Binary Alkali-Metal Silicon Clathrates by Spark Plasma Sintering: Preparation and Characterization

The binary intermetallic clathrates K8-xSi46 (x = 0.4; 1.2), Rb6.2Si46, Rb11.5Si136 and Cs7.8Si136 were prepared from M4Si4 (M = K, Rb, Cs) precursors by spark-plasma route (SPS) and structurally characterized by Rietveld refinement of PXRD data. The clathrate-II phase Rb11.5Si136 was synthesized for the first time. Partial crystallographic site occupancy of the alkali metals, particularly for the smaller Si20 dodecahedra, was found in all compounds. SPS preparation of Na24Si136 with different SPS current polarities and tooling were performed in order to investigate the role of the electric field on clathrate formation. The electrical and thermal transport properties of K7.6Si46 and K6.8Si46 in the temperature range 4–700 K were investigated. Our findings demonstrate that SPS is a novel tool for the synthesis of intermetallic clathrate phases that are not easily accessible by conventional synthesis techniques.


Introduction
Binary phases of alkali metals with silicon have been prepared by thermal decomposition of the monosilicides M 4 Si 4 (M = Na, K, Rb, Cs) for over six decades [1][2][3][4]. Structural investigations of the polycrystalline products revealed the existence of clathrate-I M 8-x Si 46 and clathrate-II M 24-x Si 136 silicides [5,6]. In both clathrate types, the alkali-metal atoms are enclosed in polyhedral cages formed in the tetrahedrally bonded silicon framework. In the clathrate-I crystal structure, the unit cell contains two Si 20 and six Si 24 cages, while in the clathrate-II structure there are sixteen Si 20 and eight Si 28 cages present in the unit cell. In the clathrate phases prepared by thermal decomposition the filling fraction of M in the silicon cages depends explicitly on both the reaction conditions and the size of the M atom relative to the size of the cage [6] (Figure 1).
Alternatively, redox techniques have been developed to obtain metastable clathrate phases [7][8][9][10] including Ge and Si allotropes with empty cages [11,12]. Defect-free single crystals of alkali metal silicon clathrates large enough for structure investigations have only been obtained in exceptional cases [4]. For example, crystals of K 8 Si 46 were grown on α-Si exposed to potassium vapor [13], while crystals of ternary clathrate-II phases Rb 8 Na 16 Si 136 and Cs 8 Na 16 Si 136 form in closed ampoules directly from the melt [14,15]. Single-crystals suitable for measurement of intrinsic transport properties were first prepared by the electrochemically-driven transformation of Na 4 Si 4 to Na 24 Si 136 The unique advantage of SPS for the synthesis of clathrates by producing scalable, compact bulk materials also continues to be of great interest [4]. In this work we discuss both the preparation and consolidation of alkali-metal silicon clathrates by SPS. We report on the preparation and crystal structure of different clathrate phases (Table 1) as well as SPS densification of K7.6Si46 and K6.8Si46 powders by SPS that allow for the investigation of their transport properties over a large temperature range. The formation and evaporation the free alkali metals during the reaction allow us to work only with a small amount (less than 200 mg) of precursors and, in the results, after processing, we obtained 10-20 mg of the materials.

Results and Discussion
As first described in reference [16], clathrates are formed during SPS redox-treatment by oxidation of Si4 4− at the anode while the alkali metal is reduced at the cathode (Figure 2a,b). Clathrate-II formation, according to The unique advantage of SPS for the synthesis of clathrates by producing scalable, compact bulk materials also continues to be of great interest [4]. In this work we discuss both the preparation and consolidation of alkali-metal silicon clathrates by SPS. We report on the preparation and crystal structure of different clathrate phases (Table 1) as well as SPS densification of K 7.6 Si 46 and K 6.8 Si 46 powders by SPS that allow for the investigation of their transport properties over a large temperature range. The formation and evaporation the free alkali metals during the reaction allow us to work only with a small amount (less than 200 mg) of precursors and, in the results, after processing, we obtained 10-20 mg of the materials.

Results and Discussion
As first described in reference [16], clathrates are formed during SPS redox-treatment by oxidation of Si 4 4´a t the anode while the alkali metal is reduced at the cathode (Figure 2a, Reactions (1) and (2) describe the basics of the clathrate formation by SPS, whereby the process is driven by the DC field and the evaporation of the alkali metal at the cathode and its absorption by the graphite die ( Figure 2b). In order to qualitatively investigate the influence of electric current on clathrate formation we performed three control experiments with Na4Si4 as the precursor. First, the polarity of the DC current was reversed, upon which it was found that the clathrate phase, Na24Si136 in this case, consistently forms at the anode (Figure 2c). In a second experiment, equal volumes of Al2O3 powder were placed above and below the precursor to create an electrically insulating barrier. The electrical current then presumably flowed through the low resistance graphite die such that the current passing through the Na4Si4 precursor is minimized. In this case no reaction was observed under the identical conditions used to form the clathrate Na24Si136 (Figure 2d). In the third experiment, the use of stainless steel punches combined with a die manufactured from boron nitride, a high electrical-resistivity material, presumably forced most of the current to pass through the precursor (as opposed to the surrounding die) (Figure 2e). A more rapid reaction than in the case of using the graphite die was observed. Optical photographs of cross-section of SPS-processed pellets (Figure 2f,h) clearly show that formation of the clathrate phases occurred at the anode during SPS processing. This was also the case for all clathrate compositions prepared in this study. The crystal structures of the SPS-prepared clathrate phases were investigated by the Rietveld method using powder X-ray diffraction (PXRD) data. Experimental, calculated, and difference Reactions (1) and (2) describe the basics of the clathrate formation by SPS, whereby the process is driven by the DC field and the evaporation of the alkali metal at the cathode and its absorption by the graphite die ( Figure 2b). In order to qualitatively investigate the influence of electric current on clathrate formation we performed three control experiments with Na 4 Si 4 as the precursor. First, the polarity of the DC current was reversed, upon which it was found that the clathrate phase, Na 24 Si 136 in this case, consistently forms at the anode (Figure 2c). In a second experiment, equal volumes of Al 2 O 3 powder were placed above and below the precursor to create an electrically insulating barrier. The electrical current then presumably flowed through the low resistance graphite die such that the current passing through the Na 4 Si 4 precursor is minimized. In this case no reaction was observed under the identical conditions used to form the clathrate Na 24 Si 136 (Figure 2d). In the third experiment, the use of stainless steel punches combined with a die manufactured from boron nitride, a high electrical-resistivity material, presumably forced most of the current to pass through the precursor (as opposed to the surrounding die) (Figure 2e). A more rapid reaction than in the case of using the graphite die was observed. Optical photographs of cross-section of SPS-processed pellets (Figure 2f,h) clearly show that formation of the clathrate phases occurred at the anode during SPS processing. This was also the case for all clathrate compositions prepared in this study.
The crystal structures of the SPS-prepared clathrate phases were investigated by the Rietveld method using powder X-ray diffraction (PXRD) data. Experimental, calculated, and difference diffraction patterns for the clathrate-I and II samples are shown on Figures 3 and 4, respectively, and the resulting crystallographic data are presented in Tables 2 and 3, (2) *: Occ-occupancy factor.
The lattice parameters for K 7.6 Si 46 , Rb 6.2 Si 46 , and Cs 7.8 Si 136 show good agreement with those synthesized by other preparation methods [7,16,18,19,[24][25][26]. The lattice parameter for Rb 11.5 Si 136 , as expected, has an intermediate value between that reported for K 17.2 Si 136 [19] and Cs 8 Si 136 [6]. In all cases the silicon framework sites were found to be fully occupied (no vacancies), in agreement with prior results for binary silicon clathrates [7,16,18,19,[26][27][28]. The stable covalent Si-Si bonds in silicon clathrates allow for a large excess of electrons that fill antibonding states. In a rigid-band model, a high concentration of conduction electrons are provided by charge transfer from the alkali-metal guests, thus inducing metallic behavior, as confirmed experimentally [15,29]. In contrast to SPS-prepared Na 8 Si 46 and Na 24 Si 136 , which show full occupancy of the polyhedral cages by sodium [19], the SPS-prepared clathrates of potassium, rubidium and caesium exhibit partial occupancy of the alkali-atom positions (the specific refined compositions are given in Tables 2 and 3).
The compositions of clathrates-II prepared in this study together with previously reported in the literature [16,19,29,30] allow for a comparison of the lattice parameter versus guest-atom content in clathrate-II phases. As shown in Figure 5, in all cases the lattice parameter increases with those increasing M content in the Si 20 (crystallographic site 16c) cage. In addition, the lattice parameter also increases with increasing alkali-ion radius [31] for a specific M concentration. Preferential occupancy of the larger Si 28 (crystallographic site 8b) cages was observed for the case of Rb 11.5 Si 136 , as well as for for Na x Si 136 [30,32].  In order to further characterize the products formed by SPS we measured the electrical and thermal transport properties of K7.6Si46 and K6.8Si46. The density of the pellets (only K8-xSi46) was approximately 90% of the calculated density based on PXRD data. Temperature-dependent electrical resistivity ρ (Figure 6, top), Seebeck coefficient S (Figure 6, middle), and thermal conductivity κ (Figure 6, bottom) in the temperature range from 10 to 700 K are shown in Figure 6. The agreement between the high-and low-temperature data, measured with two different setups on two different specimens, is an indication of the homogeneity of the materials prepared by SPS. Both materials K7.6Si46 and K6.8Si46 show metallic temperature dependence of ρ and S, in agreement with the seminal work by Cros et al. [6]. The fact that these materials show metallic conduction follows from the simple assertion that each K atom donates one electron to the silicon framework resulting in an overall high electron concentration. Measured ρ of cold-pressed K7.6Si46, obtained from the thermal decomposition of K4Si4, was reported to be much higher: ρ ~ 140 mΩ·cm at room temperature with weak temperature dependence [28]. The large discrepancy between those results and the data shown in Figure 6 may be attributed to the microstructure of the specimens and the typically poor inter-granular contact that is achieved using cold pressing [28]. Nevertheless, the ρ values shown in Figure 6 are significantly larger than expected for the presumed high electron concentration. In addition, the residual resistance ratio, ρ300K/ρ10K = 3.6, is also relatively low for a good metal and significantly smaller than the values obtained for single crystals of Na8Si46 (ρ300K/ρ10K = 36) [33] and Na24Si136 (ρ300K/ρ10K = 14) [17]. These results reiterate the need for high-density polycrystalline specimens in order to minimize the effects of grain boundary scattering on the transport properties of intermetallic clathrates [34].   [27].
In order to further characterize the products formed by SPS we measured the electrical and thermal transport properties of K 7.6 Si 46 and K 6.8 Si 46 . The density of the pellets (only K 8-x Si 46 ) was approximately 90% of the calculated density based on PXRD data. Temperature-dependent electrical resistivity ρ (Figure 6, top), Seebeck coefficient S (Figure 6, middle), and thermal conductivity κ (Figure 6, bottom) in the temperature range from 10 to 700 K are shown in Figure 6. The agreement between the high-and low-temperature data, measured with two different setups on two different specimens, is an indication of the homogeneity of the materials prepared by SPS. Both materials K 7.6 Si 46 and K 6.8 Si 46 show metallic temperature dependence of ρ and S, in agreement with the seminal work by Cros et al. [6]. The fact that these materials show metallic conduction follows from the simple assertion that each K atom donates one electron to the silicon framework resulting in an overall high electron concentration. Measured ρ of cold-pressed K 7.6 Si 46 , obtained from the thermal decomposition of K 4 Si 4 , was reported to be much higher: ρ~140 mΩ¨cm at room temperature with weak temperature dependence [28]. The large discrepancy between those results and the data shown in Figure 6 may be attributed to the microstructure of the specimens and the typically poor inter-granular contact that is achieved using cold pressing [28]. Nevertheless, the ρ values shown in Figure 6 are significantly larger than expected for the presumed high electron concentration. In addition, the residual resistance ratio, ρ 300K /ρ 10K = 3.6, is also relatively low for a good metal and significantly smaller than the values obtained for single crystals of Na 8 Si 46 (ρ 300K /ρ 10K = 36) [33] and Na 24 Si 136 (ρ 300K /ρ 10K = 14) [17]. These results reiterate the need for high-density polycrystalline specimens in order to minimize the effects of grain boundary scattering on the transport properties of intermetallic clathrates [34]. . Lattice parameters versus M content, x for clathrate-I NaxSi136 (0 < x < 24) [28], Na24Si136 [16], K17.2Si136 [19], Rb11.5Si136 (this work), and Cs7.8Si136 (this work), as well as ternary Rb8Na16Si136 and Cs8Na16Si136 [27].
In order to further characterize the products formed by SPS we measured the electrical and thermal transport properties of K7.6Si46 and K6.8Si46. The density of the pellets (only K8-xSi46) was approximately 90% of the calculated density based on PXRD data. Temperature-dependent electrical resistivity ρ (Figure 6, top), Seebeck coefficient S (Figure 6, middle), and thermal conductivity κ (Figure 6, bottom) in the temperature range from 10 to 700 K are shown in Figure 6. The agreement between the high-and low-temperature data, measured with two different setups on two different specimens, is an indication of the homogeneity of the materials prepared by SPS. Both materials K7.6Si46 and K6.8Si46 show metallic temperature dependence of ρ and S, in agreement with the seminal work by Cros et al. [6]. The fact that these materials show metallic conduction follows from the simple assertion that each K atom donates one electron to the silicon framework resulting in an overall high electron concentration. Measured ρ of cold-pressed K7.6Si46, obtained from the thermal decomposition of K4Si4, was reported to be much higher: ρ ~ 140 mΩ·cm at room temperature with weak temperature dependence [28]. The large discrepancy between those results and the data shown in Figure 6 may be attributed to the microstructure of the specimens and the typically poor inter-granular contact that is achieved using cold pressing [28]. Nevertheless, the ρ values shown in Figure 6 are significantly larger than expected for the presumed high electron concentration. In addition, the residual resistance ratio, ρ300K/ρ10K = 3.6, is also relatively low for a good metal and significantly smaller than the values obtained for single crystals of Na8Si46 (ρ300K/ρ10K = 36) [33] and Na24Si136 (ρ300K/ρ10K = 14) [17]. These results reiterate the need for high-density polycrystalline specimens in order to minimize the effects of grain boundary scattering on the transport properties of intermetallic clathrates [34].

Synthesis of Precursors
Binary precursors M4Si4 (M = Na, K, Rb, or Cs) were prepared from alkali metal pieces (ChemPur, Karlsruhe, Germany 99.9 wt %) and coarsely ground silicon (ChemPur, 99.999 wt %), by sealing of the precursor mixture under argon in small, welded tantalum ampoules that were enclosed under vacuum inside silica tubes and heated to 800 °C at an average rate of 3 °C/min. The ampoules were held at 800 °C for 4 h before being slowly cooled to room temperature (furnace cooled). The reaction products were determined to be single phase M4Si4 by powder X-ray diffraction (PXRD). All sample manipulations were performed in an argon-filled glove box (H2O < 1 ppm; O2 < 1 ppm). CAUTION: M4Si4 compounds are sensitive to air and moisture (pyrophoric) and must be kept under inert atmosphere at all times.

Spark Plasma Synthesis
SPS experiments were conducted under vacuum (background pressure <10 Pa) using a dedicated SPS setup in a protective argon atmosphere (Figure 2a; SPS: 515 ET, Syntex-Fuji, Japan; glove box: MBraun, Garching, Germany) [24]. Table 1 contains the preparation conditions for the clathrates synthesized within this study. The precursors were ground to fine powder and loaded into graphite or boron nitride (h-BN) dies with an inner diameter of 8 mm. Graphite or stainless steel punches were used to apply the uniaxial pressure. Tantalum foil was employed to surround the powder specimen on all sides, isolating the specimen from the die and punches during SPS processing. Graphite dies worked well for Na, however they were observed to react with alkali-metal vapor for precursors of K, Rb, and Cs and often cracked the graphite die during SPS processing. For these alkali metals h-BN was found to be most suitable. After SPS the products were isolated from the residual precursor by washing first with ethanol and then distilled water under flowing argon.

Synthesis of Precursors
Binary precursors M 4 Si 4 (M = Na, K, Rb, or Cs) were prepared from alkali metal pieces (ChemPur, Karlsruhe, Germany 99.9 wt %) and coarsely ground silicon (ChemPur, 99.999 wt %), by sealing of the precursor mixture under argon in small, welded tantalum ampoules that were enclosed under vacuum inside silica tubes and heated to 800˝C at an average rate of 3˝C/min. The ampoules were held at 800˝C for 4 h before being slowly cooled to room temperature (furnace cooled). The reaction products were determined to be single phase M 4 Si 4 by powder X-ray diffraction (PXRD). All sample manipulations were performed in an argon-filled glove box (H 2 O < 1 ppm; O 2 < 1 ppm). CAUTION: M 4 Si 4 compounds are sensitive to air and moisture (pyrophoric) and must be kept under inert atmosphere at all times.

Spark Plasma Synthesis
SPS experiments were conducted under vacuum (background pressure <10 Pa) using a dedicated SPS setup in a protective argon atmosphere (Figure 2a; SPS: 515 ET, Syntex-Fuji, Japan; glove box: MBraun, Garching, Germany) [24]. Table 1 contains the preparation conditions for the clathrates synthesized within this study. The precursors were ground to fine powder and loaded into graphite or boron nitride (h-BN) dies with an inner diameter of 8 mm. Graphite or stainless steel punches were used to apply the uniaxial pressure. Tantalum foil was employed to surround the powder specimen on all sides, isolating the specimen from the die and punches during SPS processing. Graphite dies worked well for Na, however they were observed to react with alkali-metal vapor for precursors of K, Rb, and Cs and often cracked the graphite die during SPS processing. For these alkali metals h-BN was found to be most suitable. After SPS the products were isolated from the residual precursor by washing first with ethanol and then distilled water under flowing argon.

Sample Characterization
Powder X-ray diffraction (PXRD) data were collected using the Guinier technique (Huber G670 camera, Rimsting, Germany, Cu-K α1 radiation, λ = 1.540598 Å, graphite monochromator, 5˝ď 2θ ď 100˝, ∆2θ = 0.005˝). Crystal structure refinements were performed using the software WinCSD [25]. For the transport measurements, the K 7.6 Si 46 and K 6.8 Si 46 powders were first washed then consolidated by SPS using tungsten carbide tool (8 mm inner diameter) at 300˝C and a uniaxial pressure of 600 MPa under vacuum (background pressure of approximately 10 Pa) for 10 min. The density of the pellets was approximately 90% of the calculated density based on PXRD data. The specimens were cut with a diamond wire saw into parallelepiped pieces 1.5 mmˆ1.5 mmˆ8.0 mm for electrical transport measurements and low-temperature thermal conductivity measurements, and 1.5 mm thick and 8 mm in diameter disks for thermal diffusivity measurements. Electrical resistivity (ρ) and Seebeck coefficient (S) measurements from 22 to 450˝C on SPS densified K 7.6 Si 46 and K 6.8 Si 46 specimens were performed with a commercial ZEM-3 setup (ULVAC-RIKO, Munich, Germany). Specific heat (C p ) measurements were performed on a Netzsch Pegasus DSC (Selb, Germany) while heating to 450˝C at a heating rate of 10˝C/min in an Ar atmosphere. Thermal diffusivity, D, was measured from 22˝C to 450˝C using the laser-flash technique (LFA 457 Micro Flash, Netzsch, Selb, Germany). The temperature dependent thermal conductivity, κ, was calculated using the relation κ = C pˆDˆd where d is the density in g/cm 3 . Low temperature ρ, S, and κ data were collected using the TTO-option (Thermal Transport Option) of a PPMS (Physical Property Measurement System) (Quantum Design, San Diego, CA, USA) using a four-point method.

Conclusions
Binary intermetallic clathrates K 8-x Si 46 (x = 0.4; 1.2), Rb 6.2 Si 46 , Rb 11.5 Si 46 , and Cs 7.8 Si 136 were synthesized by spark plasma synthesis from the respective M 4 Si 4 precursors. The formation of polycrystalline products at the anode of the SPS setup was observed for all compounds. The preparation of the new clathrate-II phase Rb 11.5 Si 136 is further evidence that SPS is useful for the inorganic synthesis. The role the electric field is in driving the chemical conversion of the precursor into the clathrate phase at the anode during SPS processing is emphasized. Conventional thermal decomposition (CTC) [6,28,35], kinetically controlled thermal decomposition (KCTD) [18,19], chemical oxidation (CO) [7,27], and SPS yield different intermetallic clathrate phases from the same M 4 Si 4 precursor (Table 4). While the influence of temperature clearly plays a role in clathrate formation, a definitive understanding of the local formation mechanisms in each of these synthetic approaches [36] is of great interest and is an ongoing task.