ERE Pore-scale heterogeneity in the reactive surface area of minerals

Transcription

1 ERE Pore-scale heterogeneity in the reactive surface area of minerals Peter Lai, Samuel Krevor 1. There has been recent interest in studying the significance of the continuum assumption to systems in which mineral reactions and groundwater transport are coupled. 2. In Li et al. (2007), heterogeneity in the distribution of reactive surface area was found to rise to large deviations in effective reaction rates as compared to continuum models with average parameters. Li, Peters, Celia (2007) American Journal of Science Vol 307 pp Clay coated quartz in Berea sandstone Mineral specific information in 3D

2 Mineral(specificreac.ve-surface-area 1. This work provides a statistical characterisation of pore-scale heterogeneity in the reactive surface area of minerals within permeable rocks based on direct observation of the pore morphology and mineral distribution in three dimensions. Probability Clay µ: σ: Quartz µ: σ: Feldspar µ: σ: Others µ: σ: Surface area / pore volume [µm 1 ] Peter Lai - PICO :52 17:00

3 Motivation 1. CO 2 interacting with subsurface water will produce carbonic acid which leads to chemical reactions with the rock 2. While water imbibing dry CO 2 region can lead to precipitation of salts. 3. The extent of mineral dissolution and precipitation will depend on the spatial distribution and nature of the minerals in the rock. 4. Spatial variation in reactive surface area can result in very large differences in overall reaction timeframes. Various minerals with distinct reactivity 3

4 Motivation 1. In Li et al. 2007, simulations were performed on a sandstone rock model composed of quartz, clay and anorthite, with the clay and anorthite constituting 6-50% of the volume fraction of the rock. 2. Heterogeneity in the distribution of reactive surface area could give rise to large deviations in effective reaction rates as compared to continuum models with average parameters. Li, Peters, Celia (2007) American Journal of Science Vol 307 pp

5 Motivation 1. At that time pore-scale observations of the surface area properties were not available and the distributions were based on reasonable assumptions inferred from published values for 1 cm 3 scale samples of various rocks, including Berea sandstone. Probability Clay µ: σ: Quartz µ: σ: Feldspar µ: σ: Others µ: σ: Mineral specific information in 3D Surface area / pore volume [µm 1 ] 5

6 Samples! 1. Several carbonate samples, Edwards, Estiallades, Guelph, Indiana, and Ketton as well as Berea sandstone were used. 50 μm Berea 10 μm Edwards 3 μm 10 μm Estaillades 2. Each represent different pore structures. 4 μm 3. Berea and Edwards comprise multiple mineral types. 50 μm 100 μm 300 μm Guelph Indiana Ketton 6

7 3D-Imaging 1. 3D micro-tomography with1µm pixel resolution. 2. Image is sub-sampled and surface area and pore volume measured. 3. Repeated to characterise heterogeneity i.e. build a probability distribution. Probability µm Probability µm Rock µ: σ: Surface area / pore volume [µm 1 ] Surface area / pore volume [µm 1 ] 7

8 Roughness 1. Comparison of x-ray imagery to surface adsorption measurements leads to a variation of the concept of the roughness parameter. BET / CT [ ] Berea 1 Berea 2 Ketton Estaillades Indiana Guelph Edward 2. The ratio of the two surfaces areas, the roughness parameter, is plotted as a function of the BET surface area BET [m 2 /m 3 ] 3. A comparison of the roughness parameter plot with the MIP based analysis of resolvable porosity provides insight into the nature of the roughness for the various rocks. Fraction of pore radius > 0.1µm Berea 1 Berea 2 Edw Gue Est Ket Ind 8

9 Roughness 4. Ketton carbonate has the highest roughness factor and also the most sub-resolution pore space, nearly 40% of the pore volume was expected to be below the resolution of the x-ray imagery. 5. This is followed by the Edwards carbonate in both roughness and microporosity. Microporosity is thus likely to be a major contributor to the roughness ratio. 6. Berea sandstone has the next highest roughness factor but among the most resolvable porosity. The roughness must be primarily attributed to the high clay content of the rock. BET / CT [ ] Fraction of pore radius > 0.1µm BET [m 2 /m 3 ] Berea 1 Berea 2 Edw Gue Est Ket Ind Berea 1 Berea 2 Ketton Estaillades Indiana Guelph Edward 9

10 Observa.on-scale- 1. For each rock five frequency histograms of the surface area distributions were generated. As the sampling size decreases the distributions broaden reflecting the expected increase in heterogeneity at smaller length scales. 2. This description of heterogeneity is less meaningful if the sub volume being used is smaller than a significant number of pores in the rock. This threshold can be seen clearly for the Berea. 3. This issue is complicated for rocks such as the carbonates that have several modes or a continuous range of pore sizes across length scales. Surface area to pore volume ratio [µm 2/ µm 3 ] Surface area to pore volume ratio [µm 2/ µm 3 ] SA/PV Normal Distributions Berea! Mean SD Sampling [ m] Sampling scale [µm] Indiana! Mean SD Sampling scale [ m] Sampling scale [µm] Sampling scale [µm] 10

11 EDS-comparison BSE Segmented 1. To guide the x-ray image processing for the Berea sandstone, segmentation was performed on locations in the rock that had been imaged using BSE. 2. The segmentation of the x-ray image identified quartz, Kfeldspar and most clays well. X-ray CT - Segmented 90µm 200µm Berea sandstone 11

12 EDS-comparison BSE Segmented 3. A large orthoclase grain that has been partially weathered to illite and was identified as feldspar and quartz in the x-ray image. 4. In this rock, the illite and albite minerals constitute less than 10% of the total segmentation grouping implying that this location saw a larger impact than would be typical for the image as a whole. 90µm X-ray CT - Segmented 200µm Berea sandstone 12

13 BET Normalisation 1. To make the comparison with Li et al (2007) s simulation study, the distributions have to be converted to values representative of those used in Li s study which have been measured by BET adsorption. 2. The average roughness ratio of BET surface area to x-ray CT derived geometric surface area for Berea was found to be, β = This roughness factor is made up of contributions from the surfaces of specific minerals. 13

14 BET Normalisation 5. Compiled literature values reported in White et al. (1996) show that weathered quartz, feldspar, oxides and carbonates tend to have specific surface areas that are of the same order of magnitude while clays are a factor of 10 or more greater. 6. Using the geometric surface area fractions of each mineral, average roughness ratio β = 39, and the constraints from White (1996), gives the roughness ratios. 14

15 BET Normalisation 7. Compared here are the surface area to pore volume distribution observed in this work before, and after normalisation to BET equivalent values, and the distribution assumed in Li et al. (2007). 8. The distribution of the surface area to pore volume ratio implied in Li et al. (2007) is more heterogeneous than what was observed in this work, with a variance an order of magnitude higher than the variance. Before normalization After normalization Distribution used in Li et al s simulations 15

16 SA:PV Constraint 1. The difference in the distributions appears to originate predominantly in the assumption common to pore network model studies that surface area and pore volumes for a given pore are uncoupled. 2. There is a weak positive correlation of surface area with pore volume. 3. The surface area to pore volume ratios fall between the bounds of the ratio that would be obtained for subvolumes filled with spheres with the quartz roughness factor (β = 13) and those filled spheres with clay roughness factor (β = 130). 16

17 SA:PV Constraint 4. The quartz boundary is a theoretical minimum limit whereas the clay boundary represents the minimum surface area of a medium that is filled entirely with spheres with clay roughness. None of our sub volumes imaged were composed of only clay and this is seen here. 17

18 SA:PV Constraint 5. The uncorrelated realisation of the plot - surface areas were plotted randomly against pore volumes shows the impact of this assumption. 6. The ratio of surface area to pore volume falls well outside of the boundaries observed, with both very large surface areas and surface areas smaller than the theoretical minimum for a given pore volume. 7. The ratio of surface area to pore volume falls well outside of the boundaries observed, with both very large2 surface areas and surface areas smaller than the theoretical minimum for a given pore volume. 18

19 Conclusion 1. This work provides a statistical characterisation of pore-scale heterogeneity in the reactive surface area of minerals within permeable rocks based on direct observation of the pore morphology and mineral distribution in three dimensions. 2. Comparison of average specific surface area from BET measurements with those inferred from x-ray images showed that the two observations could be reconciled with a variation on the use of a roughness factor. Roughness factors ranged from 10 < β < 200. Total BET surface area, the presence of clay, and microporosity were all associated with higher values of the roughness factor for a given rock. 3. In the Berea sandstone, the most important factor determining abundance of the surface area of clay and feldspar group minerals in a location was the presence of the mineral in that location - increased feldspar led to increased feldspar surface area. For quartz, however, the presence or absence of clay was the most important factor. 19

20 Conclusion 4. In the Edwards carbonate, the amount of total surface area within a sub volume was the most important factor controlling the abundance of the surface area of a given mineral group. 5. The information can be used directly in statistically-based models of reactive transport including the emerging group of pore network models focused on characterising reactive processes. 6. In pore network models it will be important to put physically based constraints on combinations of properties for individual pores. 20

21 Acknowledgements The authors gratefully acknowledge funding from the Qatar Carbonates and Carbon Storage Research Centre, provided jointly by Qatar Petroleum, Shell, and Qatar Science & Technology Park. " " Contact: " Peter Lai" " Vienna, Austria April