Thermal transport in ZnO nanocrystal networks synthesized by nonthermal plasma

Semiconductor materials with independently controlled electrical and thermal properties have a unique promise for energy-related applications from thermoelectrics and thermophotovoltaics. Here, using nonthermal plasma synthesized, direct-contact zinc oxide (ZnO) nanocrystal (NC) networks infilled with amorphous $\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$, and amorphous ZnO-$\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$ mixture, it is shown that such independent control of electrical and thermal properties is achievable. In this study, in addition to our early reports on control of the electrical properties in these two-phase nanocomposites by tailoring the contact radius between NCs, we demonstrate that the infill composition has a significant impact on the overall thermal conductivity of the NC network and can be used for thermal control. It is also shown that in these heterogeneous systems, the phonons are the dominant heat carriers, and the NC-NC contact radius has a negligible effect on thermal transport. The work suggests that this paradigm of independently controlling the electrical and thermal properties of NC-based materials through tuning the NC-NC contact radius and infill composition can be exploited even further by varying NC and infill materials with potential applications ranging from solar cells and light emitting diodes to solid-state energy converters.

Figure 1

(a) Schematic of ZnO NC networks for thermal characterization. A metal transducer layer (Al) is deposited on the top surface of each sample for time-domain thermoreflectance measurement. (b) Cross-sectional sketch of two NCs in contact through facets with radius ρ. ZnO and $\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$ are coated on NC surface from inside to outside in sequence by ALD to serve as infill materials. The interface between the NC and infill material is highlighted as orange. The nanocrystal diameter $d$ is controlled to be ∼10 nm.

Figure 2

Structural and surface characterization of ZnO NC networks: (a) A high-magnification HAADF-STEM image of sample 3-b along with atomic-resolution image of the highlighted (framed) region, which shows the structure of the interface between two NCs. Scale bars are 5 nm. (b) HAADF-STEM image and composite STEM-EDX map of the ZnO NCs film in cross section showing the distribution of elements throughout the film. The scale bar is 100 nm. The shaded region in the concentration distribution plot is the borosilicate glass substrate. (c) AFM HDCF and corresponding 3D topography (inset) over a 60-μm × 60-μm scanned area of sample 3-b. Dashed lines denote the extrapolations of the HDCF at small and large length scales for determining the in-plane correlation length (ξ ≈ 1.4 μm) of the sample morphology $({\sigma }_{\mathrm{rms}}\approx 20\mathrm{nm})$. The height variation range of the 3D topography (inset) is 120 nm. (d) X-ray-scattering measurements of samples 3-a (red) and 3-b (black). The ZnO crystallite size estimated from diffraction patterns is the same for both samples.

Figure 3

(a) Picosecond acoustic signals from a representative ZnO NC sample to determine the Al thickness. (b) TDTR ratio signal and fitting results for ZnO NC networks after doping IPL treatments with different infill material compositions of ZnO and $\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$. (c) TDTR ratio signal (symbols) and fitting results (red curves) at modulation frequencies of 18 and 1.6 MHz, respectively.

Figure 4

Thermal conductivity of ZnO NCs with different annealing (red) and doping (blue) IPL flash numbers for series 1–3 (1: circles; 2: triangles; 3: stars). The open symbols and the solid symbols are the measured network thermal conductivity ${\mathrm{\Lambda }}_{\mathrm{nw}}$ and electronic thermal conductivity ${\mathrm{\Lambda }}_{\mathrm{e}}$, respectively. There are two $x$ coordinates, the first row with red numbers presents the sintering IPL flash numbers before ALD of infill materials, and the second row with blue numbers presents the doping IPL flash numbers after 1000 flashes of sintering IPL and ALD of infill materials.

Figure 5

The network thermal conductivity ${\mathrm{\Lambda }}_{\mathrm{nw}\_\mathrm{cal}}$ as functions of various normalized parameters calculated from the modified EMA model, including ${\mathrm{\Lambda }}_{\mathrm{NC}}/{\mathrm{\Lambda }}_{\mathrm{ZnO}}$ (gold dashed line), ${\mathrm{\Lambda }}_{\mathrm{IF}}/{\mathrm{\Lambda }}_{\mathrm{Al}2\mathrm{O}3}$ (black solid line), $G/G_0$ (red dashed line), ρ/$a$ (green dashed line). Specifically, ${\mathrm{\Lambda }}_{\mathrm{NC}}$ are normalized by the bulk thermal conductivity of single-crystal ZnO (${\mathrm{\Lambda }}_{\mathrm{ZnO}}=50\mathrm{W}{\mathrm{m}}^{-1}{\mathrm{K}}^{-1}$) [52], ${\mathrm{\Lambda }}_{\mathrm{IF}}$ is normalized by the bulk thermal conductivity of single-crystal $\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$ (sapphire, ${\mathrm{\Lambda }}_{\mathrm{Al}2\mathrm{O}3}=30\mathrm{W}{\mathrm{m}}^{-1}{\mathrm{K}}^{-1}$) [57], ρ is normalized by the NC radius $\left(a=5\mathrm{nm}\right)$, and G is normalized by a reference value representing a good solid-solid thermal interface (e.g., $G_0=500\mathrm{M}\mathrm{W}{\mathrm{m}}^{-2}{\mathrm{K}}^{-1}$ between TiN and MgO) [58]. The inset shows the calculated ${\mathrm{\Lambda }}_{\mathrm{nw}\_\mathrm{cal}}$ as a function of ${\mathrm{\Lambda }}_{\mathrm{IF}}/{\mathrm{\Lambda }}_{\mathrm{Al}2\mathrm{O}3}$ in the range of 0.04–0.10, and the measured ${\mathrm{\Lambda }}_{\mathrm{nw}}$ for sample 1-a (open red circle), sample 2-a (open red triangle), and sample 3-a (open blue star) compared with the calculated ${\mathrm{\Lambda }}_{\mathrm{nw}}$ (solid blue star).