(622c) Separating Carbon Nanotubes: Connecting Single Molecule Electrical Measurements to Ensemble Spectroscopic Properties

Authors: 
Strano, M., Massachusetts Institute of Technology
Kim, W., Massachusetts Institute of Technology
Nair, N., Massachusetts Institute of Technology
Lee, C. Y., Ulsan National Institute of Science and Technology (UNIST)


The separation and sorting of single-walled carbon nanotubes (SWNTs) is one of the most significant challenges in this rapidly progressing field. A variety of methods including density gradient centrifugation1 and free solution electrophoresis2 invariably utilize Raman, photoabsorption, or photoluminescent spectroscopy to benchmark the resulting separation. However, it is not clear how to relate these qualitative numbers to absolute quantities of individual SWNTs. Moreover, for nanoelectronic applications, separation and purity standards need to be studied using single molecule electrical measurements, and ensemble spectroscopic

assessments are thus far uninformative for this purpose. There is also recent evidence that solution phase processing of the types employed for the above separation methods induces undesirable defects on SWNTs that adversely affect their electrical performance. However, there is currently no quantitative standard by wh ich the degree of defect generation on individual SWNTs from solution phase processing is accessed. In this presentation, we directly compare, for the first time, the ensemble spectroscopic measurements to a systematic and statistically rigorous counting of individual metallic and semiconducting SWNT devices using a high throughput electrical probe station. The comparison allows us for the first time to report an accurate extinction coefficient ratio for metallic and semiconducting SWNTs from HiPco and laser oven preparations. This parameter should greatly aid in the analytical chemistry of separation methods for SWNTs. The systematic counting of metallic and semiconducting types from solution also allows us to examine the variances associated with device properties and therefore provide the first measure of potential defect generation during processing.