Characterization of soluble microbial products in a partial 1 nitrification sequencing batch biofilm reactor treating high 2 ammonia nitrogen wastewater

11 In present study, the characterization of soluble microbial products (SMP) was 12 evaluated in a partial nitrification sequencing batch biofilm reactor (SBBR). During 13 the stable operation of SBBR, the NH 4+ -N removal efficiency and nitrite 14 accumulation ratio were 96.70 ±0.41% and 93.77± 1.04%, respectively. According to 15 excitation-emission matrix (EEM), the intensities of protein-like substances were 16 reduced under anoxic and aerobic phases, whereas humic-like substances had little 17 change during the whole cycle. Parallel factor analysis (PARAFAC) further 18 indentified two components and their fluorescence intensity scores were both reduced. 19 Synchronous fluorescence spectra revealed that the fluorescence intensity of 20 protein-like fraction decreased with reaction time. Two-dimensional correlation 21 spectroscopy (2D-COS) further demonstrated that protein-like fraction might decrease 22 earlier than the other fractions. The information obtained in present study is of


Introduction
Biological nitrogen removal is commonly applied for the treatment of both domestic and industrial wastewater (Ma et al., 2016).In traditional N-removal process, ammonia is oxidized to nitrate in aerobic nitrification process and nitrate is reduced to molecular nitrogen in anoxic denitrification process (Zhou et al., 2017).As a cost-effective N-removal technology, partial nitrification has been widely developed in recent years based on the fact that nitrite is an intermediate compound in both nitrification and denitrification steps (Ciudad et al., 2005).It is generally accepted that partial nitrification via nitrite could save approximately 25% oxygen in nitrification stage and 40% carbon source in denitrification stage (Guo et al., 2009).The key control strategy to achieve partial nitrification is the enrichment of ammonia oxidizing bacteria (AOB) and limitation-inhibition-washout of nitrite oxidizing bacteria (NOB).Till now, it has been successfully achieved by appropriate regulating operational parameters, such as dissolved oxygen (DO), sludge retention time, pH value, temperature and free ammonia (FA) etc. (Ciudad et al., 2007;Peng and Zhu, 2006).
Soluble microbial products (SMP) are defined as a pool of organic compounds that are derived from the substrate metabolism (usually with biomass growth) and biomass decay during complete mineralization of supplying nutrients (Jarusutthirak and Amy, 2007).Indeed, SMP are one kind of heterogeneous mixtures containing various complex organic materials, such as proteins, polysaccharides, humic acids, fulvic acids and organic acids etc. SMP have been drawn intensive attention in the field of wastewater treatment process due to its adverse impact on the effluent quality and treatment efficiency in wastewater treatment plant (WWTP).Many published literatures have confirmed that SMP constitute a major part of organic component in effluent from biological wastewater treatment (Wu et al., 2016;Xie et al., 2012).
Additionally, some SMP may cause further environmental hazard to the receiving water after the sewage treatment systems, such as toxicity and metal chelating properties (Kunacheva et al., 2017;Liang et al., 2007).The presence of SMP may also affect the viscosity, flocculating and other physical characters of sludge (Kim et al., 2016;Zhou et al., 2009).Moreover, SMP have a certain effect on the composition changes of microbial community in bioreactor (Kunacheva and Stuckey, 2014).Hence, it is essential to clearly identify the key components of SMP for better understanding the fundamental mechanisms of biological activities.However, till now, most researches are related to the full nitrification process in activated sludge reactor, and little information is available on the production of SMP in a partial nitrification system.
To date, a series of advanced analytical methods have been applied to explore the specific components in SMP produced in biological treatment process (Guo et al., 2013).Three-dimensional excitation-emission matrix (3D-EEM) has been widely utilized for determining SMP owing to its better selectivity, higher sensitivity, and more simple and convenient operation.Combined with 3D-EEM, parallel factor analysis (PARAFAC) is commonly used to interpret the fluorescence spectra by decomposing the complete map into independent components (Zhang et al., 2016).
Although PARAFAC is always applied to explore the further information of fluorescence spectra, it is not able to identify the mutual relationships between different components (Xu and Jiang, 2013).Two-dimensional correlation spectroscopy (2D-COS) could be applied as a versatile tool to express the specific variation order of any slight changes.It has a great advantage in solving the problem of overlapping peaks occurred in the original spectra by extending the overlapped bands in second dimension (Noda, 2006).Therefore, it is of great significant to provide a comprehensive analytical method for explaining the SMP formation in partial nitrification system.However, there is little information available regarding to this point in previous literature.
Based on the above discussion, the objective of present study was to evaluate SMP production in a stable partial nitrification sequencing batch biofilm reactor (SBBR) treating high ammonia nitrogen wastewater.A spectroscopic analysis based on the combination of 3D-EEM, PARAFAC, synchronous fluorescence and 2D-COS was used to characterize SMP samples in various reaction times.The results could provide insightful information on understanding the formation of SMP in a partial nitrification system.

Experimental set-up
The experiment was carried out in a cylindrical SBBR with a working volume of 3.4 L. The height and inner diameter were 30 and 12 cm, respectively.Cylindrical carriers (K3, plastic media) were applied as biomass support with a packing rate of 40% (v/v).The diameter and height of each carrier were 25 and 15 mm, respectively.
The specific gravity and the specific surface area of each carrier were 110 kg/m 3 and 500 m 2 /m 3 , respectively.
The SBBR was operated sequentially in 8h for each cycle, consisting of 5 min for influent filling, 85 min for anoxic stage, 360 min for aeration, 15 min for settling and 15 min for effluent and idle.Aeration was provided by using an air pump and controlled through a gas flow meter.The exchange volume of SBBR was 50% for each cycle.Influent wastewater was prepared in a water tank and pumped into the bottom of biofilm system.

Synthetic wastewater and seed sludge
The influent high-strength nitrogen wastewater was shown as follows: COD (as  (Tay et al., 2002).The influent pH value was adjusted to 8.0 by using NaHCO 3 and HCl.
Seed sludge for SBBR was collected from a lab-scale SBR of 17 L and mixed liquid suspended solids (MLSS) was controlled at about 3.0 g/L.In present study, high pH value (>8.0) and ammonia (200 mg/L) was controlled in the influent wastewater, which may provide a feasible influent FA inhibition (high of 70 mg/L) on the activity of NOB, as similarly reported in our previous literature (Wei et al., 2017).
After approximately 60 days operation, partial nitrification biofilm was successfully achieved, and the biomass concentration increased to 6.0 g/L.

Fluorescence analysis
SMP samples were obtained from partial nitrification system at various reaction time from 0 to 420 min.Each sample was centrifuged at 8000 rpm for 5 min to separate out the solids.The supernatant was regarded as SMP.Fluorescence spectra were measured by using a Luminescence spectrometer (LS-55, Perkin-Elmer Co., USA).3D-EEM was obtained by subsequently scanning emission from 220 to 400 nm at 10 nm increments by varying the excitation wavelength from 280 to 550 nm at 0.5 nm increments.PARAFAC analysis was obtained by using MATLAB 7.6 (Mathworks, Natick, MA, USA) with the N-way toolbox for the further analysis of EEM data (Yu et al., 2010).Synchronous fluorescence spectra were determined by ranging the excitation wavelengths from 250 to 550 nm with a constant offset ( ) of 60 nm.All samples were scanned at a speed of 1200 nm/min.2D-COS was applied to synchronous fluorescence spectra, and detailed mathematical procedure could be found elsewhere (Noda, 2006).It was found from PARAFAC model that the fluorescent SMP samples from partial nitrification system could be indentified into two components (Fig. 3).Two peaks were exhibited in component 1 (Ex/Em of 280/342.5 nm and Ex/Em of 230/342.5 nm), which were corresponded to tryptophan protein-like substances and aromatic protein-like substances, respectively (Phong and Hur, 2015).Component 2 was characterized three main peaks that represented the presence of humic-like substances (Ex/Em of 350/427 nm and 250/433 nm) and fulvic-like substances (Ex/Em of 220/433 nm) (Wu et al., 2011).

3D-EEM spectra
PARAFAC also provides additional information to quantitatively describe the variations of two components, as displayed in Fig. 4. It was found that the fluorescence intensity scores of the two components decreased from 0.67 and 0.66 to 0.06 and 0.17 during the typical cycle, respectively.In addition, the decreased degree of component 1 was greater than that of component 2, indicating that protein-like substances changed to a much higher extent than those of humic-like substances and fulvic-like substances.PARAFAC coupled with 3D-EEM has been widely used to characterize SMP or DOM in wastewater treatment process.Sanchez et al. (2013) found that the loadings of PARAFAC sample exhibited a higher association with the total EEM signal in the raw and treated water samples compared with alternative analysis techniques.The study reported by Ou et al. (2014) indicated that EEM-PARAFAC not only can rapidly identify fluorescent matter characteristics, but also can indirectly reflect the variations of representative contaminants.The analysis results showed that EEM-PARAFAC could be used as a promising monitoring tool for tracking the trends of fluorescence components in a partial nitrification process.

Synchronous fluorescence spectra
Synchronous fluorescence spectra have the advantages of higher selectivity and sensitivity in the assessment of multi-component materials.As shown in Fig. 5, three fluorescence regions (protein-like, fulvic-like and humic-like fractions) could be assigned to the wavelength ranges of 250-300, 300-380 and 380-550 nm, respectively (Chen et al., 2015).The data showed that the fluorescence intensity of SMP samples gradually decreased in the whole range of wavelengths with the reaction time from 0 to 420 min.In addition, a major peak corresponding to protein-like fluorescence fraction was identified at the wavelength of 260 nm.The observation of protein-like fluorescence fraction as the main component in SMP was consistent with the evaluation of 3D-EEM.

2D-COS
In order to resolve the overlapping peaks problem, 2D-COS was conducted to enhance the spectral resolution by distributing spectral intensity trends over a second dimension.In synchronous 2D-COS map (Fig. 6A), one positive auto-peak along the diagonal line was identified at 282 nm, suggesting that protein-like fraction was more susceptible than other fractions.All peaks identified from synchronous 2D-COS map are positive, indicating that the decrease of fluorescence intensity proceed in the same direction, as similarly with the observation reported by Hur and Lee (2011).
Asynchronous 2D-COS map was used to determine the response order of different molecules or groups along a given external perturbation.As shown in Fig. 6B, two main positive cross-peaks at 291/349.5, 291/366 nm and two negative cross-peaks at 264/291, 282.5/291 nm were observed upper the diagonal line of asynchronous map.According to Noda's rule (2005), the variation of fluorescence to reaction time took place sequentially in the order: 264 and 282.5>291>349.5 and 366 nm.The results demonstrated that protein-like fraction might occur earlier than the other fractions.

Conclusions
In summary, a partial nitrification lab-scale SBBR was achieved by controlling

+
were determined according to their respective standard methods (APHA, 2005).Total nitrogen (TN) was based on the sum of NH 4 than an independent TN test.The DO and pH values were measured by using on-line probes (3420i, WTW Company, Germany).Nitrite accumulation ratio (NAR) of partial nitrification system was calculated according to the following equation (1) reported by Wei et al. (2015): the COD measurement was based on oxidation by potassium dichromate, it was influenced by nitrite (oxidized into nitrate).Thus COD values were corrected in order to estimate the oxygen demand only due to the organic substances.=COD 0.5(NO -N)M /M − (2)

Fig. 1
Fig.1 shows the variations of nitrogen compound, DO and pH values in partial

Fig. 2
Fig.2shows the 3D-EEM spectra of SMP samples corresponding to reaction the influent high pH value and NH 4 + -N.It was found that the reactor expressed better organic matter and nitrogen removal efficiencies during stable operation.A combined use of EEM, PARAFAC, synchronous fluorescence and 2D-COS was conducted to characterize SMP samples in the typical cycle of partial nitrification system.EEM-PARAFAC indentified three fluorescence peaks and two components in SMP samples.Synchronous fluorescence spectra and 2D-COS analysis demonstrated that protein-like fraction was the dominant components in SMP and took place earlier than the other fractions with reaction time.Shandong Province (2014ZZCX05101), the Science and Technology Development Plan Project of Shandong Province (2014GGH217006), and QW thanks the Special Foundation for Taishan Scholar Professorship of Shandong Province and UJN (No. ts20130937).

Fig. 4
Fig. 4 Fluorescence intensity scores of two PARAFAC-derived components in SMP samples.

Fig. 5
Fig. 5 Changes in synchronous fluorescence spectra of SMP corresponding to reaction time.
Table 1 summarizes the contaminant removal performance of the SBBR under stable operation.The influent and effluent COD were 594.2±2.26 and 43.3±5.22mg/L, --N concentrations were 80.97 ± 2.53 and 5.38 ± 0.94 mg/L, respectively, resulting in a high NAR of 93.77± 1.04%.TN removal efficiency of partial nitrification SBBR was 52.67 ± 1.75%.Data implied that a better nitrogen removal performance in partial nitrification SBBR was obtained under stable operation.

Table . 1
The contaminant removal performance of SBBR under stable operation.