Persulfate-Based Advanced Oxidation: Critical Assessment of Opportunities and Roadblocks

Cite this: Environ. Sci. Technol. 2020, 54, 6, 3064–3081
Publication Date (Web):February 16, 2020
Copyright © 2020 American Chemical Society
ACS AuthorChoice
Article Views
PDF (3 MB)


Reports that promote persulfate-based advanced oxidation process (AOP) as a viable alternative to hydrogen peroxide-based processes have been rapidly accumulating in recent water treatment literature. Various strategies to activate peroxide bonds in persulfate precursors have been proposed and the capacity to degrade a wide range of organic pollutants has been demonstrated. Compared to traditional AOPs in which hydroxyl radical serves as the main oxidant, persulfate-based AOPs have been claimed to involve different in situ generated oxidants such as sulfate radical and singlet oxygen as well as nonradical oxidation pathways. However, there exist controversial observations and interpretations around some of these claims, challenging robust scientific progress of this technology toward practical use. This Critical Review comparatively examines the activation mechanisms of peroxymonosulfate and peroxydisulfate and the formation pathways of oxidizing species. Properties of the main oxidizing species are scrutinized and the role of singlet oxygen is debated. In addition, the impacts of water parameters and constituents such as pH, background organic matter, halide, phosphate, and carbonate on persulfate-driven chemistry are discussed. The opportunity for niche applications is also presented, emphasizing the need for parallel efforts to remove currently prevalent knowledge roadblocks.


Jump To

Advanced oxidation processes (AOPs) employ highly reactive hydroxyl radical (OH) to abate a wide range of organic pollutants in water with diffusion-limited kinetics. Since OH is short-lived, it is generated in situ during ozone- and UV-based processes (1) by activating stable precursors, such as H2O2. (2,3) Alternative AOPs utilizing peroxymonosulfate (PMS) or peroxydisulfate (PDS) (collectively referred to as persulfate; see Figure 1 for their structures) instead of H2O2 have emerged based on the same strategy. In sulfate radical-based AOPs (referred to herein as SO4•–-AOPs), highly reactive, short-lived sulfate radicals (SO4•–) are produced in situ by cleaving the peroxide bond in the persulfate molecule via energy and electron transfer reactions. (4−6) Unlike H2O2, however, persulfate can also oxidize some organics directly, without involving radical species. (7−13) In the current literature, “persulfate-AOPs” refer to any physicochemical method that enhances the oxidizing capacity of persulfate regardless of involvement of radicals.

Figure 1

Figure 1. Activation of peroxymonosulfate (PMS) and peroxydisulfate (PDS) through electron- and energy-transferring processes.

Originally introduced for soil and groundwater remediation in the late 1990s to overcome the technical limitations of H2O2, (14,15) over the past decades, persulfate-AOPs have drawn a significant attention as a viable alternative to traditional OH-based AOPs in water and wastewater treatment. A simple comparison of redox potentials of key radical species, that is, E0(SO4•–/SO42–) = +2.60 – +3.10 VNHE > E0(OH/OH) = +1.90 – +2.70 VNHE, (16) initiated a lot of optimism. (4−6,9) Other technical advantages of persulfate-AOPs over H2O2-AOPs that have been identified include: (i) the higher achievable radical formation yield, (17−20) (ii) a wider variety of methods available to activate persulfate, (4−9,11−13) (iii) less dependence of the treatment efficiency on the operational parameters (e.g., pH, initial peroxide loading, background constituents), (21−24) and (iv) lower costs of storage and transportation due to the availability of persulfate salts. A recent surge in scientific publications, mostly on developing persulfate activation strategies, also reflects the optimism that currently prevails in academic research. (5,6,25−27)
Gauging the true potential of this technology as a substitute for a process that is already well established in the industry requires a careful evaluation of compounding factors such as water matrix effects, byproduct formation and toxicological consequences, costs, and engineering challenges. However, conflicting views on the identity of major oxidants and the mechanisms of persulfate activation, pollutant degradation, and background constituent influence (e.g., humic substance, chloride, and bicarbonate) exist in persulfate literature. The chemistry and mechanisms are very different as compared to OH-AOPs, despite the similarity in the concept how they were both initially designed, to an extent that some persulfate-AOPs do not carry features that distinguish AOPs from other oxidation processes (e.g., low selectivity of OH). Such differences originate primarily from unique reactivity of PMS/PDS and the involvement of differential radical (e.g., SO4•–) and nonradical species (e.g., singlet oxygen (1O2)). Accordingly, this critical review scrutinizes the chemistry involved in these processes with special attention to similarities and differences to OH-AOPs in order to define challenges for prioritizing future studies in this field.

Sulfate Radical: Production and Properties

Jump To

In Situ Production by Persulfate Activation

The primarily intended goal of persulfate-AOPs is a release of a large amount of SO4•– by homolytically or heterolytically cleaving the peroxide bond in the persulfate molecule. (6,16,28) In water treatment, the peroxide bond activation (Figure 1) can be achieved by an input of energy in the form of photons (UV photolysis) or heat (thermolysis). In most often adopted approaches, peroxide bond-breaking redox reactions are initiated by direct electrolysis or by reduced metals, (Fe0, Co(II)), metal oxides (e.g., CuCo2O4, LnMnO3), and some composites (e.g., Fe/Co, Co3O4/C3N4). (4,29) The occurrence of SO4•– has been well-documented (e.g., emergence of the maximum transient absorption at 450 nm with a molar extinction coefficient of 460 ± 25 M–1 cm–1). (30,31) Its role in pollutant oxidation is also well-established based on observations, such as (i) delayed pollutant degradation by addition of radical scavengers, (17) (ii) formation of halogen-containing products (e.g., chlorinated phenols and BrO3) in the presence of excess halide ions, (32,33) (iii) electron paramagnetic resonance (EPR) detection of radical adducts, (34) and (iv) product distribution and substrate-specificity that align well with the reactivity of SO4•–. (31,35,36)
SO4•– can rapidly oxidize a range of organic pollutants, even leading to mineralization of select compounds upon extensive exposure. (37) But both “the range of target pollutants” and “mineralization potential” need to be advocated with caution. Although the oxidation potential of SO4•– is comparable to that of OH, (16) SO4•–-driven oxidation is much more selective. Therefore, simply comparing the oxidation potential of SO4•– versus OH to promote the SO4•–-AOP is unjustified. Considering that the AOP is often synonymously taken as broadband abatement of organic compounds because of its root in less selective OH-based processes, this distinction is significant. The selectivity of SO4•– explains why it is used for persulfate-mediated organic synthesis (e.g., free radical pathway for hydroxylation of aromatics) instead of other oxidants. (38) Below we examine both similarities and differences between SO4•– and OH that are critical to water treatment applications.

Substrate Specific Reactivity

Organic compounds react with SO4•– and OH via similar pathways; (i) hydrogen abstraction, (ii) electron transfer, and (iii) addition–elimination. The difference lies in a preferred reaction pathway and the reaction kinetics. For example, the oxidation of saturated hydrocarbons such as alkanes and aliphatic alcohols by SO4•– is known to proceed through hydrogen abstraction similar to OH. (39,40) This is supported by the substantial kinetic isotope effect observed in oxidation of deuterated compounds. (39) However, second-order rate constants for the reactions of SO4•– with typical alkanes (e.g., ethane and propane) and aliphatic alcohols (e.g., methanol and ethanol) are 2 or 3 orders of magnitude smaller than those for OH. (39,40) Furthermore, the rate of hydrogen abstraction by SO4•– varies by one or 2 orders of magnitude depending on the degree of alkylation and the type of functional groups. (39,40) This is in marked contrast to the reactivity of OH which is less sensitive to chemical surroundings. For instance, the second order rate constant for hydrogen abstraction by SO4•– drastically increases with the increasing alkyl chain length; k(ethane) = 5.6 × 106 M–1 s–1, k(propane) = 4.7 × 107 M–1 s–1, and k(2-methylpropane) = 9.9 × 107 M–1 s–1. (40) Likewise, electron-donating groups such as alkyl, allyl, and hydroxyl moieties make the α hydrogen more susceptible to abstraction by SO4•–: k(1-propanol) = 5.9 × 107 M–1 s–1, k(2-methyl-1-propanol) = 1.3 × 108 M–1 s–1, and k(2-propen-1-ol) = 1.4 × 109 M–1 s–1. (39,40) In contrast to SO4•–, the second-order rate constants for the corresponding reactions with OH only range from 2.8 × 109 M–1 s–1 to 3.4 × 1010 M–1 s–1. (41,42)
The difference in reaction mechanisms for SO4•– and OH is quite evident when the routes of reactions with aliphatic carboxylic acids are compared (Figure 2). The first step of OH-induced oxidation is predominantly hydrogen abstraction from carbons in the aliphatic chain attached to the carboxyl group, thus producing carbon-centered radicals, which react further by oxygen addition and then to various products according to the Russel- or Bennett-type reactions. (43) In contrast, SO4•– preferentially abstracts an electron from oxygen in the carboxyl group. (35,44−46) A resulting carboxyl radical (RCO2) releases CO2 and an alkyl radical. The decarboxylation is unique to SO4•– (e.g., mineralization of acetic acid to CO2 involving few intermediates) and distinguishes SO4•–-AOPs from OH-AOPs with respect to the product distribution. (35) Aromatic carboxylic acids also undergo decarboxylation by SO4•– that proceeds sequentially through (i) formation of an aromatic radical cation through direct electron abstraction and (ii) CO2 loss accompanying (substituted) an aryl radical release (Figure 2). (45) Select aromatic carboxylic acids such as benzoic and phthalic acids decarboxylate in SO4•–-AOPs whereas hydroxylation is more common in conventional OH-mediated AOPs. (36,47)

Figure 2

Figure 2. Pathways for the oxidation of acetate and benzoate by hydroxyl radical and sulfate radical.

Electron abstraction and addition–elimination mechanisms are also important routes of reactions between SO4•– and aromatic compounds. Short-lived intermediates after electron abstraction, observed by EPR or transient absorption spectra, include (i) radical cations, (ii) SO4•– adducts, or (iii) OH adducts. (46,48−50) Hydroxycyclohexadienyl-type radicals (i.e., OH adduct), found typically after the electrophilic addition of OH to aromatic rings, also result from either a hydration of the radical cation or SO4•– addition followed by elimination. (46,51) The selective nature of SO4•– makes the reactivity toward aromatic compounds highly sensitive to substituent effects. This is obvious from Hammett-type correlations of SO4•– and OH; ρ = −2.4 for SO4•– and ρ = −0.5 for OH were determined from linear correlations of the logarithm of the relative second-order rate constants with Hammett substituent constant (log(k/k0) = ρσ, where k = second-order rate constant for benzene, k0 = second-order rate constant for substituted benzene, and σ = Hammett substituent constant). (31,50,52) Depending on the type of substituents, reaction pathway and products after the initial attack by SO4•– also vary significantly. Long-lived aromatic radical cations due to electron-donating groups tend to rearrange to undergo side chain oxidation. For instance, a radical cation generated during p-toluic acid (or gallic acid (trihydroxybenzoic acid)) oxidation by SO4•– rearranges to the corresponding benzyl radical (or phenoxyl radical), which further converts to benzyl alcohol (or biphenyl). (45,53) Dimers and trimers formed from phenoxyl radicals are also characteristic products of SO4•–-AOPs. (54−56) Such products are barely observed in OH-AOPs. Haloaromatic radical cations destabilized by electron-withdrawing groups (e.g., carboxylic group in p-bromobenzoic acid) are dehalogenated after rapid hydrolysis. (45) In contrast, stable radical cations substituted with electron-donating groups (e.g., hydroxyl group in p-bromophenol) are not prone to hydrolysis, thus slowing down the dehalogenation kinetics. (45)

Sulfate Radical and Persulfate: Matrix Effects

Jump To

Interaction with Dissolved Organic Matter

A significant decrease in the treatment efficiency due to reactions of OH by background organic constituents such as natural organic matter (NOM) and effluent organic matter (EfOM) is a well-recognized drawback of OH-AOPs. (57−59) The SO4•–-AOP is not an exception (Figure 3a), but the efficiency loss occurs generally to a much smaller extent compared to OH-AOPs. (21,22) Several studies (21,60) suggested that the gross second order rate constants for the reactions between SO4•– and NOM ranges from 2.5 × 107 to 8.1 × 107 MC–1 s–1, which is around 1 order of magnitude lower than for OH (k = 1.6 – 3.3 × 108 MC–1 s–1). (61) The kinetic inhibition by the organic matrix components depends on the extent of oxidation of the fast reacting moieties by the substrate-specific SO4•–. In contrast, the inhibitory effects persist for the less selective OH until the NOM or EfOM is almost completely mineralized. (62) It is noteworthy that the SO4•–-AOP can also be affected by more significant inhibition in some cases. For instance, degradation of organics that are less reactive with SO4•– (e.g., ibuprofen or perfluorooctanoic acid (PFOA)) were found to be significantly inhibited when humic-like substances (containing aromatic and olefinic moieties that more readily scavenge SO4•– than aliphatic components (63)) are present. (64)

Figure 3

Figure 3. (a) Interaction of natural organic matter with peroxymonosulfate (PMS) and peroxydisulfate (PDS), (b) chemical transformation of halide ions by PMS and sulfate radical, and (c) generation of reactive intermediates through reaction of PMS with nucleophiles.

In addition to being a scavenger of SO4•–, some functional groups present in NOM can serve as activators for PMS and PDS (Figure 3a). For example, quinone-type compounds (e.g., p-benzoquinone) accelerates the self-decay of PMS, yielding 1O2 as reactive transient intermediate. (11,65) The nonradical PMS activation occurs predominantly in the basic pH region (11,65) since PMS self-decomposes when the pH exceeds the pKa value of PMS (∼9.3). (66) In contrast, PDS can undergo reductive transformation into SO4•– by a semiquinone radical that forms via the comproportionation between benzoquinone and hydroquinone. (67) Similarly, the phenolate anion, a dominant species at pH > pKa (∼10), activates PDS to SO4•–, whereas the neutral phenol barely activates PDS. (68) Some quinone and phenol derivatives with relatively low pKa values (e.g., pentachlorophenol; pKa = 4.70 (69)) can activate PDS even under mildly acidic and neutral conditions. (68) The overall oxidizing capacity of reactive intermediates such as SO4•– and 1O2 resulting from NOM-induced persulfate activation is marginal due to the capacity of NOM as a natural sink for reactive oxidizing species.

Interaction with Halide Ions

Radical scavenging by Cl and consequential loss of the process efficiency are a unique challenge encountered in SO4•–-AOP (Figure 3b). In conventional AOP, OH adds to Cl to form ClOH•–, but it mostly reverts to OH under neutral conditions (Cl + OH ⇆ ClOH•–) rather than forming Cl (ClOH•– + H+ ⇆ Cl + H2O). (70) In contrast, SO4•– produces Cl through one-electron abstraction from Cl (SO4•– + Cl → Cl + SO42–). (71,72) As a result, even though Cl reacts more rapidly with OH than SO4•– (k = 4.3 × 109 M–1 s–1 for OH (73) and k = 3.1 × 108 M–1 s–1 for SO4•– (74)), Cl causes a more severe retarding effect on pollutant degradation kinetics, greater production of reactive chlorine species, or a switch of the main oxidant (from SO4•– to OH) in SO4•–-AOP than OH-AOPs. (75−78)
A complication arises when halide ions are present at high concentrations (e.g., brackish groundwater, saline wastewater, and reverse osmosis concentrate). A suite of halide radicals, including X, X2•–, XY•– (mixed-halogen radical such as BrCl•–), and OX•–, form at considerable concentrations. (79) X reacts with H2O/OH at low halide levels to yield HOX•– as an intermediate, which readily transforms into OH under nonacidic conditions. (78,79) Further reactions involving X and X2•– lead to the formation of X2 (X + X2•– → X2 + X; X2•– + X2•– → X2 + 2X) and HOX (X2 + H2O → HOX + X + H+). (79) These reactive halogen species are more selective than SO4•– and OH despite relatively high standard reduction potentials (E0(Cl/Cl) = 2.5 VNHE; E0(Cl2•–/Cl) = 2.2 VNHE; E0(Br/Br) = 2.0 VNHE; E0(Br2•–/Br) = 1.7 VNHE). (79) Consequently, the treatment efficiency tends to decrease, especially for the target compounds that are less reactive to reactive halogen species (e.g., benzoic acid). (76) For the same reason, it is possible that the SO4•–-AOP becomes more efficient when target pollutants are more susceptible to oxidation by reactive chlorine species compared to SO4•– or OH. (80) This has caused some erroneous claims of the positive effects of Cl on persulfate-AOP, (76,81) since HOCl, which has a much longer lifetime, can become the main oxidant in high Cl conditions, contrary to the intention of utilizing SO4•–. Two-electron oxidation of halide ions by PMS (k(Cl) = 2.1 × 10–3 M–1 s–1, k(Br) = 7.0 × 10–1 M–1 s–1, and k(I) = 1.4 × 103 M–1 s–1) (82) leads to direct HOX formation involving no halide-containing radicals as intermediates (Figures 3b and 3c), which creates binary mixtures of PMS and halide (e.g., PMS/Cl, PMS/Br, PMS/I) that can oxidize selected electron-rich organics. (13,82−86) Note that mixtures of PMS and salts of Cl and Br are used for the synthesis of chlorinated and brominated olefins. (87)
A major challenge in treating water containing halide ions and background NOM/EfOM (or target organics) by the persulfate-AOP comes from the formation of toxic halogenated byproducts, such as trihalomethanes (THMs) and haloacetic acids (HAAs) (or halogenated intermediates), through the halogenation of NOM by HOX. (13,85,88−91) Since the reactivity of PMS toward halide increases in the order of Cl < Br < I, (82) brominated and iodinated byproducts can be formed in addition to chlorinated byproducts. They are typically more toxic than the chlorinated analogues by a factor of >10 and >100, respectively. (92,93) Whereas SO4•– dehalogenates halogenated organic byproducts to release halide ions, (88,91,94) some organics were found to become more susceptible to halogenation due to reaction with SO4•– (e.g., conversion of carboxylic substituents on the aromatic rings to hydroxyl groups). (47)
Unlike organic halogenation, the formation of toxic halogen-containing oxyanions, such as ClO3 and BrO3, majorly involves SO4•– attack because of the inability of PMS to further oxidize HOX as a precursor to oxyanions (Figure 3b). (33,90,95) Nevertheless, I can be converted by PMS to IO3 as a desired end product because of the easier oxidizability of HOI (Figure 3b). (13,82) Br is of more problematic than Cl since (i) Br is more reactive toward SO4•– and PMS than Cl; (37,82) (ii) HOBr•– that forms from the reaction of Br and OH, unlike HOCl•–, does not readily decay back to OH and Br (k = 3.3 × 107 s–1 for HOBr•– decay; (96)k = 6.1 × 109 s–1 for HOCl•– decay (79)); and (iii) bromination of phenols proceeds with second-order rate constants that are about 3 orders of magnitude higher than for chlorination. (97,98) In addition, further oxidation of HOBr/OBr by SO4•– can lead to the formation of BrO3. Note that BrO3 formation is inhibited in the presence of dissolved organic matter mainly because of the formation of superoxide radical. (99) It is also suppressed in traditional UV-based AOPs that employ H2O2, due to a reduction of HOBr to Br by H2O2. (100)

Interaction with OH

Raising the pH above ∼8.5–9 can cause a transition from SO4•–-dominated to OH-dominated oxidation process. (35,101) This results from an one-electron oxidation of OH by SO4•– (k = 6.5 × 107 M–1 s–1) (102) that is kinetically preferred over the reverse reaction (OH + HSO4 → SO4•– + H2O; k = 6.9 × 105 M–1 s–1). (41) A change of the main oxidant from SO4•– to OH leads to a more effective abatement of organic compounds that persist in persulfate activation, (101) but the enhancing effect can be offset by unwanted competitive reactions involving the less selective OH. OH is more significantly consumed than SO4•– by natural water matrix components (e.g., NOM and CO32–) and PMS/PDS. (41,77) Regardless, increasing the pH has been employed as a simple approach for persulfate activation. The major oxidant in the activation process varies depending on whether PMS or PDS is used. (103,104) Alkaline conditions initiate nucleophilic attack of PMS (SO52–) to the peroxide oxygen of PMS (HSO5), which results in the self-decay of PMS and the associated 1O2 production (Figure 3c). (104,105) In contrast, base activation of PDS occurs in two sequential reactions: (i) HO2 formation through base-catalyzed PDS hydrolysis and (ii) reduction of PDS by HO2 to SO4•– (103) (which further converts to OH in highly alkaline conditions, e.g., at pH 12). (101,103)

Interaction with Oxyanions

Anions such as phosphate (i.e., HPO4/H2PO42–) and bicarbonate/carbonate (HCO3/CO32–) can scavenge SO4•– (k = ∼ 106–107 M–1 s–1) (77,106) and decrease the overall efficiency of the SO4•–-AOPs. (60,64,80,107) The radical scavenging becomes more noticeable when pH exceeds the pKa values of the corresponding acid of an anion. For instance, HPO42–, the dominant species at pH > pKa2 = 7.2, exhibits 2 orders of magnitude greater reactivity toward SO4•– than H2PO4: k(HPO42–) = 1.2 × 106 M–1 s–1 and k(H2PO4) < 7 × 104 M–1 s–1. (77) In case of HCO3/CO32–, the second-order rate constants for the reaction with SO4•– are of the same order of magnitude but also higher for the deprotonated species: k(HCO3) = 1.6 × 106 M–1 s–1 and k(CO32–) = 6.1 × 106 M–1 s–1. (108) However, a greater scavenging at a pH above the pKa2 = 10.3 is also observed, since the main oxidant changes from SO4•– to OH, which is more readily quenched by CO32–, with k(HCO3) = 8.5 × 106 M–1 s–1 and k(CO32–) = 3.9 × 108 M–1 s–1, (41) leading to the formation of CO3•–. NO2 also rapidly reacts with SO4•– (k = 8.8 × 108 M–1 s–1), (77) thus significantly decelerating organic oxidation by activated persulfate, (107) while NO3 does not react with SO4•–. (80)
Formation of anion-derived radicals from the reaction between anions and SO4•– brings another complication to the system. (106,109−113) In addition to reducing the overall oxidation kinetics, these weaker oxidants (e.g., E0(CO3•–/CO32–) = +1.63 V) (108) are very selective and preferentially abate specific classes of electron-rich organics and lead to products that are not typically expected with SO4•–. In natural waters and wastewater effluents, CO3•– often becomes the dominant oxidant due to HCO3/CO32– oxidation not only by SO4•– but also by aforementioned halogen radicals. (106) For example, Cl and Cl2•– formed through one-electron abstraction from Cl by SO4•– have relatively high sec