Probing Nanoscale Solids at Thermal Extremes

We report a novel nanoscale thermal platform compatible with extreme temperature operation and real-time high-resolution transmission electron microscopy. Applied to multiwall carbon nanotubes, we find atomic-scale stability to 3200 K, demonstrating that carbon nanotubes are more robust than graphite or diamond. Even at these thermal extremes, nanotubes maintain 10% of their peak thermal conductivity and support electrical current densities $\sim 2\times {10}^8\mathrm{A}/{\mathrm{cm}}^2$. We also apply this platform to determine the diameter dependence of the melting temperature of gold nanocrystals down to three nanometers.

Figure 1

Design and operation of the thermal test platform. (a) A nanoscale sample is electrically contacted on a suspended membrane. (b) Nanocrystal thermometers are deposited on the sample. The sample is resistively heated via an electrical current, $I$, causing the nanocrystals to melt, yielding the temperature of the sample. (c) At higher bias, the melting nanocrystals yield an isothermal line which fans out across the membrane.

Figure 2

Operation of MWNT devices. (a)–(d) TEM images of MWNT devices, with color overlays to indicate local temperature [note (d) uses a different color scale]. (a) Zero bias, isothermal ($T=300\mathrm{K}$) conditions. (b),(c) Same device operating at biases above the threshold for melting of the gold nanoparticles. (d) A different MWNT (chosen for image clarity) operated close to the breakdown temperature of 3200 K. The darker area near the center of the nanotube shows the membrane failure.

Figure 3

Full temperature dependence of the thermal conductivity $\kappa$ of MWNTs. Experimentally determined high-temperature $\kappa$ reflecting umklapp and three-phonon processes [Eq. (2)] is plotted as a solid line. Also plotted are literature results for $\kappa$ of various carbon nanotubes at low temperatures [15, 16, 17].

Figure 4

Diameter (d) dependence of the melting point ($T_m$) of gold nanoparticles (NPs). Scales are 5 nm (a) TEM image of MWNT heater with NPs (lines indicate MWNT outer wall). Selected NPs on the heater are outlined in color: red ($d=3\mathrm{nm}$), green ($d=4\mathrm{nm}$), and blue ($d=6$, 6.5 nm). (b)–(e) TEM video images as heater bias is increased with rendered models of system. (b) $T=1165\mathrm{K}$, no NPs have melted. (c) $T=1130\mathrm{K}$, 3 nm NP melts and evaporates. (d) $T=1200\mathrm{K}$, 4 nm NP evaporates. (e) $T=1275\mathrm{K}$, 6, 6.5 nm NPs evaporate. (f) $T_m$ vs $d$. Data from this study (crosses) show strong diameter dependence, with experimental results of Koga et al. (open circles). The liquid shell model (dashed line) provides an excellent fit to the data.