Pyrogenic Carbon-Promoted Haloacetic Acid Decarboxylation to Trihalomethanes in Drinking Water

Abstract

Drinking water disinfection by chlorination or chloramination can result in the formation of disinfection byproducts (DBPs) such as haloacetic acids (HAAs) and trihalomethanes (THMs). Pyrogenic carbonaceous matter (PCM), such as activated carbon (AC), is commonly used as an ostensibly inert adsorbent to remove HAAs from water. HAA degradation has been mainly attributed to biological factors. This study, for the first time, revealed that abiotic HAA degradation in the presence of PCM could be important under water treatment conditions. Specifically, we observed complete destruction of Br₃AA, a model HAA, in the presence of powder AC at pH 7 within 30 min. To understand the role of PCM and the reaction mechanism, we performed a systematic study using a suite of HAAs and various PCM types. We found that PCM significantly accelerated the transformation of three HAAs (Br₃AA, BrCl₂AA, Br₂ClAA) at pH 7. Product characterization indicated an approximately 1:1 HAA molar transformation into their respective THMs following a decarboxylation pathway with PCM. The Br₃AA activation energy (E_a) was measured by kinetic experiments at 15–45°C with and without a model PCM, wherein a significant decrease in E_a from 25.7±3.2 to 13.6±2.2 kcal·mol⁻¹ was observed. We further demonstrated that oxygenated functional groups on PCM (e.g., −COOH) can accelerate HAA decarboxylation using synthesized polymers to resemble PCM. Density functional theory simulations were performed to determine the enthalpy of activation (ΔH^‡) for Br₃AA decarboxylation with H₃O⁺ and formic acid (HCOOH). The presence of HCOOH significantly lowered the overall ΔH^‡ value for Br₃AA decarboxylation, supporting the hypothesis that −COOH catalyzes the C-C bond breaking in Br₃AA. Overall, our study demonstrated the importance of a previously overlooked abiotic reaction pathway, where HAAs can be quickly converted to THMs with PCM under water treatment relevant conditions. These findings have substantial implications for DBP mitigation in water quality control, particularly for potable water reuse or pre-chlorinated water that allow direct contact between HAAs and AC during filtration as well as PAC fines traveling with finished water in water distribution systems. As such, the volatilization and relative low toxicity of volatile THMs may be considered as a detoxification process to mitigate adverse DBP effects in drinking water, thereby lowering potential health risks to consumers.

Graphical abstract

1. Introduction

Haloacetic acids (HAAs) and trihalomethanes (THMs) are major classes of disinfection byproducts (DBPs) that are produced when disinfectants react with organic and inorganic precursors in water (McRae et al., 2004; Parvez et al., 2019; Zhang et al., 2015). The occurrence of HAAs and THMs have been reported in drinking water (Levesque et al., 2006; Rodriguez et al., 2004; Villanueva et al., 2003). Since the identification of the first known DBPs in the 1970s, DBPs such as HAAs and THMs have been increasingly reported to be ubiquitous in chlorinated drinking water at HAA and THM concentrations ranging between 8–86 and 3–50 μg·L⁻¹, respectively (Richardson, 2002; Villanueva et al., 2003). Due to the potential genotoxicity and carcinogenicity of HAAs (Pals et al., 2011; Parvez et al., 2019; Zhang et al., 2015), the United States Environmental Protection Agency (USEPA) currently regulates five HAAs (ClAA, Cl₂AA, Cl₃AA, BrAA, Br₂AA) and four THMs (CHBr₃, CHCl₃, CHBrCl₂, CHBr₂Cl) in drinking water under Stages I/II of the Disinfectants/Disinfection Byproducts Rule at a combined maximum contaminant level (MCL) of 60 μg·L⁻¹ for HAAs and 80 μg·L⁻¹ for THMs. Four additional HAAs, namely, tribromoacetic acid (Br₃AA), bromochloroacetic acid (BrClAA), bromodichloroacetic acid (BrCl₂AA), and chlorodibromoacetic acid (Br₂ClAA), although not yet regulated, have raised concerns over their higher cytotoxicity and genotoxicity as compared to regulated HAAs (Plewa et al., 2010; USEPA, 2016). Thus, effective mitigative approaches toward controlling both regulated and emerging HAAs are needed.

Pyrogenic carbonaceous matter (PCM) is the solid residual generated from the incomplete combustion of fresh or fossilized biomass. Engineering analogs of PCM, such as granular activated carbon (GAC) and powder activated carbon (PAC), have been widely used in water treatment (Benjamin and Lawler, 2013). For instance, GAC/PAC (Altmann et al., 2016; Tung et al., 2006) are commonly used medium for water filtration, wherein their effectiveness largely depends on the contaminant’s hydrophobicity (Golea et al., 2020; Jiang et al., 2017). Therefore, they are effective towards intermediate aromatic DBPs (Jiang et al., 2020; Jiang et al., 2018; Jiang and Zhang, 2018; Jiang et al., 2017), but are ineffective towards DBPs with high polarity and low molecular weight, such as HAAs (Tung et al., 2006) and THMs (Capar and Yetiş, 2002; Summers et al., 1993). Furthermore, once the adsorption sites are saturated, used GAC/PAC require off-site disposal or regeneration (Crittenden et al., 2012; Xie, 2003), with costs ranging from $1,720 to $3,000 ton⁻¹year⁻¹ (Casey et al., 2013; USEPA, 2000). Biological AC (BAC) is often used for HAA removal during water treatment. BAC reactivity has been attributed to biotic processes of bacterial enrichment cultures (McRae et al., 2004), heterotrophic bacteria (Hozalski et al., 2001; Tung and Xie, 2009), and biofilms (Behbahani et al., 2018; Tung and Xie, 2011; Zhou and Xie, 2002), whereas AC is considered an inert attachment surface for microbial growth. Recent studies have found that PAC incorporation into porous electrodes can mimic BAC performance by facilitating the adsorption of polar DBPs, such as HAAs and N-nitrosodimethylamine (NDMA), and subsequently destroying these contaminants electrochemically (Almassi et al., 2019a; Almassi et al., 2019b).

Although the role of PCM as a passive adsorbent is well-recognized, recent studies have demonstrated that PCM has intrinsic reactivity and can participate in an array of biotic and abiotic reactions occurring in the natural environment (Pignatello et al., 2017; Xu et al., 2020a; Xu et al., 2021). For instance, PCM promotes a suite of pollutant transformation under environmentally relevant conditions, including primary amines (Aguilar et al., 2005), nitroaromatics (Ding et al., 2018; Gomes et al., 2008), phenolic compounds (Li et al., 2015; Pei et al., 2018; Xie et al., 2017), chlorinated hydrocarbons (Chen et al., 2014), explosives (Xu et al., 2010), agrochemicals (Ding et al., 2019; Ding and Xu, 2016; Xu et al., 2013; Xu et al., 2022; Xu et al., 2020b), and azo dyes (van der Zee et al., 2003; Zhao et al., 2019). Given the widely observed PCM reactivity in nature, these reactions could potentially occur in engineered systems, including water treatment processes. For instance, AC was able to convert N-DBP precursors (e.g., dimethylamine) to nitrosamines, increasing secondary amine formation in wastewater effluent (Padhye et al., 2010; Padhye et al., 2011). Previous studies also reported HAA conversion to their respective trihalomethanes under ambient (Heller-Grossman et al., 1993; Zhang and Minear, 2002) or high-temperature conditions (Chen et al., 2021; Lifongo et al., 2010; Pan et al., 2014) in the absence of PCM with relatively slow kinetics, e.g., a half-life of 17 d for Br₃AA (Zhang and Minear, 2002). However, despite the prevalence of BAC for HAA removal in water treatment, little information is available regarding whether or not PCM could promote abiotic HAA transformation kinetics or their associated product distribution. Characterizing and understanding how these reactions may affect HAA mitigation are important in water quality control, especially for potable water reuse or pre-chlorinated water, wherein HAAs and AC have direct contact during the filtration process, as well as clear water and well distribution systems that often present PAC fines that travel with finished water.

This study aims to investigate the feasibility of PCM in promoting abiotic HAA transformation under conditions relevant to water treatment and characterize the associated surface transformation reaction pathway. Specifically, we explored nine HAAs (HAA₉: Br₃AA, Br₂ClAA, BrCl₂AA, ClAA, Cl₂AA, Cl₃AA, BrAA, Br₂AA, BrClAA), including both regulated and emerging DBPs, to understand structural HAA requirements for PCM-enhanced abiotic transformation. The degradation of HAAs was monitored and their products were characterized. Kinetic modeling was performed to delineate the adsorption kinetics from surface transformation rates. The role of specific PCM functional groups (e.g., −COOH, −OH) in promoting HAA transformation (i.e., Br₃AA) was investigated by synthesizing PCM-like polymers to resemble PCM. Density functional theory (DFT) simulations were performed, where the enthalpy of activation (ΔH^‡) and overall reaction enthalpy (ΔH_r) for Br₃AA decarboxylation were determined. Lastly, the impact of solution chemistry was evaluated and HAA decay dependence on PCM type (e.g., graphite, PAC, MWCNTs) was studied. Our findings elucidated an overlooked reaction where the presence of PCM significantly enhanced abiotic HAA transformation into their respective haloforms under water treatment relevant conditions. These results can inform the water industry regarding their DBP mitigation approaches, where the increased presence of volatile THMs with relatively lower toxicity following HAA decay might help reduce the adverse health impact of DBPs in drinking water.

2. Materials and Methods

2.1. Reagents

A complete reagent list is detailed in Text S1. PCM-like polymers were synthesized by Pd(0)/Cu(I)-catalyzed cross-coupling chemistry following a method adapted from our previous study (Li et al., 2019). Three PCM-like polymers were obtained, where PCM₀, PCM_-OH, and PCM_-COOH represent polymers with no functionality, −OH, and −COOH groups, respectively. PCM₀ was obtained by reactions where a 1,4-dibromobenzene node connected to a rigid 1,3,5-triethynylbenzene strut. Briefly, 1,3,5-triethynylbenzene (3.0 mmol), 1,4-dibromobenzene (3.0 mmol), tetrakis(triphenylphosphine)palladium(0) (150 mg), and copper(I) iodide (45 mg) were dissolved in a N,N-dimethylformamide (20 mL) and triethylamine (20 mL) mixture for 30 min, stored in darkness, and heated to 80°C for 72 h under nitrogen. The reactor was cooled to room temperature. The products were washed with chloroform, water, methanol, and acetone to remove unreacted monomers and catalyst residues, and purified by Soxhlet extraction with methanol for 48 h and dried at 100°C for 24 h. PCM_-OH and PCM_-COOH were synthesized following the same route, except 2,5-dibromohydroquinone and 2,5-diiodobenzoic acid respectively substituted 1,4-dibromobenzene. PCM₀ (yield: 107%) and PCM_-OH (yield: 114%) were dark brown as previously reported (Li et al., 2019), and PCM_-COOH (yield: 86%) was light orange.

2.2. Batch adsorption experiments

A series of HAA₆ (ClAA, Cl₂AA, Cl₃AA, BrAA, Br₂AA, BrClAA) from 1–10 mg·L⁻¹ were spiked into batch reactors (KIMAX Culture Tubes filled to 10 mL with minimal headspace) with pre-weighed PCM (30 g·L⁻¹) in 10 mM KH₂PO₄/K₂HPO₄ (pH 7.0±0.1) to obtain a 1–2 log range isotherm curve. A model PCM, graphite, was chosen due to its high carbon purity (~99 wt%) and lack of functionality and porosity. Blank controls without graphite were simultaneously prepared. All samples were capped with Teflon-lined septa and placed on an end-to-end rugged rotator mixer (Glas-Col, Tere Haute, IN, USA) in darkness at 30 rpm, 25°C in a Model VRI6P incubator (VWR International, Radnor, PA, USA) for 24 h to allow adsorption equilibrium. The equilibrium time was determined separately by monitoring the aqueous phase HAA (i.e., Br₂AA) concentration until it plateaued (Fig. S1). After 24 h, all samples were centrifuged (Forma Scientific Inc. 5678, Marietta, OH, USA) at 3,000 rpm for 3 min. HAA₆ from the aqueous phase were analyzed following the methods described in Section 2.4. Adsorption isotherms were constructed by plotting the solid phase concentration (q_e; μg·g⁻¹), which was determined by a mass balance approach, versus HAA aqueous phase concentration (C_e; μg·L⁻¹) at equilibrium. The isotherms were subsequently fit against the Freundlich model (Tung et al., 2006).

2.3. Batch reaction experiments

A concentration of 50 μg·L⁻¹ of HAA₃ (Br₃AA, Br₂ClAA, or BrCl₂AA) was added to batch reactors containing pre-weighted PCM or PCM-like polymers (0.77 for PAC, MWCNTs, and PCM-like polymers; 23.08 g·L⁻¹ for graphite) with 10 mM KH₂PO₄/K₂HPO₄ (pH 7.0±0.1) or local tap water (pH 7.0±0.2; unadjusted) for HAA₃ degradation experiments. A higher concentration of graphite was used to speed up the reaction as it has the minimum surface area and functionality of the PCM tested. For pH-dependent study, a phosphate-carbonate buffer (KH₂PO₄/K₂HPO₄/NaHCO₃) was used for pH 5–9, wherein [KH₂PO₄]+[K₂HPO₄]+[NaHCO₃] was adjusted to 2 mM to ensure uniform ionic strength. DI water was adjusted with H₂SO₄ for pH-dependent study at pH 3. Controls without solids were simultaneously prepared. All samples were capped with Teflon-lined septa and placed on an end-to-end rugged rotator mixer in darkness (30 rpm), 25°C in an incubator. Samples were centrifuged at 3,000 rpm for 3 min at various time intervals. Both aqueous and solid phases were analyzed for HAA₃ and THM₃ (CHBr₃, CHBr₂Cl, CHBrCl₂) following methods described in Section 2.4.

2.4. Chemical analysis

The HAA₉ and THMs were analyzed following methods adapted from EPA method 552.3 and 551.1, respectively (USEPA, 1995; 2003). Briefly, the aqueous phase was pipetted out and adjusted to pH<0.5, and methyl tert-butyl ether (MTBE; with 200 μg·L⁻¹ 1,2,3-trichloropropane (1,2,3-TCP) as internal standard) was added at a 1:1 volumetric ratio. Subsequently, HAA₉ solid-phase extraction (SPE) proceeded by adding 5 mL MTBE to the remaining solids and 5 min vortexing. The extracted HAAs in MTBE were converted to methyl esters by adding acidic methanol (10 vol%, 18 M H₂SO₄) at a 1:1 volumetric ratio followed by 30 s vortexing and 50°C heating for 2 h. A Na₂SO₄ solution was added to remove the acidic aqueous phase and a saturated NaHCO₃ solution was used for solution neutralization. Fifty μg·L⁻¹ 2,3-dibromopropanoic acid was added as the surrogate standard. Conversely, THM₃ liquid-liquid microextraction proceeded by adding 2 mL MTBE (containing 200 μg·L⁻¹ 1,2,3-TCP) to 2-mL water samples and 2 min vortexing. THM₃ SPE proceeded by adding 5 mL MTBE, 5 min vortexing, and 3,000 rpm centrifuging for 3 min. All extracts were analyzed by gas chromatography/electron capture detection (GC/ECD) with an Rxi-5ms column (30 m×0.25 mm, 0.25 μm; Restek Inc., Bellefonte, PA, USA) and N₂ as carrier gas. GC oven program details are in Text S2. The extraction efficiencies and phase distributions for HAAs and THM₃ are in Text S3. For product identification purposes, select samples with high Br₃AA concentration (10 mg·L⁻¹) were analyzed for bromide using ion chromatography (IC; Shimadzu, Kyoto, Japan).

2.5. Characterization of PCM and PCM-like polymers

The graphite, PAC, MWCNTs, PCM₀, PCM_-OH, and PCM_-COOH surface areas were measured following a previously established method using a Brunauer-Emmett-Teller (BET) surface analyzer NOVA 3000e (Quantachrome Instruments, Boynton Beach, FL, U.S.A.) with nitrogen at 77.3 K (Li et al., 2019) (Table S1); Particle Technology Labs (Downers Grove, IL, U.S.A.) conducted PCM_-COOH analysis on a Tristar II BET surface analyzer (Micrometrics, Norcross, GA, U.S.A). All samples were outgassed at 110°C for 18 h before measurement. Surface areas calculations used data collected from 0.015–0.05 p/p₀. All nuclear magnetic resonance (NMR) experiments were performed on a Bruker Avance 400 spectrometer at 100 MHz ¹³C frequency with a double-resonance probe head with 4-mm sample rotors. A previously established ¹³C multiple cross-polarization/magic angle spinning (multi-CP/MAS) NMR technique was applied to gather PCM-like polymer quantitative structural information (Johnson and Schmidt-Rohr, 2014; Li et al., 2019), wherein analysis was performed to investigate the polymerization degree (Mao and Schmidt-Rohr, 2003; Tang et al., 2000). The quantitative spectra were recorded at a spinning speed of 14 kHz and 90° ¹³C pulse length of 4 μs. All spectra presented good signal-to-noise ratios with significantly small (<3%) spinning sidebands and minimal overlapping center bands. In addition, the multi-CP experiments were combined with 68 μs dipolar dephasing (multi-CP/DD) to generate non-protonated C and mobile group sub-spectra (Johnson and Schmidt-Rohr, 2014; Li et al., 2019).

We adapted a previously established method to determine the point-of-zero charges (pzc) for PCM and PCM-like polymers (Ni et al., 2011). Briefly, DI water, wherein all reagents and NaOH and HCl solutions were prepared, was boiled for 1 h and degassed with N₂ for 30 min for CO₂ removal. Subsequently, 0.8 g powder (graphite, PCM_-OH, or PCM_-COOH) was pre-wetted in 10 mL DI water for 48 h at 25°C, 40 rpm. The initial blank pH (no solids) with 10 mL DI water was lowered to pH~2, and recorded volumes of 0.1 or 1 N NaOH were added until pH~12. Identical HCl and NaOH volumes were added to each powder sample, and the pH at each increment was recorded. The powder pzc was obtained when the initial and final pH values were identical.

2.6. Density functional theory (DFT) simulations

Density functional theory (DFT) simulations were performed using Gaussian 16 software (Frisch, 2016). Unrestricted spin, all-electron calculations were performed using the 6–31G+(d) basis set for frequency and geometry optimizations, and the 6–311G+(3df, 2p) basis set for energy calculations. M06-2X hybrid meta exchange-correlation functionals were employed for exchange and correlation. This combination of basis sets and correlation functionals has been shown to be accurate for estimating reactions of organic compounds (Foresman and Frisch, 1996; Zhao and Truhlar, 2008). Implicit water solvation was simulated using the solvation model based on solute electron density (SMD) (Marenich et al., 2009). The transition state of a given reaction was determined by the pseudo-reaction coordinate method (Anderson and Kang, 1998), where the C-C bond length in Br₃C-COO⁻ was systemically varied. The resulting structure was confirmed to be representative of the transition state by yielding a single negative frequency during the frequency calculation (Foresman and Frisch, 1996).

3. Results and Discussion

3.1. Br₃AA, Br₂ClAA, and BrCl₂AA transformation kinetics and products

Three out of nine HAAs (HAA₃: Br₃AA, Br₂ClAA, BrCl₂AA) exhibited reactivity toward graphite. By contrast, degradation was not observed for the other six HAAs (HAA₆). Consistent with a previous study (Almassi et al., 2019b), HAA₆ adsorption can be described by the Freundlich model (adsorption isotherms in Fig. S2). HAA₃ decay was monitored over time to obtain over 90% parent compound decay. The total Br₃AA mass from both solid and aqueous phases were obtained using previously determined extraction efficiencies (Text S3). Our results suggest that the presence of graphite significantly accelerated Br₃AA decay, where (80.5±0.01)% of Br₃AA disappeared with graphite compared to (38.7±0.01)% without graphite over the experimental timeframe (Fig. 1a,1b). HAA degradation without graphite has been previously reported to follow an intramolecular electron transfer reaction (Zhang and Minear, 2002). Bromoform (CHBr₃) was identified as the main Br₃AA transformation product, with a 1:1 molar transformation of Br₃AA to CHBr₃. The average mass balance on bromine atom, calculated as 3[Br₃AA]+3[CHBr₃], was close to 100% over the entire experimental timeframe with/without graphite.

Fig. 1.

Tribromoacetic acid (Br₃AA; □) degradation and bromoform (CHBr₃; ○) formation over time. Mass balance (―) on bromine is calculated as (a,b) 3[Br₃AA]+3[CHBr₃]. Reaction conditions: [Br₃AA]₀=50 μg·L⁻¹; solid-to-liquid ratio: 23.08 g·L⁻¹ in 10 mM KH₂PO₄/K₂HPO₄ (pH 7.0±0.1); 25°C in darkness, 30 rpm; controls were without graphite. Error bar represents standard deviation of experimental duplicates. (c) Pseudo first-order degradation rate constants and enhancement factors for Br₃AA (R²_graphite=0.97; R²_{no graphite}=0.91), Br₂ClAA (R²_graphite=0.85; R²_{no graphite}=0.86), and BrCl₂AA (R²_graphite=0.92; R²_{no graphite}=0.90) with/without graphite.

Similarly to Br₃AA, graphite also accelerated Br₂ClAA and BrCl₂AA decay but with much slower kinetics. Details on HAA₃ pseudo first-order decay kinetics and R² values are in Fig. S3. Similar to Br₃AA, a 1:1 molar transformation of Br₂ClAA and BrCl₂AA to their respective THMs, CHBr₂Cl and CHBrCl₂, were observed. Approximately 100% Br mass balance, calculated as 2[Br₂ClAA]+2[CHBr₂Cl] and [BrCl₂AA]+[CHBrCl₂], were achieved for Br₂ClAA and BrCl₂AA, respectively. To characterize the formation of other transformation products, select samples with high Br₃AA concentration (10 mg·L⁻¹) were monitored with or without graphite, wherein no Br⁻, Br₂AA, BrAA, or acetic acid were observed, suggesting reductive dehalogenation was negligible (Adhikary et al., 2017; Hozalski et al., 2001) for all reaction systems with/without graphite.

As shown in Fig. 1c, k_obs for HAA₃ decay with graphite followed the trend: Br₃AA ((2.02±0.17)×10⁻¹ d⁻¹; R²=0.97) > Br₂ClAA ((4.75±0.88)×10⁻² d⁻¹; R²=0.85) > BrCl₂AA ((1.03±0.15)×10⁻² d⁻¹; R²=0.92). HAA₃ decay without graphite exhibited a similar trend with much slower kinetics: Br₃AA ((4.62±0.64)×10⁻² d⁻¹; R²=0.91) > Br₂ClAA ((1.77±0.43)×10⁻² d⁻¹; R²=0.86) > BrCl₂AA ((3.88±0.58)×10⁻³ d⁻¹; R²=0.90). HAA₃ enhancement factors by graphite were calculated by a ratio of the observed rate constants (k_obs) with/without graphite (i.e., kgraphite/k_no graphite), which followed: Br₃AA (4.37±0.25) > Br₂ClAA (2.68±0.16) > BrCl₂AA (2.65±0.01).

Although haloform formation from HAA decay without PCM has been previously reported (Heller-Grossman et al., 1993; Lifongo et al., 2010; Zhang and Minear, 2002), this is the first study to demonstrate that PCM can significantly enhance the decay kinetics of these reactions. For instance, graphite shortened Br₃AA half-lives from 15±1.1 to 3.4±0.4 d. Notably, graphite was used herein as a model PCM with the smallest surface area (orders of magnitude lower than PAC) and minimum surface functionality. It is expected that HAA decay would be much faster in the presence of PAC, which will be further evaluated in Section 3.4. Moreover, the observed HAA₃ degradation trend mirrored their logK_ow values (Br₃AA (1.71) > Br₂ClAA (1.62) > BrCl₂AA (1.53)), suggesting that non-specific interaction between HAAs and the PCM surface played a dominant role in accelerating their decay. We also investigated the possible contribution of trace metals on graphite to HAA decay, wherein no significant differences were observed in Br₃AA decay rates (k_obs) with graphite as a reference ((2.68±0.21)×10⁻¹ d⁻¹; R²=0.97)) versus HCl-washed graphite ((2.89±0.29)×10⁻¹ d⁻¹; R²=0.95)). Details on the experimental procedure and decay kinetics with HCl-washed graphite are in Fig. S4.

3.2. Modeling HAA decomposition on PCM surface

In general, HAA transformation with PCM is a combination of three processes (i.e., adsorption, desorption, and surface transformation reactions on PCM), which can be described by the Langmuir kinetic model (Ding et al., 2018; Li et al., 2019; Xu et al., 2013):

$$\frac{dq}{dt}=k_{a}C(Q-q)-k_d-k_{rxn}q$$ Equation (1)

where q is the HAA solid phase concentration (μg·g⁻¹), C is the aqueous concentration (μg·L⁻¹), Q is the total number of sorption sites (μg·g⁻¹), and k_a (L·μg⁻¹·d⁻¹), k_d (d⁻¹), and k_rxn (d⁻¹) are the adsorption, desorption, and surface transformation reaction rate constants, respectively. Depending on the rate-limiting step among these processes, k_obs may not reflect the true surface transformation reaction rate (k_rxn). Thus, we used a combined modeling and experimental approach to quantify k_rxn. Br₃AA was employed as a model compound herein. Details on the model fitting are in Text S4, Fig. S5 and obtained rate constants for Br₃AA adsorption, desorption, and surface transformation reaction with graphite are in Table 1.

Table 1.

Calculated adsorption, desorption, and surface transformation reaction rate constants of Br₃AA with graphite, PCM₀, PCM_-OH, or PCM_-COOH at 25°C. Details on Langmuir kinetic model fitting and adsorption kinetics are in Text S4, Fig. S5, and Fig. S8.

Powder	ka (L·μg−1·d−1)	kd (d−1)	krxn (d−1)
Graphite	(4.55±0.35)×10−2	(4.67±0.39)×101	(3.52±0.92)×102
PCM 0	(1.60±0.32)×10−1	(1.86±0.36)×102	(2.68±0.01)×102
PCM -OH	(9.29±0.68)×10−2	(1.33±0.22)×102	(1.97±0.03)×103
PCM -COOH	(6.56±0.13)×10−2	(1.18±0.16)×102	(4.34±0.01)×103

Our results suggest a three-fold larger k_rxn ((3.52±0.92)×10² d⁻¹) as compared to k_obs ((2.02±0.17)×10⁻¹ d⁻¹) for Br₃AA decay with graphite, suggesting that the adsorption process is rate-limiting (Table 1). Moreover, we calculated the Br₃AA mass fluxes (ϕ; μg·m⁻²·d⁻¹) for the surface transformation reaction (ϕ_rxn=k_rxnq), adsorption (ϕ_a=k_aC(Q-q)), and desorption (ϕ_d=k_dq) with graphite (Table S2). Across the experimental timeframe, all mass flux terms were comparable with each other, suggesting that all three processes contributed to Br₃AA transformation and thus the observed k_obs with graphite.

3.3. Reaction mechanism

3.3.1. Activation energy (E_a)

A previous study suggested that HAA conversion to THMs followed intramolecular electron transfer (Zhang and Minear, 2002). In our study, the presence of PCM evidently accelerated the reaction. To understand the role of PCM in accelerating Br₃AA degradation, we first quantified the activation energy (E_a) of Br₃AA decarboxylation by measuring Br₃AA decay rate constants over 15–45°C with and without graphite. Detailed decay kinetics are in Fig. S6. Eq. 2 can be linearized to calculate E_a:

$$k_{obs}=A{e}^{-\frac{E_a}{RT}}$$ Equation (2)

where k_obs is the degradation rate constant (d⁻¹), A is the pre-exponential factor (d⁻¹), E_a is the activation energy (kcal·mol⁻¹), R is the gas constant, and T is the temperature (K).

As shown in Fig. 2a, graphite accelerated Br₃AA decay from 15 to 45°C as compared to without graphite, though the PCM effect was less predominant at higher temperatures. We determined the Br₃AA decay E_a values using the linear regression slopes, which were 13.6±2.2 and 25.7±3.2 kcal·mol⁻¹ with and without graphite, respectively. Our results suggest that graphite significantly lowered E_a and thus accelerated Br₃AA decay. The obtained E_a without graphite was comparable to a previous study, which reported an E_a of 29.2 kcal·mol⁻¹ for Br₃AA decarboxylation in water (Zhang and Minear, 2002). This is the first study showing that graphite could lower Br₃AA activation energy.

Fig. 2.

(a) Arrhenius plot $k_{\text{obs}}=A{e}^{-E_a/RT}$ for Br₃AA degradation without (□) and with (○) graphite over time at 15°C, 25°C, 35°C, and 45°C in darkness. The calculated Br₃AA decay activation energies (E_a) were 13.6±2.2 and 25.7±3.2 kcal·mol⁻¹ with/without graphite. (b) Solid state ¹³C multiCP/MAS NMR spectra of PCM₀ (Li et al., 2019), PCM_-OH (Li et al., 2019), and PCM_-COOH. All the spectra were recorded at a spinning speed of 14 kHz with a 90° ¹³C pulse-length of 4 μs. (c) Pseudo first-order Br₃AA degradation rate constants with PCM₀, PCM_-OH, and PCM_-COOH. (d) Optimized structures for reactants, transition state, and products of (i) Reaction (1) and (ii) Reaction (2) based on DFT simulation.

3.3.2. Br₃AA decay with functionalized PCM-like polymers

In this study, we employed PCM-like polymers to understand the role of oxygenated functional groups in accelerating HAA decay. By doing so, we were able to isolate specific functionalities from an array of functional groups typically present in PCM. Like PCM, the PCM-like polymers are (1) highly conjugated and amorphous with no long-range molecular order, (2) contain micropores and large surface areas, and (3) have a high affinity towards organic contaminants. Different from PCM, however, specific functional groups can be introduced via a polymer synthesis approach following protocols established in our lab (Li et al., 2019), allowing us to investigate the impact of individual functionalities on HAA decay.

Advanced solid-state ¹³C NMR spectroscopy was used to provide information on the polymerization degree and confirm the presence of target functional groups. As shown in Fig. 2b, peaks at 123.7 ppm (C_Ar−C C−C_Ar), 131.8 ppm (C_Ar−H), and 90.8 ppm (C_R-C≡C-R’) suggest conjugated polymer network formation in PCM, PCM_-OH, and PCM_-COOH. Additionally, peaks at 167.8 ppm (C_Ar−OH) and 196.2 ppm (C_Ar=O) for PCM_-OH respectively verify the incorporation of phenolic and carbonyl groups in the polymer network. Peak-differentiation-imitating NMR analysis was employed to investigate the polymerization degree (Li et al., 2019). Specifically, the peak at 90.8 ppm (C_R-C≡C-R’) indicates efficient aryl halide and alkyne coupling. The low-intensity peak ratio at 80.3 ppm (C_−C≡C−H) over the peak at 90.8 ppm (C_R-C≡C-R’) provides information on the condensation degree, wherein smaller values indicate higher polymerization degrees. PCM₀ and PCM_-OH exhibited ratios of 0.18 and 0.15, respectively, as previously reported (Li et al., 2019), and PCM_-COOH presented a ratio of 0.01, suggesting a higher condensation degree, which were consistent with previous literature (Dawson et al., 2009; Jiang et al., 2008).

To understand the influence of oxygenated functional groups on HAA sorption and decay, the Br₃AA reactivity against PCM₀, PCM_-OH, and PCM_-COOH was examined. Br₃AA demonstrated pseudo first-order decay at pH 7 with PCM-like polymers (Fig. 2c): PCM_-COOH (k_obs=(4.31±0.05)×10⁻¹ d⁻¹; R²=0.92) > PCM_-OH ((3.58±0.62)×10⁻¹ d⁻¹; R²=0.84) > PCM₀ ((2.57±0.20)×10⁻¹ d⁻¹; R²=0.98). The kinetics of Br₃AA decay, product formation, and bromine mass balance in the presence of PCM₀, PCM_-OH, and PCM_-COOH are shown in Fig. S7, where ~100% Br mass balance were observed with all PCM-like polymers. Given that all polymers were synthesized following the same reaction route with an identical catalyst, we attribute the observed differences to variations in polymer surface functionalities. Similarly to Section 3.2, we obtained the Br₃AA adsorption (k_a), desorption (k_d), and surface transformation reaction (k_rxn) rate constants with PCM₀, PCM_-OH, and PCM_-COOH (Table 1). Details on Br₃AA adsorption kinetics with PCM₀, PCM_-OH, and PCM_-COOH are in Fig. S8. Br₃AA adsorption kinetics followed: PCM₀ (k_a=(1.60±0.32)×10⁻¹ d⁻¹) > PCM-_OH ((9.29±0.68)×10⁻² d⁻¹) > PCM_-COOH ((6.58±0.13)×10⁻² d⁻¹), which can be attributed to a combination of hydrophobic and electrostatic interactions between Br₃AA and PCM-like polymers. The measured pzc of PCM_-COOH and PCM_-OH were 2.57 and 7, respectively (Fig. S9). Thus, PCM_-COOH is more negatively-charged compared to PCM_-OH under circumneutral conditions, repelling negatively-charged Br₃AA (pK_a=0.72) and hindering its adsorption (Serjeant and Dempsey, 1979). Interestingly, despite the slowest adsorption rate constant for PCM_-COOH, its k_rxn was the highest: PCM_-COOH (k_rxn=(4.34±0.01)×10³ d⁻¹) > PCM_-OH ((1.97±0.03)×10³ d⁻¹) > PCM₀ ((2.68±0.01)×10² d⁻¹). We also calculated the mass fluxes (ϕ; μg·(m²·d)⁻¹) of Br₃AA adsorption, desorption, and surface transformation reactions (Table S3). For PCM₀, ϕ_a was approximately twice compared to ϕ_rxn. Interestingly, ϕ_rxn became predominant for PCM_-OH (ϕ_a<ϕ_rxn) and PCM_-COOH (ϕ_a<<ϕ_rxn). These results further confirmed that adsorption was the rate-limiting step for Br₃AA decay with PCM_-OH and PCM_-COOH and the surface transformation rate was much higher than the k_obs.

3.3.3. DFT modelling

To understand the reaction pathway for Br₃AA decarboxylation, DFT simulations were performed. The enthalpy of activation (ΔH^‡) and overall reaction enthalpy (ΔH_r) were determined for Br₃AA reaction with H₃O⁺ according to Reaction (1):

$${\text{Br}}_3\text{C-}{\text{COO}}^{-}+{\text{H}}_3{\text{O}}^{+}\to {\text{Br}}_3\text{CH}+{\text{CO}}_2+{\text{H}}_2\text{O}$$ Reaction (1)

where ΔH^‡=50 kcal·mol⁻¹ and ΔH_r=−3.8 kcal·mol⁻¹. The C-C bond in Br₃AA increased from 1.55 to 2.00 Å in the transition state structure, wherein H⁺ transfer from H₃O⁺ to −CBr₃ was characterized (Fig. 2d-i). To investigate the ability of −COOH in catalyzing the reaction, the DFT simulations added formic acid per Reaction (2):

$${\text{Br}}_3\text{C-}{\text{COO}}^{-}+\text{HCOOH}\to {\text{Br}}_3\text{CH}+{\text{CO}}_2+{\text{HCOO}}^{-}$$ Reaction (2)

where ΔH^‡=27 kcal·mol⁻¹ and ΔH_r=3.4 kcal·mol⁻¹. The Br₃AA C-C bond increased from 1.58 to 2.32 Å in the transition state structure, wherein H⁺ transfer from HCOOH to −CBr₃ was also observed (Fig. 2d-ii). The lower ΔH^‡ value for Reaction (2) relative to Reaction (1) supports our experimental observation where the E_a value for Br₃AA decreased from 25.7±3.2 to 13.6±2.2 kcal·mol⁻¹ with graphite. Moreover, our DFT simulation supports that −COOH catalyzes the C-C bond-breaking reaction of Br₃AA while facilitating the transfer of H⁺ from −COOH to −CBr₃.

We postulate that the formation of a low barrier H-bond (LBHB) between Br₃AA and the surface oxygenated functional groups, −COOH in particular, played an important role in catalyzing C-C bond breaking. The LBHB strength is highly dependent on the proton affinity (pK_a) proximities of the proton donor and acceptor groups, where smaller differences in acidity constants generate stronger H-bonds (Brück et al., 2000; Pignatello et al., 2017). Thus, the higher HAA reactivity with PCM_-COOH versus PCM_-OH can be explained by the formation of a stronger H-bond between Br₃AA and ⧙~COOH as evidenced by the closer proximity of pK_a values between Br₃AA (0.72) and ⧙~COOH (pK_a,COOH=3–6) compared to ⧙~OH (pK_a,phenol=9–10) (Pignatello et al., 2017). In addition to surface functionalities, other reaction sites such as surface edges may also contribute to accelerating HAA decay (Luo and Jean-Charles, 2006; Pignatello et al., 2017).

3.4. PCM Type

Other PCM types, including PAC and MWCNTs, were also investigated with respect to their ability to promote Br₃AA transformation at pH 7.0, wherein k_obs was normalized with respect to PCM dosage (Fig. 3a; left axis) vs. surface area (Fig. 3a; right axis). Details on the 1:1 molar transformation of Br₃AA to CHBr₃ with PAC and MWCNTs are in Fig. S10. Our results suggest that Br₃AA decay on a dosage-normalized rate constant basis was over two orders of magnitude higher with PAC than that with MWCNTs, which was one order and three orders magnitude higher than that with the PCM-like polymers and graphite, respectively. Normalizing these rate constants once more with respect to surface area, we now see the values showed overall smaller variations among different PCM: PAC was now one order of magnitude higher than MWCNTs, which was about one order of magnitude higher than functionalized PCM-like polymers and two orders of magnitude higher than graphite. We attributed the lowest reactivity of graphite to its smallest surface area and functional groups among all PCM. These results suggest that in addition to PCM surface area, other PCM properties such as the identity and density of surface functional groups as well as surface defects (Luo and Jean-Charles, 2006; Pignatello et al., 2017) might also contribute to the PCM-accelerated HAA decay. Detailed rate constants normalized against PCM dosage and surface area are presented in Table S4.

Fig. 3.

(a) Pseudo first-order Br₃AA decay kinetics with graphite, MWCNTs, and PAC normalized to PCM dosage (left axis), and surface area (right axis). (b) Br₃AA decay kinetics with graphite at pH 3.0 (k_obs=1.49±0.16 d⁻¹; R²=0.92), 5.2 (k_obs=(2.46±0.25)×10⁻¹ d⁻¹; R²=0.96), 6.2 (k_obs=(2.30±0.01)×10⁻¹ d⁻¹; R²=0.99), 7.3 (k_obs=(2.22±0.18)×10⁻¹ d⁻¹; R²=0.97), and 8.9 (k_obs=(1.97±0.18)×10⁻¹ d⁻¹; R²=0.96). (c) Br₃AA decay kinetics with 10 mM KH₂PO₄/K₂HPO₄ (k_obs=(2.26±0.21)×10⁻¹ d⁻¹; R²=0.97) and tap water (k_obs=(2.91±0.06)×10⁻¹ d⁻¹; R²=0.99). (d) Br₃AA decay kinetics in the presence of graphite with (k_obs=(1.79±0.04)×10⁻¹ d⁻¹; R²=0.99) or without (k_obs=(2.02±0.17)×10⁻¹ d⁻¹; R²=0.97) 3 mg·L⁻¹ Suwannee River NOM. Reaction conditions: [Br₃AA]₀=50 μg·L⁻¹; solid-to-liquid ratio: 0.77 or 23.08 g·L⁻¹; 25°C in darkness, 30 rpm; controls were without powder. Error bar represents standard deviation of experimental duplicates.

It is possible that there might be radicals present on the PCM surface, which could initiate the formation of a carboxylate radical and thus lead to the subsequent decarboxylation of Br₃AA. We thereby performed quenching experiments by monitoring Br₃AA degradation in the presence of PAC with and without 2-propanol-d₈, as detailed in Fig. S11 and Text S5. A statistical analysis (t-test) indicated no significant differences between the two rate constants of HAA decay (p>0.05) with ((3.26±0.19)×10⁻¹ min⁻¹) or without 2-propanol-d₈ ((3.66±0.25)×10⁻¹ min⁻¹). Moreover, the statistical analysis (t-test) indicated no significant differences (p>0.05) in CHBr₃ formation with (6.75±0.93)×10⁻⁴ μmol) or without (7.70±0.38)×10⁻⁴ μmol) 2-propanol-d₈ following a 3-min reaction of Br₃AA in the presence of PAC in a separate experiment. Our results clearly demonstrate that the radical pathway, if present, is negligible for the observed conversion between HAAs to THMs.

3.5. Effect of solution chemistry on Br₃AA degradation

We further investigated the effect of solution chemistry on Br₃AA decay kinetics under a pH range of 3–9. All pH values were measured before and after reaction, which exhibited negligible change. We observed pseudo first-order Br₃AA decay with graphite from pH 5 to 9, wherein k_obs followed: pH 5.2 (k_obs =(2.46±0.25)×10⁻¹ d⁻¹; R²=0.96) ≈ 6.2 (2.30±0.01)×10⁻¹ d⁻¹; R²=0.99) ≈ 7.3 (2.22±0.18)×10⁻¹ d⁻¹; R²=0.97) ≈ 8.9 (1.97±0.18)×10⁻¹ d⁻¹; R²=0.96) (Fig. 3b). A statistical analysis (t-test) indicated no significant differences among the four rate constants (p>0.05), suggesting negligible pH effect on Br₃AA decay with graphite at pH 5–9. By contrast, a much higher k_obs was noted (k_obs,pH3=1.49±0.16 d⁻¹; R²=0.92) for Br₃AA degradation with graphite at pH 3. Details on Br₃AA decay kinetics with/without graphite from pH 3–9 are in Fig. S12. A previous study found negligible solution pH effect for Br₃AA degradation between pH 6–9 in water (k_obs=0.04–0.043 d⁻¹) (Zhang and Minear, 2002). Our results confirmed that similar trend held true for Br₃AA decay from pH 5–9 with graphite but not when the pH was lowered to 3. We attributed the accelerated Br₃AA decay observed at pH 3 to the interaction between graphite and Br₃AA. Since the pzc charge of graphite was 3.5 (Fig. S9), graphite will be negatively-charged at pH 5–9, repelling HAA and resulting in slow adsorption in the respective pH range. However, graphite will be neutral/positively-charged at pH 3, resulting in higher affinity toward HAA due to electrostatic interaction and accelerating HAA decay (Punyapalakul et al., 2009).

We also examined the water matrix effect on Br₃AA decay using local tap water (pH 7.0±0.2; unadjusted). Br₃AA removal with graphite in tap water exhibited pseudo first-order kinetics (k_obs=0.291±0.01 d⁻¹; R²=0.99). The obtained value is comparable to the reference value in KH₂PO₄/K₂HPO₄ buffer at pH 7 (k_obs=0.226±0.02 d⁻¹; R²=0.97), suggesting negligible water matrix effects on Br₃AA decay with graphite. Details on Br₃AA decay kinetics with tap water and relevant parameters are in Fig. 3c. Similarly, the presence of Suwannee River natural organic matter (NOM; 3 mg·L⁻¹ as C; pre-equilibrated with graphite overnight) did not affect Br₃AA decay, as evidenced by the decay kinetics in Fig. 3d. Consistent with previous studies, our results suggest negligible water matrix effects on HAA decay with NOM (Zhang and Minear, 2002).

4. Conclusions

The present study, for the first time, demonstrated that PCM alone can significantly accelerate abiotic HAA decarboxylation reaction. Although HAA decarboxylation in water alone has been previously reported, the half-lives were too slow to affect engineering applications (e.g., a half-life of 15±1.1 d for Br₃AA in water at pH 7). Herein, we demonstrated that complete Br₃AA degradation removal can be achieved within the typical empty bed contact time (EBCT) of the filtration process (i.e., 30 min or less) in the presence of PAC, which suggests that the AC can significantly influence the conversion of HAAs to their respective THMs in potable water reuse or pre-chlorinated water during the filtration process where HAAs are in direct contact with AC. Moreover, PAC fines may travel with the finished water in clear well and water distribution systems with a typical residence time of hours to days (Tinker et al., 2009), increasing the contact time between HAAs and PAC and further promoting the transformation of HAAs to their respective THMs. We expect these reactions to proceed even faster in residential hot water distribution systems, resulting in increased presence of volatile DBPs in treated water following HAA decay. In addition, GAC filters are widely used as home filters, where HAAs might be further converted to THMs in the time frame of minutes to hours (WQA, 2016). It has been recently reported that volatile DBPs contributed marginally to the developmental toxicity of DBP mixtures in drinking water (Han and Zhang, 2018; Li et al., 2020; Plewa and Wagner, 2015; Wagner and Plewa, 2017). Therefore, both the volatilization and the relative low toxicity of volatile THMs might contribute as a detoxification process for mitigating the adverse effects of DBPs in drinking water and thus reduce health risks to consumers.

We also employed a combined experimental and computational approach to elucidate the role of PCM in accelerating Br₃AA decarboxylation. Specifically, this study is the first to demonstrate that the presence of PCM lowered the activation energy for Br₃AA decarboxylation. We further demonstrated that the carboxylic acid groups on PCM catalyzed Br₃AA decarboxylation using a polymer synthesis approach. Our DFT simulation supported these experimental observations and showed a lower overall ΔH^‡ value for catalyzed Br₃AA decarboxylation with HCOOH versus with H₃O⁺, indicating that −COOH catalyzed the C-C bond breaking of Br₃AA while facilitating H⁺ transfer from −COOH to −CBr₃.

Overall, this study unraveled a previously overlooked process, namely the abiotic HAA conversion to their respective THMs in the presence of PCM, which could significantly affect the DBP mitigative approaches due to the rapid conversation of hydrophilic DBPs into their respective volatile DBPs. Its implications on smaller-scale applications such as home heaters or in-home granular AC filters require future investigation.

Highlights.

• Complete abiotic transformation of Br3AA in 30 min with powder activated carbon.

• Rapid haloacetic acid conversion to volatile trihalomethanes by pyrogenic carbon.

• Pyrogenic carbon lowered the activation energy for haloacetic acid decarboxylation.

• A combined experimental and computational approach was employed.

• COOH catalyzed HAA C-C bond breaking and facilitated H+ transfer to form THMs.