Inparticular, the structural and thermoelectric properties of the ZnO-based thin filmsare modified by controlling carrier concentration, mobility, Fermi level, density-of-state effective
Quality factor oo -.-.-ssaksA
Quality factor (B) is also well-known as an effective evaluation for TE performance of a material under acoustic-phonon scattering assumption, which is defined as [4-6]: tw (—) 5/2
Kia \300 (Eq 15) where uw is the weighted carrier mobility which can be directly extracted from Seebeck coefficient (5) and electrical conductivity (ứ) measurement as the following equation [5]: ex (4 -2) 3 _ 5|
W 8re(2mokpgT)3/2 E (4 _ | (4 _ )| "B 1+exp|—5 Ep/e 1 1 +exp|5 kp/e 1 where h is the Planck constant; kg is the Boltzmann constant; e is the elementary charge; and mo 1s the rest mass of electrons.
In TE materials, electronic transport properties can be quantitatively characterized by weighted mobility [5], which relates directly to carrier mobility and band structure (DOS) of the material It can be controlled by scattering between carriers with multiscale scattering centers Normally, the energy- and temperature- dependent magnitude are characteristics of each scattering mechanism [7], which depends on the kind of lattice defects originating from dopant level and preparation conditions.
Trade-offs among TE parameters - G5 Ăn vn g erey 7 1 Electrical COMCUCTIVILY - - ôkg HH ng ng 8 2 Seebeck COeÍICITIL - G9 HH HH ng nh 8 3 Density-of-state (DOS) effective mass -ccccssseeerey 9 4 Thermal COndUCfIVIẨY c + 31331119 1E EEsrieerreeereere 10 5 Optimization prObẽ€mm - <6 + 3x13 139v re 11 1.3 Thermoelectric ZnO Imiaf€TIèẽS - óc 11212119111 1 2 1 vn ng re 12 1.3.1 Crystalline Structure
From Equation (1.4), to optimize the ZT value, it is necessary to improve theSeebeck coefficient and electrical conductivity (power factor in general), and simultaneously reduce the total thermal conductivity as much as possible.
Electrical conductivity (c) is a fundamental parameter determining how easily electrons can flow through a substance o is directly proportional to carrier concentration (7) and Hall mobility (ux) as given by [1]: ỉ = nye (Eq 1.7)
The carrier concentration can be improved by doping foreign elements into the host structure For the n-type semiconductors, each dopant atom can substitute and donate one or several valence electrons to increase the carrier concentration On the other hand, dopants can also decrease carrier mobility due to enhanced electronic scatterings on the other electrons and the ionized dopant atoms Therefore, the optimized carrier concentration is found to be 10! — 107! em [8].
Seebeck coefficient (S) is a fundamental parameter characterizing the ability of a material to produce an electric voltage when subjected to a temperature gradient.
It quantifies the magnitude of the TE effect S is defined as the ratio of induced voltage (AV) to the temperature difference (AT) across the material, as given by:
Positive and negative values of S indicate the direction of the generated voltage concerning the temperature gradient It plays a crucial role in TE materials and devices aimed at harnessing waste heat and improving energy efficiency If S is negative, the majority carriers are electrons; otherwise, they are holes In fact, doping is found to enhance the Seebeck coefficient by modifying the electron density of state(DOS) Based on the single parabolic band assumption, the temperature-dependent Š value can be described through the Pisarenko relationship, as described by Pisarenko relationship [3,9]:
SE 3eh2 mar () where ma’ is the DOS effective mass It is seen that S is directly proportional to ma” and 1/n?3 According to Equations (1.7) and (1.9), S tends to be inversely proportional to o if ne is changed Thus, enhancing ZT by increasing ne-dependent PF is limited.
1.2.3.3 Density-of-state (DOS) effective mass
DOS effective mass is a fundamental parameter that describes the behavior of carriers, specifically electrons in a material It provides insight into the mobility and transport properties of charge carriers within a solid Furthermore, the DOS effective mass plays an important role in studying phenomena like band structure, carrier concentration, and energy dispersion Overall, understanding and manipulating the DOS effective mass is essential for advancing TE performance of materials and developing innovative technologies.
DOS effective mass is determined as ma” = Mv73mằ”, where Ny is the number of degenerate carrier pockets (valley degeneracy), and mp’ is the DOS effective mass of each pocket (average band mass) [10] It is well-known that engineering band structure can enhance Ny via multi-band convergence or mpằ via band curvature modification For ZnO, the band convergence has not been demonstrated due to Ny
= 1 In this case, the mp" enlargement is, therefore, proposed to be mainly responsible for the ma’ enhancement From the energy dispersion relation, the mp’ is described as the expression [11,12]: a2E(k)\ mỹ = r( 1o ) (Eq 1.10) where fi is the reduced Planck constant, E(k) and k are the energy dispersion function and wave vector in reciprocal space, respectively Higher zm” is associated with a small curvature of the conduction band (band flattening).
Total thermal conductivity (ict) is a fundamental parameter determining how quickly heat can pass through a material Understanding dot is essential for designing efficient heat conversion systems and optimizing TE properties.
Ktot = Kear + Kat = & X Cy xp (Eq 1.11) where a, Cp, and p are the thermal diffusivity, specific heat capacity, and density of materials [13] Another strategy to enhance ZT is reducing the total thermal conductivity Among them, Acar can be considered concerning ứ via the Wiedemann— Franz law [13]:
Kear = LOT (Eq 1.12) where L is the Lorentz number (2.45x103 WOK”) for electron as majority carrier
[14] Thus, ôcar also closely depends on the carrier concentration The larger the carrier concentration is, the more carriers transfer, and the higher ôcar obtains For example, the high free-electron concentration in metals results in large electrical and thermal conductivities.
On the other hand, ôia is known to be independent of o, which is associated with the lattice structure of a material, and generally contributes more than 90% of Kot in non-degenerate semiconductors [14,15] Consequently, controlling the thermal-conductivity components of TE materials has, therefore, emerged as an effective strategy for achieving good performance Nanostructure engineering and dopant addition are two approaches for reducing lattice thermal conductivity. Nanostructure engineering is performed to shorten the phonon mean free path or to enhance phonon scattering (dimensionality effect) in low-dimensional materials, such as quantum dots, nanowires, superlattices, and thin films Dopant addition reduces thermal conductivity by introducing scattering elements, including defects, nanoprecipitate secondary phases, distortions, and stacking faults.
Figure 1.2 Dependence of Seebeck coefficient S, electrical conductivity ứ, thermal conductivity x, power factor PF, and figure of merit ZT on carrier concentration n.
In brief, according to Equations (1.4), (1.7), (1.9), and (1.12), the complex trade-offs between ứ, xcar, and S are the biggest challenge for getting high- performance TE materials, due to their opposite coupling to carrier concentration, n. Their relationships are depicted in Figure 1.2 While o and xca are directly proportional to n, S is reversed Hence, it is hard to make a breakthrough in ZT by increasing n Instead, many efforts to boost ZT have focused on reducing x by impairing xia Component, or raising S by increasing density-of-state (DOS) effective mass, ma The reduction of ô1a can be obtained by nano-structuring [16-20], and point-defect engineering [21-23]; whereas the enhancement of ma’ can be conducted through band modification [4,24-26].
At room temperature, ZnO stably exists in the form of a hexagonal wurtzite structure, as shown in Figure 1.3 In this structure, each unit cell has two ZnO molecules, including two Zn atoms at Miller indices (0,0,0) and (1/3,2/3,1/2), and two O atoms at (0,0,u) and (1/3,2/3,1/2+u) with u = 3/5 The hexagonal structure of
ZnO is considered as two interlocking hexagonal lattices; one contains Zn** cations, and the other includes O* anions Each Zn atom bonds to four O atoms to form a tetrahedron, where an O atom is located at a distance of uxc, and the others are situated at [1/342 + c2 —1/2)?]!Z Lattice parameters of ZnO are a ~0.324 nm and c
~0.521 nm, leading to a ratio of c/a ~1.6 (approximately 1.633 of ideal hexagonal structure) The apices of the tetrahedron are oriented along the c-axis; thus, the c-axis is the anisotropic axis of the ZnO crystal.
Figure 1.3 Structural model of hexagonal wurtzite of ZnO material Adapted from the Ref [27].
Suitability for power-generation appẽ1CafIO'S - -‹ ô++ô<++<<+ 12 1.3.3 Research literature and Motivation 5 55 + ke 14
Experiments and metho(è0èOĐẽ€S s- <5 e< se< se ess=seesesse 25 2.1 5.) 200i 00v ốc
Materials and composition calculation -ô++-s<++s++se++ess+ 25 2.1.2 Preparation CQUIPMENE <1 E19 190 19 1v vn ve 27 2.1.3 Bulk material synthesis 718
Commercial ZnO (purity 99.9%, Merck, Germany), Ga2O3 (purity 99.99%, Sigma-Aldrich, USA), and In2O3 (purity 99.99%, Sigma-Aldrich, USA) powders were used as starting materials All calculations for the composition of the ZnO-based bulks are based on atomic percentages between dopants (In, Ga) and host Zn atoms. Table 2.1 explains how to calculate the needed mass of component oxide powders to synthesize bulk materials with a defined ratio of InxGao.os-xZno.05sO, where x is the atomic percentage of In; and the total percentage of dopants (In and Ga) is fixed at 5 at%.
In this thesis, the total mass of powders (rot) is fixed at 25 g for the Ino.00sGao.o4sZno.95O pellets investigated in Chapter 3 On the other hand, miot is fixed at 140 g for the InxGaoosxZnòzO bulks used as sputtering targets for thin-film synthesis, with varying x from 0 to 0.015 (corresponding to 0 — 1.5 at%), as discussed in Chapters 4 and 5 Table 2.2 details the needed mass of component oxide powders for each case.
Table 2.1 General calculation of the needed mass of component oxide powders for synthesizing IGZO bulks.
Oxide powders ZnO In2O3 Ga203
Atomic percentage a (at%) = xx100% 3 a sa
Mass of ZnO Mtot X 81.389 x 95 mzno (8) 81.389x95+277.637 x a/2+187.443x[(5-a) /2]
Mass of In2O3 Mtot X 277.637 x a/2 mm2oa (8) 81.389x95+277.637 x a/2+187.443x[(5-a)/2]
Table 2.2 Detailed calculation of the needed mass of component oxide powders for synthesizing IGZO bulks with different total masses.
Ino.00sGao.o4sZn0.950 InxGao.05-xZno.9sO pellets targets ) %¢ Mot = 140 g
Symbol of Mot samples X (Used in Chapter 3) (Used in Chapters 4, 5)
Mzno Mm203 Mca203 Mzno Im203 MGa203
To synthesize bulk materials for the oxide powders, some special-purpose machines and equipment are utilized, as listed in Table 2.3 All the machines and the synthesis processes were performed at Laboratory of Advanced Materials (AM-Lab), University of Science, VNU-HCM.
Table 2.3 List of special-purpose machines and equipment for the bulk process.
Vacuum drying oven (SHELLAB SL1410, England)
Pressing dies and punches (Homemade, Vietnam)
- Purpose: weighing component oxide powders for mixing.
- Milling jar (1000 cc capacity, including nylon cap and rubber seal) is made from alumina sintered at high temperature.
- Mixed milling balls (10 and 18.5 mm in diameter) are made from non-porous 90%
- Purpose: drying wet powders slurry.
- Size: diameter 200 mm x 40 mm height.
- Purpose: Compressing and shaping bulks.
- Square shape: inside size of 30x30 mm? (used for pellets).
- Round shape: inside diameter of 90 mm (used for sputtering targets).
- Purpose: controlling solid-state reactions for sintering bulk materials at high temperature.
- Purpose: preparing bulk samples with
USA) : appropriate sizes for measurements.
- Variable speed: 0 — 600 rpm Polishing machine - Sandpaper no P100 — P800 grit.
9 (Unipol-810, MTI Corp., - Purpose: polishing and removing
USA) contamination on the surface of pellets and sputtering targets.
- Purpose: grinding bulks to powders for measurements (XRD, XPS, DSC ).
The IGZO bulk materials were prepared via solid-state reactions using different preparation processes The raw powders were mixed to obtain Ga/In in a 4.5/0.5 ratio (at%) The wet raw mixture was ball-milled for 5 hours The slurry was carefully dried at 120°C for one day to remove water and divided into three portions: e One portion was hydraulically pressed into 30-mm-square pellers under a pressure of 14 MPa for about 2 minutes After that, the green compact body was annealed directly at 1400°C, then soaked for 3 hours (process 1) The IGZO bulk obtained via process 1 was designated as IGZO-1. e A separate green compact body obtained under the same pressing conditions was calcinated at 1000°C for 3 hours The calcined pellet was then ground using an agate mortar and pestle, pressed again, and sintered at 1400°C (process 2). The IGZO bulk created via process 2 with pressure-assisted conglomeration was designated as IGZO-2. e The third portion of the originally dried slurry was calcinated at 1000°C without pressure-assisted conglomeration Following the calcination of the powder mixture, the green compact body was sintered at 1400°C (process 3) and designated as IGZO-3.
Volumetric change (%) one-step preparation, and precursor powder preparation of two-step preparation, (c) the 2TM step — sintering of two-step preparation; (d) naked-eye photograph and (e) volumetric change of the IGZO bulks sintered by various processes.
The three IGZO bulk sintering processes are summarized in Figure 2.la-c A photograph of the three sintered IGZO bulk samples is shown in Figure 2.1d, and the volumetric changes due to sintering are compared in Figure 2.1e The largest and smallest shrinkages were observed in the IGZO-1 and IGZO-2 samples, respectively. Shrinkage was closely related to the characteristics of the IGZO bulk samples. Densification of the IGZO bulk samples, density, and hardness will be discussed in Chapter 3 An undoped ZnO bulk was also prepared via the one-step process (process
1) and used as a reference sample The thermal processes are found to play crucial roles in deciding the quality of sintered bulk materials All the processes were conducted in air, with a slow temperature acceleration rate of 2.5°C/min.
Figure 2.2 Temperature setpoints and hourly schedule for the thermal processes of bulk materials.
In the temperature schedule (Figure 2.2), the thermal processes can be divided into five steps:
(1) Heating from room temperature to 200°C for about 1 hour.
(2) At 200°C, holding the temperature for 2 hours to dry and remove water from the green compact body.
(3) Heating from 200°C to 1000°C for 5.3 hours (calcination process) or 1400°C for 8 hours (sintering process).
(4) At peak temperature of 1000°C or 1400°C, holding the temperature for 3 hours to sinter partially or fully bulk materials, respectively.
(5) Cooling from the peak temperature to room temperature.
Magnetron sputtering is a commonly used physical vapor deposition (PVD) technique employed in various industries, specifically manufacturing thin films for
-30- electronic devices and coatings This technique involves the thin-film deposition onto a substrate through the sputtering process, which is the ejection of atoms from a target (bulk) material due to energetic collisions with high-speed ions A vacuum chamber is utilized to create a low-pressure ambiance Inside the chamber, a target material is located on a cathode The cathode is surrounded by a magnetic flux from permanent magnets (magnetron), forming a magnetic field near the target surface This magnetic field confines electrons near the cathode, enhances the ionization of the Ar gas, and maintains the plasma environment.
Initially, Ar gas is introduced into the chamber The Ar molecules are ionized by applying a high voltage to the cathode, creating a plasma The ions in the plasma are accelerated toward the target surface, bombarding it with large energy As a result, atoms from the target material are dislodged and ejected into the chamber as charged particles These ejected particles move through the chamber and deposit onto a substrate opposite the target By adjusting various parameters such as the sputtering power, gas pressure, and substrate temperature, magnetron sputtering allows precise control over the film’s properties, including thickness, density, and composition.
Briefly, magnetron sputtering is a low-cost, versatile, and efficient technique for producing high-quality thin films with excellent control over film characteristics, high deposition rates, good adhesion, and large coating area.
The fabrication process of the sputtering targets was followed by the one-step sintering (process 1), as shown in Section 2.1.3 above The masses of component powders for the targets were calculated above in Table 2.2 After fabrication, the targets had a high relative density (>90%) compared to the theoretical density of ZnO
(5.606 g/cm?), 75 mm in diameter, and ~5 mm in thickness, as shown in Figure 2.3.
Figure 2.3 Homemade 3-inch targets with various compositions after many times of sputtering: (a)-(b) Representative green compact body prior to sintering, (c) ZnO, (d) GZO, (e) IGZO 0.3, (f) IGZO 0.5, (g) IGZO 1.0, and (h) IGZO 1.5.
All films were deposited using a commercial sputtering system (Leybold Univex-450, Germany) located at Laboratory of Advanced Materials (AM-Lab), University of Science, VNU-HCM, as pictured in Figure 2.4.
Figure 2.4 Sputtering system Leybold Univex-450 at AM-Lab.
All films were grown on Si(200) substrates at 300°C using DC magnetron sputtering The 3-inch sputtering targets were employed Pure ZnO target was used to prepare the pristine ZnO films, whereas the Ga-doped ZnO (GZO) and In, Ga co- doped GZO (IGZO) films were deposited from In;Gao,os.xZnoosOệ compound targets with varying x from 0 to 0.015 (corresponding to 0 — 1.5 at%) Before deposition, the substrates were plasma-treated under an Ar pressure at 1.2x10° torr to clean and strengthen film-substrate adhesion Besides, the targets were pre-sputtered for 10 minutes to remove surface contamination and stabilize deposition velocity The sputtering power and working pressure were maintained at 60 W and 3.5 mTorr, respectively, during the deposition process After deposition, to avoid oxidizing the films, the substrate temperature was naturally cooled down to lower than 100°C The film-on-substrate was removed from the vacuum chamber and stored in different boxes All condition parameters for thin-film deposition are detailed in Table 2.4.
Table 2.4 Deposition parameters for TE ZnO-based films in the thesis.
One-side polished un-doped Si(200) wafers, thickness ~700 um, resistivity >10* Ocm
5 Power source DC — 60 W, auto-stabilized power mode
9 Discharge cleaning Ar plasma of 1x10” torr, for 15 min
X-ray diffraction (XRD) is a powerful analytical technique used to study crystalline structure of materials By passing X-ray beams through the samples, the method allows to analyze the arrangement of atoms and determine their spatial distribution The interaction between X-rays and crystalline lattices results in a diffraction pattern, which can be measured and analyzed to deduce valuable information, such as unit cell dimensions, atomic positions, symmetry, crystal size, strain, and stress.
In this thesis, the crystalline structure and crystallographic data of the ZnO- based materials were obtained from a XRD apparatus (Bruker D8-Advanced, Japan), with an X-ray Cu-Ka source and the 0-29 geometry configuration The system is situated at Center for Innovative Materials and Architectures (INOMAR), VNU- HCM The XRD measurement parameters for the ZnO-based materials are listed in Table 2.5.
Table 2.5 XRD operating parameters used in the thesis.
I Sample forms Powder, film on substrate
3 X-ray wavelength 0.154 nm (Cu-Ka)
4 Power of X-ray source 1600 W (40 kV, 40 mA)
X-ray photoelectron spectroscopy (XPS) is a high-sensitivity and non- destructive surface analytical technique for studying the chemical states and elemental composition of materials By bombarding a sample with an X-ray beam, it generates photoelectrons whose kinetic energies are measured to determine the binding energies of core electrons Evaluating the binding energies allows for identifying the elements present and their oxidation states XPS provides valuable insights into the electronic structure and bonding environments of materials.
Table 2.6 XPS operating parameters used in the thesis.
1 Sample forms Powder, film on substrate
3 X-ray energy 1486.6 eV (Al-Ka)
6 Survey scan mode Step size: 1 eV
7 Narrow scan mode Step size: 0.1 eV
Electrical and thermopower Properties 5 S1 re, 75 3.7 Thermoelectric quality factor and figure Of T€TII - 5+ ô<< ô++s+++2 82 3.8 Brief CONCẽUSIOTNS Ghi 86
Because of the poor properties of the ZnO, analyses of TE properties focus on the IGZO bulks Figure 3.17 shows the electrical characteristics of the IGZO bulks.
A slight decrease in o of all the samples with increasing temperature was observed (Figure 3.17a) It means the o as a negative coefficient of temperature, indicating a degenerate semiconductor behavior The IGZO-1 bulk had much lower ứ than IGZO-
The variation in o originates from the different contributions of carrier concentration (ne) and mobility (un), which is described as o = netne (e is the elementary charge) The ne, Wu, and o values of the samples obtained from the Hall- effect measurement at room temperature were considered (Figure 3.17b) The results of o collected from the simultaneous-determination conductivity and Seebeck coefficient (ZEM-3) system, and the Hall-effect measurement were reasonable The IGZO-1 sample had the lowest ne due to the strong spinel segregation according to
Equations (3.4) and (3.5), while the good Zn substitution by the dopants was responsible for the highest ne of the IGZO-2 and IGZO-3 bulks, as illustrated in Equations (3.10) and (3.11).
Figure 3.17 Electrical characteristics of the IGZO bulks: (a) o as a function of temperature; (b) room-temperature Hall-effect measurement, the inset is a thin square sample with In electrodes at four corners for the measurement; and (c) ứ as a function of carrier concentration, the dashed lines are fittings to the experimental data.
The formation of the spinel phase increased Vzn concentration which captured electrons and decreased ne In contrast, the Zn substitution by the dopants could support free electrons for electrical conduction The variation in Vzn concentration was indicated by the PL analysis An opposite tendency of variation in zy compared to me was observed It is previously explained by the highest densification of the IGZO-1 bulk and the highest porosity of the IGZO-2 sample, as discussed in Section 3.2 Furthermore, in this study, the lowest “wu of the IGZO-2 bulk could be closely associated with the many Znj- and Vo-related scattering centers, which were demonstrated in Section 3.4 The mobility compensation was insufficient to enhance the o value of the IGZO-1 bulk The contribution of ne to the conductivity is shown in the plot of o vs ne (Figure 3.17c) The o values decreased with ne Hence, the conductivity of the samples mainly depended on ne, which was attributed to the competition between the spinel segregation and the Zn substitution by the dopants.
Figure 3.18 shows the thermopower characteristics of the IGZO bulks All the samples possessed negative S values (Figure 3.18a), indicating the n-type semiconductor characteristics The |S! values increased with temperature The IGZO-
1 bulk achieved the highest ISI, followed by the IGZO-3 and IGZO-2 samples It is explained in terms of the increased ne However, to evaluate carefully contributive factors to the ISI based on Equation (1.9), the variations in both ne and DOS effective mass were considered The effect of ne was investigated first The dependence of Š on temperature can be performed as the following equation [150,151]: se (5 +A) Eq 3.12 = leer (Eq 3.12) where Es = Ey — EF is the thermopower activation energy, defined as the energy difference between bottom conduction-edge energy (Er) and Fermi-level energy (Er), kg is the Boltzmann constant, e 1s the elementary charge, and A is a constant.
Figure 3.18 Thermopower characteristics of the IGZO bulks: (a) Š as a function of temperature; (b) plot of S vs 1000/7, the inset represents variation in activation energy —Es; (c) plot of S vs me, the dashed lines are fittings to the experimental data; and (d) temperature-dependent ne.
From Equation (3.12), the Es value could be determined from the linear fitting of plot S versus inverse temperature (Figure 3.18b) The IGZO-1 bulk had the highest
Es (14.87 meV), followed by the IGZO-3 (11.47 meV) and IGZO-2 (9.19 meV) samples The large activation energy suggests the Er shifting far away from the conduction band [151] Similarly, the low ne due to low donor concentration also led to less shifting of the Fermi level towards the conduction band of degenerate semiconductors Thus, the Es results were in line with the lowest and the highest me
- 78 - in the IGZO-1 and IGZO-2 samples, respectively From these arguments, the dependence of S on ne was confirmed through plotting S versus ne (Figure 3.18c) It is seen that the IS] decreased with ne attributed to the increased temperature To make this point clearer, ne of the bulks was measured as a function of temperature (Figure 3.18d) The decreased ne of all the bulks with increasing temperature is in line with the degenerate-semiconductor characteristics Furthermore, the decreased rate in the temperature-dependent 7e was the fastest for the IGZO-2 bulk, followed by the IGZO-
3 and IGZO-1 samples At low temperature (473 K), however, ne of the IGZO-1 bulk was approximately, even higher as compared to that of the two remained samples Thus, the high IS! of the IGZO-1 bulk at high temperature could be attributed to another factor, i.e the DOS effective mass.
Figure 3.19 Analyses on DOS effective mass ma/mo of the IGZO bulks: (a)
Temperature-dependent characteristics, the inset depicts the relationship between
DOS effective mass and unit cell volume at room temperature; and (b) plot of maẽ/mo
VS Ne, the fitted dashed lines are also drawn.
Using the Pisarenko relationship, the DOS effective mass of the samples could be simply estimated, as illustrated in Figure 3.19 The ma’/mo values decreased with
- 70 - increasing temperature (Figure 3.19a) It can be due to the decreased scattering rates, such as electron-electron and electron-donor scatterings [152,153] At low temperature (473 K) Park er al demonstrated that high ma mo was associated with enlarged unit-cell volume [154] Specifically, the enlarged volume due to increased atomic bond length decreased hybridization, resulting in a narrower band dispersion In this study, the In** has a larger ionic radius (0.080 nm) and a longer In-O bond length (0.216 nm) than those of Zn”! (0.074 nm and 0.199 nm, respectively) [155,156] Hence, the large unit-cell volume due to the good Zn** substitution by In** led to high ma’/mo of the IGZO-2 bulk In contrast, the poor Zn** substitution and the strong spinel segregation could give rise to low ma no in the
The relationship between ma /mo and unit-cell volume at the low temperature, typically at room temperature is shown in the inset graph Despite the low ma‘/mo, |S| of the IGZO-1 bulk still retained the highest, thus the contribution of ma’/mo to the ISI was minor At the high temperature, however, the ma‘/mo values of the IGZO-2 and
IGZO-3 samples reduced significantly and were lower than that of the IGZO-1 bulk.
The reason could be proposed that In*+ and Ga* could be removed from the substitutional sites, leading to relaxation of the unit-cell volume and lowering the ma mo value On the other hand, the reduction of dopant ions at the substitutional sites could be compensated by the migration of dopant ions from the spinel to ZnO phases under thermal energy Specifically, the M** ions at octahedral [MOs] sites in the spinel could diffuse to tetrahedral [MOa] sites in ZnO (where M = In or Ga), as mentioned in the XPS result (Section 3.3) and Ref [133] In fact, this process was more favorable to occur in the IGZO-1 sample due to the high available Vzn concentration which was produced much from the spinel formation as indicated in the PL analysis (Section 3.4).
Consequently, at the low temperature (-sSio.4Sno.6 [172-174] In addition, the Ga and In dopants randomly occupy the Zn sites in the lattice structure The electronegativity of Ga and In are close but not equal It means that electrons will experience a fluctuating potential on the atomic cores This fluctuating potential can reduce jun, leading to high ma’ Consequently, the IGZO 0.5 thin films have a higher increased rate of Seebeck coefficient than the GZO films.
The power factor (PF), which represents the electrical contribution to the TE performance, was calculated from the results in Figure 4.5a-b As shown in Figure 4.5c, PF of all the films increased with temperature The o at 573 K is 79.0, 2185.5, and 1131.6 S/cm for the pure ZnO, GZO, and IGZO 0.5, respectively The S at 573
K is -121.8, -30.9, and -56.7 wV/K for the pure ZnO, GZO, and IGZO thin films, respectively Consequently, the PF values at 573 K are 117.2, 208.1, and 363.2 uW/mK? for the pure ZnO, GZO, and IGZO thin films, respectively Among the three prepared thin films, the IGZO 0.5 thin films provide the largest power factor at higher temperatures because of the increase in the Seebeck coefficient and electrical conductivity This PF value is higher when compared to other oxides, such as the InGaO3(ZnO)m superlattice structure [76].
4.1.4 Thermal properties and thermoelectric performance
Properties of thermal conductivity (xe, xia, and Aor) of the ZnO, GZO, and IGZO 0.5 films, respectively, measured over the temperature range of 300 - 573 K, are presented in Figure 4.6.