Study of a Magnetically Stabilized Porous Structure for

Feb 20, 2013 - Ni-ferrite with structural stability for solar thermochemical H2O/CO2 splitting. I. Teknetzi , P. Nessi , V. Zaspalis , L. Nalbandian. ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/IECR

Study of a Magnetically Stabilized Porous Structure for Thermochemical Water Splitting via TGA, High-Temperature-XRD, and SEM Analyses Kyle M. Allen,*,† Ayyoub M. Mehdizadeh,† James F. Klausner,† and Eric N. Coker‡ †

Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, Florida, United States Advanced Materials Laboratory, Sandia National Laboratories, Albuquerque, New Mexico, United States



ABSTRACT: Sintering of metal particles at high temperature in thermochemical looping processes, such as the iron-based looping process, dramatically reduces their reactivity. During the sintering process, metal particles lose their chemically active surface area, typically within only a few oxidation and reduction cycles. To cope with this problem, a uniform external magnetic field is applied to a fluidized mixture of iron and silica particles to form an organized structure of magnetic iron chains. In this study, reaction kinetics of the magnetically stabilized porous structure during oxidation has been examined using thermogravimetric analysis (TGA). High temperature X-ray diffraction (HT-XRD) has been performed in situ to determine at which temperatures structural changes occur. Scanning electron microscopy (SEM) has been performed on samples, both before and after many cycles, to show the morphological changes that occur within the sample. TGA results demonstrate that the magnetically stabilized porous structure has excellent stability and reactivity at high reaction temperatures. These results indicate that this structure is suitable for many industrial and chemical processing applications involving high temperature, such as synthetic fuel production. Recently, Go et al.9 used thermogravimetric analysis to study the kinetics of iron oxide reduction with methane and oxidation with steam. An activation energy for the oxidation of Fe using H2O is reported as 77.9 kJ/mol in the temperature range of 500−900 °C. Several kinetic models were introduced and were classified into diffusion-controlled, boundary-controlled, and random nucleation growth models. The Jander equation, a threedimensional diffusion-controlled model, gave the best fit for the oxidation of Fe. Volatile metal oxide thermochemical looping processes have been explored as methods of hydrogen production. Abanades10 recently studied the chemical kinetics of the thermochemical dissociation of water and carbon dioxide using SnO nanopowders as the reactive material via thermogravimetric analysis. An operating temperature for the dissociation of SnO2 to SnO was reported as 1627 °C. An activation energy of 51 ± 7 kJ/mol for the reduction of water to H2 in the temperature range of 525−650 °C was reported. Similarly, Weidenkaff et al.11 used TGA to study the dissociation of ZnO and the hydrolysis of Zn given various operating conditions. The activation energy for the dissociation of ZnO to metallic Zn was reported as 312−376 kJ/mol in the temperature range of 1000− 1550 °C. These processes have an advantage in that nearly full reduction of the oxidized metal can be achieved with thermal reduction, requiring no chemical reducant during the reduction phase. However, these volatile cycles require the condensation of the reactive material from a gas phase to a liquid or solid phase. These processes are energy intensive as large temperature swings are required for the condensation of the gas. Also, inconsistencies using

1. INTRODUCTION With fuel prices on the rise, the past decade has seen a renewed research interest in the steam-iron looping process as a method of producing pure hydrogen, which can be directly used in fuel cells and indirectly used for the synthesis of liquid fuels. In order to have an efficient reaction, this process requires a large reactive surface area, often in the form of small iron particles. One of the main hindrances to the advancement of the steam-iron looping process is the inability to maintain high chemically active surface area of small metallic particles in powders at high temperature. The agglomeration of these powders through sintering is one of the main technical challenges in hydrogen production using the looping process.1−3 Attempts have been made to develop longterm solutions to eliminate sintering,4−6 and it has been found that by adding dopant particles or coating reactive particles with inert materials, partial suppression of particle sintering can be achieved. These approaches continue to be active areas for research. Recently, a novel solution to exploit sintering was proposed by Mehdizadeh et al.7 By applying an external magnetic field to a fluidized mixture of iron and silica particles, a stabilized highly porous silica-iron bed is formed. During the first oxidation reaction on this bed of particles, and due to the presence of the external magnetic field, a porous structure consisting of organized iron particle chains with similar magnetic polarity is formed. Unlike sintering within conventional fluidized beds, the loss of chemically active surface area in this well controlled sintering process is very small at an operating temperature of 800 °C. A peak rate of 29 ccm of H2 per gram of material was reported. For the magnetically stabilized porous structure, Mehdizadeh et al.8 have reported that the oxidation step can be very well described by first order reaction kinetics with an activation energy of 88 ± 3 kJ/mol. © 2013 American Chemical Society

Received: Revised: Accepted: Published: 3683

October 2, 2012 January 16, 2013 February 20, 2013 February 20, 2013 dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

the two-step water splitting process acts like a “filter” and “quality booster”; hydrogen can be produced without the reductant affecting its purity. A schematic depiction of the chemical looping process is shown in Figure 1.

these processes are inherent since the physical characteristics of the reactive material will change with every redox cycle. Gálvez et al.12 considered hydrogen production using Mg/MgO as a reactive material and using solid carbon and methane as chemical reductants. Activation energies for the reduction of the MgO to metallic Mg and the oxidation of Mg were reported as 419.1−468.4 kJ/mol and 67.1−67.2 kJ/mol, respectively. The temperature range required for the reduction of MgO was reported as 1427−1627 °C. A shrinking core model13 with external mass transfer and internal diffusion for constant size spherical particles was used to predict the hydrogen yield from the oxidation of the Mg particles. In this current study, reaction kinetics of the magnetically stabilized iron-silica structure developed by Mehdizadeh et al.7 was investigated using thermogravimetric analysis for operating temperatures and water vapor mass fractions ranging from 725 to 850 °C and 3.7%−19.0%, respectively. The kinetic model presented by Mehdizadeh et al.8 is validated using this method. A series of high temperature X-ray diffraction patterns recorded in situ during reduction of the structure are presented to illustrate where conversion of the sample from a majority of magnetite (Fe3O4) to elemental iron (Fe) occurs. Scanning electron microscope (SEM) images are presented to show the physical changes that occur to the sample used in this analysis during repeated cycling. During the oxidation step of the iron-based looping process, steam is injected into a reactor containing reduced iron. The steam oxidizes iron to produce magnetite (Fe3O4) and high purity hydrogen is liberated. Hydrogen can be easily captured by condensing the excess water vapor from the output gas mixture. The ideal oxidation step reaction is 3Fe + 4H 2O ↔ Fe3O4 + 4H 2

Figure 1. Schematic representation of the steam-iron cycling process, using carbon monoxide as reductant.

2. IRON-SILICA MAGNETICALLY STABILIZED STRUCTURE FORMATION A detailed description of the synthesis of an iron-silica magnetically stabilized structure has been previously presented by Mehdizadeh et al.7,8 A brief explanation of the synthesis is provided here. A reactor containing 100 g of iron particles with a size range of 63−75 μm mixed with 105 g of silica particles in the size range of 75−106 μm is used. A uniform external magnetic field of 75 G is applied to the fluidized mixture to stabilize the bed and form magnetic chains of iron particles. The magnetic chains have the same polarity and thus repel each other in the normal direction and provide a natural spacing from neighboring chains. After the chains have formed, the stabilized bed is heated to approximately 700 °C under a low flow of steam. Between 400 and 600 °C, the iron particles will begin to oxidize and sinter with the adjacent particles, resulting in sintered chains that no longer require magnetic support. The resulting structure has high porosity and preserves chemically active surface area. While half of the material used (silica) is inert toward water splitting, the presence of the secondary nonmagnetic silica particles in the structure enhances the porosity and stability of the formed structure by providing spacing support between the iron chains. However, due to these secondary particles, a significant amount of heat during the redox cycling will be “wasted” on heating up material that does not contribute to the reaction. This would reduce the overall efficiency of the reaction if the chemically active surface area remained the same with or without the supporting particles. Since the silica particles allow for the surface area to remain constant during repeated cycling (which does not occur without the supporting particles), an overall advantage is gained by their presence. Mehdizadeh et al.7,8 demonstrated that the magnetically stabilized iron-silica structure significantly enhances the hydrogen production rate over repeated cycles, with a peak rate that is nearly double that of other iron-based reactive structures previously presented in the open literature.6,15 A schematic representation of the magnetically stabilized porous structure within the reactor and SEM images of the formed structure are shown in Figure 2. The dark gray particles are silica, and the light gray particles are iron/magnetite.

(1)

The above reaction takes place at temperatures above 400 °C, and the reaction rates increase as temperature rises. In the second step, the iron oxide (magnetite) must be reduced to elemental iron so that the water splitting process can proceed in a cyclic manner. This process can make use of any suitable chemical reductant to reduce the magnetite, for example syngas produced from the gasification of coal. During the reduction step, the magnetite reacts with carbon monoxide (CO) or hydrogen (H2) to produce both elemental iron (Fe) and carbon dioxide (CO2) or water (H2O) according to Fe3O4 + 4CO ↔ 3Fe + 4CO2

(2)

Fe3O4 + 4H 2 ↔ 3Fe + 4H 2O

(3)

or These reduction reactions take place at temperatures above 600 °C. However, caution should be applied to reduction using CO below 700 °C due to the likely formation of coke as predicted by the Boudouard reaction

2CO ↔ C + CO2

(4)

While syngas (CO + H2) is produced from the gasification of coal, it also contains a conglomeration of other gasified products, including carbon dioxide (CO2), methane (CH4), and trace amounts of hydrogen sulfide (H2S), carbonyl sulfide (COS), ammonia (NH3), and hydrogen cyanide (HCN).14 These “contaminants” have little effect on the ability of syngas to reduce magnetite back to iron and do not poison the iron; however, they prevent the direct usage of coal syngas for synthetic fuel production since sulfur compounds typically degrade the performance of catalysts. As such, 3684

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

corrosion kinetics during high temperature oxidation. TGA relies on a high degree of precision in three measurements: weight, temperature, and input gas flow rate. As such, a highly accurate balance (sensitivity: 5 μg), temperature sensor (S-type thermocouple, accurate to 1600 °C) and mass flow controllers (sensitivity: 0.1 sccm) have been used for this work. TGA was carried out using a Netzsch 409 CD simultaneous thermal analyzer (STA), using an Al2O3 crucible to contain the sample. The specimens used for TGA analysis had an initial mass of approximately 100 mg. Ultra high purity (UHP) argon was passed through an adsorbent bed (Agilent) to remove traces of oxygen, water, etc. A constant stream of UHP Ar of 0.071 g min−1 was passed through the TGA balance as a protective gas. Water vapor was produced by passing a second stream of Ar through a H2O bubbler which was thermostatically maintained at 60 °C. A schematic representation of the H2O bubbler is shown in Figure 3b. By passing Ar through a bath of H2O at a set temperature, the amount of H2O introduced into the system can be calculated. Assuming 100% saturation of water vapor into the argon gas stream through the bubbler, the amount of water vapor (in g min−1) carried by the gas stream is calculated via eq 5. The saturation pressure of water vapor at 60 °C is 19.941 kPa. The atmospheric pressure at the site of the TGA (Albuquerque, NM) was 85.3 kPa. ṁ WV = mAr ̇ 2

Psat , WV Ptotal − Psat , WV

(5)

The mass fraction of water vapor is defined as the mass flow rate of water vapor into the TGA divided by the total mass flow rate. This is calculated via eq 6 ωWV =

mAr ̇ 1

ṁ WV + mAr ̇ 2 + ṁ WV

(6)

where ṁ Ar1 is the argon mass flow rate directly into the TGA, and ṁ Ar2 is the argon mass flow rate through the bubbler. For the reduction step, UHP CO was used as received. The TGA was evacuated and backfilled with argon a minimum of three times, and then the system was allowed to equilibrate at ambient temperature and pressure for at least three hours prior to starting the measurement. This was performed to remove any excess H2O from the system after the previous experimental run and any air introduced upon changing samples. Thermochemical cycling experiments in the TGA were conducted under heating and cooling rates of 25 °C min−1. Gas flow rates were varied during studies of the concentration effect and remained constant during studies of the temperature effect. The sample was heated under Ar to the desired reduction temperature and then held isothermally for various periods under CO (until fully reduced) and various periods under water vapor (until fully reoxidized). The system was evacuated after each run before reweighing the sample to remove all excess H2O from the system. A baseline run was conducted under identical conditions to those described above but without a sample, and the data from the baseline run were subtracted from all sample runs, resulting in virtual absence of any buoyancy or gas-switching effects in the reported results. A schematic representation of the experimental set up used in the TGA experiments is shown in Figure 3a. 3.2. High-Temperature X-ray Diffraction (HT-XRD). High-temperature X-ray diffraction (HT-XRD) experiments were performed using a Scintag PAD X diffractometer (Thermo Electron Inc.). This diffactometer is equipped with a sealed-tube source (Cu Kα, λ = 0.15406 nm), an incident-beam mirror optic, a peltier-cooled

Figure 2. a) Schematic representation of the magnetically stabilized iron-silica porous structure, Mehdizadeh et al.8 and b) SEM images of the formed structure, Mehdizadeh et al.7

3. EXPERIMENTAL SECTION 3.1. Thermogravimetric Analysis (TGA). Thermogravimetric analysis (TGA) was performed on a small piece of the magnetically stabilized structure synthesized as previously described. TGA determines changes in sample weight in relation to changes in temperature and atmosphere. It is often used to estimate the 3685

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

Figure 3. a) Schematic representation of TGA experimental setup and b) schematic representation of H2O bubbler used for entrainment of water vapor.

representation of the measurement system is shown in Figure 4. A representative sample of the magnetically stabilized iron/silica structure was ground to a fine powder, dispersed in methanol, and deposited as a thin, even film onto a 10 × 10 × 0.5 mm3 platelet of single-crystal 8YSZ (yttria-stabilized zirconia containing 8 mol-% Y2O3). The methanol was allowed to evaporate fully before loading the specimen into the HT-XRD instrument.16

Ge solid-state detector, and a Buehler hot-stage with Pt/Rh heating strip and surround heater. The hot stage resides within a sealed chamber with an X-ray-transparent beryllium window. An all-metal gas manifold was attached to the inlet of the reaction cell allowing the controlled flow of hydrogen or carbon dioxide through the cell. Due to the HT-XRD instrument configuration and safety concerns, H2 is used for reduction and CO2 is used for oxidation; the redox reactions of the iron system (eqs 1 and 2) are not expected to be significantly different under these modified environmental conditions. Hydrogen (3 vol-% in He) was used as received. An oxygen- and moisturespecific adsorbent purifier bed was used in the CO2 line. A schematic

4. RESULTS AND DISCUSSION 4.1. Effect of Temperature on Oxidation. Several TGA experiments were run on the same magnetically stabilized structure 3686

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

concentration of water vapor was used to reduce the likelihood of condensation of H2O from the gas stream. As the oxidation temperature increases, the rate of oxidation of the sample also increases. In order to verify that the sample does not degrade with repeated cycling, oxidation at 725 °C was repeated after all other runs were conducted, and the data shown in Figure 5 are reproducible. The same sample in the magnetically stabilized bed reactor8 has shown to have a much faster rate of oxidation for the first 15 min of the reaction and then slows to a much lower rate of oxidation in comparison to the present data. The reason for this difference can be attributed to different test conditions for the two experimental setups. While the temperature and flow rate may be the same, the magnetically stabilized bed reactor forces inlet gas flow through the sample bed, causing more of the sample to be saturated with water vapor at one time. In the TGA, by contrast, steam has to diffuse through the sample, and this restricts the reaction kinetics. Other parameters that may affect this initial surge of hydrogen production are the lower mass fractions of water vapor in the inlet gas stream (7.2% versus 100%) and small sample size (0.1 g versus 255 g) in the TGA experiments. In order to compare the performance of the structure to other iron-based looping systems in the open literature, the thermogravimetric data were converted to hydrogen production rates based on ideal reaction stoichiometry using the molecular weights of iron (55.845 g mol−1) and hydrogen (2.0158 g mol−1) and the density of hydrogen at STP (0.08988 g L−1). The data, given in standard cubic centimeters of hydrogen produced per minute per gram of iron in the sample, are shown in Figure 6.

Figure 4. Schematic representation of HT-XRD experimental setup.

specimen with constant reduction conditions (same flow rate of CO, time, and temperature) and varied oxidation temperatures (but constant mass flow rate of H2O). The sample was originally fully oxidized and was returned to a fully oxidized condition between experiments by saturating the sample with H2O at the oxidation temperature after each experiment. The fractional iron conversion is defined as mreactive , t − mFe , specimen XFe = mFe3O4 , specimen − mFe , specimen (7) where mreactive,t is the weight of reactive material in the sample at any given time, mFe,specimen is the weight of the reactive material in the specimen when all of the material has been converted to iron, and mFe3O4,specimen is the weight of the reactive material in the specimen when all of the material has been converted to magnetite. The specimen was reduced for seven hours under 0.005 g min−1 of CO and 0.071 g min−1 of Ar to bring it to a fully reduced state. Results of the subsequent 7.5 h oxidation under a stream of 7.2% water vapor mass fraction in Ar (0.00733 g min−1 of water vapor in 0.095 g min−1 of Ar) are shown in Figure 5. This low

Figure 6. Hydrogen production rates of the magnetically stabilized structure at various temperatures exposed to an inlet gas stream containing 7.2% water vapor mass fraction.

For a water vapor mass fraction of 7.2%, a peak hydrogen production rate of 3.7 sccm per gram of iron was produced by the sample at 850 °C. The rates for higher concentrations of water will be reported in section 4.2. Hydrogen production rate data were fit to the hybrid contracting sphere/3D diffusion model developed by Mehdizadeh et al.8 The overall dimensionless reaction rate constant, k+ = k0τ exp(−Ea/RT) (k0 and τ are the reaction rate constant (s−1) and the residence time (s), respectively), is extracted from each data set by finding the smallest possible error between the

Figure 5. Effect of temperature on the oxidation of the magnetically stabilized structure exposed to an inlet gas stream containing 7.2% water vapor mass fraction. 3687

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

measured hydrogen production rate and the model prediction. The ln k+ determined for each temperature is shown in Figure 7.

Figure 8. Effect of water vapor mass fraction on the oxidation of the magnetically stabilized porous structure at 800 °C. Figure 7. Arrhenius plot of the magnetically stabilized structure at various temperatures.

As expected, as the mass fraction of water vapor increases, the rate of oxidation also increases. In order to determine the order of reaction, the above data were converted to hydrogen production rates in the same manner as described in section 4.1. The inlet water vapor concentration is computed as

An activation energy of 98 ± 3 kJ/mol was calculated from the data in Figure 9 for the specimen of the magnetically stabilized iron-silica porous structure sample. The value obtained by Mehdizadeh et al.8 is 88 ± 3 kJ/mol. While the value obtained here falls outside of the error bounds presented by Mehdizadeh et al.,8 it should be noted that several conditions present in the TGA are different than that in the laboratory reactor. Different temperature ramp-up rates might result in the formation of different solid phases of iron and hence a different effective activation energy. 4.2. Effect of Water Vapor Concentration on Oxidation. Several TGA redox cycles were run for the same magnetically stabilized structure sample with constant reduction conditions (0.005 g min−1 CO, 7 h, and 800 °C) and constant oxidation temperature (800 °C), but the inlet water vapor mass fractions were varied. The water vapor conditions for these experiments are shown in Table 1.

Psat , WV

C0 =

ρ0 M H2O

RTsat , WV

= 1+

1 − ωWV M H2O ωWV MAr

(mol/m 3) (8)

where C0 is the molar concentration of water vapor at the inlet to the TGA, ρ0 is the species density of the water vapor at the inlet to the TGA, R is the universal gas constant (J mol−1 K−1), and MH2O and MAr are the molecular weights of water and argon, respectively. The hydrogen production rates obtained at 17%, 34%, 51%, 68%, and 79% of fractional iron conversion were selected and plotted for each water vapor concentration condition and are shown in Figure 9.

Table 1. Flow Rate Conditions for Water Vapor Concentration Studies water vapor mass fraction (%)

stream 1 (Ar) mass flow rate (g min−1)

stream 2 (bubbler, Ar) mass flow rate (g min−1)

water vapor mass flow rate (g min−1)

19.0 16.1 11.3 7.2 3.7

0.071 0.071 0.071 0.071 0.071

0.237 0.120 0.051 0.024 0.0103

0.0723 0.0366 0.0157 0.00733 0.00315

The specimen was originally fully oxidized and was returned to a fully oxidized condition between experiments by saturating it with H2O at 800 °C after each experiment. The sample was reduced for ten hours at 800 °C under 0.005 g min−1 of CO and 0.071 g min−1 of Ar to bring the sample to a fully reduced state. The length of each oxidation cycle was adjusted according to the percentage of water vapor used. Lower percentages required longer run times. Results of the fractional iron conversion with time for varied water vapor mass fraction are shown in Figure 8.

Figure 9. Hydrogen production rate dependence on inlet water vapor concentration at 800 °C. 3688

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

Figure 10. HT-XRD pattern of reduction of the iron-silica magnetically stabilized structure using a reducing gas of 98.5% mass fraction He and 1.5% mass fraction H2. A change of the structure from magnetite to elemental iron occurred between 600 and 800 °C, with FeO appearing as an intermediate beginning at 700 °C.

Assuming that the hydrogen production rate is proportional to concentration to the power n (ṙH2 ∝ CnH2O), Figure 9 gives the order of the reaction as n = 1.28. This result validates the order of reaction obtained by Mehdizadeh et al.,8 which is approximately unity. 4.3. High Temperature X-ray Diffraction (HT-XRD) Patterns. In order to determine the temperatures at which changes in structure occur during reduction, High Temperature X-ray Diffraction (HT-XRD) was performed in situ under the presence of a 98.5 mass-% He and 1.5 mass-% H2 forming gas. Due to safety concerns, the forming gas is used; results are not expected to be significantly different under these modified environmental conditions. The results of this analysis are presented in Figure 10. Each horizontal stripe in the figure represents one XRD pattern recorded at the indicated temperature. The chronology of the experiment increases from the bottom of the figure to the top. Lighter bands represent higher intensity of diffracted X-rays (i.e., a peak, if plotted as intensity versus diffraction angle). The major phases that were detected are indicated in the figure. The shifting of peaks for any particular structure to lower angle as temperature increases is due to thermal expansion of the specimen. In other systems where the iron oxides have finite solubility in another component present (for instance, zirconia or yttria-stabilized zirconia), the precise peak shift (accounting for thermal expansion) can also provide information about the degree of solubility of iron in the matrix as a function of temperature and oxidation state.17 It can be seen in Figure 10 that the reactive component of the sample reduced from magnetite to elemental iron between 700 and 800 °C. Quartz silica (SiO2) was present throughout the experiment. These results indicate the importance of reducing above 700 °C, previously simulated using thermodynamic modeling,17 as seen in Figure 11.

Figure 11. Open system solid mole fractions for reduction by H2 at 1 bar pressure: (a) stoichiometric amount of H2, (b) two times stoichiometric H2, and (c) four times stoichiometric H2.

Since the sample used in the HT-XRD analysis was exposed to many times the stoichiometric amount of forming gas, Figure 11c indicates that the solid composition of the sample would contain mostly iron at 800−1000 °C, with a slight amount of wüstite 3689

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

Figure 12. HT-XRD pattern of oxidation of the iron-silica magnetically stabilized structure using a CO2 oxidation gas stream. A change of the structure from elemental iron to magnetite occurred between 600 and 800 °C.

present. This is clearly present in Figure 10. This work validates previous thermodynamic modeling presented by Klausner et al.,17 which presents reduction of iron oxide by carbon monoxide. The disappearance of the FeO peaks at high temperature is consistent with the data in Figure 11 for hyperstoichiometric H2:Fe mixes, as is the reappearance of FeO once the specimen was cooled to below about 800 °C. In order to determine the temperatures at which changes in structure occur during oxidation, HT-XRD was performed in situ under the presence of CO2. The results of this analysis are presented in Figure 12. In Figure 12, it can be seen that the reactive component of the sample oxidized from elemental iron to magnetite between approximately 600 and 800 °C. Quartz silica (SiO2) was present throughout the experiment. These results indicate the importance of oxidizing above 600 °C, previously simulated using thermodynamic modeling,17 as seen in Figure 13. Since the sample used in the HT-XRD analysis was exposed to many times the stoichiometric amount of CO2, Figure 13c indicates that the solid composition of the sample should contain mostly magnetite at 800−1000 °C, with lesser amounts of Fe2O3 and FeO present as well. This coincides with the thermodynamic modeling presented by Klausner et al.,17 which presents oxidation of iron oxide using H2O. While magnetite is clearly present at these elevated temperatures in Figure 12, hematite and wüstite do not appear at all. However, given that the intensity of XRD peaks increases with the amount of that crystal structure present in the sample, Figure 13 does predict the most dominant phase shown in Figure 12. Structures present at low concentration may not be readily detectable in the HT-XRD; rapid scan times are necessary to track changes in the sample, but these also lead to less resolution.

Figure 13. Open system solid mole fractions for oxidation by CO2 at 1 bar pressure: (a) stoichiometric amount of CO2, (b) two times stoichiometric CO2, and (c) four times stoichiometric CO2.

4.4. Scanning Electron Microscope (SEM) Images. After the thermogravimetric analysis cycles were completed on the 3690

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research

Article

Figure 14. SEM images of the iron-silica magnetically stabilized structure after 50 redox cycles.

From the collection of these images, it can be seen that the physical structure of the sample has changed. However, since no decay in the reactivity of the sample after 50 cycles of chemical reduction and oxidation was seen during TGA, these physical changes appear to be inconsequential to reactivity.

specimen of the iron-silica magnetically stabilized structure, SEM images were taken and compared to those of an uncycled specimen to demonstrate the physical changes that occurred to the sample during the cycles. Several of these images are presented in Figure 14. These images present a striking difference to those of the freshly made structure shown in Figure 2. As can be seen best in Figure 14c, the continued cycling of the sample has caused iron to migrate away from its “chains” and to produce a coating of magnetite crystals on the surface of the silica particles. A closer view of these individual crystals is shown in Figure 14d. In other locations in the sample, once a silica particle has been fully coated, the iron/magnetite seems to form dendritic extensions from the surface of the silica particle, as can be seen in Figure 14b. These dendrites are uncharacteristic for magnetite. However, utilization of the Energy-Dispersive X-ray Spectroscopy (EDS) function of the SEM on a magnetite crystal in Figure 14d and on a “magnetite dendrite” in Figure 14b yields the same element al composition−the composition corresponding to magnetite. A supposition is proposed for the formation of these magnetite dendrites. During the evacuation of the excess water vapor between redox cycles in the TGA, residual water vapor is subjected to low pressure (30 mbar). Under these conditions, goethite (FeO(OH)) formation is thermodynamically favorable if the partial pressure of water vapor is high enough.18 Goethite forms dendritic crystals, which could subsequently be turned into magnetite after a high temperature oxidation cycle. Further research is required to confirm the above hypothesis. Palumbo et al.19 have demonstrated the stepwise conversion of goethite to hematite (beginning at 210 °C) and then to magnetite (beginning at 1170 °C) under both reducing and oxidizing conditions.

5. CONCLUSIONS Thermogravimetric analysis on a 0.1 g sample of an iron-silica magnetically stabilized bed structure has been performed. An analytical study of the reaction kinetics of the structure has been carried out. Further validation of the model proposed by Mehdizadeh et al.8 has been achieved. The activation energy and the reaction rate constant of the magnetically stabilized iron-silica porous structure have been found to be 98 (kJ/mol) and 1.11 × 105 (s−1), respectively. The order of the reaction due to a varied concentration of water vapor during oxidation has been found to be 1.28. Reduction of the structure was performed in situ in an HT-XRD, and a reactive material crystalline structure change from magnetite to elemental iron was found to occur between 700 and 800 °C. SEM images of the sample after 50 redox cycles were taken, and physical changes to the sample can be seen. However, no decay in reactivity of the sample was seen via TGA, thus indicating that these physical changes do not affect the performance of the structure during repeated cycling.



AUTHOR INFORMATION

Corresponding Author

*Phone: +1 (352) 392-1086. E-mail: allek022@ufl.edu. Notes

The authors declare no competing financial interest. 3691

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692

Industrial & Engineering Chemistry Research



Article

contact mass in the RESC process for hydrogen production. Int. J. Hydrogen Energy 2006, 31, 2025−2031. (6) Otsuka, K.; Kaburagi, T.; Yamada, C.; Takenaka, S. Chemical storage of hydrogen by modified iron oxides. J. Power Sources 2003, 122, 111−121. (7) Mehdizadeh, A. M.; Klausner, J. F.; Barde, A.; Mei, R. Enhancement of thermochemical hydrogen production using an iron−silica magnetically stabilized porous structure. Int. J. Hydrogen Energy 2012, 37, 8954− 8963. (8) Mehdizadeh, A. M.; Klausner, J. F.; Barde, A.; Rahmatian, N.; Mei, R. Investigation of hydrogen production reaction kinetics for an ironsilica magnetically stabilized porous. Int. J. Hydrogen Energy 2012, 37, 13263−13271. (9) Go, K. S.; Son, S. R.; Kim, S. D. Reaction kinetics of reduction and oxidation of metal oxides for hydrogen production. Int. J. Hydrogen Energy 2008, 33, 5986−5995. (10) Abanades, S. CO2 and H2O reduction by solar thermochemical looping using SnO2/SnO redox reactions: Thermogravimetric analysis. Int. J. Hydrogen Energy 2012, 37, 8223−8231. (11) Weidenkaff, A.; Reller, A.; Wokaun, A.; Steinfeld, A. Thermogravimetric analysis of the ZnO/Zn water splitting cycle. Thermochim. Acta 2000, 359, 69−75. (12) Gálvez, M. E.; Frei, A.; Albisetti, G.; Lunardi, G.; Steinfeld, A. Solar hydrogen production via a two-step thermochemical process based on MgO/Mg redox reactionsThermodynamic and kinetic analyses. Int. J. Hydrogen Energy 2008, 33, 2880−2890. (13) Abanades, S.; Flamant, G. Thermochemical hydrogen production from a two-step solar-driven water-splitting cycle based on cerium oxides. Solar Energy 2006, 80, 1611−1623. (14) Levenspiel, O. Chemical Reaction Engineering; Wiley: New York, 1972. (15) Driscoll, D. J.; Ackiewics, M. Hydrogen from Coal Multi-Year RD&D Plan; DOE NETL, 2010. (16) Coker, E. N.; Ambrosini, A.; Rodriguez, M. A.; Miller, J. E. FerriteYSZ composites for solar thermochemical production of synthetic fuels: in operando characterization of CO2 reduction. J. Mater. Chem. 2011, 21, 10767. (17) Klausner, J. F.; Hahn, D. W.; Petrasch, J.; Mei, R.; Mehdizadeh, A. M.; Barde, A.; Allen, K. M.; Stehle, R. C.; Bobek, S. M.; Al-Raqom, F.; Greek, B.; Li, L.; Abhishek, S. Quarterly report 3: Novel magnetically fluidized bed reactor development for the looping process: coal to hydrogen production R&D; DOE NETL, Project DE-FE0001321, 2010. (18) Svoboda, K.; Slowinski, G.; Rogut, J.; Baxter, D. Thermodynamic possibilities and constraints for pure hydrogen production by iron based chemical looping process at lower temperatures. Energy Convers. Manage. 2007, 48, 3063−3073. (19) Palumbo, R.; Diver, R. B.; Larson, C.; Coker, E. N.; Miller, J. E.; Guertin, J.; Schoer, J.; Meyer, M.; Siegel, N. P. Solar thermal decoupled water electrolysis process I: Proof of concept. Chem. Eng. Sci. 2012, 84, 372−380.

ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy under Award No. DE-FE0001321 and partially supported by the National Institute for Nano Engineering program at Sandia National Laboratories. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000. We thank Mark A. Rodriguez, James NMI Griego, and Jonathan Torres (Sandia National Laboratories) for help with HT-XRD data collection and analysis. We thank Abhishek Singh (University of Florida) for help with the thermodynamic modeling.



NOMENCLATURE C0 = molar concentration of water vapor at the inlet to the TGA [-] Ea = activation energy [kJ mol−1] k+ = overall dimensionless rate constant [-] for model by Mehdizadeh et al.8 k0 = reaction rate constant [s−1] for model by Mehdizadeh et al.8 MH2O = molecular weight of water [g mol−1] MAr = molecular weight of argon [g mol−1] ṁ Ar = mass flow rate of argon from stream 2 into the TGA [g min−1] ṁ Ar2 = mass flow rate of argon from stream 1 into the TGA [g min−1] mFe,specimen = mass of the reactive material in the specimen when all of the material has been converted to iron [g] mFe3O4,specimen = mass of the reactive material in the specimen when all of the material has been converted to magnetite [g] ṁ WV = mass flow rate of water vapor into the TGA [g min−1] mreactive,t = mass of reactive material in the sample at any given time [g] n = order of reaction [-] Psat,WV = saturation pressure of water vapor at the inlet to the TGA [kPa] Ptotal = total atmospheric pressure at site of TGA [kPa] R = universal gas constant [J mol−1 K−1] ṙH2 = rate of hydrogen production [sccm H2 g−1 Fe] Tsat,WV = saturation temperature of water vapor at the inlet to the TGA [K] XFe = fractional iron conversion [-]

Greek Letters

ωWV - water vapor mass fraction at the inlet to the TGA [-] ρ0 - species density of the water vapor at the inlet to the TGA [g cc−1] τ - residence time [s] for model by Mehdizadeh et al.8



REFERENCES

(1) Lorente, E.; Peña, J. A.; Herguido, J. Cycle behaviour of iron ores in the steam-iron process. Int. J. Hydrogen Energy 2011, 36, 7043−7050. (2) Fukase, S.; Suzuka, T. Residual oil cracking with generation of hydrogen: deactivation of iron oxide catalyst in the steam-iron reaction. Appl. Catal., A 1993, 100, 1−17. (3) Knight, P. C.; Seville, J. P. K.; Kamiya, H.; Horio, M. Modelling of sintering of iron particles in high-temperature gas fluidisation. Chem. Eng. Sci. 2000, 55, 4783−4787. (4) Ettabirou, M.; Dupré, B.; Gleitzer, C. Nucleation and early growth of magnetite on synthetic and natural hematite crystals. React. Solids 1986, 1, 329−343. (5) Thaler, M.; Hacker, V.; Anilkumar, M.; Albering, J.; Besenhard, J. O.; Schröttner, H.; Schmied, M. Investigations of cycle behaviour of the 3692

dx.doi.org/10.1021/ie302691e | Ind. Eng. Chem. Res. 2013, 52, 3683−3692