Removal of trace mercury (II) from aqueous solution by in situ MnO_x combined with poly-aluminum chloride

Mercury (Hg(II)) is regarded as one of the most harmful toxic metals in the environment (Zabihi et al. 2010). Many countries have suffered from mercury pollution, including Iraq, Brazil, Indonesia, the USA, and China (Jiang et al. 2006). Mercury can enter water supplies through industrial waste from various sources such as chloralkali, mining, and metallurgical processes (Weisener et al. 2005). As a result, mercury pollution of water, in particular, has created serious environmental problems (Brown et al. 1979). It is well known that mercury can be harmful to many forms of life, including humans, through bioaccumulation of the toxic metal (Sari & Tuzen 2009). Toxicological studies have indicated that some types of mercury damage the human body through the central nervous system (Sari & Tuzen 2009). Furthermore, even very low levels of mercury in drinking water may be dangerous for humans (Blue et al. 2008; Li et al. 2011). The US Environmental Protection Agency, the World Health Organization (WHO), and China have all recognized this serious health threat and have set standards for maximum mercury levels in drinking water at 2.0, 1.0, and 1.0 μg/L, respectively (Li et al. 2011). Therefore, it is clear that mercury pollution in drinking water needs to be controlled.

The conventional methods for mercury removal from water are biosorption (Wagner-Dobler et al. 2000; Inbaraj et al. 2009), adsorption (Jeon & Park 2005; Sumesh et al. 2011; Xu et al. 2011), ion exchange (Ratto et al. 2000; Anirudhan et al. 2008), and chemical precipitation (Brown et al. 1979; Skyllberg & Drott 2010). However, these methods are used primarily for the treatment of aqueous solutions that contain high mercury concentrations; they may not be effective for achieving the low levels of mercury that meet the standards for drinking water quality. Presently, there are only a few methods that can be applied for trace mercury removal from water (Blue et al. 2008). One such method is enhanced coagulation, which has many advantages such as lower cost, shorter hydraulic retention time, and lack of complicated treatment structures. Thus, enhanced coagulation has frequently been studied as the method of choice to remove water contaminants such as algae (Wu et al. 2011), arsenic (Song et al. 2006), organic matter (Yan et al. 2008), color (Jiang & Graham 1996), and phosphorus (Zhou et al. 2008). Certain types of enhanced coagulation have even been applied in water treatment plants (Yan et al. 2006; Guo et al. 2011).

There are several types of enhanced coagulation such as increasing coagulant dosage, change of pH, addition of coagulant aid, and optimization of process. An adsorption–flocculation process as an enhanced coagulation technology is used in our work. The enhanced coagulation resulting from the combination of in situ-formed manganese (hydr)oxides (Mn (hydr)oxides) with poly-aluminum chloride (PAC) is an adsorption–flocculation process. In an early study, manganese dioxide (MnO₂) produced from permanganate oxidation was shown to enhance flocculation and filtration of contaminants from water of both high and low turbidity (Ma et al. 1997). A more recent study showed that organic pollutants are adsorbed by MnO₂ formed in situ; in addition, atomic force microscope analysis indicated that MnO₂ formed in situ has smaller particle sizes, and consequently higher adsorption capacities than the aged MnO₂ (Zhang et al. 2008). These studies highlight the ability of in situ MnO_x to deal with pollutants. However, to date, the enhanced removal of mercury from drinking water by in situ MnO_x in combination with PAC coagulation has not yet been studied. In this adsorption–flocculation process, the in situ MnO_x transfers mercury to the solid phase. However, since in situ MnO_x particles are small, PAC is used as a flocculant to settle the Hg–Mn particles.

In this work, trace mercury removal efficiency of enhanced coagulation, or adsorption–flocculation, using in situ MnO_x combined with PAC was examined. Contaminated drinking water was simulated with trace mercury solutions, and the effect on water quality was investigated by varying pH and ionic strength. Optimum test conditions for trace mercury removal were determined as a function of reaction time, chemical dosage, pH, and temperature. Moreover, a possible mechanism of trace mercury removal was determined using Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). Finally, water was obtained from a drinking water source, and trace mercury removal by in situ MnO_x combined with PAC was carried out.

Reagents of analytical grade purity were used in these experiments except for nitric acid (HNO₃), which was of metal-oxide-semiconductor (MOS) grade purity. Mercuric nitrate (Hg(NO₃)₂) was obtained from Jiangyan Huanqiu Reagent Company (China). Other chemicals used were purchased from Sinopharm Chemical Reagent Co., Ltd, China.

Freshly distilled water was used for water solutions. The stock solution of mercury (1,000 mg Hg/L) was prepared by dissolving mercuric nitrate (Hg(NO₃)₂) in 2% HNO₃ solution and then diluting with water to achieve the desired concentrations in trace mercury solutions. The solutions of sodium thiosulfate (Na₂S₂O₃) and l-cysteine were prepared immediately prior to use. PAC was purchased from Sinopharm Chemical Reagent Co., Ltd, China. For the PAC solutions in this study, the ratio of active species in the stock solution was above 10% as Al₂O₃. PAC solutions were diluted with distilled water to get 20 g Al/L dilution for use. Reagents were stored tightly capped at 4 °C until needed.

Trace mercury solutions were prepared at various concentrations by diluting the stock mercury solution with freshly distilled water. The alkalinity and ionic strength of drinking water were simulated by addition of 0.001 mol/L sodium bicarbonate (NaHCO₃) and 0.01 mol/L sodium chloride (NaCl) to these mercury solutions.

In situ-formed Mn (hydr)oxides were formed by adding potassium permanganate (KMnO₄) and Na₂S₂O₃ at a molar ratio of 1.5 into trace mercury solutions needed for water treatment.

All glassware was cleaned by first soaking in HNO₃ solution (10–20%) for 24 hours and then rinsing three times with tap water, followed by distilled water. The mercury concentration of distilled water in the glassware was detected to ensure the glassware was clean, and all detection values were less than 0.1 μg/L (for Hg). After coagulation and rinsing, the experimental wastewater containing mercury was collected and treated in the professional wastewater treatment plant to guard against mercury pollution.

Standard jar tests were performed as with a conventional coagulation process. In the first step of the experimental procedure, in situ MnO_x was added into 1 L of trace mercury solution. The solution was stirred rapidly for 2 minutes at 150 rpm and then stirred slowly for 15 minutes at 40 rpm to facilitate mercury adsorption. PAC was then added, and the solution was stirred rapidly for 2 minutes and then slowly for 15 minutes during flocculation. Lastly, the solution was left sitting for 30 minutes to allow particles (flocs) to settle; the supernatant was removed and filtered prior to taking samples.

Each experiment was conducted in duplicate. The blank checks were performed to ensure that mercury was not found in the distilled water used in the samples, and all detection values were less than 0.1 μ/L (for Hg). The control experiment was designed to investigate the ability of in situ MnO_x, in which PAC was the coagulant used to remove mercury in the distilled water system, as PAC may affect mercury removal by in situ MnO_x in the adsorption process. In the jar tests, PAC was added to in situ MnO_x solutions to accelerate the aggregation and precipitate of in situ MnO_x. However, there was no obvious removal of mercury by PAC (figures not shown).

To determine optimum test conditions, the stirring time varied from 0 to 62 minutes, the Mn dosage was varied from 0 to 10 mg Mn/L, the PAC dosage was 5 mg Al/L, pH varied from 3.0 to 9.0, temperature was varied from 5 to 35 °C, and the ionic strength was varied from 0.001 to 0.1 mol/L. In the experimental process, when one of the factors was studied, other factors were kept constant. If not otherwise specified, the solution pH, Mn dosage, temperature, and initial concentration of mercury in the trace mercury solutions were 7.0 ± 0.1, 4 mg Mn/L, 25 °C, and 30 μg/L, respectively. The solution pH was kept constant by adding a series of concentrations of hydrochloric acid (HCl) and sodium hydroxide (NaOH).

Actual drinking water was collected from the Songhua River (Harbin, China) and used as a mercury-contaminated model by the addition of mercury. In these tests, the solution pH was not kept fixed. The quality parameters of mercury concentration, temperature, pH, turbidity, and total alkalinity for this drinking water were 0.02 μg/L, 24 ± 1 °C, 6.8–7.1, 5.2–6.2 nephelometric turbidity units (NTU), and 120–133 mg/L CaCO₃, respectively. The actual water was spiked to obtain the initial mercury concentration (30 μg/L) in the experiments.

The supernatant was taken and filtered immediately through a cellulose acetate membrane of 0.22 μm pore size. The mercury concentration of contaminated water alone after filtration was tested, and the results showed that there was no change in the mercury concentration, indicating that the filters were clean. Then, the samples were acidified with HNO₃ (2%). Total mercury concentration was detected using an inductively coupled plasma mass spectrometry (ICP-MS) spectrometer (NexION 300 Q, PerkinElmer Inc.). Before every sample measurement, a solution prepared by dissolving l-cysteine in 2% HNO₃ (MOS) (1 mg/L l-cysteine solution) was used to rinse the tubes used for ICP-MS analysis. Error bars in figures represent the standard deviation of the mean.

The initial mercury concentration and Mn dosage were 500 μg/L and 20 mg Mn/L, respectively. The precipitates were collected, washed with distilled water, and then freeze–dried in vacuum before FTIR and XPS analysis. FTIR spectra were collected on a Spectrum One B spectrometer (PerkinElmer Inc.).

The XPS measurements were performed on a spectrometer (K–Alpha, Thermofisher Scientific Company) with 350 kcps sensitivity, using an AlKa source (2,000 eV). The vacuum chamber was 1.0 × 10^–8 mbar in the analysis room. All spectra were measured in a 400 μm diameter analysis area. The high-resolution scans were recorded using fixed pass energy of 50 eV and a scanning step of 0.1 eV. The method of depth profiling was Ar erosion, using energy of 2,000 eV, area of 2 mm, and speed of 0.3 nm/s. A binging energy of 284.5 eV was used for C1s peak as an inner calibration standard.

Figure 1(a) shows that trace mercury removal was affected by stirring time. As stirring time was increased from 0 to 62 minutes, mercury removal increased from 0 to 93%. For an increase from 0 to 7 minutes of stirring time, mercury removal increased rapidly to 50%. This was followed by a slow increase of mercury removal to 93% in the stirring time range of 12–42 minutes. After 42 minutes, the removal ratio remained approximately constant. When they are first formed, the particles of in situ MnO_x have been shown to be very small and unregulated, with diameters varying from 20 to 100 nm (Zhang et al. 2008). Thus, in situ MnO_x particles have a greater surface area early on after formation (Zhang et al. 2008). At shorter time scales, the particles may have many active adsorption sites on their surfaces, leading to a rapid increase of mercury removal efficiency. However, as the processes of stirring and adsorption proceed, the particles of in situ MnO_x become larger (Zhang et al. 2008) and the surface active sites likely decrease, resulting in a slower mercury removal ratio.

Thus, in situ MnO_x may have lots of active adsorption sites on the surface, leading to a rapid increase of mercury removal ratio. As the processes of stirring and adsorption proceeded, the particles of in situ MnO_x became larger (Zhang et al. 2008) and the surface active sites decreased so that the mercury removal rate became slower.

Mercury removal increased with increasing Mn dosage (Figure 1(b)). The mercury removal ratio was 10% when the Mn dosage was 2 mg/L. For the Mn dosage increase from 4 to 7 mg/L, the mercury removal ratio increased from 42 to 98%. The surface active sites on Mn may be much more abundant with increasing Mn dosage, resulting in increased mercury removal (Guan et al. 2009a, b). In these experiments, the initial mercury concentration was 30 μg/L, and a mercury concentration of less than 1 μg/L was achieved with an Mn dosage of 7 mg/L. In other studies, ‘Lemna minor’ powder (at 2 g/L) and silica–titania composites (at 1 g/L) were used to decrease mercury concentrations in solution from 20 to 1.9 μg/L and from 100 to 10 μg/L, respectively (Byrne & Mazyck 2009; Li et al. 2011). The comparatively low Mn dosages suggest that in situ MnO_x may have higher affinity for Hg(II) than both the ‘Lemna minor’ powder and the silica–titania composites.

The pH value of the solution was an important factor for mercury removal by in situ MnO_x/PAC, as seen in Figure 1(c). At pH lower than 5, the mercury removal was greater than 97% and the residual Hg concentration was less than 1 μg/L. With the increase of pH from 5 to 7, the mercury removal ratio decreased rapidly from 99 to 42%; the ratio of mercury removal remained almost constant for higher pH values. In general, the mercury removal was highly pH-dependent.

A strong dependency of mercury removal on pH was also reported by other researchers (Daughney et al. 2002; Herrero et al. 2005; Sari & Tuzen 2009; Zabihi et al. 2010; Zhang et al. 2010) and may have resulted from mercury speciation with the change of pH values. Speciation diagrams calculated with MINEQL+ for Hg(II) in the presence of chloride ions showed more than 98% of the total dissolved mercury as neutral species (HgCl₂, Hg(OH)Cl, and Hg(OH)₂) in water (Herrero et al. 2005; Carro et al. 2010). In the present study, the mercury speciation may have been different than that in the presence of chloride ions. There may have been residual Na₂S₂O₃ in solution from the in situ MnO_x formation process wherein KMnO₄ and Na₂S₂O₃ were added at a molar ratio of 8/3 for a complete redox reaction to produce MnO₂. The excess Na₂S₂O₃ may have reduced in situ MnO_x to form Mn²⁺ (Forrez et al. 2010) and may also have formed complexes with mercury (Stumm & Morgan 1996; Ma et al. 2012). However, the exact species undergoing both the reduction reaction and complexation were not determined for the given experimental conditions. Thus, the molar ratio of Na₂S₂O₃/Hg was also analyzed as a function of pH. The mercury species distribution was calculated (Visual MINTEQ 3.0) over the pH range of 0–14 for different molar ratios of Na₂S₂O₃/Hg (Figure 2), with a chloride ion concentration of 0.01 mol/L. As the molar ratio of Na₂S₂O₃/Hg is increased from 0.5 to 2.0, becomes the main species in the pH range of 3.0–8.0. According to Figure 2, a mercury concentration of 30 μg/L (0.15 μmol/L) would require 0.3 μmol/L Na₂S₂O₃ to form complexes at nearly 100%. The required Na₂S₂O₃ concentration was ca. 27.3 μmol/L at a molar ratio (8/3) of KMnO₄ (5 mg/L for Mn) to Na₂S₂O₃ for a complete redox reaction to produce MnO₂. However, the Na₂S₂O₃ concentration was ca. 48.5 μmol/L at the molar ratio of 1.5 (KMnO₄/Na₂S₂O₃) in the present paper. Since the experimental Na₂S₂O₃ concentration was greatly in excess of 0.3 μmol/L at approximately 21 μmol/L (48.5–27.3 = 21.2 μmol/L), the main mercury species may have been negatively charged ⁠, which would be easily adsorbed on the surface of in situ MnO_x.

In the process of mercury adsorption on Aspergillus versicolor biomass, high adsorption was believed to be a result of a high affinity of the mercury species for certain groups of the biomass (Das et al. 2007). Similarly, the high mercury removal at pH lower than 6 may be attributable to the presence of a mercury species with strong affinity for the active functional groups (–OH) on the surface of in situ MnO_x at pH lower than 6, because more Mn ions were produced and may be adsorbed on the surface of in situ MnO_x (Forrez et al. 2010). The positively charged Mn²⁺ on the surface of MnO₂ can adsorb negatively charged algae (cell or dissolved organic matter (DOM)) (Ma et al. 2012), in other words, algae can be indirectly adsorbed on the surface of in situ MnO_x by the action of bridge connection (Mn²⁺ as a bridge). Similarly, in the present paper, the Mn²⁺ on the surface of in situ MnO_x can act as a cation bridge to enhance mercury adsorption due to mercury species as negatively charged ⁠. The decrease in ratio of mercury removal with increasing pH may be the result of competitive adsorption between OH^– and ⁠.

Figure 1(d) shows mercury removal ratio as a function of temperature. Mercury removal increased from 11 to ∼92% as temperature was increased from 5 to 35 °C. The observed increase of mercury removal ratio with increasing temperature reveals the endothermic nature of mercury removal by in situ MnO_x. It also suggests that the available active sites on the surface of in situ MnO_x increase with the increasing temperature (Zabihi et al. 2010). In addition, the interparticle diffusion of in situ MnO_x may be enhanced in an endothermic process, also leading to the increase of mercury removal.

Ionic strength also had a significant effect on mercury removal, as shown in Figure 3. An increase in the concentration of NaCl from 0.001 to 0.1 mol/L caused a corresponding increase in mercury removal from 18 to 65%. It has been noted that light metal ions such as sodium can compete with heavy metal ions in binding to adsorbents due to electrostatic effects (Schiewer & Wong 2000). However, in this investigation, mercury removal increased as the concentration of light metal ions increased. Therefore, the effect of ionic strength on mercury removal may have been caused by the anions rather than the light metal cations (Herrero et al. 2005; Carro et al. 2010). Moreover, it is likely that the different anions (e.g. nitrate ions and chloride ions) had different degrees of influence on mercury removal. Nitrate does not form complexes with mercury, as calculated by MINEQL +; hence, mercury removal was not affected by nitrate ions (Carro et al. 2010). On the other hand, mercury (Hg(II)) can be complexed with chloride or thiosulfate ions to form neutral mercury species (e.g. HgCl₂ and Hg(OH)Cl) and negatively charged species (HgCl₃^–, HgCl₄^2–, and ⁠). As seen in Figure 2, was likely the dominant species due to excess Na₂S₂O₃ in solution. Overall, the calculated mercury species distributions suggest that the mercury removal ratio increased due to the increase in the proportion of negatively charged mercury species. According to Derjaguin-Landau-Verwey-Overbeek (DLVO) theory and extended DLVO theory (Hermansson 1999; Benítez et al. 2007), the in situ MnO_x particle and Mn²⁺ formed an electric double layer. The increase in ionic strength (NaCl) from 0.001 to 0.1 mol/L may lead to the thickness of the double layer being compressed, resulting in an increase in the interparticle distance and a decrease in the energy barrier (the point of maximum repulsive energy) between in situ MnO_x particles and Mn²⁺ (Hermansson 1999; Benítez et al. 2007). Consequently, the Coulomb attractive forces may be enhanced between in situ MnO_x particles and Mn²⁺ (Hermansson 1999). According to DLVO theory and extended DLVO theory, the Mn²⁺ on the surface of in situ MnO_x particles and Hg(S₂O₃)₂^2– may also form an electric double layer ⁠. As ionic strength increases, the thickness of the double layer may be compressed and the energy barrier decreased. As a result, the Coulomb attractive forces may be enhanced between Mn²⁺ and ⁠. Therefore, the aggregation in MnO_x, Mn²⁺, and may be enhanced and more easily form aggregate, transferring mercury to the solid phase. This was the reason for the increase in mercury removal.

Based on these results, it may be speculated that the negatively charged mercury species were easily removed by in situ MnO_x combined with PAC in the adsorption–flocculation process. Looking ahead, this adsorption–flocculation technology may also be effective for the treatment of water from drinking water sources with higher ionic strengths than were tested for in this investigation.

The influence of initial mercury concentration on mercury removal is illustrated in Figure 4. At pH 3 and 5, the mercury removal ratio remained high (with residual mercury concentration at less than 1 μg/L) for initial mercury concentration of up to 30 μg/L (Figure 4(a)). However, at pH 7 and 9, a residual mercury concentration of less than 1 μg/L was only achieved when the initial mercury concentration was less than 10 μg/L. When the initial mercury concentration was more than 30 μg/L (pH 3, 5) and 10 μg/L (pH 7, 9), the residual mercury concentration increased almost linearly with the increase of mercury concentration. This suggests that in situ MnO_x combined with PAC can be effective at removing mercury over a range of different initial mercury concentrations by adjusting the pH value.

FTIR spectroscopy was used for the determination of functional groups on the surface of Mn (hydr)oxides, where such information suggests the nature of possible adsorbent metal interactions (Sari & Tuzen 2009; Li et al. 2011). The FTIR spectrum of in situ Mn (hydr)oxides was obtained and is presented in Figure 5(a). The peak at 1,634 cm^–1 was caused by the deformation of H–O–H, due to the presence of physisorbed water on the surface of Mn (hydr)oxides. The bands at 1,179 and 979 cm^–1 were dominated by the bending vibration of the –OH on the surface of Mn (hydr)oxides. The peak at 579 cm^–1 represents stretching vibrations of Mn–O or Mn–OH (Zhou et al. 2001). The analysis of the FTIR spectral results reveals that substantial hydroxyls exist on the surface of the Mn (hydr)oxides. The hydroxyls may act as the surface active adsorption sites for the binding of mercury to the surfaces of Mn (hydr)oxides (Zhang et al. 2007).

To further detail the mercury removal process, XPS was used to analyze the chemical states of Mn (hydr)oxides. The results of XPS spectra Mn2p and Hg4f are given in Figures 5(b) and 5(c), respectively. The binding energies of 641.7 eV (representing Mn2p_3/2) and 653.2 eV (representing Mn2p_1/2) imply that the primary compound type was MnO₂ or MnOOH (Moulder et al. 1992; Biesinger et al. 2011). The binding energies at 101.2 and 105.0 eV are assigned to Hg4f_7/2 and Hg4f_5/2, respectively (Moulder et al. 1992; Behra et al. 2001). The presence of Hg4f clearly confirms that mercury transferred to a solid phase from a liquid phase. Furthermore, the Mn2p spectrum suggested that MnO₂ or MnOOH played a large role in the adsorption–flocculation process for mercury removal using in situ MnO_x combined with PAC.

Based on the above discussion, we conclude that a possible mechanism for trace mercury removal by in situ MnO_x combined with PAC is interfacial adsorption with co-precipitation, whereby mercury is transferred to the solid phase from the liquid phase by binding with surface hydroxyls (active adsorption sites), and the subsequent Mn–Hg mixture is precipitated by coagulation using PAC.

Trace mercury removal by in situ MnO_x combined with PAC was conducted on water from an actual drinking water source and the results are displayed in Figure 6. The initial mercury concentration was set at 30 μg/L by adding mercury to the water. In all experiments, pH values decreased from ca. 7.0 to ca. 5.8. When the Mn dosage was 4 mg/L, mercury removal was greater than 80% and higher than that achieved in the simulated aqueous solution under similar experimental conditions. The difference may be attributable to the large number of suspended solids (SS) in the water, which may have adsorbed mercury. Furthermore, the water contained natural organic matter (NOM), which has abundant functional moieties (e.g. hydroxyl, carboxyl, and phenol groups) that could have adsorbed mercury (Guan et al. 2009a; Ma et al. 2012). The mercury adsorbed onto SS and NOM could then have been transferred to the precipitate during coagulation. This means that, in the presence of SS and NOM, mercury might be removed even without Mn addition. So we designed experiments whereby mercury was removed by PAC alone to compare with in situ MnO_x. However, the mercury removal was ca. 70% when the PAC dosage was from 2 to 50 mg/L (data not shown), suggesting that in situ MnO_x was still important for mercury removal. For a Mn dosage of 6 mg/L, the residual mercury concentration was less than 1 μg/L and able to meet the standards for drinking water quality (China). The experimental data obtained for water from a drinking water source suggest that in situ MnO_x combined with PAC has the potential to be applied in the practical engineering of trace mercury removal for drinking water treatment.

The residual Mn concentrations after coagulation were measured. However, all Mn concentrations were less than 0.1 mg/L and met the requirements of WHO and China.

The performance of trace mercury removal by in situ MnO_x combined with PAC was affected markedly by the experimental conditions of reaction time, Mn dosage, pH, temperature, and ionic strength. In addition, the amount of mercury removed per amount of Mn added (q) increased with the increase of initial mercury concentration. Based on the FTIR and XPS analyses, the observed high-efficiency mercury removal may be facilitated by the hydroxyl groups (active sites) on the surface of MnO₂ or MnOOH.

The results of the enhanced coagulation experiments using a simulated aqueous solution indicated that in situ MnO_x combined with PAC could be used as an effective technology for trace mercury removal. Thus, under the optimized experimental conditions, enhanced coagulation was carried out in water from the drinking water source, and a residual mercury concentration that met the standard for drinking water quality (less than 1 μg/L) was achieved. Consequently, this method offers promising technology for mercury removal from drinking water and other aqueous solutions.

This work was financially supported by the National Science & Technology Pillar Program, China (No. 2012BAC05B02); the Funds for Creative Research Group of China (51121062); the National Natural Science Foundation of China (51008104); the funds of the State Key Laboratory of Urban Water Resource and Environment (HIT, 2013TS04); and the Fundamental Research Funds for the Central Universities (HIT. NS-RIF. 201188).