Boosting Thermoelectric Performance in Nanocrystalline Ternary Skutterudite Thin Films through Metallic CoTe2 Integration

Metal–semiconductor nanocomposites have emerged as a viable strategy for concurrently tailoring both thermal and electronic transport properties of established thermoelectric materials, ultimately achieving synergistic performance. In this investigation, a series of nanocomposite thin films were synthesized, embedding metallic cobalt telluride (CoTe2) nanophase within the nanocrystalline ternary skutterudite (Co(Ge1.22Sb0.22)Te1.58 or CGST) matrix. Our approach harnessed composition fluctuation-induced phase separation and in situ growth during thermal annealing to seamlessly integrate the metallic phase. The distinctive band structures of both materials have developed an ohmic-type contact characteristic at the interface, which raised carrier density considerably yet negligibly affected the mobility counterpart, leading to a substantial improvement in electrical conductivity. The intricate balance in transport properties is further influenced by the metallic CoTe2 phase’s role in diminishing lattice thermal conductivity. The presence of the metallic phase instigates enhanced phonon scattering at the interface boundaries. Consequently, a 2-fold enhancement in the thermoelectric figure of merit (zT ∼ 1.30) is attained with CGST-7 wt. % CoTe2 nanocomposite film at 655 K compared to that of pristine CGST.


XRD measurement and analysis details:
The XRD patterns of thin films were collected at the Taiwan Synchrotron Facility NSRRC by photon energy of 20 KeV (0.61992 Å), therefore, the current 2θ values are obtained at a high energy photon source.The respective 2θ values from Figure 1a, fall in the range of 12-65° in CuKα energy.Additionally, we are providing here a large 2θ spectrum from 5-40° (CuKα range: 12-116°) in Figure S3a.
The refinement fitting process was conducted on TOPAS v5 software using the Rhombohedral CIF file of the ternary skutterudite phase.In the pristine structure, this phase crystallizes in 3 symmetry.However, as we reported in our previous work (Inorganic Chemistry 2022, 61 (10), 4442-4452), the introduction of Sb doping restores the symmetry to symmetry.The 3 3 unavailability of an Im3 CIF file led us to utilize the R3 CIF file for the refinement process.
In the current XRD patterns of the thin films, an absence of Rhombohedral structure reflections at specific 2θ peaks (12.247° and 13.543°) was observed.This observation strongly indicates that the thin films have a cubic structure.Therefore, additional peaks observed in the difference pattern of

S-5
the fitting arise from the absence of these R3 symmetry reflections in the experimental data.Given our primary objective of estimating the weight fraction of CoTe 2 , we decided to use the R3 CIF file for fitting our experimental XRD patterns.This approach, while unconventional, was carefully considered and executed.We ensured the goodness of the fitting by achieving R wp values below 7%.

S-7
Table S1.Refinement parameters R wp , Lattice parameters a, c and volume of unit cell for (1-x) CGST with x CoTe 2 composite sample calculated from the Rietveld refinement of the XRD patterns.

Band alignment analysis:
The work function ( ) of respective phases were calculated from the measured UPS data using the where is the He-light source photon energy, and is the binding energy of (ℎ = 21.2 eV)   - cut-off measured by UPS measurement represented in Figures 4a and b.

S-13
TDTR measurement: In our study, we employed the Time-domain Thermo-reflectance (TDTR) method to measure the thermal conductivity of our thin films.The TDTR technique is based on a pump-probe approach.
It involves using a short laser pulse, referred to as the 'pump', to heat the surface of the known transducer layer on top of the subjected thin film sample.Subsequently, a delayed laser pulse, known as the 'probe', is used to measure the change in reflectance caused by the temperature change on the sample surface.By analyzing the temporal evolution of the surface temperature, we can derive critical information about the thermal properties of the materials under investigation.
A key aspect of accurately measuring the thermal conductivity of the thin film, while effectively excluding the influence of the substrate, lies in the use of a sophisticated three-layer heat diffusion model.The experimental data obtained from TDTR is fitted using the provided model.This model comprehensively accounts for the thermal contributions from both the thin film and the substrate.
By incorporating known properties of the substrate, such as its thickness and thermal conductivity, into the model, we can isolate and accurately determine the thermal conductivity of the thin film alone.This approach ensures that the measured thermal properties are representative of the thin

S-14
film itself, effectively minimizing the impact of the substrate on our results.Further model details can be found in the following research article {David.G Cahill Rev.Sci.Instrum., 75, 12, 2004}.
In the current study, a 200 nm thick Au layer followed by 5 nm Cr was deposited on nanocomposite thin film samples to use as a transducer layer.The measurement quality and reliability are ensured by an average of 5 data collection points at each temperature of the subjected sample.

For Bruggeman model:
The volume fraction was determined using the formula: Weight fraction CoTe2 density CoTe2 х density CGST Table S3.Temperature-dependent specific heat and density values of pristine CGST, CoTe 2 , and of x= 0, 6, 7, 8, and 11 nanocomposite films, calculated by rule of mixture formula.

Figure S9 . 12 Figure S10 .
Figure S9.Surface topology (a) and (b) and the corresponding current distribution map (c) and (d)