Research Article pubs.acs.org/journal/ascecg
Dry-Oxy Methane Reforming with Mixed e−/CO32− Conducting Membranes Peng Zhang, Jingjing Tong, and Kevin Huang* Department of Mechanical Engineering, University of South Carolina, 541 Main Street, Columbia, South Carolina 29201, United States ABSTRACT: We report here an alternative way to make syngas from dry-oxy methane reforming with CO2 and O2 directly captured from flue gas by a mixed e −/CO 32− conducting membrane in a single reactor loaded with a reforming catalyst. The mixed e−/CO32− conducting membrane is comprised of a porous electronic conductor (silver) with the skeleton overcoated with a nanoscaled ZrO2 within which a molten carbonate is withheld. The reforming catalyst is Ni supported on MgO impregnated with 1 wt % Pt (NMP). With a simulated flue gas as the CO2/O2 source and CH4−Ar as the sweep gas, the reactor can convert >82% CH4 under all conditions tested. The rates of H2 and CO production reach as high as 4.0 and 4.2 mL min−1 cm−2, respectively, at 800 °C. A long-term test shows a reasonably stable performance within 115 h. Overall, this work demonstrates the technical feasibility of a new capture-and-conversion “all-in-one” CO2 reactor for efficient and cost-effective syngas production via dry-oxy methane reforming. KEYWORDS: mixed conducting membrane, CO2−O2 cocapture, dry-oxy methane reforming, nanoscaled ZrO2, syngas production
■
INTRODUCTION Carbon capture and conversion is widely deemed the most realistic near-term solution for stabilizing the rising atmospheric CO2 concentration and enable a carbon neutral ecosystem for a sustainable future. Although carbon capture has been demonstrated with an amine-based “chemical washing” process at large scale,1,2 the majority of today’s CO2 captured is still being buried underground for either permeant storage or enhanced oil recovery (EOR).3−5 Converting the captured CO2 into useful products is a better alternative to the storage approach because it offers a sustainable solution to realize a carbon neutral ecosystem. The current efforts for CO2 conversion are mainly focused on laboratory-scale feasibility studies using chemical methods,6−9 photocatalytic conversion,10−12 and electrocatalytic reduction13−18 to name just a few. Among many CO2 conversion methods, dry methane reforming (DMR), i.e., CH4 + CO2 = 2CO + 2H2, stands out as an appealing catalytic process to upgrade methane into syngas for the synthesis of oxygenated compounds and hydrocarbons from the Fischer−Tropsch process.19−21 A distinctive feature of DMR is that it reutilizes the captured CO2 to combine with today’s widely accessible and costcompetitive CH4 to make value-added products with profound environmental implications. Unfortunately, a large-scale industrial deployment of DMR technology is hampered by the limited catalyst life caused by coking and high CO2 cost associated with capture, purification, compression, and transportation.8,22 Therefore, searching for new ways to improve the reactor lifespan and lower CO2 cost are of practical importance © XXXX American Chemical Society
to an ultimate industrial-scale implementation of the DMR technology. High-temperature concentration gradient-driven membranebased reactors are a new method that has emerged in recent years to capture CO2 and instantly convert it into syngas in a single reactor.23 An inherent advantage for such membrane reactors is the continuous removal of the captured CO2 (and O2) products, thus constantly maintaining a high concentration gradient across the membrane to achieve high flux. The reactor demonstrated in the author’s laboratory is comprised of either a mixed e−/CO32− (denoted as MECC hereafter) or O2−/CO32− (denoted as MOCC hereafter) conducting membrane capable of selectively separating CO2/O2 and CO2 from flue gas, respectively. If CH4 is used as the sweep gas, the permeated CO2/O2 or CO2 can instantly react with CH4 to produce syngas in the presence of an appropriate catalyst. A modeling analysis by Rui et al.24 on using a 0.075 mm thick La0.6Sr0.4Co0.8Fe0.2O3−σ (LSCF)-carbonate dual-phase membrane tubular reactor to capture CO2 from flue gas and instantly react with CH4 indicates a CH4 conversion of 48.06% at an average CO2 permeation flux of 1.52 mL(STP) cm−2 min−1 at 800 °C. The study also found that the methane reforming performance can be improved by adding O2 in the feed gas as a result of improved CO2 flux by concurrent O2 permeation and the reaction between permeated O2 and reduced species at the sweeping side.24,25 In 2013, Anderson et Received: March 12, 2017 Revised: April 29, 2017 Published: May 3, 2017 A
DOI: 10.1021/acssuschemeng.7b00773 ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX
Research Article
ACS Sustainable Chemistry & Engineering al.26 experimentally demonstrated the concept with an LSCFcarbonate dual-phase O2−/e−/CO32− conducting membrane with a thickness of 1.5 mm. Clearly, such a combined capture and conversion avoids the need for CO2 compression and transportation. The cost of this produced syngas is therefore expected to be low. However, the performance of this first proof-of-concept demonstration was poor because of the reaction between LSCF and carbonate. With 10% Ni/γ-Al2O3 as the catalyst, the CH4 conversion rate and syngas production rate were only 8.12% and 0.3 mL min−1 cm−2 at 850 °C, respectively. Following the same concept, our laboratory investigated alternative oxide as the matrix and recently demonstrated the greatly improved performance with a more stable CeO2carbonate composite as the mixed O2−/CO32− conducting membrane for combined CO2 capture and DMR.27 The CH4 conversion rate reached >90% with a syngas production rate of syngas >3 mL min−1 cm−2 at 850 °C in the presence of a Nibased catalyst. However, the study observed coking at higher CH4 concentration and production of syngas with a low H2/ CO ratio. In the present study, we report a new CH4 reforming membrane reactor based on a different type of mixed conductor: e−/CO32− comprised of a metal and carbonate phase. Figure 1 illustrates the working principle of such a CO2
we report the performance of such a unified reactor in the temperature range of 740−800 °C. The catalyst used for the catalytic DOMR is a commercial standard Ni supported on MgO impregnated with 1 wt % Pt (denoted as NMP hereafter); 1 wt % Pt is added to minimize coking. The feed gas to the reactor is a simulated flue gas containing 75% N2, 15% CO2, and 10% O2, and a mixture of CH4 and Ar is used as the sweep gas. It is worth mentioning that oxygen transport membranes (OTMs) based on mixed e−/O2− conductors have also been explored recently for syngas production from CH4 using compressed CO2 as an oxygen source.32−35 The reactor works in a way that the CO2 is reduced by electrons in OTM into CO and O2− on one side of the membrane. Meanwhile, CH4 at the other side of the membrane reacts with the permeated O2−, forming syngas via partial oxidation of methane (POM) reaction. However, one issue with this method is that it requires pure CO2 to be the source of oxygen, which could add to the cost. Any containment of O2 in CO2 such as in flue gas could block CO2 reduction by the faster and prevalent O2 reduction reaction (1/2O2 + 2e− = O2−). Another issue with perovskite-type OTMs employed is their poor stability in CO2containing atmosphere. The demonstrated maximum CO and H2 production rates at 1029 °C by OTM-based reactors were only 3.4 and 5.1 mL min−1 g−1LSCF, respectively, with low CH4 and CO2 conversion rates of 17 and 8.0%, respectively. As another example, Sibudjing et al.35 reported the use of a La0.6Sr0.4Co0.8Ni0.2O3−σ (LSCN) ceramic hollow fiber membrane reactor loaded with Ni-based catalyst to produce syngas through the DOMR process. Better coking resistance and higher H2/CO ratio were observed for the integrated OTM reactor than for a pure DMR process.
■
EXPERIMENTAL PROCEDURE
Preparation of MECC Membrane. The porous Ag matrix was prepared in the same way as in our previous study.23,36 Briefly, silver powders in 1−3 μm (99.9% metal basis, Alfa Aesar) were intimately mixed with carbon black as a pore former in a volume ratio of 1:1 in ethanol. The dried powder mixture was then pressed into pellets under 70 MPa, followed by sintering at 650 °C for 2 h in air to remove the carbon pore former and achieve good mechanical strength. Thus, fabricated porous silver matrix in ϕ15.5 mm and 0.82 mm thickness was then subjected to coating of ZrO2 using an ALD system (Savannah S200, Cambridge NanoTech, USA). Tetrakis(dimethylamido)zirconium(IV) ([(CH 3) 2 N] 4 Zr, Sigma-Aldrich, USA) was used as the zirconia precursor; DI water was used as the oxidant, and N2 was used as the carrier/purge gas. Different from the conventional flow mode recipe, an exposure mode recipe was particularly developed for our microporous substrate. One deposition cycle consisted of eight steps: vacuuming → pulsing DI → dosing DI → purging DI → vacuuming → pulsing Zr → dosing Zr → purging Zr with the reaction chamber temperature was controlled at 250 °C. The ZrO2 precursor pulsing, dosing, and purging times were 0.4, 20, and 60 s, respectively. The deposition rate of the film was determined to be 0.13 nm/cycle. A total of 200 cycles was performed, resulting in ∼25 nm ZrO2 layer on the silver wall.31 The image and composition of Ag and Zr were also examined by SEM and EDS at different locations of the sample, confirming the uniformity of the ZrO2 coating. The ZrO2overcoated porous silver pellet was then soaked in a 52 mol % Li2CO348 mol % Na2CO3 molten carbonate (denoted as MC) at 650 °C for 2 h to form a dense membrane; the membrane thickness is ∼800 μm. The weight increase after MC impregnation was ∼20%. After MC infiltration, the surface of the Ag-MC membrane was thoroughly cleaned by sandpaper in the presence of ethanol. Preparation of Catalyst Bed. The standard Ni-based Ni0.2Mg0.8O-1 wt %Pt (denoted as NMP) catalyst was selected for
Figure 1. Schematic illustration of the e−/CO32− (MECC) mixed conducting membrane reactor loaded with a catalyst bed for CO2 capture and catalytic DMR. TPB: triple phase boundary.
capture and methane-reforming combined reactor. The enabling surface reaction suggests that the membrane be a CO2 and O2 transporter. The presence of O2 in CO2 is expected to mitigate coking, lower the heat requirement, and produce syngas with a higher H2/CO ratio. This catalytic process is commonly known as dry-oxy methane reforming (denoted as DOMR hereafter). The state-of-the-art mixed e−/CO32− MECC membrane consists of a porous silver matrix filled with a molten carbonate (MC) phase. Because porous silver is prone to sinter at higher temperatures, its working temperature is usually limited to the 500−650 °C range to avoid unwanted flux degradation.28,29 This low working temperature is thermodynamically unfavorable to the DOMR reaction that becomes prevalent only at a temperature ≥700 °C.7,30 To expand the working temperature of silver−carbonate membranes to ≥700 °C, we recently demonstrated that a silver−carbonate membrane with the porous silver skeleton overcoated by a nanoscaled layer of ZrO2 via atomic layer deposition (ALD) can retain a stable CO2/O2 flux at 800 °C for an extended period of time.31 The enhanced high-temperature stability of ALD-ZrO2 overcoated silver−carbonate membrane offers a great opportunity to study combined CO2 (and O2) capture and catalytic DOMR within a single reactor at ≥700 °C. In the present work, B
DOI: 10.1021/acssuschemeng.7b00773 ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX
Research Article
ACS Sustainable Chemistry & Engineering
Figure 2. Schematic of the CO2 permeation cell configurations: (1) CO2 cylinder, (2) nitrogen cylinder, (3) hydrogen cylinder, (4) argon cylinder, (5) methane cylinder, (6) mass flow controllers, (7) furnace, (8) inner feed tube, (9) short alumina tube, (10) thermocouple, (11) supporting alumina tube, (12) sealant, (13) MECC membrane, (14) catalyst bed, (15) inner sweep tube, (16) gas chromatography (GC), (17) directionalcontrol valve, and (18) soap film flowmeter. and chemical equilibrium before GC sampling. The final CO2 and O2 flux densities (JCO2 and JO2) measured with pure Ar as the sweep gas were calculated by
this study, which is an easy choice based on prior work on cofed DMR and POM.27,37 To make it, Mg(NO3)2·6H2O (99%, Alfa Aesar) and Ni(NO3)2·6H2O (99%, Alfa Aesar) with a molar ratio of 8:2 were used as precursors directly dissolved in DI water. Then, 1 wt % H2Cl6Pt· 6H2O (Sigma-Aldrich) was added to the aqueous solution and thoroughly stirred for 12 h. After drying on a hot plate at 80 °C, the precipitate was finally calcined in air at 800 °C for 5 h. Finally, the NMP catalyst powder was pressed into pellets under 200 MPa, followed by pulverizing into 20−40 mesh size granules. The catalyst granules in the amount of ∼1.0 g were then loaded on top of an Al2O3 wool located ∼1 mm underneath of the surface of silver−carbonate membrane. Prior to testing, the catalyst was activated in pure H2 for 2 h at 850 °C. MECC Membrane Reactor Performance Test. The rates of CO2−O2 permeation and CH4 conversion of the membrane reactor were evaluated in a homemade permeation cell, which has been previously described in detail.38 Briefly, to assemble the cell, a Ag-MC membrane was first sealed to a supporting alumina tube by a commercial silver paste as the sealant (Shanghai Research Institute of Synthetic Resins). A short alumina tube was then mounted to the top surface of the MECC membrane to shield the feed gas as shown in Figure 2. The feed gas was a mixture of 15 mL min−1 CO2, 10 mL min−1 cm−2 O2, and 75 mL min−1 N2; N2 was used as a tracer gas for leak correction if any. The sweep gas was a mixture of CH4 and Ar, the concentration of which varied slightly for different tests. The concentrations of CO2, CH4, H2, CO, and N2 in the effluent were analyzed by an online gas chromatograph (Agilent 490). Commercial mass flow controllers (Smart-Trak, 50 Series) specifically calibrated for each gas being used was employed to control the gas flow rates. An appropriate amount of NMP catalyst supported on an Al2O3 wool in approximately 1 cm height was added to the sweeping chamber right beneath the Ag-MC membrane to form the catalyst bed. The gap between the membrane sweep-side surface and catalyst bed is ∼1 mm. The overall experimental setup is shown in Figure 2. The testing temperature for CO2 flux evaluation was varied in the range of 650−800 °C in an interval of 25 °C with pure Ar as the sweep gas, whereas for DOMR, the testing temperature range was between 740 and 800 °C with a CH4-Ar mixture as the sweep gas. At each temperature, ∼1 h was given to allow the membrane to reach thermal
CCO2
JCO =
1 − CCO2 − CO2 − C N2
2
JO = 2
CO2 1 − CCO2 − CO2 − C N2
×
×
Q A
(1)
Q A
(2)
where CCO2, CO2, and CN2 are the measured concentrations of CO2, O2, and N2, respectively, Q is the flow rate of the Ar sweep gas, and A is the effective area of the sample, 0.921 cm2. The CO2 permeation rate in the CO2−CH4 conversion experiment was calculated by a different method, which will be elaborated in the next section. The CH4 and CO2 conversion rates (XCH4, XCO2) were calculated by
XCH4 =
XCO2 =
FCH4(in) − FCH4(out) FCH4(in)
× 100 (3)
JCO (total) − JCO (unconsumed) 2
2
JCO (total)
× 100 (4)
2
where FCH4(in) and FCH4(out) are the mass flow rates of methane into and out of the system, respectively, and J CO 2 (total) and JCO2(unconsumed) are the calculated total CO2 permeation rate and the unconsumed outlet CO2 flow rate, respectively. The H2 and CO production rates were calculated by
JH = 2
JCO = C
c H2
FAr S
(5)
cCO F × Ar 1 − c H2 − cCH4 − cCO − cCO2 S
(6)
1 − c H2 − cCH4 − cCO − cCO2
×
DOI: 10.1021/acssuschemeng.7b00773 ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX
Research Article
ACS Sustainable Chemistry & Engineering
highest (CO2 + O2)/N2 selectivity of 363 is achieved at 800 °C, which corresponds to the highest JCO2 + JO2 (= 1.17 mL min−1 cm−2) and N2 concentration of 0.00058% (equivalent to JN2 = 0.0032 mL min−1 cm−2). A better sealing and higher (JCO2 + JO2) at higher temperatures are two leading reasons for the higher selectivity. DOMR Performance. After confirming the membrane’s performance with pure Ar as the sweep gas, the DOMR performance was tested against temperature in a range of 740 to 800 °C at a fixed 1.8% CH4−Ar sweep gas; the results are shown in Figure 4(a). The selection of 740 °C as the lowest
where CH2, CCH4, CCO2, and CCO are the measured concentrations of H2, CH4, CO2, and CO, respectively, in the effluent,, and FAr is the flow rate of Ar gas. The theoretical selectivity of CO2 + O2 for this electrochemical membrane should be 100% because only CO32− is allowed to pass through the dense MECC membrane. However, in reality there is always a small fraction of physical leakage associated with the membrane or gas seals, inadvertently mixing N2 into the permeated CO2 + O2 stream and lowering product purity. To better evaluate the leakage issue and thus product selectivity, we use the ratio of (CO2 + O2) flux density sum (JCO2 + JO2) over N2 flux density (JN2), (JCO2 + JO2)/JN2 as a measure of the selectivity for MECC membranes. Characterization of the MECC Membrane. The cross-sectional views of the Ag-MC membrane and morphology of NMP catalyst before and after testing were examined by a scanning electron microscope (SEM) (FESEM, Zeiss Ultra) equipped with energydispersive spectrometry (EDS).
■
RESULTS AND DISCUSSION CO2/O2 Flux Density vs Temperature with Ar as the Sweep Gas. The CO2 and O2 permeation flux densities of the as-prepared Ag-MC membrane were first evaluated in the temperature range of 650−800 °C with pure Ar as the sweep gas and simulated flue gas as the feed gas; the results are shown in Figure 3, where the ratio of JCO2 to JO2 is found to be very
Figure 3. CO2 and O2 flux densities as a function of temperature with pure Ar as the sweep gas.
Figure 4. (a) Effect of temperature on DOMR performance of ALDZrO2-Ag-MC membrane reactor with NMP catalyst with feed gas of 75 mL min−1 N2, 15 mL min−1 CO2, and 10 mL min−1 O2 and sweep gas of 0.94 mL min−1 CH4 and 50 mL min−1 Ar; (b) effect of CH4 molar fraction on the DOMR performance of the ALD-ZrO2-Ag-MC membrane reactor with NMP catalyst measured at 800 °C. GHSV = 5800−6500 h−1.
close to 2:1, confirming the stoichiometry required by the surface reaction CO32− = CO2 + 1/2O2 + 2e−. Both JCO2 and JO2 increase with temperature, suggesting a thermally activated process for the CO2 and O2 permeation through the Ag-MC membrane. The highest CO2 and O2 flux densities reached 0.78 and 0.40 mL min−1 cm−2, respectively, at 800 °C, which agrees with our previous results.31 The apparent activation energies Ea for CO2 and O2 obtained from Figure 3 are 51.6 and 53.9 kJ mol−1, respectively. The proximity of the two activation energies implies that the activation processes for CO2 and O2 transport are likely to be coupled by the same enabling reaction: CO2 + 1/2O2 + 2e− = CO32−. The selectivity calculated based on (JCO2 + JO2)/JN2 shows an increase from 121 to 363 as the temperature is increased from 650 to 800 °C. The lowest (CO2 + O2)/N2 selectivity of 121 is observed at 650 °C, which corresponds to the lowest JCO2 + JO2 (= 0.427 mL min−1 cm−2) and highest N2 concentration of 0.0064% (equivalent to JN2 = 0.0035 mL min−1 cm−2). The
testing temperature was aimed to minimize the potential of coking as suggested by DOMR thermodynamics. Overall, the CH4 conversion rate is seen to increase slightly from 82.8 to 86.6% as the temperature is increased from 740 to 800 °C, suggesting the endothermic nature of DMR. Meanwhile, the H2 production rate decreases slightly accompanied by a minor increase in CO production rate. The observed H2 and CO production rates are 1.3 and 1.6 mL min−1 cm−2, respectively, at 800 °C. According to DOMR and POM reactions, the theoretical ratio between H2 and CO of DMR and POM reactions should be 1 and 2, respectively. Thus, the produced H2/CO ratio from DOMR is expected to be within 1−2. However, Figure 4(a) indicates a lower H2/CO ≤ 1 at T ≥ 740 D
DOI: 10.1021/acssuschemeng.7b00773 ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX
Research Article
ACS Sustainable Chemistry & Engineering °C, suggesting that reverse water gas shift (RWGS) reaction H2 + CO2 = H2O + CO and possibly H2 + 1/2O2 = H2O are likely concurrent with DOMR. It is also noticed from Figure 4(a) that the CO2 conversion rate decreases from 64 to 46% as the temperature increases from 740 to 800 °C. The reason is rooted in the increase in CO2/O2 permeation rate with temperature. Because the CH4 molar fraction is fixed at 1.8% during the test, the CO2 conversion rate will decrease as CO2/ O2 permeation rate increases. Determination of the CO2 flux density (or permeation rate) requires special caution. The amount of unconsumed CO2 can be directly measured by GC, whereas the amount of consumed CO2 can only be calculated by the carbon balance with the assumption that there is no carbon deposition. On the basis of our thermodynamic calculations and post-test analysis on he catalyst (see Figure 7(d and e)), we are confident that this assumption is valid for most conditions in this study. Thus, the following equation is used to determine the total CO2 flux density (JCO2(total))
5.27 Note that the same catalyst NMP was used for both cases, but the reforming temperature was 800 °C for the MECC
Figure 5. Performance comparison between MOCC-DMR and MECC-DOMR. Solid lines: MECC at 800 °C; dashed lines: MOCC at 850 °C.27
JCO (total) = JCO (unconsumed) + JCO(produced) − JCH (consumed) 2
2
4
(7)
reactor to ensure minimal silver sintering and 850 °C for the MOCC reactor. Obviously, this is a conservative comparison because thermodynamically 800 °C favors coking more than 850 °C. However, Figure 5 shows that the MECC membrane can still handle up to 4.2% CH4 (the highest tested) even though it was operated at 800 °C. In comparison, the MOCC membrane shows signs of coking at CH4 > 2.4% at 850 °C, where the production rates for CO and H2 deaccelerate and accelerate (CH4 = C + 2H2), respectively. The coking resistance of MECC being higher than that of MOCC membranes is thus demonstrated. Stability Test of DOMR Performance. Stability is a very important performance metric for evaluating the technical feasibility of any commercial catalytic reactor. Figure 6 shows
where JCO2(unconsumed), JCO(produced), and JCH4(consumed) are the rates (flux densities) of CO2 consumed, CO produced, and CH4 consumed, respectively. Shown in Figure 4(a) is the calculated CO2 flux density, increasing expectedly from 0.87 to 1.17 mL min−1 cm−2 as the temperature is increased from 740 to 800 °C. The dependence of DOMR performance on CH4 inlet concentration is shown in Figure 4(b). Evidently, both H2 and CO production rates increase monotonically with CH4 molar fraction. The maximum H2 and CO production rates reach 4.0 and 4.2 mL min−1 cm−2 at 800 °C with a 4.4% inlet CH4. In comparison, the CH4 conversion rate does not change significantly with CH4 concentration, implying that the reactor has not yet reached its full reforming capacity. The concurrent increase of CO2 and O2 flux densities with CH4 molar fraction is the leading reason for the high CH4 conversion rate. Indeed, as shown in Figure 4(b), the CO2 flux density increases from 1.1 to 2.4 mL min−1 cm−2 as the CH4 molar fraction is increased from 1.7 to 4.4%. The same trend is also expected for the O2 flux density because of the 2:1 ratio between CO2 and O2 flux density shown in Figure 1. Also shown in Figure 4(b) is the slightly increased H2/CO ratio from 0.69 to 0.94 as the CH4 molar fraction is increased from 1.7 to 4.4%. This trend originates from the fact that increasing the CH4 molar ratio in the sweep gas will result in consumption of more O2 and CO2, thus reducing the portion of RWGS and water formation and ultimately leading to an increase in the H2/CO ratio. One can see clearly from Figure 4(b) that the CO2 conversion rate is
