Meeting multiple water quality objectives through treatment using locally generated char: improving organoleptic properties and removing synthetic organic contaminants and disinfection by-products

While microbial pathogens typically represent the most immediate threat to health, a wide variety of natural (e.g., cyanotoxins) and synthetic organic chemicals (SOCs) can adversely impact the safety of surface and shallow ground waters collected for drinking in developing communities (Kearns et al. 2014a). Long-term chronic exposure to trace quantities (ppt to ppb levels) of SOCs such as pesticides, pharmaceutical residues, and fuel compounds, and halogenated disinfection by-products (DBPs) can lead to cancer, birth defects, reproductive disorders, endocrine disruption, neurological dysfunctions, damage to internal organs, and other toxic effects (USEPA 2006; CDC 2009/2013). In addition, naturally occurring metabolites of cyanobacteria such as 2-methylisoborneol (MIB) and geosmin, and disinfectants such as chlorine, can impair the organoleptic properties of drinking water. While effective technologies for removal of SOCs and objectionable tastes/odors from drinking water such as adsorption by activated carbon (AC) are currently utilized in high-income regions, low-cost, simple and robust technologies for mitigating these contaminants in developing communities have yet to be deployed. This study and others (Kearns et al. 2014a, 2014b) have demonstrated that, when prepared under appropriate conditions, biomass chars can be as effective as ACs for adsorption of trace organic contaminants. To achieve the goal of providing safe water for all, harmonization is needed between approaches to water treatment in developing communities and the philosophy and capabilities of affluent-world treatment works through establishing parity between chemical, microbiological, and esthetic concerns for water quality. Because char adsorbents can be produced at low cost from local biomass using simple, scalable technologies (Aqueous Solutions 2014; Kearns et al. 2014a), they offer an attainable and environmentally sustainable treatment option for households in rural and developing communities to achieve this harmonization with treatment trains in modern drinking water plants.

The objective of this study was to quantify uptake of (1) trihalomethanes (THMs, the compound class of DBPs typically formed at the highest concentrations); (2) the widely occurring trace level SOC WFN, an anticoagulant pharmaceutical and rodenticide; and (3) MIB, a naturally occurring odorant from surface water by wood-based chars produced under different conditions and in comparison with commercial ACs. To the best of our knowledge, no studies have yet been published describing the adsorption of any of these contaminants by (non-activated) biomass chars.

There is ample evidence that correct and sustained usage of point-of-use (POU) chlorination improves water quality and reduces diarrheal incidence (Quick et al. 1996; Semenza et al. 1998; Sobsey et al. 2003; Chiller et al. 2006). However, surveys have demonstrated attenuation of health benefits within 6 months to 1 year from chlorine interventions (Arnold & Colford 2007; Sobsey et al. 2008; Hunter 2009). Chlorine residual tests performed in households show a high degrees of non-compliance, up to 73% (Quick et al. 1996; Mong et al. 2001; Olembo et al. 2004). Chlorination programs that do not employ continual advocacy by program staff in the form of regular visits, motivational interviews, social marketing, and community mobilization often experience low uptake and/or substantial curtailing of use over time (Kirchhoff et al. 1985; Roberts et al. 2001; Reller et al. 2003; Chiller et al. 2006). Efforts to sustain consistent household chlorination practices are costly and resource-intensive for NGOs and public health agencies (Deb et al. 1986; Semenza et al. 1998; Sobsey et al. 2003). Water esthetics, especially undesirable taste and odor from chlorine, play a major role in the decline of household chlorination practices (Kirchhoff et al. 1985; Roberts et al. 2001; Chiller et al. 2006). Removal of residual chlorine while maintaining water safety could therefore potentially increase the consistency of POU disinfection practices for many populations. AC has long been used for dechlorination (Magee 1956) as it is capable of rapidly reducing free chlorine (HOCl, OCL⁻) to chloride (Suidan et al. 1980). Column studies with AC have demonstrated reduction of up to 2.5–4 g of chlorine (as Cl₂) per gram of AC prior to exhaustion (Suidan & Snoeyink 1975; Suidan et al. 1980). In batch tests, Mahoney (2012) observed chlorine neutralization capacities by chars produced from a biomass cookstove to be 20–56% that of a commercial granular activated carbon (GAC).

For household disinfection, WHO (2011) recommends free chlorine dose of ‘about 2 mg/L to clear water (<10 NTU) and twice that (4 mg/L) to turbid water (>10 NTU)’ while acknowledging that these doses are likely to lead to chlorine residuals that exceed recommended residuals at the point of delivery for water that is centrally treated (0.2–0.5 mg/L), and can lead to formation of elevated levels of DBPs. To our knowledge, only two studies have examined THM formation during household POU disinfection with chlorine in developing communities or emergency situations (Lantagne et al. 2008, 2010). Using sodium hypochlorite solution and local waters collected from rainwater catchment, groundwater, and surface sources (3–7 mg/L TOC) in western Kenya, Lantagne et al. (2008) found that 33% of samples taken 24 hours after chlorination with doses ranging from 1.9 to 3.8 mg/L and which resulted in residual free chlorine concentrations of 0.07–3.64 mg/L exceeded the USEPA MCL for TTHMs. No sample exceeded WHO guidelines for individual or additive THMs. In a follow-up study of hypochlorite and dichloroisocyanurate treated drinking waters collected from ground and surface sources (2–10 mg/L TOC) in Tanzania, Lantagne et al. (2010) again found that no sample exceeded WHO guidelines for individual or additive THMs for chlorine doses of 2–4 mg/L with 24 hours residuals of 0.2–2 mg/L. However, several samples of chlorinated surface waters approached or exceeded the USEPA MCL for TTHMs. In both of these studies, pretreatment of water by filtration using cloth, sand, or ceramic, or settling for 12 hours and decanting did not remove TTHM precursors.

The chlorine doses (2–4 mg/L) and contact times (1, 8, 24 hours) used by Lantagne et al. (2008) did not appear to achieve full DBP formation potential for most of the Kenyan and Tanzanian surface waters. Household and community water treatment in the developing world is characterized by high degrees of variability in disinfectant dosing and contact times. For example, Lantagne (2008) reviewed recommendations by 18 nongovernmental and emergency agencies and identified variations in NaOCl dose from 0.2 to 28 mg/L. Chlorine overdosing, as well as highly variable contact times ranging from several minutes to several days, are therefore likely scenarios that can lead to increased DBP formation.

Contamination of drinking water sources by SOCs such as pharmaceutical residues, a wide variety of biocides, fuel compounds and industrial wastes is a major worldwide problem. A review in Science by Schwarzenbach et al. (2006) indicates that the >300 million tons of SOCs produced annually constitute a major impairment to water quality on a global scale. The residues of human and livestock pharmaceuticals are emerging contaminants frequently detected in wastewater treatment plant effluent, landfill leachate, reclaimed wastewater used for irrigation, and runoff from livestock facilities, and are impacting surface and groundwater worldwide (Kolpin et al. 2002; Boxhall et al. 2003; Barnes et al. 2004, 2008; Sarmah et al. 2006; Ankley et al. 2007; Larsson et al. 2007; Foaczio et al. 2008; Carlsson et al. 2009; Zhao et al. 2010; Luo et al. 2011; Fenet et al. 2012). For example, as highlighted in Nature (Lubick 2009), ‘waste flowing out of a treatment plant near Hyderabad in India pollutes the region's waters with some of the highest levels of pharmaceuticals ever detected in the environment.’

WFN is the most commonly used anticoagulant for the prevention and treatment of thromboembolic diseases (Ansell et al. 2008), and is widely prescribed in both the developed world and developing countries (e.g., in Thailand; Saokaew et al. 2012). WFN is also widely used as a rodenticide, with applications in the domestic, agricultural, municipal, and commercial sectors. It has been detected in aquifers and surface water bodies at ppt-to-low-ppb levels (Kinney et al. 2006; Godfrey et al. 2007). WFN is a polar molecule, and with a pK_a of 5.1, it occurs predominantly in anionic form in most natural waters and as a result is very water-soluble. Prior research on the adsorption of trace level SOCs, including pesticides, pharmaceuticals, fuel and industrial compounds, by AC found that WFN is difficult to adsorb relative to most other SOCs (Kennedy et al. 2015). Thus, if a char adsorbent effectively removes WFN it will likely remove many other SOCs.

MIB is a naturally occurring cyanobacterial metabolite that does not constitute a human health risk. However, at levels as low as 7–10 ng/L (7–10 ppt), MIB imparts a musty flavor and odor to drinking water. In developed countries, AC has proven to be an effective means for controlling MIB-related tastes and odors in natural waters (Suffet et al. 1996). Up to 50% of water treatment plants in the US experience periodic taste and odor episodes associated with blooms of cyanobacteria and/or algae (Suffet et al. 1996); therefore periodic MIB outbreaks are expected to also occur in many developing community drinking water sources, especially considering that eutrophication management to reduce blooms is not as widely practiced. Like WFN, MIB is very water soluble and difficult to remove via adsorption relative to other SOCs (Kennedy et al. 2015). Because MIB is readily detectable by the human nose at very low levels and is more difficult to remove than most other SOCs, it can serve as an indicator for exhaustion of household POU char adsorber units prior to the breakthrough of toxic SOCs under some circumstances (discussed below).

Herein the nonspecific term ‘char’ is used for simplicity. Related terms include ‘biochar,’ which refers to the practice of applying biomass-derived char to agricultural soils in order to improve crop yields and sequester carbon, as well as ‘charcoal,’ which refers to biomass-derived char used as cooking fuel. Char is produced by pyrolysis – the thermal degradation of biomass in an oxygen-limited environment – using a range of reactor configurations including retorts, traditional kilns, and gasifiers (Antal & Gronli 2003).

Locally produced biomass chars have been employed in water treatment for thousands of years (Mason 1916). This traditional practice continues today, particularly in rural areas of the major charcoal producing countries such as Brazil, India, China, Thailand, and throughout SE Asia (UNSD 2011). Much recent research has shown that chars can possess appreciable capacity for sorption of a variety of SOCs including industrial compounds, pesticides, and pharmaceuticals. A list of published studies is presented in Table S1 (a)–(c) (available online at http://www.iwaponline.com/washdev/005/172.pdf). The potential for use of chars as adsorbents in water contaminant management and environmental remediation has recently been reviewed by Zhang et al. (2013a, 2013b), Ahmad et al. (2014), Mohan et al. (2014), and Nartey & Zhao (2014).

The majority of research to date indicates that SOC sorption, normalized to adsorbent mass, is highest in highly micro-porous chars with large internal surface areas generated at temperatures ≥800 °C (Chun et al. 2004; Chen et al. 2008; Uchimiya et al. 2010, 2012; Ahmad et al. 2012; Graber et al. 2012; Han et al. 2013; Wu et al. 2013; Zhang et al. 2013a, 2013b; Zheng et al. 2013). Studies with chars generated from traditional kilns and top-lit up-draft (TLUD) gasifier ovens have shown high-temperature chars to be very effective for adsorbing the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) (Kearns 2012; Kearns et al. 2014a, 2014b) and the sulfonamide antibiotic sulfamethoxazole (SMX) (Shimabuku et al. in preparation). In these batch studies conducted under equilibrium and non-equilibrium conditions, uptake of 2,4-D and SMX by high-temperature chars was equivalent to uptake by commercial AC.

Batch mode equilibrium experiments were conducted to quantify uptake of THMs, WFN, and MIB from a representative simulated surface water by wood-based chars. Chars were produced at different temperatures from a retort and from gasifiers at two scales – a 3.8 L (1 gal) cookstove and a 200 L (55 gal) drum oven. Sorption by chars was compared with sorption by commercial powdered activated carbon (PAC) and GAC.

Low and intermediate temperature retort (‘R___’) chars were produced from eucalyptus wood cut into slats (15 × 10 × 1 cm), placed in a metal bin, covered with sand to exclude oxygen, and heated in a laboratory furnace to 350 °C over 96 hours (‘R350’) and 600 °C over 72 hours (‘R600’). Previous research has established that laboratory chars produced in this manner are representative of chars collected from traditional charcoal ‘beehive’ kilns (Kearns 2012). A 3.8 L (1 gal) TLUD cookstove (‘C___’) gasifier operated in natural draft and forced draft modes was used to generate intermediate (625 °C, ‘C625’) and high (900 °C, ‘C900’) temperature chars from pelletized pine forestry waste (Kearns 2012). A 200 L (55 gal) TLUD gasifier drum oven (‘D___’; Aqueous Solutions 2014) was used to produce char from chopped eucalyptus branches at around 900 °C (‘D900’). Previous studies have established peak temperature and draft conditions within the pyrolysis reactor to be the dominant controlling variables in the adsorbent characteristics of the product, with biomass precursor identity exerting a negligible influence when comparing woody feedstocks (Antal & Gronli 2003; Downie et al. 2009; Kearns 2012; Kearns et al. 2014a, 2014b). Thus, chars made from either pine or eucalyptus woods under a given pyrolysis treatment can be considered to have equivalent sorption capacities. Furthermore, the R350 < C625 < C900 series of chars has been demonstrated to encompass the range of SOC sorption capacities – low, moderate, and high adsorption capacity – expected from chars produced from a variety of woody feedstocks using low cost, small-scale traditional kilns and drum ovens (Kearns 2012; Kearns et al. 2014a, 2014b).

All chars were ground by hand with a mortar and pestle to pass a US standard 200-mesh sieve (0.074 mm) and dried at least overnight at 105 °C prior to characterization and batch experiments. To measure ash content, around 1 g of each char was accurately weighed in a ceramic crucible and then combusted at 550 °C for 6 hours, cooled to room temperature in a desiccator, and re-weighed. Reported values are averages taken of triplicate measurements for each charcoal sample. Elemental (C, N, H) analyses of each charcoal sample were completed by the Analytical Services Laboratory at North Carolina State University using a CHN Elemental Analyzer (Perkin-Elmer model 2400) calibrated with an acetanilide standard. A reference coal (NIST, 1632c standard reference material trace elements in coal – bituminous) was included as a calibration standard. Oxygen content of each sample was estimated as the remainder of mass after subtraction of C, N, H, and ash components. Brunauer, Emmett, and Teller (BET) surface areas were determined from N₂ gas adsorption (Autosorb MP-1, Quantachrome Instruments).

The GAC used for SOC uptake comparison with chars was bituminous-based Norit 1240, also ground by hand in a mortar and pestle to pass a US standard 200-mesh sieve. The PAC used for comparison was bituminous-based Calgon WPH, and used as received from the manufacturer (>90% passes US standard 325-mesh (45 μm) sieve).

Batch sorption experiments were conducted with char doses ranging from 5 to 4,000 mg/L in simulated natural water at pH 7 (maintained throughout using 20 mM phosphate buffer: 1.6 g/L KH₂PO₄ and 1.1 g/L Na₂HPO₄) containing background DOM isolated from a surface water source near Big Elk Meadows, CO, USA. This watershed is non-impacted from a standpoint of wastewater discharges or agricultural runoff. A DOM concentration of 4 mg/L total organic carbon (TOC) was used and is representative of many surface waters. All suspensions were agitated for 2 weeks to reach adsorption equilibrium (contact time to equilibrium established previously, Kearns et al. 2014b), and then were filtered through pre-fired glass fiber membranes (1.2 μm Whatman GF/C) to remove the char, and the filtrate analyzed for target SOCs.

To determine THM uptake, batch tests were initiated after the DOM solution had been in contact with free chlorine at pH 7 for 7 days in dark, headspace-free amber bottles at 25 °C. Two residual free chlorine concentrations were targeted: (1) 5 mg/L (Standard Method 5710) and (2) 13 mg/L, by following the same method but with excess free chlorine, to represent a field-relevant scenario where over-chlorination occurred. Both chlorination conditions produced a concentration of stable TTHMs of approximately 215 μg/L, a level that is roughly 2.7 times the USEPA regulatory level, simulating high but field-relevant DBP formation conditions. The solution conditions produced chloroform, DCBM and CDBM – in ratios of approximately 58%, 27%, and 15% of total THMs, respectively – but did not result in the production of bromoform. THMs were quantified using liquid–liquid extraction gas chromatography with electron-capture detection (USEPA Method 551.1). Residual free chlorine was evaluated by the USEPA DPD method (Standard Method 4500-Cl G).

As a preliminary assessment of trace contaminant desorption, the glass fiber membranes with the retained chars from batch uptake experiments were reintroduced to the same DOM background solution, absent WFN and MIB, and agitated for 4 weeks. The suspensions were then filtered and analyzed for desorbed WFN and MIB as described above.

Surface area, ash content, and elemental analysis data are presented in Table 1. The trends of increasing surface area (Rutherford et al. 2005; Keiluweit et al. 2010), increasing C content and decreasing H:C and O:C ratios (Spokas 2010; Enders et al. 2012; Schimmelpfennig & Glaser 2012) with increasing pyrolysis temperature agree with other research. As pyrolysis temperature increased, the decrease in the H:C molar ratio suggests increased char aromaticity. The decrease in the O:C and N:C ratios suggests a reduction in the amount of polar functional groups. This is in agreement with spectroscopy studies that indicate significant presence of noncarbonized aliphatic organic moieties (‘volatile matter’) in low temperature chars that are mostly or completely absent in highly aromatized high-temperature chars (Chen et al. 2008; Keiluweit et al. 2010).

Chlorine reduction by char was evaluated in batch tests with solutions containing 5 or 13 mg/L as Cl₂; i.e., with the solutions resulting from the DBP formation experiments. Upon char addition (5–4,000 mg/L char), the free chlorine residual concentration was reduced to below detection limit (0.02 mg/L) in all batch tests, independent of the starting free chlorine concentration of 5 or 13 mg/L. This result illustrates that the chlorine reduction capacity exceeded 2.5 g Cl₂/g char for the extreme case – a dose of 5 mg/L char and an initial chlorine concentration of 13 mg/L. Quantifying total chlorine reduction capacity was not an objective of this study. However, given the relatively large capacities for chlorine reduction exhibited by ACs (Suidan et al. 1980; Gillogly et al. 1998) and cookstove chars (Mahoney 2012), it is unlikely that the chlorine reduction capacity of biochars was reached under the experimental conditions used in the present study. Accordingly, char adsorbers designed and operated for the control of SOCs, as discussed below, will likely achieve complete removal of free chlorine.

Figure S1 (available online at http://www.iwaponline.com/washdev/005/172.pdf) displays the total and individual THM concentrations remaining in solution at equilibrium after contact with chars and ACs for all adsorbents studied, as representatively illustrated for the C625 char in Figure 1(a). Horizontal dashed lines indicate initial concentrations of total and individual THMs (C₀), as well as the USEPA TTHM MCL of 80 μg/L. Relative removal (on a percentage basis compared with C₀) of total and individual THMs was approximately consistent for each adsorbent over a dose range corresponding to 10–90% uptake, although CDBM appeared to be slightly better removed by some adsorbents (Figure S2, available online at http://www.iwaponline.com/washdev/005/172.pdf). Because the USEPA MCL concerns only TTHMs, the following discussion will consider TTHMs. TTHM removal efficacy was observed in the order PAC > C900 ≈ D900 >GAC ≈ R600 > C625 > R350 (Figure 1(b)). For the chars, TTHM removal strongly correlated with pyrolysis temperature (R600 > C625 can be explained by a longer heating duration in the retort). This result is consistent with our prior work as well as that of many other researchers (Chun et al. 2004; Chen et al. 2008; Uchimiya et al. 2010, 2012; Ahmad et al. 2012; Graber et al. 2012; Han et al. 2013; Wu et al. 2013; Zhang et al. 2013a, 2013b; Zheng et al. 2013) who have observed OC sorption capacity to increase with pyrolysis temperature.

Considering the approximate adsorbent dose required to bring TTHM concentrations below the USEPA regulatory standard under the solutions conditions studied, the high-temperature (900 °C) gasifier chars removed TTHMs about three times as effectively as intermediate temperature (600–625 °C) chars, and about 25 times as effectively as low temperature (350 °C) char. Furthermore, TTHM removals by the high-temperature gasifier chars were comparable to AC. To meet the USEPA standard of <80 μg/L TTHMs under these solution conditions required doses of approximately 20 mg/L, 35 mg/L, and 65 mg/L of PAC, either of the C900 or D900chars, and GAC, respectively.

WFN and MIB remaining in solution at equilibrium after contact with chars and PAC on a relative concentration basis (C/C₀, expressed as a percentage) are shown in Figure 2. Prior research has shown that, when background DOM is present, adsorption of trace SOCs on a percentage basis is independent of initial concentration (Knappe et al. 1998). WFN was slightly better removed than MIB by all adsorbents, in accordance with expectations based on previous research with AC (Kennedy et al. 2015). Removal efficacy observed was PAC ≈ C900 > C625 > R350, where PAC and the C900 char exhibits about 10 times the sorption capacity of the C625 char and 100 times the capacity of R350 char. This result is likewise in agreement with our prior work as well as that of many other researchers who have demonstrated sorption to increase with pyrolysis temperature (Kearns et al. 2014b).

Figure 3 illustrates the effect of initial free chlorine concentration and the presence of background DOM at a concentration of 4 mg/L TOC on MIB uptake by chars. The presence of DOM and chlorine exert a negative effect on MIB uptake by the intermediate and high-temperature chars. MIB adsorption by the C625 and the C900 chars is observed in the rank order: DI water > DOM in the absence of chlorine > DOM plus 5 mg/L chlorine > DOM plus 13 mg/L chlorine. Studies with ACs have shown background DOM to directly compete with trace SOCs for adsorption sites (Knappe et al. 1998; Graham et al. 2000; Corwin & Summers 2012). ACs are well graphitized, highly microporous adsorbents. As pyrolysis temperature increases, chars become more aromatic and condensed with less residual amorphous matter and more closely resemble AC physico-chemical properties such as highly developed microporosity (Antal & Gronli 2003; Chen et al. 2008; Keiluweit et al. 2010). Thus for intermediate to high-temperature chars a similar competitive effect from DOM likely occurs, as suggested in Figure 3.

The presence of chlorine can modify char capacity for MIB adsorption by either or both of two mechanisms: (1) directly, by oxidation of surface adsorption sites on the char, and (2) indirectly, by reaction with background DOM to produce low-molecular-weight compounds that compete with MIB for sorption sites. Gillogly et al. (1998) observed MIB uptake by PAC to decrease by about 50% in the presence of 3 mg/L chlorine, and to decrease further with increasing free chlorine concentration. The authors attributed this to the oxidation of MIB adsorption sites by chlorine. It can be hypothesized that oxidized surface sites attract water clusters through hydrogen bonding that physically block MIB access to surface adsorption sites and thereby reduce uptake, a phenomenon that has been observed in OC adsorption by ACs (Li et al. 2002). Cook et al. (2001) observed reduction in MIB uptake by PAC attributable to competitive adsorption by the increased concentration of low-molecular-weight compounds formed by DOM reaction with chlorine. Observations reported here do not permit discrimination between mechanisms for the decrease in MIB adsorption in the presence of chlorine – more research with biochar and various competing adsorbates is needed to investigate these phenomena.

Figure 3 indicates that MIB is sparingly taken up by R350 char, and uptake does not appear to be influenced by the presence of either chlorine or DOM under the solution conditions studied. It is possible that free chlorine reacts with the abundance of residual aliphatic ‘volatile matter’ in the char and is thereby quenched. OC uptake by low temperature chars is believed to occur mainly by partitioning and complexation with oxygen-bearing functional groups rather than pore filling and surface adsorption mechanisms (Chun et al. 2004; Chen et al. 2008; Sun et al. 2011; Li et al. 2013). Since SOCs are known to partition into bulk organic matter, it is possible that MIB interacts with the amorphous aliphatic domains of low temperature char and adsorbed organic matter similarly. This could explain the observation that MIB uptake by R350 char is similar in DOM and DI water. Further study is needed to support these hypotheses, however.

In summary, an important result for the use of intermediate and high-temperature chars in household water treatment is that typical chlorination conditions reduce batch-mode adsorbent equilibrium capacity for SOCs – in the case of MIB, by about a factor of five. However, batch mode experiments likely overestimate the extent to which chlorine reduces SOC uptake by adsorption units operated in flow-through column mode – a probable configuration for household water treatment – since chlorine reacts rapidly within the top few centimeters of the adsorbent bed (Suidan & Snoeyink 1975; Suidan et al. 1980). The presence of background DOM also reduces adsorbent capacity for target SOCs. From a research/experimental perspective, this underscores the importance of quantifying SOC uptake from solutions that are representative of natural drinking water sources, rather than from laboratory clean (DI) water which leads to overestimation of adsorbent capacity under real-world conditions.

A concern regarding the use of char adsorbers is desorption (back-diffusion, or ‘leaching’) of contaminants out of the char. Recent research using AC has shown little leaching of adsorbed contaminants to occur (Corwin & Summers 2011). The preliminary assessment of trace contaminant desorption conducted in this study showed WFN and MIB below detection levels (5 ng/L) for all chars and AC after 4 weeks of agitation in simulated surface water (data not shown). These results are encouraging; however, the rate and extent of back-diffusion of SOCs adsorbed to chars needs to be better established by further research.

Two important results from these studies relevant to application of the chars are: (1) C900 char removed all SOCs roughly equivalent to PAC, and removed TTHMs slightly better than GAC; and (2) that SOC removal by the three chars differs by about a factor of 10 each, in the order C900 > C625 > R30. For the chars, the trend of OC removal corresponds to the trends in increasing C content, decreasing H:C and O:C ratios, and increasing surface area (Table 1). It is interesting that C900 char exhibited similar OC uptake to GAC and PAC despite its significantly lower surface area (491 m²/g for C900 char compared to 697 m²/g and 902 m²/g for PAC and GAC, respectively).

To illustrate how the results of this study can be applied in household and community water treatment systems, Table 2 indicates the approximate adsorbent dose (mg/L) required to meet a given treatment objective for WFN, MIB, and TTHMs based on equilibrium contaminant uptake data. A separate study described a method for using adsorbent dose data collected in batch studies for design and operation of flow-through char adsorber units (Kearns et al. 2014a). The treatment objective of 90% removal was set for MIB because this corresponds to reducing MIB concentration to the approximate taste-and-odor threshold (∼10 ng/L) given a relatively high initial MIB concentration (C₀ = 100 ng/L). WFN is currently unregulated by USEPA and WHO, so its treatment objective was set equal to that for MIB, i.e., 90% removal. Two scenarios for TTHM removal were considered, corresponding to the USEPA MCL 80 μg/L and the WHO recommended additive guideline (Equation (1) and Figure S3, which is available online at http://www.iwaponline.com/washdev/005/172.pdf). Note that all residual free chlorine is removed using char doses that achieve WFN, MIB, and TTHM treatment objectives.

For the low and intermediate temperature R350 and C625 chars, MIB removal requires the greatest char dose and would thus govern char adsorber design and operation to meet MIB, WFN, and THM treatment objectives. This is fortuitous in climatic locations such as the tropics and in general during warm seasonal periods when MIB is likely to be present, since it can be readily detected by unaided human senses at low levels and does not pose a health concern. Detection of MIB taste/odor by users could therefore provide a convenient indicator for char replacement. With increasing char pyrolysis temperature, MIB and WFN removal efficiencies improved more dramatically than THM removal efficiency. As a result, char doses to meet the THM removal criterion were larger than those for MIB and WFN, and char adsorber design and operation would have to take this into account. Note, however, that under the experimental conditions used here, the C900 char dose of 13 mg/L required for 90% MIB removal would be expected to achieve ≥50% TTHM removal. This could be an acceptable treatment objective for waters with lower background DOM levels and/or under more moderate chlorination conditions that are likely to form fewer DBPs. The overall higher contaminant uptake capacity of high-temperature chars will reduce capital costs and adsorber unit maintenance frequency.

Starred values were derived by extrapolation beyond the range of data collected. Values for WFN and MIB were calculated using the simplified IAST (Qi et al. 2007).

Values for TTHMs were derived visually from Figures 1(b) and Figure S3 (available online at http://www.iwaponline.com/washdev/005/172.pdf).

The hatch denotes that for TTHM removal by PAC, the quantity of data points was insufficient to perform an extrapolation with confidence.

This paper has indicated that trace level organic contaminants and halogenated DBPs are a challenge to the provision of safe drinking water at the household level. Presented here is a viable low-cost treatment option for chemical and microbial risk reduction through (1) effective mitigation of chemical toxin exposure and (2) improvement of drinking water organoleptic properties that can reinforce safe water practices (i.e., disinfection using chlorine). This work has shown how traditional methods can be optimized for production of enhanced adsorbent chars. High-temperature (∼900 °C) wood chars produced from gasifier ovens were demonstrated to perform comparably with commercial ACs for equilibrium uptake of selected SOCs from representative surface water at environmentally and field-relevant concentrations. High-temperature gasifer chars have roughly 10 times the organic contaminant adsorption capacity of intermediate temperature (∼600 °C) chars and at least 100 times the capacity of low temperature (∼350 °C) retort chars. This work also revealed that chlorination and the presence of background DOM decrease the adsorption capacity for trace SOCs by biochars in a similar manner as has been observed here and elsewhere for ACs. Adsorption using local biomass chars produced at high temperature is potentially a cost-effective, sustainable, and practical option for household and community water treatment. This work promotes harmonization between approaches to water treatment in developing communities and the philosophy and capabilities of affluent-world treatment works through advancing parity between chemical, microbiological, and esthetic concerns for water quality. Further efforts are needed to increase understanding of pollutant sorption mechanisms by a wider variety of chars under dynamic (non-equilibrium) conditions, as well as for long-term monitoring and validation of chlorination and char adsorber systems in the field at the household and small community scales in the context of sustained use, costs, and health outcomes.

The authors would like to thank Dorothy Noble at University of Colorado-Boulder, Lisa Lentz at North Carolina State University, and David Rutherford at USGS-Denver for analytical services. This work was funded in part by grants from the Charles A. and Anne Morrow Lindbergh Foundation, the iBoP-Asia Partnership between the Ateneo School of Government (Philippines) and the International Development Research Centre (Canada), the University of Colorado Chancellor's Fellowship for Research, the University of Colorado Excellence in Engineering Fund, the EPA Science To Achieve Results (STAR) Graduate Research Fellowship program, the Bill & Melinda Gates Foundation, and Aqueous Solutions, a US- and Thailand-based not-for-profit appropriate technology water and sanitation research collaborative (www.aqsolutions.org).