2, Northrop Grumman Corporation, Redondo Beach, California, United States
3, Stanford University, Stanford, California, United States
4, Stanford University, Stanford, California, United States
5, Stanford University, Stanford, California, United States
Semiconducting single-wall carbon nanotubes (s-SWNTs) are expected to have high Seebeck coefficient and high electrical conductivities [1,2]. However due to the presence of metallic SWNTs, polymer wrapping, and dopants, the effective Seebeck coefficient of materials based on s-SWNTs have been measured to be much lower . In this work, we present a comprehensive study of the effect of temperature and doping (both n- and p-type) on the thermoelectric transport in ultra-pure (>99.9 %) s-SWNT networks. We measure the highest electron and hole Seebeck coefficients for polymer-free s-SWNT networks, over the 80 to 600 K temperature range.
We use nanoscale on-chip thermometry to measure the electrical conductance and Seebeck coefficient of ultra-pure s-SWNT networks (5-7 nm thick) as a function of Fermi energy by back-gating the network, from electron- to hole-transport regime. We fabricate large numbers of devices based on an electrical thermometry technique . First, we pattern and deposit Ti/Pt electrodes on 300 nm SiO2/p++ Si substrates. Next, the s-SWNT are solution-deposited onto the chips. We remove traces of polymer from the SWNTs and subsequently pattern and etch the s-SWNTs into 10–50 µm scale channels over the electrical thermometers.
We measure the Seebeck coefficient and electrical conductance of these networks under vacuum while varying the back-gate voltage over temperatures ranging from 80 K to 600 K. Due to the high quality of the s-SWNTs, the networks have high electrical on/off ratios of 106. The SWNT films transition from p-type to ambipolar transport above 450 K. Beyond 550 K, we measure both high hole and electron thermopower, reaching of up to ±550 μV/K, which is a record for a SWNT network. We attribute our results to the minimal amount of SWNT bundling, low polymer residue, and very few metallic SWNTs in our networks. Using physical models of the thermal, thermoelectric, and electrical properties of both individual SWNT and the junction between s-SWNTs, we produce a deeper understanding of the fundamental thermoelectric transport in these networks.
 J. P. Small et al., Phys. Rev. Lett., 91, 256801 (2003)
 N. T. Hung et al., Phys. Rev. B, 92, 165426 (2015)
 A. D. Avery et al., Nat. Energy 1, 16033 (2016)