Environ. Sci. Technol. 2004, 38, 932-936
In Situ Observation of CO2 Sequestration Reactions Using a Novel Microreaction System GEORGE H. WOLF,† A N D R E W V . G . C H I Z M E S H Y A , ‡,§ JASON DIEFENBACHER,‡ AND M I C H A E L J . M C K E L V Y * ,‡,§ Department of Chemistry and Biochemistry, Center for Solid State Science, and Science and Engineering of Materials Graduate Program, Arizona State University, Tempe, Arizona 85287
A novel, externally controlled microreaction system has been developed to provide the first in situ observations of the reaction processes that control CO2 sequestration via mineral carbonation. The system offers pressure (to 20 MPa), temperature (to 250 °C), and activity control suitable for investigating a variety of fluid-fluid and fluid-solid interactions of environmental interest. Mineral sequestration efforts to date have effectively accelerated carbonation, a natural mineral weathering process, to an industrial timescale. However, the associated reaction mechanisms are poorly understood, limiting further process development. Synchrotron X-ray diffraction and Raman spectroscopy have been used to provide the first in situ insight into the associated supercritical mineral carbonation process. Magnesite was found to form directly under the reaction conditions observed (e.g., 150 °C and 15 MPa CO2), facilitating geologically stable sequestration. Thermodynamic analysis of fluid-phase species concentrations in the Na+ buffered H2O-CO2 reaction system found the primary aqueous reactant species to be CO2(aq) and HCO3-, with CO2(aq) more prevalent under the reaction conditions observed. The microreactor provides a powerful new tool for in situ investigation of a broad range of environmentally, fundamentally, and commercially important processes, including the reactions associated with geological carbon dioxide sequestration.
Introduction Worldwide energy use is increasing exponentially, associated with the rising global standard of living (1, 2). As developing countries seek the economic wealth enjoyed by developed countries, future energy demands are expected to accelerate further. Fossil fuels satisfy over 80% of current demand, with the majority provided by coal due to its low cost, wide availability, and high energy density (3-5). Coal reserves alone can satisfy global energy needs for centuries to come, if the environmental challenges associated with CO2 emissions can be overcome (3, 5). Although the greenhouse effect of atmospheric CO2 was recognized in the late nineteenth century (6), fossil-fuel * Corresponding author phone: (480) 965-4535; fax: (480) 9659004; e-mail:
[email protected]. † Department of Chemistry and Biochemistry. ‡ Center for Solid State Science. § Science and Engineering of Materials Graduate Program. 932
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004
energy generation has historically assumed CO2 can be vented to the atmosphere without environmental consequences. Exponentially increasing atmospheric concentrations have brought this assumption under critical review (4, 5). A portfolio of candidate technologies is being investigated to evaluate their potential to provide permanent, environmentally benign, and economically viable CO2 sequestration (5). In situ process observations are central to developing the mechanistic understanding needed to effectively evaluate and engineer improved technologies. Observations under controlled “real-world” conditions can also eliminate artifacts associated with process quenching and thereby accelerate research and development. We have developed a microreaction system (7) specifically designed to enable in situ observations of the reaction mechanisms that govern mineral carbonation (8), as well as other reactions relevant to candidate CO2 sequestration technologies (e.g., geological sequestration) (5). Our primary interest is in mineral carbonation, in which anthropogenic CO2 is disposed of as geologically stable and environmentally benign mineral carbonates, thereby avoiding many of the potential legacy issues associated with other proposed longterm storage technologies (4, 5, 9). The potential for viable mineral sequestration process development has motivated active investigation by a number of groups, including the CO2 Mineral Sequestration Working Group, which is managed by Fossil Energy (U.S. Department of Energy) and includes members from the Albany Research Center, Arizona State University, Los Alamos National Laboratory, the National Energy Technology Laboratory, the University of Utah, Penn State University, and Science Applications International Corporation. Carbonation feedstocks being explored include common Mg-rich minerals, such as serpentine and olivine, which can be cost-effectively mined (e.g., $4-5/ton). These minerals are available in quantities that exceed known global coal reserves and offer large-scale sequestration potential (10-12). The primary challenge for viable process development is reducing process cost. Serpentine carbonation, which occurs via natural mineral weathering over geological time (e.g., 100 000 years), has been accelerated to near completion in less than an hour (e.g., by carbonating heat-pretreated serpentine, meta-serpentine, in an aqueous solution under supercritical CO2) (11). Although the potential for this sequestration process is encouraging, it is far from optimized and not yet economically viable. An understanding of the reaction mechanisms that govern carbonation is essential to engineering new processes that enhance carbonation reaction rates and lower cost by design. In situ observations are key to developing such understanding, as demonstrated in a recent investigation of the ocean sequestration of fossil fuel CO2 (13). Unfortunately, current reaction conditions (e.g., serpentine carbonation in a 180 °C aqueous solution under 15 MPa of CO2) are difficult to achieve and control during in situ observation using currently available systems. For these reasons, we have developed a new microreaction system which has enabled us to make the first in situ observations of CO2 mineral sequestration reaction processes.
Experimental Section The serpentine used in these investigations was collected in Globe, Arizona, and found to contain primarily 1T lizardite [a ) 5.30 (4) Å and c ) 7.29 (4) Å; space group P31m], with minor clino-chrysotile inclusions, as observed by X-ray powder diffraction and electron microscopy. Microprobe analysis found 43.9 wt % SiO2 and 40.5 wt % MgO compared 10.1021/es0346375 CCC: $27.50
2004 American Chemical Society Published on Web 12/30/2003
FIGURE 1. Schematic diagram and photographs of the microreaction system. (a) Side view of the microreactor assembly: gaskets (black), moissanite windows (white), reactor core (light gray), reactor frame (dark gray), and moissanite sample holder (red). (b) Schematic of the sample chamber dimensions, configuration, and components: exit window, a; sample holder, b; H2O-rich fluid, c; CO2-rich fluid, d; and entrance window, e. The incident and scattered X-ray beams are drawn as blue and orange lines, respectively. (c) Photograph of the empty microreactor viewed from the exit window (detector side) in which the location of the sample trough (yellow arrow) can be seen in relation to exit window. (d) Close-up photograph of the microreactor exit window shown after sample loading, evacuation, injection of the aqueous reaction solution, and carbon dioxide pressurization. The yellow arrow indicates the sample trough. The CO2-rich fluid phase can be clearly seen as a bubble above the sample trough. with the 43.4% and 43.6% expected theoretically, with the principal elemental impurities being 1.2 wt % FeO and 0.2 wt % Al2O3. Complete dehydroxylation of the mineral, as determined by thermogravimetric analysis (TGA) during heating at 2 °C/min to 1100 °C under helium, found 12.8% weight loss associated with hydroxyl water. This is in good agreement with the 13.0 wt % expected for the ideal serpentine composition Mg3Si2O5(OH)4. The 640 °C heat activated metaserpentine used as the solid reactant was prepared by heating the serpentine at 2 °C/min to 640 °C under helium, followed by rapid cooling to ambient temperature. This resulted in 11.3% hydroxyl weight loss, indicating that 88% of the parent serpentine’s hydroxyl groups were removed via dehydroxylation. The sodium bicarbonate and sodium chloride used in the aqueous mineral carbonation reaction solution were AR grade and obtained from Mallinckrodt and EM Science, respectively. The carbon dioxide used was electronic grade (99.998%) from Air Liquide. The meta-serpentine was loaded into the sample holder in the microreactor cell depicted in Figure 1. The cell was then sealed, evacuated to 10-3 Torr and filled with enough solution (0.64 M NaHCO3 + 1.0 M NaCl) to just immerse the meta-serpentine (∼60% full) and reproduce process reaction conditions as closely as possible (11). The cell was then pressurized to 15 MPa CO2 and, after an equilibration period, slowly heated. The ensuing reaction process was monitored via in situ synchrotron X-ray diffraction using the GSECARS bending magnet beamline at the Advanced Photon Source at Argonne National Laboratory. Raman scattering measurements were made using an ISA triple spectrometer (model S-3000) that was optically coupled to a modified Olympus BH-2 petrographic microscope. The spectrometer was equipped with a liquid-nitrogen-cooled CCD detector (Prin-
ceton Instruments) and a 5-watt Coherent (90-5) Ar+ laser for excitation. Micro-Raman measurements were taken with spatial resolutions ranging down to 5 microns.
Microreaction System The use of X-ray and vibrational spectroscopy for in situ reaction observation is well established in high-pressure science. Diamond-anvil-cells (DAC) and hydrothermal DACs have been successfully developed and used to carry out a wide range of high-pressure and temperature studies (14). These cells achieve high pressures by concentrating large forces on small areas via diamond culets. Because of the scarcity and high cost of suitable diamonds, such studies are often limited to very small reaction volumes (14, 15). Unfortunately, this severely limits the ability to reliably control critical reaction process parameters, including reactant activity, over the range of interest for mineral carbonation. Standard DAC pressure probes, such as ruby fluorescence, also lack the sensitivity needed to accurately monitor the moderate pressures of interest (from ambient pressure to 20 MPa). Although larger batch-style diamond cells (e.g., 0.04 cm3) have been developed to study solidfluid reactions, they do not offer external reactant activity control (16). External activity control is available in diamond flow cells designed to probe fluid phases under controlled temperatures and pressures (17). However, such cells are not well suited for the study of solid-fluid reactions under actual process conditions. To circumvent these limitations, we developed the microreaction system shown in Figure 1, which provides controlled temperature to 250 °C and pressure to 20 MPa. The design incorporates moissanite windows, which enable VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
933
a variety of X-ray, optical, and spectroscopic investigations to be performed in situ (15). In particular, they offer very good resistance to the mechanical stress associated with high pressures and are transparent to X-ray radiation for energies above 25 keV (15), enabling the synchrotron investigations described herein. The reaction volume (∼0.1 cm3) is several orders of magnitude greater than that of DAC systems (15), enabling external connection to a reactant source. The reaction cell is connected to an external high-pressure supply, which provides accurate pressure monitoring and control during in situ reaction observation. Importantly, this connection also allows reactant activity control throughout the reaction process. The temperature is controlled by an external heater which can be adjusted manually or programmed. A key feature of the large reaction volume is that it also enables the simultaneous microscopic observation of fluid-fluid and fluid-solid reaction regions of interest, as shown in Figure 1d. The chemical and thermodynamic conditions within the reactor such as temperature, pressure, and reactant activity can be remotely controlled independently, making synchrotron investigations possible.
Results and Discussion The optimal mineral carbonation process to date reacts an aqueous solution of 0.64 M NaHCO3 and 1.0 M NaCl with the pretreated serpentine feedstock under supercritical CO2 at moderate temperatures and pressures, e.g., 15 MPa CO2 and 180 °C (11). The potential for reducing process cost by lowering the reaction temperature and pressure, assuming carbonation reactivity can be maintained, makes the range of conditions down to ambient temperature and atmospheric CO2 pressure of particular interest. Two fluids are present under these conditions: CO2 rich and H2O rich. The microreactor design allows the pretreated serpentine to be immersed in the more reactive H2O-rich phase, while maintaining fluid-fluid contact between the H2O and CO2rich phases to facilitate CO2 diffusion and mimic mineral carbonation process conditions, as seen in Figure 1d. The microreactor provides simultaneous activity, temperature, and pressure control and, hence, fluid phase speciation control over the entire range of interest. To better understand the effect of microreactor control parameters on fluid speciation and its impact on carbonation reactivity, a detailed thermodynamic understanding of the supercritical CO2-H2O system is needed. Somewhat surprisingly, there is a paucity of such understanding. To provide some insight into fluid speciation, we investigated the effect of pressure and temperature on the concentrations of various fluid-phase species by analyzing the solution equilibrium equations for a Na+ buffered H2O-CO2 system. The pressure and [Na+] dependence of the concentration of aqueous dissolved CO2(aq) under typical mineral carbonation conditions is described by the following equilibrium equations:
CO2(g) f CO2(aq); KH ) [CO2(aq)]/γ CO2PCO2 (1) CO2(aq) + H2O f HCO3- + H+;
K1 ) [HCO3-][H+]/[CO2(aq)] (2)
HCO3- + H2O f H3O+ + CO32-;
K2 ) [H+][CO32-]/[HCO3-] (3)
H2O f H+ + OH-;
KW ) [H+][OH-] (4)
[Na+] + [H+] ) [OH-] + [HCO3-] + 2 [CO32-]
(5)
The last equation expresses charge neutrality. The temper934
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004
FIGURE 2. Equilibrium speciation in a Na+ buffered CO2-H2O system as a function of CO2 partial pressure at 50 °C (left panel) and 180 °C (right panel). The curves are color coded as follows: pH (black), OH- (blue), CO2(aq) (green), CO32- (red), and HCO3- (orange). ature dependence of the relevant equilibrium constants was obtained from Stumm and Morgan (18), except for KW(T) which was taken from Harned and Owen (19). Accurate temperature dependence for the Henry’s law constant was obtained from a recent reparametrization of the equation of state of pure CO2 appropriate for the temperature and pressure range of interest (20). This improves upon an earlier form valid at higher temperatures and pressures (21). The Na+ concentration is that used in the reactant solution (0.64 M NaHCO3 and 1.0 M NaCl), although ionic strength effects, including the activity of Cl-, are not yet explicitly taken into account. The equilibrium reaction equations 1-5 were solved for the concentrations of [CO2(aq)], [OH-], [HCO3-], [CO32-], and [H+]. The results, shown in Figure 2, reveal that temperature and pressure variations over the range of reaction conditions of interest lead to significant variations in speciation. The left and right panels of the figure show the results for T ) 50 °C and T ) 180 °C, respectively. According to our model, the largest variations in speciation are expected in the pressure range from 0 to 7 MPa P(CO2), i.e., below the critical pressure of CO2. At lower CO2 pressures the bicarbonate concentration dominates. A crossover occurs at ∼5-8 MPa, with the CO2(aq) concentration dominating at higher pressures, followed by the bicarbonate ion concentration. On the basis of this model, we expect similar variations in speciation over the range of reaction conditions observed in our experiments, namely, 15 MPa CO2 from 20 to 180 °C. Heat-activated meta-serpentine, obtained by serpentine heat treatment in helium to 640 °C, was used as the solid reactant for the first in situ observations of mineral carbonation. At present these conditions offer the best compromise between enhanced meta-serpentine carbonation reactivity and process economy (11). A unique feature of the microreaction system is that reaction conditions can be easily identified by directly varying temperature and/or pressure in real time. To identify the event(s) that lead to carbonation we gradually increased the temperature at a fixed carbon dioxide pressure of 15 MPa. As seen in Figure 3a, the poorly ordered meta-serpentine shows no observable change with increasing reaction temperature until reaction onset at 150 °C. At this temperature magnesite (104), (113), and (202) reflections are first observed (corresponding to 2.75, 2.10, and 1.93 Å, respectively). As the reaction progressed, the (006) and (110) reflections grew in at 2.51 and 2.31 Å, respectively, as seen for the 20-min scan shown in Figure 3a. Together these peaks account for all magnesite reflections expected in the observed range (1.8-3.1 Å) (22). Continued heating to 180 °C resulted in further carbonation and an
FIGURE 3. (a) Sequence of X-ray diffraction patterns observed as a function of temperature for the in situ mineral carbonation of meta-serpentine under 15 MPa CO2. Little change is observed until reaction onset at 150 °C, when the (104), (006), (110), (113), and (202) magnesite reflections are observed to form (orange curve) at 2.75, 2.51, 2.31, 2.10, and 1.93 Å, respectively. Heating to 180 °C resulted in further carbonation and confirmed the stability of the magnesite formed (red curve). The lower temperature spectra (black curves) are all 5-min scans, whereas the 150 and 180 °C spectra were obtained over 20 min. The latter intensities were normalized for the purpose of comparison. Each scan was taken under isothermal conditions ((1 °C) using an X-ray wavelength of 0.3311 Å. (b) Raman spectra of the carbonated meta-serpentine showing the characteristic magnesite shifts at 1094 and 331 cm-1 (indicated by the arrows). The features marked by asterisks are produced by the microreactor’s moissanite (hexagonal SiC) windows. overall increase in magnesite scattering intensity. No reflections associated with any silica-containing species were observed at any time, suggesting the silica-like products formed are highly structurally disordered. All of the crystalline reflections that appear from reaction onset to completion are exclusively associated with magnesite (MgCO3), indicating that magnesite forms directly under the observed reaction conditions, without observable intermediate formation, consistent with the following reaction:
Mg3Si2O6.76(OH)0.48 + 3CO2 f 3MgCO3 + 2SiO2 + 0.24H2O (6) Our initial observations indicate the nucleation of magnesite is a key step in the reaction process. In addition, because nucleation may be influenced by the wide range of compositional variation found in “as-mined” serpentine minerals (23), it can also be rate limiting under certain conditions. In this event, seeding the reaction solution with magnesite nuclei may circumvent this limitation. It should be emphasized that direct magnesite formation is preferred over the formation of other higher-energy intermediates, such as various
hydrous phases, for permanent CO2 sequestration, because of its proven environmental stability over geological time. The lower process costs associated with systems that have the potential to exhibit enhanced carbonation reactivity at lower reaction temperatures and pressures has motivated us to begin to explore related systems. Preliminary experiments on a similarly prepared meta-serpentine derived from lizardite obtained from the Philips Mine in Globe, Arizona, indicate that magnesite can form at temperatures as low as 100 °C at 15 MPa CO2. However, this may not persist for new feedstock materials that are engineered to carbonate at still lower reaction temperatures and pressures. For example, heat-activated Mg(OH)2, which is much more carbonation reactive, is known to form amorphous carbonate and/or hydroxycarbonate at lower temperatures and ambient CO2 pressure (9). The formation of other intermediate or final products, such as hydromagnesite, may also be found as reaction temperatures and pressures are reduced. Understanding the reaction mechanisms and conditions that govern the formation of these materials is a central objective in sequestration process development. The in situ microreaction system described above is ideally suited for such investigations, which are currently underway in our laboratories and in collaboration with GSECARS at the Advanced Photon Source at Argonne National Laboratory. A series of Raman spectra were acquired to explore the potential of using the microreactor with Raman spectroscopy for rapid carbonation reaction product characterization. As seen in Figure 3b, the primary magnesite Raman shifts at 1094 and 331 cm-1 are clearly observable, along with the strong background peaks from the microreactor’s moissanite windows. All of the peaks observed are associated with magnesite or moissanite. Our Raman results underscore the ability of the microreaction system to be used with a variety of analytical techniques. For example, infrared and nuclear magnetic resonance spectroscopy can also be readily accommodated for use as in situ probes. Such capabilities can be directly applied to a variety of processes, including the in situ observation of fluid-phase speciation during both solidfluid and fluid-fluid reactions. Combined with the ability to simultaneously image (e.g., X-ray and optical) the associated processes, the microreaction system provides an effective microscopic laboratory for in situ process investigations with full temperature and pressure control. The present study indicates that the microreaction system can enable detailed in situ investigations for a variety of environmental, chemical, and materials processes involving both subcritical and supercritical fluids (24). Systems to which it can be applied include organic and organometallic reactions, catalytic reaction processes, pharmaceutical materials processing, organic waste decomposition, geochemical and mineralogical reactions, and solvothermal materials synthesis reactions. For example, organic synthesis reactions using supercritical CO2 fluids can be used to eliminate the organic waste solvents used in traditional methods and the environmental impact associated with their use and waste disposal. In addition, the system can further the evaluation and development of other candidate CO2 sequestration technologies. An example is geological sequestration (5), in which the microreaction system can simulate the conditions associated with the underground injection of carbon dioxide and directly observe the associated reaction processes (25).
Acknowledgments We thank Guoyin Shen and Vitali Prakapenka for their assistance with the synchrotron experiments performed at the GSECARS beamline at the Advanced Photon Source at Argonne National Laboratory. This work was supported by Power Systems Advanced Research (Fossil Energy) via U.S. Department of Energy NETL/ANL Contract 1F-01262. VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
935
Literature Cited (1) Siegenthaler, U.; Oeschger, H. Tellus 1987, 398, 140. (2) Keeling, C. D.; Whorf, T. P.; Wahlen, M.; van der Plicht, J. Nature 1995, 375, 666. (3) United Nations. 1991 Energy Statistics Yearbook; New York, 1993. (4) Herzog, H.; Drake, E.; Adams, E. CO2 Capture, Reuse, and Storage Technologies for Mitigating Global Climate Change; DOE Report DE-AF22-96PC01257; U.S. Government Printing Office: Washington, DC, 1997. (5) Carbon Sequestration Research and Development. Offices of Science and Fossil Energy, U.S. Department of Energy; U.S. Government Printing Office: Washington, DC, December, 1999, and references therein. (6) Arrhenius, S. Philos. Mag. 1896, 41, 237. (7) McKelvy, M. J.; Diefenbacher, J.; Wolf, G.; Chizmeshya, A. V. G. Patent pending. (8) Seifritz, W. Nature 1990, 345, 486. (9) Be´arat, H.; McKelvy, M.; Chizmeshya, A.; Sharma, R.; Carpenter, R. J. Am. Ceram. Soc. 2002, 85, 742. (10) Lackner, K.; Wendt, C.; Butt, D.; Joyce, E., Jr.; Sharp, D. Energy 1995, 20, 1153. (11) O’Connor, W. K.; Walters, R. P.; Dahlin, D. C.; Rush, G. E.; Nilsen, D. N.; Turner, P. C. Proceedings of the 26th International Technical Conference on Coal Utilization & Fuel Systems; Coal Technology Association: Gaithersburg, MD, 2001; p 765. (12) Butt, D. P.; Lackner, K. S.; Wendt, C. H.; Benjamin, A. S.; Currier, R.; Harradine, D. M.; Holesinger, T. G.; Park, Y. S.; Rising, M. World Resour. Rev. 1997, 9, 324. (13) Brewer, P. G.; Friederich, G.; Peltzer, E. T.; Orr, F. M., Jr. Science 1999, 284, 943.
936
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004
(14) Mao, H. K.; Hemley, R. J. Rev. Mod. Phys. 1994, 66, 671. (15) Xu, J.; Mao, H. Science 2000, 290, 783. (16) Fulton, J. L.; Darab, J. G.; Hoffman, M. M. Rev. Sci. Instrum. 2001, 72, 2117. (17) Hoffman, M. M.; Addleman, R. S.; Fulton, J. L. Rev. Sci. Instrum. 2000, 71, 1552. (18) Stumm, W.; Morgan, J. J. Aquatic Chemistry, 3rd ed.; John Wiley & Sons: New York, 1995. (19) Harned, H. S.; Owen, B. B. The Physical Chemistry of Electrolytic Solutions; Van Nostrand: New York, 1958. (20) Duan, Z. H.; Sun, R. Chem. Geol. 2003, 193, 257. (21) Duan, Z. H.; Moller, N.; Weare, J. H. Geochim. Cosmochim. Acta 1995, 59, 2869. (22) U.S. National Bureau of Standards Circular 539; U.S. Department of Commerce, National Bureau of Standards: Washington, DC, 1957; pp 7-28. (23) Faust, G. T.; Fahey, J. J. The Serpentine-Group Minerals; Geological Survey Prof. Paper 384-A; U.S. Department of the Interior: Washington, DC, 1962. (24) Fletcher, P. D. I.; Haswell, S. J.; Pombo-Villar, E.; Warrington, B. H.; Watts, P.; Wong, S. Y. F.; Zhang, X. Tetrahedron 2002, 58, 4735. (25) Kaszuba, J. P.; Janecky, D. R.; Snow, M. G. Appl. Geochem. 2003, 18, 1065.
Received for review June 20, 2003. Revised manuscript received October 16, 2003. Accepted November 4, 2003. ES0346375
