Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2023 Jan 19;15(4):5071–5085. doi: 10.1021/acsami.2c16587

High Power Factor Nb-Doped TiO2 Thermoelectric Thick Films: Toward Atomic Scale Defect Engineering of Crystallographic Shear Structures

Xiaodong Liu †, Demie Kepaptsoglou ‡,§, Ewa Jakubczyk ∥, Jincheng Yu †, Andrew Thomas †,⊥,#, Bing Wang †, Feridoon Azough †, Zhaohe Gao †, Xiangli Zhong †,⊥, Robert Dorey ∥, Quentin M Ramasse ‡,, Robert Freer †,*
PMCID: PMC9906629  PMID: 36656149

Abstract

graphic file with name am2c16587_0012.jpg

Donor-doped TiO2-based materials are promising thermoelectrics (TEs) due to their low cost and high stability at elevated temperatures. Herein, high-performance Nb-doped TiO2 thick films are fabricated by facile and scalable screen-printing techniques. Enhanced TE performance has been achieved by forming high-density crystallographic shear (CS) structures. All films exhibit the same matrix rutile structure but contain different nano-sized defect structures. Typically, in films with low Nb content, high concentrations of oxygen-deficient {121} CS planes are formed, while in films with high Nb content, a high density of twin boundaries are found. Through the use of strongly reducing atmospheres, a novel Al-segregated {210} CS structure is formed in films with higher Nb content. By advanced aberration-corrected scanning transmission electron microscopy techniques, we reveal the nature of the {210} CS structure at the nano-scale. These CS structures contain abundant oxygen vacancies and are believed to enable energy-filtering effects, leading to simultaneous enhancement of both the electrical conductivity and Seebeck coefficients. The optimized films exhibit a maximum power factor of 4.3 × 10–4 W m–1 K–2 at 673 K, the highest value for TiO2-based TE films at elevated temperatures. Our modulation strategy based on microstructure modification provides a novel route for atomic-level defect engineering which should guide the development of other TE materials.

Keywords: thermoelectric thick film, Nb doping, crystallographic shear structure, oxygen deficiency, energy filtering

Introduction

Thermoelectric (TE) energy conversion is a solid-state, quiet, and eco-friendly technology, which can directly convert waste heat into electricity.1,2 The TE conversion efficiency is closely related to the dimensionless TE figure of merit (ZT), defined as ZT = S2σT/κ, where S is the Seebeck coefficient, T is the absolute temperature, σ and κ are electrical and thermal conductivities, respectively, and S2σ is the power factor (PF). Consequently, to optimize the TE performance and achieve high ZT values, a higher PF and/or lower thermal conductivity are required. While a number of commercial, high-performance TE materials are available, most are limited to modest temperature ranges and contain heavy, toxic, rare, and expensive elements, which significantly hinder the wide-scale exploitation of the green TE technology.

Oxide ceramics are attracting increasing interest for TE applications as they are predominantly based on earth-abundant, environmentally benign elements and are stable over a wide range of temperatures. The families of materials include, but are not limited to, p-type cobaltites (NaCo2O4,3 Ca3Co4O9,4 CaxCoO2,5 Ca3Co2O6,6 Co3O4,7 and Bi2Sr2Co2Oy8), n-type strontium titanate (SrTiO39), titania (TiO1.1,10 TiO2–x,11 and TiO212), zinc oxide (ZnO13), CaMnO3,14 Magnéli phases (TinO2n–1,15 Nb12O29,16 and WO3–x17), indium oxide (In2O318), and tungsten bronze-structured oxides (Ba5.19Nd8.54Ti18O5419 and Ba6Ti2Nb8O3020). Among these, TiO2-based oxides offer excellent thermal stability but so far modest TE performance. However, a variety of strategies are availed for improving their TE transport properties. For instance, donor-doping of TiO2 (by Nb,11,21 Al,22 Ta,23 etc.) enhances electrical conductivity and carrier concentration and reduces thermal conductivity. Similarly, oxygen-deficient Magnéli-structured TiO2–x shows much promise as the oxygen deficiency produces a high density of carriers.15,24−26 In addition, nano-scale interfaces [including twin boundaries (TBs)] are able to enhance energy-filtering effects,27−29 thereby increasing Seebeck coefficients.

TE devices are usually based on arrays of n-type and p-type elements connected electrically in series and thermally in parallel. With the growing demand for automatic fabrication processes for TE devices, film-based structures and technologies are becoming increasingly attractive due to the great potential in developing IoT sensors and devices.30−33 While film-based TE modules have limitations in terms of power generated, they offer benefits in terms of lower cost, a reduction in the quantity of starting materials, and are readily scalable through a variety of fabrication technologies, including screen-printing,34 deposition,35 plasma spray,36,37 magnetron sputtering,38 and inkjet printing.39 Among these methods, screen-printing is particularly attractive because the facile fabrication route is readily amenable to automatic fabrication. Previous work on oxide-based films has focused on pulsed-laser deposition, hydrothermal, reactive magnetron sputtering, and spin coating methods.36−38,40−45 For TiO2-based films, although strategies including donor-doping and the development of oxygen-deficient Magnéli structures have been utilized for films, their TE performance at medium to high temperatures is still inferior to that of bulk TiO2-based TE materials.

Guided by currently employed optimization strategies in combination with donor-doping and defect engineering, we present a facile, scalable, and eco-friendly screen-printing method to fabricate high-performance n-type TiO2-based TE thick films. Through advanced electron microscopy techniques, we provide direct evidence of the modified oxygen-deficient defect structures and detailed evolution of processes at the atomic level. Through combined enhancement of both electrical conductivity and the Seebeck coefficients, a peak PF of 4.3 × 10–4 W m–1 K–2 is achieved, the highest value reported for TiO2-based films at elevated temperatures.

Materials and Methods

Materials

Stoichiometric mixtures of TiO2 powder (Sigma-Aldrich, >99.99%) and Nb2O5 powder (Sigma Aldrich, >99.99%) were weighed and mixed according to the nominal chemical formulation: (1 – x) TiO2 – xNb2O5 (x = 0.01 and 0.06). The mixed powders were wet-milled with propan-2-ol (Fisher Scientific) and YSZ balls (Pi-Kem) for 24 h. After drying at 353 K for 24 h, the x = 0.01 powders were transferred into an alumina crucible with a lid and calcined at 1473 K for 4 h in air. In contrast, the x = 0.06 powders were placed in an open alumina crucible and calcined at 1473 K for 12 h in the Ar/H2 atmosphere. Both sets of calcined powders were then wet-milled and dried again, using the same procedures as described above.

The thick films were prepared by screen-printing. The ink was prepared from 55.7 wt % of the as-prepared TiO2–Nb2O5 powder mixed with 40.8 wt % terpinol (Sigma-Aldrich), 1.6 wt % polyvinyl butyral (Mowital), and 2.0 wt % ethyl cellulose (Sigma-Aldrich) using a ball-mill to form an ink with a suitable viscosity and rheology for the multilayer screen-printing process. Screen-printing was performed using a DEK 245 screen-printer with a mesh screen of 25 cm2 and a squeegee speed of 4 cm/s. Before use, the alumina substrates (96% Rubalit 710, CeramTec) were ultrasonically cleaned with acetone and then deionized water. Each film consisted of five printed layers, with each layer being dried at 403 K for 1 h to remove the solvents, prior to the deposition of the next layer.

The as-printed films were sintered without pressure at 1673 K, using two types of sintering conditions. In the first arrangement, the film was placed on top of a layer of zirconia balls inside an alumina crucible and sintered in an open environment with 95% Ar–5% H2 gas flow (Figure S1a). In the second arrangement, the alumina crucible was first filled with powder [x = 0.06 powder + 5 wt % Graphene nanoplatelets (XG science)] to provide a strongly reducing, oxygen scavenging, environment,46 into which a cube-shaped hole was dug to accommodate a single layer of zirconia balls. The film was placed in direct contact with the zirconia balls and sintered in a closed environment with 95% Ar–5% H2 gas flow (Figure S1b). Detailed information about the samples and sintering conditions is presented in Table S1.

Characterizations

The structure of each film was examined using a Philips X’Pert X-ray diffractometer (XRD) with Cu Kα radiation (λ = 1.540598 Å); the collection step size was 0.033° with 2 theta range from 10 to 100°. Phase identification and refinement were performed using TOPAS software.47 The microstructure and chemical compositions were characterized using a TESCAN MIRA LC FEG scanning electron microscope equipped with an Oxford Instrument SDD energy-dispersive X-ray (EDX) detector. Electron backscattered diffraction (EBSD) data were acquired using an Oxford Instruments Symmetry EBSD detector.

For TEM examination, samples were prepared by focus ion beam (FIB) and standard crushing methods. FIB preparation employed a FEI Helios 660 Dual Beam FIB [equipped with both Ga ion source FIB column and a FEG scanning electron microscopy (SEM) column] and a FEI Helios Plasma Cryo-FIB (equipped with a Xe ion source FIB column and a FEG SEM column). In order to reduce sample thickness for atomic-level chemical investigations, they were further milled using a Fischione Model 1040 NanoMill system. For the standard crushing method, samples were crushed in an agate mortar, dispersed in propan-2-ol, and dropped onto a holey carbon film. Phase structure analysis was carried out by selected area electron diffraction (SAED), using an FEI Tecnai 20 analytical LaB6 TEM, operating at 200 kV. Nano-structural characterization was carried out using a Thermo Fisher Talos F200X FEG TEM/STEM equipped with a Super-X EDXS detector at 200 kV. Atomically resolved scanning transmission electron microscopy (STEM) images were obtained using a FEI Titan G2 aberration-corrected ChemiSTEM (X-FEG and SuperX EDXS) (with convergence semi-angle of 21 mrad) equipped with HAADF(high angle annular dark field)/DF2/DF4/BF(bright field) STEM detectors at 200 kV. The atomically resolved, electron energy loss spectroscopy (EELS) characterization was conducted by a Nion UltraSTEM 100 aberration-corrected dedicated scanning transmission electron microscope, equipped with HAADF (semi-angle range of 86–190 mrad)/MAADF (medium angle annular dark field) (semi-angle range of 40–86 mrad)/BF detectors and a Gatan Enfina EEL spectrometer, operating at 100 kV. The convergence semi-angle and the EELS collection semi-angle were 31 and 36 mrad, respectively. Processing of the TEM data was undertaken using Gatan Microscopy Suite. The principal component analysis of the EELS and EDXS data was performed using temDM MSA Digital Micrograph plugin.48 Dynamic ball- and stick-models and simulated SAED patterns were generated using CrystalMaker software.49 The geometric phase analysis (GPA) of the atomic strain maps was conducted using Strain++ software.50

The element valence states were investigated by X-ray photoelectron spectroscopy (XPS), using a high-throughput XPS, equipped with an ESCA2SR spectrometer (Scienta Omicron) with monochromatic Al Kα radiation (Esource = 1486.69 eV). The survey scans were collected from 0 to 1200 eV (binding energy). High-resolution spectra for specific elements, Ti (2p), Nb (3d), O (1s), and C (1s), were collected separately. The C (1s) peak (284.8 eV) was used to calibrate the peak positions of the survey scans and high-resolution scan spectra. Processing of the XPS data employed CasaXPS software.51

The electrical transport properties (electrical conductivity and Seebeck coefficients) were determined using an Ulvac-Riko ZEM-3 operating in a low-pressure He atmosphere from 323 to 673 K, with a step size of 50 K. The uncertainties of the electrical conductivity, Seebeck coefficients, and PF data are 3, 5, and 8%, respectively.

Results and Discussion

Two donor-doping contents were selected in this investigation, expressed by the chemical formula: (1 – x) TiO2 – xNb2O5 (x = 0.01 and x = 0.06). The as-printed x = 0.01 (sample code: as-printed TF1) and x = 0.06 (sample code: as-printed TF6) films were sintered under two conditions (Figure S1): 95% Ar–5% H2 gas flow (sample code: as-sintered TF1 and TF6) reducing atmosphere, and oxygen scavenging, strongly reducing atmosphere (sample code: as-sintered TF6G). For clarity, these sample codes (Table S1) will be used in the following sections.

As-Printed Thick Films

The as-printed thick films were prepared successfully with no visible cracks (Figure S2). Both TF1 and TF6 films are predominantly rutile structures (Figure 1a and Table S2) and exhibit same porous structure (Figure S3), while TF6 films show slightly higher level of reduction (Figures S4, S5 and Table S3) (further details are given in the Supporting Information).

Figure 1.

Figure 1

Open in a new tab

(a) XRD patterns for the as-printed thick films; (b) XRD patterns for the as-sintered thick films, the blue arrows indicate the additional phases present in the films.

As-Sintered Thick Films

X-ray diffraction patterns confirmed that all the sintered films are predominantly rutile structured (tetragonal, space group P42/mnm) (Figure 1b). TF1 and TF6 films reveal single phase rutile, while TF6G films exhibit several new, high-intensity peaks (marked by blue arrows) between 25° and 30° (close to (110) peak). According to the TiO2–Al2O3 phase diagram, TiO2 is able to react with Al2O3 to form Al2TiO5 above 1573 K.52 Notably, these new peaks cannot be indexed using the standard database, but their position suggests a structure somewhat similar to that Al2TiO5.53 Their origin will be addressed when discussing the SEM data (Scanning Electron Microscopy Section). The lattice parameters extracted by Rietveld refinement are presented in Table 1; it is noted that a, b, and c increase sequentially from TF1 to TF6G. The increase from TF1 to TF6 results mainly from the replacement of Ti ions by the larger Nb ions (ionic radii: RTi4+ = 0.605 Å and RTi3+ = 0.67 Å, RNb5+ = 0.64 Å and RNb4+ = 0.68 Å).54 However, the increase from TF6 to TF6G is dominated by the stronger reducing atmosphere (for TF6G), generating more low state Nb and Ti ions with higher ionic radii.

Table 1. Lattice Parameters for the Rutile Phase in Sintered Thick Films.

sample a, b (Å) c (Å)
TF1 4.6199(2) 2.9732(1)
TF6 4.6206(1) 2.9744(1)
TF6G 4.6236(2) 2.9749(2)
Open in a new tab

Scanning Electron Microscopy

All the sintered films were flat, smooth, and crack-free, typically 20–60 μm thick (Figure S6). Porosity [determined from EBSD band contrast (BC) images] increased slightly with Nb content (TF1 ∼ 12%; TF6 ∼ 24%; TF6G ∼ 17%; Table S1). High density, parallel, step-like, facetted growth bands are visible in high-resolution images (Figure S7). The microstructures of the interfaces between the films (upper, lighter-colored regions) and the substrates (dark gray regions) for the different samples can be seen in Figure 2. However, it is noted that there are also light gray regions at the interface between the film and the substrate, and indeed some inside the films, as a result of reactions between the film and the substrate. It is clear that the TF1 and TF6G films are much denser than the TF6 films. In TF1, there is a concentration of pores at the interface layer (highlighted by blue circles) (Figure 2a); this is also the case in TF6 (highlighted by green circles) but less evident due to higher overall levels of porosity (Figure 2b). In contrast, there is less porosity, and closed pores can be observed in TF6G films (Figure 2c). SEM EDS line scans indicate that, in comparison to the majority of the films, the mid-gray areas are enriched in Al but depleted in Ti. Indeed, the Al and Ti content changes with distance from the substrate across the interphase region (highlighted by the dark gray background in EDX line scans in Figure 2). Therefore, the light gray regions appear to consist of complex, intermediate phases. In addition, high density, stripe-like, sub-grain features occur inside the TF6G film (marked by yellow dash lines). High-magnification and high-contrast BSE images clearly reveal that the stripe-like features (dark gray in color) are 2–5 μm wide and interlaced with each other (Figure S8). Between these wide primary features are secondary, stripe-like features, approximately 50–100 nm wide, at intervals up to 200 nm. As these features are widely distributed in the film matrix, some new peaks in TF6G XRD patterns should originate from the presence of these features. The primary and secondary features will be examined in more detail in the Transmission Electron Microscopy Section.

Figure 2.

Figure 2

Open in a new tab

SEM backscattered electron micrographs and the corresponding EDX line scans of cross-sections of the thick films: (a) TF1; (b) TF6; and (c) TF6G.

To provide information about internal stress and orientation distributions in the thick film grains, EBSD analysis was performed on the interface layer between the films and the substrates (Figure 3). In the BC and phase overlay map (denoted as Phase) for TF6 films, the red regions are rutile, the green regions are Al2O3 substrate, and the light gray regions are the interphase TiO2–Al2O3 (Figure 3a). As the intermediate phases could not be matched to standard databases, they cannot be indexed using existing crystal structure models. The corresponding Kernel Average Misorientation (KAM) map (Figure 3b) reveals low average misorientation values, which indicate relatively low lattice distortion and stored strain energy.55−57 Although the film and substrate exhibit different coefficients of thermal expansion,58,59 the low stored strain energy inside the film and substrate is beneficial for the structural integrity of the film; this is attributed to the formation of intermediate phases, which act as a transition “buffer” layer to distribute stress. In addition, pores, twins, and phase changes can also help mitigate the stress. The EBSD neighboring pair misorientation maps (Figure 3h) for the same region reveal two special neighboring pair misorientation angles (approximately 55 and 66°), with significantly high relative frequency (∼10% for 66°) compared to the theoretical value (∼1.75% for 66°). These special angles, 66 and 55°, match well with the rotation angles for typical {101} and {310} TBs for the rutile structure.60,61 The BC and special boundary ({101}TBs with 66 ± 2° and {310} TBs with 55 ± 2°) overlay map (Figure 3c) shows a high density of TBs inside the film; the measured misorientation angles of 66 and 55° confirm the twin types (Figure 3d–g). In contrast, TF6G films contain fewer TBs (Figure 3j) but exhibit larger grain sizes (TF6: ∼1.46 μm; TF6G: ∼3.85 μm). Large slender primary features (marked by red arrows in Figure 3i) can be clearly observed in the BC map (with different contrast compared to the matrix grains). The KAM map (Figure 3j) indicates marginally higher misorientations inside the primary features, which might be associated with a high level of lattice distortion; this will be further investigated at the atomic level by TEM techniques.

Figure 3.

Figure 3

Open in a new tab

EBSD analysis of the TF6 (a–h) and TF6G (i–j) samples. (a) BC + Phase map; (b) BC + KAM map; (c) BC + special boundary map (blue lines marked the {101} TBs, red lines marked the {310} TBs); (d,e) enlargement of the regions denoted by the yellow boxes in (c); (f–h) neighboring pair misorientation angle distribution details for the blue and red chosen paths along {101} and {310} TBs in (d,e); (i) BC map; and (j) BC + KAM map.

Transmission Electron Microscopy

The crystal structure, sub-grain features, and special boundaries of the thick films were investigated by TEM techniques. The SAED patterns and the corresponding HRTEM images collected along the major [111]R and [001]R (R represents rutile) zone axes confirmed the rutile structure for TF1 thick films (Figure S9). Similarly, atomically resolved BF STEM images and atomic number (Z)-sensitive HAADF STEM images, acquired along major zone axes for TF6 and TF6G samples, also confirm the rutile phase (Figure 4); the enlarged STEM images match well with the overlaid simulated ball- and stick-models for rutile structures. The intensity of HAADF cation columns along all zone axes is uniform, reflecting the uniform distribution of cations and oxygen vacancies. The corresponding SAEDs along the major zone axes also match well with the simulated SAED patterns (Figure S10). The results of electron diffraction and atomic-scale studies are thus consistent with data from XRD and SEM investigations concerning the nature of the primary rutile phase in film samples.

Figure 4.

Figure 4

Open in a new tab

Atomically resolved HAADF and BF STEM images of the matrix rutile phase in TF6 and TF6G samples. (a–d) HAADF and BF STEM images collected along [100]R and [111]R zone axes in TF6 films; (e–h) HAADF and BF STEM images collected along [101]R and [111]R zone axes in TF6G films. The corresponding ball- and stick-models were simulated using the rutile P42/mnm structure.

Intragranular features in TF1 films include a high density of bright, parallel lines inside the grains (Figure S11a); atomically resolved HAADF STEM images (Figure 5a,b) collected along the Inline graphic zone axis indicate that the parallel structures are formed at (121)R planes [matching well with the superlattice directions in SAED patterns (Figure S11b)], with an interplanar spacing of approximately 2–10 nm. The enlarged noise-filtered HAADF STEM image (Figure 5c) reveals that the features are typical (121)R crystallographic shear (CS) planes (CSPs) with Inline graphic CS vectors; the CS structure matches well with the simulated ball- and stick-model using Ti8O15I-1 structure.62 The formation of the (121)R CSPs is highly dependent on the degree of oxygen deficiency. Therefore, a higher oxygen vacancy content locally, together with local lattice distortion, disrupting channelling, could generate the bright contrast observed for (121)R CSPs in the HAADF image. Furthermore, the HAADF STEM image (Figure 5b) shows that two Inline graphic CSPs and one Inline graphic anti-phase boundary (APB) formed a closed loop structure. Details in the HAADF STEM image (denoted by the blue box in Figure 5b) reveal the interaction mechanism between the CSPs (highlighted by red dash lines) and the APB (highlighted by green dash lines). These two defects share the same displacement vectors of Inline graphic. The (121)R CSP and the (011)R APB perfectly connect with each other, and the two vectors are superimposed to form a Inline graphic displacement (compensated at the (011)R plane), thereby forming the joint structure. This is the first atomic resolution observation of the complex CSP/APB connecting structure. In addition, edge twins can be observed; an atomically resolved HAADF STEM image (Figure S12) reveals a typical {101}R twin-stacking fault (SF) structure along the Inline graphic zone axis. The cation columns show mirror symmetry at the Inline graphic TB, with rotation angles of approximately 137°, consistent with that of typical {101} TBs in previous work.63 For the upper TB, the insertion of one (011)R plane created the SF alongside the TB and formed the {101}R Twin-SF structure. These nano-sized CSPs, APBs and TBs can effectively control the TE properties through enhancing phonon scattering; the high density of oxygen vacancies introduced by CSPs can also optimize carrier transport in the films.

Figure 5.

Figure 5

Open in a new tab

Atomically resolved STEM analysis of the defect structure in TF1 and TF6 samples. (a,b) HAADF STEM images of the defect structures in TF1 sample along the [11̅1]R zone axis; (c) noise-filtered HAADF STEM image of the region denoted by the yellow box in (b). (d) HAADF STEM image of single TF6 grain along the [010]R zone axis; (e) HAADF STEM image and corresponding FFT pattern of the region denoted by the blue box in (d); (f,g) HAADF/ABF STEM images of the region denoted by the purple box in (d) showing a full dislocation; (h,i) inverse Fast Fourier transformation of (101) and Inline graphic reflections of the full dislocation; and (j,k) corresponding GPA maps of horizontal normal strain (εxx) and shear strain (εxy).

From EBSD boundary analysis (Figure 3), approximately 10% of grain boundaries in TF6 films are actually {101}R TBs. Nevertheless, a high density of intragranular {101}R TBs can be observed in TF6 films. Figures S13 and S14 show the {101}R TBs along [131]R and [111]R zone axes; these indicate that the intervals between the nano-twins are approximately 5 to 20 nm. Aberration-corrected (AC) HAADF STEM images show a typical {101}R TB close to a dislocation along the [010]R zone axis (Figure 5d). The AC HAADF STEM image reveals the nature of the (101)R TB along the [010]R zone axis (Figure 5e). The upper and lower twin domains have mirror symmetry to the (101)R TB and share the (101)R planes; the corresponding (200)R and (101̅)R planes from both twin domains indicate that the rotation angle is approximately 114°, consistent with the rotation angle of the typical (101)R TBs.61 The interplanar spacings across the (101)R twin planes are uniform, indicating no large lattice distortions at the TB interface. The corresponding FFT patterns confirm mirror symmetry between (200)R and Inline graphic planes to the (101)R plane. A dislocation core can be observed very close to the (101)R TB (with a distance of approximately 5 nm) (denoted with a purple box in Figure 5d). The AC HAADF and BF STEM images present the nature of a dislocation core with Burgers’ vectors of 1/2d(101)R (Figure 5f,g). The corresponding Inverse Fast Fourier Transformation along (101)R and Inline graphic planes confirm the presence of a 1/2d(101)R(200)R dislocation (Figure 5h,i). It should be noted that in the STEM images, the position of a Ti column near the dislocation core is not sufficiently clear compared to the parent rutile structure; this is attributed to the highly distorted lattice, which consequently affects the alignment of the Ti column and the electron channelling of the HAADF images. The highly distorted lattice is confirmed by the corresponding GPA maps (Figure 5j,k). The presence of intragranular nano-twins inside the films could be beneficial to enhance the scattering of low-energy carriers and improve the Seebeck coefficients through carrier-filtering effects.64 Furthermore, the nano-TBs and the dislocation cores can effectively enhance phonon scattering65−67 and reduce lattice thermal conductivity, thereby improving the TE performance.

The microstructure of TF6G samples was investigated in more detail using the FIB samples. STEM EDX analysis (Figure S15) shows the elemental distribution across one sample. In TF6G films, large bright/dark stripes with widths of approximately 0.5–1.5 μm are present as primary features in HAADF/BF STEM images (Figure S16). At higher magnification, a high density of secondary features is observed, having the form of stripes and parallelograms typically 20–100 nm in size (Figure S16). The corresponding EDX maps confirm that all the features are enriched with Al but depleted in Ti, Nb, and O. The nature of these features was subsequently determined at the atomic level.

Aberration-corrected (AC) HAADF and BF STEM images of the primary features along the [001]R zone axis reveal an ordered (001)R plane (Figure 6a,b); linear features with intervals of approximately 2–3 nm can be observed. The corresponding FFT patterns are in good accordance with the HAADF image (Figure 6c), and the periodic bright lines from the Inline graphic plane to the (110)R plane and from (000)R to (210)R planes reveal that the linear features are located at (210)R planes, matching well with the SAED patterns of the primary features (Figure S17). Atomic-level EELS and EDX mapping of the linear features along the [001]R zone axis show clear Al segregation at the (210)R plane (Figure 6d,e). The EELS and EDX maps indicate that the linear features are enriched in Al but depleted in Ti and O, consistent with the low-magnification EDX results (Figure S17).

Figure 6.

Figure 6

Open in a new tab

Aberration-corrected STEM analysis of the primary features in TF6G samples. (a–c) Atomically resolved HAADF/BF STEM images and the corresponding FFT pattern of the primary features along the [001]R zone axis; (d,e) atomically resolved EELS/EDX data for the primary features along the [001]R zone axis; (f,g) HAADF STEM image and corresponding schematic map of a single CSP; and (h–j) atomically resolved EELS maps of a single CSP.

An AC HAADF STEM image (Figure 6f) and corresponding schematic map (Figure 6g) of a single (210)R linear structure reveal that in the rutile matrix, the cations are octahedrally surrounded by O ions forming TiO6 octahedra; they are linked by sharing edges and corners to make up the rutile structure. In the (210)R linear features, the cations are still located at the center of the octahedra, but because of oxygen vacancies, the two octahedra on each side of the (210)R plane undergo a 1/2[020]R displacement vector and become face sharing to accommodate the oxygen deficiency. The presence of low-state cations can be distinguished in the EELS data and the fine structure of the Ti L3, 2 edge (Figure S18) by comparing the peak positions with those of Ti2O3 (Ti3+) and TiO2 (Ti4+): the eg peaks are between those for pure Ti3+ and Ti4+. The EELS O K edge signals extracted from both the matrix rutile and (210)R linear structure reveal clear differences in peak intensity ratios (Figure S19): the peak A and B fall at the linear feature, in good agreement with the reported O K energy loss near-edge structure of oxygen-deficient titanium oxide systems.68 Thereby, the (210)R linear structures can be indexed as: 1/2[020]R(210)R; they can be further resolved into a (110)R shear plane (1/2[020]R(110)R) and a (200)R shear plane (1/2[020]R(200)R), with index relation of: 2(210)R = (110)R + (110)R + (200)R. As a result, the (210)R linear features can be defined as oxygen-deficient 1/2[020]R(210)R CS structures. This is the first observation and formal identification of this new, complex CS structure in TiO2-based ceramics.

The chemically sensitive Z contrast HAADF STEM image enables a qualitative distinction between different cations and shows clear intensity differences at the (210)R CS structure. The HAADF intensity profiles across different regions in Figure S20 reveal three distinctly different intensities [from high to low, corresponding to matrix cation sites, (C, D) sites and (A, B) sites at CS planes], which can be attributed to different cation occupancies. Atomically resolved EELS maps (Figure 6h–j) reveal that the matrix cation sites are dominated by Ti (Z is 22), the (A, B) sites are dominated by Al (Z is 13), while the (C, D) sites contain both Ti and Al; these are in good agreement with the intensity profiles. The Al-segregation and the site preferences can be ascribed to the unique CS structure. The face-sharing causes considerable octahedral distortion, and the cations are pushed away from the ideal octahedral centers in the HAADF image. Consequently, the smaller Al ions (RAl3+ = 0.48 Å) diffuse into the CS plane and substitute for larger Ti ions (RTi3+ = 0.67 Å and RTi4+ = 0.605 Å), thus effectively stabilizing the structure and reducing internal stress. Hexagonal tunnels are formed every three Inline graphic planes; hence, the Al cations exhibit a high site occupancy at the edge of hexagonal tunnels due to the low lattice distortion.

In TF6G films, a high density of primary features with different orientations can be observed along the [001]R zone axis (Figure S21); the interfaces between them (containing a high-density CS planes) were investigated using AC STEM and EDX techniques. On the basis of the angle between the CS planes (CSPs), three types of interfaces can be clearly distinguished (Figure 7); they are ∼53° interface (Figure 7a), ∼37° interface (Figure 7b), and ∼90° interface (Figure 7c). EDX maps confirm the single atomic layer thickness Al segregation (enriched in Al but depleted in Ti) at different CSPs (Figure 7a5–a10,b5–b10,c5–c10), consistent with the results of atomically resolved EDX analysis in Figure 6. The ∼53° interface is formed between the 1/2[020]R(210)R CSPs (Figure 7a2) and Inline graphic CSPs (Figure 7a3); the calculated angle between these two CSPs is 126.9°, which is consistent with the complementary angle of ∼53° (Figure 7a4). These two CSPs are intersected at angles of both ∼53 and ∼127° (Figure 7a5). Similarly, the ∼37 and ∼90° interfaces are formed between 1/2[020]R(120)R CSPs (Figure 7b2) and 1/2[020]R(210)R CSPs (Figure 7b3), 1/2[020]R(120)R CSPs (Figure 7c2) and Inline graphic CSPs (Figure 7c3), respectively. The calculated angles between these CSPs (36.87 and 90°) match well with the measured angles (Figure 7b4,c4). It is noted that at ∼37 and ∼90° interfaces, these CSPs are rarely connected; they seem to prefer to associate with one ∼53° and one ∼127° interface between the neighboring planes (Figure 7b5,c5).

Figure 7.

Figure 7

Open in a new tab

Aberration-corrected HAADF images and STEM EDX analyses of the primary feature interfaces in TF6G samples. (a) ∼53° interface; (b) ∼37° interface; and (c) ∼90° interface.

In the rutile structure (tetragonal, space group P42/mnm), the (210)R, Inline graphic, (120)R, Inline graphic, Inline graphic, Inline graphic, Inline graphic, and Inline graphic planes can be classified into the family of the {210}R crystal planes; similarly, the displacement vectors [200]R, [020]R, Inline graphic, and Inline graphic can be classified into ⟨200⟩R. Therefore, these CS planes in the primary features can be classified as 1/2⟨200⟩R{210}R CS planes, and the index relation is 2{210}R = {110}R + {110}R + {200}R. Therefore, the primary features are high-density, ordered Al-segregated 1/2⟨200⟩R{210}R CS planes.

HAADF STEM images collected along the [001]R zone axis show the presence of secondary features (Figure 8a,c). The corresponding FFT patterns indicate that these secondary features are established by connecting single linear features having {210}R orientations. AC HAADF/MAADF STEM images show a single linear section of a secondary feature, which exhibits the same atomic arrangements as that of the CS structure (1/2[020]R(210)R) in primary features (Figure 8k–m). This suggests that the secondary features are formed by connecting disordered 1/2⟨020⟩R{210}R CSPs with preferred ∼37, ∼53, and 90° interfaces. The proposed nature and interpretation of the CSPs are supported by EDX and EELS maps, confirming that the secondary features are enriched in Al and deficient in Ti and O at the CS planes (Figure 8e–j,n–q). However, it is worth noting that the EDX Al maps show the planar segregation features along [113]R and [101]R zone axes (Figures S22f,g and S16g–l), while they show linear segregation features along the [001]R zone axis (Figure 8e–j). This can be explained in terms of the ball- and stick-model, as shown in Figure S23. Aluminum which segregated to the {210}R planes formed the CSPs; the {210}R planes are perpendicular to the {001}R plane and indeed can be observed as linear features. However, there is no vertical orientation relationship between the {210}R and {113}R, {101}R planes, thereby only planar segregation is observed along ⟨113⟩R and ⟨101⟩R zone axes. Therefore, the 1/2⟨020⟩R{210}R CSPs are essentially the basic unit of the primary and secondary features in the TF6G films. The primary features are composed of ordered CSPs, and secondary features are randomly distributed. These carefully characterized features (CSPs) play an important role in controlling the carrier transport, which will be addressed in the TE property sections.

Figure 8.

Figure 8

Open in a new tab

Aberration-corrected STEM analysis of the secondary features in TF6G samples. (a–d) HAADF STEM images and the corresponding FFT patterns of the secondary features along the [001]R zone axis; (e–j) HAADF STEM image and the corresponding EDX maps of the secondary features; and (k–q) atomically resolved EELS analysis (k—survey image; l,m—HAADF/MAADF STEM images) of a single CS plane in secondary features.

X-ray Photoelectron Spectroscopy

XPS survey scans confirm the presence of all primary elements in the films; the higher Nb 3d peak intensity of TF6 and TF6G films are ascribed to the higher Nb doping level (Figure S24). Figure 9 shows high-resolution XPS spectra for the Ti 2p transitions and Nb 3d transitions; the doublet in Ti 2p and Nb 3d spectra results from spin orbit-splitting, with binding energy spitting values of ∼5.669 and ∼2.6 eV,70 respectively. Notable asymmetry of the peaks can be attributed to the presence of low-oxidation state cations,71−73 Ti3+ and Nb4+, which are confirmed by Ti 2p and Nb 3d peak fittings (Figure 9). The increasing content of low-oxidation state cations, [Ti3+] and [Nb4+], exhibits an upward trend: TF1 < TF6 < TF6G (Table S4); the increase from TF1 to TF6 film is mainly due to the higher Nb content, indicating the positive effect of Nb on the reduction of cations in this system. In contrast, the increase from TF6 to TF6G is attributed to the stronger reducing atmosphere during sintering, promoting the enhanced reduction of the cations. Since the concentration of Nb4+ only slightly increased in TF6G samples, the relatively large enhancement in the Ti3+ concentration reveals a higher level of oxygen deficiency, which confirmed that the strongly reducing atmosphere is beneficial for the formation of oxygen vacancies.

Figure 9.

Figure 9

Open in a new tab

High-resolution XPS spectra for Ti 2p transition and Nb 3d transition for the thick films: (a,b) TF1; (c,d) TF6; and (e,f) TF6G.

Electrical Conductivity

Figure 10 shows the carrier transport properties for the thick films as a function of temperature. The electrical conductivity (σ) values increase with increasing temperature, confirming semiconducting behavior for all thick films (Figure 10a). The σ values of TF1 films are relatively low (1.1–5.7 S cm–1), while TF6 films exhibit much higher σ (6.9–30.6 S cm–1), indicating the benefit of Nb doping74 which gives rise to increased carrier concentrations (n) and thereby higher σ. In contrast to TF6, the TF6G films exhibit significant enhancement (18.2–68.8 S cm–1) in σ, which can be attributed to the strongly reducing atmosphere and the formation of Al-segregated CS structures. The latter are accompanied by a high density of oxygen vacancies. The XPS data confirm that TF6G films contain a higher level of oxygen vacancies than TF6 films, contributing to the enhancement of n and σ. It is worth noting that the σ value for TF6G films at 673 K is approximately the same as that for densified bulk samples with the same composition,11 and this high σ value is still achieved with films containing approximately 17% porosity.

Figure 10.

Figure 10

Open in a new tab

Carrier transport properties of TiO2-based thick films as a function of temperature: (a) electrical conductivity; (b) Seebeck coefficients; (c) PF; and (d) PF compared with published data.

Seebeck Coefficients

Figure 10b presents the temperature-dependent Seebeck coefficients (S) for the thick films; all are negative, indicating n-type conduction with electrons as carriers. The absolute S values decrease with increasing temperature and are consistent with previous studies.21 The TF1 films show the highest absolute S values (approximately 360 to 380 μ V K–1), in good agreement with earlier investigations employing Nb doping.21 In contrast, TF6 films exhibit much lower absolute S values (approximately 255 to 295 μ V K–1). The sharp decrease in S for TF6 film can be understood through the Mott formula (eq 1)75

graphic file with name am2c16587_m023.jpg 1
graphic file with name am2c16587_m024.jpg 2

where S is the Seebeck coefficient, Ce is the electron specific heat, n is the carrier concentration, k is the Boltzmann constant, T is the absolute temperature, e is the electron charge, and μ(ε) is the energy correlated carrier mobility. Typically, the first term of the eq 1 is considered to be dominant and thus is similar to the Drude model (eq 2).75 Therefore, the higher Nb doping of TF6 films gives rise to more carriers, and thus, the sharp decrease of S can be attributed to variations in n. However, the absolute S values of TF6 films are slightly higher than the previously reported values for similar compositions;11 this can be ascribed to the presence of significant amounts of TBs and different processing atmospheres. The higher density of TBs contributes to S by energy filtering effects;76 while the less reducing atmosphere results in lower oxygen deficiency, which reduces n and thereby enhances S.

While TF6G films exhibit much higher electrical conductivity than TF6 films, the absolute S values are only slightly lower than those of TF6 films (Figure 10b). This indicates that S is not only controlled by the number of carriers but also by the energy of the carriers. Therefore, the Seebeck coefficient can be further expressed as

graphic file with name am2c16587_m025.jpg 3

where ρ is the electrical resistivity, e is the electron charge, ϵ, μ, and σ(ϵ) are the chemical potential, the energy of the carrier, and the differential conductivity, respectively.77 In TF6G films, the high-density ordered and disordered CS planes are interlaced, having a scale of 1–2 nm; this might be able to produce a high density of interfaces with potential barriers. Moizhes et al.78 argued that the contribution to the Seebeck coefficient from low-energy carriers (ϵ < μ) is in contrast to that of high-energy carriers (ϵ > μ), indicating that absolute S values are reduced by compensation effects between the high- and low-energy carriers. Therefore, the high density of CS planes in TF6G films are believed to induce carrier-filtering effects, encouraging high absolute S values compared to TF6 films. The high-energy carriers (ϵ > μ) can pass through the CS planes without being affected, which ensures high carrier transport and high electrical conductivity. However, the low-energy carriers (ϵ > μ) can be strongly scattered by these CS planes, minimizing any reduction in absolute S values.

Power Factor

The significantly enhanced electrical conductivity for TF6G thick films, along with the high Seebeck coefficients, results in a high-PF value of approximately 430 μW m–1 K–2 at 673 K, which is approximately twice than that for TF6 films (Figure 10c). Most importantly, the high PF values for TF6G are achieved in spite of the high porosity of approximately 17%. The good reproducibility of the PF data is demonstrated in Figure S25. Although marginally higher PF values have been reported for transparent (TiO2:Nb) thin films (∼150 nm thick) near room temperature,38 the transport properties of our thick films are exceptional and significantly exceed the high-temperature properties of recently reported TiO2-based films (Figure 10d).22,36,37,40,41

Conclusions

High-performance, high stability Nb-doped TiO2 TE thick films have been prepared by a scalable, low-cost screen-printing technique and traditional pressureless sintering methods. Utilizing advanced characterization methods, we have defined the first time the nature of the CS structures and infer from these observations how they might enhance TE performance.

The as-sintered thick films are predominantly rutile phase, with secondary phases resulting from reactions between the substrates and films. Nano-sized point and planar defects control the TE performance of the films. Compared to TF1 films, the higher Nb content of TF6 films suppresses the formation of oxygen vacancies and promotes the formation of planar defects including {121} CS planes and {101} and {301} TBs. For TF6G films, the strongly reducing processing atmosphere lowers the density of TBs, while promoting Al diffusion to the film and the formation of a high density of Al-segregated {210} CS structures.

Through the use of advanced aberration-corrected STEM techniques, we demonstrate for the first time the nature of the defects at the atomic level: the {121} CSP/{101} APB/{121} CSP connecting structure has been observed and resolved; the nature of {101} TBs and 1/2d(101)R(200)R dislocations have been explored. In addition, we discovered a new Al-segregated {210} CS structure and presented the detailed structure and nature of its chemistry at the atomic level.

There exist significant energy-filtering effects at the oxygen-deficient {210} CS interfaces in the films, leading to synchronized enhancement of electrical conductivity and Seebeck coefficients, and hence, a remarkable PF value of 430 μW m–1 K–2 is achieved at 673 K. This is exceptionally high and exceeds the high-temperature properties of recently reported TiO2-based films. The outstanding TE performance and stability of the thick films prepared by a low-cost route open up a new approach for fabrication of TE films. The newly discovered CS structure broadens the CS family and provides a successful strategy for modifying atomic-level structures and enhancing the TE performance of such materials.

Acknowledgments

The authors are grateful to the EPSRC for the provision of funding for this work (EP/H043462, EP/I036230/1, EP/L014068/1, and EP/L017695/1 acknowledged by R.F.). The work was also supported by the Henry Royce Institute for Advanced Materials, funded through EPSRC Grants EP/R00661X/1, EP/S019367/1, EP/P025021/1, and EP/P025498/1. Part of the electron microscopy work was carried out at SuperSTEM, the UK National Research Facility for Advanced Electron Microscopy, supported by EPSRC (EP/W021080/1). We gratefully acknowledge the support from X-ray facilities in the Department of Materials in the University of Manchester. X.L. thanks China Scholarship Council for their financial support during his Ph.D. program. All research data supporting this work are directly available within this publication.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.2c16587.

  • Experimental procedures; lattice parameters, XPS component analysis of the as-printed films; SEM, experimental, and simulated TEM SAED, HRTEM, STEM EDX, STEM EELS, HAADF, and BF AC-STEM data of the sintered thick films; and XPS survey spectra and reproducibility of TE performance of the sintered thick films (PDF)

The authors declare no competing financial interest.

Supplementary Material

am2c16587_si_001.pdf (2.1MB, pdf)

References

  1. He J.; Tritt T. M. Advances in Thermoelectric Materials Research: Looking Back and Moving Forward. Science 2017, 357, eaak9997 10.1126/science.aak9997. [DOI] [PubMed] [Google Scholar]
  2. Koumoto K.; Funahashi R.; Guilmeau E.; Miyazaki Y.; Weidenkaff A.; Wang Y.; Wan C. Thermoelectric Ceramics for Energy Harvesting. J. Am. Ceram. Soc. 2013, 96, 1–23. 10.1111/jace.12076. [DOI] [Google Scholar]
  3. Jakubczyk E. M.; Mapp A.; Chung C. C.; Sansom C. L.; Jones J. L.; Dorey R. A. Enhancing Thermoelectric Properties of NaCo2O4 Ceramics through Na Pre-Treatment Induced Nano-Decoration. J. Alloys Compd. 2019, 788, 91–101. 10.1016/j.jallcom.2019.02.199. [DOI] [Google Scholar]
  4. Lim C. H.; Seo W. S.; Lee S.; Lim Y. S.; Kim J. Y.; Park H. H.; Choi S. M.; Lee K. H.; Park K. Anisotropy of the Thermoelectric Figure of Merit (ZT) in Textured Ca3Co4O9 Ceramics Prepared by Using a Spark Plasma Sintering Process. J. Korean Phys. Soc. 2015, 66, 794–799. 10.3938/jkps.66.794. [DOI] [Google Scholar]
  5. Takahashi K.; Sakai A.; Adachi H.; Kanno T. The Off-Diagonal Thermoelectric Effect in Inclinedly Oriented Thin Films of Layered Cobaltite CaxCoO2. J. Phys. D: Appl. Phys. 2010, 43, 165403. 10.1088/0022-3727/43/16/165403. [DOI] [Google Scholar]
  6. Mikami M.; Funahashi R. The Effect of Element Substitution on High-Temperature Thermoelectric Properties of Ca3Co2O6 Compounds. J. Solid State Chem. 2005, 178, 1670–1674. 10.1016/j.jssc.2005.03.004. [DOI] [Google Scholar]
  7. Delorme F.; Diaz-Chao P.; Guilmeau E.; Giovannelli F. Thermoelectric Properties of Ca3Co4O9–Co3O4 Composites. Ceram. Int. 2015, 41, 10038–10043. 10.1016/j.ceramint.2015.04.091. [DOI] [Google Scholar]
  8. Hsu H. C.; Lee W. L.; Wu K. K.; Kuo Y. K.; Chen B. H.; Chou F. C. Enhanced Thermoelectric Figure-of-Merit ZT for Hole-Doped Bi2Sr2Co2Oy through Pb Substitution. J. Appl. Phys. 2012, 111, 103709. 10.1063/1.4720075. [DOI] [Google Scholar]
  9. Wang J.; Zhang B. Y.; Kang H. J.; Li Y.; Yaer X.; Li J. F.; Tan Q.; Zhang S.; Fan G. H.; Liu C. Y.; Miao L.; Nan D.; Wang T. M.; Zhao L. D. Record High Thermoelectric Performance in Bulk SrTiO3 via Nano-Scale Modulation Doping. Nano Energy 2017, 35, 387–395. 10.1016/j.nanoen.2017.04.003. [DOI] [Google Scholar]
  10. Okinaka N.; Akiyama T. Latent Property of Defect-Controlled Metal Oxide: Nonstoichiometric Titanium Oxides as Prospective Material for High-Temperature Thermoelectric Conversion. Jpn. J. Appl. Phys. 2006, 45, 7009–7010. 10.1143/jjap.45.7009. [DOI] [Google Scholar]
  11. Liu X.; Kepaptsoglou D.; Gao Z.; Thomas A.; Maji K.; Guilmeau E.; Azough F.; Ramasse Q. M.; Freer R. Controlling the Thermoelectric Properties of Nb-Doped TiO2 Ceramics through Engineering Defect Structures. ACS Appl. Mater. Interfaces 2021, 13, 57326–57340. 10.1021/acsami.1c18236. [DOI] [PubMed] [Google Scholar]
  12. Thébaud S.; Adessi C.; Bouzerar G. Investigating the High-Temperature Thermoelectric Properties of n-Type Rutile TiO2. Phys. Rev. B 2019, 100, 195202. 10.1103/physrevb.100.195202. [DOI] [Google Scholar]
  13. Labégorre J.-B.; Lebedev O. I.; Bourgès C.; Rečnik A.; Košir M.; Bernik S.; Maignan A.; le Mercier T.; Pautrot-d’Alencon L.; Guilmeau E. Phonon Scattering and Electron Doping by 2D Structural Defects in In/ZnO. ACS Appl. Mater. Interfaces 2018, 10, 6415–6423. 10.1021/acsami.7b19489. [DOI] [PubMed] [Google Scholar]
  14. Wang Y.; Sui Y.; Fan H.; Wang X.; Su Y.; Su W.; Liu X. High Temperature Thermoelectric Response of Electron-Doped CaMnO3. Chem. Mater. 2009, 21, 4653–4660. 10.1021/cm901766y. [DOI] [Google Scholar]
  15. Liu Y.; Yang J.; Liu Y.; Zheng J.; Lee W.; Shi J.; Horlyck J.; Xie J.; Tay Y. Y.; Tan T. T.; Yu D.; Mole R.; McIntyre G.; Zhang C.; Toe C. Y.; Waite T. D.; Scott J.; Wang Y.; Wu T.; Han S.; Li S. Manipulation of Planar Oxygen Defect Arrangements in Multifunctional Magnéli Titanium Oxide Hybrid Systems: From Energy Conversion to Water Treatment. Energy Environ. Sci. 2020, 13, 5080–5096. 10.1039/d0ee02550j. [DOI] [Google Scholar]
  16. Backhaus-Ricoult M.; Rustad J.; Moore L.; Smith C.; Brown J. Semiconducting Large Bandgap Oxides as Potential Thermoelectric Materials for High-Temperature Power Generation?. Appl. Phys. A: Mater. Sci. Process. 2014, 116, 433–470. 10.1007/s00339-014-8515-z. [DOI] [Google Scholar]
  17. Tran N. Q. M.; Ohtaki M.; Suekuni K. High-Temperature Thermoelectric Performance of (W1–xTix)18O49. J. Eur. Ceram. Soc. 2022, 42, 1486–1492. 10.1016/j.jeurceramsoc.2021.12.014. [DOI] [Google Scholar]
  18. Klich W.; Ohtaki M. Thermoelectric Properties of Mo-Doped Bulk In2O3 and Prediction of Its Maximum ZT. Ceram. Int. 2021, 47, 18116–18121. 10.1016/j.ceramint.2021.03.129. [DOI] [Google Scholar]
  19. Azough F.; Freer R.; Yeandel S. R.; Baran J. D.; Molinari M.; Parker S. C.; Guilmeau E.; Kepaptsoglou D.; Ramasse Q.; Knox A.; Gregory D.; Paul D.; Paul M.; Montecucco A.; Siviter J.; Mullen P.; Li W.; Han G.; Man E. A.; Baig H.; Mallick T.; Sellami N.; Min G.; Sweet T. Ba6–3xNd8+2xTi18O54 Tungsten Bronze: A New High-Temperature n-Type Oxide Thermoelectric. J. Electron. Mater. 2016, 45, 1894–1899. 10.1007/s11664-015-4275-6. [DOI] [Google Scholar]
  20. Jiang D.; Ekren D.; Azough F.; Day S. J.; Chen K.; Mahajan A.; Kepaptsoglou D. M.; Ramasse Q. M.; Reece M. J.; Freer R. The Structure and Thermoelectric Properties of Tungsten Bronze Ba6Ti2Nb8O30. J. Appl. Phys. 2019, 126, 125115. 10.1063/1.5119962. [DOI] [Google Scholar]
  21. Verchère A.; Pailhès S.; Le Floch S.; Cottrino S.; Debord R.; Fantozzi G.; Misra S.; Candolfi C.; Lenoir B.; Daniele S.; Mishra S. Optimum in the Thermoelectric Efficiency of Nanostructured Nb-Doped TiO2 Ceramics: From Polarons to Nb–Nb Dimers. Phys. Chem. Chem. Phys. 2020, 22, 13008–13016. 10.1039/d0cp00652a. [DOI] [PubMed] [Google Scholar]
  22. Xu L.; Garrett M. P.; Hu B. Doping Effects on Internally Coupled Seebeck Coefficient, Electrical, and Thermal Conductivities in Aluminum-Doped TiO2. J. Phys. Chem. C 2012, 116, 13020–13025. 10.1021/jp302652c. [DOI] [Google Scholar]
  23. Zaitsev S. v.; Moon J.; Takagi H.; Awano M. Preparation and Characterization of Nanocrystalline Doped TiO2. Adv. Powder Technol. 2000, 11, 211–220. 10.1163/156855200750172321. [DOI] [Google Scholar]
  24. Fan Y.; Feng X.; Zhou W.; Murakami S.; Kikuchi K.; Nomura N.; Wang L.; Jiang W.; Kawasaki A. Preparation of Monophasic Titanium Sub-Oxides of Magnéli Phase with Enhanced Thermoelectric Performance. J. Eur. Ceram. Soc. 2018, 38, 507–513. 10.1016/j.jeurceramsoc.2017.09.040. [DOI] [Google Scholar]
  25. Harada S.; Tanaka K.; Inui H. Thermoelectric Properties and Crystallographic Shear Structures in Titanium Oxides of the Magnéli Phases. J. Appl. Phys. 2010, 108, 083703. 10.1063/1.3498801. [DOI] [Google Scholar]
  26. Portehault D.; Maneeratana V.; Candolfi C.; Oeschler N.; Veremchuk I.; Grin Y.; Sanchez C.; Antonietti M. Facile General Route toward Tunable Magnéli Nanostructures and Their Use as Thermoelectric Metal Oxide/Carbon Nanocomposites. ACS Nano 2011, 5, 9052–9061. 10.1021/nn203265u. [DOI] [PubMed] [Google Scholar]
  27. Yu Y.; He D. S.; Zhang S.; Cojocaru-Mirédin O.; Schwarz T.; Stoffers A.; Wang X. Y.; Zheng S.; Zhu B.; Scheu C.; Wu D.; He J. Q.; Wuttig M.; Huang Z. Y.; Zu F. Q. Simultaneous Optimization of Electrical and Thermal Transport Properties of Bi0.5Sb1.5Te3 Thermoelectric Alloy by Twin Boundary Engineering. Nano Energy 2017, 37, 203–213. 10.1016/j.nanoen.2017.05.031. [DOI] [Google Scholar]
  28. Xiong Z.; Chen X.; Huang X.; Bai S.; Chen L. High Thermoelectric Performance of Yb0.26Co4Sb12/YGaSb Nanocomposites Originating from Scattering Electrons of Low Energy. Acta Mater. 2010, 58, 3995–4002. 10.1016/j.actamat.2010.03.025. [DOI] [Google Scholar]
  29. Xie W. J.; Yan Y. G.; Zhu S.; Zhou M.; Populoh S.; Gałązka K.; Poon S. J.; Weidenkaff A.; He J.; Tang X. F.; Tritt T. M. Significant ZT Enhancement in P-Type Ti(Co,Fe)Sb–InSb Nanocomposites via a Synergistic High-Mobility Electron Injection, Energy-Filtering and Boundary-Scattering Approach. Acta Mater. 2013, 61, 2087–2094. 10.1016/j.actamat.2012.12.028. [DOI] [Google Scholar]
  30. Petsagkourakis I.; Tybrandt K.; Crispin X.; Ohkubo I.; Satoh N.; Mori T. Thermoelectric Materials and Applications for Energy Harvesting Power Generation. Sci. Technol. Adv. Mater. 2018, 19, 836–862. 10.1080/14686996.2018.1530938. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Burton M.; Howells G.; Atoyo J.; Carnie M. Printed Thermoelectrics. Adv. Mater. 2022, 34, 2108183. 10.1002/adma.202108183. [DOI] [PubMed] [Google Scholar]
  32. Tarancón A. Powering the IoT revolution with heat. Nat. Electron. 2019, 2, 270–271. 10.1038/s41928-019-0276-4. [DOI] [Google Scholar]
  33. Nandihalli N.; Liu C. J.; Mori T. Polymer Based Thermoelectric Nanocomposite Materials and Devices: Fabrication and Characteristics. Nano Energy 2020, 78, 105186. 10.1016/j.nanoen.2020.105186. [DOI] [Google Scholar]
  34. Han C.; Tan G.; Varghese T.; Kanatzidis M. G.; Zhang Y. High-Performance PbTe Thermoelectric Films by Scalable and Low-Cost Printing. ACS Energy Lett. 2018, 3, 818–822. 10.1021/acsenergylett.8b00041. [DOI] [Google Scholar]
  35. Sun Y.; Rogers J. A. Inorganic Semiconductors for Flexible Electronics. Adv. Mater. 2007, 19, 1897–1916. 10.1002/adma.200602223. [DOI] [Google Scholar]
  36. Lee H.; Seshadri R. C.; Pala Z.; Sampath S. Optimizing Thermoelectric Properties of In Situ Plasma-Spray-Synthesized Sub-Stoichiometric TiO2–x Deposits. J. Therm. Spray Technol. 2018, 27, 968–982. 10.1007/s11666-018-0731-1. [DOI] [Google Scholar]
  37. Lee H.; Han S. J.; Chidambaram Seshadri R.; Sampath S. Thermoelectric Properties of In-Situ Plasma Spray Synthesized Sub-Stoichiometry TiO2-x. Sci. Rep. 2016, 6, 36581. 10.1038/srep36581. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Ribeiro J. M.; Correia F. C.; Kuzmin A.; Jonane I.; Kong M.; Goñi A. R.; Reparaz J. S.; Kalinko A.; Welter E.; Tavares C. J. Influence of Nb-Doping on the Local Structure and Thermoelectric Properties of Transparent TiO2:Nb Thin Films. J. Alloys Compd. 2020, 838, 155561. 10.1016/j.jallcom.2020.155561. [DOI] [Google Scholar]
  39. Lu Z.; Layani M.; Zhao X.; Tan L. P.; Sun T.; Fan S.; Yan Q.; Magdassi S.; Hng H. H. Fabrication of Flexible Thermoelectric Thin Film Devices by Inkjet Printing. Small 2014, 10, 3551–3554. 10.1002/smll.201303126. [DOI] [PubMed] [Google Scholar]
  40. Lee H.; Chidambaram Seshadri R.; Han S. J.; Sampath S. TiO2–x Based Thermoelectric Generators Enabled by Additive and Layered Manufacturing. Appl. Energy 2017, 192, 24–32. 10.1016/j.apenergy.2017.02.001. [DOI] [Google Scholar]
  41. Kurita D.; Ohta S.; Sugiura K.; Ohta H.; Koumoto K. Carrier Generation and Transport Properties of Heavily Nb-Doped Anatase TiO2 Epitaxial Films at High Temperatures. J. Appl. Phys. 2006, 100, 96105. 10.1063/1.2362990. [DOI] [Google Scholar]
  42. Nishio Y.; Hirano T. Improvement of the Efficiency of Thermoelectric Energy Conversion by Utilizing Potential Barriers. Jpn. J. Appl. Phys. 1997, 36, 170–174. 10.1143/jjap.36.170. [DOI] [Google Scholar]
  43. Mele P.; Saini S.; Honda H.; Matsumoto K.; Miyazaki K.; Hagino H.; Ichinose A. Effect of Substrate on Thermoelectric Properties of Al-Doped ZnO Thin Films. Appl. Phys. Lett. 2013, 102, 253903. 10.1063/1.4812401. [DOI] [Google Scholar]
  44. Hirose Y.; Tsuchii M.; Shigematsu K.; Kakefuda Y.; Mori T.; Hasegawa T. Thermoelectric Properties of Amorphous ZnOxNy Thin Films at Room Temperature. Appl. Phys. Lett. 2019, 114, 193903. 10.1063/1.5089679. [DOI] [Google Scholar]
  45. Duployer B.; Tenailleau C.; Thimont Y.; Lenormand P.; Barnabé A.; Presmanes L. Preparation and Study of CuBi2O4 Thin Films by RF Magnetron Sputtering. Mater. Res. Bull. 2020, 130, 110940. 10.1016/j.materresbull.2020.110940. [DOI] [Google Scholar]
  46. Cao J.; Ekren D.; Peng Y.; Azough F.; Kinloch I. A.; Freer R. Modulation of Charge Transport at Grain Boundaries in SrTiO3: Toward a High Thermoelectric Power Factor at Room Temperature. ACS Appl. Mater. Interfaces 2021, 13, 11879–11890. 10.1021/acsami.0c21699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Coelho A. A. Whole-Profile Structure Solution from Powder Diffraction Data Using Simulated Annealing. J. Appl. Crystallogr. 2000, 33, 899–908. 10.1107/s002188980000248x. [DOI] [Google Scholar]
  48. Potapov P.; Longo P.; Okunishi E. Enhancement of Noisy EDX HRSTEM Spectrum-Images by Combination of Filtering and PCA. Micron 2017, 96, 29–37. 10.1016/j.micron.2017.02.001. [DOI] [PubMed] [Google Scholar]
  49. Yin Y. CrystalMaker 6.3 CrystalMaker Software, Begbroke Science Park, Building 5, Sandy Lane, Yarnton OX5 1P5, U.K. http://www.crystalmaker.com. $599.00 (single user), $415.00 (academic, single user). Contact company for other pricing options. J. Am. Chem. Soc. 2004, 126, 14996. 10.1021/ja040947r. [DOI] [Google Scholar]
  50. Hÿtch M. J.; Snoeck E.; Kilaas R. Quantitative Measurement of Displacement and Strain Fields from HREM Micrographs. Ultramicroscopy 1998, 74, 131–146. 10.1016/S0304-3991(98)00035-7. [DOI] [Google Scholar]
  51. Walton J.; Wincott P.; Fairley N.; Carrick A.. Peak Fitting with CasaXPS : A Casa Pocketbook; Acolyte Science: Knutsford, 2010. [Google Scholar]
  52. Das S. The Al-O-Ti (Aluminum-Oxygen-Titanium) System. J. Phase Equilib. 2002, 23, 525–536. 10.1361/105497102770331271. [DOI] [Google Scholar]
  53. Skala R. D.; Li D.; Low I. M. Diffraction, Structure and Phase Stability Studies on Aluminium Titanate. J. Eur. Ceram. Soc. 2009, 29, 67–75. 10.1016/j.jeurceramsoc.2008.05.037. [DOI] [Google Scholar]
  54. Shannon R. D. Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides. Acta Crystallogr., Sect. A: Found. Adv. 1976, 32, 751–767. 10.1107/s0567739476001551. [DOI] [Google Scholar]
  55. Takayama Y.; Szpunar J. A. Stored energy and Taylor factor relation in an Al-Mg-Mn alloy sheet worked by continuous cyclic bending. Mater. Trans. 2004, 45, 2316–2325. 10.2320/matertrans.45.2316. [DOI] [Google Scholar]
  56. Wright S. I.; Nowell M. M.; Field D. P. A Review of Strain Analysis Using Electron Backscatter Diffraction. Microsc. Microanal. 2011, 17, 316–329. 10.1017/s1431927611000055. [DOI] [PubMed] [Google Scholar]
  57. Saraf L. Kernel Average Misorientation Confidence Index Correlation from FIB Sliced Ni-Fe-Cr Alloy Surface. Microsc. Microanal. 2011, 17, 424–425. 10.1017/s1431927611002996. [DOI] [Google Scholar]
  58. Meagher E. P.; Lager G. A. Polyhedral thermal expansion in the TiO2 polymorphs; refinement of the crystal structures of rutile and brookite at high temperature. Can. Mineral. 1979, 17, 77–85. [Google Scholar]
  59. Yates B.; Cooper R. F.; Pojur A. F. Thermal expansion at elevated temperatures. II. Aluminium oxide: Experimental data between 100 and 800 K and their analysis. J. Phys. C: Solid State Phys. 1972, 5, 1046. 10.1088/0022-3719/5/10/009. [DOI] [Google Scholar]
  60. Suzuki K.; Ichihara M.; Takeuchi S. High-Resolution Electron Microscopy of Lattice Defects in TiO2 and SnO2. Philos. Mag. A 1991, 63, 657–665. 10.1080/01418619108213906. [DOI] [Google Scholar]
  61. Lu W.; Bruner B.; Casillas G.; He J.; Jose-Yacaman M.; Farmer P. J. Large Scale Synthesis of V-Shaped Rutile Twinned Nanorods. CrystEngComm 2012, 14, 3120–3124. 10.1039/c2ce06564a. [DOI] [Google Scholar]
  62. Le Page Y.; Strobel P. Structural Chemistry of the Magnéli Phases TinO2n-1, 4 ≤ n ≤ 9 : II. Refinements and Structural Discussion. J. Solid State Chem. 1982, 44, 273–281. 10.1016/0022-4596(82)90374-7. [DOI] [Google Scholar]
  63. Zheng J. G.; Pan P.; Schweizer M.; Zhou F.; Weimar U.; Göpel W.; Rühle M. Growth Twins in Nanocrystalline SnO2 Thin Films by High-Resolution Transmission Electron Microscopy. J. Appl. Phys. 1996, 79, 7688. 10.1063/1.362434. [DOI] [Google Scholar]
  64. Minnich A. J.; Dresselhaus M. S.; Ren Z. F.; Chen G. Bulk Nanostructured Thermoelectric Materials: Current Research and Future Prospects. Energy Environ. Sci. 2009, 2, 466–479. 10.1039/b822664b. [DOI] [Google Scholar]
  65. Zhou Y.; Yang J.-Y.; Cheng L.; Hu M. Strong Anharmonic Phonon Scattering Induced Giant Reduction of Thermal Conductivity in PbTe Nanotwin Boundary. Phys. Rev. B 2018, 97, 085304. 10.1103/physrevb.97.085304. [DOI] [Google Scholar]
  66. Ming H.; Zhu G.; Zhu C.; Qin X.; Chen T.; Zhang J.; Li D.; Xin H.; Jabar B. Boosting Thermoelectric Performance of Cu2SnSe3 via Comprehensive Band Structure Regulation and Intensified Phonon Scattering by Multidimensional Defects. ACS Nano 2021, 15, 10532–10541. 10.1021/acsnano.1c03120. [DOI] [PubMed] [Google Scholar]
  67. Wang H.; Zheng S.; Wu H.; Xiong X.; Xiong Q.; Wang H.; Wang Y.; Zhang B.; Lu X.; Han G.; Wang G.; Zhou X. Realizing Enhanced Thermoelectric Performance and Hardness in Icosahedral Cu5FeS4–xSex with High-Density Twin Boundaries. Small 2022, 18, 2104592. 10.1002/smll.202104592. [DOI] [PubMed] [Google Scholar]
  68. Stoyanov E.; Langenhorst F.; Steinle-Neumann G. The Effect of Valence State and Site Geometry on Ti L3,2 and O K Electron Energy-Loss Spectra of TixOy Phases. Am. Mineral. 2007, 92, 577–586. 10.2138/am.2007.2344. [DOI] [Google Scholar]
  69. Armitage D. A.; Grant D. M. Characterisation of Surface-Modified Nickel Titanium Alloys. Mater. Sci. Eng., A 2003, 349, 89–97. 10.1016/s0921-5093(02)00818-3. [DOI] [Google Scholar]
  70. Atashbar M. Z.; Sun H. T.; Gong B.; Wlodarski W.; Lamb R. XPS Study of Nb-Doped Oxygen Sensing TiO2 Thin Films Prepared by Sol-Gel Method. Thin Solid Films 1998, 326, 238–244. 10.1016/s0040-6090(98)00534-3. [DOI] [Google Scholar]
  71. Kurtz R. L.; Henrich V. E. UHV-Cleaved Single Crystal Ti2O3 (101̅2) by UPS and XPS. Surf. Sci. Spectra 1998, 5, 182–185. 10.1116/1.1247849. [DOI] [Google Scholar]
  72. Darlinski A.; Halbritter J. Angle-resolved XPS studies of oxides at NbN, NbC, and Nb surfaces. Surf. Interface Anal. 1987, 10, 223–237. 10.1002/sia.740100502. [DOI] [Google Scholar]
  73. Jackman M. J.; Thomas A. G.; Muryn C. Photoelectron Spectroscopy Study of Stoichiometric and Reduced Anatase TiO2 (101) Surfaces: The Effect of Subsurface Defects on Water Adsorption at near-Ambient Pressures. J. Phys. Chem. C 2015, 119, 13682–13690. 10.1021/acs.jpcc.5b02732. [DOI] [Google Scholar]
  74. Di Valentin C.; Pacchioni G.; Selloni A. Reduced and n-Type Doped TiO2: Nature of Ti3+ Species. J. Phys. Chem. C 2009, 113, 20543–20552. 10.1021/jp9061797. [DOI] [Google Scholar]
  75. Wang Y.; Sui Y.; Wang X.; Su W.; Liu X. Enhanced High Temperature Thermoelectric Characteristics of Transition Metals Doped Ca3Co4O9+δ by Cold High-Pressure Fabrication. J. Appl. Phys. 2010, 107, 033708. 10.1063/1.3291125. [DOI] [Google Scholar]
  76. Dharmaiah P.; Lee K.-H.; Song S. H.; Kim H. S.; Hong S.-J. Enhanced Thermoelectric Performance of Bi0.5Sb1.5Te3 Composites through Potential Barrier Scattering at Heterogeneous Interfaces. Mater. Res. Bull. 2021, 133, 111023. 10.1016/j.materresbull.2020.111023. [DOI] [Google Scholar]
  77. Zhou Y.; Matsubara I.; Shin W.; Izu N.; Murayama N. Effect of Grain Size on Electric Resistivity and Thermopower of (Ca2.6Bi0.4)Co4O9 Thin Films. J. Appl. Phys. 2004, 95, 625–628. 10.1063/1.1631732. [DOI] [Google Scholar]
  78. Moizhes B. Y.; Nemchinsky V. A.. 11th Int. Conf. Thermoelectrics, 1992; University of Texas Press: Texas, 1993; p 232.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

am2c16587_si_001.pdf (2.1MB, pdf)

Articles from ACS Applied Materials & Interfaces are provided here courtesy of American Chemical Society

RESOURCES