1160
Ind. Eng. Chem. Process Des. Dev. 1985, 2 4 , 1160- 1 165
third-order model can be used to approximate other models in some cases. Of course, in using the ZieglerNichols method, one does need to know the process model, but as noted in the paper, the Ziegler-Nichols may fail to give satisfactory performance, whereas the method proposed here will not fail. One might well question what the advantages are of the method in comparison with such other classical methods as root-locus and frequency response (Le., Bode plots) since these latter methods are not limited to third-order models. The root-locus method is a rather complicated method and becomes especially so if a pure delay is involved. In using the Bode method, one would need to construct several frequency response plots to determine reasonable controller settings in terms of gain and phase margin. Some trial and error would be involved. On the other hand, the method proposed here is computationally simple and straightforward. The initial controller settings provided by the method will be better than those given by the method of Ziegler and Nichols. If desired, some trial and error could be used to find the "best" value of p, but this additional step may not always be necessary. Nomenclature
g, = gain margin G = transfer function K = product of controller and process gains Kc = controller proportional gain
K,, K , = process gains Kp = product of process gains KpI= maximum loop gain under proportional-integralcontrol K,,, = maximum loop gain s = Laplace operator R1 = TIIT3 R2 = T2IT3 T I ,T,, T3 = process time constants
TR = reset time Greek Symbols a = tuning parameter for gain in eq 21a
P = tuning parameter for reset time in eq 21b { = damping factor 7 = dimensionless reset time defined by eq 5b 0 = system parameter defined by eq 9c X = dimensionless frequency defined by eq 5c A, = dimensionless critical frequency E, = dimensionless process parameter defined by eq 5a Xp, = dimensionless critical frequency defined by eq 9b u, (TI, u2 = frequency-dependentparameters defined in eq 6a-c w = frequency, radltime w, = natural frequency of second-order component of process L i t e r a t u r e Cited Cohen, G. H.; Coon, G. A. Trans. ASME 1953, 75, 827. Coughanowr, D. R.; Koppel, L. B. "Process Systems Analysis and Control"; McGraw-Hill: New York, 1965. Harriott, P. "Process Control"; McGraw-Hill: New York, 1964. Latour, P. R.; Koppel, L. B.; Coughanowr, D. R . Ind. Eng. Chem. Process Des. Dev. 1967, 6 , 452. Lopez, A. M.; Miller, J. A.; Smith, C. L.; Murrill, P. W. Instrum. Techno/. 1967, 14, 57. McAvoy, T. J.; Johnson, E. F. Ind. Eng. Chem. Process Des. Dev. 1967, 6 , 440. Miller, J. A.; Lopez, A. M.; Smith, C. L.; Murrill, P. W. ControlEng. 1967, 14, 72. Murrill, P. W.; Smith, C. L. Hydrocarbon Process. 1966, 4 5 , 105 Smith, C. L.; Murrill, P. W. ISA J. 1966, 13, 50. Weber, T. W. "An Introduction to Process Dynamics and Control"; Wiley-Interscience: New York, 1973. Weber, T. W.; Bhalodia, M. Ind. Eng. Chem. Process Des. Dev. 1979, 18, 217. Weigand, W. A.; Kegerreis, J. E. Ind. Eng. Chem. Process Des. Dev. 1972, 11, 86. Yuwana, M.; Seborg, D. E. AIChE J. 1982, 28, 434 Ziegler, J.; Nichols, N. Trans. ASME 1942, 6 4 , 759.
Received for review June 8, 1984 Revised manuscript received December 5 , 1984 Accepted December 12, 1984
Infinite Dilution Activity Coefficients by Inert Gas Stripping Method: Extension to the Study of Viscous and Foaming Mixtures Domlnique Rlchon Ecole Nationale Sup6rieure des Mines de Paris, Centre R6acteur.s et Processus, Equipe de recherche associ6e au CNRS n O 768, Laboratoire de Thermodynamique, 77305 Fontainebleau, France
FranGols Sorrentlno and Andrde Vollley €cole Nationale Sup6rieure de Biologie Appliquh a la Nutrition et B L 'alimentation, Centre Universitaire Montmuzard, Laboratoire de Biologie Physico-Chimique, 2 1 100 DJon, France
The inert gas stripping method allows rapid determination of accurate infinite dilution activity coefficients. Unfortunately, previous works were limited to the study of nonfoaming mixtures with low viscosity (less than 50 cP) due to the design of the equilibrium cell. In this paper, a new cell design is proposed to extend the validity of the method up to 1000 cP. In addition, a special device was developed to break foams without disturbing phase equilibrium inside the equilibrium cell. Preliminary tests prove that the new apparatus is well suited to investigations of aqueous mixtures of polyols, glucides, and proteins.
Rapid determination of vapor-liquid equilibria is of great interest for industry: for example, in the selection of solvents to be used in extractive distillation, liquid extraction, or optimal conditions in food processing. In the
last case, one important question is how to provide consumers with the food having the best sensory qualities at the lowest prices. To answer this question, it is necessary to known the retention of aroma in food. A first approach
0196-4305/85/1124-1160$01.50/00 1985 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
1181
Ln S
I.
CD-
Figure 1. Equilibrium cell: A, vapor-phase outlet; B, deflector; C, pivot; D, gasket; E, foam-breaking device; F, permanent magnet; G , plug; H, bladed screw; I, Archimede's screw; J, internal cylinder; K, dilutor still; L, carrier gas capillary injectors; and M, carrier gas inlet (helium).
to the problem is to measure infinite dilution activity coefficients of aroma in different substrates composing real food (Lebert and Richon, 1984). In order to measure infinite dilution coefficients close to real conditions, aqueous mixtures with high concentrations of sugars (viscosity up to lo00 cP)must be examined as well as aqueous mixtures with a high concentration of proteins (foaming mixtures) or polyols (viscous and foaming mixtures). This paper is therefore specially devoted to the design of a new equilibrium cell which works well under the aforesaid difficult experimental conditions. Equipment and Procedure The principles and equipment have been fully described in papers by Leroi et al. (1977) and Richon et al. (1980). A few improvements only have been made: an automatic sampler was set up, and a GL chromatograph (Girdel Model 3000, with flame ionization detector) was used. The equilibrium cell "dilutor cell" is completely new. Its design results from a mass-transfer study presented in the Appendix section. Equilibrium Cell. A drawing of the equilibrium cell is given in Figure 1. The cell is composed of a glass tube (K) closed at each extremity by plugs (G), the corresponding sealings being achieved by using soft gaskets (D). At the lower extremity of the glass tube is found the carrier gas inlet (M) holding the capillary injectors (L). At the other extremity are the vapor-phase outlet (A), a deflector (B), a foam-breaking device (E),and a bladed screw (H). The bladed screw is used to prevent liquid rotation in the cell and promote the coalescence of gas bubbles. Between the two extremities, an Archimede's screw is maintained by two pivots. It is activated by means of a magnet (F) and is used to circulate the liquid from top to bottom inside the internal cylinder (J). Experimental Procedure. The substrate is prepared by weighing within g. The amount of substrate introduced in the cell is then deduced from double-weighing. The solute (about 50 ML)is injected directly into the cell by using a chromatographic syringe; its mole fraction is always less than 5 x The cell is connected to the chromatographic circuit. The Archimede's screw rotation speed and carrier gas flow rate are adjusted in such a way
12
1
t !'AI
ma m a 7588 Figure 2. Logarithm of solute peak area as a function of time.
as to obtain very small gas bubbles less than 1mm. The temperature is regulated within K by using a liquid bath and measured by using a platinum probe. Samples of vapor phases continuously flowing out from the cell are periodically analyzed through the GLC instrument which allows the elution law of the solute to be determined and the infinite dilution activity coefficients at a temperature T to be calculated with the equation (Leroi et al., 1977)
where N = CiNiif the solvent(substrates) is a multicomponent solvent. This expression may be used when activity coefficients are not too high or vapor-phase volumes in the dilutor cell are small enough to avoid the correction proposed by Duhem and Vidal(1978). For all measurements taken in this experiment, the vapor-phase volumes were always chosen to be very small so as to fulfill the last requirement. In Figure 2, we have plotted the logarithm of the solute peak area, S , as a function of time, t. The results is a straight line with only the three first experimental points which are too high. These points allow the estimate of the time necessary to reach equilibrium conditions in the whole circuit of the apparatus from the time zero where a given carrier gas flow rate is set up. Accuracies of activity coefficient determinations depend heavily on temperature measurements because of the calculation of the solute vapor pressure An extensive study of the temperature influence is available in Sorrentino (1984). Under good experimental conditions, the accuracy of 7," is estimated to be 170, Experimental Results The purities and origin of the solutes are given in Table I. The water is bidistilled. The characteristics of glucides are presented in Table IT. The poly(ethy1ene glycols) (PEG) are Mercks products. The whole bovine casein was obtained from skimmed milk diluted with the same quantity of water. The pH of this solution is adjusted to 4.6 by using a 1N hydrochloric acid aqueous solution to obtain a precipitate which is filtrated on a Buchner. The precipitate is then washed 3 times to flush out lactose and inorganic salts and is freeze-dried (freezer-drier SMJ usifroid) for 44-48 h after setting its pH to 7 by using a 1 N aqueous solution of soda.
e(T).
1162
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
Table I. Vapor Pressure, Purity, and Origin of the Solutes vap press P,s(T), compon temp, K mmHg min pur origin methanol 303.15 163.8O 99.7 Merck methanol 298.15 126.9" ethanol 303.15 78.6" 99.85 Merck ethanol 298.15 59.1" I-propanol 303.15 28.57" 99.5 Merck 1-propanol 298.15 20.72" 303.15 285.2" 99 Merck 1-propanone 303.15 114.1n 99.5 1-butanone Merck 303.15 ethyl formate 305.8* 97 Merck 303.15 120.1" ethyl acetate 99.7 Merck 28.57" propyl acetate 303.15 99 Eastman Kodak 303.15 ethanal 11OO.OQ 99 Merck 303.15 440.0' propanal 99 Carlo Erba
Table IV. Infinite Dilution Activity Coefficients Compared to Literature Data at 298.15 K activity coeff solute solv a b c d e hexane hexadecane 0.889 0.904 0.892 0.888 methanol water 1.64 1.65 ~
"This work. bFrom Leroi et al. (1977). 'From Lenoir et al. (1971). dFrom Richon et al. (1979). eFrom Lebert (1984). :f
20
15
"From Kojima and Tochigi (1979). bFrom Perry (1963). 'From Hala et al. (1968).
Table 11. Characteristics of Glucides mean molar compon formula massp g C6H1206 180 g1u cose (anhydrous) C12H22OI1(+HzO) 360 maltose (monohydrate) glucose syrup (C6H11O5In 324 DE 61.5 glucose syrup (C6HI1o5), 1170 DE 20 ,a From
10
origin Merck Merck
5
Roquette FrBres Roquette FrBres number o f Carbon a l o m s
freezing point depression method.
Table 111. Influence of the Carrier Gas Flow Rate on the Determined Values for Activity Coefficients of Propanone at 303.15 K flow rate, substrate cm3/s YpropanoneD water 1.6 7.71 water 1.9 7.65 water 4.0 7.72 water 5.1 7.69 water(1)-glucose syrup DE 20(2) (W2= 0.5) water(1)-glucose syrup DE 20(2) (W2 = 1.4 10.8 0.5) water(1)-glucose syrup DE 20(2) ( W , = 0.5) water(1)-glucose syrup DE 20(2) (W2 = 0.5) water(1)-glucose syrup DE 20(2) (W, = 0.5)
i
G
1.8
11.0
3.0
10.7
4.3
9.9
Some measurements have been made by using the dilutor technique with several different carrier gas flow rates. The values of the limiting activity coefficient of propanone are presented in Table I11 as a function of flow rate in a low viscous solvent (water) and a substrate (water + glucose syrup DE 20), with a viscosity of 112.0 cP. When the viscosity is low, equilibrium is reached no matter what the flow rate (in the example) is, but for high viscosities, the flow rate has to be limited (at about 3 cm3 s-l in the example). By adjusting the rotation speed of the Archimede's screw, the size of the gas bubbles was always less than 1 mm in diameter for all the y measurements, except for the last one of Table 111, when a flow rate of 4.3 cm3 s-' was used. In this last case, the maximum rotation speed of the Archimede's screw was reached, and the bubble
2
3
Figure 3. Alcohol infinite dilution activity coefficients as a function of the number of carbon atoms (contained in their chain) in different substrates: (x) water; (0) water(l)-glucose(2) (W, = 0.5); (+) water(l)-maltose(2) (W2 = 0.5); (0)water(1)-glucose syrup DE 61.5(2) ( W , = 0.5); (A)water (l)-glucose syrup DE 20(2) ( W 2 = 0.5); (e) water(1)-poly(ethy1ene glycol 35000) ( W , = 0.3); (*) water(1)-whole bovine casein(2) (W2= 0.15). 0
r
rb
distance
,r
Figure 4. Concentration profile of a solute inside and around a gas bubble immersed in a solvent.
diameters nevertheless remained too big (about 3 mm). Comparisons between activity coefficients obtained with the new dilutor and the literature's are presented in Table IV. They display a very good agreement. Infinite dilution activity coefficient determinations (see Figure 3) were then performed for several solutes in different substrates: alcohol in viscous substrates/waterglucides systems (see Table V); solutes in foaming sub-
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 1163 Table V. Substrate Viscosity and Alcohol Infinite Dilution Activity Coefficients at 298.15 K infin dilut activ coeff substrate viscos CP methanol ethanol 1-propanol 1 1.64 water 3.55 11.2 water(1)-glucose(2); W z = 0.5 5.85 13.5 2.39 19.4 water(l)-maltose(2); Wz = 0.5 14.4 1.93 4.38 13.8 4.81 water(1)-glucose syrup DE 61.5(2); W z = 0.5 15.0 1.93 15.1 water(1)-glucose syrup DE 20(2); Wz = 0.5 4.41 112.0 2.01 13.4 water(1)-PEG SOOO(2); W , = 0.3 31.5 1.61 3.38 8.8 water(1)-PEG lOOOO(2); Wz = 0.3 86.2 1.63 3.40 9.1 water(1)-PEG 35000(2); Wz = 0.3 798.5 1.60 3.45 9.9 Table VI. Infinite Dilution Activity Coefficients of Solutes in Water(1)-Whole Bovine Casein(%)at 303.15 K and Solute Infinite Dilution Henry's Constants in Pure Water infin dilut activ coeff y i m in water(1)-whole bovine casein(2) with solute = 0.10e W z = 0.15* Hi", atm methanol 1.70 1.72 0.27' ethanol 3.70 3.80 0.28" 11.15 11.1 1-propanol O.3On 2.94b 9.4 10.3 1-propanone 4.50b 35.9 39.1 1-butanone 75.6 84.0 ethyl formate 23.4b ethyl acetate 12.6b 102 115 610 propyl acetate 17.8b 560 ethanol 6.6b 6.0 6.8 propanol 9.0b 21.9 25.1
/ i
w,
"Values at 298.15 K. bValues at 303.15 K. cViscosity = 38 CPat 298.15 K. dViscosity = 829.5 CPat 298.15 K.
Figure 5. Influence of viscosity on the limiting bubble speed for different gas bubble diameters: (-) 2rb = 2 mm; (- -) 2rb = 1.5 mm; (-e-) 2rb = 1 mm; and (--) 24, = 0.5 mm.
-
I
.
.
.
.
,
500
.
.
.
.
,ob0
VlSCOSlTV (CP)
Figure 6. Influence of viscosity on the mass-transfer coefficient for different gas bubble diameters: (-) 2rb = 2 mm; (- - -) 2rb = 1.5 mm; (-e--) 2rb = 1 mm; and (--) 2rb = 0.5 mm.
/
20.
0'
1
2
/
/
3
4
2 ,b(")
Figure 7. Influence of bubble diameter on the time necessary to reach equilibrium for different liquid viscosities: (-) pL = 50 cP; ( - - - ) pL = 200 cP; and (--) pL = 1000 cP.
strates/water-whole bovine casein systems (see Table VI); and alcohol in viscous and foaming poly(ethy1eneglycol) system (Table V). Conclusion The new dilutor equipment is very well adapted to solute activity coefficient determinations in either viscous (up to 10oO cP) or foaming and viscous substrates. It is therefore an apparatus of great interest to the food industry where aromas are of high importance. Acknowledgment We are grateful to Prof. D. Simatos from E.N.S.B.A.N.A. and H. Renon from Ecole des Mines de Paris for their interest in this study. This work was performed with financial support of French Government: D.D.S.T.I., aid no 83.C.0166. Appendix Mass Transfer in the Equilibrium Cell. The stripping method gives accurate results only if the vapor phase which leaves the equilibrium cell is in equilibrium with the liquid phase contained in the cell. Then, the bubbles of the carrier gas have to be saturated with the solute during the time they spend inside the liquid mixture. A t a time t (far from equilibrium), the solute concentration around the center of a bubble can be represented by the scheme in Figure 4 where CL(t) is the solute concentration in the liquid, C t ( t ) is the solute concentration at the interface on the liquid side, C:(t) is the solute concentration at the interface on the vapor side, and C,(t,r) is the solute concentration in the vapor. The three first quantities are only time dependent while the last one is also dependent on the distance r , to the center of the gas bubble. It is a difficult task to get an accurate analytical expression for CG(t,r). Some simplifying hypotheses are needed; we used the following ones: the solute is considered as the only component which is exchanged between
1164
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985
the liquid and the gas bubble; the bubble diameter is constant; the modification of the liquid concentration is negligible while the gas bubble is in the solution supposedly perfectly stirred; and at the gas-liquid interface, equilibrium is achieved. According to these hypotheses, we can write
Sh =
($)Li21(Re)P2f2
where I(Re) = 2(4/3- 6.82Re)-lI2, Re = 2PLV"rb/pL,Sh = 2KL (rb/DL), and Pe = 2V"(rb/D~).v" is the limiting speed of the bubble V" = [(7.2 X 10-2)v~-o.6(2r~)'.6g]-1.4
where KL is the mass-transfer coefficient, rb is the bubble radius, and nb is the number of solute molecules inside the bubble. A t the interface, concentrations C&(t)and Ci(t) are related by the equilibrium relation A2 where p is the density
and M is the molar mass. For the diffusion in the gas phase, Crank ,(1956)gives the following relation which is only valid if Cl, is constant Cdt,r) = -Dij,GU2X2t 1 + -rbE "- ( - 1 ) U sin E ex.( rb2 ))CL (A3) Tr U = l u rb where D,,q is the diffusion coefficient of solute i in gas j . dnb is given by dt dt
dt
r2 dr
('44)
By combining eq A2 with eq A4, we get
where ff
= 1/KLrb2
and
P
PGML
= -- ymPLMG
The problem now is to solve eq A5 and then find t which satisfies eq A6. +Jrb
CG(t,r) r2 dr = 0.999flCL
rb
Several parameters have to be calculated in order to use eq A5 and A6. a. The diffusion coefficient Dj,, is estimated from the correlation of Slattery and Bird (1958) D.. = iJ,G
with a = 2.745 X and b = 1.823. b. The mass-transfer coefficient KL is estimated from the Cheh and Tobias (1968) equation relating three adimensional numbers
vL is the liquid kinematic viscosity. D, is the diffusion coefficient in the liquid phase; it is estimated from the correlation of Wilke and Chang (1955). c. Example of Calculation. The solute is propanone, the carrier gas is helium, and the solvent has the following properties: pL = 1; Pc = 100 atm, and Tc = 400 K. The activity coefficient of propanone in this solvent should be 10. Working temperatures and pressures are 298.15 K and 1 atm. Numerical solutions are obtained in order to define the influence of solvent viscosity and bubble diameter on the limiting bubble speed (Figure 5), the mass-transfer coefficient (Figure 6), and the time necessary to reach equilibrium (Figure 7). From Figures 5 and 6, it is shown that the viscosity influences the bubble speed more than the mass-transfer coefficient. Then, the most important problem occuring when increasing the viscosity is the long time spent by bubbles in the solution, limiting the carrier gas flow t o a very slow rate. To diminish the time spent by bubbles in the solution, an Archimede's screw is used in the equilibrium cell (Figure 1)to circulate the liquid from the bottom to top outside of the internal cylinder which surrounds the Archimede's screw. From Figures 5,6, and 7, it is seen that when the bubble diameter (2rb)is increased, (1)the time spent by bubbles in the solution diminishes, (2) KL diminishes, and (3) the minimal diffusion distance of the solute increases. In the actual design, the total length of the equilibrium cell was limited to 7 cm and the time to be spent by a bubble in the solution was set a t 4-5 s. The number of capillary injectors is 10 (internal diameter of capillary = 0.1 mm). Under these conditions, the solute transfer is achieved (vapor and liquid in equilibrium cell are in equilibrium) if the bubbles are less than 1mm in diameter. Such a diameter can be easily obtained for viscosities up to 1000 CPby independently adjusting the speed of rotation of the Archimede's screw and the carrier gas flow rate.
Nomenclature C = concentration, mol cm-3 D = diffusion constant, m2 s-l F = carrier gas flow (at pressure P, temperature T ) ,cm3s-l g = gravitational constant, cm H = Henry's constant, atm K = mass-transfer coefficient, cm s-l M = molar mass, g mol-' n = number of moles of solute N = number of moles of solvent P = total pressure, atm Pe = Peclet's number r = distance, cm R = gas constant, cm3 atm g-mol-' K-' Re = Reynold's number S = GLC peak area, arbitrary unit Sh = Sherwood's number t = time, s T = temperature K V = speed, cm W = weight fraction Greek Letters y = activity
coefficient
1165
Ind. Eng. Chem. Process Des. Dev. 1085, 2 4 , 1165-1 168
p
Duhem, P.; Vidal. J. Fluid Phase Equiillb. 1978,2 (31),231. Hala, E.; Wichterle, I.; Porak, J.; Boublik, T. "Vapor-Liquid Equilibrium Data"; Pergamon Press: London, 1968. Kojima, K.;Tochigi, K. "Prediction of Vapor-Liquid Equilibria by the ASOG Method"; Eisevier: New York, 1979. Lebert, A. Thesis Dissertation, Ecole Nationale SupCieure des Industries Agricoles et Aiimentaires, Massy, France, 1984. Lebert, A.; Richon, D. J. FoodSci. 1984, 49, 1304. Lenoir, J. Y.; Renault. P.; Renon, H. J . Chem. Eng. Data 1971, 16, 340. Leroi, J. C.;Masson, J. C.; Renon, H.; Fabrles, J. F.; Sannier, H. Ind. Eng. Chem. Process Des. Dev. 1977, 16, 139. Perry, J. H. "Chemical Engineers Handbook", 4th ed.: Mc Graw-Hill: New York, 1963. Richon, D.; Antoine, P.; Renon, H. ind. Eng. Chem. Process Des. Dev. 1980, 19, 144. Slattery, J. C.; Bird, R. B. AiChEJ. 1958,4, 137. Sorrentino, F. Thesis Dissertatlon, Ecole Nationale Superieure de Biologie Appiiquee 6 la Nutrition et B I'Alimentatlon, Dijon, France, 1984. Wiike, C. R.; Chang, P. AIChEJ. 1955, 1 , 264.
= dynamic viscosity, CP
v = kinematic viscosity, cSt p
= density
Superscripts i = at interface S = saturated property m = infinite property Subscripts b = bubble c = critical property i , j = component i and j G = gas L = liquid L i t e r a t u r e Cited Cheh, Y. H.; Tobias, C. W. Ind. Eng. Chem. Fundam. l9S8, 7 , 48. Crank, J. "The Mathematics of Diffusion"; Oxford University Press: London,
Received for review June 20, 1984 Accepted December 18, 1984
1956.
Comparison of CaO, ZnO, and Fe,O, As H,S Adsorbents at High Temperatures Motoo Yumura and Edward Furlmsky' Energy Research Laboratorh, Department of Energy, Mines and Resources, Ottawa, Ontario, Canada K1A OG I
CaO, ZnO, and Fe,03 oxides were compared as solid adsorbents for H,S removal from hot gas. Increasing the temperature from 600 to 800 OC increased the H,S removal in the presence of CaO but decreased it in the presence of Fe203. For ZnO, the temperature change had little effect on its adsorption. The bulk adsorption capacity was the largest for Fe,O, followed by CaO and ZnO. When the results were normalized to a unit of surface area, the adsorption capacity for ZnO was the largest followed by Fe,O3 and CaO. The adsorption of H2S was accompanied by its decomposition. I n the early stages, decomposition was the most extensive in the presence of CaO. With time on stream, the difference in H,S decomposition, measured by H, yields, became smaller.
Purification is an essential step prior to utilization of combustible gases. In this respect, the particulate matter and corrosive components must be removed to avoid detrimental effects on material and the environment. Among the latter, a great deal of attention has been directed to sulfur-containing species, where H2S is usually the most abundant and the most stable compound. The formation of H2S accompanies many industrial processes, e.g., pyrolysis, cracking, hydrocracking, and hydrorefining, in which H2Sis usually in the mixture with H2, CHI, and higher hydrocarbons. After H2S removal, these gases could be used as valuable fuels or as a source of petrochemicals. On a commercial scale, H2Sis removed by wet-scrubbing techniques. This step results in a loss of H2bound in H2S. It would be a significant achievement if H2S could be decomposed in order to reuse the H2. Inevitably, such a route would require a solid catalyst and temperatures higher than those applied in wet purifications processes. Another group of gases containing H2S are those from gasification. Under reducing conditions, such as those usually applied during moving bed and fluidized bed gasification, a substantial portion of S present in the feedstock is converted to H2S. The H2S removal is essential prior to utilization of gasification products either as fuel gas or as synthesis gas. In the former, the use in
combined cycle power generation has been thoroughly investigated. The present status of this route indicates that the purification at a near gasification temperature is required to make this route economically viable. Attempts to commercialize the combined cycle prompted the search for solid adsorbents which can efficiently and economically remove S-containing species. Most of the effort has been devoted to materials containing Zn, Co, and Fe. Among those CaO, ZnO, and Fe203have been frequently tested. However, a study comparing the performance of these oxides on a similar basis has not yet been published. The approach used in the present study involves a normalization of the H2S adsorption to a unit of surface area of the oxides. Experimental Section Materials. The CaO and ZnO oxides were certified
Fisher products, and Fe203was a reagent product of J. T. Baker Co. The powders of these oxides were pelletized by using 2% binder (stearic acid) to produce pellets of 1.5 mm in diameter and 1.5 mm high. To remove the binder, the pellets were roasted at 500 " C overnight. The mixture of N2 H2S (10 vol % of H2S)was of the UHP grade and was supplied by the Liquid Air Co. Apparatus and Procedure. The experimental system consisted of a vertical reactor made of Vycor glass (10-mm
+
0196-4305/85/1124-1165$01.50/00 1985 American Chemical Society
