Combined Ceria Reduction and Methane Reforming in a Solar-Driven Particle-Transport Reactor.

We report on the experimental performance of a solar aerosol reactor for carrying out the combined thermochemical reduction of CeO2 and reforming of CH4 using concentrated radiation as the source of process heat. The 2 kWth solar reactor prototype utilizes a cavity receiver enclosing a vertical Al2O3 tube which contains a downward gravity-driven particle flow of ceria particles, either co-current or counter-current to a CH4 flow. Experimentation under a peak radiative flux of 2264 suns yielded methane conversions up to 89% at 1300 °C for residence times under 1 s. The maximum extent of ceria reduction, given by the nonstoichiometry δ (CeO2-δ), was 0.25. The solar-to-fuel energy conversion efficiency reached 12%. The syngas produced had a H2:CO molar ratio of 2, and its calorific value was solar-upgraded by 24% over that of the CH4 reformed.

Introduction

Thermochemical redox cycles driven by concentrated solar energy are a promising route to split H2O and CO2 and produce syngas,[1,2] a mixture of H2 and CO that can be further processed to liquid hydrocarbon fuels via established gas-to-liquid processes. Nonstoichiometric ceria is currently considered the state of the art among nonvolatile redox materials because of its rapid kinetics and crystallographic stability.[3−6] The two-step cycle comprises the endothermic reduction of ceria generally operated at above 1400 °C, followed by the lower-temperature exothermic reoxidation of ceria to its initial state with H2O and CO2 to form syngas. Various solar reactor concepts have been proposed for effecting this cycle, some of them experimentally demonstrated, including packed beds,[7,8] porous structures,[9−13] rotating components,[14−16] and moving particles,[17−20] that incorporated heat recovery during the temperature-swing mode or operated under isothermal mode. An intriguing approach to decrease the reduction temperature and thus the temperature swing between reduction and oxidation is to combine this redox cycle with the reforming of methane,[21,22] according to the following: During the endothermic reduction step (eq ), CeO2 is reduced in the presence of CH4 to an extent given by the nonstoichiometry δ, where the required high-temperature heat is delivered by concentrated solar energy. The δ moles of oxygen released from the ceria partially oxidize CH4 to form CO and H2. In the subsequent exothermic oxidation step (eq 2), CeO2−δ reacts with CO2 or H2O to reincorporate oxygen into the lattice and form additional CO or H2, respectively. Reactions and 2b intrinsically assume full oxidation at thermodynamically favorable temperatures below 1000 °C.[23,24] Such a combined reduction-reforming processing using concentrated solar heat was previously proposed in the context of the co-production of metals and syngas.[25,26] The introduction of a reducing agent such as CH4 during the reduction step effectively lowers the oxygen partial pressure and shifts the equilibrium to lower temperatures, below 1000 °C.[21,22] Thus, it enables the operation of the two-step cycle isothermally, as demonstrated with a fixed bed reactor in the range 900–1000 °C.[27] Using a similar fixed bed reactor in the range 400–800 °C, Pt was shown to catalyze the surface-controlled kinetics.[28]

This concept is similar to chemical looping combustion (CLC) or chemical looping reforming (CLR) processes that also utilize metal oxides as an intermediate.[29−34] The main advantage of these concepts compared to direct combustion or reforming is that the fuel and oxidant do not come into direct contact. Thus, in the case of CLC, CO2 can be generated without dilution in air, and in the case of CLR, high purity H2[27] or CO may be produced in the oxidation reactor without contamination by trace combustion gases. The main difference to the proposed concept is that, in CLC/CLR, the energy for endothermic reduction of the metal oxide is typically supplied by fuel combustion while the proposed concept aims at utilizing concentrated solar energy as the source of high-temperature process heat. The advantage to using ceria as a redox intermediate compared to other oxides is related to its rapid kinetics, favorable thermodynamics, and selectivity, as discussed in depth by Krenzeke et al.[27] and Warren et al.[22]

Dry reforming and steam reforming of methane driven by solar energy have also been investigated intensively,[35] most recently by Wegeng et al.[36] In contrast, the proposed combined reduction-reforming process does not require the use of catalysts, even for dry reforming. For example, the oxidation of ceria in the presence of CO2 is known to be totally selective to CO production.[9−11,13] As such, a wide range of synthesis gas ratios (CO:H2 between 1:1 and 1:3) are achievable without the need for a downstream water–gas shifting reactor by co-feeding CO2 and H2O during the oxidation step.[11]

We recently proposed a solar particle-transport reactor concept based on a cavity-receiver enclosing an array of alumina tubes, each containing a downward flow of ceria particles counter to an inert sweep gas flow.[17,18] This concept offers in situ separation of the solid and gas products, enhanced heat and mass transfer, good scalability due to the modular tubular configuration, and continuous operation of the reduction step. It further offers the possibility to individually design the (nonsolar) oxidation reactor and operate it independently and round-the-clock in combination with a particle storage. Possible configurations for an oxidation reactor include but are not limited to riser reactors and moving bed reactors. However, because of the indirect heat transfer to the reaction site by conduction through the Al2O3 tube and convection–radiation to the particle flow, tube temperatures of 1500 °C and above were required for the effective reduction of ceria particles in short residence times. This critical temperature requirement can be significantly alleviated by the combined reduction-reforming approach.

In this work, we report on the experimental performance of the solar particle-transport reactor for the thermal reduction of ceria particles with CH4. We study the impact of temperature, ceria particle mass flow rate, gas flow rate, and CH4 concentration on the extent of ceria reduction, methane conversion, and syngas quality. An energy balance is carried out for each experimental run to determine the solar-to-fuel energy conversion efficiency. We further compare counter-current and co-current gas–particle flow configurations and investigate the reactor stability under steady-state operating conditions.

Experimental Methods

The schematic of the lab-scale solar reactor is shown in Figure . It is composed of an Al2O3 tube (Haldenwanger, ALSINT 99.7, Dout = 52 mm, Din = 40 mm, L = 850 mm) vertically positioned inside a cavity receiver (100 mm × 100 mm × 250 mm). The cavity is lined with 76.2 mm-thick Al2O3 insulation (Zircar Zirconia, Type BusterM-35) supported by a 5 mm-thick aluminum shell. High-flux radiation enters the cavity through a polished Al compound parabolic concentrator[37] (CPC, half acceptance angle of 45°) positioned in front of the windowless 30 mm-diameter circular aperture.[38] The incident radiative power is absorbed by internal multiple reflections, resulting in a cavity’s apparent absorptivity of 88%, determined by Monte Carlo ray tracing performed with the in-house code VEGAS.[39] Ceria particles (Chempur, 99.9% purity, Dv50 = 40 μm) contained in a 2 kg reservoir were delivered to the alumina tube by means of a screw feeder with adjustable rotational speed and subsequently carried to the reaction zone by gravity. The feeder design limited the particle mass flow rate to 0.6 g s–1 to enable reasonable dispersion and avoid severe fluctuations. A balance (Kern FKB 6L0.02, uncertainty < ±0.02 g) below the reaction zone at the exit of the tube enabled online measurements of the mass flow rate of reduced ceria particles ṁCeO. Electric mass flow controllers (Bronkhorst, F-201C, uncertainty < ±2%) were used to deliver Ar purge gas to the balance housing at the bottom and the feeder at the top. Additionally, two lateral gas connections above and below the reaction zone were implemented to obtain either a co-current or counter-current gas–particle flow through the reaction zone (material flows for both configurations indicated in Figure ). The product gas composition was analyzed by infrared-based detectors (Siemens Ultramat 6, uncertainty < ±1%) for CO/CO2/CH4, a paramagnetic alternating pressure O2 detector (Siemens Oxymat 6, uncertainty < ±1%) and a thermal conductivity-based H2 detector (Siemens Calomat 6) and validated with gas chromatography (Varian 490, uncertainty < ±1%) measuring H2, O2, CO, CO2, and CH4. Undetectable species H2O and C were calculated though molar balances of the supplied CH4 and measured product gases. The tube temperature distribution was measured on the outside by six shielded thermocouples type-B (uncertainty <0.25%); the average yielded the nominal tube temperature Ttube. The temperature distribution of particles, the loading distribution of particles, and the velocity distribution of particles across the radius were not measured. These measurements are complex and require optical techniques to avoid interference with the flow, but optical access was not possible with this solar reactor configuration. Prior to each experimental run, the particles were exposed to air at 300 °C for more than 8 h to ensure a fully oxidized state (δi = 0). Carbon contamination in the ceria particles was below 0.004 molC molCeO–1, as also verified by thermogravimetric analysis in air at 800 °C.

Figure 1

Schematic of the solar particle-transport reactor. Material flows are indicated by the colored arrows for either co-current (dashed) or counter-current (solid) flow configuration.

The experiments were carried out at the high-flux solar simulator (HFSS) of ETH Zurich. An array of seven Xe-arcs, close-coupled to truncated ellipsoidal reflectors, provided an external source of intense thermal radiation, mostly in the visible and infrared spectra, that closely approximated the heat transfer characteristics of highly concentrating solar systems such as towers and dishes.[40] The radiative flux distribution at the aperture plane was measured with a calibrated CCD camera focused on a refrigerated Al2O3 plasma-coated Lambertian (diffusely reflecting) target. The total solar radiative power input Psolar at the exit of the CPC was calculated by flux integration and verified by water calorimetry.

During a typical experimental run, the cavity receiver was heated by concentrated radiation to the desired Ttube in the range of 1150–1350 °C, while being purged with Ar to reduce the oxygen partial pressure below 200 ppmv. Once steady-state Ttube was reached, Psolar was maintained within 1.1–1.6 kW for mean solar concentration ratios of 1556–2264 suns (1 sun = 1 kW m–2). Ar purge flows of 0.5 and 1 LN min–1 (SLPM, gas flow rates calculated at 273.15 K and 1 atm) were delivered to the feeder and the balance housing, respectively. An Ar/CH4 gas mixture was delivered to the lateral gas inlet at variable flow rates and concentrations to obtain the desired flow conditions in the reaction zone. Typical Re numbers of the gas flow were in the range 25–40, indicative of laminar regime. The CH4 concentration was limited to 10% because of lab safety regulations.

Results and Discussion

Representative Experiment

Figure shows the variation of the ceria mass flow rate (left axis) and CH4 inlet molar flow rate (right axis) alongside H2, CO, CO2, and CH4 outlet molar flow rates as a function of time during a representative experimental run. In this run, ṁCeO = 0.13 g s–1 (ṅCeO = 44.2 mmol min–1) of ceria particles were fed for 9.5 min with a V̇CH = 2 LN min–1 flow of 10% CH4–Ar (ṅCH = 9.0 mmol min–1) through the reaction zone at a constant Ttube of 1303 °C. After a short stabilization period, outlet flows of H2, CO, CO2, and unreacted CH4 reached steady-state even though ṁCeO fluctuated due to poor dispersion by the particle screw feeder. ṁCeO was calculated by ṁCeO = ṁCeO + MO (ṅCO + 2ṅCO + ṅH), where ṁCeO is the mass flow rate of reduced ceria (online balance measurement), MO denotes the molar mass of monatomic oxygen, and ṅi denotes the molar flow rate of species i (GC and feed gas flows measurements). During steady-state ṅH = 14 mmol min–1, ṅCO = 6 mmol min–1, ṅCO = 0.24 mmol min–1, ṅH = 0.75 mmol min–1, ṅC = 1.2 mmol min–1, and ṅCH = 1.4 mmol min–1. The final nonstoichiometry, δfinal = (ṅCO + 2ṅCO + ṅH)/ṅCeO, was 0.16. The methane conversion, XCH = 1 – ṅCH/ṅCH, was 0.85.

Figure 2

Ceria mass flow rate at average ṁCeO = 0.13 g s–1 (ṅCeO = 44.2 mmol min–1) (left axis) and CH4 inlet/outlet, H2, CO, and CO2 molar flow rates (right axis) as a function of time during a representative experimental run. The subscript 0 indicates the inlet condition. Experimental conditions: Ttube = 1302 °C, δfinal = 0.16, V̇CH = 2 LN min–1, co-current flow configuration.

Effect of Ceria Mass Flow Rate

The effect of ṁCeO on δfinal at Ttube = 1302 °C, V̇CH = 2 LN min–1, and xCH = 0.1 is shown in Figure a and b by filled squares for co-current and counter-current gas–particle flows, respectively. For both gas–particle flow configurations, δfinal decreased with increasing ṁCeO. Additionally, for the same ṁCeO, counter-current flow resulted in a higher δfinal. For example, for ṁCeO = 0.13g s–1, δfinal was 0.18 and 0.23 for co-current and counter-current gas–particle flow, respectively. For the investigated temperature range, a closed-system thermodynamic analysis[22] indicates that, at equilibrium, δeq approaches the stoichiometric ratio δeq ≈ ṅCH/ṅCeO. δeq is plotted in Figure a and b by the dashed red line. Interestingly, δfinal > δeq for ṁCeO > 0.2 g s–1 with the co-current flow configuration and for all ṁCeO with the counter-current flow configuration. This apparent inconsistency is explained by comparing the measured product composition with the thermodynamic equilibrium composition. Figure a displays the equilibrium composition and corresponding ceria nonstoichiometry δeq for the system CeO2 + 0.25CH4 as a function of temperature at 1 atm. Computations were carried out following the methodology outlined in Warren et al.[21] and considered the following species: CeO2, CH4, H2, CO, CO2, C, and O2. For ni,CH = 0.25 molCH molCeO–1, δeq increases with temperature until 1027 °C where it plateaus due to the complete CH4 conversion. Below this temperature, C formation derived from CH4 decomposition is thermodynamically favorable. Above this temperature, a shift from C to CO occurs as oxygen evolves from ceria. Syngas constitutes more than 99 mol % of the products at equilibrium with molar ratio H2/CO approaching 2. H2O and CO2 are minimal throughout the considered temperature span. Figure b evaluates the impact of syngas for initial molar fractions corresponding to the complete CH4 conversion (ni,H = 0.5 molH2 molCeO–1 and ni,CO = 0.25 molCO molCeO–1) on the ceria nonsoichiometry. Similar to Figure a, δeq increases with temperature. However, in the absence of CH4, the gas composition contains increasing amounts of H2O and CO2. Additionally, for a given temperature, δeq is lower for reduction with syngas compared to that with CH4. Consequently, in the presence of CH4 (Figure a), no further reduction of ceria with products H2 and CO is possible in a closed system.

Figure 3

Nonstoichiometry δfinal (closed symbols) and δCH (open symbols) alongside CH4 conversion as a function of ceria mass flow rate for the co-current (a) and counter-current (b) flow configurations. δeq and XCH, indicated by the dashed curves, correspond to thermodynamic equilibrium solutions for a closed system. Experimental conditions: Ttube = 1302 °C, V̇CH = 2 LN min–1, xCH = 0.1.

Figure 4

Equilibrium composition and corresponding ceria nonstoichiometry as a function of temperature for the following systems: (a) CeO2 + 0.25CH4 and (b) CeO2 + 0.5H2 + 0.25CO.

The measured product composition is plotted in Figure a and b for the co-current and counter-current flow configurations, respectively. For both configurations, xH decreased and xH increased with increasing ṁCeO, while all other molar fractions remained relatively constant. This shift from H2 to H2O, which is more accentuated under the counter-current gas flow configuration, contradicts the thermodynamic prediction (Figure a). The formation of water is attributed to the reduction of ceria with H2, which is thermodynamically favorable in the absence of CH4 (Figure b). Thus, we hypothesize that ceria is reduced initially by H2 and then by CH4 to an extent larger than δeq. For the co-current flow configuration, this is only possible if H2 diffuses counter to the gas flow, which can be expected because of a low Peclet number of 0.77. For the counter-current flow configuration, unreduced ceria inherently comes in contact with produced syngas, resulting in a higher value of δfinal at the same ṁCeO as well as a stronger decrease in xH2 with increasing ṁCeO. We recognize that δfinal as defined previously is a superposition of the influence of different reducing species, namely, CH4, H2, and CO, and thus can be divided into their respective influence according toδCH isolates the nonstoichiometry resulting from the reduction with methane without considering the additional reducing agents H2 and CO, and thus δCH ≤ δeq. δCH4 is plotted in Figure a and b by the open red squares and closely follows the trend of δeq for both flow configurations. Both configurations result in the same δCH for a given mass flow rate and consequently equal XCH4. This indicates that the reaction is not kinetically limited. The gain obtained in δfinal by operating with counter-current flow configuration is a result of undesired reactions between ceria and product syngas. While the consumed syngas is regained during a subsequent oxidation step, these side reactions result in dilution of the product syngas with H2O and CO2, requiring energy intensive postprocessing. Consequently, the co-current gas particle flow configuration is considered to be superior and will be the focus of further investigation.

Figure 5

Product composition as a function of ceria mass flow rate for the co-current (a) and counter-current (b) flow configurations (Ar is omitted). Experimental conditions: Ttube = 1302 °C, V̇CH = 2 LN min–1, xCH = 0.1.

Effect of Tube Temperature

Figure a shows the impact of the nominal temperature of the Al2O3 tube on ceria nonstoichiometry (left axis) and corresponding methane conversion (right axis) for a ceria mass flow rate of ṁCeO = 0.13 g s–1, gas flow of V̇CH = 2 LN min–1, inlet methane concentration of 10%, and co-current gas–particle flow. As Ttube was raised from 1150 to 1350 °C, δfinal increased from 0.025 to 0.22, respectively, and XCH increased from 0.39 to 0.89, respectively. With similar nonstoichiometries of 0.2 and 0.25, however, lower XCH of 0.6 and 0.52 were obtained in packed bed reactors at ceria temperatures of 1000 and 1120 °C, respectively.[22,27] This points to a significant difference between the measured tube wall temperature and the actual particle temperature. Similar to the results shown in Figure , δfinal exceeds the predicted closed system equilibrium (dashed line), presumably due to the reaction of ceria particles with product gases H2 and CO at Ttube = 1350 °C. All syngas consumed in this manner during the reduction step is reformed during the oxidation step and, consequently, results in no additional gain/loss in fuel yield. The uncertainty in δfinal is estimated to be below 10% based on the standard deviation of measurements taken for 1250 and 1300 °C on consecutive days (Figure ). It is mainly due to fluctuations in the ceria mass flow rate with time, as shown in Figure . These fluctuations are inherent to the feeder design and can be as high as ±30% of the average mass flow rate delivered during steady-state operation. This high extent of ceria reduction and CH4 conversion was realized in very short particle residence times (<1 s) inherent to the present lab-scale reactor. Since complete conversion is thermodynamically favorable at the investigated temperatures. For example, at a methane to ceria ratio of 0.25, complete conversion is favorable at 1000 °C as indicated in Figure ; thermodynamic limitations were not expected at tube temperatures above 1300 °C. Equilibrium curves for δeq and XCH are indicated in Figure a. Figure b shows the molar fraction of product species, xi = ṅi/ṅtot, where ṅtot corresponds to the sum of all molar flows except Ar but including unmeasured C and H2O. As expected, both xH and xCO increased with Ttube, while xCH decreased, yielding at 1350 °C a H2:CO molar ratio of 2, suitable for gas-to-liquid processing via Fischer–Tropsch. Trace amounts of CO2 and H2O were indicative of the high selectivity from CH4 to syngas. Formation of solid carbon, attributed to CH4 cracking on hot nucleation surfaces (tube walls and ceria particles), reached a maximum solid phase molar fraction of xC = 0.13 at 1150 °C but was eliminated at above 1300 °C.

Figure 6

(a) Nonstoichiometry and methane conversion as a function of the nominal temperature of the Al2O3 tube and the corresponding equilibrium values δeq and XCH. (b) Corresponding product compositions (Ar is omitted). (c) Corresponding upgrade factor. Experimental conditions: ṁCeO = 0.13 g s–1, V̇CH = 2 LN min–1, xCH = 0.1, co-current flow configuration.

The upgrade factor U is defined as ratio of the energy contained in the outlet flow to the energy content of the inlet flow, given bywhere HVi corresponds to the (high) heating value of species i. This definition intrinsically assumes complete reoxidation of ceria with CO2. Note that the energy content of carbon deposited is not considered because it is an undesirable product. U increased from 0.92 to 1.2 when Ttube was raised from 1150 °C 1350 °C, as shown in Figure c. At Ttube < 1250 °C, U < 1 partly because of carbon deposition. At Ttube > 1250 °C, U > 1 which indicates solar energy stored in the form of syngas. For comparison, U < 0.92 for autothermal steam-based reforming of methane without the involvement of ceria, fueled by combustion of excess methane.

Effect of Gas Flow Rate

Figure a shows δfinal and XCH as a function of the CH4/Ar gas flow rate while keeping xCH4,0 = 0.1, Ttube = 1300 °C, and ṁCeO = 0.14 g s–1 and for the co-current flow configuration. The corresponding product composition is shown in Figure b. With increasing V̇, the amount of CH4 available for the reaction increased proportionally, resulting in a peak δfinal = 0.19 for V̇CH = 3 LN min–1. A further increase in V̇CH resulted in a decrease in δfinal and xCO due to the carbon formation associated with CH4 cracking, possibly caused by shorter gas and particle residence times and the resulting heat transfer and kinetic limitations. The residence time limitation is further supported by the decrease in XCH with V̇CH shown in Figure a. Consequently, U monotonically decreased as well (Figure c).

Figure 7

(a) Nonstoichiometry and CH4 conversion as a function of the gas flow rate. (b) Corresponding product composition (Ar is omitted). (c) Corresponding upgrade factor. Experimental conditions: Ttube = 1302 °C, ṁCeO = 0.14 g s–1, xCH = 0.1, co-current flow configuration.

Effect of Methane Concentration

Figure a shows δfinal and XCH as a function of xCH for Ttube = 1303 °C, ṁCeO = 0.12 g s–1, V̇CH = 2 LN min–1, and co-current flow configuration. The corresponding product compositions are shown in Figure b. δfinal increased monotonically from 0.088 to 0.18 when xCH increased from 2.5% to 10%, while XCH remained nearly constant at 0.86 ± 0.02 over this range, corresponding to xCH = 0.053, presumably due to mass transfer effects in the co-current flow configuration. xH and xCO increased with xCH4,0 and correlated with a decrease in xH and xCO. It is likely, that the particles reach conditions were the reduction with syngas is thermodynamically favorable for low methane to ceria rations, resulting in increased H2O and CO2 formation. Carbon formation was not observed under these conditions, resulting in a constant U = 1.25 ± 0.02.

Figure 8

(a) Nonstoichiometry alongside methane conversion for varying inlet methane concentration. (b) Corresponding product composition (Ar is omitted). Experimental conditions: Ttube = 1303 °C, ṁCeO = 0.12 g s–1, V̇CH = 2 LN min–1, co-current flow configuration.

Solar-to-Fuel Energy Conversion Efficiency

The solar-to-fuel energy conversion efficiency, ηsolar-to-fuel, is defined as the ratio of the (high) calorific value of syngas (H2 and CO) produced to the summation of the solar radiative energy input and the calorific value of the converted CH4. It is thus given byNote that the third term in the numerator accounts for the energy stored in the reduced ceria; i.e., it assumes that reduced ceria is completely reoxidized with CO2 (eq ) to generate additional CO. The heating value of carbon is not included because it is an undesired byproduct. No energy penalty is accounted for inert gas consumption because Ar dilution was used only for safety lab regulations. ηsolar-to-fuel is plotted in Figure a–d as a function of Ttube, ṁCeO, V̇CH, and xCH, respectively. Despite the associated reradiation losses,ηsolar-to-fuel increased from 3.2% at 1150 °C to 9.4% at 1350 °C. As expected, ηsolar-to-fuel also increased monotonically with xCH. In contrast, ṁCeO had no significant influence regardless of the flow configuration, yielding ηsolar-to-fuel = 8.4 ± 0.4%. ηsolar-to-fuel increased with V̇ peaked at 12% for V̇CH = 3 LN min–1 and decreased because of solid carbon deposition and residence time limitations. At this point, it is not possible to compare these results directly to other studies in the literature. ηsolar-to-fuel would be a relevant indicator to compare, for example, to the value obtained for the solar reforming of methane without ceria (CH4 + H2 → 3H2 + CO). The PNNL’s reforming system,[36] which uses heat exchangers to recover the sensible heat of the hot outlet stream, reports an efficiency of 69% but is defined based on the enthalpy change of the reaction and without considering the calorific value of methane (HVCH) as energy input in the denominator, i.e., different definition than eq and thus difficult to compare. Since we are only driving half of the redox cycle without any heat recovery, the particles are heated from ambient to the reaction temperature, affecting detrimentally ηsolar-to-fuel. In a complete cycle with efficient heat management between reduction and oxidation steps, the particles should enter the reactor near the reaction temperature. To maximize ηsolar-to-fuel, the reactor should be operated in the upper range of Ttube, while V̇CH needs to be selected to maximize the CH4 supply while avoiding residence time limitations.

Figure 9

Solar-to-fuel energy conversion efficiency as a function of the nominal temperature of the Al2O3 tube (a), ceria mass flow rate (b), gas flow rate (c), and inlet methane concentration (d). Experimental conditions if not stated otherwise in the graph: Ttube = 1303 °C, ṁCeO = 0.13 g s–1, V̇CH = 2 LN min–1, xCH = 0.1, co-current flow configuration.

Steady-State Syngas Production

Figure shows the continuous syngas production for over 60 min for an experimental run with co-current flow configuration and for Ttube = 1291 °C, V̇CH = 2 LN min–1, xCH = 0.1 (ṅCH = 9.03 mmol min–1), and mean ṁCeO = 0.15 g s–1 (ṅCeO = 53.7 mmol min–1). Shortly after the particle feeder and the inlet gas flow were initiated and despite the fluctuation in ṁCeO, the CO and H2 flow rates reached steady-state conditions at about 6 and 13.7 mmol min–1, respectively, resulting in a H2:CO ratio of 2.3. The deviation from 2 is due to trace amounts of CO2 production (xCO = 0.015) and carbon deposition (xC = 0.029). For this run, XCH = 0.85 and ηsolar-to-fuel = 8.3%. No evidence of a reaction between ceria particles and the Al2O3 tube was observed. Carbon deposition was observed but did not exceed xC = 0.029 and may be avoided by operating at above 1300 °C (Figure b) and/or by co-feeding H2O.

Figure 10

Continuous syngas production for over 60 min with stable outlet gas composition. Experimental conditions: Ttube = 1291 °C, δ = 0.15, V̇CH = 2 LN min–1, xCH = 0.1, mean ṁCeO = 0.15 g s–1 (ṅCeO = 53.7 mmol min–1), co-current flow configuration.

Summary and Conclusions

The technical feasibility of the solar particle-transport reactor was experimentally demonstrated for performing the combined CeO2 reduction and CH4 reforming process using both counter-current and co-current flow configurations. Experiments driven by high-flux irradiation resulted in peak ηsolar-to-fuel of 12% and an upgrade factor of 14% at Ttube = 1303 °C, ṁCeO = 0.13 g s–1, V̇CH = 3 LN min–1, and xCH = 0.1 with a co-current flow configuration. Further, a peak upgrade factor of 24%, ηsolar-to-fuel = 9%, and methane conversion of 89% were obtained at Ttube = 1303 °C, ṁCeO = 0.13 g s–1, V̇CH = 1 LN min–1, and xCH = 0.1 with a co-current flow configuration.

Ceria nonstoichiometry (reduction extent), efficiency, and methane conversion increased with temperature, while carbon formation was suppressed at tube temperatures above 1300 °C. In contrast, nonstoichiometry decreased with ceria mass flow rate, while efficiency and methane conversion were unaffected. Interestingly, H2 molar fractions decreased with increasing ceria mass flow rate, while the H2O molar fractions increased correspondingly. This trend was more pronounced for counter-current flow configuration and was attributed to the reaction of unreacted ceria with the produced syngas. Continuous steady-state operation for over 1 h yielded syngas of quality suitable for gas-to-liquid processing. Reaction kinetics over short residence times are the main limitation at the current scale. We conclude that the solar particle-transport reactor is a robust and scalable concept for effecting the combined CeO2 reduction and CH4 reforming process.