amorphous carbon particles. At a higher voltage of 2.6 V, the product was a mixture of honeycomb SiC nanowires and carbon particles. When the electrolysis voltage was increased to 2.8 V, carbon flakes and interconnected nanowires were observed. Moreover, the authors found that lower potential is favorable to hindering SiC formation.  Thus, Si/C composites can be obtained in NaCl-KCl-MgCl 2 molten salts with electrolytic voltages below 2.5 V, which has excellent rate performance for Li-ion battery, maintaining a high capacity of 926 mAh g −1 even at a current density of 500 mA g −1 . However, the fact that pyrolysis and electrolysis are carried out in 2 steps complicates the whole reaction route and reduces the treatment efficiency.
Pang et al. [129] argue that CaCl 2 -based molten salts are still a better choice for the electrochemical reduction of SiO 2 than MgCl 2 -based ones. The reason behind this is the sustained and efficient conversion and the reduced anodic evolution of corrosive Cl 2 . Concerning the uncontrollable formation of SiC in CaCl 2 -based molten salts, the authors propose that this can be countered by precise modulation of the microstructure and composition of RHs. The feasibility of controlling the conversion of SiO 2 and organic matter in RHs to form SiC/C or Si/C by superheated electrolysis in CaCl 2 -NaCl molten salts has also been verified. RHs with a highly cross-linked structure with a low SiO 2 /C molar ratio and high content of metal impurities were converted to SiC-NW/C. The product exhibited potential as microwave-absorbing material (Fig. 11B). Acid-impregnated RHs (A-RHs) containing isolated SiO 2 , with high SiO 2 /C molar ratio and marginal metal impurities, were converted to Si-NP/C (Fig. 11C). This was attributed to the bubble phenomenon caused by CO and H 2 released in the molten salt during the thermal electrolysis of organic matter that effectively hindered SiC formation. Nevertheless, the process described earlier yields greenhouse gases (CO 2 emissions). The same team recently reported that the SiC-NW/C obtained from RHs could be used as an efficient CO 2 reduction photocatalyst [130]. As shown in Fig. 11D, RHs are electrochemically converted to SiC-NW/C and CO 2 in molten CaCl 2 -NaCl. Due to the strong coupling between the components of SiC-NW/C, the photoelectrons generated by SiC-NW through excitation by light absorption will rapidly migrate to the porous carbon. The conversion of the adsorbed CO 2 to CO occurs (Fig. 11E), providing sustainable power for the molten salt electrolyzer. Hence, the conversion of RHs to SiC-NW/C is theoretically a closed-loop carbon cycle bridge in abundant light. Undoubtedly, these studies provide new insights into the direct access to SiC or Si/C mixtures from RHs by molten salt electrolysis while opening a new path for the sustainable production of Si-related energy storage materials.

Ionic liquids-An alternative to high-temperature molten salts?
Ionic liquids are molten salts with melting points below 100 °C, which consist of only cations and anions [131,132]. Compared to conventional simple metal halides (e.g., AlCl 3 ), the cations and anions of ionic liquids are relatively complex. The charges are often delocalized or shielded by side groups [133]. Besides, ionic liquids exhibit good electrical conductivity, a wide electrochemical window (>5 V), and high thermal stability, so they can be used to electrodeposit silicon at room or milder temperatures. In this case, ionic liquids are called "room-temperature molten salts" and are considered an alternative to high-temperature molten salts [20].
Reportedly, Katayama et al. [134] found that silicon can be electrodeposited at low temperatures (90 °C) in ionic liquids (1-ethyl-3-methylimidazolium hexafluorosilicate). However, the product completely converted to SiO 2 when exposed to air. Thus, it remained an open question whether the obtained silicon had semiconducting properties. In 2004, El Abedin and colleagues [135,136] reported for the first time that efficient nanoscaled electrodeposition of elemental silicon was achieved at room temperature using SiCl 4 as the silicon source in 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl) imide ([BMP][Tf 2 N]). However, the purity of the deposits was unsatisfactory. The authors found that electrodeposition can severely disturb interfacial processes even at a ppm concentration atmosphere. Besides, silicon nanoparticles easily interact with metal substrates during deposition to form intermetallic silicides. Recently, Tsuyuki et al. [137] have also tried to electrodeposit silicon thin films for solar cells in ionic liquids and systematically investigated the doping of the film structure. N-trimethyl-N-hexylammonium bis(trifluoromethanesulfonyl) imide (TMHA-TFSI) and SiCl 4 were used as the ionic liquid and silicon source, respectively, to promote the growth of continuous and dense silicon films by adjusting the light time. The addition of AlCl 3 resulted in the formation of p-type silicon films. However, these films were primarily oxidized, which may be attributed to the films' exposure to air or during annealing in an attempt to remove impurities generated by ionic liquids.
Generally, a high temperature (>700 °C) is required for the electrodeposition process to produce crystalline silicon. While prepared at low temperatures, it is always amorphous and requires additional thermal annealing and purification. Gu et al. [138] proposed a new strategy for electrochemical synthesis using a liquid metal electrode (e.g., Ga). The latter can act as a source of electrons for dissolved species and as a recrystallization solvent. This strategy is called "electrochemical liquid-liquid-solid" (ec-LLS) crystal growth (Fig. 12A  and B), which allows the direct production of crystalline silicon at near ambient temperatures [139]. Recently, Zhang et al. [140] reported an efficient method to fabricate crystalline silicon films using SiCl 4 on a liquid gallium surface at low temperatures in the presence of tri-1-butylmethylammonium bis((trifluoromethyl)sulfonyl)amide ([N4441][TFSI]) as the electrolyte. The silicon films had 2 sides, one containing polycrystalline silicon particles and the other smooth amorphous silicon (Fig. 12C to E). The authors suggested that the possible growth mechanisms for crystalline Si include the formation of liquid Si in liquid Ga followed by nucleation and growth of crystalline Si on reduced amorphous Si films. Zhao et al. [141] obtained excellent-quality crystalline silicon films in a similar system by adjusting the reaction temperature, deposition potential/duration, and different substrates. However, as an attractive strategy for the fabrication of crystalline Si in lowtemperature ionic liquids, ec-LLS strategies still have many problems to be addressed.
As discussed above, considerable progress has been made in the electrodeposition of silicon in ionic liquids. However, the incompatibility of low-temperature and pure crystalline products severely limits the attractiveness of the process. In the future, the silicon oxidation contamination problem deserves further attention. Moreover, further investigation into the controllable deposition and application of liquid electrodes is necessary to facilitate the development and popularization of this strategy for producing crystalline silicon at room temperature.

New opportunities for designing Si structures: Controlled tuning of templates and parameters
Electrodeposition from room-temperature ionic liquids offers a potential alternative to traditional physical vapor or chemical vapor deposition as a cost-effective technique in preparing class IV thin films or nanostructures such as silicon. Thomas et al. [142] described the effects of potential, the concentration of electroactive substances, temperature, and organic additives on silicon deposits in 1-butyl-1-methylpyrrolidinium bis(trifluoromethanesulfonyl)imide ionic liquids. It was found that the concentration of electroactive substances or additives in the solvent has an almost negligible effect on the synthesis of silicon. However, the temperature and the applied potential were crucial. The cross-sectional morphology of the silicon films at different temperatures (25,50, and 100 °C) is interesting (Fig. 13A to C). Compared to silicon films deposited at room temperature, the one obtained at 100 °C is rougher and exhibits lower adhesion to Au substrates, indicating that temperature plays a crucial role in silicon growth. Thus, the deposition of silicon films at 50 °C can be balanced by adjusting the deposition rates, adhesion, and growth roughness. Furthermore, the authors note that prolonged electrodeposition of silicon in ionic liquids will be limited by the maximum charge that can be passed. However, the charge limitation can be improved by increasing the temperature, potential, or concentration of electroactive substances. This study offers new insights into silicon electrodeposition in room-temperature ionic liquids. This is crucial for controlling and improving the synthesis of silicon for specific applications.
An exciting aspect of silicon electrodeposition in roomtemperature ionic liquids is the ability to grow differently shaped nanomaterials in a template-assisted way. Mallet and colleagues [143,144] obtained pure amorphous Si-NWs and Si-NTs by electrodeposition at room temperature using SiCl 4 as the precursor in 1-butyl-1-methylpyrrolidinium (trifluoromethylsulfonyl) imide (P 1,4 ) ( Fig. 13D and E). The diameter or wall thickness of the deposit can be precisely controlled by simply changing the electrodeposition parameters or the size of the template (polycarbonate film). Annealing can crystallize silicon deposits without changing their shape and structure. Such a strategy may replace the more restrictive high-vacuum approach, and the synthesized homogeneous silicon nanomaterials have promising applications in microelectronics and optoelectronics [145,146].
Some recent publications have demonstrated that synthesizing ionic liquids can yield Si-based anode materials by surface plating or co-deposition at room temperature [147][148][149]. Kowalski et al. [147] deposited a continuous, compact silicon shell on the surface of TiO 2 nanotubes via electrodeposition from SiCl 4 in room-temperature ionic liquid. The thickness of the silicon shell was adjustable by the charge density. Silicon layer thicknesses of 20, 60, and 100 nm can be obtained at charge densities of 0.3, 1.0, and 1.8 C cm −2 , respectively. Li et al. [150] fabricated porous sheet-like Si-Cu composites. Its structure contained alternating Si-Cu layers on a Cu substrate. The applied potential was 1.9 to 2.1 V during electrodeposition in ionic liquids containing 0.002 mol l −1 Cu(TfO) 2 and 1 mol l −1 SiCl 4 -[BMP]Tf 2 N. The porous structure was caused by the SiCl 4 bubbles attached to the Si-Cu films. The authors pointed out that introducing copper helps to obtain a specific silicon morphology and improves its conductivity. The as-obtained Si-Cu porous sheets used as an anode material have a capacity of 1,042.8 mAh g −1 after 600 cycles at a current density of 21 A g −1 .

Challenges and Potential Solutions
Notwithstanding the recent promising progress in silicon electrochemistry research, several scientific and technical obstacles remain to achieve practical applications, including developing inert anodes, improving electroreduction rates, and production of high-purity silicon. Hence, the challenges of silicon electrochemistry are summarized in Fig. 14, and potential solutions are presented to advance silicon electrolysis in real-world applications.

Development of inert anodes
Graphite has been widely employed as an anode material for high-temperature molten salt electrochemical extraction of silicon due to its excellent electrical conductivity and thermal stability [151]. However, there are several effects on the graphite anode during the electrochemical production of silicon: (a) O 2− released from the silica or silicate during reduction will inevitably react with the graphite anode, generating undesirable CO or CO 2 [152]; (b) CO and CO 2 will partially dissolve in the molten salt, generating CO 3 2− and reducing at the cathode, resulting in a significant decline in the purity and performance of the product silicon; (c) the carbon powder peeled off from the graphite anode will float on the surface of the molten salt during longterm electrolysis, thus causing a short circuit between the cathode and anode. A daunting challenge is developing stable, inert anodes not consumed in high-temperature molten salts.
Various prospective anode materials have been used to assemble molten salt electrolysis, including alloys, oxides, and ceramics [153][154][155][156][157][158][159][160]. The Cr-Fe alloy anode seems promising because it shows excellent stability in electrolytic processes involving molten iron at temperatures up to 1,565 °C [153]. This is a crucial reference for silicon electrolysis at even lower temperatures. Another potential anode material is CaTiO 3 / CaRuO 3 -based composite, which shows extremely low corrosion rates and oxygen precipitation in CaCl 2 -based molten salts [158,159]. However, the usage of CaRuO 3 -based anodes may be restricted by the limited reserves of Ru. Recently, a Magnéli-phase titanium oxide (Ti 4 O 7 ) inert anode with conductivity close to that of graphite was obtained, which produced only oxygen during silicon electrolysis [160]. More importantly, this anode can be regenerated by current reversal or chemical reduction of H 2 in the molten salt, thereby enabling recycling. However, such excellent performance was merely tested under laboratory conditions, and many of the characteristic values obtained are unrepresentative of actual production conditions.

Improvement of electroreduction rate
The reduction rate is an important indicator to determine the feasibility of a silicon electrochemical extraction technology. In the case of electrodeposition, continuous electrolysis will result in a significant decrease in the silica reduction rate because the deposited silicon precipitates as a solid on the substrate. To address this challenge, an experimental approach was suggested in which the deposition of liquid silicon should be carried out in the BaO-SiO 2 -BaF 2 system [39]. However, the relatively high melting point of silicon (1,414 °C) is a drawback that hinders the scaling up of the process. However, it is not impossible to overcome this obstacle. For example, its melting point is much lower than that of iron. Regarding the silicon electro-deoxidation technology, the reduction rate of SiO 2 is extremely low due to the deoxidation reaction mainly occurring near the 3PIs. Even porous particles pressed from SiO 2 (2 to 7 μm) require several hours to achieve complete reduction [65]. Future research may focus on 2 directions: (a) Finding a suitable additive that not only improves the electrical conductivity of SiO 2 precursor but also allows complete separation from the product at the end of the reaction, and (b) exploring appropriate liquid metal electrodes instead of common ones would significantly increase the contact area between the electrode and the silica precursor and

Production of high-purity Si
The key challenge in silicon electrochemistry is purity, as the silicon purity for PV devices is at least 6 N (≥99.9999), while that used in electronic devices is higher (EG-Si or 11 N). In most cases, silicon precursors, reaction vessels, electrode-connected metals, electrolytes, anodes, and cathode materials can introduce impurities during the electrochemical reduction process. Without a one-step purification, the reduced silicon will hardly meet the above requirements.
Pre-electrolysis is a typical purification method because it effectively reduces impurities in the electrolyte. When silicon precursors are present (e.g., fluorosilicate/fluoride systems), impurities with reduction potentials lower than silicon can be removed by setting a pre-electrolysis potential. However, impurities with a reduction potential higher than the pre-electrolysis potential remain in the electrolyte. In this regard, Xu and Haarberg [26] suggested that precise regulation of the concentration of electroactive substances and the applied potential can improve the purity of the product. In the case of SiO 2 or CaSiO 3 as precursors, there is an opportunity to periodically pre-electrolyze the electrolyte so that all impurities are decreased to tolerable thresholds (e.g., Mg < 0.05 ppm, Na < 0.05 ppm, and W < 0.05 ppm). This was confirmed by Zou et al. [73] as they obtained ~6 N purity silicon films.
Nohira's research team suggested that using gaseous SiCl 4 as a silicon precursor for the electrochemical extraction of high-purity Si might be very promising [161] since the wellknown Siemens purification technology also employed a gaseous silicon source (SiHCl 3 ). Although SiCl 4 is almost insoluble in chloride electrolytes, its solubility in KCl-KF has exceeded 80%. It has the advantage that the anode reaction is the evolution of Cl 2 rather than the emission of CO or CO 2 . Moreover, the generated Cl 2 can be collected and used for the chlorination of SiO 2 and then distilled to prepare high-purity SiCl 4 . In addition, using liquid Si to control the impurity content is also a promising means of purification. During the separation of the solid Si from the liquid alloy, the impurities will remain in the liquid phase. Yasuda et al. [162] proposed that using liquid Si-Zn cathodes to produce SoG-Si in molten CaCl 2 , the molten salt effectively inhibited the evaporation of Zn, thus ensuring the long-time operation of the Zn cathodes. The reduction of SiO 2 on Zn proceeds at E < 1.45 V (versus Ca 2+ /Ca), which is much more positive than the reduction of SiO 2 on Si. Pure Si particles can be collected during the cooling of the Si-Zn system.

Advantages and drawbacks/limitations of various strategies
Silicon electrochemistry is a constantly evolving field that provides eco-friendly and energy-efficient silicon extraction and processing strategies. This paper reviews the history and recent advances in the electrochemical production of silicon. As discussed in the review, several promising strategies have been attempted. Each has its advantages and drawbacks, so the characteristics (Si purity, scalability, cost-efficiency, reduction rate, and low energy consumption) of these strategies will be summarized from the point of view of the electrolytes (Fig. 15).
The electrolysis of Si in molten oxide is also referred to as the electrowinning of liquid Si. Since Si is electrodeposited in liquid form, it is reduced at a speedy rate. As shown in Table, the purity of the silicon ingots obtained from the barium silicate solution is about 99.98 wt.%. This purity is close to the quality required to produce solar cells with 10% efficiency by a single directional solidification stage. However, the applied ultra-high temperatures (~1,400 °C) also result in high energy consumption and  Molten fluoride is recognized as the most promising electrolyte for Si electrodeposition because of its inherent etching properties. The most extensively investigated systems are fluorosilicates dissolved in molten fluoride salts (e.g., LiF, NaF, and KF). The electrolysis temperature is generally between 450 and 800 °C, with relatively low energy consumption. Fluorosilicates have been thoroughly studied as precursors, but their continued availability as an inexpensive source depends on the fertilizer industry. However, fluorosilicates can also be synthesized in situ by SiO 2 with alkaline earth fluoride baths. Although it offers significant advantages in terms of costeffectiveness and energy efficiency, it involves problems mainly linked to highly corrosive HF, including control of impurity levels, severe Si pitting, and surface staining and clouding of Si. The purity of Si obtained in fluoride-chloride electrolytes (e.g., KCl-KF) is only around 4 N.
Chloride molten salts are the most widely utilized electrolytes, especially CaCl 2 , which has been intensively studied for the electrochemical reduction of silica as an environmentally friendly, water-soluble, and inexpensive salt. The temperature of electrolytic Si in molten chloride is generally 500 to 900 °C, so the energy consumption is relatively low. The solid-to-solid reaction drives the electroreduction process of SiO 2 , and the reduction reaction occurs mainly near the 3PIs, so the reduction rate is very slow. Its potential ways of improvement have been mentioned in the "Improvement of electroreduction rate" section. Despite its very low solubility in molten chloride salts, SiO 2 can also be reduced by an SLS reaction in CaCl 2based molten salts containing CaO. This method is based on a dissolution-electrodeposition mechanism. SiO 2 is dissolved initially in CaCl 2 with a high concentration of O 2− , and then Si is electrodeposited from the resulting silicate ions. This strategy has now been shown to be helpful in depositing Si films with purity up to nearly 6 N. Ionic liquids and organic solvents used as electrolytes opened the door to silicon electrolysis at room temperature. They are very economical and convenient compared to high-temperature electrolysis techniques. The main advantages of ionic liquids are their good electrical conductivity, high thermal stability, and wide electrochemical window (>5 V), which allows them to be used for the electroreduction of Si at room temperature. Besides, ionic liquids are highly safe and eco-friendly due to their low vapor pressure, non-flammability, non-toxicity, and biodegradability. The advantages of organic solvents also include a wide electrochemical window, which makes them sufficient for the electroreduction of Si. However, Si conducts poorly at low temperatures, and the Si deposited on the substrate surface will act as an insulating layer hindering the continuous reduction of Si. In addition, Si deposited at room temperature is susceptible to oxidation in air and incorporation into other substances in the electrolyte in organic solvents; thus, the purity is low.

Electrodeposited Si film for PV cells
In principle, electrodeposition of silicon from molten salts is highly cost-effective. The reason behind it is the production of silicon wafers for solar cells directly from raw materials in a single-stage process, which is unmatched by any competing process (Fig. 16A). In terms of feedstock and product, the electrochemical production of silicon is analogous to industrial carbothermal deoxidation reduction. However, the carbothermal reduction steps require significant energy consumption and emit huge amounts of CO 2 (Fig. 16B). For reference, each kilogram of polysilicon consumes approximately 20 kWh of electricity and emits at least 84 kg of CO 2 [21]. An additional 90 to 200 kWh kg Si −1 is required when further purification is carried out using the Siemens process. In contrast, the electrical energy for the electrodeposition of silicon would be reduced to approximately 13 kWh kg Si −1 . The CO 2 emissions from silicon electrodeposition are mainly associated with the most commonly used graphite anodes (already discussed in the "Development of inert anodes" section), which are reported to consume approximately 0.19 kg of carbon and emit 0.7 kg of CO 2 for every 1 kg of polysilicon produced [160]. Despite this challenge, the CO 2 emission is significantly lower than the industrial methods.
The fluorosilicate/fluoride system is ideal for the direct electrodeposition of silicon films because of its cost and energy efficiency. Nevertheless, this process is still a long way from actual industrial production due to the high aggressiveness of fluorides. Those compounds make it difficult to control the level of impurities in the deposit. Electrolysis of the SiO 2 -CaO-CaCl 2 system has been considered a promising strategy for depositing high-purity silicon and p-type, n-type, and p-n junction silicon films. Efforts toward achieving better photoactive silicon films have intensified over the past decade. In particular, Bard's team has assembled PV cells with near-solar grade (SoG-Si) purity with a power conversion efficiency of 3.1%, confirming the potential of this strategy for real-world applications. Figure 16C summarizes the manufacturing cost per watt of solar cell module versus module efficiency based on electrodeposition technology [163][164][165]. The production cost of electrodeposited silicon wafers is significantly lower than that of conventional methods (Fig. 16D), and a module cost of nearly 0.2 $Wp −1 can be achieved by improving the energy conversion efficiency of the cell to 10%. These data show the potential of electrochemical methods for the direct preparation of low-cost solar cells; however, its processes for practical applications still need to be significantly developed. It is worth mentioning here that despite recent good progress in the electrochemical production of silicon with a purity close to solar grade, the method is still a long way from producing higher-purity silicon (such as EG-Si) and will hardly be able to compete with silicon produced by the semiconductor industry in a short term. Improving purity is therefore a key challenge for future work. Amorphous silicon, a prospective material for large-scale PV devices, can, in principle, be fabricated by low-temperature electrodeposition from ionic liquids and organic solvents. Still, research to date looks less promising than alternative molten salt processes that yield crystalline silicon. A primary reason is the low conductivity of silicon at low temperatures, which limits electrodeposition and prevents it from forming a continuous dense film.

Electrochemical synthesis of various Si-based materials
The electrochemical strategy provides the feasibility of the design and synthesis of various silicon-based materials. Electrodeoxidation offers a simple and effective route for the synthesis of micro-and nanostructured silicon and other silicon compounds for many applications, including Li-ion batteries, PV cells, semiconductors, photocatalysts, integrated optoelectronic devices, and electromagnetic absorption materials (Fig. 17). Of particular concern is the application of silicon nanomaterials in Li-ion batteries, due to the fact that silicon has the highest specific capacity to store lithium ions. However, the severe structural degradation caused by the expansion-shrinkage of silicon during charging and discharging makes the lifetime of silicon-based cells difficult to meet. Thus, designing and creating multicomponent micro-nanostructures is an effective strategy to mitigate or suppress structural degradation of silicon anodes. The controlled fabrication of a broad range of nanostructures for silicon, such as Si-NP, Si-NW, and Si-NT, is now possible on a laboratory scale by electrochemical methods. Even more exciting is the fact that a pilot ton-scale production of silica-based anode materials for Li-ion batteries by electrolysis of silica in molten salt has been established in China [85]. Therefore, more emphasis should be placed on improving the electrochemical properties of silica-based anode materials based on this approach in the future. In addition, how the various electrolytic parameters for high-performance silica-based anodes achieved in laboratory scale can be transferred to industrial-scale production should be further explored. Moreover, future efforts should be aimed at expanding the electrochemical preparation of black silicon. This process is fairly simple and fast, avoiding the emission of toxic or corrosive chemicals or harmful substances. In addition, the direct extraction of SiC and Si/C composites by electrochemical methods from RH, an agricultural waste rich in silicon resources, should also attract more attention. supported by the Shanghai Education Development Foundation and the Shanghai Municipal Education Commission (no. 21SG42). Author contributions: F.T. wrote the manuscript. X.Zh., W.N., F.W., and X.X. provided valuable suggestions. G.L., Q.X., and X.L. revised the manuscript. Z.P., S.H., L.J., X.Zo., and X.L. supervised and revised the manuscript. All authors approved the final manuscript. Competing interests: The authors declare that they have no competing interests.