76
Energy & Fuels 1987,1, 76-79
Use of X-ray Computed Tomography To Examine Microbial Desulfurization of Lump Coal Clifford L. Spiro,* David S. Holmes, John Lobos, and Donald H. Maylotte General Electric Corporate Research and Development Laboratory, Schenectady, New York 12301 Received June 5, 1986. Revised Manuscript Received September 25, 1986
The bacterial removal of pyritic sulfur from in situ deposits of coal deserves attention. The process has the advantages of economy and extended periods of time to carry out the sulfur removal. For example, some projected coal reserves are not expected to be mined for another 25-50 years. Possibly a culture and appropriate medium could be admitted to the seam and be allowed to slowly permeate the coal for sufficient time that significant desulfurization occurs. A fair assessment of the in situ desulfurization kinetics and efficiency has not been made for wont of an adequate assay. Therefore, as a preliminary step to examine the feasibility of in situ pyrite removal we have investigated the bacterial removal of the pyrite from lump coal by using X-ray computed tomography to follow the dissolution of pyrite inclusions. It is concluded that X-ray computed tomography is an effective tool to assay in situ phenomena. In particular, mixed cultures of Thiobacillus ferrooxidans and associated acidophilic bacteria were used to treat a Lower Kittanning seam high-volatile A bituminous coal. I t was demonstrated that pyrite inclusions can be removed at a depth of 17 mm within the lump over a 4-month period.
Introduction The removal of pyritic sulfur from pulverized suspensions of coal by microorganisms such as Thiobacillus ferrooxidans is well established.’P2 In the laboratory, up to 90% of the pyrite can be removed by bacteria under optimal conditions of pH, nutrient supply, and temperature.’-* Reaction rates can be enhanced by the use of finely ground coal, generally 200 mesh or less, and a maximum of 20% pulp density.’-4 Such reaction conditions are compatible with the emerging technology of coal-water slurries; hence, this has increased interest in the use of microorganisms as an alternative to the existing methods of pyritic sulfur removal. Another approach involves the use of microorganisms to remove or at least reduce pyritic sulfur from in situ deposits of coal. The rate of pyritic sulfur removal from in situ deposits does not need to be extremely fast since certain extensive coal reserves are not expected to be exploited for another 25-50 years. The crucial questions are how fast and to what extent the pyritic sulfur can be removed during this period and what factors affect these parameters. Obviously, the accessibility of the coal to aqueous solutions and the removal of the resulting acid mine drainage are of prime importance. In order to facilitate a feasibility study of in situ pyrite removal, we have explored the use of X-ray computed tomography as a potential candidate assay. A program to explore the use of X-ray computed tomography (CT) as a diagnostic tool for fossil fuels has been initiated in our laboratory. CT is of profound importance to the field of medical diagnostics, representing a marriage of advanced radiology with computing technology. With this noninvasive technique, a two-dimensional map of X-ray attenuation coefficients is obtained for a given plane (1) Dugan, P. R.; Appel, W. A. Metallurgical Applications of Bacterial Leaching and Related Microbiological Phenomea; Murr, L. E., Torma, A. E., Brierley, J. A., Eds.; Academic: New York, 1978; pp 223-250. (2) Andrews, G. F.; Maczuga,J. Biotechnol. Bioeng. Symp. 1982, No. 12, 337.
(3) Detz, C. M.; Barvinchak G. Min. Congr. J. 1979, 65, 75. Hoffman, M. R.; Faust, B. C.; Panda, F. A.; Koo, H. H.; Tsuchiya, H. M. Appl. Environ, Microbiol. 1981, 42, 259. (4)
0887-0624/87/2501-0076$01.50/0
within the object of interest. This is in contrast to a conventional radiograph which is a two-dimensional shadow graph obtained by projection through a three-dimensional object. In CT, images are not degraded by planes adjacent to the plane of interest, and therefore, higher resolution and low-contrast details are re~overed.~ The experiment consists of passing a flat, 1.5-mm-thick plane of X-rays through the object of interest. The X-rays are attenuated by the object, the extent of which is measured by a linear array of 517 gas-filled detectors. The entire array of detectors and the X-ray source rotate about the stationary object, and a second series of readings are made, each reading referred to as a “view”. Within 2-9 s, 576 such views are taken, each consisting of 517 individual X-ray attenuation data points. The images are reconstructed by tomographic mathematics by using a complex procedure whose description is beyond the scope of this text.6 The resultant image consists of an arry of “pixels”. Each 0.25 X 0.25 X 1.5 mm pixel has an x,y coordinate corresponding to its actual location within the plane of the object, as well as a 12-bit CT number, which measures the net X-ray attenuation coefficient within that specific region in space. A CT number of 0 means that all X-rays from every angle passed through that region of space unimpeded, while a CT value of 4095 represents a maximum X-ray attenuation. For reference, distilled water has a value of 1025. The X-ray tube spectrum ranges from 20 to 80 keV, so that the dominant mode of X-ray attenuation in coal is from Compton scattering. Because Compton scattering is dependent on average electron density and electron density is nearly a linear function of mass density for low atomic number elements, the CT technique represents a measure of mass density in coal to a high degree of accuracy. When CT numbers are plotted as a function of mass density for a broad range of organic liquids, a nearly straight line is ~ b t a i n e d . ~ (5) Newton, T. H.; Potts, D. G. Radiology of the Skull and Brain; C. V. Mosby: St. Louis, MO, 1981; Vol. 5. (6)Swindell, W.; Barett, H. H. Phys. Today 1977, 30, 32. (7) Maylotte, D. H.; Kosky, P. G.; Lamby, E. J.; Spiro, C. L. DOE
Annual Report, Contract DE1AC21-82MC19210, in press.
1987 American Chemical Society
Energy & Fuels, Vol. 1, No. 1, 1987 77
Microbial Desulfurization of Lump Coal Table I. Ultimate and Proximate Analysis of the Coal Used in This Study ?% anal. as received drv MAF Ultimate Analysis 1.32 0
moisture carbon hydrogen nitrogen chlorine sulfur ash
oxygen (diff)
68.21 4.28 1.03 0.1 1 1.72 19.30 4.03
69.12 4.34 1.04 0.11 1.74 19.56 4.09
Figure 1. X-ray computed tomographic images of lump coal incubated in control media no. 5 (Table 11), a t time zero (left) and after 4 months (right). The circled white spot is an iron pyrite inciusion, which is apparently unchanged after a 4-month incubation.
Proximate Analysis moisture ash volatile fixed carbon tot
2.32 19.30 26.50 52.88
0
100.00
100.00
Btu/lb sulfur
12 153 1.72
12316 1.74
19.56 26.85 53.59
1 15311 1900
Germane to this discussion is that pyrite buried deeply within the surface of lump coal can be detected, and by virtue of its high density, can be readily distinguished from the organic constituents of coal. Thus, this research was undertaken to determine the effectiveness of the CT technique as a means of analyzing the microbial dissolution of pyrite inclusions embedded in lump coal.
Experimental Section The coal used was a high-volatile A bituminous type from the Lower Kittanning nonmarine seam,Dean mine, Dean, PA. It was mined by the research staff and stored under anaerobic conditions. The coal was known to have large and extensive inclusions of iron pyrite. Composition analysis of the coal is shown in Table I. Epoxy was prepared and introduced into sterilized polyethylene vessels, which had been previously secured with aluminum registration wires. The skewed wires show up as points in the tomographic image, and can be used to ensure that the vessels are precisely realigned between scans. After the epoxy set about the aluminum, the set of flasks and lids were autoclaved to ensure sterility. A second layer of epoxy was prepared, and coals, presterilized by heating to 120 "C overnight, were anchored in the layer above the embedded registration wires. Handling was kept to a minimum to avoid contamination. The vessels were sealed until inoculation. 7'. ferrooxidans (ATCC 19859, ATCC 33020 and ATCC 13598) were obtained from the American Type Culture Collection. Growth media containing either bacteria or no added bacteria in the case of the controls was introduced into the vessels containing the embedded coal. A 1/10 volume of bacterial culture was used as an inoculum and contained initially approximately lo7 cells of T. ferrooxidans. Growth medium for the 7'. ferrooxidans was a standard minimal salts 9K medium, pH 2.2, with or without FeS04.9 X-ray computed tomography was carried out by usinga General Electric Model 8800 CT scanner. Equipment and technical assistance were generously made available by General Electric Medical Systems, Milwaukee, WI. Images were reconstructed and written to magnetic tape on a Data General S/200 computer (8) Lundgren, D. G.; Valkova-Valchanova, M.;Reed, R. Biotechnol. Bioeng. Symp. 1986,16,8-82,1986. (9) Silverman, M. P.; Lundgren, D. G. J. Bacteriol. 1959, 77, 642. (10) Dugan, P. R. Microbial Chemoautotrophy; The Ohio State University Press: Columbus, OH, 1980; pp 3-9. (11) Schnaitman,C.; Lundgren, D. G. Can. J. Microbiol. 1962,11,23. (12) Matin, A. Annu. Rev. Microbiol. 1978, 32, 433. (13) Harrison, A. P. Annu. Reo. Microbiol. 1984, 38, 265. (14) Grimes W. R. Coal Sci. 1982, 1,21. (15) Van Krevelen, D. W. Coal; Elsevier: Amsterdam, 1961. (16) Wu, S.-K.; Kispert, L. D. Fuel 1985,64,1681.
R 1800 I G H 17OO T I1600 M ;1500 E 1400 1300
1200 1300
I
I
1400
1500
I
I
1600 1700 LEFT IMAGE
1
1
1800
1900
I
I
2000 2100
Figure 2. Pixel-by-pixel plot of the encircled pyrite inclusion from Figure 1, in which CT values for each (x,y) location in the left image are plotted against the corresponding ( x ',y? image in the right image. The slope equals 1.0. augmented with a floating point system AP-12OB array processer and a General Electric "IRP" special processor. Magnetic tapes were transported to the research laboratory, and images were displayed and analyzed on a Lexidata display terminal, which arbitrarily assigned gray scales to CT values in order to maximize display and analyses. Typically, regions of high CT values were assigned in the white while lower CT values were assigned to black. Intermediate values were set according to a standard linear-array spectrum.
Results X-ray computed tomographic scans of coals containing numerous pyritic inclusions were taken immediately after inoculation with the biological growth medium with and without added .'2 ferrooxidans. Scans were also taken 4 months subsequent to inoculation. Figure 1 shows a CT image of a lump of coal incubated with the medium in the absence of 7'. ferrooxidans. The left-hand CT image was taken a t time zero, while the right-hand image was taken after a 4-month incubation without 7'. ferrooxidans. The lump of pyrite appears as a solid white area in both images (encircled in Figure l),indicating that no gross depletion of the pyrite inclusion has occurred. The identification of the lump as pyrite is based on the high CT number (4096), its morphology, and sectioning of order lumps from the same seam. It is our experience' that only calcite and pyrite inclusions in coal approach the maximum 4096 CT value, and calcite occurs along cleats in this coal, rather than in discrete lumps.
Spiro et al.
78 Energy & Fuels, Vol. 1, No. 1, 1987
cells
T. ferrooxidans, ATCC 19859 T . ferrooxidans, ATCC 33020 T. ferrooxidans, ATCC 13598 Thiobacillus acidophilus, ATCC 27807 Acidiphilium cryptum, ATCC 33463 Acidiphilium organovorum, ATCC 43141
Table 11. Media for Coal Studies" 1 2 3 4 9K/FeS04 9K/FeS04 9K/-FeS04 9K/FeS04
+ + +
+ + + + + +
5 9K/-FeS04
+ + +
6 NYPE
7 NYPE
+ + +
OLump coal was inoculated with one of the media (1-7) containing either bacterial cells (media 1,2,3,6) or no cells (media 4,5,7). A indicates the presence of a particular strain of bacteria. A full description of the media is given in the Experimental Section.
Figure 3. X-ray computed tomographic images of lump coal incubated with media no. 1 containing 7'. ferrooxidans (Table 11),a t time zero (left) and after a 4-month incubation (right). The rectangular images beneath the lump coal slices are 5 X magnifications of the region of interest. The white regions on the left image correspond to pyrite (arrow) and appear gray on the right image (arrow) due to the depletion of pyrite. The bar a t the bottom of the figure is the gray scale employed (see text).
Figure 2 shows a plot of pixel-by-pixel comparisons between the scans shown in Figure 1. Here CT values for each point (x,y) in the encircled spot on the time-zero image is compared with the corresponding( x ' j ? CT value after 4-month incubation. For two identical images, a straight line with slope = 1 is expected. Indeed, the resultant plot shows nearly a straight line with a slope = 1.0. Slight deviations from unity are likely to be the result of minute misalignment effects associated with removing the sample and replacing it with precision to the identical location. However, since essentially featureless image resulted from the digital subtraction of images taken at time zero and 4 months, it demonstrated that acceptable realignment of the samples has been accomplished. Incorrect realignment does not result in featureless difference images but yields a scan consisting of brightened features and adjacent shadows. Correct realignment is crucial for the interpretation of the data. Our experimental design allows alignment to be made within a half-pixel, reducing, though not completely eliminating, the possibility of misinterpretation of results. In contrast to the results just described, we observed significant dcpletion of large pyritic inclusions when the lump coal was incubated in the medium with added T. ferrooxidans and its energy source FeS04. An example of the dissolution of a large pyritic inclusion buried 17 mm below the surface of a lump of coal in media containing T. ferrooxidans and added FeS04 is shown in Figure 3. The left-hand image coinsists of data taken at time zero, while the right image was taken after a 4-month incubation. For clarity, the pyritic inclusions have been magnified 5 times and appear as white spots and lighter regions in the rectangles beneath the lx magnification images above. The spots in the left-hand image appear much lighter and more solid (arrow, Figure 3), while the corresponding region in the right-hand image (arrow, Figure 3) includes gray throughout. This indicates that for the left image, there
+
LEFT IMAGE
Figure 4. Pixel-by-pixel plot of the regions outlined in Figure 3, for the coal incubated in media no. 1,containing T.ferrooxidans. CT values for the various ( x , y ) locations in the left image are plotted vs the corresponding (x'y?values in the right image. The resultant two lines in the figure correspond to a region that was unaffected by the medium (slope = 1) and another that was depleted of pyrite (slope = 2.0).
was considerably greater X-ray attenuation, a likely result of microbial activity. Finally, a quantitative determination of the pyritic dissolution can be made from the pixel-by-pixel comparison shown in Figure 4. If the two images were identical, a straight line with slope = 1 would result, as in Figure 2. Were the images random, or even shifted by only three or more pixels with respect to one another, the resulting plot of left vs. right would show uninterpretable scatter. The two lines that appear together in Figure 4 are best interpreted as consisting of one region that was unaffected (slope = l),and a second region corresponding to the pyritic inclusion that showed a significant loss of material (slope = 2.0). it is noteworthy that in the absence of FeS04, no depletion of the inclusion was observed. This may be fundamental to the process, i.e. necessary to enable sufficiently rapid growth of T. ferrooxidans within the time frame of the experiment. on the other hand, it may be that this particular sample had insufficient porosity for ions or organisms to contact the inclusion. Clearly, a more thorough experimentalmatrix is warranted, though outside the scope of this effort.
Discussion
7'. Ferrooxidans is an aerobic, chemoautotrophic bacterium that grows in extremely acidic environments (pH 1.0-3.0) such as acid coal mine waste or mineral ore leach dumps. It oxidizes ferrous iron or reduced forms of sulfur including mineral sulfides to form sulfuric acid. The energy derived from this oxidation is used to assimilate carbon dioxide into all the necessary cell carbon via the Calvin Cycle.
Microbial Desulfurization of Lump Coal
Energy &Fuels, Vol. 1, No. 1, 1987 79
It is important to note that the most effective removal of pyrite occurred when the coal sample was incubated in the presence of T. ferrooxidans and its energy source FeSO,. T. ferrooxidans catalyzes the reactiona 2FeS04 + ' / , 0 2+ HzSO4
(bacteria)
Fe2(S04)3+ H,O (1)
The resulting Fe+3ions of the ferric sulfate can then oxidize the pyrite1' according to FeS, + 2Fe3+ 3Fe2+ 2S0 (2)
- + -
T. ferrooxidans is also capable of direct oxidation of pyrite: 2FeSz + 71/,02
+ H20
(bacteria)
Fe2(S04),+ H2S04 (3) Undoubtedly, reaction 3 accounts for some of the dissolution of pyrite because such dissolution was observed in the absence of added FeS04. However, since the extent of dissolution is much greater in the presence of added FeS04, we believe that indirect oxidation by eq 1and 2 is the predominant means of the observed dissolution of the pyrite inclusions. It is well known that bituminous coals have an extensive micropore system.14 These pores are up to 1.2 nm in width, although bottlenecks in the pores may be only 0.4-0.5 nm.15J6 Clearly, T. ferrooxidans cells, which are 0.5 X lo3 nm wide by 1 X lo3 nm long could not penetrate the micropores. I t is not known whether Fe3+ions with appropriate counterions could penetrate the micropore system. It is possible that the larger pyrite inclusions observed by the CT scanning method are associated with more open spacings, cracks or cleats, permitting access to solventcarrying Fe+3 or, in certain cases, even T. ferrooxidans cells. For example, Zwietering and van Krevelanls reported 25% of the pore volume in a British coal had pore radii greater than lo3 nm, probably associated with cracks. In an extensive study of pore distribution, Gan et al.lg reported pore volumes of between 11.9% and 87.77% greater (17) Olsen, T. M.; Ashman, P. R.; Torma, A. E.; Murr, L. E. Biogeochemistry of Ancient and Modern Enuironments; Australian Academy of Science: Canberra Australian, 1980; p 693. (18)Zweitering, P.; van Krevelen, D. W. Fuel 1954,33,331. (19) Gan, H.; Nandi, S.P.; Walker, P. L., Jr. Fuel 1972, 51, 272.
than 30-nm radii, strongly dependent on coal character. An alternative explanation is that, in the neighborhood of surfaces or planes, the T. ferrooxidans cells either directly or indirectly (via production of Fe+3)establish a galvanic effect. This in turn effects that redox chemistry of the more inaccessible pyrite inclusions, promoting oxidation of the pyrite. The (CT) technique, described in the paper, has a spatial resolution of less than 0.09 mm3 and can readily resolve density differences less than 0.03 g/cm3. For species with strong photoelectric absorption such as iron, silver, or any number of high atomic number elements, sensitivity increases dramatically with mass adsorption coefficients. Thus pyrite in coal or silver in aluminosilicate matrices are ideal candidates for the technique. In addition, with scan times of 9 s, slower kinetics are readily determined.
Conclusions X-ray computed tomography provides a unique and valuable, nondestructive assay for microbial activity well beneath the surface of lump coal. Depletion of pyrite from an inclusion 17 mm beneath the surface of the coal was observed within 4 months of inoculation with T. ferrooxidans. Depletion of pyrite does not occur in the absence of T. ferrooxidans. Although T. ferrooxidans in the absence of an added energy source can decrease the size of pyrite inclusions, the most effective treatment for pyrite removal is T. ferrooxidans together with its energy source FeS0,. This would suggest that pyrite dissolution is occurring predominantly by indirect oxidation due to Fe+3ions that are produced by bacterial catalysis. The possibility of galvanic effects on the pyrite should also be considered. Acknowledgment. This research was sponsored by the
U.S. Department of Energy, Contract DE-AC2182MC19210. The authors greatly appreciate the use of the facilities at General Electric Medical Systems and the technical assistance of Chuck Shiley, Herb Peters, and Ed Lamby. We would like to thank Carolyn Meyer and Fran Dohring for their help in preparing this manuscript. The authors express appreciation for officials at the Dean Mine, Dean, PA, for providing coal sampling access and assistance.
