sulfur emission rates. Adjustment of our model for the anticipated low-emission desert areas will tend to reduce the sulfur emissions projected by our model which does not contain sulfur emission measurement input from deserts. A third factor not considered in our model is the biogenic sulfur emitted from the surface of the open oceans. Emissions from this source would increase our projected global biogenic sulfur burden. Work by Junge ( 2 5 )in North America, Lodge (26) in the North Pacific, Lovelock in the Atlantic ( 2 7 ) ,and Cadle (28) in Antarctica all indicate the possible contributions of gaseous sulfur from the ocean. Acknowledgment Special thanks are extended to A1 Stankunas and Charles Hakkarinen, EPRI project officers, and to the many persons and organizations providing field sites and electrical power. Reference to commercial products is for identification purposes only and does not constitute an endorsement of these products by EPRI or the University of Idaho (RP856-2) or Washington State Univerity (RP856-1). Literature Cited (1) Junge, C. E.; Chagnon, C., W.; Manson, J . W. J. Meteorol. 1961, 18,81. (2) Mueller, P. K.; Hidy, G. M. “Implementation and Coordination of the Sulfate Regional Exueriment (SURE) and Related Research Projects”, Interim Repori EPRI No. EA-1066, EPRI Project No. RP-862, June 1979. (3) Adams, D. F.; Farwell, S. 0.;Robinson, E; Pack, M. R. “Biogenic Sulfur Emissions in the SURE Region”, Final Report EPRI No. EA-1516, EPRI Project No. RP-856-1, Sept 1980. (4) Russell, T. S.; Washington State University, Pullman, WA, personal communication, 1976. ( 5 ) Farwell, S. 0.;Gluck, S. J.; Bamesberger, W. L.; Schutte, T. M.; Adams, D. F. Anal. Chem. 1979,51,609. (6) Barr, A. J.; Goodnight, J. H.; Sall, J. P.; Helwig, J. T. “A Users Guide to SAS 76”; SAS Institute: Raleigh, NC, 1979. 17) Tauber. H. “The Chemistrv and Technoloev -“ of Enzvmes”: Wilev: New York, 1949. ( 8 ) “Climatic Atlas of the United States”: U S . DeDartment of Commerce: Washington, DC, 1968.
(9) Delmas, R.; Bandet, J ; Servant, J.; Baziard, Y. J:Geophy. Res. 1980,85,4468. (10) Hill, F. B.; Aneja, V. P.; Felder, R. M. J . Enuiron. Sci. Health 1978.13. 199. (11) Duncan, D. B. Biometrics 1955,31,1. (12) Kramer, C. Y. Biometrics 1956,32,307. (13) Jaeschke, W.; Georgii, H. M.; Claude, H.; Malewski, H. Pure A P d GeoDhvs. 1978.116.463. (14)’ Aneja, V. P. M.S. Thesis, North Carolina State University, Raleigh, NC, 1975. (15) Aneja, V. P. In “Atmospheric Deposition: Environmental Impact and Health Effects”; Ann Arbor Science Publishers: Ann Arbor, MI, 1980; Chapter 7. (16) Hansen, M. H.; Ingrown, K.; Jorgensen, B. B. Lirnnol. Oceanogr. 1978,23,66. (17) Maroulis, P. J.; Bandy, A. R. Science 1977,196,647. (18) Bandy, A. R.; Maroulis, P. J. In “Atmospheric Sulfur Deposition: Environmenal Impact and Health Effects”; Ann Arbor Science Publishers: Ann Arbor, MI, 1980; Chapter 8. (19) Hitchcock, D. R.; Spiller, L. L.; Wilson, W.E. “Biogenic Sulfides in the Atmosphere in a North Carolina Tidal Marsh”, paper presented a t the American Chemical Society meeting, New Orleans, LA, March 1977. (20) Steudler, P. A.; Peterson, B. J. “Gaseous Sulfur Release from a Salt Marsh”, paper 80-40.3, presented at the Air Pollution Control Association meeting, Montreal, Quebec, June 1980. (21) Hitchcock, D. R. “Biogenic Sulfur Sources and Air Quality in the United States”, Final Report NSF RANN No. AEN-7514571, Aug 1977. (22) Eriksson, E. Tellus 1960,12,63. (23) Granat, L.; Rodhe, H.; Hallberg R. In SCOPE Report 7, Ecol. Bull. 1976,89. (24) “Smithsonian Meteorological Tables”; Smithsonian Institute: Washington, DC, 1965; Vol. 114. (25) Junge, C. E.; Werby, R. T. J . Meteorol. 1958,15,417. (26) Lodge, J. P.; MacDonald, A. J.; Vihman, E. Tellus 1960, 12, 184. (27) Lovelock, J. E.; Maggs, R. J.; Rasmussen, R. A. Nature (London) 1972.237.452. (28) Cadle,R. D.; Fisher, W. H.; Frank, E. R.; Lodge, J. P. J . Atmos. Sci. 1968,25,100.
Received for review April 21,1981. Accepted August 17,1981. This work was primarily supported by the Electric Power Research Institute under contract RP856-1-2.
Role of Fly Ash in Catalytic Oxidation of S(IV) Slurries Sidney Cohen, Shih-Ger Chang, * Samuel S. Markowltz, and Tihomir Novakov Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720
The rate of catalytic oxidation of S(1V) in aqueous fly-ash slurries has been examined by monitoring loss of SOz2- and 0 2 , and S042- formation. Evidence from both filtering and extraction studies indicates that dissolved iron which has been leached from the ash is the chief catalyst. This homogeneous reaction dominates any catalytic effect of the ash surface.
Introduction The catalytic behavior of fly ash in the oxidation of S(1V) in aqueous systems has been reported (1-3). Prior work has considered the potential of fly ash as a scrubber medium ( 2 , 4 ) and suggested how fly ash may function in promoting acid rain formation in droplets ( 3 ) .Although several properties of the ash-surface characteristics, metal oxide content, tracemetal dissolution---have been proposed to account for its catalytic role, no systematic investigation of the nature of the catalysis has been made. It is important to distinguish between surface and homogeneous catalysis-for example, presence of chelating agents in wet scrubbers containing fly ash should inhibit catalysis by leached metals ( I , 5 ) . 1498
Environmental Science & Technology
In this study the oxidation of S(1V) in slurries of fly ash has been examined after different ash pretreatment and under varying experimental conditions. Our evidence indicates that dissolved iron is a prime factor in fly-ash-catalyzed oxidation of s02. Experimental Section
Reagents. Mallinckrodt or Baker reagent-grade chemicals were used without further purification. The fly-ash types were NBS Standard Reference Material 1633a and four lot samples, referred to as A-D. A is from the TVA Shawnee Power Plant; B, the Duke Power Steam Plant; C, the Jim Bridger Plant, Wyoming; and D, an untreated ash sample obtained from Lawrence Livermore Laboratory. Nanopure water from a Barnstead Deionizer was used throughout. S(1V) Oxidation. The net reaction being studied was S042- H+. The proportional quantities HS03- l/202 of HS03-, S02aH20, and S0s2- vary with pH, so knowledge of the working pH is essential. Unless otherwise stated, the reaction pH was 3.1 f 0.1. Reactions were monitored by using three different methods: (1) in open-air Erlenmeyer flasks agitated with a magnetic stirrer, S(1V) loss being monitored
+
-
+
0013-936X/81/0915-1498$01.25/0
@ 1981 American Chemical Society
by iodometry; (2) reaction setup as above, except S032- loss and S042- appearance monitored by ion chromatography; (3) in a closed system-allowing no diffusion of 0 2 into>solution during the course of the reaction-with 0 2 decay from initially saturated solutions being monitored with a Clark-type oxygen electrode. The initial S(1V) concentration in the reactions studied M. The pH was varied varied from 9.0 X 10-4 to 1.5 X from 2.5 to 3.9 by using appropriate mixtures of S02.H20 and NazSz06. The fly-ash concentration was varied from 0.10% to 0.36% by weight. Most rates reported are the average of 2-3 runs agreeing within 10%. Leaching Experiments. All glassware and containers were acid washed before use and rinsed 5-6 times with deionized water. Leachate solutions were stored in Nalgene linear polyethylene containers before analysis. For the leaching experiments, 2 g of ash was suspended in 100 mL of the appropriate solvent and subjected to 20 min of ultasonic agitation. In the NH20H.HCl extraction, the solutions were shaken over a period of 2 h while pH equilibration occurred. The duration of treatment for the other solutions was chosen after a separate set of experiments revealed that soluble surfaceassociated species (S042-,CaO, As2O3) are almost entirely removed after only 5-10 min of treatment. Longer treatment periods in dilute (0.02 N) acid slowly solubilizes the matrix species (Fe, AI, Si),indicating breakdown of the aluminosilicate framework. Concentrations of the leaching solutions were as follows. Acid concentrations varied from 0.02 to 1.8 M. HC1, HzS04, and H N 0 3 all affected the ash reactivity similarly, indicating there is no anion or concentration effect within the bounds studied. Base concentrations were either 0.05 or 1.5 M NaOH. The NH20H.HCl concentration was 0.1 M, and the pH was adjusted to 2.9. When extracted ash was to be used in subsequent experiments, it was dried at room temperature and then stored in sealed containers. Equipment and Materials. Elemental and chemical compositions of fly ash were characterized by X-ray fluorescence (XRF), atomic absorption (AA), and electron spectroscopy for chemical analysis (ESCA). AA work was performed on a Perkin-Elmer Model 360 atomic absorption instrument. The XRF procedure, designed to minimize matrix and particle-size effects, has been described elsewhere (6).A 0.22-pm pore size Millipore filter was used to filter solutions for leachate analysis and reaction aliquots. This pore size was chosen as the maximum size which appears to trap all ash particles to which metals might absorb (7). ESCA work was done on an AEI ES200 instrument with an A1 anode. A position-sensitive photoelectron detector was installed in our lab. Samples were mounted on the probe with double-backed sticky tape. Binding energies were referenced to the hydrocarbon C(1s) peak at 284.5 eV. Peak growth could be monitored continuously on a real time display and plotted when the signal-to-noise ratio was deemed sufficient. The analytical lines used for Fe and Mn were the 2P1/2/2P3/2 doublets. Results and Discussion Care was taken to measure the true reaction rate, unrestricted by 0 2 mass transfer into solution. The results described here underscore the need for such caution. For the S(IV) concentration range from 2 X to 4 X M, reaction curves were consistent with zeroth-order dependence on [S(IV)] when reactions were run in open-air Erlenmeyer flasks (i.e., plots of [S(IV)] vs. time were linear). Correlation coefficients on the straight-line fits were generally 0.98 or higher. Some tailing was observed a t the lowest concentrations. It has been noted that mass transfer of 0 2 into solution can be rate limiting for S(IV) consumption in the range of our experiments (8).Use of a fritted-glass gas-dispersion bubbler
to ensure air saturation effected no significant rate change compared to a similar run using only the magnetic stirrer. However, the zeroth-order dependence described above, coupled with almost identical rates being observed for the four most active ash types (Table I), prompted further consideration of this problem. SO2 evolution or precipitation of Cas03 could conceivably provide a S(1V) rate loss mimicking zeroth-order rate behavior. Ion chromatography work, showing formed, material balance between SO32- lost and eliminated this possibility. To avoid entirely the problem of 0 2 saturation, we used the oxygen electrode monitoring system described in the Experimental Section. The rates, which slowed as the reaction proceeded, were extrapolated to zero time in order to make them representative of the reaction rate in an air-saturated solution. Table I shows that this clearly defines a reactivity scale for the ashes and increases rates for all but ash type B. The rate increase indicates that mass transfer of oxygen had previously been a problem. The slower rate for ash type B is not surprising because initial rates (first 3-4 min of reaction) were generally slower regardless of the monitoring setup used. This phenomenon will be discussed more fully below. Role of Surface in Catalysis. T o determine the relative importance of surface-mediated or solution-phase reactions, we filtered a reaction slurry after the reaction was partially complete. Loss of sulfite was monitored in the filtrate; as can be seen in Figure 1, the reaction rate is not affected by removing particulate matter. Because the catalysis is apparently due to trace metals which can be leached from the ash, an attempt was made to remove such species from the ash surface by leaching with different solvents. Leaching with deionized water or strongly
Table I. Comparison of the Catalytic Activity of Different Ash Types io5 (rate measured with iodometrlc technique) a NBS A B
no reaction 5.5 5.8
C
6.2 5.0
D a
io5 rate measured by 0 2 b uptake 1.92 12.0
3.6 8.0 6.8
-d[S(IV)]/dt (mol/(Lmin)),zeroth-order rate. Converted to -d[S(IV)]/dt
(mol/(L rnin)).
,
I
I
I
I
4.Oq
1 -\
10
0
3'5r'L,
I
r
1
Reaction slurry1
filtered after 19 minutes
1 1
Time (rnin) Figure 1. Comparison of rate of S(IV) oxidation before and after filtering reaction slurry. Initial pH is 2.9; [D fly ash] = 0.18%. Volume 15, Number 12, December 1981
1499
basic solution did not affect the ash reactivity. However, an acid wash in 0.02 N acid completely deactivated the ash. Metals commonly implicated in SO2 oxidation-Cu, Fe, Mn, Co, V-can be wholly or partially removed by an acid wash (depending on their chemical state), so it is not surprising that washing with acid deactivates the fly ash. Identification of Catalytic Species. To identify which of these transition metals were most responsible for the catalysis, we carried out XRF analysis for two ash types: NBS fly ash (only sparingly active toward S(1V)oxidation) and type D (catalytically active). Results are displayed in Table 11. Surprisingly, the inactive ash contains a greater concentration of all five catalysts in the dry matter. Therefore, the bulk metal content is not a good measure of catalytic reactivity. More important is the metal solubility, which could vary widely among ash types as it depends upon both chemical state and location in the ash particle (Le., surface or matrix). Because the catalysis by fly ash depends on metal dissolution, leaching studies were deemed more relevant to this work. Manganese and iron are the two catalysts present in reaction slurries at concentrations high enough to dominate the reaction rate. Theis and Wirth (9) found that leaching with hydroxylamine hydrochloride a t pH 3 removed most of the soluble Mn, leaving (on average) all but 2% of the total Fe undissolved. We used this treatment to determine which of the two catalysts-Fe or Mn-controls the reaction. Table I11 confirms that the hydroxylamine wash more selectively removes the extractable Mn. In addition, an interesting difference between the catalytically active (D) and catalytically inactive (NBS) ashes is detected; the NBS ash contains only l/lo the amount of soluble Mn that the catalytically active ash does. Inspection of Tables I1 and I11 reveals that much of the Mn in the NBS ash is in an insoluble form, either a higher oxide or an interstitial cation in the aluminosilicate matrix.
Because of the nonuniformity of fly ash, we felt it important to involve several ash types in relating rates to any specific ash property. In Table IV Mn and Fe content of filtered reaction slurries and hydroxylamine extracts are compared to reaction rates. Although the NHZOH-HCl treatment did affect ash reactivity, reducing it by a factor of 5-10, the correlation between reactivity and Mn content of solutions is not good. Fe concentration shows a much better correlation with reaction rates. As the hydroxylamine wash was intended to remove Mn selectively, these results seem contradictory: if the NHzOH. HC1 wash reduces ash reactivity, the implication is that Mn and not Fe is the responsible catalyst. The contradiction is easily explained if one considers the following: iron in the ash is quite insoluble. It occurs predominantly in the fly-ash matrix, rather than in an easily leached surface site (IO). A small amount of Fe must be easily and rapidly leached from the ash, as it is found in both the filtered reaction slurries and the NH20H.HC1 wash. It is entirely possible that a small fraction of the total Fe is present in a soluble form-perhaps as a surface-associated ferrous or ferric salt. The fact that the reaction reaches its maximum rate within 3-4 min of initiation indicates that the species responsible for the catalysis is rapidly leached from the ash. Although a higher oxide such as Fez03 should not be reduced by the NHZOH-HCl wash at pH 3, it is likely that the small fraction of Fe which is readily soluble is catalytically important. In support of this hypothesis, it was found that removing Fe from solution reduced the reaction rate. When the pH of reaction slurry filtrates was increased to pH 4.5, a rust-colored precipitate (Fez03.H2Ox) formed, which was filtered off. The freshly filtered solutions had significantly lower catalytic activity. In a further set of experiments, we hoped to increase the Fe concentration, thereby augmenting the reaction rate. Unfortunately, stirring reaction slurries over long time periods-doubling the Mn and Fe concentrations-did not always increase the reaction rate. Chemical saturation ( I I , I 2 ) is one possible explanation for this behavior. Hudson (11) and Matteson et al. (12) found a zeroth-order dependence on Fe3+ and Mn2+concentrations, respectively, when these metal ions increase to a certain value. pH Studies. The pH of the reaction mixture is important in a t least two respects. Metal release is strongly pH dependent; and the relative fractions of SOz-HZO,HS03-, and S032are functions of pH. Many studies implicate S032- as the re-
Table 11. XRF Comparison of Type-D and NBS Ashes concn, ppm element
D ash
NBS ash
Mn
200 f 50 25000 f 2000
