Resonance-Filtered Imaging of Collective pi-States in a Carbon Nanotube using Electron Energy -Loss Spectra Acquired in an Enfina / STEM System *A. Seepujak, *U. Bangert, *A. Gutiérrez-Sosa, *A. J. Harvey, **V. D. Blank, **B. A. Kulnitskiy, **D. V. Batov * Department of Physics, UMIST, PO Box 88, Manchester M60 1QD, UK ** Technological Institute for Superhard and Novel Carbon Materials, 7a Centralnaya Street, 142092 Troitsk, Moskow, Russia Collective states of a multi-wall carbon nanotube (MWCNT) delocalised pi-system have been recently identified in individually-acquired electron energy-loss (EEL) spectra [1]. Collective pi- states propagating within the bulk / surface of the MWCNT are referred to as volume / surface plasmons, respectively. Acquiring EEL spectra using a Gatan UHV Enfina system, as opposed to individually-acquired EEL spectra, allows sophisticated post-measurement processing and analysis of EEL spectra e.g. filtering of the volume and surface plasmon contributions [1]. This has enabled striking 3D spatial distributions, termed EEL spectral images (SI maps), to be observed for the first time, of MWCNT volume / surface plasmons. The Enfina system was attached to a VG HB 601 dedicated STEM (scanning transmission electron microscope). An acquisition area was divided into a specified number of pixels (x,y) = (60,9), as shown in Fig. 1(a). In order to improve statistics, at each pixel, 10 EEL spectra were acquired, recorded and summed, each spectrum being acquired with dwell-time of 0.07 seconds. These spectra composed the SI map. Discrimination of individual plasmons in a SI map is not a trivial process. Plasmons are typically close in energy e.g. MWCNT bulk and surface pi-plasmons have a resonance energy <1.5 eV of each other [1]. Aiding discrimination of contributions, the STEM was equipped with a cold field emission source, providing an energy resolution of 0.34 eV, given by the FWHM of the ZLP, with the point spread function ensuring a sharp cut-off of the ZLP tail. Separation of the ZLP from the signal was further aided by use of a high dispersion (0.02 eV/channel) for spectral acquisition. The ZLP was extracted using a power law function, fitted to the EEL window between 2.16 – 3.68 eV, in each pixel of the SI map. This window extracted the ZLP whilst minimising the introduction of artefacts, or inadvertently removing structure [2]. Volume and surface plasmons were each sequentially isolated and extracted from the ZLP-extracted SI map at each pixel. In order to identify these contributions, a spectrum standard consisting near- exclusively of volume contributions, and a spectrum standard consisting near-exclusively of surface contributions, were required. The pixel at position V, identified in Fig. 1(a), was chosen as representative of a near-pure volume excitation, due to its close proximity to the MWCNT axis, and its position in a region where the inner and outer diameter are relatively invariant. The pixel at position S was chosen as representative of a near-pure surface excitation, due to isolation of excited surface modes in the aloof geometry, and the proximity of S to the thinnest part of the tube, ensuring any residual volume contributions were small in amplitude. Extraction of the volume contribution was performed by fitting a single Gaussian G v , to the ZLP- extracted SI map at each pixel. The energy of G v was constrained to that of the volume standard. For a volume excitation energy of ~7 eV, the wave vector of the volume mode has a value of Microsc Microanal 10(Suppl 2), 2004 Copyright 2004 Microscopy Society of America DOI: 10.1017/S143127604882539 844