With recent improvements in carbon nanotube separation methods, the accurate determination of residual metallic carbon nanotubes in a purified nanotube sample is important, particularly for those interested in using semiconducting single-walled carbon nanotubes (SWCNTs) in electronic device applications such as thin-film transistors (TFTs). This work demonstrates that Raman microscopy mapping is a powerful characterization tool for quantifying residual metallic carbon nanotubes present in highly enriched semiconducting nanotube networks. Raman mapping correlates well with absorption spectroscopy, yet it provides greater differentiation in purity. Electrical data from TFTs with channel lengths of 2.5 and 5 μm demonstrate the utility of the method. By comparing samples with nominal purities of 99.0% and 99.8%, a clear differentiation can be made when evaluating the current on/off ratio as a function of channel length, and thus the Raman mapping method provides a means to guide device fabrication by correlating SWCNT network density and purity with TFT channel scaling.As-prepared single-walled corbon nanotube (SWCNT) raw materials contain bundles of nanotubes with a 2:1 semiconducting (sc-) to metallic (m-) ratio and other impurities such as catalysts and amorphous carbon [1][2][3]. For electrical devices, these raw materials must be debundled, purified, and enriched [4][5][6][7][8][9][10][11]. It has been demonstrated that for network thin-film transistors (TFTs), the m-tube content should be less than 2% [3]. For more demanding applications, such as high-frequency logic circuits and display backplanes, the m-tube content should be less than a few ppm given the need for high mobility and high current on/off ratios [12]. In recent years, multiple efforts have led to significant progress in solution-based enrichment techniques, with sc-SWCNT purities in excess of 99% routinely achieved. This leap, however, has now started to reveal the limits of common characterization methods, and other tools are required to provide accurate purity assessment [13,14].Typically, the purity of enriched sc-SWCNTs is Nano Research 2 Nano Res. estimated from the UV absorption spectrum by comparing the peak areas associated with the m-or sc-species [12,[15][16][17]. This method works well for samples whose peaks are well defined and for which background absorption is not dominant. However, as the sc-species purity increases, the features associated with m-tube absorption will gradually disappear and the precise subtraction of the background absorption will dramatically influence the calculated results and introduce uncertainty [18]. As an alternative, we have previously used a different purity metric, denoted as φ, which is based on the ratio of the sc-peak area over the total absorption background from the metallic (M 11 ) and semiconducting (S 22 ) absorption bands [16].Although we presume that φ correlates well with purity for highly pure samples, it does not provide a quantitative assessment. Furthermore, a solution sample is not nece...