(244a) Multiscale Structural Analysis of Activated Carbon Materials From HRTEM Images by Morphology Modelling Approach

The chemical and porous structure of activated carbon materials has a key influence on the physical adsorption capacity and its selectivity. Applied to volatile organic compound on fixed carbon bed, we have recently determined mathematical relationships between structure and physical adsorption interactions [1] and between structure and spontaneous ignition reactivity [2]. Those two approaches show how important the characterisation of the structure is for the design of better material and more efficient-specific adsorption applications. Unfortunately, the real structure of activated carbons is not so well understood. Because of the complexity of the material, characterization by the mean of direct measurement is not possible and indirect methods like DFT on absorption isotherm are used to extract structural information. Experimental parameters like specific surface area or pore size distribution can not be considered as precise and absolute representation of the true structure even though they are extremely valuable to compare closely related materials. Attempts to rebuild molecular model of activated carbon fitting scattering data to a three-dimensional distribution of atoms still remains a challenging approach that do not provide so far valid models mainly because of the difficulties encountered when interpreting diffraction profiles within chemical diversity.

One of the ways to achieve direct observations of the nano-structure relies on high resolution transmission electron microscopy. With the help of state-of-the-art microscope, we can acquire images where it becomes possible to spot the poly-aromatic layers that constitute the core of the structure. Little effort has been put in trying to extract quantifiable structural information from HRTEM images [3]. Here we present our research results on the development of a novel automatic HRTEM image analysis procedure based on the mathematic morphology tools from Morph'M platform [4]. Our analysis consists on series of morphological transformations and topological measurements. From grey level image with resolution of 0.0277nm, we were able to extract major components of the structure. A thinning algorithm called ?Zhang' reduced the image to a skeleton [5]. In order to identify individual branches, segmentation and cleaning algorithm were completed. From the resulting skeleton, we were able to realize several measurements on individual fringe but also on the surrounding neighbourhood when calculating parallelism between isolate fringes.

Primary results show clearly the existence organized zones called Basic Structural Unit (BSU) like those proposed in Oberlin's model where the structure of activated carbon is though to be composed of small units of stacked graphene sheets spread in the material. Following this finding, we completed some more measurements on the shape and distribution of those BSU and compare the data between four different commercial activated carbons. The identification of BSU highlighted two other levels of nano-structural organisation which can be defined as semi-organized (graphene sheets out of BSU) and totally disorganised (no graphene sheet) zones. The relative fraction volume and their diverse distribution functions revealed some more details to better characterise activated carbons. Beyond this nano-structure analysis, we recently extended our field of investigation toward larger scale by the mean of electron tomography in order to identify by 3D reconstruction the meso- and macro-pore topologies in term of size, shape and also inter-connectivity network.

This study brings new insights about the morphology of adsorption sites and the possible mechanisms of adsorption that could take place on the surface of those porous materials. Looking backward at adsorption isotherms, we can now raise new questions about the real meaning of the current structural interpretation in term of specific surface, pore size distribution and porosity network. Our morphological analytical tools for HRTEM images will also open new opportunities to better characterise activated carbon materials in order to detect key structural differences.

[1] Giraudet S, Pré P, Tezel H, Le Cloirec P, Carbon 44(10): 1873-1883, 2006

[2] Jayabalan T, Pré P, Héquet V, Le Cloirec P, Adsorption 14: 679-686, 2008

[3] Rouzaud J-N, Clinard C, Fuel Processing Technology 77-78:229-235, 2002

[4] Jeulin D, Statistics and Computing 10: 121-132, 2000; Morph-M Software, http://cmm.ensmp.fr/Morph-M/index_en.html , 2007

[5] Zhang TY, Suen CY, Communications of the ACM, 27(3): 236-239, 1984