From urban municipalities to polar bioremediation: the characterisation and contribution of biogenic minerals for water treatment

Globally, heavy metals and organic pollutants such as petroleum hydrocarbons remain problematic contaminants in water (Hennebel et al. 2009; Edwards & Kjellerup 2013). The presence of these constituents in drinking water, soils and groundwater presents a significant ecological and human health risk, and these compounds are actively managed through processing in water treatment plants (WTPs) and various remediation technologies. Biological treatment operations have become an integral part of the water treatment industry because of the economic advantages of this approach, in terms of both capital investment and operating costs, over other treatment methods such as chemical oxidation (Hennebel et al. 2009). Where biological water treatment is employed, the formation of biogenic minerals offers a matrix for the sorption, or oxidation and reduction, of inorganic and organic compounds (Fortin & Langley 2005).

Biogenic minerals are defined as minerals produced by organisms, where these mineralised products are composite materials comprising mineral and organic components (i.e. metabolites and exopolysaccharides) (Weiner & Dove 2003). The immobilisation of organic and inorganic compounds in biogenic minerals can occur by precipitation and crystallisation or by sorption, uptake and intracellular sequestration (Gadd 2004). Negative charges on the surfaces of most bacterial cells and extracellular polymeric substances (EPS) can facilitate the binding of cations through non-specific electrostatic interactions (Hennebel et al. 2009). Microorganisms interact with minerals in order to create a more hospitable surrounding through utilisation of nutrients and sequestration of toxic substances (Gadd 2004). Mineral oxides of biological origin have been reported to have a higher specific surface area than equivalent mineral products of synthetic origin, supporting higher sorption capacities for heavy metals such as lead, chromium and nickel (Frankel & Bazylinski 2003).

The presence and formation of biogenic iron oxides (i.e. hematite), oxyhydroxides (i.e. goethite) and amorphous, poorly ordered phases (i.e. two-line ferrihydrite) have been well documented for water treatment applications (Fortin & Langley 2005; Omoregie et al. 2013; Fleming et al. 2014; Kondratyeva & Golubeva 2014; Posth et al. 2014; Picard et al. 2015). A further mineral of relevance to water treatment is manganese. The heightened metal adsorption capacities of biogenic manganese oxides, over minerals of abiotic origin, have attracted significant interest in biologically enhanced water treatment (Hennebel et al. 2009; Johnson et al. 2012). For example, Nelson et al. (2002) report the Brunauer, Emmett and Teller surface area of biogenic manganese oxides exceeding 200 m²/g, while abiotic manganese oxides were less than 60 m²/g when examined for lead adsorption capacity. Manganese-oxidising bacteria can oxidise manganese and reduce its solubility, thereby providing a protective mechanism against toxic levels of soluble manganese.

In water treatment applications employing reactive materials, such as zeolites and biological activated carbon (hereafter abbreviated to BAC), the formation of biogenic iron and manganese oxides has been shown to occur (Kondratyeva & Golubeva 2014). These biogenic structures have also demonstrated the capacity for ongoing removal of inorganic and organic compounds from water following saturation, or exhaustion, of the reactive material (Thiel & Aldridge 2010; Thiel 2014). The dependence of manganese-oxidising activity on iron concentrations demonstrates the importance and interaction of iron and manganese in water treatment (Gheriany et al. 2009). It is these properties and functions that have seen biogenic minerals receive increased attention in a variety of bioreactor configurations for municipal water treatment and bioremediation (Gadd 2004).

The metabolic activity of microbial biofilms has been shown to increase significantly following adsorption onto zeolites and activated carbon (Herzberg et al. 2003; Simpson 2008; Freidman et al. 2016). The heightened resistance of cells within biofilms to fluctuations in the surrounding environment suggests that zeolites and activated carbon may prove advantageous for the construction of biogenic minerals (Gadd 2004). Facultative microaerophilic iron- and manganese-oxidising bacteria known to colonise activated carbon include Leptothrix discophora, Pseudomonas putida, Gallionella ferruginea, Sphaerotilus natans and Pedomicrobium manganicum, as well as several other Bacillus, Erythrobacter and Arthrobacter spp. (Tebo et al. 2004). As the adsorptive capacity of activated carbon begins to decrease over time, the ability of these microorganisms to uptake organic carbon and oxidise trace contaminants becomes increasingly important when considering aspects of filter breakthrough and water quality. It is therefore crucial to understand how microbial communities form on the surface of filter media and how biogenic mineral formation contributes to the removal of iron, manganese and other trace contaminants in the presence of filter media.

This study will examine the formation of iron and manganese biogenic minerals on granular water treatment media employed for the removal of aqueous pollutants. The research encompasses findings from BAC filters at two drinking WTPs, specifically the Kyneton WTP and the Bendigo WTP in Central Victoria, south-eastern Australia. The WTPs employ microfiltration and ozone-BAC filtration to minimise the concentrations of taste and odour compounds, such as 2-methylisoborneol and geosmin, in distributed water (Northcott et al. 2010). Additionally highly soluble manganese in the raw water of many storage reservoirs can pass through treatment processes into the supply system and thus microbial oxidation of manganese on filter media is paramount to downstream water quality (Thiel & Aldridge 2010).

Recent results have shown that the performance of BAC filters for organics reduction decreases at lower water temperatures (Warren & Thiel 2013), with similar studies revealing lower adsorption kinetics on activated carbon at low temperatures (Arora et al. 2011). With a reduction in microbial metabolic activity at lower temperatures (Andersson et al. 2001; Ferguson et al. 2003), iron and manganese biogenic minerals may provide a crucial contribution to water treatment under these conditions. To understand the prevalence of particle-attached biogenic minerals across different applications and temperature regimes, cold region sites were targeted for comparative analysis. This study couples the findings of BAC media at the Kyneton WTP and the Bendigo WTP with filter media for the adsorption and bioremediation of metal and petroleum hydrocarbon contaminants, mainly Special Antarctic Blend (SAB) diesel, within permeable reactive barriers (PRBs) at Old Casey Station, East Antarctica (Revill et al. 2007). The PRB at Old Casey Station employs controlled release nutrient materials, such as nutrient-amended zeolites, to promote the formation of BAC which would otherwise be limited in the presence of low nutrient groundwater (Mumford et al. 2013).

The PRB at Old Casey Station is a sequenced design with a mixed bed of 1:2 (v/v) activated carbon and Zeopro^TM, for petroleum hydrocarbon adsorption and nutrient release, preceding a mixed bed of 1:2 (v/v) activated carbon and natural zeolite, for petroleum hydrocarbon adsorption and capture of excess nutrients. Temperatures within these PRBs commonly fall below freezing eight months of the year with intermittent freeze-thaw cycling experienced in the summer months (Gore et al. 2006). The influent to BAC filters at the Kyneton WTP and the Bendigo WTP also demonstrate low nutrient conditions (<1 mg/L NH₃) with water temperatures as low as 8 °C at the Kyneton WTP and as low as 10 °C at the Bendigo WTP in the winter months (Northcott & McCormick 2014).

While previous studies have reported the formation of biogenic minerals in water treatment applications (Gülay et al. 2014), the significance of iron and manganese biogenic minerals on filter media, ranging from municipal drinking water treatment to remote groundwater remediation requires further investigation. A detailed biological and chemical assessment of these biogenic minerals will follow, supporting the contribution of biogenic minerals to the lifetime and effective longevity of filter media in all forms of water treatment.

The BAC media employed at the Kyneton WTP and the Bendigo WTP is a mixture of thermally activated sub-bituminous coal (GS1300) and anthracite (GA1000N) (Table 1) (Activated Carbon Technologies Pty Ltd). BAC media was collected at the surface, 1 m depth and 2 m depth from one filter unit (BAC 3) at the Kyneton WTP during May 2015. BAC was also collected at the surface, 1 m depth and 2 m depth from one filter unit (BAC 6) at the Bendigo WTP during May 2015. All samples were immediately transferred to 4 °C storage. Over the last 13 years, iron concentrations in the BAC filter influent at the Kyneton WTP and the Bendigo WTP have been in the range 0.4 mg/L to 0.5 mg/L while manganese concentrations have been between 0.001 mg/L and 0.002 mg/L.

Activated carbon installed for the adsorption of petroleum hydrocarbons in Antarctic PRBs is derived from coconut husk (GC1200) and is thermally activated (Table 1) (Activated Carbon Technologies Pty Ltd). This media was collected by coring during the 2013/14 summer field season from a PRB at Old Casey Station, East Antarctica (Mumford et al. 2013). All samples were immediately transferred to −20 °C storage and returned to Australia prior to analysis. Additionally a controlled release nutrient material, ammonium exchanged zeolite, was examined from laboratory flow cells contacted with soil from Old Casey Station, East Antarctica (Mumford et al. 2013; Freidman et al. 2016). Previous reports have detailed widespread heavy metal and hydrocarbon contamination at Old Casey Station (Ferguson et al. 2003; Northcott et al. 2003). Analysis of ammonium exchanged zeolite demonstrated that sorption of organic and inorganic contaminants to iron and manganese biogenic minerals may not be restricted to BAC and can be formed on a wide range of substrates.

Imaging of bacterial communities on BAC and zeolite was achieved using a field emission scanning electron microscope (FESEM) with a backscatter electron detector at 10 kV (Philips XL30 FEG, Melbourne Advanced Microscopy Facility, University of Melbourne). Activated carbon and zeolite particles were fixed with 2.5% glutaraldehyde in a 1 M phosphate buffer solution for 12 hours prior to analysis. Samples were then dehydrated in a graded acetone series: 10, 30, 50, 70, 90, 100, 100, 100% acetone (reagent grade) at 15 minute intervals. Samples were critical point dried (Balzers CPD 030) and fixed to 12 mm Ø mounts using double-sided carbon tape. Samples were sputter coated with gold (Dynavac) for 120 seconds prior to imaging. Where pore imaging of activated carbon was conducted, samples were contacted with liquid nitrogen and carefully dissected with tweezers under a light microscope prior to gold coating.

Biogenic mineral surface chemistry was analysed by energy dispersive spectroscopy (EDS) microanalysis (AZtec software, Oxford Instruments) fitted to an environmental scanning electron microscope (FEI Quanta, Melbourne Advanced Microscopy Facility, University of Melbourne). Analysis was conducted using a backscatter electron detector. As heavy elements, such as lead, strongly backscatter electrons, these elements appear brighter in the micrographs. Alternatively lighter elements, such as carbon, absorb electrons and appear darker in the micrographs.

Samples of 0.1 g of media at field moisture were placed in 1 ml of a 1 M phosphate buffer solution. The samples were gently sonicated (Soniclean 120HT) for 10 minutes and vortexed for a further 2 minutes to release bacterial cells into solution. FESEM confirmed high cell detachment following sonication and vortexing. A 700 μl sub-sample was placed onto a density gradient medium (Histodenz^TM, Sigma-Aldrich) and centrifuged at 13,000 × g for 5 minutes to remove particulates. The supernatant was then removed and a sub-sample (20 μl) was applied to both Tryptone Soy Agar plates and R2A Agar plates using the streak plate method. Inverted plate cultures were incubated for 48 hours at 23 °C prior to analysis.

The identification of bacteria on the water treatment media was conducted using matrix assisted laser desorption ionisation-time of flight (MALDI-TOF) mass spectrometry (Microflex LT, Bruker Daltonics Inc.). Bacterial strains isolated on Tryptone Soy Agar plates and R2A Agar plates were applied to a sterile MALDI 96-well target using a wooden applicator stick. An aliquot of 1 μl of 70% formic acid was applied to each well and allowed to air dry at room temperature. Then 1 μl of αCyano-4-hydroxycinnamic acid (HCCA), suspended in an aqueous mixture of 50% acetonitrile, 47.5% water and 2.5% trifluoroacetic acid (Sigma-Aldrich), was applied to each well. The HCCA matrix was placed in the dark and air dried at room temperature before insertion into the MALDI-TOF mass spectrometer. Escherichia coli, Enterococcus faecalis and Candida albicans were run as standards for quality control. Spectra were analysed in the range m/z 2,000 to 20,000 using the MALDI BioTyper RTC wizard software and MALDI BioTyper reference library (Bruker Daltonics Inc.).

X-ray diffraction (XRD) and Raman spectroscopy were both employed because of the potential amorphous structure of iron and manganese biogenic minerals (Weiner & Dove 2003). XRD was conducted on dried activated carbon samples using a Bruker D8 Advance Diffractometer using Ni-filtered CuK_α radiation. The diffractogram was acquired over 5–75° 2θ with instrument settings of 0.02° step size, 1.2 seconds per step.

Raman spectroscopy was conducted using a frequency doubled Nd:YAG laser (532 nm). The laser was directed onto the sample with a confocal microscope equipped with a 20 × , long working distance objective lens with a numerical aperture of 0.4 (or a 100 × , 0.9 NA objective lens). The Raman scattered light was collected back through the objective lens into a Renishaw InVia Reflex 0.25 m working distance micro-Raman spectrometer and a charge coupled device array detector. The spectral resolution is approximately 0.2 meV. The exposure time was 50 s and the laser power was kept to 0.5 mW at the sample to avoid thermal degradation.

Acid digestions were conducted on reactive media to determine the concentrations of bioavailable iron and manganese (Snape et al. 2004). Samples were dried at 80 °C for 24 hours. Samples (1 g) of reactive material were then contacted with 100 ml of 1 M hydrochloric acid for 24 hours while mixed at 100 rpm. Extracts were then filtered with a 0.45 μm syringe filter and analysed using a Varian 720-ES inductively coupled plasma-optical emission spectrometer.

The precipitates initially form sheets less than <5 μm² in size that then form into porous rosette sheeted spheres with hexagonal symmetry, comprising organic and inorganic constituents (Figure 5(a)). The filaments of Pedomicrobium sp. contribute further extracellular material resulting in the formation of spheres as large as 100 μm in diameter on BAC media and up to 50 μm in diameter on zeolite media. Figure 5(f) demonstrates the high density and tight binding of these biogenic minerals to the media surface and the potential alteration of BAC and zeolite sorption properties following the development of a biogenic layer. Microscopy suggests that the morphology of biogenic mineral deposits is largely consistent across rod-shaped bacterial communities and filamentous bacterial communities, and between zeolites and activated carbon (Figure 5). While the biogenic minerals can cover the surface, they were not observed to grow specifically within the pore structure of municipal or remote region BAC media (Figure 4).

It is important to highlight that while the microbial community is similar between municipal and remote water treatment media, the biodegradation pathways for SAB diesel in the Antarctic and dissolved organic carbon at the Kyneton WTP and the Bendigo WTP are different. Similarly, the higher microporous fraction of Antarctic BAC media over municipal BAC media may regulate the rates of pore colonisation by microbes, as well as the bioregeneration rate of SAB diesel (Aktas & Cecen 2007). Along with the seasonal variability in iron and manganese concentrations in influent streams, the presence or absence of particular electron donors and electron acceptors may facilitate localised changes in biogenic mineral morphology and abundance, as observed in this study (e.g. Figure 5).

Peaks in the XRD profile of samples of ammonium exchanged zeolite contacted with Old Casey Station soil occurred at 42.1° 2Ɵ, which may be attributed to the presence of hematite (Figure 8(b)) (Shopska et al. 2012). Smaller peaks at 30.5° 2Ɵ, 30.7° 2Ɵ and 48.5° 2Ɵ align with those of magnetite (Shopska et al. 2012). The high peak intensities at 28.2° 2Ɵ and 61.2° 2Ɵ may also be the result of minerals of biological origin that remain bound to the surface of ammonium exchanged zeolite.

Raman spectroscopy revealed the presence of two sharp peaks at 1,600 cm⁻¹ and 1,350 cm⁻¹ on raw GA1000N activated carbon, associated with crystalline graphite (Figure 8(c)). The peaks in the range 500 to 510 cm⁻¹ in both media contacted with water can be assigned to birnessite mineral material, while peaks in the range of 600 to 800 cm⁻¹ have previously been reported to correspond to bixbyite-like mineral (Julien et al. 2003; Bohu et al. 2015). The presence of other elements adsorbed to manganese oxide deposits may explain the variability in peak position and intensity between BAC media from the Kyneton WTP and the Bendigo WTP below 1,000 cm⁻¹ in the Raman spectra. The bands at 1,060 cm⁻¹, 1,160 cm⁻¹ and 1,560 cm⁻¹ in both the Bendigo WTP and the Kyneton WTP samples can be assigned to carotenoids present in the bacterial cells (Ciobotă et al. 2012). These peaks assigned to carotenoids support the biogenic origin of the minerals present on the surface of the BAC and the zeolite media, highlighting an important interplay that can impact on bioremediation performance.

The prevalence of biogenic minerals on activated carbon and zeolite media observed here, and the confirmation of their chemical structure, suggests that these formations have significant implications for the removal of organic and inorganic pollutants across varying water treatment operations (Hennebel et al. 2009). Where adsorption and ion exchange capacities approach exhaustion, high surface area mineral formations may facilitate the continuing acclimation of target contaminants. While uncertainty remains as to the stability of biogenic minerals attached to particles under backwashing or high flow events, enhanced performance of filter media may be feasible beyond timeframes of BAC saturation (Thiel 2014). This has significant benefits for groundwater treatment technologies in remote regions where the replacement of adsorption material is expensive and difficult and where biofilm formation is challenged by low temperature and freeze-thaw cycling (Mumford et al. 2013). While the minerals formed by biological activity are not easily re-dissolved, the stability of biogenic minerals and potential for detachment with exposure to freeze-thaw cycling requires further investigation (Konhauser & Fyfe 1994) as such activity could reduce the effectiveness of bioremediation.

The bacterial species identified on BAC and zeolite media sourced from both Australia and Antarctica is consistent with previous findings (Hennebel et al. 2009) and with the formation of iron and manganese biogenic minerals by these species (Tebo et al. 2004; Kondratyeva & Golubeva 2014). The different biofilm structures and manganese concentrations on BAC media at varying filter depths align with previous findings from observation of samples from the Kyneton WTP and the Bendigo WTP (Thiel & Aldridge 2010; Thiel 2014). Biogenic mineral morphologies were similar across all media despite the scarcity of manganese on the Antarctic PRB media and the low concentrations of iron within BAC filter beds at the Kyneton WTP and the Bendigo WTP. This suggests that iron- and/or manganese-oxidising bacteria may be capable of catalysing the formation of biogenic minerals, even where a system is deficient in iron or manganese, or exposed to flow with a low nutrient concentration, as occurs in a cold region PRB (Freidman et al. 2016). The abundance of Gallionella sp. within the pores of the BAC media obtained from the Kyneton WTP and the Bendigo WTP suggests that not all iron and manganese is retained at the surface for construction of biogenic minerals (Picard et al. 2015). The higher surface area to volume ratio (16:1) of twisted stalks, associated with Gallionella sp., over rod-shaped bacterial cells (10:1) also supports increased sequestration of iron and manganese within the BAC pores (Konhauser 1998). The entrance of iron and manganese into the BAC pores also potentially explains the variability in biogenic mineral density between particles as this phenomenon is expected to increase heterogeneity across the surface and between surfaces.

In laboratory flow cells comprising ammonium exchanged zeolite, biogenic mineral formation was more prevalent as the concentration of total petroleum hydrocarbons in solution decreased below ∼2 mg/L (Freidman et al. 2016). Similarly, surface BAC media from the Kyneton WTP and the Bendigo WTP demonstrated low surface coverage by biogenic minerals, despite the high biomass, high manganese concentrations and pre-ozone treatment, while mineral formation was high at 1 m and 2 m depths. The scanning electron microscopy images in this study indicate that EPS and cellular material contribute to biogenic mineral formation. Bacterial cells were not always present with the biogenic minerals suggesting that cells may detach from the reactive media shortly after biogenic mineral formation (Figure 5(f)). Recent reports suggest that as biogenic mineral formation often commences at the base of the biofilm, expansion of precipitated deposits can result in the displacement of an equivalent volume of biomass (Li et al. 2015). Similarly, Coombs et al. (2013) observed that even in low nutrient environments, biogenic mineral precipitates have a much higher chemical and physical stability than the biofilm, often persisting on the surface long after biofilm detachment. This finding disagrees with previous discussions of iron and manganese biogenic oxides being coated with adherent microorganisms (Nelson et al. 2002; Gülay et al. 2014), although detailed time series maybe required to draw firm conclusions. The detachment of biofilms has implications for the removal of organic carbon by biofiltration, suggesting that as reactive media becomes covered by biogenic minerals the number of available sites for bacterial adhesion and biofilm formation may decrease (Andersson et al. 2001). It is therefore probable that the observation of biogenic minerals, coupled with low nutrient concentrations, may be indicatory of hindered biological performance within water treatment filter beds (Li et al. 2015). The requirement for controlled nutrient release sources to promote biodegradation of petroleum hydrocarbons in low nutrient environments such as Antarctica is well known (Ferguson et al. 2003; Gore and Snape 2008; Mumford et al. 2013; Freidman et al. 2016).

The hexagonal symmetry of biogenic minerals supports the EDS, XRD and Raman spectra of predominantly birnessite manganese oxides with bixbyite and vernadite deposits also detected on BAC media at the Kyneton WTP and the Bendigo WTP. Several studies have demonstrated adsorption of heavy metals, such as copper, nickel, chromium and zinc, when contacted with biogenic manganese oxides (Nelson et al. 2002; Toner et al. 2006; Edwards & Kjellerup 2013; Zhang et al. 2015). The initial oxidation of manganese by exopolymers of iron- and manganese-oxidising bacteria results in the formation of a colloidal birnessite phase of hexagonal symmetry, as observed in this study (Tang et al. 2014). This initial hexagonal birnessite phase is highly reactive and can induce secondary abiotic oxidation of manganese and other metals of environmental and human health concern (Tang et al. 2014). Specifically, Matocha et al. (2001) reported a maximum lead adsorption on birnessite of 200 mmol/mol of manganese. The absence of metals such as lead in appreciable concentrations in BAC filters at the Kyneton WTP and the Bendigo WTP likely resulted in the sole detection of manganese by EDS analysis.

Alternatively, the presence of zinc, chromium, nickel and lead on biogenic hematite and magnetite formations reinforces the findings of previous works (Hennebel et al. 2009). However where iron and manganese biogenic oxides have been combined in solution, lead adsorption by biogenic manganese oxides has been shown to be much greater than that of biogenic iron oxides (Nelson et al. 2002). Tang et al. (2014) has also shown that biogenic oxidation of metal compounds is accelerated in the presence of organic carbon. Therefore at contaminated sites comprising heavy metal and petroleum hydrocarbon contamination, such as the Thala Valley and Wilkes Landfill in east Antarctica (Northcott et al. 2003; Freidman et al. 2014), biogenic iron minerals may provide an important natural contribution to the capture and removal of contaminants alongside soil and groundwater remediation technologies.

This study demonstrates the importance of microbe–metal interactions for water treatment and the contribution of biogenic minerals to ongoing abiotic contaminant removal. Acid digestions, bacterial identification as well as XRD and Raman spectra, illustrate that the mineral symmetry is largely the result of biofilm-embedded iron and manganese being converted into rosette sheeted spheres. SEM imaging successfully demonstrated the different pathways of mineral development where manganese was limited in Antarctic settings and where iron was limited at the Kyneton WTP and the Bendigo WTP. This study shows the overall importance of iron- and manganese-oxidising bacteria to water treatment and the adaptability of these species to low temperature and low nutrient settings. Further research will be required to examine whether these attached biogenic minerals actively enhance the adsorption properties of exhausted BAC media or whether biogenic minerals accelerate exhaustion by preventing organics from entering into the pores of BAC media.

We would like to acknowledge Australian Antarctic Science Project 4029 and 4036, and the Particulate Fluids Processing Centre at the University of Melbourne for financial support. We would like to thank the Materials Characterisation and Fabrication Platform at the University of Melbourne and the Victorian Node of the Australian National Fabrication Facility for XRD analysis, and Bioscreen Medical for allowing usage of their MALDI Biotyper. We would also like to thank Roger Curtain of the Bio21 Institute, University of Melbourne for assistance with energy dispersive spectroscopic analysis and Dr Brett Johnson of the Department of Physics, University of Melbourne for assistance with Raman spectroscopy.