Synthesis and Transport Properties of ZnSnP2-yAsy Chalcopyrite Solid Solutions

This work focuses on the synthesis and properties of quaternary ZnSnP2-yAsy chalcopyrite solid solutions. Full miscibility of the solid solution is achieved using ball milling followed by hot press sintering. The measured electrical conductivity increases substantially with As substitution from 0.03 S cm−1 for ZnSnP2 to 10.3 S cm−1 for ZnSnAs2 at 715 K. Band gaps calculated from the activation energies show a steady decrease with increasing As concentration from 1.4 eV for ZnSnP2 to 0.7 eV for ZnSnAs2. The Seebeck coefficient decreases significantly with As substitution from nearly 1000 μV K−1 for ZnSnP2 to −100 μV K−1 for ZnSnAs2 at 650 K. Thermal conductivity is decreased for the solid solutions due to alloy phonon scattering, compared to the end members with y = 0 and y = 2, with the y = 0.5 and y = 1.0 samples exhibiting the lowest values of 1.4 W m−1 K−1 at 825 K. Figure of merit values are increased for the undoped solid solutions at lower temperatures when compared to the end members due to the decreased thermal conductivity, with the y = 0.5 sample reaching zT = 1.6 × 10−3 and y = 1 reaching 2.1 × 10−3 at 700 K. The largest values of the figure of merit zT for the undoped series was found for y = 2 with zT = 2.8 × 10−3 at 700 K due to the increasing n-type Seebeck coefficient. Boltztrap calculations reveal that p-doping could yield zT values above unity at 800 K in case of ZnSnAs2, comparable with ZnSnP2.

For the most part, phosphides and arsenides are neither among the best performing thermoelectrics nor among the heavily investigated ones, mostly because they typically comprise higher thermal conductivity than the more traditional antimonides and tellurides.Notable exceptions exist however [30][31][32], with figure of merit values of the order of 1 for both pand n-doped phosphides [33].Several Zn-based chalcopyrite phosphides and arsenides (ZnBPn 2 with B = Si, Ge, and Sn; Pn = P and As) were predicted to have high thermopower [34].We recently experimentally demonstrated that despite the high symmetry crystal structures and relatively light constituent elements, the solid solutions ZnGe 1-x Sn x P 2 can achieve reasonably low thermal conductivity, and ultimately high figure of merit values when properly doped [35].Here, we report on the solid solutions ZnSnP 2-y As y , focusing on Sn instead of Ge because of its lower price, higher abundancy, and (partially) substituting As for P because of its higher weight despite its higher toxicity, as higher weight typically occurs with lower lattice thermal conductivity [13].

Materials and Methods
All reactions began from the elements (Zn powder (99.9% Alfa Aesar, Tewksbury, MA, USA, −100 mesh), Sn powder (99.998%Alfa Aesar, −100 mesh), Ge pieces (99.999%STREM Chemicals, Newburyport, MA, USA), P powder (99% Alfa Aesar, −100 mesh), and As powder (99.98% Alfa Aesar, −100 mesh), which were loaded into zirconia lined ball mill jars with ~10 g of 1 mm zirconia balls in an argon filled glove box.The jars were milled at 600 rpm for 5 min increments with 1 min rest times, with the direction reversed after each rest time, using the Fritsch Pulverisette 7, (acquired from Laval Lab, Laval, QC, Canada).Three milling steps were employed, all as described above; after the first milling step of 5 h, the jars were opened in the argon glovebox and agglomerated materials were mechanically reincorporated before the two final 2 h milling steps.After those steps, the reacted materials were ground by hand to yield a uniform micro-crystalline sample.Finally, high-pressure sintering was performed as a final reaction step in graphite dies of a diameter of 12.7 mm under a pressure of 56 MPa using an Oxy-Gon Industries (Epsom, NH, USA) hot press; the temperature was ramped up over two hours to at 850 K and held there for 6 h, followed by a pressure-free cooldown.
Powder X-ray diffraction (PXRD) was performed on the ground samples as well as polished pellets at room temperature using the Inel (Artenay, France) powder X-ray diffractometer, which utilizes a position sensitive detector and Cu Kα 1 radiation.Rietveld refinements were performed using the GSAS-II (v.5761) analysis software [36].
An FEI (Hillsboro, OR, USA) Quanta FEG ESEM microscope was used for the energydispersive analysis of X-ray (EDAX) measurements with an acceleration voltage of 20 kV.Five-point measurements were taken for each sample and then averaged, and area scans and elemental mapping were performed on a 150 µm × 150 µm area.
Thermal diffusivity was measured on the pressed pellets under argon using the TA Instruments (Hillsboro, OR, USA) DLF-1200 system.The Seebeck coefficient was measured by the direct method, and electrical conductivity measured by a standard 4-point method, both with the ULVAC RIKO ZEM-3 apparatus under helium on rectangular pellets of 8 × 2 × 2 mm, cut from the original round pellet after the thermal diffusivity measurements.Estimated measurement errors are 3% for the Seebeck coefficient, 5% for the electrical conductivity [37], and 5% for the thermal conductivity measurements [38], resulting in 10% for the figure of merit.The error bars were included in the corresponding figures.
Electronic structure calculations were carried out using the WIEN2k_21.1 package that employs the full potential linearized augmented plane wave (LAPW) method [39][40][41].Here, we used the generalized gradient approximation (GGA) from Perdew, Burke and Ernzerhof (PBE) [42].We used a grid of 1000 k points evenly distributed throughout the first Brillouin zone, which resulted in 99 symmetry independent k points for ZnSnAs 2 .As convergence criterion, we used 10 −4 Ry for the maximum energy change.In addition, we employed the BoltzTraP2 (v22.3.1)package [43] that uses the Boltzmann transport equations to calculate the thermoelectric properties.Although assuming a constant relaxation time results in an additional uncertainty [44,45], this often leads to good agreement between theory and experiment [46].

Chemical and Structural Characterization
After thermal diffusivity measurements were performed, the samples were cut into bars for transport measurements, and PXRD was performed on the leftover pellet pieces hand ground into powders.Long time (>12 h) PXRD measurements for ZnSnP 2-y As y were performed for y = 0, 0.5, 1, 1.5, and 2 (Figure 1).The patterns are very consistent and do not exhibit any signs of any side products.
As convergence criterion, we used 10 −4 Ry for the maximum energy change.In addition, we employed the BoltzTraP2 (v22.3.1)package [43] that uses the Boltzmann transport equations to calculate the thermoelectric properties.Although assuming a constant relaxation time results in an additional uncertainty [44,45], this often leads to good agreement between theory and experiment [46].

Chemical and Structural Characterization
After thermal diffusivity measurements were performed, the samples were cut into bars for transport measurements, and PXRD was performed on the leftover pellet pieces hand ground into powders.Long time (> 12 h) PXRD measurements for ZnSnP2-yAsy were performed for y = 0, 0.5, 1, 1.5, and 2 (Figure 1).The patterns are very consistent and do not exhibit any signs of any side products.A shift of the characteristic peaks to lower angles occurs with increasing As concentration due to the unit cell expansion, caused by the larger size of the As atoms, compared to the P atoms.The expansion of unit cell parameters is illustrated in Figure 2a,b, and summarized in Table 1, namely a relatively steady increase with increasing As concentration from y = 0 to y = 1.5, followed by a larger increase when moving from y = 1.5 to y = 2.The tetragonality of the system, defined as c/(2a), slowly increases from 0.998 at y = 0 to 1.000 at y = 2 (Figure 2c).A shift of the characteristic peaks to lower angles occurs with increasing As concentration due to the unit cell expansion, caused by the larger size of the As atoms, compared to the P atoms.The expansion of unit cell parameters is illustrated in Figure 2a,b, and summarized in Table 1, namely a relatively steady increase with increasing As concentration from y = 0 to y = 1.5, followed by a larger increase when moving from y = 1.5 to y = 2.The tetragonality of the system, defined as c/(2a), slowly increases from 0.998 at y = 0 to 1.000 at y = 2 (Figure 2c).The chemical compositions of the series were evaluated by refining the occupancy parameters during Rietveld refinements as well energy dispersive X-ray spectroscopy (EDAX) analysis.The y values (As content) of the solid solutions were refined to 0.59(1), 1.06(3), and 1.66 (1) for the solid solutions with nominal y values of 0.5, 1.0, and 1.5.EDAX results for the solid solutions and end members can be found in Table 2.The concentrations determined from EDAX measurements match well for the solid solutions with expected atomic percent values showing differences of less than 8%, while the end member ZnSnAs 2 displayed significantly lower than expected Zn and As (higher Sn) concentration.EDAX atomic mapping for these materials can be found in the Supplementary Information (Figure S1).

Experimental Physical Properties
Electrical conductivity versus temperature measurements were carried out for all members of the series (Figure 3).We verified the stability of the samples under the measurement conditions by measuring a few additional datapoints during cooldown.These datapoints (open circles in Figure 3) match the corresponding points obtained during heating to approximately 800 K very nicely, confirming the samples' stabilities.The electrical conductivity of the P-rich members was too low to be measured at room temperature, but became measurable around 450 K or above, depending on y.As expected for semiconductors, the electrical conductivity increases with temperature for all members.An overall increase in conductivity with increasing As concentration is evident, in line with the expected decreasing band gap and higher degree of covalency (which increases carrier mobility), both caused by the lower electronegativity of As, compared to P. The conductivity for ZnSnP 2 rises from σ = 0.01 S cm −1 at 650 K to 0.1 S cm −1 at 800 K, and for ZnSnAs 2 from σ = 0.03 S cm −1 at 300 K to 7.0 S cm −1 at 650 K.An overall increase in conductivity with increasing As concentration is evident, in line with the expected decreasing band gap and higher degree of covalency (which increases carrier mobility), both caused by the lower electronegativity of As, compared to P. The conductivity for ZnSnP2 rises from σ = 0.01 S cm −1 at 650 K to 0.1 S cm −1 at 800 K, and for ZnSnAs2 from σ = 0.03 S cm −1 at 300 K to 7.0 S cm −1 at 650 K.For comparison, ZnSnAs2 crystals grown by chemical vapor transport displayed a σ value of only 1.4 × 10 −4 S cm −1 at 300 K, much lower than observed in this work, likely due to the low level of impurities typically found in perfect single crystals [47].On the other hand, single crystals synthesized by the Bridgman method displayed σ = 24 S cm −1 at 295 For comparison, ZnSnAs 2 crystals grown chemical vapor transport displayed a σ value of only 1.4 × 10 −4 S cm −1 at 300 K, much lower than observed in this work, likely due to the low level of impurities typically found in perfect single crystals [47].On the other hand, single crystals synthesized by the Bridgman method displayed σ = 24 S cm −1 at 295 K [48].ZnSnAs 2 bulk samples studied under various heat treatments had σ values ranging from 0.1 S cm −1 to 1200 S cm −1 at 300 K, with most values around 400 S cm −1 [49].Heat treatments of the quenched samples tended to decrease conductivity, indicating healing of possible defects in these samples, which then resulted in lower electrical conductivity.Chalcopyrite CuInSe 2 single crystals exhibited σ values ranging from 0.15 S cm −1 at 300 K to 0.46 S cm −1 at 575 K [50], likely low because of their low amounts of intrinsic defects.Studies of vacancy doped p-type Cu 0.99 InSe 2.05 -which adopts the same structure type and has the same number of electrons-displayed σ values ranging from 1.5 S cm −1 at 325 K to 6.0 S cm −1 at 760 K with degenerate semiconducting behavior [51].
An Arrhenius plot of ln(σ) versus T −1 is shown in Figure 4, where a linear trend is expected for intrinsic semiconductors and curved trends for extrinsic semiconductors.The series members displayed mostly intrinsic semiconducting behavior, except for ZnSnAs 2 with its extrinsic behavior indicative of a significant amount of defects.Band gaps of the series members are calculated from the slope (in the high temperature linear region) using the Arrhenius expression for the activation energy.A clear trend of decreasing band gaps with As concentration is observed as postulated above (inset of Figure 4).The previously determined band gaps of E g = 1.4 eV for ZnSnP 2 and 0.7 eV for ZnSnAs 2 are comparable to the literature values of 1.68 eV for ZnSnP 2 [52] and 0.59 eV for ZnSnAs 2 [48].The Seebeck coefficient data measured for the full series are displayed in Figure 5.The values decrease with increasing As concentration, for example at the highest temperatures from S = 744 µV K −1 for y = 0 down to −125 µV K −1 for y = 2, in line with the opposing trend in the electrical conductivity.The temperature dependence for the materials from y = 0 to y = 1.5, decreasing steadily with increasing temperature, is typical of p-type intrinsic behavior.A turnover of the slope in Seebeck versus temperature is indicative of bipolar conduction, which is seen at 475 K for the y = 2 end member, which ultimately results in a p-type to n-type transition.The Seebeck coefficient data measured for the full series are displayed in Figure 5.The values decrease with increasing As concentration, for example at the highest temperatures from S = 744 µV K −1 for y = 0 down to −125 µV K −1 for y = 2, in line with the opposing trend in the electrical conductivity.The temperature dependence for the materials from y = 0 to y = 1.5, decreasing steadily with increasing temperature, is typical of p-type intrinsic behavior.A turnover of the slope in Seebeck versus temperature is indicative of bipolar conduction, which is seen at 475 K for the y = 2 end member, which ultimately results in a p-type to n-type transition.
Previous studies of bulk ZnSnAs 2 showed degenerate p-type semiconducting behavior, with Seebeck coefficient values ranging from S = 41 µV K −1 at 300 K to 60 µV K −1 at 440 K similar to the results found in this work for the same temperature range.Various heat treatments and synthesis methods produced room temperature S values ranging from 26 µV K −1 (slowly cooled sample) to 224 µV K −1 (after annealing at 883 K) [49].Single crystals displayed larger overall S values with degenerate semiconducting behavior, ranging from 310 µV K −1 at 300 K to 400 µV K −1 at 600 K [48].The analogous I-III-VI chalcopyrite CuInSe 2 single crystals grown by vapor deposition displayed S values ranging from 542 µV K −1 at 300 K down to 300 µV K −1 at 400 K, and then increasing again to 600 µV K −1 at 625 K [50].Vacancy doped p-type Cu 0.99 InSe 2.05 had S values of 300 µV K −1 at 300 K, which to 625 µV K −1 at 620 K to finally decrease to 500 µV K −1 at 775 K [51].A similar p-type to n-type transition was observed in one study of CuInSe 2 , with 100 µV K −1 at 300 K increasing slightly to 200 µV K −1 at 390 K and then decreasing rapidly to −200 µV K −1 at 560 K [53].The Seebeck coefficient data measured for the full series are displayed in Figure 5.The values decrease with increasing As concentration, for example at the highest temperatures from S = 744 µV K −1 for y = 0 down to −125 µV K −1 for y = 2, in line with the opposing trend in the electrical conductivity.The temperature dependence for the materials from y = 0 to y = 1.5, decreasing steadily with increasing temperature, is typical of p-type intrinsic behavior.A turnover of the slope in Seebeck versus temperature is indicative of bipolar conduction, which is seen at 475 K for the y = 2 end member, which ultimately results in a p-type to n-type transition.Previous studies of bulk ZnSnAs2 showed degenerate p-type semiconducting behavior, with Seebeck coefficient values ranging from S = 41 µV K −1 at 300 K to 60 µV K −1 at 440 K similar to the results found in this work for the same temperature range.Various heat treatments and synthesis methods produced room temperature S values ranging from 26 µV K −1 (slowly cooled sample) to 224 µV K −1 (after annealing at 883 K) [49].Single crystals displayed larger overall S values with degenerate semiconducting behavior, ranging from 310 µV K −1 at 300 K to 400 µV K −1 at 600 K [48].The analogous I-III-VI chalcopyrite CuInSe2 single crystals grown by vapor deposition displayed S values ranging from 542 µV K −1 at 300 K down to 300 µV K −1 at 400 K, and then increasing again to 600 µV K −1 at 625 K [50].Vacancy doped p-type Cu0.99InSe2.05had S values of 300 µV K −1 at 300 K, which increased to 625 µV K −1 at 620 K to finally decrease to 500 µV K −1 at 775 K [51].A similar p-type to ntype transition was observed in one study of CuInSe2, with 100 µV K −1 at 300 K increasing slightly to 200 µV K −1 at 390 K and then decreasing rapidly to −200 µV K −1 at 560 K [53].
The thermal conductivity data are shown in Figure 6, revealing a typical decrease with temperature due to increased phonon frequencies.Here as well, the data collected  Considering the low electrical conductivity values, the electronic contribution to the measured total thermal conductivity remains always below 0.01 W m −1 K −1 based on the Wiedemann-Franz law.Therefore, the measured thermal conductivity is basically directly equal to the lattice thermal conductivity.Gasson et al. measured between κ = 4.3 W m −1 K −1 (sintered sample) and 14.1 W m −1 K −1 (slowly cooled sample) for ZnSnAs2 [49], demonstrating again how the properties, e.g., charge carrier concentration and thus electronic thermal conductivity, depend on the heat treatment.
The calculated thermoelectric figure of merit is obtained from combining Seebeck coefficient, electrical and thermal conductivity, as well as temperature via zT = TσS 2 κ −1 .As the thermal data were obtained for a larger temperature range, the zT values are limited to the temperature ranges of the electrical measurements.The largest and smallest zT values are observed in the y = 2 end member with zT = 1.4 × 10 −9 at 300 K to 0.003 at 700 K and hypothetically zero at the p-n transition temperature (Figure 7).At high temperatures, the Considering the low electrical conductivity values, the electronic contribution to the measured total thermal conductivity remains always below 0.01 W m −1 K −1 based on the Wiedemann-Franz law.Therefore, the measured thermal conductivity is basically directly equal to the lattice thermal conductivity.Gasson et al. measured between κ = 4.3 W m −1 K −1 (sintered sample) and 14.1 W m −1 K −1 (slowly cooled sample) for ZnSnAs 2 [49], demonstrating again how the properties, e.g., charge carrier concentration and thus electronic thermal conductivity, depend on the heat treatment.
The calculated thermoelectric figure of merit is obtained from combining Seebeck coefficient, electrical and thermal conductivity, as well as temperature via zT = TσS 2 κ −1 .
As the thermal data were obtained for larger temperature range, the zT values are limited to the temperature ranges of the electrical measurements.The largest and smallest zT values are observed in the y = 2 end member with zT = 1.4 × 10 −9 at 300 K to 0.003 at 700 K and hypothetically zero at the p-n transition temperature (Figure 7).At high temperatures, the relatively large electrical conductivity, increasing negative Seebeck coefficient, and decreasing thermal conductivity contribute to improving thermoelectric performance.Gasson et al. determined zT of slowly cooled ZnSnAs 2 to be 0.004 at 300 K, with a carrier concentration of 5.7 × 10 20 cm −1 [49].The maximum zT values for all solid solution members (i.e., containing both P and As) occur at 700 K with 2.0 × 10 −3 for y = 1, 1.6 × 10 −3 for y = 0.5, and 4.3 × 10 −4 for y = 1.5.The ZnSnP 2-y As y solid solutions outperform the y = 0 end member.

Calculated Physical Properties
The band structure and density of states of ZnSnAs2 are shown in Figure 8, as the corresponding results for ZnSnP2 were published before [35].A small direct band gap exists at the Γ point, with several bands converging at the top of the valence band, in line with the high degree of tetragonality.There are strong resemblances to the calculated band structure of ZnSnP2 [35], with the smaller band gap being the most noticeable difference.Using BoltzTraP2, the Seebeck coefficient S, the electrical conductivity σ, the power factor σS 2 , and the electronic contribution to the thermal conductivity κ0 were calculated (relative to the relaxation time τ) for temperatures up to 800 K, as displayed in Figure 9. Depending on the carrier concentration (relative energies), high Seebeck values around

Calculated Physical Properties
The band structure and density of states of ZnSnAs 2 are shown in Figure 8, as the corresponding results for ZnSnP 2 were published before [35].A small direct band gap exists at the Γ point, with several bands converging at the top of the valence band, in line with the high degree of tetragonality.There are strong resemblances to the calculated band structure of ZnSnP 2 [35], with the smaller band gap being the most noticeable difference.

Calculated Physical Properties
The band structure and density of states of ZnSnAs2 are shown in Figure 8, as the corresponding results for ZnSnP2 were published before [35].A small direct band gap exists at the Γ point, with several bands converging at the top of the valence band, in line with the high degree of tetragonality.There are strong resemblances to the calculated band structure of ZnSnP2 [35], with the smaller band gap being the most noticeable difference.Using BoltzTraP2, the Seebeck coefficient S, the electrical conductivity σ, the power factor σS 2 , and the electronic contribution to the thermal conductivity κ0 were calculated (relative to the relaxation time τ) for temperatures up to 800 K, as displayed in Figure 9. Depending on the carrier concentration (relative energies), high Seebeck values around +500 µV K −1 and −400 µV K −1 may be achievable at 400 K, and power factor values of S 2 σ τ −1 = 6 × 10 11 W m −1 K −2 s −1 for both p-and n-type at 800 K.For comparison, the correspond- Using BoltzTraP2, the Seebeck coefficient S, the electrical conductivity σ, the power factor σS , and the electronic contribution to the thermal conductivity κ 0 were calculated (relative to the relaxation time τ) for temperatures up to 800 K, as displayed in Figure 9. Depending on the carrier concentration (relative energies), high Seebeck values around +500 µV K −1 and −400 µV K −1 may be achievable at 400 K, and power factor values of S 2 σ τ −1 = 6 × 10 11 W m −1 K −2 s −1 for both pand n-type at 800 K.For comparison, the corresponding peak values were 13 × 10 11 W m −1 K −2 s −1 for ZnGeP 2 and 11 × 10 11 W m −1 K −2 s −1 for ZnSnP 2 (both at 900 K).As shown in Figure 10, the zT values increase steadily from room temperature up 800 K, reaching values slightly above unity (zT = 1.07, at 1.4 × 10 20 carriers per cm 3 , or 0. carriers per formula unit) for p-type and 0.48 for n-type doped ZnSnAs2 at 1.4 × 10 20 ca ers per cm 3 .The maximum for the p-type compares well with ZnSnP2, where a zTmax = was obtained at 800 K.   To obtain estimated figure of merit values for ZnSnAs 2 , we used a standard relaxation time of τ = 10 fs, as typical for these materials [54], and the experimentally obtained lattice thermal conductivity values κ lat (Equation ( 1)): As shown in Figure 10, the zT values increase steadily from room temperature up to 800 K, reaching values slightly above unity (zT = 1.07, at 1.4 × 10 20 carriers per cm 3 , or 0.006 carriers per formula unit) for p-type and 0.48 for n-type doped ZnSnAs 2 at 1.4 × 10 20 carriers per cm 3 .The maximum for the p-type compares well with ZnSnP 2 , where a zT max = 1.0 was obtained at 800 K.
While the work from Gasson et al. implies that a large range of different charge carrier concentrations of between 0.2 × 10 20 cm −1 and 33 × 10 20 cm −1 can be obtained for ZnSnAs 2 [49], a systematic doping study has not yet been performed on ZnSnAs 2 .p-type doping could be systematically achieved by partial replacements of Zn with Cu or Sn with In, with formulae of Cu 0.006 Zn 0.994 SnAs 2 and ZnIn 0.006 Sn 0.994 As 2 , respectively, corresponding to a hole carrier concentration of the order of 10 20 cm −3 .

Conclusions
We successfully synthesized and characterized the solid solution series ZnSnP 2-y As y via a mechanochemical route.Full miscibility exists, and the increasing As concentration causes a smaller band gap and higher electrical conductivity.The lowest thermal conductivity values were measured for the solutions with y = 0.5 and y = 1.0.
The undoped as-prepared samples all exhibit poor thermoelectric performance (low figure of merit).Our calculations showed that proper p-type doping of ZnSnAs 2 should lead to outstanding performance with figure of merit values exceeding zT = 1, while ndoping would be less successful with peak zT values of the order of 0.5.As previously demonstrated [35], the other end member, ZnSnP 2 , should be able to achieve comparable performance when p-doped, and better performance when n-doped.
Author Contributions: D.R.: conceptualization, data curation, formal analysis, investigation, visualization, writing-original draft; L.T.M.: data curation, formal analysis; H.K.: conceptualization, funding acquisition, supervision, writing-review & editing.All authors have read and agreed to the published version of the manuscript.

Figure 4 .
Figure 4. Arrhenius plots of the electrical conductivity for ZnSnP 2-y As y .

Figure 5 .
Figure 5. Seebeck coefficient for ZnSnP 2-y As y .Open circles: data collected during cooldown.The thermal conductivity data are shown in Figure 6, revealing a typical decrease with temperature due to increased phonon frequencies.Here as well, the data collected after the first heating cycle during the cooldown (open circles) match the data during heating, showing stability of the materials under the measurement conditions.The end members exhibit the largest values with κ = 4.2 W m −1 K −1 and 5.6 W m −1 K −1 at 300 K to 2.4 W m −1 K −1 and 2.7 W m −1 K −1 at 825 K for y = 0 and y = 2, respectively.The lowest values are observed in y = 0.5 and y = 1 with κ = 2.2 W m −1 K −1 and 2.3 W m −1 K −1 at 300 K, respectively, which have equal values of 1.4 W m −1 K −1 at 825 K. Slightly larger than the other solid solutions, the material with y = 1.5 displays κ values between 2.8 W m −1 K −1 at 300 K and 1.7 W m −1 K −1 at 825 K. Increased phonon scattering due to mass fluctuation effects and disorders thus reduces the thermal conductivity by more than a factor of two at room temperature for the solid solutions compared to the end members.

Figure 7 .
Figure 7. Thermoelectric figure of merit for ZnSnP 2-y As y .

Figure 10 .
Figure 10.Estimated figure of merit for ZnSnAs2.Figure 10.Estimated figure of merit for ZnSnAs 2 .

Figure 10 .
Figure 10.Estimated figure of merit for ZnSnAs2.Figure 10.Estimated figure of merit for ZnSnAs 2 .

Table 2 .
EDAX analysis results in atomic-% obtained using five-point measurements for ZnSnP 2-y As y .