Assessing trichloromethane formation and control in algal-stimulated waters amended with nitrogen and phosphorus

Nitrogen (N) and phosphorus (P) enrichments can stimulate algal growth in drinking water 27 sources, which can cause increased production of disinfection byproduct (DBP) precursors. 28 However, the effect of systematic N and P enrichments on DBP formation and control has not 29 been adequately studied. In this work, we enriched samples from a drinking water source – 30 sampled on April 5, May 30, and August 19, 2013 – with N and P to stimulate algal growth at 31 N:P ratios covering almost five orders of magnitude (0.2-4,429). To simulate DBP-precursor 32 removal processes at drinking water treatment plants (DWTPs), the samples were treated with 33 ClO 2 followed by alum coagulation prior to free chlorine addition to assess the DBP formation 34 potential (FP). Trichloromethane (TCM) was the predominant DBP formed and the TCMFP was 35 the highest at intermediate N:P molar ratios (~10-50), which corresponded with the peak in algal 36 biomass, as measured by chlorophyll- a (Chl- a ). Algal biomass was P-limited throughout the 37 study period, and co-limited by N for the August 19 sampling set. The differences in TCMFP 38 between the raw and treated waters decreased with increasing P amendment, indicating that ClO 2 39 and alum coagulation became less effective for TCM precursor removal as algal biomass 40 increased. This study highlights the impact of nutrient enrichments on TCM formation and 41 control and has implications for nutrient management strategies related to source water 42 protection and for DWTPs that use source waters increasingly enriched with N and P. Results from study can can used to assess the impact of and

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal's standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

Environmental Science
Processes & Impacts rsc.li/process-impacts Introduction gradient spanned almost five orders of magnitude, while the N and P enrichment gradients 113 spanned more than one order of magnitude each. 114 After N and P amendment, samples were placed in a 30°C water bath under artificial 115 lighting. Lights were controlled by a 12-hour on/off timer and measured to be 500 µmol photons 116 m -2 s -1 during illumination. The cubitainers used were transparent and were inverted during 117 incubation to prevent shading from the opaque lids. Each cubitainer was opened to the 118 atmosphere and shaken daily by hand to aid in aeration and minimize attached growth. Algal 119 biomass was estimated daily as raw water fluorescence measurements using a Turner Design 120 Trilogy fluorometer (Turner Designs, Sunnyvale, CA) at 880 nm. Once the samples had achieved 121 their maximum biomass (~4 days), the cubitainers were shaken vigorously and 2-L were poured 122 into prepared HDPE containers. These containers were stored in the dark at 4°C for DBPFP 123 experiments. The remaining cubitainer volume was divided evenly for analyses of phytoplankton 124 biomass and particulate nutrients. Aliquots were filtered onto Whatman glass fiber filters (GFFs) 125 and stored frozen for measurement of phytoplankton biomass as extracted chlorophyll-a (Chl-a). 126 Chl-a was measured to estimate phytoplankton biomass according to Standard Methods 127 10200 H, 24 with modifications. One filter from each sample was protected from light and 128 transferred to a 15 mL test tube containing 7 mL of 90% acetone solution. The samples were 129 placed in a dark freezer for 24 hours to further enhance pigment extraction. In a dark room, 3 mL 130 of each sample extract were then transferred into disposable test tubes and were analyzed using 131 the Turner Design fluorometer at 880 nm. To adjust for the chlorophyll degradation product 132 pheophytin, each sample was re-measured 90 seconds after addition of 0.1 mL of 0.1 N HCl. scans from 600-to 270-nm were performed on a Shimadzu UV-Vis 2450 (Kyoto, Japan) 146 spectrophotometer using a 1-cm path length low volume quartz cell.

148
Chlorine dioxide was generated using methods described previously. 26 Before dosing, 149 raw water samples were poured into prepared 1-L amber glass screw top bottles and placed in a 150 water bath at 24°C. The stock chlorine dioxide concentration was measured by absorptivity at 151 360-nm after dilution with Milli-Q water, using an assumed molar absorptivity of 1,225 M -1 cm -152 1 . The nutrient amended samples generated from source water collected on May 30, 2013 were 153 dosed with chlorine dioxide at 1 mg L -1 , whereas the August 19 samples were dosed at 2 mg L -1 .

154
After dosing, samples were capped headspace-free and placed in the dark at room temperature 155 for 24 hours. and then diluted to a lower concentration (between 2-and 4 g L -1 as Cl 2 ) for dosing with a 187 micropipette. The free chlorine dose required to achieve 7-day chlorine residuals of 3-to 5 mg L -188 1 as Cl 2 was estimated based on raw water DOC. Free chlorine doses were stair-stepped with 189 nutrient loading and ranged from 9-to 22 mg L -1 as Cl 2 . After addition of free chlorine, samples Milli-Q water for measurement of chlorine residual to measure high residuals.

197
Precisely 30 mL of the remaining sample was withdrawn for DBPFP testing as described

204
Algal biomass, nutrient concentrations, and N:P ratios 205 Algal biomass, measured as Chl-a, increased proportionally along the P enrichment 206 gradient when N availability was high in experiments from all three months (Fig. 1a). Similarly, Raw water quality results for the April 5 sample collection are shown in Table 1. DOC 219 increased with P dose from an average of 2.26-to 2.77 mg L -1 as C, suggesting the increased 220 algal biomass (Fig. 1a)  given the aromatic carbon fraction has been shown to be a significant source of THM  precursors. 32 In contrast with the trends in P dose, DOC, UV 254 , and SUVA did not change 225 across the range of N doses. Taken together, these results suggest P-limited growth for the April 226 5 sampling set, which is consistent with the biomass data (Fig. 1). The free chlorine residuals 227 after 7 days (FC-7d) were between 4-and 7 mg L -1 as Cl 2 , with no trends based on the N or P  in all 6 cases across the P gradient, but only in 3 of 5 cases across the N gradient. This indicates 237 that DOC produced by N enrichment was more resistant to removal by alum coagulation. It is 238 worth noting that the average FC-7d residuals in Table 2 were between 10-and 16 mg L -1 as Cl 2 , 239 above the target window of 3-5 mg L -1 as Cl 2 for the DBPFP tests. Ongoing experiments in our 240 laboratory suggest these higher residuals will enhance formation of chlorinated THMs at the 241 expense of bromine-substituted species and haloacetonitriles.

242
Raw and treated water quality results for the August 19 sample collection are shown in 243   Table 3. For the P-gradient, the raw water DOC ranged from 2.96-to 3.35 mg L -1 as C, but in  THMs.

259
As expected based on the high free chlorine residuals (Tables 1, 2 samples. Fig. 2c shows that treatment with 1 mg L -1 ClO 2 increased the average TCMFP relative 283 to the raw waters for the lowest two N amendments, and was similar to the raw waters for the 284 higher N doses. Fig. 2d  August 19 samples, a ClO 2 dose of 2 mg L -1 decreased the average TCMFP by 20-30 µg L -1 288 across the N amendments (Fig. 2e) and 22-47 µg L -1 across the P amendments (Fig. 2f). Further, 289 Fig. 2f shows that the differences in TCMFP between the raw and ClO 2 treated samples 290 decreased with increasing P amendment, presumably because the biomass produced (Fig. 1a) 291 exerted a demand for ClO 2 , more so than directly contributing to the TCM precursor pool. Alum coagulation following ClO 2 treatment lowered the average TCMFP, an expected 293 result based on previous research. 26 The one exception to this trend occurred for the May 30 294 samples at an N amendment of 1000 µg L -1 (Fig. 2c), in which the average TCMFP values were 295 similar for both treatments. Fig. 2d shows that alum coagulation decreased the average TCMFP 296 by 34-64 µg L -1 compared to ClO 2 -only, but the difference between treatments decreased as the 297 P amendment increased. For the August 19 samples, alum coagulation decreased TCMFP by 10-298 20 µg L -1 relative to ClO 2 -only for both nutrient amendments ( Fig. 2e and f)  concentrations that were 150% greater (e.g., 300-700 µg L -1 ) than the highest value in the GC 311 standard curve, one sample had no measurable FC-7d residual, and the other sample was 312 determined to be an outlier during the PARAFAC modeling process. The comparatively strong 313 TCMFP:DOC correlation (r 2 = 0.72, Fig. 4a) was unexpected because ClO 2 treatment increased 314 DOC (Tables 2 and 3) but decreased TCMFP (Fig. 2). The high TCMFP:UV 254 correlation (r 2 = 0.88, Fig. 4b) is in agreement with prior research, 34 supporting the contention that released DOC 316 from nutrient stimulated biomass was both low in aromatic carbon and did not contribute 317 significantly to the pool of TCM precursors. The comparatively weak correlations between 318 TCMFP and the fluorescence metrics ( Fig. 4c and 4d) were unexpected based on previous 319 research 26, 29 and suggest that dissolved species present in the samples from the nutrient 320 enrichments (e.g., algal extrudates and intracellular organic matter) may have interfered with 321 fluorescence measurements more so than UV 254 .

323
The experiments presented here demonstrate that nutrient-driven increases in algal 324 biomass reduced the effectiveness of two common DBP-control measures, ClO 2 oxidation and 325 alum coagulation. Algal biomass in nutrient amended waters was shown to be P-limited for the 326 April 5, May 30, and August 19 sampling sets, with an N co-limitation for the August 19 327 samples. For the nutrient amended raw waters, algal biomass, measured as Chl-a, was a 328 maximum at molar N:P ratios of ~10-50, which following chlorination corresponded to a 329 measurable increase in the TCMFP. Oxidation of the sample waters with chlorine dioxide 330 increased the DOC with aromatic-depleted compounds that were not significant TCM precursors.

331
Across the experimental P-gradient, the differences in TCMFP between the raw and ClO 2 +alum 332 coagulated waters decreased with increasing P amendment, indicating the algal biomass exerted 333 a demand for ClO 2 and alum. Results from this study can be used to guide nutrient management 334 strategies for source water protection and can be used by DWTPs to assess the impact of N and P 335 enrichments on TCM formation and control.  Table 1 for details on N:P ratio.   Table 1 for details on N:P ratio. Dashed lines encompass the upper and lower 95% prediction intervals for the linear models. DOC is the dissolved organic carbon, UV 254 is the ultraviolet absorbance at 254 nm, I 344/425 is the fluorescence intensity at an excitation of 344 nm and an emission of 425 nm, and C2 F MAX is the maximum fluorescence intensity for PARAFAC Component 2 (see Table 4 for description of the fluorescence-PARAFAC components). Seven samples (out of 244) were excluded from this figure because they were determined to be outliers as described in the Results and Discussion -DBPFP section.  FC-7d = free chlorine residual after 7-day hold time; N:P = molar nitrogen to phosphorus ratio based on amended doses, with the exception of two values (4429 and 0.2) which were calculated using the initial background concentrations of 2,700 µg N L -1 and 11 µg P L -1 (initial molar N:P = 539); NA = not applicable. Note: Free chlorine was dosed after all other reported measurements.  Values in parentheses are secondary and tertiary maxima; r 2 values describe the linear correlations between trichloromethane formation potential (TCMFP) and the fluorescence maximum values (F MAX ) for each parallel factor (PARAFAC) component