Coating techniques for glass beads as filter media for removal of manganese from water

Manganese (Mn) is naturally occurring in drinking water resources worldwide (Frisbie et al. 2012). It is frequently present in many surface water and groundwater sources, particularly in anaerobic or low oxidation conditions (WHO 2003). For a long time, Mn in drinking water had only been considered as an aesthetic water quality problem that was responsible for a metallic taste, discolouration and the cause of black stains on plumbing fixtures (Kohl & Medlar 2006). Recent studies have documented the neurotoxic effects of Mn in children and adults (Frisbie et al. 2012). Thus, several authors have recommended the reexamination of the 400 μg/l health-based guideline recommended by the WHO for drinking water (Ljung & Vahter 2007).

The removal of soluble manganese is commonly achieved through oxidation by potassium permanganate, chlorine dioxide, ozone, or other strong oxidants. The objective of this process is to oxidize Mn²⁺ to manganese oxide [MnO_x(s)], a solid precipitate that can be removed by solid–liquid separation processes such as sedimentation or deep-bed-filtration through granular media such as quartz sand or anthracite (Chiswell & Huang 2006; Kohl & Medlar 2006). Removal is also possible using granular filter media for microbial oxidation under aerobic conditions (Mouchet 1992; Chiswell & Huang 2006). In the presence of soluble Mn²⁺, granular filter media used in deep-bed filtration naturally develops an exterior coating of MnO_x(s). Mainly depending on the pH and O₂ concentration, this oxide-coated material removes Mn²⁺ from water by an abiotic adsorption and oxidation process known as the ‘natural greensand effect’ (NGE), which has been extensively studied (Knocke et al. 1991, 2010; Hargette & Knocke 2001). NGE is a natural process in which deposits of MnO_x(s) develop in situ within weeks to months on new uncoated filter media during normal filter operations (Knocke et al. 1988). Various products that are rich in MnO_x(s), such as Manganese Greensand^® or granular filter media composed of natural occurring manganese ores such as GENO^®-Fermanit, are commercially available and commonly applied in filters to shorten or even avoid the start-up period of the filter for manganese and iron removal.

Since 2007, glass beads have been used for filtration purposes (Klaus 2015). They are commercially applied as filter pack media and have been proved to be successful as filter media in deep-bed filtration (Klaus et al. 2014; Klaus 2015). In contrast to conventional granular filter media, glass beads possess considerable advantages regarding their physico-chemical stability as well as their hydraulic performance (Treskatis 2011; Klaus et al. 2014). Although the large-scale production of glass beads for filtration purposes is limited to the areas of application stated above, various coating techniques have been applied to glass beads so that new materials for the selective removal of metals from water were successfully prepared (Liu et al. 2002; Ozmen et al. 2009; Torkshavand et al. 2014).

MnO(OH)_2(s) molecules on the media surface are capable of repeatedly adsorbing Mn²⁺. The result is a self-regenerating process, which enables the continuous adsorption of Mn²⁺ onto the media surface. Cerrato et al. (2011) showed that in the presence of sufficient chlorine, Mn²⁺ adsorbed to MnO_x(s)-coated media is quickly oxidized to Mn(IV), which does not dissolve back into the water. Over time, this self-regenerating process produces an adequate surface coverage of manganese oxides.

This NGE can be used for synthetic coating. Merkle et al. (1997) successfully coated anthracite with a layer of synthetic manganese oxides by a wet chemical process in a batch reactor. Coating with the required stoichiometric concentrations of reactants proved unsuccessful. Precipitation of colloidal MnO_x(s) in solution was immediate, and the media remained uncoated. Merkle et al. (1997) eventually developed methods based on the gradual addition of oxidants in less than stoichiometric increments.

The success of a particle filtration process and its economic performance depend on the type of filter material and its physical properties. Every new filter material must fulfil various requirements before it should be used. These include form, grain size, number of grains, grain surface, density, pore volume, solubility, durability, settling rate and purity (Mörgili & Ives 1979). The general suitability of coated glass beads for filtration purposes was not an objective in this particular study and needs to be further investigated.

The objective of this study was the laboratory-scale production of robust manganese oxide-coated glass beads as filter media and the assessment of their efficiency to remove Mn²⁺ from water.

Two principally different coating techniques were examined and compared.

• 1. Dry coating methods

A catalytically active manganese ore was pulverized and fixed on the glass bead surface.

• 2. Wet chemical coating methods

NGE was utilized to produce a layer of synthetic oxides on the surface of modified glass beads. Batch experiments were modelled after the approach described by Merkle et al. (1997). To minimize MnO_x(s) precipitation from solution, further experiments were conducted by applying an alternative experimental setup that uses a packed-bed column.

For microbial oxidation and the removal of Mn²⁺ under aerobic conditions, quartz sand with grain-sized fractions of 0.71–1.25 mm to 1.0–2.0 mm are used (DVGW Regelwerk 2005). Commercially available products such as Manganese Greensand^® have a grain-sized fraction of 0.25–1.25 mm (DIN EN 12911 2012). Thus, as raw materials, glass beads with a grain-sized fraction of 0.75–1.0 mm (product of Sigmund Lindner GmbH) were used to further prepare the media.

• 1. Dry coating techniques

• (a) Coating with a binding agent: in total, 2 kg of glass beads were tumbled in a rotating barrel (model: Baier's Enkel #71050) with 20 g of a sol-gel material (silane) as a coating agent at 30 rpm for 1 h. At the beginning of the silane hardening process, 10 g of coating material was added and tumbled again under the same conditions for 1 h. As coating material, a granular manganese ore (GENO^®-Fermanit) was finely ground (d₅₀ ∼15 μm) with a vibratory discmill (model: Siebtech, type T250). With a proportion of 78% manganese oxides, subsequently termed as MnO_x-78, GENO^®-Fermanit is commonly used in water treatment to remove iron and manganese from water. To remove excess coating material, the beads were sieved (mesh size 400 μm) and packed for further testing.

• (b) Adhesion coating: to achieve surface roughening, 40 kg of glass beads were tumbled in a rotating barrel (model: Walther Trowal A220) with 1 kg silicon carbide [SiC (110 μm)] at 40 rpm for 12 h. Following this, the material was thoroughly rinsed to remove fine particles with demineralized water and air-dried. The matted glass beads were mixed with 10 g MnO_x-78 and 2 ml of water to prevent dust formation and tumbled again by a rotating barrel (model: Baier's Enkel #71050) for 1 h. Subsequently the beads were sieved (mesh size 400 μm) and packed for further testing.

Conditions of media preparation using dry coating techniques are summarized in Table 1.

• 2. Wet chemical coating techniques

• (a) Synthetic MnO_x(s)-coating in a batch reactor: the experiments were modelled according to Merkle et al. (1997). Conditions of the media synthesis process are summarized in Table 2 (samples 3–5). The general procedure regarding samples 3 and 4 can be described as follows: a buffered (HCO³⁻) solution of Mn²⁺ (added as MnSO₄·H₂O) was prepared, and a defined quantity of matted glass beads was added. In total, 500 ml of an oxidant solution (HOCl) was prepared and gradually added to the gently stirred manganese solution at 50 ml/min. After this addition, stirring continued for a certain amount of time. To reduce the formation of MnO_x(s) precipitates from the solution, the pH of each stirred solution was periodically checked and adjusted to be maintained between 7.2 and 8.0 by dropwise addition of 0.5 mol/l NaOH or 0.5 mol/l HCl as required. For the production of sample 3, a stepwise approach was pursued. To apply this sequential batch method, two solutions were prepared: solution 1 was 250 ml of a Mn²⁺ solution (1,000 mg/l), and solution 2 was an oxidant solution (HOCl 138 mg/l). In the first step, 25 g of matted glass beads were added to solution 1. The solution was gently stirred for 2 min, drained off and saved. In the second step, solution 2 was added, stirred, drained off and saved. The procedure was repeated for 15 min.

Conditions for media synthesis by wet chemical coating are summarized in Table 2.

Before Mn²⁺ determination, the sample volume was filtered by a syringe filter (0.2 μm) to remove particulate matter. Partitioning of the sample was performed for threefold determination. The Mn²⁺ concentration of each 5-ml volume was determined by a rapid test (Merck 1.114770.0001), followed by an absorption measurement at 450-nm wavelength using a photometer (Analytik Jena, model: Specord 200).

Analogous to each sampling pH, temperature and O₂ concentration were monitored.

Temperature and pH were measured with a pH meter (Mettler Toledo, model: FG2/EL2). The O₂ concentration was determined using an oxygen sensor (PreSens, model: Fibox 3 trace v3).

For scanning electron microscopy (SEM) analysis of the bead surface, samples were air-dried, glue-mounted and coated with a gold-layer to a thickness of 8 nm by a Sputter Coater (Bal-TEC SCD005 70S). Electron micrographs were recorded by detecting secondary electrons at nominal magnifications of 80 × , 300× and 2,500× using a JEOL JSM-5900LV electron microscope operated at 15 kV. For better contrast of the coated areas in comparison with the uncoated areas, additional images were taken at 300 × magnification by detection of backscattered electrons (BSEs). Energy-dispersive X-ray spectroscopy (EDX) was performed at 300 × magnification used for the chemical characterization of the top 1–2 μm of the bead surface.

Representative images of the media surface (300-fold magnification) from the BSE microscopy analysis were examined by the Java-based image processing program, ImageJ. The total coated area was determined using false-colour imaging.

To determine the efficiency of the coated glass beads in terms of their efficiency to remove manganese from water, a reactor modelled after a system that was first described by Carberry (1964) was used. The reactor vessel (2 l) was placed in a heatable water bath. The vessel was filled with a buffered 20 mmol/l TRIS solution and adjusted to pH 8.5 by addition of H₂SO₄. Then a concentration of 8 mg/l Mn²⁺ was adjusted by addition of MnSO₄. The solution was heated to 35 °C and continuously aerated by a glass frit with pressurized air. The three wire cages of the rotating shaft were filled with 6 g of coated glass beads each, to form a thin layer of active material in each cage. The stirrer shaft was rotated at 200 rpm. In total, 15 ml of sample volume was withdrawn at the beginning and every 20 min until the end of the experiment at 80 min.

As part of SEM analysis, samples were viewed in secondary and BSE modes. The glass beads consist of soda-lime glass with its main constituents SiO₂ (72.5%) and other diverse types of metal oxides. The powdered coating material MnO_x-78 consists of 78% MnO_x as well as Fe₂O₃, SiO₂ and Al₂O₃ as the main minor components. In the wet coating process, pure MnO_x(s) is deposited onto the media surface, as confirmed by EDX analysis. When examining the BSE image, brighter areas correspond with a higher relative nuclear density and atomic weight, i.e. in this case, the coating appears brighter than the bead. Thus, visual evaluation using ImageJ as an image processing software gives a measure for the success of the coating process. For elemental identification of the surface or surface features, EDX analysis was conducted.

EDX analysis was performed on a representative dark and bright spot of the image (Figure 5). The X-rays emitted from the sample surface, plotted as a spectrum, illustrate that there are different elements present on the bead's surface. Si, Na, O, Ca and Au are detected on dark as well as bright areas. Other than Au, which is present because of sample preparation (gold coating), these are the primary elements of the glass bead. Except for O, which is present not only in the bead but also in the coating material MnO_x-78, these elements are more prevalent on dark spots. Mn is absent on dark spots, in contrast to bright areas where Mn is detected. Thus, it can be concluded that the flaky and finely ground material seen on those bright spots is MnO_x-78, proving the occurrence of the coating material on the bead surface even after rinsing the beads. Thus, provided laminar conditions, it can be assumed that those beads are suited for deep bed filtration.

BSE images of samples 5 and 7 are not shown in these findings. Because of the omnipresence of Mn on all areas of the representative BSE image, as detected by EDX analysis, a complete covering of the bead surface by synthetic MnO_x(s) can be concluded. EDX results of samples 3–6 (batch approach) showed a higher presence of Si, Na, Mg and Ca than those of samples 6–8 (column approach), indicating a generally thinner layer of MnO_x(s) on the bead surface.

With regard to the column approach, the controlled dosage of an oxidizing agent directly before entering the filter bed causes the sample solution to immediately contact the glass beads. The fluidization of the filter bed to a defined bed height ensures optimal mixing and contact of the sample solution with the filter media, which results in uniform coating. The simple operation as well as an uncomplicated rinsing by backwashing at the end of the coating procedure indicates that the column approach may be economically advantageous for scale-up.

Different techniques for coating glass beads with a powdered sorptive material as well as methods to synthetically coat glass beads with MnO_x(s) were newly developed or refined. Glass beads prepared by dry adhesive coating successfully remove Mn²⁺ from water at a given pH of 8.5 and an O₂ concentration of 5.8 mg/l. By refining the experimental setup from a batch approach described by Merkle et al. (1997) to a setup where the filter media are coated in a packed bed column, potent filter media with an even and homogenous coating can be produced. In contrast to the use of conventional filter media such as quartz sand, where foreign inclusions such as feldspar will reduce the adsorption capacity of the filter media after a number of backwashings (Mörgili & Ives 1979), glass beads are uniformly composed. In conjunction with their high physico-chemical stability as well as their optimal hydraulic performance (Treskatis 2011; Klaus et al. 2014), MnO_x(s)-coated glass beads promise to be a viable alternative to conventional filter media, such as quartz sand and granular manganese ores. The ability of Mn oxides to adsorb heavy metals such as Pb²⁺, Cu²⁺, Co²⁺, Cd²⁺, Zn²⁺ and Tl⁺ (Gadde & Laitinen 1974; Xiong Han Fenga et al. 2007) suggests additional applications of glass beads for the adsorption of contaminants in water, which are commonly associated with mine drainages or landfill leachates.