Synthesis of MnSi_1.7 nanosheet bundles from CaSi₂ crystal powders using MnCl₂ in molten salt

The MnSi_1.7 nanosheet bundle with improved structural homogeneity was synthesized by thermal annealing at 800 °C for 10 h (Sample 3). To examine the growth evolution of the nanosheets, the structural characterization of the nanosheets at 600 °C for 0 h (Sample 1) and 800 °C for 0 h (Sample 2) was performed. In Sample 1, the CaSi₂ phase remained, and a few Si and MnSi phases were formed, as shown in Fig. 2(a). In Sample 2, the CaSi₂ peaks disappeared, and the major products were also Si and MnSi phases [Fig. 2(b)]. When the annealing time was increased to 10 h (Sample 3), the intensities of the MnSi_1.7 peaks increased, even though the small MnSi and Si peaks remained [Fig. 2(c)]. In contrast, the Si peaks were dominant in the XRD pattern of Sample 4 [Fig. 2(d)], which was synthesized at 800 °C for 10 h by using the source-material configuration (B). In this case, the MnSi_1.7 and MnSi peaks were almost invisible (Sample 4). The geometric configurations of the source materials affect the gas flow-path in the vapor phase. If the configuration is not appropriate, the Mn deposition on the Si nanosheets would be suppressed. The results indicate the growth of MnSi_1.7 with improved homogeneity in Sample 3 because of the high temperature and liquid molten salt environment.

Zoom In Zoom Out Reset image size

The effect of additional 10 h annealing at 600 °C was compared with the structural characterization of the vapor-phase-synthesized silicide nanosheet bundles. When the annealing time is increased to 10 h at this temperature, the CaSi₂ peaks disappeared, and the intensity of Si peaks increased (Sample 5), as shown in Fig. 3(b). However, the compositional inhomogeneity remained in the nanosheet bundles caused by the additional formation of Si and MnSi under the vapor-phase synthesis condition when compared with those of Sample 1 (600 °C, 0 h). In addition, peaks of the MnSi phase did not appear if NH₄Cl was added (Sample 6), when compared with those of Sample 5 (600 °C, 10 h without NH₄Cl), as shown in Fig. 3(c). The effect of NH₄Cl on the thermal reaction of CaSi₂ with MnCl₂ will be discussed later.

Zoom In Zoom Out Reset image size

All the products were synthesized in powder form (Fig. 4), which maintained the dimensions of the original CaSi₂ powders reported in a literature. ²⁸⁾ The low magnification SEM images and EDS mappings of Samples 1, 2, and 3 with the source-material configuration (A) are shown in Figs. 4(a)–4(c), respectively. In these cases, Ca extraction was successfully obtained with the increasing temperature and annealing time. The observed sheet thickness of the bundles, annealed at 600 °C for 0 h (Sample 1), was approximately a few micrometers in appearance. The bundles became thinner to approximately sub-micrometers in appearance when synthesized at 800 °C for 0 h and 10 h (Samples 2 and 3). The sheets were alternately stacked with small void spaces to form a bundle. The sheets observed in the image were easily exfoliated and divided into to further thinner nanosheets, as shown later in Fig. 5. Mn atoms started to deposit on the nanosheet bundles synthesized by thermal annealing at 600 °C for 0 h (Sample 1). The Mn atoms were inhomogeneously distributed throughout the nanosheet bundles synthesized by thermal annealing at 800 °C for 0 h (Sample 2), as shown in the EDS mappings of the bundle in Fig. 4(b). The homogeneity of the Mn atom distribution is improved by thermal annealing at 800 °C for 10 h [Sample 3; Fig. 4(c)]. The images in Figs. 4(b) and 4(c) and XRD patterns in Figs. 2(b) and 2(c) suggest that MnSi and Si phases are formed at the initial stage of the synthesis with an inhomogeneous Mn distribution within the nanosheet bundles. Further, the phases are transformed into MnSi_1.7 nanosheet bundles with improved homogeneous Mn distribution after 10 h annealing in the molten salt.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The detailed fine structures of the exfoliated nanosheets were examined using TEM. Plan-view, cross-sectional view, and enlarged plan-view TEM images of pieces of the nanosheet, annealed at 800 °C for 10 h (Sample 3), are shown in Figs. 5(a)–5(c), respectively. The plan-view image shows that relatively uniform sheets were exfoliated, including domains with diameters of approximately 100 nm. The cross-sectional TEM image shows nanosheets with a thickness of several tens of nanometers, and up to approximately 200 nm are stacked with small void spaces to form a bundle. The nanosheet had a layered structure over a wide range of areas through the nanosheet, and stepped structures were observed near the edge of the nanosheets. A stacked structure, consisting of the nanosheets, was also observed in Fig. 5(c). In the image, broad, wavy, and moiré-like fringes were observed, and the features of the fringes are explained as follows. ²⁹⁾ Spots, which occur in the diffraction pattern where two satellite sequences meet or overlap, can also be used to make an image. The fringes are more widely spaced, and perpendicular to the segment ${\rm{\Delta }}{\boldsymbol{g}}$ connecting the two interfering satellite spots. The distance is inversely proportional to $\left|{\rm{\Delta }}{\boldsymbol{g}}\right|.$ The direction of ${\rm{\Delta }}{\boldsymbol{g}}$ is highly sensitive to slight changes in the direction of the satellite sequence. These moiré-like fringes provide a map of the vector ${\rm{\Delta }}{\boldsymbol{g}}$ and thus of the spacing and orientation anomalies. ²⁹⁾

The long-range periodic structure of MnSi_1.7 was confirmed by HRTEM images and their FFT patterns (Fig. 6). Distinct symmetrical FFT patterns were obtained for the nanosheets, synthesized by the thermal annealing at 800 °C for 10 h (Sample 3). The basic spots are marked with circles. ³⁰⁾ The FFT spots, indexed using subcell notation, correspond to the sections of the reciprocal lattice with zone axes [210] and [110] of the basic structures, respectively.

Zoom In Zoom Out Reset image size

The distinct FFT patterns are a consequence and signature of the mismatched sublattices of the intricate MnSi_1.7 Nowotny chimney ladder structures. Each bright "main" spot, associated with the Mn sublattices, is superimposed with a series of more closely spaced "satellite" reciprocal spots associated with a larger Si sublattice spacing. ^29,31) In the FFT patterns, differences among several HMS structures cause additional weak spots in the patterns for the special zone axis. ³²⁾ Furthermore, the positions of some spots slightly differed slightly for different HMS phases. ³²⁾ However, because of the limited spatial resolution of the TEM images of MnSi_1.7 in Fig. 6, identifying the particular MnSi_1.7 phase is difficult.

The HRTEM image and corresponding FFT pattern of Sample 3 are shown in Fig. 7. The crystalline direction and plane of the layers were denoted using a MnSi_1.7 subcell crystalline lattice. As previously reported, the crystallographic orientation relationship between MnSi_1.7 and Si is [111] MnSi_1.7 subcell//[111]Si and $[\bar{1}10]\,$MnSi_1.7 subcell// $[\bar{1}10]\,$Si, using the MnSi_1.7 subcell notation. ³³⁾ Considering that the surface of the Si nanosheet formed from CaSi₂ crystals is Si(111), ¹⁸⁾ three types of variants possibly exist in the sheet because of the three-fold symmetry of the Si(111) surface. These variants are rotated by 120° with respect to each other along the growth direction. ³³⁾ For the tetragonal MnSi_1.7 subcell crystalline lattice, the [111] direction was nearly parallel and inclined by just 1.5° with respect to the direction perpendicular to the (332) plane. Using a conventional notation, the predominant epitaxial relationship between MnSi_1.7 and Si is given by (332), [110]MnSi_1.7//(111), [110]Si, where the (332) MnSi_1.7 plane is equivalent to (3322) and (3330) for Mn₁₁Si₁₉ and Mn₁₅Si₂₆, respectively. ³³⁾ The HRTEM image in Fig. 7(a) was obtained from the nanosheet, which was observed along the direction tilted by ∼20° from the [111]MnSi_1.7 subcell direction roughly toward the [100] subcell direction. By tilting the observation direction, the lattice fringes of {210} and {122} appear. In addition, two stacked domains with moiré-like fringes, rotating each other at approximately 120°, were observed. In the FFT pattern, additional unmarked spots from domains A and B, and others were also observed. Some spots appeared because of the satellite sequences of other further inclined domains.

Zoom In Zoom Out Reset image size

MnSi_1.7 nanosheet bundles were successfully synthesized from CaSi₂ crystal powders with improved homogeneous Mn distribution by thermal annealing with MnCl₂ in a molten salt liquid-phase environment. However, nanosheet bundles with inhomogeneous Mn distribution through the powders and the existence of Si were also observed at the initial stage of the synthesis procedure. The major products of each sample are listed in Table I. The synthesis mechanism of Mn silicide and its nanosheet structure is as follows.

As the cell temperature is gradually increased from the room temperature [Fig. 1(a)], MnCl₂ evaporates during the initial thermal annealing stage. For example, the vapor pressures are 0.21 Pa at 527 °C and 6.3 Pa at 627 °C, ³⁴⁾ providing enough impingement rate for deposition. The CaSi₂ powders are exposed only to the MnCl₂ vapor, when the cell temperature is lower than the melting temperature of MnCl₂ of 654 °C. CaCl₂ forms via the vapor-phase reaction of evaporated MnCl₂ with CaSi₂ powders and also possibly a solid-phase reaction between them where these powders are attached to each other. Based on the enthalpy of the formation of CaSi₂ (−12.0 kcal/g-atom), MnCl₂ (−38.4 kcal/g-atom), and CaCl₂ (−63.4 kcal/g-atom), ³⁵⁾ the formation of CaCl₂ is thermodynamically favored as compared to that of CaSi₂, which would be a driving force for the reaction. ¹⁹⁾ The formation of CaCl₂ was confirmed in the products before washing, as discussed in Appendix A. The Mn atoms were inhomogeneously deposited on the nanosheet bundles. The MnSi phase, rather than the MnSi_1.7 phase, was formed first, and the Si remained as nanosheets. The source powders were sealed under Ar at 1 atm (∼1 × 10⁵ Pa) and the total pressure inside the container at the annealing temperature was ∼3 × 10⁵ Pa. Thus, the mean free path of the molecules in the vapor is of the order of 10–100 nm. ³⁶⁾ At the synthesis temperature of 600 °C, the partial vapor pressure of MnCl₂ is approximately 1 Pa. ³⁴⁾ Although the mean free path of the molecules in the vapor is too short, the Mn atoms would be preferentially deposited on the Si nanosheet surface facing and nearly attached to the MnCl₂ source powders. The CaSi₂ and MnCl₂ source powders were randomly mixed. For MnCl₂, the following reactions were considered for the MnSi growth: ³⁷⁾

$$\begin{eqnarray}&&{{\rm{MnCl}}}_{2}\left({\rm{g}}\right)\,+\,{\rm{Si}}\left({\rm{s}}\right)\,\to \,{\rm{MnSi}}\left({\rm{s}}\right)\,+\,{{\rm{Cl}}}_{2}\left({\rm{g}}\right)\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&2{{\rm{MnCl}}}_{2}\left({\rm{g}}\right)\,+\,3{\rm{Si}}\left({\rm{s}}\right)\,\to \,2{\rm{MnSi}}\left({\rm{s}}\right)\,+\,{{\rm{SiCl}}}_{4}\left({\rm{g}}\right).\end{eqnarray} \tag{ 2 }$$

In addition, MnSi deposition is also expected as

$$\begin{eqnarray}&&{{\rm{SiCl}}}_{4}\left({\rm{g}}\right)\,+\,{{\rm{MnCl}}}_{2}\left({\rm{g}}\right)\,\to \,{\rm{MnSi}}\left({\rm{s}}\right)\,+\,3{{\rm{Cl}}}_{2}\left({\rm{g}}\right).\end{eqnarray} \tag{ 3 }$$

It is reasonable to consider that Mn atoms are deposited and MnSi is formed. The deposited MnSi could further react with Si at elevated temperature, leading to the formation of MnSi_1.7. On the other hand, it was reported that Mn atoms were deposited on Si substrates, the MnSi and MnSi_1.7 phases were formed at the synthesis temperature of approximately 500 °C–600 °C using MnCl₂ as a source material. The result indicates that Mn atoms were practically deposited on Si around the vapor-phase synthesis temperature using MnCl₂. ⁸⁾ In addition, Cl atoms are adsorbed just on Si surfaces and the Si surfaces are terminated by the chlorine. It was observed that chlorine or silicon dichloride were desorbed approximately at 600 °C and higher from Si surfaces. ³⁸⁾ According to literatures, ^8,38) the sticking rate of Cl atoms is considered to be lower than that of Mn atoms. Chlorine or chloride molecules would wrap around the powders to react with CaSi₂ to form CaCl₂ and Si nanosheets around the other side of the nanosheets or bundles under the vapor-phase condition with those relatively shorter mean free path. In contrast, Mn atoms are deposited on the CaSi₂ or Si nanosheet surfaces, directly facing to the MnCl₂ powders. Thus, the inhomogeneous distribution of Mn in the nanosheet bundle was observed, despite the long annealing time of 10 h at 600 °C (Sample 5), as shown in Fig. 3(b), which revealed the additional formation of Si and MnSi. Although Mn atoms were the dominant diffusion species for the practical case of Mn deposition on Si using MnCl₂, ⁸⁾ Mn diffusion into the Si nanosheets was suppressed, owing to the existence of the void space between the nanosheets.

Because the cell temperature is higher than the melting temperature of MnCl₂ (654 °C) and CaCl₂ (772 °C), the liquid-phase growth took place for the source-material configuration (A). The MnCl₂ molten salt was followed by the formation of MnCl₂–CaCl₂ eutectic molten salts because CaCl₂ was formed as a byproduct of the reaction between CaSi₂ and MnCl₂. Although CaSi₂ was consumed by increasing the annealing temperature to 800 °C, Mn distribution remained inhomogeneous for an annealing time of 0 h [Sample 2, Fig. 4(b)]. However, a drastic improvement in the Mn distribution was observed with an annealing time of 10 h for source-material configuration (A) (Sample 3). Note that, using the material configuration (B) at a temperature of 800 °C for 10 h (Sample 4), the Si formation was dominantly observed because the Mn atoms would be evaporated again from the nanosheets without capping materials in this configuration to the vapor-phase.

The results indicate that the homogeneity of the Mn distribution in Mn-silicide nanosheet bundles was remarkably improved by long annealing in the molten salt (Figs. 2 and 4). The structures of molten MnCl₂ ³⁹⁾ and CaCl₂ have been reported previously. ⁴⁰⁾ One possible arrangement of ions in molten MnCl₂ is the ionic orientation of a 180° corner share for each MnCl₄ tetrahedron around the shared Cl atoms. It is reasonable to consider that the interactions between Mn²⁺ and Cl⁻ ions are not strong. ³⁹⁾ The Mn²⁺ ions were densely and homogeneously distributed in the molten salt. The Mn atoms might be homogeneously incorporated and diffused anywhere into the Si nanosheets from the liquid molten salt. Although Mn atoms were inhomogeneously distributed during the initial stage of the vapor-phase synthesis procedure, the existence of the MnSi was suppressed, and MnSi_1.7 was dominantly formed and further homogeneously synthesized with Si consumption through the nanosheet bundles under the liquid molten salt environment. MnSi might be decomposed and redistributed to form MnSi_1.7 through the nanosheet bundles in the molten salt. Hence, the following two possibilities are considered: (1) the interdiffusion of Mn atoms becomes more active at the interface between the nanosheets and the liquid molten salt, compared with the surface of the nanosheets under a vapor atmosphere. (2) Mn atoms are excessively deposited to form MnSi under non-equilibrium conditions, which is dominated at the initial stage of the synthesis procedure. These atoms would be dissolved into the liquid molten salt. Then, Mn atoms are deposited again on the nanosheet to form MnSi_1.7, as an equilibrium reaction facilitated by the molten salt as a pass of the Mn redistribution. These speculations agree with the discussion highlighted in a literature, as "The CaCl₂ and NaCl eutectic mixture melt was speculated to promote the uniform formation of Pt₂Ca nanoparticles." ⁴¹⁾ The phenomenon of the uniform formation in the molten salts, revealed in the literature, was also observed in our experimental results, although the material systems vary. To obtain structurally homogeneous nanosheet bundles, bundles in the molten salt should be synthesized for liquid-phase synthesis. During the thermal annealing treatment at 800 °C for 0 and 10 h, the liquid phase contained MnCl₂–CaCl₂ (primarily CaCl₂) molten salts, because CaCl₂ is a byproduct of the reaction between CaSi₂ and MnCl₂. One of the advantages of forming a CaCl₂-based molten salt is that it dissolves the Mn and/or Si in the molten salt. ^42,43) CaCl₂ is a byproduct of a series of the reactions; however, it plays an important role in the facile synthesis of silicides, because MnSi alloy was prepared from SiO₂–MnO₂ in CaCl₂–NaCl molten salt. ⁴⁴⁾ Further investigations will be required to clarify the role of the MnCl₂–CaCl₂ (primarily CaCl₂) molten salt. Thus, the reaction between MnSi and molten CaCl₂ was examined at an annealing temperature of 800 °C for 10 h, as shown in Appendix B. Evidently, the MnSi surface was damaged and etched by CaCl₂ molten salts. The results indicate that morphological modification was observed before and after annealing, implying that Mn and Si atoms were more mobile on the MnSi surfaces or removed from the surface in the molten salt. The results support the above speculations and show that CaCl₂ molten salt is useful for dissolving or decomposing MnSi, which should be avoided, as well as the formation of MnSi_1.7, by the interdiffusion between MnSi and Si at elevated temperatures.

Previously, the formation of Si nanosheets by Ca extraction from CaSi₂ crystals via thermal annealing with chlorides and metal chlorides has been investigated. Although metal chlorides, such as SnCl₂, SbCl₃, BiCl₃, PbCl₂, and HgCl₂ ¹³⁾ were used as source materials, metal deposition or the incorporation of metals into Si nanosheets was rarely reported. The details of the structural properties of the treated nanosheets have not yet been clarified. It was reported that fully oxidized silicon nanosheets were synthesized in a CuCl₂ aqueous solution at room temperature, whereas partially oxidized silicon nanosheets were synthesized in SnCl₂ ethanol solution at 60 °C. Hardly oxidized silicon nanosheets were synthesized in LiCl–KCl molten salt at 400 °C. The metal nanoparticles in the silicon nanosheets were removed by rinsing the as-prepared silicon nanosheets with a solution. ²⁴⁾ Note that metal nanoparticles were formed between the nanosheets, and metal diffusion into the Si nanosheets has not been reported. In addition, Si-based nanocomposites derived from CaSi₂ were synthesized via a solid-state reaction using NiCl₂, FeCl₂, MnCl₂, and TaCl₅. ^19–22) In a series of the experiments, Ni or Fe silicide particles were formed. However, Ni, Fe, and Ta atoms were not incorporated into the Si nanosheets, and silicide nanosheet formation was not reported. For Mn-silicides, detailed nanostructures have not been reported. When CrCl₂ was used as a Ca extraction agent, Cr atoms were not incorporated into the Si nanosheets, and CrSi₂ nanosheets were not formed. ²³⁾ Moreover, Si-based nanosheet bundles using CaSi₂ crystals as a template were synthesized by thermal treatment of metal chloride vapors, including FeCl₂, FeCl₃, deliquescent MgCl_2, and NH₄Cl. ²⁵⁾ Mg₂Si/Si nanocomposites comprising Si nanosheet bundles and Mg₂Si deposits have been synthesized; however, homogeneous single-phase Mg₂Si nanosheets have not been formed using MgCl₂, which was carefully stored to avoid deliquescence. ²⁶⁾ The Mg₂Si nanosheet bundles were synthesized by annealing in MgCl₂/Mg mixed vapor. ²⁷⁾ An additional Mg source was required to form homogeneous single-phase Mg₂Si nanosheet bundles.

The difficulty of silicide nanosheet formation was discussed based on the assumption that metal chlorides are decomposed, and HCl, Cl₂, and their oxides or hydroxides are formed. The formation is caused by the reactions of metal chlorides with residual moisture and/or oxygen due to thermal instability and deliquescence of the chloride compounds, as reported in a literature. ²⁵⁾ The HCl molecules are transported to the CaSi₂ crystals to extract Ca atoms. The excess HCl itself etches the silicide nanosheets, and removes the metal atoms from the nanosheet to form Si nanosheets. In addition, once the compounds are formed, it becomes difficult to form silicides via reactions with Si-nanosheets. A detailed discussion has been reported in the literature. ²⁵⁾ Thus, this nanosheet formation model was proposed to explain the dependence of the structural and morphological properties of Si-based nanosheets on the thermal annealing conditions, including the reactant.

For FeCl₂ and MgCl₂, the decomposition of the chloride compounds by reaction with residual moisture at an elevated temperature was discussed for the Si nanosheet bundle synthesis. ²⁵⁾ However, to the best of our knowledge, the formation of HCl via the thermal decomposition of MnCl₂ has rarely been reported. The thermal stability and thermodynamics of MnCl₂ ⁴⁵⁾ and the decomposition behavior of several hydrous transitions have been reported. ⁴⁶⁾ In addition, a differential thermal analysis of MnCl₂·4H₂O was performed. ⁴⁷⁾ The formation of HCl is less pronounced because of the relatively more stable Mn chloride compared to other chlorides, such as MgCl₂ or FeCl₂ which chlorides causes the formation of their oxides or hydroxides by reactions with residual moisture and/or oxygen. ²⁵⁾ The incorporation of Mn into the Si nanosheets originating from CaSi₂ crystals would not be suppressed compared with the metal incorporation of Mg or Fe into the Si nanosheets. In the past, MnSi_1.7 with a variety of morphologies, namely, MnSi_1.7 bulk crystals, ⁴⁸⁾ layers, ⁸⁾ powders, ⁹⁾ and nanowire arrays ¹⁰⁾ have been synthesized using MnCl₂ as a Mn-source material. Furthermore, Si nanowires were also synthesized using MnCl₂. In this case, MnCl₂, MnSi_1.7, or other Mn-based compounds acted as catalyst for the growth of Si nanowires when the excess Si atoms were supplied. ^10,49,50)

It is interesting to compare the results with and without NH₄Cl source for the nanosheets synthesized at 600 °C for 10 h (Samples 5 and 6). When NH₄Cl was used as the source material, the dissociation of NH₄Cl provided HCl formation, similar to the case of using deliquescent chloride; the HCl molecules were significantly transported to the CaSi₂ crystals at elevated temperatures. The discussion is consistent with the cases of the metal incorporation of Mg or Fe into Si nanosheets in the vapor phase. ²⁵⁾ MnSi formation was observed in nanosheet bundles, synthesized without NH₄Cl. In contrast, MnSi formation was suppressed by the addition of NH₄Cl (Fig. 3). Therefore, the HCl provided by NH₄Cl was considered to remove the Mn atoms or suppress the additional Mn deposition in the vapor-phase synthesis environment.

The phase selection of Mn-silicide formation has been extensively discussed. The Mn–Si phase diagram indicates the presence of Mn₆Si, Mn₃Si, Mn₅Si₂, Mn₅Si₃, MnSi, and MnSi_1.7. ⁵¹⁾ For the Mn–Si diffusion couple, the first nucleating phase was predicted to be Mn₅Si₃. ⁵²⁾ In contrast, three phases, i.e. Mn₃Si, Mn₅Si₃, and MnSi, were formed through a layered growth process. The formation sequence for these silicides was Mn₃Si, MnSi, and Mn₅Si₃. ⁵³⁾ Unusual phenomena of coexistence of these three phases and simultaneous growth of two phases (MnSi and Mn₅Si₃) were also observed. ⁵³⁾ The phase formation depended not only on the equilibrium formation energy, but also on the interfacial composition, and lattice mismatch between the silicide phase and Si matrix. ⁵⁴⁾ Furthermore, the growth strongly depended on the boundary conditions, in addition to the free energy of formation. ⁵⁴⁾ The selective growth of a particular phase was reported to be triggered mainly by a specific interface composition and governed by the diffusion flux at the interface. ⁵⁵⁾ In this study, MnSi and MnSi_1.7 phases were formed during the silicidation reaction between CaSi₂–MnCl₂ and Si–MnCl₂ reaction couples. The MnSi_1.7 phase was the preferred nucleation phase in the reaction system of the MnCl₂ vapors with the Si substrates, particles, or wires under these growth conditions. ^8–10) The thermal treatment condition of ∼600 °C provides an appropriate dissociation ratio for MnCl₂ molecules. ^8–10) At elevated temperatures, following the reactions, shown in Eqs. (1)–(3) in the literature, ³⁷⁾ manganese silicide and deposited manganese metal could further react with the silicon to form MnSi_1.7. The molten salt synthesis using metal chlorides on Si substrates was also demonstrated for β-FeSi₂ formation. The simultaneous formation of FeSi and β-FeSi₂ was observed at the initial stage of the growth, and single-phase β-FeSi₂ was obtained after a long annealing time. ⁵⁶⁾

Thinner nanosheet bundles were formed in the case of molten salt synthesis compared to those formed during the vapor-phase synthesis (Fig. 2). The annealing temperature dependence of the observed thickness distribution of the nanosheets was reported. The density of the nucleation sites of the Si crystals by Ca atom extraction on the side wall of the CaSi₂ crystals increased with increasing thermal annealing temperatures, thereby leading to the formation of thinner nanosheets. ⁴⁹⁾ The morphology evaluation indicated that the thickness of MnSi_1.7 sheets is thinner than that of Mg₂Si nanosheet bundles. ²⁷⁾ A small void space was formed between the nanosheets owing to the reduction in volume caused by the Ca atom extraction. An additional small volume reduction as V_MnSi1.7/V_Si = 0.96 was caused by the incorporation of Mn atoms into the Si nanosheets. In addition, Mn atoms preferentially diffused into the Si nanosheets because the dominant diffusion species is Mn for the practical case of Mn deposition on Si using MnCl₂. ⁸⁾ However, the dominant diffusion species were reported to be Mn for Mn-rich silicides, such as Mn₃Si and Si for MnSi. ⁵³⁾ Morphological shrinkage was observed in the case of nanosized island formation. ⁸⁾ These features allow the Si nanosheets to retain their morphology after the silicidation and facilitate the simple and straightforward synthesis of thinner silicide nanosheets.

The silicide phase formation on the nanosheets causes the specific nanostructure of the nanosheets. The formation of these multiple growth variants stacked in the nanosheet is caused by Mn silicide nucleation on both sides of the nanosheet. A multiple-layered structure was observed in the nanosheets, where the (001) planes were inclined by nearly 60° to the nanosheet surface. The formation of ∼(332)_subcell plane layered structures provides further nanostructuring to the HMS, resulting in complex crystal structures with Si ladders inside the Mn chimneys. As reported in a literature, ⁵⁷⁾ further improvement in the thermoelectric properties is expected. This structure is considered to affect the phonon properties and electronic structure of the crystal, thereby encouraging us to modify the thermoelectric properties of MnSi_1.7. The synthesis of MnSi_1.7 nanosheet bundles on Si substrates is currently under investigation. The molten salt method is a cost-effective and simple bottom-up technique for fabricating sophisticated nanomaterials with controlled size, morphology, and surface ⁵⁸⁾ using the preferable selections of source materials and reaction paths.