Unconventional thermal conductivity of suspended zigzag graphene nanomesh

TL;DR

The paper investigates how nano-structuring graphene into zigzag-edged nanomeshes affects thermal transport at room temperature. Using an opto-thermal Raman approach on suspended zGNMs and zGNRs with atomically precise zigzag edges, it reveals a nonclassical dependence: for thin zGNMs ($N=2$–$3$ MLs) $κ$ scales inversely with width $W$, while for thicker samples ($N=5$–$10$ MLs) $κ$ is nearly width-independent down to $W=30$ nm, unlike zGNRs which show conventional suppression with reduced $W$. The work attributes these effects to the mesh structure, suggesting constructive phonon interference akin to a phononic crystal-like transport at room temperature, and demonstrates higher $κ$ in zGNRs than in GNRs with atomically rough edges due to edge order. This points to a new avenue for graphene-based thermal management and phononic crystal designs at room temperature.

Abstract

Compared to the study of graphene itself, the study of nano-structured graphene is rather limited because it is difficult to prepare atomically ordered edges. In this study, we have fabricated a periodically patterned mesh structure of graphene with atomically precise zigzag edges (zGNM: zigzag graphene nanomesh) and studied its thermal conductivity ($κ$) by opto-thermal Raman measurement. Unintuitively, it is found that the $κ$ of zGNM of 2,3 monolayers (MLs) thick is inversely proportional to the nanoribbon width ($W$), while that of zGNM of 5$\sim$10 MLs thick is independent of $W$ down to 30 nm. Since the $κ$ of suspended zigzag graphene nanoribbons (zGNRs) is suppressed by decreasing $W$, this nonclassical behavior of zGNM is due to the mesh structure. In addition, zGNRs show a higher $κ$ than GNRs with atomically rough edges. This is probably due to the atomically ordered zigzag edges.

Figures (5)

• Figure 1: Schematic diagram of (a) zGNM and (b) zGNR devices and (c) the experimental configuration. The nanoribbon width ($W$), device length ($L$) and objective lens height difference ($z$) were defined as illustrated. The red zigzag line in (a) and red straight line in (b) are the assumed quasi-1D and 1D heat paths, respectively.
• Figure 2: (a) Schematic diagram of a hexagonal network of resistance to simulate electronic current that flows in the mesh structure, and (b) simulated current map. Potential $V$ at the position, which corresponds to the laser-irradiated area, is set as $V_0$, while $V$ is set as $0$ at electrodes.
• Figure 3: The (a) $\omega_\mathrm{G}$ and (b) $I_\mathrm{G}$ obtained by changing $z$ during 90 successive measurements for a zGNM device at $P_\mathrm{abs}=0.19$ mW. The right-hand figures illustrate the configuration of the objective lens and graphene device. The $\omega_\mathrm{G}$ becomes minimum, while the $I_\mathrm{G}$ becomes maximum, where the laser through the objective lens focuses at the sample position. The focused $z$ changes gradually due to thermal drift.
• Figure 4: The $t$ dependence of the $\kappa$ of suspended zGNMs. The data are color-coded by $N$. $\kappa$ is independent of $t$ for zGNMs of $N\geq5$ MLs, while it is scattered for zGNMs of $N=3$ MLs.
• Figure 5: The $W$ dependence of the $\kappa$ of (a) suspended zGNMs and (b) suspended zGNRs ($N\ge5$ MLs). The $\kappa$ of supported GNR reported in ref. Bae2013 is overlaid in (b). The inset in each figure shows a typical scanning electron microscope image of each device. The thick blue and red shaded lines show the trend of $W$ dependence of thin and thick devices, respectively.