Variance of mercury speciation in municipal sewage treatment plant: effects of mercury on the atmosphere

The speciation of mercury in various processing units of sewage treatment in autumn and winter were studied to understand the conversion and fate of mercury. The results show that the average concentrations of total mercury (THg) in the inuent were 130.5 ± 69.8 ng/L and 231.3 ± 107.2 ng/L in autumn and winter, respectively, and the particulate mercury (PHg) was the main speciation (accounting for 59.3% and 86.9%, respectively). The proportion of dissolved mercury (DHg) increased after treatment, and the total removal eciencies of THg were 78.9% and 90.8%, respectively. The release of mercury into the atmosphere during wastewater treatment was studied for the rst time. The dissolved gaseous mercury (DGM) levels in the inuent in autumn and winter were 0.60 ± 0.40 ng/L and 0.34 ± 0.21 ng/L, respectively. The estimated mercury from aeration was 3.94 kg per year in China. DGM will be released to the air if the sewage treatment plant is open-air. Closed sewage treatment and collection of waste gas treatment are necessary to reduce the inuence of released mercury. Mercury releases into the atmosphere in the process of sewage treatment is one of the fates of mercury in sewage. Closed sewage treatment and collection of waste gas treatment are necessary to reduce the inuence of released mercury. The reactive mercury (RHg) levels in the inuents of autumn and winter were 1.28 ± 0.49 ng/L and 1.96 ± 0.43 ng/L, and these levels account for a small proportion of THg, only 1.7% and 0.8%. Hg 2+ were released by the degradation of organic matter in the secondary biological treatment. The THg levels in dehydrated sludge were higher than those in biochemical sludge but lower than the maximum limit of THg in agricultural sludge.


Introduction
Hg, a highly toxic heavy metal element that causes health and environmental problems, has been a point of great concern for decades ( Kyrre et al., 2017). With accelerated urbanization, municipal sewage treatment plants (MSTPs) have become a gathering place for domestic sewage. The mercury discharged from municipal sewage into the environment has been underestimated or even neglected, but in fact, it is an important part of China's anthropogenic mercury emissions (Liu et al., 2016;Hui et al., 2017). In the process of water treatment, species of mercury change and are redistributed in various phases to re-enter the environment. The total mercury released into the aquatic system through municipal sewage worldwide was approximately 16-81 t/a (Ullrich et al., 2001). In 2015, the analysis of the materials of 62 MSTPs in 24 cities in China showed that the amounts of THg released to the aquatic environment, soil and atmosphere from sewage were 23 t (17 t and 6.2 t from untreated and treated sewage, respectively), 120 t and 15 t, respectively . There are many mercury species in water (Ullrich et al., 2001); among them, DGM mainly exists as elemental mercury (Hg 0 ), which can be directly exchanged with the atmosphere; RHg mainly exists as free Hg 2+ , which has high activity and can complex with inorganic ions or organic matter. Different mercury species have different environmental impact mechanisms and bioavailability for the environment due to their different physical and chemical properties. Only a small part of the mercury entering the sewage treatment plant is discharged with the e uent; most of it is absorbed by the activated sludge and discharged with the remaining sludge (Mao et al., 2016;. The remaining sludge is used in agriculture, placed in land lls and incinerated (Takaoka et al., 2012; Pudasainee et al., 2013); then, mercury re-enters to the environment. Previous studies have shown that agricultural use of sludge brings large amounts of mercury into farmland (Kocman et al., 2017). However, the mercury released by sludge incineration contributes the most to mercury emissions from sewage treatment plants. In 2010, 12 t of Hg was emitted by municipal waste incineration (including sludge and other domestic waste) in China (Zhang et al., 2015), and 37 t of mercury was estimated by Tian et al. (Tian et al., 2012). It must be pointed out that burying mercury in the sludge in land lls is not entirely without negative consequences. Gaseous mercury from municipal solid waste in land lls is released into the atmosphere and groundwater, which has attracted people's attention for many years Many studies have been conducted on the migration and removal e ciency of THg and methylmercury (MeHg) and the methylation of mercury in wastewater treatment processes (Balogh and Liang, 1995;Bodaly et al., 1998;Gbondo-Tugbawa et al., 2010;Li et al., 2014;, focusing on dissolved mercury (DHg) and particulate mercury (PHg); no studies focused on DGM and RHg. In the natural water environment, Hg 2+ can be reduced to Hg 0 by photoreduction and microbial reduction (Mason et al., 1995;Nascimento and Chartone-Souza, 2003; Gonzalez-Raymat et al., 2017); then, Hg 0 is released to the air. Since many microorganisms in the biological treatment decompose organic substances in sewage water, we speculate that elemental mercury will also be generated during this process, which will be released into the air through aeration. By studying the speciation of DGM, RHg, DHg, and PHg in the sewage treatment process, the possible conversion process and removal mechanisms of different mercury species were analysed to learn their possible impacts on the environment.

Materials And Methods
The samples were collected at the World Garden Water Puri cation Plant (WGWPP) in Qingdao, China. The WGWPP mainly receives domestic sewage from surrounding universities and residential areas; its sewage treatment capacity is 6000 m 3 /day. The anaerobic-anoxic-aerobic traditional activated sludge method and membrane ltration reactor (A 2 /O-MBR) are used. The main process con guration is shown in Fig. 1.

Sampling and storage
Sampling was performed for 5 consecutive days from October 30 to November 3 in 2019 and January 11 to 15 in 2020, representing autumn and winter samples, respectively. The sampling time was 9:00 am, and 7 water sampling points were set ( ~ ) along with 4 sludge sampling points ( ), (Fig. 1). The temperature, pH and dissolved oxygen (DO) were measured at the same time. GEM was collected in aerobic tank( ) in autumn. SKC 5000 pump was used to collect the gas with a ow rate of 600 mL/min for 20 minutes. The Te on tube was penetrated to about 50 cm from the water surface, and the GEM was nally trapped by a gold sand tube, a soda-lime tube was connected to remove water vapor and acidic gases. Three parallel samples were collected every day. Finally, GEM was measured by a cold atomic uorescence system spectrometer (CVAFS, Model III, Brooks Rand Labs, USA). The similar details of the sampling methods and determination of GEM were given in (Chong et al., 2019) After the wastewater samples were brought back to the laboratory in the dark, they were divided into 4 parts. One portion of the water was immediately used to measure DGM and RHg rst. The second part was transferred to a 250ml Te on bottle, and trace-metal grade HCl was added for storage to measure THg. After 30 min of natural settlement, the supernatant water was ltered through a 0.45-µm cellulose acetate lter membrane. Then, the ltrate was transferred to a 250-ml Te on bottle, and trace-metal grade HCl was added for storage to determine DHg, while sulfuric acid was added into another ltrate for storage to determine parameters, such as COD, TN and TP. The precipitated sludge samples were centrifugally dehydrated and lyophilized for preservation. After grinding through a 100-mesh sieve, the sludge was sealed and stored.

Sample analysis method
DGM in a 400-mL water sample was bubbled using mercury-free nitrogen at a ow rate of 350 mL/min for 30 min.
Dissolved gaseous elemental Hg was adsorbed on a gold sand tube and then measured by a cold atomic uorescence spectrometer (Brooksrand Lab, Model III) . SnCl 2 solution was added in the bubbler bottle to reduce the Hg 2+ to Hg 0 , and then, Hg 0 was absorbed by the gold sand tube bubbling with mercury-free nitrogen at a ow rate of 350 ml/min for 30 min The gold sand tube was heated to measure mercury by the cold atomic uorescence spectrometer. This portion was RHg.
The un ltered and ltered water samples were digested with nitrohydrochloric acid (HCl: HNO 3 = 3:1) at 98℃ for 1 h to measure THg and DHg. Then, the digested water samples were measured according to the method of EPA1631 (USEPA, 2007). Brie y, 5 mL digested water was oxidized by bromine monochloride solution (BrCl). Then, hydroxylamine hydrochloride (NH 3 OHCl) was used to reduce excess oxidant prior to quanti cation by tin chloride reduction. Mercury was determined by an atomic uorescence spectrometer (Brooksrand Lab, Merx-T). The concentration of PHg was the difference between THg and DHg.
The chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP) in sewage were determined referring to "Water and Wastewater Monitoring and Analysis Method"(SEPA, 2002). The sludge sample was digested by nitrohydrochloric acid at 98℃ for 1 h to determine the total mercury by atomic uorescence photometer (AFS-920).
All the samples were collected in acid-washed Te on bottles (4 mol/L and 1% HCl treated successively in 60℃ for at least 12 h) and stored double bagged. The glassware was soaked in 4 mol/L HNO 3 acid for more than 12 h, rinsed with deionized water, heated at 500 °C for 1 h, then sealed in a bag for use. Two blanks, 10% parallel and 2 spiked recovered samples, were measured to check the accuracy. The relative standard deviations (RSD) of the measurements of the parallel samples were all less than 10%, and the recovery rate was approximately 92%.

Modeled Hg Saturation and Flux
T is the water temperature in K.
Water/air ux of Hg was calculated using a thin lm gas exchange model developed by (Wanninkhof and Rik, 1992) and the water/air exchange ux of Hg, F (ng·m − 2 ·h − 1 ), was calculated using Eq. (3): K w -the gas transfer velocity of a Hg 0 in the water/air surface (cm·h − 1 ) and was calculated according to Eq. (4): (Wanninkhof and Rik, 1992) K w =0.31 × U 10 2 ×(Sc Hg /600) −0.5 (4) U 10 -the wind speed normalized to 10 m above water surface; Sc Hg -the Schmidt number for Hg, is de ned as follows: Sc Hg = ν /D Hg (5) ν-the kinematic viscosity (cm 2 ·s − 1 ) of freshwater; D Hg -the Hg diffusion coe cient (cm 2 ·s − 1 ) in freshwater, which was calculated by the molecular dynamics simulation, as described by (Kuss et al., 2009). 36.3 mg/L in autumn and winter, respectively, resulting in a lower total mercury concentration in raw sewage. THg in winter was higher than that in autumn, which may be affected by higher the concentration of organic pollutants. The total amount of sewage treatment in winter was smaller than that in autumn and was affected by less bathing water.  3.2 Change in mercury speciation in the sewage treatment process

DHg and PHg
In autumn, the average DHg concentration in in uent was 50.3 ± 27.8 ng/L, ranging from 22.4 ~ 93.7 ng/L with 55% CV, while it was 24.1 ± 3.2 ng/L, ranging from 20.8 ~ 28.1 ng/L with 13.3% CV, in winter, which was lower and more stable (Fig. 3). DHg decreased from 24.1 ± 3.2 ng/L in in uent water to 16.  (Fig. 3). The DHg/THg ratio was higher in autumn; therefore, DHg was the main speciation of mercury in the wastewater treatment. In winter, although the PHg concentration of wastewater in the biological tank decreased obviously, PHg was much still higher than that of DHg in winter, in contrast to autumn. In winter, the water temperature was low (approximately 13℃); sludge bulking, poor sedimentation performance and high sludge particle content could lead to high PHg concentration, which was also the reason for the high THg in the biological tank. Some studies have found that sludge bulking caused by lamentous bacteria can be successfully induced when the operating temperature of the traditional process is below 14 ± 1℃ (Gao et al., 2020); then, PHg decreased to 13.4 ± 10.3 ng/L after the membrane lter. The removal rates of DHg and PHg were 50% and 87%, respectively, in winter. Although DHg/THg also showed an upward trend in winter (Fig. 3B), in contrast to autumn, the PHg was always higher than that of DHg in the sewage treatment process, and the nal e uent DHg/THg was 44.4%, which was higher than reported previously (Balogh and Nollet, 2008a) (mean 36%, 24%~49%); thus, PHg was the main mercury species in sewage in winter.

DGM and RHg
DGM is gaseous mercury dissolved in water, which is easily released into the air during aeration and water ow. The DGM levels in the in uent were 0.60 ± 0.40 ng/L and 0.34 ± 0.21 ng/L in autumn and winter, respectively (Fig. 5), accounting for 0.5% and 0.15% of THg; these values were much higher than those of other natural water bodies (  There was a slight decrease in the content after the primary treatment because part of the DGM was released to the air with the water ow. We found that the DGM levels in the anaerobic, anoxic and aerobic biochemical tanks were relatively high, especially in autumn; DGM in the anaerobic tank reached 2.48 ± 1.25 ng/L (Fig. 5A). The organic matter in water was decomposed under anaerobic conditions, and part of the mercury was transformed into Hg 0 by microorganisms (Robinson and Tuovinen, 1984). The dissolved oxygen was low in these three biochemical tanks; Hg 0 was apt to form under anaerobic conditions. Hg 0 was stable, and Hg 0 exchange between air and water was small under reduction conditions, resulting in high Hg 0 . The highest DGM concentration in the anaerobic tank indicated that more Hg 0 was produced under anaerobic conditions.
The average concentration of DGM in the three biological tanks was 0.86 ± 0.47 ng/L during winter, and this concentration was signi cantly higher than that in the in uent. However, there was no signi cant difference among them, which was different from the case in autumn. In winter, the average water temperature was 13℃, and the lower water temperature was not conducive to the reduction of divalent mercury because the activity of anaerobic To verify whether DGM continues to be produced in the biochemical tanks, the sewage in the three biological tanks was continuously bubbled for 2.5 h, and the released DGM was measured every 0.5 h during the period (Fig. 4). The amount of DGM produced gradually decreased over 2.5 h, and it remained stable after 1.5 h but was still produced after 2.5 h.
Therefore, DGM was produced continuously by some microbial activities under continuous nitrogen conditions. The DGM amount produced in the aerobic tank water was larger than that in the anaerobic and anoxic tanks; therefore, the formation of DGM was more favourable under the aerobic conditions.
The RHg levels in autumn and winter were 1.28 ± 0.49 ng/L and 1.96 ± 0.43 ng/L respectively, accounting for 1. These signi cant differences show that, in autumn, there was no signi cant difference in DGM among other units, except for the anaerobic tanks, although there was higher DGM in the biological tanks (Fig. 5A). The RHg in the e uent was signi cantly different from the other units. In winter, signi cant differences occurred in DGM between other nonbiological tanks and the three biological tanks (Fig. 5B). There was large difference of RHg in the water treatment process from the error bars, which might be related to the small volume of sewage treated. The RHg was lower in the biological tanks compared with the other processes, although there was no signi cant difference of RHg among most treatment processes.

Generation of GEM in aerobic tank
GEM in water is easily released into the atmosphere under aeration. In view of the continuous aeration of air into the aerobic tank in the secondary biological treatment, we speculated that a considerable amount of gaseous mercury will be generated at this stage. As shown in Fig. 6, the average GEM concentration in aerobic tank air was 6.34 ± 0.49 ng/m³, ranging from 4.28 to 9.02 ng/m³, The average concentration of GEM in Qingdao was 2.18 ng/m³ during autumn. (Nie et al., 2020). The aeration rate of the air-blower was stable at 14.2 m 3 /min. After subtracting the original GEM from the air, GEM production was 18.1 ng/m 3 by aeration according to the average sewage treatment capacity (4708 m 3 /d) of WGWPP. Therefore, 3% of mercury in the raw sewage was released into the air by aeration. GEM in aerobic tank of a large sewage treatment plant (Maidao) in Qingdao was also measured, and the average GEM concentration was 15.45 ± 4.11 ng/ m³, ranging from 11.49 to 22.67 ng/ m³ (Fig. 6) ( 1) indicates that the DGM in water was oversaturated and the higher the value, the easier the release to the atmosphere. As shown in Table 3, under the assumption the waste water treatment plant was openair, the average water/air Hg uxes were 71.6 ± 36.7 ng·m − 2 ·h − 1 and 29.4 ± 13.6 ng·m − 2 ·h − 1 in autumn and winter respectively, much higher than that the natural water bodies (Feng et al., 2004;Ci et al., 2011;Wang et al., 2017 ) because of the higher DGM. Lower DGM concentration and lower wind speed in winter make mercury released to air less signi cant than in autumn. When the tanks were covered by shelters and waste air was collected, the released mercury might be removed by the treatment materials partly. However, mercury was released to the air in waste water treat plant without shelters could not be ignored. At present, there are still many open-air sewage treatment plants in the world. If the waste air released was not treated, much mercury will be released to the air which was also an important sink for mercury in sewage. Most sewage treatment plants are located in or near densely populated urban areas, therefore, closed sewage treatment and collection of waste gas treatment are necessary to reduce the in uence of released mercury.

THg in sludge (STHg)
After treatment in each unit, only a small part of the mercury was discharged with the e uent; most of the mercury entered the activated sludge and was discharged with the remaining sludge (Gilmour and Bloom, 1995;Balogh and Nollet, 2008b). The STHg levels in the biochemical tanks were 0.80 ± 0.12(0.57 ~ 1.01) mg/kg in autumn and 0.88 ± 0.14(0.57 ~ 1.09) mg/kg in winter (Fig. 7 STHg in the anaerobic tank was slightly higher than that in the anoxic and aerobic tanks, which was affected by the higher THg in the water of anaerobic tank (Fig. 2). The STHg levels in dewatered sludge were 1.45 ± 0.10 mg/kg in autumn and 1.65 ± 0.08 mg/kg in winter, which was signi cantly higher than those in the biochemical sludge. The decomposition of organic matter in the sludge during dehydration caused a concentration effect (Lu et al., 2008). THg in dewatered sludge was lower than the maximum limit of total mercury for the agricultural use of sludge in Discharge Standard of Pollutants for Municipal Wastewater Treatment Plants (DSPMWTP, GBl8918-2002) (alkaline and neutral soil ≤ 15 mg·kg − 1 , acid soil ≤ 5 mg·kg − 1 ); therefore, the sludge can be used in agriculture.

Conclusion
The concentrations of THg in the in uent of the sewage treatment plant varied greatly from 55.4 to 230.2 ng/L in autumn and 83.6 to 335.8 ng/L in winter. The in uent THg was higher in winter. However, the e uent levels in autumn and winter were 27.5 ± 4.1 ng/L and 21.4 ± 4.0 ng/L, respectively. The removal e ciencies of THg were 78.9% and 90.8%. The removal of THg mainly occurred in the secondary biological treatment process. THg content was signi cantly correlated with the content of organic pollutants in sewage.
The ratios of PHg/THg were 59.3% and 86.9% in autumn and winter, respectively. Therefore, PHg was the main mercury speciation in the in uent, and the removal rate was relatively high in the primary physical treatment process. DHg/THg generally showed an upward trend following sewage treatment; DHg and PHg were the main mercury species in the sewage treatment process in autumn and winter, respectively. PHg contributed greatly to the removal of THg.    The relationship between the amount of DGM released from the biological sewage tanks and the continuous bubbling time.

Figure 7
THg concentration in sludge from sludge production units in autumn and winter.