Intrinsic Kinetic Modeling of Hydrogen Production by Photocatalytic Water Splitting Using Cadmium Zinc Sulfide Catalyst

International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015

Hyacinth Mae G. Tambago and Rizalinda L. de Leon

 1 H O H  O G  238 kJ/mol (1) Abstract—Photocatalytic water splitting is an alternative 2 22 2 method for hydrogen production. Cadmium zinc sulfide is a photocatalyst active under visible light that is effective for Simultaneous generation of hydrogen and oxygen on a hydrogen generation in the presence of sulfide and sulfite as single photocatalyst particle under visible light irradiation is sacrificial reagents. Design of photocatalytic reactors for such a difficult to achieve since the photocatalyst must be able to system requires a kinetic model of the photocatalytic reaction. satisfy band position requirements and its band gap must not In this study, the effect of catalyst loading, sulfide and sulfite be too large. In addition, fast electron-hole recombination concentration, and incident light intensity on the rate of and backward reaction of hydrogen and oxygen to water hydrogen production using a cadmium zinc sulfide catalyst suspended in an externally irradiated batch reactor was reduce the efficiency of overall water splitting. The investigated. The variation of the local volumetric rate of generation of a mixture of hydrogen and oxygen also calls for photon absorption (LVRPA) in the batch reactor was taken into a separation process that still has to be developed in order to account by modeling the reactor radiation field using the obtain pure hydrogen. two-flux theory of Kubelka and Munk. A kinetic model In order to enhance photocatalytic hydrogen generation, reflecting the influence of sacrificial reagent concentrations and sacrificial reagents such as alcohols and sulfide ions are the LVRPA was developed and fit to the experimental data. The added to the aqueous suspension of the photocatalyst. Also model parameters obtained are independent of irradiation form and reactor geometry and may thus be used for the design and called hole scavengers, these reducing agents enrich electrons scale-up of photocatalytic reactors. by being irreversibly oxidized instead of water by photo-generated holes. Recombination of electron and hole is Index Terms—Cadmium zinc sulfide, intrinsic kinetic minimized and thus hydrogen production is enhanced. modeling, photocatalytic hydrogen production, water splitting. However, when the sacrificial reagent becomes used up, hydrogen production is inhibited. This kind of photocatalytic hydrogen-generating system thus calls for a continual I. INTRODUCTION addition of hole scavengers and is meaningful if industrial The environmental problems caused by the current polluting by-products, biomass, and compounds abundant in consumption of fossil fuels as a major energy source, which nature are used as reducing reagents [2]. are particularly being associated with global warming, and its Metal oxide photocatalysts such as TiO2, SrTiO3, and other eventual depletion, have led the move towards clean and perovskite titanium and tantalum compounds are commonly renewable energy sources. Hydrogen is a promising energy used for water splitting due to their stability during the carrier as it undergoes clean combustion. At present, however, photoreaction in water. These photocatalysts, however, have majority of the world’s hydrogen is generated from steam too wide band gaps that make them responsive only to UV reforming of natural gas, a fossil fuel, which also results in light, which comprises only about 6% of solar radiation. carbon dioxide emissions [1]. Modifications such as doping with cations or anions and Alternative methods of hydrogen production include water loading noble metal co-catalysts are conducted to make these splitting, which is the dissociation of water into hydrogen and catalysts active under visible light and more effective for oxygen gas, as shown in (1), using light in the presence of a hydrogen generation. photocatalyst. Since the discovery of water splitting on a Cadmium sulfide (CdS) is a well-known metal sulfide catalyst that is active under visible light and that satisfies the photoelectrochemical cell comprised of a TiO2 photo-anode and a Pt photo-cathode under ultraviolet (UV) irradiation by band position requirements for water splitting. However, this Honda and Fujishima in 1972, a lot of research on catalyst undergoes photocorrosion in which the sulfur anion of the catalyst becomes oxidized by the photogenerated holes photocatalytic water splitting has been conducted. 2- to S or S2 as water is reduced to hydrogen. Sulfur may 2- deposit on the catalyst surface and S2 competes with the

Manuscript received August 10, 2014; revised November 7, 2014. This catalyst for absorption of photons in the visible light region. 2- 2- work was supported in part by the University of the Philippines Engineering In order to prevent these, sulfide (S ) and sulfite (SO3 ) ions Research and Development Foundation, Inc. (UPERDFI) Research and are added to the catalyst suspension as sacrificial reagents. Development Fellowship and Incentive Award and the Engineering 2- The sulfide species becomes oxidized to S or S2 instead of Research and Development for Technology (ERDT) under the Republic of 2- the Philippines Department of Science and Technology (DOST). the sulfide ion in the catalyst and sulfite reacts with S and S2 2- H. M. G. Tambago and R. L. de Leon are with the Department of to form thiosulfate (S2O3 ), which is a colorless species. Chemical Engineering, University of the Philippines, Diliman, 1108 Quezon Modifications on the CdS catalyst are performed to improve City, Philippines (e-mail: [email protected], [email protected]). its efficiency and stability such as loading Pt and other noble

DOI: 10.7763/IJCEA.2015.V6.485 220 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015 metals as co-catalyst or incorporating ZnS, another metal hydrogen production was found to be highest among three sulfide catalyst that is used for hydrogen production but only values of x in CdxZn1-xS considered, was synthesized via the active under UV light, into CdS to form a solid solution of co-precipitation method [6]. Cadmium acetate dihydrate, CdxZn1-xS. Synthesized CdxZn1-xS without loaded noble Cd(CH3COO)2·2H2O and zinc acetate dihydrate, metals has been found to be effective for hydrogen Zn(CH3COO)2·2H2O were dissolved in 100 mL of deionized production [3], which is beneficial for cost-effective water. The solution was constantly stirred and added drop photocatalyst production without addition of expensive metal wise with 0.1 M sodium sulfide, Na2S·H2O. The amounts of co-catalysts [4]. the reagents were such that the Cd:Zn:S atomic ratio is A practical application on the use of sulfide and sulfite as 0.4:0.6:1. The formed precipitate was separated from the sacrificial reagents in metal sulfide catalytic systems is the solution by vacuum filtration, washed with deionized water, utilization of waste streams containing these species. and dried at 348 K for eight (8) hours. Hydrogenation and flue-gas desulfurization processes generate hydrogen sulfide and sulfur dioxide as by-products, B. Photocatalytic Reactor which exist in sulfide and sulfite forms, respectively, in Photocatalytic reactions were carried out in a Pyrex alkaline solutions. Use of these waste streams for hydrogen cylindrical vessel with an O-ring joint seal. The catalyst production is interesting since it simultaneously suspension is held in the portion of the vessel below the joint accomplishes both waste treatment and renewable energy having an inner diameter of 5 cm and height of 5 cm. The generation [2]. For the design of reactors for this joints make up the headspace with a volume of 126 cm3. On photocatalytic system, a kinetic model that shows the the top of the reactor is a sample port with a influence of the system parameters such as the sacrificial polytetrafluoroethylene (PTFE) septum for extraction of gas reagent concentration and light intensity is required. samples from the headspace. This study attempts to address the need of a complete The reactor is irradiated through one side with a 400-W photoreactor design for hydrogen production using a tubular halogen lamp (Osram Haloline Eco SST). Between CdxZn1-xS photocatalyst with sulfide and sulfite for an the lamp and the reactor is a UV cut-off filter. The surface of intrinsic kinetic model that will be coupled with mass transfer, this side of the vessel facing the lamp was made to have a hydrodynamics, and radiation transfer models. The model ground glass texture to realize the assumption of a diffuse should reflect the influence of sulfide and sulfite incoming radiation used in modeling the radiation field inside concentrations and photon absorption in the form of the local the reactor. The irradiance of the lamp was measured using a volumetric rate of photon absorption (LVRPA). The LVRPA DayStar DS 05 solar meter and the lamp spectral irradiance corresponds to the amount of photons incident on a region of was determined by an Ocean Optics HR2000+ the reactor that is absorbed by the photocatalyst. Due to high-resolution spectrometer. Temperature in the reactor attenuation of light caused by the absorbing species such as was maintained at 25±1 °C by a Peltier cell cooling system. the photocatalyst in suspension, the radiation field and the LVRPA is not uniform throughout the reactor. The LVRPA C. Photocatalytic Reaction Runs and its variation in the reactor are dependent on several The photocatalyst suspension is composed of the system properties such as the intensity and spectral irradiance Cd0.4Zn0.6S photocatalyst particles in an aqueous solution of of the light source, the geometrical configuration of the sodium sulfide (Na2S) and sodium sulfite (Na2SO3) prepared reactor, the amount of the catalyst, and the absorption and by dissolving prescribed amounts of sodium sulfide hydrate scattering properties of the catalyst and other species in the flakes and anhydrous Na2SO3 in 100 mL of deionized water. system. In order for the kinetic model to be used for Hydroxide concentration in the catalyst suspension was photoreactor design and scale-up, the model parameters, determined from pH measurements using a pH meter which are derived from kinetic studies carried out in an (CyberScan pH 110). The photocatalytic reaction runs were irradiated batch reactor, must be independent of the conducted at varying catalyst loading, sulfide concentration, experimental configuration; hence, the radiation field in the sulfite concentration, and incident light intensity, as shown in photoreactor used in the kinetic studies must be modeled in Table I. Variation of the incident light intensity was done by order to consider the variation of the LVRPA in the varying the distance of the lamp from the reactor. The photoreactor. This was performed in [5] in order to determine experimental error was estimated to be the standard deviation the intrinsic kinetic model of photocatalytic oxidation of of three replicates of Run 3. cyanide in water using a TiO2 catalyst under UV irradiation. Prior to each run, the reactor suspension was purged with This study aims to develop a model for the intrinsic nitrogen gas for 15 minutes to remove oxygen. Initial rates of kinetics of photocatalytic hydrogen production via water photocatalytic hydrogen production were determined by splitting on a synthesized Cd0.4Zn0.6S photocatalyst in the taking hourly measurements of the amount of hydrogen in the presence of sulfide and sulfite ions as sacrificial reagents. reactor headspace for four (4) hours. A 1-mL gas sample is Radiation absorption will be considered for estimation of extracted from the headspace through the PTFE septum using parameters that will be independent of experimental a 2.5-mL gas-tight syringe (Hamilton HD-type) and injected conditions. onto a gas chromatograph (GC) with a thermal conductivity

detector (Shimadzu GC 2014) using a molecular sieve column (Supelco Mol Sieve 5A) and nitrogen as the carrier II. EXPERIMENTAL gas. A calibration curve relating the chromatogram peak area A. Photocatalyst units to the amount of hydrogen (in micromoles) was created

The Cd0.4Zn0.6S photocatalyst used in this study, for which by injecting different volumes of a 1.2% H2-in-N2 gas standard onto the GC.

221 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015

TABLE I: PHOTOCATALYTIC REACTION RUNS photocatalyst. Sulfite ions are needed to convert the sulfide Mass of Incident light oxidation products to thiosulfate, which is colorless. catalyst, g [Na S], [Na SO ], [OH-], Run 2 2 3 intensity, However, as can be seen in Fig. 3, increasing sulfite ions in 100 mL M M M mW/cm2 solution decreases the hydrogen generation rate due to competitive 1 0.025 0.1 0.1 0.92 54.95 adsorption of sulfite on the catalyst sites with sulfide [7], [8]. 2 0.075 0.1 0.1 0.92 54.95 As the incident light intensity is increased, the rate of 3 0.1 0.1 0.1 0.92 54.95 hydrogen generation becomes higher as shown in Fig. 4 due 4 0.2 0.1 0.1 0.92 54.95 5 0.1 0.005 0.1 0.07 54.95 to the photogeneration of more electrons and holes. 6 0.1 0.01 0.1 0.14 54.95 7 0.1 0.025 0.1 0.32 54.95 3.5 8 0.1 0.05 0.1 0.61 54.95 9 0.1 0.3 0.1 1.28 54.95 3 mol/h)

10 0.1 0.1 0.01 0.65 54.95  2.5 11 0.1 0.1 0.025 0.66 54.95 12 0.1 0.1 0.05 0.73 54.95 2 13 0.1 0.1 0.3 0.99 54.95 1.5

14 0.1 0.1 0.1 0.92 69.00 ( production 2 15 0.1 0.1 0.1 0.92 42.91 1 16 0.1 0.1 0.1 0.92 34.84 0.5

H of Rate 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 3 Na S concentration (M) 2

2.5 Fig. 2. Initial rates of hydrogen production at various Na2S concentrations 2 mol/h)

 (0.1 g catalyst in 100 mL solution with 0.1 M Na2SO3, 54.95 mW/cm incident irradiance). 2 7

production ( 1.5 2

mol/h) 6  1

RateH of 5 0.5 0 0.05 0.1 0.15 0.2 0.25 4 production ( production Catalyst mass (g) in 100 mL solution 2

Fig. 1. Initial rates of hydrogen production at various amounts of catalyst 3 2

(100 mL solution with 0.1 M Na2S, 0.1 M Na2SO3, 54.95 mW/cm incident H of Rate irradiance). 2 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Na SO concentration (M) 2 3

III. EXPERIMENTAL RATES OF HYDROGEN PRODUCTION Fig. 3. Initial rates of hydrogen production at various Na2SO3 concentrations 2 (0.1 g catalyst in 100 mL solution with 0.1 M Na2S, 54.95 mW/cm incident Figs. 1 to 4 show the experimental rates of hydrogen irradiance). production at varying operating parameters. It can be 4

observed from Fig. 1 that increasing the catalyst loading mol/h)  results in higher rates of hydrogen production due to 3 increased rates of photon absorption; increasing the concentration of catalyst particles further, however, does not 2 production ( improve the hydrogen generation rate as much due to the 2 1 shadowing effect of the catalyst particles. The high photon absorption rate by catalyst particles near the irradiated 0

RateH of 30 40 50 60 70 surface of the reactor prevents penetration of light further into 2 Incident irradiance (mW/cm ) the reactor, lowering the total photon absorption that would Fig. 4. Initial rates of hydrogen production at various incident irradiances have supposedly increased due to the higher concentration of (0.1 g catalyst in 100 mL solution with 0.1 M Na2S, 0.1 M Na2SO3). light-absorbing particles. As shown in Fig. 2, increasing sulfide concentrations promotes hydrogen evolution due to higher rates of hole consumption, preventing the IV. RADIATION FIELD MODELING AND DETERMINATION OF recombination of electrons and holes. Too high sulfide THE LVRPA concentrations, however, decreases the rate of hydrogen The photocatalytic reactor is not homogeneously irradiated production due to the reduction of adsorption sites needed for and thus the radiation distribution must be considered in hydrogen generation and the absorption of light in the visible determining intrinsic kinetic parameters independent of the region by trace polysulfides and sulfide oxidation products, irradiation form, lamp-reactor configuration, and competing with the photocatalyst. This results in the bell photoreactor geometry. The intensity profile is determined by shape of the plot of the gas production rate vs. sulfide solving the radiative transfer equation (RTE) applied to the concentration that is typical for reactions following this type photoreactor. The RTE at steady-state condition and with the of mechanism, as was also found in [7], [8] for a CdS assumption of negligible radiation emission is given by the

222 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015 following energy balance along a path 푑푠 (e.g., in cm) [9]: dJ/. dx  K   S  J   S  I  (4)

dI, () x The parameters 퐾휆 and 푆휆 are defined in terms of the  I ,,()() x   I  x ds absorption and scattering coefficients of the layer as follows:  (2) p '()'  I,' x d  KS2   2  . (5) 4 '4 The wavelength-dependent specific absorption and 2 where 퐼휆,Ω (e.g., in mW/cm sr) is the wavelength- and scattering coefficients of Cd0.4Zn0.6S suspensions in water direction-dependent intensity, 푥 (e.g., in cm) is the position were determined from diffuse transmittance vector in a three-dimensional space, 훺 is the unit vector in the spectrophotometric measurements by solving the RTE direction of radiation propagation, 훺 (sr) is the solid angle of applied to the spectrophotometric cells. These optical radiation propagation about the direction 훺, 휅 (e.g., in cm-1) properties, along with the absorption coefficient of the 휆 sacrificial reagent solution and the lamp spectral irradiance, is the wavelength-dependent absorption coefficient, 휎 (e.g., 휆 were used to determine the intensity and the LVRPA in cm-1) is the wavelength-dependent scattering coefficient, distributions in the photoreactor. and 푝 (훺′ ∙ 훺) is the scattering phase function, which 휆 The modeling procedure for determining the irradiance describes the angular distribution of scattered radiation. The and LVRPA profile in the reactor is based on a previous work left-hand term of (2) represents the change of energy along that applied the Kubelka and Munk model to a cylindrical the direction 훺. The first term of the right side of the equation photoreactor externally irradiated by a radially emitting light represents the energy loss due to absorption of light along the source [11]. Their procedure, in which the backward flux 퐽휆 direction 훺 , the second is the energy loss due to was neglected, was modified in this work in order to take into scattering-out of light from the direction 훺, and the third term account both directional fluxes. The photoreactor was is the gain of energy along the direction 훺 due to the discretized into segments with length Δ푟 and angular width scattering-in of light from all directions 훺′ in space. Δ휃, for which the fluxes 퐼휆 and 퐽휆 entering and leaving the Application of the complete form of the RTE shown in (2) layer were computed. A non-reflecting background was results in a set of simultaneous integro-differential equations, assumed, in which 퐽휆 at the reactor wall far from the lamp is the solution for which requires a rigorous numerical zero, and the measured incident irradiance at the irradiated procedure. One simplification to the RTE is given by the wall of the reactor was the other boundary condition used. In two-flux theory of Kubelka and Munk, which originated from addition, the assumption of diffuse incoming radiation used the Schuster equation for isotropic scattering. The Schuster in the radiation field model was realized by the ground glass equation considers the radiation field to be consisting of two texture of the reactor wall facing the lamp. -1 -2 -1 -1 -2 -1 oppositely directed radiation fluxes, 퐼휆 (e.g., in J s m nm ) The total incident irradiance 퐺휆 (e.g., in J s m nm ) at in the positive or incident radiation propagation direction and one wavelength 휆 on a particular segment 𝑖 is the sum of -1 -2 -1 퐽휆 (e.g., in J s m nm ), which arises from scattering, in the 퐼휆 and 퐽휆 entering that segment. negative propagation direction. The propagation direction is GIJ (6) perpendicular to an irradiated layer of thickness 푑 (defined i  i  i along the x-axis), as shown in Fig. 5. The width and length The wavelength-dependent local volumetric rate of energy (along the 푦푧-plane) of this layer are assumed to be larger absorption (LVREA, e.g., in J s-1 m-3 nm-1) by the than 푑 such that edge effects can be ignored. photocatalyst particles at each segment is computed as:

LVREA K* C G (7) ii cat 

-3 d where 퐶푐푎푡 (e.g., in g cm ) is the catalyst concentration and

∗ 2 -1 퐾휆 (e.g., in cm g ) is the 퐾휆 parameter per unit catalyst

concentration. -1 -3 -1

The LVRPA (e.g., in photons s m nm ) at each wavelength is computed from the LVREA and the energy of

a photon, which varies with wavelength, as follows:

Fig. 5. Irradiated layer modeled by the Schuster equation for isotropic LVRPA  LVREA / hc (8) scattering [10]. ii 

−34 The theory assumes isotropic distribution of scattering, where ℎ = 6.626 × 10 J s is the Planck’s constant, 8 −1 diffuse incident irradiation, random distribution of the 푐 = 2.998 × 10 m s is the speed of light, and the particles in the layer, and that the particles are much smaller bracketed term [휆 (ℎ푐)⁄ ] is the energy per photon with than the layer thickness. Under these assumptions, radiation wavelength 휆. Since the light source irradiating the balance within the layer results in the following simultaneous photocatalytic system is polychromatic, the total incident irradiance 퐺 (e.g., in J s-1 m-2) and LVRPA (e.g., in photons differential equations for 퐼휆 and 퐽휆 [10]: -1 -3 s m ) at each segment are obtained by integrating 퐺휆 and 퐿푉푅푃퐴휆, respectively, over the considered wavelength range dI/ dx   K   S  I   S  J  (3) 휆1 − 휆2:

223 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015

2 postulated for cadmium sulfide catalysts in [8], [12]. The G G d (9) i   i reactions are assumed to take place among adsorbed species 1 on the photocatalyst, following the Langmuir-Hinshelwood

2 mechanism. For brevity, the CdxZn1-xS photocatalyst site is LVRPA LVRPA d. (10) i   i represented as CdZnS. The symbol CdZnS is used to denote 1 the site on the negatively charged 푆 atom while SCdZn for The wavelength range considered was 360-500 nm, in the site on the positively charged Cd and Zn atoms. which the absorption band of the catalyst lies [6]. These The expression for the rate of hydrogen production is calculations were repeated for all segments in the reactor to derived from the mechanism with the following assumptions: determine the intensity and LVRPA at each point. (i) reactions take place among species adsorbed on the catalyst surface, (ii) reactions are elementary and irreversible, (iii) concentrations of electrons, holes, and radicals are at V. INTRINSIC KINETIC MODEL OF HYDROGEN PRODUCTION steady-state, (iv) species adsorbed on the photocatalyst are in USING CADMIUM ZINC SULFIDE CATALYST WITH SULFIDE equilibrium with those in the bulk solution, (v) sulfide, sulfite, AND SULFITE and hydroxide ions compete for the same adsorption site, (vi) concentration of water on the catalyst surface is constant, and TABLE II: REACTION MECHANISM FOR HYDROGEN PRODUCTION FROM (vii) the rate of electron-hole generation is proportional to WATER WITH SULFIDE AND SULFITE IONS [8], [12] 훾푎,푣, the LVRPA or the rate of photon absorption per unit Absorption of photon and electron-hole pair generation volume of catalyst suspension, by the average quantum CdZnS + hv → CdZnS + e− + h+ efficiency 휙 [5]. A modification to the rate expression was Recombination of electron and hole introduced such that the rate retains its dependence on species concentrations when it is reduced to a form having linear e− + h+ → ϕ dependence with the LVRPA at low intensities. The resulting Adsorption of reactants kinetic rate expression is as follows: CdZnS + H O → CdZnS − H O v -- 2 2 rk {K-- [HS ] / (1+ K [HS ] H2 HS HS 2− − − (11) S + H2O → HS + OH 2- - 2av , + K2- [SO ]+ K - [OH ]) } SO3 OH SCdZn + HS− → SCdZn − HS− 3 푣 2− 2− where 푟퐻2 is the rate of hydrogen production per unit volume SCdZn + SO3 → SCdZn − SO3 of the suspension, 퐾 −, 퐾 2−, and 퐾 − are the adsorption 퐻푆 푆푂3 푂퐻 Reactions on the catalyst surface − 2− − equilibrium constants for 퐻푆 , 푆푂3 , and 푂퐻 , respectively, Reduction of water to hydrogen and the parameter 푘 is: − − CdZnS − H2O + SCdZn + e → CdZnS − H + SCdZn − OH 2 k kred[H2 O] k ox [SCdZn] tot  (12) 2 CdZnS − H • → 2 CdZnS + H2

Oxidation of sulfide where 푘푟푒푑 is the rate constant for reduction of water to hydrogen, 푘표푥 is the rate constant for oxidation of sulfide, SCdZn − HS− + h+ → SCdZn − HS • [H2O] is the concentration of water on the catalyst surface, − − SCdZn − HS • + SCdZn − OH → CdZnS + SCdZn − S • + H2O and [SCdZn]푡표푡 represents the total amount of SCdZn sites.

− − SCdZn − HS • + SCdZn − S • → SCdZn − HS2 + CdZnS B. Estimation of Kinetic Parameters Using the experimental parameters and LVRPA profiles, Desorption of products the model is fit to the experimental rates to estimate the − − SCdZn − HS2 + CdZnS + HS2 model parameters. By assuming negligible mass transfer

Reaction of disulfide with sulfite in the liquid phase limitations since the catalyst particles are kept in suspension by magnetic stirring of the batch reactor contents and that the − 2− 2− − 퐻푆2 + 푆푂3 → 푆2푂3 + 퐻푆 amount of hydrogen dissolved in the suspension is negligible, Adsorption of other species in the system the experimentally measured rates of hydrogen production 푆퐶푑푍푛 + 푂퐻− → 푆퐶푑푍푛 − 푂퐻− are also taken to be the experimentally determined intrinsic rates of production of this gas. 2− 2− 푆퐶푑푍푛 + 푆푂3 → 푆퐶푑푍푛 − 푆푂3 The experimentally determined rates of hydrogen production presented in Fig. 1 to Fig. 4 are for the entire A. Kinetic Model reactor volume. In order to relate these rates to (11), the The photocatalytic hydrogen production on semiconductor 푣 expression for 푟H2 in (11) must be integrated over the reactor particles involves the absorption of photons with energy volume 푉 to get the total rate of hydrogen production 푟 in higher than the catalyst band gap, generating electron-hole H2 moles of hydrogen per unit time. pairs, and the reduction by electrons and oxidation by holes of water and sacrificial species, respectively, on the surface -- of the photocatalyst. A competing step is the recombination rk {K-- [HS ] / (1+ K [HS ] H2  HS HS of electrons and holes. Table II shows the reaction V (13) mechanism for hydrogen production on a cadmium zinc 2- - 2av ,  K2- [SO3 ]+ K - [OH ]) } dV sulfide catalyst, which is assumed to be the same as that SO3 OH

224 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015

Since the suspension was constantly kept under stirring, 4 the concentration of sulfide, sulfite, and hydroxide ions is /h) assumed to be uniform throughout the reactor. The LVRPA, 2 however, varies with reactor position and must remain in the 3 mol H mol integral term. Equation (13) may thus be written as:  2 -- rk {K-- [HS ] / (1+ K [HS ] 0.025 g catalyst H2 HS HS 1 0.075 g catalyst 2- - 2av , (14) 0.100 g catalyst  K2- [SO3 ]+ K - [OH ]) }  dV ( rate Predicted SO3 OH  0.200 g catalyst V 0 0 1 2 3 4 Experimental rate (mol H /h) 2 The reactor cross-sectional area was divided into several Fig. 6. Plot of predicted rate vs. experimental rate of hydrogen production at segments with length Δ푟 and angular width Δ휃 for the different catalyst amounts (in 100 mL solution with 0.1 M Na2S, 0.1 M 2 LVRPA profiling. Each of these segment areas is multiplied Na2SO3, 54.95 mW/cm incident irradiance). by the reactor height to obtain the volume Δ푉 of each of the 4 differential reactor element for which the LVRPA has been /h) determined. The integration term in (14) is thus carried out 2 numerically as follows. 3 mol H mol

 0.005 M Na S 2 2 0.010 M Na S -- 2 rk {K-- [HS ] / (1+ K [HS ] H2 HS HS 0.025 M Na S (15) 2 2- - 2av , 0.050 M Na S K2- [SO3 ]+ K - [OH ]) }  V 1 2 SO3 OH  0.100 M Na S 2

Predicted rate ( rate Predicted 0.300 M Na S 2 0 The kinetic parameter 푘 and the adsorption equilibrium 0 1 2 3 4 Experimental rate (mol H /h) constants 퐾푖 of species 𝑖 were estimated by fitting the rate 2 expression given in (15) to the experimental rates (micromol Fig. 7. Plot of predicted rate vs. experimental rate of hydrogen production at -1 − different Na2S concentrations (0.1 g catalyst in 100 mL solution with 0.1 M H h ) at the corresponding concentrations (taken to be 2 2 HS Na2SO3, 54.95 mW/cm incident irradiance). 2− equal to Na2S concentrations, M), SO3 concentrations (M), − and OH concentrations (M) given in Table I, and LVRPA 7 (micromol photons h-1 cm-3) profiles. A /h) 6 Marquardt-Levenberg non-linear regression algorithm was 2 5 used. The resulting values of the parameters are shown in mol H mol

 4 Table III. These parameters are for the temperature of 25 °C, 0.010 M Na SO 2 3 the temperature at which the experimental photocatalytic 3 0.025 M Na SO 2 3 reaction rates were obtained. 0.050 M Na SO 2 2 3 0.100 M Na SO TABLE III: ESTIMATED MODEL PARAMETERS FOR (11) AT 25 °C 1 2 3 Predicted rate ( rate Predicted 0.300 M Na SO 푘 118.27 2 3 0 퐾HS−, L/mol 9,208.58 0 1 2 3 4 5 6 7 퐾SO2−, L/mol 7,618.42 Experimental rate (mol H /h) 3 2 퐾OH−, L/mol 621.03 Fig. 8. Plot of predicted rate vs. experimental rate of hydrogen production at

different Na2SO3 concentrations (0.1 g catalyst in 100 mL solution with 0.1 Coefficient of determination, 푟2 = 0.8873 2 M Na2S, 54.95 mW/cm incident irradiance). Correlation coefficient, 푟 = 0.9401

3.5

The coefficient of determination and correlation /h)

2 3 coefficient values indicate good fit of the model to the 2.5 experimental data. Fig. 6 to Fig. 9 show the parity plots of H mol  2 predicted rates using (15) against the experimental rates for 69.00 mW/cm2 each operating parameter variation. Most of the points 1.5 54.95 mW/cm2 pertaining to the predicted rate scatter along the predicted rate 1 42.91 mW/cm2 = experimental rate line. The relatively large deviations of 0.5 2

Predicted rate ( rate Predicted 34.84 mW/cm some points are attributed to radiation field model limitations 0 0 0.5 1 1.5 2 2.5 3 3.5 and to reaction mechanisms that were not taken into account Experimental rate (mol H /h) 2 in the kinetic modeling. Fig. 9. Plot of predicted rate vs. experimental rate of hydrogen production at Error is relatively large at 0.2 g/100 mL catalyst loading, as different incident irradiances (0.1 g catalyst in 100 mL solution with 0.1 M shown in Fig. 6. This may be attributed to the limitations of Na2S, 0.1 M Na2SO3). the radiation field model used due to its simplifications, At varied sulfite concentrations shown in Fig. 8, error is particularly in the adoption of an isotropic phase function, relatively large at 0.300 M. The higher rate of hydrogen such that the deviation of the actual distribution of scattering production than that predicted may be attributed to the from the assumed form may have become significant at this participation of sulfite in the consumption of holes when high catalyst concentration.

225 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015 present in high amounts in the solution as indicated in the conditions such as the irradiation form, reactor geometry, and following reaction, which was not included in the kinetic lamp-reactor configuration were defined and taken into modeling. account in the determination of the LVRPA, the estimated

2 - + 2- model parameters may be used to provide information for the SO3 2OH + 2h SO 4 + H 2 O (16) photoreaction kinetics component in photocatalytic reactor In addition to the consumption of holes by sulfide ions, design and scale-up using the Cd0.4Zn0.6S catalyst for the oxidation of sulfite ions by the photogenerated holes further photocatalytic production of hydrogen. prevents electron-hole recombination, increasing the rate of Validation of the computed absorption and scattering hydrogen production. coefficients of the cadmium zinc sulfide particles in water and the developed intrinsic kinetic model to other C. Limitations of the Model photoreactor configurations is recommended. The optical The developed kinetic model considers only the species properties and the model parameters may be further initially present in the system and does not include the improved by solving the complete form of the RTE applied to influence of products such as thiosulfates and disulfides on the irradiated photocatalyst suspensions. The model may also the rate as they are formed. Only the photo-oxidation of be extended to take into account the influence of the products sulfide is considered in the mechanism; the participation of such as thiosulfate and the participation of sulfite in the sulfite in reactions with photogenerated holes at high sulfite oxidation reactions at high sulfite concentrations. concentrations is not taken into account. Since the model parameters were estimated using the ACKNOWLEDGMENT LVRPA values derived from the model of the radiation field The authors thank the Department of Chemical in the reactor, the parameters may be further refined by Engineering Laboratory, the Department of Mechanical applying more rigorous radiation modeling procedures Engineering Machine Shop, the College of Engineering employing an anistropic phase function. The resulting model Building Maintenance Office, and the Photonics Research in this work is only valid for catalyst concentrations less than Laboratory of the National Institute of Physics of the the optimum, which could be attributed to the adoption of an University of the Philippines Diliman for their assistance in isotropic phase function in the determination of the LVRPA. the experimental parts of this work. The model is limited to the linear-dependence of the rate of hydrogen production on the LVRPA. This linear-dependence REFERENCES regime has also been observed in other water splitting [1] M. R. Pai, A. M. Banerjee, A. K. Tripathi, and S. R. Bharadwaj, systems irradiated at intensities in the same order of “Fundamentals and Applications of the Photocatalytic Water Splitting magnitude as solar radiation in the visible light region. This Reaction,” in Functional Materials: Preparation, Processing and Applications, S. R. Banerjee and A. K. Tyagi, Eds. London: Elsevier, model may not be able to predict the rate of hydrogen 2012, pp. 579-581. production at high intensities in the UV region, at which [2] X. Chen, S. Shen, L. Guo, and S. S. Mao, “Semiconductor-based studies have shown the rate of hydrogen production on Photocatalytic Hydrogen Generation,” Chemical Reviews, vol. 110, no. 11, pp. 6503-6570, May 2010. UV-active photocatalysts to have a sub-linear dependence on [3] C. Xing, Y. Zhang, W. Yan, and L. Guo, “Band structure-controlled the LVRPA. solid solution of Cd1−xZnxS photocatalyst for hydrogen production by water splitting,” International Journal of Hydrogen Energy, vol. 31, pp. 2018-2024, March 2006. [4] D. Jing, et al., “Efficient solar hydrogen production by photocatalytic VI. CONCLUSION AND RECOMMENDATIONS water splitting: From fundamental study to pilot demonstration,” International Journal of Hydrogen Energy, vol. 35, pp. 7087-7097, The rate of hydrogen production on a Cd0.4Zn0.6S catalyst January 2010. was investigated at varying catalyst amounts, Na2S [5] J. Marugán, R. van Grieken, A. E. Cassano, and O. M. Alfano, “Intrinsic Kinetic Modeling with Explicit Radiation Absorption Effects concentrations, Na2SO3 concentrations, and incident light of the Photocatalytic Oxidation of Cyanide with TiO2 and intensities. The dependence of the hydrogen generation rate Silica-Supported TiO2 Suspensions,” Applied Catalysis B: on these parameters was explained by their effect on the rate Environmental, vol. 85, pp. 48-60, 2008. of electron-hole generation, suppression of electron-hole [6] E. del Rosario, L. Doon, P. N. Perez, R. de Leon, and H. Ramos, “Hydrogen Production by Visible Light-driven Photocatalytic Water recombination, irradiance profile, and competitive adsorption Splitting using Plasma-enhanced CdZn1-xS,” unpublished manuscript, on catalyst sites. The intensity and LVRPA profiles in the University of the Philippines Diliman, PH. photoreactor were determined using an improved procedure [7] R. Priya, “Solar Photocatalytic Generation of Hydrogen From Hydrogen Sulfide in an Alkaline Solution using CdS-ZnS/TiO2 that applies the theory of Kubelka and Munk. Nanocomposites.” Ph.D. dissertation, Centre for Environmental An intrinsic kinetic model reflecting the influence of the Studies, Anna University, Chenai, 2011. sulfide, sulfite, and hydroxide (from the hydrolysis of the [8] J. Sabate, S. Cervera-March, R. Simarro, and J. Gimenez, “Photocatalytic production of hydrogen from sulfide and sulfite waste sacrificial reagents) concentrations and the LVRPA was streams: A kinetic model for reactions occurring in illuminated developed. The model shows Langmuir-Hinshelwood suspensions of CdS,” Chemical Engineering Science, vol. 45, pp. 3089-3096, January 1990. characteristics in the concentration-dependent term and linear [9] M. N. Ozisik, Radiative Transfer and Interactions with Conduction dependence of the rate on the LVRPA valid at low intensities. and Convection, New York: Wiley, 1973. The model, with four parameters - a kinetic parameter and the [10] G. Kortum, Reflectance Spectroscopy: Principles, Methods, Applications, New York: Springer, 1969, pp. 103-127. adsorption equilibrium constants of sulfide, sulfite, and [11] G. Palmisano, V. Loddo, and V. Augugliaro. (June 2013). hydroxide ions - estimated from experimental data, shows Two-Dimensional Modeling of an Externally Irradiated Slurry good fit to the experimental data. Since the experimental Photoreactor. International Journal of Chemical Reactor Engineering. [Online]. 11(2). pp. 675-685. Available:

226 International Journal of Chemical Engineering and Applications, Vol. 6, No. 4, August 2015

http://www.degruyter.com/dg/viewarticle/j$002fijcre.2013.11.issue-2 Engineers – Metro Manila Academe Chapter. $002fijcre-2012-0049$002fijcre-2012-0049.xml [12] S. Y. Ryu, J. Choi, W. Balcerski, T. K. Lee, and M. R. Hoffmann. “Photocatalytic Production of H2 on Nanocomposite Catalysts,” Industrial & Engineering Chemistry Research, vol. 46, no. 23, pp, Rizalinda L. de Leon earned her bachelor’s degree 7467-7488, June 2007. in chemical engineering, her master’s degree in energy engineering, and her PhD degree in chemical engineering at the University of the Hyacinth Mae G. Tambago earned her bachelor’s Philippines, Diliman, Quezon City. degree and master’s degree in chemical engineering She is currently an assistant professor of from the University of the Philippines, Diliman, chemical engineering at the University of the Quezon City, Philippines. Philippines Diliman. Her research interests include She has been an instructor of the Department of photocatalytic hydrogen production, nanofluids, Chemical Engineering at the University of the and bioethanol production. Philippines Diliman since 2010. Her research Dr. de Leon is a member of the Philippine Institute of Chemical interests are photocatalytic hydrogen production and Engineers. photocatalytic reaction engineering. Ms. Tambago is a member of the Philippine Institute of Chemical

227