- Email: [email protected]
Manganese removal from acid coal-mine drainage by a pond containing green algae and microbial mat
~
Pergamon
0273-1223(95)OO503-X
Wal. Sci. T,c!L Vol. 31, No. 12, pp. 161-170, 1995. CopyrighlC 1995 lAWQ Pnnled in Greal Britain. AU rights reserved.
0273-1223,'95 $9'50 .. 0-00
MANGANESE REMOVAL FROM ACID COAL-MINE DRAINAGE BY A POND CONTAINING GREEN ALGAE AND MICROBIAL MAT P. Phillips*, J. Bender*, R. Simms*, S. Rodriguez-Eaton* and C. Britt** • Research Center for Science and Technology, Clark Atlanta University. Atlanta. GA 30314. USA ** Tennessee Valley Authority, 601 West Summit Hill Drive. Knoxville. TN 37902. USA
ABSTRACT Acid mine drainage flowing from an underground seep, through an anoxic drain to an oxidation pond and then trickling filter was distributed among three treatment ponds for manganese reduction before releasing the mining water to a surface stream. Manganese was more effectively removed from the integrated green algae and microbial mat with limestone substrate pond (algae mat system) than from control ponds containing limestone or pea gravel substrates without mat. Although there was some Mn-cell binding, manganese was primarily deposited as precipitates at the pond bottom. Day/night and winter/summer manganese removal was essentially the same. The greater efficiency in Mn removal by the algae mat pond was most pronounced with higher flow rates and during the night. At 2 metres from the influent point of each pond, the algae mat pond removed 2.59 g manganese/day/M2 , respectively, compared to 0.80 in the limestone pond and 0.37 in the pea gravel pond. Control ponds showed either Mn breakthrough or near breakthrough (Mn outflow releases > US Environmental Protection Agency regulations of 2 mg/L) during night-time sampling or when mine drainage flow exceeded 4.5 L/min. KEYWORDS Acid mine drainage; blue-green algae; cyanobacteria; manganese; metal removal; microbial mats. INTRODUCTION Manganese removal from acid mine drainage is a challenge due to the solubility of manganese sulfide and the high pH required to precipitate manganese as an oxide or carbonate. Therefore, it is common to find drainage with manganese above state or US Environmental Protection Agency (US EPA) standards (Gordon et al., 1989). The Tennessee Valley Authority utilizes constructed wetlands technology to treat acid mine drainage. These wetlands have generally been effective in removing Fe (0.4-21.3 g/day/m 2 of wetlands (GOM» and Mn (0.15-1.87 GDM) (Brodie, 1993). At one site within the Fabius Coal Mine in northeast Alabama, USA, Mn levels are approximately 8 mg/L after leaving an oxidation pond and before draining toward an extensive constructed wetland. At this point a pilot-scale field test was conducted to determine if an 161
162
P. PHILLIPS el a/.
integrated microbial (dominated by cyanobacteria or blue-green algae) and green algae mat would effectively remove residual manganese in a very small pond surface area. Microbial mats are natural heterotrophic and autotrophic communities dominated by cyanobacteria (blue-green algae). They are self-organized laminated structures annealed tightly together by slimy secretions from various microbial components. The surface slime of the mats effectively immobilizes the ecosystem to a variety of substrates, thereby stabilizing the most efficient internal microbial structure. Since mats are both nitrogen-fixing and photosynthetic (Paerl et aI., 1989), they are self• sufficient, solar-driven ecosystems with few growth requirements. In our laboratory, we have developed microbial mats, constructed with specific microbial components, for a variety of bioremediation applications (Bender and Phillips, 1993). Mats have been found to reduce selenate to elemental selenium (Bender et al., 1991a), remove Pb, Cd, Cu, Zn, Co, Cr, Fe and Mn from water (Bender, 1992~ Bender et a1 .• 1991b. in press) and to remove Pb from sediments (Bender et al., 1989). Degradation of recalcitrant organic contaminants has also been observed under both dark and light conditions (Bender and Phillips, 1993). The following contaminants have been degraded in water and/or soil media by constructed mats: TNT (Mondecar et aI., 1992), chrysene, naphthalene, hexadecane, phenanthrene (Phillips et al., 1993), PCB (Bender, 1993), TeE and the pesticides, chlordane (Bender et al., 1993) , carbofuran and paraquat. Radio-labeled experiments with mat-treated carbofuran, petroleum distillates and TeE show that these three compounds are mineralized by mats and mat products, such as biofilms and bioflocculents. Recently,our data has confirmed that the mats effectively treat mixtures of organics and heavy metal. Mats simultaneously sequestered Zn and mineralized TCE and chrysene. Research currently in progress shows that U 238 can be removed from ground water samples. This project was designed to examine the feasibility of applying mats in a field remediation pilot project. The broad goals were to assess the performance of mats under environmental conditions, such as determining seasonal survival and the efficiency of the mats in removing metals under day-night conditions. Comparisons in Mn removal were made among the three ponds~ algae mat pond with limestone substrate, a pond with limestone substrate and a pond with pea gravel substrate, these latter two without mat. Specific experiments in the present study compared the effectiveness of the algae mat and limestone pond compared to controls of limestone or pea gravel alone in removing manganese. METHODS Three 40 m2 ponds were constructed by first lining with PVC. Two ponds had limestone substrate and one had pea gravel substrate layered to create four rises separated by five troughs. Maximum trough water depth was 1 foot. Rises nearly broke the water surface. Mine drainage flowed from an oxidation pond to a the trickling filter, both designed to precipitate Fe(OH)]. From the trickling filter, one pipe fed all three ponds (Figure 1). Each pond was sampled for manganese at 6-7 points. The ponds operated continuously from August 1992-March 1993 and from June 1993. Cyanobacteria_algae Mat Pond (CGM) A cyanobacterial-dominated microbial mat is an entire ecosystem containing several bacteria species, but dominated by cyanobacteria, in the
Manganese removal from acid coal
163
OVERFLOW I
ANOXI~ DRAIN
OXIDATION POND
C C
A
1F
A
LOS B
C
D E
COM B
C
D E
~ ~
Fig. 1. Oxidation pond~ TFI trickling filter~ INI influent water~ A-E, additional sample points for manganese~ CGMI cyanobacteria-algae mat pond, LOSI limestone/oscillatoria pond~ PGOSI pea gravel/Oscillatoria pond. Not drawn to scale. multilayered mat structure. In this pond, microbial mat became enmeshed with volunteer green algae, thus forming an integrated green algae and microbial mat (CGM). Microbial strains, including Oscillatoria spp. and purple bacteria (likely a complex mixture of sulfur and non-sulfur species) were selected from the Alabama site during February 1992. Microbial mats were developed from these strains according to Bender and Phillips (1993). A microbial mat slurry (blended in water) was broadcast over the experimental pond at a rate of 1-1.5 L in three applications during a four-week period. Silage, prepared from grass clippings, added organic acids and several species of ensiling bacteria to the water. Within two months a thick green mat covered the entire pond surface. Underneath this floating layer, the limestone itself was covered with a heavy green coating of cyanobacteria. Limestone/Oscillatoria Pond (LOS) Although no biological material was intentionally added, CGM pond cyanobacteria spread into this pond. Limestone was covered with a thin «1 rom) film of cyanobacteria. Approximately 10% of the pond surface became covered with green algae within two months. The green algae died during December 1992. Precipitated iron covered the entire limestone substrate of the pond including the pond periphery beyond the effluent pipe. Pea Gravel/Oscillatoria Pond (PGOS) The pea gravel substrate in this pond also become thinly «1 mm) covered with cyanobacteria. Additionally, there was a rapid buildup of
164
P. PHILLIPS el at.
precipitated iron at the effluent pipe by December 1992. Flow Rate Initial flow was set at 1 L/min and incrementally increased over a period of two months. Flow was adjusted on a daily basis due to clogging by Fe(OHh· Ranges of flow rates (L/min) were 0.8-6.0, with the following mean values for each ponda CGM - 4.2, LOS = 3.8, PGOS = 2.8. Since Mn could be removed chemically in all ponds, by increasing the flow rate, we expected to reach a break point whereby the two ponds without substantial biological material would no longer remove manganese as efficiently as the CGM pond. Water Chemistry Water samples were collected at the oxidation pond, the trickling filter and at five points within each pond in a horizontal profile from influent to effluent (Figure 1). If an effluent sample could not be effectively collected, the last point within the pond was considered the effluent sample. In the field, influent and effluent water samples were tested for pH (Orion GX Series pH electrode; Orion 200 Series portable pH and Eh meter), Eh (Corning platinum redox combination electrode; Orion 200 Series portable pH and Eh meter), conductivity (Fisher Scientific Digital conductivity meter), temperature and dissolved oxygen (Otterbine Sentry III dissolved oxygen meter) • For metal determination, water samples were filtered (0.45 um cellulose acetate) and stabilized for transport with HNO J • These were later analyzed for manganese by atomic absorption spectrometry (Varian, Spectra AA-20 BQ, double beam). Reliability of manganese analysis was regularly verified by comparing with a synthetic solution containing 20 mg/L Mn and diluted with field pond water. RESULTS A floating mat (1-2 cm thick), composed of filamentous green algae mat and cyanobacteria, predominantly Oscillatoria spp., grew well in the CGM pond after addition of silage/mat inocula. A secondary mat of cyanobacteria also covered the limestone at the bottom. Thus the metal-contaminated water flowed between the double-layered mats. Approximately six weeks were required to establish a full pond mat cover that lasted nearly one year, but effective metal removal began in the early stages of mat growth. The thin layer of cyanobacteria covering the substrate of the two ponds designed as controls were colonized with an oscillatoria strain resembling that of the inoculated CGM pond cyanobacteria. Small amounts of green algae produced a floating mat, but the biomass remained low compared to the experimental pond. Through February 1993, the CGM pond maintained a viable algae mat. An approximately 1m2 bleached area (indicating cell death) developed during this time. Beneath this bleached algae, limestone was covered with viable cyanobacteria. Iron was evident in patches over the entire surface area of the pond, yet there were also large areas of viable green algae and microbial mat over the pond surface. In contrast to the control ponds, there was no ViSible evidence of iron precipitates in the pond periphery beyond the effluent pipe.
Manganese removal from acid coal
165
In March 1993, after a 50-em snow fall, the mat appeared to be decimated after the snow thawed. This may have been due to wash-out or snow shading. A snail infestation was also observed at that time. It is possible that snails were always present as part of the balanced ecosystem of the mat pond, but in the normal conditions of a healthy, highly productive algal biomass, their effect was irrelevant. The algae mat pond was drained and re-inoculated in June (metal deposits were left in the troughs). Mat regenerated and the pond is currently operating. Table I summarizes daytime water quality parameters from August 1992-April 1993. Separately, dissolved oxygen values for the single nighttime sample weres CGM, 4.2 mg/L; LOS, 6.1; and PGOS, 6.8. The low night-t~me value for the algae mat pond is likely due to high oxygen consumption by the biological material. TABLE I
T(sd)* CGM LOS PGOS
MEAN DAYTIME WATER QUALITY PARAMETERS PER POND
DO(sd)
14.1(7.6) 7.3(1.3) 14.7(8.4) 8.3(1.8) 14.8(8.5) 8.2(1.9)
pH(sd)
ORP(sd)
Cond(sd)
Alk(sd)
Flow(sd)
7.4(0.4) 7.6(0.4) 7.2(0.5)
413(56.4) 424(73.8) 410(74.4)
648(136.7) 618( 94.7) 580( 92.3)
167(50.8) 168(53.2) 156(49.1)
4.2(2.4) 3.2(1.0) 2.2(1.6)
*Ta temperature, celsius1 001 dissolved oxygen, mg/L, ORPa oxidation-reduction potential, mV1 Conda conductivity, umbo/cm, Alka alkalinity, mg/L equivalent CaCOJ 1 Flowa L/min, sda standard deviation.
Manganese is clearly more efficiently removed in the CGM pond. This was demonstrated by the gradual increase in water flow rate (Figure 2). This is especially significant because Figure 2 data came from the winter months when water temperatures were between 4 and 6"C. In December 1992, the flow rate was 4.5-6 L/min in LOS and PGOS ponds, and was likely the reason for low Mn and Fe removal in those ponds. The CGM pond continued high-rate removal of Mn until 5-8 m from the inflow, while LOS and PGOS show much lower removal rates, even though the metal concentration in the water remains high (Table II). At lower metal loading, the slower removal rates of LOS and PGOS are sufficient to remove most of the Mn before the effluent point. However, at flow rates of >4.5 L/min, the mechanisms present in these two ponds are spent and only the CGM pond continues effective removal. TABLE II POND FLOW L/min
MANGANESE REMOVAL RATES (G/DAY/H2 )* INFLUENT
2H
5H
8H**
3.87
0.74
Hn
CGM
4.2
4.79
2.59
LOS
3.8
3.78
0.80
0.55
0.52
4.47
0.37
0.86
0.58
PGOS
2.8
*BecaUBe of the plumbing design, it was not possible to standardize the three pond flows to the 8ame rate. **Eight metres represent8 point E in Figure 1.
P. PHILLIPS tl at.
166
Another way to distinguish the superior CGH pond metal removal efficiency was to determine differences between daytime and nighttime removal. Figure 3 shows the metal removal profiles in all ponds at a flow rate of 5 L/min, during light and dark periods. Metals were effectively removed at approximately 1-2 m from the inflow in the CGH pond. These data represent a time period after four months of continuous flow. The two ponds with cyanobacteria covering only their rock substrate did demonstrate metal removal, although not as efficiently.
8
A
6 4
2
2
..--
4
6
8
~
e
'-'
§
"g
10 B
---
6
--.-.-
4
s:: 8s:: 2 0 u s::
CGM LOS
roos
~
2
4
6
8
10 C
4
6
8
10
6 4
2
2
Fig. 2.
Distance (M)
Comparison of Mn removal at varying flow ratesl A-I.S L/min, S-3 L/min, C-4.5-6 L/min.
M..V1ganese removal from aCid coal
S
167
A
4 3
2
~e
......,
c: .9 ~ c
-
2
4
6
8
10
8c s 8 4
B
c:
~
-----
CGM LOS
--A- PGDS
3 2
2
Fig. 3.
4
6
Distance (M)
8
10
Day-night comparison of Mn removal among three pondsr A=day~ B=night.
Dried, ground and hydrolyzed mat samples from CGM pond revealed a similar decreased in manganese concentration from influent to effluent points (Table III). TABLB
III
CONCENTRATION (KG METALlo ORIEl) KAT) FROM CCM POND, JANUARt 1993*
MANGANESE
Location
Mn
1 M from Influent
2.67
Pond Center
0.98
Point B (near Effluent)
0.45
a.
*Th. highe.t conc.ntration of metal wa. pr••ant cry.tallin. dapo.it. in the pond trough.. Metal .peciation of th••• precipitate. i. currently in proqre•• (J. Neil, US Geological Survey).
P. PHILLIPS el at.
I~
DISCUSSION AND CONCLUSIONS Although there was some Mn-cell binding, manganese was primarily deposited as precipitates at the pond bottom. Unlike the control ponds, there was no evidence of metal release from the mat pond. Day/night and winter/summer metal removals were essentially the same. Fe entered the treatment pond primarily as a flocculated precipitate which became entrapped in the filamentous algae. Control ponds showed Mn breakthrough (Mn outflow releases > US EPA regulations of 2 mg/L) during Di~ht-time sampling or when mine drainage flow exceeded 4.5 L/min. Since no soil was layered in the pond the predictable microbial ecology characterizing the sediment region may not be present in this system. It is assumed that biological processes contributed the controlling factors because all exposed limestone surfaces in the CGM pond were effectively covered by the cyanobacteria mat. Metals are known to complex with a wide range of organic material, including microbes and their organic releases. Dunbabin and Bowmer (1992) identify four dominant binding processes that incorporate metals into organic materials I (1) cation exchange, (2) adsorption, (3) precipitation and co-precipitation and (4) complexation or chelation. Although metals, which are adsorbed, precipitated or complexed, can be released back into solution in an equilibrium response, no such fluctuations were detected in the duration of the experiment thus far. Additionally, conditions of neutral pH with high D.O. and redox levels (mediated by the biology of the algae mat) favor the chemical precipitation of manganese oxides and iron hydroxides. These, in turn, act as reservoirs for additional metal deposition. It is curious that, although LOS and PGOS do not function as well as CGM in they also show high ORP and 00. It may be important to consider the microzones present in the thick algal mat, which are absent in the LOS and PGOS ponds. While the gross water quality measurements were similar in the three ponds, the micro-conditions showed large variability. The algae mat of the CGM pond was annealed together by slime into a 1-2 em sheet over the pond. The slime entrapped photosynthetic gases creating zones that measured as high as 12 mg/L (bubbles could be compressed to released the gases, revealing the regions of concentrated oxygen). As the metal-laden water passed through the labyrinth of algae filaments, it was likely exposed to these elevated oxygen regions, which rapidly precipitated the Mo. Examination of the underside of the mat showed that initial precipitation may occur in that location. As precipitation continues, sections of mat break and fall to the bottom, collecting in the troughs. These regions of concentrated manganese oxide deposits function as autocatalytic nuclei for further deposition of the Mn.
Mn removal,
Although the conditions of high oxygen and high Eh generated in algae mat microzones as well as autocatalysis may be central to the deposit of Mn oxides, other factors may be functional as well. Flocculents were identified in the water column under the mat. Laboratory research showed that specific bioflocculents were released by the mat in response to the presenCe of Mn+ z (Rodriguez-Eaton et al. 1993). These materials carried surface charges ranging from -58.8 to -65.7 mV. The charges changed to +1.8 in the presence of divalent metal, indicating metal-binding to the bioflocculent. The potential bioavailability of metals is favored by increases in acidity, reducing power and salinity. Since the mat generally reduces acidity and sequesters certain anions from the water column, including chloride, the stability of the metal deposit 1s favored. Although anaerobic zones can be identified within the mat in laboratory experiments, the redox conditions of field water remained high. The purple Bulfur bacteria, Chromatium spp.,
Manganese removallrom acid cOld
169
was inoculated as part of the mat microbial group in order to maintain low sulfide conditions. These autotrophs sequester H2S instead of H2 0 for photosynthesis, thereby decreasing the reducing power of the aqueous environment. Since Chromatium spp. typically colonizes in natural mats found in the environment, it can be expected to persist in the constructed field mats. purple photosynthetic bacteria (with similar characteristics to Chromatium) was identified in the CGM pond. It is likely several mechanisms, including flocculation, cell sorption, autocatalysis and mediation of the environmental conditions of pH, Eh and oxygen concentration,contribute to the metal removal. ACKNOWLEDGEMENT Funding for this project was received from the Tennessee Valley Authority, Chattanooga, TN (TVA Contract No. TV-8972lV) and the US Environmental Protection Agency Grant No. CR8190l, both to Clark Atlanta University. REFERENCES Bender, J. (1992). Recovery of heavy metals with a mixed microbial ecosystem. Final report to the U.S. Bureau of Mines, Grant fG0190028. Bender, J. (1993). Biological effects of cultured cyanobacteria mats on water contaminants. Report 11 to the US Environmental Protection Agency, Grant fCR816890l. Bender, J., Archibold, E.R., Ibeanusi, V. and Gould, J.P. (1989). Lead removal from contaminated water by a mixed microbial ecosystem. Wat. Sci. & Tech., 21(12), 1661-1664. Bender, J., Gould, J.P., Vatcharapijarn, Y. and Saha, G. (1991a). Uptake, transformation and fixation of Se(VI) by a mixed selenium-tolerant ecosystem. water, Air and Soil Pollution, 59, 359-367. Bender, J., Gould, J.P. and Vatcharapijarn, Y. (1991b). Sequester of zinc, copper and cadmium from water with a mixed microbial ecosystem. Proceedings of the U.S. Army USATHEHA Conference, Williamsburg, Virginia. Bender, J., Murray, R. and Phillips, P. (1993). Microbial mat degradation of chlordane. In Situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Bender, J. and Phillips, P. (1993). Implementation of microbial mats for bioremediation. In situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Bender, J., washington, J.R. and Graves, B. (In press). Metal sequestering with immobilized cyanobacteria mats. International Journal of water, Air and Soil Pollution. Brodie, G. (1993). Aerobic constructed wetlands and anoxic limestone drains to treat acid drainage. Constructed wetlands Workshop for Electric Power Utilities, Chattanooga, Tennessee, Aug. 24-26, 1993. Gordon, J.A. and Burr, J.L. (1989). Treatment of manganese from mining seep 3;~~g packed columns. Journal of Environmental Engineering 115(2), 386-
170
P. PHD-LIPS tl ul.
Mondecar, M., Bender, J., Ross, J., George, W. and Dummons, K. (1992). Bioremediation of TNT by a mixed microbial ecosystem. Abstr. 92nd Ann. Meet. Am Soc. Microbial. Washington, D.C. Paerl, H.W., Bebout, B.M. and Prufert, L.E. (1989). Naturally occurring patterns of oxygenic photosynthesis and Nz fixation in a marine microbial mati physiological and ecological ramifications. Inl Microbial Mats, Y. Cohen and E. Rosenberg (Eds.), American Society for Microbiology, washington, D.C., pp. 326-341. Phillips, P., Bender, J., Word, J., Niyogi, D. and Denovan, B. (1993). Mineralization of naphthalene, phenanthrene, chrysene and hexadecane with a constructed silage-microbial mat. In Situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Rodriguez-Eaton, S., Ekansmesanq, U. and Bender, J. (1993). Release of metal-binding flocculents by microbial mats. In Situ and on;Site Bioreclamation, The Second International symposium Proceedings, San Diego, California, Apr. 4-7, 1993.
Pergamon
0273-1223(95)OO503-X
Wal. Sci. T,c!L Vol. 31, No. 12, pp. 161-170, 1995. CopyrighlC 1995 lAWQ Pnnled in Greal Britain. AU rights reserved.
0273-1223,'95 $9'50 .. 0-00
MANGANESE REMOVAL FROM ACID COAL-MINE DRAINAGE BY A POND CONTAINING GREEN ALGAE AND MICROBIAL MAT P. Phillips*, J. Bender*, R. Simms*, S. Rodriguez-Eaton* and C. Britt** • Research Center for Science and Technology, Clark Atlanta University. Atlanta. GA 30314. USA ** Tennessee Valley Authority, 601 West Summit Hill Drive. Knoxville. TN 37902. USA
ABSTRACT Acid mine drainage flowing from an underground seep, through an anoxic drain to an oxidation pond and then trickling filter was distributed among three treatment ponds for manganese reduction before releasing the mining water to a surface stream. Manganese was more effectively removed from the integrated green algae and microbial mat with limestone substrate pond (algae mat system) than from control ponds containing limestone or pea gravel substrates without mat. Although there was some Mn-cell binding, manganese was primarily deposited as precipitates at the pond bottom. Day/night and winter/summer manganese removal was essentially the same. The greater efficiency in Mn removal by the algae mat pond was most pronounced with higher flow rates and during the night. At 2 metres from the influent point of each pond, the algae mat pond removed 2.59 g manganese/day/M2 , respectively, compared to 0.80 in the limestone pond and 0.37 in the pea gravel pond. Control ponds showed either Mn breakthrough or near breakthrough (Mn outflow releases > US Environmental Protection Agency regulations of 2 mg/L) during night-time sampling or when mine drainage flow exceeded 4.5 L/min. KEYWORDS Acid mine drainage; blue-green algae; cyanobacteria; manganese; metal removal; microbial mats. INTRODUCTION Manganese removal from acid mine drainage is a challenge due to the solubility of manganese sulfide and the high pH required to precipitate manganese as an oxide or carbonate. Therefore, it is common to find drainage with manganese above state or US Environmental Protection Agency (US EPA) standards (Gordon et al., 1989). The Tennessee Valley Authority utilizes constructed wetlands technology to treat acid mine drainage. These wetlands have generally been effective in removing Fe (0.4-21.3 g/day/m 2 of wetlands (GOM» and Mn (0.15-1.87 GDM) (Brodie, 1993). At one site within the Fabius Coal Mine in northeast Alabama, USA, Mn levels are approximately 8 mg/L after leaving an oxidation pond and before draining toward an extensive constructed wetland. At this point a pilot-scale field test was conducted to determine if an 161
162
P. PHILLIPS el a/.
integrated microbial (dominated by cyanobacteria or blue-green algae) and green algae mat would effectively remove residual manganese in a very small pond surface area. Microbial mats are natural heterotrophic and autotrophic communities dominated by cyanobacteria (blue-green algae). They are self-organized laminated structures annealed tightly together by slimy secretions from various microbial components. The surface slime of the mats effectively immobilizes the ecosystem to a variety of substrates, thereby stabilizing the most efficient internal microbial structure. Since mats are both nitrogen-fixing and photosynthetic (Paerl et aI., 1989), they are self• sufficient, solar-driven ecosystems with few growth requirements. In our laboratory, we have developed microbial mats, constructed with specific microbial components, for a variety of bioremediation applications (Bender and Phillips, 1993). Mats have been found to reduce selenate to elemental selenium (Bender et al., 1991a), remove Pb, Cd, Cu, Zn, Co, Cr, Fe and Mn from water (Bender, 1992~ Bender et a1 .• 1991b. in press) and to remove Pb from sediments (Bender et al., 1989). Degradation of recalcitrant organic contaminants has also been observed under both dark and light conditions (Bender and Phillips, 1993). The following contaminants have been degraded in water and/or soil media by constructed mats: TNT (Mondecar et aI., 1992), chrysene, naphthalene, hexadecane, phenanthrene (Phillips et al., 1993), PCB (Bender, 1993), TeE and the pesticides, chlordane (Bender et al., 1993) , carbofuran and paraquat. Radio-labeled experiments with mat-treated carbofuran, petroleum distillates and TeE show that these three compounds are mineralized by mats and mat products, such as biofilms and bioflocculents. Recently,our data has confirmed that the mats effectively treat mixtures of organics and heavy metal. Mats simultaneously sequestered Zn and mineralized TCE and chrysene. Research currently in progress shows that U 238 can be removed from ground water samples. This project was designed to examine the feasibility of applying mats in a field remediation pilot project. The broad goals were to assess the performance of mats under environmental conditions, such as determining seasonal survival and the efficiency of the mats in removing metals under day-night conditions. Comparisons in Mn removal were made among the three ponds~ algae mat pond with limestone substrate, a pond with limestone substrate and a pond with pea gravel substrate, these latter two without mat. Specific experiments in the present study compared the effectiveness of the algae mat and limestone pond compared to controls of limestone or pea gravel alone in removing manganese. METHODS Three 40 m2 ponds were constructed by first lining with PVC. Two ponds had limestone substrate and one had pea gravel substrate layered to create four rises separated by five troughs. Maximum trough water depth was 1 foot. Rises nearly broke the water surface. Mine drainage flowed from an oxidation pond to a the trickling filter, both designed to precipitate Fe(OH)]. From the trickling filter, one pipe fed all three ponds (Figure 1). Each pond was sampled for manganese at 6-7 points. The ponds operated continuously from August 1992-March 1993 and from June 1993. Cyanobacteria_algae Mat Pond (CGM) A cyanobacterial-dominated microbial mat is an entire ecosystem containing several bacteria species, but dominated by cyanobacteria, in the
Manganese removal from acid coal
163
OVERFLOW I
ANOXI~ DRAIN
OXIDATION POND
C C
A
1F
A
LOS B
C
D E
COM B
C
D E
~ ~
Fig. 1. Oxidation pond~ TFI trickling filter~ INI influent water~ A-E, additional sample points for manganese~ CGMI cyanobacteria-algae mat pond, LOSI limestone/oscillatoria pond~ PGOSI pea gravel/Oscillatoria pond. Not drawn to scale. multilayered mat structure. In this pond, microbial mat became enmeshed with volunteer green algae, thus forming an integrated green algae and microbial mat (CGM). Microbial strains, including Oscillatoria spp. and purple bacteria (likely a complex mixture of sulfur and non-sulfur species) were selected from the Alabama site during February 1992. Microbial mats were developed from these strains according to Bender and Phillips (1993). A microbial mat slurry (blended in water) was broadcast over the experimental pond at a rate of 1-1.5 L in three applications during a four-week period. Silage, prepared from grass clippings, added organic acids and several species of ensiling bacteria to the water. Within two months a thick green mat covered the entire pond surface. Underneath this floating layer, the limestone itself was covered with a heavy green coating of cyanobacteria. Limestone/Oscillatoria Pond (LOS) Although no biological material was intentionally added, CGM pond cyanobacteria spread into this pond. Limestone was covered with a thin «1 rom) film of cyanobacteria. Approximately 10% of the pond surface became covered with green algae within two months. The green algae died during December 1992. Precipitated iron covered the entire limestone substrate of the pond including the pond periphery beyond the effluent pipe. Pea Gravel/Oscillatoria Pond (PGOS) The pea gravel substrate in this pond also become thinly «1 mm) covered with cyanobacteria. Additionally, there was a rapid buildup of
164
P. PHILLIPS el at.
precipitated iron at the effluent pipe by December 1992. Flow Rate Initial flow was set at 1 L/min and incrementally increased over a period of two months. Flow was adjusted on a daily basis due to clogging by Fe(OHh· Ranges of flow rates (L/min) were 0.8-6.0, with the following mean values for each ponda CGM - 4.2, LOS = 3.8, PGOS = 2.8. Since Mn could be removed chemically in all ponds, by increasing the flow rate, we expected to reach a break point whereby the two ponds without substantial biological material would no longer remove manganese as efficiently as the CGM pond. Water Chemistry Water samples were collected at the oxidation pond, the trickling filter and at five points within each pond in a horizontal profile from influent to effluent (Figure 1). If an effluent sample could not be effectively collected, the last point within the pond was considered the effluent sample. In the field, influent and effluent water samples were tested for pH (Orion GX Series pH electrode; Orion 200 Series portable pH and Eh meter), Eh (Corning platinum redox combination electrode; Orion 200 Series portable pH and Eh meter), conductivity (Fisher Scientific Digital conductivity meter), temperature and dissolved oxygen (Otterbine Sentry III dissolved oxygen meter) • For metal determination, water samples were filtered (0.45 um cellulose acetate) and stabilized for transport with HNO J • These were later analyzed for manganese by atomic absorption spectrometry (Varian, Spectra AA-20 BQ, double beam). Reliability of manganese analysis was regularly verified by comparing with a synthetic solution containing 20 mg/L Mn and diluted with field pond water. RESULTS A floating mat (1-2 cm thick), composed of filamentous green algae mat and cyanobacteria, predominantly Oscillatoria spp., grew well in the CGM pond after addition of silage/mat inocula. A secondary mat of cyanobacteria also covered the limestone at the bottom. Thus the metal-contaminated water flowed between the double-layered mats. Approximately six weeks were required to establish a full pond mat cover that lasted nearly one year, but effective metal removal began in the early stages of mat growth. The thin layer of cyanobacteria covering the substrate of the two ponds designed as controls were colonized with an oscillatoria strain resembling that of the inoculated CGM pond cyanobacteria. Small amounts of green algae produced a floating mat, but the biomass remained low compared to the experimental pond. Through February 1993, the CGM pond maintained a viable algae mat. An approximately 1m2 bleached area (indicating cell death) developed during this time. Beneath this bleached algae, limestone was covered with viable cyanobacteria. Iron was evident in patches over the entire surface area of the pond, yet there were also large areas of viable green algae and microbial mat over the pond surface. In contrast to the control ponds, there was no ViSible evidence of iron precipitates in the pond periphery beyond the effluent pipe.
Manganese removal from acid coal
165
In March 1993, after a 50-em snow fall, the mat appeared to be decimated after the snow thawed. This may have been due to wash-out or snow shading. A snail infestation was also observed at that time. It is possible that snails were always present as part of the balanced ecosystem of the mat pond, but in the normal conditions of a healthy, highly productive algal biomass, their effect was irrelevant. The algae mat pond was drained and re-inoculated in June (metal deposits were left in the troughs). Mat regenerated and the pond is currently operating. Table I summarizes daytime water quality parameters from August 1992-April 1993. Separately, dissolved oxygen values for the single nighttime sample weres CGM, 4.2 mg/L; LOS, 6.1; and PGOS, 6.8. The low night-t~me value for the algae mat pond is likely due to high oxygen consumption by the biological material. TABLE I
T(sd)* CGM LOS PGOS
MEAN DAYTIME WATER QUALITY PARAMETERS PER POND
DO(sd)
14.1(7.6) 7.3(1.3) 14.7(8.4) 8.3(1.8) 14.8(8.5) 8.2(1.9)
pH(sd)
ORP(sd)
Cond(sd)
Alk(sd)
Flow(sd)
7.4(0.4) 7.6(0.4) 7.2(0.5)
413(56.4) 424(73.8) 410(74.4)
648(136.7) 618( 94.7) 580( 92.3)
167(50.8) 168(53.2) 156(49.1)
4.2(2.4) 3.2(1.0) 2.2(1.6)
*Ta temperature, celsius1 001 dissolved oxygen, mg/L, ORPa oxidation-reduction potential, mV1 Conda conductivity, umbo/cm, Alka alkalinity, mg/L equivalent CaCOJ 1 Flowa L/min, sda standard deviation.
Manganese is clearly more efficiently removed in the CGM pond. This was demonstrated by the gradual increase in water flow rate (Figure 2). This is especially significant because Figure 2 data came from the winter months when water temperatures were between 4 and 6"C. In December 1992, the flow rate was 4.5-6 L/min in LOS and PGOS ponds, and was likely the reason for low Mn and Fe removal in those ponds. The CGM pond continued high-rate removal of Mn until 5-8 m from the inflow, while LOS and PGOS show much lower removal rates, even though the metal concentration in the water remains high (Table II). At lower metal loading, the slower removal rates of LOS and PGOS are sufficient to remove most of the Mn before the effluent point. However, at flow rates of >4.5 L/min, the mechanisms present in these two ponds are spent and only the CGM pond continues effective removal. TABLE II POND FLOW L/min
MANGANESE REMOVAL RATES (G/DAY/H2 )* INFLUENT
2H
5H
8H**
3.87
0.74
Hn
CGM
4.2
4.79
2.59
LOS
3.8
3.78
0.80
0.55
0.52
4.47
0.37
0.86
0.58
PGOS
2.8
*BecaUBe of the plumbing design, it was not possible to standardize the three pond flows to the 8ame rate. **Eight metres represent8 point E in Figure 1.
P. PHILLIPS tl at.
166
Another way to distinguish the superior CGH pond metal removal efficiency was to determine differences between daytime and nighttime removal. Figure 3 shows the metal removal profiles in all ponds at a flow rate of 5 L/min, during light and dark periods. Metals were effectively removed at approximately 1-2 m from the inflow in the CGH pond. These data represent a time period after four months of continuous flow. The two ponds with cyanobacteria covering only their rock substrate did demonstrate metal removal, although not as efficiently.
8
A
6 4
2
2
..--
4
6
8
~
e
'-'
§
"g
10 B
---
6
--.-.-
4
s:: 8s:: 2 0 u s::
CGM LOS
roos
~
2
4
6
8
10 C
4
6
8
10
6 4
2
2
Fig. 2.
Distance (M)
Comparison of Mn removal at varying flow ratesl A-I.S L/min, S-3 L/min, C-4.5-6 L/min.
M..V1ganese removal from aCid coal
S
167
A
4 3
2
~e
......,
c: .9 ~ c
-
2
4
6
8
10
8c s 8 4
B
c:
~
-----
CGM LOS
--A- PGDS
3 2
2
Fig. 3.
4
6
Distance (M)
8
10
Day-night comparison of Mn removal among three pondsr A=day~ B=night.
Dried, ground and hydrolyzed mat samples from CGM pond revealed a similar decreased in manganese concentration from influent to effluent points (Table III). TABLB
III
CONCENTRATION (KG METALlo ORIEl) KAT) FROM CCM POND, JANUARt 1993*
MANGANESE
Location
Mn
1 M from Influent
2.67
Pond Center
0.98
Point B (near Effluent)
0.45
a.
*Th. highe.t conc.ntration of metal wa. pr••ant cry.tallin. dapo.it. in the pond trough.. Metal .peciation of th••• precipitate. i. currently in proqre•• (J. Neil, US Geological Survey).
P. PHILLIPS el at.
I~
DISCUSSION AND CONCLUSIONS Although there was some Mn-cell binding, manganese was primarily deposited as precipitates at the pond bottom. Unlike the control ponds, there was no evidence of metal release from the mat pond. Day/night and winter/summer metal removals were essentially the same. Fe entered the treatment pond primarily as a flocculated precipitate which became entrapped in the filamentous algae. Control ponds showed Mn breakthrough (Mn outflow releases > US EPA regulations of 2 mg/L) during Di~ht-time sampling or when mine drainage flow exceeded 4.5 L/min. Since no soil was layered in the pond the predictable microbial ecology characterizing the sediment region may not be present in this system. It is assumed that biological processes contributed the controlling factors because all exposed limestone surfaces in the CGM pond were effectively covered by the cyanobacteria mat. Metals are known to complex with a wide range of organic material, including microbes and their organic releases. Dunbabin and Bowmer (1992) identify four dominant binding processes that incorporate metals into organic materials I (1) cation exchange, (2) adsorption, (3) precipitation and co-precipitation and (4) complexation or chelation. Although metals, which are adsorbed, precipitated or complexed, can be released back into solution in an equilibrium response, no such fluctuations were detected in the duration of the experiment thus far. Additionally, conditions of neutral pH with high D.O. and redox levels (mediated by the biology of the algae mat) favor the chemical precipitation of manganese oxides and iron hydroxides. These, in turn, act as reservoirs for additional metal deposition. It is curious that, although LOS and PGOS do not function as well as CGM in they also show high ORP and 00. It may be important to consider the microzones present in the thick algal mat, which are absent in the LOS and PGOS ponds. While the gross water quality measurements were similar in the three ponds, the micro-conditions showed large variability. The algae mat of the CGM pond was annealed together by slime into a 1-2 em sheet over the pond. The slime entrapped photosynthetic gases creating zones that measured as high as 12 mg/L (bubbles could be compressed to released the gases, revealing the regions of concentrated oxygen). As the metal-laden water passed through the labyrinth of algae filaments, it was likely exposed to these elevated oxygen regions, which rapidly precipitated the Mo. Examination of the underside of the mat showed that initial precipitation may occur in that location. As precipitation continues, sections of mat break and fall to the bottom, collecting in the troughs. These regions of concentrated manganese oxide deposits function as autocatalytic nuclei for further deposition of the Mn.
Mn removal,
Although the conditions of high oxygen and high Eh generated in algae mat microzones as well as autocatalysis may be central to the deposit of Mn oxides, other factors may be functional as well. Flocculents were identified in the water column under the mat. Laboratory research showed that specific bioflocculents were released by the mat in response to the presenCe of Mn+ z (Rodriguez-Eaton et al. 1993). These materials carried surface charges ranging from -58.8 to -65.7 mV. The charges changed to +1.8 in the presence of divalent metal, indicating metal-binding to the bioflocculent. The potential bioavailability of metals is favored by increases in acidity, reducing power and salinity. Since the mat generally reduces acidity and sequesters certain anions from the water column, including chloride, the stability of the metal deposit 1s favored. Although anaerobic zones can be identified within the mat in laboratory experiments, the redox conditions of field water remained high. The purple Bulfur bacteria, Chromatium spp.,
Manganese removallrom acid cOld
169
was inoculated as part of the mat microbial group in order to maintain low sulfide conditions. These autotrophs sequester H2S instead of H2 0 for photosynthesis, thereby decreasing the reducing power of the aqueous environment. Since Chromatium spp. typically colonizes in natural mats found in the environment, it can be expected to persist in the constructed field mats. purple photosynthetic bacteria (with similar characteristics to Chromatium) was identified in the CGM pond. It is likely several mechanisms, including flocculation, cell sorption, autocatalysis and mediation of the environmental conditions of pH, Eh and oxygen concentration,contribute to the metal removal. ACKNOWLEDGEMENT Funding for this project was received from the Tennessee Valley Authority, Chattanooga, TN (TVA Contract No. TV-8972lV) and the US Environmental Protection Agency Grant No. CR8190l, both to Clark Atlanta University. REFERENCES Bender, J. (1992). Recovery of heavy metals with a mixed microbial ecosystem. Final report to the U.S. Bureau of Mines, Grant fG0190028. Bender, J. (1993). Biological effects of cultured cyanobacteria mats on water contaminants. Report 11 to the US Environmental Protection Agency, Grant fCR816890l. Bender, J., Archibold, E.R., Ibeanusi, V. and Gould, J.P. (1989). Lead removal from contaminated water by a mixed microbial ecosystem. Wat. Sci. & Tech., 21(12), 1661-1664. Bender, J., Gould, J.P., Vatcharapijarn, Y. and Saha, G. (1991a). Uptake, transformation and fixation of Se(VI) by a mixed selenium-tolerant ecosystem. water, Air and Soil Pollution, 59, 359-367. Bender, J., Gould, J.P. and Vatcharapijarn, Y. (1991b). Sequester of zinc, copper and cadmium from water with a mixed microbial ecosystem. Proceedings of the U.S. Army USATHEHA Conference, Williamsburg, Virginia. Bender, J., Murray, R. and Phillips, P. (1993). Microbial mat degradation of chlordane. In Situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Bender, J. and Phillips, P. (1993). Implementation of microbial mats for bioremediation. In situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Bender, J., washington, J.R. and Graves, B. (In press). Metal sequestering with immobilized cyanobacteria mats. International Journal of water, Air and Soil Pollution. Brodie, G. (1993). Aerobic constructed wetlands and anoxic limestone drains to treat acid drainage. Constructed wetlands Workshop for Electric Power Utilities, Chattanooga, Tennessee, Aug. 24-26, 1993. Gordon, J.A. and Burr, J.L. (1989). Treatment of manganese from mining seep 3;~~g packed columns. Journal of Environmental Engineering 115(2), 386-
170
P. PHD-LIPS tl ul.
Mondecar, M., Bender, J., Ross, J., George, W. and Dummons, K. (1992). Bioremediation of TNT by a mixed microbial ecosystem. Abstr. 92nd Ann. Meet. Am Soc. Microbial. Washington, D.C. Paerl, H.W., Bebout, B.M. and Prufert, L.E. (1989). Naturally occurring patterns of oxygenic photosynthesis and Nz fixation in a marine microbial mati physiological and ecological ramifications. Inl Microbial Mats, Y. Cohen and E. Rosenberg (Eds.), American Society for Microbiology, washington, D.C., pp. 326-341. Phillips, P., Bender, J., Word, J., Niyogi, D. and Denovan, B. (1993). Mineralization of naphthalene, phenanthrene, chrysene and hexadecane with a constructed silage-microbial mat. In Situ and On-Site Bioreclamation, The Second International Symposium Proceedings, San Diego, California, Apr. 5-7, 1993. Rodriguez-Eaton, S., Ekansmesanq, U. and Bender, J. (1993). Release of metal-binding flocculents by microbial mats. In Situ and on;Site Bioreclamation, The Second International symposium Proceedings, San Diego, California, Apr. 4-7, 1993.