Dissociation Behavior of CO2 Hydrate in Sediments during Isochoric

Oct 20, 2008 - As CO2 is sequestered into sediments in the oceanic environment, CO2 hydrate can form as a byproduct. This study explored the dissociat...
0 downloads 0 Views 5MB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 8571–8577

Dissociation Behavior of CO2 Hydrate in Sediments during Isochoric Heating TAE-HYUK KWON Graduate Student, Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, Korea HAK-SUNG KIM Graduate Student, Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, Korea GYE-CHUN CHO* Associate Professor, Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, Korea

Received April 22, 2008. Revised manuscript received July 1, 2008. Accepted September 8, 2008.

As CO2 is sequestered into sediments in the oceanic environment, CO2 hydrate can form as a byproduct. This study explored the dissociation behavior of CO2 hydrate in sediments in relation to pore fluid pressure evolution and sediment particle size. We synthesized CO2 hydrate in three types of particulate sediments: glass beads, fine sand, and crushed silt. We then dissociated them through isochoric heating. We observed the excess pore fluid pressure build-up and self-preservation behavior, in which the pressure-temperature state evolves along the hydrate phase boundary until either it reaches the second quadruple point or all hydrates dissociate. The pore fluidpressureevolutionislimited,however,bytheCO2 vapor-liquid phase equilibrium boundary due to the liquefaction of CO2. The presence of CO2 liquid in sediments forces the pressuretemperature evolution to follow the CO2 vapor-liquid phase equilibrium boundary, regardless of hydrate formation and dissociation processes. CO2 hydrate in fine-grained sediments experiences capillary pressure-induced melting point depression, but this effect vanishes when the pores exceed ∼1 µm, such as in coarse-grained sediments. In particular, any fracture generation in sediments which involves the local release of confinement eliminates the melting point depression induced by the capillary effect.

1. Introduction Storing carbon dioxide (CO2) captured from fossil fuel plants in deep marine sediment formations is a promising method for stabilizing atmospheric greenhouse gas concentrations (1, 2). An essential physical trapping mechanism in the geologic formation is the presence of a cap rock which acts as an upper seal to prevent fluid flow out of a reservoir. As CO2 in liquid or supercritical phase is injected in an oceanic sediment environment, diffused and transported CO2 mol* Corresponding author phone: +82-42-869-3622; fax: +82-42869-3610; e-mail: [email protected]. 10.1021/es801071e CCC: $40.75

Published on Web 10/18/2008

 2008 American Chemical Society

ecules will produce carbon dioxide hydrate (CO2 hydrate) as a byproduct in shallower sediments (2). CO2 hydrate will form preferentially in uncemented sediments within 200 m below the seafloor which are mostly at geothermal temperatures less than about 283 K. The CO2 hydrate formation serves as an additional seal cap, impeding the flow of CO2 liquid and reducing the diffusion rate of dissolved aqueous CO2. Thus, this phenomenon invokes an attractive idea for the permanent storage of greenhouse gases into geologic formations. Gas hydrates are solid compounds which favor high pressure and low temperature. Since natural methane hydrate has been found, a plethora of research related to gas hydrates (mostly methane hydrate) has been conducted, focusing on topics such as new energy resources, hydrate-related geohazards, and global warming. Interest is also growing in CO2 hydrate and CO2 hydrate-bearing sediments, both of which are potential byproducts of carbon sequestration. Structurally, CO2 hydrate is similar to methane hydrate. Both are structure I, with a unit cell consisting of two small cavities (i.e., pentagonal dodecahedra, 512) and six large cavities (i.e., tetrakaidecahedra, 51262) with the same ideal hydration number of 5.75 (3, 4). The structural similarities between methane and carbon dioxide hydrate open the possibility for CO2 sequestration to be used simultaneously as a potential method for methane recovery via methane replacement by CO2 in hydrate, and the mitigation of global warming via long-term sequestration of CO2 in hydrate form (4-6). In addition, CO2 hydrate may be a suitable analogue for natural methane hydrates in terms of the mechanical and thermal behaviors of hydrate-bearing sediments. Although numerous hydrate studies related to CO2 disposal into oceans have been conducted on topics such as the formation and dissolution of CO2 hydrate (7-12), the fate of CO2-rich water, CO2 liquid and CO2 hydrate in oceans (13-15), and ecological studies around CO2 reservoirs (16), the dissociation behavior of CO2 hydrate-bearing sediments has received little attention. In fact, CO2 hydrate-bearing sediments may unavoidably destabilize as a result of geological processes, including heat transport from highlatitude to low-latitude by abnormal thermohaline circulation (17), bottom water warming induced by global warming (18), mantle-derived intrusion (19), uplift by plate tectonic mechanism (20), and volcanism (21); or as a result of drilling operations (22, 23). However, the consequences of CO2 hydrate dissociation in sediments remains to be solved. Thus, this study explored the dissociation behavior of CO2 hydratebearing sediments in relation to fluid pressure evolution and sediment particle size. We synthesized CO2 hydrates in three types of particulate sediments (i.e., glass bead, fine sand, and crushed silt), and then dissociated them through heating. All samples were tested under isochoric conditions (i.e., constant volume condition) with no fluid flux, in which the rate of hydrate dissociation was much faster than the rate of pore pressure dissipation. This condition is expected in various oceanic sediment environments where the thermal diffusion coefficient is typically higher than the pressure diffusion coefficient and when the expansion of the sediment is restricted by the stiffness of the surrounding medium. The manuscript describes the experimental procedures and results, and it discusses important physical phenomena, including self-preservation and the capillary effect.

2. Experimental Program 2.1. Experimental Setup and Testing Materials. Our experiments were designed to explore the dissociation behavior VOL. 42, NO. 22, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8571

TABLE 1. Properties of the Tested Soilsa

a D50 is the mean particle diameter, Gs is the specific gravity, Sa is the specific surface area, and nmax and nmin are the maximum and the minimum porosity, respectively. Sphericity is defined as the surface area of a sphere with the same volume as the particle relative to the actual surface area. Roundness is the average radius of curvature of surface features relative to the radius of the maximum sphere that can be inscribed. *Porosity represents the porosity of the sediment sample used in the study. †The pore sizes of the glass bead sample and the fine sand sample are calculated by assuming simple cubic packing and tetrahedral packing of the spherical particles with the mean particle diameter. ††The pore size of the crushed silt sample is inversely calculated based on the experimental result of the melting point depression. Data: Cho et al. (46) and Yun et al. (47).

of gas hydrate-bearing sediments (refer to Supporting Information Figure S1). The experiments were conducted in a transparent, cylindrical, and rigid-walled reaction cell (polycarbonate; volume 3.17 cm3; internal diameter 6.35 mm; height 100 mm). The reaction cell was instrumented with one thermocouple (TMTSS; grounded sheath T-type; OMEGA) and one pressure transducer (PX302; OMEGA). The temperature and pressure inside the reaction cell were measured during the formation and dissociation processes. The temperature of the reaction cell was controlled by circulating temperature-controlled fluids from a refrigerating circulator (Fisher Scientific). The pressure of the pore fluid in the sediments was controlled by the water pump (up to 10 MPa) and the compressed gas pressure (up to 4.4 MPa) of the CO2 gas cylinder. We prepared glass beads (uniform grain size; mean diameter ) 1.5 mm), fine sands (Ottawa F110; uniform grain size; mean diameter ) 0.12 mm), and crushed silts (silica flour; mean diameter ) 20 µm) as host sediments to study how hydrate interacts with various soils. The chosen sediments represent a broad range of the lithologies found in natural hydrate-bearing sediments. Particle images and the physical properties of the tested sediment samples are listed in Table 1. All samples were tested with a porosity φ consistently around 0.4. The pore sizes (i.e., effective pore diameter for spherical pore shape) of the glass bead sample and the Ottawa F110 sand sample are calculated under the assumption that the mineral particles are spherical and have a uniform grain size. Simple cubic packing corresponds to the largest pore size while tetrahedral packing applies to the smallest pore size. Meanwhile, the crushed silt sample contains various sizes of mineral particles. Small particles will occupy the pore spaces among large particles and significantly reduce their pore size. The pore size of the crushed silt sample is inversely calculated based on the experimental result involving capillarity. (Note that the details are described in the Results and Discussion section.) The gas hydrate former in the experiment was a scientificpurpose CO2 gas (purity ∼ 99.9%). Distilled water was prepared to saturate the sediments and to form gas hydrate inside the sediments. No electrolyte was added in this study 8572

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

in order to exclude the effect of aqueous electrolytes on thermodynamic phase boundaries and on dissociation kinetics induced by decreased water activity. 2.2. Preparation of Hydrate-Bearing Sediments. Each sediment sample was compacted in the reaction cell by handtamping so as to achieve a final and uniform porosity of 0.4. It was partially water-saturated with distilled water, and no effective stress was applied (i.e., only a self-weight stress without external confining stress). CO2 gas was injected and pressurized to about 3.2 MPa, making sure that the pressure did not exceed the liquefaction pressure. The sample was cooled down and maintained at 274.15 K until hydrate formation took place. Hydrate nucleation was presumed from a temperature jump which was caused by the exothermic reaction of the phase transformation. After the hydrate nucleation, water and CO2 gas were sequentially flushed through the specimen to make sure that a considerable quantity of CO2 hydrate formed in the pores. Finally, cool water (∼278.15 K) was flown into the reaction cell to saturate the sediment and eliminate CO2 bubbles. Then, we adjusted the fluid pressure to the equilibrium pressure (i.e., around 3 MPa for the glass bead sample and around 1.5 MPa for the fine sand sample and the crushed silt sample) before commencing gas hydrate dissociation. The gas hydratebearing sediment samples in the reaction cell were left to stabilize for more than one day. 2.3. ProceduressHeating/Cooling Cycles. A constant volume condition was made by closing the valves around the reaction cell (i.e., isochoric condition with no mass flux) before heating. The samples were heated at 1 K every 6 h stepwise without allowing fluid flux. The CO2 hydrate-bearing samples were heated up to 289.15 K (except for the glass bead sample, which was heated to 293.15 K), monitoring the liquefaction of CO2 vapor. Then, the samples were continuously cooled again to experience hydrate formation (∼2.5 K/hr). These heating-cooling cycles were conducted to confirm the repeatability of the experiments (refer to Supporting Information Table S1).

FIGURE 1. P-T diagrams during heating-cooling cycles of CO2 hydrate (a) in the glass bead sample, (b) in the fine sand sample, and (c) in the crushed silt sample. Phase equilibrium boundary for bulk CO2 in unconfined pores are replotted based on HWHYD software (a demo version of this software is available at http://www.pet.hw.ac.uk/reasearch/hydrate).

3. Results and Discussion 3.1. Gas Hydrate Dissolution in Sediments. The glass bead sample experienced a slight pressure increase with rising temperature when the pressure-temperature (hereafter P-T) condition was in the hydrate stability region (i.e., from 276.6 to 281.2 K; the first heating in Figure 1a). The small increase in the fluid pressure with temperature (i.e., approximately 0.1 MPa/K) was induced by the dissolution of CO2 hydrate and the thermal expansion of each constituent.

The presence of gas hydrates promotes further consumption of dissolved gas in water, and accordingly, the gas solubility decreases with the temperature decrease in the hydrate stability region (24-29). The increase in temperature within the stability zone leads to a breakdown of the hydrate structure due to the increased gas solubility in the surrounding pore water. Hydrate dissolution releases water and dissolved gas; there is no free gas produced. The dissolution generates smaller pore pressure changes than dissociation, which produces free gas (30-32). 3.2. Self-Preservation Behavior during Gas Hydrate Dissociation. Hydrate dissociation started when the P-T state reached the equilibrium boundary, and heat was supplied fast enough to prevent a temperature drop due to the endothermic hydrate dissociation process. The pressure evolution during temperature increase for each sample is shown in Figure 1. The glass bead sample containing CO2 hydrate began the dissociation process from 281.2 K and 3.4 MPa in the first heating test (Figure 1a). CO2 hydrate that was formed in the Ottawa F110 sand sample commenced dissociating from 276.2 K and 1.7 MPa in the first heating test (Figure 1b), whereas the hydrate formed in the crushed silt sample started dissociating from 273.7 K and 1.42 MPa in the first heating test (Figure 1c). The dissociation processes for all the samples ended at approximately 283.7 K and 4.5 MPa which was expected to correspond to the second quadruple point of the CO2-water system (Q2). The pore fluid pressure of the coarse-grained sediment samples (i.e., glass bead and Ottawa F110 sand) evolved with temperature increase along the water-hydrate-gas phase equilibrium of bulk CO2 in unconfined pores. No capillarity effect in coarsegrained sediment samples was found during the dissociation since their pore sizes were much greater than 1 µm (see Figures 1a and b). However, the dissociation in the finegrained sediment sample (i.e., crushed silt) started at a lower temperature than in the coarse-grained sediments, and pressure evolved along the shifted phase boundary (Figure 1c; the detailed calculation of the shifted phase boundary is discussed in the later section). This result reveals an apparent capillary effect in the fine-grained sediments. Gas hydrate dissociation releases gas-saturated water and water vapor saturated gas. High excess pore fluid pressure is anticipated because fluid flow and sediment volume expansion are restrained. However, the high excess fluid pressure generated by dissociation facilitates the reformation of gas hydrate, and hinders further dissociation, showing “self-preservation.” As a result, the pressure increases along the CO2 hydrate P-T boundary (LW-H-VCO2 curve in Figure 1) during the thermally driven dissociation under a constant volume condition until either the P-T state reaches the second quadruple point or all hydrates dissociate (33, 34). A complementary effect takes place during dissociation induced by depressurization under isothermal conditions where the released gas from hydrate breakdown boosts the pressure and recovers the equilibrium pressure, which ceases further hydrate dissociation. Furthermore, dissociation under adiabatic conditions causes a decrease in temperature by the endothermic reaction, which slows hydrate dissociation (35). 3.3. Upper Limit of Pressure EvolutionsBeyond Hydrate Dissociation. At the second quadruple point (Q2, 283.7 K and 4.5 MPa in this study), all the hydrate remaining was dissociated (Figure 1). The plateau region in the temperature (i.e., from 740 to 780 min in Figure 2a) infers the rapid dissociation of gas hydrate due to an endothermic reaction. Meanwhile, the rate of the pore fluid pressure increase drops relative to the rate at lower temperatures because the CO2 gas released from hydrate dissociation is liquefied and mass transition from the CO2 vapor phase to the liquid phase hinders a rapid increase in pore fluid pressure (Figure 2b). VOL. 42, NO. 22, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8573

FIGURE 2. Rapid dissociation of CO2 hydrate at the second quadruple point (Q2) in the glass bead sample. Due to the endothermic reaction of hydrate dissociation, a plateau region in the temperature is shown. After the second quadruple point (Q2), pressure evolution by hydrate dissociation is diminished by the liquefaction of released CO2 vapor. Once dissociation is complete, the pore fluid pressure is limited by the CO2 liquid-vapor phase equilibrium boundary. Consequently, water (with dissolved gas), CO2 gas (with water vapor), CO2 liquid (with dissolved water), and the solid mineral phase are present concurrently in an equilibrium state (LW-VCO2-LCO2). Thereafter, an additional increase in the temperature under a constant volume condition causes a slight increase in the pressure which is induced by the thermal expansion of the phases. This phenomenon can be found in other gas hydrates which have the second quadruple point, such as ethane hydrate (287.8 K, 3.4 MPa; data from Sloan (3)), propane hydrate (278.8 K, 0.56 MPa; data from Sloan (3)), and iso-butane hydrate (275.0 K, 0.17 MPa; data from Sloan (3)). 3.4. Additional Cooling-Heating CyclesHigh Hydrate Saturation. We confirmed the hydrate formation from the exothermic signals at around 274 K during the subsequent cooling processes. CO2 hydrate is expected to form in three ways: (1) from dissolved CO2 in an aqueous solution; (2) at the interface between CO2 liquid and water; and (3) at the interface between CO2 gas and water. As the hydrate formation proceeded and consumed CO2, the pore fluid pressure of the sand sample and the silt sample decreased as shown in Figures 1b and c. However, the pore fluid pressure of the glass bead sample did not show a pressure drop due to the presence of CO2 liquid (Figure 1a). It is expected that the glass bead sediment sample initially contained much more CO2 hydrate, which had been formed at the beginning of sample preparation and produced more CO2 liquid after complete decomposition than the other samples, such that CO2 liquid could not be completely consumed for hydrate formation for a given time (i.e., 72 h at 274 K and 3.5 MPa). During the third heating experiment of the glass bead sample which contained water, hydrate, CO2 gas, and CO2 liquid, its P-T state stayed along the CO2 liquid-vapor phase equilibrium boundary (i.e., H-VCO2-LCO2 and Lw-VCO2-LCO2), as shown in Figure 1a. This behavior implies that if CO2 liquid is present, CO2 liquid in sediments governs the pressuretemperature evolution regardless of gas hydrate presence. 3.5. Melting Point Depression in Fine-Grained Sediments. In Figure 3a, the P-T trace of the crushed silt sample during the isochoric heating deviated from the phase equilibrium boundary of bulk CO2 hydrate. This phenomenon is called “melting point depression,” and is attributed to the capillary effect of small pores in the fine-grained sediment sample. Water is thermodynamically preferred over the hydrate phase on mineral surfaces. Therefore, water wets 8574

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

FIGURE 3. (a) Capillary effect of the crushed silt sample on phase equilibrium boundary. (b) Digital images of the sample. As a fracture was generated and CO2 hydrate was released into an unconfined condition, the effect of capillarity disappeared, and the hydrate-bearing sediment was destabilized along the phase boundary of bulk CO2 hydrate. The red circles indicate the region where the fracture generated. the mineral surface at the hydrate-water interface and encloses the hydrate mass within a concave surface. As the pore size in sediments decreases, water activity decreases and the hydrate phase boundary shifts toward lower temperatures (i.e., the freezing/melting point depression, see data in refs 36-38 and the discussion in ref 39). The melting point depression ∆Tdep from the equilibrium temperature in unconfined conditions Tbulk can be calculated using the Gibbs-Thomson equation with consideration of the total curvature of the solid surface in terms of the effective pore size d (40-44). ∆Tdep ) -

(

)

2 γhwmhcos θ Tbulk Cylindrical hydrate shape d Fh0Lf (1a)

∆Tdep ) -

(

)

4 γhwmhcos θ Tbulk Spherical hydrate shape d Fh0Lf (1b)

where Fh0 is the density of CO2 hydrate () 1046 ∼ 1130 kg/m3 depending on cage occupancies, pressure, and temperature; herein 1065 kg/m3 from Anderson et al. (38)), mh is the molar mass of the CO2 hydrate (herein 174 g/mol with hydration

number of 7.23 from Anderson et al. (38)), Lf is the latent heat of dissociation of CO2 gas hydrate (herein 65.2 kJ/mol from Anderson et al. (38)), and γhw is the surface tension between CO2 hydrate and water (herein 0.030 N/m from Anderson et al. (38)). Disseminated gas hydrate crystals may be encapsulated by spherical water films and experience a melting point depression twice as large as that of cylindrical gas hydrates (44). The shifted phase boundaries of CO2 hydrate in small pores are calculated and superimposed with the experimental results, as shown in Figure 3a, in order to find the smallest pore that contains gas hydrate as suggested by Anderson et al (43). We disregard changes in pore size during heating and consider the contact angle between the water and the pore wall to be θ ) 0 by assuming that minerals are coated by a water film. The analysis suggests that the smallest pore size dmin that contains gas hydrate with a cylindrical surface is approximately 50 nm in the first heating test. In the same manner, the smallest pore size dmin in the third heating test is approximately 40 nm. However, as the hydrate surface is considered to be spherical in shape, the smallest pore size is around 80 ∼ 100 nm. The prediction of the maximum pore size is restricted because the sample has experienced fracturing and the capillarity has disappeared. CO2 hydrate in fine-grained sediments experiences capillary pressureinduced melting point depression, but this effect vanishes as pores exceed ∼1 µm, such as in coarse-grained sediments. No capillary effect is anticipated when hydrates are formed in nodules, veins, or lenses in which hydrate chunks push sediment skeletons and expand pore spaces. 3.6. Fracture Generation during Hydrate Dissociation. Hydraulic fracturing in unconsolidated sediments primarily depends on the confining stress state and the relative rate between the pressure increase and the pressure dissipation. Although all tested sediment samples were subjected to the same amount of pore fluid pressure increase, only the crushed silt sample experienced fracture generation (refer to Figure 3b). The fluids released from the hydrate render overpressurized local pores; consequently overpressurized pore fluids produce a fracture in the sediment due to the low permeability of the crushed silt sample, contrary to coarse-grained sediment samples. In addition, as the tested sediment sample was under no vertical confining stress, the lack of a confining stress apparently facilitated the fracture generation. As the dissociation proceeded, showing the melting point depression in the crushed silt sample, the pore fluid pressure approached and increased along the bulk CO2 hydrate equilibrium boundary after about 282.15 K (Figure 3a). A similar phenomenon was also observed during the third heating test, in which the P-T trace stayed away from and merged into the phase boundary of bulk CO2 hydrate at a temperature similar to that in the first heating test. This merging phenomenon coincided with the fracture generation in the sample, as observed in Figure 3b. We note that the pressure stayed nearly constant as the temperature increased until the P-T state reached the bulk hydrate phase boundary (refer to the third heating test in Figure 3a). While there was indeed a local release of confinement due to the fracture generation, the rest of the sample behaved as it would be in the absence of a fracture. Thus, the P-T state in the fracture zone was within the hydrate stability region so that reformation of hydrates took place in the fracture zone. Meanwhile, the hydrates far from the fracture zone (no fracture zone) still experienced local confinement induced by the capillary effect and were out of the hydrate stability so that they dissociated. Conclusively, while other hydrates in the neighboring sediments dissociated, hydrate newly formed in the fractured space, which buffered the pressure increase that should occur. This continued until the sample was so warm that hydrate had to dissociate in the fracture as well as in the

small pores, and the pressure followed the bulk hydrate equilibrium curve. The fracture created a situation where continued heating can cause dissociation in one location (all those fine-grained pores), but hydrate formation in another location (the fracture). As a result, the fracture generation in sediments which involves the local release of confinement eliminates the melting point depression; hence, the pore fluid pressure is governed by the bulk hydrate phase boundary (i.e., lower equilibrium pressure for a given temperature) over the phase equilibrium boundary of confined small pores. 3.7. Implications to CO2 Sequestration. CO2 hydrate, in general, is stable when the temperature is lower than 283 K and the pressure is more than approximately 1.5 MPa. Therefore, CO2 hydrate will preferentially nucleate and form in sediments within ∼200 m below the seafloor (assuming the bottom water temperature is 277.15 K and the geothermal gradient is 40 K/km). If any geologic perturbation is large enough to take pressure or temperature out of the stability region (17-23), dissociation of the CO2 hydrate in the sediment may cause geologic hazards such as submarine slope failures (30, 45). The amount of excess pore pressure generated by hydrate dissociation depends on whether CO2 is released in vapor phase or liquid phase, which water depth determines. CO2 hydrate dissociation releases CO2 liquid when the hydrostatic pressure is more than 4.5 MPa (e.g., 450 m water depth). Assuming that the densities of CO2 liquid and CO2 hydrate at 5 MPa are 900 kg/m3 and 1100 kg/m3, respectively, one unit volume of CO2 hydrate V dissociates at a constant pressure of 5 MPa, occupying a combined volume of 0.88 V where water and CO2 liquid volumes are 0.64 V and 0.24 V, respectively. The dissociation under this circumstance renders the contraction of the pore fluids, therefore it does not involve the excess pore pressure generation. On the contrary, when the water depth is less than 350 m (