Hexavalent chromium removal from municipal wastewater using ferrous salts

Chromium (Cr) is an abundant and useful transition metal, considered a micronutrient in its trivalent form and employed in a multitude of industrial processes in its other oxidation states (especially the hexavalent), due to its hardness and resistance to tarnishing. For example, when it is added to iron at levels greater than 11%, it forms stainless steel, and since 1849 it has been applied as a protective and aesthetically pleasing coating on machine and furniture parts by the process of electroplating (Dennis & Such 1993). The trivalent form, Cr(III), is the most stable oxidation state of chromium and compounds/complexes containing Cr(III) are common in natural systems (Vaiopoulou & Gikas 2012). Cr(III) exhibits a low solubility at a neutral pH range (Qin et al. 2005) and is therefore mostly particulate in nature, resulting in low mobility. In contrast, Cr(VI) occurs as a dissolved anion in the aqueous phase, making it highly mobile (Qin et al. 2005). Whilst Cr(III) is considered to be relatively innocuous, Cr(VI) is both an acute and chronic aquatic toxicant (Mount & Hockett 2000) and a known carcinogen in humans via both inhalation and ingestion (Costa & Klein 2006). Both Cr(III) and Cr(VI) can be present in municipal wastewaters, with varying impacts on the performance of activated sludge processes. Cr(III) appears to stimulate activated sludge growth at concentrations up to 15 mg L⁻¹, with lethal doses laying above 160 mg L⁻¹ (Vaiopoulou & Gikas 2012). In contrast, Cr(VI) appears to be toxic to activated sludge at much lower concentrations than Cr(III), at levels above 5 mg L⁻¹ (Vaiopoulou & Gikas 2012). To some extent, Cr(VI) reduction is possible during wastewater treatment via adsorption on the functional mass of either live or dead cells (Pradhan et al. 2017), however, this is not likely to be enough to meet discharge consents of <10 μg Cr L⁻¹. It is expected that under the Water Framework Directive (European Commission 2000), many more wastewater treatment works will be regulated for chromium discharges (United Kingdom Technical Advisory Group on the Water Framework Directive 2008). The present work uses an operational sewage treatment works as a case study to address its particular chromium challenge and to understand how a solution could be rolled out across multiple sites. The sewage treatment works in question has been chosen because it is a typical site with no specific treatment process installed for chromium removal. It already has a discharge consent for chromium, and receives occasional spikes in chromium that are not removed through its activated sludge process, which pose a threat to compliance. Consequently, a new method for chromium removal needed to be implemented at the site.

The sewage treatment works serves a population equivalent of approximately 40,000 and employs an activated sludge process constructed as a biological nutrient removal plant with anaerobic and anoxic zones. Polyaluminium chloride is dosed into the final zone of the aeration plant for phosphorus removal, as the sewage works has a discharge consent for total phosphorus. The rest of the treatment process is a conventional installation with flow reaching the inlet through a mixture of gravity sewers and pumped flows. Screening and grit removal take place at the inlet works and then the flow passes into primary settling tanks. Settled sewage gravitates to a pre-anoxic zone where it is mixed with return activated sludge from the final settlement tanks before splitting between four aeration lanes. Mixed liquor then passes to four final settlement tanks and out to the watercourse. Surplus activated sludge is belt thickened, pumped to a holding tank then taken off site to a local sludge centre for further treatment. Current consent standards are 20 mg L⁻¹ suspended solids, 9 mg L⁻¹ biochemical oxygen demand, 2 mg L⁻¹ ammoniacal nitrogen, 2 mg L⁻¹ total phosphorus, and a maximum permitted flow of treated effluent of 30,000 m³ day⁻¹. The catchment for the works includes an atypically large number of electroplating companies and other industrial activities and has consequently been given additional discharge consents of 4 mg L⁻¹ iron, 68 μg L⁻¹ nickel, 250 μg L⁻¹ zinc, 77 μg L⁻¹ copper, 1 μg L⁻¹ cadmium and 8 μg L⁻¹ chromium. All these metals are adequately removed during treatment with the exception of chromium.

Additional justification for using ferrous iron is its relative economy and ease of use when compared to the other reduction agents. It also has the potential to replace polyaluminium chloride at the site for both solids and phosphorus removal. The present work investigates the feasibility of using ferrous iron to reduce the concentration of chromium in the site's final effluent to levels below 8 μg L⁻¹, first by employing laboratory scale jar tests with wastewater samples collected from the site in question, followed by full scale implementation of ferrous salts dosing. Whilst previous studies such as Eary & Rai's (1988) work have shown that ferrous could be used to reduce chromium in a municipal wastewater setting, very few, if any, have moved beyond the laboratory to test the reaction at an operational site.

Laboratory scale tests were carried out using a Phipps & Bird PB700 jar tester using 2 L square sided jars on real wastewater samples collected from the sewage works. Each test followed a fixed protocol as follows: in experiments where chromium spikes were added, an instantaneous addition of hexavalent chromium as potassium dichromate followed by a 1 minute rapid mix at 200 rpm was employed, then in all experiments, addition of the reducing agent was followed by 3 minutes of rapid mixing at 200 rpm, then 15 minutes at 20 rpm as a flocculation stage, followed by 20 minutes with no agitation as settlement time. Potassium dichromate solution (0.48% chromium) (Fisher Scientific) was used where chromium spikes were required in the experiments, and the reducing agents used in the work were: iron(II) sulphate heptahydrate solid (20% iron) (ARCOS Organics); and then from Industrial Chemicals Limited: ferric sulphate solution (12.5% iron); ferrous chloride solution (10–14% iron); ferric chloride solution (40% iron); and polyaluminium chloride solution (18% aluminium). Liquid samples were analysed for suspended solids using a protocol equivalent to Standard Method 2540, and for chromium using Inductively Coupled Plasma Optical Emission Spectroscopy (ICPOES) in a UKAS certified laboratory to meet the requirements of the Environment Agency Monitoring Certification Scheme performance standard. This method has a limit of detection for chromium of 0.9 μg L⁻¹ and an uncertainty of 12.8% between 10 μg L⁻¹ and 300 μg L⁻¹, rising to 50% for concentrations <10 μg L⁻¹. Some of the samples for these analyses were pre-filtered using 0.45 μm syringe filters.

Full scale trials employed a bulk container of concentrated ferrous chloride and a peristaltic pump to dose 10 mg Fe L⁻¹ into the wastewater flow immediately upstream of the primary settlement tanks. Liquid samples were collected from the final effluent stream and analysed for chromium using the same method outlined above.

The direct measurement of Cr(VI) is possible using a variety of techniques (McLean et al. 2012), however, they are predominantly set up for the analysis of drinking water or environmental samples (Kimbrough et al. 1999) and are difficult to apply to wastewater matrices. Furthermore, the regulatory consent at the sewage treatment works is for total chromium, with no differentiation made between oxidation states. However, through the persistence or removal of total chromium through the treatment works, it is possible to make some informed assumptions about the oxidation state of the chromium, since Cr(VI) is dissolved and Cr(III) particulate. For example, if the chromium entering the treatment works was mostly Cr(III), its presence in the final effluent should be related to the amount of solids in the final effluent, and the converse for Cr(VI). Analysis of more than two years’ data from the site (Figure 1) shows that there is no significant correlation between suspended solids and total chromium in the final effluent, with no seasonal variations observed. This suggests that Cr(VI) may be responsible for the occasional breaches of the 8 μg Cr L⁻¹ discharge consent. Whilst the site is >98% compliant for chromium, with an average concentration of 1.84 μg Cr L⁻¹ in the 230 samples analysed, five samples from the period returned values close to or above the discharge limit. Since these elevated levels occurred over a wide range of solids concentrations, any improvement in solids removal at the works is unlikely to be correlated with an enhanced removal of chromium.

Initial jar tests were carried out using a crude sewage sample from the site and ferrous sulphate as a reducing agent. The total chromium concentration in the crude sewage was 14.3 μg Cr L⁻¹. Supernatant samples were collected from the jars after the settling period and analysed for total chromium. A 43% reduction of total chromium is evident even in the blank sample, suggesting that some of the chromium is present as particulate Cr(III) and settles out of the liquid phase independently of the addition of a reducing agent (Figure 2). Substantially greater reductions are possible using ferrous sulphate, especially at doses >10 mg Fe L⁻¹ (Figure 2). Average percentage Cr removal rates for the Fe masses dosed from the tests are: 47, 63, 72, 81 and 79 respectively for the 5, 10, 20, 30, and 40 mg Fe L⁻¹ doses. Whilst the stoichiometry of the reduction mechanism in Equation (1) shows that three moles of Fe(II) are required to reduce one mole of Cr(VI), a much larger excess of Fe(II) is needed in practice to remove Cr(VI) from a wastewater matrix, a consequence of the suspended solids, phosphorus, organic load and other interfering compounds in crude sewage that are also coagulated by the Fe(II). At an addition of 10 mg Fe L⁻¹, the mass ratio of iron to chromium is around 700 Fe: 1 Cr. However, this level of metal concentration is equivalent to levels dosed at other sewage treatment works for the removal of solids, organic load and phosphorus. Doses in excess of 30 mg Fe L⁻¹ do not appear to impart any advantage in the reduction of chromium. Consequently, the subsequent tests used 30 mg Fe L⁻¹ as the maximum dose.

A strong correlation between solids removal and chromium removal is apparent with the use of ferrous sulphate (Figure 3), suggesting that the chromium in this sample was already in the trivalent, particulate form and hence easily settleable, and/or that the ferrous is reducing any hexavalent chromium present in the crude sewage to the trivalent form. To investigate this further, some parallel jar tests were carried out using a crude sewage sample with a background concentration of 6.4 μg Cr L⁻¹ (oxidation state unknown) with an addition of Cr(VI) in the form of potassium dichromate, which brought the total concentration of Cr in the sample up to 36.4 μg L⁻¹. Polyaluminium chloride and ferrous sulphate were used on subsamples of this spiked sewage, with metal salts doses ranging from 5 to 30 mg L⁻¹. Strong correlations (R² > 0.92) were found between solids removal and chromium removal in both experiments, but the polyaluminium chloride only removed a maximum of 12.6% of the total chromium, in comparison to 60% when ferrous sulphate was used. This suggests that the polyaluminium chloride can only affect the portion of chromium that is present in the trivalent form, consistent with the 16.5% of the total chromium in the spiked sample that could have already been in the trivalent form. This is also in agreement with historic results from the site's final effluent (Figure 1), where spikes of chromium can persist in the aqueous phase despite passing through two settlement stages and the addition of polyaluminium chloride, making it highly likely that the discharge consents were breached due to the presence of Cr(VI).

A convention in wastewater treatment is to consider substances that have passed through a 0.45 μm filter to be dissolved (Hens & Merckx 2002), and those retained by the filter as particulate. This is not strictly accurate, as the size range of colloids overlaps this distinction, ranging between 0.001 and 1 μm, and truly dissolved substances are defined as being smaller than 0.001 μm (Hiemenz & Rajagopalan 1997). However, it is a useful division to use in wastewater operations as particles >0.45 μm are likely to settle during treatment, whilst colloids and dissolved species <0.45 μm are likely to pass out of the final clarifiers to the receiving watercourse. Consequently, to provide an estimate of the efficiency of Cr(VI) reductions in the present study, samples were passed through 0.45 μm syringe filters prior to ICPOES analysis. These samples were considered to contain only dissolved Cr(VI), with the particulate Cr(III) retained on the filters.

To further compare the performance of ferrous salt as a reducing agent with another commonly used coagulant, a jar test was undertaken using crude sewage from the site with a background concentration of 50 μg L⁻¹ total chromium, of which 2 μg L⁻¹ was dissolved (<0.45 μm). A spike of Cr(VI) was added to bring the dissolved concentration of chromium up to 27 μg L⁻¹, resulting in a total chromium concentration of 75 μg L⁻¹. Ferrous chloride was used in this test as it is easier to use at operational scale, being available as a liquid for bulk purchase, in contrast to ferrous sulphate, which is only available as a solid that must be saturated on site before it can be dosed as a liquid, requiring additional capital expenditure and taking longer to implement. Ferric sulphate was used as a comparator in this test as its use in the water industry is as commonplace as ferric chloride.

Ferrous chloride outcompetes ferric sulphate for Cr(VI) removal (Figure 4), achieving an 89% reduction in comparison to only a 19% reduction at a dose of 5 mg Fe L⁻¹. However, at higher concentrations, there is a significant reduction in the chromium concentration after treatment with ferric sulphate. To illustrate: at a dose of 30 mg Fe L⁻¹, the removal rates are 96% and 84% for ferrous and ferric respectively. A possible mechanism for Cr(VI) removal via ferric is that the flocs formed by it may have iron hydroxide surfaces available for chromium to adsorb to. Previous work has demonstrated successful removals of hexavalent chromium using ferric hydroxide based adsorption systems (Bailey et al. 1992; Altundogan 2005; Hu et al. 2005).

As a consequence of the oxygen demand, ferric chloride is more commonly used as a coagulant than ferrous sulphate, even though it requires more alkalinity (9.2 mg CaCO₃ L⁻¹ for a 10 mg FeCl₃ dose in comparison to 3.6 mg CaCO₃ L⁻¹ for a 10 mg FeSO₄ dose). However, the addition of oxygen (or alkalinity) should not be required at the site in question if ferrous is used because dissolved oxygen is likely to be around 3 mg L⁻¹, and alkalinity levels in its screened sewage around 200 mg L⁻¹ ± 100 as CaCO₃ (inferred from measurements taken at a similar site in the region). This means that there could be enough alkalinity and dissolved oxygen in the site's crude sewage to accommodate a dose of ferrous sulphate in excess of 30 mg Fe L⁻¹. In addition, the reaction rate between ferrous and hexavalent chromium will not be limited by these conditions, and should be complete in less than 30 seconds after addition of the ferrous (Fendorf & Li 1996), with the amount of dissolved oxygen present also unlikely to be limiting (Schlautman & Han 2001).

A final laboratory scale trial was performed to directly compare the Cr(VI) removal abilities of ferrous sulphate with ferrous chloride. This was carried out as a risk/benefit exercise to inform the choice of coagulant to take forward to full scale testing. If ferrous sulphate significantly outperforms ferrous chloride and would help to safeguard chromium compliance, despite needing additional capital expenditure in the form of an onsite saturator, then it would be selected for implementation. A crude sewage sample was collected from the site, which had a concentration of 55.6 μg L⁻¹ total chromium, of which 6 μg L⁻¹ was dissolved (<0.45 μm). A spike of Cr(VI) was added to bring the dissolved concentration of chromium up to concentration of 36 μg L⁻¹ ± 4. At all doses of Fe, ferrous sulphate removed greater quantities of Cr(VI) (Figure 5). At a dose of 5 mg Fe L⁻¹, neither the sulphate or chloride form removed enough Cr(VI) to produce an effluent quality of <8 μg Cr(VI) L⁻¹. However, at a dose of 10 mg Fe L⁻¹, both forms achieved an effluent quality <3.5 μg Cr(VI) L⁻¹, with the sulphate form outperforming the chloride form by only 20%. This outperformance appears to decrease linearly at the higher Fe doses of 20 and 30 mg L⁻¹, reducing to 15 and 10% respectively.

Whilst ferrous sulphate does outperform ferrous chloride, both forms are capable of achieving Cr(VI) removal to below the target concentration of 8 μg L⁻¹ using environmentally relevant starting Cr(VI) concentrations in laboratory tests. The additional removal rates possible using ferrous sulphate are not enough to justify the extra infrastructure required to saturate the bulk chemical onsite. Consequently, ferrous chloride was chosen for the full scale trial. This was dosed into the wastewater immediately upstream of the primary settlement tanks at a fixed rate to achieve an average dose of 10 mg Fe L⁻¹. Spot samples were collected at random intervals and demonstrate consistent total Cr removals to below 8 μg L⁻¹ (Figure 6). The crude sewage sample from the 18^th August contained a relatively high concentration of total chromium, at 304 μg L⁻¹, whilst the other eight samples were all below 20 μg L⁻¹. This spike is the key type of event that the ferrous dosing has been installed to mitigate, and on this occasion, appeared not to be carried over from the primary tanks as the concentrations in the settled sewage and final effluent were 8.6 μg Cr L⁻¹ and 1.8 μg Cr L⁻¹ respectively. The concentrations of total chromium in the crude sewage, settled sewage, and final effluent are non-linear and poorly correlated, even without the spike in the dataset. However, the average values of total chromium are 43.9, 9.81 and 2.4 μg total Cr L⁻¹ in the crude, settled, and final effluent respectively. This suggests either that the chromium entering the sewage treatment works is already in the trivalent form and settles in the primary and final tanks without the need of additional reduction, and/or that the ferrous chloride is able to reduce any hexavalent chromium present in the influent to the works to the trivalent form and hence remove it from the waste stream.

Whilst the relationship between total chromium and dissolved chromium in the crude sewage samples is non-linear and poorly correlated, the removal performance of ferrous chloride against the dissolved fraction does suggest that it is reducing hexavalent chromium levels (Figure 7). The spike of the 18^th August appears to comprise 36% dissolved chromium, at 110 μg Cr L⁻¹. This was not carried over from the primary tanks at the time of sampling, as chromium levels were 1.6 μg L⁻¹ in the settled sewage and 1.3 μg L⁻¹ in the final effluent. The average values of dissolved chromium across these samples are 15.1, 1.4 and 1.5 μg L⁻¹ in the crude, settled, and final effluent respectively, suggesting good removal performance of this fraction with the use of ferrous chloride. The average ratio of total: dissolved chromium transitions from 6.2 in the crude sewage to 1.75 in the final effluent, which when considered with the total Cr reductions, suggests that Cr(VI) is being reduced to Cr(III) and settled out of the waste stream. Whilst the ferrous chloride does remove enough total chromium to comply with the 8 μg L⁻¹ discharge consent at an average final effluent concentration of 2.4 μg total Cr L⁻¹, approximately 60% of this is <0.45 μm, and hence likely to be Cr(VI). Further work at the sewage works is required to assess the long term chromium removal performance of ferrous chloride, in addition to its impact on the removal of solids, organic load, and phosphorus.

• Occasional spikes of chromium entering a sewage treatment works are able to pass through to the final effluent and pose a risk to compliance. These spikes are likely to comprise a significant hexavalent chromium fraction, which is dissolved and difficult to remove from wastewater streams using polyaluminium chloride, the coagulant presently used at the site.

• The use of ferrous salts as an alternative to aluminium has been trialled with success at laboratory scale, showing that these salts are able to reduce dissolved hexavalent chromium to insoluble particulate trivalent chromium.

• Removal performances at laboratory scale, at Fe doses of 10 mg L⁻¹, displayed reductions of total and dissolved chromium ranging from 63 to 93%, depending on the starting concentration of chromium and the form of ferrous (sulphate or chloride).

• In all laboratory trials, regardless of the starting concentration of chromium (up to 40 μg L⁻¹) or the particular ferrous salt, a dose of 10 mg Fe L⁻¹ was sufficient to reduce the concentration of total chromium to significantly less than the discharge consent of 8 μg L⁻¹.

• A ferrous chloride dosing system has been implemented at an operational sewage works and has produced a final effluent with consistently low chromium levels containing an average of 2.4 μg total Cr L⁻¹, never exceeding 4.2 μg total Cr L⁻¹, despite influent spikes in excess of 300 μg total Cr L⁻¹. This technique can be readily implemented at any site with a chromium discharge consent.