Introduction

Atrazine (ATZ) is a chloro-s-triazine herbicide that has been widely used for the control of broad leaf and grassy weeds by interfering with the normal function of photosynthesis. Due to its high leaching potential, resistance to microbial degradation, slow hydrolysis, and moderate solubility, ATZ can accumulate in soil, surface water, and groundwater for a long period of time after being applied. ATZ can cause genotoxicity, thyroid gland problems, endocrine disruption, and serious human birth defects such as low birth weights and menstrual problems (Kong et al. 2016). U.S. EPA has classified ATZ as a possible human carcinogen, priority hazardous substance, and endocrine-disrupting compound. Aside from leaching after application, another important source of ATZ comes from the inadequate disposal of wastewater during its production process as its removal through traditional wastewater treatment processes is often incomplete (da Costa Filho et al. 2016; Aquino et al. 2017). Therefore, reliable and effective techniques for removing ATZ from industrial wastewater, especially from its production wastewater, are much desirable as a source control measure.

Recently, many research efforts have been focused on the removal of ATZ from aqueous environments using either stand-alone or hybrid advanced oxidation processes (AOPs), such as UV irradiation (Silva et al. 2014; Aquino et al. 2017), ozonation (Acero et al. 2000; Beltrán et al. 2000; Yang et al. 2014; Zhou et al. 2016; Gomes et al. 2017), UV/ozonation (Beltrán et al. 1994), sonolysis (Xu et al. 2014), Fenton process (Zhao et al. 2014), UV/H2O2 (Beltrán et al. 1996; Luo et al. 2015a), UV/chlorine (Kong et al. 2016), and UV/persulfate (Khan et al. 2014; Bu et al. 2016). The degradation of ATZ by AOPs usually involves de-chlorination, de-alkylation, and de-amination with subsequent hydroxylation, leading to the main products including cyanuric acid, ammelide, and ammeline (Bianchi et al. 2006). AOPs cannot achieve complete mineralization because opening the s-triazine ring (e.g., cyanuric acid) is difficult and can only be achieved under hydrothermal conditions (Horikoshi and Hidaka 2003). Nonetheless, effluent from AOPs can be readily treated by following biological treatment in order to completely mineralize ATZ and its end products (Lester et al. 2013). It should be noted that most of the previous studies reported in the literature have been performed in the background of ultrapure water, drinking and natural water, or synthetic wastewater. The effectiveness of AOPs in treating ATZ containing industrial wastewater particularly ATZ production wastewater has not been documented.

Given the industrial synthesis of ATZ needs a high amount of sodium hydroxide to avoid acidification, the ATZ production wastewater can be considered as a concentrated solution of sodium hydroxide contaminated with ATZ and other organics. The recovery of high-purity sodium hydroxide is thus an attractive option from a waste management perspective and therefore requires the reduction of impurities such as chemical oxygen demand (COD) in industry practice (Prieto-Rodríguez et al. 2013; Manenti et al. 2015). Ammonia nitrogen (NH3-N) present in ATZ production wastewater also needs to be reduced in order to protect aquatic organisms at receiving water bodies (Huang et al. 2008; Khuntia et al. 2012; Wen et al. 2016). To the best of our knowledge, there exists no previous report of the simultaneous removal of COD, NH3-N, and ATZ from ATZ containing industrial wastewater by AOPs. In addition, kinetic modeling has been proposed as an efficient tool to enhance the understanding of the mechanisms of AOPs and to aid the design and optimization of industrial applications (Yang et al. 2014; Bu et al. 2016; Zhou et al. 2016). However, there have been no such studies thus far in modeling the treatment of ATZ containing wastewater by AOPs.

To help fill the knowledge gaps, the objectives of the present work were (1) to investigate the efficacy of ATZ production wastewater treatment, particularly the removal of COD and NH3-N by a typical AOP, namely UV/O3; (2) to develop a mathematical model for the removal kinetics of COD, NH3-N, and ATZ; and (3) to compare the contributions of direct O3 oxidation and indirect radical oxidation in the treatment system.

Materials and methods

Chemicals and photoreactor

Potassium dichromate, sulfuric acid, silver sulfate, mercuric sulfate, ferroin, ammonium iron sulfate, sodium hydroxide, salicylic acid, potassium sodium tartrate, sodium pentacyanonitrosylferrate, and sodium hypochlorite were purchased from Beijing Chemical Works, China. Atrazine-d5 standard and sodium thiosulfate were purchased from Anpel Laboratory Technologies (Shanghai), China. Trichloromethane (Thermo Fisher Scientific, China) was used for aqueous sample extraction. All chemicals were of analytical reagent grade (> 99% pure) and used as received without further purification. Ultrapure water was produced on-site from a Direct-Q 3 UV unit (Millipore, France). ATZ production wastewater was collected from a pesticide manufacturer. Detailed location information is not available due to client confidentiality. Its physical-chemical characteristics are listed in Table 1.

Table 1 Detailed physical-chemical characteristics of the ATZ production wastewater
Full size table

As shown in Fig. S1 (Supplementary Material), the bench-scale photoreactor has an inner 4 L quartz jar and an outer stainless steel jacket. The outer jacket has an aluminum lid that can be sealed to provide heat and light insulation. The inner diameter, height, and wall thickness of the quartz jar are 20, 25, and 0.4 cm, respectively. Eight 3.5 W low-pressure UV lamps, emitting exclusively at 254 nm, are evenly mounted inside the quartz jar near the wall. Incident irradiance for different numbers of lamps was measured at a wavelength of 254 nm with a calibrated radiometer (Sentry Optronics Corp, Model ST-512). A 300-W ozone generator with dedicated ozone flow rate monitor is used to produce ozone on-site from ambient air. The inner quartz jar has a PTFE lid equipped with a stirring rod on which two PTFE six-bladed paddle impellers are mounted to stir the wastewater sample at 80 rpm.

Experimental procedure and analytical methods

The experimental procedure started by transferring 2 L ATZ production wastewater into the quartz jar and stirring it for 20 min to reach the thermal and volatilization equilibria. Then, the ozone generator was switched on and O3 was bubbled into the bottom of the quartz jar through a distributor nozzle at a fixed rate of 15 g h−1 (measured by the ozone generator). UV lamps were switched on 30 min before filling the jar, and power was provided at three different levels of 0, 7, and 14 W by using 0, 2, and 4 lamps, respectively. Temperature was maintained at room temperature and the system was magnetically stirred during the experiment. At various time intervals during a 180-min period, a 20-mL water sample was collected from the reactor using a peristaltic pump, transferred into a 20-mL amber vial and immediately quenched by adding 1.0 mL saturated Na2S2O3 solution to remove residual oxidants. All experiments were triplicated and averaged to ensure reproducibility.

Given the removal efficiency of ATZ by UV/O3 was investigated in our previous studies (Jing et al. 2015, 2017), the focus of this study was on the reduction kinetics of COD and NH3-N. Therefore, the measurement of COD and NH3-N was conducted for samples collected at all time points (i.e., 20, 40, 60, 90, 120, 150, and 180 min). COD was measured by the potassium dichromate method after 100-fold dilution due to the high chloride content. NH3-N was determined by the spectrophotometer method with salicylic acid. Contrastingly, ATZ was only measured at the ending point (i.e., 180 min) by a pretreatment method proposed by Zheng et al. (2015a) and a gas chromatograph (GC) (Agilent 7890A) equipped with a HP-5MS column (30 m × 5 mm × 0.25 μm) and a mass selective detector (MS) (Agilent 5975C). Oven temperature was initially set as 70 °C for 2 min and then increased by 20 °C min−1 to 230 °C and maintained for 20 min. Analysis was performed in the selected ion monitoring mode at an electron energy of 70 eV and a source temperature of 230 °C.

Kinetic modeling

Table 2 summarizes the possible reactions in the UV/O3 batch reaction system along with rate constants obtained from the literature. To develop the kinetic model, the following assumptions were made: (1) only reactions tabulated in Table 2 occurred in the system; (2) to account for light scattering and absorbing effect caused by suspended solids, a linear attenuation correlation suggested by Luo and Al-Dahhan (2004) and Benson and Rusch (2006) was adopted to modify the Beer-Lambert Law equation as shown in Eq. 5; (3) the reactions of COD with O3 and OH· were second order; (4) the removal of COD caused by direct photolysis was negligible as compared with oxidation by OH· and O3 (Gassie et al. 2016); (5) NH3-N was expressed in the form of NH3 because at pH greater than 11, over 90% NH3-N exists as free ammonia (Khuntia et al. 2012); (6) the radical scavenging effect caused by \( {\mathrm{CO}}_3^{2-} \) and \( {\mathrm{HCO}}_3^{-} \) was negligible due to the complex matrix (Luo et al. 2015a); (7) the change of pH was negligible; (8) given the flow rate of 15 g h−1 and the total volume of 2 L in this study, an ozone injection rate of 4.34 × 10−5 M s−1 was obtained. Ozone mass transfer rate in semi-batch reactors usually stabilizes after a certain period of time. However, in this study, the real-time ozone concentration was not measured due to turbidity and technical difficulties. To address this concern, as shown in Eq. 7, a fixed ozone utilization rate of 0.7 (Pophali et al. 2011; Quero-Pastor et al. 2014) was adopted to represent the increase of ozone concentration; (9) according to Park and Kim (2015), a pseudo-first-order rate constant of 3.47 × 10−6 s−1 was used for ammonia stripping; (10) because the wastewater pH (i.e., 12) was greater than the pK a (i.e., 11.8) of H2O2, the existence of its conjugated species (\( {\mathrm{HO}}_2^{-} \)) and the high extinction coefficient of \( {\mathrm{HO}}_2^{-} \) (i.e., 240 M−1 cm−1) cannot be overlooked.

Table 2 Reactions occurred in the UV/O3 system
Full size table

The kinetic expressions of ATZ, O3, OH·, H2O2, HO2, \( {\mathrm{HO}}_2^{-} \), Cl−, Cl, \( {\mathrm{Cl}}_2^{-} \), ClOH−, COD, NH3, NH2, and NHOH· are shown in Eqs. 1–13 as below and Eqs. S1–S6 in the Supplementary Material, with the reaction rate constants listed in Table 2.

$$ {f}_{ATZ}=\frac{\varepsilon_{ATZ}\left[ ATZ\right]}{\varepsilon_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]} $$
(1)
$$ {f}_{O_3}=\frac{\varepsilon_{O_3}\left[{O}_3\right]}{\varepsilon_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]} $$
(2)
$$ {f}_{H_2{O}_2}=\frac{\varepsilon_{H_2{O}_2}\left[{H}_2{O}_2\right]}{\varepsilon_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]} $$
(3)
$$ {f}_{{\mathrm{HO}}_2^{-}}=\frac{\varepsilon_{{\mathrm{HO}}_2^{-}}\left[{\mathrm{HO}}_2^{-}\right]}{\varepsilon_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]} $$
(4)
$$ I={I}_0{e}^{-\alpha b}={I}_0{e}^{-\left({k}_b{c}_b+{k}_w\right)b} $$
(5)
$$ {\displaystyle \begin{array}{l}d\left[ ATZ\right]/ dt=-{\varPhi}_{ATZ}{If}_{ATZ}\left(1-{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)-{k}_1\left[{O}_3\right]\left[ ATZ\right]-\\ {}\kern36.5em {k}_9\left[ ATZ\right]\left[ OH\cdot \right]-{k}_{16}\left[ ATZ\right]\left[{Cl}_2^{-}\cdot \right]\end{array}} $$
(6)
$$ {\displaystyle \begin{array}{l}d\left[{O}_3\right]/ dt=0.7\times 4.34\times {10}^{-5}-{\varPhi}_{O_3}{If}_{O_3}\left(1-{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)-\\ {}\kern1em {k}_1\left[{O}_3\right]\left[ ATZ\right]-{k}_2\left[{O}_3\right]\left[{HO}_2^{-}\right]-{k}_3\left[{O}_3\right]\left[{OH}^{-}\right]-{k}_4\left[{O}_3\right]\left[{H}_2{O}_2\right]-{k}_5\left[{O}_3\right]\left[ OH\cdot \right]-{k}_{17}\left[{O}_3\right]\left[{Cl}^{-}\right]-\\ {}\kern24.5em {k}_{18}\left[{O}_3\right]\left[{Cl}_2^{-}\cdot \right]-{k}_{20}\left[{O}_3\right]\left[ COD\right]\kern0.5em -4\times {k}_{21}\left[{O}_3\right]\left[{NH}_3\right]\end{array}} $$
(7)
$$ {\displaystyle \begin{array}{l}d\left[ OH\cdot \right]/ dt=2{\varPhi}_{H_2{O}_2}{If}_{H_2{O}_2}\left(1-{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)+{\varPhi}_{HO_2^{-}}{If}_{HO_2^{-}}\left(1-\right.\\ {}\left.{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)+{k}_2\left[{O}_3\right]\left[{HO}_2^{-}\right]-{k}_4\left[{O}_3\right]\left[{H}_2{O}_2\right]-{k}_5\left[{O}_3\right]\left[ OH\cdot \right]-\\ {}\kern4.5em {k}_6\left[{H}_2{O}_2\right]\left[ OH\cdot \right]-{k}_7\left[{HO}_2^{-}\right]\left[ OH\cdot \right]-{k}_8\left[{HO}_2\cdot \right]\left[ OH\cdot \right]+{k}_9\left[ ATZ\right]\left[ OH\cdot \right]+{k}_{10r}\left[{Cl OH}^{-}\cdot \right]-\\ {}{k}_{10f}\left[{Cl}^{-}\right]\left[ OH\cdot \right]-{k}_{19}\left[ COD\right]\left[ OH\cdot \right]-{k}_{22}\left[{NH}_3\right]\left[ OH\cdot \right]+{k}_{24}\left[ NHOH\cdot \right]\left[{HO}_2^{-}\right]\end{array}} $$
(8)
$$ {\displaystyle \begin{array}{l}d\left[{H}_2{O}_2\right]/ dt={\varPhi}_{O_3}{If}_{O_3}\left(1-{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)-{\varPhi}_{H_2{O}_2}{If}_{H_2{O}_2}\left(1-\right.\\ {}\left.{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)\kern0.5em -{k}_4\left[{O}_3\right]\left[{H}_2{O}_2\right]-{k}_6\left[{H}_2{O}_2\right]\left[ OH\cdot \right]-\\ {}{k}_{14}\left[{H}_2{O}_2\right]\left[ Cl\cdot \right]-{k}_{15}\left[{H}_2{O}_2\right]\left[{Cl}_2^{-}\cdot \right]-{k}_{23}\left[{NH}_2\cdot \right]\left[{H}_2{O}_2\right]\end{array}} $$
(9)
$$ {\displaystyle \begin{array}{l}d\left[{HO}_2\cdot \right]/ dt={k}_4\left[{O}_3\right]\left[{H}_2{O}_2\right]+{k}_5\left[{O}_3\right]\left[ OH\cdot \right]+{k}_6\left[{H}_2{O}_2\right]\left[ OH\cdot \right]+{k}_7\left[{HO}_2^{-}\right]\left[ OH\cdot \right]-\\ {}\kern10em {k}_8\left[{HO}_2\cdot \right]\left[ OH\cdot \right]+{k}_{14}\left[{H}_2{O}_2\right]\left[ Cl\cdot \right]+{k}_{15}\left[{H}_2{O}_2\right]\left[{Cl}_2^{-}\cdot \right]\end{array}} $$
(10)
$$ {\displaystyle \begin{array}{l}d\left[{HO}_2^{-}\right]/ dt=-{\varPhi}_{HO_2^{-}}{If}_{HO_2^{-}}\left(1-{e}^{-2.303b\left({\varepsilon}_{ATZ}\left[ ATZ\right]+{\varepsilon}_{O_3}\left[{O}_3\right]+{\varepsilon}_{H_2{O}_2}\left[{H}_2{O}_2\right]+{\varepsilon}_{HO_2^{-}}\left[{HO}_2^{-}\right]\right)}\right)-{k}_2\left[{O}_3\right]\left[{HO}_2^{-}\right]+\\ {}\kern25em {k}_3\left[{O}_3\right]\left[{OH}^{-}\right]-{k}_7\left[{HO}_2^{-}\right]\left[ OH\cdot \right]-{k}_{24}\left[ NHOH\cdot \right]\left[{HO}_2^{-}\right]\end{array}} $$
(11)
$$ d\left[ COD\right]/ dt=-{k}_{19}\left[ COD\right]\left[ OH\cdot \right]-{k}_{20}\left[ COD\right]\left[{O}_3\right] $$
(12)
$$ d\left[{NH}_3\right]/ dt=-{k}_{21}\left[{O}_3\right]\left[{NH}_3\right]-{k}_{22}\left[{NH}_3\right]\left[ OH\cdot \right]-{k}_{25}\left[{NH}_3\right] $$
(13)

where f ATZ , \( {f}_{O_3} \), \( {f}_{H_2{O}_2} \), and \( {f}_{HO_2^{-}} \)are the fractions of UV irradiation absorbed by ATZ, O3, H2O2, and \( {\mathrm{HO}}_2^{-} \), respectively; b is the optical path length (cm); I and I 0 are the attenuated average irradiance and incident irradiance, respectively (E L−1 s−1); α is light attenuation coefficient; k b and c b are the extinction coefficient accounting for suspended solids (cm−1) and the number of suspended solids, respectively; k w is the extinction coefficient accounting for water (cm−1); ε ATZ , \( {\varepsilon}_{O_3} \), \( {\varepsilon}_{H_2{O}_2} \), and \( {\varepsilon}_{HO_2^{-}} \)are the molar extinction coefficients of ATZ, O3, H2O2, and \( {\mathrm{HO}}_2^{-} \) at 254 nm, respectively (M−1 cm−1); Φ ATZ , \( {\varPhi}_{O_3} \), \( {\varPhi}_{H_2{O}_2} \), and \( {\varPhi}_{HO_2^{-}} \) are the quantum yields of ATZ, O3, and H2O2, respectively (mol E−1).

Model parameters and initial conditions are listed in Table 3 by considering (1) incident irradiance was assumed to be homogeneous in the reactor; (2) given that k w was set as 0.002 cm−1 (Luo and Al-Dahhan 2004), the term k b c b can be estimated at 0.02 cm−1 according to Benson and Rusch (2006); (3) chlorides were assumed to all be NaCl (197,500 mg/L = 5.56 M); (4) the rate constants of COD with O3 and OH· were set as 0.32 M−1 s−1 (Rivas et al. 2009; Kwon et al. 2012) and 4.9 × 109 M−1 s−1 (Mohajerani et al. 2012; Sekaran et al. 2014) based on literature recommendations and trial-and-error testing, respectively; (5) the initial concentration of COD in M was converted from 14,300 mg L−1 in Table 1.

Table 3 Parameters and initial conditions used for kinetic modeling
Full size table

The degradation of ATZ, COD, and NH3 and the individual contributions of various reactive species under different experimental conditions were simulated in MATLAB using function ode15s. The modeling results were then compared with those obtained from experiments to test the accuracy of the proposed kinetic model.

Results and discussion

UV/O3 treatment efficiency

The removal of COD and NH3-N by UV/O3 was found to be best described by pseudo-first-order regression as depicted in Fig. 1. The pseudo-first-order rate constants shown in Eq. 14 were calculated using standard least squares procedures, and all regression coefficients (R 2) were greater than 0.90, indicating reasonable goodness of fit (Table 4).

$$ \mathit{\ln}\left(\frac{C_t}{C_0}\right)=- kt $$
(14)

where c t and c 0 are the instant and initial concentrations of COD or NH3-N (M), respectively; t is time (min); and k is the pseudo-first-order rate constant (min−1). When UV power was set at 0 W, the treatment process was equivalent to sole ozonation and the results showed that 24.6% NH3-N was removed after 180 min (Fig. 2a), which was consistent with the literature. Zheng et al. (2015b) applied microbubble-ozonation to acrylic fiber manufacturing industry wastewater and found 21% NH3-N removal efficiencies after 120 min at pH = 8 and an O3 dose of 5 g h−1. Although O3 itself can oxidize free NH3 to nitrate in a relatively slow process (Lester et al. 2013), the abundant hydroxyl ions (OH−) in the ATZ production wastewater (i.e., pH = 12 in this study) can initiate the decomposition of aqueous O3 into OH·, which can subsequently degrade NH3-N (Ozturk and Bal 2015). Similar trend has been reported by Luo et al. (2015b) in treating ammonia-containing wastewater. There was no removal of NH3-N at a pH of 8 after 120 min ozonation, while an 85% removal was observed at a pH of 12.

Fig. 1
figure 1

Pseudo-first-order regression of the removal of COD and NH3-N using 0, 7, and 14 W UV light (O3 flow rate at 15 g h−1)

Full size image
Table 4 Pseudo-first-order rate constants of the removal of COD and NH3-N (ozone dose at 15 g h−1)
Full size table
Fig. 2
figure 2

Comparison between measured and modeled removal of COD and NH3-N using a 0 W, b 7 W, and c 14 W UV light (O3 flow rate at 15 g h−1)

Full size image

Contrastingly, the removal of COD was only about 2% at a rate of 0.0001 min−1. Malik et al. (2017) found a COD removal efficiency of 10–38% to treat complex textile wastewater using 0.1–0.5 g h−1 O3. Wu et al. (2017) also reported that the ozonation of petrochemical secondary effluent can only reduce 13% of the COD with an O3 dose if 7.9 g h−1 within 1 h. Nonetheless, in this study, a 2% reduction of COD at 15 g h−1 O3 dose was considered to be significantly low. This may be attributed to the fact that most organics in the complex wastewater matrix were oxidized selectively by O3. Some refractory organic matters (e.g., acetic acid and toluene) cannot be detected by the potassium dichromate method but may be degraded to detectable COD by UV photolysis and larger amount of OH· when UV was applied. It also suggests that the generation of OH· via OH− induced O3 decomposition (i.e., \( {\mathrm{HO}}_2^{-} \) as an intermediate) may not be sufficient for an effective mineralization of the organics in ATZ production wastewater.

In real wastewater containing high alkalinity and organic content, the differences between the performance of AOPs are usually more marked (Carra et al. 2016). Therefore, when UV was applied, it can be seen that the removal of both COD and NH3-N was appreciably enhanced with increasing UV power. The direct photolytic degradation of COD and NH3-N by UV is generally considered insignificant (Lucas et al. 2010). Therefore, such enhancement is believed to be ascribed to a synergistic effect that the self-decomposition of O3 is accelerated by UV irradiation, resulting in the generation of more OH·. Hong et al. (2016) also reported COD removal efficiencies of 12.4, 38.9, and 61.6% for the treatment of bio-treated textile wastewater by UV, O3, and UV/O3. As OH· formed from O3 decomposition can considerably oxidize NH3-N to \( {\mathrm{NO}}_2^{-} \) and further to \( {\mathrm{NO}}_3^{-} \) in alkaline conditions (Huang et al. 2008; Schroeder et al., 2011), the elevation of UV power from 0 to 14 W was able to raise the rate constant of NH3-N from 0.0015 to 0.0056 min−1. Moreover, applying UV with O3 was able to substantially enhance the BOD5/COD ratio. For example, when 14 UV was used, the BOD5/COD ratio was observed to increase from 0.15 to 0.37 after 180 min treatment, indicating an increase of the biodegradability.

As for ATZ, its removal efficiency was determined to be over 95% after 180 min treatment, regardless the level of UV power. ATZ undergoes fast degradation and is transformed into totally de-alkylated intermediates by ozonation alone, where both direct O3 attack to the nitrogen or α-carbon atom of the side chains and OH· attack to the α-carbon are involved. ATZ can also be degraded by UV alone through de-chlorination first, where the cleavage of the C–Cl bond occurs at its excited state and then a relatively slow de-alkylation, leading to a greater detoxification rate (Bianchi et al. 2006). Silva et al. (2014) also reported 65–80% removal efficiency with significantly reduced toxicity after 30 min 36 W UV exposure. When applied together, the synergistic effect between UV and O3 can result in a relatively faster de-alkylation and slower de-chlorination, resulting in the final product of cyanuric acid.

Kinetic modeling of the UV/O3 system

Figure 2 plots the measured and modeled removal efficiencies of COD and NH3-N at each sampling point when 0, 7, and 14 W UV was employed. It can be seen that the modeled results agreed well with the measured ones, with root mean square errors (RMSE) of 12.2 and 13.1% for COD and NH3-N, respectively. The largest discrepancy can be observed in predicting the removal of COD in Fig. 2a and that of NH3-N in Fig. 2c. In Fig. 2a, the modeled removal efficiencies of COD were higher than the observed ones with a RMSE of 20.3%. One possible reason could be ascribed to the existence of some refractory organic matters, which cannot be detected by the potassium dichromate method but may be degraded to detectable COD when UV was applied (Fig. 2b, c). Another possible explanation is that the amount of OH· formed by OH− induced O3 decomposition was not enough, in the experimental system, to achieve a significant mineralization of organics. In addition, pH was not measured during the experiments and was assumed to be unchanged for the ease of model development. Hydroxyl ion-catalyzed O3 decomposition is highly dependent on the pH of the aqueous system. Treating wastewater by O3 may slightly reduce pH due to the formation of small molecule organic acids and carbonic acids from the mineralization, which may affect the OH· degradation pathway (Lucas et al. 2010). However, these possible mechanisms were not supported by kinetic modeling as the reduction of COD was modeled as a whole with fixed second-order rate constants, thus causing the discrepancy in Fig. 2a. In Fig. 2c, when 14 W UV was applied, the modeled removal process of NH3-N was much slower than the measured one. This may be caused by the synergistic effect between UV dose and ozone such that the actual rate constants for NH3-N removal were higher than the ones listed in Table 2.

The modeled removal efficiencies of ATZ at 0, 7, and 14 W UV after 180 min treatment period were all greater than 95%, which were in good agreement with experimental results. As shown in Fig. 3, when UV was applied, even at a limited power of 7 W, the modeled removal process of ATZ was appreciably accelerated as compared to that of O3 alone. Such an elevated rate of ATZ degradation can be attributed to two possible mechanisms. One is the extra OH· generated from O3 decomposition by UV irradiation. According to the insert of Fig. 3, the modeled concentration of OH· increases sharply with the level of UV power applied and is in a range that agrees reasonably with previous research (Zhao et al. 2014; Meng et al. 2017). The other is that higher UV irradiation can increase the possibility of photons being absorbed by ATZ rather than being absorbed, reflected, and scattered by suspended solids. The modeled overall contribution of \( {\mathrm{Cl}}_2^{-} \) ·, OH·, and direct O3 oxidation to the removal of ATZ was not as comparable as that of direct UV photolysis. Figure 3 plots this contribution by only considering \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 oxidation in Eq. 6. For example, the removal efficiency of ATZ at 20 min with 7 W UV was predicted at 99%, whereas this number decreased to 31% by only considering \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 oxidation. The contribution of \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 oxidation to the pseudo-first-order rate constants for the removal of ATZ was around 10% when UV was applied.

Fig. 3
figure 3

Modeled removal of ATZ under different experimental conditions (O3 flow rate at 15 g h−1); calculation of \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 contributions was done by removing the first item UV direct photolysis in Eq. 6)

Full size image

Contrastingly, the contributions of \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 oxidation to the removal of ATZ without UV irradiation are plotted in Fig. 4. It can be seen that as long as OH· was taken into account, not considering the contribution from either \( {\mathrm{Cl}}_2^{-} \)· or direct O3 oxidation did not much affect the removal. This observation implies that in the alkaline environment, the decomposition of O3 into OH· dominates the direct O3 oxidation pathway. In addition, the concentration of Cl− decreased by 0.05% due to scavenging of O3 and OH·, whereas the contribution of \( {\mathrm{Cl}}_2^{-} \)· to the degradation of ATZ was determined to be negligible, which is consistent with previous reports (Kong et al. 2016).

Fig. 4
figure 4

Contributions of \( {\mathrm{Cl}}_2^{-} \)·, OH·, and direct O3 oxidation to the removal of ATZ (UV 0 W, O3 flow rate at 15 g h−1)

Full size image

Contributions of O3 and OH· to the removal of COD and NH3-N

To better understand the degradation mechanisms of both COD and NH3-N in real wastewater systems, the contributions of O3 and OH· were quantified by the developed kinetic model. The interactions between O3 and OH· were assumed to be negligible. Figure 5a shows that the removal of COD and NH3-N caused by direct O3 oxidation alone appears to be steady at a final proportion of 6.4 and 13.3%, respectively. As a comparison, OH· degrades up to 32.4 and 25.1% of the original amount of COD and NH3-N in 180 min, respectively. Direct O3 oxidation is a selective reaction with typical reaction rate constants of 1–1000 M−1 s−1 (Deng and Zhao, 2015) and usually predominates at pH less than 9. In this study, due to the high pH at 12, OH·-based radical reactions originated from O3 decomposition would thus dominate.

Fig. 5
figure 5

Modeled contributions of O3 and OH· on the removal of COD and NH3-N using a 0 W, b 7 W, and c 14 W UV light (O3 flow rate at 15 g h−1)

Full size image

As depicted by Fig. 5b and c, the higher the UV power, the faster is the decomposition of O3 and the formation of OH· radicals and thereby the greater the contribution of OH·. The contributions of OH· to the removal of COD and NH3-N were determined to be 46.7 and 35.5% with 7 W UV and then further increased to 61.4 and 47.5% with 14 W UV. Given the stable contribution of direct O3 oxidation under different experimental conditions, this finding is in accordance with many previous studies (Laera et al. 2011; Moussavi and Mahdavianpour 2016; Cheng et al. 2016) demonstrating that OH· plays a dominant role for the removal of both COD and NH3-N in the UV/O3 system, especially in alkaline environments. By comparing Fig. 5a with Fig. 5b and c, it can be concluded that the contributions of O3 and OH· at ozonation alone are not as distinguishable as those observed with UV irradiation. This indicates that many other solutes (e.g., Cl−) may consume OH· in competition with COD and NH3-N, and such deficiency can be overcome by applying UV irradiation.

Conclusions

The removal of COD, NH3-N, and ATZ from ATZ production wastewater via UV/O3 was examined by bench-scale experiments and modeled by reaction kinetics. The following conclusions were drawn and could be valuable for potential scale-up process optimization:

  1. 1)

    The removal of COD and NH3-N by UV/O3 followed pseudo-first-order kinetics. For O3 alone, the low COD removal may be attributed to the existence of certain refractory organic matters, which cannot be detected by the potassium dichromate method but can be degraded by OH· and converted to detectable COD afterwards. When UV was applied, the removal of both COD and NH3-N was appreciably enhanced.

  2. 2)

    Modeling results agreed reasonably with the experimental data. For COD, the discrepancy maybe due to the presence of refractory organic matters and possible fact that the amount of OH· formed by OH− induced O3 decomposition was not enough. As for NH3-N, the modeling error may be due to the underestimation of its reaction rate constants.

  3. 3)

    According to the modeling results, UV direct photolysis contributed appreciably to the degradation of ATZ, while OH· played a dominant role for the removal of both COD and NH3-N, especially in alkaline environments.

  4. 4)

    Some possible improvements that could be made for better prediction include continuous O3 concentration and pH measurement, in situ light attenuation determination, and a detailed analysis of organic composition of the wastewater samples.