All PCBs (99% purity) were purchased from Fischer Scientific and used without further purification. All organic solvents were HPLC grade from Sigma-Aldrich. Non-porous P-25 TiO₂ powder (80% anatase, average diameter 200 nm and surface area ~50 m²/g) was a gift from Degussa Corporation (Akron, Ohio, USA). H₂O₂ (30%, reagent grade) was from Fischer Scientific.

PCBs structures were optimized using density functional theory (DFT) with B3LYP functional and the second-order Möller-Plesset perturbation approach. No symmetry constrains were imposed during geometry optimization. Geometry searches for a variety of possible configurations were performed to obtain the global minimum. DFT and MP2 vibrational frequency calculations were carried out to verify the true minimum.

PCBs were dissolved in acetonitrile/water (80:20) and stored at 4 °C. The PCB solution was poured into a flat-bottomed quartz cuvette (1 cm light path) with a stopper to prevent solvent evaporation. Control samples were prepared using the same conditions as the test samples but covered with aluminum foils and kept in dark. These samples were analyzed with HPLC to validate the photo aided degradation. All experiments with TiO₂ were kept in the dark for 180 min to establish equilibrium between adsorption and desorption. The amount of PCBs adsorbed by TiO₂ at equilibrium was measured in suspensions containing 3 mg of TiO₂ in 3 mL of acetonitrile solution containing 100 μM of the PCB. After 180 min in the dark, the suspensions were centrifuged and the residual PCB concentrations in the supernatant were determined by HPLC. The samples after dark absorption were exposed to simulated sunlight in a reactor with a 1,000 W xenon lamp (Suntest CPS+ lamp system, Atlas MTT BV, Germany). The photoreactor lamp emits light in the range of 300–800 nm with an irradiance of 250–765 W/m² and the temperature was maintained at 35 °C. Samples were collected at regular intervals of irradiation. After irradiation, the samples containing TiO₂ were centrifuged at 9000 RPM for 10 min and the supernatant was filtered through 0.45 μm Millipore filter discs before HPLC analysis. The pH of the solutions was adjusted with 0.01 M NaOH solution. Photodegradation experiments carried out were: (1) At pH 7 and 9; (2) In the presence of TiO₂ at pH 7 and 9; (3) In the presence of H₂O₂ alone; (4) In the presence of both TiO₂ and H₂O₂. The amount of TiO₂ was 1 mg/mL and the final concentration for H₂O₂ was 12 mM.

The concentration of PCBs was assessed using HPLC (Shimadzu Corp., Kyoto, Japan). The samples were filtered with 0.25-μm Teflon syringe filter (Millipore) prior to analysis and eluted with a 4.6 × 150 mm reverse phase C₁₈ (Zorbax XDB-C₁₈) column using acetonitrile/water (80:20) as the mobile phase at a flow rate of 1 mL/min. PCBs were detected by absorbance at 250 nm. Replicates gave an error of ±5%. In the case of direct photolysis, samples for 0 min irradiation were collected just before the irradiation, whereas 0 min samples for TiO₂ photocatalysis and peroxide catalysis were collected before and after addition of the catalyst to study the effect of the catalyst on degradation. The peak height (mAu) at 0 min is considered as maximum concentration (100%) of the PCB. The change in peak height with time was plotted to show the degradation of PCBs. Complete destruction of a PCB was considered when the parent peak disappeared. For all samples, triplicates were run and peak height was presented as an average of these values.

Photodegradation rate constant was calculated for pseudo-first order reactions based on the plot of ln (A0/At) = kt. A plot of ln (A₀/A_t) versus irradiation time (t) was fitted to a straight line whose slope equals to the first-order rate constant k (min⁻¹). Half-life (t_1/2, min) values were calculated by k/0.693.

All five PCBs were structurally optimized to obtain the lowest energy conformation with the computed global minimal energy structure shown in Table 1. The dihedral angle between the two benzene rings for each PCB was determined for the optimized structure. As expected, PCBs without chlorine on either of the ortho-positions, 3,3′,4,4′-TCB and 3,4′,5-TCB, are near planar with the dihedral angles between the two benzene rings being about 37°. For 2,2′,4,4′-TCB, which has one chlorine each on each of the benzene ring’s ortho-positions, the dihedral angle is 68°. If both the ortho-positions of the same benzene ring are occupied by chlorines, like in 2,6-DCB and 2,3,5,6-TCB, the two benzene rings are nearly perpendicular to each other with dihedral angles of 77° and 90°, respectively.

Photodegradation of PCBs was monitored using HPLC. From the chromatogram for 2,6-DCB in Figure 1, one can see that 2,6-DCB appears at an elution time of 6.0 ± 0.2 min. At pH 7, 120 min irradiation yielded intermediates between 4–6 min (Figure 1, left), but in the presence of TiO₂, more complex products were detected with elution time of 4–6 min and 10–11 min. The peaks at 10.5 and 11.0 min are probably high molecular weight compounds that degrade on further irradiation. At alkaline pH 9 (Figure 1, middle), a major peak appeared between 8.0–10.5 min which we assume to be a higher molecular weight compound and seemed to be different from the ones under pH 7. In the presence of H₂O₂ alone (Figure 1, right), no significant reduction in the parent peak was observed. Further, addition of TiO₂ slowed down the degradation and the peak at 10.5 min was split to two peaks after 120 min irradiation. This might be attributed to the dechlorination of the PCBs resulting in the formation of lower biphenyls.

HPLC analysis of 2,2′,4,4′-TCB (Figure 2) showed the initial compound peak at 9.4 ± 0.2 min. At pH 7 (Figure 2, left), photolysis resulted in probably lower biphenyls at 4.0, 5.5 and 8.0 min. Similar photoproducts were observed in the case of light + H₂O₂ but with much faster degradation rate, indicating that peroxide speeds up the degradation. Addition of TiO₂ as a catalyst with or without peroxide might have transformed the parent compound into a higher molecular weight product (Figure 2, right). A peak at 10.3 min elution time was seen in both cases along with the other peaks at 4.5 min, 8.0 min and 8.7 min. Degradation with H₂O₂ went to near completion after 30 min of irradiation.

HPLC chromatogram f 3,3′4,4′-TCB has a peak at 11.3 ± 0.2 min. At pH 7, light irradiation for 240 min produced peaks at 6.0, 7.0, 7.5 and 9.5 min (Figure 3, left). At pH 9, the parent compound was more efficiently degraded, but no photoproducts were detected (Figure 3, middle). There was no evidence for the formation of intermediates when H₂O₂ was used (Figure 3, right).

3,4′,5-TCB has an initial peak at 9.8 ± 0.2 min. Irradiation at pH 7 (Figure 4, left) showed no degradation intermediates, but addition of TiO₂ resulted in a shoulder on the original peak, and light irradiation produced 4 degradation products at 4.5, 5.8, 7.4 and 8.0 min. At pH 9.0, a major peak at 7.0 min was observed; whereas addition of TiO₂ resulted in the formation of a peak around 11.0 min which may be a higher molecular weight compound (Figure 4, middle). Peroxide accelerates the degradation without formation of intermediates (Figure 4, right).

From the chromatograms of 2,6-DCB, 2,2′,4,4′-TCB and 3,3′,4,4′-TCB we observed that light-irradiation resulted in peaks eluting at lower elution times indicating a possible dechlorination of the parent PCBs, but addition of TiO₂ catalyst generated peaks at lower and higher elution times hinting a possible generation of both dechlorinated biphenyls and higher molecular weight compounds. We have also observed the peaks with higher elution time disappeared indicating a possible degradation of the higher molecular weight compounds upon further irradiation. As an example, in Figure 2 (left) the peaks below 10 min are probably the dechlorinated products of the parent 2,2′,4,4′-TCB, whereas the peak at 12.5 min must be a higher molecular weight compound. The peak around 9.5 min is normally a trichlorobiphenyl while the peak around 7.0 min may be a dichlorobiphenyl. Previous reports showed the formation of hydroxylated-DCB, methylated-DCB and mono-chlorobiphenyl for the degradation of 2,6-DCB [51,52]. Another report confirmed that 2,6-DCB photodegradation yielded 2-chlorobiphenyl, 6-chlorobiphenyl and biphenyl [26]. Similar products may have formed in our study (Figure 1). As seen in Figure 4 for 3,4′,5-TCB, some of the photoproducts are due to dechlorination to less substituted biphenyls [26,53–55].

The degradation of the 5 selected PCBs followed a pseudo-first order reaction in general [56,57]. The best linear fit yielded R² > 0.92. However, for TiO₂ photocatalyzed degradation, the linear fit was not optimal due to adsorption of PCBs onto the catalyst surface. The degradation kinetics involving H₂O₂ showed very good first-order fits. Their pseudo-first order degradation rate and half-lives were determined and listed in Table 2. Literature values for degradation half-lives in hexane irradiated by 254 nm UV lamp for some of the compounds in the present study are 2,6-DCB, 23 min; 2,2′,4,4′-TCB, 925 min; 3,3′,4,4′-TCB, >1,000 min. 2,2′,4,4′-TCB was efficiently degraded under all conditions, but under the condition with light + TiO₂ + H₂O₂ being the fastest. The slow degrading compounds are 2,6-DCB, 3,4′5-TCB and 3,3′4′4′-TCB in both neutral and alkaline conditions. Peroxide can efficiently catalyze complete mineralization of most PCBs.

A representative graph for the degradation kinetics with direct photolysis, TiO₂ photocatalysis and peroxide catalysis is presented in Figure 5. 3,3′,4,4′-TCB which has all the 4 ortho positions vacant was highly resistant to photodegradation with very large half-lives (Figure 5A). Similar results were reported by Miao et al. [24] when the degradation was performed in UV light with hexane as solvent. However, peroxide as catalyst has greatly accelerated the degradation rate of 3,3′,4,4′-TCB with a t_1/2 of 20 min and rate constant of 0.034 min⁻¹. The efficiency of hydrogen peroxide in achieving faster degradation is due to the production of larger number of hydroxyl radicals [58,59]. Complete mineralization of 3,3′,4,4′-TCB could be achieved in a shorter time in the presence of hydrogen peroxide suggesting the success of this methodology.

It can be clearly seen in Table 2 that the two planar PCBs, 3,4′,5-TCB and 3,3′,4,4′-TCB, degrade very slowly with degradation half lives of >4 hours. While the non-planar 2,6-DCB has a half-life around 2 hours and the perpendicular 2,3,5,6-TCB and 2,2′,4,4′-TCB degrades even faster with half lives of 107 and 38 min respectively. 2,3,5,6-TCB, and 2,2′,4,4′-TCB degrade very fast at pH 9 with estimated half lives of less than 10 min. Lin et al. reported that the number and location of chlorine atoms are important for PCB degradation [60]. Hong et al. [26] showed that photodegradation of 2,2′,4,4′-TCB in n-hexane with t_1/2 of 926 min. In contrast, its degradation half life is <66 min in our study in the aqueous solvent. The difficulty in degrading planar PCBs like 3,3′,4,4′-TCB is due to the extended conjugation between the two phenyl rings leading to a longer half-life as reported by Miao et al. [24] It has been widely reported that ortho chlorine confers greater photolability on a PCB [27,61]. Similar results were observed for 3,3′,4,4′-, 3,3′,4,4′,5-, and 3,3′,4,4′,5,5′-PCBs by de Felip et al. [51]. All these three PCBs showed t_1/2 values of >300 min which is in good agreement with our results. Our results further confirm that photoreactivity of PCBs depends on the position of the chlorines: ortho > meta > para positions of PCB rings [20].

The presence of H₂O₂ has a profound effect on photodegradation of non-planar PCBs which degrade generally very slow. The degradation rate enhancement is 20 times for 3,3′,4,4′-TCB and 35 times for 3,4′,5′-TCB. This great enhancement could be utilized as a possible “green” method for environmental remediation of planar PCBs. TiO₂ alone does not have a strong effect on the photodegradation in aqueous solution. However, the combination of H₂O₂ and TiO₂ speeds up the degradation for all the five tested PCBs. H₂O₂-catalyzed photodegradation of all 5 PCBs showed photoproducts of lower substituted biphenyls. The degradation was relatively fast (<30 min) for 2,2′,4,4′-TCB and 3,3′,4,4′-TCB without intermediate photoproducts detected. It was reported that irradiation of H₂O₂ with λ < 360 nm light leads to efficient generation of OH^• radicals [54]. The OH^• can dechlorinate PCBs and produce lower molecular weight products.

The use of light/H₂O₂/TiO₂ for the degradation of many hazardous compounds has been reported in recent years. However, PCBs degradation using hydrogen peroxide as a catalyst was not reported to our knowledge [57,62–64]. H₂O₂ can absorb light energy from UV irradiation and the O-O bond is ruptured leading to the production of activated ^•OH which is the crucial step in improving the rate of the reaction [62]. Other mechanisms proposed by Barkat et al. [65] suggested that H₂O₂ is a better electron acceptor than oxygen decreasing the chances of an electron/holes recombination [65]. The hydroxyl radicals formed by the photolysis of H₂O₂ might react with the holes generated during excitation which in turn will reduce the probability to undergo electron-hole recombination. The superoxide radical (^•O₂ ⁻) formed as the result is less dominant in affecting the rate of the photocatalytic reaction as compared with hydroxyl radicals. The general mechanisms of radical formation in various photocatalytic reactions are presented below.

In the present study the application of peroxide in the degradation of selected PCBs was found to be the most effective methodology. Degradation of 2,6-DCB in H₂O₂/UV and H₂O₂/UV/TiO₂ showed 68% and 83% respectively. While 100% degradation of 2,3,5,6-TCB was seen using both methods. TiO₂ presence has aided the degradation to completion in 60 min which otherwise took 90 min. For degradation of 2,2′,4,4′-TCB, UV/H₂O₂/TiO₂ facilitated 100% degradation after 210 min of irradiation whereas 98% degradation was achieved in 270 min in the absence of the semiconductor catalyst. The effect of peroxide on coplanar PCBs, 3,4′,5-TCB and 3,3′4,4′-TCB, was slightly different from the previously discussed PCBs. Both these PCBs were highly resistant to degradation at pH 7 and partially degraded at pH 9. When peroxide was used in the presence of light, 100% degradation of the compound was achieved after 60 min of irradiation with no intermediates seen, while using H₂O₂/UV/TiO₂ yielded 76% reduction at the end of 150 min.