Comparison of ferrate and ozone pre-oxidation on disinfection byproduct formation from chlorination and chloramination

12 This study investigated the effects of ferrate and ozone pre-oxidation on disinfection 13 byproduct (DBP) formation from subsequent chlorination or chloramination. Two natural 14 waters were treated at bench-scale under various scenarios (chlorine, chloramine, each with 15 ferrate pre-oxidation, and each with pre-ozonation). The formation of brominated and 16 iodinated DBPs in fortified natural waters was assessed. Results indicated ferrate and ozone 17 pre-oxidation were comparable at molar equivalent doses for most DBPs. A net decrease 18 in trihalomethanes (including iodinated forms), haloacetic acids (HAAs), dihaloacetonitrile, 19 total organic chlorine, and total organic iodine was found with both pre-oxidants as 20 compared to chlorination only. An increase in chloropicrin and minor changes in total 21 organic bromine yield were caused by both pre-oxidants compared to chlorination only. 22 However, ozone led to higher haloketone and chloropicrin formation potentials than ferrate. 23


33
While crucial in the prevention of waterborne disease, drinking water disinfection with 34 chlorine has caused concern as more than 600 disinfection byproducts (DBPs) with  (Diehl et al., 2000). However, chloramines can cause other water quality and operational 49 problems and their use has not always been viewed as the best option. Instead, many 50 utilities have chosen to make process changes to improve the removal of DBP precursors, 51 and use of pre-oxidation features prominently in these strategies. Application of oxidants 52 (e.g. ozone) can help by partially oxidizing DBP precursors within natural organic matter 53 (NOM), and by having a beneficial impact on subsequent processes that affect final DBP 54 concentrations (Camel and Bermond, 1998). 55 The use of oxidants, and ozone in particular, for abatement of THM precursors was an 56 early area of research even in advance of the 1979 THM rule. Many of these early studies 57 were poorly documented (Rice, 1980;Rosen, 1980;Stevens and Symons, 1976) and as a 58 result, it's difficult to draw quantitative conclusions from this work. Subsequent research 59 on the role of radical scavengers and chlorination protocols (Reckhow et al., 1986)   Ozone was found to destroy chlorination DBP precursors with the efficacy order of 67 DHANs > THMs ≈ THAAs > DHAAs, and also decreased the yields of DHAAs and THMs 68 from chloramination (Hua and Reckhow, 2013;Wert and Rosario-Ortiz, 2011). However, 69 pre-ozonation followed by chlorination was found to increase CP formation by 2−10 times 70 in different waters (Hoigné and Bader, 1988;Jacangelo et al., 1989). 71 Pre-or intermediate ozonation may affect the performance of downstream treatment 72 processes and thereby further impact final DBP concentrations. Although studied by 73 several groups, the many effects of pre-ozonation on coagulation is still not easy to 74 characterize or predict (Becker and O'Melia, 1995;Edwards and Benjamin, 1992; 75 Reckhow et al., 1986;Schneider and Tobiason, 2000). Perhaps most important are the 76 impacts on granular media filtration, either due to changes in biodegradation or adsorption. 77 It has long been known that precursors to many DBPs are subject to biological degradation 78 on acclimated media filters (Tobiason et al., 1993). Accordingly, ozone has been shown to 79 enhance precursor removal in subsequent filtration much as it does removal of bulk NOM 80 (Speitel et al., 1993). This impact is substantial and it affects precursors to different DBPs  producing fewer regulated halogenated byproducts as compared to other oxidants more 87 commonly deployed in water treatment (Sharma, 2013(Sharma, , 2011(Sharma, , 2010. Also, ferric iron 88 resulting from ferrate decomposition may support coagulation (Goodwill et  to a small-scale drinking water treatment system showed positive impacts on finished water 95 quality without negative impacts on downstream processes .

96
Compared to ozone, ferrate oxidation seems to produce lower levels of bromate due to 97 the slow reaction rate between ferrate and bromide (Jiang et al., 2016b). Ferrate can be 98 considered a simpler alternative to ozone, as it can be produced offsite without ancillary 99 systems, or produced onsite using common water treatment inputs. In-situ electrochemical  The reactions of strong oxidants like ozone with NOM and the ways in which these 109 reactions propagate through subsequent treatment processes are difficult to study without 110 simplifying the systems in some way. A reductionist approach has been commonly used 111 whereby sub-components of full treatment system are been studied in isolation, and then 112 the pieces assembled to get a better picture for the complete system. This study represents 113 a reductionist study of the direct impacts of pre-oxidation on DBP precursors.

114
The principal objectives of this research were (1)  were conducted just once, due to the large experimental matrix. As such, only larger (> 147 10%) differences between conditions is discussed, supported by monotonic changes from 148 multilevel tests. The pH was adjusted to 7.0 by dropwise addition of sodium hydroxide 149 (NaOH) or sulfuric acid (H2SO4) solutions as needed, prior to the preoxidation step. For 150 each raw water, two typical doses of ferrate and ozone (low and high, see Table 2) were 151 added under rapid mixing (G ~ 350 sec -1 ). The pH was monitored during preoxidation and 152 only adjusted during rapid mixing if it deviated ± 0.1 from the set point, which occurred 153 infrequently, following a previously published procedure  an ozone and oxygen mixture through 2-L chilled (~6 degrees C) reactor filled with Milli-162 Q water. The Milli-Q water was made slightly acidic (Reckhow et al., 1986) to decrease 163 potential decay of O3 via the formation pathway of OH-radical (Staehelin and Holgné, 1982) 164 using ~50 µL of 500 mM nitric acid, and also shielded from light.  replicates measured in parallel (see Text S1, Tables S1 -S4 and Figure    pre-oxidation (see Figure S3). In addition, bromine is more reactive with hydrophilic and  concentrations (see Figure S4).

L -F e ( V I) /C l2 H -F e ( V I) /C l2 L -O 3 /C l2 H -O 3 /C l2
hydrophilic and low MW precursors produced more I-THMs, whereas hydrophobic and 449 high MW precursors were more reactive with iodine in TOI and UTOI formation. Our 450 results also showed that the specific TOI and UTOI yields (normalized to the DOC 451 concentration) were higher for the GL water, whereas the specific I-THM yields were 452 similar for the two waters. Without pre-oxidation, the UTOI/TOI ratio was also higher for 453 the GL water, and decreased with the iodide concentration.  were observed at 0.2 mg/L iodide (see Figure 6). The UTOI/TOI ratio generally increased 469 by pre-oxidation, which was due to the greater decreases in I-THM yield than the TOI yield 470 caused by pre-oxidation. Ferrate pre-oxidation led to slightly higher UTOI/TOI ratio than   bromochloramine is very labile and reactive (Valentine, 1986). Therefore, 491 bromochloramine may play important roles in DBP formation and the products are similar 492 to those produced by free bromine (Diehl et al., 2000). This formation of reactive 493 bromochloramine might also have caused the higher DBPFPs at elevated bromide levels 494 in the NW water (see Figure S7). 495 Figure S8 shows the change in yield of THM, DHAA, and DHAN derived by pre-496 oxidation compared to chloramination only (see Figure 7 for concentrations in GL water). should be similar to that for enhanced THM formation, either due to the formation of free 531 bromine or increased reactivity of NOM toward bromochloramine or chloramine. Figure   532 S8 shows that different doses of ferrate and ozone increased the DHANFP for the GL water, 533 and decreased the DHANFP in the NW water at the higher dosages. These large differences 534 between the two waters indicated that the effect of pre-oxidation on DHAN formation from 535 chloramination was also greatly affected by water quality. Hua and Reckhow (2013) also 536 found that DHAN precursor removal was site-specific, and the ability of ozone to destroy 537 DHAN precursors depends on water quality and precursor properties. Ferrate and ozone 538 pre-oxidation also increased the BSFs of THMs, DHAAs, and DHANs with 539 chloramination (see Figure S9).  The dominant HK species with chloramination was DCP, differing from the chlorination 546 result. TCP was below detection limit due to the inability of chloramine to produce 547 trihalogenated byproducts (Hua and Reckhow, 2007a). Figure 8 shows that for DCP, the 548 lower doses of ferrate slightly increased the DCP formation potential (DCPFP). DCPFP 549 decreased with increasing ferrate dose. In contrast, the DCPFP was increased by ozone pre-550 oxidation under all conditions. The lower doses of ozone increased the DCPFP by 34 and 551 74%, and the higher doses of ozone increased the DCPFP by 54 and 130% for the GL and 552 NW waters, respectively. These results indicate that the byproducts from oxidation, e.g. 553 ketones, might be able to react with chloramine forming DCP. Ozone led to higher DCPFP 554 than ferrate under all conditions. 555 Similar to chlorination, the CP formation from chloramination was also increased by 556 both pre-oxidants. Ozone pre-oxidation generally led to higher CP yields than ferrate. The 557 CP formation potential (CPFP) of the GL water without pre-oxidation was below detection 558 limit. Ferrate and ozone pre-oxidation followed by chloramination produced CP of 0.19 559 and 0.13 µg/L at the lower doses, and 0.25 and 0.33 µg/L at the higher doses, respectively.

560
For the NW water, both the low and high oxidant doses yielded increased CPFFP. CP yield 561 for ozone was higher than that of ferrate, with the high ozone dose more than doubling the 562 CP yield. produced some amounts of iodoform (CHI3), which was not detected with chlorination (see 571 Figure 6). Total I-THM yields for the GL were slightly lower with chloramination than 572 with chlorination, whereas chloramine produced more I-THMs than chlorine in the NW 573 water (compare Figures 6 and 9). Chloramine oxidized iodide to iodine and did not further 574 oxidize iodine to iodate (Hua and Reckhow, 2007a;Kumar et al., 1986). In contrast, free chlorine was able to partially oxidize iodide to iodate and thus decrease the I-DBPFP.

576
Iodinated THMs could form during chlorination, especially when there was significant 577 competition from NOM for iodine (Hua and Reckhow, 2007b). Therefore, no consistent 578 trend regarding whether chlorine or chloramine produced more I-THMs was observed. In 579 contrast, the TOI yields were much higher with chloramination than with chlorination for 580 both waters, attributable to the greater oxidation of iodide to iodate by chlorine than greater for the GL water, than for the NW water. As previously noted, hydrophobic and 584 high MW precursors are more reactive with iodine in TOI and UTOI formation. Therefore, 585 the specific TOI and UTOI yields with chloramination were also higher in the high SUVA 586 GL water than the NW water.

587
Ferrate and ozone greatly decreased the formation potential of I-THMs, UTOI, and TOI, 588 by oxidizing iodide to iodate. For I-THMs, the formation of CHCl2I was decreased by pre-589 oxidation, whereas some increases in CHBr2I yield were observed.    The relative cytotoxicity resulting from Fe(VI) and O3 generally followed a similar pattern 606 (see Figure S10) proportional to the concentration of each DBP. Toxicity of regulated 607 THMs (e.g. THM4) and iodinated THMs following chlorination were decreased by 608 increasing dosages of both ozone and ferrate. THM4 following chlorination was much 609 lower following chloramination, however, preoxidation with both O3 and Fe(VI) increased

617
Ferrate and ozone pre-oxidation were generally comparable at equivalent doses for the 618 abatement of numerous DBP precursors, following both chlorination and chloramination.

619
The estimated relative cytotoxicity of THM4 and I-THMs resulting from Fe(VI) and O3   Text S1. HAA and THM analyses were conventional, following USEPA 552.2 and 551.1.

929
These methods are well established with typical method detection limits and limits of 930 quantification well below the concentrations measured in this study (see Tables S1 through   931 Tables S3, and Figure S1. Standard deviations are commensurate with the method detection 932 limits. Figure S1 includes additional bromodichloromethane measurements collected    Figure S1. Linear relationship between mean concentration and standard deviation for 948 replicate bromodichloromethane measurements.