Silicon electrodeposition from chloride – fluoride melts containing K 2 SiF 6 and SiO 2

Silicon electrodeposition on glassy carbon from KF–KCl–K2SiF6, KF–KCl–K2SiF6–KOH and KF–KCl–K2SiF6–SiO2 melts was studied by cyclic voltammetry. The electroreduction of Si(IV) to metallic Si was observed as a single 4-electron wave under all the considered conditions. The reactions of cathode reduction of silicon from fluoride and oxyfluoride complexes were suggested. It was shown that the process could be controlled by the preliminary transformation of SiO4 to SiF6 and SiOxFy. The influence of the current density on the structure and morphology of silicon deposits obtained during galvanostatic electrolysis of the KF–KCl–K2SiF6–SiO2 melt was studied.


INTRODUCTION
Development of renewable technologies for power generation is directly connected with the achievements in the field of the manufacture of new constructive and functional silicon materials.2][3][4][5] The advantages of the electrochemical method are the possibility to control the structure and properties of deposits, [4][5][6][7][8] relatively simple equipment and low energy consumption.
The effective control of the electrocrystallization process requires a detailed study of many aspects, including those related to the regularities of nucleation/ /growth stage on an indifferent cathode, 9,10 the influence of the electrodeposition conditions (melt composition, substrate material and structure, temperature, etc.) on the process mechanism and kinetics, and the structure and morphology of cathode deposits.10][11][12][13][14][15][16][17][18][19][20] The KF-KCl-K 2 SiF 6 is one of the most prospective electrolytes, as it has sufficient thermal stability, it is less aggressive than a purely fluoride one, and it is water soluble, which facilitates separation of the salt and silicon phases of the cathode deposit. 7,8,18,19The mechanism of silicon electroreduction in this melt was studied previously. 10,18- 20It was found that during silicon electrodeposition from a KF-KCl (2:1)-0.2mol K 2 SiF 6 melt on a glassy carbon at 1023 K 10 and from KF-KCl (45:55 mole ratio)-(0.5-5mol %) K 2 SiF 6 on silver at 923 K, 18,19 a single stage Si(IV) discharge with the simultaneous transfer of 4 electrons occurred.However, another paper 20 reported a complex mechanism with the formation of intermediate silicon compounds during deposition from (KF-KCl) eutectic -0.4 mol % K 2 SiF 6 on silver at 933 K.
Silicon electrodeposition from a KF-KCl-K 2 SiF 6 -SiO 2 melt requires further study.There are a few works devoted to the study of the physico-chemical properties of this melt 21 and the interaction between SiO 2 and KF-KCl-K 2 SiF 6 . 22ilicon deposits on graphite were obtained during the electrolysis of a KF-KCl--K 2 SiF 6 -SiO 2 melt. 23,24It was shown that deposits with a high specific surface were formed when silicon dioxide was present in the melt.A more detailed study of the electrocrystallization silicon from the KF-KCl-K 2 SiF 6 -SiO 2 melt is of great interest both for defining the influence of oxygen-containing additions on the mechanism of the cathode process and for evaluating the possibility of performing the process in air and using more accessible raw materials (SiO 2 ).
The purpose of the present work was to define the influence of oxygen-containing additions (SiO 2 , KOH) on the mechanism of silicon electroreduction from the KF-KCl-K 2 SiF 6 -based melts and the impact of the current density on structure and morphology of the silicon deposits obtained during electrodeposition from a KF-KCl-K 2 SiF 6 -SiO 2 melt under galvanostatic conditions.
The prepared electrolyte was placed into the glassy carbon crucible (3), heated in vacuum to 573 K and maintained at this temperature for 3 h.Then the cell was filled with argon, the temperature increased and a stepwise potentiostatic purifying electrolysis was performed on the graphite electrode to remove all admixtures and traces of moisture in the electrolyte.Subsequently, the graphite electrode was removed from the cell via the sluice device (13) and the glassy carbon working electrode was inserted (15).
The cyclic voltammograms were recorded by an Autolab PGStat 302N potentiostat/galvanostat with Nova 1.5 software.Before measuring, the electrodes were kept for 30 min in the melt to set the potential.The initial values of the electrode potential were controlled before each series of measurements (the difference did not exceed 5 mV).The resistance of the measuring circuit was detected by the impedance method and compensated instrumentally.The temperature of the melt was controlled by a Pt-Pt/Rh thermocouple with an accuracy of ±1 °C.
The silicon deposits were obtained during electrolysis of the KF-KCl (2:1)-(10 mol %) K 2 SiF 6 -(2-3 mol %) SiO 2 melt under galvanostatic conditions.The experimental technique was described in detail previously. 23The obtained deposits were separated from the electrode using a puller made of instrumental steel, ground, washed with an aqueous HCl (1.0 mol L -1 ) solution at 353 K until complete removal of the electrolyte and then dried in a Pro-Analytical centrifuge at a rotation speed to 6000 rpm (Centurion Scientific Ltd., UK).
The contents of the admixtures in the melt (before and after the experiment) and in the deposits were detected by inductively coupled plasma atomic emission spectroscopy using an iCAP 6300 Duo instrument (Thermo Scientific, USA).A Sorbi N.4.1 (Meta, RF) was used to determine the specific surface of the powders.X-Ray spectral microanalysis (EDS) and SEM imaging of the samples were performed using a JMS-5900LV scanning electron microscope (Jeol, Japan).XRD analysis of the silicon powders with recognition of the crystallite sizes was realized by means of Rigaku D/MAX 2200VL/PC defractometer (Rigaku, Japan).

Study of the silicon electroreduction mechanism
Typical cyclic voltammograms obtained during silicon electroreduction from a KF-KCl-K 2 SiF 6 melt are presented in Fig. 2. All i-E dependencies demonstrated only one cathode peak, which was related to diffusion to the deposited silicon, and a corresponding anode peak of silicon dissolution.Such a shape of the curves is typical for a single stage Si(IV) discharge.Increasing the temperature (Fig. 2A) expectedly led to a growth of the cathode peak, i p , which shifted to the region of the less negative potentials due to mass transfer acceleration.Increasing the silicon concentration in the melt (Fig. 2B) caused a proportional increase in i p , which enabled the cathode peaks to be associated with the process of silicon electrodeposition.The cyclic voltammograms recorded at different scan rates, v, are illustrated in Fig. 3.The i p dependence on v 1/2 , which was obtained over a wide range of scan rates (from 0.01 to 1 V s -1 ), was not linear.Such behavior may be observed in systems with low standard rate constant of the electrode process. 25Then, at a relatively low scan rates, the process may be controlled by diffusion.Previously, it was shown that at low Si ions concentration in the melt and v < 0.1 V s -1 , the process was controlled by diffusion. 10Under diffusion control conditions, some authors 12,13 used the Berzins-Delahey 26 Equation to calculate the diffusion coefficient of the deposited ions, and the Mamantov Equation 27 to detect the number of electrons.However, these equations are based on the assumption that a layer of the new phase was initially present on the electrode. 26,27This condition did not exist in the present case as the processes of silicon crystals nucleation and growth occurs on the surface of the indifferent electrode after application of the potential. 10The initial stages of the electrocrystallization (nucleation/growth processes) may influence significantly the value and current peak location and for this reason, the experimental curves were not analyzed quantitatively using the Berzins-Delahey and Mamantov Equations.The in situ Raman spectroscopy 22 in the KF-KCl-K 2 SiF 6 melt revealed that silicon is stable in the Si 4+ state.Therefore, the cathode process in this case may be presented as follows: The cyclic voltammograms obtained during silicon electrodeposition from a KF-KCl-K 2 SiF 6 -SiO 2 melt are shown in Fig. 4. The i p increase corresponded to the increase in the silicon concentration in the melt containing 3.11•10 -5 mol cm -3 of SiO 2 .However, a larger addition of silicon dioxide (curve 3, Fig. 4) did not result in a proportional increase of the signal.Moreover, in both cases, the common shape of the curve did not change.

Z H U K et al.
A previous paper 22 reported that during the interaction of silicon dioxide with a fluoride-chloride melt, fluoride, oxyfluoride and silicate silicon complexes are formed.In a KF-KCl-K 2 SiF 6 melt saturated with silica, the dissolution process was decelerated and a noticeable number of oxyfluoride and fluoride silicon complexes appeared only after the melt was sustained for 100 min.Thus, it may be assumed that when the large amount of SiO 2 (curve 3, Fig. 4) was added, some of the silicon ions belonging to the SiO 4 4-groups did not participate in the electrode process.To prove this hypothesis, an extra test to define the effect of the SiO 2 exposure time in the melt on the value of the current peak was performed.The experimental results (Fig. 5) showed that increasing the exposure time progressively increased the current peak.This data is in a good agreement with the conclusions reached in a previous study.During electrodeposition from a KF-KCl-K 2 SiF 6 -SiO 2 melt, the shape of the cyclic voltammograms was affected by both increasing silicon concentration and the addition of oxygen-containing ions.To evaluate the impact of the later factor, silicon electrodeposition from a KF-KCl-K 2 SiF 6 -KOH melt was performed.It could be seen from Fig. 6 that the addition of 2.2 mol % KOH had no significant influence on the shape of the cathode part of the i-E curve.However, current peaks in the potassium hydroxide-containing melt were noticeably smaller than those in KF-KCl-K 2 SiF 6 under otherwise identical conditions.Probably, this was due to the appearance of SiO 4 4-in the melt, according to the reaction: According to the literature, 22 the interaction between SiO 4 4-and F -should lead to Si-O bond breakage and the formation of Si-F bonds, i.e., to the formation of oxyfluoride and fluoride silicon groups.To prove this mechanism, the influence of the exposure time of potassium hydroxide in the melt on the cyclic voltammograms was studied (Fig. 7).As expected, the increase in exposure time resulted in an increase in the current peak.
Therefore, in KF-KCl-K 2 SiF 6 -KOH and KF-KCl-K 2 SiF 6 -SiO 2 , the cathode reduction of silicon was preceded by the chemical transformation of SiO 4 4-into oxyfluoride and fluoride complexes.If the K 2 SiF 6 concentration was low when compared to the concentration of the oxygen-containing component, the following process scheme may be suggest: Z H U K et al.

Silicon electrodeposition
Silicon deposits were obtained during electrolysis of a KF-KCl (2:1)-10 mol % K 2 SiF 6 -(2-3 mol %) SiO 2 melt under galvanostatic conditions (0.02-1.5 А cm -2 ) at 1023 K.A silicon-salt "pear" formed on the cathode; cleaned from the electrolyte, the powder-like deposit had a color from dark yellow to light brown.XRD analysis showed that the deposits were a homogeneous silicon phase.A typical diffraction pattern is shown in Fig. 8.The SEM images of the deposits obtained at different current densities are presented in Figs. 9 and 10.Silicon was found to crystallize into fibers with an average diameter of 150-200 nm at current densities below 0.1 A cm -2 (Fig. 9).The formation of fiber and spongy-fibrous deposits was observed at current densities of 0.1-0.25 А cm -2 (Fig. 10А and B) and at higher current densities, the fiber structures degenerated (Fig. 10C and D).
The data such as those presented in Fig. 9B, allowed the suggestion that the silicon fibers are polycrystalline.The sizes of metal and semiconductor grains in  the 10-200 nm range can be accurately detected by the X-ray crystallography method.The deposit characteristics obtained by X-ray crystallographic analysis are presented in Table I.The calculated values of the densities of the deposits corresponded to those for elementary silicon, within the error of the method.The Z H U K et al.
lattice parameter of deposits obtained at current densities less than 0.1 А•cm -2 , corresponded to the reference value.For deposits obtained at higher current densities, the lattice parameter and the elementary cell volume slightly increased, which was probably due to character changes of the electrode process (at cathode current densities higher than 0.1 А•cm -2 , silicon and potassium co-deposition occurs). 22,23The data regarding the specific surface of the deposits totally agreed with the electron microscopy data.The deposits obtained at low current densities had a higher specific surface.

CONCLUSIONS
The influence of oxygen-containing additions on the mechanism of silicon electroreduction on glassy carbon during electrodeposition from KF-KCl--K 2 SiF 6 based melts was studied by cyclic voltammetry.The electroreduction of Si(IV) to metallic Si was observed as a single 4-electron wave under all considered conditions.However, at silicon electrodeposition from KF-KCl-K 2 SiF 6 --SiO 2 and KF-KCl-K 2 SiF 6 -KOH melts, both fluoride (SiF 6 2-) and oxyfluoride (SiO x F y z-) complexes may reduce on the cathode.Moreover, the process may be limited by preliminary SiO 4 4-transformation into SiF 6 2-and SiO x F y z-.
Silicon deposits were obtained by electrolysis of a KF-KCl (2:1)-10 mol % K 2 SiF 6 -(2-3 mol %) SiO 2 melt under galvanostatic conditions at 1023 K.It was found that polycrystalline silicon nanofibers form at low cathode current densities (0.02-0.1 А cm -2 ).Increasing the current density resulted in the formation of spongy-fibrous and powder deposits, and the grain size decreased.The deposits obtained at current densities up to 0.1 А cm -2 had the highest specific surface.
52Z H U K et al.
54 Z H U K et al.

Fig. 5 .
Fig. 5. Influence of the SiO 2 exposure time in the melt on the current peak.

Fig. 7 .
Fig. 7. Influence of the KOH exposure time in the melt on the current peak.

TABLE I .
Characteristics of the silicon deposits obtained at different current densities