5846
Ind. Eng. Chem. Res. 2004, 43, 5846-5853
KINETICS, CATALYSIS, AND REACTION ENGINEERING Mass-Transfer Characteristics for Gas-Liquid Reaction of H2S and Sulfuric Acid in a Packed Column Hui Wang,† Ivo G. Dalla Lana, and Karl T. Chuang* Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta T6G 2G6, Canada
The mass-transfer characteristics of a gas-liquid reaction system in a packed column filled with ceramic Raschig rings were studied using the reaction between hydrogen sulfide (H2S) and sulfuric acid solutions. An analysis based on two-film theory shows that the mass-transfer resistance consists of two consecutive steps: gas-side mass transfer and surface reaction. The resistance from the liquid side was negligible because the concentration of sulfuric acid was above stoichiometric and can be regarded as a constant. Onda et al.’s correlations (Onda, K.; Takeuchi, H.; Okumoto, Y. J. Chem. Eng. Jpn. 1968, 1, 56) are able to estimate the effective interfacial area as well as the mass-transfer coefficients for our reactor system. Because the reaction between H2S and concentrated sulfuric acid is a pseudo-first-order reaction with respect to H2S under the experimental conditions used, the approximate equality between the measured overall mass-transfer coefficient and the reaction rate constant suggests the regime of reaction rate control. In other words, the comparison between the rate constants and mass-transfer coefficient is able to show the rate-controlling regimes in terms of operating conditions such as acid concentration, temperature, and acid and gas flow rates. Tests were also carried out with gaseous compounds often found in industrial H2S streams. No reaction was observed for methane, carbon dioxide, carbonyl sulfide, and carbon disulfide. However, the conversion of ethylene was about 20%, and those of mercaptan and thiophene were nearly 100%. This study provides useful data that can facilitate scale-up calculations of this potential sulfur removal and recovery technology. Introduction A new sulfur removal and recovery technology is being sought based on the gas-liquid reactions of H2S and sulfuric acid, i.e., the reaction between H2S and concentrated sulfuric acid, denoted as reaction 1, and the reaction between H2S and sulfur dioxide (SO2), which is produced in the former reaction, denoted as reaction 2. The reaction kinetics of these two reactions were studied separately using a gas-liquid batch reactor under the extreme conditions where only one of the reactions occurred at a time and the reaction rate was controlling.1,2 The results suggested that the two reactions predominantly occur at the interface between the gas and liquid phases. The rate equations of the two reactions per unit interfacial area are
rH2S,1 ) kP1PH2S
(1)
rH2S,2 ) kP2PH2SPSO2
(2)
where rH2S,1 and rH2S,2 are the reaction rates per * To whom correspondence should be addressed. Tel.: (780) 492-4676. Fax: (780) 492-2881. E-mail: karlt.chaung@ ualberta.ca. † Present address: Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Saskatchewan S7N 5C5, Canada.
interfacial area (mol s-1 m-2) for reactions 1 and 2, respectively; kP1 and kP2 are the corresponding rate constants; and PH2S and PSO2 are the partial pressures of H2S and SO2, respectively. A column reactor packed with ceramic Raschig rings is technically favored to carry out these gas-liquid reactions when the technology based on the kinetic results is developed. For scaleup, two additional issues have to be taken into consideration: the mass-transfer characteristics of this reaction system in the packed column reactor and the reactivity of the reaction agent, i.e, sulfuric acid, to possible species contained in the to-be-treated gas streams. This paper reports the results of a study of these two respects. For the reaction between H2S and sulfuric acid, H2S from the bulk gas and H2SO4 from the bulk liquid diffuse to the interface, where they react. For the reaction between H2S and SO2, after the latter has been generated from the former reaction at the interface, SO2 must dissolve in the liquid phase to initiate the reaction with H2S.2 However, when fresh concentrated (g90 wt %) sulfuric acid is used, the reaction between H2S and sulfuric acid is overwhelming, and the reaction between H2S and SO2 is negligible. Meanwhile, when a dilute H2S gas mixture (e5 mol %) is fed, sulfuric acid can be considered to be present in excess, so that its concentration remains essentially constant during the course of the reaction. As a result, the mass-transfer resistance to the diffusion of H2SO4 in the liquid phase is also
10.1021/ie030845u CCC: $27.50 © 2004 American Chemical Society Published on Web 07/10/2004
Ind. Eng. Chem. Res., Vol. 43, No. 18, 2004 5847
Figure 1. Mass-transfer profile of the gas-liquid reaction of H2S and H2SO4.
negligible. According to the two-film model,3,4 the masstransfer profile, as shown in Figure 1, is that H2S diffuses from gas phase to the gas-liquid interface, where it undergoes a pseudo-first-order reaction with H2SO4. At steady state, the flux of H2S being transported from the bulk gas to the interface is equal to the reaction rate per unit surface area i PH ) 2S
i kP1PH 2S
where y is the mole fraction of H2S, a is the gas-liquid interface area per unit volume packing, P is the total pressure of the gas phase, G is the superficial molar flow rate of the gas, and z is the height of packing bed. To use eq 8 to calculate KGa from experimental data, the experiments have to be carried out such that the amount of mass transport along the axial direction is negligible compared to the amount of mass transported by convection and the acid concentration change resulting from the reaction is insignificant. The first condition is satisfied for almost all situations involving flow in packed-column reactors,5 and so is the second if the proper flow rates of H2S and sulfuric acid are chosen. The prerequisite for using eq 8 to separate KG from KGa is to have the value for the mass-transfer area per unit volume of the packed bed, a. After comparing the applicability of the available models to our reactor system in terms of packing material and operating conditions, we chose to use Onda et al.’s correlation (eq 9) for the estimation of this parameter. To further confirm the validity of Onda et al.’s equations for the packed-bed reactor that was used, the liquid-side masstransfer coefficient, kL, for the H2S-water system was measured and compared with the values estimated according to eq 11. Subsequently, the gas-side masstransfer coefficient values calculated from the experimental data using eq 8 and that estimated from Onda et al.’s correlation (eq 10) were compared, and the ratecontrolling regime was determined from the comparison. Onda et al.’s empirical equations6 are as follows
aw/at ) 1 - exp[-1.45(σc/σ)0.75(L/atµL)0.1
(3)
(L2at/FL2g)-0.05(L2/FLσat)0.2] (9)
i The partial pressure of H2S at the interface, PH , is 2S not as easily measured as the bulk partial pressure, i PH2S. To eliminate PH , solving eq 3 yields 2S
kGRT/atDG ) c(G/atµG)0.7(µG/FGDG)1/3(atDp)-2.0 (10)
rH2S ) kG(PH2S -
i ) PH 2S
)
kGPH2S kG + kP1
(4)
and the rate of reaction at the interface becomes
rH2S ) KGPH2S
(5)
where KG is the overall mass-transfer coefficient and can be defined as
1 1 1 ) + KG kG kP1
(6)
for this reaction system. For the packed column reactor, a steady-state mole balance on H2S in a differential segment of the bed along the axial direction of the reactor results in the differential equation
dy KGaP ) dz y G
(7)
Integration of eq 7 from y1, the inlet mole fraction of H2S, to y2, the outlet mole fraction of H2S, gives rise to the expression
ln
y2 KGaPz ) y1 G
(8)
kL(FL/µLg)1/3 ) 0.0051(L/awµL)2/3(µL/FLDL)-1/2(atDp)0.4 (11) where aw is the wetted surface area of the packing, m2 m-3; at is the total surface area of the packing, m2 m-3; σc is the critical surface tension of the packing material, dyn cm-1; σ is the surface tension, dyn cm-1; L is the superficial mass velocity of the liquid, kg m-2 h-1; µL is the viscosity of the liquid, kg m-1 h-1; FL is the density of the liquid, kg m3; g is the gravitational constant, m h-2; kG is the gas-phase mass-transfer coefficient, kgmol m-2 h-1 atm-1; R is the gas constant, m3 atm kgmol-1 K-1; T is the absolute temperature, K; G is the superficial mass velocity of the gas, kg m-2 h-1; DG is the diffusivity of H2S in the gas, m2 h-1; µG is the viscosity of the gas, kg m-1 h-1; FG is the density of the gas, kg m-3; Dp is the nominal size of the packing, m; and c is 5.23 for the larger packing and 2.00 for the smaller (nominal size < 15 cm) packing. In addition, to determine whether the developing technology would interact with the useful constituents and other sulfides in possible to-be-treated sour gas streams, this study also includes the reactions between concentrated sulfuric acid and possible components in sour gas streams: methane, ethylene, carbon dioxide, carbonyl sulfide, carbon disulfide, mercaptan, and thiophene. A semibatch reactor was used to measure the conversion and determine whether these compounds react with sulfuric acid.
5848 Ind. Eng. Chem. Res., Vol. 43, No. 18, 2004
Physicochemical Properties
binary system of H2S and N2
Some physicochemical properties of the gases and liquids involved in this study are discussed as follows: Density, Gravitational Constant, and Surface Tension. The density of sulfuric acid solutions and water was found from Perry’s Chemical Engineers’ Handbook.7 The value for a particular acid concentration and temperature was determined using interpolation from the available data if it was not available in the handbook. The density of the gas was calculated using the equation of state for an ideal gas. The density of water was calculated in terms of the temperature used in the experiments. The gravitational constant is 9.807 m s-2. The critical surface tension, σc, of ceramic rings is 56 dyn cm-1 from Table 18-11 of Perry’s Chemical Engineers’ Handbook.7 The surface tensions, σL, of water and sulfuric acid have been reported by Myhre et al.8 Although the surface tension of a liquid varies with temperature, the dependence is not significant. For example, the surface tension of 96 wt % sulfuric acid changes from 0.05544 N m-1 at 0 °C to 0.05308 N m-1 at 50 °C. The value of surface tension at a particular temperature was estimated by interpolation or extrapolation using the data available. Viscosity. The viscosities of sulfuric acid solution and water at a given temperature were chosen from the nomograph for viscosities of liquids.7 The viscosities of gases such as nitrogen can be found in CRC Handbook of Chemistry and Physics.9 Diffusivity. The diffusivity of H2S in water was calculated by the Wilke-Chang correlation10
T DH2S,L ) 1.173 × 10-16(φML)0.5 µLVH2S0.6
(12)
where DH2S,L is the diffusivity of H2S diffusing through water, m2 s-1; φ is the “association parameter” of the solution, which is 2.6 for water, as suggested by Wilke and Chang; ML is the molar molecular weight of water, g mol-1; µL is the viscosity of the solution, Pa s or kg m-1s-1; and VH2S is the molar volume of H2S at the boiling point, 0.0329 L mol-1. Thus, eq 12 can be simplified to read
T DH2S,L ) 1.339 × 10-14 µL
(13)
The diffusivity of H2S in the liquid phase, DH2S,L, is used for the calculation of kL, and a mass-transfer resistance analysis indicates that the resistance in the liquid can be negligible when reaction occurs between a dilute H2S gas stream and a concentrated sulfuric acid solution. Therefore, the calculation of DH2S,L in sulfuric acid was unnecessary. The diffusivity of H2S in the gas phase was determined using the Fuller equation10
DAB )
1 × 10-7T 1.75(1/MA + 1/MB)1/2
∑vA)1/3 + (∑vB)1/3]2
P[(
(14)
where A represents H2S and B represents N2 for most of the runs in this study; DAB is the diffusivity, m2 s-1; T is the temperature, K; MA and MB are the molar weights of A and B, respectively, g mol-1; P is the total pressure, atm; and ∑vA and ∑vB are the sums of the structural volumes of A and B, respectively. For the
DAB ) 8.840 × 10-10T 1.75/P
(15)
Experimental Method A schematic diagram of the experimental setup is shown in Figure 2. Because both liquid (sulfuric acid) and gas (containing H2S) involved in this study are corrosive, ceramic Raschig rings with a nominal size of 6 mm, which is the magnitude of the outside diameter and ring height, were used as the packing. The column was made of Pyrex glass with an inner diameter of 0.06 m and a height of 1 m. The bottom section of the column consisted of a packing holder made of Teflon, a gasliquid separation chamber, a liquid outlet flow-controlling valve, and a gas inlet/outlet tube. The top section consisted of a liquid inlet tube connected to a liquid distributor and a gas inlet/outlet tube. The three sections were joined together by two standard joints and sealed with Teflon gaskets. Curved aluminum blocks, inside which the cartridge heaters were installed, were placed around the column to protect the reactor and to heat it as well. The temperature of the heaters was controlled by an Omron temperature controller, model E5CX (Omron Corporation, Tokyo, Japan) with 0.1 °C resolution. In addition to the thermocouples that measured the temperature of the column skin, two other thermocouples with glass covers were located in the inlet and outlet acid solutions to monitor the temperature inside the column. Installing different amount of packing allowed a change in the height of the packing in the column. The position of the liquid distributor was adjustable such that it could be positioned just above the packing when the height of the packing was changed. The gas stream was introduced into the column either from the top or from the bottom so that either co-current or countercurrent operation was possible. Most of the runs were carried out in co-current flow. The flow rates of the cylinder gases were regulated by mass flow controllers (Sierra Instruments, Inc., Monterey, CA), which were calibrated using a BIOS primary air flow meter, model DryCal DC-2M (BIOS International Corporation, Pompton Plains, NJ). The fresh acid solution was stored in a 5-L tank of Pyrex glass, which was heated to a preset temperature by a hot plate (model 200T, Fisher Scientific). The heated acid solution was pumped into the column using a variable-speed acid pump (Cole Parmer Instruments Co., Chicago, IL). The preheater was located after the pump if temperatures higher than 100 °C were required. This is because the acid pump could not be operated above 100 °C. The spent acid solution was collected in another 5-L tank. For the sake of safety, the glass tanks were placed into stainless steel pockets. For a typical run, sulfuric acid solution of the desired concentration was prepared and heated to a preset temperature. The inlet gas mixture was prepared according to the chosen composition and flow rate. The composition was also analyzed using a GC equipped with TCD and SCD.11 Then, the solution was pumped into the column at a set flow rate. The gas mixture was directed to the column once steady acid flow had been established. After the reaction, both gas and liquid were directed to the gas-liquid separation chamber below the packing bed (for co-current flow). A valve was used to control the liquid level in the chamber. When steady state was reached, the composition of the effluent gas
Ind. Eng. Chem. Res., Vol. 43, No. 18, 2004 5849
Figure 2. Schematic diagram of the experimental setup for the packed-bed column reactor: 1, mass flow controllers; 2, fresh acid storage tank; 3, protecting pockets; 4, temperature controllers; 5, valves; 6, acid pump; 7, acid preheater; 8, heating-oil storage tank; 9, heatingoil pump; 10, packed-bed column reactor; 11, temperature indicator; 12, gas-liquid separator; 13, acid flow control valve; 14, spent acid storage tank; 15, pressure gauges; 16, 4-way valve; 17, needle valves; 18, GC coupled with TCD and SCD.
was analyzed three times during a run. Usually, it took 15 min for the system to reach steady state. Sulfuric acid solutions at various concentrations were prepared by diluting ∼96 wt % sulfuric acid (Fisher Scientific, Nepean, Ontario, Canada) with deionized water. The 100 wt % acid was prepared by mixing 20% free SO3 fuming sulfuric acid (Acros Organics, Morris Plains, NJ) with 96 wt % sulfuric acid solution. The concentrations of the solutions were determined by titration with a standard 0.1 N sodium hydroxide solution (Fisher Scientific, Nepean, Ontario, Canada), using 0.1% methyl orange solution (Fisher Scientific, Fair Lawn, NJ) as an indicator. Cylinders of anhydrous hydrogen sulfide (CP grade), SO2, and prepurified nitrogen were provided by Praxair Products Inc. (Mississauga, Ontario, Canada). The specifications for other chemicals can be found in ref 11. Results and Discussion Estimation of the Liquid-Side Mass-Transfer Coefficient. The H2S-water system was used to estimate the liquid-side mass-transfer coefficient of our packed-bed reactor. In such experiments, a high degree of purity of the gas phase is usually used such that the gas-side resistance is considered to be negligible in comparison with the liquid-side resistance.12 However, for safety reasons, these experiments employed a feed gas containing H2S (1-5 mol %) and nitrogen (inert). Thus, the H2S flux from gas to liquid in terms of the
overall liquid-side mass-transfer coefficient can be expressed as * rH2S ) KL(CH - CH2S) 2S
(16)
1 1 1 ) + KL HkG kL
(17)
where
KL is the overall liquid-side mass-transfer coefficient; * is the H2S concentration in equilibrium with a CH 2S pressure of PH2S in the gas phase; CH2S is the H2S concentration in the liquid phase; H is the Henry’s law constant; and kG and kL are the gas-side and liquid-side mass-transfer coefficients, respectively. A molar balance for a differential segment of the bed along the axial direction of the reactor at steady state gives
L * dCH2S ) KLa(CH - CH2S) dz 2S FL
(18)
where L is the superficial mass flow rate of the liquid, FL is the density of the liquid phase, a is the effective interfacial area, and z is the axial direction of the column. L could be regarded as a constant because the amount of H2S transferred from the gas phase to the liquid phase was small. To easily integrate eq 18, the experiments were carried out such that PH2S did not * could be change significantly (