The sugarcane slag used in the experiment was taken from a sugar mill in Yunnan Province. The corn cob and sunflower straw were taken from Fuxin local farmland. Biomass materials were washed with sterile deionized water, crushed and sieved. Iron scrap, taken from the school training factory, washed with acid of low concentrations and then washed with sterile deionized water. It was dried, ground and screened.

According to the optimized data of single factor and orthogonal before the experiment, the dynamic columns were designed. The dynamic column 1 was filled with 32 mesh sugarcane slag and 60 mesh iron scrap, the mass percentage of them being 2.1% and 97.9%, respectively. The dynamic column 2 was filled with 32 mesh corn cob and 60 mesh iron scrap, the mass percentage of them being 6.6% and 93.4%, respectively. The dynamic column 3 was filled with 10 mesh sunflower straw and 60 mesh iron scrap, the mass percentage of them being 0.9% and 99.1%, respectively.

The concentration of SO₄^2–, Cr⁶⁺ and Cr³⁺ were 800–900, 8–10 and 20–25 mg/L respectively and the pH value ranged from 4.0 to 4.5 in simulated AMD.

SRB [18]: Take the wet mud under one of the coal gangue piles in Fuxin City as seed sludge, and put it into Starkey medium for anaerobic culture to obtain SRB. The SRB (NCBI accession number: MT804386) obtained belongs to Desulfotomaculum and the 16S rDNA sequence is:

TGGGGAATCTTCCGCAATGGACGAAAGCCTGACGGAGCAACGCCGCGTGAGGGAAGAAG GCCTTCGGGTTGTAAACCTCTGTCCTAAAGGAAGAAAGAAATGACGGTACTTTAGGAGGAAGC CCCGGCTAACTACGTGCCAGCAGCCGCGGTAAGACGTAGGGGGCAAGCGTTGTCCGGAATTAC TGGGCGTAAAGGGCGCGTAGGTGGTCCATTAAGTTAGAGGTGAAAGTGCGGGGCTTAACCCCG TTATTGCCTCTCATACTGGTGGACTTGAGTGCTGGAGAGGGGAGTGGAATTCCCACTGTAGCG GTGAAATGCGTAGAGATTGGGAGGAACACTAGTGGCGAAGGCGGCTCTCTGGACTGCAACTG ACACTGAGGCGCGAAAGCGTGGGGAGCAAACAGG.

Three Plexiglas’s tubes with 500 mm height and 80 mm internal diameter were used as the experimental dynamic columns. The quartz sand layer with 50 mm height and 3–5 mm diameter, mixture (iron scrips and biomass material) layer with 200 mm height and quartz sand layer with 50 mm height and 3–5 mm diameter were filled in the columns in order. The three dynamic columns are injected with the equal amount of SRB and Starkey broth, that is, 2 L SRB at the concentration of 1 × 10⁶/mL and 1 L Starkey medium. Incubated the columns in a constant temperature water bath at 37°C for one week until the surface of the biomass material and the iron scrap were covered with a layer of black SRB biofilm. The water was injected from bottom with the peristaltic pump and flew out from the top. The influent flow rate, controlled by a flowmeter, was 0.7 mL/min. The experimental device was shown in Fig. 1. The hydraulic retention time of the three dynamic columns was all 24 h. The devices ran continuously for 30 d.

The concentration of Cr⁶⁺ was measured by diphenyl carbazide spectrophotometry (GB 7467-87), the concentration of Cr³⁺ was measured by potassium permanganate oxidation-diphenyl carbazide spectrophotometry (GB 7466-87), the concentration of SO₄^2– was measured by barium chromate spectrophotometry (HJ/T 432-2007), the value of pH was measured by glass electrode, and the COD was measured by rapid digestion spectrophotometry (HJ/T 399-2007) [19].

The morphology of the filling sample was observed by ZEISS Sigma 500 scanning electron microscope (SEM), in the meantime the chemical substances on the surface of the filling sample were measured by energy-dispersive X-ray spectroscopy probe (EDS). Phase analysis was carried out by Bruker D8 Advance X-ray Diffraction (XRD) with a scan step of 10°–90°.

The concentration and removal rate of SO₄^2– in the three dynamic columns are shown in Fig. 2. Fig. 2 shows that the average residual amount of SO₄^2– in dynamic columns 1, 2 and 3 were 465.2, 304.4 and 662.8 mg/L and the average removal rates of SO₄^2– were 54.3%, 74.9% and 32.5% respectively. In the initial stage, the SO₄^2– removal rates in three dynamic columns were lower, but their curve of removal rate increased rapidly. The maximum removal rates of SO₄^2– were 58%, 86.2% and 44.1% on the 4th, 6th and 5th days, respectively, and then the removal rate curve show a trend of fluctuation downtrend. The removal process of SO₄^2– is mainly the dissimilation reduction of SRB [20]. In those dynamic columns, the carbon source, which was required for the dissimilatory reduction of SRB, was generated by degradation of the decomposition of natural cellulose and lignin contained using the anaerobic fermentation microorganism [21]. In the initial stage, the microorganisms have poor biological activity due to the poisoning of heavy metal ions and the inhibition of acidity. Therefore, the removal rate of SO₄^2– is lower. The iron scrap in dynamic columns can reduce Cr⁶⁺ and adjust acidity, so it can reduce the inhibition of microorganisms by heavy metal ions and acidity [22]. Meanwhile the hydrogen released during the reaction can provide electricity for the growth of microorganisms. Due to rapid recovery of biological activity in the three dynamic columns, the capacity of the dissimilation reduction of SRB was also increased quickly. According to Fig. 6, because of COD release decreased gradually in sugarcane slag, corn cob and sunflower straw and the surface of iron scrap gradually passivated in the three columns [23], the biological activity of SRB decreased gradually. Therefore, the curves of SO₄^2– removal rate were fluctuating downward.

The concentration and removal rate of Cr⁶⁺ in the three dynamic columns are shown in Fig. 3. Fig. 3 shows that the average residual amount of Cr⁶⁺ were 0.3, 0.4 and 0.5 mg/L and the average removal rates of Cr⁶⁺ were 96.9%, 96.1% and 95.2% in dynamic columns 1, 2 and 3, respectively. From the initial stage to the 15th day, the average removal rates of Cr⁶⁺ kept a high level in the three dynamic columns, and then the removal rate curves were a trend of fluctuation downtrend, But the scale of downtrend is smaller, the removal rate is more than 92%. The removal process of Cr⁶⁺ has three main processes in dynamic columns, the first one is the reduction of iron scrap and their reaction products [22], and the second one is the adsorption of biomass materials and the reduction of their hydrolysate [24]. Thirdly H₂S as a metabolite of SRB has a strong reducibility, along with extracellular polymeric substance of SRB both can absorb Cr⁶⁺. Due to the stronger reduction ability of iron scrap, high-valent Cr⁶⁺ can be rapidly reduced to low-valent Cr³⁺. At the same time, the FeS generated by the reduction of iron scrap has been proved to be highly capable of removing heavy metals [25]. In addition, EPS generated by SRB contain a great amount of anionic group (carboxyl, hydroxyl and amidogen, etc.), which can improve the combination of cell surface and heavy metal [26,27]. The mercapto group, amino group, amide group and hydroxyl group in sugarcane slag, corn cob and sunflower straw have a strong affinity for heavy metal ions [28]. And its anaerobic fermentation products such as glucose, fructose have the characteristics of reducibility to Cr⁶⁺ [29]. Based on the above, the average rate that three dynamic columns remove Cr⁶⁺ is higher. Due to the surface passivation of iron scrap, the adsorption of biomass tending to saturation and the decrease of SRB metabolism SO₄^2– ability, the removal ability of Cr⁶⁺ was gradually weakened in the three dynamic columns. So the removal rate curves of Cr⁶⁺ was a trend of small fluctuations downtrend. Bagasse is capable of absorbing Fe²⁺ [30]. The studies have shown that the rank of the amount of glucose and fructose produced by unit mass biomass materials is different: bagasse (0.089 g/g), corn cob (0.037 g/g), sunflower straw (0.014 g/g) [31]. The Fe²⁺ absorbed on the bagasse surface as well as glucose and fructose from anaerobic fermentation conducted a redox reaction with Cr⁶⁺, which made the removal rate in dynamic column 1 is slightly better than the ones in dynamic columns 2 and 3.

The concentration and removal rate of Cr³⁺ in the three dynamic columns are shown in Fig. 4. Fig. 4 shows that the average residual amount of Cr³⁺ in dynamic columns 1, 2 and 3 were 6.9, 3.0 and 8.8 mg/L and the average removal rates of Cr³⁺ were 67.1%, 85.6% and 57.8% respectively. In the initial stage, the removal rate of Cr³⁺ decreased and then rose rapidly in the three dynamic columns. The maximum removal rates were 74.0%, 91.7% and 63.9% on the 5th days, respectively, and then the removal rate curves were a trend of fluctuation downtrend. The removal process of Cr³⁺ is mainly the precipitation of S^2– by SO₄^2– reduction, at the same time, the reduction process of Cr⁶⁺ will generate new Cr³⁺ [32]. Because the amount of S^2– generated was lower due to the lower dissimilatory reduction activity of SRB, and a large amount of Cr³⁺ was generated by the iron scrap reduction, the removal rate curves of Cr³⁺ decreased in the initial stage. However, due to the enhancement of dissimilatory reduction activity of SRB, a large amount of S^2– had been generated to precipitate more Cr³⁺, the removal rate curve of Cr³⁺ raised. When the amount of S^2– generated by dissimilation reduction of SRB was gradually reduced due to the reduction of C/S in dynamic columns, the removal rate curves of Cr³⁺ were a trend of fluctuation downtrend. Compared with Figs. 2 and 4, the order of the average removal rates of SO₄^2– (2 > 1 > 3) is highly consistent with the order of the average removal rate of Cr³⁺ (2 > 1 > 3). It can be seen that the removal capability of Cr³⁺ is positively correlated with dissimilation reduction ability of SRB in the three dynamic columns.

The pH values in the three dynamic columns are shown in Fig. 5. The average pH of the effluent was 6.98, 6.23 and 7.96 respectively in dynamic columns 1, 2 and 3. At the initial stage, the pH value of the AMD solution increased quickly in the three dynamic columns. The pH value was affected by the reduction of iron scrap and the combination of H⁺ with ligands, which in the biomass materials on the surface. The difference between ligands binding and H⁺ directly results in the obvious difference of the pH of effluent. On the first day the pH had ranged from the original 4.23 to 6.37, 6.14 and 8.32 respectively. It can be seen that the metabolic activity of SRB hasn’t reached its optimal state according to the effect of SO₄^2– removal. Because the iron scrap had the reduction reaction with the H⁺ from AMD in the acid environment, the pH value increased rapidly. The reason why pH value in dynamic column 3 was higher than the ones in dynamic columns 1 and 2, is that the interior pores of sunflower straw are thin and loose, and there is a larger specific surface area and more points to adsorb H⁺. the sunflower straw is superior to bagasse [33]. The previous studies show that the combination of bagasse with H⁺ is superior to the corn cob [14]. For H⁺, when pH is low, it is easy to bind to adsorption sites on the surface of sunflower [34]. After sunflower straw adsorbs a large amount of H⁺, the unit area of the remaining H⁺ from the contact reaction with iron filings in dynamic column 3 system increases, which promotes the further consumption of H⁺, improves the pH value of the system, reduces the toxicity of acidity to SRB, and promotes the growth of SRB and alkaline substances. In the middle stage of the reaction, the H⁺ binding site on the surface of biomass material decreases. The reaction between iron filings and H⁺ tends to be stable. The SRB activity in the system is enhanced, and the SRB metabolism also produces alkalinity, which makes the pH value of the effluent from dynamic columns 1–3 tends to be stable. However, the removal rates of H⁺ is reduced due to the passivation of iron scrap surface and the decrease of SRB bioactivity in the later stage. Therefore, the pH value curves of the effluent showed a slow downward trend.

The released amount of COD is shown in Fig. 6 in the three dynamic columns. Fig. 6 showed that the average COD concentration of effluent were 287.3, 636.5 and 158.1 mg/L respectively in columns 1, 2 and 3. In the initial stage, the effluent COD values of the three dynamic columns (especially columns 1 and 3) were lower, but then the curve of COD concentration value rose more rapidly. On the 6th day, the COD values of the effluent reached the maximum value of 332, 708 and 233 mg/L, respectively, and then the COD value of the effluent showed a downward trend. The process of COD released is mainly caused by the decomposition of natural cellulose and lignin in biomass materials by anaerobic fermentation microorganisms [35], and the hydrogen released from iron reaction can promote microbial fermentation activity, but the metabolism of SRB can consume part of COD. Therefore, in the initial stage, the anaerobic fermentation products of biomass materials increased gradually with the enhancement of anaerobic microorganism activity, and the COD value curve of effluent increased rapidly. With the progress of the reaction, the amount of cellulose (hemicellulose) that was easily degraded in the biomass material gradually decreased [36], and the output of hydrogen was also reduced owing to the passivation of the iron scrap surface. The amount of biodegradable cellulose and hemicellulose from the biomass material rank as followings: corn cob (0.37 and 0.32 g/g) [37], bagasse (0.35 and 0.25 g/g) [38], and sunflower straw (0.27 and 0.16 g/g) [39]. It can be seen from Fig. 6 that the amount of the COD value in the three dynamic columns is dynamic column 2 > dynamic column 1 > dynamic column 3. The gap of COD value is due to the different amount of biodegradable cellulose and hemicellulose. SRB needed to consume COD when the SO₄^2– was transformed in the anaerobic environment. The appropriate amount of COD can improve the growth of SRB, while the excess COD may lead to the secondary pollution. The COD released from the corn cob in dynamic column 2 is desirable for the growth of SRB according to the SO₄^2– removal in the three dynamic columns. The content of COD is still higher after the utilization of SRB, which can lead to the secondary pollution. The research also shows that the bagasse in dynamic column 1 can provide a more durable carbon source compared with the sunflower straw in dynamic column 3 and help SRB metabolize SO₄^2–. In the meantime, the less amount of COD in dynamic column 1 can reduce the possibility of secondary pollution.

The released amount of TFe is shown in Fig. 7 in the three dynamic columns. The average TFe concentration of effluent were 4.4, 15.8 and 1.1 mg/L respectively in dynamic columns 1, 2 and 3. On the first day, the TFe concentration of the effluent in the three dynamic columns were 5.37, 17.38 and 1.42 mg/L respectively. Compared with dynamic column 1 and dynamic column 3, the TFe concentration from dynamic column 2 was significantly higher. There were three main reasons for this phenomenon. Firstly, the effluent pH value of dynamic column 2 is lower than that of dynamic column 1 and dynamic column 3. Secondly, iron filings in the acidic environment of dynamic column 2 formed Fe²⁺ and Fe³⁺, and Fe²⁺ and Cr⁶⁺ generated Fe³⁺ and Cr³⁺ through redox reaction, which promoted the iron filings to move towards the reaction direction of Fe²⁺. Finally, the activity of SRB in dynamic column 2 was significantly higher than that in dynamic column 1 and dynamic column 3. The metabolism of SRB with high activity produced a large amount of S^2–, and the reaction of S^2– with Cr³⁺ generate precipitation, which promoted the iron filings to react in the direction of Fe²⁺ formation. In the initial stage, the effluent TFe concentration of the dynamic columns 1 and 2 (especially column 2) were higher, but then the curve of the TFe concentration decreased slowly. At the 11th and 17th days, the effluent TFe concentration reached the minimum of 2.6 and 14.0 mg/L respectively, and then the TFe concentration of the effluent showed a rising trend. However, the effluent TFe concentration of dynamic column 3 remained relatively low and the curve fluctuated gently. The release process of TFe in the dynamic columns was mainly the redox reaction of iron scrap with Cr⁶⁺ and H⁺, and the consumption of TFe was mainly the adsorption of biomass materials and precipitation of S^2– and OH^–. Therefore, due to the rapid reaction of a large number of iron scrap and relatively weak biological activity in the early stage, the release of TFe is higher in effluent. With the progress of the reaction, the amount of S^2– produced by the dissimilation reduction become smaller because of the gradual weakening of the biological activity from strong. Therefore, the TFe concentration curves of the dynamic column 1 and 2 first dropped and then rose. At the same time, because the effluent of dynamic column 3 was weakly alkaline, and the sunflower straw has sparse pores and loose pores, which has larger specific surface area and more adsorption sites, the effluent TFe concentration of the dynamic column 3 is always kept low.

The fillings of dynamic columns 1, 2 and 3 were dried and grinded to be analyzed by EDS. The results were shown in Fig. 8.

Fig. 8 showed that the fillings of the dynamic column 1 mainly contained elements C, O, Al, Si and Fe before the reaction, and their weight percentage was 39.26%, 11.51%, 0.30%, 1.69% and 47.23%, respectively. After reaction, the fillings mainly contained elements C, O, S, Ca, Cr, Mn and Fe, and their weight percentages were 13.67%, 23.93%, 3.04%, 0.32%, 1.18%, 0.12% and 57.74%, respectively. The fillings of the dynamic column 2 mainly contained elements C, O, Si and Fe before the reaction, and their weight percentages were 24.42%, 6.17%, 1.98% and 67.44% respectively. After reaction, the fillings mainly contained elements C, O, Si, S, Cl, Ca, Cr and Fe, and their weight percentages were 12.83%, 27.39%, 1.29%, 6.24%, 0.30%, 0.44%, 5.00% and 46.51% respectively. The fillings of the dynamic column 3 mainly contained elements C, O, Si and Fe before the reaction, and their weight percentages were 26.92%, 6.06%, 2.02% and 65.00% respectively. After reaction, the fillings mainly contained elements C, O, Al, Si, S, Cr and Fe, and their weight percentages were 3.03%, 5.11%, 0.48%, 2.68%, 1.03%, 0.58% and 87.09% respectively.

The fillings of dynamic columns 1, 2 and 3 were dried and grinded to be analyzed by XRD. The results were shown in Fig. 9.

Fig. 9 showed that the fillings of dynamic columns 1, 2 and 3 mainly contained elements iron and organic matter before the reaction. After reaction, FeS and Cr₂S₃ appeared in the fillings.

The fillings of dynamic columns 1, 2 and 3 were dried and analyzed by SEM. The results were shown in Fig. 10.

Fig. 10 showed that the surface texture of the filling was smooth and regular, and their pores were clear and smooth before the reaction. After reaction, the surface of the filling appeared obvious folds, the pores became smaller, and there was obvious granular sediment on the surface.