Raw FA samples used in this research were collected from the main boiler of a local pulp and paper manufacturing company. The source water samples used in the jar tests and column filtration experiments were collected from a local community that highly concerns about the DBPs in their drinking water systems. Other materials and chemicals used in this research are: Nitric acid (HNO₃, ACS reagent, 70%); Sulfuric acid (H₂SO₄, ACS reagent, 95.0% - 98.0%); Hydrochloric acid (HCl, ACS reagent, 37%); Iodine (USP, 99.8% - 100.5%); Potassium iodide (KI, anhydrous, free-flowing, Redi-Dri™, ACS reagent, ≥99%); Potassium iodate (KIO₃, ACS reagent, 99.5%); Copper (II) sulfate (CuSO₄, ReagentPlus®, ≥99%); Potassium phosphate monobasic (KH₂PO₄, powder, ≥99.0%); Sodium phosphate dibasic (Na₂HPO₄, powder, ACS reagent, ≥99.0%); Potassium dichromate (K₂Cr₂O₇, ACS reagent, ≥99.0%); Sodium thiosulfate (Na₂S₂O₃, ReagentPlus®, 99%); Methylene blue (powder, certified by the Biological Stain Commission); Sodium carbonate (Na₂CO₃, ACS reagent, anhydrous, ≥99.5%); Sodium hypochlorite (NaClO, 5% usp solution, A&C Ltd., assay: 4.0% - 6.0%); potato starch (powder). All of them were purchased from Sigma-Aldrich Canada Co., ON, Canada.

The raw FA sample was firstly grinded into powder before characterization. The raw powder sample was then washed with 5% - 10% HNO₃ solution on a carbon to acid ratio of 1 g per 10 mL at 60˚C for 2 hrs to recover unburnt carbon and remove impurities from the FA. After filtration, the particles were cleaned with distilled water for several times to eliminate any acid residue. At last, the filtered clean FA was dried at 110˚C overnight before activation use. Characterization of particle size, pH, moisture, ash content, carbon content, Iodine number (IN), methylene blue value (MBV), and metal elements were conducted for both raw and clean FA samples.

Comparing with chemical activation, physical activation usually generates cleaner activated carbon (AC) products which are more suitable for water treatment technology development. Therefore, this study only focuses on physical activation of this wood-based fly ash. Iodine number, methylene blue value, scanning electron microscopy (SEM), as well as Brunauer, Emmett and Teller (BET) analyses were all used to assess the surface area and porosity development.

In this part, a programmable Lindberg/Blue M tube furnace was used for both carbonization and activation processes. The temperatures examined were 650˚C, 700˚C, 750˚C, 800˚C, 850˚C and 900˚C, respectively. In carbonization process, about 6 grams of the clean FA was firstly heated from room temperature to a desired temperature at a rate of 10˚C/min under N₂ flow (75 cm³/min). Once reaching the desired temperature, the carbonization was continued for another hour. This was then followed by a 1-hour activation process under CO₂ flow (75 cm³/min) at the same temperature. The AC samples generated at different temperatures together with raw and clean FA samples were analyzed for loss of ignition, iodine number, and methylene blue value. The sample with high iodine number and methylene blue value was selected to evaluate the effect of activation time.

In this part of experiment, the clean FA sample was carbonized and activated under the selected activation temperature for 1 hour, 2 hours, and 3 hours, respectively. The AC samples generated in this part were analyzed for iodine number and methylene blue value. The one with high iodine number and methylene blue value was employed in the following water filtration experiments.

After evaluating the adsorption capacity of iodine and methylene blue by fly ash derived AC, real water samples were applied to further evaluate the adsorption of organic matter through batch test. The source water sample was collected from a local community that highly concerns about the DBPs in their drinking water systems. The natural organic matter in the source water was firstly characterized by measuring the total organic carbon (TOC). Other parameters of pH, alkalinity, color, turbidity, ultraviolet (UV) at the wavelength of 254 nm, and total dissolved solids (TDS) were also determined. Jar tests were conducted to evaluate the adsorption of organic matter by different AC doses and for different contact times. In this experiment, different amount of the AC sample was mixed with a series of 500 mL source water at ratios of 0.5 g/L, 1 g/L and 2 g/L. The mixtures were thoroughly mixed on magnetic stirrers for adsorption time of 15 mins, 30 mins and 60 mins, respectively. The mixtures were filtered by 0.45 µm membrane filters and the filtrates were analyzed for TOC and UV. Since natural organic matter is commonly measured as TOC, it is used to quantify the natural organic matter in the water samples. It is believed that some natural organic matter compounds have certain chemical structures that can absorb UV light especially at the wavelength of 254 nm [13] [25] [26]. Hence, UV₂₅₄ is also applied in the quantification of the natural organic matters. Besides AC dose and contact time, temperature and pH can have influences on the adsorption of organic matter as well. However, the adsorption tests in real cases were usually conducted at room temperature, and the pH of the source water does not have much fluctuation in different seasons. The effects of temperature and pH are therefore not considered in this study.

A lab scale fixed bed carbon filtration system was developed. When the water enters the column, and passes through the carbon barrier, the natural organic matter in water can be adsorbed by the activated carbon so that the formation of DBPs can be avoided or mitigated. An acrylic glass column with 30 cm length and 40 mm diameter is made to implement the carbon filtration technology in the lab. In general, a 0.45 µm membrane filter paper is placed at the bottom of the column and 5 g AC sample is then packed into the column. The column top part is connected to a water reservoir which contains water samples to be treated, and the water will pass through the column in a down flow mode. A peristaltic pump is connected at the column outflow to control the filtration speed. A sketch of the filtration system is displayed in Figure 1. A 24-hr column filtration test was conducted to evaluate the adsorption performance of the AC sample. The source water passed through the column in a down flow mode. The flow rate of this test was kept in the range of 6 to 8 mL/min and about 10 L in total of the source water has been filtered within the 24-hr period. In this study, the filtrate samples collected at the column outlet at the 1^st, 2^nd, 3^rd, 6^th, 12^th and 24^th hr were analyzed for both TOC and UV₂₅₄.

In this part of study, the formation potential of two major groups of DBPs (THMs and HAAs) were examined for the local community and for both raw and filtered water samples. A sodium hypochlorite 5% usp solution (A & C Ltd., NaClO assay: 4.0% - 6.0%) was used as a disinfectant. Both raw and column filtered source-water samples were chlorinated by sodium hypochlorite and stored in a dark place for 24 hours. According to the Guidelines for Canadian Drinking

Water Quality (GCDWQ), free chlorine in the majority of drinking-water distribution systems is in the range of 0.04 mg/L to 2 mg/L [8]. To keep the concentration of free chlorine after 24 hours in the same range, different amounts of sodium hypochlorite solution were added to the water samples. The formation of THMs and HAAs was monitored after 1 hour, 4 hours, 12 hours, and 24 hours of chlorination. Meanwhile, the concentrations of four typical THM compounds—chloroform (CHCl₃), bromodichloromethane (CHCl₂Br), dibromochloromethane (CHClBr₂), and bromoform (CHBr₃)—and all nine HAA compounds, which are monochloroacetic acid (MCAA), monobromoacetic acid (MBAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromochloroacetic acid (BCAA), bromodichloroacetic acid (BDCAA), dibromoacetic acid (DBAA), chlorodibromoacetic acid (CDBAA), and tribromoacetic acid (TBAA), were analyzed.

The particle size of raw FA sample was analyzed. Figure 2 shows the particle size distribution of the tested sample. From Figure 2, more than 25% of raw FA has the particle size of 15 to 31 µm; about 60% of it was evenly distributed in the range of 37 to 300 µm; the minimum and maximum particle sizes were 7.8 and 710 µm, respectively.

The characteristics of pH, moisture content, ash content, carbon content, iodine number and methylene blue value for both raw and clean FA samples were examined. Results are shown in Table 1. Since the raw FA sample was cleaned

by HNO₃ solution, the clean FA shows strong acidity while the raw FA is alkaline. The acids left in the FA samples could be removed in the following activation process. Most moisture and ash content were reduced after the cleaning process, which indicates the effectiveness of using nitric acid solution as cleaning reagent. Meanwhile, the reduction of ash content also resulted in the increase of carbon content of the FA sample, which would further benefit the activation process. The iodine number decreased, and the methylene blue value increased after cleaning, which means some impurities in the raw sample reacted with the iodine other than the methylene blue. Accordingly, it is very necessary to conduct the cleaning process before any further treatment so that the evaluation of the carbon adsorption capacity will be accurate.

The metal elements in FA were analyzed before and after the cleaning process, and results are shown in Table 2. The raw FA is rich in calcium, aluminum, iron and magnesium. After cleaning, about 44% of calcium was removed. The metal elements in FA were analyzed before and after the cleaning process, and results are shown in Table 2. The raw FA is rich in calcium, aluminum, iron and magnesium. After cleaning, about 44% of calcium was removed. Also, more than 60% of aluminum, and more than 70% of iron and magnesium were removed from the FA, which illustrates that the cleaning process is efficient in metal removal. Also, more than 60% of aluminum, and more than 70% of iron and magnesium were removed from the FA, which illustrates that the cleaning process is efficient in metal removal.

The fly ash derived activated carbon samples were evaluated by determining the loss of ignition, iodine number and methylene blue value. Results are shown in Table 3. Iodine number and methylene blue value are also shown in Figure 3. From Table 3, the loss of ignition basically increases when the activation temperature increases. The loss in the activation process is mainly due to the reaction between C and CO₂ [27]. When the activation temperature increases, the reaction will be accelerated, which will therefore result in a higher loss.

Iodine number and methylene blue value are two parameters that are commonly used to assess the performance of activated carbon. Iodine adsorption implies the micropore content of an AC sample, while the adsorption of methylene blue can give a hint about the mesopore structure [28]. From Figure 3, it is clearly that both iodine number and methylene blue value increase when the activation temperature increases. It illustrates that higher temperature results in higher level of micropore and mesopore development. However, the higher activation temperature applied, the higher ignition loss obtained. For the sample

activated under 900˚C, it has the most micropores and mesopores while the loss of ignition could be higher than fifty percent. It is not cost-effective to generate activated carbon with high energy consumption and loss of ignition. Comparing with activated carbon generated under 900˚C, the one obtained at 850˚C shows relative high level of micropore and mesopore, and relative low level of ignition loss. The 850˚C is then selected to evaluate the effect of activation time.

After testing different activation temperatures, different activation time periods were also examined. Results of iodine number and methylene blue value were shown in Table 4 and Figure 4. The results of iodine number and methylene blue value illustrate that the sample being activated under 850˚C for 2 hours has the most micropores and mesopores. It is hence selected to conduct the adsorption tests on real water samples.

The surface microstructure images of FA-Clean and AC-850˚C-2hr are observed by SEM and shown below. Comparing Figure 5(a) & Figure 5(b) with Figure 5(c) & Figure 5(d), a dense porous structure was formed after activation, which is in accordance with the results of iodine and methylene blue tests. The BET surface area and porosity analysis were conducted to further assess how well the carbon has been activated. As listed in Table 5, there is a significant improvement

on surface area, micropore area and pore volume in the sample after activation. Nitrogen (N₂) adsorption-desorption isotherm curves for both samples are also plotted in Figure 6 and Figure 7, respectively. It is revealed that the adsorption rate of the carbon sample is highly improved after activation, which is also in accordance with the iodine and methylene blue test results.

The characteristics of the source water sample are listed in Table 6. The pH, turbidity and TDS of the analyzed source water are all within the GCDWQ. Only the level of color exceeds that in the guideline. Color is prevalent in natural source waters especially surface waters. The higher level of color indicates more organic matter in the water.

The adsorption capacity of fly ash derived AC was examined through batch tests. TOC removal and UV deduction were investigated and shown in Figure 8 and Figure 10. As can be seen in Figure 8, 75% of TOC was removed after 60 mins adsorption at a dose of 0.5 g/L. The trend of TOC removal at dose of 1 g/L is similar to that at 2 g/L from 15 mins to 60 mins. About 85% of TOC removal was found from 30 mins to 60 mins at both doses, which indicates that the maximum adsorption of TOC is around 85% and it can occur within 30 mins of

(*Source: Newfoundland and Labrador Water Resources Portal).

adsorption. As shown in Figure 10, a 72% of UV₂₅₄ deduction was observed after 15 mins adsorption at the dose of 0.5 g/L, meanwhile a 95% of UV₂₅₄ deduction was obtained after 60 mins at the same dose. For the doses of 1 g/L and 2 g/L, the UV₂₅₄ was dramatically decreased to zero within 15 mins of adsorption, which illustrates the AC is extremely effective in UV₂₅₄ deduction in the adsorption experiments. Comparing TOC removal with UV₂₅₄ deduction at the carbon dose of 1 g/L with 15 mins adsorption time, 74% of TOC was while 100% UV₂₅₄ deduction was detected. As stated earlier, both TOC and removed UV were commonly

used to quantify the organic matter. The difference in the TOC removal and UV deduction results implied that the composition of TOC was not only contributed by the UV-C absorptive organics (e.g., aquatic humic substances), some of its organic components might not be sensitive to the UV light [13] [29] [30]. From another point of view, the generated AC was more effective in the adsorption of light-sensitive organic matters. The same phenomenon was also discovered in the column experiment in the next section.

Figure 9 displays the TOC adsorption isotherm by AC with different doses. From the isotherm, the adsorbed TOC by each gram of activated carbon is 3.7 mg/g, 1.1 mg/g and 0.6 mg/g for doses of 0.5 g/L, 1 g/L and 2 g/L, respectively. The use of 0.5 g (AC) /L (water) in the batch test has the highest adsorption effectiveness and the lowest cost of AC. This dose is thus determined to be appropriate in the following experiment.

From Figure 11, the TOC of the column effluent samples decreases from 8.472 mg/L to 3.2 mg/L in the first two hours, and then increases to 7.8 mg/L at the twenty-fourth hour. The TOC in column effluent could have such a trend is mainly due to the unstable flow rate. Since the AC used in column test is in powder form, it can cause blocking problems sometime in the middle of the experiment. The flow rate is usually high in the beginning of the filtration, and it will gradually become slow. When the flow rate slows down, the contact time between carbon and water sample will be extended. Therefore, the TOC in the column effluent was decreasing in the first few hours.

Comparing the original source water with the column effluent sample at the end of the second hour, more than 60% of the organic matter has been adsorbed or removed by the carbon barrier. After 2 hours, the organic matter in the column effluent starts to increase but still below the original level, which implies that the adsorption rate of the carbon barrier is decreasing. After 24 hours, the TOC of the column effluent sample is close to 8 mg/L, which reveals that the carbon barrier becomes almost saturated.

From Figure 12, the result of UV₂₅₄ has a similar trend as that of TOC, because of the blocking problem in the column. The UV₂₅₄ of the column outflow

sample decreases from 0.271 A to 0.04 A within three hours filtration, and then it increases to 0.164 A at the end of twenty-four hours. The TOC of the 24^th-hour sample is about 92% of the original water sample, which indicates the carbon barrier is 92% saturated. However, the UV₂₅₄ of the same sample is only 61% of the original water sample, which implies the carbon barrier is only 61% saturated. The big difference in the saturation degree from both results could possibly because some natural organic matters in the source water do not have UV-absorbing chemical structures. It also proves that the AC is more effective in adsorbing light sensitive organic matters. This phenomenon is also observed in the batch test, which may hint that the TOC results are more reliable than the UV results in this study.

Figure 13 displays the total THM concentrations of both raw and column filtered water samples during chlorination; the total THM concentrations of all the samples increases as chlorination time increases. For the source water, the majority of THMs are formed within 12 hours of chlorination. For the column filtered water samples, THM concentrations gradually increase at relatively lower rates during the whole chlorination process. The source water has a THM concentration of 935 µg/L after 24 hours, while the THM concentrations of the column 6-hour, 12-hour, and 24-hour effluent samples are about 404 µg/L, 345 µg/L, and 521 µg/L, respectively. This indicates that column filtration is effective in NOM removal and can significantly reduce the amount of THM formed during chlorination. Among the column 6-hour, 12-hour, and 24-hour effluent samples, the 12-hour sample holds the lowest level of THM throughout chlorination; this implies that the 12-hour sample contains the least NOM. This is possibly due to the blocking problem that occurred in the middle of the column filtration, as blocking can lead to longer contact time between water and carbon

barrier and more NOM can be adsorbed. Table 7 shows that chloroform plays a predominant part in the total THM concentrations. In comparing the source water and the column filtered water results, it is interesting to find that the bromine-containing THM compounds are only formed in the raw water but not in the effluent water samples during chlorination. Figure 14 presents the total HAA concentrations of Sunnyside raw and column filtered water samples during different periods of chlorination. As with the THM results, the total HAA concentrations of all samples continuously increase within 24 hours. The raw water sample always shows a higher total HAA than that of column filtered effluent samples during chlorination; this proves that carbon filtration can prohibit

HAA formation. In a comparison of all the column filtered effluent samples, the HAA of 6-hour column effluent sample is close to that of the 12-hour column effluent sample, and they are slightly lower than the 24-hour column effluent sample. This reveals that the adsorption rate of the carbon barrier slowly decreases in the column filtration process. Even though carbon filtration can result in lower DBP concentrations, when both THM and HAA concentrations of the raw and column effluent water samples at the 24^th hour of chlorination are considered, THM formation was prohibited at a higher degree than the HAA formation. Therefore, column filtration is more efficient in reducing THM formation potential than that of HAA. The concentrations of each of the nine HAA compounds for all chlorinated water samples are listed in Table 8, with DCAA exhibiting the highest concentration. TCAA is also an important component in total HAAs. The formation of BCAA, BDCAA, DBAA, CDBAA, and TBAA is very limited during chlorination for all samples. In some, such as source water,

the MCAA concentration decreases in the first 12 hours, but those of DCAA and TCAA increase. This phenomenon may suggest that MCAA is transformed into DCAA and/or TCAA during chlorination and indicate that more C-H bonds in acetic acid were substituted by Cl atoms and formed C-Cl bonds with the presence of free chlorine. A similar phenomenon found in other water samples demonstrates that a longer contact time can lead to a higher degree of chlorination.