Characterization of the 3-dimensional microstructure …

1 Characterization of the 3-dimensional microstructure of a graphite negativeelectrode from a Li-ion Shearinga, Howardb, J rgensenc, Brandona,*, HarrisdaDepartment of Earth Science and Engineering, Imperial College London, London, United KingdombDepartment of Mineralogy, The Natural History Museum, South Kensington, London, United KingdomcRis National Laboratory, Roskilde, DenmarkdGeneral Motors R&D Centre, Warren, Michigan, USAarticle infoArticle history:Received 10 December 2009 Received in revised form 23 December 2009 Accepted 24 December 2009 Available online 4 January 2010 Keywords:X-ray tomographyMicrostructureRepresentative volume elementBatteryGraphite electrodeabstractThe 3-dimensional microstructure of a porous electrode from a lithium-ion battery has been charac-terized for the first time.

2 We use X-ray tomography to reconstruct a 43 348 478lm samplevolume with voxel dimensions of 480 nm, subsequent division of the reconstructed volumes intosub-volumes of different sizes allow us to determine microstructural parameters as a function ofsub-division size. We show that the minimum size for a representative volume element is about43 60 60lm for volume-specific surface area, but as large as the full sample volume for porosityand tortuosity. 2009 Elsevier All rights IntroductionLi-ion batteries are generally analyzed using a macro-homoge-neous porous electrode model[1,2].

3 Model input includes particleradius and porosity; otherwise, it assumes that the electrode is anisotropic, homogeneous, 1-dimensional porous material con-structed from mono-disperse, non-porous isotropic spherical parti-cles small compared to the electrode thickness. This model hasbeen highly successful in optimizing electrode properties such asfilm thickness and porosity[3].The ability to predict cell degradation is a challenge, however,because so many seemingly unrelated degradation mechanismshave been identified[4].

4 Analysis of specific degradation mecha-nisms can in some cases provide a rationale for experimentally ob-servedphenomena[5 8], butwithoutadditional experimentaldata,quantitative cause-and-effect relationships between observationand degradation pathway are difficult to develop. Measurementsshowing the evolution of the 3-dimensional microstructure willhelp enable deconvolution of these are a number of experimental techniques that can pro-vide such information. For example, Yoshizawa et al.[9] used elec-tron tomography (3D TEM) to observe the internal structure andconnectivity of carbon nanospheres.

5 Thorat et al.[10] have devel-oped a novel technique to determine electrode and separator tor-tuosity, in contrast to its more common treatment as anadjustable parameter[2].The use of tomographic techniques in the field of fuel cell re-search has provided unprecedented access to microstructuralinformation. Two commonly used techniques are high resolutionX-ray computerized tomography[11]and focused ion beamtomography[12,13]. Of these, focused ion beam milling has beenused to examine battery electrodes[14 19]; however, to datethere have been no 3-dimensional reconstructions of a batteryelectrode paper presents the results of tomography experiments tocharacterize graphite electrode microstructures with subsequentgeometrical analysis of the reconstructed volume.

6 Any tomographyprocedure must balance the dual requirements of reconstructing asufficient sample volume while maintaining imaging high stopping distance of ions in graphite means that graphitespecimens are highly resistant to ion beam milling, thus limitingthe sample volume that can be reconstructed. After preliminaryinvestigations, X-ray nano-CT has been selected as the most appro-priate technique to characterize our sample. As material failure isoften a result of local inhomogeneities and defects[20,21], we fo-cus here on the validity of the widely used assumptions of homo-geneity and isotropy in battery electrode $ - see front matter 2009 Elsevier All rights *Corresponding author.

7 Tel.: +44 Brandon).Electrochemistry Communications 12 (2010) 374 377 Contents lists available atScienceDirectElectrochemistry Communicationsjournal homepage: ExperimentalAbout 1 cm2from a graphite negative electrode was harvestedfrom a Lishen 18650 battery of A h capacity. The copper currentcollector was dissolved in nitric acid, and an area of the electrodewas identified for examination and mounted onto an aluminumpin using silver paint. Dissolution of metals from carbon using ni-tric acid is a standard process ( [22]).

8 The Gatan X-ray ultramicroscope (XuM) system was used forhigh resolution computerized tomography (nano-CT)[23,24]. Pro-jected X-ray images were acquired at 1 rotation increments over190 and reconstructed using Gatan s cone-beam back-projectionalgorithm to generate a 3D volume. The images were acquired withan 80 s exposure time (total acquisition time of h) and a totalmagnification of , corresponding to 480 nm total reconstructed sample volume was 43 348 478lm; subsequent geometrical analysis was conducted to ex-tract porosity, pore-connectivity, particle and pore size distribu-tion, surface area and tortuosity.

9 The analysis considered theentire bulk volume as well as a number of sub-volumes (which,when combined, represent the entire bulk volume); sub-volumedimensions are provided inTable 1. We then extracted standarddeviations at each sub-volume dimension, allowing us to suggestthe minimum representative volume element (RVE) surface area was calculated by creating polygons alongthe surface defined by the pore-graphite interface. Tortuosity iscalculated based on the geometrical definition provided by Shen[25] and is determined as follows: for a given pore structure theminimum distance from one side to the other is termed D.

10 Theshortest distances (L1) from every pixel on one side to the closestpixel on the opposing side is calculated, ensuring that the path ismaintained within the 3-dimensional pore structure. Tortuosity,calculated for each of theseL1 values is defined as tortuos-ity =L1/D, and tortuosity factor is defined as the square of tortuos-ity. Pore size distribution was calculated by the 3-dimensional continuous PSD method described in[26].3. Results and discussionFig. 1a shows an individual slice from the reconstructedtomography sequence;Fig.