international journal of energy 40 (2015) 1639e1650

Available online at www.sciencedirect.com ScienceDirect

journal homepage: www.elsevier.com/locate/he

Solar hydrogen production via thermochemical oxideeiron sulfate cycle

* Rahul R. Bhosale , Anand Kumar, Leo J.P. van den Broeke, Shahd Gharbia, Dareen Dardor, Mehak Jilani, Jamila Folady, Mashail Shaif Al-Fakih, Mahsa Ali Tarsad

Department of Chemical Engineering, College of Engineering, Qatar University, Doha, Qatar article info abstract

Article history: This paper reports the thermodynamic analysis of solar H2 production via two-step ther- Received 18 September 2014 mochemical iron oxideeiron sulfate (IOeIS) water splitting cycle. The first step belongs to

Received in revised form the exothermic oxidation of FeO via SO2 and H2O producing FeSO4 and H2 and second step

19 November 2014 corresponds to the endothermic reduction of FeSO4 into FeO, SO2, and O2. The products,

Accepted 25 November 2014 FeO and SO2 can be recycled to step 1 and hence, reutilized for the production of H2 via Available online 20 December 2014 water splitting reaction. Thermodynamic equilibrium compositions and variations in enthalpy, entropy and Gibbs free energy of the thermal reduction and water splitting re- Keywords: actions were computed as a function of reaction temperatures. Furthermore, the effect of _ Solar fuel molar flow rate of inert Ar (nAr) on thermal reduction temperature (TR) and equilibrium

Iron oxideeiron sulfate water compositions during the thermal reduction of FeSO4 was also examined. Second law h splitting cycle thermodynamic analysis was performed to determine the cycle efficiency ( cycle) and solar h Thermodynamics to fuel energy conversion efficiency ( solartofuel) attainable with and without heat recu- _ e e Hydrogen production peration for varying nAr (0 30 mol/s) and TR (1280 1510 K). Results obtained indicate h ¼ h ¼ h ¼ cycle 39.56% and solartofuel 47.74% (without heat recuperation) and cycle 51.77% and h ¼ ¼ solartofuel 62.43% (by applying 50% heat recuperation) at TR 1510 K. Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

favored thermodynamically as it requires process heat at Introduction temperatures above 2723 K for obtaining a significant degree of dissociation of water [1]. Furthermore, to avoid the forma- H2 is considered as one of the potential energy carriers due to tion of an explosive mixture comprised of H2 and O2, a gas its capacity of producing 143 MJ/kg of energy (higher in separator unit (for the separation of H2 and O2 at high comparison with the oil, gas, and coal individually) and its temperatures) needs to be equipped near to the water disso- environmentally friendly nature (burning of H2 produces ciation reactor which further enhances the overall cost of the water with no other polluting gases). Water splitting reaction H2 production process. is considered as one of the most promising ways for the pro- As the direct dissociation of water is not practical due to duction of H2. However, direct water thermolysis is not the high operating temperature and gas separation problem,

* Corresponding author. Tel.: þ974 4403 4168; fax: þ974 4403 4131. E-mail addresses: [email protected], [email protected] (R.R. Bhosale). http://dx.doi.org/10.1016/j.ijhydene.2014.11.118 0360-3199/Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. 1640 international journal of hydrogen energy 40 (2015) 1639e1650

attempts are currently underway to achieve H2 production via investigated metal catalysts were loaded either on non- water-splitting reaction at lower operating temperatures and sulfating substrates or substrates forming very stable sul- via bypassing the formation of H2 and O2 explosive mixture. fates. At high reaction temperatures, most of the transition Several thermochemical cycles such as iron oxide cycle [2e6], metal catalysts (such as nickel, manganese based) undergo /zinc oxide cycle [7e10], tin/tin oxide cycle [8,11e13], formation of stable sulfates and resulted as inefficient in e e e mixed ferrite cycle [14 21], ceria cycle [22 24], sulfur iodine terms of their catalytic activity towards the reduction of SO3 e cycle [25 27], and [28,29] were investigated into SO2 and O2 [31]. In contrast, the noble metal catalysts (e.g., e towards H2 production via water-splitting reaction. Among platinum based) supported on BaSO4 TiO2, ZrO2, or SiO2 were these, the sulfureiodine cycle (reaction set I) and its variation observed to be active towards the endothermic dissociation of the hybrid sulfur cycle (reaction set II) are more appealing as SO3. Although, the noble metal catalysts are attractive for the required operating temperatures are lower as compared to such reactions, they are less preferable due to the limited other thermochemical cycles. availability and high cost. Reaction set I: sulfureiodine cycle Utilization of metal oxides as the catalytic materials (instead of noble metal catalysts) and converting the sul- H SO /SO þ H O ð673 KÞ (1) 2 4 3 2 fureiodine and hybrid sulfur cycle into a ‘metal oxideemetal sulfate’ cycle operated using concentrated solar energy is one SO34SO2 þ 1=2O2 ð1123 1273 KÞ (2) of the alternative to achieve H2 production at moderate tem- peratures. Solar ‘metal oxideemetal sulfate’ thermochemical þ þ / þ ð Þ SO2 2H2O I2 H2SO4 2HI 393 K (3) cycle utilizes solar energy, metal oxide (MO), SO2 and H2O for

the production of H2 and O2. It is a two-step process in which / þ ð Þ 2HI H2 I2 573 723 K (4) the first non-solar step belongs to the exothermic oxidation of Reaction set II: hybrid sulfur cycle MO by SO2 and H2O producing metal sulfate (MSO4) and H2 (reaction 8). The endothermic step two corresponds to the / þ ð Þ H2SO4 SO3 H2O 673 K (5) solar thermal reduction of MSO4 into MO, SO2, and O2. The MO

and SO2 produced in step 2 are recycled back to step 1 and

SO34SO2 þ 1=2O2 ð1123 1273 KÞ (6) hence can be used in multiple cycles.

MO þ SO2ðgÞþH2OðgÞ/MSO4 þ H2ðgÞðNon solarÞ (8) SO2 þ 2H2O/H2SO4 ð353 393 KÞ (7)

For both sulfureiodine cycle and hybrid sulfur cycle, the / þ ð Þþ ð Þð Þ MSO4 MO SO2 g 1=2O2 g Solar (9) most energy consuming step is the dissociation of SO3 into

SO2 and O2. The reduction of SO3 into SO2 and O2 is possible According to the previous investigations [2e24], among the only under catalytic conditions and takes place at high tem- several MO systems examined for the solar thermochemical perature due to its endothermic nature. Several metal based water splitting application, iron oxide based redox materials catalytic systems have been investigated previously towards are considered as one of the most favorable materials. In this the decomposition of the gaseous SO3 [30]. As sulfation viewpoint, this paper proposes the utilization of the ‘iron e ’ e poisoning is a major concern related to such reactions, the oxide iron sulfate (IO IS) cycle for the production of solar H2

Fig. 1 e Typical two-step solar thermochemical IOeIS water splitting cycle. international journal of hydrogen energy 40 (2015) 1639e1650 1641

via thermochemical water splitting reaction. Typical two-step the IOeIS cycle were performed with the help of commercial solar thermochemical IOeIS water splitting cycle is presented thermodynamic HSC Chemistry software and databases [32] in Fig. 1. In this study, the thermodynamic feasibility of this by assuming a) solar reactor operated in continuous mode cycle was investigated and obtained results are presented in with inlet concentration of FeSO4 equal to 1 mol/s and b) by detail. At first, the thermodynamic equilibrium compositions considering FeO, SO2(g), H2O(g), FeSO4,H2(g), and O2(g) as the during the solar thermal reduction of FeSO4 under inert at- reactive species. mosphere and oxidation of FeO via water splitting reaction were determined. The variation of the reaction enthalpy, en- H2 production via water splitting reaction tropy and Gibbs free energy for the thermal reduction and water splitting steps with respect to the operating tempera- Fig. 2a and b represent variations in the equilibrium compo- tures were studied. Furthermore, the maximum theoretical sition for step 1 with respect to water splitting temperature solar energy conversion efficiency of the IOeIS cycle is (TS) (in absence of Ar). According to the results presented, the determined by performing the second law thermodynamic production of H2 via oxidation of FeO by SO2 and H2Ois analysis over different solar reactor temperatures and with/ possible below 483 K. It is further observed that at lower without considering the heat recuperation. temperatures, FeSO4 and H2 are the major species, whereas

the molar concentration of FeO, H2O and SO2 increases with

the increase in TS. D D D Chemical thermodynamic equilibrium Variations in HWS, SWS, and GWS as a function of TS for the production of H2 via oxidation of FeO by SO2 and H2Ois Simplified chemistry of the solar thermochemical IOeIS water presented in Fig. 3a and b. Following are the mathematical correlations derived from the obtained HSC simulation results splitting cycle involving thermal reduction of FeSO4 and water D D D splitting via FeO oxidation is shown below: for the estimation of HWS, SWS, and GWS at different TS. À Á À Á D ¼ 6 2 þð : Þ ð : Þ FeO þ SO2ðgÞþH2OðgÞ/FeSO4 þ H2ðgÞ (10) HWS 9 10 Ts 0 0129 Ts 123 08 300 900 K (12) / þ ð Þþ = ð Þ FeSO4 FeO SO2 g 1 2O2 g (11) À Á À Á DS ¼ 9 106 T2 þð0:0499ÞT 257:35 300 900 K The equilibrium thermodynamic calculations related to WS s s (13) À Á À Á D ¼ 5 2 þð : Þ : GWS 2 10 Ts 0 2532 Ts 119 36 300 900 K (14)

Fig. 2 e a) and b) Variations in equilibrium composition as Fig. 3 e Variations in a) enthalpy and entropy, and b) Gibbs a function of TS during H2 production via oxidation of FeO free energy as a function of TS during H2 production via ¼ _ ¼ ¼ _ ¼ by SO2 and H2O(P 1 atm, nAr 0 mol/s). oxidation of FeO by SO2 and H2O(P 1 atm, nAr 0 mol/s). 1642 international journal of hydrogen energy 40 (2015) 1639e1650

À Á À Á According to the results presented in Fig. 3a, DHws is varied D ¼ 6 2 ð : Þ þ : STR 3 10 TR 0 029 TR 297 6 300 900 K (16) by 11.35% with the increase in the TS from 300 to 900 K. Pre- À Á À Á sented results further indicate that more amount of 5 2 DGTR ¼ 1 10 T ð0:296ÞTR þ 361:29 300 900 K (17) exothermic heat will be released if the water splitting reaction R is carried out at 300 K (117.94 kJ/mol) instead of 900 K According to Fig. 4a, energy required for performing the D (104.56 kJ/mol). Likewise, DSws is also observed to be altered endothermic thermal reduction of FeSO4 i.e. HTR is decreased D by 10.79% (from 242.39 J/K mol to 216.23 J/K mol) due to the by 6.50% as the TR increased from 300 to 1500 K. Similarly, STR increase in the TS from 300 K to 900 K. As the water splitting is also reduced by 9.047% with the increase in TR (from 300 to reaction has a negative enthalpy and entropy change, this 1500 K). It is further realized that the thermal reduction of reaction seems to be more feasible at lower temperatures. The FeSO4 is more viable at high temperature as the reaction D data related to change in the Gws with respect to TS presented enthalpy and entropy has positive variation. This realization in Fig. 3b indicate a decrease of 149.40% in DGws with the is further confirmed from the data reported in Figs. 4b and 5. D decrease in the TS from 900 to 300 K. Furthermore, the results According to Fig. 4b, GTR for the thermal reduction of FeSO4 reported confirm that the production of H2 via oxidation of becomes negative above 1290 K (in absence of Ar) which

FeO by SO2 and H2O is feasible below 483 K. confirms that this reaction is possible at higher temperatures as compared to water splitting reaction. Furthermore, the

Step 2: thermal reduction of FeSO4 variations in equilibrium composition related to the thermal reduction of FeSO4 also confirm the above mentioned

Similar to the H2 production step, equilibrium thermody- observation. namic analysis of the thermal reduction of FeSO4 (in absence According to the results reported in previous investigations of Ar) was also simulated with the help of HSC Chemistry [2e24], the thermal reduction of MO can be achieved at lower software and databases. Variations in DHTR, DSTR, DGTR, and temperatures if inert carrier gas is used during the thermal equilibrium compositions as a function of TR are presented in reduction step. Based on this observation, we examined the Figs. 4(a and b) and 5, respectively. The mathematical ex- effect of Ar as an inert carrier gas during the thermal reduc- pressions (15)e(17) represents the change in the DHTR, DSTR, tion of FeSO4 in a solar reactor. The results reported in Fig. 6a _ DGTR with respect to TR. and b shows the influence of molar flow rate of Ar (nAr)on À Á À Á equilibrium composition of gaseous and solid species during DH ¼ 7 106 T2 ð0:0079ÞT þð363:93Þ 300 900 K TR R R thermal reduction of FeSO4. It is observed that the slope of the

(15) decrease in the molar concentration of FeSO4 and increase in

the molar concentration of FeO, SO2 and O2 is shifted towards _ ' lower TR as the nAr increases. As per the Le-Chatelier s prin- ciple, due to the inclusion of the inert gas during the solar

thermal reduction of FeSO4, the thermodynamic equilibrium

shifts towards the lower TR as the entropy of the product gaseous species decreases. _ Effect of nAr on TR required for complete conversion of

FeSO4 into FeO, SO2 and O2 is shown in Fig. 7. The results re- ported indicate that in absence of Ar, the complete conversion _ is possible at TR ¼ 1510 K. As the nAr increases upto 5 mol/s, the

required TR for complete conversion decreases to 1410 K. A

further decrease in the TR down to 1280 K is observed with the

Fig. 4 e Variations in a) enthalpy and entropy, and b) Gibbs Fig. 5 e Variations in equilibrium composition as a free energy as a function of TR during reduction of FeSO4 function of TR during reduction of FeSO4 into FeO, SO2 and ¼ _ ¼ ¼ _ ¼ into FeO, SO2 and O2 (P 1 atm, nAr 0 mol/s). O2 (P 1 atm, nAr 0 mol/s). international journal of hydrogen energy 40 (2015) 1639e1650 1643

Effective emissivity and absorptivity equal to unity All the products separates naturally without expending any work All reactions reach 100% completion

The exergy analysis was carried out by following the methodology and governing equations derived and used pre- viously for other MO based solar thermochemical cycles [7,33]. Thermodynamic properties were extracted from HSC soft- ware and databases and the analysis is normalized to the

FeSO4 molar flow rate (1 mol/s) entering the solar reactor.

Exergy analysis

Solar energy absorption efficiency The solar energy absorption efficiency is defined as the net rate at which energy is being absorbed by the solar reactor divided by the solar energy input through the aperture of the solar reactor. The solar energy absorption efficiency can be calculated as:   sT4 h ¼ 1 R (18) absorption IC

_ where, Fig. 6 e Effect of nAr on equilibrium composition of a) gaseous and b) solid species as a function of TR during I ¼ direct-normal solar irradiance (normal bean reduction of FeSO4 into FeO, SO2 and O2. insolation) ¼ 1000 (W/m2) C ¼ solar flux concentration ratio (ratio of the solar flux _ intensity achieved after concentration to the normal beam increase in nAr upto 30 mol/s. However, additional increase in _ insolation, dimensionless number) (suns) nAr upto 100 mol/s is of not much use as required TR reduces TR ¼ solar reactor temperature required for the thermal only by 30 K. reduction of FeSO4 (K) s ¼ StefaneBoltzmann constant ¼ 5:6705 108 (W/m2 K4)

h _ Second law thermodynamic analysis Fig. 7 represents variation in absorption as a function of nAr ¼ 2 ¼ and TR at constant I 1000 (W/m ) and C 5000 suns. Simu- h Process configuration and assumptions lation results indicate that the absorption increases with the _ increase in nAr and decrease in the TR. For instance, at ¼ _ ¼ h To assess the maximum possible energy conversion efficiency TR 1510 K (nAr 0 mol/s), absorption is equal to 94.10%. How- _ of the solar thermochemical IOeIS water splitting cycle, a ever, as nAr increases from 0 to 30 mol/s and corresponding TR h second law thermodynamic analysis (exergy analysis) was decreases from 1510 to 1280 K, absorption is increased upto carried out and obtained results are reported in this section. A 96.96%. process flow configuration for the continuously operated solar thermochemical IOeIS water splitting cycle is shown in Fig. 8.

This process is comprised of a solar reactor, FeO oxidizer (H2 generator), theoretical H2/O2 ideal fuel cell, gas separator, and few heaters and coolers. Production of H2 is carried out at 1 atm and steady state conditions with negligible viscous losses and changes in kinetic and potential energies. It is worthy to note that, the energy penalty associated with the separation of SO2 from O2 and Ar are not accounted during the efficiency calculation. Also, the heat exchangers required for recovering the sensible latent heat are omitted from thermo- dynamic considerations. In addition, several other assump- tions are made as follows:

The solar reactor is assumed to be a perfectly insulated e h _ blackbody absorber Fig. 7 absorption and required nAr at chemical equilibrium No convective/conductive heat losses as a function of TR. 1644 international journal of hydrogen energy 40 (2015) 1639e1650

e e Fig. 8 Process configuration for H2 production via two-step solar thermochemical IO IS water splitting cycle.

h _ Effect of C on absorption is also examined at different nAr and Heat energy required to produce steam by heating the

TR and findings are reported in Fig. 9. According to Fig. 9, liquid water can be determined as: h _ ¼ absorption increases with the increase in C. At nAr 5 mol/s and ¼ _D j ¼ h Qheater2 n H ð Þ@ / ð Þ@ (21) TR 1410 K, an increase in the absorption from 77.58 to 97.76% is H2O l 298 K H2 O g 398 K observed as C increases from 1000 to 10,000 suns. Similarly, at _ _ Energy penalty associated with the heating of steam from different TR such as 1510 K (nAr ¼ 0 mol/s), 1375 K (nAr ¼ 10 mol/ _ _ 398 K upto TS can be estimated as: s), 1340 K (nAr¼ 20 mol/s), and 1280 K (nAr ¼ 30 mol/s), the h _ increases by 26.53%, 18.24%, 16.45%, and 13.70% with Q ¼ nDHj ð Þ@ / ð Þ@ (22) absorption heater 3 H2O g 398 K H2O g 473 K the increase in C from 1000 to 10,000 suns. Equation (23) gives the amount of energy needed to heat Net solar energy absorbed for the operation of IOeIS cycle the inert Ar from ambient conditions upto TR: e To operate the solar thermochemical IO IS water splitting ¼ _D j QArheating n H ArðgÞ@298 K/ArðgÞ@T (23) cycle, several energy requirements associated with following R steps needs to be fulfilled by solar energy input. The Energy required for the complete reduction of FeSO4

into FeO, SO2, and O2 is given by following equation: Energy required for the complete reduction of FeSO4 into _ ¼ D j QFeSO4 reduction n H FeSO @473 K/FeOþSO ðgÞþ 1=2O ðgÞ@T (24) FeO, SO2, and O2 4 2 2 R Energy required for heating of Ar from ambient conditions

upto TR Energy required for heating of SO2 from ambient condi-

tions upto TS Energy required for the conversion of liquid water (298 K) into steam (398 K)

Energy required for the heating of steam from 398 K upto TS

The net solar energy absorbed for the operation of solar thermochemical IOeIS water splitting cycle can be calculated as:

¼ þ þ þ þ Qreactor net QFeSO4 reduction QAr heating Qheater 1 Qheater 2 Qheater 3 (19)

The energy required for the heating of SO2 from ambient conditions upto TS is given by:

¼ _D j Qheater1 n H so ðgÞ@298 K/so ðgÞ@473 K (20) e h 2 2 Fig. 9 Effect of C on absorption. international journal of hydrogen energy 40 (2015) 1639e1650 1645

e _ _ ¼ ¼ _ ¼ ¼ _ ¼ Fig. 10 Effect of nAr and TR on Qreactornet:a)nAr 0 mol/s, TR 1510 K, b) nAr 5 mol/s, TR 1410 K, c) nAr 10 mol/s, ¼ _ ¼ ¼ _ ¼ ¼ ¼ ¼ ¼ TR 1375 K, d) nAr 20 mol/s, TR 1340 K, and e) nAr 30 mol/s, TR 1280 K (1 Qheater1,2 Qheater2,3 Qheater3, ¼ ¼ 4 QArheating,5 QFeSO4reduction).

; ; ; _ Variations of Qreactor net, QFeSO4 reduction QAr heating Qheater 1 enhances from 0 to 613.35 kW as nAr increases from 0 to ; _ Qheater2 and Qheater3 as a function of nAr and TR are shown 30 mol/s. Overall Qreactornet increases by 50.665% (from _ in Fig. 10. The results presented indicate that the 563.522 kW to 1142.243 kW) with the increase in nAr (from 0 to ; ; Qheater1 Qheater2 and Qheater3 does not vary with respect to 30 mol/s) and decrease in the TR (from 1510 to 1280 K). _ nAr and TR and observed to be fixed at equal to 8.072 kW, 47.366 kW, and 2.625 kW, respectively. As per the results ob- Solar energy input _ tained, with the increase in nAr, required TR decreases and The amount of solar energy input required for the operation of solar thermochemical IOeIS water splitting cycle can be correspondingly QFeSO4 reduction also decreases. For example, as _ calculated as: nAr increases from 0 to 30 mol/s and TR decreases from 1510 K Qreactornet to 1280 K, correspondingly QFeSO reduction reduces from ¼ 4 Qsolar h (25) absorption 505.46 kW to 470.83 kW. In contrast to this, the QArheating 1646 international journal of hydrogen energy 40 (2015) 1639e1650

From the results reported in Fig. 11, Qsolar is observed to be maximum (1178.12 kW) at TR ¼ 1280 K as the large amount of energy is required for Ar heating. Furthermore, Qsolar decreases with the increase in the TR as less energy is required to in- crease the temperature of the decreasing Ar molar flow. As the _ TR increased from 1280 K (nAr ¼ 30 mol/s) upto 1510 K _ (nAr ¼ 0 mol/s), Qsolar decreases by 49.17%. To avoid the excessive energy penalty associated with the heating of the _ Ar, lower nAr or a larger reactor aperture and concentration system should be used in practice.

Radiation heat losses The amount of loss of heat energy due to re-radiation from the e Fig. 12 Effect of TR on Qreradiation and e solar reactor associated with the solar thermochemical IO IS % re radiation losses. water splitting cycle can be calculated as:

Qreradiation ¼ Qsolar Qreactornet (26) Opposite to this, Q increases with the increase in the T Also, % re-radiation losses can be determined as: cooler 2 R as the heat energy that can be recovered from the solid FeO

Q will be higher if the thermal reduction is carried out at higher % re radiation losses ¼ re radiation 100 (27) Qsolar temperatures. According to the simulation results presented in Fig. 12, Rate of heat rejected to the surrounding by FeO oxidizer and Qreradiation remain constant in the range of 30e37 kW with h cooler 3 respect to the increase in the TR. However, as the absorption and FeO generated via thermal reduction of FeSO4 is transferred to Qsolar decreases with the increase in TR,%re radiation losses the FeO oxidizer and reacted with the combined stream of SO increases from 3.045 upto 5.90%, respectively. 2 and H2O at 473 K for the production of H2. By assuming 100% oxidation of FeO producing FeSO and H via water splitting Rate of heat rejected to the surrounding by cooler 1 and cooler 2 4 2 reaction, the rate of heat rejected to the surrounding by FeO Gaseous (SO2,O2, and Ar) and solid (FeO) products generated oxidizer is estimated as 118.43 kW according to equation (30). during thermal reduction of FeSO4 exits the solar reactor and further cooled down via passing through cooler 1 and cooler 2. ¼_D j QFeOoxidizer n H FeOþH OðgÞþSO ðgÞ@473 K/FeSO þH ðgÞ@473 K (30) The rate of heat rejected by cooler 1 and cooler 2 is given by: 2 2 4 2 H generated at 473 K via water splitting reaction is allowed _ 2 Q ¼nDHj ð Þþ : ð Þþ ð Þ@ / ð Þþ : ð Þþ ð Þ@ (28) cooler 1 SO2 g 0 5O2 g Ar g TR SO2 g 0 5O2 g Ar g 298 K to pass through cooler 3 before entering the ideal fuel cell (operated at 298 K). According to equation (31), 5.035 kW of _ Q ¼nDHj @ / @ (29) heat is released by cooler 3 due to the cooling of H2 from 473 K cooler 2 FeO TR FeO 473 K to 298 K. As per the results reported in Fig. 13, with the increase in _ the TR from 1280 upto 1510 K, Qcooler1 reduces from 1135.52 kW Q ¼nDHj ð Þ@ / ð Þ@ (31) cooler 3 H2 g 473 K H2 g 298 K to 92.72 kW. This radical drop in the Qcooler1 is because of the _ decrease in the required nAr as the thermal reduction of FeSO4 Rate of entropy produced across the solar reactor, cooler 1, and _ is carried out at higher temperatures. As the utilization of nAr cooler 2 during the thermal reduction decreases, the heat energy that The rate of entropy produced across the solar reactor (during can be recovered from the gaseous products also decline. thermal reduction of FeSO4), cooler 1 (during cooling of

e _ e Fig. 11 Effect of TR and nAr on Qsolar. Fig. 13 Effect of TR on Qcooler1 and Qcooler2 . international journal of hydrogen energy 40 (2015) 1639e1650 1647

Rate of entropy produced across the FeO oxidizer and cooler 3 Similar to the previous section, this section is related to the estimation of the rate of entropy produced across the FeO

oxidizer (during H2 generation via water splitting reaction)

and cooler 3 (during cooling of H2 from 473 to 298 K). The en- tropies produced across the FeO oxidizer and cooler 3 are calculated with the help of equations (35) and (36) and observed to be equal to 0.010598 kW/K and 0.0035 kW/K.   QFeOoxidizer s_ ¼ FeO oxidizer 473 _ þ nDSj þ ð Þþ ð Þ@ / þ ð Þ@ (35) FeO H2 O g SO2 g 473 K FeSO4 H2 g 473 K e s_ s_ s_ Fig. 14 Effect of TR on reactor, cooler1, and cooler2.   _ Qcooler3 _ scooler3 ¼ þ nDSj ð Þ@ / ð Þ@ (36) 298 H2 g 473 K H2 g 298 K

Rate of theoretical work performed and heat rejected by the gaseous products), and cooler 2 (during cooling of solid prod- ideal fuel cell ucts) are given by following equations: Solar thermochemical IOeIS water splitting cycle is completed     with the addition of an ideal H /O fuel cell with 100% work 2 2 s_ ¼ Qsolar þ Qreradiation efficiency (assumption). This ideal fuel cell is added to the reactor T 473 R cycle to extract the maximum work from the net H produced _ 2 þ nDSj þ ð Þ@ / ð Þþ ð Þþ : ð Þþ ð Þ@ (32) FeSO4 Ar g 473 K FeO s SO2 g 0 5O2 g Ar g TR by this process. The rate of theoretical work performed by the   ideal fuel cell is calculate by equation (37) and observed to be s_ ¼ Qcooler1 equal to 237.05 kW. Likewise, the rate of heat energy released cooler1 298 by the ideal fuel cell is determined as equal to 48.56 kW [ac- _ þ nDSj ð Þþ : ð Þþ ð Þ@ / ð Þþ : ð Þþ ð Þ@ (33) cording to equation (38)]. SO2 g 0 5O2 g Ar g TR SO2 g 0 5O2 g Ar g 298 K   ̇ WFCIdeal ¼n DGj ð Þþ : ð Þ@ / ð Þ@ (37) _ Qcooler2 _ H2 g 0 5O2 g 298 K H2O l 298 K scooler2 ¼ þ nDSj @ / @ (34) 473 FeO TR FeO 473 K ̇ Q ¼ð298Þn DSj ð Þþ : ð Þ@ / ð Þ@ (38) Fig. 14 represents the change in the rate of entropy pro- FC Ideal H2 g 0 5O2 g 298 K H2 O l 298 K duced across the solar reactor, cooler 1, and cooler 2 as a function of TR. According to the obtained simulation results, Process efficiency s_ e reactor decreases by 32.96% due to the increase in the TR from Cycle efficiency of the solar thermochemical IO IS water s_ 1280 upto 1510 K. Similarly, 93.79% reduction in cooler1 is seen splitting process is defined as the ratio of theoretical work _ with the similar increase in the TR. nAr plays an integral role as performed by the ideal fuel cell to the solar energy input: _ _ both sreactor and scooler1 decreases until Ar dilution is no longer s_ h ¼ WFCIdeal required. As Ar dilution is not in the picture, cooler2 increases cycle (39) Qsolar by a factor of 1.489 as TR increases from 1280 to 1510 K.

e h h Fig. 15 Effect of TR on cycle and solartofuel. 1648 international journal of hydrogen energy 40 (2015) 1639e1650

h capable of achieving higher cycle 34.09% as compared to the Table 1 e h and h of solar solar to fuel other thermochemical cycles if operated at and above 1410 K. IOeIS water splitting cycle. Furthermore, if the energy required for the heating of inert Ar is h h TR (K) cycle (%) solartofuel(%) h ¼ neglected, this cycle can attain maximum cycle 44.83% at Recuperation ¼ 0% lower temperatures of 1280 K. 1510 39.56 47.73 1410 34.09 41.11 Heat recuperation 1375 31.12 37.52 The h and h for the solar thermochemical IOeIS 1340 23.54 28.40 cycle solar to fuel 1280 20.12 24.26 water splitting cycle can be enhanced further if the heat Recuperation ¼ 10% rejected by the cooler 1, cooler 2, cooler 3, and FeO oxidizer is 1510 41.54 50.09 recuperated and utilized to operate this cycle. The total 1410 36.07 43.50 amount of heat that can be recuperated is given as: 1375 33.23 40.08

1340 25.22 30.42 Qrecuperable ¼ Qcooler1 þ Qcooler2 þ Qcooler3 þ QFeOoxidizer (42) 1280 23.59 28.45 Recuperation ¼ 25% Furthermore, the solar energy input required for the 1510 44.86 54.10 operation of this cycle after heat recuperation can be esti- 1410 39.52 47.66 mated as: 1375 37.02 44.65

1340 28.35 34.18 Qsolar; with recuperation ¼ Qsolar ð% recuperationÞ Qrecuperable (43) 1280 27.86 33.60 h h Recuperation ¼ 40% The cycle and solartofuel after heat recuperation can be 1510 48.77 58.81 determined as: 1410 43.71 52.70 1375 41.79 50.39 h ¼ WFCIdeal cycle (44) 1340 32.36 39.01 Qsolar;with recuperation 1280 36.21 43.67 Recuperation ¼ 50% HHV h ¼ H2 1510 51.77 62.43 solartofuel (45) Qsolar;with recuperation 1410 47.02 56.63 h h 1375 45.71 55.12 Influence of heat recuperation on cycle and solartofuel is 1340 35.72 43.07 listed in Table 1. Reported data indicate that as the heat 1280 45.26 54.58 h h recuperation increases, both cycle and solartofuel increases ¼ h significantly. For instance, at TR 1280 K, the cycle and h Furthermore, the solar-to-fuel energy conversion effi- solartofuel increases by 25.14% and 30.32% if 50% heat recu- ciency of the solar thermochemical IOeIS water splitting peration is applied to the solar thermochemical IOeIS water process is defined as the ratio of higher heating value (HHV) of splitting cycle. the H2 produced to the solar energy input: Verification HHV h ¼ H2 solartofuel (40) The exergy analysis performed for the solar thermochemical Qsolar IOeIS water splitting cycle was further verified by performing an Where, energy balance and by evaluating the maximum available pro-

̇ cess efficiency from the total available work and from the total HHV ¼n DHj ð Þþ : ð Þ@ / ð Þ@ (41) H2 H2 g 0 5O2 g 298 K H2O l 298 K solar energy input [7]. The energy balance conducted for the e h h solar thermochemical IO IS water splitting cycle confirms that As per the HSC simulations, both cycle and solartofuel in- creases with the increase in T . For instance, at 1280 K, h is R cycle W ¼ Q ðQ þ Q þ Q þ Q h FC Ideal solar re radiation cooler 1 cooler 2 cooler 3 equal to 20.12% and solartofuel is equal to 24.26%, and as the TR þ þ Þ h h QFeOoxidizer QFCIdeal (46) increases upto 1510 K, cycle and solartofuel also enhanced upto ¼ ¼ _ ¼ 39.56% and 47.74% (Fig. 15). The energy balance at TR 1510 K, TS 473 K, and nAr 0mol/

According to previous studies, the maximum cycle efficiency s confirms the accuracy of the analysis as WFCIdeal ¼ 237.05 kW reported for different thermochemical cycles are as follows: 29% calculated from Eq. (37) equals WFCIdeal determined from Eq. (46). in case of ZnO/Zn cycle [7], 29.8% in case of SnO2/SnO cycle [11], The available work can be calculated as the sum of the

20.2% in case of ceria cycle [22], and 30% in case of Fe3O4/FeO work done by the ideal fuel cell, and the lost work due to cycle [34]. It is worthy to note that the cycle efficiencies of all the irreversibility's in solar reactor, cooling units, and the FeO- above mentioned thermochemical cycles were calculated by oxidizer. For the solar thermochemical IOeIS water splitting neglecting the amount of energy required for heating the inert cycle, the maximum cycle efficiency was calculated by Eq. (47) gas. The results obtained during the presented investigation and it was observed to be equal to the Carnot heat engine indicate that solar thermochemical IOeIS water splitting cycle is operating between hot and cold temperature reservoirs.

þ ðs_ þ s_ þ s_ þ s_ þ s_ Þ h ¼ WFCIdeal TS reactor cooler1 cooler2 cooler3 FeOoxidizer ¼ TS ¼ h cycle;maximum 1 carnot (47) Qsolar TR international journal of hydrogen energy 40 (2015) 1639e1650 1649

¼ ¼ _ ¼ For instance, at TR 1510 K, TS 473 K, and nAr 0 mol/s, Qcooler1 heat rejected to the surrounding from cooler 1, kW h ¼ h ¼ cycle;maximum carnot 68.67%. Qcooler2 heat rejected to the surrounding from cooler 2, kW

Qcooler3 Heat rejected to the surrounding from cooler 3, kW

QFCIdeal heat rejected to the surrounding from ideal fuel cell, Summary and conclusions kW QFeOoxidizer heat rejected to the surrounding from FeO Two-step solar thermochemical iron oxideeiron sulfate oxidizer, kW e Q energy required for the thermal reduction of (IO IS) water splitting cycle for the production of H2 have been FeSO4 reduction thermodynamically investigated. Thermodynamic equilib- FeSO4,kW rium calculations performed at 1 atm pressure and in absence Qheater1 energy required for heating of SO2,kW Q energy required for production of steam, kW of inert Ar indicate that the H2 production via water splitting heater 2 reaction is feasible below 483 K and thermal reduction of Qheater3 energy required for heating of steam, kW Q net energy input required for the operation of IOeIS FeSO4 is possible above 1290 K. The change in the enthalpy reactor net and entropy for the water splitting reaction is observed to be cycle, kW negative (indicating that this reaction is more feasible at lower Qreradiation radiation heat loss from the solar reactor, Kw temperatures), whereas the variation in enthalpy and entropy Qrecuperable total amount of heat that can be recuperated, kW Q solar energy input, kW associated with the thermal reduction of FeSO4 is observed as solar positive (confirming that this reaction is more feasible at Qsolar;with recuperation solar power input after heat recuperation, kW higher temperatures). The TR required for the complete con- T thermal reduction temperature, K version of FeSO4 into FeO, SO2, and O2 is observed to be R _ T water splitting temperature, K decreased from 1510 to 1280 K with the increase in the nAr S from 0 to 30 mol/s. WFCIdeal work output of an ideal fuel cell, kW h The second law thermodynamic analysis indicate that the absorption solar absorption efficiency h h carnot efficiency absorption decreases by 2.86% with the increase in TR from 1280 carnot h to 1510 K (at C ¼ 5000 suns) and increases by 26.53% with the cycle cycle efficiency h ; maximum cycle efficiency increase in C from 1000 to 10,000 suns (at TR ¼ 1510 K). Simi- cycle maximum h solar to fuel energy conversion efficiency larly, Qreactornet and Qsolar are observed as increased by 50.665% solar to fuel DG Gibbs free energy change, kJ/mol and 49.17% with the decrease in the TR from 1510 to 1280 K. D The reason for this increase is the high energy requirement for GWS Gibbs free energy change for water splitting reaction, s_ s_ kJ/mol the heating of inert Ar. A decrease in the reactor and cooler1 is D realized until Ar dilution is no longer required. The achieve- GTR Gibbs free energy change for thermal reduction h ¼ h ¼ reaction, kJ/mol ment of maximum cycle 39.56% and solartofuel 47.74% is DH enthalpy change, kJ/mol observed at higher TR ¼ 1510 K. At similar TR, increase in the h h DHWS enthalpy change for water splitting reaction, kJ/mol cycle and solartofuel upto 51.77% and 62.43% is possible by h DHTR enthalpy change for thermal reduction reaction, kJ/ applying 50% heat recuperation. Furthermore, the cycle and h mol solartofuel of this cycle are higher as compared to other solar D $ thermochemical water splitting cycles. The results obtained S entropy change, J/mol K D $ provide a foundation for pursuing an experimental study for SWS entropy change for water splitting reaction, J/mol K e DS entropy change for thermal reduction reaction, J/ producing solar H2 via thermochemical IO IS water splitting TR $ cycle. mol K s StefaneBoltzmann constant, 5.670 10 8 (W/m2$K4) s_ cooler1 rate of entropy produced across cooler 1, kW/K s_ cooler2 rate of entropy produced across cooler 2, kW/K s_ Acknowledgments cooler3 rate of entropy produced across cooler 3, kW/K s_ FeOoxidizer rate of entropy produced across FeO oxidizer, kW/ The authors gratefully acknowledge the financial support K provided by the Qatar University internal grant QUUG-CENG- s_ reactor rate of entropy produced across the solar reactor, CHE-13/14-4. kW/K

Nomenclature references C solar flux concentration ratio, suns CSP concentrated solar power HHV higher heating value I normal beam solar insolation, W/m2 [1] Agrafiotis C, Pagkoura C, Lorentzou S, Kostoglou M, IOeIS iron oxideeiron sulfate Konstandopoulos AG. Hydrogen production in solar reactors. Catal Today 2007;127:265e77. MO metal oxide _ [2] Steinfeld A, Sanders S, Palumbo R. Design aspects of solar n molar flow rate, mol/s thermochemical engineering e a case study: two-step water- _ nAr molar flow rate of Ar, mol/s splitting cycle using the Fe3O4/FeO redox system. Sol Energy e QArheating energy required for heating of Ar, kW 1999;43:43 53. 1650 international journal of hydrogen energy 40 (2015) 1639e1650

[3] Kodama T, Nakamuro Y, Mizuno T. A two-step [19] Bhosale RR, Khadka RP, Shende RV, Puszynski JA. H2 thermochemical water spitting by iron-oxide on stabilized generation from two-step thermochemical water-splitting e zirconia. J Sol Energy Eng 2006;128:3 7. reaction using sol-gel derived SnxFeyOz. J Renew Sustain [4] Gokon N, Murayama H, Umeda J, Hatamachi T, Kodama T. Energy 2011;3. 063104-1e063104-16.

Monoclinic zirconia-supported Fe3O4 for the two-step water- [20] Varsano F, Padella F, Alvani C, Bellusci M, Barbera A. splitting thermochemical cycle at high thermal reduction Chemical aspects of the water-splitting thermochemical temperatures of 1400e1600 C. Int J Hydrogen Energy cycle based on sodium manganese ferrite. Int J Hydrogen 2009;34:1208e17. Energy 2012;37:11595e601. [5] Ishihara H, Kaneko H, Hasegawa N, Tamaura Y. Two-step [21] Bhosale RR, Shende RV, Puszynski JA. Thermochemical e e water-splitting at 1273 1623 K using yttria-stabilized water-splitting for H2 generation using sol gel derived Mn- zirconia-iron oxide solid solution via co-precipitation and ferite in a packed bed reactor. Int J Hydrogen Energy solid-state reaction. Energy 2008;33:1788e93. 2012;37:2924e34. [6] Miller J, Allendorf M, Diver R, Evans L, Seigel N, Stuecker J. [22] Scheffe JR, Steinfeld A. Thermodynamic analysis of cerium- Metal oxide composites and structures for ultra-high based oxides for solar thermochemical fuel production. temperature solar thermochemical cycles. J Mater Sci Energy Fuels 2012;26:1928e36. 2008;43:4714e28. [23] Abanades S, Flamant G. Thermochemical hydrogen [7] Steinfeld A. Solar hydrogen production via a two-step water- production from a two-step solar-driven water-splitting splitting thermochemical cycle based on Zn/ZnO redox cycle based on cerium oxides. Sol Energy reactions. Int J Hydrogen Energy 2002;37:611e9. 2006;80:1611e23.

[8] Abanades S. Thermogravimetric analysis of CO2 and H2O [24] Gal A, Abanades S. Dopant incorporation in ceria for reduction from solar nanosized Zn powder for enhanced water-splitting activity during solar thermochemical fuel production. Ind Eng Chem Res thermochemical hydrogen generation. J Phys Chem C 2012;51:741e50. 2012;116:13516e23. [9] Ernst F, Steinfeld A, Pratsinis S. Hydrolysis rate of submicron [25] Lin X, Zhang Y, Wang Z, Wang R, Zhou J, Cen K. Hydrogen e e Zn particles for solar H2 synthesis. Int J Hydrogen Energy production by HI decomposition over nickel ceria zirconia 2009;34:1166e75. catalysts via the sulfureiodine thermochemical water- [10] Vishnevetsky I, Epstein M. Production of hydrogen from solar splitting cycle. Energy Convers Manag 2014;84:664e70. zinc in steam atmosphere. Int J Hydrogen Energy [26] Murphy IV J, O'Connell J. Process simulations of HI 2007;32:2791e802. decomposition via reactive distillation in the sulphur-iodine [11] Abanades S, Charvin P, Lemont F, Flamant G. Novel two-step cycle for hydrogen manufacture. Int J Hydrogen Energy e SnO2/SnO water-splitting cycle for solar thermochemical 2012;37:4002 11. production of hydrogen. Int J Hydrogen Energy [27] Dehghani S, Sayyaadi H. Energy and exergetic evaluations of 2008;33:6021e30. Bunsen section of the sulfureiodine thermochemical [12] Charvin P, Abanades S, Lemont F, Flamant G. Experimental hydrogen production plant. Int J Hydrogen Energy e study of SnO2/SnO/Sn thermochemical systems for solar 2013;38:9074 84. production of hydrogen. AIChE J 2008;54:2759e67. [28] Lin S, Flaherty R. Design studies of the sulphur trioxide [13] Vishnevetsky I, Epstein M. Tin as a possible candidate for decomposition reactor for the sulphur cycle hydrogen solar thermochemical redox process for hydrogen production process. Int J Hydrogen Energy 1983;8:589e96. production. J Sol Energy Eng 2009;131. 021007-1e021007-8. [29] Hinkley J, O'Brein J, Fell C, Lindquist S. Prospects for solar [14] Scheffe JR, Li J, Weimer AW. A spinel ferrite/hercynite water- only operation of the hybrid sulphur cycle for hydrogen splitting redox cycle. Int J Hydrogen Energy 2010;35:3333e40. production. Int J Hydrogen Energy 2011;8:589e96.

[15] Bhosale RR, Shende RV, Puszynski JA. H2 generation from [30] Brittan R, Hildenbrand D. Catalytic decomposition of gaseous thermochemical water-splitting using solegel derived Ni- sulphur trioxide. J Phys Chem 1983;87:3713e7. ferrite. J Energy Power Eng 2010;4:27e38. [31] Normal JH, Mysels KJ, Sharp R, Williamson D. Studies of the [16] Neises M, Roeb M, Schmucker M, Sattler C, Pitz-Paal R. sulphureiodine thermochemical water-splitting cycle. Int J Kinetic investigations of the hydrogen production step of a Hydrogen Energy 1982;7:545e56. thermochemical cycle using mixed iron oxides coated on [32] Roine A. Outokumpu HSC chemistry for windows, version ceramic substrates. Int J Energy Res 2010;34:651e61. 7.1. Pori, Finland: Outokumpu Research Oy; 2013.

[17] Bhosale RR, Shende RV, Puszynski JA. H2 generation from [33] Glavez ME, Loutzenhiser PG, Hischier I, Steinfeld A. CO2 thermochemical water-splitting using sol-gel synthesized splitting via two-step solar thermochemical cycles with Zn/ e Zn/Sn/Mn-doped Ni-ferrite. Int Rev Chem Eng 2010;2:852 62. ZnO and FeO/Fe3O4 redox reactions: thermodynamic [18] Lorentzou S, Zygogianni A, Tousimi K, Agrafiotis C, analysis. Energy Fuels 2008;22:3544e50. Konstandopoulos AG. Advanced synthesis of nanostructured [34] Diver RB, Miller JE, Allendorf MD, Seigel NP, Hogan RE. Solar material for environmental applications. J Alloys Compd thermochemical water-splitting ferrite-cycle heat engines. J 2009;483:302e5. Sol Energy Eng 2008;130. 041001-1e041001-8.