Earth Abundant Element Type I Clathrate Phases

Abstract

Earth abundant element clathrate phases are of interest for a number of applications ranging from photovoltaics to thermoelectrics. Silicon-containing type I clathrate is a framework structure with the stoichiometry A_8-xSi₄₆ (A = guest atom such as alkali metal) that can be tuned by alloying and doping with other elements. The type I clathrate framework can be described as being composed of two types of polyhedral cages made up of tetrahedrally coordinated Si: pentagonal dodecahedra with 20 atoms and tetrakaidecahedra with 24 atoms in the ratio of 2:6. The cation sites, A, are found in the center of each polyhedral cage. This review focuses on the newest discoveries in the group 13-silicon type I clathrate family: A₈E₈Si₃₈ (A = alkali metal; E = Al, Ga) and their properties. Possible approaches to new phases based on earth abundant elements and their potential applications will be discussed.

1. Introduction

Inorganic clathrate phases which can be described as all-inorganic frameworks with guest atoms have been of interest for a wide variety of applications, including thermoelectrics, Li-ion batteries, hydrogen storage, and photovoltaics [1]. In particular, silicon (Si)-containing clathrate phases are of significant interest because of being earth abundant and environmentally benign. The alkaline earth-containing group 13-Si light element clathrates have been reviewed recently [2]; therefore, this short review provides an update on the alkali metal-containing group 13-Si compounds crystallizing in the clathrate type I structure.

Na_8-xSi₄₆ is the first observed silicon-based compound discovered to crystalize in the type I clathrate structure. The structure can be described as having a covalently bonded silicon (Si) framework with sodium (Na) as the guest atom centered within the polyhedra [3]. The structure has been described in detail many times and consists of Na cations encapsulated in tetrahedrally bonded silicon framework (cubic space group, Pm-3n), shown in Figure 1. The framework contains 46 atoms with three unique sites typically referred to by their Wyckoff positions 6c, 16i, and 24k. The eight guest atoms in the unit cell are located at the Wyckoff positions 2a and 6d, which centers the two types of polyhedra found in type I clathrate: 20-vertex pentagonal dodecahedra and 24-vertex tetrakaidecahedra. Following the discovery of Na_8-xSi₄₆, the alkali metals series (A = K, Rb, Cs) of type I clathrate were synthesized from the elements by annealing, with the Cs clathrate being a high pressure phase [4,5]. The alkali metals´ guest atoms in the silicon-containing clathrate phases donate electrons, and these phases are metallic [1].

Clathrate phases with the composition A₈E₈Si₃₈ (A = alkali metal, E = Al, Ga, In) are considered to be electron-balanced phases as the compounds contain an equivalent amount of group 13 elements to accept the donated electrons from guest atoms and were predicted to show semiconducting behavior [7,8]. The phases prepared to date are A₈E₈Si₃₈ (A = Na [9,10], K [7,10,11,12], Rb [8], Cs) [13,14] and have been shown experimentally to be semiconducting when the composition is near the expected; the E = In and A = Na, E = Ga phases have not been prepared.

Table 1. Experimental Unit Cell Parameter of Type I Clathrate Phase A₈E₈Si₃₉ (A = Na, K, Rb, Cs; E = Al, Ga).

Compound	a (Å)		Reference
	Powder (293 K)	Single Crystal (T)
Na8Al8Si38	10.3260(1)	10.3170(1) (100 K)	[9]
K8Al8Si38	10.48072(2) ‡	10.4802(16) (90 K)	[11]
Rb8Al8Si38	10.53069(1)	10.517(1) (298 K)	[14]
Cs8Al8Si38	10.5848(4)	10.569(1) (298 K)	[14]
Na8Ga8Si38	-	-	-
K8Ga8Si38	10.424916(10) ‡	10.4270(3) (200 K)	[13]
Rb8Ga8Si38	10.470174(13)	10.4433(15) (200 K)	[13]
Cs8Ga8Si38	10.535069(15)	10.5339(3) (200 K)	[13]

^‡ Synchrotron data.

provides the lattice parameters of the A₈E₈Si₃₈ compounds reported to date. The only borosilicide compound with type I clathrate structure is K₇B₇Si₃₉ [15], although boron has been incorporated into a type I Ba₈Al₁₆Si₃₀ clathrate at dopant levels [16]. Besides group 13 substitutions in the Si framework, silicon clathrates containing earth abundant transition metal substitutions A₈M^II₄Si₄₂ have been prepared with M^II = Zn: K₈Zn_3.5Si_42.5 and Rb_7.9Zn_3.6Si_42.4 [17]. While the heavier group 14 elements are not a topic of this review, it is noteworthy that, while most of the compositions described are n-type semiconductors, there is at least one example of a p-type compound with Sn, K₈Ga_8.5Sn_37.5 [18]. Since thermoelectric devices need both n- and p-type semiconductors, the possibility that both can be prepared provides additional incentive for developing the chemistry of the silicon clathrates further. Type II clathrate phases composed of group 13 elements and Si have also been prepared, Cs₈Na₁₆Ga₂₃Si₁₁₃, Rb_8.4Na_15.6Ga₂₀Si₁₁₆ [19], and Cs₈Na₁₆Al₂₄Si₁₁₂ [20], demonstrating new horizons for further development of alkali metal-containing clathrates. Aside from structure, only the thermoelectric transport properties of the type II clathrate Cs₈Na₁₆Al₂₄Si₁₁₂ have been reported to date [20].

2. Theory

Interest in these electron-precise clathrate structures, A₈E₈Si₃₈, has resulted in theoretical investigations of their formation energies, bandgaps and other transport properties [11,21,22,23]. These silicon-containing clathrates are of interest for both photovoltaics and thermoelectrics. The interest in photovoltaics is a result of the possibility of a direct bandgap in the appropriate range for solar energy conversion [11,12,13]. Even without being direct gap semiconductors, the idea that charge recombination can be reduced or minimized has led to a resurgence of interest [22]. Clathrate structures for thermoelectric applications have been considered ideal Phonon Glass Electron Crystal (PGEC) materials [24]. PGEC has been an important concept for designing efficient thermoelectric materials [23]. The idea is that a material can have properties similar to a glass for phonons and as a crystal for electrical transport. This allows for low thermal conductivity and high electrical conductivity: two important parameters for efficient thermoelectric materials [23]. Silicon and earth abundant clathrate phases have been particularly desirable as possible PGECs, as many of the most efficient materials for direct heat to electrical conversion contain critical or environmentally unfriendly elements.

In general, clathrate structures form a cage network that encapsulates an element referred to as the guest. The optimal material for thermoelectric applications would be where the guest atom is weakly bound to the cage and provides a vibrational mode referred to as a “rattling mode” that can scatter the lattice phonon mode. While reduction in thermal conductivity in these type I clathrate phases has been attributed to the guest vibrational modes, first principle calculations also demonstrate that structural disorder of the covalently bonded cages plays an important role in reducing the thermal conductivity [11,22].

The band structures of the series A₈E₈Si₃₈ (A = Na, K, Rb, Cs; E = Al, Ga, In) were calculated using density functional theory [25,26]. The calculations indicate that the bandgap remained indirect, but became larger with the heavier guest alkali atom with an estimated gap of ~0.9 eV for A = Na and ~1.36 eV for A = Cs. The calculations indicate that the bandgap was not very sensitive to the group 13 element, E, but again followed the principle of becoming slightly larger with heavier elements. The authors noted that the theoretical predictions might not be valid because of site-specific positioning of the atoms, vacancies, or deviations from ideal stoichiometry, but they are fairly close to the experimental values. Additional theoretical calculations on the series, focusing on the group 13 site preferences, was reported [21]. Site specificity of the group 13 element within the clathrate I framework has been noted experimentally, so detailed calculations are of great interest. As mentioned above, there are three framework sites that the group 13 element could occupy. In a simplistic view, the group 13 element occupies the site most likely to avoid any E-E bonding interactions in order to better satisfy valence [27]. Figure 2 shows how the specific sites that will be discussed further are interconnected with the Wyckoff nomenclature indicated. In these type I clathrates, the group 13 element favors the 6c framework site with a small amount on the 24k site in the ratio of approximately 5:3, when A = K, Rb, Cs. This is consistent with experimental observations for these alkali metal clathrates [7,8,9,11,13].

Theoretical calculations suggest that these phases should have applications in photovoltaics with K₈Al₈Si₃₈ showing a quasi-direct bandgap of ~ 1 eV with relatively high mobilities determined [22]. Additionally, investigation of the valence band and conduction band-wave functions revealed clear distinction in spatial location, thereby suggesting low probability of charge carrier recombination and efficient energy conversion. The transport properties related to thermoelectrics have also been calculated for the E = Ga phases and efficiencies predicted with the best energy conversion phase being hole-doped Cs₈Ga₈Si₃₈ [21]. Overall, there has been significant progress in both theoretical and experimental investigations suggesting that fundamental understanding of the physics and chemistry will be advanced along with the practical applications of these materials [23].

3. Structure

Powder X-ray diffraction was performed on the binary type I clathrate phases K_7.62(1)Si₄₆ and Rb_6.15(2)Si₄₆ to examine their crystal structure [4]. The host atom sites have no vacancies revealed by Rietveld refinement of the X-ray powder diffraction pattern; the Si frameworks are fully occupied similar with that of Na₈Si₄₆ [28]. Fully occupied framework sites are commonly noticed in alkali metals-Si clathrate phases; in the contrast, Ge or Sn clathrates with alkali metals have been reported to possess vacancies at the framework’s 6c site [8,29,30,31,32,33]. Ramachandran et al. attributed the difference in the framework vacancy formation to the weaker homoatomic bonding descending the group in the periodic table [4].

Compounds with the stoichiometry A₈E₈Si₃₈, A = Na, K, R, Cs and E = Al and Ga have been prepared to date and their lattice parameters are provided in Table 1. In ternary clathrate phases, the substitution preference of silicon framework sites (6c, 16i, 24k) can affect the band structure and transport properties significantly. In the A₈E₈Si₃₈ framework, Ga occupancy can be distinguished by single-crystal X-ray diffraction as Ga and Si atoms have electron densities that provide significantly different scattering factors. For A = K, Rb, Cs compounds, the Ga occupancy ratio at the 6c site is in the range of 61%~46% and 17%~20% at the 24k site. At the 16i site, the Ga occupancy ratio is low, only 1%~2%, which indicates Ga preference for the 6c site (

Table 2. Comparison of Al and Ga’s site occupancy fraction (s.o.f.) at the Si framework sites, 6c, 24k, 16i of A₈Tr₈Si₃₈ (A = K, Rb, Cs; Tr = Al, Ga) compounds.

Compounds	Atom	s.o.f. at 6c	s.o.f. at 24k	s.o.f. at 16i	Reference
K8Al8Si38	Al	75(3)%	12(1)%	4(1)%	Neutron powder diffraction [14]
Rb8Al8Si38	Al	60(4)%	15(2)%	5(3)%
Cs8Al8Si38	Al	41(3)%	16(1)%	10(2)%
K8Ga8Si38	Ga	61.0(7)%	17.7(4)%	3.2(3)%	Single crystal XRD [13]
Rb8Ga8Si38	Ga	52.0(9)%	17.8(4)%	0.4(4)%
Cs8Ga8Si38	Ga	46.2(5)%	20.2(3)%	1.07(3)%

) [13]. Similarly, Zn also shows site preference for the 6c site and avoids substitution on the 16i site [17]. Distinguishing Al from Si in A₈Al₈Si₃₈ (A = K, Rb, Cs) is difficult, so neutron powder diffraction was performed to study the Al/Si mixed occupancy ratios [14]. Even with neutron diffraction, it is difficult to distinguish Al from Si. Rietveld refinement of the neutron powder diffraction gave the total chemical formula of A/Al/Si close to the nominal stoichiometry 8/8/38. Al’s site occupancy fraction (s.o.f.) is shown in Table 2 to compare with the Ga’s s.o.f. The group 13 atoms’ preference to occupy 6c site is attributed to the fact that there is no direct bonding between 6c sites (shown in Figure 2), and group 13 atoms occupy these specific sites in order to avoid the disfavored 13-13 bonds [27]. Consistent with what is found in the A₈Ga₈Si₃₈ samples, Al prefers 6c site and avoids 16i. Additionally, as the unit cell expands with larger guest atoms from K to Cs, the s.o.f. of Al at the 6c site decreases. This is similar to what has been discovered in Ba₈Al_xSi_46-x samples by neutron diffraction study; when x increases from 6 to 15, to accommodate more Al atoms, more Al atoms go to 24k site [6,34].

The theoretical calculations discussed above also identified the two most stable configurations as being consistent with what is found experimentally: 6 Al atoms at 6c and 2 Al atoms at 16i, or 5 Al atoms at 6c, 2 Al atoms at 24k and 1 Al atom at 16i.

Na₈Al₈Si₃₈ has only been prepared by kinetically controlled thermal decomposition (KCTD) of a mixture of NaSi [35] and NaAlSi [36]. This phase cannot be prepared by conventional synthetic routes so may be a metastable or kinetically stable phase. Small single crystals and phase-pure sample were obtained. The lattice parameter vs. unit cell volume, shown in Figure 3, is in good agreement with that expected for A = Na when compared to A₈Al₈Si₃₈ (A = K, Rb, Cs). Rietveld and single-crystal refinements indicate that Al preferentially resides at the 6c site in the ratio Al:Si 0.79:0.21. The 24k site to a significantly lesser extent with the ratio Al:Si 0.13:0.87 and no evidence for Al on the 16i site. This result is similar to what has been determined for A₈Al₈Si₃₈.

In the case of K₇B₇Si₃₉, where B is a much smaller atom than Si, B is found to substitute 16i sites by single-crystal X-ray diffraction refinement, while only Si is found at 6c and 24k sites [15]. This is very different than that observed for A₈E₈Si₃₈ (E = Al, Ga). In Figure 2, it can be seen that the six-membered ring of the 24-vertex tetrakaidecahedral cage, composed of 6c and 24k sites, has bond angles deviating from an ideal tetrahedron and therefore creates chemical stress if substituted by the small B atom. Therefore, B atoms have a preference for the 16i site [15].

In the type I clathrate crystal structure, the guest atom Wyckoff nomenclature is 2a inside the 20-vertex pentagonal dodecahedral cages and 6d inside the 24-vertex tetrakaidecahedral cages. Large anisotropic atomic displacement parameters (ADPs) have been reported for the 6d sites guest atoms, and therefore they are considered as rattler and contribute to phonon scattering. For compounds K_7.62(1)Si₄₆ and Rb_6.15(2)Si₄₆, when guest atom vacancy is present, occupancy at 6d site is preferred and 2a site has lower s.o.f, 88.5(4)% at K_7.62(1)Si₄₆ 2a site and 21.8(5)% at Rb_6.15(2)Si₄₆ 2a site [4]. Similarly, in sample Cs_7.8(1)Si₄₆, s.o.f of 2a site is 0.881(4) while 6d site is close to fully occupied [5].

Although this review is concerned with the alkali metal guest group 13 Si clathrate type I phases, it is informative to review the guest ADPs for Ba₈Al_xSi_46-x, since the amount of Al can vary from x ~ 8–15. Neutron powder diffraction at 35 K of Ba₈Al_xSi_46-x showed that when Al composition increases from 8 to 15, the anisotropic ADPs behave differently as presented in Figure 4. In the Figure, U₁₁ and U₃₃ are the two thermal displacement parameters parallel to the six-membered ring in the E₂₄ cage, and U₂₂ is direction perpendicular to six-membered ring. As the Al composition increases and the cage volume increases as well, U₂₂ changed from a value larger than U₁₁, U₃₃ to a value smaller than those two. As shown in Figure 4, the thermal ellipsoid changes critically when Al compositions increased [6]. However, compare the anisotropic ADPs of 6d site in the case of A₈Ga₈Si₃₈ and A₈Ga₈Si₃₈ (A = K, Rb, Cs): the alkali metals at 6d sites behave similarly with those of Ba₈Al₁₅Si₃₁ instead of Ba₈Al₈Si₃₈, with the larger U elongated in the directions parallel to the six-membered ring (U₁₁, U₃₃) [13,14].

Unlike Ba₈Al_xSi_46-x, the composition for A₈Al₈Si₃₈ appears to be fixed [14,37], although minor adjustments appear to be possible [10,38]. Therefore, in order to tune the electronics, the composition K_8-xBa_xAl_8+xSi_38-x was investigated using metal hydrides for the alkali and alkaline earth metals and Al/Si arc melted mixtures heated together. The nominal compositions of x = 1.0, 1.5, and 2.0 were investigated. The structures were investigated for each composition with synchrotron radiation at 295 and 100 K. The 2a site with the smaller volume polyhedron was mostly substituted by Ba²⁺ rather than K⁺ guest atoms, consistent with the smaller ionic radius of Ba²⁺ [37]. Similar to the anisotropic ADP described above for the guest atom at the 6d site, the guest atom in the 6d site shows a larger displacement than the direction perpendicular to the six-membered ring of the 24-vertex cage. This is also observed for the Zn-containing phases [39].

4. Properties

The band structures of A₈Ga₈Si₃₈ were first calculated using density-functional theory [25,26] and determined to be indirect bandgap semiconductors with the calculated bandgaps increasing with increasing mass of A. K₈E₈Si₃₈ for E = Al, Ga, In was also calculated and showed a slight increase in bandgap with decreasing mass of E. Overall, the gaps were calculated to be 0.45 eV for Na₈Ga₈Si₃₈ to 0.89 for Cs₈Ga₈Si₃₈ while Eg value of Si with the diamond structure is calculated to be 0.65 eV. Since the experimental value of Eg of Si is 1.12 eV, these calculated values are adjusted according, corresponding to 0.92 and 1.36 eV for Na₈Ga₈Si₃₈ and Cs₈Ga₈Si₃₈, respectively. The experimental determination was for K₈Ga₈Si₃₈ and the optical bandgap was estimated to be indirect at 1.2 eV [12,40]. With a higher level of theory and site preferences of E similar to that found experimentally, the bandgap of K₈Al₈Si₃₈ was found to be larger than 1 eV [11]. The experimental measurements of A₈Ga₈Si₃₈ follow the predicted trend, but are larger than the original calculations [13]. The values of the bandgap for K₈Al₈Si₃₈ and A₈Ga₈Si₃₈ (A = K, Rb, Cs) were determined from surface photovoltage spectroscopy [41,42] indicated as n-type and are in the range of 1.1–1.4 eV [13]. Na₈Al₈Si₃₈ indicated an optical bandgap of 0.64 eV [9].

There has been significant interest and recent progress in the measurement of thermoelectric properties of these clathrates. Theoretical calculations indicated that that the dimensionless figure of merit, zT, for optimally doped compounds can reach 0.5 at 600 K. Hole-doped Cs₈Ga₈Si₃₈ is predicted to provide a zT of 0.75 at 900 K. The figure of merit is a dimensionless parameter that provides a way to distinguish a favorable material for high performance as it combines the following thermoelectric parameters: Seebeck coefficient, resistivity, and thermal conductivity for a specific absolute temperature.

Table 3. Experimental Seebeck coefficient, thermal conductivity, electrical resistivity and figure of merit zT for A₈E₈Si₃₈ and related clathrates estimated at 300 K.

Phase	$\mathit{\rho }$ $(\mathsf{\mu }\Omega ·m)$	$\mathit{S}$ $(\mathsf{\mu }\mathit{V}{\mathit{K}}^{-1})$	$\mathit{\kappa }$ $(\mathit{W}{\mathit{K}}^{-1}{m}^{-1})$	zT	Reference
Na8Al8Si38	4.3 × 103	−22	2.03	1.7 × 10−5	[9]
K8Al8Si38	6.4 × 103	−222	1.65	1.4 × 10−3	[14]
	1 × 103	−90	1.80	1.4 × 10−3	[13]
	625	−210	1.80	1.2 × 10−2	[37]
K8Al7Si39	78.5	−65.5	1.7	9.6 × 10−3	[38]
Rb8Al8Si38	3.2 × 104	−252	1.24	4.8 × 10−4	[14]
Cs8Al8Si38	5.4 × 106	−300	0.93	5.3 × 10−6	[14]
K8Ga8Si38	2.0 × 103	−25	0.51	1.8 × 10−4	[13]
Rb8Ga8Si38	1.0 × 104	17	1.1	8.0 × 10−6	[13]
Cs8Ga8Si38	1.2 × 105	30	1.7	1.3 × 10−6	[13]

shows the data reported at 300 K for comparison. The thermal conductivity in all cases is low and the Seebeck coefficients are reasonable, but the electrical resistivity is too high in most cases. In the example of K₈Al₈Si₃₈, it is apparent from the reported results that the precise stoichiometry greatly affects the electrical resistivity, as can be seen in Table 3 with the various entries [13,14,37] and for the one example that is slightly off stoichiometry, K₈Al₇Si₃₉ [38].

Since the K₈Al₈Si₃₈ phase showed the most promise with ease of synthesis, further optimization of that phase was considered. As indicated above, one method to optimize the transport properties is to focus on the Al-Si ratio and K₈Al₇Si₃₉ shows the promise of this approach. Another method for decreasing the resistivity of the stoichiometric A₈E₈Si₃₈ samples is to attempt to control the electron donation to the framework with mixed cations. The mixed alkali metal-alkaline earth Al-Si clathrate phases, K_8-xBa_xAl_8+xSi_38-x (x = 0, 1, 1.5, 2) were prepared and their thermoelectric properties measured. The K₈Al₈Si₃₈ used in this study [37] had slightly different lattice parameters than those measured previously attributed to slightly different stoichiometry [13,14]. Thermogravimetric/differential scanning calorimetry (TG/DSC) measurements indicated that the K₈Al₈Si₃₈ phase was stable up to 900 K (Figure 5), so temperature-dependent studies up to 900 K were performed on about 90% dense pressed pellets. The samples showed significantly lower electrical resistivities compared with stoichiometric A₈E₈Si₃₈ samples.

Preparing dense pellets of clathrates via spark plasma sintering is challenging, especially those with low melting points or potentially volatile guest atoms. In the case of K₈Al₈Si₃₈, the samples decompose as K volatiles with the application of SPS. Therefore, while pellets with density of 80%–85% could be prepared for low temperature measurements (4 mm diameter pellets), high temperature spark plasma sintering of larger pellets provided samples with lower densities thereby making it difficult to obtain reliable experimental measurements of thermal conductivity. Pressed pellets of K_8-xBa_xAl_8+xSi_38-x (x = 0, 1, 1.5, 2) were prepared and the Seebeck coefficient and electrical resistivity measured (Figure 6) from room temperature to 1000 K.

Figure 6 shows that the K_8-xBa_xAl_8+xSi_38-x samples are all n-type with the lowest resistivity for x = 1.5. The resistivities do not correlate with x, as precise control over the Al-Si content is difficult to achieve with the hydride synthesis described in the manuscript [37]. The power factor for the same samples is provided in Figure 7, indicating that x = 1.5 is the best sample in the series.

If lattice thermal conductivities are extrapolated based on the low temperature measurements, and electronic thermal conductivities are calculated according to the Wiedemann-Franz relationship, the zT can be estimated at ~0.35 at 873 K, a significant value for an all earth abundant material, providing inspiration for those working on novel clathrates for high zT in a non-toxic, earth abundant material [37].

5. Synthesis

The simplest routes to the phases A₈E₈Si₃₈ involve simply reacting the element in sealed Nb/Ta tubes at high temperatures. In particular, the gallium phases of A = K, Rb, Cs; E = Ga can be prepared in this manner. K₈Ga₈Si₃₈ and Rb₈Ga₈Si₃₈ were synthesized from on-stoichiometry reactions between pure elements in Nb/Ta tube heated to 1270 K and annealed at 970 K and their type I clathrate structure were reported [7,8]. K₇B₇Si₃₉ can also be prepared from the elements, heating the mixtures in a Ta tube, encapsulated in fused silica, at about 1100 K. However, the reaction was not complete, and the impurities were removed by reaction with sodium hydroxide and concentrated acid [7,12]. To date, there has been little or no follow-up of the boron-containing phases.

Na₈Al₈Si₃₈ can be synthesized in spark plasma sintering (SPS) and kinetically controlled thermal decomposition (KCTD) reactions with NaAlSi and NaSi mixture [9]. This route can lead to single crystals or phase-pure microcrystalline powders. Using different precursor rations, Na₈Al_xSi_46-x compositions were investigated, but while the diffraction peaks broadened, there was no direct evidence for a composition other than x ~ 8. Figure 8 shows a schematic of how the reaction is performed. The NaAlSi and NaSi mixture is placed in a custom-designed stainless steel apparatus under a uniaxial pressure of 20 MPa and heated for 9 h under dynamic vacuum (10⁻⁶ torr) at 953 K. During this reaction, the Na is removed in a chemically controlled manner from the NaAlSi/NaSi mixture via interaction with a graphite flake which is spatially separated from the mixture via NaCl. The NaCl stops a direct reaction of the Na with the graphite, so it is presumed to be a vapor-phase interaction of Na with the graphite flake. This procedure has been described in detail for the synthesis of Na-Si clathrate phases [43]. The product is carefully washed with ethanol and distilled water and dried in air. Caution should be used as NaSi can react violently with water.

High pressure synthesis of Na₈Al_xSi_46-x has also been investigated employing a mixture of NaSi, Al and Si as reagents, but only x~0.5 was prepared at 5.5 GPa and 1570 K [10]. This low x phase could not be produced as a phase-pure product and further synthetic efforts are necessary to provide a conclusion concerning whether x can be varied in this system. K₈Al₈Si₃₈ was reported to be synthesized from KH and arc-melted Al/Si, sealed in a Ta tube jacketed in fused silica at 973 K. Ball milling KH and Al/Si together subsequent to heating enhanced the sample purity [13]. K_8-xBa_xAl_8+xSi_38-x could be prepared from ball milling BaH₂ with KH and arc-melted Al/Si in the same manner as described above [37]. While this route provided fairly good control over the stoichiometry, caution should be taken when handling the metal hydrides that can react violently with water. A₈Ga₈Si₃₈ (A = K, Rb, Cs) can be prepared from the pure elements reaction in a Ta tube jacketed in a fused silica ampoule at 973 K [13]. Excess alkali metals were used to obtain phase-pure samples and carefully removed with ethanol from samples. Caution when handling these samples is necessary as unreacted alkali metal can react violently with ethanol. Typically, the samples were initially allowed to oxidize slowly by exposing them over time to air, then the samples were covered with hexane to protect from water and ethanol slowly added before the mixture was sonicated. In the case of Cs₈Cd₄Sn₄₂, phase-pure samples could be prepared by simple heating and annealing in a sealed metal tube [44]. It is possible that the Cd-containing Si or Ge phases can be prepared in a similar manner. Single-phase A₈Al₈Si₃₈ (A = K, Rb, Cs) can also be prepared from Al and the respective alkali-metal halide salt flux, as reported by Baran et al. Similarly, Cs_8-xGa_8-ySi_38+y was synthesized from a stoichiometric reaction of the elements in an alkali metal halide flux, while K₈Zn_3.5Si_42.5 and Rb_7.9Zn_3.6Si_42.4 were synthesized using a combined alkali metal halide/zinc flux [14,17]. Excess halides flux can be removed by dissolving in water as all of the clathrate phases are air stable. Unlike the Sn [45] or Ge clathrates [46], this is the only flux route reported to date for the synthesis of alkali metals containing group 13-silicon clathrates and, while it is reported to provide enough sample for pressed powders, it results in small crystals which are not sufficient for thermoelectric property measurements [14,17]. Therefore, there is still a significant challenge to grow large single crystals of good quality for property measurements of the Si clathrates with alkali metal guests.

6. Conclusions

New synthetic processes to prepare single-phase clathrates of all earth abundant and non-toxic elements have provided new clathrate type I phases that have potential for applications in energy storage and energy conversion. Metal hydrides, salt fluxes, and kinetically controlled decomposition of ternary phases provide methods for precise control over composition. All of the compositions discussed above show reasonable bandgaps for photovoltaics and low thermal conductivity, important to thermoelectrics. Controlling the carrier concentration by alkali metal-alkaline earth metal solid solutions resulted in a reasonable zT, suggesting that further improvements are possible. Controlling the guest atoms by means of solid solutions of K and Ba have been shown to be a productive way to systematically change the n-type carrier concentration. Further investigations of Si-Ge, Si-Sn, and more complex solid solutions (Al-Ga; Ga-In, for example) of these phases should be investigated with guidance from theory. The solid solutions of Zn(Cd)-E (E = Al, Ga, In) might provide a way to access hole-doped (p-type) clathrates. This is a rich area of chemistry, and progress towards other clathrate structures along with both guest and framework substitutions, provides a rich playground for new materials with technologically relevant properties.