Advanced nitrogen removal from the biological secondary effluent of dyeing wastewater via a biological–ferric–carbon nitrification and denitrification process

Abstract

It is often difficult for the ammonia nitrogen (NH₄⁺-N) concentration in biological secondary effluent of dyeing wastewater to reach the discharge standard by biological treatment due to the nitrogen-containing auxiliary application. Therefore, a biological–ferric–carbon nitrification and denitrification process is applied in this study. In the ferric carbon system, either under the condition of sludge seeding or not, nitrifying bacteria are accumulated to form a biological–ferric–carbon system, in which the nitrification process is initiated. Under the condition of sludge seeding, a higher NH₄⁺-N removal efficiency is reached in a shorter period. In addition, the biological–ferric–carbon system has advantages both in NH₄⁺-N and COD removal because of mutual promotion effects. In the subsequent denitrification process, an extra carbon source is necessary, and the Fe³⁺ in the nitrification liquid is beneficial for denitrification, thus a denitrification efficiency of nearly 90.2 ± 2.1% is achieved in the short period of 2 h, however, only 44.6 ± 1.5% NO₃⁻-N is denitrified during 6 h when Fe³⁺ is removed by coagulation. Under optimal operation conditions for the biological–ferric–carbon nitrification and denitrification process, the NH₄⁺-N concentration decreased from 32.3 ± 1.5 mg L⁻¹ to 2.5 ± 0.2 mg L⁻¹, the effluent total nitrogen was less than 8.8 ± 1.0 mg L⁻¹, and COD was reduced from 120.0 ± 6.0 mg L⁻¹ to 49.4 ± 2.1 mg L⁻¹, which presents a good performance both in advanced nitrogen removal and COD degradation.

---

1 Introduction

Textile dyeing wastewater is typical refractory industrial wastewater because of its complicated compositions, deep color, high turbidity, high content of organic pollutants and fluctuating pH.¹ In China, textile-dyeing wastewater contributes a large proportion of discharge and pollutant emission. In 2013, the discharge of textile industry wastewater accounted for 21.5 billion m³ per year, which is ranked third among the different industries, and large amount of chemical oxygen demand (COD) and ammonia nitrogen (NH₄⁺-N) were discharged.² In order to protect the water body from pollution, especially from eutrophication, more and more stringent discharge standards have been enforced in recent years, and the effluent NH₄⁺-N and TN concentrations of municipal wastewater treatment plants (WWTPs) are limited to less than 5 and 15 mg L⁻¹, respectively (GB 18918-2002).

To save on treatment cost, dyeing wastewater is always pretreated in the dyeing enterprises and then discharged to the WWTPs for centralized treatment. However, most WWTPs apply cheap biological treatment technologies, such as anaerobic/aerobic (A/O) or anaerobic/anoxic/aerobic (A2O),^3–5 which make it difficult to reach the discharge standard especially COD and NH₄⁺-N because of the low biodegradable and toxic characteristics of dyeing wastewater, and the application of urea and aqua ammonia during the production process. Therefore, WWTPs face the challenge of upgrading and reforming. Advanced treatment, which uses biological secondary effluent as influent for further disposal, is a good selection since it can keep the existing equipment. Therefore, low cost and efficient advanced treatment methods are required urgently.⁶ However, most of the advanced treatment methods are focused on COD removal, and advanced nitrogen removal is less documented.

Regarding nitrogen removal, biological methods are often used due to their low cost. In recent years, many state of the art technologies have been introduced, such as short-cut nitrification–denitrification, and the anammox process. These novel biotechnologies have the advantages of energy saving and less greenhouse gas emission,^7,8 especially the anammox-based treatment technology, which is the focus of more and more researchers and is thought to be a promising way of making sewage treatment energy-neutral or energy-generating. However the anammox technology is mainly applied in ammonium rich wastewater, such as anaerobic sludge digester and landfill leachate,^9–11 with ammonium-nitrogen concentrations ranging between 500–3000 mg L⁻¹. In addition, there are still some challenges in the practical application of anammox-based technology at full-scale, such as long start-up period, limited application to mainstream municipal wastewater, poor effluent water quality, and various affecting factors.^12,13 Thus, these novel technologies are not suitable for advanced nitrogen removal in large centralized wastewater treatment plants at the present stage. The traditional nitrification and denitrification process is suitable to adopt due to its easy operation and good nitrogen removal performance for low NH₄⁺-N containing wastewater.

For centralized WWTPs, which receive dyeing wastewater from dyeing and printing industrial parks, the influent NH₄⁺-N concentration is often higher than municipal sewage. In addition the toxicity of dyeing wastewater might inhibit bacteria activity, which results in the biological effluent not achieving the nitrogen discharge standard. Sometimes the effluent COD does not achieve the discharge standard as well. In our previous pilot-scale study of advanced COD removal via the ferric–carbon microelectrolysis (Fe–C ME) process, obvious NH₄⁺-N removal and NO₃⁻-N appearance were observed after Fe–C ME was continuously operated for about half a month under neutral pH conditions without sludge seeding. In addition, it is documented that activated sludge with added Fe³⁺ has a higher capacity to remove nitrogen and phosphate.^14,15 Thus, the hypothesis that the Fe–C system is beneficial for nitrification bacteria accumulation was posed, and a biological–ferric–carbon system, in which ferric, carbon and microbial coexist, is established for advanced nitrogen removal in this study, and ferric–carbon microelectrolysis degrades pollutants and reduces toxicity. Microbes adhere to the activated carbon, and biological activated carbon (BAC) is formed gradually, which is beneficial for pollutant degradation and the nitrification process.^16,17 In addition, the iron shavings used in the Fe–C ME system release Fe³⁺, which could be used as a component of enzymes and is beneficial for microbial activity, since it has been reported in the literature that the bio-ferric process formed by ferric scraps and microbes has advantages in the nitrification process.¹⁸ Therefore relationships of mutual promotion between ferric, carbon and microbes might be formed, which have the role of simultaneous pollutant degradation and nitrification. Then, the denitrification process follows and is optimized to achieve total nitrogen removal.

The objective of this paper is to find an effective and practical method for advanced nitrogen removal from biological secondary effluent in WWTPs receiving dyeing wastewater. Firstly, a biological ferric carbon system is structured for nitrification, and different nitrification bacteria accumulation methods are compared to confirm the advantage of this system. Then, the denitrification process is optimized with regard to carbon source selection, iron ion contribution and carbon source dosage. Since the biological secondary effluent being treated is a real sample from a WWTP that treats dyeing wastewater, the COD was outside of the discharge standard in addition to the NH₄⁺-N concentration, thus the COD removal performance was considered as well during the whole nitrification and denitrification process.

2 Material and methods

2.1 Actual dyeing wastewater

The wastewater used in this study came from the biological secondary effluent of the WWTP in the Zhejiang Province of China, which applied the technology of anaeroxic–anoxic–oxic (A2O) to treat 80% of dyeing water, 10% of industrial wastewater and 10% domestic sewage. The characteristics of this secondary effluent were: COD 110–140 mg L⁻¹, ammonia nitrogen (NH₄⁺-N) 30–35 mg L⁻¹, and pH 7–8. This secondary effluent was used as the influent of advanced nitrogen treatment in this study.

2.2 Ferric and carbon

The ferric used in this study was iron scraps, which came from the waste of lathe cutting, with lengths of 5–10 cm and shape of spiral bends. Before the experiment, the iron scraps were washed and immersed in NaOH (0.2 M) for 24 h to remove surface oil. The composition of the iron scrap was measured via inductively coupled plasma optical emission spectrometry (ICP-OES, PerkinElmer Optima 2100 DV, USA) after digestion and the result is shown in Table S1.† The granule activated carbon (GAC) used in this study was immersed in the wastewater to be treated, and the COD was detected continuously. When the COD was stable, it was considered that the adsorption by GAC achieved saturation, and this saturated GAC was applied in the Fe–C ME process to avoid possible deviation induced by adsorption.

2.3 Initiation of the nitrification process

It was observed that the nitrification process could be initiated in the Fe–C ME system with no extra sludge seeding under the neutral conditions in our previous pilot-scale advanced COD removal study. In the present study the nitrification processes in the ferric–carbon system were initiated under the conditions of no sludge seeding and sludge seeding, respectively. Two reactors (Reactor N and Reactor S) with an inner shelve having the working volume of 1 L were used (the reactor sketch map is illustrated in Fig. S1†), and the ferric and GAC were put on the inner shelve in order to avoid the effect of mixture stirring. 60 g L⁻¹ of ferric and 180 g L⁻¹ of GAC were implemented in the reactor according to our previous study,¹⁹ and a magnetic stirrer was placed at the bottom of the reactor. The influent pH (7–8) was not adjusted, and one liter of biological secondary effluent was added to each reactor. Seeding activated sludge withdrawn from the Songjiang wastewater treatment plant in Shanghai was centrifuged and washed three times using a 0.9% NaCl solution, then was inoculated in Reactor S with 3000 ± 300 mg L⁻¹ of mixed liquid suspended solids (MLSS). The reactor was maintained at room temperature and constantly mixed using the magnetic stirrer except during the settling and decanting period. Air was provided by an aerator using an on/off control system and the dissolved oxygen (DO) concentration was maintained at around 3 mg L⁻¹. After an 11 h reaction period, the mixture was settled for 1 h, and the supernatant was discharged and 1 L of fresh biological secondary effluent was supplemented during the initial 5 min of the next reaction cycle. With regard to Reactor N, its operation conditions were the same as that of Reactor S except that no activated sludge was seeded. After both reactors reached a stable nitrification performance, the optimal reaction time and microbial structure were analyzed.

2.4 Experiments to determine the advantages of biological ferric–carbon for nitrification

When the biological ferric–carbon system was acclimated to stable, 400 of mL sludge mixture was withdrawn from Reactor S and centrifuged and washed three times using a 0.9% NaCl solution. The pellet was remixed in 40 mL 0.9% NaCl solution as the stock sludge solution. Five 250 mL beakers (beakers A, B, C, D, and E) were used, and 10 mL of stock sludge solution was added to beakers A, B, C and D. Then, 6 g Fe and 18 g GAC were supplemented to beaker A to form the biological ferric–carbon system, 6 g Fe was supplemented to beaker B to form the biological ferric system, 18 g GAC was added to beaker C to form the biological carbon system, and no ferric or carbon was added to beaker D, which was used as a single biological system. Pertinent to beaker E, only 6 g Fe and 18 g GAC were added to the form ferric–carbon microelectrolysis system. Then all the five beakers were supplemented with the wastewater to be treated to a final volume of 100 mL. The operation conditions were the same as the acclimation conditions. After three nitrification reaction cycles, the nitrification efficiencies and COD removal rates were detected. All the experiments were conducted in triplicate.

2.5 Denitrification process

After the nitrification process, the denitrification step followed. The sludge was seeded in the denitrification reactor with a working volume of 1 L, and the MLSS was maintained at about 2500 ± 50 mg L⁻¹. The reactor was maintained at room temperature and constantly mixed using a magnetic stirrer except during the settling and decanting period. After bubbling with nitrogen gas for 10 min to keep the DO less than 0.5 mg L⁻¹, the reactor was sealed and anaerobically stirred for 6 h. In order to obtain a better denitrification efficiency, the effects of iron ions contained in the nitrification liquid, the carbon source, and optimum reaction time on the denitrification performance were investigated.

2.6 DNA extraction and MiSeq sequencing

One gram of sludge in the nitrification reactors (Reactor S and Reactor N) was weighed, and the total DNA was extracted using the FastDNA SPIN kit (Qbiogene, Carlsbad, CA, USA) according to the manufacturer's protocol. The purity and concentration of the DNA extracts were measured using a NanoDrop microvolume spectrophotometer (Thermo Fisher Scientific, Wilmington, Delaware, USA). Then the procedure FOR MiSeq sequencing was conducted. The fusion primers (515F: 5′-GTGCCAGCMGCCGCGGTAA-3′ and 806R: 5′-GGACTACVSGGGTATCTAAT-3′) were used to amplify the bacteria 16S rDNA V4 region. The PCR reaction system contained a 25.0 μL mixture, which included 1.0 μL of 20 ng μL⁻¹ template, 5 μL of buffer (5×), 2.0 μL of 2.5 mM dNTP, 5.0 μL of GC high enhancer (5×), 1.0 μL of each primer (10 μM), 0.25 μL of Pyrobest DNA polymerase, and 9.75 μL of ultra-pure water. The PCR conditions were as follows: initial denaturation at 98 °C for 5 min, followed by 25 cycles of denaturation at 98 °C for 30 s, annealing at 50 °C for 30 s, and extension at 72 °C for 30 s, and a final extension at 72 °C for 5 min. The sludge bacterial 16S amplicon sequencing was performed under the Illumina MiSeq platform by the Personal Biotechnology Co. (Shanghai, China).

2.7 Processing of sequencing data

The raw sequences from this research were sorted based on the specific barcodes of sludge samples using the Pipeline Initial Process tool at the Ribosomal Database Project (RDP). The raw sequencing data were deposited to the NCBI Sequence Read Archive with the accession numbers SRR3590404 and SRR3404605. In order to obtain high-quality sequences, ambiguous and homologous sequences, or those with a length shorter than 150 bp were trimmed. The remaining sequences were checked for potential chimeras using QIIME and deleting the chimeras using mothur, which were grouped into OTUs using a 97% identity threshold (3% dissimilarity level). The reads flagged as chimeras were extracted, and the non-chimera reads formed the database of effective reads for each sample. Rarefaction curves were generated, and Shannon, Chao and abundance-based coverage estimator (ACE) indices were calculated to compare the microbial diversity and richness between Reactor N and Reactor S according to the literature.^20,21

2.8 Other analytical methods

According to the Standard Method,²² the COD was measured by dichromate via the open reflux method, and NH₄⁺-N, NO₃⁻-N, and NO₂⁻-N were detected via a spectrophotometric method. The contents of Fe³⁺ and Fe²⁺ were determined by the phenanthroline spectrophotometric method using a UV-Vis spectrophotometer (PerkinElmer, Lambda 35).

2.9 Statistical analysis

All tests were performed in triplicate and the results are expressed as mean ± standard deviation. Analysis of variance (ANOVA) was used to test the significance of the results and p < 0.05 was considered to be statistically significant.

3 Results and discussion

3.1 Biological–ferric–carbon nitrification performances

3.1.1 Nitrification initiation. In our previous pilot-scale advanced COD removal experiment, it was found that NH₄⁺-N was removed significantly and the NO₃⁻-N concentration was increased in the Fe–C system when the influent pH was near neutral. Therefore, the initiation of nitrification process in the Fe–C system was studied by comparing the natural acclimation (without sludge seeding) and sludge seeding acclimation, and the NH₄⁺-N removal efficiency was tracked every day, as illustrated in Fig. 1.

Fig. 1 Performance of NH₄⁺-N removal efficiency in the two ferric–carbon reactors with or without sludge seeding.

It can be seen that when there is no sludge seeding, nitrification in Reactor N occurred after several days culturing, and then the NH₄⁺-N removal efficiency increased gradually to a stable state after 30 days. When activated sludge was seeded in Reactor S, nitrification started immediately. Since the seeding activated sludge in reactor S was withdrawn from the Songjiang wastewater treatment plant, which applies the technology of A2O, the seeding activated sludge had the capacity for nitrification already, thus leading to some nitrification on the first day. Stable, high NH₄⁺-N removal efficiency could be reached after 16 days. Therefore, high NH₄⁺-N removal efficiency could be obtained under the conditions of either the sludge seeding or not, and the biological ferric–carbon system might be formed. For Reactor N, the bacteria in the influent might be selected and accumulated in the Fe–C system, and for Reactor S, the nitrification bacteria came from the seeded activated sludge. Although the time for the nitrification process to become stable was different, the interesting question of whether the microbial communities are similar since the nitrifying bacteria origins were different is put forward.

3.1.2 Microbial communities analysis. Under different acclimation conditions of sludge seeding or not, the microbial communities in the two biological ferric–carbon reactors were investigated and compared using high-throughput pyrosequencing on the Illumina MiSeq platform owing to the huge numbers of sequences it generates.²³ To compare the compositions of the bacterial population involved in Reactor S and Reactor N, it can be seen from Table 1 that a total of 121 649 trimmed high quality sequences were generated after raw sequence processing and denoising. Fig. S2† shows that when the sequencing depth of each sludge sample was >10 000, the Shannon diversities were not obviously increased, which indicates that the high quality sequences with a library size of >55 000 in this study are satisfactory to fully characterize the bacterial communities in the different acclimation reactors.

Table 1 Summary of pyrosequencing data for the reactors with or without sludge seeding

Sample ID	α = 0.03
	Sequence number	OTUs	Chao^a	ACE^a	Shannon^b
a Chao/ACE richness estimator: a higher number indicates higher richness.b Shannon diversity index: an index to characterize species diversity. A higher value represents higher diversity.
Reactor S (sludge seeding)	55 970	1758	6213	9431	5.33
Reactor N (no sludge seeding)	65 679	1826	8861	14 257	5.39

---

To assess the phylogenetic diversities of the bacterial communities in the biological–ferric–carbon Reactor S and Reactor N, the trimmed sequences were grouped into operational taxonomic units (OTUs) using a 97% identity threshold. Table 1 shows that the bacteria sequences from Reactor S and Reactor N were identified to be 1758 and 1826 OTUs, respectively. Although species richness estimators, such as ACE and Chao, indicate that the bacteria in Reactor N is little higher than that of Reactor S, the Shannon index and Venn diagram, which are shown in Fig. S3,† of these two reactors are similar, and 1470 OTUs or 40.9% of the total OTUs (3594) were shared in these two different reactors. Thus, it could be considered that the bacteria richness and diversity in these two different reactors are similar although their bacteria origins are varied.

During the nitrification process, large numbers of relative bacterial population are involved. The information of Phylum and Class distributions was further analyzed, as illustrated in Fig. 2. There were 9 main phyla detected in these two reactors, including Proteobacteria, Chloroflexi, Planctomycetes, Bacteroidetes and Nitrospirae, which accounted for 74.6% (Reactor N) and 82.8% (Reactor S) of the total reads. Pertinent to the class level, both of the reactors were highly enriched with alpha-, beta-, and gamma-proteobacteria, Chloroflexi and Nitrospira, which have been reported to be responsible for nitrification.²⁴ These results indicate that nitrification-relative bacteria could be easily accumulated and grown in the Fe–C system. Although both Reactor S and Reactor N could achieve a good nitrification performance, the nitrification initiation via sludge seeding was adopted due to its shorter time to achieve stability.

Fig. 2 Phylum and Class level distributions of the bacteria populations in Reactor S (with sludge seeding, outer ring) and Reactor N (without sludge seeding, inner ring).

3.1.3 Optimization of nitrification reaction time. In the nitrification initiation stage, the HRT was set long enough as 11 h and NH₄⁺-N could be removed totally in the stable stage. Then the optimum HRT was further investigated. As shown in Fig. 3, the effluent NH₄⁺-N concentration decreased significantly with reaction time. When the HRT was 6, the NH₄⁺-N concentration in the nitrification liquid was low at less than 5 mg L⁻¹, which could achieve the NH₄⁺-N discharge standard of first level A (GB 18918-2002). In addition, considering that the remaining NH₄⁺-N could be used as the nutrient element for microbial growth in the following denitrification period, the optimum reaction time for nitrification was selected as 6 h.

Fig. 3 Effect of reaction time on NH₄⁺-N removal efficiency during the nitrification process. Error bars represent the standard deviations of triplicate measurements.

3.1.4 Advantages of biological–ferric–carbon for nitrification. In the nitrification reactor, ferric, carbon and microbial coexisted, and Fe–C microelectrolysis, and biological activated carbon are formed simultaneously, which are beneficial for pollutant degradation and the nitrification process.^16,17 In addition, the released Fe³⁺ from iron shavings could be used as a component of enzymes and is favorable for microbial activity.¹⁸ It could be assumed that the relationships of mutual promotion between ferric, carbon and microbes might be structured. Based on the above theoretical hypothesis, the advantages of biological ferric–carbon for nitrification were investigated by comparing the nitrification efficiencies in systems containing sludge–ferric–carbon, sludge–ferric, sludge–carbon, ferric–carbon and only sludge. In addition, COD removal efficiencies were also compared since Fe–C ME and BAC in the biological ferric carbon system might function to degrade pollutants.

Experiments to determine the advantages of biological–ferric–carbon were conducted as described in the previous section. Under all the aforementioned conditions, the nitrification effluent had a low concentration of NO₂⁻-N (data not shown), and the majority of NH₄⁺-N was converted to NO₃⁻-N. It can be seen from Fig. 4 that the nitrification efficiency was only 44.7 ± 1.8% in the reactor only containing sludge. When ferric or activated carbon was added to form the biological ferric or BAC system, the nitrification efficiency was increased significantly to 66.3 ± 3.0% and 83.7 ± 3.8%, respectively. For the biological ferric carbon system, the nitrification efficiency was 84.7 ± 4.0%, which is similar to that of the BAC system, and the effluent NH₄⁺-N concentration was less than 5 mg L⁻¹. Therefore, when only considering the NH₄⁺-N nitrification efficiency, both biological–ferric–carbon and biological carbon show good performances. With regard to COD removal, the microbes could not use the refractory pollutant, and thus the COD in the nitrification effluent was not decreased obviously in the reactor only containing sludge. However, when carbon and ferric coexisted with sludge, COD removal occurred, and the COD declined from 120.0 ± 6.0 to 63.2 ± 3.0 mg L⁻¹, thus showing a higher COD removal efficiency than the other conditions. When compared with the COD removal efficiency in the reactors of biological ferric carbon and Fe–C, it could easily be found that the Fe–C system can be responsible for partial COD removal and the microbes functioned in COD removal, as well in the biological ferric carbon system. Above all, considering both the COD and NH₄⁺-N removal efficiencies, the biological ferric carbon system is more suitable than the others, with the effluent characteristics of COD of 63.2 ± 3.0 mg L⁻¹, NH₄⁺-N of 4.8 ± 0.2 mg L⁻¹ and NO₃⁻-N of 28 ± 1.5 mg L⁻¹.

Fig. 4 Removal performance of NH₄⁺-N and COD during the nitrification process in the reactors containing only sludge, biological ferric (sludge–Fe), biological activated carbon (sludge–C), biological ferric carbon (sludge–Fe–C) and ferric carbon (Fe–C). Error bars represent the standard deviations of triplicate measurements.

3.2 Denitrification process optimization

It should be noted that in the nitrification step, the majority of NH₄⁺-N is transferred to NO₃⁻-N, then denitrification step follows in order to remove the total nitrogen. Since the denitrification process demands a carbon source, it is not necessary to remove the COD in the influent during the nitrification period if part of the COD could be utilized as the carbon source in the denitrification step. However, some of the COD came from the biological secondary effluent in this study, with the characteristic of low biodegradability, thus it might be difficult for denitrification bacteria to use. Therefore, whether part of the COD could be used as a denitrification carbon source should be discussed, and the optimal denitrification conditions explored.

3.2.1 Carbon source selection in denitrification process. The carbon source is often supplemented in the denitrification process, however cost will decline if the COD contained in the influent could be used as the denitrification carbon source. Thus, the anaerobic–oxic (AO) process and internal recycle of the nitrification liquid from the nitrification tank to denitrification tank are often adopted. Herein, the nitrification liquid obtained from the biological ferric carbon system is used for denitrification. In order to avoid the probable impact of Fe³⁺ on the denitrification performance, the nitrification liquid was coagulated by adding Ca(OH)₂ to adjust the pH to 8 and implementing 1 mg L⁻¹ of PAM to remove Fe³⁺. The coagulated liquid mixed with influent was refluxed for the denitrification step, and the reflux ratio (nitrification liquid/influent) was investigated. Since the empirical mass ratio of COD to NO₃⁻-N (C/N) for denitrification was 5, the reflux ratios were set as 100% and 200%, thus making the C/N around 6.5 ± 0.3 and 4.4 ± 0.2, respectively. Considering the organic pollutants contained in the nitrification liquid and that the influent might be difficult to use by denitrification microorganisms, an experiment of adding methanol to the nitrification liquid was conducted, and 133 mg L⁻¹ methanol (200 mg L⁻¹ COD) was supplemented. Furthermore, this nitrification liquid was directly used in the denitrification process and compared with the nitrification liquid without methanol. During the denitrification period, the MLSS was maintained at about 2500 ± 50 mg L⁻¹, and the reaction time was set as 6 h. The denitrification performance details under various types of carbon sources are summarized in Table 2.

Table 2 Denitrification performance under various types of carbon sources^a

		Denitrification influent			Denitrification effluent		Denitrification efficiency (%)
		COD (mg L^−1)	NO3^−-N (mg L^−1)	C/N	COD (mg L^−1)	NO3^−-N (mg L^−1)
a All experiments were conducted triplicate.
Reflux ratio	100%	91.6 ± 4.2	14 ± 0.5	6.5 ± 0.3	84 ± 3.8	13.1 ± 0.5	6.4 ± 0.3
	200%	82.1 ± 3.2	18.7 ± 0.6	4.4 ± 0.2	70 ± 3.3	16.8 ± 0.7	10.0 ± 0.4
Nitrification liquid + methanol		263.2 ± 6.5	28 ± 1.5	9.4 ± 0.4	160.0 ± 6.0	15.5 ± 0.7	44.6 ± 1.5
Nitrification liquid		63.2 ± 3.0	28 ± 1.5	2.3 ± 0.1	60.0 ± 2.6	27.3 ± 1.1	2.5 ± 0.1

---

Obviously, the carbon source contained in the nitrification liquid was scarcely used in the denitrification process. Only a small fraction of carbon source in the refluxing liquid was used for denitrification, and the denitrification efficiency was less than 10%. However, when methanol was supplemented in the nitrification liquid, denitrification increased significantly to 44.6 ± 1.5%, which indicates that an external carbon source is necessary, and it is difficult for the carbon source contained in the nitrification liquid and influent to be used. It should be noted that the external supplemented methanol could not be totally consumed and there was still 15.5 ± 0.7 mg L⁻¹ NO₃⁻-N left in the effluent. Therefore, enhancing the denitrification efficiency should be further considered.

3.2.2 Effect of Fe³⁺ on denitrification efficiency. It was reported that the ferric ion is beneficial for microbial growth and could promote the denitrification efficiency.¹⁵ In the biological ferric carbon nitrification process, ferric ions are released from the iron shavings, and Fe³⁺ is the main existing form under the neutral aeration condition. Thus, impact of the presence of Fe³⁺ in the nitrification liquid on the denitrification process was investigated from the aspects of denitrification efficiency, denitrification time and carbon source utilization rate.

Based on the above experiment, 200 mg COD per L of methanol supplemented in the coagulated nitrification liquid could result in a 44.6 ± 1.5% denitrification efficiency at the reaction time of 6 h. Then, the denitrification experiment using uncoagulated nitrification liquid, which contained 9.8 ± 0.6 mg L⁻¹ dissolved Fe³⁺, was compared under the same methanol dosage and same operation conditions. It can be seen from Fig. 5(A) that when the nitrification liquid contained Fe³⁺, the denitrification efficiency achieved was nearly 90.2 ± 2.1% in the short period of 2 h, which implies that Fe³⁺ could significantly promote the denitrification process.

Fig. 5 Effect of Fe³⁺ on the denitrification efficiency, denitrification time and carbon source utilization rate in the denitrification process.

It is well known that a certain amount of COD is required when 1 mg of NO₃⁻-N is denitrified to N₂, specifically, the amount of COD consumed and the amount of NO₃⁻-N being denitrified has a linear relationship. In addition, it was reported that different amounts of COD are consumed under different conditions or different wastewater quality.²⁵ Herein, the effect of Fe³⁺ on this linear relationship is illustrated in Fig. 5(B). It could be seen that when the nitrification liquid was coagulated, 8.19 g of COD was needed to denitrify 1 g of NO₃⁻-N, whereas only 6.80 g of COD was required when Fe³⁺ existed in the nitrification liquid. Therefore, Fe³⁺ could increase the carbon source utilization rate in the denitrification process, which implies that a smaller amount of external carbon source is needed. Then, the optimal dosage of the methanol supplement in the presence of Fe³⁺ was further investigated.

3.2.3 Optimal dosage of methanol supplement. It was observed that when 200 mg COD per L of methanol was added to the nitrification liquid, there was still some methanol left in the denitrification effluent. Therefore, different amounts of methanol (150, 175, and 200 mg COD per L) were added in the uncoagulated nitrification liquid, making the denitrification initial COD 213.2 ± 5.7, 238.2 ± 6.6 and 263.2 ± 7.8 mg L⁻¹, respectively, and the optimal dosage was selected. Since the concentration of NO₂⁻-N in the whole denitrification process was low, it was not considered here.

As shown in Fig. 6, when the supplemented external COD was 150 and 175 mg L⁻¹, the effluent NO₃⁻-N was 4.1 ± 0.2 and 1.1 ± 0.1 mg L⁻¹, respectively, which correspond to the denitrification efficiency of 85.4 ± 3.1% and 96.1 ± 2.9%. With regard to COD consumption, the methanol could not be consumed completely when 200 mg COD per L of methanol was added, however, when the amount of methanol was decreased to 150 and 175 mg COD per L, the denitrification effluent COD was 51.5 ± 2.2 and 49.4 ± 2.1 mg L⁻¹, respectively, which were lower than the original COD in the nitrification liquid (63.2 ± 1.5 mg L⁻¹). It should be noted that when there was no external carbon source added, the organic matter in the nitrification liquid was refractory, and could hardly be used in the denitrification process. Whereas, some refractory COD was utilized when there was external methanol added, which implies that co-metabolism between refractory organic pollutants and methanol might occur in denitrification microorganisms, thus resulting in a further decrease in the COD in the effluent. Thus, when comprehensively considering the COD and denitrification efficiencies, 175 mg COD per L of added methanol was selected.

Fig. 6 COD consumption and NO₃⁻-N removal efficiency in the denitrification process under different dosages of methanol supplement. Error bars represent the standard deviations of triplicate tests.

In conclusion, Fe³⁺ in the nitrification liquid is beneficial for the denitrification process, and the extra carbon source of 175 mg COD per L methanol is required. Under the above conditions, NO₃⁻-N could be denitrified to low concentrations during a 2 hour period.

3.3 Advanced nitrogen removal performance

Under the optimal nitrification and denitrification conditions, as discussed above, the performance of COD removal and NH₄⁺-N, NO₃⁻-N, and TN transformations were measured during the whole process, as illustrated in Fig. 7. It is found that the biological ferric carbon system could achieve a good advanced nitrogen removal performance, in which the NH₄⁺-N concentration decreased from 32.3 ± 1.5 mg L⁻¹ in the influent to 2.5 ± 0.2 mg L⁻¹ in the effluent. In addition, the system also functioned to degrade pollutants, which resulted in a decline in the COD from 120.0 ± 6.0 mg L⁻¹ to 49.4 ± 2.1 mg L⁻¹. Under the same conditions, the nitrification–denitrification system was long-term operated for about 45 days, and the good advanced nitrogen removal performance was maintained. Therefore, the biological–ferric–carbon system is adaptable for advanced nitrogen removal.

Fig. 7 Performance of advanced nitrogen and COD removal during the whole nitrification and denitrification process. Error bars represent the standard deviations of triplicate tests.

4 Conclusion

In summary, the nitrification process could be initiated and a high NH₄⁺-N removal efficiency achieved both under the condition of sludge seeding or not in the biological ferric carbon system, in which ferric, carbon and sludge microbes coexist. The bacteria community composition and diversity in these two different conditions are similar although their bacteria origins are varied, however the sludge seeding system reached a stable high nitrification efficiency in a shorter period. An extra carbon source is required, and the released Fe³⁺ in the nitrification liquid from the biological ferric carbon system is beneficial for the subsequent denitrification process, which shows a higher denitrification efficiency, shorter reaction time and higher carbon source utilization rate. Under the optimal nitrification and denitrification conditions, the NH₄⁺-N concentration decreased from 32.3 ± 1.5 mg L⁻¹ in the influent to 2.5 ± 0.2 mg L⁻¹ in the effluent, and the system also functioned to degrade pollutants, which resulted in a decline in the COD from 120.0 ± 6.0 mg L⁻¹ to 49.4 ± 2.1 mg L⁻¹. Therefore, the biological ferric carbon nitrification and denitrification process shows a good performance in advanced nitrogen removal.

Acknowledgements

This work was financially supported by National Natural Science Foundation of China (Grant no. 51508081), China Postdoctoral Science Foundation (2015M581499), Special Financial Grant of China Postdoctoral Science Foundation (2016T90322) and the Fundamental Research Funds for the Central Universities (2232015D3-23).

References

1. E. Erdem, G. Çölgeçen and R. Donat, J. Colloid Interface Sci., 2005, 282, 314–319.
2. Department of Pollution emission control, State environmental statistic report, 2014, http://zls.mep.gov.cn/hjtj/nb/2013tjnb/.
3. J. Wang, M. C. Long, Z. J. Zhang, L. N. Chi, X. L. Qiao, H. X. Zhu and Z. F. Zhang, Chemosphere, 2008, 71, 195–202.
4. V. Murali, S.-A. Ong, L.-N. Ho and Y.-S. Wong, Bioresour. Technol., 2013, 143, 104–111.
5. N. Liu, X. Xie, H. Jiang, F. Yang, C. Yu and J. Liu, RSC Adv., 2016, 6, 25624–25632.
6. P. Jin, X. Jin, X. C. Wang and X. Shi, Chemosphere, 2013, 91, 558–562.
7. H. Yoo, K.-H. Ahn, H.-J. Lee, K.-H. Lee, Y.-J. Kwak and K.-G. Song, Water Res., 1999, 33, 145–154.
8. B. Ma, S. Wang, S. Cao, Y. Miao, F. Jia, R. Du and Y. Peng, Bioresour. Technol., 2016, 200, 981–990.
9. F. Veuillet, S. Lacroix, A. Bausseron, E. Gonidec, J. Ochoa, M. Christensson and R. Lemaire, Water Sci. Technol., 2014, 69, 915–922.
10. U. van Dongen, M. S. Jetten and M. C. van Loosdrecht, Water science and technology: a journal of the International Association on Water Pollution Research, 2001, vol. 44, pp. 153–160.
11. K. Wang, S. Wang, R. Zhu, L. Miao and Y. Peng, Bioresour. Technol., 2013, 134, 212–218.
12. M. Ali and S. Okabe, Chemosphere, 2015, 141, 144–153.
13. Y.-C. Yu, Y. Tao and D.-W. Gao, RSC Adv., 2014, 4, 54798–54804.
14. W. Sui, Master thesis, Qingdao Technology University, 2011.
15. X. Zhou, Master thesis, Qingdao Technology University, 2014.
16. A. Andersson, P. Laurent, A. Kihn, M. Prévost and P. Servais, Water Res., 2001, 35, 2923–2934.
17. V. Abromaitis, V. Racys, P. van der Marel and R. J. W. Meulepas, Chemosphere, 2016, 149, 183–189.
18. X.-H. Zhi, Doctor of Philosophy thesis, Tongji University, 2006.
19. C. Zheng, H. Chen, G. Xue, X. Li and Y. Liu, Chin. J. Environ. Eng., 2016, 10, 2478–2484.
20. X. Zheng, Y. Su, X. Li, N. Xiao, D. Wang and Y. Chen, Environ. Sci. Technol., 2013, 47, 4262–4268.
21. M. L. Sogin, H. G. Morrison, J. A. Huber, D. Mark Welch, S. M. Huse, P. R. Neal, J. M. Arrieta and G. J. Herndl, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 12115–12120.
22. APHA, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, Washington, DC., 20th edn, 1998.
23. Y.-S. Jeon, S.-C. Park, J. Lim, J. Chun and B.-S. Kim, J. Microbiol., 2015, 53, 60–69.
24. D. Y. Sorokin, S. Lucker, D. Vejmelkova, N. A. Kostrikina, R. Kleerebezem, W. I. C. Rijpstra, J. S. S. Damste, D. Le Paslier, G. Muyzer, M. Wagner, M. C. M. van Loosdrecht and H. Daims, ISME J., 2012, 6, 2245–2256.
25. J.-J. Her and J.-S. Huang, Bioresour. Technol., 1995, 54, 45–51.

---

Footnote

† Electronic supplementary information (ESI) available: A sketch map of the reactor used in this study, bacterial Shannon diversity curves in the biological ferric–carbon systems, Venn diagram of the bacterial communities from different reactors, and metal element composition and content of the ferric scrap used in this study. See DOI: 10.1039/c6ra15130b

---