Recovery of Metals from Waste Lithium Ion Battery Leachates Using Biogenic Hydrogen Sulﬁde

: Lithium ion battery (LIB) waste is increasing globally and contains an abundance of valuable metals that can be recovered for re-use. This study aimed to evaluate the recovery of metals from LIB waste leachate using hydrogen sulﬁde generated by a consortium of sulfate-reducing bacteria (SRB) in a lactate-fed ﬂuidised bed reactor (FBR). The microbial community analysis showed Desulfovibrio as the most abundant genus in a dynamic and diverse bioreactor consortium. During periods of biogenic hydrogen sulﬁde production, the average dissolved sulﬁde concentration was 507 mg L − 1 and the average volumetric sulfate reduction rate was 278 mg L − 1 d − 1 . Over 99% precipitation e ﬃ ciency was achieved for Al, Ni, Co, and Cu using biogenic sulﬁde and NaOH, accounting for 96% of the metal value contained in the LIB waste leachate. The purity indices of the precipitates were highest for Co, being above 0.7 for the precipitate at pH 10. However, the process was not selective for individual metals due to simultaneous precipitation and the complexity of the metal content of the LIB waste. Overall, the process facilitated the production of high value mixed metal precipitates, which could be puriﬁed further or used as feedstock for other processes, such as the production of steel.


Introduction
Globally, there has been an exponential growth in lithium ion battery (LIB) consumption, with production expected to increase 520% from 2016 to 2020 [1]. LIBs have become an ideal energy storage component for portable devices and electric vehicles (EVs) due to their high energy density and lightweight nature and are expected to dominate the battery market for the next 20 years [2].
Crude recycling and disposal practices of electronic waste (e-waste; such as LIBs) often result in the production of toxic substances (e.g., dioxins) and the leaching of toxic heavy metals into the environment [3][4][5]. Currently, there are only a few commercial operations capable of recovering metals from LIB waste, largely located in Asia and Europe [6]. In 2012 and 2013, only 2% of LIBs were collected for recycling in Australia, with the remainder still largely disposed of to landfill [2]. Considering the abundance of valuable metals LIBs contain (Table 1) [7], as well as the potentially damaging impact of LIB waste if disposed to the environment, recovering metals from LIB waste is both environmentally and economically attractive. Table 1. Metals recoverable from LIB waste, along with the percentage composition, commodity prices and total value of metals contained in LIBs (Table modified from [7]). Commodity prices correct as at 13 August 2019 and were sourced from Metalary. To recover metals from e-waste such as LIBs, the techniques of pyrometallurgy and hydrometallurgy are traditionally employed [8]. Typically, pyrometallurgy is attractive for large-scale operation as LIBs can be processed with other forms of e-waste [9]. The process generally aims to predominantly recover nickel (Ni) and copper (Cu) at high temperatures, which make these processes economically feasible [9]. Therefore, traditional metal recovery processes may not recover all metals associated with complex waste products, such as waste LIB. For example, high temperature treatment does not allow for the recovery of other metals (such as lithium (Li), iron (Fe), aluminium (Al), and manganese (Mn) and the processing can be energy intensive [9].

Metal
In comparison, hydrometallurgical processing allows the recovery of most of the metals within LIBs to be extracted, separated and recovered [10]. Leaching of metals from LIB waste have been widely investigated using various lixiviants to promote metal solubilisation [7,11,12]. It has been demonstrated that up to 99% solubilisation of cobalt (Co) and Li can be achieved with LIB wastes using strong inorganic acids such as hydrochloric, sulfuric and nitric acids [9,13,14]. Often in these studies, other recoveries for base metals, Al and Fe are also reported, but the recovery is dependent on the leach conditions implemented for the spent LIB [9]. Similarly, literature demonstrates that between 96% and 100% of Co and Li can be recovered in leach experiments using common organic acids such as citric, succinic and oxalic acids [15][16][17].
Regardless of the metal solubilisation method employed, the recovery of metals from the complex metal-containing leachate solutions can be challenging. Solutions often contain various metals of complex chemistries and co-digested compounds associated with binders and other materials. Various methods to recover metals from LIB waste have also been investigated including precipitation [18], solvent extraction, [19,20], membrane separation [21] and electrodeposition [22][23][24]. Using these methods, it has previously been shown that the separation of metal combinations such as Ni and Co and Cu, and Co and Mn are notoriously challenging due their similar chemical and physical properties [18,21].
When combined, these metal extraction and recovery processes are associated with the consumption of large amounts of reagents or high energy inputs. Accordingly, there has been an increasing interest in developing more environmentally benign processes for metal recovery [25][26][27].
Recently, the recovery of metals from various metal containing waste streams using biogenic sulfide produced by sulfate reducing bacteria (SRB) has been explored as an alternative to the conventional metal precipitation processes [28][29][30][31]. In biogenic sulfide generation, the SRB use sulfate as a terminal electron acceptor to oxidise organic compounds or hydrogen to form hydrogen sulfide (Equation (1)) [29]. The hydrogen sulfide gas that is produced is then used to precipitate metals as low solubility metal sulfides (Equation (2)) [29].
The organic substrate (electron donor) is represented by CH 2 O. In both Equations (1) and (2), the soluble metal cation is represented by M 2+ .
Biogenic sulfide precipitation of metals has been demonstrated at scale, with various sulfate-and metal-rich source waters from contaminated and environmental sources, including acid mine drainage, spent refinery waste waters, and final slag leachates [32]. Using sulfide to precipitate metals from low concentration, highly complex, metal containing solutions allows more selective precipitation of target metals as compared to hydroxide precipitation [31], and hence is more beneficial for complex metal containing solutions such as those generated by the leaching of waste LIBs and other e-wastes.
To the best of our knowledge, biogenic sulfide precipitation of metals from LIB waste leach liquors has not yet been investigated. Therefore, this study explored the feasibility of using biogenic sulfide precipitation to recover metals from leach solutions generated from LIB waste. Further, it was anticipated that the use of biogenic sulfide precipitation may provide an alternative pathway for recovering metals from complex metal containing leach liquors generated from other forms of e-waste, such as printed circuit boards and fluorescent lights. As resource recovery is becoming increasingly important for alleviating the environmental impact of e-wastes, the development of alternative resource recovery processes that consume less materials and energy is crucial to assist the transition towards a circular economy for these products.

LIB Waste Processing and Composition
Waste LIBs from laptops were collected from a local e-waste processing facility (Total Green Recycling Ltd. Co., Perth, WA, Australia) and were manually dismantled into the plastic housings and battery cells as previously described [7]. Briefly, the battery cells were further processed by course shredding (Shredder Type 600, Kovia Engineering Works, Coimbatore, India) and milling (Essa LM5 mill, FLSmidth, Copenhagen, Denmark). Oversized metal casings were then removed by screening at 16 mm, 10 mm and 5 mm and material <5 mm was re-milled. The particle size fraction of <500 µm was used in the study. The composition of the battery waste (Table 1) was previously determined by X-ray diffraction (XRD) and electron scanning microscopy at CSIRO Mineral Resources Analytical Services Laboratory (Perth, WA, Australia).

Leaching of LIBs
Processed LIB powder (pulp density 10% w/v) was leached at 25 • C using 2 N HCl, with shaking at 200 rpm for 24 h. At the completion of leaching, the metal-rich leach solution was recovered by vacuum filtration (0.22 µm) and used for subsequent metal precipitation experiments. The pH and oxidation-reduction potentials (ORP) were recorded at the start and completion of the leaching using a Smart-CHEM meter (Ag/AgCl reference electrode; calibrated with 4.01 and 7.01 standard buffers TPS, Brendale, QLD, Australia). At the completion of leaching, samples were obtained for soluble metals analysis (Al, Ni, Co, Li, Fe, Mg, Mn, Cd, Zn, Cu) by inductively coupled plasma optical emission spectrometry (ICP-OES), undertaken at CSIRO Mineral Resources Analytical Services Laboratory.

Bioreactor Operation
The FBR (height 36 cm, internal diameter 5 cm, total volume 0.78 L) and associated sparging reactor (SR) (height 32 cm, internal diameter 5 cm, total volume 0.46 L) had a total liquid volume of 1.24 L (Figure 1). Granular activated carbon (GAC; 0.5-1 mm; 50% v/v) was added to the FBR as the biomass carrier to facilitate biofilm formation, and the reactor was fluidised between 5% v/v and 10% v/v [30,35]. Fluidisation was maintained by recirculation of the bioreactor medium at flow rates determined by the fluidisation rate. Glass beads were placed at the bottom of the reactor to prevent any backflow of the carrier material into the recirculation line.
The FBR was maintained at 35 • C by circulating heated water through tubing wrapped around the reactor, which was also well insulated to minimise heat loss. The FBR was initially operated in batch mode for 15 days to facilitate biofilm formation at the GAC surface. The FBR was then continuously fed with the modified, minimal Postgate medium, at a hydraulic retention time (HRT) between 2.09 days and 2.16 days. The HRT was calculated using the active fluidised bed volume (1.24 L). The pH was monitored continuously by an in-line pH probe (Ionode IJ44A, Tennyson, QLD, Australia; Ag/AgCl reference electrode). The pH of the influent medium was not corrected prior to addition to the reactor and had a starting pH of 4.3. The reactor pH was naturally maintained at approximately 7 because of biogenic alkalinity generation [36,37], but on some occasions when process perturbations were encountered, the pH was manually adjusted to neutral by adding 1 M H 2 SO 4 or 1 M NaOH. Microbial activity was measured as dissolved sulfide generation using the colorimetric method described by Cord-Ruwisch [38].

Bioreactor Operation
The FBR (height 36 cm, internal diameter 5 cm, total volume 0.78 L) and associated sparging reactor (SR) (height 32 cm, internal diameter 5 cm, total volume 0.46 L) had a total liquid volume of 1.24 L (Figure 1). Granular activated carbon (GAC; 0.5-1 mm; 50% v/v) was added to the FBR as the biomass carrier to facilitate biofilm formation, and the reactor was fluidised between 5% v/v and 10% v/v [30,35]. Fluidisation was maintained by recirculation of the bioreactor medium at flow rates determined by the fluidisation rate. Glass beads were placed at the bottom of the reactor to prevent any backflow of the carrier material into the recirculation line.
The FBR was maintained at 35 °C by circulating heated water through tubing wrapped around the reactor, which was also well insulated to minimise heat loss. The FBR was initially operated in batch mode for 15 days to facilitate biofilm formation at the GAC surface. The FBR was then continuously fed with the modified, minimal Postgate medium, at a hydraulic retention time (HRT) between 2.09 days and 2.16 days. The HRT was calculated using the active fluidised bed volume (1.24 L). The pH was monitored continuously by an in-line pH probe (Ionode IJ44A, Tennyson, QLD, Australia; Ag/AgCl reference electrode). The pH of the influent medium was not corrected prior to addition to the reactor and had a starting pH of 4.3. The reactor pH was naturally maintained at approximately 7 because of biogenic alkalinity generation [36,37], but on some occasions when process perturbations were encountered, the pH was manually adjusted to neutral by adding 1 M H2SO4 or 1 M NaOH. Microbial activity was measured as dissolved sulfide generation using the colorimetric method described by Cord-Ruwisch [38].

DNA Extraction and Analysis
Microbial community composition of the FBR was characterised at various times of operation. DNA was extracted from both suspended and attached biomass. Suspended biomass from the bioreactor liquor was collected in a known liquid volume and concentrated prior to DNA extraction. The biomass carrier samples (5 mL) were taken from the fluidised bed. The wet weight of the collected GAC sample was recorded (Table 2), and the samples were sonicated in the bioreactor liquor for three

DNA Extraction and Analysis
Microbial community composition of the FBR was characterised at various times of operation. DNA was extracted from both suspended and attached biomass. Suspended biomass from the bioreactor liquor was collected in a known liquid volume and concentrated prior to DNA extraction. The biomass carrier samples (5 mL) were taken from the fluidised bed. The wet weight of the collected GAC sample was recorded (Table 2), and the samples were sonicated in the bioreactor liquor for three  Cells from both suspended and attached biomass were concentrated to 250 µL by centrifuging at 10,000× g for 8 min (Centrifuge 5424R, Eppendorf, Macquarie Park, NSW, Australia). DNA was extracted using the DNeasy Powersoil Kit (Qiagen, Chadstone, VIC, Australia) as per the manufacturer's instructions. The protocol was amended to incubate the membrane filter in elution buffer for an extended period of 10 min before centrifugation. The DNA concentrations were analysed using a Qubit 3.0 Fluorometer (Invitrogen, Thermo Fisher Scientific, Scorseby, VIC, Australia).
DNA samples were mixed in 4:1 ratio with DNA Stable (Sigma-Aldrich, Castle Hill, NSW, Australia) and sent to Molecular Research Laboratory (MR DNA, Shallowater, TX, USA) for sequence analysis using Illumina sequencing. Briefly, DNA was amplified using the HotStarTaq Plus Master Mix Kit (Qiagen) with universal 16S rRNA gene primers, illCUs515F (5 GTGYCAGCMGCCGCGGTAA 3) and new806RB (5 GGACTACNVGGGTWTCTAAT 3). The conditions for amplification were as follows: initial denaturation at 94 • C for 3 min, followed by 30-35 cycles of denaturation at 94 • C for 30 s, primer annealing at 53 • C for 40 s and elongation at 72 • C for 1 min, after which a final elongation step at 72 • C for 5 min was performed. Following amplification, the PCR products were examined on a 2% agarose gel. Samples were purified using calibrated Ampure XP beads. Purified PCR products were used to prepare an Illumina DNA library. Sequencing was performed on a MiSeq following the manufacturer's guidelines. Sequence data was processed using MR DNA's analysis pipeline, and percentage abundance of dominant community members was calculated for both bacterial and archaeal genera in the suspended and attached biomass communities.

Biogenic Sulfide Precipitation of Metals
The metal-rich leachate (100 mL) was added to a 250 mL precipitation flask and magnetically stirred for the duration of the experiment. During the initial 24 h (Period I), the pH was uncontrolled. Thereafter, the pH of the leach solution was sequentially increased by computer-feedback dosing of 1 M NaOH to two different set-points, namely pH 5 (Period II) and pH 10 (Period III). Feedback control was via a customised LabVIEW program and a controllable peristatic pump (77120-52, Cole Parmer Instrument Company, Vernon Hills, IL, USA).
During the experiment, the SR unit was continuously sparged with an analytical grade N 2 gas at a flow rate of 0.1 L min −1 . The recovered H 2 S gas was continuously added to the leachate for the duration of the experiment to precipitate metals at each pH set point (Periods I-III). Samples were routinely taken to measure the dissolved sulfide in the FBR [38] and the soluble metal concentrations (ICP-OES) in the precipitation flask. Samples for soluble metal analysis were prepared by membrane filtration (0.22 µm; Merck Millipore, Bayswater, VIC, Australia) before fixing with 7 M HNO 3 . During the sparging process, the FBR was maintained in batch mode, and the pH was adjusted to pH 7 using 1 M H 2 SO 4 , as required.

Biogenic Sulfide Precipitation Efficiency
The precipitation efficiencies for each metal were calculated using Equation (3), where [m] soluble is the soluble metal concentration [29]: The purity of metal precipitates obtainable, for each metal, were calculated as purity indices using Equation (4), where m i is the mass of an individual metal precipitated between pH points, and m total is the mass of all metals precipitated between the same pH points [29]. The mass of metals was calculated based on metal concentrations in solution: Precipitate purity indices = m/m total (4)

Bioreactor Performance and Troubleshooting
The FBR performance over a period of approximately 200 days is shown in Figure 2.

Biogenic Sulfide Precipitation Efficiency
The precipitation efficiencies for each metal were calculated using Equation (1), where [m]soluble is the soluble metal concentration [29]: Metal precipitation efficiency (%) = ([m]soluble (before H2S addition) − [m]soluble (after H2S addition)) × 100/[m]soluble (before H2S addition) Equation (3) The purity of metal precipitates obtainable, for each metal, were calculated as purity indices using Equation (2), where mi is the mass of an individual metal precipitated between pH points, and mtotal is the mass of all metals precipitated between the same pH points [29]. The mass of metals was calculated based on metal concentrations in solution: Precipitate purity indices = m/mtotal Equation (4)

Bioreactor Performance and Troubleshooting
The FBR performance over a period of approximately 200 days is shown in Figure 2. In general, the performance varied as reflected by the fluctuations in both the measured dissolved sulfide concentrations and pH. Sulfate reduction rates at times were measured as negative values. This is likely due to the carryover of substrates from inoculum, concentration of sulfate due to evaporation during sparging events and other bioreactor events and/or an artefact of operation of the bioreactor whilst in batch mode.
3.1.1. Dissolved Sulfide, pH and Temperature as Indicators of Microbial Activity The FBR produced high dissolved sulfide concentrations after approximately two months of operation, reaching a level of up to 860 mg L −1 at day 72 ( Figure 2A). During periods of sulfide production (dissolved sulfide concentrations >100 mg L −1 ), the average dissolved sulfide concentration was 507 mg L −1 , which is within the typical range reported for sulfidogenic FBRs [30,37,39]. Although this level of dissolved sulfide is close to those reported as inhibitory to SRB (ca. 550 mg L −1 ) [40], it has also been shown that SRB can oxidise substrates at high dissolved sulfide concentrations (up to 1000 mg L −1 ), albeit at a lower rate [41]. Therefore, the dissolved sulfide concentration achieved by the FBR was considered satisfactory for this work.
During periods of sulfide production, the bioreactor pH and temperature were relatively stable ( Figure 2C). However, when the process was not actively producing sulfide (days 90 and 160), notable fluctuations in temperature or pH were recorded, which resulted in delayed recovery of the bioreactor. To expedite the recovery of the bioreactor, the bioreactor was re-inoculated with an active SRB culture on three occasions (on days 61, 92 and 163). Following re-inoculation, the reactor recovered to dissolved sulfide concentration >100 mg L −1 within 8-10 days.

Installation of a Separate Sparging Reactor Facilitated Biogenic H 2 S Production in the FBR
Prior to installing the sparging reactor (SR; Figure 2: day 104), the sparging device was housed directly within the FBR bed. However, sparging the FBR directly with N 2 gas to strip H 2 S resulted in a process failure of the FBR as dissolved sulfide concentration sharply declined (Figure 2: days 61 and 76). The vigorous nature of sparging within the reactor bed may have detached SRB from the carrier material and consequently, the SRB may have been washed out of the reactor system, while bacteria with greater adherence to the carrier material remained [41,42]. The loss of SRB from the bioreactor system would account for the lengthy bioreactor recovery times following sparging events. As the aim of the study was to maximise sulfide generation and not to evaluate the suitability of other biomass carrier materials, the biomass carrier used in this study was not changed to promote stronger attachment of cells. Activate carbon has been reported to be an efficient biomass carrier for FBR [43,44].
To remove H 2 S from the reactor without displacing SRB from the biocarrier material, a separate SR was installed in-line with the FBR on day 104 ( Figure 1). After installing the SR, two sparging events were carried out (Figure 2: days 155 and 181). Following the sparging on day 155, the bioreactor failed to recover for over 10 days. However, this was due to simultaneous interruption to influent flow, and hence, the likely starvation of the microbial community [34]. The bioreactor performance rapidly increased with no intervention following the sparging event on day 181. This indicated that the use of a separate SR allowed the recovery of H 2 S without impacting the activity of the microbial community in the FBR.

Ratio of Sulfate to Lactate was Important for Optimising Sulfide Generation
For the first 56 days of bioreactor operation, the sulfate:lactate mass ratio of the growth medium was 1:2 to enrich and promote biomass growth. However, during this period of operation, only low dissolved sulfide concentrations were recorded, which suggested that growth conditions were not optimal for biogenic sulfide production. The initially low sulfate to lactate ratio may have led to sulfate limitation, allowing acetogens and methanogens to outcompete SRB for substrates [35]. Low sulfate to organic carbon ratio can also result in the accumulation of propionate and acetate in the effluent, which may inhibit sulfidogenisis [34,45,46]. At 56 days, the growth medium was modified to have a sulfate: lactate mass ratio of 1:0.35. Higher sulfate concentrations when compared to lactate have been shown to promote the growth of SRB such as Desulfovibrio and Desulfomicrobium and is closer to the stoichiometric molar ratio of 3:2 (i.e., mass ratio of 1:0.63) for the complete sulfidogenic oxidation of lactate to bicarbonate (Equation (5)) [29,34]: Within 24 h of changing to the modified, minimal Postgate medium, a detectable dissolved sulfide concentration was recorded, and the concentration continued to increase until day 76 when N 2 sparging disrupted the microbial activity (Figure 2: days 61-76). The installation of the SR (day 104), in combination with the use of a minimal, modified Postgate medium allowed the reactor to stabilise over time, and the reactor to recover activity more quickly in times of perturbation.

Sulfate Reduction Rate and Efficiency
During periods of sulfide production (dissolved sulfide concentrations >100 mg L −1 ), the volumetric sulfate reduction rate varied between 170 and 633 mg L −1 d −1 , with an average of 378 mg L −1 d −1 ( Figure 2B) and the sulfate reduction efficiency varied between 21% and 86% with an average of 50% ( Figure 2B). Although with lower sulfate reduction rates, the sulfate reduction efficiencies were similar to those reported for other FBRs [30,35,37,39,47]. This is likely due to the longer HRT used in this study (2 days), when compared to these other studies. However, Kaksonen et al. [41] indicated that the relationship between sulfate reduction rate and HRT was not linear. Instead, the sulfate reduction rate likely depends on several variables, including sulfate concentration, organic substrate concentration, dissolved sulfide concentration, biomass microbial community, pH and temperature [41]. An increased sulfate reduction rate in the sulfidogenic bioreactor indicated an increased rate of sulfide production. As the bioprecipitation of metals from solution depends on available H 2 S, a greater sulfate reduction rate may increase the capacity of sulfidogenic bioreactors for metal recovery.

Microbial Community Structure
The performance of the FBR was dependent on the microorganisms present, their ability to form consortia on the carrier material and their regeneration capacity [48]. A better understanding of the microbial community present in sulfidogenic bioreactors may provide insight into the reproducibility of this process and the potential to tailor bioreactor conditions for increased performance. In this study, samples for DNA sequencing were taken from both biomass that was suspended in the reactor volume and that attached to the carrier material. The samples were taken 20 days apart during variable bioreactor performance (Table 2). Samples A and B were taken during a period of bioreactor recovery where the dissolved sulfide concentration was undetectable, and the sulfate reduction efficiency was only 1.6%. Samples C and D were taken during an uninterrupted period of sulfide production, when the dissolved sulfide concentration was 762 mg L −1 and, the sulfate reduction efficiency was 65%. Finally, samples E and F were taken within 24 h of an interruption to influent flow, when the dissolved sulfide concentration was 590 mg L −1 and, the sulfate reduction efficiency was 55%.
The relative abundance data for bacteria and archaea from the suspended biomass samples (Table 2: samples B, D and F) showed minor changes in community composition as bioreactor performance varied. A biofilm consortium can provide a greater resistance to environmental changes compared to a suspended culture [48]. Also, microbes in suspension can easily be washed out from the reactor system [49].
DNA sequencing and relative abundance calculations showed that a diverse, changing and complex bacterial community was present within the FBR (Figure 3; Table 3). Nonetheless, the FBR microbial community, previously enriched from environmental sources and industrial samples, contained common SRB from the genera Desulfovibrio, Desulfomicrobium and Desulfococcus. The most dominant genus present in the bioreactor was Desulfovibrio (ranging from 15% to 22%), followed by Desulfomicrobium (ranging from 8% to 27%), whilst Desulfococcus was a minor representative of the SRB community (ranging from 4% to 15%), depending on the sample and time taken from the reactor (Figure 3). These genera have been previously shown to dominate sulfidogenic bioreactors [35,48]. In general, SRB made up between 31% and 58% of the FBR bacterial community, depending on the sampling time (Figure 4). Minerals 2019, 9, x FOR PEER REVIEW 9 of 19 contained common SRB from the genera Desulfovibrio, Desulfomicrobium and Desulfococcus. The most dominant genus present in the bioreactor was Desulfovibrio (ranging from 15% to 22%), followed by Desulfomicrobium (ranging from 8% to 27%), whilst Desulfococcus was a minor representative of the SRB community (ranging from 4% to 15%), depending on the sample and time taken from the reactor (Figure 3). These genera have been previously shown to dominate sulfidogenic bioreactors [35,48]. In general, SRB made up between 31% and 58% of the FBR bacterial community, depending on the sampling time (Figure 4).  Table 2.   Table 2.
Minerals 2019, 9, x FOR PEER REVIEW 9 of 19 contained common SRB from the genera Desulfovibrio, Desulfomicrobium and Desulfococcus. The most dominant genus present in the bioreactor was Desulfovibrio (ranging from 15% to 22%), followed by Desulfomicrobium (ranging from 8% to 27%), whilst Desulfococcus was a minor representative of the SRB community (ranging from 4% to 15%), depending on the sample and time taken from the reactor (Figure 3). These genera have been previously shown to dominate sulfidogenic bioreactors [35,48]. In general, SRB made up between 31% and 58% of the FBR bacterial community, depending on the sampling time ( Figure 4).  Table 2.   Table 2. Table 3. Summary of the key bacterial and archaeal genera detected in DNA samples taken from the FBR. All genera listed were present in at least two of the analysed samples and with a relative abundance >2%.

Bacteria
Dechloromonas B, C, D, E, F Anaerobic, Gram-negative bacteria capable of complete acetate oxidation by dissimilatory (per)chlorate reduction to chloride. [50,51] Clostridium B, C, D, E, F A diverse genus of anaerobic, spore forming, Gram-positive bacteria found in soils, animals and humans (both pathogenically and commensally). [52] Longilinea A, B, C, E, F Anaerobic Gram-negative bacteria which utilise proteins and some sugars for growth. Growth is enhanced in co-culture with hydrogen scavenging methanogens. [53,54] Bellilinea A, E, F Anaerobic Gram-negative bacteria which utilise carbohydrates for growth. Growth is enhanced in co-culture with hydrogen scavenging methanogens. [54] Desulfovibrio All Anaerobic Gram-negative SRB. They incompletely oxidise organic substrates to acetate. [55][56][57] Desulfomicrobium All Anaerobic Gram-negative SRB. They incompletely oxidise organic substrates to acetate. [58,59] Azonexus All Facultative aerobic Gram-negative bacteria. Capable of nitrogen fixation. [60,61] Desulfococcus All Strictly anaerobic Gram-negative SRB. They completely oxidise organic substrates. [62][63][64] Rhodobacter A, B Gram-negative bacteria with a range of metabolic capabilities. Main species of this genus are photosynthetic purple bacteria. [65,66] Spirochaeta A, B, E, F Anaerobic, saccharolytic and helical shaped bacteria. [67,68] Acholeplasma A, B Wall-less bacteria with uncertain classifications as either saprotrophic or pathogenic. [69] Bacteroides C, E, F Anaerobic Gram-negative bacteria largely found in the mammalian gastrointestinal microbiota. They break down complex molecules (such as plant and host-derived glycans) to simpler ones. [70,71] Burkholderia B, C, E, F Gram-negative bacteria which are aerobic but can often tolerate anaerobic conditions. They include animal, human and plant pathogens, as well as environmentally important species involved in nitrogen fixation, promoting plant growth and killing pest organisms. [72,73] Archaea Methanobacterium All Methanogenic archaea known to grow on hydrogen and carbon dioxide. They grow without oxygen and are often present in anaerobic bioreactors. They exhibit similar growth kinetics to Desulfovibrio. [74,75] Methanosarcina A, C, E Archaeal genus containing species of acetoclastic methanogens that play a pivitol role in anaerobic consortia. [76] Thermogymnomonas C, E Aerobic, acidophilic and heterotrophic archaea which lack a cell wall. [77] The FBR archaeal community consisted mainly of H 2 -and CO 2 -utilising methanogens, belonging to the genus Methanobacterium. This genus comprised between 88% and 94% of the archaeal community of the attached biomass taken from the FBR carrier samples (Table 2: samples A, C, and E). The carrier samples contained multiple genera of methanogenic archaea (Table 2). A much less diverse archaeal community was observed in the suspended biomass samples (Table 2: samples B, D, and F). The diversity of the archaeal community (in the carrier samples) showed that methanogenic archaea adhered to the carrier material [74]. The presence of methanogenic archaea in the attached biomass consortium indicated that the conditions within the FBR were conducive for the growth of methanogens, which may in turn affect the activity of the SBR for sulfate reduction [53,54]. Microbial community analysis also showed that the FBR contained many non-SRB (Figure 4; Table 3). It is likely that some non-SRB were responsible for degrading complex organic matter produced by the biocarrier consortium, providing additional energy and carbon sources, or nutrients (e.g., nitrogen source) for the SRB [41]. Azonexus and Desulfococcus were most abundant within 24 h of an interruption to influent flow (Figure 3: sample E at day 146). This suggested that when there is a lack of available substrates for growth nutrients, Azonexus and Desulfococcus remain attached to the carrier material. In contrast, SRB such as Desulfovibrio and Desulfomicrobium are more likely to detach and wash out from the bioreactor [48].
Four genera, namely Clostridium, Dechloromonas, Burkholderia and Bacteroides were only detected at a relative abundance of <10% when the reactor was producing sulfide (Figure 3). Bacterial genera of low abundance may still play a pivotal role in maintaining a stable consortium and supporting the growth of SRB by contributing to nutrient cycling and removal of secondary substrates and metabolites [50,52] (Table 3). Members of the genus Longilinea were least abundant when the bioreactor was experiencing uninterrupted sulfide production and most abundant during bioreactor recovery. The growth of Longilinea is often supported by methanogens, which can outcompete SRB under certain conditions [53,54]. The abundance of Longilinea may indicate unfavourable bioreactor conditions for SRB and an abundance of methanogens. The number of genera detected at relative abundance of less than 2%, decreased by 5.8% between samples A and C (21 days) and 5.6% between samples C and E (20 days) (Figure 3). This suggested that over time, the diversity of the microbial community decreased.

Sparging by the SR Immediately Removed All H 2 S, but the FBR Remained Active
Sulfide speciation in solution is dependent on pH. The dominant sulfide form in acidic solution (pH < 7) is H 2 S, whereas the dominant form becomes HS − in an alkaline solution (pH > 7) [31]. Since the experiment required the removal of H 2 S from the bioreactor, it was critical that reactor pH was maintained at pH not exceeding 7. Figure 5A shows that bioreactor pH increased gradually following N 2 sparging, due to H + removal from solution as H 2 S forms (Equation (6)). During the precipitation experiment, the reactor pH was manually lowered with 1 M H 2 SO 4 to increase the dissolved H 2 S concentration, facilitating the stripping of H 2 S to the gas phase for precipitating metals from the leach solution: The precipitation experiment ran for 75 h with continuous sparging of the SR. Figure 5A shows that the dissolved sulfide concentration of the bioreactor decreased rapidly to undetectable levels at the start of Period II. At the end of Period I, the leachate was filtered to remove any precipitates. Throughout Period II, the leachate became noticeably turbid and darkened in appearance as metal sulfides continued to precipitate from solution. Although dissolved sulfide concentrations had decreased to undetectable levels, the SRB had remained metabolically active. However, as soon as the SRB produced H 2 S, it was stripped from the SR by the N 2 gas sparging. This would account for the undetectable dissolved sulfide concentrations observed for the remainder of the experiment. A dissolved sulfide concentration of 10 mg L −1 was recorded shortly after the end of the experiment, further indicating that the SRB was metabolically active ( Figure 5A:~73 h). dissolved sulfide concentration of 10 mg L −1 was recorded shortly after the end of the experiment, further indicating that the SRB was metabolically active ( Figure 5A: ~73 h).

Recovery of Metals was pH Dependent
The H2S produced in the sulfidogenic bioreactor was used to precipitate and recover metals from the LIB waste leachate. Biogenic sulfide was added to the leachate via gas sparging, as opposed to directly adding the sulfide containing effluent from the bioreactor, as practiced in previous studies [30]. The use of gas transfer was favored in this study as it may enable better pH control for metal recovery than when using bioreactor effluent for precipitation. The leach solution had an initial pH of 3.5, and this continued to spontaneously decrease with the addition of H2S and without pH control in Period I ( Figure 5A), likely due to the formation of copper sulfide (Reaction (5)): H2S + Cu 2+ → CuS(s) + 2H + Reaction (5) In Period I, 38% Cu and 12% Al were precipitated, while other metals remained largely in solution ( Figure 5B). Soon after the addition of H2S to the leachate, the color of the solution changed from a transparent pink-red to a turbid black-brown, which indicated the formation of metal sulfides. By the end of Period II, Cu and Al had precipitated from solution at efficiencies of 93% and 98%, respectively. Li, Ni and Co appeared to precipitate slightly, but efficiencies were low and variable throughout this period of precipitation ( Figure 5B). At the end of Period III, Ni (99.9%), Co (99.9%), Mn (98.9%), Cd (98.6%) and Zn (98.4%) precipitated from solution, which indicated that a complex metal precipitate was formed as the pH was increased to a set point of 10 ( Figure 5A,B). Cibati et al.

Recovery of Metals was pH Dependent
The H 2 S produced in the sulfidogenic bioreactor was used to precipitate and recover metals from the LIB waste leachate. Biogenic sulfide was added to the leachate via gas sparging, as opposed to directly adding the sulfide containing effluent from the bioreactor, as practiced in previous studies [30]. The use of gas transfer was favored in this study as it may enable better pH control for metal recovery than when using bioreactor effluent for precipitation. The leach solution had an initial pH of 3.5, and this continued to spontaneously decrease with the addition of H 2 S and without pH control in Period I ( Figure 5A), likely due to the formation of copper sulfide (Equation (7)): In Period I, 38% Cu and 12% Al were precipitated, while other metals remained largely in solution ( Figure 5B). Soon after the addition of H 2 S to the leachate, the color of the solution changed from a transparent pink-red to a turbid black-brown, which indicated the formation of metal sulfides. By the end of Period II, Cu and Al had precipitated from solution at efficiencies of 93% and 98%, respectively. Li, Ni and Co appeared to precipitate slightly, but efficiencies were low and variable throughout this period of precipitation ( Figure 5B). At the end of Period III, Ni (99.9%), Co (99.9%), Mn (98.9%), Cd (98.6%) and Zn (98.4%) precipitated from solution, which indicated that a complex metal precipitate was formed as the pH was increased to a set point of 10 ( Figure 5A,B). Cibati et al. [29] and Kaksonen et al. [30] also reported low precipitation efficiencies for Ni and Co as sulfides within the pH ranges as recorded in Periods I and II. The low precipitation efficiencies for Ni and Co until Period III showed that increasing the leachate pH to 10 was essential to recover most of these metals. Based on solution analysis alone, it was not possible to determine if the metals had precipitated as metal sulfides or hydroxides at more alkaline pH values.
As shown in Table 4, all metals had precipitated from solution at efficiencies greater than 98% at the end of the experiment (pH 10), except for magnesium (49%) and lithium (0%). Mg does not readily precipitate as a sulfide, and MgS spontaneously decomposes in aqueous systems. The precipitation of Mg observed in this study is likely to be attributed to the precipitation of MgO 2 and Mg(OH) 2 , which occurs over a range of pH 8 to 14 in aqueous systems as shown by the Mg Pourbaix diagram [78]. The precipitation efficiency of Li reached percentages greater than 30% at times throughout the precipitation experiment. However, by the conclusion of the experiment all the precipitated lithium had returned to solution. This may be explained by the high solubility of lithium sulfide or the reaction of lithium sulfide with HCl, producing H 2 S (Equation (8)): Lithium does not precipitate as a metal sulfide within the pH range used in this study. Therefore, other methods, such as metal carbonate precipitation, are necessary to adequately recover lithium from the leachate solution. For instance, Meshram et al. [79] successfully recovered Li from LIB waste leachate as lithium carbonate (98% purity) upon the addition of sodium carbonate (pH 14). Hence, it is likely that the recovery of Li. This process is beneficial as it results in the production of a lithium-rich solution that can be further processes to recover lithium for secondary use. Recovery of lithium will require an additional carbonation or hydroxide precipitation step after the sulfide precipitation of other metals from solution.

Purity Indices of Metal Precipitates
The ability to selectively recover metals with H 2 S was assessed by calculating the purity indices for each metal as selected pH values ( Figure 6) [29]. The purity index illustrates the relative amount of each individual metal precipitated from solution between pH points. For example, 35% of the total metal precipitated at a pH of 3 was Cu (shown in Figure 6 as a purity index of 0.35 at pH 3). The highest purity index detected was 0.75 for Co between pH 7 and pH 10. The highest purity indices for Cu and Ni were obtained at pH between 5 and 7, namely 0.42 and 0.31, respectively, whereas the highest purity index of 0.29 for Al was recorded at pH between 4 and 5. The results also suggested that the biogenic sulfide precipitation of metals from LIB waste leachate was not truly selective, as multiple metals were precipitated from solution at all pH values tested ( Figure 6). Nonetheless, relatively high purity indices for Co (0.75 at pH 7-10) and Cu (0.42 at pH 5-7) were calculated. This is significant because Co and Cu make up 90.4% of the metal value contained in LIBs ( Table 1). The similar precipitation properties of Ni and Co make it difficult for high purity Ni to be selectively recovered [29]. The majority of Al precipitated between pH 3 and 5, which was when most metals began to precipitate from solution, making it difficult to selectively recover high purity Al. Al precipitation in the acidic pH range could likely be attributed to the precipitation of hydroxysulfate minerals, as described by Falagan et al. [80]. However, in the alkaline pH range, Al precipitation is more likely to be in the form of Al(OH 2 ) [78]. Finally, the precipitation of Li proved difficult due to the solubility and reactivity of lithium sulfide in the leachate, as previously discussed (Equation (7)). began to precipitate from solution, making it difficult to selectively recover high purity Al. Al precipitation in the acidic pH range could likely be attributed to the precipitation of hydroxysulfate minerals, as described by Falagan et al. [80]. However, in the alkaline pH range, Al precipitation is more likely to be in the form of Al(OH2) [78]. Finally, the precipitation of Li proved difficult due to the solubility and reactivity of lithium sulfide in the leachate, as previously discussed (Reaction (5)).  (2)).

Implications of Findings
Some previous studies have investigated the selective precipitation of metals using biogenic sulfide [29]. However, the use of biogenic sulfide for precipitating metals from waste LIB leach liquors has to our knowledge not been previously demonstrated. The precipitation efficiencies of Cu and Zn in this study are in line with those reported previously for sulfidogenic FBR metal recovery processes [30,39]. The present study also showed Al, Co, Ni, and Fe precipitating at efficiencies over 99%, in line with the findings of Sahinkaya et al. [81]. However, it should also be noted that the initial metal concentrations in this study (Table 4) were much higher than the concentrations used in the previously mentioned studies. The purity indices for individual metals reported by Cibati et al. [29], were much higher than those obtained from this study. However, the present study attempted to recover a greater number of metals (10) from a leach solution than Cibati et al. [29] (four metals only) and some other mixed metal precipitation studies [48,82].
The metals obtained from solution in this study were not selectively separated. There is an opportunity to optimise this process to promote selective metal separation, and this can potentially be achieved by refining the pH set points, reagent concentrations, scale and time course of the experiment. Nonetheless, it is anticipated that the separation of metals such as Co, Cu, Mn and Ni will remain a challenge due to their similar chemical and physical properties [29]. The present study demonstrated that high purity metal sulfide mixtures could be generated, and these can potentially be further refined into high value alloys. Since Al, Ni, Co and Cu represent over 96% of the metal value contained in LIB waste, it was concluded that most of the metal value of LIB waste can be successfully recovered with biogenic sulfide precipitation as demonstrated in this study. However, further techno-economic assessment is required to quantify costs associated with processing, and

Implications of Findings
Some previous studies have investigated the selective precipitation of metals using biogenic sulfide [29]. However, the use of biogenic sulfide for precipitating metals from waste LIB leach liquors has to our knowledge not been previously demonstrated. The precipitation efficiencies of Cu and Zn in this study are in line with those reported previously for sulfidogenic FBR metal recovery processes [30,39]. The present study also showed Al, Co, Ni, and Fe precipitating at efficiencies over 99%, in line with the findings of Sahinkaya et al. [81]. However, it should also be noted that the initial metal concentrations in this study (Table 4) were much higher than the concentrations used in the previously mentioned studies. The purity indices for individual metals reported by Cibati et al. [29], were much higher than those obtained from this study. However, the present study attempted to recover a greater number of metals (10) from a leach solution than Cibati et al. [29] (four metals only) and some other mixed metal precipitation studies [48,82].
The metals obtained from solution in this study were not selectively separated. There is an opportunity to optimise this process to promote selective metal separation, and this can potentially be achieved by refining the pH set points, reagent concentrations, scale and time course of the experiment. Nonetheless, it is anticipated that the separation of metals such as Co, Cu, Mn and Ni will remain a challenge due to their similar chemical and physical properties [29]. The present study demonstrated that high purity metal sulfide mixtures could be generated, and these can potentially be further refined into high value alloys. Since Al, Ni, Co and Cu represent over 96% of the metal value contained in LIB waste, it was concluded that most of the metal value of LIB waste can be successfully recovered with biogenic sulfide precipitation as demonstrated in this study. However, further techno-economic assessment is required to quantify costs associated with processing, and identify potential end uses for the mixed metal sulfide products generated.

Conclusions
Biogenic hydrogen sulfide produced in the FBR and an incremental pH increase to 10 with NaOH, precipitated Al, Ni, Co, and Cu at recovery efficiencies greater than 98%, resulting in high (96%) recovery of the metal value from LIB waste. The purity indices were highest for Co and Cu, achieving 0.75 at pH 10 and, 0.42 at pH 7, respectively. The mixed metal sulfides produced by non-selective biogenic sulfide precipitation could be further refined for the manufacture of high value metal products, but further process optimisation for improving metal recovery selectivity and purity of the mixed metal sulfides is required. The economic feasibility of this process for the recovery of mixed metals from waste streams also needs to be further investigated. Additional work to analyse metal precipitates generated would provide further insight to the mechanism of metal precipitation using biogenic sulfide precipitation as the recovery method.