Characterisation of the porous structure of Gilsocarbon graphite using pycnometry, cyclic porosimetry and void-network modelling. Giuliano M. Laudone a , Christopher M. Gribble a , G. Peter Matthews a & Luca Lucarelli b a School of Geography, Earth and Environmental Sciences, Plymouth University, UK b ThermoFisher Scientific, 20090 Rodano, Milan, Italy Summary The cores of the fourteen Advanced Gas-cooled nuclear Reactors (AGRs) within the UK comprise Gilsocarbon graphite, a manufactured material surrounded predominantly by CO 2 at high pressure and temperature to provide heat exchange. The intense ionising radiation within the reactors causes radiolytic oxidation, and the resulting mass loss is a primary factor in determining reactor lifetime. The void structure of the porous Gilsocarbon graphite affects the permeability and diffusion of the CO 2 , and the sites of oxidation. To model this void structure, we have measured the porosities and densities of ten Gilsocarbon graphite samples by powder and helium pycnometry 1 . Samples have been examined at a range of magnifications by electron microscopy. Total porosities and percolation characteristics have been measured by standard and cyclic mercury porosimetry up to an applied mercury pressure of 400 MPa. Inverse modelling of the cyclic intrusion curves, using PoreXpert (www.porexpert.com), produces simulated void structures with characteristics which closely match those of experiment. The structures are based on infinitely repeating unit cells, in which the porous network is simplified to cubes connected by cylindrical voids. The simulated void networks provide the bases for future simulations of the radiolytic oxidation process itself, extending the work reviewed by Best et al. 2 . Objectives Investigate the porous structure of Gilsocarbon graphite. Investigate whether cyclic mercury porosimetry provides significant additional information when com- pared with traditional mercury porosimetry. Create a void network model of Gilsocarbon graphite to investigate mercury hysteresis. Materials & Methods Ten virgin Gilsocarbon graphite samples were supplied by EDF Energy, from spare bricks destined for UK nuclear reactors. Pycnometry was carried out using a Micromeritics GeoPyc 1360 and a Ther- moFisher Pycnomatic ATC. Mercury porosimetry and cyclic porosimetry experiments were performed using Thermo Fisher PASCAL 140 and 440 mercury porosimeters. All experiments were carried out following local ISO 9001:2008 protocols. SEM images of the Gilsocarbon graphite are shown in Figure 1. (a) SEM image of P37E9E with a 100 μm scale bar (b) SEM image of P37E9E with a 10 μm scale bar Figure 1: SEM images of P37E9E at two different magnifications. Modelling PoreXpert is a commercially available software package for the construction of void networks by inverse modelling of percolation characteristics. The void structure of a porous material is represented as a series of interconnected unit cells with periodic boundary conditions. The network model can generate structures with porosity and percolation properties such as mercury porosimetry, porometry or water retention, closely matching experimental measurements. The model fits experimental data using an annealed ameoboid simplex 3 . A new algorithm for the simulation of extrusion of non-wetting fluids when the external pressure is reduced was developed in order to simulate the experimental results obtained by cyclic mercury porosimetry. There are different types of mercury intrusion-extrusion curves shown in Figure 2. Figure 2a below shows the situation when there is no trapping or contact angle hysteresis. Figure 2b occurs when the contact angle changes between intrusion and extrusion. For most samples Figures 2c & 2d are more likely because mercury almost always gets trapped because of snap off. 1 st intrusion 1 st extrusion 2 nd intrusion Applied mercury pressure Accessible void volume (a) No trapping and no contact angle hysteresis. 1 st intrusion 1 st extrusion 2 nd intrusion Applied mercury pressure Accessible void volume contact angle hysteresis (no trapping) (b) Contact angle hysteresis. 1 st intrusion 1 st extrusion 2 nd intrusion Applied mercury pressure Accessible void volume contact angle and shielding hysteresis partial trapping (c) Partial trapping. 1 st intrusion 1 st extrusion 2 nd intrusion Applied mercury pressure Accessible void volume complete trapping (d) Complete trapping. Figure 2: Mercury intrusion-extrusion cycles. The extrusion algorithm works by checking the difference in diameter between the pore and adjoining throats. When the ratio exceeds a value of 5, the ”snap-off ratio”, the algorithm retains the mercury in the structure 1 . Results & Discussion Figure 3a below shows the experimental cyclic porosimetry results for sample P37E9E, with the porosity determined by pycnometry shown as a dashed horizontal line. Gilsocarbon graphite is an unusual material in that it has a high percentage of closed porosity, up to 30 %, so the pycnometry line crosses below the maximum of the porosimetry curve. Figure 3b is a graph that compares the experimental and simulated mercury intrusion and extrusion curves up to the accessible porosity. The experimental data is shown in black and the simulated data is in grey, with solid lines indicating intrusion and dashed lines representing extrusion. It can be seen that there is a good agreement between the experimental and simulated data. 10 -1 10 0 10 1 10 2 Pressure /MPa 20 40 60 80 100 normalised mercury volume /mm 3 g -1 Void volume measured by pycnometry operating pressure in AGRs intrusion extrusion (a) Full cyclic porosimetry curve for sample P37E9E, with the equivalent pycnometry void volume identified and oper- ating pressure of AGR (4 MPa). 10 -2 10 -1 10 0 10 1 Pressure /MPa 0 20 40 60 80 100 accessible void volume /% experimental intrusion experimental extrusion simulated intrusion simulated extrusion (b) Simulated and experimental cyclic porosimetry curve for sample P37E9E, trimmed to the pycnometry porosity shown in Figure 3a. Figure 3: Experimental and simulated mercury porosimetry results. Figure 4 below shows a PoreXpert unit cell for sample P37E9E half filled with mercury. The unit cell was created using the non-cyclic mercury porosimetry intrusion curve, up to a maximum pressure equivalent to the open porosity for Gilsocarbon graphite measured by pycnometry, Figure 3a. Figure 4: PoreXpert unit cell for sample P37E9E. The small bar bottom left has length 164.8 μm. The void features cover four orders of magnitude so some are invisibly small. The transparent region corresponds to the solid matrix, the grey void features indicates the presence of mercury, injected to a pressure of 0.17 MPa. Figure 3a shows that there are steps in intrusion at both low and high pressure. This is explained by the complexities of the void structure, Figures 1a & 1b. At high pressure, sample compression can be discounted, because the intrusion and extrusion curves diverge. Figure 2a is typically encountered with compression only. The simulation correctly predicts, based on the intrusion curves and sample porosity, that total or almost total trapping occurs, Figure 2d, over the entire pressure range. In the present case, it has done so because it has generated a priori a void network in which nearly all of the pores are more than 5 times larger than their adjoining throats in the intrusion/extrusion direction, and which therefore trap all of the mercury. Natural samples normally exhibit characteristics such as that shown in Figure 2c. The differences between the natural sample and the man-made Gilsocarbon can be explained by its manufacture. During the graphitisation process coal tar pitch is added which causes the entries to be blocked or almost blocked. Mercury therefore enters through openings which are much smaller than the adjacent pores, so mercury is trapped and extrusion is prevented. An additional complication of the sample from the manufacturing process is the production of gas that forms voids in the structure. The pressurisation during porosimetry may be able to break through ’narrow walls’ separating the open and closed porosity. Graphitisation also forms Mrozowski cracks, which can lead to total trapping of intruded mercury. Conclusions Pycnometry results have shown that while the density remains constant, the open, closed and total porosities varied widely 1 . The electron micrographs and mercury intrusion measurements confirm that the void structure of Gilsocarbon graphite is complex at every size level. Inverse modelling of the intrusion provides a much more realistic estimate of the void sizes than available previously, and provides the basis for a quantitative estimate of radiolytic oxidation and hence mass loss. References (1) Laudone, G.M., Gribble, C.M., & Matthews, G.P. (2014). Characterisation of the porous structure of Gilsocarbon graphite using pycnometry, cyclic porosimetry and void-network modeling. Carbon, 73, 61-70. (2) Best, J.V., Stephen, W.J., & Wickham, A.J. (1985). Radiolytic graphite oxidation. Prog. Nucl. Energy, 16(2),127 - 178. (3) Laudone, G.M., Matthews, G.P., Gregory, A.S., Bird, N.R.A., & Whalley, W.R. (2013). A dual-porous, inverse model of water retention to study biological and hydrological interactions in soil. EJSS, 64, 345 - 356. Acknowledgements The financial support and advice of EDF Energy (Gloucester, UK) is gratefully acknowledged. However the views expressed in this paper are those of the authors and do not necessarily represent the views of the sponsors. The authors are grateful for the advice of Anthony Wickham, Visiting Professor in the School of Mechanical, Aerospace and Civil Engineering, University of Manchester, UK. The authors would also like to thank the University of Plymouth Electron Microscopy Centre. Additional information For further information about the work presented in this poster, please see reference 1. The corresponding authors email is [email protected]. To find out more about PoreXpert which was used to create the models, please visit http://www.porexpert.com, where you can download a trial copy of the software. You can also contact PoreXpert Ltd by using [email protected].