Introduction

Acid mine drainage (AMD), characterized by high acidity (pH 2.0–4.0), high concentration of sulfate, metals and metalloids ions, is the creation of coal and metal mine industries mostly (Johnson and Hallberg 2005). It is a potential hazard to the environment, and the overall impact is to reduce ecological stability (Gray 1997). Hence, that should be treated before discharge.

Different technologies have been applied to mitigate the problem, in which biological treatment strategies using sulfate reducing bacteria (SRB) also find favor (Johnson and Hallberg 2005; Das et al. 2009). Under anaerobic condition, SRB oxidize organic compounds and reduce sulfate to hydrogen sulfide and thereby facilitate the precipitation of metal sulfides. Bicarbonate production in this reaction neutralizes the acidity. This type of bioremediation is generally achieved through anaerobic wetlands/compost bioreactors, permeable reactive barriers as well as in successive alkalinity producing system (SAPS) (Johnson and Hallberg 2005).

Limestone is often used to generate alkalinity in AMD treatment systems because of its low cost and availability to close proximity to mining areas. However, its effectiveness reduces when mainly Fe3+ of AMD precipitate, causing armoring and clogging. In SAPS, this problem is partially overcome as influent water flow vertically through organic layer, into limestone beds, and ultimately discharged (Demchak et al. 2001). In the organic layer (mainly compost), microbiological respiration removes oxygen and produces the anaerobic zone where Fe+3 are reduced to Fe+2. Fe+3 generally precipitate as pH rises, but Fe+2 are expected to be soluble at neutral pH. The reduced water then passes through limestone layer where the acidity is being neutralized.

Spent mushroom compost is widely used for the creation of organic layer in SAPS (Nairn and Mercer 2000; Demchak et al. 2001; Bhattacharya et al. 2008). It showed higher relative performance compared to other waste materials as electron donor for SRB (Chang et al. 2000). It can provide sufficient dissolved organic carbon (DOC), nitrogen (DON), and phosphate (Guo et al. 2001). Polysaccharide content of SMC is generally degraded by hydrolytic fermentative anaerobes to alcohols and fatty acids that support the growth of SRB (Chang et al. 2000). Because of its slow degradation rate and bulk physical property, it can be chosen as a suitable substrate for long-term operation in AMD bioremediation (Dvorak et al. 1992; Bhattacharya et al. 2008). SMC has also been applied for treating metals in AMD and coal mine drainage in passive bioreactor system (Dvorak et al. 1992; Stark et al. 1994). However, no systematic study has been carried out about its organic carbon–liberating capacity and to support SRB activity.

SAPS was reconstructed in the form of chemo-bioreactor and its performance to treat acidic and metal laden wastewater was reported in our previous study (Cheong et al. 2010). Spent mushroom compost was used as organic matrix in this study. However, the extent of nutrient supply and the biochemical aspects in such a setting is still unclear and is necessary for longer treatment process design. The objective of this study was to evaluate how long SMC could liberate available organic carbon, bacterial abundance (especially sulfate reducing bacteria) in this environment, and utilization of organic matrix. In addition, evaluation of limestone in terms of exhaustion for better performance of chemo-bioreactor in continuous treatment and the fate and the composition of metal precipitates have been analyzed.

Materials and methods

Chemo-bioreactor operation

The study was performed with a pilot-scale gravity fed chemo-bioreactor system reported earlier (Cheong et al. 2010). SMC and limestone were reported to be characterized for alkalinity generation, DOC liberation, and metal release. The source of the materials used in this study was exactly the same as earlier. The amount of these material and some parametric values of synthetic AMD was slightly modified. Main reacting tank D (145 × 30 cm) of the chemo-bioreactor was packed with 4.8 kg of limestone (4.2 cm) at the base followed by 5.12 kg of SMC (14 cm) and a gravel layer (made up of quartz type of rock) of 5.3 kg (3.5 cm) at the top, with a final reacting bed height of 21.7 ± 0.2 cm. The mass density of the reactor bed was 0.77 g cm−3. The porosity was calculated to be 0.59 ± 0.01, using the final reacting bed volume (without gravel layer) and the calibrated liquid volume used to pack the bed. The chemo-bioreactor was operated in room temperature at 25 ± 5°C. Synthetic acid mine drainage was prepared alike the previous study. The recipe was based on reported literature values of originally characterized AMD. It contains metal ions (mg/L) Fe (460), Mn (220), Mg (267) Cu (264), and SO4 2− (2,611) with pH 2.8 and acidity of 8 mg/L as CaCO3. All chemicals used to prepare solutions were reagent grade. Eighty liters of this mixture solution was fed to the chemo-bioreactor (D) initially, filling other three tanks (A, B and C), and was left to settle. Blackish pond sediment (80 mL) having most probable number (MPN) of 48 ± 4 SRB cells/100 mL was added as inoculums. After 15 days, continuous flow was started. A flow rate of 535.7 mL/h (0.0706 h−1 dilution rate with respect to liquid volume) was maintained during the experiment, which is equal to 0.758 cm h−1 of superficial flow velocity and with a hydraulic retention time of 24 h through the bed. During this period, the water volume was kept constant by controlling flow rate until end (18 weeks).

Sampling and analysis

Influent (sample port in tank D) and effluent (tank F) water were collected at 7 days interval regularly. Samples for analytical purpose were filtered through a 0.45-μm nylon membrane filter to remove precipitates and other solid materials. Metal concentrations were measured by ICP-MS (Varian 820-MS) after pretreatment with concentrated HNO3 (1.5 μl/mL sample; v/v) (Greenberg et al. 1992). Sulfate was measured by titration method in electrometric auto-titrator (Orion 950 Ross FASTQC, USA) in the fixed millivolt (mv) increment mode for first-derivative endpoint detection. pH and oxidation reduction potential (ORP) were measured by multi-parameter water analyzer (Orion 5–Star BENCHTOP MULTI, Thermo Scientific, USA). Total organic carbon (TOC) was analyzed by TOC analyzer (TOC-V CPH, Shimadzu, Japan) using the filtered sample and reported as DOC (Khan et al. 1998). Limestone exhaustion in terms of calcium ion concentration was measured before and after the experiment. Whenever possible, pH, metals, and sulfate were determined on the same day the samples were collected. Total bacterial count of the reactor matrix was performed by standard plate count method on Trypticase soy agar. SRB count of the matrix was done by MPN method described by Fortin et al. (2000). Reactor matrix was taken during sampling, weighted, and diluted into respected dilution before plating or MPN experiment. Total bacterial population is expressed in CFU/g matrix, whereas SRB population is expressed in MPN/g matrix (Cochran 1950). Epifluorescence microscopic method was performed for observing bacterial morphology and live-dead differentiation (Leica DMR coupled with Leica MPS60 photomicrographic system). In brief, 1 g of matrix was homogenized in 100 mL 1x PBS buffer. One and half milliliter of it was centrifuged for 5,000 rpm × 3 min, and supernatant was discarded. The pellet was dissolved in 1 mL PBS buffer and again centrifuged under similar condition. Acridine orange (AO) and propidium iodide (PI) of 2 μl each were added to the pellet, and volume was adjusted to 25 μl by 1x PBS buffer. The tubes were then incubated for 20 min at dark. After incubation, the stains were washed thrice with PBS buffer and then centrifuged for 4 min at 5,000 rpm to pellet the cells. The pellet was dissolved again in PBS buffer to make the volume 500 μl. Living bacterial produce green fluorescence (due to AO), and dead would produce red (due to PI). The counting was done by standard procedure (Mesa et al. 2003)

SMC and precipitate analysis

SMC samples (before and after experiment) and the precipitate formed in the E tank were analyzed for precipitate characterization. Samples were examined by a Zeiss field emission scanning electron microscope (FESEM) with attached energy dispersive X-ray (EDAX) analyzer. Mineral phases were identified by X-Ray Diffractometry (XRD) (Philips) with CuKα radiation (λ = 1.5406). Peaks were identified by standard software (JCPDS-ICDD 1999). Infrared spectra of SMC and precipitates were recorded on a Nexus TM 870 FT-IR (Thermo Nicolet, USA) spectrophotometer equipped with deuterated triglycine sulfate detector thermo electric cool (DTGS TEC). Each sample spectra was performed using KBr disc in the range 400–4,000 cm−1.

Results

Chemo-bioreactor performance

Figure 1 shows the time course of removal efficiency of all tested metals (Fe, Mn, Mg, and Cu), and the average reactor performance is summarized in Table 1. Removal of metal ions in the effluent was found to be started from 3rd week, but the value of Mg reduction percentage was negative, indicating its initial addition into the effluent (Fig. 1). Magnesium was the least reduced metal compared to other tested metals, while Cu was reduced maximum in the same time. Mg (93.6%) and Mn (96.4%) were reduced maximum at 3rd week, Cu (97.3%) at 6th week, Fe (97.9%), and sulfate (98.6%) at 12th week of chemo-bioreactor performance. The pH was maintained over 6- up to 13-week period. ORP was less than −100 mV during the experiment. However, removal of all tested metals and sulfate was decreased sharply after 14th week, indicating some kind of process failure occurred at this stage (Figs. 1 and 2). DOC level was also dropped from 400 to 50 mg/L after 14th week (Fig. 2). Drop of alkalinity after 12th week and pH after 13th week was also found (Fig. 3). Calcium remained in the exhausted limestone is 36.93 ± 0.68%, implying about 63% was used.

Fig. 1
figure 1

Status of metal concentration during continuous flow. Data are the mean of triplicates performance result ± standard error of the mean (S.E.M.)

Full size image
Table 1 Chemo-bioreactor performance during continuous operation
Full size table
Fig. 2
figure 2

Relation of sulfate reduction with available DOC and bacterial load. Symbol indicates DOC (open triangle); Sulfate (filled square); total bacterial population (filled circle); and sulfate reducing bacterial population (open circle), and data are the mean of triplicates performance result ± S.E.M

Full size image
Fig. 3
figure 3

pH and alkalinity profile during continuous run. Data are the mean of triplicates performance result ± S.E.M

Full size image

Bacterial abundance

Double fluorescence stained samples (Fig. S1) showed both live and dead cells were present in the effluent, in which rod-shaped cells were predominated. Figure 2 illustrates the status of total bacterial and SRB population density in the chemo-bioreactor. Throughout the experiment, both population was found to be declined except at 3rd week when SRB (≈108 MPN/g) as well as total bacterial population (≈109 CFU/g) reached maximum. Results indicate that initially there was a prevalence of other bacteria over SRB in the chemo-bioreactor. Gradually, growth of SRB predominated, and hence, SRB population curve follows the same pattern like total bacterial population. From 9th to 12th week, SRB population began to comprise most of the bacterial population and at this period, total bacterial population was 1.73 and 2.5 times of SRB population, respectively. Total bacterial population was far over 107 CFU/g, whereas SRB population was near 107 MPN/g at DOC higher than 100 mg/L. However, both total bacterial and SRB population decreased exponentially after 12th and 14th week when available DOC level was as low as 99 and 35 mg/L, respectively. Noticeably, at that point, SRB population decreased sharply correlating significantly with released DOC level. SRB population dropped down to 104 MPN/g at DOC level less than 50 mg/L, at the end of the experiment comparing a constant population of total bacteria near 106 CFU/g at that period.

Microscopic and mineralogical analysis

SEM images of SMC samples before and after the experiment (exhausted SMC) revealed structural differences (Fig. 4). Biofilm formation on the SMC bed was observed after experiment. Attached EDAX analysis demonstrates that the major elements in SMC before loading were Si, Al, Mn, Fe, Mg, and O (H is not detectable) and minor elements included Cu, C, and K. Major elements in exhausted SMC were Si, Fe, Ca, and O, whereas minor elements being Cu, Mg, and Al. Blackish green precipitates were formed in the first effluent tank (E) from the beginning of the experiment and continued to form till the completion of the experiment. SEM analysis of the precipitates (Fig. 4c) shows the formation of some geometrical-shaped particles associated intermittently with organic matter in low concentration. Major elements of this precipitates were Fe, Mn, Ca, C, S, and O and minors were Si, Cu, Mg, Zn, and Ni.

Fig. 4
figure 4

SEM and EDAX analysis of SMC and precipitates. SEM and EDAX micrograph of samples, respectively, (A and a) SMC before loading, (B and b) SMC after experiment, and (C and c) precipitates in effluent tank

Full size image

No sulfide mineral has been detected in the powder XRD spectra of raw SMC (Fig. 5). However, sulfide compound of all tested metals has been recognized in exhausted SMC. This sample contains digenite (Cu9S5), magnesium sulfide (MgS), manganese oxide (MnO), iron sulfide (FeS), mackinawite (FeS), hausmannite (Mn3O4), and periclase (MgO). Precipitates formed in the effluent tank were predominantly sulfides, namely iron sulfide (FeS), pyrite (FeS2), magnesium sulfide (MgS), hauerite (MnS2), greigite (Fe3S4), covellite (CuS), and manganese sulfide (β-MnS).

Fig. 5
figure 5

X-ray diffraction pattern of SMC and precipitates. Diffraction pattern of a SMC before loading; b SMC after experiment; and c precipitate formed in the effluent tank

Full size image

FTIR analysis

FTIR spectrum (Fig. 6) of raw and exhausted SMC and effluent tank precipitate showed broad peaks at 3,418, 3,415, and 3,405 cm−1, respectively, corresponding to the OH stretching of polymers or phenolic group and N–H stretching of primary amides that might be associated with cellulosic cell wall of mushroom and plant debris (Bellamy 1975; Lau et al. 2003; Chen et al. 2005). Decrease in intensity of peak near 3,400 cm−1 in exhausted SMC can be attributed to the involvement of the H-bonded OH and N–H group in metal biosorption. Shifting of peak in the region of 1,650 cm−1 [C=O stretching of primary amide] (Bellamy 1975; Pagnanelli et al. 2000) of SMC to lower frequencies at 1,620 cm−1 in effluent precipitate can be due to interaction of this group with metal ions. Peaks near 1,100 cm−1 [C=O; stretching and O–H deformation vibration of secondary alcohol] (Bellamy 1975) are present in both SMC sample but absent in precipitates. A new peak with small intensity at 2,360 cm−1 region in sample of exhausted SMC and precipitate was found [υ3 band of CO2] (Nakamoto 1986; Taraschewski et al. 2005). New peaks in precipitate at 1,252 and at 1,345 cm−1 can be assigned to SO3 stretching, (Winie and Arof 2006; Kang et al. 2007) and inorganic NO2 stretching (Bellamy 1975), respectively. It seemed the deposition of sulfate ions also occurred with metal ions. No perceptible change of the peak position and intensity at 2,920 cm−1 (the peak signifying CH2 asymmetric stretch in aromatic and alkenes present in lignin) in exhausted SMC was found (Bellamy 1975). Absorption peaks at 796, 798, and 939 cm−1 in raw SMC, exhausted SMC and in precipitate, respectively, could be due to the presence of residual quartz (Snyder et al. 1983; Misaelides et al. 1996; Han-xu et al. 2008).

Fig. 6
figure 6

Overlaid FTIR spectra of SMC and precipitates. a Control SMC before loading to experimental tank; b SMC after experiment showed new peak at 2,360 cm−1 (presence of CO2) and decrease in peak intensity near 1,100 cm−1 (utilization of secondary alcohol), and c precipitate formed in the effluent tank showed along with CO2, new peaks formed at 1,252 and 1,345 cm−1 representing SO3 stretching

Full size image

Discussion

The results illustrate that a total 1,620 liters of synthetic AMD has been treated successfully during long-term running. This treatment process was established as a cost effective scheme in previous study (Cheong et al. 2010), and SMC used to water treated ratio was much less (3.16 g SMC used/liter AMD treated) in this study. Iron and sulfate removal efficiency was better than other reported chemo-bioreactor (Dvorak et al. 1992; Stark et al. 1994; Demchak et al. 2001). However, reduction in Mg was limited in as much to reach the expected level compared to other metals, even some addition of it to the effluent also observed. Mineralogical study shows the presence of magnesium compound in effluent tank precipitate (Figs. 4c and 5c). In our previous study, we have shown that SMC releases Mg in similar condition (Cheong et al. 2010). From these observations, it can be stated that both addition of Mg from SMC and its transformation to insoluble complexes in the chemo-bioreactor occurred simultaneously. Deposition of metals onto bacterial cell or in cell-bound polysaccharide is a well-known phenomenon (Iyer et al. 2005). Functional groups such as carboxylic, amino, thiol, hydroxyl, and hydroxy-carboxylic groups present in bacterial cell wall are reported to interact with metal ions (Sigg 1987). Phenolic group takes a major part of metal deposition which was often found in microbes and distributed widely in the lignin, cellulose, and semi-cellulose containing fungal cell wall (Chen et al. 2005). Metal sorption by various organic groups treating AMD is reported earlier (Gibert et al. 2005). FTIR analysis shows the involvement of OH group of polymers or phenolic compounds, N–H and C=O group of primary amides, in metal binding. It may be hypothesized that metal ions are deposited onto the SMC and bind to its phenol-rich constituent and cell walls of bacteria. Periodic fluctuations of the metal removal as shown in Fig. 1 were probably due to the simultaneous adsorption–desorption of metals into the matrix. Formation of biofilm on SMC from SEM images and presence of SRB by MPN test have been established here. FTIR study reveals the presence of CO2 in exhausted SMC and precipitate, and those points toward either microbial metabolism-evolved-CO2 in the chemo-bioreactor or residual limestone-evolved-CO2 in precipitate. No significant change was observed of the peak position, and intensity remained at 2,920 cm−1 in exhausted SMC. Therefore, it can be ascribed that lignin of SMC was neither utilized as nutrient source nor responsible for biosorption or it might only be used as microstructure for attachment site of microbes.

XRD analysis of both exhausted SMC and precipitates revealed the presence of sulfide complex of all tested metals. Diverse SRB species are reported to be present in sulfidogenic reactor feeding simple organic carbon source (Sigg 1987; Gibert et al. 2005). Majority and diversity of metal sulfide formations in effluent precipitate (Fig. 5c) indicate the diverse mechanism of sulfate reduction in the reactor. There is a high likelihood of SRB-mediated H2S production, and thus metal sulfide formation could constitute the major reactions.

Solubility of metal sulfides was highly reduced at pH 5.0–9.0 and hence got precipitated. There is a positive correlation between SRB abundance and sulfate reduction. At primary stage of continuous flow when adequate DOC was present in the chemo-bioreactor, total bacterial population was nearly tenfold greater than SRB at 9th week, whereas this difference became very close after that. FTIR data showed that (decrease in band intensity at 1,100 cm−1 in exhausted SMC compared to raw SMC) microorganism in the chemo-bioreactor could have utilized the alcoholic group in SMC for their growth. SRB are only able to utilize complex organic carbon when synergic effect of other species results SRB degradable one (Neculita et al. 2007; Zagury et al. 2006). Therefore, initially there might be diverse microbial species including some aerobes present, which provided available organic carbon to SRB by synergic action. Microbial metabolism led CO2 and limestone-evolved-CO2 below the long water column helped to make the environment anoxic and supported anaerobic microbial growth. Negative ORP value confirms the oxygen deficient condition during the experiment. When steady state condition was being established, sulfate became the major electron acceptor and SRB might outcompete other microbes, as they have a thermodynamic advantage in sulfate-rich condition (Fernandez-Polanco and Garcia Encina 2006; Neculita et al. 2007).

After 13th week, a process failure of chemo-bioreactor was observed. It might be due to the lower level of DOC and this reduced DOC level contributed to the reduction in SRB abundance. However, at that time, total bacterial count was not reduced as rapidly as SRB. This is might be due to the exhaustion of SRB degradable DOC. However, released DOC was sufficient to promote the growth of other bacteria present at that stage. Sulfate reduction in the chemo-bioreactor dropped abruptly with falling DOC level. Therefore, a drought of degradable organic carbon for SRB started to occur after 13th week. In the absence of sufficient electron donors, metabolism of SRB stopped and growth ceased resulting population decrease. An enormous fall of alkalinity and slight drop of pH might be due to the cessation of microbial alkalinity. Sudden change of alkalinity is reported to be results on the change of microbial community in sulfidogenic reactor (Zhao et al. 2010). That may be a reason of less effect on total microbial population after 13th week. Whatever little amount of metal sulfide was formed at this stage became soluble at that pH. Hence, organic carbon drought and insufficient alkalinity in chemo-bioreactor caused process failure. SRB degradable organic carbon dosing after that instance can be thought as a sustainable treatment solution at large scale and longer treatment.

Conclusion

The developed pilot scale SAPS-based chemo-bioreactor showed better performance for long-term AMD treatment than other earlier developed. SMC used was a good host for SRB growth and activity. However, unavailability of DOC after some instance was the major reason for process failure combined with restricted alkalinity in the chemo-bioreactor. In addition to that, SMC is a good absorbent for direct metal removal also. Hence, the developed chemo-bioreactor is cost effective and useful for long-term AMD treatment with better performance.