A Carbon-Free Way for Obtaining Nanoscale Silicon

: The nanosized silicon powder has been produced by reduction of silica with magnesium in an argon medium using both the mechanically activated self-propagating high-temperature synthesis and the direct mechanochemical synthesis and has been investigated by X-ray phase analysis, Infrared spectroscopy, electron scanning microscopy, and energy dispersive X-ray spectroscopy. The optimal Mg:SiO 2 ratio has been found to provide the minimum content of contaminant impurities of magnesium silicide and silicate in mechanically activated self-propagating high-temperature synthesis. For the ﬁrst time, direct mechanochemical synthesis of Si via reduction of silica with magnesium has been implemented. Optimal component ratio and mechanical activation parameters have been determined, yielding Si/MgO composites without impurity phases (magnesium silicide and silicate). A puriﬁcation procedure has been proposed for separating silicon obtained from magnesium oxide and other impurity phases. The ratio of initial components has been determined, at which puriﬁed silicon has the least amount of impurities. The particle size of silicon powder obtained was 50–80 nm for the mechanically activated self-propagating high-temperature synthesis, and 30–50 nm for the direct mechanochemical synthesis.


Introduction
Nanostructured silicon is a promising material for lithium-ion batteries [1,2], photovoltaics systems [3,4], photocatalysis [5], nanoenergetics materials [6], and thermoelectrics [7,8]. Metallurgical-grade silicon is industrially produced by carbothermal reduction of silica [9]. However, this process cannot fabricate nanostructured material as the temperature of carbothermic reduction (over 1900 • C) is higher than the melting point of silicon (1414 • C). In addition, this process is multistage, energy consuming, and rather dangerous ecologically. Therefore, the development of alternative time-and energy-saving technologies that also reduce the cost of the final product is an extremely desirable task.
One of the most common carbon-free methods for obtaining silicon for most highly exothermic systems is self-propagating high-temperature synthesis (SHS) [10][11][12]. However, in systems where the combustion temperatures are significantly higher than the melting temperatures of the reactants (first of all, magnesium-and aluminothermic processes), the preparation of powder nanosized products is a serious problem. Usually, a decrease in the combustion temperature in SHS is achieved by diluting the reaction mixture with the reaction products [13][14][15][16].
It was shown that for some high-energy systems, reactions can proceed in SHS or mechanically activated self-propagating high-temperature synthesis (MA SHS) mode but cannot be implemented purely mechanochemically (in an activator). For example, the reduction of SiO 2 with aluminum was carried out by the MA SHS process [27,31], but it was not possible to reduce SiO 2 mechanochemically to silicon under any MA regimes [31].
The purpose of this work was to study the possibility of direct mechanochemical reduction of silica with magnesium to obtain nanosized silicon powder, as well as to compare the structural and morphological characteristics of products obtained by different methods using mechanochemistry.
Mixtures of silica and metallic magnesium with different component ratios were processed in a high-energy planetary ball mill AGO-2 with water cooling (drum volume 250 cm 3 , steel ball diameter 5 mm, ball loading 200 g, sample weight 10 g, and the rotation speed of drums around the common axis~1000 rpm or~600 rpm, which corresponds to a centrifugal acceleration 60 g and 20 g respectively).
An upper-drive mixer RZR 1 (HEIDOLPH, Schwabach, Germany) with rotation speed 1000 rpm was used during the acid treatment of MA SHS samples. The separation of silicon from impurities for the products of both MA SHS and direct mechanochemical reduction was carried out by three-stage acid treatment [32]. The sample was sequentially treated with a 2 M solution of HCl; with a mixture of 2 M HCl and 20-25% CH 3 COOH; with a mixture of 5% HF and 20-25% CH 3 COOH. Acid treatment conditions were τ = 1 h, T = 70-80 • C. After each acid treatment stage, the suspension was centrifuged and the precipitate was washed with distilled water and extracted from the aqueous phase by decantation.
Infrared (IR) spectra were registered on a Tensor-27 spectrometer (Bruker Optik GmbH, Ettlingen, Germany) in the wavenumber range of 4000-400 cm −1 . The samples for the study were compacted with calcinated potassium bromide [33].
X-ray investigations were carried out on a D8 Advance diffractometer (Bruker AXS GmbH, Karlsruhe, Germany, CuK α -radiation, λ = 1.5418 Å). Phase composition and crystal structure of the samples were determined by X-ray diffraction data using the database of the International Centre for Diffraction Data (ICDD) PDF4 [34].
Morphology of obtained samples was studied using scanning electron microscopes (SEM): Hitachi TM 1000 (Tokyo, Japan) (at accelerating voltage of 15 kV) equipped with a TM1000 EDS detector to determine the chemical composition of the sample and highresolution SEM by MIRA\TESCAN with an energy-dispersive X-ray (EDX) attachment by OXFORD INSTRUMENTS (Abingdon, Oxfordshire, UK).
The content of impurity elements in the obtained silicon powders was studied using an atomic emission spectrometer with inductively coupled plasma AKTIVA M («HORIBA Jobin Yvon S.A.S», Longjumeau, France) (elements from F to U, detection limit from 0.1 ppb).

Study of the MA SHS Reduction of Silica by Magnesium
Mechanochemical reactions of silica reduction with magnesium were carried out under different modes of mechanical activation and with different component ratios.
According to IR spectroscopy data, when processed for 40 s in a high-energy ball mill with a maximum load of 60 g, the product mainly contains magnesium silicate Mg 2 SiO 4 ( Figure 1, curve 2).

Study of the MA SHS Reduction of Silica by Magnesium
Mechanochemical reactions of silica reduction with magnesium were carried out under different modes of mechanical activation and with different component ratios.
According to IR spectroscopy data, when processed for 40 s in a high-energy ball mill with a maximum load of 60 g, the product mainly contains magnesium silicate Mg2SiO4 (Figure 1, curve 2).
When the rotation speed decreases to 600 rpm and the activation duration is less than 2 min, Mg2SiO4 is not formed and SiO2 bands are kept in the spectrum. Under such treatment conditions, formation of silicates begins after 4 min of activation. The mechanical activation mode (duration 40 s at a load of 20 g) worked-off for the stoichiometric composition Mg/SiO2 = 2:1 was applied for the preparation of Mg/SiO2 precursors with other component ratios also used for further SHS processes: 1. Molar ratio of Mg:SiO2 = 1.5:1; 2. Molar ratio of Mg:SiO2 = 2:1, stoichiometry; 3. Molar ratio of Mg:SiO2 = 2.5:1; 4. Molar ratio of Mg:SiO2 = 3:1; 5. Molar ratio of Mg:SiO2 = 4:1.
The X-ray diffraction (XRD) patterns of Mg/SiO2 precursors with different component ratios ( Figure 2) are consistent with the IR spectroscopy data. When the rotation speed decreases to 600 rpm and the activation duration is less than 2 min, Mg 2 SiO 4 is not formed and SiO 2 bands are kept in the spectrum. Under such treatment conditions, formation of silicates begins after 4 min of activation.
The mechanical activation mode (duration 40 s at a load of 20 g) worked-off for the stoichiometric composition Mg/SiO 2 = 2:1 was applied for the preparation of Mg/SiO 2 precursors with other component ratios also used for further SHS processes:
The X-ray diffraction (XRD) patterns of Mg/SiO 2 precursors with different component ratios ( Figure 2) are consistent with the IR spectroscopy data.
X-ray phase analysis of the mechanochemically obtained precursors did not show any traces of phase transformations during mechanical activation at these durations. Only intensities of magnesium diffraction reflections change with MA ( Figure 2), and SiO 2 used in this work is X-ray amorphous. X-ray phase analysis of the mechanochemically obtained precursors did not show any traces of phase transformations during mechanical activation at these durations. Only intensities of magnesium diffraction reflections change with MA ( Figure 2), and SiO2 used in this work is X-ray amorphous.
The excessive magnesium content was added to speed up the heat removal during the SHS process, resulting in reduction of combustion temperature.
Investigation of the SHS process in mechanochemically obtained Mg/SiO2 composites as the precursors showed that even a short-term MA (20-60 g) allows igniting the combustion in powder mixtures without preheating. For all compositions studied, a very rapid heat evolution takes place in the combustion front (due to the magnesiothermal reaction) (Figure 3).   Figure 3 shows that an increase in the MA duration leads to a decrease in the ignition delay time td. Thus, after MA for 20 s (at 1000 rpm), the td ≈ 80 s, while after 60 s MA (at 600 rpm), the initiation of the reaction is observed after ≈40 s. The maximum combustion temperature (TC ~ 1283-1288 °C) is achieved for the stoichiometric composition (sample The excessive magnesium content was added to speed up the heat removal during the SHS process, resulting in reduction of combustion temperature. Investigation of the SHS process in mechanochemically obtained Mg/SiO 2 composites as the precursors showed that even a short-term MA (20-60 g) allows igniting the combustion in powder mixtures without preheating. For all compositions studied, a very rapid heat evolution takes place in the combustion front (due to the magnesiothermal reaction) ( Figure 3). X-ray phase analysis of the mechanochemically obtained precursors did not show any traces of phase transformations during mechanical activation at these durations. Only intensities of magnesium diffraction reflections change with MA ( Figure 2), and SiO2 used in this work is X-ray amorphous.
The excessive magnesium content was added to speed up the heat removal during the SHS process, resulting in reduction of combustion temperature.
Investigation of the SHS process in mechanochemically obtained Mg/SiO2 composites as the precursors showed that even a short-term MA (20-60 g) allows igniting the combustion in powder mixtures without preheating. For all compositions studied, a very rapid heat evolution takes place in the combustion front (due to the magnesiothermal reaction) (Figure 3).     Figure 3 shows that an increase in the MA duration leads to a decrease in the ignition delay time t d . Thus, after MA for 20 s (at 1000 rpm), the t d ≈ 80 s, while after 60 s MA (at 600 rpm), the initiation of the reaction is observed after ≈40 s. The maximum combustion temperature (T C~1 283-1288 • C) is achieved for the stoichiometric composition (sample no. 2) ( Figure 3). According to [35], the adiabatic temperature of the reaction 2Mg (s) + SiO 2 (s) → 2MgO (s) + Si for the stoichiometric composition (45% Mg) is 2123 K (1850 • C). At the same time, the calculation of the adiabatic temperature in the ISMAN-TERMO program, presented in [13], gives a temperature of 1900 • C (it decreases when diluents are used). In our case, for a stoichiometric composition, the maximum combustion temperature is T C~1 283-1288 • C, which is lower than the calculated adiabatic temperature given above. This is due to the complex and not yet fully disclosed mechanism of the influence of MA on the combustion parameters [22,36,37].
It is evident from Figure 3 that an increase in the MA duration results in the essential decrease in the ignition delay time t d even for a lower intensive mode of MA. Thus, t d is about 80 s for 20 s of MA at a load of 60 g, while the delay time of about 40 s is observed after MA for 1 min at a load of 20 g. Herewith, the propagation velocity of the combustion wave for a longer but less intensive MA mode is significantly lower (0.35 mm/s compared to 1.03 mm/s for a load of 60 g and 20 s of MA).
In a sample with a lack of magnesium (no. 1) T C is slightly lower (~1100 • C). Combustion temperature in samples with excess magnesium content is about 1020-1050 • C.
Analysis of the phase composition of combustion products in the systems under investigation showed that at a molar ratio of Mg:SiO 2 from 1.5 to 4, the products of MA SHS, except Si and MgO, contain a small amount of Mg 2 Si and Mg 2 SiO 4 .
Herewith, the duration of MA has a greater effect on the relative content of magnesium silicate than the mill power in the studied range of values (20-60 g), and the phase composition of the synthesis products is mainly dependent on the component ratio of the charge mixture ( Figure 4).
SiO2 (s) → 2MgO (s) + Si for the stoichiometric composition (45% Mg) is 2123 K (1850 °C). At the same time, the calculation of the adiabatic temperature in the ISMAN-TERMO program, presented in [13], gives a temperature of 1900 °С (it decreases when diluents are used). In our case, for a stoichiometric composition, the maximum combustion temperature is ТC ~ 1283-1288 °С, which is lower than the calculated adiabatic temperature given above. This is due to the complex and not yet fully disclosed/described mechanism of the influence of MA on the combustion parameters [22,36,37].
It is evident from Figure 3 that an increase in the MA duration results in the essential decrease in the ignition delay time td even for a lower intensive mode of MA. Thus, td is about 80 s for 20 s of MA at a load of 60 g, while the delay time of about 40 s is observed after MA for 1 min at a load of 20 g. Herewith, the propagation velocity of the combustion wave for a longer but less intensive MA mode is significantly lower (0.35 mm/s compared to 1.03 mm/s for a load of 60 g and 20 s of MA).
In a sample with a lack of magnesium (no. 1) TC is slightly lower (~1100 °C). Combustion temperature in samples with excess magnesium content is about 1020-1050 °C.
Analysis of the phase composition of combustion products in the systems under investigation showed that at a molar ratio of Mg:SiO2 from 1.5 to 4, the products of MA SHS, except Si and MgO, contain a small amount of Mg2Si and Mg2SiO4.
Herewith, the duration of MA has a greater effect on the relative content of magnesium silicate than the mill power in the studied range of values (20-60 g), and the phase composition of the synthesis products is mainly dependent on the component ratio of the charge mixture ( Figure 4). According to the ХRD data, the main products of SHS are silicon and magnesium oxide. The formation of some amounts of silicate and magnesium silicide is also confirmed. The Mg2Si content significantly increases with the excess of magnesium, and the greatest intensity of its reflexes is achieved at Mg/SiO2 = 2.5:1 (Figure 4, curve 3). The strongest peak of Mg2Si in curve 3 can be explained by the high rate of silicon formation at the presence of the excessive magnesium and rapid interaction of small particles of the reduced silicon with magnesium in the mechanically activated material (nanocomposite). With an increase in the magnesium excess, the relative silicon content and the rate of silicon formation, as well as contact surface between the reactants, decrease. Therefore, the formation of magnesium silicide is suppressed, which is observed in curves 4 and 5. The According to the XRD data, the main products of SHS are silicon and magnesium oxide. The formation of some amounts of silicate and magnesium silicide is also confirmed. The Mg 2 Si content significantly increases with the excess of magnesium, and the greatest intensity of its reflexes is achieved at Mg/SiO 2 = 2.5:1 (Figure 4, curve 3). The strongest peak of Mg 2 Si in curve 3 can be explained by the high rate of silicon formation at the presence of the excessive magnesium and rapid interaction of small particles of the reduced silicon with magnesium in the mechanically activated material (nanocomposite). With an increase in the magnesium excess, the relative silicon content and the rate of silicon formation, as well as contact surface between the reactants, decrease. Therefore, the formation of magnesium silicide is suppressed, which is observed in curves 4 and 5. The concentration of magnesium silicate is the lowest at the component ratio 2.5:1, i.e., the lowest intensity of Mg 2 SiO 4 reflexes is identified in X-ray diffraction patterns of MA SHS products as compared with that of other Mg:SiO 2 ratios (Figure 4, curve 3). It should be noted that it is easier to perform chemical separation of silicon from silicide than from magnesium silicate. concentration of magnesium silicate is the lowest at this component ratio, i.e., the lo intensity of Mg2SiO4 reflexes is identified in X-ray diffraction patterns of MA SHS p ucts as compared with that of other Mg:SiO2 ratios (Figure 4, curve 3). It should be n that it is easier to perform chemical separation of silicon from silicide than from ma sium silicate. Figure 5a-c show the microstructure and elemental mapping of the MA SHS pro in a stoichiometric reaction mixture after MA at 60 g for 20 s. Rounded silicon par ranging in size from 1 to 15 μm are clearly distinguished in the matrix of magne oxide. Relatively large regions containing all the elements of the composition (obvio Mg2SiO4) with the diffused boundaries of individual Si and MgO particles are reve after MA at 20 g for 60 s (Figure 5d-f).
The IR spectra of the MA SHS products indicate the formation of magnesium sili of larger or smaller quantities in all mixtures investigated. Based on the XRD results, a certain amount of silicates and magnesium silicide identified in the products after leaching with hydrochloric acid. For a more comple moval of magnesium compounds, the second stage of treatment of the sample with a ture of acetic and hydrochloric acids was carried out. The addition of acetic acid red the adsorption of metal cations on silicon powder.
The treatment of precipitations with a mixture of hydrofluoric and acetic acids used at the final purification stage.
According to the X-ray phase analysis of silicon powders obtained from all the positions, MgO, Mg2Si, and Mg2SiO4 reflexes are not observed in the XRD patterns three-stage purification ( Figure 6). The IR spectra of the MA SHS products indicate the formation of magnesium silicates of larger or smaller quantities in all mixtures investigated.
Based on the XRD results, a certain amount of silicates and magnesium silicide was identified in the products after leaching with hydrochloric acid. For a more complete removal of magnesium compounds, the second stage of treatment of the sample with a mixture of acetic and hydrochloric acids was carried out. The addition of acetic acid reduces the adsorption of metal cations on silicon powder.
The treatment of precipitations with a mixture of hydrofluoric and acetic acids was used at the final purification stage.
According to the X-ray phase analysis of silicon powders obtained from all the compositions, MgO, Mg 2 Si, and Mg 2 SiO 4 reflexes are not observed in the XRD patterns after three-stage purification ( Figure 6).
EDX analysis of all the investigated silicon powders after purification did not reveal a noticeable presence of impurities (within the limits of detection errors).
Further studies of the impurity content in the as-obtained silicon powders were carried out using a certified atomic emission spectrometer with inductively coupled plasma "ACTIVA M" (France). The error of the method was 3-5% relative.
The data on the content of impurity elements in silicon powders after the threestage acid treatment are presented in Table 1. As can be seen from Table 1, the purity of silicon strongly depends on the ratio of the initial components in the reaction mixture. In the powder obtained from the reaction mixture of stoichiometric composition, a certain amount of magnesium (1.1%) is retained after leaching. In the sample obtained at the ratio Mg/SiO 2 = 2.5:1, the magnesium content is at the level of other impurity elements. Obviously, this is due to the predominant formation of magnesium silicide and insignificant content of Mg 2 SiO 4 silicate in the synthesis product. Thus, the purity of the silicon powder obtained at the stoichiometric ratio of the components of the reaction mixture is 98.45%. The powder obtained at the ratio Mg/SiO 2 = 2.5:1 is much more pure (99.57%) with the same purification technology.

EDX analysis of all the investigated silicon powders after purification did no a noticeable presence of impurities (within the limits of detection errors).
Further studies of the impurity content in the as-obtained silicon powders w ried out using a certified atomic emission spectrometer with inductively coupled "ACTIVA M" (France). The error of the method was 3-5% relative.
The data on the content of impurity elements in silicon powders after the thre acid treatment are presented in Table 1. As can be seen from Table 1, the purity of strongly depends on the ratio of the initial components in the reaction mixture powder obtained from the reaction mixture of stoichiometric composition, a amount of magnesium (1.1%) is retained after leaching. In the sample obtained at t Mg/SiO2 = 2.5:1, the magnesium content is at the level of other impurity elements ously, this is due to the predominant formation of magnesium silicide and insig content of Mg2SiO4 silicate in the synthesis product. Thus, the purity of the silicon p obtained at the stoichiometric ratio of the components of the reaction mixture is 9 The powder obtained at the ratio Mg/SiO2 = 2.5:1 is much more pure (99.57%) w same purification technology.   Preliminary studies of morphology of purified silicon samples by the SEM at the lower magnifications showed that the average size of the particle agglomerates is about 1-10 µm. SEM micrographs at the higher magnifications show that the agglomerated powders consist of smaller particles with a small size range (Figure 7).
Powders 2022, 2, FOR PEER REVIEW 8 Preliminary studies of morphology of purified silicon samples by the SEM at the lower magnifications showed that the average size of the particle agglomerates is about 1-10 μm. SEM micrographs at the higher magnifications show that the agglomerated powders consist of smaller particles with a small size range (Figure 7).

Study of the Mechanochemical Reduction of Silica by Magnesium
The effects of mechanical treatment duration (in the range from 1 to 8 min at a load value of 60 g) and the component ratios in the Mg-SiO2 system on the reduction of silica with magnesium were studied by XRD analysis. After mechanical activation for 1 min of the sample of stoichiometry composition (Mg:SiO2 = 2:1) the reflections of the initial magnesium, magnesium oxide, silicon, silicide, and magnesium orthosilicate can be identified in the XRD patterns (Figure 9a, curve 1). An increase in activation duration to 2-4 min (Figure 9a, curves 2 and 3) leads to a decrease in the intensity and broadening of the reflexes of all these phases. When the activation time is increased to 8 min, widened reflexes of silicon and magnesium oxide as well as reflexes of magnesium silicide of low intensity are identified (Figure 9a, curve 4). Preliminary studies of morphology of purified silicon samples by the SEM at the lower magnifications showed that the average size of the particle agglomerates is about 1-10 μm. SEM micrographs at the higher magnifications show that the agglomerated powders consist of smaller particles with a small size range (Figure 7). The agglomerates of samples with molar ratio Mg:SiO2 = 2:1 and 2.5:1 consist of nanodispersed particles with the particle size about 50-80 nm (Figure 8).

Study of the Mechanochemical Reduction of Silica by Magnesium
The effects of mechanical treatment duration (in the range from 1 to 8 min at a load value of 60 g) and the component ratios in the Mg-SiO2 system on the reduction of silica with magnesium were studied by XRD analysis. After mechanical activation for 1 min of the sample of stoichiometry composition (Mg:SiO2 = 2:1) the reflections of the initial magnesium, magnesium oxide, silicon, silicide, and magnesium orthosilicate can be identified in the XRD patterns (Figure 9a, curve 1). An increase in activation duration to 2-4 min (Figure 9a, curves 2 and 3) leads to a decrease in the intensity and broadening of the reflexes of all these phases. When the activation time is increased to 8 min, widened reflexes of silicon and magnesium oxide as well as reflexes of magnesium silicide of low intensity are identified (Figure 9a, curve 4).

Study of the Mechanochemical Reduction of Silica by Magnesium
The effects of mechanical treatment duration (in the range from 1 to 8 min at a load value of 60 g) and the component ratios in the Mg-SiO 2 system on the reduction of silica with magnesium were studied by XRD analysis. After mechanical activation for 1 min of the sample of stoichiometry composition (Mg:SiO 2 = 2:1) the reflections of the initial magnesium, magnesium oxide, silicon, silicide, and magnesium orthosilicate can be identified in the XRD patterns (Figure 9a, curve 1). An increase in activation duration to 2-4 min (Figure 9a, curves 2 and 3) leads to a decrease in the intensity and broadening of the reflexes of all these phases. When the activation time is increased to 8 min, widened reflexes of silicon and magnesium oxide as well as reflexes of magnesium silicide of low intensity are identified (Figure 9a, curve 4).
Investigation of the effect of activation duration (at a load of 60 g and component ratio of Mg:SiO 2 = 2.5:1) showed that magnesium reflexes are mainly identified in the X-ray diffraction pattern of the sample after the activation for 1 min (Figure 9b, curve 1). After 2 min activation (Figure 9b, curve 2), intensive reflections of silicon and magnesium oxide together with low intensive peaks of silicide and magnesium orthosilicate appear in the X-ray diffraction patterns, while the intensity of magnesium reflections sharply decreases. A further increase in the activation time from 4 to 8 min (Figure 9b, curves 3 and 4) results in broadening and a decrease in the intensity of reflections of silicon, magnesium oxide, and magnesium. Investigation of the effect of activation duration (at a load of 60 g and component ratio of Mg:SiO2 = 2.5:1) showed that magnesium reflexes are mainly identified in the Xray diffraction pattern of the sample after the activation for 1 min (Figure 9b, curve 1). After 2 min activation (Figure 9b, curve 2), intensive reflections of silicon and magnesium oxide together with low intensive peaks of silicide and magnesium orthosilicate appear in the X-ray diffraction patterns, while the intensity of magnesium reflections sharply decreases. A further increase in the activation time from 4 to 8 min (Figure 9b, curves 3 and 4) results in broadening and a decrease in the intensity of reflections of silicon, magnesium oxide, and magnesium.
To separate silicon from MgO and other byproducts presented in the composition of mechanocomposites, investigations of the treatment of samples with acid solutions were carried out. As in the case of MA SHS, a three-stage acid treatment of the products of mechanochemical reduction of silica with magnesium was carried out.
The XRD data show that after three-stage acid treatment, silicon is purified from the main impurities, such as MgO, Mg2Si, and Mg2SiO4 ( Figure 10). To separate silicon from MgO and other byproducts presented in the composition of mechanocomposites, investigations of the treatment of samples with acid solutions were carried out. As in the case of MA SHS, a three-stage acid treatment of the products of mechanochemical reduction of silica with magnesium was carried out.
The XRD data show that after three-stage acid treatment, silicon is purified from the main impurities, such as MgO, Mg 2 Si, and Mg 2 SiO 4 ( Figure 10). The results of SEM analysis of the powder morphology ( Figure 11) have shown that after separation from byproducts, silicon powders consist of primary particles with a size of ~30-50 nm, making the larger secondary particles -aggregates with sizes from 1 to 10 μm. The results of SEM analysis of the powder morphology ( Figure 11) have shown that after separation from byproducts, silicon powders consist of primary particles with a size of~30-50 nm, making the larger secondary particles -aggregates with sizes from 1 to 10 µm.
The results of SEM analysis of the powder morphology ( Figure 11) have shown that after separation from byproducts, silicon powders consist of primary particles with a size of ~30-50 nm, making the larger secondary particles -aggregates with sizes from 1 to 10 μm. Figure 11. SEM micrograph of purified highly dispersed silicon, produced by the direct mechanochemical reduction of silica by magnesium.

Discussion
It is well known that both self-propagating high-temperature synthesis (SHS) reactions [10][11][12] and mechanically stimulated reactions [38,39] can occur in high-energy systems. The main disadvantage in such systems is extremely high synthesis temperatures, as a result of which the reaction products are melted and coarse. The preliminary joint mechanical activation of initial components allows one to reduce the temperature in the combustion wave and to increase the rate of the overall process [21,36,37,40]. The main reasons for the decrease in the combustion temperature and the increase in the rates of Figure 11. SEM micrograph of purified highly dispersed silicon, produced by the direct mechanochemical reduction of silica by magnesium.

Discussion
It is well known that both self-propagating high-temperature synthesis (SHS) reactions [10][11][12] and mechanically stimulated reactions [38,39] can occur in high-energy systems. The main disadvantage in such systems is extremely high synthesis temperatures, as a result of which the reaction products are melted and coarse. The preliminary joint mechanical activation of initial components allows one to reduce the temperature in the combustion wave and to increase the rate of the overall process [21,36,37,40]. The main reasons for the decrease in the combustion temperature and the increase in the rates of chemical reactions in mechanically activated mixtures are structural changes, namely, the achievement of a high density of interfacial boundaries between initial components, an extremely high contact surface, and a very high concentration of defects due to a large number of atoms on juvenile surfaces and in near-surface layers [41][42][43][44][45]. As a result, the changes in mass transfer and phase formation mechanisms are observed. In the case of composite formation with several levels of heterogeneity, the rate of chemical reactions in the region of interfacial boundaries with nanocomposite structure is much higher compared to the rate of reactions with a micron scale of heterogeneity. These structural changes which occur as a result of mechanical activation contribute to solid-phase reactions in mechanically activated mixtures. A change in the chemical routes in these mixtures may include direct synthesis of the target phase without the formation of intermediates formed in non-activated systems, as well as a change in the type and sequence of reactions instead of those occurring during self-propagating high-temperature synthesis without prior mechanical activation. Thus, the finely dispersed composite particles formed at the stage of mechanical activation contribute to a heterogeneous crystallization in combustion wave and prevent the growth of product grains behind its front.
Our previous studies for the 4Al-3SiO 2 high-energy system revealed the effect of the size of SiO 2 nanoparticle aggregates on the starting temperature of chemical interaction and phase formation processes in SHS. It was found that, despite the nanosize of the initial silica powder, preheating is necessary to initiate the reaction in the non-activated reaction mixture of 4Al-3SiO 2 . The formation of Al/SiO 2 composite particles during the MA process allows initiating the combustion at room temperature. In this case, the phase composition of MA SHS product depends on the degree of dispersion of SiO 2 inclusions in the aluminum matrix. The presence of relatively large (micrometric and submicron) aggregates of SiO 2 nanoscale particles leads to the interaction of aluminum oxide formed at the initial stage of SHS with silica to form mullite. When SiO 2 is finely dispersed in aluminum, forming aggregates with dimensions of about 55 nm, the almost instantaneous (the life time of the liquid phase is ≈1 s) one-stage chemical reaction is observed which leads to the formation of Si and α-Al 2 O 3 .
A significant decrease of the starting temperature for the reaction and an increase of the reaction rate can ensure the realization of mechanostimulated reactions in the activator, as, for example, in the Fe 2 O 3 -Al [46], Fe 2 O 3 -Zr [47], and CuO-Zr [48] systems. In this work, a comparative study of MA SHS and MSR in the SiO 2 -Mg system was carried out.
The prospects to obtain silicon in the interaction of silica with magnesium depend primarily on the thermodynamic possibility of the reaction, as shown in [49][50][51].
The reactions that can occur in the Mg-SiO 2 system are presented in Table 2 [52]. Based on thermodynamic data, it should be expected that reactions 1-3 with the highest formation heats are preferable; the products of reactions 1-3, in addition to Si, are MgO, Mg 2 Si, and magnesium silicates. Therefore, there is a task of separating silicon from byproducts. The results of our studies of the combustion process in the Mg-SiO 2 system showed that in a non-activated reaction mixture with a reagent molar ratio Mg:SiO 2 = 2:1, the ignition without preheating of samples is almost impossible. The preheating temperature needed to initiate and realize the SHS process is about 400-550 • C.
The study of the SHS process, where mechanochemically obtained Mg/SiO 2 composites were used as precursors, showed that even a short-term MA (20-60 g) allows initiating the combustion process in powder mixtures without preheating. For all compositions studied, very rapid heating was observed due to the course of the magnesium-thermic reaction in the combustion front.
Our studies have shown that the combustion modes during the SHS of mechanically activated mixtures depend on the conditions of the preliminary MA, primarily energy intensity and duration of treatment. The MA of the mixture must be carried out at a rotation speed of 600 rpm for 1-2 min. An increase in the activation time at this rotation speed leads to a decrease in the ignition delay time.
Taking into account the fact that T C in all cases studied is above the melting point of magnesium (650 • C), but below the melting point of silica (1713 • C) and combustion products (melting points of silicon and magnesium oxide are 1410 • C and 2825 • C, respectively), it is obvious that in the process of SHS using the preliminary formed mechanocomposites, the so-called "combustion with an intermediate molten layer" is realized [22].
The main products of MA SHS are silicon and magnesium oxide; the formation of some amounts of magnesium silicate and silicide is also observed. Since the aim of this work was to obtain nanoscale silicon powder, it was necessary to separate it from byproducts. Hydrochloric acid was used to dissolve magnesium oxide. Some magnesium silicates and silicide were present in the products after leaching with hydrochloric acid. For a more complete removal of magnesium compounds, the second stage of sample treatment with a mixture of acetic and hydrochloric acids was carried out. The addition of acetic acid made it possible to reduce the adsorption of metal cations on silicon powder. For the final purification of silicon, treatment with a mixture of hydrofluoric and acetic acids was introduced. Three-stage acid treatment [32] is technologically simple to implement, allows one to scale the cleaning process, and provides the production of single-phase nanosize silicon with a particle size of about 50-80 nm and a high degree of silicon purification from Al, Ca, Cd, Cr, Fe, K, Mg, Mn, Ni, and Zn.
Previously obtained results on mechanostimulated reactions have shown that direct mechanochemical reduction of oxides is possible for highly exothermic systems [53][54][55][56][57]. In this paper, the influence of mechanical activation modes on the direct mechanochemical reduction of silica by magnesium was studied. It was shown that the reduction process depends on the energy intensity of the activator, the duration of activation, and the ratio of the components in the mixture. Complete reduction of silica by magnesium can be achieved at a drum rotation speed of ≥1000 rpm and activation duration of at least 8 min. The reaction products, as in the case of MA SHS, are silicon, magnesium oxide, and small amounts of Mg 2 Si and magnesium silicates. The three-stage acid treatment was carried out to purify Si from the main impurities, such as MgO, Mg 2 Si, and Mg 2 SiO 4 . The silicon powder obtained consists of primary particles with sizes of 30-50 nm, forming secondary particles-aggregates with sizes from 1 to 10 µm.

Conclusions
For the first time, direct mechanochemical synthesis of Si via reduction of silica with magnesium has been implemented. The optimal component ratio (Mg:SiO 2 = 2.5:1) and the MA parameters (1000 rpm, duration 8 min) have been determined, yielding Si/MgO mechanocomposites without impurity phases of magnesium silicide and magnesium orthosilicate.
SHS has been carried out with preliminary MA of the SiO 2 + 2Mg mixture. MA conditions (600 rpm, duration 40 s) have been determined, providing a decreased combustion temperature (1283-1288 • C) for the formation of SiO 2 /Mg composite structures. The optimal ratio of components (Mg:SiO 2 = 3:1) for the MA SHS with minimum content of accompanying impurity phases (magnesium silicide and magnesium silicate) has been found.
Three-stage acid treatment of products of both mechanochemical and MA SHS reduction of silica with magnesium allows the obtaining of single-phase nanosize silicon.
The size of powdered silicon particles obtained by the MA SHS is 50-80 nm. In the case of direct mechanochemical synthesis, the size of silicon particles is 30-50 nm.
Thus, the direct mechanochemical reduction of silica with magnesium is preferable compared to mechanically activated, self-propagating high-temperature synthesis, since the process is carried out in one stage and in a short time (no more than 8 min), the reduced products do not contain the Mg 2 Si and Mg 2 SiO 4 impurity phases, and purified silicon has smaller particle sizes.