curves and MPD schedules discussed in Cases 1 and 2 used 0.75 for the stability parameters, If in actuality they are really 0.8, then all of the SO2 concentration curves are too high and the MPD schedules somewhat in error, This case was included to vividly point out the importance of using a dispersion model which has been tailored to describe a given area over a given period of time.
Comlusion The MPD algorithm was used to determine the meteorologically sensitive MPD schedules for the three plants whose SO2 emission rates were determined from the emission rates of the individual generators in each plant. MPD does appear to provide quantitative information about plant dispatching schedules for alleviating a pollution episode, given an accurate, deterministic, dispersion model for the area being studied. Since the MPD algorithm was developed to be compatible with present-day dispatch computer capability, it could be easily implemented as an alternate dispatching strategy in the event of a pollution episode. In addition, it could be a valuable aid to system planners in making decisions concerning future power plant sites. Because the MPD algorithm depends on a deterministic dispersion model, it is mandatory that the meteorological parameters in the model be obtained in the area being modeled and that they be updated when necessary. The accuracy and effectiveness of MPD are no better than the dispersion model used. Since the work described in this paper represents an initial effort, it should be obvious that many refinements could and should be made to the basic approach. Some of the refinements are: The availability of power from pool members should be included so that load shedding may be avoided, with the amount and source being determined based on lowest dollar cost. The environmental cost function in Equation 1 could be modified to reflect the auerage concentration over a given area rather than reflect the concentration only a t one point. Minimization techniques other than the one used here should be considered.
Nomenclature L = environmental cost function or total SO2 concentration a t a given point in ppm X = downwind distance in feet Y = crosswind distance in feet PGl = electrical power out of generator i in per unit p G l m a x = maximum electrical power out of generator i in per unit PGlmin = minimum electrical power out of generator i in per unit Plead = total electrical load to be supplied in per unit t L = ground-level SO2 concentration due to the output of the i t h generator in ppm M , = meteorological coefficient for ith generator C, = horizontal diffusion coefficient C, = vertical diffusion coefficient U = mean wind speed in ft/sec X i = downwind distance in feet Y , = crosswind distance in feet Z, = effective stack height Q i = SO2 emission rate from the ith generator in ft3/sec K , = conversion coefficient Greek Letters = Lagrange multiplier p,min
p p x
= Kuhn-Tucker dual variable = Kuhn-Tucker dual variable
Literature Cited Bibbero, R. J., “System Approach Toward Nationwide Air Pollution Control,” IEEE S p e c t r u m , November 1971, pp 73-81. Corn, M.. Monteomerv, T. L.,“Adherence of SO2 Concentrations in the Viciniti of a s t e a m Plant to Plume Dispersion Models,” J a p c a , 17,512 (1967). Federal Register. Environmental Protection Aaencv. Vol. :36. Fri~” day, April 30, 1971. Sullivan, R. L.. “Minimum Pollution DisDatchine,” IEEE DaDer . . C72468-7, 1972. Wells, C . H., Lau, R. W. J., “Stochastic Modeling and Control of Ambient Air Quality: A Kew Approach,” Systems Control Inc., Palo Alto, Cal., Weston Technical Papers, 1971.
Received for r e v i e u J a n u a r y 2, 1973. Accepted J u n e 25, 1973. Work supported by the Engineering and Industrial E x p e r i m e n t S t a t i o n IEIES). L‘niivrsit?’ of Florida.
Absorption of Sulfur Dioxide by Molten Carbonates Robert A. Mcllroy The B a b c o c k Wilcox Co., Alliance, Ohio
Glenn A. Atwood’ D e p a r t m e n t of Chemical Engineering, The University of Akron, Akron, Ohio 44325
Coleman J. Major College of Engineering, The University of A k r o n , Akron, Ohio 44325
The pollution of the atmosphere by sulfur dioxide is one of the greatest air pollution problems facing the country today. The United States Department of Health, Education, and Welfare, Public Health Service, estimated that in 1963 more than 23 million tons of SO2 were discharged into the atmosphere and by 1980, the discharge will be more than double unless efficient control methods are develoDed and put into wide scale use (Squires, 1967). The To whom correspondence should be addressed. 1022
Environmental Science & Technology
largest single source of sulfur dioxide pollution is the fossil fuel-fired power boiler.
Molten Carbonate Process One approach to the control of sulfur dioxide emission is the absorption of sulfur dioxide by molten carbonates. This process appears to have certain advantages over other suggested processes. Since the absorption medium is a liquid material, transport problems would be considerably simpler than in the solid absorption processes. The absorption would take place a t a relatively high tempera-
Absorption of sulfur dioxide by molten carbonates is one of the many processes proposed to reduce atmospheric pollution by fossil-fueled power plants. In this work the absorption of sulfur dioxide by a ternary eutectic mixture of sodium, potassium, and lithium carbonates was studied using three small absorbers. A synthetic flue gas was bubbled through a pool of molten salt, and the concentration of sulfur dioxide in the effluent was measured. The fractional absorption could be correlated by the equation
ln
[$I
tration, t is the superficial contact time, DIM is the diffusivity of sulfur dioxide in the flue gas, and R is an effective drop radius which is dependent on the reactor geometry. The equation was developed for transfer from a stagnant bubble of constant volume. This model applies as long as the carbonate fraction is greater than 0.3. The coefficient
P%l L
=
A -
~~D,.ii t
where p L ois the inlet concentration, FL is the exit concenture, 800-1100°F, which requires that part of the conventional boiler cycle be located after the absorption system. As a result, the flue gas would be discharged from the unit a t its normal temperature and humidity with no less of buoyancy. The molten carbonate process for a modern coal-fired power boiler is shown in Figure 1. Flue gas from the boiler enters the absorber a t a temperature of 800-~100'F. In the absorber, sulfur dioxide from the flue gas reacts with the molten carbonates according to the reactions
SO?+ M2S03 M2SO3 + CO, SO?+ M2C0,j-t X 0,e M2S0, + CO,
(1)
(2)
where M stands for the metallic ions. The flue gas, essentially free of SO2 is returned to the boiler to pass through the conventional economizer and air heater. The absorption solution is transferred to a regeneration system where a reducing agent, such as producer gas, regenerates the carbonate. For the producer gas the chemical reactions are
+ 2CO + H2 M&O, i- H2S + CO, (3) M2S0, i- 2CO + 2H2 6M&O, + H2S + CO, + H20 M,SO,
-.
4
ranged from 0.2 to 0.8 and over 97% of the sulfur dioxide was absorbed as long as the contact time was greater than 0.015 sec. cess is operated between 800' and 1000"F, before solidification of the mixture becomes a problem. All experiments in this work were performed between 800" and 1100"F, with the majority being run near 1000°F. The physical properties of the molten carbonate eutectic have been evaluated by Janz (1967) a t 1000°C. The density, viscosity, and surface tension of the eutectic are 1.926 g/cm3, 4.33 cP, and 211.8 dyn/cm, respectively. These properties indicate that the eutectic will be easily transported. No information was found for the equivalent sulfate and sulfite systems.
Experimental Apparatus and Procedures The experimental apparatus used in the study is shown schematically in Figure 2. A synthetic flue gas was bubbled through a pool of molten carbonate eutectic mixture contained in a reactor located within an electrically heated furnace. The reactor consisted of a gas-liquid disengaging section, a removable carbonate chamber, and a gas inlet probe. The reactor (Figure 3) was constructed from 316 stainless steel. Three different carbonate chambers were used to vary the cross-sectional area and superficial gas velocity through the molten carbonate pool. The important dimensions of the chambers are given in Table I.
(4) The hydrogen sulfide can be converted to elemental sulfur or sulfuric acid by conventional processes. Botts and Oldenkamp (1972) and Oldenkamp (1969) describe the process, describe a pilot plant for the process, and present an economic evaluation of the process. The Atomic International Division of North American Rockwell Co. has been granted patents on several variations of the basic process (Grantham, 1967; Grantham and Larsen, 1967; Heredy, 1967; Heredy et al. 1969). The work described in this paper was undertaken to determine the conditions under which the absorption of sulfur dioxide by molten carbonates is feasible. Since no previously published work was found for the absorption stages, it was decided to study the effect of conversion, temperature, and contact time on the absorption of SO2 using a number of small absorbers.
Molten Carbonate, Molten Sulfate Properties The melting points of pure carbonates are too high for them to be a practical absorption medium for conventional boiler design. The ternary eutectic of sodium, potassium, and lithium carbonate was chosen because of its low melting point, 397°C (747°F). The melting point of the sulfate system corresponding to the composition of the ternary carbonate eutectic is 640°C (1184°F) (Lavin, 1964). Therefore, a high degree of conversion for the sulfates should be possible if the pro-
-
Absorber MtCOs+ S O I + 1/20)
Bakr
-MISO,+
cooc
Cot
-
Regenerator
MISO,+tCO+rHt
M$O.
-MtC03+ C ~ * H ~ O * H , S Producar Goa
"Dl5
",,h
na.n,*r 6 0 , CII#"dt!
l,,h R I g Y I ~ 1 m
Gar
131 .
Pl...
owp,
"~,",,
no.
Convol
BIWk
Supply System
Figure 2.
And
Reector 8 Furnace System
GOS P ~ O I ~ SSystem ~S
Schematic diagram of experimental apparatus Volume 7 , Number 11, November 1973 1023
Table I. Reactor Dimensions Chamber A
0
C
0 . d . in.
Wall thickness, in.
Cross-sectional flow area, ft2
0.625 0.875 1.250
0.065 0.065 0.095
10.99 x 10-4 27.86 X 58.80 X
The molten carbonate chamber, including the union was located entirely within the vertical Hoskin's tubular furnace. The temperature was controlled by a 5-amp Variac. The disengaging section of the reactor vessel consisted of the reactor, an enlarged air-cooled section which supported the reactor. The enlarged section was provided to disengage any liquid from the pool entrained in the gas stream and to allow any alkali salts vapor to condense. The top of the reactor could be water cooled to further cool the exit gas and protect the analysis instrument. Water cooling was not required during these experiments. Gas was introduced through a 0.375-in. 0.d. tube placed in the center of the reactor. The probe entered the reactor through a Conax packing gland on the top plate which allowed the probe position to be controlled%.The gas probe was reduced to 0.188-in. 0.d. just above the absorption chamber to maximize the free area in the lower chamber. The inlet tip was located 0.125 in. above the bottom of the carbonate chamber during all of the experiments. The temperature of the molten carbonate pool and the flue gas were recorded using 0,063-in. diameter unground-
Figure 3.
Reactor vessel with three molten carbonate chambers
ed chromel-alumel sheathed thermocouples. All thermocouples entered the reactor through a Conax packing gland a t the top of the gas inlet probe. The thermocouple in the melt and gas a t the bottom of the reactor passed from inside the probe a t the point where the transition from a 0.375-in. to a 0.188-in. tube occurs. The synthetic flue gas for the study was supplied from premixed cylinders. It contained 15% carbon dioxide, 2% oxygen, sulfur dioxide (three different concentrations were used), and nitrogen. The pressure to the system was maintained a t 20 psig. The gas flow rate was controlled using a calibrated rotameter and needle valve. The gas leaving the reactor was split with a small portion going to the sulfur dioxide analysis instrument, and the rest was vented. The concentration of sulfur dioxide in the exit gas was monitored using a Barton Electrolytic Titrator, Model 286. This instrument (Thoen et al., 1968) is basically a bromine microcoulometric cell. For each run the reactor vessel with a weighed quantity of mixed carbonates in the lower chamber was heated to the reaction temperature. Nitrogen was purged through the system a t the flow rate which would be used for the flue gas. The test was initiated by rapidly shutting off the nitrogen block valve and opening the synthetic flue gas block valve. The test conditions were maintained for 2-3 hr unless the exit sulfur dioxide concentration reached the upper limit (2500 ppm) of the Barton detector. Thermodynamics The sulfur dioxide vapor pressure over the carbonatesulfate melt must be extremely low to ensure that equilibrium will not give an unacceptably high lower limit on the concentration of sulfur dioxide in the exit gas and thereby limit the removal efficiency of the process. A detailed thermodynamic study of the process is not possible due to a lack of data for some compounds. Both sulfites and sulfates (Equations 1 and 2) are possible reaction products since normal boiler flue gas will contain 2-670 oxygen (dry volume bases). Unfortunately, no reliable thermodynamic data have been found for the three sulfites or for lithium sulfate. Rosen 1960, however, has estimated the heat capacity of NaZS03 and used it to estimate the Gibbs free energy. Thermodynamic data for KzS04 were obtained from the work of Rossini (1952) and Kelly (1960). The data for all other compounds were from the JANAF Thermochemical Tables (Stull, 1965, 1966, 1967). The equilibrium constants for the following reactions of sulfur dioxide with the constituents of the carbonate melt are shown in Figure 4.
Na2COj + SO, + X02 e NaSO,
+ CO,
+ SO? + %O, K,SO, + C02 Na,CO, + SO2 e Na2S0, + CO,
K,CO,
constants for reactions of SO2 with molten
carbonates 1024
(6)
(7)
If the sodium carbonate reactions (i.e., Equations 5 and 7) are used as representative of the entire melt, the equilibrium sulfur dioxide vapor pressure over the melt can be estimated. The equilibrium constant for Equation 5 is given by
Log K
Figure 4. Equilibrium
(5)
Environmental Science & Technology
At 1000"F, the constant has a value of about The equilibrium SO2 partial pressure is 1.07 X atm or about parts per million (vppm), assuming that the ratio of the activities of the two liquid species is unity and
that the carbon dioxide and oxygen concentrations are 15% and 2%, respectively. The COz and 0 2 concentrations in the flue gas are representative of those used in this work. The equilibrium constant for Equation 7, however, is only about 5 X lo2 a t 1000°F. Using the same assumptions as above gives a n equilibrium sulfur dioxide concentration of about 300 vppm for the sodium sulfite reactions. The large positive value of log K for Equation 5 in comparison with that of Equation 7 indicates that the sulfate will predominate when equilibrium is achieved, and that the exit SO2 concentration will be less than 0.1 ppm. On the other hand, the maximum SO2 concentration a t equilibrium could be on the order of 300 ppm if considerable sulfite were present. Since Equation 5 can be interpreted as a n addition of Equation 7 and a sulfite oxidation reaction, an interpretation of the results of this study in terms of reaction mechanism appears warranted.
Reaction Mechanisms Although the overall stoichiometry of the absorption of SO2 by molten carbonates is expressed by Reactions 1 and 2, a number of reactions must be considered in order to develop a model for the absorption process. If we use the notation ( d ) to indicate a gas dissolved in the melt and (1) to indicate the ionic constituents of the liauid melt, the
LW I D GAS
Figure 5 . Gas absorption
model
ide absorption can be developed. The following simplifying assumptions were made in developing this model: (1) The bubbles were noncirculating spheres with an average radius, R . ( 2 ) The volume of the bubble is constant. (3) The reaction a t the gas liquid interface can be represented by Reaction 14. (4) The reaction occurs instantaneously on the surface. Assumption 3 means that for each mole of SO2 diffusing to the liquid surface, 1 mole of COz is liberated and diffuses into the bubble. Hence, the diffusion can be considered as equimolar counterdiffusion. The assumption of an instantaneous surface reaction is supported by the work of Ruthven and Kenney (19671, who studied the absorption of oxygen by molten copper chloride. In the reactor, the bubble rises through a t most 2 in. of fluid. For a n exit pressure of 1 a t m , this would cause a bubble volume change of less than 1% which confirms Assumption No. 2. Writing a material balance for sulfur dioxide over a differential volume in the bubble (Figure 5) and assuming spheiical symmetry give: _ d(rN _, ) - r*c
dr
[$I
Substitution of Fick's first law of diffusion and the ideal gas law into Equation 20, assuming a constant diffusivity gives:
with the boundary conditions
Insufficient information is available to choose which of the above equations controls the absorption of SO2 by the carbonate. However, the work of Ruthven and Kenney (1967) and the limited physical solubility of gases in high temperature melts indicate that diffusion Reactions 9 and 17 are of limited importance since in normal boiler flue gas only a small amount of SO3 is present, about 1% of the sulfur oxides. Oldenkamp (1969) indicates that sulfite is the major reaction product with only about 10% sulfate formation. He states that Reactions 13 and 15 are relatively slow compared to the absorption reactions. Thus, based on the information in the literature, the rate-limiting mechanism is probably the one described by Reaction 14. This is followed by the conversion of sulfite to sulfate (Reaction 15). The formation of sulfate following Reaction 13 is probably very slow.
p,(r,O) = p,O
(22)
p , ( R , t )= 0
(23) (24)
The solution of Equation 21 with this set of initial and boundary conditions is
(25) This equation gives the variation of the concentration with respect to both the radial position and time. The experimental values of concentration are the average for the bubble so that ,Equation 25 must be integrated to obtain the volume average Concentration. The volume average concentration of a species in the spherical coordinates is given by
Model f o r Mass Transfer in Reactor If one assumes that the flue gas mixture bubbles through the pool of molten carbonates in small spheres, a mathematical model to predict the amount of sulfur diox-
(26)
Volume 7,
Number 11, November 1973
1025
Table 11. Summary of Experimental Data and Results Initial M 2 C 0 3 Run no 2 3 4 5 6
a 9 10 11 12 13 14 15 16 17 18 19
Chamber A A A A
A A A
A A A A B B C B C C
Weight, gram 5.0 5.0 5.0 10.0 5.0 5.0 5.0 5.0 5.0 10.0 5.0 12.6 12.6 30.0 12.6 30.0 30.0
Depth, in. 1 1 1 2 1 1 1 1 1 2 1 1 1 1 1 1 1
Temp of melt, O F 1000 1095 ai0 1010 1010 1005 1010 990 1000 1000 1015 1010 1015 1010 1020 985 1015
Gas flow, std I./m 2 2 2 2 3 2 3 4 1 2 0.5 1 .o 2.0 4.4 4.0 1.5 3.0
Substituting Equation 25 into Equation 26 and integrating give
The series approaches r 2 / 6 as t approaches zero so t h a t the model satisfies the initial conditions. For the exposure times used in this work, the series converges rapidly. If one retains only the first term of the series, Equation 27 can be rearranged to give (28) Equation 28 indicates that the logrithm of the fraction remaining is proportional to the residence time. Discussion of Experimental Results The test conditions and results are summarized in Table 11. The results obtained during this exploratory experimental investigation of the absorption of SO2 by molten carbonates have been correlated by Equation 28. The ratio of the steady state exit concentration to inlet SO2 concentration vs. the contact time .between the gas and molten carbonate is shown in Figure 6. The contact time is defined as the depth of carbonate divided by the superficial velocity where the superficial velocity is the gas flow rate a t the operating temperature and pressure of the system divided by the empty chamber cross-sectional flow area for the respective chamber as previously defined. The straight lines of Figure 6 show t h a t the model can be used to correlate the experimental data. A separate straight line was required for each carbonate chamber. This means that the value of r2D,.v/R*, is a function of the chamber dimensions. As the free flow area increases, as shown in Table 111, this term decreases. This can be interpreted to mean that the effective bubble radius in the reactor increases with increasing flow area which is reasonable. It is significant that the concentration of SO2 leaving the reactor was lower than 30 ppm for runs 5 , 11, 12, 13, 14, 15, 18 and 19 which would indicate that the major form of sulfur in the melt must be the sulfate, based on the previously calculated equilibrium sulfur dioxide vapor pressures which is in direct conflict with Oldenkamp's (1969) statement that sulfite is the major product. This would indicate t h a t the contact times of 0.05 sec mini1026
Environmental Science 8 Technology
Superficial velocity, ft/sec 3.09 3.29 2.67 3.14 4.83 3.14 4.73 6.19 1.55 3.09 0.78 0.61 1.23 1.27 2.47 0.43
0.85
SO2 concn. Yo
Inlet 0.2675 0.2675 0.2675 0.2675 0.2675 0.4064 0.4064 0.4064 0.4064 0.4064 0.4064 0.4064 0.4064 0.4064 0.2620 0.2620 0.2620
Exit (av) 0.0087 0.0081 0.0065 0.0005 0.0071 0.01 10 0.0166 0.0302 0.001 2 0.001 6 0.0005 0.0002 0.0010 0.0077 0.0090 0.0004 0.0026
Rate of absorption, g mol/m X 2.31 2.32 2.34 2.39 3.49 3.54 5.24 6.75 1.81 3.61 0.90 1.79 3.57 7.88 4.50 1.73 3.46
mum was apparently sufficient to allow Reactions 13 and 15 to occur to a significant extent or that the estimated thermodynamic properties upon which the prior estimates were made are in error. Therefore, the thermodynamic data for the sodium, potassium, and lithium sulfates and sulfites should be obtained and then a further kinetic study should be undertaken to determine the reaction mechanism. Although the depth of molten carbonate was accounted fdr by tht contact time as defined above, there is apparently a minimum depth required for good contact. At the lowest carbonate depth used, 0.5 in. in Run N o . 7, a steady state SO2 absorption rate was never obtained. It is believed t h a t the gas flow, 5.82 liters per minute a t the reactor temperature, was too high for the shallow pool and permitted channeling of the gas through liquid This means that there was little if any contact between the gas and the carbonate which invalidates the model given in Equation 28. The temperature of the molten carbonate can be varied from the melting point about 750°F up to the practical operating limits of about 1200°F. Temperature had no significant effect on the SO2 absorption under the conditions studied, nominal temperatures of 800, 1000, and 1100"F,as seen by comparison of Runs 2, 3. and 1 in Table 111. This would be expected for those systems in which the gas phase resistance is controlling. Typical results in terms of the rate of sulfur dioxide absorption, mol/min., are shown as a function of time in Figure 7 for chamber A . These curves are characterized by a short initial period of falling absorption rate, a long period in which the rate is essentially constant, and a terminal period in which the rate of absorption decreases rapidly. The high absorption rate during the initial period is probably the result of dilution of the sulfur dioxide in the sampling train with residual purge gases, so that analyses during this period are probably in error. Therefore, this section of the curve should be ignored. The rapid drop in the absorption rate a t the end is probably caused by a change in the rate-controlling mechanism. If one looks a t runs 8, 9, 10 (Figure 8), it is noted that the rate of transfer is a function of the amount of carbonate reacted. Figure 8 shows t h a t as long as the mole fraction of carbonate in the melt is greatei than 0.3, the model presented in Equation 28 can be used to predict the exit sulfur dioxide concentration. When the carbonate concentration is less than 0.3, the model cannot be used to predict the exit sulfur
Table 111. Effect of Free Flow Area on T‘DAB/R* Chamber A 0
C
aZDas/R2
0.789 0.436 0.196
Free flow area, ft2 10.99 x 10-4
Wall to injector distance, ft2
27.86 X 56.86X
0.0128 0.0243 0.0363
dioxide concentration. In this region, the diffusion of carbonate to the gas-liquid interface may become the controlling rate step rather than the diffusion of sulfur dioxide from the gas, which was the basis for the model. Thus it appears that the practicable upper limit for operation would be in the region of about 70% conversion of carbonate to sulfate and sulfite, if one wants to minimize the size of equipment required to reach a given sulfur dioxide removal. Conclusions
5
0
-
-
\
\ A
0 ” -
v)
\
t 0.01
Contact Time, sac
x 10’2
Figure 6. Effectof contact time on gas concentration
The following conclusions can be made based on the results of the exploratory investigation. The absorption of sulfur dioxide by molten carbonates in the reactors used in this study follows the model represented by the Equation 28. The c,ontrolling absorption mechanism changes when the melt contains less than 0.3 mole fraction carbonate. The sulfur in the melt appears to be present as the sulfate rather than the sulfite, provided the estimated thermodynamic properties are realistic. The rate of sulfur dioxide absorption is very rapid with 97-99.570 being absorbed with contact times in the range 0.015 to 0.10 sec. The use of molten carbonate for removal of SO:, from boiler flue gas appears to be a practical process with commercial potential. Acknowledgment The authors wish to acknowledge the support of the Babcock Wilcox Research Center in providing laboratory equipment and facilities for the experimental portion of the study and allowing the results to be published. Nomenclature
= molar concentration of the ith species, mol/l. = diffusivity of component i in a gas mixture, L 2 / t N , = molar flux of the ith species mol/LZt pi = partial pressure of the ith species F/L2 r = radialpositionL R = average bubble radius L t = time yi = mole fraction of the ith species c,
DiM
Greek Letter i~ = 3.14159
Literature Cited Astarita, G., Ind. Eng. Chem., 58, (8), 18 (1966). Astarita, G., Gioia, F., Chem. Eng. Sci., 19, 963 (1964). Botts, M. U., Oldenkamp, R. D., “The Molten Carbonate Process Time ( m i d
Figure 7. Rate of SO? absorption in chamber A
Figure 8. Effect of carbonate conversion on rate of absorption
for Sulfur Dioxide Removal from Stack Gases,” Process Description Economics, Pilot Plant Design, presented a t APCA 65th annual meeting, Miami, Fla., Paper 72-115, June 18, 1972. DeNevers, N., A.I.Ch.E. J., 14 ( 2 ) , 222 (1968). Fair, J . R., Lambright, A. J., Anderson, J. W., Ind. Eng. Chem. Process Des. Deuelop., 1,34 (1962). Grantham, L. F., “Two Stage Regeneration of Absorbent for Sulfur Oxides,” U.S. Patent 3,438,728, May 15, 1967. Grantham, L. F., Larsen, C. M., “Sulfur Production Using Carbon Regenerant,” ibid.,3.438,733, May 15, 1967. Heredy, L. A , , “Absorbent Regeneration Using Carbon Regenerant,” ibid.,3,438,727, May 15, 1967. Heredy, L. A,, McKenzie, D. E., Yoxim, S. J., “Removal of Sulfur Oxide from Flue Gas,” ibid.,3,438,722, April 15, 1969. Hodgeman, C. D., ed., “Handbook of Chemistry and Physics,” 36th ed., p 1610, Chemical Rubber Publishing Co., Cleveland, Ohio, 1955. Janz, G. J., “Molten Carbonate Electrolytes as Acid-Base Solvent Systems,” presented at Symposium on Fused Salt Reactions, Toronto, Canada, February 1967. Kelly, K. K., U.S. Bur. Mines, Bull. 584, p 148, 1960. Lavin, E. J., RoPPjns, C. R., McMurdie, H . F., “Phase Diagrams for Ceramists, American Ceramic Society, Columbus, Ohio, 1964.
Volume 7,Number 11, November 1973 1027
McIlroy, R. A,, “Absorption of Sulfur Dioxide by Molten Carbonates,” M S Thesis, The University of Akron, Ohio, December 1969. Oldenkamp, R. D., “Development of the Molten Carbonate Process For Control of Sulfur, Oxide Emissions,” presented a t AIChE National Meeting, Cleveland, Ohio, May 5-7, 1969. Rohrmann, F. A., Ludwig, J . H., Chem. Eng. Progr., 61 (9), 59 (1965). Rosen, G., “Data and Calculations for Gasification of Spent Cooking Liquors from the Pu,l,p Industry, Part 1: Fundamental Thermodynamic Quantities, Trans. Roy. Insti. Technol., Stockholm, Sweden (159). 1960.
Rossini, F. D., Wagman, D. D., Evans, W. H., Levine, S., Jaffe I., Nat. Bur. Std. Circ. 500, p 496 (1952). Ruthven, D. M., Kenney, C. N., Chem. Eng. Sci., 22,1561 (1967). Squires, A. M., Chem. Eng., 74 (24) 260 (November 6, 1967). Stull, R. D., Ed., JANAF Thermochemical Tables, the Dow Chemical Co., Midland, Mich., August 1965; First Addendum August 1966; Second Addendum August 1967. Thoen, G. N., DeHass, G. C., Austin, R., TAPPI, 51, 6, 246 ( 1968).
Received for review January 24, 1973. Accepted July 27, 1973
Adsorption of Hg(ll) by Hydrous Manganese Oxides R. Addis Lockwood and Kenneth Y. Chen’ Environmental Engineering Programs, University of Southern California, Los Angeles, Calif. 90007 The adsorption of Hg(I1) by hydrous manganese oxides suspended in solutions with wide ranges of pH, chloride concentration, and ionic strength was studied to evaluate the possible importance of manganese oxides in mercury scavenging in natural waters. Manganese dioxide and possibly other oxides were precipitated with the addition of M n S 0 4 solution to KMn04 solution. Low crystallinity was verified by X-ray. Hg(I1) was adsorbed rapidly when added to aged suspensions of MnOz with low ionic strength. The equilibrium values in the pH range 6-8 fit a Freundlich isotherm:
0.6M NaCl repressed adsorption below pH 9, but not above pH 10. Apparently 0.6M NaC104 changed equilibrium values only slightly, but adsorption rates were many orders of magnitude lower, particularly above p H 8. The uncharged Hg(OH)2 is the adsorbed species. HgC12 was also adsorbed, but not so strongly as Hg(0H)z. It is concluded that MnOz may be important as a mercury scavenger in fresh or brackish water, either in the natural environment or in the treatment of water and waste water. The enrichment of manganese nodules with rare metals, including mercury, has been reported by various authors. Krauskopf (19561, after observing the results of laboratory adsorption studies and calculating solubility product limitations, stated that at least 13 of the metal concentrations in the sea are controlled by adsorption. Furthermore, of the various adsorbents, manganese dioxide appears to be the most effective, with iron oxide to a lesser degree. Though mercury was not included in his manganese dioxide experiments, the appearance of mercury in manganese nodules suggests that manganese oxide may be effective for scavenging mercury (Goldberg, 1963). According to Krauskopf‘s (1956) estimate, between 0.01 and 0.07% of the amount of mercury entering qceans has remained in the water. The role of manganese in removing mercury from ocean water seems to be important (Harris, 1968). Manganese, though not as abundant as iron, is nevertheless nearly ubiquitous in its appearance in natural waters. Jenne
T o whom correspondence should be addressed. 1028
EnvironmentalScience 8 Technology
(1968) suggested that the activities of earth materials in adsorbing various metals might be localized on manganese and iron oxide sites. Early work in this laboratory indicated that clay materials of nearly pure aluminum silicate are poor adsorbing agents for mercury, but material containing iron showed greater adsorptive activity. The adsorption of unhydrolyzed cations on manganese dioxide has been attributed to ion exchange in the surface of the precipitate (Posselt et al., 1968). However, mercury is different from the divalent metals used in many adsorption studies because above pH 3 the uncharged metal oxide hydrate Hg(0H)z is the dominant species. Also, mercury is easily reduced to the elemental state, which makes the problem more complicated if the redox potential is not maintained a t a high value to prevent the formation of the reduced forms. The main purpose of this work was to evaluate the role of hydrous manganese oxide in the scavenging and transport of mercury in water. The degree to which these solids remove Hg(I1) from water solution was studied using pH, ionic strength, and sodium chloride concentration as variables. The 25°C isotherm function for the adsorption of aqueous mercury by MnO2 solids was also developed.
Experimental Equipment. Temperature was maintained at 25 f 0.5”C in a constant temperature and humidity environmental chamber. The Perkin Elmer Model 124 doublebeam spectrophotometer was used for manganese analysis. The Unicam Model SP90 atomic absorption spectrophotometer was used in the flameless mode with a 15-cm path quartz absorption cell for the mercury analysis. pH was determined with a Corning Model 110 pH meter using a Corning glass electrode and an Orion double-junction standard calomel electrode. Suspensions were kept agitated with an Eberbach shaker operating a t about 120 strokes per minute. Reagents. All reagents were analytical reagent grade or primary standard grade except for sodium perchlorate, which was Fischer Scientific purified grade. The analytical reagent red mercuric oxide used as the mercury standard was dried at 110°C and stored over calcium chloride. The hydroxylamine hydrochloride contained mercury. Solutions were sparged overnight with nitrogen to remove the mercury before use in the analytical procedures. All other reagents were checked for mercury and were free of interfering quantities. Sodium hydroxide was rendered carbonate-free by centrifuging the saturated solution.
