1688
Ind. Eng. Chem. Res. 2003, 42, 1688-1697
Zinc Oxide Sorbents for the Removal of Hydrogen Sulfide from Syngas Ilaria Rosso,* Camilla Galletti, Massimo Bizzi, Guido Saracco, and Vito Specchia Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, Corso Duca degli Abruzzi, 24-10129 Turin, Italy
Several pure zinc oxide materials, prepared with different methods (combustion synthesis and a modified version of the citrates method) and calcined at different temperatures, were investigated as low-temperature desulfurizing sorbents from gaseous streams (syngas, in particular). Comparative tests with a commercial sorbent were also carried out. The sulfidation performance was investigated in a fixed-bed reactor in terms of breakthrough curves at 250 °C. Fresh and sulfided samples were characterized by X-ray diffraction, scanning electron microscopy-energy dispersion spectroscopy, BET, pore volume, pore size, and pore size distribution analyses. The ZnO sorbent prepared by the citrated method calcined at 400 °C showed the most durable effectiveness in reducing sulfur from 100 to less than 1 ppm: its breakthrough time is about 9 h measured at a space velocity of about 105 h-1. Its regenerability was evaluated by subsequent sulfidation-thermal regeneration cycles. A numerical model was also developed and validated on the experimental data of the ZnO sorbent calcined at 400 °C; a good agreement was obtained. The internal mass-transfer resistance resulted in the ratelimiting step of the process. Sulfur sorption was found to be confined to the external layers of the pellets (because it was difficult for H2S to reach the adsorbent core), and a maximization of internal open porosity to improve the system performance was pointed out as the main route for further developments and improvements. Introduction Fuel cells are receiving attention from the thermal efficiency1 and/or environmental points of view. Their high efficiency reduces CO2 emission (greenhouse gas) per unit of electric power without the emission of NOx and SOx. Numerous types of fuel cells have been developed: the differently used materials and operation temperatures make them suitable for several uses.2 Polymeric-electrolyte-membrane fuel cells (PEMFCs) work at low temperature (80-100 °C) and are promising for vehicle applications. The use of fuel cells on cars has been subjected to intense development efforts in recent years because of the significant advantages of the longterm possibility to reach zero emissions of pollutants, particularly important in large metropolitan areas. Gasoline and other hydrocarbon fuels do not have adequate electrochemical reactivity to be used directly in PEMFCs, so a catalytic fuel processor is required to convert these fuels to hydrogen-rich, sulfur-free fuel gases.3 Because the hydrogen demand of the fuel cell stack varies with the electric load required, fuel processors for automotive fuel cell engines must be able to start up quickly, follow the power demand rapidly, and operate efficiently over a wide range of conversion rates. Fuel conversion needs to be essentially complete over the entire load range, and, no less important, automotive fuel processors must be very compact and low in cost. The development of a complete on-board fuel processor for the production of clean hydrogen from commercial gasoline is the specific objective of an industrial-type EU project, PROFUEL, which involves, beyond our group, several partners from the automotive * Corresponding author. Tel: +39-011-5644710. Fax: +39011-5644666. E-mail:
[email protected].
Figure 1. Schematic description of a gasoline for hydrogen production. The average operating temperatures of each process stage are indicated.
and catalyst manufacturing industries. The fuel processor is schematically described in Figure 1. It consists of the following units: a dosing system, an autothermal reformer, a high-temperature water-gas-shift (HTWGS) reactor, a desulfurizer, a low-temperature water-gasshift (LTWGS) reactor, and a selective CO-oxidation (PROX) unit. Although the oil companies are expected to follow the lead entered by BP-Amoco and progressively reduce the sulfur content in gasoline (in Europe, 50 ppm are planned for 2005 against the actual 150 ppm), the complete removal of sulfur is unlikely to happen in the short term. It has been observed that more than 1 ppm of sulfur could dramatically shorten the lifetime of LTWGS catalysts as well as that of the noble metals (Pt and Pt-Ru) used for the selective CO oxidation and for the fuel cell electrodes. In this scenario, the development of long-life, low-cost adsorbers effective in reducing H2S in the reformed syngas to less than 1 ppm is essential. In this context, the trapping efficiency and capacity should be maximized to reduce as much as possible the size and weight of the on-board desulfurizer. The literature reports several studies on metal oxides as candidate desulfurization sorbents: pure oxides (e.g.,
10.1021/ie0208467 CCC: $25.00 © 2003 American Chemical Society Published on Web 03/13/2003
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003 1689
ZnO, CuO, CaO, Fe2O34-6), solid mixtures of metal oxides that react with H2S (e.g., ZnO-Fe2O3, ZnOMnO, CuO-Fe2O36-9), or mixtures of an inert oxide with a solid reactant (ZnO-TiO2, MnO-Al2O3, Fe2O3-Al2O3, CaO-MgO6,10,11) have been investigated as sorbents for the removal of H2S, especially at high temperatures. The addition of an inert solid in the ZnO sorbent is, in fact, expected to stabilize the metal oxide against its reduction to the metal form and/or volatilization.9 Among the different tested oxides, zinc oxide has the highest equilibrium constant for sulfidation, yielding H2S removal down to even fractions of 1 ppm. Its principal limitation is that in the highly reducing atmosphere of synthesis gas it is partially reduced to elemental zinc, volatile above 600 °C, with consequent sorbent loss; for this reason the addition of TiO2, which has a stabilizing effect, is recommended for hightemperature applications. In contrast, if the operating temperature is low, pure zinc oxide has been singled out as the sorbent of choice for desulfurization of coal gas. Considering the fuel processor described in Figure 1, the desulfurizer could be located either before or after the HTWGS unit. In the first case, a more complex sorbent, stable at high temperature, i.e., ZnO-TiO2, would be required; protection of HTWGS catalysts against sulfur poisoning would also be guaranteed. However, in the second case, as represented in Figure 1, a simpler and lower in cost sorbent, i.e., pure ZnO, can be used and, anyway, protection against sulfur poisoning of the most sulfur-sensitive catalysts of the fuel processor (LTWGS and CO-PROX catalysts) is secured. Furthermore, the low-temperature operation brings about another important advantage for vehicle applications: the consequent low gas viscosity implies low pressure drops. Taking into account the quite high sulfur tolerance of the HTWGS catalysts, the location of the desulfurizer after the HTWGS unit was chosen by the PROFUEL partnership, and an adsorption material able to reduce the sulfur content down to 1 ppm at about 250°C with the highest possible capacity was fixed as the main research target. In this paper the development of a long-life zinc oxide sorbent is described. Specifically, the effect of the procedure followed to prepare the sorbents on their H2S removal ability is evaluated, in comparison with a commercial zinc oxide sorbent, by means of sulfidation experiments performed on a fixed bed of sorbent particles. A model of the H2S adsorber based on a ZnO pellets bed was developed as well, with the main purpose of enlightening the mechanisms governing the system performance and outlining pathways for further improvements. Experimental Section Sorbent Preparation. A commercial sample of ZnO was obtained from Aldrich. Conversely, a series of ZnO sorbents were prepared following two different experimental procedures: (i) The glycerine method, a modified version of the so-called “citrates method”:12 an amount of Zn(NO3)2‚ 6H2O (from Aldrich) was mixed with a 40 wt % amount of glycerin (from Fluka) and a 40 wt % amount of water. The mixture was slowly heated to 120 °C until a slight NOx emission started, then rapidly poured into a stainless steel vessel, and kept in an oven at 180 °C for 30 min. Under such conditions, NOx, CO2, and water vapor
form in huge amounts, thus causing the formation of a solid scum, quite friable and porous. Each sorbent was then finely ground in an agate mortar and calcined in an electric oven in calm air for 2 h at 300 °C (ZnO300), at 400 °C (ZnO400), at 500 °C (ZnO500), and at 600 °C (ZnO600). (ii) The combustion synthesis method13 (the urea method): appropriate amounts of Zn(NO3)2‚6H2O and CO(NH2)2 (both from Aldrich) were dissolved in the minimum amount of water possible (about 5 mL for 1 g of final ZnO). After a few minutes of stirring on a heating plate, to ensure proper homogeneity, the soprepared solution was transferred in a capsule, which was placed into an oven kept at the constant temperature of 500 °C (ZnOu500) or of 700 °C (ZnOu700). First, the aqueous solution underwent dehydration, and then the mixture frothed and swelled, until a fast and explosive reaction took off:
3[Zn(NO3)2 + 6H2O] + 5CO(NH2)2 f 3ZnO + 8N2 + 5CO2 + 28H2O (1) and large amounts of gases evolved. The whole process was over after 5-6 min, but the time occurring between the actual ignition and the end of the reaction was less than 10 s. A foamy and easily crumbled material was obtained to give a fine and volatile powder. Each sorbent was then finely ground in an agate mortar and stabilized by calcination in an electric oven for 1 h at 600 °C in calm air. Sorbent Characterization. X-ray diffraction (XRD) analyses (Philips PW1710 apparatus equipped with a monochromator for the Cu KR radiation) were performed on all fresh sorbents to check the crystallization of zinc oxide, on the sulfided ones (see the next section) to check the formation of zinc sulfide, and on the regenerated ones (see the following section) to check the disappearance of zinc sulfide. The specific surface area, determined by the Brunauer-Emmett-Teller (BET) method using N2 (Micromeritics ASAP 2010M apparatus), together with the total pore volume, the pore size, and the pore size distribution, was measured on all of the sorbents in pelletized form, as prepared, after sulfidation, and after regeneration. Fresh, sulfided, and regenerated sorbents were also examined by scanning electron microscopy (SEM) and energy dispersion spectroscopy (EDS) (Philips 515 SEM equipped with an EDAX 9900 EDS) to investigate the possible morphology variation before and after sulfidation as well as the elemental distribution over the pellets. Because the morphological analysis (SEM) needs gold metallization of the sample and the X-ray emission (EDS) of gold overlaps those of sulfur, simultaneous EDS and SEM measurements on the same sample were not possible. Sulfidation Apparatus and Procedure. The desulfurization performance of all of the sorbents was investigated in a bench-scale reactor. A fixed bed of about 2 cm in length containing 0.25 g of sorbent particles (obtained by pressing at 125 MPa the sorbent powders into tablets, by crushing the tablets, and by sieving the produced particles to separate 0.25-0.425 mm granules) was enclosed in a glass tube (i.d. of 4 mm) and sandwiched between two glass-wool layers. The reactor was placed in a PID (PID)-regulated oven and operated at 1 bar of outlet pressure. A thermocouple was inserted in the packed bed for oven regulation purposes.
1690
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003
After 30 min at 250 °C in nitrogen flow, as a common pretreatment, a gas flow rate of 400 Ncm3 min-1 (composition: H2S, 100 ppm; He, balance) was fed to the reactor. The composition and the flow rate of the feed gas were controlled by mass flow controllers, and a pressure transducer checked the pressure of the inlet and outlet gases in order to overlook any possible undesired packing of the sorbent fixed bed. A constant temperature of 250 °C was maintained during the test. The outlet gas stream was sampled every 15 min and analyzed through a gas chromatograph-mass spectrometer (GC-MS; Varian 3400 Saturn 4D). Three aliquots of 1000 µL of the same sampled volume were injected into a fused silica capillary column (Poraplot Q; 25 m length × 0.32 mm i.d.) kept at 40 °C and eluted with a helium flow of about 2 mL/min. The appearance of a peak relative to the mass/charge ratio equal to 34 and 36, due to the molecular ions H2S+ and H234S+, respectively (retention time of 4 min), monitored by the mass spectrometer, permitted a quantitative evaluation of H2S. Therefore, H2S concentration plots versus time were obtained. The time of the abrupt change of the H2S concentration (from 0 to 2 ppm, the minimum detector sensitivity) in the product gas was called “breakthrough time”, whereas the so-obtained time-dependent curves were called “breakthrough curves”. Regeneration Procedure. Sulfided sorbents were regenerated by flowing air at 625 °C for 15 min in the same apparatus as that used for sulfidation experiments. Six successive sulfidation-regeneration cycles were performed on the same sample of the ZnO400 sorbent. Model Equations In the system under investigation, a reaction takes place between the gas-phase reactant and the solidphase surface:
H2S + ZnO f H2O + ZnS
(2)
As the reaction proceeds, the solid particle storage capacity is progressively consumed and the unreacted core of the particle shrinks, thus increasing the difficulty of the H2S molecules to reach the fresh sorbent material. The model developed in the present study consisted of the mass conservation equations for the gas and solid phases. Because the system was operated isothermally and the H2S concentration was very low, the energy balance equation was neglected. The momentum equation was also neglected because of the considerably low pressure drop that occurred in the experimental setup under investigation (a few tens of pascals). The spatial velocity was about 105 h-1. Under these conditions, the longitudinal Peclet number is as high as 300, and a plug-flow assumption was considered to be acceptable in the model formulation.14 Therefore, the following partial differential equation represented the H2S balance in the gas phase:
∂CH2S RH2SAi F ∂CH2S (1 - fb) + fb ) S ∂z ∂t Vp
(3)
The oxygen mass balance on the solid phase was employed to determine the amount of fresh sorbent on a single catalyst pellet. By stating that the variation of the oxygen content of the sorbent particle equals the net oxygen consumption due to the heterogeneous reaction, one obtains the equation describing the variation of the radius of the unreacted particle core:
(Ka)ovcH2S ∂ri )∂t 4πWF r 2
(5)
s i
In eq 5, the density value should be that of the unreacted core of the sorbent particle, whereas the W parameter represents the oxygen content of the unreacted core, expressed in moles of oxygen per unit mass of fresh solid. The overall kinetic constant of eqs 4 and 5 was obtained by considering that the oxygen consumption rate should be determined by accounting for the contribution of several transport resistances in series. To reach the fresh sorbent core, the H2S molecule must be transferred from the bulk of the gas phase to the particle external surface (external mass transport), then must diffuse across the internal pores of the particle (internal mass transfer), and finally undergo a chemical reaction with the unreacted core. By considering these processes to be first order versus the H2S concentration and by assuming the Fick’s law with an appropriate effective diffusivity for the internal diffusion, one obtains the following expression for the overall coefficient:
re - ri 1 1 1 ) + + 2 2 (Ka)ov 4πri Kr 4πre Kext Deff4πrire
(6)
An exhaustive analysis of the reaction kinetics of H2S removal by oxide sorbents was thoroughly performed in the work of Garcia et al.15 A first-order kinetic expression with the following Arrhenius-type kinetic constant was suggested:
Kr ) 1107 exp(-3257.17/T)
(7)
The external mass transport coefficient Kext of H2S in the fixed-bed system was evaluated by means of the following equations reported in the work of Yoshida et al.:16
{
Jd ) 0.91Re-0.51ψ 0.01 < Re < 50 Jd ) 0.61Re-0.41ψ 50 < Re < 1000
(8)
where17
Re ) Jd )
Gdp 6µ(1 - fb)
(9)
Sh Re × Sc1/3
(10)
Sh ) Kextdp/D
(11)
The RH2S that appears in this equation is the specific adsorption rate of H2S that can be expressed as
Sc ) µ/FD
(12)
AiRH2S ) -(Ka)ovcH2S
As far as the internal mass transport is concerned, the pore dimension of the sorbent particles could be such
(4)
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003 1691
Figure 2. XRD patterns of fresh ZnO sorbents prepared with the glycerin method: (a) ZnO300; (b) ZnO400; (c) ZnO500; (d) ZnO600.
that both molecular and Knudsen mechanisms could play an important role in the diffusion process. Therefore, the following equation to calculate the diffusion coefficient18 was employed:
1 1 1 ) + D Dm Dk
(13)
The molecular diffusivity Dm value was calculated by the Chapman-Enskog equation.17 The Knudsen diffusivity was calculated with the following expression, used also by other authors in the present research field:18
Dk )
xMT
19400i AsFs
(14)
The porosity used in this expression should be the intraparticle value, namely, the internal open porosity of the sorbent pellet. It must be noticed, moreover, that the density value to be employed in this expression should be that of the reacted external layers, which have to be crossed by the H2S molecules in order to reach the unreacted core. Once the D value was determined, the effective diffusivity in a sorbent particle could be calculated depending on the diffusion coefficient and on the sorbent tortuosity:
Deff ) Di/τ
(15)
Because no means to evaluate τ were available, the random pore model was then employed19 as a first approximation. This model relates tortuosity and porosity and states that τ is the inverse of . The effective diffusivity expression therefore becomes
Deff ) Di2
(16)
Results and Discussion Experimental Procedures. The diffraction patterns of sorbents prepared with the glycerin method (ZnO300, ZnO400, ZnO500, and ZnO600) are shown, as an example, in Figure 2. The diffraction peaks of ZnO
Table 1. Breakthrough Time, Specific Surface Area (BET), and Pore Volume of Fresh and Sulfided Pellets of ZnO Sorbents fresh
sorbent ZnO300 ZnO400 ZnO500 ZnO600 ZnOu500 ZnOu700 ZnO (Aldrich)
sulfided
pore pore BET volume breakthrough BET volume 2 3 2 (m /g) (cm /g) time (min) (m /g) (cm3/g) 50.3 43.3 13.1 10.1 2.5 2.2 7.8
0.20 0.30 0.13 0.13 0.06 0.04 0.07
550 430 20 10 5 5 20
12.4 12.2 11.3 11.3 2.5 2.2 6.5
0.07 0.10 0.08 0.04 0.06 0.04 0.06
(JCPDS card 80-0075) are clearly seen in each pattern, but they become higher and sharper from ZnO300 to ZnO600 (from curve a to curve d, respectively), because the increasing calcination temperature brings about larger crystals. No other phases were observed. The calcination temperatures were chosen considering the decomposition temperature of zinc carbonate (300 °C20), which forms in large quantity during the reaction of the glycerin method. The diffraction patterns of commercial ZnO and of sorbents prepared via combustion synthesis, the urea method (ZnOu500 and ZnOu700), not shown, are similar to the ZnO600 diffraction pattern presenting well-defined ZnO diffraction lines. The BET data of all fresh sorbents are listed in Table 1. ZnOu500 and ZnOu700 sorbents, prepared with the urea method, have a very low specific surface area (about 2.5 m2/g), whereas the sorbents prepared with the glycerin method have a higher specific surface area, which increases by decreasing the calcination temperature, in good agreement with their diffraction patterns. The urea method, even if very quick, does not permit one to obtain immediately the ZnO phase at both 500 and 700 °C reaction temperatures, so that 1 h of calcination at 600 °C is required and low specific surface area materials are consequently obtained. On the contrary, the glycerin method is more time-consuming but permits one to obtain sorbents with higher specific surface area even when calcination is performed at 600 °C (about 10 m2/g of ZnO600 against the 2-2.5 m2/g of ZnOu500 and ZnOu700 obtained with the urea method).
1692
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003
Figure 3. Pore size distribution of the ZnO300 sorbent in the fresh (solid line) and sulfided (dashed line) states calculated by the BJH adsorption method.
Moreover, the BET area of all of the sorbents prepared with the glycerin method is higher than that of the commercial ZnO. Pore-volume data of pellets of fresh sorbents, also listed in Table 1, are directly related to BET data: sorbents with the highest specific surface area have generally also the highest pore volume. The breakthrough times obtained in sulfidation experiments on each sorbent are listed in Table 1, as well. ZnOu500 and ZnOu700 sorbents show the shortest breakthrough times (5 min), whereas ZnO600, ZnO500, ZnO400, and ZnO300 sorbents show progressively longer times (from 10 to 550 min, respectively). The commercial ZnO has a breakthrough time of only 20 min. By comparison of these results with the BET data and the pore volume of fresh sorbents, the important role of pellet structural properties, i.e., pore volumes and specific surface areas, in sulfur trapping capacity can be easily deduced: as expected, the higher are the specific surface area and the pore volume, the higher is the storage capacity. Table 1 also lists the BET data and the pore volumes of sulfided sorbents. The sorbents with low BET area, low pore volume in their fresh state, and consequently low breakthrough time show a negligible variation in the BET area and pore volume after the sulfidation experiments. On the contrary, the sorbents with high BET area, pore volume, and breakthrough time in their fresh state (specifically ZnO300 and ZnO400) show a strong reduction in both the specific surface area and pore volume after sulfidation. A pure thermal effect at 250 °C in the absence of H2S was evaluated on the specific surface area and pore volume of ZnO300 (the sorbent calcined at the lowest temperature, 300 °C, close to the temperature of sulfidation experiments, 250 °C): the BET area diminished from 50.3 to 37.0 m2/g over a period of 1450 min, whereas the pore volume shows a negligible variation. Hence, sulfidation seems to be the main cause of BET and pore-volume reduction. It is well assessed that the chemical reaction involved in sulfidation is only the one reported in eq 2. Because the molar volume of the sulfide product is greater than that of the oxide reactant (the ratio between the molar volume of ZnS and ZnO is 1.67), morphological and structural property changes do accompany the sulfidation process.
Figure 4. SEM micrographs of pellets of the ZnO400 sorbent: (A) fresh; (B) sulfided.
The relationship between the characteristics of pore and H2S sorptivity has been investigated by calculation of the pore size distribution by the BJH method from the adsorption isotherm. As an example, Figure 3 reports the ratio of the pore volume per unit mass/pore diameter, which can be considered to be proportional to the specific surface area, versus the pore diameter of the ZnO300 sorbent both in the fresh state (solid line) and after sulfidation (dashed line). The average pore
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003 1693
Figure 5. XRD patterns of sulfided ZnO sorbents prepared with the glycerin method: (a) ZnO300; (b) ZnO400; (c) ZnO500; (d) ZnO600. Legend: (2) ZnO; (b) ZnS; ([) sample support (Al).
diameter, calculated by the BJH adsorption method, increases from 17.2 nm in the fresh state to about 26 nm after sulfidation, confirming that the average pore diameter is essentially inversely proportional to the surface area. In particular, Figure 3 shows that in the fresh state (solid line) the pore area is provided by a large amount of small pores of about 1.8 nm and by larger pores of about 20 nm, whereas in the sulfided state (dashed line) the pore area is mainly given by few large pores of about 25 nm. This means that small pores are completely blocked by H2S and only a few large pores are kept after sulfidation. Figure 4 shows SEM micrographs of the ZnO400 sorbent in its fresh state (Figure 4A) and after sulfidation (Figure 4B): the small, quite uniform grains of ZnO in its fresh state become larger and less uniform after sulfidation, whereas EDS analyses confirmed the presence of a large amount of sulfur in the sulfided samples (about 20 wt %). XRD analyses, performed on all of the sulfided sorbents, show the appearance of peaks of the ZnS phase (JCPDS card 72-0162). Figure 5 shows, as an example, the diffraction patterns of sulfided ZnO300-ZnO600 sorbents (curves a-d, respectively). ZnO and ZnS phases are present in all curves, but the intensity of the ZnS peaks is very low for the ZnO600 sorbent (curve d) and higher for the other sorbents. At the same time, if the ZnO peaks of fresh and sulfided sorbents are compared (curves a-d of Figure 2 with curves a-d of Figure 5), it can be noticed that their intensity is lower in the sulfided sorbents; the formation of the ZnS phase involves a progressive disappearance of the ZnO peaks, according to reaction (2). This occurrence is more pronounced for the ZnO300 and ZnO400 sorbents (curves a and b of Figure 5) thanks to their higher sulfur trapping capacity (Table 1). The breakthrough curves of ZnO300-ZnO600 sorbents are shown in detail in Figure 6. The outlet H2S concentration rises from 0 ppm and reaches 100 ppm rapidly after only 269 min for ZnO600 and after 388 min for ZnO500. The sorbents ZnO400 and ZnO300, instead, show a much higher sulfur capture capability as the outlet H2S concentration reaches about 100 ppm at about 1500 min for ZnO400 and well above 1600 min for ZnO300.
Figure 6. Breakthrough curves of ZnO sorbents prepared with the glycerin method. Table 2. Total (Outlet S Concentration Equal to the Inlet One) and Breakthrough Time Amount of Sulfur Adsorbed by ZnO Sorbents during Sulfidation Experiments
sorbent
total S adsorbed (mg of S/g of sorbent)
S adsorbed at breakthrough time (mg of S/g of sorbent)
ZnO300 ZnO400 ZnO500 ZnO600 ZnOu500 ZnOu700 ZnO (Aldrich)
48 48 12 7.0 1.7 2.5 12.5
31.4 24.5 1.14 0.57 0.28 0.28 5.0
The total and breakthrough time amounts of sulfur adsorbed by each sorbent can be calculated considering the total amount of H2S fed to the sorbents during the time of sulfidation experiments (the time required by the outlet H2S concentration to pass from 0 to 100 ppm), the one at breakthrough conditions, and the trend of the breakthrough curves. The results, listed in Table 2, confirm that ZnO300 and ZnO400 sorbents have the highest capacity of sulfur trapping and are in very good agreement with the data obtained by EDS analyses on the sulfided sorbents. The ZnO300 sorbent has the
1694
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003
an oxygen atmosphere gives the following regeneration reaction:
ZnS + 3/2O2 f ZnO + SO2
Figure 7. Breakthrough curve. Comparison between the experimental ZnO400 curve (b) and model calculations (s). Fitting parameter Deff ) 5 × 10-8 m2/s.
highest breakthrough time (550 min), according to its highest specific surface area; the ZnO400 sorbent, however, is singled out as the best and more reliable for application in syngas purification because it has a higher structural stability than ZnO300, deriving from the higher calcination temperature (400 °C), and its sulfur trapping capacity (breakthrough time of 430 min) is very high as well. On the selected sorbent, ZnO400, the regeneration performance was investigated. It is well-known21 that
(17)
Preliminary experiments were carried out in order to investigate the regeneration performance as a function of temperature, time, and O2 concentration. The best regeneration of the ZnO400 sorbent was achieved in 15 min at 625 °C in 20% by volume of O2. Multicycle tests consisting of seven sulfidations with six intervening regenerations were carried out on the same aliquot of a new preparation of the ZnO400 sorbent (breakthrough time in the fresh state of 390 min). The breakthrough times after each sulfidation (186, 92, 73, 60, 57, and 57 min, respectively) show that the ZnO400 sorbent recovers about 50% of its starting sulfur trapping capacity after the first regeneration and about 25% after the second regeneration, and it stabilizes on a value of about 15% from the third to the sixth regeneration. BET data demonstrate a progressive diminution of the specific surface area after each cycle. Starting from a value of 29.2 m2 g-1 in its fresh state, it becomes 21.3 m2 g-1 after the first sulfidation and 20.7 m2 g-1 after the first regeneration; it maintains 20.7 m2 g-1 after the second sulfidation, diminishes to 19.4 m2 g-1 after the second regeneration, and at the end of the seventh sulfidation is equal to 19.2 m2 g-1. If sulfidation brings about inevitably morphological and structural property changes, such changes should
Figure 8. Results of model calculations: concentration profiles (a) and breakthrough curve (b) without internal mass transfer (Deff equal to 5 × 10-5 m2/s); concentration profiles (c) and breakthrough curve (d) in the presence of internal mass transfer (Deff equal to 5 × 10-8 m2/s). Concentration profiles along the sorbent bed are calculated every 2.5 h.
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003 1695
Figure 9. Comparison between the different terms of the overall transport process: (a) ri/re ) 99.95%; (b) ri/re ) 99.5%; (c) ri/re ) 99%; (d) ri/re ) 97%.
be at least partially reversed during the regeneration step. XRD analyses on regenerated sorbents show the total disappearance of the zinc sulfide phase, according to reaction (17); however, small traces of sulfur are detected by EDS analysis on the same materials. More importantly, the unsuccessful recovery of the specific surface area after the regeneration step shows that irreversible structural changes occur. The high temperature requested by the regeneration step (625 °C), much higher than the calcination temperature of the sorbent (400 °C), is likely the main reason for the unavoidable progressive sintering of the sorbent, which, however, stabilizes after the second regeneration step with consequent progressive asymptotic behavior of the breakthrough times of successive sulfidation steps. Modeling and Related Issues. The first step in the numerical analysis was the model validation, carried out on a data set relative to the most promising ZnO sorbent: ZnO400. The comparison between the model calculations and the experimental breakthrough curve is presented in Figure 7. The model was solved considering the effective diffusivity as a fitting parameter that was tuned to achieve the best agreement between experiments and calculations. The optimal Deff value was 5 × 10-8 m2/s. Considering that the sample had a surface area of 43.3 m2/g, this diffusivity value entails, according to eq 16, an internal open porosity of 25%, which appears to be quite a reasonable value if compared with the original 45% porosity of the sorbent pellets measured by helium picnometry. This is a clear sign that sulfidation affects the internal mass-transfer properties of the sorbent despite the fact that sulfur appears to be preferentially captured in the nanosized pores instead of the larger ones, which should likely host most of the diffusive flow rate toward the core of the pellets. The rather flat shape of the breakthrough curve suggests that the internal diffusion process should limit the system performance. Figure 8 shows the calculated
concentration profiles and the corresponding breakthrough curves obtained by computer simulation with Deff values of 5 × 10-5 m2/s (Figure 8a,b) and 5 × 10-8 m2/s (Figure 8c,d) closely related to the presented experimental data (Figure 7). The former Deff is equal to the molecular diffusivity of oxygen in air at 523 K, and the simulation with this value represents the case of diffusion within the sorbent pores without the rate limitation due to the Knudsen regime and to internal tortuosity. In this case, the gas stream would easily access the sorbent material at the particle core that must be completely consumed before the H2S concentration profile can move downstream. As a result, plugshaped profiles can be observed, and the breakthrough curve reflects this behavior, taking the same qualitative shape. On the other hand, in the case of considerable transport resistance due to internal mass transfer, the sorbent material is mainly employed at its external layers, and its core can hardly be reached. As a result, the concentration profiles turn into a different shape, which reveals how the H2S molecules need a fresh external surface to be readily adsorbed and that the core of the particle needs longer residence times to be reached. To further enlighten this point, the three terms of eq 6 were separately calculated and represented in an Arrhenius plot in Figure 9, at different values of the ri parameter. An overall transport coefficient, deduced from (Ka)ov according to eq 18, was represented on the same diagram:
Kov ) (Ka)ov/4πre2
(18)
Moreover, for a direct comparison with the other transport coefficients, the internal mass transport coefficient was defined as
Ki )
Deff re - ri
(19)
1696
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003
Figure 10. Effect of open porosity (i) and space velocity (GHSV) on the sorbent performance: (b) experimental conditions of our tests.
It can be noticed that with the fresh sorbent particles (Figure 9a) the transport resistance due to internal diffusion is relatively negligible. Under these conditions, the system is limited by reaction kinetics at low temperature and by external mass transport at higher temperature. In particular, Figure 9a shows that, at the operating temperature of 523 K (1/T ) 0.0019), the fresh sorbent particles operate under an external transport controlled regime. Parts b-d of Figure 9, obtained by progressively decreasing the ri value, show that the relative importance of the internal mass transport increases as the sorbent particles are consumed. In particular, it can be observed that already when the ri reaches the 99-97% of the initial particles radius (Figure 9d), the system operates in an internal mass transport controlled regime. This remark provides a further enlightenment on the shape of the breakthrough curve because it reveals that even a minor particle consumption already leads the system under the control of internal mass transport. Therefore, the importance of the internal morphology of the sorbent particles on the system performance has to be underlined. Obviously, the most remarkable improvement on the system performance could be achieved by an appropriate preparation method of the sorbent particles, intended to maximize their internal open porosity and to minimize their tortuosity, in line with the observations derived by the investigation on the pore size distribution. More in detail, a multimodal pore structure, i.e., a sort of “clusterized” pore distribution with macropores entering the pellets from its surface with side micropores exploiting the high specific surface area, should improve the effectiveness factor of the pellets and consequently the system performance. The calculations shown in Figure 8 indicate that the breakthrough time could even be doubled in theory. Finally, this optimized pore structure could determine the possibility of a further increase in the performance of the system if it is operated at higher space velocities (see Figure 10, where the experimental conditions of our texts are also drawn). Experimental studies are in progress on this perspective. Conclusions Several pure zinc oxide materials were prepared following two different routes, characterized and tested
as low-temperature desulfurizing sorbents; breakthrough curves in a fixed-bed reactor equipped with sorbent pellets of about 0.34 mm in diameter were obtained. ZnO300 and ZnO400 sorbents, calcined at the lowest temperatures and with the consequent highest specific surface area and pore volume, showed the highest sulfur trapping capacity, with breakthrough times varying from 430 to 550 min measured at a space velocity of 105 h-1; these are very good results, especially if compared to the breakthrough time of the commercial sorbent: 20 min at the same space velocity. The ZnO400 sorbent is singled out as the best and more reliable for application in syngas purification because of its higher structural stability than ZnO300. On the ZnO400 sorbent, muticycle tests of sulfidation and regeneration were carried out to provide information concerning its durability. Regeneration performed for 15 min at 625 °C in air brings about only a partial recovery of the starting sulfur trapping capacity, because of an unavoidable progressive sintering of the sorbent. A numerical model of the system was developed and validated, with a good agreement, by fitting the experimental data set of the ZnO400 sorbent. Under the considered operating conditions, the rate of ZnO pellet sulfidation was limited by the internal mass-transfer resistance, because of the difficulty of H2S molecules reaching the adsorbent core, while the influence of the external mass transfer became rapidly irrelevant. This means that only a portion of the sorbent pellets, the external layers, is consumed and that a maximization of their internal open porosity could improve the sorbent performance. The increase of the internal porosity of sorbent pellets, by exploiting a multimodal pore distribution with macro- and micropores, represents the major pathway for our future research efforts. Nomenclature A ) area [m2] cH2S ) molar concentration of H2S in the gas phase [kmol/ m3] d ) diameter [m] D ) diffusivity [m2/s] F ) gas flow rate [m3/s] G ) specific mass flow rate [kg/s‚m2] Jd ) mass-transfer factor
Ind. Eng. Chem. Res., Vol. 42, No. 8, 2003 1697 K ) kinetic constant (gas-phase transport or surface reaction) [m/s] (Ka)ov ) overall kinetic constant multiplied by a mean exchange area [m3/s] M ) molecular weight of gas-phase species [kg/mol] ri ) radius of the internal unreacted core of a sorbent particle [m] re ) initial radius of the sorbent particle [m] RH2S ) specific adsorption rate of H2S [kmol/m2‚s] Re ) Reynolds number S ) reactor cross section [m2] Sh ) Sherwood number Sc ) Schmidt number T ) temperature [K] t ) time [s] V ) volume [m3] W ) oxygen content of the unreacted core [kmol/kg] z ) reactor axial coordinate [m] Greek Letters ) porosity µ ) gas viscosity [kg/m‚s] F ) density [kg/m3] τ ) tortuosity ψ ) shape factor in the mass-transfer coefficient equation Subscripts eff ) effective ext ) external mass transfer fb ) fixed bed i ) intraparticle value (limit of the internal unreacted core) m ) molecular k ) Knudsen ov ) overall p ) particle r ) chemical reaction s ) solid phase
Literature Cited (1) Haynes, C. Clarifying reversible efficiency misconceptions of high-temperature fuel cells in relation to relation to reversible heat engines. J. Power Sources 2001, 92, 199. (2) Ahmed, S.; Krumpelt, M. Hydrogen from hydrocarbon fuels for fuel cells. Int. J. Hydrogen Energy 2001, 26, 291. (3) Ogden, J. M.; Steinbugler, M. M.; Kreutz, T. G. A comparison of hydrogen, methanol and gasoline as fuels for fuel cell vehicles: implications for vehicle design and infrastructure development. J. Power Sources 1999, 79, 143. (4) Swisher, J. H.; Schwerdtfeger, K. Review of Metals and Binary Oxides as Sorbents for Removing Sulfur from Coal-Derived Gases. J. Mater. Eng. Perform. 1992, 1, 399. (5) Miura, K.; Mae, K.; Inoue, T.; Yoshimi, T.; Nakagawa, H.; Hashimoto, K. Simultaneous Removal of COS and H2S form Coke
Oven Gas at Low Temperature by Use of an Iron Oxide. Ind. Eng. Chem. Res. 1992, 31, 415. (6) Tamhankar, S. S.; Bagajewicz, M.; Gavalas, G. R.; Sharma, P. K.; Flytzani-Stephanopoulos, M. Mixed-Oxide Sorbents for High-Temperature Removal of Hydrogen Sulfide. Ind. Eng. Chem. Process Des. Dev. 1986, 25, 429. (7) Christoforou, S. C.; Efthimiadis, E. A.; Vasalos, I. A. Sulfidation of Mixed Metal Oxides in a Fluidized-Bed Reactor. Ind. Eng. Chem. Res. 1995, 34, 83. (8) Kobayashi, M.; Flytzani-Stephanopoulos, M. Reduction and Sulfidation Kinetics of Cerium Oxide and Cu-Modified Cerium Oxide. Ind. Eng. Chem. Res. 2002, 41, 3115. (9) Gupta, R.; Gangwal, S. K.; Jain, S. C. Development of Zinc Ferrite for Desulfurization of Hot Gas in a Fluid-Bed Reactor. Energy Fuels 1992, 6, 21. (10) Lew, S.; Sarofim, A. F.; Flytzani-Stephanopoulos, M. Sulfidation of Zinc Titanate and Zinc Oxide Solids. Ind. Eng. Chem. Res. 1992, 31, 1890. (11) Lew, S.; Jothimurugesan, K.; Flytzani-Stephanopoulos, M. High-Temperature H2S Removal for Fuelo` Gases by Regenerable Zinc Oxide-Titanium Dioxide Sorbents. Ind. Eng. Chem. Res. 1989, 28, 535. (12) Rosso, I.; Garrone, E.; Geobaldo, F.; Onida, B.; Saracco, G.; Specchia, V. Sulfur poisoning of LaMn1-xMgxO3 catalysts for natural gas combustion. Appl. Catal. B 2001, 30, 61. (13) Civera, A.; Pavese, M.; Saracco, G.; Specchia, V. Combustion synthesis of perovskite-type catalysts for natural gas combustion. Catal. Today 2003, in press. (14) Scott Fogler, H. Elements of chemical reaction engineering, 2nd ed.; Prentice Hall International Editions: Upper Saddle River, NJ, 1992. (15) Garcia, E.; Cilleruelo, C.; Ibarra, J. V.; Pineta M.; Palacios, J. M. Kinetic Study of High-Temperature removal of H2S by Novel metal Oxide Sorbents. Ind. Eng. Chem. Res. 1997, 36, 846. (16) Yoshida, F.; Ramaswami, D.; Hougen, O. A. Temperature and partial pressures at the surface of catalyst particles. AIChE J. 1962, 5, 8. (17) Perry, R. H.; Green, D. Perry’s Chemical Engineers’ Handbook; McGraw-Hill International Editions: Singapore, 1984. (18) Jothimurugesan, K.; Harrison, D. P. Reaction between H2S and Zinc Oxide-Titanium Oxide Sorbents. 2. Single-Pellet Sulfidation Modeling. Ind. Eng. Chem. Res. 1990, 29, 1167. (19) Wakao, N.; Smith, J. M. Diffusion in catalyst pellets. Chem. Eng. Sci. 1962, 17, 825. (20) Pascal, P. Noveau Traite` de Chimie Minerale; Masson: Paris, 1958. (21) Focht, G. D.; Ranade, P. V.; Harrison, D. P. Hightemperature Desulfurization Using Zinc Ferrite: Regeneration Kinetics and Multicycle Testing. Chem. Eng. Sci. 1989, 44, 2919.
Received for review October 25, 2002 Revised manuscript received February 10, 2003 Accepted February 13, 2003 IE0208467