Enhanced removal of nutrients and heavy metals from domestic-industrial wastewater in an academic campus of Hanoi using modified hybrid constructed wetlands

Vietnam, like many developing countries, is facing serious water quality issues due to discharging wastewaters of industrial, agricultural and municipal activities, which can constitute a potential risk for aquatic ecosystems, food safety and human health. The treatment technologies such as bioreactors, membrane filtration, nanotechnology and biodegradable material have been developed for the removal of nutrients, organic pollutants and heavy metals (Singh & Thomas 2012; Bavandpour et al. 2018). However, due to the relatively high construction and operation cost for treating complex wastewaters, developing an effective and cost-efficient system for the removal of nutrients and heavy metals remains a critical issue. Particularly, mixing industrial and domestic wastewater that contains both nutrients and heavy metals increases the complexity of wastewaters because of their interactions and variability of the physio-chemical forms. Therefore, there is an urgent need to develop and implement efficient and cheap treatments to depurate the increasing amounts of wastewaters released in the environment. The wastewaters from Vietnam Academy of Science and Technology (VAST) are known to mix releases from laboratories and from domestic activities. Although they have been pretreated using biological methods combined with physical–chemical treatment, the concentrations of some nutrients and heavy metals are still high and do not meet discharge requirements (QCVN 14:2008/BTNMT; QCVN 40:2011/BTNMT). Therefore, the need to enhance drained water treatment efficiency has become an important issue.

Constructed wetlands (CWs) are wastewater treatment systems designed to utilize the natural processes and interaction of wetland plants, bacteria and substrates in a controlled environment (Vymazal 2013). In terms of the complexity of the processes occurring, constructed wetlands create a more diverse environment compared to conventional treatment technologies (Vymazal 2013). Thus, constructed wetlands are now attracting more attention and are considered promising for being an environmentally-friendly method with low investment and operating cost for treating wastewater (Wang et al. 2017). Over the past few decades, CWs have proved to be functional in treating domestic municipal wastewaters (Kivaisi 2001), agricultural effluents (Wood et al. 2007), landfill leaches (Justin & Zupančič 2009), and industrial sewage waters (Guittonny-Philippe et al. 2015). In the wetlands, the processes of eliminating nutrients and heavy metals from domestic–industrial wastewaters are determined by interacting processes, which include retention in substrates, plant uptake and microbial metabolism (Bilgin et al. 2014).

The selection of plants is an important matter regarding wetlands. Selected plants must be tolerant to toxicity and changes in the entering wastewater characteristics. Phragmites australis and Cyperus alternifolius are freshwater wetland plants, which have been widely used in CWs (Ebrahimi et al. 2013; Rezania et al. 2019). The heavy metals and nutrient removal efficiencies by CW vegetated with P. australis and C. alternifolius were studied and discussed. According to Rezania et al. (2019), P. australis is a great plant for phytoremediation of contaminants due to its ability to accumulate the different types of heavy metals, while the plant C. alternifolius had high efficiency in the removal of nutrients and organic matter of wastewater by its rhizomes through the growing season (Ebrahimi et al. 2013; Bilgin et al. 2014).

Earthworms are widely used to treat wastewater in CW systems due to their functions in decomposing of biodegradable materials, as well as increasing enzyme activity and nitrification potential (Xu et al. 2015). Perionyx excavatus is an epigeic earthworm found commonly over a large tropical area and stays active in the surface layers. P. excavatus was well known for breaking down and processing organic wastes (Biruntha et al. 2013). Earthworms can grind the silt and sand particles, increasing the total specific surface area, which enhances the ability to ‘adsorb’ the organics and inorganics from the wastewater (Suthar 2007).

According to the hydraulic conductivity and flow paths, various CW designs were developed using different materials as substrate for CWs. Traditionally, wetlands had been constructed with local soils as substrate. However, there was a problem with overland flow and short-circulating of the wastewater between inlet and outlet because of the low hydraulic conductivity of soils. To overcome this disadvantage, the selection of materials with high removal capacity of contaminants plays a crucial role in CW design. The materials used as substrate in CWs should have good hydraulic characteristics and high pollutant removal capacity, and be cheap and locally available, such as sand, gravel, rice straw and wood chips (Wu et al. 2015; Xu et al. 2015)

The present study aimed to examine the effectiveness of different types of substrate arrangement in hybrid constructed wetlands (HCWs) to treat the wastewater of an academic campus using plants (P. australis and C. alternifolius) with and without earthworm P. excavatus. Thus, one classic hybrid CW with four layers of substrates and two modified hybrid constructed wetlands with seven layers of substrates were designed to confirm the scientific hypothesis that: (1) when wastewater passes through more layers of materials/substrates, particles, pollutants and bacteria in the wastewater can be trapped on the surface or between the gaps of the materials/substrates; (2) more layers can increase the diversity in filtering that may lead to improved efficiency of the treatment. In the modified hybrid CW with earthworm addition, it was expected that it would improve the pollutant removal, because earthworms can contribute as a biological process (bioturbation) in combination with the physical process of wetland to remove pollutants. The finding could provide appropriate design of HCWs for wastewater treatment and significance of substrate arrangement for optimization of CWs.

One HCW and two modified hybrid constructed wetlands (MHCW) were built on campus, Vietnam Academy of Science and Technology, Hanoi, Vietnam. Table 1 described the characteristics of HCWs. Those constructed wetlands were exposed to natural conditions and were made of concrete. A waterproof cement (SIKA, Switzerland) has been applied in two layers to ensure waterproofness for all HCWs. The outside of walls is covered with regular cement. Each pilot dimension was 2.0 × 2.0 × 1.5 m (length × width × height) in total. Total height of the filter was 150 cm with 60 cm saturated with water and 120 cm of filtering materials in total. As a filter media, woodchip, gravel and quartz sand (filtration sand) were selected for all CW systems. Figure 1 provides a schematic diagram of wetland dimensions and operational arrangement.

An outlet made from a 90 cm PVC pipe was placed in the middle of each tank to connect the outlet. Ventilation pipes had then been set up to increase the circulation of air and oxygen in the pilots. About 90 mm drains were implemented on three sides of each tank after filling each tank with 60 cm of filtering material. A ventilation pipe went up and has been cut to be at least 15 cm above the last layer (15 cm above woodchip). They were covered with a PVC T pipe to avoid falling of materials inside the ventilation pipe. Overflow spots were created with a PVC pipe of 20 cm diameter. It has been cut to be at least 10 cm above the last layer (10 cm above woodchip). Partition walls have been installed by cutting and using a magnesium oxide panel. Dimensions of partition walls were 1.95 × 0.65 m.

Pilot 1 (HCW), which was a classical design of hybrid constructed wetland, was designed with four layers of substrates (Figure 2). The surface layer of the pilot HCW consisted of a 5-cm-thick woodchip, which was used as the mulch for the plants in the wetlands. The following was 40 cm of filtration sand (gain diameter of Ø 2/4 mm). The lower horizontal layer of bed was 20 cm thick and consists of medium gravel with a granulation range of Ø 6–10 mm. Underneath, there was a layer of a 60-cm-thick coarse gravel (gain diameter ranging from Ø 20 to 40 mm).

Pilots 2 and 3, which were modified HCWs (MHCW-1 and MHCW-2), were made with seven layers and consisted of imitating a riverbank matrix. MHCW-1 and MHCW-2 pilots were modified regarding the thickness and arrangement of layers from HCW pilot (Figure 3). The thickness of materials was different from that of the HCW; it was designed with seven layers for each pilot. The first layer was a 5-cm-thick woodchip. The second and fifth layers consisted of 20 cm of filtration sand (granulation range of Ø 2–4 mm). The third and sixth layers were 20 cm of medium gravel (gain diameter of Ø 6/10). The first and seventh layers were coarse gravel (gain diameter of Ø 20–40 mm) and those layers was filled into the filters 2 and 3 with a thickness of 20 cm.

Both HCW and MHCWs were planted with two plants. Half of wetland surface was planted with reed P. australis and half with cyperus C. alternifolius with 5 plants per m² for each plant species. Earthworm P. excavatus was added into the filter of MHCW-2 with a density of 5 kg m⁻³.

The wastewater was collected from the campus of VAST. Wastewater from baths, showers, wash basins, and research institutes was collected separately and drained into a common settling tank. The acclimation period lasted 90 days (from January to March 2019) for plants (P. australis and C. alternifolius) and earthworm P. excavatus (juvenile stage) prior to experimentation. After acclimation, the experimental period lasted 153 days (from April to August 2019). Wetlands during the acclimation and experimental periods were fed with wastewater collected from research buildings of VAST. Wastewater was pumped from the adjacent stabilization tank to filters every day. The wastewater was supplied to the HCW with a frequency of six cycles per day (feeding every 4 hours) and with the capacity of 100 liters per filter per cycle. Hydraulic residence time was 7 days in each wetland to optimize the removal efficiency of pollutants by physical and biological processes. The growth status of plants, such as sprouting, survival and wilting, was recorded during the acclimation and experimental periods.

During this period, wastewater samples were taken biweekly from the inlet pipe (INLET) and outlet of each filter (HCW, MHCW-1 and MHCW-2). Forty water samples of INLET, HCW and MHCWs were collected and tested in this experiment. The pH was measured and recorded using a multiparameter instrument (Hanna Hi-98,194). The samples were refrigerated at 4 °C in labelled polystyrene bottles for chemical analysis.

The water samples were further used for the physio-chemical and biological parameters as per standard methods for the examination of water and wastewater (APHA 2012). Chemical oxygen demand (COD) was measured using a closed reflux chromate titrimetric method. Biochemical oxygen demand (BOD₅) was measured using the 5-day incubation method. Total Kjeldahl nitrogen (TKN), ammonium nitrogen (⁠-N), nitrate nitrogen (⁠-N), phosphate (⁠-P) and total phosphate (TP) were analyzed using a spectrometer (UV Spectrophotometer, UV-1800). Total suspended solid (TSS) was determined by the standard paper filtration and direct gravimetric method.

For chemical analysis of Cu, Cr and Pb, a 50-mL water sample was immediately acidified to a final concentration of 5% nitric acid (nitric acid ≥65%, Sigma-Aldrich). Samples were then analyzed by Inductively-Coupled Plasma Mass Spectrometry (ICP-MS, Thermofisher Scientific). The standard solutions were prepared from a multielement calibration solution (Astasol-Mix M010, Analytika, Czech Republic). The samples were diluted in a 3% final nitric acid solution made from nitric acid 65%, TraceMetal^™ Grade – Fisher Chemical, with 1:3 (v/v) for ICP-MS analysis. Quantification limits were 0.5 μg L⁻¹ for Cu, 1.0 μg L⁻¹ for Cr and 0.1 μg L⁻¹ for Pb.

All statistical analyses were performed using GraphPad Prism 8.3.0. An analysis of variance (ANOVA) test (Tukey's test) was used to determine statistical differences of the experimental results among treatments at 95% confidence level. All data was expressed as Mean ± S.E (standard error).

The results of the HCW and MHCW systems after 5 months of operation showed that the nutrient removal efficiency and removal loading rate differed depending on the modified HCWs (Figures 4 and 5). The concentrations of -N, TKN, -P and TP of HCW and MHCWs were significantly lower compared with the INLET samples (Table 2).

Throughout the monitoring period, -N concentrations in HCW and MHCWs were lower than those in INLET samples (Table 2). -N removal reached efficiencies above 66% (1.99–2.05 mg -N m⁻² day⁻¹) in all CWs and has shown no significant difference between HCW and MHCWs (p > 0.05) (Figure 4). This result was probably due to the uptake of -N by the plants australis and C. alternifolius in HCW and MHCWs (Saeed et al. 2018).

Although a significant reduction in the -N concentration was found between inlet and outlets of HCW and MHCWs, removals of - N were low, only 34.2% for HCW, 32.5% for MHCW-1 and 41.8% for MHCW-2. This result showed that the removal efficiency of -N in MHCW-2 reached 41.8% (with removal loading rate of 2.48 mg m⁻² day⁻¹) and was significantly higher than those in HCW and MHCW-1 (p < 0.05) (Figure 4). This is consistent with stated results in the literature in that mostly -N and partly organic nitrogen are converted to -N in aerobic conditions (Vymazal 2013; Vymazal & Kröpfelová 2015). Processes of ammonification and nitrification-denitrification also aid in the removals of -N and -N (Cooper 1996). However, it is difficult to maintain the optimal conditions for the complex processes of nitrification and denitrification in HCW.

The highest removal of TKN was achieved in MHCW-2 (41.3%), followed by MHCW-1 (34.2%), and lowest in HCW (24.9%). The removal loading rate of TKN in HCW was 1.77 mg m⁻² day⁻¹, which was significantly lower than those in MHCW-1 (2.1 mg m⁻² day⁻¹) (p < 0.05) and in MHCW-2 (2.85 mg m⁻² day⁻¹) (p < 0.05). Comparatively, three systems showed low removal of TKN as compared with those in the literature (Nguyen et al. 2017; Haydar et al. 2020). The low removal in the TKN could possibly be due to no further uptake of TKN by the plants. Furthermore, the presence of organic matters (such as rice husk, coco peat, wood mulch, and biochar) is also proportional to the TN removal due to media oriented internal C generation (Saeed & Sun 2011; Saeed et al. 2012; Tee et al. 2012; Zhou et al. 2017). However, both HCW and MHCWs had lower amounts of organic matters, having only 5 cm of woodchip layer in the surface layer of the filters. Thus, low concentration of organic carbon could negatively affect the removal of TKN (Nguyen et al. 2017).

All data taken into account, the concentrations of -P and TP treated by HCW and MHCWs significantly decreased (p < 0.05) (Table 2). Although the -P removal rates in this study showed no significant differences between MHCW-1 and MHCW-2, the removal rate of TP in MHCW-2 (35%; 0.25 mg m⁻² day⁻¹) was considerably higher than those in MHCW-1 (27.5%; 0.2 mg m⁻² day⁻¹) and HCW (19.9%; 0.15 mg m⁻² day⁻¹) (p < 0.05). Previous studies demonstrated that phosphorus is removed from wastewater by various processes such as sorption, precipitation and plant uptake (Vymazal 2007). In addition, the presence of earthworms can result in some phosphorus being converted to form that are plant available (Suthar 2006), because microorganisms in the intestinal organ of the earthworms have the ability to convert insoluble P into soluble forms (Adhami et al. 2014). Thus, some TP was probably released in forms available to plant uptake in MHCW-2.

In general, MHCW-2 showed highest efficiency in removing -N, TKN and TP from wastewater. Previous studies demonstrated that the presence of earthworms can increase TN and TP removal efficiencies in wastewater (Xu et al. 2015). Thus, it can be explained that the earthworm P. excavatus in MHCW-2 had efficient biological potential for conversion of insoluble N and P into soluble forms, which are high-value useful plant growth media (Suthar 2006, 2007). In addition, the reed P. australis and cyperus C. alternifolius had high efficiency in the removal of -N, TKN and TP (Ebrahimi et al. 2013; Shahi et al. 2013; Bilgin et al. 2014; Rezania et al. 2019; Hu et al. 2020).

TSS concentrations significantly differed between the influents of INLET and effluents of HCW and MHCWs (p < 0.05) (Table 2). The efficiency for TSS reached 77.6% in HCW, 78.3% in MHCW-1 and 81.1% in MHCW-2 (Figure 7). Therefore, there were no significant differences in the effluents TSS removal between HCW and MHCWs (p > 0.05). The concentrations of COD significantly decreased after treatments by HCW and MHCWs (Table 2). Average COD concentrations in HCW and MHCWs were below 50 mg L⁻¹, which were significantly lower than INLET (p < 0.0001). The average COD removal efficiencies in HCW and MHCWs were lower than 80% (Figure 6); an insignificant difference was obtained between the three filter units of this study (p > 0.05). The results are in line with the COD removal values for floating treatment wetlands reported by Chen et al. (2016), whereas they are lower than the concentration reductions reported by Ali et al. (2018) for HCWs (80%) and Hu et al. (2020) for sludge treatment wetlands (>98%). The presence of earthworms in MHCW-2 unit did not improve the treatment efficiency for COD, indicating that the low COD removal efficiency was mainly due to physical processes such as sedimentation and filtration (Maucieri et al. 2016). In addition, a positive relationship between COD concentration and water turbidity was proved in the previous study by Barco & Borin (2017). The relatively high removal efficiencies of TSS could result in the relatively high COD removal efficiencies (Ali et al. 2018). In this study, TSS was removed from 77.6 to 81.1% by HCW and MHCW units, and 77–78.5% COD was also removed by those units.

In comparison with Vietnam's discharge limits (QCVN 40:2011/BTNMT and QCVN 14:2008/BTNMT) for industry and domestic wastewater (50 mg L⁻¹ of BOD₅), the average effluent of BOD₅ in MHCW-2 was less than 45 mg L⁻¹ and suitable for disposal into water bodies that are used for agricultural purposes or the equivalent. The results are in line with the removals reported in literature for BOD₅ removal in HCWs (Nguyen et al. 2017). In contrast, BOD₅ values in HCW and MHCW-1 were above Vietnamese's discharge standard for wastewater into environment (above 50 mg L⁻¹). As a consequence, the best removal performance was seen in the MHCW-2 (70.2%), followed by MHCW-1 with 52.9% and only 34.8% for HCW (Figure 6). By ANOVA test (Tukey's test), there were significant differences in the removal efficiency and removal loading rate of BOD₅ between the HCW and MHCWs (p < 0.05). It is suggested that earthworms can enhance the microbial activities, hydraulic conductivity and natural aeration by granulating the clay particles during the vermicomposting process (Suthar 2007; Sinha et al. 2008). However, the removals of BOD₅ of both HCW and MHCWs were low, according to the previous studies (Vymazal & Kröpfelová 2015; Ali et al. 2018; Haydar et al. 2020). The low removal efficiency of BOD₅ could be explained due to the long hydraulic residence time (HRT) to be set up for those filters. BOD₅ efficiencies in this study are similar to those found by Ahmad et al. (2013), who reported that an increase in HRT led to a reduction in BOD₅ removal.

The results showed that the concentrations of Cu and Pb in HCW and MHCWs were significantly lower (p < 0.05) than those in INLET, indicating that those wetlands had efficiently removed Cu and Pb from VAST wastewaters (Table 3). In HCW and MHCWs, the Cu concentrations ranged from 0.5 to 1.2 mg L⁻¹, which were within the permissible limit in Vietnam (2.0 mg L⁻¹; QCVN 40:2011/BTNMT). Although the removal loading rates for Cu showed no significant differences between three HCW and MHCWs (p > 0.05), effective removals of Cu in MHCW-1 (73.5%; 0.28 mg m⁻² day⁻¹) and MHCW-2 (79.4%; 0.31 mg m⁻² day⁻¹) were significantly higher than that obtained in HCW (48%; 0.22 mg m⁻² day⁻¹) (p < 0.05). The removal efficiencies of Cu in this study are in agreement with those obtained from CW treating natural gas storage produced water (Kanagy et al. 2008). This result indicates that more substrate layers in MHCWs resulted in increased removal efficiency of Cu. However, it seems that the presence of the earthworm P. excavatus in MCHW-2 did not increase the effective removal of Cu from wastewater.

Similarly, HCW was significantly less effective for Pb removal (52%) (p < 0.05) as compared with MHCW-1 (71.5%) and MHCW-2 (67.9%) (Figure 8), indicating that the number of substrate layers did affect the absorbing capacity of Pb from wastewater of those wetlands. The results also showed that there were insignificant differences in removal efficiencies between MHCW-1 and MHCW-2 (p > 0.05). Previous study proved that Pb was removed mainly by the roots of the plants and poorly by earthworms (Das & Osborne 2018). Although Pb concentration was over four times higher in the INLET samples (2.2 mg L⁻¹) and reduced to 0.6 mg L⁻¹ in HCW and MHCW-1 (without earthworms), it is slightly above the permissible limit set for industry wastewaters in Vietnam (0.5 mg L⁻¹) (Table 3). The high Pb concentration of INLET wastewaters could reduce the microbe population in both plants and earthworms, leading to a decreased removal efficiency of Pb (Das & Osborne 2018). In addition, the plants used for wastewater treatment would show variable degrees of uptake and accumulation of Pb. According to the study of Khan et al. (2009), P. australis showed lower Pb uptake and accumulation compared with other plants such as Pistia stratiotes, Typha latifolia and Ceratophyllum demersum. Moreover, the C. alternifolius plant in this study has high efficiency in the removal of chemical parameters such as TSS, COD, -N and NH₃-N, and -P (Shahi et al. 2013).

In this study, the concentrations of Cr did not significantly decrease after treating the wastewaters by HCW and MHCWs (Table 3). However, Cr concentrations of INLET wastewater and OUTLET samples (HCW and MHCWs) were lower than 1.0 mg L⁻¹, which did not exceed the permissible limit (1.1 mg L⁻¹) set for industrial wastewaters in Vietnam. The results showed that both HCW and MHCWs had less effective removal of Cr, ranging from 24.6 to 34.6% (0.02–0.03 mg m⁻³ day⁻¹) for both HCW and MHCWs (Figure 8). It is unclear whether HCW and MHCWs can efficiently remove Cr when INLET wastewaters have a low concentration or not. It is because previous studies have proved that CW has effectively removed the Cr (over 80%) from wastewater (Hadad et al. 2006; Khan et al. 2009). Moreover, the input loading of wastewaters is a critical factor for achieving higher removal rates employing wetland systems (Saeed & Sun 2017). In the next study, an assessment of Cr removal efficiency with high concentration of INLET should be conducted to confirm the Cr removal efficiencies of HCW and MHCWs.

Overall, it has been hypothesized that more substrate layers and a synergistic association of the plants and earthworms would enhance the removal of heavy metals. However, more substrate layers (seven layers) showed higher removal efficiency of Cu and Pb in comparison to fewer substrate layers (four layers). In contrast, earthworm addition into MHCW did not significantly affect the removal efficiency of heavy metals. Chen & Hu (2019) indicated that earthworms had an insignificant effect on heavy metal accumulation in the plants and a significant effect on heavy metals removal from wastewater in unplanted wetland systems.

The present study focused on the efficiencies of the arrangement of three substrates and organisms that were used in HCWs for treatment of mixed domestic-industrial wastewater of an academic campus (VAST) before discharging into the urban water bodies. The removals of -N, TKN, TP, BOD, Cu and Pb were more efficient in the wetlands that were arranged in seven substrate layers (MHCW-1 and MHCW-2) than in those with an arrangement of four substrate layers (HCW). Particularly, MHCW-2 (seven substrate layers and earthworm addition) showed higher removals of BOD and Cu than MHCW-1 (seven substrate layers and without earthworms) and HCW (four substrate layers).

The funding support from Vietnam Academy of Science and Technology within the framework of the project entitled: ‘Research on applying SMART GARDEN, an ecological engineering platform, to improve wastewater quality from wastewater treatment plant of Vietnam Academy of Science and Technology’ (contract no. ĐL0000.04/18-19) is gratefully acknowledged. The authors would like to thank Prof. Christian Valentin and Prof. Paco Bustamante for their advice and comments during the development of this manuscript.

All relevant data are included in the paper or its Supplementary Information.