Introduction

Reducing ammonium nitrogen concentration in wastewater effluent is one of the major goals to prevent aquatic eutrophication. In general, autotrophic nitrogen removal by combined partial nitritation (PN) and anaerobic ammonium oxidation (anammox) has attracted considerable attention in recent years as an alternative to convention nitrogen removal processes, because the PN-anammox is a cost-effective and energy-saving potential process. Nitrogen removal with PN-anammox is possible for low C/N ratio wastewater treatments, and approximately 30–40% of the overall nitrogen removal cost will be saved compared to nitrification or denitrification and does not require organic carbon source (Desloover et al. 2011). In the PN-anammox process, half of the ammonium (NH4 +-N) is oxidized with oxygen to nitrite (NO2 −-N) by PN, which provides NO2 −-N for anammox. Subsequently, NH4 +-N and NO2 −-N are directly converted to nitrogen gas (N2) by anammox (Huang et al. 2016; Lotti et al. 2014).

Phenol and thiocyanate (SCN−) are toxic compounds to most living things, and landfill leachates as well as several industries such as coke plants and refineries produce complex wastewater containing NH4 +-N, phenol, and SCN− (Milia et al. 2012; Morita et al. 2007). The phenol concentration ranges between 0.35 and 1040 mg L−1 in landfill leachate and coke wastewater (Aziz et al. 2010; Toh and Ashbolt 2002), whereas the SCN− concentration has been reported as high as 535 mg L−1 (Toh and Ashbolt 2002). Satisfactory treatment of leachate is not trivial due to its high concentrations of ammonium nitrogen and toxic compounds. Thus, it is vital to study phenol/SCN−-mediated inhibition of the PN-anammox process to guide the research and application of this method to treat both nitrogen- and phenol/SCN−-containing wastewater.

Several studies have reported phenol inhibition on anammox activity or biomass (Hou et al. 2014; Jin et al. 2013b; Pereira et al. 2014; Toh and Ashbolt 2002; Yang et al. 2013; Yang and Jin 2012), and the inhibition of SCN− on anammox has also been assessed by Chen et al. (2017a, b) and Toh and Ashbolt (2002). However, systems based on granular biomass tend to perform better in the presence of inhibitory or toxic compounds because the granule architecture creates diffusion gradients that help protect sensitive bacteria (Maszenan et al. 2011). In this sense, some studies have suggested the possibility that anammox granules acclimate in phenol-containing wastewater (Toh and Ashbolt 2002). Moreover, as the first step of the PN-anammox process, PN is the fundamental basis for a successful PN-anammox operation (Fudala-Ksiazek et al. 2014). PN sludge mainly contains two different autotrophic groups of bacteria, ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB). It is well known that AOB and NOB are sensitive to environmental factors such as pH, temperature, heavy metals, and even organic solvents (Gilbert et al. 2014). However, little research has examined the effect of toxic substances (phenol and SCN−) present in the landfill leachate or coke wastewater on the PN process or their combined toxicity. Mixture exposure is a universal phenomenon; however, assessing the potential toxic effects and quantifying the exposure risk to chemical mixtures in landfill leachate and coke wastewater still remains a major challenge. Consequently, the joint, cumulative effects of toxic mixtures might cause microbial risk. Models for predicting the effect of mixtures are urgently needed.

A certain composite design (CCD) coupled with response surface methodology (RSM) is an efficient design tool to determine the optimal conditions for a desirable response. These methods were successfully applied for wastewater treatment to optimize hydrogen production and identify the inhibitory effects of toxicants (Rastegar et al. 2011; Xing et al. 2014).

Therefore, this study aims to (1) quantify the IC50 of phenol and SCN− on PN sludge with short-term effects, (2) investigate the inhibition of phenol and SCN− on AOB and NOB, and (3) evaluate the interactive effects of phenol and SCN− on PN sludge activity using RSM with CCD.

Materials and methods

Activated sludge and synthetic medium

The activated sludge used in this study was collected from a 6.8-L internal-loop airlift reactor operated in continuous-flow mode at 30 ± 1 °C with a hydraulic retention time of 26 h. The DO concentration was maintained at 0.8–1.3 mg L−1, and the level of free ammonia in the system was between 48.4 and 95 mg L−1, which is favorable for partial nitrification. The synthetic wastewater and the element solution were made according to a modified formula described by Xing et al. (2013); the feed medium was devoid of organic carbon and contained 900–950 mg NH4 +-N L−1. The system ran with an NH4 +-N removal efficiency of approximately 43–59% with the molar ratio of NO2 −-N/NH4 +-N in the effluent from 0.8 to 1.2, and nitrite accumulation ([NO2 −-N]/[NO x −-N] × 100%) ranged from 44.1 to 54.5%.

PN activity with batch tests

Batch assays were performed in triplicate, and the activity of PN sludge was measured using the respiration rate. The respirometric tests were used because the respiration rate of activated sludge is reduced in the presence of toxic substances and they provided an experimental assessment of the oxygen uptake rate (OUR). A Strathtox respirometer (Strathkelvin Strathtox, Scotland, UK) was used to measure the toxicity of trade effluents entering the wastewater treatment plant. This equipment was based on the respirometry applications in the biomedical field and used six oxygen electrodes simultaneously. The use of six oxygen electrodes allowed the respiration rate of a control sample of the same sludge mixed with five other different concentrations of toxic compounds. Tests were carried out in six 20-mL glass tubes. The tubes were kept stirred with a magnetic stir bar in a water bath of Strathtox unit. Toxic compounds at different concentrations were added to five testing glass tubes each with a total volume of 18 mL. After reaching the constant temperature of 25 °C, 2 mL of activated sludge was quickly added to each tube and oxygen electrodes were inserted into the tubes for recording the respiration rate values. The test was stopped as soon as the dissolved oxygen content of the mixed liquor in the tube with the fastest respiration rate had fallen to near zero. The concentration of each component is listed in Table 1.

Table 1 The compositions of the batch tests for respirometric calculations
Full size table

The specific respiration rate (SRR) was calculated according to Eq. (1) and was used to express PN sludge activity.

$$ SRR=\frac{OUR}{VSS}\times {10}^{-3} $$
(1)

in which OUR is the oxygen uptake rate (mg L−1 h−1) and VSS is the volatile suspended solids (g L−1).

The relative specific respiration rate (RSRR), which was calculated according to Eq. (2), was used to express the relative PN sludge activity with toxic compounds at different concentrations compared to the control.

$$ RSRR=\frac{SRR_e}{SRR_c}\times 100\% $$
(2)

in which SRR e represents the SRR of the experimental groups (with inhibitors or toxic compounds added during the experiment), and SRR c represents the SRR of the control group.

The respiration inhibition ratio (RIR) was calculated as follows:

$$ RIR=100\%- RSRR $$
(3)

Inhibition of AOB and NOB

Selective inhibitors of AOB and NOB were used to test the variable activity responses of AOB and NOB to phenol and SCN−. Allylthiourea (ATU) was selected to inhibit AOB (Wang and Gu 2014), and NaClO3 was chosen as an NOB inhibitor in this study.

First, the original specific oxygen uptake rate (SOUR) was tested and recorded as α. After 3 min, as the DO level decreased, NaClO3 was added at a concentration of approximately 2.13 g L−1, and the SOUR was measured and recorded as β. Three minutes later, ATU was added to each glass tube at a final concentration of approximately 5 mg L−1 to suppress AOB metabolism, and γ represented the SOUR at this point. As described above, the AOB activity was defined as (β − γ), and the NOB activity was (α − β).

The interactive effects of phenol and SCN−

A CCD was used to study the interactive effects of phenol and SCN− on the PN activity response. Design Expert software version 8.0.6.1 (STAT-EASE Inc., Minneapolis, USA) was used to design the CCD experiment, to perform the regression analysis of the experimental data, and to plot the response surface and contour plots. In the experiment, both variables were assessed at five different coded levels, marked −α, −1, 0, +1, and +α. The ranges and levels of the independent input variables for the phenol and SCN− concentrations were set as listed in Table 2 based on the results from PN activity with batch tests. The RSRR was selected as the dependent output variable. The CCD applied in this study is presented in Table 3, which provides the experimental conditions and their responses. As the response of the PN sludge activity, the RSRR (Y) was modeled as a second-order polynomial equation, showing the mathematical relationship between the independent variables phenol and SCN− and Y, as Eq. (4) presents.

Table 2 Experimental ranges and levels of the independent test variables
Full size table
Table 3 Experimental plan showing the different experimental sets for the CCD
Full size table
$$ Y={\beta}_0+\sum_{i=1}^n{\beta}_i{X}_i+\sum_{i=1}^n{\beta}_{i i}{X}_i^2+\sum_{i< j}^n\sum_{j=1}^n{\beta}_{i j}{X}_i{X}_j $$
(4)

in which Y is the predicted response, n is the number of factor variables, X i and X j are the coded levels of the independent variables, β 0 is the offset term, β i is the ith linear coefficient, β ii is the iith quadratic coefficient, and β ij is the ijth interaction coefficient.

The model obtained from regression analysis was used to generate the response surface and contour plots. The coefficient of determination R 2 was used to express the quality of the polynomial model equation fit, and the significance was analyzed using an F test in the same program. A t test was used to test the significance of the regression coefficient (Xing et al. 2014). An analysis of variance (ANOVA) was used to evaluate the interaction between the process variables and the response, and p <0.05 was considered statistically significant.

The experiments were designed (Table 4) to validate the regression model equation obtained from the CCD experiment. The RSRR tests at phenol concentrations of 1.5, 3.0, and 4.5 mg L−1 with varying SCN− concentrations were used to evaluate the effect of SCN− on PN sludge activity. The validation experiments were performed in triplicate.

Table 4 Experiments to validate the regression model equation
Full size table

Assessment of the combined toxicity of phenol and SCN−

Independent, additive, synergistic, and antagonistic effects describe the four main types of combined toxicities (Ding et al. 2015). Isobole plots (IP) are considered a simple and straightforward method for characterizing the combined toxicities of bicomponent mixtures (Ding et al. 2015).

Analytical methods and mathematical model

VSS levels were determined using standard methods (APHA, AWWA, AEF, 2012). The statistical tests were performed with the F test for ANOVA.

The modified non-competitive inhibition model (Eq. (5) was used to represent the inhibitory characteristics of the toxic compounds in the activated sludge (Jin et al. 2013a), and the model was fit to the experimental data using the minimum squared errors method.

$$ RIR=\left(1-\frac{1}{1+{\left(\mathrm{s}/{k}_a\right)}^b}\right)\times 100\% $$
(5)

in which RIR is the inhibition response, s is the concentration of phenol or SCN− in the experiment, k a is the IC50 value of phenol or SCN− on PN sludge, and b is the fitting parameter.

Results and discussion

Individual toxicity

The acute toxicities of phenol and SCN− individually under fixed initial substrate level conditions were obtained.

The modified non-competitive inhibition model described the acute-term effects of phenol on PN sludge activity expressed with RIR. Accordingly, the RSRR decreased as the phenol concentrations increased, and the IC50 of phenol on PN sludge was 5.60 mg L−1 (RIR = 100 %  × (1 − 1/(1 + (s/5.60)1.91))(R 2 = 0.9772)). In a study by Liu et al. (2005), nitrifying sludge tolerated the toxicity of an initial phenol concentration below 10 mg L−1, although a partial irreversible disruption of the nitrification was observed when the exposure to phenol was above 15 mg L−1. However, large differences were observed in a pure culture experiment to simultaneously remove toxic pollutants. In a pure culture study by Lauchnor et al. (2011), the ammonia oxidation rates were inhibited by 50% when exposed to 5.27 mg L−1 phenol, whereas in a study using adapted activated sludge to treat coke wastewater, significant nitrification inhibition was observed at a phenol concentration of 200 mg L−1 (Kim et al. 2008). The tremendous differences between the IC50 levels might be caused by variations in the microorganism species, the sludge structure, or the operating conditions.

The toxic effect of SCN− on PN sludge was initially apparent as the concentrations increased. During the acute exposure batch tests, the IC50 of SCN− on PN sludge was 351 mg L−1 (RIR = 100 %  × (1 − 1/(1 + (s/351)1.05))(R 2 = 0.8792)). An inhibitory effect on nitrification was observed when the concentration of SCN− was above 200 mg L−1 as well as increased ammonia loading (Kim et al. 2008). However, in this study, 200 mg L−1 SCN− inhibited specific respiration by approximately 44.0%.

The sensitivity of PN sludge activity was characterized by the IC10 (concentration inhibiting the activity by 10%). Concentrations below this were regarded as relatively safe, and the IC10 values were obtained using the modified non-competitive inhibition model mentioned above. Accordingly, the IC10 of phenol and SCN− on PN sludge were 1.77 and 43.3 mg L−1, respectively, suggesting the peak concentrations of PN sludge were not inhibited when treating wastewater containing phenol and SCN−. Moreover, considering the practical concentration of phenol and SCN− in industrial effluents such as coke wastewater and landfill leachate (Table 5), the concentration of phenol in coke wastewater is generally above 15 mg L−1 and ranges between 0.35 and 33 mg L−1 in landfill leachate. Thus, serious inhibition of PN sludge occurred during coke wastewater treatment, whereas slight inhibition was observed during the treatment of landfill leachate. However, the diversity of the sludge structure and the acclimatization of the microorganisms contribute to sludge tolerance in unfavorable conditions (Cho et al. 2014; Kim et al. 2013).

Table 5 Phenol and SCN− levels in landfill leachate and coke wastewater and the removal method used
Full size table

Phenol and SCN− inhibition of AOB and NOB

By adding ATU and NaClO3, selective inhibitors of AOB and NOB, the inhibition of phenol and SCN− on AOB and NOB was quantified.

The IC50 values for phenol on AOB and NOB were 4.52 and 7.56 mg L−1, respectively, as shown in Fig. 1a. The results show that AOB is more sensitive to phenol stress than NOB, and this phenomenon is caused by a DO concentration imbalance in the bulk liquid. Phenol reduced the NH4 +-N removal rate, and the consumption of nitrite was not affected (Toh and Ashbolt 2002), as demonstrated by the substrate consumption results. Dyreborg and Arvin (1995) demonstrated that the pseudo-critical concentration of phenol was 3.7 mg L−1 in a pure culture of nitrifying microorganisms. As Morita et al. (2007) reported, 5.0 mg L−1 phenol completely inhibited the ammonia oxidation capability of suspended Nitrosomonas europaea cells.

Fig. 1
figure 1

Inhibition of different phenol (a) and SCN− (b) concentrations on the activity of AOB and NOB

Full size image

In the presence of SCN−, the IC50 values for AOB and NOB were 270 and 37.2 mg L−1, respectively, as shown in Fig. 1b. Both AOB and NOB were sensitive to SCN− exposure, which might be caused by the SCN− inhibition of several enzyme systems and Mg2+ ATPases (Gould et al. 2012). NOB is relatively more sensitive to thiocyanate than AOB.

RSM experiment and statistical analysis

Thirteen experiments were performed randomly to minimize the interference of uncontrolled variables on the obtained responses. The quadratic multinomial regression model is summarized in Eq. (6):

$$ RSRR=197-44.6\times \left[ Phenol\right]-0.892\times \left[{SCN}^{-}\right]+8.94\times {10}^{-3}\times \left[ Phenol\right]\times \left[{SCN}^{-}\right]+5.25\times {\left[ Phenol\right]}^2+2.16\times {10}^{-3}\times {\left[{SCN}^{-}\right]}^2 $$
(6)

Statistical tests involving the reduced quadratic models for the RSRR were performed with the F test for the ANOVA analysis, and the results are shown in Table 6. Accordingly, the intercept coefficient (β 0), the linear coefficient (β i ), the quadratic coefficient (β ii ), and the interaction coefficient (β ij ) all exhibited significant effects. Furthermore, the influence of the individual operational parameters on the RSRR followed the order of phenol > SCN−, and both of the linear coefficients (−44.6 and −0.892) were below zero, indicating that they negatively influence (inhibiting or toxic) the RSRR. The model was highly significant, with an F value of 86.62 and probability values of less than 0.05 for a 95% confidence interval. A “Pred R 2” of 0.9022 is in reasonable agreement with the “Adj R 2” of 0.9727. “Adeq Precision” measures the signal-to-noise ratio, and the calculated value of 24.085 was greater than 4, suggesting adequate signal, meaning that the model can be used to navigate design space (Xing et al. 2014).

Table 6 ANOVA on the response surface quadratic model
Full size table

Good convergence between the experimental and predicted values are also depicted in Fig. 2a, and the predicted values versus the actual values almost graphed on the line “y = x.” In Fig. 2b, a random distribution was observed for the residual plots for the models and the RSRR dataset, and the absolute value of the internally studentized residuals remained low. Moreover, the six additional experiments (Table 5) confirmed that the RSRR experimental values are very close to the predictive model values. Based on these results, the model is suitable for the design space.

Fig. 2
figure 2

Predicted versus actual values for RSRR (a) and internally studentized residuals in the residuals versus the predicted (b)

Full size image

Interactive effects of phenol and SCN− on PN sludge activity

Three-dimensional response surface plots and two-dimensional contour plots were constructed to get a better visualization of the levels of the independent variables (phenol and SCN−). The output of the predictive model for RSRR is shown in Fig. 3a (three-dimensional response surface) and Fig. 3b (two-dimensional contour plots).

Fig. 3
figure 3

Three-dimensional (a) and contour plots (b) for the response variable (RSRR) with respect to the independent variables (phenol and SCN−)

Full size image

It can be concluded from Fig. 3a that the RSRR increased slightly when the concentration of phenol and SCN− were below 0.79 and 54.6 mg L−1, respectively. However, the RSRR decreased with further increases in the phenol and SCN− concentrations. The minimum RSRR was identified as the surface confined to the smallest curve of the contour plot. In the range of independent variables, the most severe inhibition was observed with a valley value (17%) for the RSRR when the phenol and SCN− concentrations were 4.08 and 198 mg L−1, respectively.

The shape of the contour reflects the intensity of the interaction between the factors, and an ellipse represents a significant interaction, whereas a circle indicates insignificance (Chen et al., 2014). The phenol and SCN− contours are elliptical, indicating that the interaction between phenol and SCN− is a mostly significant interaction. In the perturbation study, phenol played a dominant role when its concentration was below 2.8 mg L−1 and the concentration of SCN− did not exceed 175.5 mg L−1; however, when the phenol concentration was above 2.8 mg L−1 and SCN− exceeded 175.5 mg L−1, SCN− was the more toxic compounds. These findings agree with the sharp curve for phenol and SCN−. In Fig. 3b, for a constant phenol concentration of 0.6 mg L−1 and a concentration range for SCN− of 51–300 mg L−1, the RSRR response was maintained above 81% and only slightly fluctuated. However, when the concentration of SCN− was fixed at 51 mg L−1, the RSRR response ranged between 64 and 124% (Fig. 3b).

Because the joint toxicities of multicomponent mixtures are usually different from the toxicities of the individual chemicals (Ding et al. 2015), an IP was used to judge the combined toxicity of phenol and SCN− as shown in Fig. 4. The joint toxic effect of phenol and SCN− shifted from synergistic to additive, then to independent and finally to antagonistic with increasing concentrations. Accordingly, they shared more space independently.

Fig. 4
figure 4

Equivalent effect picture of the joint toxicity of phenol and SCN−. A 1 and A 2 represent the 95% confidence intervals for the IC50 of phenol, and B 1 and B 2 represent the 95% confidence intervals for the IC50 of SCN−

Full size image

Although both phenol and SCN− exhibited toxic effects, the sensitivity of PN sludge does not necessarily mean that the activity of PN sludge is absent in a reactor treating substances that inhibit PN. For low concentrations of phenol and SCN−, the effective concentration of the inhibitor can be lowered by absorption, precipitation, chelation, and biodegradation (Vázquez et al. 2006). As the concentration of phenol and SCN− increased, the absorption sites reached saturation, and an additive effect was observed. No cross-inhibition of PN sludge activity was apparent in the independent area, and when the phenol concentration exceeded 5.95 mg L−1, the presence of SCN− aggravated the toxic effects on PN sludge.

In a PN process for treating wastewater from coke plants, refineries, and landfill leachates, some major compounds present in the wastewater would cause an inhibitory effect on microbial activity in activated sludge. Particularly, phenol most seriously inhibits PN in activated sludge; the influx of toxic compounds into the process should be controlled below their threshold concentrations.

Conclusions

The IC50 values of phenol and SCN− on PN sludge were 5.60 and 351 mg L−1, respectively. The relatively sustainable phenol and SCN− concentrations for PN sludge were 1.77 and 43.3 mg L−1, respectively. Based on batch tests, AOB is more sensitive to phenol than NOB, whereas NOB is more seriously suppressed by SCN−. The RSM analysis indicated a significant interaction between phenol and SCN−. The most severe inhibition, resulting in an RSRR of 17.0%, occurred when the concentrations of phenol and SCN− were 4.08 and 198 mg L−1, respectively. The joint inhibition effect of phenol and SCN− tended to vary based on the different concentration of the toxic components.