Key Word: thermoelectric conversion / semiconducting carbon nanotube / electrical conductivity / Seebeck coefficient / thermal conductivity Introduction It is reported that nearly 15 terawatt of energy has been wasted as low temperature heat [1, 2] per year. Thus, conversion of heat into electricity so called thermoelectric (TE) conversion has become an important technology. Typical approach for the TE conversion is to utilize the Seebeck effect that enables the temperature gradient into the carrier movement, resulting in the generation of the electricity. TE efficiency of the materials is usually characterized by the figure of merit (ZT) given by the equation (1) ZT = (ÏS2/Îș) ă» T (1), in which Ï, S, Îș, T is electrical conductivity, Seebeck coefficient, thermal conductivity and temperature, respectively. To achieve a high ZT value, the material having high Ï and S, low Îș is desired. Up to date, inorganic semiconducting materials such as Bi2Te3 are used as the thermoelectric materials because of their high ZT value (ZT>1) [3]. However, these materials have problems such as toxicity, high cost and low processability, which prevent these materials from the practical use. In this background, recently, single-walled carbon nanotubes (SWNTs) have attracted strong attention due to the extremely high electrical conductivity as well as their non-toxicity, light weight, mechanical toughness, flexibility and low cost, all of which are advantageous as the thermoelectric material. SWNTs are produced as a mixture of metallic (m-) and semiconducting SWNTs (s-SWNTs) and s-SWNT sheet was proved to have large Seebeck coefficient theoretically and experimentally [4, 5]. However, the improvement of ZT value by separating s-SWNT from the as-produced SWNT mixture is still uncertain. In this study, we characterized the TE properties of s-SWNT sheets with different metal impurity and compared their ZT values with that of unsorted SWNT sheet. Experiment s-SWNTs 2.0 mg (98% Nanointegris) and m-SWNT 1.0 mg (98% Nanointegris) were added to 0.5% SDBS (sodium dodecylbenzenesulfonate) solution and dispersed by bath sonicator (BRANSON, 1 hr) and probe sonicator (TOMY UD-200). The dispersion were filtrated and the free-standing s-:m-SWNT=2:1 sheet was obtained (2:1 mix). s:m-SWNT=4:1 sheet (4:1 mix), s- and m-SWNT sheets were made in the same fashion. In-plane electrical conductivity and Seebeck coefficient were measured by ZEM-3 (ADVANCE RIKO) from 30 to 100 ÂșC. The specific heat capacity (Cp) was measured by differential scanning calorimetry (DSC) method using EXSTAR DSC 6200 (SII Nanotechnology) at the heating rate of 10 K min-1. In-plane thermal diffusivity (α) were evaluated using a Thermowave Analyzer TA (Bethel Co., Ltd.,). The density of the sheets (Ï) was calculated from the weight and volume of the sheets. Results and discussion Table 1 summarized the in plane Ï, S and power factor (PF) of the unsorted SWNT, s-SWNT, m-SWNT and 2:1 mix sheets at 30 ÂșC. Unsorted SWNT sheet exhibited nearly 1.9 times higher electrical conductivity than the m-SWNT sheet. This was due to the defects induced during the separation process of the m-SWNT, which was confirmed by the larger D band of m-SWNT than that of unsorted SWNT in Raman spectra (data not shown). s-SWNT sheet showed higher Seebeck coefficient, however, the powerfactor (PF) value was similar to that of the unsorted SWNT sheet due to the large decrease of electrical conductivity. In addition, ZT value of s-SWNT was almost comparable with that of unsorted SWNT, which implied the good TE properties of unsorted SWNT in the practical point of view. Reference [1] Hochbaum, A. I. et al, Nature, 451, 163-167 (2008) [2] Ren, Z. F. and Chen, G. et al, Energy Environ. Sci. 5, 5147â5162 (2012) [3] G. J. Snyder and E. S. Toberer et al, Nature Mater., 7, 105-114 (2008). [4] Blackburn, J. L. and Ferguson, A. J. et al. Nature Energy, 1 (2016) [5] Nakai, Y., Maniwa, Y.et al, Applied Physics Express 7, 025103 (2014) Figure 1