# Fileset

[Mechanistic Insight into the Effect of Cu Doping on Thermoelectric Properties of Sintered Wet-Chemically Synthesised SnSe2 Nanosheets.pdf](https://mdr.nims.go.jp/filesets/008056d0-bc48-4de4-8167-e5a15e1ef261/download)

## Creator

Simon Moore, [Mari Takahashi](https://orcid.org/0000-0001-9252-0721), [Philipp Sauerschnig](https://orcid.org/0000-0003-4666-5262), Keiji Kobayashi, Koichi Higashimine, Masanobu Miyata, Takahiro Baba, [Jun Uzuhashi](https://orcid.org/0000-0003-2023-8158), [Michihiro Ohta](https://orcid.org/0000-0002-9093-7117), [Takao Mori](https://orcid.org/0000-0003-2682-1846), [Tadakatsu Ohkubo](https://orcid.org/0000-0003-3548-1951), [Shinya Maenosono](https://orcid.org/0000-0003-2669-8219)

## Rights

This document is the Accepted Manuscript version of a Published Work that appeared in final form in ACS Applied Energy Materials, copyright © 2024 American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see https://doi.org/10.1021/acsaem.4c01698[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Mechanistic Insight into the Effect of Cu Doping on Thermoelectric Properties of Sintered Wet-Chemically Synthesized SnSe<sub>2</sub> Nanosheets](https://mdr.nims.go.jp/datasets/35c4e4ad-b743-4a5b-aa96-56ef917d6d55)

## Fulltext

1 Mechanistic Insight into the Effect of Cu Doping on Thermoelectric Properties of Sintered Wet-Chemically Synthesised SnSe2 Nanosheets Simon Moore,1 Mari Takahashi,1 Philipp Sauerschnig,2 Keiji Kobayashi,1 Koichi Higashimine,3 Masanobu Miyata,1 Takahiro Baba,4,5 Jun Uzuhashi,6 Michihiro Ohta,2 Takao Mori,4,5 Tadakatsu Ohkubo,6 Shinya Maenosono1* 1 School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan 2 Global Zero Emission Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8569, Japan 3 Center for Nano Materials and Technology, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan 4 Research Center for Materials Nanoarchitectronics, National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan 5 Graduate School of Pure and Applied Science, University of Tsukuba, Tennoudai 1‑1‑1, Tsukuba, Ibaraki 305‑8671, Japan 6 Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan  Corresponding Author* Email: shinya@jaist.ac.jp Phone: +81-761-51-1611 mailto:shinya@jaist.ac.jp 2 ABSTRACT  SnSe2 – an uncommon, sustainable n-type thermoelectric material – has previously been doped with elements including Cu to improve its thermoelectric figure of merit, ZT value. However, the effects and mechanisms behind Cu doping remain unclear. To provide important mechanistic insight, SnSe2:Cu nanosheets were synthesised, sintered, and subjected to multifaceted analysis using a variety of analytical techniques. The results suggested that intercalated Cu+ ions contribute to the density of states of the valence band, shifting the Fermi energy towards the conduction band. The carrier concentration was raised, causing a shift to conduction across two conduction band minima. Additionally, a change to ionised impurity scattering dominated carrier mobility meant a high mobility at high temperatures, since electron-phonon interactions were insignificant. The result was a significantly increased electrical conductivity, while the thermal conductivity decreased due to scattering at nanosheet grain boundaries and Cu+ ions. High ZT values of 0.2 in the in-plane, and 0.6 in the out-of-plane directions were achieved at 667 K.  Keywords: waste heat harvesting, sustainable materials, XAFS, nanoparticles, chalcogenide      3 INTRODUCTION Converting directly between temperature differences and electrical energy via the Seebeck and Peltier effects, thermoelectric (TE) materials have garnered much attention, with a variety of semiconductors investigated for applications in waste heat harvesting,1,2 energy generation,3,4 cooling devices,5 and others.6 Of these, waste heat harvesting is considered a key technology in the development of sustainable energy systems.6,7 The conversion efficiency of TE materials is related to the dimensionless TE figure of merit, 𝑍𝑍𝑍𝑍 = 𝜎𝜎𝑆𝑆2𝑇𝑇 𝜅𝜅⁄ , where 𝜎𝜎 is electrical conductivity; 𝑆𝑆 is the Seebeck coefficient; 𝑇𝑇 is the absolute temperature; and 𝜅𝜅 = 𝜅𝜅e + 𝜅𝜅l is the total thermal conductivity, which includes contributions from charge carriers, 𝜅𝜅e  (the carrier thermal conductivity), and phonons, 𝜅𝜅l  (the lattice thermal conductivity). Current successfully implemented TE devices use materials such as Bi-Te8,9 and Pb-Te,10,6 for which recent developments have seen high ZT values of 1.24 at 350 K for Bi0.5Sb1.5Te3 – 20wt% Te11 and 2.8 at 850 K for Na0.03Pb0.97Te – 2% MgTe – 0.75% GeTe.12 However, the use of rare or highly toxic elements such as Te, Sb and Pb means that such materials are considered unsustainable for future, large-scale, practical use.13 On the other hand, more sustainable materials, such as copper tin sulphide14 and metal chalcogenides other than tellurides,15,16 have lower intrinsic ZT values, so research has focused on their improvement. The path to higher ZT lies in increasing the power factor (PF= 𝜎𝜎𝑆𝑆2), while simultaneously decreasing 𝜅𝜅. The methods to do so include dopant atoms to increase 𝜎𝜎;17,18 band engineering to maximise PF;19,20 and defect engineering (for example, the introduction of controlled nanograin boundaries), which decreases 𝜅𝜅 through increased phonon scattering.21,22  4 One highly discussed sustainable TE material is tin monoselenide (SnSe), a p-type semiconductor with a layered, orthorhombic crystal structure (Figure S1a). Made of relatively abundant elements with low toxicity (further discussion of the sustainability of Sn-Se can be found in the Supporting Information and Figure S2), it caused a stir with a reported maximum ZT value of 2.6 along the b-axis at 923 K for the single crystal, thanks to its anharmonic structure creating a low intrinsic 𝜅𝜅 (0.35 W∙m−1∙K−1 along the b-axis at 923 K).23 Polycrystalline SnSe generally showed lower TE performance but recent developments have seen significant improvements,24 with strategies such as Ge alloying resulting in a high ZT of 1.75 at 873 K.25 Another strategy focused on the removal of a tin oxide layer, which often forms on the outside of SnSe grains thanks to the exposed Sn atoms and thus has a detrimental effect on the TE performance of polycrystalline SnSe. Removal of this layer by a reducing gas led to a reported maximum ZT of 3.1 at 783 K.26 Although intrinsically a p-type semiconductor, doping to create n-type SnSe has also borne fruit with high ZT values reported,27,28 including a value of 2.23 at 873 K for Cd and Pb-doped SnSe,29 which represents a record high for all reported n-type polycrystalline materials.   In contrast, the n-type, hexagonal, Van der Waals (VdW) structured (Figure S1b) tin diselenide (SnSe2) has a much lower ZT value, (0.10 and 0.15 at 673 K along the ab-plane and c-axis, respectively) due to its rigid hexagonal crystal structure giving a high intrinsic 𝜅𝜅, and a relatively low carrier concentration (approximately 2.3×1018 cm−3 at 300 K) resulting in low PFs of 3.5 and 1.0 µW∙cm−1∙K−2 at 673 K for the ab-plane and c-axis, respectively.30 However, SnSe2 has shown promise as a sustainable n-type TE material, with significant increases in PF and ZT reported for Cl and Br doping, which substitute in Se sites and increase the carrier concentration. High ZT values were reported of 0.3 (in-plane, perpendicular to the pressing direction) and 0.4 (out-of-plane, parallel to the pressing direction) at 673 K for 12% Cl-doping;31 and 0.25 (in-plane) and  5 0.55 (out-of-plane) at 670 K for 10% Br-doping.32 Interestingly, although n-type SnSe has demonstrated very high ZT values in the higher temperature ranges (> 600 K), the reports of halogen-doped n-type SnSe2 have shown improved ZT values at more moderate temperatures (between 400 and 600 K). Given that the value of 𝜅𝜅 remains high and 𝜎𝜎 remains low for these materials suggests that with further improvement, SnSe2 holds potential as a TE material for this temperature range. Additionally, thanks to the complete surrounding of Sn atoms by Se in the crystal structure of SnSe2 (Figure S1b), it does not suffer from the surface oxidation problem of polycrystalline SnSe, simplifying the fabrication process.    As another way to improve the electrical properties of SnSe2, small amounts of Cu doping have also been trialled. Wang et al. reported a remarkably high ZT of 0.75 at 300 K for polycrystalline SnCu0.01Se2.33 Although, they did not account for the anisotropy in the TE properties, so the practical ZT value is likely to be much lower. Zhou et al. reported an ab-plane ZT of approximately 0.35 at 770 K for SnCu0.005Se2. They also co-doped Cu with Br to achieve a maximum ab-plane ZT of 0.6 at 770 K for SnCu0.005Se1.98Br0.02.34 While both reports demonstrated that Cu could be an effective dopant, their contrasting results also suggested that different mechanisms were in play. For Wang et al., Cu doping increased the Hall carrier concentration, 𝑛𝑛H , and decreased Hall mobility, 𝜇𝜇H, across the measured temperature range (10–300 K). On the other hand, Zhou et al.’s SnCu0.005Se2 showed no change in 𝑛𝑛H until the temperature exceeded 600 K. Based on atom probe tomography, scanning transmission electron microscopy (STEM) and density functional theory (DFT) calculations, they concluded that Cu atoms intercalated in the VdW gap of SnSe2, and that their ionization at high temperatures increased 𝑛𝑛H. Bridging of the VdW gap by Cu atoms and electron delocalisation in the surrounding Se atoms also enhanced 𝜇𝜇H. In contrast, Wang et al. did not investigate Cu’s position in the crystal lattice nor the mechanism of electron donation.  6 Additionally, other reports have included Cu as a co-dopant alongside Br,35 with one report combining this with heat treatment of the sample to produce a high ZT of approximately 1.1 at 770 K.36 However, these reports did not focus on investigating the effects of the Cu doping itself. As a result, the exact mechanism of Cu doping in SnSe2 remains unclear and we are unable to reconcile or reproduce the contrasting results of Wang et al. and Zhou et al. Cu is also a popular dopant in other n-type and p-type TE materials, such as SnSe,37 ZnSb,38 Bi2Te3,39 and Bi2Te2.7Se0.3.40 Cu often increases 𝑛𝑛H, and thus seems to be able to act as both an electron and hole donor, depending on the material. However, the complex mechanisms are either not fully clear or not investigated in detail, especially for nanostructured materials and SnSe2. To provide a mechanistic insight into Cu doping of SnSe2 and, by extension, other n-type TE materials, we present a thorough investigation into the TE properties of Cu-doped SnSe2 nanosheets (NSs), synthesised based on our previous report,41 while Cu doping was achieved through a Cu impurity in the Sn precursor. The NSs were sintered into pellets by hot press, after which the TE properties and the electronic transport properties were investigated by light flash analysis (LFA), ZEM-3, and a physical properties measurement system (PPMS). A thorough structural analysis was performed by X-ray diffraction (XRD), scanning transmission electron microscopy (STEM), scanning electron microscopy (SEM), energy dispersive x-ray spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS), and three-dimensional atom probe (3DAP). X-ray absorption fine structure (XAFS) Sn and Se K-edge measurements and DFT calculations helped investigate the mechanism in more detail. High ZTs of approximately 0.2 in the in-plane (IP) direction, and 0.6 in the out-of-plane (OOP) direction at 667 K were observed – among the highest values reported for SnSe2.17,30-36,42 The characterisation and analysis provide new insights into the mechanisms of Cu doping in SnSe2, which help to explain the observed properties.  7 EXPERIMENTAL SECTION Chemicals  Oleylamine (OLA, technical grade 70%), tin (IV) acetate (Sn(OAc)4), selenium powder (Se, 100 mesh, purity 99.99% metals basis), trioctylphosphine oxide (TOPO, technical grade 90%), and tin (IV) fluoride (SnF4) were purchased from Sigma Aldrich. Thiourea (purity > 99.0%) was purchased from Tokyo Chemical Industry Co. Ltd. Toluene was purchased from Fujifilm Wako and methanol was purchased from Kanto Chemical. All chemicals were used without further modification.  Synthesis of SnSe2:Cu and SnSe2 NSs SnSe2 NSs were synthesised using the method in our previous report41 with the following changes. The scale was increased to use five times the reported amounts and the ratio of the Sn to Se precursors was set at 1:10 to produce pure phase SnSe2 NSs. SnSe2 NSs without the Cu impurity were synthesised by changing the Sn precursor to SnF4, which was included in the flask from the start of the synthesis instead of being injected. The total amount of OLA, TOPO, and Se, as well as the temperature profile were kept identical.  Ligand exchange  Approximately 1.20 g of as-synthesised SnSe2 NSs were dispersed in 18.75 mL of toluene in a conical beaker by sonication. At the same time, 3.75 g of thiourea was dissolved in 56.25 mL of methanol. After the NSs were completely dispersed and the thiourea completely dissolved, the thiourea-methanol solution was added to the NS-toluene mix and the two sonicated together for two hours, with occasional swilling by hand. Then, the mixture was separated into four 50 mL centrifuge tubes and centrifuged at 4,640 × g for 5 mins, after which the supernatant was discarded. After this, washing was performed by redispersing the NSs using 22.5 mL of methanol in each  8 tube, followed by centrifugation at 4,640 × g for 5 mins and the disposal of the supernatant. This washing step was performed four times, and the final product was dried in a vacuum. Thermogravimetric analysis (TGA, TGA/SDTA851, Mettler Toledo) was used to confirm the success of the ligand exchange. Pelletisation   After ligand exchange, the NS powder was sintered into a pellet using a hot press at a pressure of 300 MPa for 30 mins at 400 °C in an Ar atmosphere. In total, three 10 mm diameter cylindrical pellets were produced – two that were 2 mm thick and one that was 8 mm thick. Characterisation  As synthesised NSs were characterised as shown in our previous report.41 After ligand exchange the NSs were characterised using powder X-ray diffraction (XRD, Mini-Flex600, Rigaku) using a Cu Kα source; scanning electron microscopy equipped with energy dispersive X-ray spectroscopy (SEM-EDS, Miniscope TM3030Plus, Hitachi) at an acceleration voltage of 15 kV; transmission electron microscopy (TEM, H-7650, Hitachi) at an acceleration voltage of 100 kV; scanning TEM (STEM, JEM-ARM200F, JEOL) at an acceleration voltage of 200 kV, equipped with EDS and a high-angle annular dark field detector (HAADF); and X-ray photoelectron spectroscopy (XPS, AXIS-ULTRA DLD, Shimadzu/Kratos) using a monochromated Al Kα source.  X-ray absorption fine structure (XAFS) analysis was performed by measuring the Sn and Se K-edge X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) in transmission mode at the NW-10A beamline of the Photon Factory – Advanced Ring (PF-AR) facility at the High Energy Accelerator Research Organisation (KEK) in Tsukuba, Japan. The energy and current of the electrons in the storage ring were 6.5 GeV and 60 mA, respectively.  9 After sintering, the pellets were first polished using silicon carbide abrasive paper (grit size 2,000) and then had their densities measured using a gas pycnometer (Shimadzu AccuPyc II 1340) and by calculation based on the measured mass and dimensions of each pellet. Powders of the pellets were characterised by XRD, SEM-EDS and XPS. The pellets’ outer surfaces were also characterised by SEM-EDS. For further investigation an offcut from one of the 2 mm pellets had its cross-section investigated. It was first polished using an Ar beam polisher (IM4000, Hitachi High-Tech), followed by observation by SEM-EDS (CrossBeam550, CarlZeiss) at an acceleration voltage of 15 kV. Following this a small, thin rectangular sheet and a small cone were cut from the cross section of the pellet, as shown in Figure S3, using an FIB-SEM dual beam system (Helios 5UX, Thermo Fisher Scientific). The rectangular sheet was further thinned and mounted onto a sample holder for observation by STEM (Titan G2 80-200, Thermo Fisher Scientific) equipped with HAADF and EDS. The cone was then used for measurement by 3D atom probe (3DAP, LEAP 5000XS, CAMECA) using a 355 nm wavelength laser pulsing mode at a repetition rate of 250 kHz and a laser energy of 20 pJ with the sample temperature set to 30 K. TE properties measurement Each pellet had its density measured using a gas pycnometer (Shimadzu Accupyc II 1340 pycnometer) in a He atmosphere. The thermal diffusivity, D (m2·s−1) was measured using light flash analysis (LFA, Netzsch LFA467). The specific heat, c (J·kg−1·K−1), was derived by measuring the sample and a standard sample (Pyroceram 9606, Netsch) under the same conditions using the same LFA. κ was then calculated from the relationship 𝜅𝜅 = 𝑑𝑑𝑑𝑑𝑑𝑑, where d is density (kg∙m−3). Electrical properties (𝜎𝜎 and 𝑆𝑆) were measured using a ZEM-3 instrument (Advance Riko) for a temperature range of 350–667 K. PF and ZT were calculated using 𝑃𝑃𝑃𝑃 = 𝜎𝜎𝑆𝑆2 and 𝑍𝑍𝑍𝑍 = (𝜎𝜎𝑆𝑆2 𝜅𝜅⁄ )𝑇𝑇. The carrier thermal conductivity, 𝜅𝜅e was calculated from the Wiedermann-Franz  10 Law 𝜅𝜅e = 𝐿𝐿𝐿𝐿𝐿𝐿 where L is the Lorentz number (2.44 × 10−8 W∙Ω∙K−2) and 𝜅𝜅l was calculated by 𝜅𝜅l = 𝜅𝜅 − 𝜅𝜅e. The pellet preparation and measurement technique are shown in Figure S4. One of the 2 mm thick pellets had its out-of-plane (OOP, parallel to the pressing direction) 𝜅𝜅 measured by LFA, before it was cut into a rectangular bar and the in-plane (IP, perpendicular to the pressing direction) 𝜎𝜎 and 𝑆𝑆 were measured by ZEM-3. The 8 mm thick pellet was first sanded down into a 6 mm × 8 mm × 2 mm pellet in which the 6 mm side was parallel to the pressing direction. The IP 𝜅𝜅 and OOP 𝜎𝜎 and 𝑆𝑆 were then measured in the same manner as the first pellet. The error bars for the TE properties were calculated using known measurement errors of 5% for the values measured by ZEM-3 (σ and S), and 7% for κ measured by LFA. Based on these, the maximum and minimum possible values of σ, S, and κ were calculated, these were then used to calculate the maximum and minimum possible values of PF and ZT. To investigate the possibility of measuring 𝜅𝜅 in the IP direction using a thin pellet, the second 2 mm thick pellet first had its OOP D measured by LFA (Netzsch LFA467) (Figure S5a). The pellet was then cut into a rectangular bar, and the same LFA device was used to heat the bar at both ends while the thermal response was measured in the middle. A COMSOL simulation was then used in which the OOP D was fixed by the previous measurement and the IP D was varied until it matched the measured signal (Figure S5b). Then IP κ was calculated from the equation 𝜅𝜅 = 𝑑𝑑𝑑𝑑𝑑𝑑. Electronic transport properties measurement After measurement of the TE properties by LFA and ZEM-3, 𝑛𝑛H , 𝜇𝜇H , and temperature dependence of the electrical resistivity, 𝜌𝜌 , were measured using a Physical Properties Measurement System (PPMS, Quantum Design). For the measurement, gold wire electrodes were attached to the pellet using a gold paste. A thin layer of gold was applied via sputtering to the locations which would be used for the electrodes to enhance adhesion and reduce contact resistance.  11 The pellet was placed on a copper plate and then onto the sample puck, at which point the gold wires were connected to the terminals of the sample puck using a silver paste. The Hall voltage, 𝑉𝑉H was measured over a temperature range of 10–340 K, under a magnetic field of ± 5 T and a current of 30 mA. 𝜌𝜌 was measured over a temperature range of 3–340 K at a heating rate of 1 K∙min−1, under a current of 30 mA. The Hall coefficient, 𝑅𝑅H, was calculated using 𝑅𝑅H = 𝑉𝑉H𝑡𝑡 𝐵𝐵𝐵𝐵⁄  where t is the thickness of the pellet (m), B is the magnetic field (T), and I is the applied current (A). 𝑛𝑛H was then calculated using 𝑛𝑛H = 1 𝑒𝑒𝑅𝑅H⁄  where e is the elementary charge, and 𝜇𝜇H was found using 𝜇𝜇H = 1 𝑒𝑒𝑒𝑒𝑛𝑛H⁄ . The weighted mobility, 𝜇𝜇W, was calculated using the values for 𝑆𝑆 and 𝜎𝜎 measured by ZEM-3 and the following equation:43  𝜇𝜇W = 3ℎ2𝜎𝜎8𝜋𝜋𝜋𝜋(2𝑚𝑚𝑒𝑒𝑘𝑘B𝑇𝑇)3 2��exp� |𝑆𝑆|𝑘𝑘B 𝑒𝑒⁄ −2�1+exp�−5� |𝑆𝑆|𝑘𝑘B 𝑒𝑒⁄ �−1�+3𝜋𝜋2 |𝑆𝑆|𝑘𝑘B 𝑒𝑒⁄1+exp�5� |𝑆𝑆|𝑘𝑘B 𝑒𝑒⁄ �−1�� (Eqn. 1)  where h is the Planck constant, e is the elementary charge, 𝑚𝑚𝑒𝑒 is the electron rest mass, 𝑘𝑘B is the Boltzmann constant, and |𝑆𝑆| is the magnitude of the Seebeck coefficient (V∙K−1). Density functional theory calculations The electronic structure of an SnSe2(P-3m1) primitive cell and a Cu intercalated SnSe2 supercell, which included eight Sn, sixteen Se, and a Cu atom, were calculated using the OpenMX software package44,45 based on density-functional theory (DFT), norm-conserving pseudopotentials, and pseudo-atomic localized basis functions. We applied the generalized gradient approximation proposed by Perdew, Burke, and Ernzerhof.46 The contribution of spin-orbit coupling was also considered. The pseudo-atomic orbital basis functions are specified as Se7.0-s3p2d1, Sn7.0- 12 s3p2d1 and Cu6.0S-s2p1d1, where the symbols respectively denote the element, cutoff radius (Bohr), and specification of the optimized orbitals. We applied the unit cell with k-point grids (k grids) of 14×14×8 and 7×7×4 for the SnSe2 cell and the Cu intercalated SnSe2 supercell. We set an energy cutoff of 220 Ry for numerical integration. We performed an optimization of the lattice constants and atomic positions using the steepest descent method, with a force criterion of less than 2.0×10−4 Hartree Bohr−1.  RESULTS AND DISCUSSION Structural characterisation of Cu-doped SnSe2 nanosheets and pellets Cu-doped, pure-phase, hexagonal SnSe2 NSs (SnSe2:Cu) were synthesised using a scaled-up version of our previously reported method.41 XRD and TEM of the produced NSs are shown in Figure S6. The Cu impurity was initially discovered by SEM-EDS and STEM-EDS (Figure S7) of the NSs and confirmed by XPS analysis of the pellet with a detected amount of 1.2 at% relative to Sn. The Cu 2p and Cu L3M4,5M4,5 Auger regions of the XPS spectra are show in Figure S8. The Sn precursor, Sn(OAc)4, was the source of the Cu impurity, with SEM-EDS and XPS analysis suggesting a composition of approximately 2.4 at% Cu (Figure S9 and Table S1). Multiple syntheses with different bottles produced the same results, suggesting that Cu is always present, though the manufacturing process of Sn(OAc)4 has not been made public. The weak XPS signal for the Cu L3M4,5M4,5 Auger region made full deconvolution difficult, but the Cu 2p peaks’ binding energies suggested the presence of Cu+ in the SnSe2:Cu pellet and Cu metal in Sn(OAc)4, indicating oxidation taking place during the synthesis.   13  Figure 1. Characterisation of the SnSe2:Cu pellets. (a) XRD pattern with after ligand exchange (ALE) NSs as a comparison. Red is reference pattern for hexagonal SnSe2 (PDF: 01-089-2939). The blue star is a peak originating from a damaged XRD substrate. (b) SEM and (c) STEM-HAADF images showing the arrangement of NSs in the pellet. In the SEM and STEM images, the vertical direction is the OOP direction. (d) Schematic image of the pellet showing how the IP and OOP directions relate to the ab-plane and c-axis of SnSe2.  Since large ligands can influence the TE properties,47 a ligand exchange was performed from trioctylphosphine oxide (TOPO) and oleylamine (OLA) to thiourea before sintering by hot press. Investigation by thermogravimetric analysis (TGA) before (BLE) and after (ALE) the ligand exchange (Figure S10), confirmed the successful exchange. Powder XRD patterns of ALE NSs and the pellet are shown in Figure 1a. All peaks except for one (at approximately 28°) could be indexed to hexagonal SnSe2, though only the main peaks have been labelled in the figure. The remaining small peak at approximately 28° was found to originate from a damaged Silicon XRD substrate, the XRD pattern for which can be found in Figure S11. Mean crystalline sizes (𝐷𝐷XRD)  14 of the (001) and (110) peaks, representing the c-axis and ab-plane, respectively, were calculated by Scherrer formula, producing 13.7 nm and 35.2 nm for the (001) and (110) peaks of the ALE NSs, and 33.7 nm and 43.5 nm for the (001) and (110) peaks of the pellet. Thus, we can see that the pellet retains the NSs’ pure-phase SnSe2, though some grain growth has taken place.  Cross-sectional SEM and STEM-HAADF images of the pellet are shown in Figures 1b,c and additional images can be found in Figure S12. Note that in all such images the vertical and horizontal directions represent the OOP and IP directions of the pellet, respectively. We know from our previous work on the synthesis of these SnSe2:Cu NSs that the layers of the SnSe2 crystal structure lie parallel to the top and bottom surface of each NS,41 and the clearly visible atomic structure in Figure 1c confirms this to be the case in the pellet as well. From Figures 1b,c and S12, we can see that the general alignment of the NSs appears to be along the IP direction of the pellet, an observation supported by the larger grain growth evident in the (001) direction. This would suggest that the pellets’ IP and OOP directions can represent the ab-plane and c-axis of SnSe2, respectively (Figure 1d). However, there are a significant number of NSs with different orientations, so we can expect that, while there will be some anisotropy in the TE properties of the pellet, the anisotropy will not be as strong as that found in a single crystal. For this reason, we will continue to refer to the pellet’s IP and OOP directions rather than the lattice directions of SnSe2. The voids visible in Figures 1b and S12 raise questions about the possibility of a low density or high porosity of the pellets leading to an effect on the measured TE properties. Therefore, the densities of all three pellets were determined using a gas pycnometer and the measured mass and dimensions of each pellet. With the density of single-crystal SnSe2 reported to be 5.54 g∙cm-3,48 the relative densities of the three pellets were found to be 98.3%, 92.6%, and 95.6% for the two 2 mm thick pellets and the one 8 mm thick pellet, respectively. Since a minimum value of 95% is  15 generally considered desirable, as above this value any effects of porosity should be low, we can see that two of our pellets meet this requirement. The lower density pellet was only used for the determination of the IP 𝜅𝜅 by COMSOL simulation, the results of which – as we will see later – agreed well with the other pellets. So, we concluded that the lower density would not have a significant effect on the conclusions drawn in the rest of this paper.  The distribution of Cu was investigated using STEM-EDS (Figure S12), SEM-EDS (Figure S13), and 3DAP (Figure 2a), the results of which are summarized in the schematic diagram in Figure 2b. Segregated nanoscale Cu-rich domains were visible, with the region in the 3D atom maps having a composition of 42.41 at% Sn, 56.12 at% Se, and 1.47 at% Cu, compared to 33.36 at% Sn, 66.63 at% Se, and 0.01 at% Cu in the surrounding material. Note that 3DAP’s accuracy is known to be high enough for the at% to be stated to two decimal places.49 The nanoscale Cu-rich regions were also visible in the STEM-EDS mapping conducted on the pellet cross-section (Figure S12), although their small size makes them undetectable on the cross-sectional SEM-EDS mapping in Figure S13b. From the STEM-EDS images the volume fraction of these nanoscale Cu-rich regions can be calculated to be 0.18%. The low volume fraction of such regions suggests that they are unlikely to have a significant effect on the TE properties of SnSe2:Cu and, as such, further discussion of their distribution and nature can be found in the Supporting Information. The remaining Cu, though low in concentration, is homogeneously distributed as Cu+ ions, so we focused our investigations on their effect on the TE properties.  16  Figure 2. Characterisation of the distribution of Cu in the SnSe2:Cu pellets. (a) 3D atom maps of a sample taken from the cross-section of the pellet. Green, blue, and orange represent Sn, Se, and Cu, respectively. (b) schematic image of the distribution of Cu ions (black dots) and Cu-rich regions (blue shapes) in the pellet as determined by SEM, STEM, and EDS analysis. Not to scale.  TE properties of Cu-doped SnSe2 Since SnSe2 has an anisotropic structure and due to the difficulty of producing thick pellets for cutting into multiple specimens, the IP and OOP direction TE properties were measured using 3 comparable pellets (Figure S4). One 2 mm thick pellet had its OOP 𝜅𝜅 measured by LFA, and IP 𝜎𝜎 and 𝑆𝑆 measured using a ZEM-3. A second 2 mm thick pellet was used to estimate the IP 𝜅𝜅 at 300 K using LFA and a COMSOL simulation. An 8 mm thick pellet was cut into a plate parallel to the pressing direction for measurement of the IP 𝜅𝜅 and OOP 𝜎𝜎 and 𝑆𝑆. The results were combined to give the full IP and OOP direction TE properties of SnSe2:Cu shown in Figure 3 alongside bulk,  17 single crystal SnSe2.30 The thermal cycling stability of the samples was confirmed by measurement over a full heating and cooling cycle for both LFA and ZEM-3, the resulting σ, S, and κ values for which are shown in full in Figure S14. The heating and cooling stages of the measurements show good agreement indicating the thermal stability of the material. It’s worth noting that, as shown in the TGA results in Figure S10, SnSe2 is thermally stable until approximately 550 °C (823 K), above which Se evaporation rapidly occurs. It is for this reason that the TE properties measurement temperature was limited to 393 °C (667 K).  The electrical properties of SnSe2:Cu in both directions are shown in Figure 3a-c alongside those of bulk, single crystal SnSe2 for comparison. An increased 𝜎𝜎 in both directions (Figure 3a), which also increased with temperature, along with a decrease in the magnitude of 𝑆𝑆  (Figure 3b), suggested at an increased carrier concentration for SnSe2:Cu compared with bulk SnSe2. Combining these, a significantly higher PF in the OOP direction and a slight increase in PF in the IP direction (Figure 3c) was achieved. It is notable that the electrical properties in both the IP and OOP directions of SnSe2:Cu are remarkably similar to each other, in contrast with the high anisotropy shown in the bulk, single crystal SnSe2. As discussed earlier, and shown in Figure 1b, the anisotropy in the SnSe2:Cu pellets is not as strong as that in a single crystal, which may be the reason for these results. Additionally, it should be noted that the IP and OOP direction properties were measured using different pellets, which – while they were fabricated under the same conditions – may not be completely identical, thus making some additional variation in the TE properties expected, and the comparison of the two directions difficult.   Looking at the thermal transport properties in Figures 3d,e, the IP direction 𝜅𝜅 can be clearly seen to have decreased significantly (approximately half at 350 K), while the OOP direction has only seen a slight decrease. Importantly, there was good agreement between the IP 𝜅𝜅 measured by LFA  18 and that estimated by our COMSOL simulation (Figure 3e), indicating that this is a valid method for the measurement of IP 𝜅𝜅 in thin pellets.  We can also see that, despite the apparent increase in carrier concentration, the carrier thermal conductivity, 𝜅𝜅𝑒𝑒 remains extremely low to the point of being negligible, thus meaning that 𝜅𝜅 = 𝜅𝜅𝑙𝑙 and the reductions in 𝜅𝜅 are a direct result of reduced 𝜅𝜅𝑙𝑙. We have seen from Figures 1b,c that the sintering of our NSs into a single pellet has caused the creation of a large number of additional grain boundaries, which are known to reduce 𝜅𝜅𝑙𝑙 due to increased phonon scattering. The anisotropic crystal structure of SnSe2 includes a rigid lattice extending across the ab-plane, but a layered structure including VdW gaps, which hinder the propagation of phonons, stretching along the c-axis (Figure S1). This accounts for the intrinsic high ab-plane and low c-axis 𝜅𝜅𝑙𝑙  of SnSe2 and also the reduction we see in our nanostructured SnSe2:Cu pellet. The large number of additional grain boundaries in the IP direction significantly increases phonon scattering and reduces 𝜅𝜅𝑙𝑙. In the OOP direction, on the other hand, the number of additional grain boundaries, with an average spacing of about 30 nm based on the mean crystalline size, still does not compare to the number of VdW gaps that would exist along the c-axis of a single crystal, thus explaining the very slight reduction. Overall, the ZT value (Figure 3f) in both directions significantly increased compared to bulk, single crystal SnSe2, with maximums of 0.2 in the IP direction and 0.6 in the OOP direction at 667 K.  From these results we can see that although both directions experienced an increase in PF and a reduction in 𝜅𝜅, the main contributors to the increases in ZT were the decreased 𝜅𝜅 in the IP direction, but the increased PF in the OOP direction. These results compare very favourably with those of other reports,35,36 as shown in Figure S15, where maximum ZT values (regardless of measurement direction) of several reports of Cu-doped SnSe2 have been reproduced. We can see that only the Cu and Br co-doped and heat treated SnSe1.97Br0.03 – 3% Cu AHE (after heat  19 treatment) reported by Wei et al. showed a higher result.36 Note that, as we mentioned previously, the notably high ZT at 300 K reported by Wang et al.33 does not take into account the anisotropic nature of SnSe2, making this value unreliable, and thus has not been included in this comparison.    Figure 3. TE and electronic transport properties of the SnSe2:Cu pellets (red) and bulk, single crystal SnSe2 (black, reported in reference 30) in the IP (solid circles and lines), and OOP (open squares and dashed lines) directions. (a) electrical conductivity, 𝜎𝜎; (b) Seebeck coefficient, 𝑆𝑆; (c) power factor, PF; (d) IP thermal conductivity; (e) OOP thermal conductivity; (f) ZT value; (g) Hall carrier concentration, 𝑛𝑛H; (h) Hall mobility, 𝜇𝜇H; and (i) weighted mobility, 𝜇𝜇W. The carrier thermal conductivity, 𝜅𝜅e, is also shown in blue in (d) and (e). The green triangle in panel (e) is the value of IP 𝜅𝜅 estimated at 300 K using the COMSOL simulation.    20 The carrier transport properties of SnSe2:Cu were investigated by PPMS, with 𝑛𝑛H (Figure 3g) and 𝜇𝜇H  (Figure 3h), measured between 10 and 340 K. For higher temperatures, the weighted mobility, 𝜇𝜇W (Figure 3i), was calculated from the 𝜎𝜎 and 𝑆𝑆 measured by ZEM-3 and using Eqn. 1 in the experimental section.43 SnSe2:Cu has a higher 𝑛𝑛H than bulk SnSe2 across the measured temperature range, indicating easily-activated shallow donor levels. 𝜇𝜇H, on the other hand, is lower in both directions and significantly disagrees with the calculated 𝜇𝜇W (Figure 3h,i). The increase of carrier scattering at grain boundaries and the increased 𝑛𝑛𝐻𝐻  may contribute to the observed reduction in 𝜇𝜇H . Wang et al. reported a 𝜇𝜇H  which decreased as the amount of the Cu dopant increased,33 while Zhou et al. saw a large decrease in 𝜇𝜇H and a shift to ionised impurity scattering at temperatures between 300−400 K.34 Indeed, logarithmic plots of both mobilities against temperature (Figure S16) show ionised impurity scattering to be the dominant scattering mechanism across the measured temperature range, which is likely a result of the presence of the homogeneously distributed Cu+ ions. The upshot being a mobility which increases approximately proportional to 𝑇𝑇1.5, as the contribution from carrier-phonon scattering remains negligible. Comparison of Hall mobility and weighted mobility in SnSe2:Cu   To resolve the disagreement between 𝜇𝜇H and 𝜇𝜇W we considered the relationship between the two as well as the band structure of SnSe2. The relationship is sometimes quoted as being 𝜇𝜇W ∝𝜇𝜇H(𝑚𝑚∗ 𝑚𝑚𝑒𝑒⁄ )3 2⁄  where 𝑚𝑚∗ is the density of states (DOS) effective mass, and 𝑚𝑚𝑒𝑒 is the electron rest mass,43 although drift mobility is sometimes quoted in place of 𝜇𝜇H . Pisarenko plots were constructed to estimate 𝑚𝑚∗ for both SnSe2:Cu and bulk SnSe2 at 350 K (Figure S17). The obtained 𝑚𝑚∗ for bulk SnSe2 were 1.0𝑚𝑚𝑒𝑒 (IP) and 1.2𝑚𝑚𝑒𝑒 (OOP), while SnSe2:Cu had lower values of 0.4𝑚𝑚𝑒𝑒 (IP) and 0.6𝑚𝑚𝑒𝑒 (OOP). However, calculating 𝜇𝜇W using the measured 𝜇𝜇H and 𝑚𝑚∗ does not account for the significant disagreement between the two measures of mobility. The answer comes from  21 electronic band structure calculations by Ding et al.50 and others51-53 who showed that SnSe2 has two conduction band minima (CBM), at the 𝐿𝐿 and Γ points. Furthermore, Ding et al. demonstrated that at lower carrier concentrations (𝑛𝑛 ≈ 1018 cm−3), conduction occurs at the 𝐿𝐿 point CBM, but once 𝑛𝑛 reaches approximately 1019 cm−3, the Fermi energy increases allowing conduction to also occur in the Γ point CBM. This means that as n increases, there is a shift away from single band conduction and the two divergent CBM both contribute to conduction. The Hall mobility, 𝜇𝜇H, depends on the assumptions of the single parabolic band (SPB) model and is inversely proportional to n, whereas weighted mobility, 𝜇𝜇W, is relatively independent of 𝑛𝑛 and is known to be able to represent materials with more complex band structures,43 thus the two differ when the SPB model no longer applies. This is clear in the mobility of bulk SnSe2 (Figure 3h,i), where 𝜇𝜇W and 𝜇𝜇H closely agree until the temperature reaches approximately 470 K, above which the values diverge as 𝑛𝑛H increases from 3.4 to 5.3 × 1018 cm−3 at 520 K. After this point, 𝜇𝜇H continues to decrease as  𝑛𝑛𝐻𝐻 increases, due to the fact that it is inversely proportional to carrier concentration. This 𝑛𝑛H value falls in the range suggested by Ding et al. for the transition point between conduction in the 𝐿𝐿 and Γ point CBMs. In SnSe2:Cu, the value of 𝑛𝑛H throughout the measured temperature range lies above this value, suggesting that conduction should be occurring in both CBMs, and thus 𝜇𝜇W is likely to be the more reliable measure of mobility.  Returning to the question of effective mass, we can see that the calculated values of 𝑚𝑚∗ in our SnSe2:Cu samples are notably lower than those of the bulk SnSe2. Ding et al. calculated that the Γ point CBM has a significantly lower 𝑚𝑚∗ than the 𝐿𝐿 point with values of 0.3038me for (L-H) and 0.083me for (Γ-M). Given that both CBM are active in the high carrier concentration SnSe2:Cu, this could explain the observed lower effective mass of the whole sample. It is important to note that the two CBM in this case remain separate and that no band convergence is observed. This is  22 in contrast to other reports of band structure engineering in TE materials, which aim to cause band convergence as a way to increase 𝑚𝑚∗ and thus S.54,55 In fact, it is more analogous to reports of band divergence causing a reduction in 𝑚𝑚∗.28 We can therefore say that the reduced effective mass lends weight to the idea that the increase in carrier concentration causes a shift from conduction in a single band, to conduction across the two divergent CBM at the 𝐿𝐿 and Γ points.  Overall, we can conclude that, as a result of Cu doping, the increase in carrier concentration causes conduction across two divergent CBM, reducing the value of  𝑚𝑚∗ . The effect of this reduction is to increase the mobility, despite the increase in grain boundaries and carrier concentration, as can be seen in the value of 𝜇𝜇W. Additionally, the presence of Cu+ ions caused a shift to an ionised impurity scattering dominated system, in which mobility increases with temperature. The result is that the higher values of 𝑛𝑛H  and 𝜇𝜇W  contribute to a significantly increased 𝜎𝜎 in both directions, with the increase in 𝜇𝜇W particularly important as it can increase 𝜎𝜎 without having a detrimental effect on the value of S.    The mechanisms of Cu-doping in SnSe2 We have seen that even a small amount of Cu in SnSe2 can cause a significant increase in 𝑛𝑛H, as well as a change in the conduction behaviour and scattering mechanism. To investigate in more detail the role which Cu+ ions play we conducted XAFS measurements of SnSe2:Cu NSs and comparable Cu-free SnSe2 NSs, for which the characterisation by XRD TEM and XPS is shown in Figure S18. Sn and Se K-edge XANES and EXAFS measurements were conducted and, while the Se spectra showed no significant change between SnSe2 and SnSe2:Cu (Figure S19), the Sn spectra showed some interesting differences. The Sn K-edge XANES spectra (Figure 4a) and its 1st derivative (Figure 4b) show that the SnSe2:Cu sample had a 0.39 eV lower edge jump energy than the SnSe2 sample. Since it is well known that the K-edge energy of Sn shifts to higher photon  23 energies when the oxidation state increases,56,57 this result suggests that SnSe2:Cu contains a small amount of Sn2+ (possibly Sn3+, although unlikely) in addition to Sn4+. The reduction of Sn4+ ions may be an indication of the presence of Cu+ ions in interstitial sites, as a reduction in the oxidation state of Sn4+ would then be required to maintain the charge neutrality of the lattice. The k3-weighted EXAFS oscillations and their Fourier transforms are shown in Figures 4c,d. In Figure 4d the two largest peaks represent the Sn-Se and Sn-Sn paths, respectively. Fitting of the oscillations by FEFF calculation showed no significant change in the Sn-Se bond length between SnSe2 and SnSe2:Cu. However, the Sn-Sn bond length had shrunk by 0.025 Å (from 3.844 to 3.819 Å) – a significant difference within the reported capabilities of EXAFS analysis.58-60 Next, we looked at the increase in peak amplitude in SnSe2:Cu compared to SnSe2, particularly for the Sn-Se path. Since peak intensity is related to coordination number, the significant increase indicates a higher coordination number within the Sn-Se bond distance. This suggests the presence of Cu, which, as suggested by the DFT calculations of Zhou et al., is likely to intercalate between the layers of SnSe2.34 If Cu ions are intercalated between the layers, but within the Sn-Se bond distance of an Sn atom, Coulombic attraction will pull Se2- anions to the Cu+ cation in a direction perpendicular to the Sn-Se bond (Figure 5a). The neighbouring Sn atoms’ positions may then shift towards the space vacated by Se (Figure 5b). This would explain the reduction in the Sn-Sn path length, but the insignificant change in the Sn-Se bond length.   24  Figure 4. XAFS Sn K-edge spectra of SnSe2 (black) and SnSe2:Cu (red) NSs. (a) XANES, (b) 1st derivative of XANES, (c) k3-weighted EXAFS oscillations, (d) Fourier transform (FT) of (c) (dotted lines) and FEFF calculation fittings (solid lines) for SnSe2 and SnSe2:Cu. The fitting window is indicated by the blue line.  To shed light on the effect of these intercalated Cu atoms on the electronic states of SnSe2, a DFT calculation was performed using a simple 1×1×1 cell of SnSe2 and a 2×2×2 supercell of SnSe2:Cu, containing a single intercalated Cu atom. This does represent a slightly higher concentration of Cu than experimentally measured but was chosen to provide an insight into the potential effects while reducing the calculation time. The total density of states (DOS) for SnSe2 and SnSe2:Cu are shown in Figures 5c,d, respectively, and the partial DOS calculations are shown in Figure S20. We can see clearly in Figure 5d the contribution of the Cu orbitals to the valence  25 band of SnSe2, which is mainly from the outer d orbital (Figure S20). The result is a significant shift in the Fermi level towards, and indeed into, the conduction band, indicating that n-type doping has taken place. Although the concentration is slightly higher than in our actual SnSe2:Cu sample, we can clearly see that Cu+ ions can cause significant electron doping, which is directly responsible for the observed increase in 𝑛𝑛𝐻𝐻 in our SnSe2:Cu sample.  Figure 5. (a,b) Simple schematic image of Cu doping of SnSe2. (a) Cu+ ions intercalate in the VdW gap. Nearby Se atoms are attracted by Coulombic interaction in a direction perpendicular to the Sn-Se bond. (b) neighbouring Sn atom is drawn towards the vacated space, shortening the Sn-Sn distance. (c) DOS of a 1×1×1 cell of SnSe2. (d) DOS of a 2×2×2 supercell of SnSe2 containing one intercalated Cu atom. Black, grey, green, and blue represent total DOS and contributions from Sn, Se, and Cu, respectively. Fermi level (grey dashed line) is set at 0 eV.   26 CONCLUSIONS By synthesising and sintering Cu doped SnSe2 NSs we have produced SnSe2:Cu pellets which have shown significantly higher TE properties than bulk, single crystal SnSe2. This is down to both a reduced 𝜅𝜅𝑙𝑙, due to phonon scattering at grain boundaries, and an increased 𝜎𝜎, due to an increased 𝑛𝑛H and 𝜇𝜇w.  A maximum ZT of 0.6 at 667 K in the OOP direction demonstrates the potential of small amounts of Cu doping for practical, sustainable n-type TE materials, such as SnSe2. The result also compares very favourably with other reports of Cu doping of SnSe2, with only one report which included Br as a co-dopant and further heat treatment showing a higher TE performance. We then thoroughly investigated the mechanism of Cu doping and found that the Cu+ ions are likely to have intercalated between the VdW layers, within the Sn-Se bond length. The presence of Cu+ shifts the Fermi energy into the conduction band, leading to a significantly increased carrier concentration across the measured temperature range. This causes a change in the conduction mechanism from a single CBM to two divergent CBMs, thus reducing 𝑚𝑚∗. When combined with a shift to ionised impurity scattering under which mobility increases with temperature, the result is a significantly higher 𝜎𝜎 and PF across the measured temperature range. With the increased ZT value and mechanistic insight, we hope that this work will provide an important step towards practical, more sustainable TE materials.  ASSOCIATED CONTENT The following files are available free of charge. Experimental and DFT calculation details; schematic diagrams of the preparation process of the pellet for STEM and 3DAP analysis; pellet preparation for TE properties measurement; IP 𝜅𝜅 measurement using a thin pellet and COMSOL simulation; XRD and TEM analysis of the large-scale SnSe2 NSs; STEM-EDS analysis of the SnSe2:Cu NSs; XPS spectra of Cu 2p and Cu  27 L3M4,5M4,5 Auger regions of the SnSe2:Cu NSs; SEM-EDS and XPS composition results and XPS spectra of Cu 2p and Cu L3M4,5M4,5 Auger regions of Sn (IV) acetate; TGA analysis of the SnSe2:Cu NSs before and after the ligand exchange; STEM-EDS and SEM-EDS maps of the SnSe2:Cu pellet; log-log plots of carrier mobility against temperature; Pisarenko plots; XRD, TEM, and XPS results of Cu-free SnSe2 NSs; XAFS Se K-edge spectra; partial DOS calculations (PDF) AUTHOR INFORMATION Corresponding Author *Shinya Maenosono − School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan; E-mail: shinya@jaist.ac.jp Authors Simon Moore − School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan Mari Takahashi − School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan Philipp Sauerschnig − Global Zero Emission Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8568, Japan Keiji Kobayashi − School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan Koichi Higashimine − Center for Nano Materials and Technology , Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan Masanobu Miyata − School of Materials Science, Japan Advanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan Takahiro Baba − Research Center for Materials Nanoarchitectronics, National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan  28 Jun Uzuhashi − Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan Michihiro Ohta − Global Zero Emission Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8568, Japan Takao Mori − Research Center for Materials Nanoarchitectronics, National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan Tadakatsu Ohkubo − Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan Author Contributions The work was initiated by S.Ma. (Shinya Maenosono); S.Mo. (Simon Moore), M.T. and S.Ma. devised the method; S.Mo. constructed the experimental setup and performed all the experiments under the supervision of M.T. and S.Ma.; K.K. assisted in the experimental setup; K.H. conducted STEM observation of SnSe2 nanosheets; P.S. measured thermoelectric properties of pellets under the supervision of M.O.; J.U. conducted STEM observation and 3DAP analysis of pellets under the supervision of T.O.; T.B. estimated the in-plane thermal conductivity of the SnSe2:Cu pellet using the COMSOL simulation under the supervision of T.M.; S.Mo., M.T. and S.Ma. carried out XAFS measurements; M.M. performed the DFT calculation; S.Mo., M.T. and S.Ma. drafted the manuscript. All authors reviewed the manuscript and have given approval to the final version of the manuscript. Funding Sources This work was supported by JSPS KAKENHI (Grant Number 19H02440) and the Thermal and Electric Energy Technology Foundation. This work has also been partly supported by The  29 Research Grant against Global Warming of the Ichimura Foundation for New Technology. T.M. and T.B. thank JST Mirai JPMJMI19A1. Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS This work was supported by JSPS KAKENHI under Grant No. 19H02440. This work was partially supported by the Thermal & Electric Energy Technology Foundation. T.M. and T.B. thank JST Mirai JPMJMI19A1. S. Mo thanks Fukue Kotegawa for her help with the XAFS measurements. REFERENCES [1] Fitriani; Ovik, R.; Long, B. D.; Barma, M. C.; Riaz, M.; Sabri, M. F. M.; Said, S. M.; Saidur, R. A Review on Nanostructures of High-Temperature Thermoelectric Materials for Waste Heat Recovery. Renew. Sustain. Energy Rev. 2016, 64, 635−659.  [2] Champier, D. Thermoelectric Generators: A Review of Applications. Energy Convers. Manag. 2017, 140, 167-181. [3] Zoui, M. A.; Bentouba, S.; Stocholm, J. G.; Bourouis, M. A Review on Thermoelectric Generators: Progress and Applications. Energies 2020, 13, 3606. [4] Palaporn, D.; Tanusilp, S. A.; Sun, Y.; Pinitsoontorn, S.; Kurosaki, K. Thermoelectric Materials for Space Explorations. Mater. Adv. 2024, DOI: 10.1039/d4ma00309h. [5] Mao, J.; Chen, G.; Ren, Z. Thermoelectric Cooling Materials. Nat. Mater. 2017, 20, 454-461.  30 [6] Shi, X. L.; Zou, J.; Chen, Z. G. Advanced Thermoelectric Design: From Materials and Structures to Devices. Chem. Rev. 2020, 120, 7399−7515. [7] Yan, Q.; Kanatzidis, M. G. High-Performance Thermoelectrics and Challenges for Practical Devices. Nat. Mater. 2022, 21, 503−513.  [8] Chen, Y. X.; Zhang, J. Z.; Nisar, M.; Abbas, A.; Li, F.; Liang, G. X.; Fan, P.; Zheng, Z. H. Realizing High Thermoelectric Performance in n-type Bi2Te3 Based Thin Films via Post-Selenization Diffusion. J. Materiomics 2023, 9, 618-625. [9] Lee, H.; Kim, T.; Son, S. C.; Kim, J.; Kim, D.; Lee, J.; Chung, I. Unique Microstructures and High Thermoelectric Performance in n-type Bi2Te2.7Se0.3 by the Dual Incorporation of Cu and Y. Mater. Today Phys. 2023, 31, 100986. [10] Jia, B.; Huang, Y.; Wang, Y.; Zhou, Y.; Zhao, X.; Ning, S.; Xu, X.; Lin, P.; Chen, Z.; Jiang, B.; He, J. Realizing High Thermoelectric Performance in Non-Nanostructured n-type PbTe. Energy Environ. Sci. 2022, 15, 1920-1929. [11] Deng, R.; Su, X.; Zheng, Z.; Liu, W.; Yan, Y.; Zhang, Q.; Dravid, V. P.; Uher, C.; Kanatzidis, M. G.; Tang, X. Thermal conductivity in Bi0.5Sb1.5Te3+x and the Role of Dense Dislocation Arrays at Grain Boundaries. Sci. Adv. 2018, 4:eaar5606. [12] Jai, B.; Wu, D.; Xie, L.; Wang, W.; Yu, T.; Li, S.; Wang, Y.; Xu, Y.; Jiang, B.; Chen, Z.; Weng, Y.; He, J. Pseudo-Nanostructure and Trapped-Hole Release Induce High Thermoelectric Performance in PbTe. Science 2024, 384, 81-86.  31 [13] Caballero-Calero, O.; Ares, J. R.; Martín-González, M. Environmentally Friendly Thermoelectric Materials: High Performance from Inorganic Components with Low Toxicity and Abundance in the Earth. Adv. Sustain. Syst. 2021, 5, 210095. [14] Shen, Y.; Li, C.; Huang, R.; Tian, R.; Ye, Y.; Pan, L.; Koumoto, K.; Zhang, R.; Wan, C.; Wang, Y. Eco-Friendly P-Type Cu2SnS3 Thermoelectric Material: Crystal Structure and Transport Properties. Sci. Rep. 2016, 6, 32501. [15] Wu, H.; Wang, G.; Peng, K.; Chi, H.; Zhang, B.; Chen, Y.; Li, C.; Yan, Y.; Guo, L.; Uher, C.; Zhou, X.; Han, X. Sodium-Doped Tin Sulfide Single Crystal: A Nontoxic Earth-Abundant Material with High Thermoelectric Performance. Adv. Energy Mater. 2018, 8, 1800087. [16] Shi, Y.; Sturm, C.; Kleinke, H. Chalcogenides as Thermoelectric Materials. J. Solid State Chem. 2019, 270, 273−279. [17] Li, F.; Zheng, Z.; Li, Y.; Wang, W.; Li, J. F.; Li, B.; Zhong, A.; Luo, J.; Fan, P. Ag-Doped SnSe2 as A Promising Mid-Temperature Thermoelectric Material. J. Mater. Sci. 2017, 52, 10506−10515. [18] Duong, A. T.; Nguyen, V. Q.; Duvjir, G.; Duong, V. T.; Kwon, S.; Song, J. Y.; Lee, J. K.; Lee, J. E.; Park, S. D.; Min, T.; Lee, J.; Kim, J.; Cho, S. Achieving ZT=2.2 with Bi-Doped N-Type SnSe Single Crystals. Nat. Commun. 2016, 7, 13713. [19] Hong, M.; Chen, Z. G.; Yang, L.; Chasapis, T. C.; Kang, S. D.; Zou, Y.; Auchterlonie, G. J.; Kanatzidis, M. G.; Snyder, G. J.; Zou, J. Enhancing the Thermoelectric Performance of SnSe1−xTex Nanoplates Through Band Engineering. J. Mater. Chem. A 2017, 5, 10713.  32 [20] Pei, Y.; Wang, H.; Snyder, G. J. Band Engineering of Thermoelectric Materials. Adv. Mater. 2012, 24, 6125−6135. [21] Biswas, K.; He, J.; Blu, I. D.; Wu, C. I.; Hogan, T. P.; Seidman, D. N.; Dravid, V. P.; Kanatzidis, M. G. High-Performance, Bulk Thermoelectrics with All-Scale Hierarchical Architectures. Nature 2012, 489, 414−418. [22] Ortega, S.; Ibanez, M.; Liu, Y.; Zhang, Y.; Kovalenko, M. V.; Cadavid, D.; Cabot, A. Bottom-Up Engineering of Thermoelectric Nanomaterials and Devices from Solution-Processed Nanoparticle Building Blocks. Chem. Soc. Rev. 2017, 46, 3510−3528. [23] Zhao, L. D.; Lo, S. H.; Zhang, Y.; Sun, H.; Tan, G.; Uher, G.; Wolverton, C.; Dravid, V. P.; Kanatzidis, M. G. Ultralow Thermal Conductivity and High Thermoelectric Figure of Merit in SnSe Crystals. Nature 2014, 508, 373−377. [24] Kim, T.; Lee, H.; Chung, I. SnSe: The Rise of the Ultrahigh Thermoelectric Performance Material. Bull. Korean Chem. Soc. 2024, 45 (3), 186-199. [25] Chandra, S.; Biswas, K. Realization of High Thermoelectric Figure of Merit in Solution Synthesized 2D SnSe Nanoplates via Ge Alloying. J. Am. Chem. Soc. 2019, 141, 6141-6145. [26] Zhou, C.; Lee, Y. K.; Yu, Y.; Byun, S.; Luo, Z. Z.; Lee, H.; Ge, B.; Lee, Y. L.; Chen, X.; Lee, J. Y.; Cojocaru-Miredin, O.; Chang, H.; Im, J.; Cho, S. P. Wuttig, M.; Dravid, P. V.; Kanatzidis, M. G.; Chung, I. Polycrystalline SnSe with a Thermoelectric Figure of Merit Greater than the Single Crystal. Nat. Mater. 2021, 20, 1387-1384.  33 [27] Chandra, S.; Bhat, U.; Dutta, P.; Bhardwaj, A.; Datta, R.; Biswas, K. Modular Nanostructures Facilitate Low Thermal Conductivity and Ultra-High Thermoelectric Performance in n-type SnSe. Adv. Mater. 2022, 34, 2203725. [28] Chang, C.; Wu, M.; He, D.; Pei, Y.; Wu, C. F.; Wu, X.; Yu, H.; Zhu, F.; Wang, K.; Chen, Y.; Huang, L.; Li, J. F.; He, J.; Zhao, L D. 3D Charge and 2D Phonon Transports Leading to High Out-of-plane ZT in n-type SnSe Crystals. Science 2018, 360, 778-783. [29] Byun, S.; Ge, B.; Song, H.; Cho, S. P.; Hong, M. S.; Im, J.; In C. Simultaneously Engineering Electronic and Phonon Band Structures for High-Performance n-type Polycrystalline SnSe. Joule 2024, 8 (5), 1520-1538. [30] Pham, A. T.; Vu, T. H.; Cheng, C.; Trinh, T. L.; Lee, J. E.; Ryu, H.; Hwang, C.; Mo, S. K.; Kim, J.; Zhao, L. D.; Duong, A. T.; Cho, S. High-Quality SnSe2 Single Crystals: Electronic and Thermoelectric Properties. ACS Appl. Energy Mater. 2020, 3, 10787−10792. [31] Xu, P.; Fu, T.; Xin, J.; Liu, Y.; Ying, P.; Zhao, X.; Pan, H.; Zhu, T. Anisotropic Thermoelectric Properties of Layered Compound SnSe2. Sci. Bull. 2017, 62, 1663−1668. [32] Pham, A. T.; Vu, T. H.; Nguyen, Q. V.; Vu, M. T.; Park, J. H.; Park, S. D.; Cho, S. Br-Doped N-Type SnSe2: Single-Crystal Growth and Thermoelectric Properties. ACS Appl. Energy Mater. 2012, 4, 2908−2913. [33] Wang, J.; Jian, X.; Lou, S.; Li, G.; Zhou, S. Cu-Embedded SnSe2 with A High Figure of Merit at Ecofriendly Temperature. ACS Omega 2020, 5, 12409−12414. [34] Zhou, C.; Yu, Y.; Zhang, X.; Cheng, Y.; Xu, J.; Lee, Y. K.; Yoo, B.; Cojocaru-Miredin, O.; Liu, G.; Cho, S. P.; Wuttig, M.; Hyeon, T.; Chung, I. Cu Intercalation and Br Doping to  34 Thermoelectric SnSe2 Lead to Ultrahigh Electron Mobility and Temperature-Independent Power Factor. Adv. Funct. Mater. 2020, 30, 1908405. [35] Nisar, M.; Qin, W.; Zhang, J.; Zheng, Z.; Li, F.; Liang, G.; Fan, P.; Chen, Y. X. Synergistic Optimization of Thermoelectric Performance in SnSe2 through Co-doping: Anionic Vacancy Formation and Band Engineering. Mater. Lab 2023, 2, 230023. [36]  Wei, Y.; Li, J.; Zhang, D.; Zhang, B.; Zhou, Z.; Han, G.; Wang, G.; Prestipino, C.; Lemoine, P.; Guilmeau, E.; Lu, X.; Zhou, X. Phase-dependent Microstructure Modification Leads to High Thermoelectric Performance in n-type Layered SnSe2 Acta Mater 2024, 263, 119504. [37] Singh, N. K.; Bathula, S.; Gahtori, B.; Tyagi, K.; Haranath, D.; Dhar, A. The Effect of Doping on Thermoelectric Performance of P-Type SnSe: Promising Thermoelectric Material. J. Alloys Compd. 2016, 668, 152−158. [38] Zheng, Z. H.; Fan, P.; Luo, J. T.; Liang, G. X.; Liu, P. J.; Zhang, D. P. Enhanced Thermoelectric Properties of Cu Doped ZnSb Based Thin Films. J. Alloys Compd. 2016, 668, 8−12. [39] An, J.; Han, M. K.; Kim, S. J. Synthesis of Heavily Cu-doped Bi2Te3 Nanoparticles and Their Thermoelectric Properties. J. Solid State Chem. 2019, 270, 407−412. [40] Liu, W. S.; Zhang, Q.; Lan, Y.; Chen, S.; Yan, X.; Zhang, Q.; Wang, H.; Wang, D.; Chen, G.; Ren, Z. Thermoelectric Property Studies on Cu-doped N-Type CuxBi2Te2.7Se0.3 Nanocomposites. Adv. Energy Mater. 2011, 1, 577−587.  35 [41] Moore, S.; Kobayashi, K.; Takahashi, M.; Higashimine, K.; Maenosono, S. Facile Phase-Selectable Chemical Synthesis of Tin Selenide Nanosheets Towards Nanostructured Materials for Energy-Related Applications. J. Phys. Chem. C 2023, 127, 12394−12403. [42] Lee, S. W.; Park, O.; Kim. S. I. Enhanced Thermoelectric Transport Properties of N-Type SnSe2 Polycrystalline Alloys by Te Doping. Int. J. Energy Res. 2023, 2900242. [43] Snyder, G. J.; Snyder, A. H.; Wood, M.; Gurunathan, R.; Snyder, B. H.; Niu, C. Weighted Mobility. Adv. Mater. 2020, 32, 2001537. [44] Ozaki, T. Variationally Optimized Atomic Orbitals for Large-Scale Electronic Structures. Phys. Rev. B 2003, 67, 155108. [45] Ozaki, T.; Kino, H. Efficient Projector Expansion for the ab initio LCAO Method. Phys. Rev. B 2005, 72, 045121. [46] Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865-3868. [47] Yazdani, S.; Pettes, M. T. Nanoscale Self-Assembly of Thermoelectric Materials: A Review of Chemistry-Based Approaches. Nanotechnology 2018, 29, 432001.  [48] Jaih, A.; Ong, S. P.; Hautier, G.; Chen, W.; Richards, W. D.; Dacek, S.; Cholia, S.; Gunter, D.; Skinner, D.; Ceder, G Commentary: The Materials Project: A Materials Genome Approach to Accelerating Materials Innovation. APL Mater. 2013, 1, 011002.    36 [49] Uzuhashi, J.; Chen, J.; Kumar, A.; Yi, W.; Ohkubo, T.; Tanaka, R.; Takashima, S.; Edo, M.; Sierakowski, K.; Bockowski, M.; Sakurai, H.; Kachi, T.; Sekiguchi, T.; Hono, K. Atomic-Scale Investigation of Implanted Mg in GaN Through Ultra-High-Pressure Annealing. J. Appl. Phys. 2022, 131, 185701. [50] Ding, Y.; Xiao, B.; Tang, G.; Hong, J. Transport Properties and High Thermopower of SnSe2: A Full Ab-Initio Investigation. J. Phys. Chem. C 2017, 121, 225−236. [51] Fong, C. Y.; Cohen, M. L. Electronic Energy-Band Structure of SnS2 and SnSe2. Phys. Rev. B 1972, 5, 3095−3101. [52] Xia, C.; An, J.; Wei, S.; Jia, Y.; Zhang, Q. Electronic Structures and Optical Properties of SnSe2(1-x)O2x Alloys. Comput. Mater. Sci. 2014, 95, 712-717. [53] Lochocki, E. B.; Vishwanath, S.; Liu, Z.; Dobrowolska, M.; Furdyna, J.; Xing, H. G.; Shen, K. M. Electronic Structure of SnSe2 films grown by molecular beam epitaxy. Appl. Phys. Lett. 2019, 114, 091602. [54] Shenoy, U. S.; Goutham, K. D.; Bhat, D. K. Resonance States and Hyperconvergence Induced by Tungsten Doping in SnTe: Multiband Transport Leading to a Propitious Thermoelectric Material. J. Alloys Compd. 2022, 905, 164146. [55] Zhao, L. D.; Wu, H. J.; Hao, S. Q.; Wu, C. I.; Zhou, X. Y.; Biswas, K.; He, J. Q.; Hogan, T. P.; Uher, C.; Wolverton, C.; Dravid, V. P.; Kanatzidis, M. G. All-scale Hierarchical Thermoelectrics: MgTe in PbTe Facilitates Valence Band Convergence and Suppresses Bipolar Thermal Transport for High Performance. Energy Environ. Sci. 2013, 6, 3346.  37 [56] Lützenkirchen-Hecht, D.; Scotti, N.; Jacobs, H.; Frahm, R. XAFS Investigations of Tin Nitrides. J. Synchotron Rad. 2001, 8, 698−700. [57] McKeown, D. A.; Buechele, A. C.; Gan, H.; Pegg, I. L. Tin Valence and Local Environments in Silicate Glasses as Determined from X-ray Absorption Spectroscopy. J. Non-Cryst. Solids 2008, 354, 3142−3151. [58] Gainar, A.; Stevens, J. S.; Jaye, C.; Fischer, D. A.; Schroeder, S. L. M. NEXAFS Sensitivity to Bond Lengths in Complex Molecular Materials: A Study of Crystalline Saccharides. J. Phys, Chem. B 2015, 119, 14373−14381. [59] Hennig, G.; Reich, T.; Funke, H.; Rossberg, A.; Rutsch, M.; Bernhard, G. EXAFS as a Tool for Bond-Length Determination in The Environment of Heavy Atoms. J. Synchotron Rad. 2001, 8, 695−697. [60] Newville, M.; Kas, J. J.; Rehr, J. J. Improvements in Modeling EXAFS with Many-Pole Self-Energy and FEFF8.5. J. Phys.: Conf. Ser. 2009, 190, 012023.