Breakpoint Chlorination Chemistry in a Chlorine-Cyanurate System and Trade-Offs between Nitrosamine Formation and Micropollutant Removals

Abstract

Breakpoint chlorination occurs in swimming pools and generates •OH and nitrosating agents. While •OH facilitates micropollutant removal, nitrosating agents promote nitrosamine formation. Cyanuric acid, a common chlorine stabilizer, reacts reversibly with free chlorine to form chlorinated cyanurates, effectively “locking” free chlorine and altering breakpoint chemistry. Accurate estimation of free chlorine is essential for evaluating these impacts and depends on the hydrolytic dissociation constants of chlorinated cyanurates, yet reported values vary widely. This study re-evaluates these constants and examines how cyanuric acid influences breakpoint reactions, focusing on the trade-off between •OH-driven micropollutant degradation and nitrosamine formation. Using phenolic probes, we show that electrochemically determined hydrolytic dissociation constants more accurately predict free chlorine concentrations under pool-relevant conditions than spectrophotometric values. Kinetic experiments reveal that chlorinated cyanurates participate in breakpoint reactions, chlorinating NH₂Cl and NHCl₂ with rate constants approximately half of those of HOCl. Cyanuric acid also catalyzes NHCl₂ formation from NH₂Cl. Under simulated pool conditions, cyanuric acid enhanced micropollutant removal, suppressed NCl₃ formation, but promoted nitrosamine formation. A refined kinetic model captured these trends and provided mechanistic insights. Cyanuric acid, while mitigating scavenging of •OH and nitrosating agents by oxidants, prolongs NCl₃–NHCl₂ interactions, thereby increasing •OH and nitrosating agent yields while lowering residual NCl₃.

Introduction

Chlorine disinfection remains the standard practice for maintaining microbial safety in swimming pools. In both outdoor and indoor pools, cyanuric acid is often present either as a deliberately added stabilizer to reduce chlorine photodegradation or as a byproduct of stabilized chlorine disinfectants such as trichloroisocyanuric acid and sodium dichloroisocyanurate. − Typical cyanuric acid concentrations in swimming pools range from 39 to 595 μM, with a median of 311 μM. Cyanuric acid forms reversible complexes with free chlorine species (i.e., HOCl and OCl^–), resulting in a dynamic equilibrium among free chlorine, cyanuric acid, and chlorinated cyanurates (Scheme ). This equilibrium is influenced by pH, total chlorine concentration, and the cyanuric acid-to-chlorine molar ratio, with higher ratios favoring the formation of chlorinated cyanurates. Prior studies have reported that elevated cyanuric acid concentrations reduce disinfection efficacy attributed to a decline in HOCl. , Given that HOCl is the principal oxidant responsible for chlorination of cellular and organic targets, these findings also suggest that chlorinated cyanurates are significantly less reactive as chlorinating agents.

1. Equilibria among Free Chlorine, Chlorinated Cyanurates, and Cyanurates in a Chlorine-Cyanuric Acid System .

^a The hydrolytic dissociation constants (K₁, K₃, K₅, K₇, K₉, and K₁₁) represent reactions that release HOCl, while the acid dissociation constants (K₂, K₄, K₆, K₈, K₁₀, and K₁₂) correspond to the dissociation of acidic species (values provided in Table S4).

While chlorination effectively mitigates acute risks of waterborne diseases, it raises chronic exposure concerns attributed to formation of disinfection byproducts (DBPs). Over 100 DBPs have been detected in pool waters. , Among them, N-nitrosamines like N-nitrosodimethylamine (NDMA) occur at relatively low levels but carry significant risks due to their high toxicity. For example, Walse and Mitch reported NDMA levels ranging from 12–44 ng/L in indoor pools and 5–7 ng/L in outdoor pools, whereas a field study in Korea found a wider range of 0.7–210 ng/L in indoor pools. ,

In contrast to drinking water, swimming pool water typically contains far higher concentrations of DBP precursors because of human-derived substances such as sweat and urine, resulting in DBP concentrations up to 200 times higher than those in tap water. − Swimmers are estimated to release 78–250 mL of body fluids (mixtures of sweat and urine) during a swim event, equivalent of 1,400 mg-C of dissolved organic carbon (DOC). These excretions also introduce micropollutants such as pharmaceuticals and personal care products in addition to DBP precursors. Swimmers may be exposed to DBPs or micropollutants through inhalation, dermal absorption, and accidental ingestion.

Since ammonia is a major component of human body excretions, release of body fluids during swimming triggers breakpoint reactions in pools wherein free chlorine is steadily maintained. A cascade of reactions is initiated when the chlorine-to-ammonia (Cl₂/NH₄ ⁺) molar ratio exceeds ∼ 1.5, resulting in the stepwise formation of monochloramine (NH₂Cl), dichloramine (NHCl₂), and trichloramine (NCl₃) through their interactions with HOCl (Scheme ). Subsequently, NCl₃ reacts with either NHCl₂ or NH₂Cl to form N₂ and other byproducts (reaction U12 in Scheme ), with this reaction predominantly contribute to total chlorine loss relative to other reactions. While kinetic models have been developed to describe these processes, − recent studies have expanded this understanding by identifying highly reactive species such as nitrosyl chloride (ClNO; via U12b), a potent nitrosating agent, and •OH (via U12c) as additional products of breakpoint chlorination. , •OH plays a beneficial role in degrading recalcitrant micropollutants, yet ClNO can drive the formation of toxic nitrosamines. Furthermore, recent work has shown that •OH and NCl₃ enhance haloacetonitrile production, likely through NCl₃-mediated nitrogen incorporation into hydroxylated moieties of natural organic matter. These results indicate that breakpoint chlorination can inadvertently accelerate the formation of nitrogenous DBPs, which exhibit cytotoxicity and genotoxicity orders of magnitude greater than trihalomethanes.

2. . Reaction Pathways during the Free Chlorine-Ammonia Interactions .

^a The reaction scheme is based on Jafvert and Valentine’s model, with additional reactions regarding formation of nitrosating agent and •OH. ,,

The presence of cyanuric acid in pools complicates breakpoint chlorination chemistry in several ways. First, it reduces the equilibrium concentration of free chlorine, thereby lowering the transformation rates of ammonia and chloramines (U1, U3, and U11). Accurate quantification of free chlorine in HOCl–cyanuric acid systems is critical for evaluating the impact of cyanuric acid on the dynamics of breakpoint reactions. However, the widely used DPD (N,N-diethyl-p-phenylenediamine oxalate) method cannot distinguish free chlorine from chlorinated cyanurates, necessitating indirect estimation based on equilibrium and hydrolytic dissociation constants, particularly for dichloroisocyanurate (Cl₂Cy^–, K₇) and hydrochloroisocyanurate (HClCy^–, K₉), which are dominant under typical pool conditions. ,− Reported values for these constants, however, vary considerably. Early spectrometric measurements employing curve-resolving techniques reported pK ₇ and pK ₉ of 4.51 and 5.62, respectively. Later, Pinsky and Hu revised these values to 4.11 and 4.92 using linear sweep voltammetry to address potential spectral overlaps at mM concentrations. In a follow-up study, Jensen and Johnson suspected that interference from chlorinated cyanurates may have affected prior measurements and re-evaluated pK ₉ as 4.8 (nearly one log unit lower than previously reported) using an amperometric membrane electrode. Despite these efforts, the debate over the accuracy of hydrolytic dissociation constants persists in recent studies, , leaving the accuracy of pK ₇ and pK ₉ values an open question. While recognizing these disparities, a recent study argued in favor of the constants reported by the early spectrometric measurements, though without clearly articulating the underlying rationale. Notably, the U.S. EPA’s web-based Free Chlorine and Cyanuric Acid Simulator also employs equilibrium constants derived from these early spectrometric data.

In addition to altering dynamics, cyanuric acid also appears to influence reaction pathways during breakpoint chlorination. One study indicates that cyanuric acid enhances the conversion of NH₂Cl to NHCl₂, suggesting the presence of additional pathways beyond direct HOCl-mediated transformation (U3) or NH₂Cl disproportionation (U5). These pathways remain largely uncharacterized and are not yet represented in current kinetic models.

Moreover, cyanuric acid may influence the balance between contaminant degradation and byproduct formation during breakpoint reactions. Previous studies have shown that both micropollutant removal and nitrosamine formation peak at a Cl₂/NH₄ ⁺ molar ratio of approximately 1.8–2 (the ‘breakpoint’), then decline as the ratio increases. ,,− This trend, though no explanation was provided in previous works, may be attributed to enhanced scavenging of •OH and ClNO by excess free chlorine under postbreakpoint conditions. However, these scavenging reactions are likely diminished in the presence of cyanuric acid due to the lower reactivity of chlorinated cyanurates. For example, k _•OH for chlorinated cyanurates (<1.9 × 10⁷ M^–1s^–1) are orders of magnitude lower than that for HOCl/OCl^– (1.2–6 × 10⁹ M^–1s^–1). This reduced reactivity may extend the lifetime of •OH and ClNO, thereby enhancing removal of recalcitrant micropollutants while also elevating the risk of nitrosamine formation. These dual effects underscore the complex trade-offs introduced by cyanuric acid in chlorinated pool systems, which however remain largely uncharacterized.

The main objective of this study is to evaluate how cyanuric acid stabilizers affect micropollutant removal and DBP formation, particularly nitrosamines, under typical chlorinated pool conditions. To support this assessment, we first re-evaluated the equilibrium constants governing the hydrolytic dissociation of chlorinated cyanurates to improve the accuracy of free chlorine quantification. We then investigated potential interactions among cyanuric acid, chlorinated cyanurates, and chloramines, and determined their associated rate constants. Finally, we examined the trade-offs between nitrosamine formation and micropollutant removal during breakpoint chlorination and provided mechanistic insights using refined kinetic model simulations.

Materials and Methods

Chemicals and Reagents

All chemicals used in this study are listed in Table S1 in the Supporting Information (SI). All inorganic chloramine stock solutions were prepared freshly. NH₂Cl stock solution was prepared by adding NaOCl dropwise to a NH₄Cl solution stirred continuously at a 1:1 Cl₂:N molar ratio. Stock solutions of NHCl₂ were prepared by adjusting the pH of a 3.5 mM NH₂Cl solution to 3.7 with 0.2 M phosphoric acid and maintaining this condition for 30 min. The solution was then stored in an ice bath for an additional hour to allow the NH₂Cl disproportionation reaction (2 NH₂Cl + H⁺ → NHCl₂ + NH₄ ⁺) to complete, leaving a solution containing NHCl₂ and NH₄ ⁺ at approximately a 1:1 molar ratio. The ammonia-containing NHCl₂ stock solution was passed through an Amberlite IRC120 H resin to remove ammonia, yielding a final solution with a pH of 1.9 and minimal ammonia content (<5% molar ratio). Spectrophotometric analysis confirmed that the spectra (200–400 nm) for the resin-treated NHCl₂ solution remained stable for over an hour; experiments associated with NHCl₂ were conducted within 1 h. A Cary 60 UV–visible spectrophotometer was used to standardize the concentrations of inorganic chloramine stock solutions (Text S1).

Experiments for Re-evaluating Equilibrium Constants

Phenol and p-chlorophenol were used as probes for determining the equilibrium concentration of free chlorine in cyanuric acid-HOCl system. These compounds were chosen because their chlorination kinetics are well characterized, − and their reaction pathways have been extensively studied. , While both compounds undergo rapid and well-defined reactions with free chlorine, their reactivity is not identical (Table S5), making them complementary probes. Using both therefore provided a means to cross-validate the experimental results and ensured that the conclusions were not dependent on a single compound.

Experiments were carried out by treating phenol or p-chlorophenol at 2 μM with HOCl and cyanuric acid at target concentrations and at target pH. Samples were taken periodically with a stoichiometric amount of chlorine quencher (sodium thiosulfate). All experiments were conducted in duplicate.

Experiments for Evaluating Interaction between Chloramines and (Chlorinated) Cyanurates

A total of 17 experiments were conducted to investigate the interaction between cyanuric acid or chlorinated cyanurates with NH₂Cl or NHCl₂ (as listed in Table S2). Experiments were carried out in a 10 cm path length quartz cuvette mounted in a spectrophotometer for in situ measurements of UV–vis spectra at wavelengths between 230–400 nm. A 10-mM or 20-mM phosphate buffer was used to maintain the pH throughout the experiments, and UV–vis spectra were taken periodically during the reaction (scan time was ∼ 1 s). An UV absorbance/simultaneous equations method was employed to evaluate oxidant evolution and decomposition during the reactions, as detailed in Text S1. Experiments were conducted in duplicate, and the spectra were highly consistent; for simplicity, only one representative spectrum is shown.

Experiments for Evaluating Trade-Offs between Nitrosamine Formation and Micropollutant Removals

Experiments evaluating the trade-offs between micropollutant removals and nitrosamine formation in breakpoint reactions with or without cyanuric acid were carried out separately. Five model compounds (benzoate, nitrobenzene, 1,4-dioxane, N, N-Diethyl-meta-toluamide (DEET), and caffeine) that exhibit different reactivity toward •OH, •Cl, •Cl₂ ^–, •ClO, and reactive nitrogen species (Table S3), − which are potentially generated in breakpoint reactions, were chosen for evaluating micropollutant removals. Their reactivity toward •OH fell within 2.5 × 10⁹–7.5 × 10⁹ M^–1s^–1. Among them, 1,4-dioxane showed relatively low reactivity with •Cl (k = 4.4 × 10⁶ M^–1s^–1), whereas nitrobenzene and caffeine exhibit the rate constants of 5.2 × 10⁸ and 3.9 × 10¹⁰ M^–1s^–1, respectively. In addition, caffeine is susceptible to •ClO attack, with a rate constant of 1.03 × 10⁸ M^–1s^–1. DEET was also included because it reacts appreciably with reactive nitrogen species (HOONO/OONO^–, •NO, and •NO₂ ^–), with a combined rate constant on the order of 10⁹ M^–1s^–1. Our recent study demonstrates that •OH is the primary reactive species for their degradation in breakpoint reactions except caffeine.

N-Chloro-dimethylamine (Cl-DMA) was utilized as a model precursor for nitrosamines as suggested by previous studies. , Dimethylamine, excreted in human body fluids, is readily chlorinated to Cl-DMA under typical pool conditions.

Breakpoint chlorination experiments were conducted in deionized water buffered with phosphates at pH 7, using NaOCl and NH₄ ⁺ at target Cl₂/NH₄ ⁺ molar ratios (ranging from 0 to up to 5) with micropollutants at sub-μM levels (0.2–0.5 μM) or with 7.5 μM Cl-DMA. A mixture of micropollutants or Cl-DMA and cyanuric acid in 10 mM phosphates was adjusted pH using NaOH or phosphoric acid. The mixture was then subsequently dosed with ammonia and HOCl at desired concentrations to initiate the reactions. Samples were collected periodically and immediately quenched with a stoichiometric amount of chlorine quencher. Two quenchers were used depending on the subsequent analysis: ascorbic acid was applied for NDMA samples, while sodium thiosulfate was used for micropollutant samples because ascorbic acid eluted early in the HPLC-UV chromatogram and slightly interfered with quantification. The pH change during the experiments was consistently <0.2 unit.

Reaction rate constant of ClNO with Cl-DMA was experimentally determined by monitoring the decay of Cl-DMA in a mixture of NO₂ ^– and Cl^– at acidic pH. Additionally, a competition kinetics approach was employed to determine the rate constant of ClNO with H₃Cy/H₂Cy^– at pH 7 using Cl-DMA as a reference compound. Text S2 provides the experimental details.

Analytical Method

Total chlorine concentration in the breakpoint chlorination experiments, with or without cyanuric acid, was measured by the DPD method. Nitrite, nitrate, and ammonia were analyzed using a Dionex Aquion ion chromatograph with a conductivity detector. NDMA, 1,4-dioxane, and DEET were analyzed using an Agilent GC (7890B)-MS (5977A). Nitrobenzene, benzoic acid, and caffeine were analyzed using an Agilent HPLC (1260 II) coupled with a UV detector. Analytical details are available in previous studies. , Phenol, p-chlorophenol, and Cl-DMA were analyzed using an Agilent HPLC (1260 II) coupled with a UV detector (using 275 nm for phenol, 225 nm for p-chlorophenol, and 262 nm for Cl-DMA). Details for Cl-DMA analyses were provided in Text S2.

Kinetic Modeling

A kinetic model encompassing 25 elementary reactions was adapted from our recent study (Reactions S1–S25 in Table S4). This model is based on the widely used unified (UF) model, with revised reactions and rate constants for NCl₃–NHCl₂ interactions. Scheme provides the reactions. Additional revision was made for the bad reversible loop found in previous model, by revising the reversed reaction rates as suggested by a recent work. Moreover, this model was further refined with incorporating 24 reactions regarding the acid-dissociation reactions for (chlorinated) cyanuric acid/cyanurates, and six primary reactions between (chlorinated) cyanurates and NH₂Cl or NHCl₂ (Table ), as discussed below. Kinetic modeling was implemented using Kintecus 6.8. Rate constants for the additional 24 reactions were either adapted from the literature, determined in this study, or reasonably assumed. For those parameters obtained through data fitting, optimization was performed sequentially rather than simultaneously, as elaborated in a later section.

1. Primary Reactions and Rate Constants for the (Chlorinated) Cyanurates-Chloramines Interactions .

reaction		rate constant
H3Cy + NH2Cl → HClCy– + NH4 +	[R1]	k R1 = 2 M–1s–1
H2Cy– + NH2Cl → HClCy– + NH3	[R2]	k R2 < 0.01 M–1s–1
HClCy– + NH2Cl → H2Cy– + NHCl2	[R3]	k R3 = 166 M–1s–1
H3Cy + NHCl2 → H2ClCy + NH2Cl	[R4]	k R4 = negligible
H2Cy– + NHCl2 → HClCy– + NH2Cl	[R5]	k R5 = negligible
HClCy– + NHCl2 → H2Cy– + NCl3	[R6]	k R6 = 135 M–1s–1

^aReactions and rate constants were determined in this study.

Results and Discussion

Chemical Equilibrium Constants for Chlorine-Cyanuric Acid System

Initial experiments focused on evaluating the hydrolytic dissociation constants for Cl₂Cy^– (K₇ in Scheme ) and HClCy^– (K₉), driven by the fact that the wide variation in reported values introduces significant uncertainty in the predictions of free chlorine concentration at circumneutral pH (>5-fold differences; Text S3). To address this, we choose to examine the degradation kinetics of phenol in chlorine-cyanuric acid systems; phenol chlorination is one of the best-studied reactions, with established bimolecular rate constants well-established for HOCl/OCl^– reactions (Table S5). −

The degradation of phenol followed pseudo-first-order kinetics (Figure S6) due to the presence of free chlorine in substantial excess relative to phenol (i.e., 100 μM total Cl₂ relative to the 2 μM phenol). In this case, the apparent rate constant for phenol degradation (k _app) can be expressed by eq , where f _PhOH and f _PhO‑ represent the fractions of total phenol in the conjugate acid (PhOH) and phenolate (PhO^–) forms, respectively.

$$k_{\mathrm{app}}=k_{\mathrm{HOCl},\mathrm{PhO}-}{[\mathrm{HOCl}]}_{\mathrm{eq}}{f}_{\mathrm{PhO}-}+k_{\mathrm{HOCl},\mathrm{PhOH}}{[\mathrm{HOCl}]}_{\mathrm{eq}}{f}_{\mathrm{PhOH}}+\mathrm{\Sigma }{k}_{\mathrm{Cl}-\mathrm{cyanurate},\mathrm{PhO}-}{[\mathrm{Cl}-\mathrm{cyanurate}]}_{\mathrm{eq}}{f}_{\mathrm{PhO}-}+\mathrm{\Sigma }{k}_{\mathrm{Cl}-\mathrm{cyanurate},\mathrm{PhOH}}{[\mathrm{Cl}-\mathrm{cyanurate}]}_{\mathrm{eq}}{f}_{\mathrm{PhOH}}$$ 1

While k _{HOCl, PhO‑} are orders of magnitude larger than k _HOCl, PhOH (Table S5), experimental results in this study indicated that Cl-cyanurates exhibit negligible reactivity with phenol and phenolate (Figure S7). Furthermore, the reaction involving OCl^– is excluded due to its significantly lower reactivity. , As a result, eq simplifies to eq :

$$k_{\mathrm{app}}=k_{\mathrm{HOCl},\mathrm{PhO}-}{[\mathrm{HOCl}]}_{\mathrm{eq}}{f}_{\mathrm{PhO}-}+k_{\mathrm{HOCl},\mathrm{PhOH}}{[\mathrm{HOCl}]}_{\mathrm{eq}}{f}_{\mathrm{PhOH}}$$ 2

In this simplified form, k _app is solely dependent on the equilibrium concentration of HOCl ([HOCl]_eq). The equilibrium constants K₇ and K₉ from various studies along with other relevant constants were used to calculate [HOCl]_eq and the corresponding k _app using eq . These calculated values were then compared with experimentally observed k _app to determine which set of K₇ and K₉ provided the best match.

Figures a–b demonstrates that the values of pK ₇ (4.1) and pK ₉ (4.8) reported by Pinsky and Hu and Jensen and Johnson provided the most accurate k _app predictions from experiments with a variety of conditions, including those with different cyanuric acid concentration (0–300 μM) and different pH (6–8). The relative percentage errors remained within 15%, whereas predictions using pK ₇ and pK ₉ from other studies exhibited errors as high as 164%. Further validation using p-chlorophenol as a probe (k _HOCl and k _OCl‑ provided in Table S5) also demonstrated a strong agreement between model predictions and experimental k _app with low errors (∼20%, Figure c). These findings support the accuracy of K₇ and K₉ values determined via electrochemical methods, and highlight the importance of refining these constants to enhance the robustness of equilibrium-based approaches for chlorine speciation.

1.

Experimentally observed k (k _app) for the degradation of (a) phenol at pH 7 with different cyanuric acid concentration, (b) phenol at pH 6–8 with 100 μM cyanuric acid concentration, and (c) p-chlorophenol at pH 7 with different cyanuric acid concentration, and the predictions calculated using eq in which the equilibrium concentration of HOCl were calculated using K₇ and K₉ reported from the literature. A 10-mM phosphate buffer was used to maintain the pH, with 2 μM phenol or p-chlorophenol treated by 100 μM HOCl and 0–300 μM cyanuric acid. The K₇ and K₉ reported from different works are summarized in Figure S4. References: P. and H.; J. and J.; O’Brien et al.; Gardner; Brady et al. (d) Experimental and modeled total chlorine concentrations during the reactions of 100 μM HOCl with 50 μM NH₄ ⁺ and 50 or 450 μM cyanuric acid (Cy). Model simulations were implemented with and without inclusion of reactions R1–R6 from Table . Error bar represents data range from experimental duplicates.

Development of Kinetic Model: Interactions between Cyanuric Acid and Chloramines

Kinetic models for evaluating micropollutant degradation and DBP formation need to reliably predict (1) radical formation and (2) the dynamic concentrations of oxidants during breakpoint chlorination in the presence of cyanuric acid. These predictions are essential to elucidate the competition among target contaminants, oxidants, and matrix components for reactions with these radicals. In this study, the chlorine-cyanuric acid reactions (Reactions S26–S49 in Table S4) were integrated into a previously developed breakpoint chlorination kinetic model. As the equilibrium constant (K) for an elementary reversible reaction is defined by the ratio of forward (k _f) and reverse (k _r) reaction rate constants (i.e., K = k _f/k _r), the knowledge of any one parameter enables the derivation of the others. Acid–base reactions are generally rapid, such that we assumed a 1 × 10¹⁰ M^–1s^–1 (diffusion-controlled limit) for the k values of base association reaction (e.g., H⁺ + ClCy^2– → HClCy^–; k = 1 × 10¹⁰ M^–1s^–1). Although exceptions exist (e.g., nitronate anions), sensitivity analysis showed that varying these rate constants from 1 × 10⁶ to 1 × 10¹⁰ M^–1s^–1 had negligible effects on predicted chlorine loss (see later section).

However, the rate constants that have been reported involve reactions of nonchlorinated cyanuric acid species (H₂Cy^– and HCy^2–) with HOCl to form the corresponding monochlorocyanurates, with values of 7.27 × 10⁴ M^–1s^–1 and 2.16 × 10⁷ M^–1s^–1, respectively. By contrast, the kinetics of reactions between HOCl and chlorinated cyanurates (leading to higher chlorinated derivatives) remain largely unreported. We adopted a reasonable estimate of 2 × 10⁵ M^–1s^–1 for the remaining reactions between (chlorinated) cyanurates and HOCl. This estimate was based on findings by Jensen and Johnson, who used stopped-flow spectroscopy to determine that the reaction of DPD with HOCl (k _HOCl for DPD = 3.15 ± 0.03 × 10⁶ M^–1s^–1) had a reaction half-time of <0.4 s in HOCl-cyanuric acid systems. Our chosen rate constant (2 × 10⁵ M^–1s^–1) yielded a reaction half-time of 0.4 s under their experimental conditions (Text S4). When incorporating these rate constants, the kinetic model successfully predicted the degradation kinetics of phenol and p-chlorophenol in HOCl-cyanuric systems across all experimental conditions (Figure S9).

The kinetic model was employed to simulate total chlorine losses during breakpoint reactions. While the model accurately captured chlorine loss kinetics in the absence of cyanuric acid, it significantly underestimated the rate when cyanuric acid was present (Figure d and Figure S10). For instance, in experiments with 50 μM and 450 μM cyanuric acid, the measured total chlorine losses after 10 min were 74% and 70%, respectively, substantially higher than the model predictions of 46% and 5%. The discrepancy between the modeled and experimental results suggests that cyanuric acid or chlorinated cyanurates (particularly H₃Cy, H₂Cy^–, and HClCy^–, the dominant species under pool disinfection conditions) may actively facilitate the chlorination of ammonia, NH₂Cl, or NHCl₂ (i.e., Reactions R1–R6 in Table ), counteracting the slow conversions due to the low equilibrium concentration of free chlorine.

Further experiments were carried out to investigate the interactions among HOCl, cyanuric acid, and chloramines. Spectral measurements of a mixture containing 400 μM cyanuric acid (in 10 mM phosphates at pH 7) treated with 50 μM NH₂Cl showed that absorbance in the 278–330 nm range increased over time, while absorbance below 278 nm decreased (Figure ). Since all cyanurate species and NH₂Cl exhibit limited absorbance above 280 nm (Figure S1a), the emerging signal in the 278–330 nm range potentially indicates the formation of NHCl₂ which has a characteristic absorbance peak at 292 nm (Figure S1a). The spectra also revealed an isosbestic point at 277 nm, where the molar absorption coefficient of NHCl₂ (ε_{NHCl2, 277 nm} = 169 M^–1cm^–1) is twice that of NH₂Cl (ε_{NH2Cl, 277 nm} = 84 M^–1cm^–1). This suggests that two NH₂Cl molecules are consumed to generate one NHCl₂ during the cyanuric acid-mediated NH₂Cl conversion, leading to a total absorbance change at 277 nm (ΔA_277 nm = – 2×ε_{NH2Cl, 277 nm} + 1×ε_{NHCl2, 277 nm}) of zero. Similar spectral patterns were obtained when repeating the experiments at pH 6, 6.5, and 7.5, where NH₂Cl remains stable (<2% change) over a 20 min time scale (Figures S11–S13).

2.

Time-dependent (a) UV spectra of a mixture containing 400 μM cyanuric acid in 10 mM phosphates at pH 7 treated with 50 μM NH₂Cl, and (b) Concentrations of NH₂Cl and NHCl₂ over time, determined by the UV absorbance/simultaneous equations method. Symbols represent experimental data and lines are model predicted results. The hollow circle in (a) denotes the isosbestic point. Error bar represents data range from experimental duplicates.

A UV spectra/simultaneous-equation approach used to assess the evolution and decomposition of chloramines confirmed that NH₂Cl decomposed with the concomitant formation of NHCl₂ (Figure b and Figures S11b–S13b). Across the tested pH range of 6–7.5, the reaction followed a consistent stoichiometric ratio of 2.1(±0.1) NH₂Cl moles depleted per mole NHCl₂ formed, aligning with the isosbestic point at 277 nm observed in the spectra. The data further revealed that as pH increased from 6.0 to 7.5, both the decomposition of NH₂Cl and the formation of NHCl₂ slowed substantially. This trend points to the protonated form of cyanuric acid (H₃Cy, pK _a = 6.88) being more effective than its deprotonated form (H₂Cy^–) in driving the conversion of NH₂Cl to NHCl₂. In line with this interpretation, kinetic analysis (Text S5) showed that NH₂Cl decomposition in phosphate-buffered NH₂Cl–cyanuric acid systems followed second-order kinetics, with apparent rate constants (k _app) decreasing from 26.4 ± 0.1 M^–1s^–1 at pH 6.0 to 1.7 ± 0.1 M^–1s^–1 at pH 7.5.

Previous work by Valentine and Jafvert demonstrated that NH₂Cl disproportionation to NHCl₂ (NH₂Cl + NH₂Cl → NHCl₂ + NH₃; Reaction S5 in Table S4) proceeds via a general acid-catalyzed mechanism. In this pathway, protonation of NH₂Cl produces NH₃Cl⁺, which rapidly chlorinates a second NH₂Cl to generate NHCl₂, NH₃, and H⁺. Accordingly, the overall second-order rate constant for NH₂Cl disproportionation (i.e., d[NH₂Cl]/dt = k _S5×[NH₂Cl]²) was expressed as the sum of contributions from proton-donating species. We extended this framework by recognizing that both H₃Cy and H₂Cy^– bear exchangeable protons and may therefore catalyze NH₂Cl disproportionation. Incorporating these terms yields eq .

$$k_{\mathrm{S}5}=k_{\mathrm{H}+}[{\mathrm{H}}^{+}]+k_{\mathrm{H}3\mathrm{PO}4}[{\mathrm{H}}_3{\mathrm{PO}}_4]+k_{\mathrm{H}2\mathrm{PO}4-}[{\mathrm{H}}_2{{\mathrm{PO}}_4}^{-}]+k_{\mathrm{H}3\mathrm{Cy}}[{\mathrm{H}}_3\mathrm{Cy}]+k_{\mathrm{H}2\mathrm{Cy}-}[{\mathrm{H}}_2{\mathrm{Cy}}^{-}]$$ 3

We estimated k _H3Cy and k _H2Cy‑ using linear free-energy relationships which relate the specific rate constant of a proton donor to its pK _a, the number of exchangeable protons, and the binding capacity of its conjugate base (Text S5). The resulting values, 0.37 M^–2s^–1 for H₃Cy and 4.4 × 10^–4 M^–2s^–1 for H₂Cy^–, were on par with those for H₂CO₃ (0.68 M^–2s^–1) and HCO₃ ^– (1.7 × 10^–3 M^–2s^–1). When the concentrations of each acid were substituted into the expanded expression, the calculated k _S5 values (1.4 × 10^–3–1.4 × 10^–2 M^–1s^–1) were more than 3 orders of magnitude lower than the experimental k _app values obtained in the NH₂Cl–cyanuric acid systems (1.7–26.4 M^–1s^–1). This large discrepancy indicates that acid-catalyzed disproportionation alone cannot explain the rapid NH₂Cl decay observed in the presence of cyanuric acid.

To test this further, we performed control experiments in carbonate buffer (pH 7), where the H₂CO₃ concentration (173 μM) matched the H₃Cy concentration present in NH₂Cl–cyanuric acid experiments (50 μM NH₂Cl and 400 μM cyanuric acid at pH 7). Under these conditions, NH₂Cl remained essentially stable over 20 min (Figure S15), in stark contrast to the rapid decay observed in the presence of cyanuric acid. These findings demonstrate that cyanuric acid promotes NH₂Cl conversion through additional pathways beyond simple acid catalysis.

Zhang and von Gunten’s recent investigations into chlorine reactions with amides demonstrate that OCl^– reacts with amides to form N-chloroamides through the formation of a stable complex. This structure features a hydrogen bond between the oxygen in OCl^– and the hydrogen in amide, enabling the amide N to attack the partially positively charged – Cl in OCl^–. By analogy, the interaction between H₃Cy and NH₂Cl may proceed through the formation of a hydrogen bond between the nitrogen in NH₂Cl and the hydrogen in H₃Cy. This interaction facilitates a nucleophilic attack by the nitrogen in H₃Cy on the chlorine in NH₂Cl, generating H₂ClCy and NH₃ (Reaction R1 and Scheme S1a). Once formed, H₂ClCy undergoes Cl­(+1) transfer reactions with another NH₂Cl molecule to produce NHCl₂ and regenerates cyanuric acid (Reaction R3 and Scheme S1b). These reactions follow an overall stoichiometry in which two NH₂Cl molecules are consumed to produce one NHCl₂ molecule. Cyanuric acid thus appears to act as a catalyst that promotes NH₂Cl disproportionation and facilitates NHCl₂ formation, mirroring with the catalytic role of amides in amides-OCl^– interactions.

Our results suggest a two-step process in which H₃Cy catalyzes the conversion of 2 NH₂Cl to 1 NHCl₂, though the rate-limiting step remains unknown. To investigate this, we treated a solution containing 20 μM HOCl pre-equilibrated with 400 μM cyanuric acid by 20–125 μM NH₂Cl. The large excess of cyanuric acid relative to HOCl ensured that more than 96% of the chlorine remained cyanurate-bound, with HClCy^– accounting for 93% of the total Cl₂. The results (Figure S16) revealed a biphasic pattern in both NHCl₂ formation and NH₂Cl decomposition, with an initial rapid phase lasting approximately 2 min followed by a slower phase. The kinetics of NHCl₂ formation and NH₂Cl decomposition were significantly enhanced in the HOCl-equilibrated cyanuric acid solution compared to a solution containing only cyanuric acid and NH₂Cl during the first 2 min (Figure S17), suggesting Reaction R1 is the rate-determining step (i.e., k _R3 > k _R1). Therefore, this biphasic behavior arises from the interplay between a rapid reaction of HClCy^– with NH₂Cl to form NHCl₂ (R3) and a concurrent slower, cyanuric acid-facilitated transformation of NH₂Cl to NHCl₂ (R1–R3). The increasing stoichiometry of NH₂Cl consumption per NHCl₂ formed over time (Figure S16) supports the progressive contribution of both mechanisms.

Although NHCl₂ may interact with H₃Cy in a similar fashion to NH₂Cl, our results suggest that H₃Cy is less effective at catalyzing Cl­(+1) transfer between two NHCl₂ molecules than between two NH₂Cl molecules. When NHCl₂ was treated with H₃Cy alone at pH 7 for 5 min, only 10% decomposition was observed (Figure S18), which was half the 20% decomposition seen when NH₂Cl was treated under the same conditions (Figure b). This can be explained by the relative electron-withdrawing nature of – Cl on NHCl₂ compared to – H on NH₂Cl, lowering the tendency of forming the hydrogen bond between the – N on NHCl₂ and – H on H₃Cy. However, NHCl₂ decomposition was significantly accelerated in the 20 μM HOCl-equilibrated 400 μM cyanuric acid solution, reaching 70–80% within 5 min (Figure S19). While NCl₃ is likely formed as an intermediate (R6), the newly formed NCl₃ rapidly reacts with NHCl₂, further accelerating NHCl₂ decomposition (U12 in Scheme ). As a result, the spectra did not display the characteristic absorption peak of NCl₃ at 365 nm (Figure S19a).

Rate Constants for Reactions between Cyanuric Acid and Chloramines

Plotting the initial decay rates for NH₂Cl or NHCl₂ (i.e., – d­[NH₂Cl]/dt or – d­[NHCl₂]/dt) against [NH₂Cl]_initial (or [NHCl₂]_initial)×[HClCy^–]_initial from all experiments yielded linear lines (Text S6 and Figure S20). Accordingly, the rate constants of k _R3 (HClCy^– + NH₂Cl) and k _R6 (HClCy^– + NHCl₂) were determined to be 166 and 135 M^–1s^–1, respectively. These rate constants are approximately half of those for HOCl with NH₂Cl (278 M^–1s^–1) and NHCl₂ (330 M^–1s^–1 at pH 7), reflecting that HClCy^– is less effective than HOCl in chlorine transfer reactions to chloramines.

These reactions were incorporated into the kinetic model to determine species-specific reaction rate constants for k _R1 and k _R2, by implementing the refined model to optimize k ₅ and k ₆ against experimental results for cyanuric acid-NH₂Cl reaction at pH 6–7.5 (Text S7). The results showed that the rate constant for reaction of NH₂Cl with H₃Cy (k _R1 = 2 M^–1s^–1) or H₂Cy^– (k _R2 < 0.01 M^–1s^–1) is several orders of magnitude lower than that for the reaction between HOCl with H₂Cy^– (7.3 × 10⁴ M^–1s^–1).

When incorporating Reactions R1–R6 and the respective rate constants, the refined kinetic model successfully predicted NH₂Cl and NHCl₂ concentrations across all HOCl–cyanuric acid–chloramine experiments listed in Table S2 (as shown in Figure b and Figure S11–S13, S16, and S21), with most deviations between predicted and measured values remaining within twice the method detection limit (2 × MDL ∼ 1.6 μM). Furthermore, the model significantly improved its ability to predict the total chlorine losses (relative errors <5%; Figure d and Figure S10), emphasizing the critical role of chlorinated cyanurates in the chlorination of ammonia, NH₂Cl, and NHCl₂. Our results also suggest that chlorinated cyanurates are more selective chlorination agents than HOCl, as they showed negligible reactivity with phenols despite their similar molecular weights to NH₂Cl/NHCl₂.

Impact of Cyanuric Acid in Model Micropollutant Removals and DBP Formation

In the absence of cyanuric acid, the degradation of the five HOCl-resistant micropollutants during breakpoint chlorination followed a distinct volcano-shaped trend. Micropollutant removals increased with rising Cl₂/NH₄ ⁺ molar ratios, peaking around the breakpoint (1.8–2.0) before tapering off at higher ratios (Figure and Figure S22), consistent with prior studies. ,,

3.

Experimental and modeled removals for (a) 1,4-dioxane, (b) DEET, and (c) caffeine during the chlorination of a mixture of 50 μM NH₄ ⁺ and five micropollutants at 0.2 or 0.4 μM at pH 7 by 0–200 μM HOCl with or without cyanuric acid. (a)–(c) share the same legend. (d) Modeled cumulative formation concentration for •OH, and modeled and experimental NCl₃ formation concentration at 60 min after the treatments (experimental details and analyses provided in Text S1-D). (e) NDMA formation concentration at 60 min after the treatments of a solution containing 7.5 μM Cl-DMA and 50 μM NH₄ ⁺ by HOCl at pH 7.0 with 0–300 μM cyanuric acid. Error bar represents data range from experimental duplicates.

Kinetic modeling well captured these trends and offered mechanistic insights. The observed removal pattern is primarily governed by the interplay between •OH generation and •OH scavenging by free chlorine. At the breakpoint, cumulative •OH formation reaches its maximum due to a series of reactions through which NHCl₂ is the hub. Following its formation, NHCl₂ reacts with HOCl to form NCl₃, which subsequently reacts with NHCl₂ to yield •OH, N₂, ClNO while regenerating NHCl₂ through parallel pathways (Reactions U12, U12b, and U12c in Scheme ). This cyclic process results in an observed stoichiometry where 2.6 mol of NCl₃ are consumed per mole of NHCl₂. The optimal Cl₂/NH₄ ⁺ ratio (i.e., at breakpoint) enhances NHCl₂–NCl₃ interactions while minimizing the residual levels of both NHCl₂ and NCl₃ after reaction (as reported in the literature , and in Figure d), thus maximizing •OH production and micropollutant degradation (Figure a–c).

Beyond the breakpoint (Cl₂/NH₄ ⁺ > 2), the chemical dynamics shift. Excess HOCl rapidly drives the conversion of NHCl₂ to NCl₃ (via U11 in Scheme ), reducing the availability of NHCl₂ for •OH-generating reactions. As a result, NCl₃ accumulates and •OH formation declines (Figure d). Additionally, free chlorine competes with micropollutants as a major •OH scavenger. Model simulations indicate that free chlorine scavenges approximately 75% of the •OH formed at Cl₂/NH₄ ⁺> 2.5, compared to 53% at the breakpoint (Figure S23).

The presence of cyanuric acid significantly altered this behavior by enhancing micropollutant removals and mitigating the postbreakpoint decline. At Cl₂/NH₄ ⁺ = 4, removals for micropollutants except caffeine increased from <15% without cyanuric acid to 56–88% with 450 μM cyanuric acid. Two key mechanisms contribute to this enhancement. First, cyanurate-bound HClCy^– converts NHCl₂ to NCl₃ (i.e., reaction R6) at a rate approximately one-third that of HOCl, thereby prolonging the lifetime of NHCl₂ and expanding the window for NHCl₂–NCl₃ interactions; model predictions show 12–36% higher cumulative •OH formation within the Cl₂/NH₄ ⁺ range of 2.5–4 when cyanuric acid is present (Figure d). Second, cyanuric acid markedly reduced •OH scavenging by free chlorine by more than 50% over the same Cl₂/NH₄ ⁺ range (Figure S23), thus increasing the fraction of •OH available for micropollutant degradation. Cyanuric acid also suppressed NCl₃ accumulation; modeled NCl₃ concentrations were more than 40% lower with cyanuric acid than without (Figure d), consistent with the extended NHCl₂ reaction window and increased •OH generation. Experimental measurements confirmed these predictions (Figure d).

Structural trade-offs among micropollutants emerged in both systems. •OH scavenging by HOCl generates •ClO, a selective oxidant that is of high reactivity with imidazole moieties. This accounts for the nearly complete removal of caffeine postbreakpoint (Figure c). With cyanuric acid, reduced HOCl scavenging lowered •ClO formation and thus decreasing caffeine removal, although the model predictions did not capture this behavior due to the lack of comprehensive •ClO-related reactions. When the experiments were repeated with real pool water, a similar but less pronounced removal pattern was observed, attributable to radical scavenging by additional inorganic and organic constituents (Figure S24).

Lastly, cyanuric acid significantly affected nitrosamine formation. In separate experiments using Cl-DMA as the NDMA precursor, NDMA formation in the absence of cyanuric acid followed a volcano-shaped trend with respect to the Cl₂/N molar ratio (Figure e), consistent with patterns observed in micropollutant degradation. Increasing cyanuric acid concentrations to 300 μM broadened the NDMA formation profiles, with peak concentrations shifting toward higher Cl₂/N ratios and overall maximum NDMA levels decreasing. These trends suggest that ClNO generated during breakpoint chlorination may react not only with HOCl/OCl^– but also with (chlorinated) cyanurates, and the observed NDMA formation behavior likely reflects the competitive consumption of ClNO among oxidants, cyanuric acid, and precursors. Indeed, competition kinetics using ClNO vapor revealed a k _ClNO for H₃Cy/H₂Cy^– of 5.8 × 10⁵ M^–1s^–1 at pH 7 (Text S2), approximately an order of magnitude lower than that for Cl-DMA (6.8 × 10⁶ M^–1s^–1). Although the k _ClNO for HOCl/OCl^– could not be determined due to limitations of the vapor-phase approach, model simulations suggest it lies between the values for ClNO with cyanuric acid and ClNO with Cl-DMA (additional discussion in Text S8). This inference is supported by the closer agreement between model-predicted concentration of products from the ClNO–Cl-DMA reactions and experimentally observed NDMA formation concentrations. Nonetheless, further investigation is warranted to refine the kinetic model by comprehensively characterizing the aqueous-phase reactivity and fate of ClNO.

Environmental Implications

Cyanuric acid is widely used in outdoor swimming pools as a stabilizer to reduce chlorine photodegradation. While its role in preserving chlorine residuals is well recognized, its broader impact on chlorination chemistry and DBP formation in ammonia-containing water has remained underexplored. Our experimental results show that chlorinated cyanurates (e.g., HClCy^–) function as effective chlorinating agents for NH₂Cl and NHCl₂, with reaction rates approximately half those of HOCl. These findings suggest that chlorinated cyanurates may contribute to chlorination of low-molecular-weight amines, which are important precursors for nitrogenous DBPs, although they exhibit low reactivity toward phenolic compounds.

Accurate estimation of free chlorine concentrations, which is critical for assessing disinfection efficacy, depends on reliable hydrolytic dissociation constants for chlorinated cyanurates. Using phenols as probes, we confirm that the electrochemically determined values for Cl₂Cy^– (pK ₇ = 4.1) and HClCy^– (pK ₉ = 4.8) better predict free chlorine levels than previously reported spectrophotometric values, yielding concentrations nearly five times higher under typical pool conditions. These insights are particularly important given the limitations of standard DPD-based chlorine measurements, which cannot differentiate HOCl from chlorinated cyanurates.

Although cyanuric acid is primarily used in outdoor pools, it is occasionally applied in indoor settings in some countries (e.g., China). , Because cyanuric acid slows the formation and accumulation of trichloramine, its use in indoor pools may help mitigate exposure to this volatile irritant, which is associated with skin, eye, and respiratory symptoms.

Swimmer-derived inputs such as ammonia, organic nitrogen, and micropollutants can lead to substantial local variation in chlorine-to-nitrogen ratios within pool water. The presence of cyanuric acid alters the fate of chlorine-derived radicals and nitrosating agents in such microenvironments, leading to complex trade-offs between disinfection efficacy, contaminant removal, and DBP formation. This is supported by our kinetic modeling (Text S9), which simulated micropollutant removal and nitrosamine formation potential under realistic pool conditions where free chlorine is steadily maintained while ammonia is continuously introduced by swimmers. The prolonged lifetime for •OH, while enhancing micropollutant degradation, may also compensate the reduced disinfection efficacy due to low HOCl equilibrium concentration. Prior research highlights that breakpoint-chlorination-induced reactive species improve disinfection of hard-to-kill organisms like Amoeba spores and their internal bacteria.

Lastly, given that cyanuric acid is predominantly applied in sunlit outdoor pools, the potential influence of irradiation on the kinetics and mechanisms of chlorine–ammonia–cyanuric acid reactions, as well as their implications for DBP formation, warrants further investigation.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c09618.

• Chemical lists, additional experimental details and analytical methods, figures for micropollutant removals during the treatments of breakpoint chlorination with or without cyanuric acid, kinetic model development, and additional experimental results and discussion. (PDF)

The authors declare no competing financial interest.

Published as part of Environmental Science & Technology special issue “Celebrating the 50th Anniversary of the Discovery of Drinking Water Disinfection Byproducts”.