Oxidation Protective Hybrid Coating for Thermoelectric Materials

Two commercial hybrid coatings, cured at temperatures lower than 300 °C, were successfully used to protect magnesium silicide stannide and zinc-doped tetrahedrite thermoelectrics. The oxidation rate of magnesium silicide at 500 °C in air was substantially reduced after 120 h with the application of the solvent-based coating and a slight increase in power factor was observed. The water-based coating was effective in preventing an increase in electrical resistivity for a coated tethtraedrite, preserving its power factor after 48 h at 350 °C.

A considerable effort has been made to produce materials with high figure of merit, and several compounds have been identified and improved using different approaches: doping elements; composites; nanostructuring; and mesostructuring [3,4,8]. The effect of conventional synthesis and sintering techniques have been evaluated [9] and innovative methods are constantly under development [10][11][12][13][14]. One of the main challenges in the thermoelectrics field is the identification of efficient materials that are inexpensive, easy to be produced, and formed of earth-abundant and environmentally friendly elements. In this respect, magnesium silicide [15] and tetrahedrite [16] are considered attractive and sustainable candidates for n and p-type thermoelectrics, respectively. One important aspect for the development of high temperature thermoelectric generators is their long-term stability in air at high temperature [17].
Magnesium silicide is a semiconductor of the Mg 2 X (X = Si, Ge, Sn and Pb) compounds family. It possess an anti-fluorite structure with a bandgap of 0.784 eV [18]. It can be doped to achieve good thermoelectric properties (ZT of 0.86 at 862 K with Bi-doping [19]) but it is limited by its relatively high thermal conductivity. N-type solid solutions of Mg 2 Si with Mg 2 Sn [20][21][22][23] and/or Mg 2 Ge [24,25] have been produced in an attempt to reduce their thermal conductivity. One of the best values has been reported for Mg 2.08 Si 0.364 Sn 0.6 Sb 0.036 (ZT of 1.5 at 723 K [26]).
Magnesium silicide and its solid solutions are prone to oxidation above~400 • C; Skomedal et al. [27] reported breakaway oxidation behavior for Mg 2 Si 1−x Sn x (0.1 < x < 0.6) at

Materials and Methods
Mg 2.1 Si 0.487 Sn 0.5 Sb 0.13 (Mg-silicide) powders were provided by European Thermodynamics Ltd (Leicester, UK). Powders were then sintered into 30 mm diameter discs using a Spark Plasma Sintering furnace (FCT HPD 25; FCT Systeme GmbH, Rauenstein, Germany). The sintering of Mg-silicide was carried out at a temperature of 750 • C with a heating and cooling rate of 100 • C/min, a dwell time of 5 min and a pressure of 50 MPa. Cu 11.5 Zn 0.5 Sb 4 S 13 (THD) was prepared starting from single elements powders: Cu (Alpha Aesar, 150 mesh, purity 99.5%), Sb (Alpha Aesar, 100 mesh, purity 99.5%), S (Sigma Aldrich, 100 mesh, purity reagent grade) and Zn (Sigma Aldrich, ≥ 99%). They were weighted in the appropriate stoichiometry and sealed in a stainless steel jar in an argon filled glove box, processed in a ball milling machine (QM-3SP2, Nanjiing University, China) employing stainless steel balls at 360 rpm for 96 h, with a ball to powder ratio of 30:1. The sintering of tetrahedrite was carried out at a temperature of 400 • C with a heating and cooling rate of 50 • C/min, a dwell time of 5 min and a pressure of 50 MPa. The density of the pellets was measured using the Archimede's method. Each pellet was cut into bars having square base of 3 mm per side and 10 mm height.
After preliminary tests, a solvent-based resin (CP4040-S1) was chosen for Mg-Silicide and a water-based resin (CP4040) for tetrahedrite, both purchased from AREMCO SCIENTIFIC COMPANY (Los Angeles, USA). They were applied using a foam brush and subsequently cured in a tubular furnace (Carbolite Gero STF/180, Neuhausen, Germany) for 45 min at 250 • C with a heating and cooling rate of 1.6 • C/min. Aging tests was performed in a muffle oven (Manfredi OVMAT 2009, Pinerolo, Italy) in air at a temperature of 500 • C for 120 h (for Mg-silicide) and at 350 • C for 48h (for THD) with a heating rate of 1.2 • C/min. The choice of oxidation temperatures was guided by previous literature and the potential operating temperatures of the materials. Both Mg-Silicide and tetrahedrite are oxidized in air at these temperatures without being subjected to any phase transformations and their properties are near their optimum values. The tests would provide an initial benchmark for the tested hybrid coatings.
XRD data were collected using X'Pert Pro MRD diffractometer with Cu Kα radiation (PANalytical X'Pert Pro, Philips, Almelo, The Netherlands, with the aid of the X-Pert HighScore software) and the different phases were identified using the JCPDS data base. Field emission scanning electron microscope (FE-SEM, Merlin electron microscope, ZEISS, Oberkochen, Germany) and energy dispersive X-ray Spectroscopy (EDS, ZEISS Supra TM 40, Oberkochen, Germany)were used to characterize the microstructure morphology and chemical composition of uncoated and coated samples, before and after ageing. The measurements of the electrical properties were carried out using a Linseis LSR-3 (Linseis Messgeraete GmbH, Selb, Germany) with Pt thermocouples and electrodes. The oxide layer of the aged samples was removed before measuring their properties.

Mg-Silicide (Solvent-Based Coating)
Ball milling of the elemental powders effectively produced a single phase solid solution (Mg 2 Si 0.4 Sn 0.6 ) of Mg 2 Si and Mg 2 Sn, and no peak splitting was observed in the XRD pattern ( Figure 1a). After sintering, no phase separation was visible in the XRD pattern and the peaks simply became sharper (Figure 1b).
The density of the as sintered sample was about 96% of the theoretical one. The aging at 500 • C for 120 h in air had a very clear effect on the uncoated sample; it was completely burned and turned into powder ( Figure 2). The coated sample did not suffer such a catastrophic effect despite the fact that the applied coating was damaged at the edges.
The XRD pattern of the uncoated sample after aging (Figure 1c) shows the decomposition of Mg 2 Si 0.4 Sn 0.6 into a mixture of compounds (MgO, SiO, SnO 2 and Sn), as already observed by Skomedal et al. [27]. Figure 3 shows the cross-section of a coated sample after the curing. The interface between Mg-silicide and the hybrid coating is continuous, without cracks. However, the coating shows a few cracks parallel to the surface; this was likely due to a mismatch in CTE between the matrix and ceramic filler or more likely an effect due to solvent evaporation during curing, with consequent shrinkage effects. The thickness of the layer was found to be about 30-100 µm, being thinner at the edges.

Mg-Silicide (Solvent-Based Coating)
Ball milling of the elemental powders effectively produced a single phase solid solution (Mg2Si0.4Sn0.6) of Mg2Si and Mg2Sn, and no peak splitting was observed in the XRD pattern ( Figure  1a). After sintering, no phase separation was visible in the XRD pattern and the peaks simply became sharper ( Figure 1b).
The density of the as sintered sample was about 96% of the theoretical one.  The aging at 500 °C for 120 h in air had a very clear effect on the uncoated sample; it was completely burned and turned into powder ( Figure 2). The coated sample did not suffer such a catastrophic effect despite the fact that the applied coating was damaged at the edges. The XRD pattern of the uncoated sample after aging ( Figure 1c) shows the decomposition of Mg2Si0.4Sn0.6 into a mixture of compounds (MgO, SiO, SnO2 and Sn), as already observed by Skomedal et al. [27]. Figure 3 shows the cross-section of a coated sample after the curing. The interface between Mgsilicide and the hybrid coating is continuous, without cracks. However, the coating shows a few cracks parallel to the surface; this was likely due to a mismatch in CTE between the matrix and ceramic filler or more likely an effect due to solvent evaporation during curing, with consequent shrinkage effects. The thickness of the layer was found to be about 30-100 µm, being thinner at the edges. After aging for 120 h at 500 °C in air, the coated sample did not experience significant oxidation, The aging at 500 °C for 120 h in air had a very clear effect on the uncoated sample; it was completely burned and turned into powder ( Figure 2). The coated sample did not suffer such a catastrophic effect despite the fact that the applied coating was damaged at the edges. The XRD pattern of the uncoated sample after aging (Figure 1c) shows the decomposition of Mg2Si0.4Sn0.6 into a mixture of compounds (MgO, SiO, SnO2 and Sn), as already observed by Skomedal et al. [27]. Figure 3 shows the cross-section of a coated sample after the curing. The interface between Mgsilicide and the hybrid coating is continuous, without cracks. However, the coating shows a few cracks parallel to the surface; this was likely due to a mismatch in CTE between the matrix and ceramic filler or more likely an effect due to solvent evaporation during curing, with consequent shrinkage effects. The thickness of the layer was found to be about 30-100 µm, being thinner at the edges. After aging for 120 h at 500 °C in air, the coated sample did not experience significant oxidation, and was still mainly composed of a single phase (Figure 1d). However, a small amount of MgO was present and the coating showed cracks on the edges and peeled off in some areas. Figure 4 shows the cross-section of a coated Mg-silicide after the oxidation test: an uneven oxide scale formed on the TE surface, thus indicating that the coating effectively reduced the oxidation reaction rate compared to the uncoated sample but did not prevent it as an oxide layer growing at the coating/sample interface. After aging for 120 h at 500 • C in air, the coated sample did not experience significant oxidation, and was still mainly composed of a single phase (Figure 1d). However, a small amount of MgO was present and the coating showed cracks on the edges and peeled off in some areas. Figure 4 shows the cross-section of a coated Mg-silicide after the oxidation test: an uneven oxide scale formed on the TE surface, thus indicating that the coating effectively reduced the oxidation reaction rate compared to the uncoated sample but did not prevent it as an oxide layer growing at the coating/sample interface. The comparison between the electrical properties of the as-sintered sample and coated sample after ageing is a useful tool to understand the effectiveness of the coating ( Figure 5).
It was clearly impossible to measure the properties of the uncoated sample after the oxidation test since it was completely destroyed (Figure 2). The initial properties of the sample are comparable to those of a similar composition reported in the literature [20,22]. The electrical resistivity (ρ) of the coated sample after aging was increased by about 50%, while the Seebeck coefficient (S) increased by about 10%. Different stoichiometric ratio of Mg2Si and Mg2Sn as well as Mg vacancies or interstitial (due to the fact that Mg diffuses towards the surface) can influence the electronic properties [15,22,45,46].
The coating provides a very good degree of protection of Mg-silicide, against a complete burning of the Mg-silicide substrate.
From the data, the power factor (S 2 /ρ) of the coated samples seems to be slightly increased. On the other hand, it is also clear that the effectiveness is time-limited and longer ageing times may likely determine the total oxidation of the sample, as occurred for the uncoated sample.
Further work is needed to fully understand the implications of defects or non-homogeneous areas through the coating.
Further work should focus on the production of a homogeneous coating with a controlled optimal thickness, which should prevent the surface damage and remove easy path for oxygen diffusion. Due to the sample shape, it was not possible to evaluate thermal conductivity and, therefore, ZT. The comparison between the electrical properties of the as-sintered sample and coated sample after ageing is a useful tool to understand the effectiveness of the coating ( Figure 5).

Tetrahedrite (Water-Based Coating)
The Ball milled powders produced starting from single elements consisted of single phase Cu12Sb4S13 ( PDF Card n.00-024-1318) (Figure 6a). The density of the as sintered THD measured with the Archimede's method was found to be about 98% of the theoretical density. The XRD pattern of the as-sintered THD (Figure 6b) confirmed that the main phase is Cu12Sb4S13 with a minor amount of Cu3SbS4 (Famatinite PDF Card n. 01-071-0555). It was clearly impossible to measure the properties of the uncoated sample after the oxidation test since it was completely destroyed (Figure 2). The initial properties of the sample are comparable to those of a similar composition reported in the literature [20,22]. The electrical resistivity (ρ) of the coated sample after aging was increased by about 50%, while the Seebeck coefficient (S) increased by about 10%. Different stoichiometric ratio of Mg 2 Si and Mg 2 Sn as well as Mg vacancies or interstitial (due to the fact that Mg diffuses towards the surface) can influence the electronic properties [15,22,45,46].
The coating provides a very good degree of protection of Mg-silicide, against a complete burning of the Mg-silicide substrate.
From the data, the power factor (S 2 /ρ) of the coated samples seems to be slightly increased. On the other hand, it is also clear that the effectiveness is time-limited and longer ageing times may likely determine the total oxidation of the sample, as occurred for the uncoated sample.
Further work is needed to fully understand the implications of defects or non-homogeneous areas through the coating.
Further work should focus on the production of a homogeneous coating with a controlled optimal thickness, which should prevent the surface damage and remove easy path for oxygen diffusion. Due to the sample shape, it was not possible to evaluate thermal conductivity and, therefore, ZT.

Tetrahedrite (Water-Based Coating)
The Ball milled powders produced starting from single elements consisted of single phase Cu 12 Sb 4 S 13 (PDF Card n.00-024-1318) (Figure 6a). The density of the as sintered THD measured with the Archimede's method was found to be about 98% of the theoretical density. The XRD pattern of the as-sintered THD (Figure 6b) confirmed that the main phase is Cu 12 Sb 4 S 13 with a minor amount of Cu 3 SbS 4 (Famatinite PDF Card n. 01-071-0555).

Tetrahedrite (Water-Based Coating)
The Ball milled powders produced starting from single elements consisted of single phase Cu12Sb4S13 ( PDF Card n.00-024-1318) (Figure 6a). The density of the as sintered THD measured with the Archimede's method was found to be about 98% of the theoretical density. The XRD pattern of the as-sintered THD (Figure 6b) confirmed that the main phase is Cu12Sb4S13 with a minor amount of Cu3SbS4 (Famatinite PDF Card n. 01-071-0555). After the ageing at 350 • C for 48 h in air, the uncoated THD was oxidised with a darker surface than the as-sintered sample. Cross sectional SEM images of the uncoated THD (Figure 7) shows the formation of an inhomogeneous layer (around 3-5 µm) on the whole surface of the thermoelectric. The point indicated with the black arrow can be attributed to antimony oxide.
The XRD analysis of the uncoated sample surface after ageing (Figure 6c) shows that the main phase was Sb 2 O 3 (PDF Card n. 00-043-1071), confirming the SEM analysis, with the presence of Cu 3 SbS 4 and Cu 2 S (PDF Card n. 01-072-1071) as secondary phases, as also reported by Chetty et al. and Harish at al. [16,47]. The cross-section of the water-based resin coated THD after curing at 250 • C for 45 min (Figure 8) shows crystals of different shape and composition well dispersed in the silicone resin matrix, and no pores, cracks or other defects are visible at the coating/THD interface. h without coating d) aged THD at 350 °C for 48 h with coating and e) Cu12Sb4S13 PDF Card n.00-024-1318.
After the ageing at 350 °C for 48 h in air, the uncoated THD was oxidised with a darker surface than the as-sintered sample. Cross sectional SEM images of the uncoated THD (Figure 7) shows the formation of an inhomogeneous layer (around 3-5 µm) on the whole surface of the thermoelectric. The point indicated with the black arrow can be attributed to antimony oxide. The XRD analysis of the uncoated sample surface after ageing (Figure 6c) shows that the main phase was Sb2O3 (PDF Card n. 00-043-1071), confirming the SEM analysis, with the presence of Cu3SbS4 and Cu2S (PDF Card n. 01-072-1071) as secondary phases, as also reported by Chetty et al. and Harish at al. [16,47]. The cross-section of the water-based resin coated THD after curing at 250 °C for 45 min (Figure 8) shows crystals of different shape and composition well dispersed in the silicone resin matrix, and no pores, cracks or other defects are visible at the coating/THD interface. The cross-sectional image of the water-based coated THD after ageing at 350 °C for 48 h ( Figure  9) evidenced the absence of cracks within the coating, which is still well-adhered to the substrate. As can be seen in the SEM image, no evidence for the formation of oxidation layers was found at the coating/THD interface. Furthermore, XRD analysis (Figure 6d) shows that after the ageing at 350 °C for 48 h in air there were no apparent changes in the THD compared to the as-sintered sample, confirming that the hybrid coating provided an effective protection, inhibiting the oxidation of THD under thermal ageing.  The XRD analysis of the uncoated sample surface after ageing (Figure 6c) shows that the main phase was Sb2O3 (PDF Card n. 00-043-1071), confirming the SEM analysis, with the presence of Cu3SbS4 and Cu2S (PDF Card n. 01-072-1071) as secondary phases, as also reported by Chetty et al. and Harish at al. [16,47]. The cross-section of the water-based resin coated THD after curing at 250 °C for 45 min (Figure 8) shows crystals of different shape and composition well dispersed in the silicone resin matrix, and no pores, cracks or other defects are visible at the coating/THD interface. The cross-sectional image of the water-based coated THD after ageing at 350 °C for 48 h ( Figure  9) evidenced the absence of cracks within the coating, which is still well-adhered to the substrate. As can be seen in the SEM image, no evidence for the formation of oxidation layers was found at the coating/THD interface. Furthermore, XRD analysis ( Figure 6d) shows that after the ageing at 350 °C for 48 h in air there were no apparent changes in the THD compared to the as-sintered sample, confirming that the hybrid coating provided an effective protection, inhibiting the oxidation of THD under thermal ageing. The cross-sectional image of the water-based coated THD after ageing at 350 • C for 48 h (Figure 9) evidenced the absence of cracks within the coating, which is still well-adhered to the substrate. As can be seen in the SEM image, no evidence for the formation of oxidation layers was found at the coating/THD interface. Furthermore, XRD analysis ( Figure 6d) shows that after the ageing at 350 • C for 48 h in air there were no apparent changes in the THD compared to the as-sintered sample, confirming that the hybrid coating provided an effective protection, inhibiting the oxidation of THD under thermal ageing. A comparison between the properties of the as-sintered sample and the uncoated and coated counterparts after ageing at 350 °C for 48 h in air (Figure 10), further confirm the effectiveness of the coating. The Seebeck coefficients of the three samples did not show any differences, but the values of the aged THD with coating were slightly higher than those without coating, at least starting from 150 A comparison between the properties of the as-sintered sample and the uncoated and coated counterparts after ageing at 350 • C for 48 h in air (Figure 10), further confirm the effectiveness of the coating. The Seebeck coefficients of the three samples did not show any differences, but the values of the aged THD with coating were slightly higher than those without coating, at least starting from 150 • C. The coating is able to prevent the increase in electrical resistivity noticed in the uncoated sample, as it maintains the original chemical composition. Consequently, the power factor of the uncoated sample suffers a significant reduction, while the coated sample maintains a similar value. A comparison between the properties of the as-sintered sample and the uncoated and coated counterparts after ageing at 350 °C for 48 h in air (Figure 10), further confirm the effectiveness of the coating. The Seebeck coefficients of the three samples did not show any differences, but the values of the aged THD with coating were slightly higher than those without coating, at least starting from 150 °C. The coating is able to prevent the increase in electrical resistivity noticed in the uncoated sample, as it maintains the original chemical composition. Consequently, the power factor of the uncoated sample suffers a significant reduction, while the coated sample maintains a similar value.
. Figure 10. Temperature dependence of the a) Electrical resistivity b) Seebeck coefficient and c) Power Factor of as-sintered THD and aged THD, with and without coating after ageing at 350 °C for 48 h.

Conclusions
The effectiveness of two hybrid protective coatings for Mg 2.1 Si 0.487 Sn 0.5 Sb 0.13 and Cu 11.5 Zn 0.5 Sb 4 S 13 thermoelectric materials was reviewed and discussed.
The solvent-based resin significantly reduced the oxidation rate of magnesium silicide at 500 • C in air. Even with some imperfections (incomplete adhesion, few cracks within the resin after curing), the coated sample was not significantly oxidised after 120 h and the electrical properties were not severely modified.
The water-based hybrid coating was effective at providing a barrier coating to avoid the oxidation. Consequently, the values of the power factor did not decrease in the presence of the hybrid coating, indicating that it is a promising candidate for protecting THD against high temperature oxidation in air.