Silicon-Rich Oxide Superlattices with High Thermoelectric Figure

Subscriber access provided by UNIV OF TEXAS DALLAS

Functional Nanostructured Materials (including low-D carbon)

ZnO/Silicon-Rich Oxide Superlattices with High Thermoelectric Figure of Merit: A Comprehensive Study by Experiment and Molecular Dynamic Simulation Hsuan-Ta Wu, You-Chun Su, Chun-Wei Pao, and Chuan-Feng Shih ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b20725 • Publication Date (Web): 12 Mar 2019 Downloaded from http://pubs.acs.org on March 12, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

ZnO/Silicon-Rich Oxide Superlattices with High Thermoelectric Figure of Merit: A Comprehensive Study by Experiment and Molecular Dynamic Simulation Hsuan-Ta Wu1, You-Chun Su1, Chun-Wei Pao2*, and Chuan-Feng Shih13* 1Department 2Research

of Electrical Engineering, National Cheng Kung University, No.1, University Road, Tainan, 70101, Taiwan Center For Applied Sciences, Academia Sinica, No.128, Academia Road, Section 2, Nankang, Taipei 11529,

Taiwan 3Hierarchical Green-Energy Materials (Hi-GEM) Research Center, National Cheng Kung University, No.1, University Road, Tainan 70101, Taiwan KEYWORDS: Thermoelectric Figure of Merit, Silicon-Rich Oxide, ZnO, Superlattices, Molecular Dynamics

ABSTRACT: ZnO is a direct band gap material that has numerous optoelectronic applications. Recently, the thermoelectric behavior of ZnO has drawn much attention because that is expected to enrich the multifunctional application of ZnO. However, the high thermal conductivity nature of ZnO (~50W/m·K) is a challenge to further increase its thermoelectrtic figure of merit (ZT). In this paper, a way to increase the ZT of ZnO thin films by insertion silicon-rich oxide (SRO) interlayers is reported. All of the constituents are earth-abundant and environmental friendly. The effects of the number of SRO layers, thickness, grain size, heat treatment and crystallinity of ZnO of the superlattices on the thermoelectric behaviors of ZnO were investigated. The thermoelectric ZT was determined by the transient Harman method by measuring the Seebeck voltage. The thermal conductivity of the ZnO/SRO superlattices that is crucial to elucidate the ZT behaviors is calculated using molecular dynamic simulation, in which the Zn−O and Zn−Zn interactions were described by the Born-Mayer potential and the short-range non-Coulombic O−O interaction was described by the Morse potential. For a given total ZnO/SRO thickness, the grain size of the ZnO decreases monotonically with increasing the number of SRO layers, thus leading to a decrease of the thermal conductivity and an increase of the ZT of the superlattices. As the best result, the annealed 45 nm-thick ZnO thin film with three SRO interlayers presents a high ZT~0.16 in room temperature. A comprehensive study on the ZnO/SRO superlattice-based thermoelectrtic devices was carried out by the experiment and theoretical simulation. The results imply potential thermoelectric application of the ZnO/SRO superlattices.

 INTRODUCTION Thermoelectric materials that convert thermal to electrical energy are getting important for sustainable environment and recycling of waste energy. The state-of-the-art Te-based superlattices have shown high thermoelectric figure of merit (>1)1-2. However, it is still urgent to search for alternatives because of the Te is rare and the use of environmental-polluted Pb is crucial for n-type Te-based alloys. ZnO is a earthabundant material and a direct band gap semiconductor that has been widely applied for various optoelectronic applications, such as thin-film transistor3-4, ultra-violet photodetector5-7, and transparent electrode8-9. Recently, ZnO thin film has been reported as a thermoelectric material with a room temperature (RT) figure of merit by appropriate Al doping (~2%)10. Because of the manifold applications of the ZnO, its promising thermoelectric properties has great potential for integration with the ZnO-based devices to demonstrate multi-functional applications. However, the high thermal conductivity nature of ZnO (~50W/m·K) is the main challenge to further improve its thermoelectric ZT. Three main approaches have been reported to improve the ZT values for thin-film thermoelectric materials. One is the use of

the quantum-confined structure that enhances the density of state near the Fermi energy and hence increase the carrier transport11-13; the second is the use of superlattices that reduce the carrier scattering14; and the third is the used of 2 heterostructures15. The ZT is defined as 𝑆 𝜎𝑇 𝜅, where S is the Seebeck coefficient, σ is the electrical conductivity, T is the temperature, and κ is the thermal conductivity. Therefore, phonon blocking interfaces are required for ZnO for thermoelectric applications. Several approaches have been adopted to enhance the ZT of ZnO. For example, decreasing the thickness of the Al-doped ZnO increased the ZT by decreasing the grain size and increasing the area of grain boundaries, yielding a ZT greater than 0.110. Inserting Al2O3 interlayers increased the RT ZT up to 0.1414. However, further increase the ZT of the ZnO is difficult, because the carrier transport through the highly-insulated Al2O3 interlayers is restarted. Additionally, increasing the conductivity of the ZnO by doping is limited by the activated dopant and the crystalline ZnO becomes amorphous when the grain size is further reduced that greatly degrades the thermoelectric properties. Silicon, the most earth abundant material, has a high thermal conductivity (>150 W/m·K), high Seebeck coefficient, and tunable conductivity. However, the ZT value is low (ZT~0.01)

ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

due to the long phonon mean free path of transport. Boukai reports that the Si nano-wires and nanocrystals embedded within oxide matrix can effectively reduce the κ16. The ZT of polycrystalline Si can be increased by reducing the grain size17. The dominant mechanism of the improvement is that the carriers transport between the Si nanostructures can be enhanced by the quantum mechanical tunneling that avoids the scattering effect at the interfaces. High ZT has been achieved in SiO2/SO2+Ge superlattices with ion implanted Si15. In this research, Si-rich oxide (SRO) was used as the phonon blocking interlayers to increase the ZT of ZnO films. The SiO2/SRO superlattices have been demonstrated potential applications for lighting devices and memory devices18-19, Based on the previous experience, the electrons are expected to tunnel through the quantum barriers in the ZnO/SRO superlattices in a low driving voltage. As a result, high ZT values were obtained in the ZnO/SRO superlattices. The 45 nmthick ZnO with 3 SRO interlayers presents a ZT~0.16 in room temperature. A comprehensive study on the ZnO/SRO superlattices based thermoelectric devices is presented both by experiment and simulation.

Page 2 of 7

Figure 1. Illustration of ZnO/SRO multilayered thermoelectric devices with (a) 0 (b) 1(c) 2 (d) 3 Si-rich oxide interlayers.

 EXPERIMENTAL PROCEDURE 0.8 μm-thick Mo was deposited on a 2 cm × 2 cm glass as the bottom electrode. The ZnO/SRO superlattices were deposited on the Mo substrate using two sputtering guns alternatively. The ZnO was deposited by the 100 W RF magnetron sputtering process and the SRO was deposited by the 160 W pulsed DC magnetron sputtering. The flow rate of the working gas argon was 12 sccm and the pressure during deposition was 2E-3 torr. ZnO/SRO multilayers with two different total thin-film thickness were prepared (100 nm and 45 nm), in which the thickness of the SRO layer was identical (5 nm). The SRO layer divided the ZnO layer into equal segments, so the thickness of each ZnO layer decreased as the number of the SRO increased. For example, the thickness of each of the two ZnO layers in a 45 nm-thick films with one SRO layer was 20 nm. The illustration of the device structures were shown in Figure 1. Finally, the films were annealed at 400 °C for 1 hour in N2+H2 ambient (N2:95%; H2:5%) using a horizontal furnace. Composited metal electrode composed of Cr/Al/Ti/Pt/Au (1.8nm/250nm/5nm/40nm/1810 nm) was deposited on the top of the films by electron-beam evaporation. The thickness and microstructure were identified by the high-resolution transmission electron microscopy (HRTEM) and scanning electron microscopy (SEM). The crystallinity of the ZnO was characterized by x-ray diffraction patterns (XRD). ZT values were analyzed by a source meter (Keysight B2962A) using transient Harman method20. The measurement approach and setup follows Juang’s report21, which shows that the ZT can be directly measured by applying a transient direct current vertically across the device and recording the voltage-to-time response. The electrical conductance was extracted from the stable region between the rise and fall regions of the voltage versus time profile for transient Harman measurement. The positive voltage was biased on the top electrode, and the injection current was chosen as small as possible to minimize the joule heating effect.

Figure 2. Cross-section TEM images of the ZnO/SRO multilayers with (a) 0, (b) 1, (c) 2 and (d) 3 Si-rich oxide interlayers. (scale bar: 20 nm)

 EXPERIMENTAL RESULTS Material properties of ZnO/SRO superlattices. Figure 2 shows the cross-section TEM images of the as-prepared ZnO/SRO superlattices with (a) 0, (b) 1, (c) 2, and (d) 3 SRO interlayers. The SRO layers can be clearly observed with a brighter contrast than the ZnO layers, and the interfaces between them were smooth. All of the SRO layers were found amorphous without nano-crystallites because the heat-treatment temperature (400 °C) was lower than their phase separation temperature around 700°C22. Contrarily, the ZnO layers were crystalline. Figure 3 shows the XRD patterns of the as-prepared (Figure 3(a)) and annealed (Figure 3(b)) ZnO/SRO superlattices with an identical total thickness of 45 nm. It was found that the intensity of the ZnO XRD peaks were increased by the annealing but decreased by increasing the number of the SRO layers. For the ZnO/SRO superlattices with more than 2 SRO interlayers, the XRD peaks were very weak. Using the XRD data in Figure 3, Figure 4 plots the grain size (D) of the ZnO calculated by the Scherrer’s equation (D = 0.9𝜆 𝛽cos 𝜃), where β is the full-width at half maximum of the XRD peaks of ZnO (002), θ is the Bragg’s angle, and λ is the wavelength of the xray. It was found that the grain size of the ZnO in the superlattices decreased with the number of SRO layers as a result of decreasing the thickness of each ZnO layer. The grain size of ZnO was increased by annealing markedly, and the grain


2

Page 3 of 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces size of the ZnO of the 100 nm-thick samples was larger than that of the 45 nm-thick samples. The grain size of the ZnO of the 45 nm samples decreased rapidly by increasing the number of SRO layers; while the grain size of the 100 nm samples was less influenced by that.

caused by measurement environment, and large joule heat that was injected to the thin film with a small heat capacity. In our case using the Harman method, the Seebeck voltage (Vα) can be resolved in both the ramping up and ramping down sides in the time-dependent voltage profile. Figure 5 shows the transient voltage response of the Herman measurement for the 45 nm and 100 nm thick supaterlattices with and without annealing. Because the curves were very symmetry in rising and falling sides, only the falling side data was shown in the figure. Actually the data in falling side is more frequently used than the rising side in literature because before stopping injecting current the sample has achieved the thermal equilibrium. Initially, the linear voltage drops (VR) as soon as the injection current was off followed the ohm’s law. Then, the voltage drop caused by the Seebeck effect turned to exponentially decay. A threshold voltage was determined by the intersection of the linear fitting and exponential fitting curves from the two regions, labeled as a solid diamond in the figure. The ZT value was calculated by 𝑉𝛼 𝑉𝑅. Obviously, the Vα was increased by annealing and increasing of the number of SRO layers.

Figure 3. XRD patterns of the (a)as-prepared and (b) annealed ZnO/SRO multilayers with various number of SRO interlayers. Thickness of the total superlattices is 45 nm.

Figure 5. Time-dependent voltage response measured using Harman method. (a) and (b) show 45 nm-thick ZnO/SRO superlattices with and without annealing, respectively. (c) and (d) show 100 nm-thick ZnO/SRO superlattices with and without annealing, respectively.

 MOLECULAR DYNAMICS SIMULATIONS

Figure 4. Grain size of ZnO in ZnO/SRO multilayers determined using XRD patterns and Scherrer’s equation. The thickness of the ZnO/SRO multilayers were 45 nm and 100 nm, in which the thickness of the SRO is 5 nm. The annealing temperature is 400 °C.

Thermoelectric properties. Figure 5 shows the thermoelectric properties of the 45/100 nm-thick ZnO/SRO superlattices with and without annealing measured by the transient Harman method. The Harman method has been reported as a simple and accurate way to obtain the ZT of thinfilm and 2-D materials21, 23 because the ZT can be measured directly. For traditional method, the ZT was calculated by substituting Seebeck coefficient, thermal conductivity, and 2 electric conductivity into the equation ZT = 𝑆 𝜎𝑇 𝜅, but those data were measured separately. Therefore, errors could be

One primary factor influencing ZT value of thermoelectric material is the thermal conductivity. The thermal conductivity of the thermoelectric material should be as low as possible. To examine the effects of insertion of thin phonon scattering layers on the thermal conductivities of the superlattice, we performed a series of molecular dynamics (MD) simulations to compute the thermal conductivities of the multilayers instead of measurement using optical method such as time-domain thermoreflectance24. Note that in MD simulations silicon were used as the material of the phonon scattering layer instead of the SRO because it is difficult to accurately determine the detailed structure and composition of oxygen in the SRO layer (we only know the SRO layer is Si-rich); nonetheless, this will not alter the conclusion of the present study since the primary objective of MD simulations is to verify the effects of phonon scattering layer on thermal conductivities of the superlattice. Furthermore, the SRO in general has lower thermal conductivities than Si crystal, implying that the MD simulations provide an upper


3


bound on the superlattice thermal conductivities and further enhancement of ZT value is anticipated by replacing silicon with SRO. Construction of the atomic model of ZnO, Si, and ZnO/Si multi-layer simulation supercells. The wurtzite ZnO (WZZnO) was chosen for thermal conductivity calculation. We employed the potential parametrized by Wang for molecular dynamics simulations25. In this potential, the short-range Zn−O and Zn−Zn interactions were described by the Born-Mayer potential: ― 𝑟𝑖𝑗

𝜑(𝑟𝑖𝑗) = 𝐴𝑒

(1)

𝜌

The short-range non-Coulombic O−O interaction was described by the Morse potential 𝜑(𝑟𝑖𝑗) = 𝐴[(1 - 𝑒

-𝐶(𝑟𝑖𝑗 ― 𝜌) 2

) - 1)]

(2)

All the potential parameters were compiled in Table 1. A cutoff radius of 12.0 Å was set for all short-range nonCoulombic interactions, and the Ewald method was used to calculate the long-range Coulombic interactions. The Tersoff potential were employed to describe the Si-Si interactions. Finally, the Lennard-Jones potential was employed for the ZnSi interaction: 𝐸 = 4𝜖

𝜎 12

𝜎 6

𝑟

𝑟

[( ) ― ( ) ],𝑟 < 𝑟

(3)

𝑐

where rc is cutoff radius, and ϵ and σ denote the depth and location of the potential well, respectively. The parameters are set as 𝜖𝑍𝑛 ― 𝑆𝑖=3.442 meV, 𝜎𝑍𝑛 ― 𝑆𝑖=3.001 Å to ensure stable interfaces. The cutoff distance for interatomic interaction is set to be 10.0 Å. Table 1. Potential parameters for atomistic calculations. Effective charge

Short-range interactions Intera ction Zn-Zn Zn-O O-O

Type BornMayer BornMayer Morse

A(eV)

ρ(Å)

78.91

0.5177

257600

0.1396

0.1567

3.405

C(Å-1)

q(e)

±1.14 1.164

Calculations of thermal conductivity of superlattices. The thermal conductivity was computed by reverse non-equilibrium molecular dynamics (RNEMD) method proposed by MullerPlathe26. Figure 6(a) displays system setups for thermal conductivity computation. The upper panel of Figure 6(a) displays typical atomistic systems for RNEMD simulations. The whole system was divided into N slabs for kinetic energy swapping as well as computing slab temperature distributions, see the lower panel of Figure 6(b). The slab indexed as N/2 was set as the hot area and slab indexed as 0 was set as the cold area. In the RNEMD approach, the atoms with the lowest kinetic energy in slab N/2 (namely, the “coldest” atom in slab N/2) was exchanged with those with the highest kinetic energy in slab 0. This kinetic energy exchange process introduced temperature gradient (shown at Figure 6(c)) and the heat flux can be simply expressed as the amount kinetic energies exchanged during simulation. Once obtaining heat flux Δ𝑄/ΔT and temperature gradient ΔT/Δ𝑍, the thermal conductivity can be computed according to the Fourier’s law:

∆𝑄

Page 4 of 7

∆𝑍

(4)

к = 2𝐴𝑟𝑒𝑎∆𝑡∆𝑇

where Δ𝑄/ΔT is heat flux, A is area of simulated model , ΔT/Δ𝑍 is reciprocal of temperature gradient. In carrying out RNEMD simulations, a time step of 0.1 fs was used. The system was first pre-optimized as a canonical ensemble (NVT) for 1 ns and then switched to a microcanonical ensemble (NVE) for 10 ns. To eliminate the potential system drifts during MD simulations, the center-of-mass momentum was rescaled to zero every 10 steps. We began with computing the thermal conductivities of ZnO and Si to validate the force field. To take finite size effects into account 27, we computed the thermal conductivities of both ZnO and Si systems of different thickness Lz (namely, cell size along the horizontal direction in Figs. 6 and 7), and extrapolated the computed thermal conductivities to Lz=∞ as the bulk thermal conductivities for direct comparison with experiments. Figure S1(a) and (b) show the atomic model of ZnO and Si, respectively. The thermal conductivities of ZnO and Si models with different simulation cell size in the Z direction LZ were compiled in Table 2. In Table 2, it is evident that thermal conductivities of both ZnO and Si are correlated with cell size Lz, which is in agreement with REF 27. Furthermore, we also compared the thermal conductivities of Si from MD simulations with those of polycrystalline Si of similar thickness measured from experiments (REF. 17). MD simulation results are in reasonable agreements with experimental measurements (~15 and ~30 w/m·k for 100 and 200 nm thick films, respectively), and the discrepancy can be attributed to the existence of high density of grain boundaries and defects. We then plotted the thermal conductivities with respect to inverse of simulation cell size (namely, 1/LZ) (see Figure S1(b) for details). The thermal conductivity of ZnO in the bulk can be obtained by extrapolated the curve in Figure S1(b) to 0 (corresponding to LZ = ∞). The estimated thermal conductivity of bulk ZnO is 52.7 W/m ·K at 300 K, which is close to the bulk ZnO (~50 W/m·K). The atomistic model used for computing thermal conductivity of Si is displayed in Figure S1(c). Similar to ZnO, we computed the thermal conductivities of systems with different simulation cell size in the Z direction and plotted the computed thermal conductivities with respect to 1/LZ (see Figure S1(d) for details), and obtained the bulk thermal conductivity by extrapolating to 0. The thermal conductivity of bulk Si was 153.14 W/m·K by extrapolation and it is closed to the thermal conductivity of bulk Si ~150 W/m·K. Hence, the force field employed for ZnO and Si are robust since the bulk thermal conductivities of bulk ZnO and Si extracted from molecular dynamics simulations are in good agreements with experiments. Table 2. Calculated thermal conductivity with different thickness of ZnO and Si κ(W/m·K) Thickness(nm)

ZnO

Si

34

18.8

14.2

45

22.3

18.3

100

32.7

35.4

220

41.2

60.7


4

Page 5 of 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces Table 3. RNEMD simulation results of κ of ZnO/Si multilayers

Figure 6. (a) RNEMD simulation system setups; (b) heat flux(∆Q/∆t) (c) temperature gradient(∆T/∆Z) of ZnO model.

The atomic models of ZnO with one, two, and three SRO layers (~5 nm) were built, see Figure 7(a), (b) and (c), respectively. The thickness of the ZnO/Si superlattice slab for thermal conductivity calculation (namely, half of the simulation cell in the Z direction) in all cases was fixed at 45nm.

Figure 7. Atomistic simulation models of (a) one (b) two and (c) three Si interlayers in ZnO at 300K. Green: Si; Blue and red: ZnO.

 DISCUSIONS The thermal conductivities of ZnO/Si multilayers with 1-3 Si layers were simulated. The detailed simulation results can be found in Figure S2. The results were summarized in Table 3. It was found that the thermal conductivities were around 1.9-2.6 W/m·K, which is much smaller than those of ZnO (22.34 W/m·K) and Si (18.26 W/m·K) with identical slab thickness. The significant decrease in thermal conductivities of ZnO/Si multilayer superlattice can be attributed to the phonon scattering effect at ZnO/Si interfaces.

Si layer 1

ΔQ/Δt (W) 1.19x10-7

A (m2) 3.11x10-18

ΔT/ΔZ (K/m) 7.4x109

κ (W/m·K)

2

1.1x10-7

3.11x10-18

8.6x109

2.1

3

1.2x10-7

3.11x10-18

1x1010

1.9

2.6

Figure 8(a) plots the ZT values and thermal conductivities as functions of the thickness of ZnO. It was found that the ZT values increased with decreasing thermal conductivity and ZnO layer thickness. The decrease of thermal conductivities with decreasing ZnO slab thickness can clearly be attributed to the phonon scattering at interfaces10. The enhancement of ZT with decreasing grain sizes has been reported in some nanocrystalline material system28. Figure 8(b) plots the ZT values and thermal conductivities as functions of the number of SRO layers of the 45 nm-thick superlattices. The ZT increased monotonically with the number of SRO layers. Variation of ZT for the ZnO/SRO superlattices with different total thickness show similar trend, as shown in Figure S3. However, the thermal conductivity of the 45 nm thick sample decreased markedly when one SRO interlayer was inserted, and decreased slightly when more SRO interlayers were added. Therefore, the increase of ZT when the SRO layers exceeds two cannot be explained by the variation of thermal conductivity. Instead, the reduction of grain size is a more dominate factor than the thermal conductivity. Moreover, annealing also further increased the ZT. Although the annealing increased the grain size, the elimination of defects and improvement of the film and interfacial quality should also contribute to the ZT by increasing electrical conductivity. This has been approved by the conductance measurement, as shown in Figure S4. The conductance was doubled by the annealing. Note that the electrical conductance increased slightly but decreased with the number of SRO layers for the as-prepared and annealed SRO films, respectively. The fact could be understood by the following effects. Since the as-prepared SRO has lots of point defects , the conductivity increased with the amount of SRO layers. On the other hand, annealing increased the conductivity of ZnO but decreased the conductivity of SRO by passivation, therefore the conductivity varied in the opposite direction. As a best result, the annealed 45 nm thick ZnO/SRO superlattices with three SRO interlayers has a high ZT value of 0.16. This value is close to the Al-doped ZnO and much higher than the related material systems. Figure 8(c) shows the ZT of ZnO/SRO superlattices with different total thickness, i.e. different ZnO thickness, as functions of the number of SRO layers. Similarly, annealing and increasing the number of SRO layers increased the ZT of the 100 nm-thick superlattices. The ZT of the 45 nmthick samples were larger than that of the 100 nm-thick samples in the whole region, indicating again the reduction of grain size of ZnO a dominate factor of the ZT. Table S1 shows the ZT vales for the superlattices with different total thickness and number of SRO layers. Noted that the simulation of the thermal conductivity of the superlattices used Si instead of the SRO. The reason is that to build a SRO model that exactly matched to the experimental SRO layer is very difficult. On the other side, it was also hard to prepare a pure Si layer without oxidation in the sputtering system. In fact, the Si content within the SRO layer was ~1.4


5


(Si/O ratio). The influence of the oxide and charged defects were not concerned in the simulation. Finally, it is expected that the ZT of ZnO using the proposed superlattices structure can be further upgraded by utilizing doping to increase the conductivity of ZnO and Si. Moreover, enhancing the appearance of Si nanocrystals by hydrogen reduction is also possible to increase the ZT by introducing tunneling path for electrons and phonon blocking interfaces. Because the polarity of the Si can be controlled by doping, it is possible to fabricated both n-type and p-type thermoelectric devices based on the ZnO/SRO structure. The process is capable of integrated with the large-scale Si technology.

Page 6 of 7

the Mo glass substrate by rf-sputtering. SRO has a ~5 nm thickness and various SRO layers were inserted into the ZnO with equidistance. It was found that for a given ZnO/SRO thickness, the grain size of the ZnO decreases with increasing the number of the SRO layers. Annealing increased the ZnO grain size largely. The ZT values increased with the number of the SRO interlayers. The atomic model of ZnO, Si, and ZnO/Si multi-layer simulation supercells were build. The thermal conductivity was computed by RNEMD method. The calculation results indicated that the thermal conductivity decreased markedly when one SRO interlayer was inserted, and then decreased slightly when more SRO interlayers were added. The calculated thermal conductivity was ~1.59 (W/m K), while a high ZT of 0.16 measured by the transient Herman method was obtained for the 45 nm ZnO/SRO superlattices with 3 SRO interlayers.

SUPPORTING INFORMATION Atomic model and к simulation of Si and ZnO, temperature gradient and heat flux in ZnO with Si interlayers, ZT for the ZnO/SRO superlattices, electrical conductance of ZnO/SRO superlattices, and summary of ZT.

ACKNOWLEDGMENT The authors are grateful for the support of the Ministry of Science and Technology of the Republic of China under Contract No. MOST 105-2221-E-006-220-, MOST 106-2221-E-006-225- and Instrument Center, National Cheng Kung University, Taiwan.

REFERENCES

Figure 8. ZT and simulated thermal conductivity as functions of (a) thickness of ZnO and (b) number of Si interlayers. (c) ZT as functions of number of Si interlayers and total thickness of ZnO/SRO superlattices. The total thickness of the superlattices in (b) is 45 nm.

 CONCLUSION The thermoelectric properties of ZnO/SRO supterlattice were studied. Experimentally, the ZnO thin films were prepared on

(1) Ettenberg, M. H.; Jesser, W. A.; Rosi, F. D. A New N-type and Improved P-type Pseudo-ternary (Bi2Te3)(Sb2Te3)(Sb2Se3) Alloy for Peltier Cooling. Fifteenth International Conference on Thermoelectrics. Proceedings ICT '96 1996; 52-56, DOI: 10.1109/ICT.1996.553255. (2) Sales, B. C.; Mandrus, D.; Williams, R. K. Filled Skutterudite Antimonides: A New Class of Thermoelectric Materials. Science 1996, 272 (5266), 1325-1328, DOI: 10.1126/science.272.5266.1325. (3) Masuda, S.; Kitamura, K.; Okumura, Y.; Miyatake, S.; Tabata, H.; Kawai, T. Transparent Thin Film Transistors Using ZnO as an Active Channel Layer and Their Electrical Properties. Journal of Applied Physics 2003, 93 (3), 1624-1630, DOI: 10.1063/1.1534627. (4) Carcia, P. F.; McLean, R. S.; Reilly, M. H.; Jr., G. N. Transparent ZnO Thin-film Transistor Fabricated by RF Magnetron Sputtering. Applied Physics Letters 2003, 82 (7), 1117-1119, DOI: 10.1063/1.1553997. (5) Liang, S.; Sheng, H.; Liu, Y.; Huo, Z.; Lu, Y.; Shen, H. ZnO Schottky Ultraviolet Photodetectors. Journal of Crystal Growth 2001, 225 (2), 110-113, DOI: 10.1016/S0022-0248(01)00830-2. (6) Zhou, J.; Gu, Y.; Hu, Y.; Mai, W.; Yeh, P.-H.; Bao, G.; Sood, A. K.; Polla, D. L.; Wang, Z. L. Gigantic Enhancement in Response and Reset Time of ZnO UV Nanosensor by Utilizing Schottky Contact and Surface Functionalization. Applied Physics Letters 2009, 94 (19), 191103, DOI: 10.1063/1.3133358. (7) Chen, K. J.; Hung, F. Y.; Chang, S. J.; Young, S. J. Optoelectronic Characteristics of UV Photodetector Based on ZnO Nanowire Thin Films. Journal of Alloys and Compounds 2009, 479 (1), 674-677, DOI: 10.1016/j.jallcom.2009.01.026. (8) Nanto, H.; Minami, T.; Shooji, S.; Takata, S. Electrical and Optical Properties of Zinc Oxide Thin Films Prepared by RF Magnetron Sputtering for Transparent Electrode Applications. Journal of Applied Physics 1984, 55 (4), 1029-1034, DOI: 10.1063/1.333196. (9) Schropp, R. E. I.; Madan, A. Properties of Conductive Zinc Oxide Films for Transparent Electrode Applications Prepared by RF Magnetron Sputtering. Journal of Applied Physics 1989, 66 (5), 20272031, DOI: 10.1063/1.344341.


6

Page 7 of 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces (10) Loureiro, J.; Neves, N.; Barros, R.; Mateus, T.; Santos, R.; Filonovich, S.; Reparaz, S.; Sotomayor-Torres, C. M.; Wyczisk, F.; Divay, L.; Martins, R.; Ferreira, I. Transparent Aluminium Zinc Oxide Thin Films With Enhanced Thermoelectric Properties. Journal of Materials Chemistry A 2014, 2 (18), 6649-6655, DOI: 10.1039/C3TA15052F. (11) Hicks, L. D.; Dresselhaus, M. S. Effect of Quantum-well Structures on the Thermoelectric Figure of Merit. Physical Review B 1993, 47 (19), 12727-12731, DOI: 10.1103/PhysRevB.47.12727. (12) Hicks, L. D.; Dresselhaus, M. S. Thermoelectric Figure of Merit of a One-dimensional Conductor. Physical Review B 1993, 47 (24), 16631-16634, DOI: 10.1103/PhysRevB.47.16631. (13) Hicks, L. D.; Harman, T. C.; Dresselhaus, M. S. Use of Quantum ‐ well Superlattices to Obtain a High Figure of Merit from Nonconventional Thermoelectric Materials. Applied Physics Letters 1993, 63 (23), 3230-3232, DOI: 10.1063/1.110207. (14) Lee, J.-H.; Park, T.-H.; Lee, S.-H.; Park, N.-W.; Lee, W.-Y.; Lim, J.-H.; Lee, S.-K.; Park, J.-S. Enhancing the Thermoelectric Properties of Super-lattice Al2O3/ZnO Atomic Film via Interface Confinement. Ceramics International 2016, 42 (13), 14411-14415, DOI: 10.1016/j.ceramint.2016.06.032. (15) Budak, S.; Parker, R.; Smith, C.; Muntele, C.; Heidary, K.; Johnson, R. B.; Ila, D. Superlattice Multinanolayered Thin Films of SiO2/SiO2+Ge for Thermoelectric Device Applications. Journal of Intelligent Material Systems and Structures 2013, 24 (11), 1357-1364, DOI: 10.1177/1045389x13483022. (16) Boukai, A. I.; Bunimovich, Y.; Tahir-Kheli, J.; Yu, J.-K.; William A. Goddard , I.; Heath, J. R. Silicon Nanowires as Efficient Thermoelectric Materials. In Materials for Sustainable Energy; pp 116119. (17) Valalaki, K.; Vouroutzis, N.; Nassiopoulou, A. G. Significant Enhancement of the Thermoelectric Figure of Merit of Polycrystalline Si Films by Reducing Grain Size. Journal of Physics D: Applied Physics 2016, 49 (31), 315104, DOI: 10.1088/00223727/49/31/315104. (18) Fu, S.-W.; Chen, H.-J.; Wu, H.-T.; Chuang, B.-R.; Shih, C.-F. Effect of Hydrogen Ion Beam Treatment on Si Nanocrystal/SiO2 Superlattice-based Memory Devices. Applied Surface Science 2016, 367, 134-139, DOI: 10.1016/j.apsusc.2016.01.076. (19) Fu, S.-W.; Chen, H.-J.; Wu, H.-T.; Chen, S.-P.; Shih, C.-F. Enhancing the Electroluminescence Efficiency of Si NC/SiO2 Superlattice-based Light-emitting Diodes through Hydrogen Ion Beam Treatment. Nanoscale 2016, 8 (13), 7155-7162, DOI: 10.1039/C5NR08470A. (20) Harman, T. C. Special Techniques for Measurement of Thermoelectric Properties. Journal of Applied Physics 1958, 29 (9), 1373-1374, DOI: 10.1063/1.1723445. (21) Juang, Z.-Y.; Tseng, C.-C.; Shi, Y.; Hsieh, W.-P.; Ryuzaki, S.; Saito, N.; Hsiung, C.-E.; Chang, W.-H.; Hernandez, Y.; Han, Y.; Tamada, K.; Li, L.-J. Graphene-Au Nanoparticle Based Vertical Heterostructures: A Novel Route towards High-ZT Thermoelectric Devices. Nano Energy 2017, 38, 385-391, DOI: 10.1016/j.nanoen.2017.06.004. (22) Chen, X. Y.; Lu, Y. F.; Tang, L. J.; Wu, Y. H.; Cho, B. J.; Xu, X. J.; Dong, J. R.; Song, W. D. Annealing and Oxidation of Silicon Oxide Films Prepared by Plasma-enhanced Chemical Vapor Deposition. Journal of Applied Physics 2005, 97 (1), 014913, DOI: 10.1063/1.1829789. (23) Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, B. Thin-Film Thermoelectric Devices with High Room-temperature Figures of Merit. Nature 2001, 413, 597-602, DOI: 10.1038/35098012. (24) Cahill, D. G.; Ford, W. K.; Goodson, K. E.; Mahan, G. D.; Majumdar, A.; Maris, H. J.; Merlin, R.; Phillpot, S. R. Nanoscale Thermal Transport. Journal of Applied Physics 2003, 93 (2), 793-818, DOI: 10.1063/1.1524305. (25) Wang, S.; Fan, Z.; Koster, R. S.; Fang, C.; van Huis, M. A.; Yalcin, A. O.; Tichelaar, F. D.; Zandbergen, H. W.; Vlugt, T. J. H. New Ab Initio Based Pair Potential for Accurate Simulation of Phase Transitions in ZnO. The Journal of Physical Chemistry C 2014, 118 (20), 11050-11061, DOI: 10.1021/jp411308z. (26) Müller-Plathe, F.; Bordat, P. Reverse Non-equilibrium Molecular Dynamics. In Novel Methods in Soft Matter Simulations;

Karttunen, M.; Lukkarinen, A.; Vattulainen, I., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg, 2004; pp 310-326. (27) Schnell, S. K.; Vlugt, T. J. H. Thermal Conductivity in Zeolites Studied by Non-equilibrium Molecular Dynamics Simulations. International Journal of Thermophysics 2013, 34 (7), 1197-1213, DOI: 10.1007/s10765-013-1467-2. (28) Nakamura, Y.; Isogawa, M.; Ueda, T.; Yamasaka, S.; Matsui, H.; Kikkawa, J.; Ikeuchi, S.; Oyake, T.; Hori, T.; Shiomi, J.; Sakai, A. Anomalous Reduction of Thermal Conductivity in Coherent Nanocrystal Architecture for Silicon Thermoelectric Material. Nano Energy 2015, 12, 845-851, DOI: 10.1016/j.nanoen.2014.11.029.

TOC Graphic


7

Silicon-Rich Oxide Superlattices with High Thermoelectric Figure

Recommend Documents