Surface Functionalization of Surfactant‐Free Particles: A Strategy to Tailor the Properties of Nanocomposites for Enhanced Thermoelectric Performance

Abstract The broad implementation of thermoelectricity requires high‐performance and low‐cost materials. One possibility is employing surfactant‐free solution synthesis to produce nanopowders. We propose the strategy of functionalizing “naked” particles’ surface by inorganic molecules to control the nanostructure and, consequently, thermoelectric performance. In particular, we use bismuth thiolates to functionalize surfactant‐free SnTe particles’ surfaces. Upon thermal processing, bismuth thiolates decomposition renders SnTe‐Bi2S3 nanocomposites with synergistic functions: 1) carrier concentration optimization by Bi doping; 2) Seebeck coefficient enhancement and bipolar effect suppression by energy filtering; and 3) lattice thermal conductivity reduction by small grain domains, grain boundaries and nanostructuration. Overall, the SnTe‐Bi2S3 nanocomposites exhibit peak z  T up to 1.3 at 873 K and an average z  T of ≈0.6 at 300–873 K, which is among the highest reported for solution‐processed SnTe.


SnTe nanoparticle (NP) synthesis
Large scale SnTe nanoparticles (NPs) were prepared by reacting sodium stannite (Na2SnO2) with sodium hydrogen telluride (NaHTe) in water following a procedure previously reported by Guang Han et al with slight modifications. 1 Firstly, NaBH4 (4.54 g, 120 mmol) was first dissolved in 200 ml deionized water, and then Te powder (5.104 g, 40 mmol) was quickly added into the solution. An Ar flow was introduced to prevent the air. The reaction rate between NaBH4 and Te is very slow and takes ~2 h to react entirely. The suspension was stirred until the solution becomes transparent light purple, indicating the formation of NaHTe. In parallel, NaOH (16 g, 400 mmol) and SnCl2· 2H2O (9.826 g, 40 mmol) were mixed with 200 ml of deionized water. The mixture was stirred at room temperature under Ar flow until the solution became transparent, indicating the formation of Na2SnO2. When NaHTe solution is ready, the Na2SnO2 solution was heated to its boiling point at ca. ~101 o C with a condenser to assure reflux. Then the freshly prepared NaHTe solution was rapidly injected. Since NaHTe is sensitive to the air, two 100ml syringes are prepared to inject NaHTe. Upon injection, the solution color changed from transparent to black, indicating the SnTe NP formation. The mixture was heated again to 101 o C and maintained at this temperature for 30 min. Then the mixture was cooled to room temperature with cooling water. NPs precipitated quickly when the stirring stopped. The transparent supernatant solution was carefully removed. The remaining crude was transferred into centrifuge tubes for further purification.
The SnTe NPs were washed four times with ethanol and acetone alternatively. At each step, SnTe NPs were dispersed by ethanol/ acetone first and then centrifuged at 8000 rpm for 5 min. At last, the washed NPs were dried under vacuum overnight at room temperature and kept in the glovebox for further use.

S4
The Bi2O3/Bi2S3 molecualr complex preparation method applied in this work was developed by R. L. Brutchey et al. 2 The solubilities of Bi2O3 and Bi2S3 in en+EDT solvent (1:10) is ~15-20 % and ~10 %, respectively. Here, we dissolved 150 mg Bi2O3 or 90 mg Bi2S3 with 1.1 ml en+EDT solvent (1 ml en, 0.1 ml EDT) in N2-filled vial. The mixture was sonicated for 1 min to accelerate the dissolving processing until complete dissolution. All the Bi2O3 and Bi2S3molecular complex were prepared fresh before blending with SnTe NPs in MFA.
Samples for LC/Q-TOF were prepared from fully dissolved solutions of Bi2O3 (150 mg/mL, 1:10 (vol/vol) EDT/en) filtered through a 0.45 m syringe filter and diluted with DMF to nanomolar concentrations. Fully dissolved solutions of Bi2S3 (90 mg/mL, 1:10 (vol/vol) EDT/en) were filtered through a 0.45 m syringe filter and diluted with DMSO to nanomolar concentrations. Both solutions were mixed right before injection into the LC/Q-TOF instrument to avoid possible decomposition products. Samples were direct injected into an Agilent 1290 Infinity ll-6545XT AdvanceBio LC/Q-TOF. UV-vis absorption spectroscopy was performed in a 1-cm path length quartz cuvette placed within a 150 mm integrating sphere on a PerkinElmer Lambda 950 UV-vis-NIR spectrometer with roughly equimolar solutions of Bi2O3 (70 mg/mL) diluted in DMF and Bi2S3 (80 mg/mL) diluted in DMSO (2 L ink in 3 mL of either DMF or DMSO). Spectra were collected immediately to avoid any possible absorption from decomposition products.

X-Ray Diffraction (XRD)
X-ray diffraction analyses were carried out on a Bruker AXS D8 ADVANCE powder diffractometer in a Bragg-Brentano geometry with Cu-Kα1 radiation (1.5406 Å). The height of the sample was aligned, and diffraction results were collected from 20° to 80° in 2θ at a rate of 5 °/min. The powder XRD of SnTe NPs with Bi2O3 and Bi2S3 molecular complex before and after annealing processing were carried out, as shown in Figure S3.

Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM)
The size and morphology of initial NPs, annealed NPs, and sintered pellets were examined by field-emission SEM on an Auriga Zeiss operated at 5.0 kV. STEM characterization of the Bi2O3 molecular complex surface coated SnTe samples has been performed using a JEOL JEM2800 microscope operated at 200 kV with a point to point resolution of 0.14 nm. Besides, STEM-EDS elemental mapping was also performed using a JEOL JEM-2800 microscope at an accelerating voltage of 200 kV with a JEOL silicon drift detector (SDD).

Elemental analysis
The overall material composition was investigated by using an Oxford energy dispersive X-ray spectrometer (EDX) attached to the Zeiss Auriga SEM at 15.0 kV.

Electrical Properties
Seebeck coefficients were measured by using a static DC method. Electrical resistivity data was obtained by a standard four-probe method. Both the Seebeck coefficient and the electrical resistivity were simultaneously measured in an LSR-3 LINSEIS system from room temperature to 873 K, under a helium atmosphere. The temperature was increased at a rate of 10 K min -1 . Data approximately was taken every 30 degrees.

S6
Room-temperature hall charge carrier concentrations (nH) and mobilities (μH) were measured with the Van der Pauw method using a magnetic field of 0.6 T (ezHEMS, NanoMagnetics). Each sample is measured five times to get the average value. Temperaturedependent Hall charge carrier concentrations (nH) and mobilities (μH) from 300 K to 873 K were measured utilizing Hall Effect Analyzer by Linseis Company.

Thermal properties
An XFA 500 Xenon Flash Apparatus and an LFA 1000 Laser Flash were used to determine the thermal diffusivities (α) of the samples with an estimated error of ca. 5 %. The total thermal conductivity was calculated by κ = αCpρ, where α is the thermal diffusivity, Cp is the heat capacity, and ρ is the mass density of the specimen. The temperature-dependent Cp values were derived from reference 3. Notably, we applied the Cp of bare SnTe to estimate Cp in this work becasue the real Bi content in all samples are below 3.5%. In Figure S15, we compared Cp values measured by different groups.
We estimated the lattice thermal conductivity (κL) simply by subtracting the electronic thermal conductivity (κe) from the measured total thermal conductivity (κtot): The electronic part κe is directly proportional to the electrical conductivity σ by the Wiedemann-Franz law; An estimation of L can be made using a single parabolic band (SPB) model with acoustic phonon scattering. The calculations based on SPB model results in an L with a deviation of less than 10% as compared with a more rigorous single non-parabolic band and multiple band model calculations. It is well known that the L value is used to estimate the lattice thermal conductivity, which will not change the total thermal conductivity and final zT values. The Lorenz number is given by the formula: where kB is the Boltzmann constant and η represents the reduced Fermi energy, which can be derived from the measured Seebeck coefficients via the following equations:

Error estimation
Taking the system accuracy and the measurement precision (including the measurement of sizes/distances) into account, we estimate an error of ca. 4 % in the measurement of both electrical conductivity and Seebeck coefficient. Combining the uncertainties of electrical conductivity and Seebeck coefficient, so the uncertainty of the power factor is ca. 12%. An LFA 1000 Laser Flash were used to determine the thermal diffusivities (α) of the samples with an estimated error of ca. 5 %. The density (ρ) error is ~2%. To avoid cluttering the plots, error bars were not included in the figures except zT. Therefore, the combined uncertainty for all measurements involved in zT determination shown in the plot is estimated to be ca. 17%.

Klemens model
At a temperature above the Debye temperature ΘD, the ratio of the lattice thermal conductivities including point defects and that of parent material can be expressed as the following equation: where κlat and κlat,p are the lattice thermal conductivities of the defected and parent materials, respectively. In this work, κlat,p = 2.43 W m -1 K -1 . U is defined as: where ΘD, Ω, h and va, represent the Debye temperature, average atom volume, Planck constant and average sound velocity, respectively. Г, the imperfection scaling parameter is a weighted sum of the mass fluctuation ГM and strain field fluctuation ГS and can be written as: where vl and vs denote longitudinal and shear sound velocities, respectively. Then, we can obtain va as 1 3 33 ls Setting Bi alloying as an example. Г is defined as: ( ) The mass fluctuation ГM and strain field fluctuation ГS of S alloying can be obtained using the same method. Figure S1. Annular dark-field scanning transmission electron microscope (ADF-STEM) images of SnTe-1.5%Bi2O3 molecular complex NPs.          Figure S14. EDS mapping of (a) SnTe-2.5% Bi2S3 and (b) SnTe-3.0% Bi2S3 nanocomposites produced with Bi2O3 molecular complexes as precursor. The EXD resolution is too low to detect the element distribution at the grain boundary. S16 Figure S15. EDS mapping of SnTe-1% *Bi2S3* nanocomposite produced with Bi2S3 molecular complexes as precursor.    Cp values is taken from previous work by Zhao et al. 36 Cp in some other references are listed for comparison; 7-10 The temperature dependent TE properties of SnTe-xBi2S3 nanocomposites produced with Bi2O3 molecular complexes as precursor (x=0, 0.5%, 1%, 1.5%, 2%, 2.5%, 3%). (b) thermal diffusivity; (c) Lorenz number; (d) electronic thermal conductivity. Figure S20. The reproduced TE properties measurement of SnTe-xBi2S3 nanocomposites (x=1.5, 3%) produced with Bi2O3 molecular complexes as precursor. 1 st and 2 nd indicates the first and second heating measurement processing (a) electrical conductivity; (b) Seebeck coefficient; (c) power factor; (d) thermal diffusivity; (e) total thermal conductivity; (f) zT. Figure S21. The reproduced TE properties measurement of SnTe-1.5% Bi2S3 nanocomposites produced with Bi2O3 molecular complexes as precursor. R1 and R2 indicates the SnTe-1.5% Bi2S3 nanocomposite synthesized in another two separate batches. (a) electrical conductivity; (b) Seebeck coefficient; (c) power factor; (d) thermal diffusivity; (e) total thermal conductivity; (f) zT. Tables   Table S1. Composition of SnTe-xBi2S3/*Bi2S3* nanocomposites produced with 2O3/Bi2S3 molecular complexes as precursor obtained from SEM-EDS analysis.  Table S2. Room temperature Hall carrier concentration and mobility of SnTe-Bi2S3/*Bi2S3* nanocomposites produced with Bi2O3/Bi2S3 molecular complexes as precursor. Table S3. Densities of SnTe-xBi2S3/*Bi2S3* pellets produced with Bi2O3/Bi2S3 molecular complexes as precursor measured with the Mass-and-volume method.