Chlorination and bromination of 1,3-diphenylguanidine and 1,3-di-o-tolylguanidine: Kinetics, transformation products and toxicity assessment

Journal Pre-proof Chlorination and bromination of 1,3-diphenylguanidine and 1,3-di-o-tolylguanidine: Kinetics, transformation products and toxicity assessment Benigno J. Sieira, Rosa Montes, Arnaud Touffet, Rosario Rodil, Rafael Cela, Herv´ e Gallard, Jos´ e Benito Quintana PII: S0304-3894(19)31544-4 DOI: https://doi.org/10.1016/j.jhazmat.2019.121590 Reference: HAZMAT 121590 To appear in: Journal of Hazardous Materials Received Date: 30 July 2019 Revised Date: 17 October 2019 Accepted Date: 31 October 2019 Please cite this article as: {doi: https://doi.org/ This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier. This is the postprint (accepted manuscript) version of the article published in the Journal of Hazardous Materials https://doi.org/10.1016/j.jhazmat.2019.121590 © 2019. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Page 1 CHLORINATION AND BROMINATION OF 1,3-DIPHENYLGUANIDINE AND 1,3-DI-O- TOLYLGUANIDINE: KINETICS, TRANSFORMATION PRODUCTS AND TOXICITY ASSESSMENT Benigno J. Sieira 1, Rosa Montes 1, Arnaud Touffet 2, Rosario Rodil 1, Rafael Cela 1, Hervé Gallard 2*, José Benito Quintana 1* 1 Department of Analytical Chemistry, Nutrition and Food Sciences, Institute of Food Analysis and Research (IIAA), Universidade de Santiago de Compostela, R/ Constantino Candeira S/N, 15782 – Santiago de Compostela (Spain) 2 Institute de Chimie des Milieux et des Matériaux de Poitiers (IC2MP), École Nationale Supérieure d’Ingénieurs de Poitiers (ENSIP), Université de Poitiers, 1, rue Marcel Doré, TSA 41105, 86073 – Poitiers (France) * Corresponding authors: Hervé Gallard: [email protected] José Benito Quintana. [email protected] Graphical abstract Journal Pre-proof Page 2 Highlights  DPG and DTG react rapidly with chlorine and bromine  The pH dependance of the reaction was modelled  Several transformation products were identified by LC-QTOF  Chloroform and dichloroacetonitrile are also produced  TPs are more toxic than DPG and DTG Abstract This works investigates the chlorination and bromination of two rubber and polymer related chemicals, which have emerged as relevant water contaminants, i.e. 1,3-di-o- tolylguanidine (DTG) and 1,3-diphenylguanidine (DPG). Kinetic constants at different pH values were obtained and modelled, taking into account the pKa values of DTG/DPG and HClO, showing that the maximum reaction rate (kapp > 104 M-1 s-1) is obtained at pH values 8.8 for DPG and 9.1 for DTG. Bromination is also very fast, although unlike chlorination, deviation from the model was observed at neutral pH, which was Journal Pre-proof Page 3 attributed to formation of metastable transformation product (TP). A total of 35 TPs, corresponding to halogenation, hydroxylation, formation of monophenylguanidine derivatives and cyclization reactions, were tentatively identified. Furthermore it was found that chloroform can be formed up to a 25% molar yield, while dichloroacetonitrile was formed into less than a 3% yield. Several ecotoxicological endpoints were predicted by quantitative structure–activity relationship models (QSAR) for the TPs, some of which were predicted to be more toxic than DPG/DTG. Also a chlorinated solution investigated by a Vibrio Fisheri acute toxicity test, confirmed that toxicity increases with chlorination. Keywords Halogenation; transformation products; high-resolution mass spectrometry (HRMS); ecotoxicity; disinfection by-products. 1. Introduction Polar organic compounds, if persistent, spread along the water cycle, even becoming a human health problem if that substances reach drinking waters (Hernández et al. 2015, Reemtsma et al. 2016, Schulze et al. 2019). 1,3-Di-o-tolylguanidine (DTG) and 1,3-diphenylguanidine (DPG) are chemicals used as accelerators in the vulcanization processes of rubber and other polymers manufacture, with a registered production in Europe, according to the REACH dossiers, in the 100-1000 tons (DTG) and 1000-10,000 tons (DPG) (ECHA 2018a, b). However, so far, little information about their Journal Pre-proof Page 4 environmental (particularly water) occurrence is available. A first study from Montes et al. (Montes et al. 2017), in the frame of the project PROMOTE(Metcalfe et al. 2003), reported DTG occurring in environmental water samples across Europe by liquid chromatography-high resolution mass spectrometry (LC-HRMS) screening. Later on, within the same project, it has been shown that DPG as well as DTG occur in different water compartments at the ng L-1 level (Montes et al. 2019, Schulze et al. 2019). Zahn et al. have recently shown that, when considering natural processes, DPG (DTG was not considered in that study) photolyzes and reacts with manganese oxide, but does not biodegrade and is stable to hydrolysis (Zahn et al. 2019). Furthermore, this compound has been identified in drinking water in China at 0.7 mg L-1, migrating from polyethylene pipes (Tang et al. 2015) and, more recently, has also been recently identified as being the major chemical leaching from tire wear particles (Hübner et al. 2019, Zahn et al. 2019). Furthermore, there is some literature that describes DTG and DPG toxicity and pharmacological activity in mice and rats (Jaramillo-loranca and Esram 2015, Lamy et al. 2010). Yet, the possible reaction of both chemicals with chemical oxidants used in drinking water treatment plants (DWTPs) and wastewater treatment plants (WWTPs) has not been studied so far. Several disinfection techniques and processes are employed in DWTPs and WWTPs. Among them, chlorine is the oxidant used in the vast majority of DWTPs in Europe, and also in some WWTPs to a minor extent (Benitez et al. 2011, Quintana et al. 2014). Although chlorine is effective to inactivate bacteria, the formation of possible harmful transformation products (TPs), including well-known disinfection byproducts (DBPs), as e.g. trihalomethanes, needs to be taken into account (Acero et al. 2013, Postigo and Richardson 2014, Quintana et al. 2014, Rodil et Journal Pre-proof Page 5 al. 2012). Such TPs (including DBPs) may in some cases be more (eco)toxic than the precursor chemicals themselves (Postigo and Richardson 2014, Quintana et al. 2014). Therefore, their identification is necessary in order to obtain a relevant interpretation of the reaction. Thus, the aim of this work was to perform a comprehensive study about the chlorination of DTG and DPG in water. This includes a kinetic study and modelling, identification of TPs (including those formed by bromination, since hypobromite is rapidly formed from bromide into solution during chlorination (Benitez et al. 2011)) by LC-HRMS, quantification of the yield of known DBPs formed and preliminary (eco)toxicological assessment. Journal Pre-proof Page 6 2. Materials and methods 2.1. Chemicals and stock solutions DTG (99%) and DPG (97%) were purchased from Sigma-Aldrich (Steinheim, Germany) and stock solutions were prepared in ultra-pure water. Ultra-pure water was obtained directly in the lab from a Milli-Q Gradient A-10 system (Millipore, Bedford, MA, USA). All solutions and dilutions necessary for the experiments were done in ultra-pure water until desired concentration. Sodium hypochlorite (8-14% Cl2), ammonium chloride (> 99%) and potassium phosphate dibasic trihydrate (> 99%) were obtained from Sigma-Aldrich. Sodium thiosulfate (99.5%) and potassium bromide were from ACS Acros Organics (Thermo Fisher Scientific, Waltham, MA, USA) and potassium di-hydrogen phosphate (99.5%) was from Panreac (Barcelona, Spain). Standard solutions of chloroform and haloacetonitriles (HANs) were prepared from EPA 551B Halogenated Volatiles Mix supplied from Supelco. The exact nominal free chlorine content employed was regularly determined spectrophotometrically by measuring the hypochlorite anion absorption at 292 nm (ɛ = 350 L-1 cm-1) (Johnson and Melbourne 1996) of the stock solution (pH >10). 2.2. Real samples Two samples were used to study the extent of the chlorination reaction with a real matrix. A surface water sample was collected from the River Sarela in Santiago de Compostela (pH 6.8, Dissolved Organic Carbon: 2.42 mg L-1, chloride: 7.98 mg L-1, bromide: 0.043 mg L-1). A wastewater effluent was collected from a WWTP comprising a primary and a secondary conventional sludge treatment (pH: 7.5, Dissolved Organic Journal Pre-proof Page 7 Carbon: 14.1 mg L-1, chloride: 23.1 mg L-1, bromide: 0.075 mg L-1). Dissolved organic carbon was measured with a Shimadzu 5000A TOC analyzer (Duisburg, Germany), while bromide and chloride were determined with a Metrohm 850 Professional Ion Chromatograph (Zofingen, Switzerland). 2.3. Chlorination experiments Chlorination of DTG and DPG were performed individually in 100 mL amber closed vials at room temperature. Also, experiments without chlorine were prepared as a control. Experiments to study chlorination kinetics were performed in a similar way, but with lower compound concentrations (1 µM), an excess of chlorine (10 µM, 20 µM or 50 µM) and different pH of sample (5-12) being considered in 10 mM phosphate buffer and NaOH for very basic pHs. Aliquots of 1 mL were taken at different reaction times and the reaction stopped with 20 µL of 0.01 M sodium thiosulfate before residual concentration of guanidine was analysed by liquid chromatography-photodiode array detection (LC-PDA). Ammonium chloride (1 mM) was used in some experiments as “soft” quenching method as being selective to free chlorine and as to avoid any Na2S2O3-induced back reaction that could interfere with determination of rate constant (Dodd and Huang 2004). An experiment was also performed without stopping the reaction and aliquots were manually injected at different reaction times using a Rheodyne valve in the LC-PDA system. Experiments were performed at room temperature (22 ±1°C). The pH was measured before and after the experiment, and variation was less than 0.1 unit. Free active chlorine was analysed by DPD colorimetric method (Clesceri et al. 1998) at the end of Journal Pre-proof Page 8 reaction time. Chlorine consumption was usually below 10% and pseudo-first-order plots were always linear (see Figure S1 for examples). Additional experiments for the identification of TPs (performed in triplicate) were carried out with a similar procedure. For these experiments, ultrapure water adjusted at pH 7.0 was used, spiked with the compound at 10 µM, and initial chlorine dose set to 100 µM. TPs were identified after reduction by ascorbic acid for reaction times of 30 s, 1 min, 2 min, 5 min, 10 min and 30 min. DBPs formation potentials (chloroform and haloacetonitriles) were determined for a reaction time of 2 days at pH 7.0 in ultrapure water with an initial concentration (of either DPG or DTG) of 10 µM and molar chlorine to guanidine ratios of 1, 10 and 100 in headspace-free conditions. 2.4. Bromination experiments Bromination experiments were performed under the same conditions as chlorination, in order to determine apparent rate constants and detect TPs that can be produced in bromide containing waters. Bromine was generated in the lab according to the procedure described by Benitez et al. (Benitez et al. 2011). Briefly, bromine was produced from the reaction between 9 mM HOCl and 10 mM potassium bromide. The yield of this reaction was followed spectrophotometrically (hypobromite anion maximum absorption wavelength at 329 nm with an ɛ = 332 M-1 cm-1 at pH above 11.5) (Benitez et al. 2011). Kinetics of bromination were studied as for chlorination for pH values ranging from 5 to 9 using direct method in batch reactor with an excess of bromine (10 to 100 µM) compared to 1 µM DPG or DTG solution (see Figure S2 for examples of bromination of Journal Pre-proof Page 15 The value of the rate constant k was determined by a non-linear least-square regression of the experimental pH profile of the kapp values using Sigma Plot 11.0 (Systat Software Inc., San Jose, CA, USA). The pH dependence of kapp is shown in Figures 1a and 1b for DPG and DTG, respectively. For both compounds, the pH profile exhibits a maximum between pH 8 and pH 10. This maximum corresponds to the concomitant presence of both HOCl and neutral guanidine. The maximum pH value is equal to the average value of the pKa of HOCl and guanidines i.e. 8.8 for DPG and 9.1 for DTG. As shown in Figure 1, the model fits well with the experimental data considering only the reaction of HOCl with neutral guanidine. No improvement was obtained by including the reaction of ClO- with neutral guanidine and the reaction of HOCl with protonated guanidine (see Text S1 and Table S5), which is in accordance with the literature (Deborde and von Gunten 2002). The rate constants, k, of the reactions between HOCl and DPG and DTG neutral species determined from model fitting to the experimental values are 4.1 (±0.3) × 106 M-1 s-1 and 2.6 (±0.1) × 107 M-1 s-1 for DPG and DTG, respectively. The apparent rate constants at neutral pH (~103 M-1 s-1) and intrinsic rate constants (k) of neutral species (~106 - 107 M-1 s-1) are in the range of rate constants of secondary amines with chlorine (Deborde and von Gunten 2008). However, N-chloroamino compounds were not detected during kinetic experiments and identification of TPs would indicate that initial reactive site is the aromatic ring. Lower rate constant of 19 M-1 s-1 was obtained for ethyl guanidine at pH 7.2 – 7.4 (Pattison and Davies 2001), which can be explained by the stronger basic character of alkyl guanidines. 3.2. Bromination kinetic study Journal Pre-proof Page 16 The apparent rate constants of bromination determined by using direct and competition kinetics methods are listed in Table S3 and S4 for DPG and DTG, respectively and are plotted versus pH in Figure 2. Examples of pseudo-first-order and competition kinetics plots are given in Figure S2 and Figure S3 for DPG. The apparent rate constants range from 76 to 2.89 x 105 M-1 s-1 for DPG and from 36 to 5.87 x 104 M- 1 s-1 for DTG. In contrast to chlorine, lower rate constants were determined for DTG, which could be attributed to steric effects between the bulky bromine atoms and the methyl groups in DTG. The experimental results fit well with the proposed model at pHs below 6 and above 9, while a strong deviation and even discrepancies between rate constants determined by the direct kinetics method and the competition kinetics method using BP as reference compound in the pH 7 – 9 range. Such deviations are attributed to a metastable TP with oxidizing properties which could not be identified by LC-HRMS in that pH range (see detailed discussion in Text S2 and Figures S3-S6). Excluding those pH values, and compared to chlorination, maximum kapp values are slightly shifted to higher pH values due to the higher pKa of HOBr (pKa = 8.8). Calculated intrinsic rate constants for the reaction of HOBr with neutral DPG and DTG were 8.3 (±0.4) x 106 and 5.5 (±0.7) x 106, respectively. While HOBr reacts usually much faster than HOCl with organic compounds (Heeb et al., 2014), the rate constant of HOBr with DPG was only twice as high as the reaction of HOCl with DPG and the rate constant for the reaction of HOBr with DTG was lower than that of HOCl. Steric hindrance (as mentioned), different reactive sites and type of reaction (oxidation vs Journal Pre-proof Page 17 substitution) might explain this unexpected result, which requires further investigation. 3.3. Transformation products Identification of TPs was performed for both DTP and DPG considering chlorination and bromination. Experiments were carried out as described in sections 2.3 and 2.4. All samples were analysed in the LC-QTOF equipment as detailed in 2.6. The proposed structures of the TPs are presented in Figures 3 and 4. Further details on formulas, mass errors and scores of the TPs, as well as individual structures derived from the interpretation of MS/MS spectra are presented in Tables S6 and S7. TPs were named with the precursor compound abbreviation followed by the nominal mass of its [M+H]+ ion. As it can be observed the empirical formula could be proposed with a high degree of certainty, with score values higher than 95% and mass errors lower than 5 ppm, except for DTG-274, whose score was 79% and mass error was 9.7 ppm due to its low intensity. The proposed structures are based on the interpretation of the MS/MS spectra, which are presented into Figures S7 and S8, for DPG and DTG TPs, respectively. Moreover, DPG-136 (i.e. monophenylguanidine) was unequivocally identified by purchasing its authentic standard from Sigma-Aldrich. There are four main types of reactions occurring during chlorination, i.e. ipsochlorination to produce monoguanidine derivatives, introduction of chlorine atoms (bromine when samples are brominated) into an aromatic ring, hydroxylation and intramolecular cyclization. Thus, DPG-136 was easily identified by its spectrum (Figure S7), and because an authentic standard was available, as mentioned. This TP further Journal Pre-proof Page 18 reacts by halogenation to the corresponding chlorinated or brominated derivatives (DPG-170 and DPG-214), easily identified because their MS/MS spectra is similar to DPG-136 and exhibit the halogen isotopic pattern (Figure S7). DPG-119 is also produced at long reaction times. Hydroxylation of DPG produces DPG-228, while halogenation produces DPG-246 (chlorination) and DPG-290 (bromination), all of them easily identified by their MS/MS spectra (Figure S7). A key TP is DPG-210, which a similar empirical formula than DPG itself but with one further double-bond equivalent (i.e. 2 atoms of H less), Table S6. Its MS/MS spectrum exhibits first the loss of ammonia to m/z 192.0671 and also the elimination of CH3N2 to m/z 167.0720 from the protonated molecular ion (Figure S7e). This second ion would not be possible unless a cycle is formed. Because of this, we hypothesize here that DPG-210 corresponds to the structure shown in Figure 3, by formation of the 7- membering cycle from DPG-228 (hydroxylated DPG) by elimination of water. In fact the maximum intensity of DPG-228 was observed at 0.5 min and then its intensity rapidly drops, while DPG-210 maximum is reached at 1 min and then drops more slowly (see Figure S9a). The fact that DPG-210 has also been observed as a photolysis TP by Zahn et al. (Zahn et al. 2019), although no structure was proposed in that publication, further supports this hypothesis. Once DPG-210 is formed, this molecule further reacts to yield a mono-hydroxylated-TP (DPG-226), a hydroxychloro-TP (DPG-260) or a dichloro,hydroxy-TP (DPG-294). In the case of DTG, the reaction was similar to DPG, as expected, but a larger number of TPs could be identified (23 vs. 11 TPs). In general, the main difference in the TPs produced is that a greater degree of hydroxylation and halogenation is observed, e.g. Journal Pre-proof Page 19 DTG-308 (a dichlorinated derivative of DTG), which can be explained by the electrondonor effect of the methyl group on aromatic ring. Although no MS/MS spectra (Figure S8) was obtained for all TPs, due to the low intensity of some of them, structures compiled into Figure 4 and Table S7 (those TPs without MS/MS data are marked with a * symbol in the Table) were assigned on the basis of the MS/MS spectra (when available) and by analogy to DPG TPs. As regards of the most relevant TPs, Figures S9b and 10b present the normalized amount of each TP and their precursors for a reaction with 100 µM chlorine and up to 30 min of contact time. Normalization was performed by using the signal of the original compound (DPG or DTG) as a surrogate to calculate an approximate yield, except for DPG-136, DPG-119, DPG-170, DTG-150, DTG-166, DTG-184 and DTG-228, where monophenyl-guanidine (DPG-136) was used instead, as being considered structurally closer. In the case of DPG (Figure S9), the most intense TP is DPG-136 with a yield of ca. 5% at 0.5-2 min, dropping down to 3% at 30 min. The second most relevant TP is DPG-246 (monochloro-DPG), with a yield of ca. 3% at 0.5-2 min, dropping down to ca. 0.2% at 30 min. It is noteworthy that at 30 min, there was no DPG detectable and the sum of all TPs intensities would approximately represent a 5% yield. This could likely be attributed to the formation of ring-opening products like chloroform with a relatively high yield (see 3.4) and the uncertainty of the semiquantitative approach, due to the lack of authentic standards for most TPs. In the case of DTG (Figure S10), the most intense TP is DTG-254 (hydroxylated cyclic product) with a yield of ca. 30% at 0.5-2 min, followed by DTG-290 (hydroxy,chloro- DTG) with a yield of ca. 15% at 0.5-2 min. These two TPs are also the most relevant at Journal Pre-proof Page 20 30 min, representing an estimated yield of 4 and 8% respectively. Also, in the case of DTG, the total yield of TPs at 30 min is ca. 20%. 3.4. Disinfection by-products Subsequently to TPs identification, we investigated the formation potential of classical DBPs (chloroform and HANs). Chloroform was measured as the only trihalomethane that can be formed without bromide and representing the most relevant group of DBPs, while HANs is another important group of DBPs which can be produced from N- containing chemicals, as it is the case of DPG and DTG. Figure S11 shows the molar yields of CHCl3 and dichloroacetonitrile (DCAN), the only HAN detected, produced from the chlorination of DPG and DTG after 2 day reaction time. Similar yields were obtained for both guanidines. For CHCl3, the yield of ca. 25% for a molar guanidine/Cl2 ratio of 1:100 is similar to CHCl3 yields ranging from 10 to 32% already described for hydroxylated and chlorinated aromatic compounds (Gallard and von Gunten 2002). This is consistent with the formation of chloro or/and hydroxy DPG and DTG that further react with chlorine leading to CHCl3 as end-product after ring cleavage. Among HANs, only DCAN was detected at significant levels and with yields much lower than CHCl3. For a ratio of 1:100, molar yields were 4.4% and 2.3% for DPG and DTG, respectively. 3.5. Reaction in real sample matrices The reactivity of both guanidine compounds was tested by spiking two real matrices (a surface water and a wastewater effluent) with 1 µM (i.e., ca. 200 µg L-1) of either DPG or DTG and 10 µM chlorine and the reaction kinetics followed for 20 min (reaction quenched with ascorbic acid) by LC-HRMS. Journal Pre-proof Page 21 In the case of the surface water, DPG and DTG reacted rapidly, being below 1% of their initial concentration after 2 min (Figure S12a-b). Conversely, with the effluent wastewater (with a higher TOC), still an 87% of DPG and 73% of DTG remained after 20 min (Figure S12c-d). Indeed, formation of TPs is easier to happen during drinking water production than by chlorination of wastewater (after secondary treatment). Even so, the chlorinated wastewater was analysed for the TPs previously identified in ultrapure water and several of them could be detected. The amount of bromide in those samples is very low (<0.1 mg L-1, see 2.2), therefore no brominated TPs were detected. When excluding those brominated TPs, 6 out of 9 TPs where detected for DPG (DPG-136, DPG-228, DPG-210, DPG-226, DPG-260 and DPG-294) and 7 out 16 where detected for DTG (DTG-150, DTG-256, DTG-290, DTG-239, DTG-254, DTG-270 and DTG-288). 3.6. (Eco)toxicity assessment To obtain a preliminary estimation of the ecotoxicological implications of the chlorination reaction, the US-EPA TEST software was used in order to predict the toxicity of the two guanidines and their TPs. This prediction was performed only for Daphnia Magna LC50 (48 h), Tetrahymena Pyriformis LC50 (48 h) and oral rat LD50 (as a proxy of human toxicity), since the software was unable to produce an estimation for other endpoints. The results obtained are summarized in Tables 1 and 2. As it can be appreciated, oral rat toxicity LD50 values are in the 602-805 mg Kg-1 bw for the two guanidines, which would classify them as Category 4 (i.e. the less toxic category) according to the ECHA Guidance (ECHA 2017). The TPs would also be classified as Category 4. Hence human toxicological hazard is expected to be low. Journal Pre-proof Page 22 The predicted acute aquatic toxicity endpoint values lie in the 5-28 mg L-1 and 3-6 mg L- 1 ranges, for DPG and DTG respectively (Tables 1 and 2). Thus, they would not be classified as Category Acute 1 (the only acute aquatic toxicity category) according to the ECHA Guidance (ECHA 2017). As regards the TPs, the predicted aquatic toxicity of the monoguanidine TPs is lower than the precursor guanidines for the crustacean Daphnia Magna, while it could not be predicted for the fish T. pyriformis. On the other hand, particularly TPs which are halogenated are predicted to be more toxic. Thus, DPG-294 and 14 TPs from DTG (see Table 1 and 2) would have a predicted toxicity endpoint <1 mg L-1 and would thus be classified in the Category Acute 1 for aquatic organisms. Care must be taken with these data, since the third trophic level (algae) toxicity could not be predicted and values obtained for algae can likely result into more ecotoxicity. In fact DTG is classified in the REACH dossier as Aquatic Acute 1 (ECHA 2018a). Furthermore, the acute toxicity of DPG and DTG was assessed using the bioluminescent Vibrio fisheri test. The EC50 and EC20 values of Vibrio fisheri test were estimated from a series of geometrical dilutions with dilution factors from 1 to 256 and initial guanidine concentration of 100 mg L-1. Figure S13 shows two dose-response curves of DPG after an incubation time of 30 min at 15°C. The initial EC20 of DPG was 40 ±2 mg L-1 and the EC50 was estimated (as detailed in Text S3) to be 245 ±33 mg L-1 from extrapolation of the dose-response curves. The toxicity of DTG was lower and only an EC20 of 80 mg L-1 could be determined. These results confirm that both guanidines have a low acute aquatic toxicity. Due to solubility limitations, the effect of chlorination on acute toxicity Journal Pre-proof Page 23 was only tested with a DPG solution of 40 mg L-1 corresponding to the EC20. Chlorination was performed with chlorine doses of 40 and 400 mg Cl2 L-1 (i.e. molar Cl2/DPG ratio of 3 and 30). Toxicity tests were conducted after the absence of chlorine residual was checked. Results in Figure 5 shows that the bioluminescence inhibition strongly increases from 14% before chlorination to 45 and 99% for Cl2/DPG ratios of 3 and 30, respectively. Similar results were generally observed in the literature after chlorination and were assigned to more toxic halogenated TPs (El Najjar et al. 2013, Tawk et al. 2015). Even though the increase of toxicity could not be assigned to specific TPs/DBPs, results of bioluminescent Vibrio fisheri test were in agreement with predicted aquatic toxicity endpoints obtained by QSAR. 4. Conclusions DPG and DTG rapidly react with chlorine and bromine at natural water pH values. This reaction leads to the formation of several TPs via ipso-halogenation, hydroxylation, halogenation and cyclization. Several of these TPs are predicted to be more toxic than the original guanidine compounds, which was confirmed by measuring the acute toxicity of a chlorinated mixture by a Vibrio Fisheri acute toxicity assay. Moreover, chlorination leads to the production of the traditional/regulated DBPs chloroform and, to a minor extent, dichloroacetonitrile, when the molar ratio of chlorine to DPG/DTG is high. Declaration of interests Journal Pre-proof Page 24 ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Acknowledgements This work was supported by the Water Challenges for a Changing World Joint Program Initiative (Water JPI) Pilot Call (ref. WATERJPI2013 – PROMOTE), funded by the Spanish Ministry of Economy and Competitiveness/Spanish Agencia Estatal de Investigación (refs. JPIW2013-117 and CTM2017-84763-C3-2-R) and French Office National de l’Eau et des Milieux Aquatiques (ref. PROMOTE). We also acknowledge the Galician Council of Culture, Education and Universities (ref. ED431C2017/36), Région Nouvelle Aquitaine and FEDER/EDRF funding. Journal Pre-proof Page 31 pH 4 6 8 10 12 kapp (M-1 s-1) 1e+1 1e+2 1e+3 1e+4 1e+5 DPG experimental DPG model a) pH 4 6 8 10 12 kapp (M-1 s-1) 1e+1 1e+2 1e+3 1e+4 1e+5 DTG experimental DTG model b) Figure 1. pH dependence of the experimental and modelled apparent rate constants of chlorination of (a) DPG and (b) DTG. Journal Pre-proof Page 32 pH 4 6 8 10 12 kapp (M-1 s-1) 1e+1 1e+2 1e+3 1e+4 1e+5 1e+6 DPG model pH 4 6 8 10 12 kapp (M-1 s-1) 1e+1 1e+2 1e+3 1e+4 1e+5 1e+6 DTG model a) b) Figure 2. pH dependence of the experimental (symbols) and modelled apparent rate constants of bromination of (a) DPG and (b) DTG. Direct kinetic method (circle) and competitive kinetic method using 4-bromophenol (square) or 2,4,6-tribromophenol (triangle) as reference compounds were used for kapp determination. Only full symbols were used for model calculation. Journal Pre-proof Page 33 Figure 3. Schematic representation of DPG TPs Journal Pre-proof Page 34 Figure 4. Schematic representation of DTG TPs Journal Pre-proof Page 35 Figure 5. Evolution of bioluminescence inhibition measured using Vibrio fisheri Microtox® test during DPG chlorination. [DPG]o = 40 mg Cl2 L-1, molar DPG:Cl2 ratios of 1:3 and 1:30, chlorination time of 4 days at pH 7.0. Journal Pre-proof Page 36 LIST OF TABLES Table 1. QSAR Predicted toxicity values for DPG and its TPs. Daphnia magna T. pyriformis Oral rat LC50 (48 hr) (mg/L) IGC50 (48 hr) (mg/L) LD50 (mg/kg) DPG 5.09 28.5 805 DPG-119 16.4 np 493 DPG-136 30.6 np 500 DPG-170 11.5 np 455 DPG-210 5.29 11.4 908 DPG-214 1.14 9.53 1320 DPG-226 2.46 10.4 1143 DPG-228 5.24 21.1 2433 DPG-246 1.22 7.21 886 DPG-260 1.89 4.78 1241 DPG-290 1.14 9.53 1320 DPG-294 0.75 2.88 1019 np: no prediction possible Journal Pre-proof Page 37 Table 2. QSAR Predicted toxicity values for DTG and its TPs. Daphnia magna T. pyriformis Oral rat LC50 (48 hr) (mg/L) IGC50 (48 hr) (mg/L) LD50 (mg/kg) DTG 2.92 5.53 602 DTG-150 27.5 np 498 DTG-166 23.5 np 993 DTG-184 22.7 np 429 DTG-228 5.00 np 519 DTG-238 3.14 6.35 553 DTG-254 1.68 8.26 1036 DTG-256 3.43 9.19 1259 DTG-270 2.16 6.52 1121 DTG-272 2.52 11.2 2352 DTG-274 0.70 2.29 1178 DTG-286 8.28 9.80 527 DTG-288 0.76 1.98 721 DTG-290 0.93 2.14 1014 DTG-306 0.95 2.06 1870 DTG-308 0.49 1.24 1109 DTG-318 0.39 4.35 1103 DTG-324 0.74 1.11 940 DTG-332 0.51 1.55 1627 DTG-334 0.87 3.21 927 DTG-340 0.31 1.11 1074 DTG-250 0.79 2.38 1485 DTG-396 0.19 1.46 563 DTG-412 0.28 1.21 823 DTG-426 0.03 0.55 384 np: no prediction possible Journal Pre-proof Page1  SUPPORTINGINFORMATIONTO:  CHLORINATION AND BROMINATION OF 1,3‐DIPHENYLGUANIDINE AND 1,3‐DI‐O‐ TOLYLGUANIDINE: KINETICS, TRANSFORMATION PRODUCTS AND TOXICITY ASSESSMENT Benigno J. Sieira 1, Rosa Montes 1, Arnaud Touffet 2, Rosario Rodil1 , Rafael Cela 1, HervéGallard2*,JoséBenitoQuintana1* 1DepartmentofAnalyticalChemistry,NutritionandFoodSciences,InstituteofFood AnalysisandResearch(IIAA),UniversidadedeSantiagodeCompostela,R/Constantino CandeiraS/N,15782–SantiagodeCompostela(Spain) 2InstitutdeChimiedesMilieuxetdesMatériauxdePoitiers(IC2MP)UMRCNRS7285, ÉcoleNationaleSupérieured’IngénieursdePoitiers(ENSIP),UniversitédePoitiers,1, rueMarcelDoré,TSA41105,86073–Poitiers(France) *Correspondingauthors: HervéGallard:herve.gallard@univ‐poitiers.fr JoséBenitoQuintana.[email protected]  Page2  TABLEOFCONTENTS: CONTENT Page TextS1.Fullkineticmodel5 TextS2.Furtherexperimentsinvestigatingguanidinesbromination 7 TextS3.DescriptionofMicrotox®testinhibitioncalculations. 9 TableS1.ExperimentallyobtainedkappforDPGatthedifferentpHvaluesand correspondinghalf‐livescalculatedfor10µMCl2(i.e.0.71mgCl2L‐1). 10 TableS2.ExperimentallyobtainedkappforDTGatthedifferentpHvaluesand correspondinghalf‐livescalculatedfor10µMCl2(i.e.0.71mgCl2L‐1). 11 TableS3.Experimentallyobtainedkapp,kineticsmethodsforbrominationof DPGatthedifferentpHvaluesandcorrespondinghalf‐livescalculatedfor10 µMBr2(i.e.0.71mgBr2L‐1).BPandTBPare4‐bromophenoland2,4,6‐ tribromophenol,respectively.ValuesofkrefareapparentrateconstantsofBP andTBPcalculatedfromHeebetal.(2014). 12 TableS4.Experimentallyobtainedkapp,kineticsmethodsforbrominationof DTGatthedifferentpHvaluesandcorrespondinghalf‐livescalculatedfor10 µM Br2(i.e.0.71mgBr 2L ‐1). BP and TBP are 4‐bromophenol and 2,4,6‐ tribromophenol,respectively.ValuesofkrefareapparentrateconstantsofBP andTBPcalculatedfromHeebetal.(2014). 13 Table S5.SpecificrateconstantsofhalogenationofDPGandDTG determinedbykineticmodellingconsideringsimpleorfullkineticmodel. 14 TableS6.ListofchlorinationandbrominationDPGTPs. 15 TableS7.ListofchlorinationandbrominationDTGTPs. 17 FigureS1.Examplesofpseudo‐fist‐orderkineticsplotsobtainedduringthe chlorinationofDPG(Phosphatebuffer10mM).(a)InfluenceofpHand quenchingmethods([DPG]01µM,[chlorine]010µM).Thereactionwas 21 Page3  stoppedbythiosulfate(fullcircle),ammonium(opencircle)ormanualdirect injectionwasused(opensquare).Linearregressionsareplottedfor reductionbythiosulfate.(b)Influenceofchlorineconcentrations([DPG]01 µM,pH6.1).Thereactionwasstoppedbythiosulfate. FigureS2.Examplesofpseudo‐fistorderkineticsplotsobtainedfor brominationofDPG(Phosphatebuffer10mM).(a)InfluenceofpH([DPG]01 µM).Thereactionwasstoppedbythiosulfate.(b)Influenceofbromine concentrationandquenchingmethod([DPG]01µM,pH5.6).Thereaction wasstoppedbythiosulfate(fullcircle)ormanualdirectinjectionwasused (opencircle).Linearregressionisplottedforreductionbythiosulfate. 22 Figure S3. Determination of apparent second order rate constants for bromination of DPG by using competition kinetics method with (a) 4‐ bromophenol (BP) and (b) 2,4,6‐tribromophenol (TBP) as reference compound ([DPG]05µM,[BP] 0or[TBP] 05µM,[bromine] 00to10µM, phosphatebuffer10mM) 23 FigureS4.DeterminationofapparentrateconstantofbrominationofTBP (pH6.94,[TBP]01µM,[bromine]020µM,10mMphosphatebuffer) 24 FigureS5.DecayofoxidantresponseduringbrominationofDPG(pH6.9, [DPG]050µM,[bromine]05µM,phosphatebuffer10mM) 25 Figure S6. UV/visible spectra of DPG solution before and after bromine addition (pH 6.9, 10 mM phosphate buffer, [DPG]050µM,[Br 2]0 50 µM, reductionbyanexcessofthiosulfate) 25 FigureS7.ChromatogramsandMS/MSspectraofDPGanditsTPs. 26 FigureS8.ChromatogramsandMS/MSspectraofDTGanditsTPs. 38 FigureS9.PlotsummarizingtheformationofTPsfromDPG(10µMDPG+ 100µMCl2)atdifferentreactiontimes:(a)resultsnormalizedtothetime whentheTPreacheditsmaximum;(b)resultsnormalizedbyassumingthat theresponseoftheTPswasequaltoDPG,exceptforDPG‐136,DPG‐119and  Page10  Table S1. Experimentally obtained kappforDPGatthedifferentpHvaluesand correspondinghalf‐livescalculatedfor10µMCl2(i.e.0.71mgCl2L‐1). pH kapp(M‐1s‐1) t1/2(s) 5.0 32 2203 5.6 73 950 5.9 170 408 6.0 194 357 6.1 230 301 6.4 580 120 6.5 670 104 6.7 1310 53 7.0 2400 29 7.5 3923 18 8.0 9760 7 8.4 11061 6 9.0 10776 6 9.5 4567 15 10.0 5999 12 11.0 1269 55 11.7 372 186   Page11  Table S2. Experimentally obtained kapp for DTG at the different pH values and correspondinghalf‐livescalculatedfor10µMCl2(i.e.0.71mgCl2L‐1). pH kapp(M‐1s‐1) t1/2(s) 4.9 25 2803 5.5 148 469 6.0 364 190 6.5 1764 39 7.0 5599 12 7.5 7941 9 8.0 11327 6 9.9 18297 4 11.0 8205 9   Page12  TableS3.Experimentallyobtainedkapp,kineticsmethodsforbrominationofDPGatthe differentpHvaluesandcorrespondinghalf‐livescalculatedfor10µMBr2(i.e.0.71mg Br2L‐1).BPandTBPare4‐bromophenoland2,4,6‐tribromophenol,respectively.Values ofkrefareapparentrateconstantsofBPandTBPcalculatedfromHeebetal.(2014). pH Kineticmethod kref (M‐1s‐1) kapp (M‐1s‐1) t½ (s) 5.00 direct ‐ 76 916.9 5.18 direct ‐ 84 827.0 5.50 direct ‐ 314 220.7 5.63 direct ‐ 195 354.7 5.74 direct ‐ 334 207.5 6.00 direct ‐ 228 304.0 6.83 direct ‐ 344 201.6 7.00 competitionwithTBP 2112 311 222.9 7.08 direct ‐ 363 190.9 7.57 direct ‐ 460 150.7 8.00 direct ‐ 940 73.7 8.00 competitionwithTBP 2842 4276 16.2 8.05 competitionwithBP 26900 16829 4.1 8.50 direct ‐ 962 72.1 8.46 competitionwithBP 50700 61934 1.1 8.95 competitionwithBP 93300 158254 0.4 9.06 direct ‐ 2240 30.9 9.65 competitionwithBP 439000 289338 0.2 10.00 competitionwithBP 245910 246000 0.3 10.48 competitionwithBP 93300 96164 0.7 11.03 competitionwithBP 27700 37977 1.8 11.75 competitionwithBP 5360 8245 8.4 Page13  TableS4.Experimentallyobtainedkapp,kineticsmethodsforbrominationofDTGatthe differentpHvaluesandcorrespondinghalf‐livescalculatedfor10µMBr2(i.e.0.71mg Br2L‐1).BPandTBPare4‐bromophenoland2,4,6‐tribromophenol,respectively.Values ofkrefareapparentrateconstantsofBPandTBPcalculatedfromHeebetal.(2014). pHKineticmethod kref (M‐1s‐1) kapp (M‐1s‐1) t½ (s) 5.22 direct ‐ 36 1918.4 5.55 direct ‐ 48 1438.6 6.04 direct ‐ 113 611.7 6.50 direct ‐ 277 250.3 6.85 direct ‐ 318 217.8 6.99 competitionwithTBP 2112 127 545.8 7.50 direct ‐ 417 166.2 8.02 competitionwithTBP 2842 797 86.9 9.15 competitionwithBP 698000 47857 1.4 9.90 competitionwithBP 294000 58707 1.2 10.55 competitionwithBP 80300 42185 1.6 10.94 competitionwithBP 33900 25430 2.7 11.80 competitionwithBP 4780 8355 8.3  Page14  Table S5. Specific rate constants of halogenation of DPG and DTG determined by kineticmodellingconsideringsimpleorfullkineticmodel.  Fullmodel Simplemodel k1k 2k 3R 2k R 2 Chlorination DPG 4.1x106 1.4x10‐7 2.4x10‐6 0.898 4.1x106 0.898 DTG 2.4x107 2.5x10‐8 3.9x103 0.975 2.6x107 0.960 Bromination DPG 8.3x106 2.2x10‐8 0 0.950 8.3x106 0.950 DTG 5.5x106 1.2x10‐6 0 0.997 5.5x106 0.997 Page15  TableS6.ListofchlorinationandbrominationDPGTPs. Name Experimental m/z Molecular formula Theoretical m/z Error (ppm) Error (mDa) DBE Score (%) Structure DPG 212.1182 C13H13N3‐ ‐‐9‐  DPG‐119 119.0604 C7H6N2 119.0604 ‐0.21 ‐0.03 6 100.00  DPG‐136 136.0867 C7H9N3 136.0869 1.66 0.22 5 99.69  DPG‐170 170.0475 C7H8N3Cl 170.0480 2.67 0.45 5 98.90  DPG‐210 210.1025 C13H11N3 210.1026 0.35 0.07 10 99.97  DPG‐214 213.9968 C7H8N3Br 213.9974 2.99 0.64 5 98.14  N H NH2 NH Br Page16  DPG‐226 226.0975 C13H11N3O 226.0975 ‐0.05 ‐0.01 10 100.00  DPG‐228 228.1131 C13H13N3O 228.1131 0.17 0.04 9 99.99  DPG‐246 246.0795 C13H12N3Cl 246.0793 ‐1.01 ‐0.25 9 99.74  DPG‐260 260.0579 C13H10N3OCl 260.0585 2.38 0.62 10 98.51  DPG‐290 290.0287 C13H12N3Br 290.0287 0.13 0.04 9 100.00  DPG‐294 294.0186 C13H9N3OCl2 294.0195 3.22 0.94 10 96.91  N H N H NH Br Page17  TableS7.ListofchlorinationandbrominationDTGTPs. Name Experimental m/z Molecular formula Theoretical m/z Error (ppm) Error (mDa) DBE Score(%) Structure DTG 240.1495 C15H17N3‐‐9‐  DTG‐150 150.1026 C8H11N3 150.1026 ‐0.18 ‐0.03 5 100.00  DTG‐166 166.0971 C8H11N3O 166.0975 2.35 0.39 5 99.17  DTG‐184 184.0632 C8H10N3Cl 184.0636 2.19 0.40 5 99.17  DTG‐228 228.0131 C8H10N3Br 228.0131 ‐0.06 ‐0.01 5 100.00  DTG‐238 238.1336 C15H15N3 238.1339 1.16 0.27 10 99.68 N H NH2 NH Br HN HN HN Page18  DTG‐254 254.1285 C15H15N3O 254.1288 1.14 0.29 10 99.66  DTG‐256 256.1436 C15H17N3O 256.1444 3.29 0.84 9 97.23  DTG‐270* 270.1238 C15H15N3O2 270.1237 ‐0.36 ‐0.10 10 99.96  DTG‐272 272.1391 C15H17N3O2 272.1394 0.93 0.25 9 99.75  DTG‐274* 274.1079 C15H16N3Cl 274.1106 9.71 2.65 9 79.46  DTG‐286* 286.1182 C15H15N3O3 286.1186 1.47 0.42 10 99.36  DTG‐288* 288.0895 C15H14N3OCl 288.0898 1.10 0.32 10 99.64  HN HN HN OH Cl Page19  DTG‐290 290.1042 C15H16N3OCl 290.1055 4.38 1.27 9 95.41  DTG‐306* 306.0995 C15H16N3O2Cl 306.0931 2.89 0.88 9 97.39  DTG‐308 308.0706 C15H15N3Cl2 308.0716 3.19 0.98 9 96.81  DTG‐318 318.0600 C15H16N3Br 318.0600 0.12 0.04 9 100.00  DTG‐324 324.0657 C15H15N3Cl2O 324.0665 2.46 0.79 9 97.98  DTG‐332* 332.0393 C15H14N3OBr 332.0393 0.00 0.00 10 100.00  DTG‐334 334.0550 C15H16N3OBr 334.0550 ‐0.15 ‐0.05 9 99.99     Figu r   r eS7.Chro m m atograms andMS/M  Page26 SspectraofDPGandi  tsTPs.   Figu r   r eS7.Chro m m atograms andMS/M  Page27 SspectraofDPGanditsTPs.Con t  tinued.   Figu r  r eS7.Chro m m atograms andMS/M  Page28  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page29  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page30  SspectraofDPGanditsTPs.Con t  tinued.   Figu r  r eS7.Chro m m atograms andMS/M  Page31  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page32   SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page33  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page34  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS7.Chro m m atograms andMS/M  Page35  SspectraofDPGanditsTPs.Con t   tinued.  Figu r  r eS8.Chro m m atograms andMS/M  Page42  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page43  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page44  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page45  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page46  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page47  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page48  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page49  SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page50   SspectraofDTGanditsTPs.Con t   t inued.  Figu r  r eS8.Chro m m atograms andMS/M  Page51  SspectraofDTGanditsTPs.Con t   t inued.