Article pubs.acs.org/est
Influence of High-Temperature Steam on the Reactivity of CaO Sorbent for CO2 Capture Felix Donat,† Nicholas H. Florin,†,* Edward J. Anthony,†,‡ and Paul S. Fennell† †
Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, United Kingdom CanmetENERGY, Natural Resources Canada, 1 Haanel Drive, Ottawa, Canada K1A 1M1
‡
S Supporting Information *
ABSTRACT: Calcium looping is a high-temperature CO2 capture technology applicable to the postcombustion capture of CO2 from power station flue gas, or integrated with fuel conversion in precombustion CO2 capture schemes. The capture technology uses solid CaO sorbent derived from natural limestone and takes advantage of the reversible reaction between CaO and CO2 to form CaCO3; that is, to achieve the separation of CO2 from flue or fuel gas, and produce a pure stream of CO2 suitable for geological storage. An important characteristic of the sorbent, affecting the cost-efficiency of this technology, is the decay in reactivity of the sorbent over multiple CO2 capture-and-release cycles. This work reports on the influence of hightemperature steam, which will be present in flue (about 5−10%) and fuel (∼20%) gases, on the reactivity of CaO sorbent derived from four natural limestones. A significant increase in the reactivity of these sorbents was found for 30 cycles in the presence of steam (from 1−20%). Steam influences the sorbent reactivity in two ways. Steam present during calcination promotes sintering that produces a sorbent morphology with most of the pore volume associated with larger pores of ∼50 nm in diameter, and which appears to be relatively more stable than the pore structure that evolves when no steam is present. The presence of steam during carbonation reduces the diffusion resistance during carbonation. We observed a synergistic effect, i.e., the highest reactivity was observed when steam was present for both calcination and carbonation.
■
INTRODUCTION Carbon capture and storage (CCS) is an important part of the strategy for mitigating the risks of climate change, recognizing the continuing global reliance on fossil fuels. CCS technology is in the early demonstration stages, likely to last at least a decade, with governments currently developing and implementing strategies to accelerate commercial deployment (including significant activity in the U.S., E.U., and Australia).1 This work is focused on Calcium Looping CO2 capture technology. Calcium Looping is based on the reversible reaction between CaO and CO2, according to eq 1: CaO + CO2 ↔ CaCO3 (1) Shimizu et al.2 described a system consisting of two interconnected fluidized bed reactors which used the reversible reaction to achieve CO2 separation from a power plant flue gas and produce a pure stream of CO2 suitable for storage (Figure 1). Flue gas enters the first fluidized bed reactor (carbonator), where CO2 is captured by CaO and converted to CaCO3. CaCO3 is then transported to the second reactor (calciner/oxyfuel combustor) where it is decomposed to regenerate CaO and produce a concentrated stream of CO2. The regeneration step requires the input of energy to increase the temperature (from about 650 °C to about 900 °C) and drive the endothermic © 2011 American Chemical Society
Figure 1. Calcium looping technology applied to capturing CO2 from a combustion flue gas (after Shimizu et al.)2.
calcination reaction. This is achieved by burning additional fuel in pure oxygen, to avoid dilution of the CO2 stream with N2 from the air. The advantages of the proposed scheme compared Received: Revised: Accepted: Published: 1262
August 3, 2011 November 2, 2011 December 14, 2011 December 14, 2011 dx.doi.org/10.1021/es202679w | Environ. Sci. Technol. 2012, 46, 1262−1269
Environmental Science & Technology
Article
quantities of deionised water directly through a heated line into the quartz tube. A fraction of the exhaust gas was continuously sampled using a stainless steel tube inserted into the top of the quartz linear; the sampled gas passed through a heated section of tubing (about 110 °C) housing a relative humidity probe (Vaisala HMP75), then through an ice bath to condense out the water and a static particle trap (quartz wool). A nondispersive infrared CO2 gas analyzer (ADC SB100) was used to measure the CO2 concentration every second and the relative humidity probe was used to measure the steam concentration. The sorbent reactivity was determined by regularly measuring and recording the concentration of CO2 (and/or H2O in the exhaust gas) and is presented in terms of the carrying capacity per cycle N (CN), defined as the number of moles of CO2 captured (nCO2) per mole of CaO (nCaO) loaded in the reactor, according to eq 2 (Here, MWCaCO3 is the molecular weight of CaCO3, mlimestone,initial is the mass of limestone initially loaded to the reactor and xCa is the corresponding content of Ca, as shown in Table S1 of the SI). An increase in the CO2 concentration corresponded with calcination and a decrease corresponded with carbonation.
to technologies closer to market include: (1) use of cheap and nontoxic sorbent derived from limestone; (2) use of mature fluidized bed technology; (3) lower energy penalty and operating costs; and (4) potential synergy by using the spent CaO in cement manufacture. There are drawbacks, most significantly the decay in the reactivity of the sorbent through multiple CO2 capture-and-release cycles. Factors known to influence the sorbent reactivity include sintering, attrition, and reaction with impurities in the flue or fuel gas, primarily sulfur species (SO2, H2S). Steam will be present in combustion flue gas (5−10%), in the oxy-fuel combustor/calciner and fuel gases (∼20%) and so the influence of steam during calcination and carbonation is relevant for determining the reactivity through multiple cycles. However, to date, most previous work in the field has examined the reactivity of CaO sorbents using dry gas mixtures. That said, there is a significant body of relevant literature from previous studies carried out in different contexts and there is general agreement that the presence of steam, even at low concentrations: (i) increases the rate of sintering of CaO;3−5 (ii) increases the rate of calcination;6,7 and (iii) influences the conversion of CaO to Ca(OH)2, CaS, CaSO4, and CaCO3.6,8−19 There is no consensus in the literature in terms of the effect of steam on the carbonation reaction, summarized in Table 1. Clearly, further work is required to better elucidate the effect of high-temperature steam on the reactivity of CaO for CO2 capture. To this end, we investigated the influence of steam on the reactivity of four limestones in a fluidized bed reactor. We present results for experiments conducted with steam concentrations from 0−20% for calcination and/or carbonation up to 30 cycles.
CN =
nCO2,captured nCaO,initial
≈
nCO2,captured · MWCaCO3 mlimestone,initial ·xCa
(2)
Typical experiments were carried out at atmospheric pressure with an inlet gas stream containing 15% (v/v) CO2. The steam concentration was varied from 0−20% and the N2 flow rate was adjusted to maintain a constant total flow rate. The total flow rate for the inlet gas stream was 170 mL/s (800 °C) for all experiments, such that the U/Umf for the limestone was about 8, calculated for calcination conditions. Calcination was conducted at 900 °C and carbonation at 650 °C. The duration of the carbonation and calcination cycles was 600 s, including time for heating and cooling at an average rate of about 0.9 °C/s. CaO-sorbents derived from four limestones were tested: Havelock and Cadomin from quarries located in Canada, and Purbeck and Longcliffe from the U.K. The elemental compositions, determined by X-ray fluorescence (XRF) analysis (Bruker AXS S4 Explorer) are given in Table S1 of the SI. For the majority of experiments, the fresh limestone was sieved to a size fraction of 500−710 μm, however a number of additional experiments were carried out using limestone particles sieved to two smaller size fractions (250−355 μm and 150−250 μm). About 4 g of limestone was used per experiment; mixed with about 12.5 g of sand (355−425 μm). A known mass of limestone and sand was loaded into the reactor at 900 °C, and the final bed mass was measured to determine the loss of sorbent material owing to attrition, assuming only limestone fines were elutriated. On the basis of this method, the results reported for cycles 2−10, unless otherwise stated, are likely slightly lower than the actual carrying capacity. (Figure S2 of the SI shows carrying capacities, normalized for mass loss, for the sorbents tested with and without steam.) Particle size distributions were measured using an optical microscope (Seben Stereo Microscope Incognita III) and an electronic eyepiece (Hangzhou Opto Electronics MD 130) for Longcliffe sorbent after the initial calcination, and after 10 and 30 cycles with and without steam. Surface area and pore volume measurements were taken for a selection of calcined sorbents by N2 adsorption/desorption (Micromeritics TriStar 3000), in all cases, the sand and the calcined limestone were separated by
■
EXPERIMENTAL SECTION In this work, an atmospheric pressure bench-scale bubbling fluidized bed (BFB) reactor (quartz tube) was used (Figure S1 of the Supporting Information, SI) to simulate continuous operation of a Calcium Looping CO2 capture process. This was achieved by periodically changing the reaction temperature between that required for the carbonation and calcination of a bed of limestone mixed with sand. (Sand was used to ensure temperature uniformity owing to the exothermic and endothermic reactions, respectively.) An external resistanceheated tube furnace was usedcontrolled using a K type thermocouple positioned within the reactor bed. The fluidizing gas, which consisted of a mixture of N2, CO2, and steam, was metered using rotameters. The gas was introduced at the base of the quartz tube, via a heated line to avoid condensation. The lower section of the quartz tube, below a sintered glass disk distributor (with average porosity ∼70 μm), was stuffed with quartz wool to aid in preheating the gas and to reduce temperature gradients within the bed. Water vapor was introduced using a heated bubbler (saturator) system, whereby the dry gas (N2 and CO2) was passed through a gas washing bottle filled with deionised water. The amount of water vapor was controlled by varying the temperature of the bubbler (using heating tape) and using a rotary valve and three solenoid valves. This arrangement enabled the diversion of some, or all, of the dry gas past the bubbler, such that steam injection could be switched on and off during an experiment without significant upstream changes in pressure. This system achieved partial vapor pressures of about 60−70%, relative to the theoretical saturated vapor pressure. For calibration of the humidity measurements, a syringe pump was used which injected known 1263
dx.doi.org/10.1021/es202679w | Environ. Sci. Technol. 2012, 46, 1262−1269
1264
Lu et al.19
Yang and Xiao18
Dou et al.17
Wang et al.16
Sun et al.15
Lu et al.14
Bandi et al.13
Dobner et al.12
Manovic and Anthony11
author (date)
comments
Carbonation of calcines from seven limestones (250−425 μm) using a TGA; carbonation temperature varied from 350−800 °C with steam concentrations of 10 and 20% Reported improved reactivity observed (most significant at lower temperatures and with most sintered and 20% CO2; Calcination under N2 (800 °C) or CO2 (950 °C). samples); authors observed that steam accelerated the solid-state diffusion-controlled reaction phase. Carbonation of dolomite particles (∼60 μm) in a TGA in the presence of steam; during calcination without steam. Reported a “catalytic” effect of steam on the carbonation of “dolomite” observing an increase in conversion. They demonstrated that the influence was not permanent as it was not sustained when the steam was switched off, i.e., there was no “memory effect”. Carbonation and calcination of “dolomite” sorbent (
