Process Safety and Environmental Protection

The presence of toxic chlorinated compounds in drinking water, generated during the disinfection step in water treatment plants is of great concern for public health. Therefore, special attention has been given to the development of effective organochlorine-removal techniques. The reductive degradation via zero-valent-metals is recognized as a promising alternative. In this study, the capacity of zero-valent-copper (ZVC) containing materials to degrade 2,4,6-thichlorophenol (TCP) was investigated, using a bench-scale recirculating packed column system. The results indicate that this metal is effective for TCP degradation and dechlorination, even when derived from scrap. The kinetic model that better suits the degradation proﬁles is a second-order model, with an average normalized surface area rate constant (k SA ’) of (2.44 ± 1.27) × 10 − 3 L 2 min − 1 m − 2 for ZVC-containing materials. The ZVC scrap-derived material was found attractive for ﬁeld applications due to its reusability and low leachability, despite its performance being affected in the presence of water natural constituents. The degradation by-products elucidated conﬁrm that dechlorination is the main degradation pathway, leading to the formation of totally dechlo-rinated by-products such as phenol-like compounds and cyclohexanone. However, these may still pose a threat to aquatic organisms as revealed by toxicity assays and activity-structure relationship model (ECOSAR USEPA) predictions. Further investigation is therefore required aiming at following by-products formation with degradation time in order to ﬁnd the best residence time that generates innocuous and/or adequate efﬂuents for environmental disposal. © 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.


Introduction
Water contamination can occur through several processes, including those used for water disinfection, as is the case of the chlorination step in water treatment plants. In this process, the remaining organic matter can bind to the free residual chlorine, leading to the generation of organochlorines in drinking water supplies, which are generally harmful to human health (Agency for Toxic Substances and Disease Registry, 1999). Among these Chemical oxidation-reduction technologies are well developed as pretreatment alternatives for the degradation of chlorinated compounds and are able to convert these compounds into simple and biodegradable by-products (Yazdanbakhsh et al., 2018), being suitable for practical application. However, although oxidation processes have shown to be very efficient in organic compounds degradation, molecules with electron deficient groups (halogens) might reveal some resistance to oxidation (Munter, 2001;Juretic et al., 2014). In fact, oxidation of chlorinated organic compounds may lead to the generation of even more toxic products, as already mentioned in the case of TCP (Juretic et al., 2014). Given that, for these types of substances, an initial reductive treatment is preferred to a direct oxidative treatment (Satapanajaru and Anurakpongsatorn, 2008). Among the most common reducing agents for the dechlorination of organochlorines are zero-valentmetals (ZVM), which are a relatively low-cost option and simple to apply (Tratniek et al., 2010). Zero-valet-iron (ZVI) has been the most explored ZVM so far, given iron abundance, low price and low toxicity in comparison to other metals. In the particular case of TCP, besides ZVI (Dorathi and Kandasamy, 2012;Shengyang et al., 2010), zero-valent-zinc (ZVZ) (Kim and Carraway, 2003;Choi and Kim, 2009) and zero-valent-magnesium (ZVMg) (Morales et al., 2002) were also tested as reducing agents, presenting better reductive performances as the reduction potential decreases (E 0 Mg = −2.2 V < E 0 Zn = −0.76 V < E 0 Fe = −0.44 V, standard hydrogen electrode -SHE). However, recent investigations indicated zero-valent-copper (ZVC), a metal with positive reduction potential (E 0 Cu = +0.334 V, SHE) (Bratsch, 1989), as very reactive against other aromatic organochlorines, namely 4-chlorophenol (Duan et al., 2016) and atrazine (Hollanda et al., 2019), thus contradicting the expected trend and deserving further investigation. For the exception of ZVI, all these metals were only tested in the suspension form, requiring separation afterwards. To avoid this inconvenience, the best approach for field application of this kind of technology is in permeable reactive barriers, in which ZVI has gained much attention (Guan et al., 2015;Zhang, 2003;Dorathi and Kandasamy, 2012). To test the feasibility of this operational design in the dechlorination of TCP with ZVC, in the present investigation different ZVC-containing materials, including some derived from scrap, were used in a packed-column through which a TCP solution was recirculated. The efficiency of these materials was compared with similar ZVI-containing materials. Moreover, the effect of some typical influencing factors on ZVM-mediated reactions, such as metal origin, metal dosage, contaminant concentration and water matrix constituents were also explored and optimized for the system containing ZVC derived from scrap. The degradation products resulting from the reaction between this material and TCP were identified by liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS), not only to infer possible degradation mechanisms, but also to predict their toxicity based on a quantitative structure activity relationship (QSAR) model. Lastly, these toxicity predictions were compared with real toxicity tests performed with Vibrio fischeri.

Experimental procedure
A glass column, with inner diameter of 1.5 cm and length of 16 cm, packed with glass balls (diameter of 0.1 cm) and reactive material, was used to perform the experiments (Fig. 1). A 50-mL TCP solution with the desired initial concentration was pumped through the column using a peristaltic pump, at a flow rate of 2 mL min −1 . Every 15 min (time that the solution takes to pass through the system), a 300-L sample was collected for analysis, over 150 min of reaction, totalizing 10 complete cycles.
The reactive material was selected between ZVI powder (ZVIp), commercial iron wool (ZVIw), scrap steal wires (ZVIsw), ZVC powder (ZVCp), commercial copper wires (ZVCcw), or copper wires from obsolete electrical cables (ZVCow). Iron-containing materials were acid washed with a 0.4% (v/v) H 2 SO 4 solution for 5 min in an ultrasonic bath and rinsed with deionized water before use. This procedure was only performed with iron containing materials, since these materials had a thick oxide layer (ZVIp) or zinc protection (ZVIsw) that could affect the reaction between TCP and the reactive metal (ZVI). Experiments to compare metals reactivity were conducted under the same experimental conditions, using 10 mg L −1 of TCP and 16.4 g of the respective metal placed in the middle of the column, as schematized in Fig. 1. The metal wires were cut in pellets, with average lengths of 1,72 ± 0,26 cm (copper) and 1,31 ± 0,20 cm (iron).
To optimize TCP degradation with ZVCow, TCP concentration and metal dosages were varied according to an experimental design, as described in Section 2.5. All the experiments were performed at free pH, which varied from about 5 to 6 during the experiments.
For the reusability tests, deionized water passed through the system (glass balls + ZVCow) before starting a new degradation cycle, in order to avoid any TCP accumulation from one cycle to another. After washing, the column was dried with compressed air.

Analytical methods
Copper content of each ZVCow lot was determined by the acid digestion of 1 g of this material, following the methodology described by Skoog and Leary (1994).
TCP degradation was monitored by high pressure liquid chromatography (HPLC) system (Shimadzu, series 20) equipped with a C18 column (ACE, 250 mm × 4.6 mm, 5 m) and a diode array detector (SPD-M20A). An isocratic elution consisting of methanol and 1% acetic acid (70:30, v/v), at a flow rate of 1.2 mL min −1 was used. The injection volume was 50 L, the column temperature was maintained at 40 • C, and the wavelength for TCP detection was set at 215 nm. This analytical method gave a TCP retention time of 8.65 min, a limit of detection (LOD) of 0.57 mg L −1 , and a limit of quantification (LOQ) of 1.71 mg L −1 .
The TCP degradation products were identified by HPLC coupled to a hybrid ion trap/time-of-flight mass spectrometer (Shimadzu HPLC/MS-IT-TOF), equipped with an electrospray ionization source (ESI), operated in negative mode and in tandem configuration (MS/MS).
The same aforementioned mobile phase was used, although eluted with a flow rate of 0.2 mL min −1 through a different C18 column (Shimadzu, 250 mm × 2.0 mm, 5 m). Sample drying was performed with nitrogen gas with a minimum purity of 99%, pressure of 180 kPa and 1.5 L min −1 . The electrospray probe (ESI) was operated at 4.5 kV. The curved desolvation line (CDL) interface was operated at 200 • C. Molecules fragmentation in tandem mass mode (MSn) was performed by collision-induced dissociation with Ar (> 99.99%), with a collision-induced dissociation (CID) energy of 50%. The error between experimental and calculated mass-to-charge ratios (m/z) was lower than 1 mg L −1 for all the compounds identified.
TOC and IC analyses were performed in a TOC-L analyzer (Shimadzu).
The metal cation content of the tap water was determined by ICP-OES, using the model 710 equipment from Agilent Technologies, while anion concentrations were determined by ion chromatography, using a Metrohm equipment with a conductivity detector (850 Professional IC 1) and a Metrosep A Supp 5-150/4 column.

Toxicity assays
The toxicity of the samples was determined using the marine bacterium V. fischeri. Before performing any toxicity assay, the sam-ples pH was adjusted between 6.0 and 8.0, followed by filtration through a 0.22 um PVDF membrane to remove any precipitates that could have been formed during pH adjustment. Analyses were carried out in screening mode according to the basic 81.9% Microtox ® test protocol, using a Microtox Model 500 analyzer and Microtox-Omni v. 4.2 software (Modern Water, Inc.). The contact time was 30 min. The results were treated and the evolution of toxicity was expressed in terms of luminescence inhibition percentage, following the established Microtox protocol.

Experimental designs
Firstly, a 2 3 factorial design was followed to investigate the system general trends and especially the influence of ZVCow origin, varying TCP concentration and ZVCow dosage. These three variables were varied at two levels: 5 and 15 mg L −1 for the initial TCP concentration; 6.4 and 26.4 g for ZVCow dosage and lot 1 or lot 2 for ZVCow lot, described in Table 2. Experiments A and H were replicated for statistical validation.
Then, for the ZVCow lot that revealed the best degradation performance, a more detailed investigation was undertaken for ZVCow dosage and initial TCP concentration ([TCP] 0 ), by following a central composite design. Here, the ZVCow lot variable was fixed and the high and low levels of the previous design were maintained. Three replicates of the central point (10 mg L −1 TCP and 16.4 g ZVCow) were used for statistical validation, together with four star points (2.9 mg L −1 TCP and 16.4 g ZVCow, 10 mg L −1 TCP and 30.6 g ZVCow, 10 mg L −1 TCP and 2.2 g ZVCow, 17.1 mg L −1 TCP and 16.4 g ZVCow), see Table 3. The values of ZVCow dosage and TCP concentration that promoted the highest TCP degradation after 150 min were optimized following a response surface methodology (Geravand et al., 2015). Multiple regression analysis based on the least-squares method was performed using STATISTICA 7.0 software, in order to obtain the quadratic response surface models (Eq. (1)), where X 1 and X 2 refer to the codified values of [TCP] 0 and ZVCow dosage, respectively.

Comparison between different ZVI and ZVC-containing materials on TCP degradation
The TCP removal promoted by each reactive material is shown in Fig. 2. TCP degradation profiles are very similar for all the metals tested, enabling an almost complete removal (> 97.5%) at 150 min of recycling through the column, except for ZVIp that promotes a slower degradation compared with the other materials tested. ZVCcw reveals a slightly better performance than the other metals.
In order to provide a fair comparison between metals reactivity, different surface areas have to be taken into account, given their role during the reductive reaction between the zero-valent-metal and the organochlorine. For that, the observed degradation kinetic constants (k obs ) have to be normalized with respect to the metal surface area, modifying the equation suggested by Johnson et al. (1996), see Eq. (2): Where k SA ' is the normalized rate constant, to be compared for the different metal-containing materials; k obs is the observed degradation rate constant and a s is the specific surface area of metal Table 1 Values of k obs , R 2 , specific surface area and k SA ' for each metal-containing material.
Material 2 nd order k obs R 2 Surface Area k SA ' (L 2 min −1 m −2 ) (L g −1 min −1 ) (as, m 2 g −1 )   (m 2 g −1 ), which was obtained by BET analysis. All the degradation profiles fitted better to second-order degradation kinetics (see Supplementary Material, Figs. S1 and S2); therefore, the second-order k obs was used to calculate the k SA ' (Table 1).
Results show that iron containing materials (ZVIp, ZVIw and ZVIsw) present higher k SA ' values, indicating that they are more reactive than copper-containing materials. This was expected, since iron has a lower standard reduction potential than copper (E 0 Fe = −0.440 V, SHE; E 0 Cu = +0.339 V, SHE) (Bratsch, 1989), meaning that it donates electrons more easily and, consequently, provides a faster TCP reduction. Studies dealing with the reductive degradation of organochlorines under anaerobic conditions are more suitable for comparison purposes, given the limited air exposure of both solution and materials during the degradation experiments performed in this work. Similar results were thus obtained by Tratniek et al. (2010) regarding the reactivity of zero-valent-zinc (E 0 Zn = −0.762 V, SHE) (Bratsch, 1989) in comparison to ZVI towards carbon tetrachloride. On the other hand, Lee et al. (2010) reported very close reductive performances for ZVI and ZVC nanoparticles towards chlorobenzene, which contradicts the general trend in which lower reduction potentials promote faster degradation reactions.
Comparing the materials of the same metal, it was expected to observe higher degradation rates (k obs ) with those presenting higher surface areas, although this was only valid for ZVI-containing materials and for ZVCw in relation to ZVCow. Probably ZVCp did not exhibit the expected degradation rate due to the presence of oxides onto its surface, as confirmed by DRX (see Supplementary Material  Fig. S3), which are known to hamper the reductive reaction (Guan et al., 2015;Lee et al., 2010).
Comparing obsolete/scrap metal with commercial materials of the same metal, a very similar reactivity was observed for ZVCow in relation to ZVCw, and also for ZVIsw in relation to ZVIw, meaning that it is possible to use scrap materials in this kind of reactions to turn this process more environmentally and economically friendly.
Even though ZVC-containing materials have promoted slower degradations than ZVI-containing materials, these are still considered effective for TCP removal given their relatively high degradation rates (in a timescale of minutes) in comparison to those normally reported in the literature for zero-valent-metals (in a timescale of days or hours) (Choi and Kim, 2009;Han et al., 2012). Actually, given the high copper standard reduction potential, it was very unlikely that this metal could promote such a fast TCP degradation, which suggests that some other degradation mechanism than solely surface electron transfer is occurring between these materials and TCP. In fact, Duan et al. (2016) reported that the high degradation efficiency of copper against organochlorines containing aromatic rings in their structures, which is the case of TCP, is due to the conjugation between ZVC electron orbitals and the -bonds of the aromatic ring present in the contaminant structure. Therefore, this ZVC feature would partially explain the relatively high reactivity of ZVC-containing materials against TCP.
Fe and Cu leaching can consist of a potential problem because of the limits for wastewater discharge. However, the amount of these metals leached after 150 min of reaction was quite low (< 2 mg L −1 for Fe and < 0.5 mg L −1 for Cu), and therefore the zerovalent-metal-driven processes studied can be considered safe from the environmental regulation point of view (EPA, 2002).

Evaluation of the influencing factors on ZVCow-driven TCP degradation
Both ZVCow and ZVIsw revealed to be effective in the reductive degradation of TCP, which makes them very promising for practical application both in an economical and environmental point of view, due to the fact that a material is recycled and gains a new economic value, by the same time that a TCP-contaminated water is treated. Between the two materials, we opted to conduct a more detailed investigation for the ZVCow, since no similar studies were found in the literature regarding organochlorines degradation with this kind of scrap material, thus being a more valuable contribution to the knowledge in this area than exploring scrap iron, for which the knowledge is wider (Fan et al., 2011).
Given that, a 2 3 factorial design was firstly followed to evaluate the influence of ZVCow provenience (lot) on TCP degradation, varying both the TCP concentration and ZVCow dosage at two extreme levels, in order to infer about the system general trends. As revealed in Table 2, the second lot of ZVCow (lot 2) generally promoted higher TCP removals after 150 min of reaction. This trend is also clear in the interaction plots shown in Fig. 3A and B. The distinct responses obtained with each ZVCow lots indicate that scrap materials of different origins may result in different TCP removals, which is a non-desirable feature for practical application. However, this can be surpassed as long as high ZVCow dosages are used, as indicated in Fig. 3B. The higher reactivity of lot 2 can be attributed to its higher specific surface area, pore volume and copper content (see Supplementary Material, Table S1).
As expected, higher TCP concentrations led to lower degradations ( Fig. 3A and C), due to an increase in the competition between contaminant molecules for ZVCow active sites. In the same way, an increase in the ZVCow dosage leads to higher TCP percent removals, since there are more active sites for the reductive reaction ( Fig. 3B and C). The analysis of variance (ANOVA) for this experimental  Table S2) shows that the interaction [TCP] 0 ×lot is not significant (95% confidence level).

Optimization of TCP percent degradation
To model and optimize the percentage of TCP degradation as a function of [TCP] 0 and ZVCow dosage, the lot 2 was selected, and a central composite design was followed. The results are shown in Table 3 and the model is described by Eq. (3). The model determination coefficient, R 2 , was equal to 0.996, which shows an accurate fitting of the model to the experimental data. The effects of [TCP] 0 (codified variable X 1 ), [TCP] 0 2 (X 1 2 ), ZVCow (codified variable X 2 ) and ZVCow 2 (X 2 2 ) were the most important for TCP degradation, as indicated by the Pareto chart (Fig. S4 of the Supplementary Material) (95% confidence level). The pure error of the model is equal to 0.66%, confirming the small variability between replicates, close to experimental uncertainties. The response surface and contour plot resulting from this model are shown in Fig. 4A and B, respectively.
TCP % degradation at 150 min = 71.16 − 8.10 X 1 − 6.94 X 2 The model adequacy was evaluated using the Fisher test (Geravand et al., 2015). The ANOVA gives the regression and the lack-of-fit F-values (Table 4). If the regression F-value is greater than the related F tab , the model is considered statistically significant; if the lack-of-fit F-value is lower than the corresponding F tab , the model is considered statistically predictive. Then, if the model matches these two requirements, it is thus considered adequate. The degree of freedom (DF) values for the regression and residual error allow determining the related F tab using the F-distribution table. Similarly, the lack-of-fit related F tab is determined using the DF for the lack-of-fit and pure error sources. By doing so, it is possible to conclude that the model is statistically significant and predictive, thus being considered adequate for the response prediction.
Given that, the model coefficients can be used to foresee the influence of each variable in the degradation process, with the signs (plus or minus) indicating the positive or negative effects on the response factor (TCP% degradation at 150 min). The negative coefficient value for the effect of [TCP] 0 (X 1 ) indicates that the response is favored for lower pollutant concentrations, as already predicted by the 2 3 factorial design. The positive contribution of ZVCow dosage is also confirmed in this experimental design by the positive value of the X 2 linear coefficient. The optimum conditions for this treatment system were calculated using the software STATISTICA 7.0, resulting in X 1 = − 0.58 ([TCP] 0 = 7.1 mg L −1 ) and X 2 = 2.74 (ZVCow dosage =43.8 g), for which 85.5% TCP removal is expected, albeit not possible to be experimentally reproduced due to column size limitations (maximum ZVCow filling possible is 30.6 g). Therefore, the feasible experimental conditions that give the best response were adopted as those of assay 5 (Table 3) and were the values selected to perform further experiments regarding the influence of water matrix.
Again, no considerable copper leaching was detected after each assay, even when high ZVCow dosages were used (approximately 0.5 ppm of leached copper in these cases).

Evaluation of the ZVCow reusability and water matrix influence on TCP degradation
One of the main advantages of packed-column systems in comparison to slurry batch systems is the possibility of reusing the reactive material without expensive and energy-consuming separation processes. The reusability study was carried-out by using a fresh 10 mg L −1 TCP solution every cycle, 16.4 g of ZVCow and performing a simple washing procedure with ultrapure water and compressed air between cycles, just to ensure that the starting point was similar between two consecutive reaction cycles. As indicated by Fig. 5, the degradation efficiency does not seem to be significantly affected over 10 consecutive cycles, although with a slight decrease from an average TCP degradation from about 76% to 70% in the 5 th cycle, followed by stagnation of the degradation efficiency from there on. Other similar studies report the need of regenerating the zero-valent-metal between cycles to maintain its reductive capacity, which requires the use of reactants and is a time-consuming step (Dorathi and Kandasamy, 2012;Mortazavian et al., 2018;Zhao et al., 2008). In this way, ZVCow is considered more attractive for practical application, since despite the small efficiency loss after 5 cycles, it does not require complex regenerating procedures besides washing with water, being able to be reused more than 10 times. The explanation behind the small variation in ZVCow efficiency can rely on the formation of a passivation/protective layer onto ZVCow surface due to copper corrosion, thus inhibiting the electron transfer. This phenomena is often reported as one of the main reasons behind the deterioration of ZVI performance in reducing reactions (Sun et al., 2016).  A test using tap water instead of deionized water was performed under the conditions of run 5 (Table 3), i.e. using 10 mg L −1 of [TCP] 0 and 30.6 g of ZVCow. The results shown in Fig. 6 reveal that TCP degradation was two times slower in tap water than in deionized water (k obs deionized = 2.23 × 10 -3 L g -1 min -1 ; k obs tap water = 1.10 × 10 -3 L g -1 min -1 ), indicating that other water constituents are hampering the reduction reaction. The tap water used in this study contains Ca 2+ , Cl-and organic carbon (see Section 2.1), which are constituents known for affecting zero-valent-metalsmediated reactions due to competition for or blockage of reactive sites (Guan et al., 2015), in agreement with our observations. Also, these same water constituents might be behind the pronounced error bars of the tap water experiment, as they can interact with the metal at different rates every time a replicate is performed, leading to the disparity of the results. Anyway, our results clearly indicate that real TCP-contaminated waters may require more time or higher ZVCow dosages to achieve the target TCP removal, in order to overcome the effects of water constituents.

By-products determination and proposed degradation mechanism
The elucidation of TCP degradation by-products was performed after each cycle of all the degradation experiments, in order to have a better understanding of intermediates formation along the reaction. In addition, a degradation mechanism was proposed, as illustrated in Fig. 7. It is worth observing that TOC values remained constant over time in all experiments. Five different m/z values were detected and associated to each structure, and, in some cases, the same m/z was associated to more than one structure, originating nine different degradation products (Table 5). Total ion chromatogram profiles and mass spectra of each identified m/z are shown in Figs. S5-S8 of the Supplementary Material. The structures of the intermediates proposed were based not only on their molecular ion peak and cleavage pattern, but also on structures already proposed in other investigations dealing with TCP degradation via reductive processes (Dorathi and Kandasamy, 2012;Yazdanbakhsh et al., 2018). It is therefore possible to confirm that ZVCow promotes the progressive dechlorination of TCP. Compounds B1 and B2 are commonly reported as the first TCP dechlorination products, that are formed by the substitution of a chlorine atom by hydrogen (Dorathi and Kandasamy, 2012;Yazdanbakhsh et al., 2018). Products C1 and C2 were hypothesized as possible degradation products derived from aromatic ring reduction of compounds B1 and B2, respectively, since cyclohexanone (structure G) was detected in our samples and it is an indicator of such degradation mechanism (Choi and Kim, 2009). Subsequent dechlorination of C1 or C2 compounds leads to the generation of products D and G, while the dechlorination of compound E would have led to the generation of phenol, which is a common TCP degradation product originated from zero-valent-metals mediated reactions (Choi and Kim, 2009;Dorathi and Kandasamy, 2012). Although the latter has not been detected in our samples, compounds F and H are likely to be derived from the subsequent reduction of the phenol aromatic ring.

Toxicity assessment
More than the target contaminant removal efficacy, it is also important to evaluate the remaining toxicity after treatment, since more toxic by-products may be generated by the treatment itself.
Firstly, serial dilutions of a stock solution containing 100 mg L −1 of TCP in Milli-Q water were tested against V. fischeri, in order to find the effective dose at which 50% luminescence inhibition was achieved (EC 50 ). By a dose-response curve (see Supplementary Material, Fig. S9) it was possible to estimate an EC 50 value of 10.35 mg L −1 , which is a bit higher than that reported in the literature for the same contact time (2438 mol L −1 = 4.815 mg L −1 ) (Altenburger et al., 2000), although of the same order of magnitude, thus being considered similar values (Soares and Calow, 1993).
Then, toxicity assays were carried out with samples resulting from each run of the central composite design, in order to evaluate different degradation stages. The results were compared to the toxicity of the respective parent TCP solution, as illustrated by Fig. 8. It is possible to conclude that, in general, high TCP degradations (% TCP degradation 150 min > 50%), correspond to solutions more toxic that their parent TCP solution. On the other hand, solutions resulting from Exp. 1, Exp, 9, Exp. 10 and Exp. 11 revealed to be less toxic than the respective initial TCP solution. With the exception of Exp. 1, these experiments are associated to low degradation percent removals after 150 min (< 50%). These results suggest that the first TCP degradation products (from product B1 to E) are probably less toxic than the last (from F to H). In order to strengthen this hypothesis, toxicology data of each degradation product was predicted by using the ECOSAR V2.0 software. This software has the capacity to predict the toxicity of a specific chemical towards aquatic organisms based on structurally similar organic chemicals for which experimental information is available. In cases where TCP by-products fitted more than one chemical class (e.g., allyl halides and allyl ketones), the same classification was selected (allyl halides), for the sake of comparison. Only predicted acute toxicity endpoints [LC 50 (96 h) for fish, LC 50 (48 h) for daphnid, LC 50 (14 days) for earthworms and EC 50 (96 h) for green algae] were considered.
The predicted toxicological data suggest that the toxicity of compounds B1 to E would be comparable to or lower than that of TCP. According to ECOSAR, two compounds have different toxicities if the predicted values differ by at least one order of magnitude, i.e., one logarithmic unit in Fig. 9; Mayo-Bean et al., 2012. This is coherent with what was previously hypothesized. Compounds F and H are indicated as more toxic than TCP towards daphnids and earthworms, which would explain the increase in the acute toxicity in solutions containing higher concentrations of these by-products. Again, this is coherent with what was hypothesized regarding solutions associated with a high TCP removal, since compounds F and H are among the last TCP degradation products. However, for product G, which is also a final TCP by-product, a lower toxicity is predicted in comparison with that of TCP. Since the opposite was observed in samples containing this product, two conclusions can be drawn based on ECOSAR predictions: (i) product G was not formed in the reaction, thus prevailing the upper degradation route illustrated in Fig. 7, or (ii) product G exists at lower concentrations than compounds F and H at the final degradation stages. In sum, this toxicological evaluation leads to the conclusion that there is no advantage on performing long degradation treatments, since more toxic products can be originated. The ideal operating time would be that capable of converting TCP totally into cyclohexanone (product G). For that, the intermediates formation or toxicity must be performed along the degradation time.

Conclusions
Scrap-derived zero-valent metals revealed to be effective for TCP degradation. A detailed study regarding the influencing factors on ZVCow-mediated degradation indicates that the metal origin, metal concentration, TCP concentration and water constituents  (ions and organic carbon) are important factors affecting the pollutant removal efficacy. Besides being a low-cost material, ZVCow has also advantages of reusability and low leachability, which are very attractive for practical application. The mechanism in which this metal promoted TCP degradation was fundamentally by dechlorination, leading to the progressive generation of less-chlorinated compounds. However, this not necessarily led to less-toxic products, given the higher toxicity that phenol-like compounds present towards aquatic organisms. Therefore, it is recommended that toxicity and/or degradation products should be followed along the reaction to avoid the formation of more toxic solutions. In addition, combined reductive-oxidative processes can also be used for residual toxicity removal.