Exceptional Thermoelectric Performance of Cu2(Zn,Fe,Cd)SnS4 Thin Films

High-quality Cu2(Zn,Fe,Cd)SnS4 (CZFCTS) thin films based on the parent CZTS were prepared by aerosol-assisted chemical vapor deposition (AACVD). Substitution of Zn by Fe and Cd significantly improved the electrical transport properties, and monophasic CZFCTS thin films exhibited a maximum power factor (PF) of ∼0.22 μW cm–1 K–2 at 575 K. The quality and performance of the CZFCTS thin films were further improved by postdeposition annealing. CZFCTS thin films annealed for 24 h showed a significantly enhanced maximum PF of ∼2.4 μW cm–1 K–2 at 575 K. This is higher than all reported values for single-phase quaternary sulfide (Cu2BSnS4, B = Mn, Fe, Co, Ni) thin films and even exceeds the PF for most polycrystalline bulk materials of these sulfides. Density functional theory (DFT) calculations were performed to understand the impact of Cd and Fe substitution on the electronic properties of CZTS. It was predicted that CZFCTS would have a smaller band gap than CZTS and a higher density of states (DoS) near the Fermi level. The thermal conductivity and thermoelectric figure of merit (zT) of the CZFCTS thin films have been evaluated, yielding an estimated maximum zT range of 0.18–0.69 at 550 K. The simple processing route and improved thermoelectric performance make CZFCTS thin films extremely promising for thermoelectric energy generation.


Synthesis of Precursors
The synthesis of the diethyldithiocarbamate precursors was conducted using a standard Schlenk system under an inert atmosphere of dry nitrogen.All reagents were purchased from Sigma-Aldrich and used as received.Elemental analysis was conducted by the micro-analytical laboratory at The University of Manchester.decomposes between 250-320 ℃ with a residual mass of ~23% (expected value for CuS: ~27%).

Synthesis
[Cd(S 2 CN(C 2 H 5 ) 2 ) 2 ] decomposes between 260-310 ℃ with a residual mass of ~32% (expected value for CdS: ~35%).[Fe(S 2 CN(C 2 H 5 ) 2 ) 3 ] decomposes between 260-325 ℃ with a residual mass of ~20% (expected value for FeS: ~25%).The measured residual masses of the precursors after decomposition are all lower than the calculated values due to sublimation, as observed in previous work. 1e DSC data show obvious endothermic peaks for the decompositions of all precursors.The sharper endothermic peaks might result from the melting of the precursors, whilst the broader endothermic peaks possibly result from the decomposition.After these endothermic events, heat flow increases steadily for all the precursors, indicating continued crystallization. 2 The TGA and DSC characterisation of the precursors is consistent with published data.

Effect of Precursor Ratios on Film Composition
Figures S4a and S4b show the stoichiometries of the prepared CZTS and CZFCTS thin films.
From Figure S4a, it is clear the stoichiometries of the CZTS thin films are close to the theoretical values, indicating that close-to-stoichiometric CZTS thin films can be prepared by AACVD.In contrast, the CZFCTS thin films have similar ratios of Cu:Sn:B-site (Figure S4b), with differences below 6%, but show Sn sufficiency and Cu deficiency compared to the CZTS thin films.Figure S4b shows the Fe/B-site, Zn/B-site and Cd/B-site ratios in the CZFCTS thin films.The CZFCTS-1 thin films prepared with equimolar ratios of the Cd, Fe and Zn precursors show a larger Fe/B-site ratio than Cd/B-site and Zn/B-site ratios, possibly as a result of the higher solubility of Fe in the CZTS lattice. 5- 6 ompared with the CZFCTS-1 thin film, the CZFCTS-2 thin film prepared with a lower concentration of the Fe precursor exhibits obviously reduced Fe/B-site ratio and increased Zn/B-site ratios.The CZFCTS-3 thin film, prepared with a 2:1:2 ratio of Zn:Fe:Cd and a higher total concentration of the B-site metal precursors, has a larger Fe/B-site ratio and smaller Cd/B-site and Zn/B-site ratios than the CZFCTS-2 thin film.Figure S4c shows the correlation between the mole fractions of the metal atoms in the precursor feed and those in the obtained thin films.The CZFCTS S6 thin films have a similar Cu percentage, whereas the CZFCTS-1 thin film exhibits the highest Fe content.Thus, the stoichiometry of CZFCTS thin films can be controlled to some degree by adjusting the ratios of the corresponding precursors in the feed.The optical band gap energies of the CZTS and CZFCTS films were estimated from the absorption spectra using the relation , 7 where is the absorption coefficient,

Raman and UV-VIS Analysis
is photon energy, is a constant, and is the bandgap.In the present work, the equation is satisfied    for = 2, implying direct allowed transitions in both the CZTS and CZFCTS thin films.1] The CZFCTS-1 thin film, which has higher Fe content, exhibits the the smallest bandgap of ~1.3 eV.

Estimation of Thermal Conductivity and zT
As described in the text, the electronic thermal conductivity was estimated from the   electrical transport measurements and combined with the total thermal conductivity measured  _ using TFA to estimate .

𝜅 𝑙𝑎𝑡_𝑇𝐹𝐴
The morphology of the thin film deposited on the TFA test chip (Figure S15a) was slightly different from the films deposited on glass and contains small cracks, probably as a result of thermal stresses caused by the large difference in thermal expansion coefficients between the film and the Si 3 N 4 membrane on the test chip.These cracks could lead to underestimation of the thermal conductivity.We used the reported for bulk CZTS/1 wt% Ag samples with a similar grain size to our  _ thin films 12 to calculate a from our measured . _ The estimated zT values were calculated from and from the relationship  _  _  =

Figure S3 .
Figure S3.In situ XRD patterns for the CZFCTS-1 thin film from room temperature (RT) to 300 ℃.Subplot (b) shows an expansion of the region marked by the blue box in (a).

Figure S4 .
Figure S4.(a)/(b) Stoichiometries of the CZTS and CZFCTS thin films prepared using different precursor ratios (B-site = Zn + Fe + Cd).The orange stars in (a) and (b) represent stoichiometric Cu 2 ZnSnS 4 and Cu 2 (Zn 1/3 Fe 1/3 Cd 1/3 )SnS 4 respectively.(c) Relationship between the mole fractions of the metal atoms in the films and mole fractions of the precursors in the feed.

Figure S7 .
Figure S7.(a) Raman spectra and (b) optical bandgap measurements for the CZTS and CZFCTS thin films prepared using different precursor ratios.

Figure S8 .Figure S9 .
Figure S8.Repeat measurements of the temperature dependence of (a) the Seebeck coefficient ( ), (b)  electrical conductivity ( ) and (c) power factor (PF) of the CZFCTS-1 thin film.There is neglectable  differences between the results obtained from repeat measurements, confirming the stability of the samples at the temperature of < 300℃.

Figure S12 .S13Figure S13 .
Figure S12.SEM image (a) and EDX elemental mapping (b), performed at the locations marked b and c in (a), for CZFCTS-1 thin films annealed for 36 h.

Figure S14 .
Figure S14.Repeat measurements of the temperature dependence of (a) the Seebeck coefficient ( ),  (b) electrical conductivity ( ) and (c) power factor (PF) of a CZFCTS-1 thin film annealed at 390 ℃  for 24 h.

Figure S15 .
Figure S15.(a) SEM and cross-sectional images of the CZFCTS-1 thin film deposited on the TFA test chip.(b) Temperature dependence of the calculated Lorenz number for the CZFCTS-1 thin film  annealed for 24 h.
was prepared using a similar procedure to that outlined above.Na(S 2 CN(C 2 H 5 ) 2 )•3H 2 O (0.04 mol, 9.0 g) was dissolved in 200 mL methanol and stirred for 30 min.Zn(CH 3 COO) 2 (0.02 mol, 3.6 g) was mixed with methanol and the resulting Zn(CH 3 COO) 2 suspension was added dropwise to the S3 Na(S

Table S1 .
Binding energies determined from XPS measurements on the CZTS and CZFCTS thin films.