Evaluation of advanced oxidation processes for β-blockers degradation: a review

β-Blockers belong to the class of pharmaceuticals that are widely used for treatment of cardiovascular diseases such as hypertension and cardiac arrhythmias. They are β-andrenergic antagonists inhibiting the effect of adrenaline and noradrenaline levels in the body, primarily in the heart. β-Blockers possess a chiral carbon atom in the propanol unit, therefore there are two enantiomeric forms for each pharmaceutical. The most frequently used medicines of the β-blockers are atenolol (ATE), metoprolol (MET) and propranolol (PRO). Their pK_a values are between 9.42 and 9.70 and they have logK_ow values between 0 and 4. Some studies have demonstrated that the long-term hazard of each pharmaceutical product to humans is irrelevant below the μg-ng dm⁻³ concentration level. However, the ecotoxicological context of these pharmaceuticals is still unclarified (Jones et al. 2005). In the case of β-blockers the acute toxicity data have not been investigated in detail yet, with the exception of PRO. Acute toxicity test was done only on aquatic vertebrates and invertebrates (Huggett et al. 2002; Hernando et al. 2004; Cleuvers 2005). PRO has the highest acute toxicity that can be explained by the high K_ow value and strong membrane stabilizer character (Fent et al. 2006).

Their excretion mode from the human body and their half-lives being related to the degree of lipophilicity are key parameters from an environmental point of view: the more are lipophilic these compounds, the smaller are their half-lives in the human body. The most lipophilic and hydrophilic compounds are PRO and ATE, respectively. Metabolism of each drug takes place to various extents in human applications (Alder et al. 2010). In some cases, the parent compound is excreted unchanged via the urine. Additionally, the conventional wastewater treatment techniques are not effective enough for complete elimination of β-blockers. In this way these drugs and their metabolites may also occur in influent and effluent streams of wastewater treatment plants in the μg dm⁻³ concentration range in some countries (Maurer et al. 2007; Alder et al. 2010; Nödler et al. 2013; Duan et al. 2018). It is thus not surprising, that β-blockers and their metabolites have been detected also in surface waters: for example, ATE and MET were found at up to ∼ 8,000 ng dm⁻³ in the Glatt Valley Watershed (Switzerland) or in the Llobregat River (Spain) (Alder et al. 2010; López-Roldán et al. 2010).

ATE, MET and PRO are structurally similar compounds; their common feature is the oxypropanolamine side chain that contains an asymmetric carbon atom, an amino and an OH group (Figure 1). In ATE and MET this side chain is attached to a benzene ring, whereas PRO contains a naphthalene ring instead of the benzene ring. Their physicochemical properties are summarized in Table 1 (Hernando et al. 2007; Liu & Williams 2007; Kim & Tanaka 2009; Quinten et al. 2012; Liu et al. 2013).

Based on their chemical structure, and UV and Raman spectra, ATE and MET are highly similar compounds (Farcas˛ et al. 2016). By contrast, PRO differs slightly from ATE and MET due to its naphthalene unit. Basically, β-blocker molecules have two vulnerable reaction sites in radical reactions: the aromatic ring and the amino moiety. Difference in their reactivity is due predominantly to the structure of their aromatic part: a condensed ring possesses more attackable sites and higher electron density than a simple aromatic ring. Based on the cyclic voltammetry measurements the reduction potential of ATE is estimated to be around 1.15 eV vs. NHE (Patil et al. 2009; Hegde et al. 2011). The value for PRO, which has a naphthalene unit with extended conjugation, is expected to be smaller.

The presence of β-blockers in natural waters is an essential environmental problem and development of effective water purification treatments is required. Advanced oxidation processes (AOP) are chemical treatments based on generation of several reactive oxidizing radicals and reactants (Table 2) that are effective for the elimination of these pollutants (Glaze et al. 1987).

The basic reactions between radicals and organic molecules are suggested to be one or two of the following three possibilities: electron transfer, radical addition to a double bond and H-atom abstraction from the saturated parts of molecules. Direct oxidation is a feasible reaction in which the one-electron reduction potential of the attacking radical is higher (more positive) than the potential of the organic molecule. The forming radicals, depending on their reduction potentials (Table 3), may also have an important role in oxidative reactions. Due to their high reduction potential values SO₄^•−, Cl₂^•−, CO₃^•− and ClO^• may react with the β-blocker molecules in electron transfer reactions. ^•OH mainly reacts in radical addition to the double bonds or in H-atom abstraction reaction.

The reaction mechanism of β-blocker degradation depends on the forming reactive species in the applied AOP, the substrates and the operating parameters. Predominantly, ^•OH is responsible for the degradation reactions excluding direct photolysis. In SROP, SO₄^•− also acts as a strong oxidizing reactant. In addition ^•OH and SO₄^•−, O₃, h⁺, O₂^•−, ¹O₂, O₂^•−/HO₂^• and chlorine-containing species may have a contribution to the degradation.

During ^•OH attack, the first step can be hydroxylation on the aromatic ring or the secondary amine (Romero et al. 2011; Santiago-Morales et al. 2013; Gao et al. 2018a, 2018b). Mono- (m/z values of 283, 284 and 276), di- (m/z values of 299, 300 and 292), tri- (m/z values of 315, 316 and 308) and tetra-hydroxylated products (m/z values of 331 and 332) may form in ATE, MET and PRO degradation, respectively. More isomers were observed in the case of PRO with the naphthalene ring than in the cases of ATE and MET with the simple benzene ring. The most vulnerable sites of PRO have been determined (Xie et al. 2019). The largest negative charge is on the carbon atom in the ortho-position, thus hydroxylation is the most favoured reaction in this position (Figure 2). Based on partial charge calculations, ortho- and para-additions are the most likely reactions leading to the major photoproducts (Piram et al. 2012). Further oxidation of hydroxylated isomers can lead to the ring-opening reactions forming aldehyde and ketone derivatives (Ji et al. 2012). Hydroxylation of the amine moiety results in a hydroxylamine product.

In SROP, based on erythrocyte sedimentation rate measurements, it was suggested that SO₄^•− has a higher contribution to the oxidation reactions than ^•OH in the degradation of PRO (Gao et al. 2018a, 2018b; Minhui et al. 2019; Xie et al. 2019; Yang et al. 2019; Chen et al. 2020). SO₄^•− favours the attack on the aromatic ring and the amino moiety via electron transfer (Gao et al. 2018a, 2018b). Because of its electrophile character, hydroxylation on the aromatic ring was preferred over hydroxylation on the amino moiety. Following hydroxylation on the aromatic ring, the oxidized naphthalene moiety remains reactive to further SO₄^•− attack (Yang et al. 2019).

In the presence of chlorine-containing radicals, chlorinated degradation products (DPs) may form during degradation (Gao et al. 2020; Pan et al. 2020; Xiong et al. 2020). Chlorination takes place on the benzene/naphthalene ring or on the amino moiety.

During ozonolysis, the main reactive agents are O₃ and ^•OH. β-Blockers have two sites to be attacked by molecular ozone: the benzene/naphthalene and the secondary amino moieties. O₃ can react with β-blockers in a direct way via 1,3-cycloaddition following the Crigée mechanism. Below pH 4, the reactions of O₃ are preferred as the generation of ^•OH is limited. In this media, the amino moiety is in a protonated form that is less reactive. In the first step, DP 301 and DP 300 may form from ATE and MET, respectively (Wilde et al. 2014).

Figure 3 shows a general mechanism of ^•OH-induced degradation. In the literature, three degradation pathways are proposed for the degradation of β-blockers: (i) ^•OH attack on an ipso carbon atom resulting in 3-(isopropylamino)propane-1,2-diol (m/z 134); (ii) electrophile ^•OH addition on the aromatic ring or on the amino moiety, and (iii) loss of the side chain depending on the compound. After hydroxylation, ring opening, ether cleavage and H abstraction may take place. The amino moiety is less reactive towards ^•OH than the aromatic ring (Romero et al. 2011). In the cases of ATE and MET the elimination of the isopropyl moiety via N-dealkylation may take place forming DP 194 and 226, respectively (not shown on Figure 3) (Wilde et al. 2014).

Mutual diffusion of the reactants to each-others vicinities is a precondition for the reaction, thereby the rate of diffusion limits the rate constant values. Table 3 contains the suggested diffusion controlled rate constants (k_diff) together with the reduction potentials and the measured rate constant values. The rate constants with the same radical measured for the studied β-blockers in different laboratories by using different techniques agree with each other reasonably.

The rate constant values with ^•OH (k_•OH) were determined for the three molecules by competition technique using p-hydroxybenzoic acid as reference compound with k_•OH = 2.19 × 10⁹ M⁻¹ s⁻¹ (determined also in competitive experiments) (Sirés et al. 2010). These k_•OH values are just one-third of the values measured by other authors. This can be explained that this applied reference value is unrealistically small, based on pulse radiolysis experiments k_•OH = 7.0 × 10⁹ M⁻¹ s⁻¹ is suggested for the p-hydroxybenzoic acid + ^•OH reaction (Wojnárovits et al. 2018). In Table 3 the recalculated values are listed.

As we discussed previously one of the basic ^•OH reactions is radical addition to the aromatic ring. This conclusion is based on the transient absorption spectra observed in pulse radiolysis experiments, which have a wide absorption band with maximum around 330 nm, characteristic to the ^•OH adduct cyclohexadienyl radicals (Song et al. 2008).

Based on the published k_•OH values for ATE, MET and PRO we calculated as averages 7.19 ± 0.89 × 10⁹, 8.0 ± 0.37 × 10⁹ and 1.1 ± 0.18 × 10¹⁰ M⁻¹ s⁻¹, respectively. They are close to the diffusion controlled limit of 1.1 × 10¹⁰ M⁻¹ s⁻¹ (Wojnárovits & Takács 2017). The rate constant of ATE is consistent with the ^•OH reaction being predominantly at the ring moiety, the electron withdrawing NH₂COCH₂-side chain slightly decreases the rate constant as compared to that of benzene (7.8 × 10⁹ M⁻¹ s⁻¹). The slightly higher rate constant for MET than for ATE is consistent with the weaker electron-withdrawing CH₃OCH₂CH₂-side chain group. The high value for PRO reflects the fact that the naphthalene group has both higher electron density and more potential sites for hydroxyl radical reaction (Song et al. 2008).

The rate constant of the sulfate radical anion reaction (k_SO4•−) determined by Liu et al. (2013) for ATE, 1.3 × 10¹⁰ M⁻¹ s⁻¹, is much higher than the value of Lian et al. (2017), 5.11 ± 0.26 × 10⁹ M⁻¹ s⁻¹. The same is true for the values determined for MET 5.11 ± 0.12 × 10⁹ and 1.0 ± 0.1 × 10¹⁰ M⁻¹ s⁻¹, respectively (Tay & Ismail 2016; Lian et al. 2017). The value for PRO is 4.79 ± 0.26 × 10⁹ M⁻¹ s⁻¹ (Lian et al. 2017). All other k_SO4•− values in the Table 3 for this molecule are above 1 × 10¹⁰ M⁻¹ s⁻¹. k_SO4•− values above 1 × 10¹⁰ M⁻¹ s⁻¹ are unrealistic, they are much higher than the rate constant limit dictated by the diffusion (k_diff = 6.5 × 10⁹ M⁻¹ s⁻¹) (Wojnárovits & Takács 2016). It seems that the applied methods were not adequate for rate constant determination. We suggest to accept the k_SO4•− values of Lian et al. (2017) for the three compounds mentioned.

The reactivity of the chlorine atom (Cl^•) is similar to that of ^•OH and the basic mechanism of this reaction is radical addition to the double bonds. It reacts with ATE and MET with a rate constant above 1 × 10¹⁰ M⁻¹ s⁻¹ (Lei et al. 2019; Mangalgiri et al. 2019).

Cl₂^•− is suggested to react with many aromatic molecules in direct oxidation with rate constants in the 10⁸–10⁹ M⁻¹ s⁻¹ range (Lei et al. 2019). As a result of the reaction, two chloride ions (Cl⁻) are left behind plus the radical cation of the organic molecule (Hasegawa & Neta 1978). The k_Cl2•− values of ATE, MET and PRO reactions are between 4 × 10⁸ and 2 × 10⁹ M⁻¹ s⁻¹ (Jasper et al. 2016; Lei et al. 2019; Pan et al. 2020).

The chlorine monoxide radical (ClO^•) is also considered to be a one-electron oxidant, the reduction potential is suggested to be 1.39 V (Armstrong et al. 2015). Similar rate constant values for ATE and MET reaction have been published, 8.68 × 10⁷ and 1.34 × 10⁸ M⁻¹ s⁻¹, respectively (Gao et al. 2018b).

The dibromide radical anion (Br₂^•−) is a milder one-electron oxidant than Cl₂^•− (Armstrong et al. 2015). It reacts with ATE and MET with two orders of magnitude smaller rate constants (1.3 ± 0.2 × 10⁶ and 6.1 ± 0.2 × 10⁶ M⁻¹ s⁻¹, respectively) as Cl₂^•− (Szajerski et al. 2006).

The k_CO3•− values of reactions with the protonated forms (pH 6–8) of ATE, MET and PRO are in the 10⁶ M⁻¹ s⁻¹ range (Jasper & Sedlak 2013; Wols et al. 2014; Lian et al. 2017). CO₃^•− reacts with higher rate constant (5.9 ± 1.6 × 10⁷ M⁻¹ s⁻¹ for ATE and 4.6 ± 0.7 × 10⁸ M⁻¹ s⁻¹ for PRO) with the deprotonated molecules at pH 11.4 as with the protonated forms at lower pH (Jasper & Sedlak 2013). CO₃^•− is suggested to react with aromatic molecules in one-electron oxidation (Wojnárovits et al. 2020). Using the analogy of ^•OH reactions, Augusto et al. (2002) proposed CO₃^•− addition to the benzene ring as the rate determining step in the reaction of aromatic molecules. However, the addition complexes are suggested to be unstable and decay too fast to be observed (Moore et al. 1977; Augusto & Miyamoto 2011). The rate constants for H-abstraction reactions are generally low (alcohols, primary amines, thiols), e.g., for ethanol k_CO3•− = 2.0 × 10⁴ M⁻¹ s⁻¹ (Clifton & Huie 1993).

Chemical reactions of the two-electron oxidant ¹O₂ (singlet excited dioxygen) generally occur with rate constant in the 10⁷ M⁻¹ s⁻¹ range. Chen et al. (2009) published a value of (9.3 ± 0.4) × 10⁶ range for ¹O₂ reaction with PRO.

Ozone (O₃) reacts with the higher rate constant with molecules containing double bonds by addition to these bonds. The k_O3 values for ATE and MET are ∼2 × 10³ M⁻¹ s⁻¹; a higher value is suggested for PRO ∼1 × 10⁵ M⁻¹ s⁻¹ (Benner et al. 2008).

Considering only the realistic values the rate constants of the one-electron oxidation reactions are increasing in the CO₃^•− < ClO^• < Cl₂^•− < SO₄^•− order in agreement with the increasing one-electron reduction potential of the oxidant.

The Z constant is usually taken as 1 × 10¹¹ M⁻¹ s⁻¹ (Jonsson et al. 1994; Meisel 1975; Armstrong et al. 1996). In Equation (1) R is the gas constant; T is the absolute temperature ln Equation (2) k_diff and k_act stand for the diffusion and activation controlled rate constant. Transfer of an electron is accompanied by the reorganization of solvent shell around the donor and acceptor molecules. λ⁰ is the energy needed for this reorganization.

Figure 4 shows the log k vs. ΔE plot. The reduction potentials of ATE, MET and PRO were taken as 1.15, 1.1 and 1.05 V, respectively. The solid line was calculated according to Equation (1) using λ⁰ = 180 kJ mol⁻¹. This is a rather high reorganization energy, but as we mention, in electron transfer between (O₂/O₂^•−) and (PhO^•/PhO⁻) Jonsson et al. (1994) found also a high value, λ⁰ = 155 kJ mol⁻¹. The fact that the measured rate constants show Marcus-type behaviour also reflects that the main reaction in these processes is the electron transfer.

Photolysis is an effective technique in water treatment in which artificial UV light or solar light is applied as the light source. The effectiveness of photolysis is the function of the UV absorbance spectra of the treated molecules, the quantum yield and the lamp geometry. The photochemical properties of starting molecules determine the results. Table 1 summarizes the physical properties of the three β-blockers. All three compounds possess a low molar absorption coefficient (ɛ_{254 nm} < 1,000 M⁻¹ cm⁻¹) at 254 nm (low-pressure mercury lamp). Direct photolysis of the three β-blockers was investigated using a UV light source and sunlight (simulated or artificial light source). The photodegradation obeyed pseudo-first order kinetics in all three cases.

In the direct photolysis experiments of Liu & Williams (2007) the degradation of ATE and MET was slower than that of PRO due to the presence of the naphthalene unit in the latter. During 46 h exposure, the forming DPs of PRO reflected oxidation, deoxygenation and rearrangement. In the experiments of Piram et al. (2012) different hydroxylated isomers were detected in the presence of dissolved O₂. They showed that during direct photolysis ^•OH also forms and studied the position of ^•OH attack by calculations based on the Mulliken and Hirshfeld models. Based on the charge distributions, the probabilities of attacks on each atom were predicted. Ortho- and para-additions were the favoured ^•OH reactions. There are more positions for ^•OH attack in the case of PRO due to the naphthalene ring. Mono-, di- and tri-hydroxylated products were identified.

The effect of the photoreactor material on PRO removal was investigated (De la Cruz et al. 2013). The photoreactors were made of either photochemical quartz or Duran glass (borosilicate glass). They showed different photochemical properties; the light transmittance of quartz and Duran glass starts at 200 and 300 nm, respectively. Quartz was more efficient in the degradation of PRO.

In natural waters and wastewaters there are a variety of inorganic ions and dissolved organic matter (DOM) present. Among them, nitrate and DOM (mainly fulvic acid derivatives) may serve as photosensitizers in indirect photolysis. Indirect photolysis of the three compounds was investigated in the presence of fulvic acid (Chen et al. 2012; Makunina et al. 2015). During PRO degradation, laser flash photolysis (355 nm) and steady-state (365 nm) photolysis were used (Makunina et al. 2015). They suggested that mainly triplet state fulvic acid contributed to the photodegradation. Degradation can take place via electron transfer. Dissolved oxygen and the forming singlet dioxygen (¹O₂) do not contribute to the photosensitized degradation of β-blockers. Indirect photolysis of ATE was also investigated in nitrate (NO₃⁻)-containing solution under simulated sunlight (Ji et al. 2012). In photochemical reactions of NO₃⁻, ^•OH forms and plays a key role in the degradation reactions. The NO₃⁻ concentration and the pH strongly influence the rate of photooxidation: increasing NO₃⁻ concentration increased the photodegradation of ATE. Moreover, the NO₃⁻-induced photogeneration of ^•OH is a pH-dependent process; the acid–base equilibrium reaction between HOONO and OONO⁻ (pK_a = 6.5 ± 0.1) decreased the formation rate of ^•OH (Løgager & Sehested 1993; Vione et al. 2009).

In the UV/H₂O₂ method ^•OH reactive species induce the degradation (Rivas et al. 2010; Yang et al. 2019). The UV/H₂O₂ treatment was more efficient than direct photolysis in the degradation of MET (Rivas et al. 2010). Removal efficiency of MET increased with the rising H₂O₂ concentration. However, because H₂O₂ reacts with ^•OH, transforming it to HO₂^•, at high H₂O₂ concentration the degradation rate decreased. The degree of mineralization was low (∼10%).

It is relatively easy to eliminate β-blockers by O₃ treatment (Benner et al. 2008; Quispe et al. 2011; Wilde et al. 2014). Ozone may react with the target molecules in direct (O₃) and indirect (^•OH) reactions. In direct reactions, O₃ is a highly selective oxidant (E⁰ = 2.07 V) and can act as a nucleophilic or electrophilic reagent in the degradation. The reaction between O₃ and the aromatic ring is a pH-independent process (Benner et al. 2008), the reactivity of the amino moiety is affected by both the pK_a values of the starting molecules and the solution pH in the reactions of both O₃ and ^•OH. The ^•OH formation rate is increasing with the increasing pH.

The effectiveness of O₃ treatment was studied at different pHs (3–11) in aqueous solutions of the three β-blockers (Wilde et al. 2014). At acidic pH, O₃ treatment was less efficient due to the formation of less ^•OH. In the 7–11 pH range, all three compounds were eliminated in 10 min; there were no differences in the degradation rates. PRO showed higher reactivity towards both O₃ and ^•OH which is explained by the presence of the electron rich naphthalene ring. This treatment was very effective in brine and in wastewater treatment plants (WWTP) effluents using higher O₃ (5–10 mg dm⁻³) concentrations (Benner et al. 2008).

Degradation mechanisms of β-blockers (ATE, MET and PRO) were examined at pHs 5, 7 and 9 in details (Wilde et al. 2014). Different reaction mechanisms were suggested at the applied pHs with molecular ozone and ^•OH. Degradation of ATE and MET came to pass in a similar way by reason of their similar structure. The representative reaction of O₃ with 1,3-dipolar cycloaddition via Crigée mechanism could be observed in the course of ATE and MET decomposition. The main products of ATE degradation were aromatic ring hydroxylated molecules, products reflecting ring opening and cleavage of acetamide moiety. By contrast, in MET degradation ^•OH reacted on the secondary amine moiety and N-dealkylation was suggested. Products showing the cleavage of the ether side chain are represented among the final products. In the case of PRO, there was a more complex picture about the degradation processes under reductive and oxidative reactions. Hydroxylated and decarboxylated products were detected among the DPs.

Photocatalysis is a low-cost and eco-friendly technique for removal of different organic pollutants. In comparison with other AOP, photocatalysis is suggested to be one of the most effective methods for removal of β-blockers (Rivas et al. 2010; Yang et al. 2010; Abramović et al. 2011; Ioannou et al. 2011; Romero et al. 2011; Rey et al. 2012; Czech & Rubinowska 2013; De la Cruz et al. 2013; Golubović et al. 2013; Ji et al. 2013; Santiago-Morales et al. 2013; Píšt'ková et al. 2015; Pinedo et al. 2016; Leyva et al. 2019; Ponkshe & Thakur 2019).

TiO₂ is the most widely used semiconductor catalyst. The degradation efficiency is affected by the specifications of the applied photocatalyst, such as crystal composition, surface area, particle size distribution, band gap, etc. The most often-applied photocatalysts used in studies on β-blockers’ decomposition are summarized in Table 4: they differ in the crystal form, BET surface area, particle size and the manufacturer. TiO₂ has three crystal modifications: anatase (A), rutile (R) and brookite (B). Generally, anatase and rutile are used as the photocatalyst. Anatase is more active and has a larger surface area and higher energy gap value than rutile (E_B(anatase) = 3.2 eV; E_B(rutile) = 3.0 eV) (Chong et al. 2010; Ji et al. 2013). In spite of this, the commercial P25 TiO₂ contains anatase and rutile phases in a ratio of about 3:1 (depending on the producer) (Ioannou et al. 2011; Ji et al. 2013; Píšt'ková et al. 2015; Ponkshe & Thakur 2019). In the catalyst, smaller sized rutile particles are spread in the anatase phase and electron transfer takes place between the two crystal forms, resulting in retarded electron/hole recombination (Haque et al. 2006; Ji et al. 2013).

Surprisingly, the commercial TiO₂ Wackherr and synthesized mesoporous anatase nanopowders were more effective in MET degradation than Degussa P25 (Abramović et al. 2011; Golubović et al. 2013). Compared to Degussa P25, TiO₂ Wackherr has much larger particles (300 nm) resulting in lower surface area. In spite of this, the holes and the back reactions have higher importance in the course of reactions on Degussa P25. In the case of mesoporous anatase nanopowders, the greater mean pore diameter led to a degradation efficiency. Although, slower decomposition was observed in the case of Degussa P25, a higher degree of mineralization was achieved (total organic carbon (TOC) removal 96 and 83% for Degussa P25 and TiO₂ Wackherr, respectively). Furthermore, it is interesting that the degradation reactions take place through different pathways on the applied photocatalysts.

The initial pH strongly influences the rate of photocatalytic degradation (Ji et al. 2013; Yang et al. 2010; Ponkshe & Thakur 2019), it determines the ionization state of both photocatalyst and the substrate. The ionization state can be characterised by the pH of the point of zero charge (PZC) for the photocatalyst and by the pK_a value for the organic substrate. The PZC value of TiO₂ is at pH ∼6.7. The effect of pH on the photocatalytic degradation rate of the three β-blockers was investigated (Yang et al. 2010). The rates became higher with the increasing pH, the changes were similar in the cases of ATE and MET due to their similar structures. At acidic pH (pH < 6.7), both the catalyst and the β-blockers are in the protonated form, thus electrostatic repulsion takes place between catalyst and substrate. In the 6.7 < pH < pK_a range, electrostatic attraction evolves between the negatively charged TiO₂ and positively charged β-blockers enhancing the rate. The interpretation of reactions at highly basic media is more complex. At high pH, on the one hand, the catalyst's surface is negatively charged and the substrate is neutral, so there is no attraction, on the other hand, the generation of ^•OH is more favoured.

The effect of TiO₂ loading was studied in homogeneous (Píšt'ková et al. 2015) and heterogeneous (Abramović et al. 2011; Ioannou et al. 2011; Romero et al. 2011; De la Cruz et al. 2013) systems during the degradation of the three compounds. In heterogeneous photocatalysis the reaction rate until a certain extent is linearly proportional to the amount of TiO₂. Above the so-called saturation level, this value becomes independent on TiO₂ loading. At very high catalyst concentrations, light scattering may take place, the TiO₂ particles may aggregate in the suspension leading to attenuation of the contact surface area and the number of active sites. This phenomenon was observed in degradation of the three β-blockers. The aim of these experiments was to establish the optimum TiO₂ loading in favour of the best accomplishment. Using an immobilized photocatalyst, the light scattering effect and aggregation of particles can be avoided. The effect of immobilized TiO₂ loading on glass slide for PRO removal was studied (Píšt'ková et al. 2015). Immobilization resulted in shorter illumination time and more efficient elimination.

The variation of the degradation rate was investigated also in for function of initial substrate concentration in the decomposition of the three compounds using P25 photocatalyst (Yang et al. 2010; Abramović et al. 2011; Ioannou et al. 2011; Ponkshe & Thakur 2019). ^•OH with its short lifetime can react with organic substrates close to the place of generation. Higher substrate concentration resulted in greater probability of collisions between ^•OH and the substrate. However, more substrate molecules can occupy more active sites on the photocatalyst surface suppressing the formation of ^•OH. Consequently, after a certain concentration the degradation rate decreased with increasing substrate concentration for all three compounds. However, this expected tendency did not depend only on the initial substrate concentration, but depended also on TiO₂ loading.

Degradation of PRO and ATE was studied in the concentration range between 5 and 20 mg dm⁻³ (∼0.018 and 0.08 mM) at 0.25 g dm⁻³ TiO₂ loading (Ioannou et al. 2011). After 120 min treatment, they observed a decrease in the rate of conversion for both PRO and ATE. The effect of initial substrate concentration in the 0.01–0.4 mM range was investigated at 0.3 g dm⁻³ TiO₂ loading (Ponkshe & Thakur 2019). Maximum conversion was accomplished at concentrations of 0.05 mM and 0.1 mM for PRO and ATE, respectively. The effect of the catalyst loading was studied on the degradation rate (Yang et al. 2010; Abramović et al. 2011). Using 1 g dm⁻³ loading with increasing MET concentration the degradation rate increased and maximum rate was at 0.08 mM. By contrast, 2 g dm⁻³ TiO₂ loading caused a decrease in the reaction rate at 0.05–0.2 mmol dm⁻³ concentrations for PRO, ATE and MET (Yang et al. 2010). Presumably, the applied catalyst concentration was too high for an efficient degradation.

During photocatalysis UV light (in the UV-A and UV-C ranges) and sunlight are also capable of generating electron–hole pairs and ^•OH. Degradation of PRO was studied in the presence of both artificial and solar light (De la Cruz et al. 2013). As artificial light, a Xe-OP (1 kW) lamp with a filter (cut-off < 280 nm) was used in a Solarbox using the so-called compound parabolic concentrators (CPCs). The effect of light sources was tested at 0.1, 0.2 and 0.4 g dm⁻³ P25 TiO₂ loadings in PRO degradation. Using solar equipment, the PRO conversion was 64.8%, 65.4% and 80.5% at the end of the 240 min treatment, respectively. In the Solarbox experiments, the percentages of removals were 81.6%, 88.0% and 94.1% for the three catalysts loadings. The authors observed 1.2–1.4 times faster reaction by using the Xe-OP lamp.

The efficiency of UV light emitting at 254 nm and artificial light (Xe-OP lamp) was compared for the degradation of PRO and ATE (Ponkshe & Thakur 2019). They measured the substrate, chemical oxygen demand (COD) and TOC removals for both light sources. The efficiency of applied light sources was about the same. In the case of PRO, the removal percentage was 96 and 94%, the COD removal was 82 and 73%, while TOC declined by 67 and 66% using UV light and simulated solar light, respectively. Similar values were measured in degradation of ATE: the removal percentages were 94 and 95%, COD reductions were 72 and 74%, while 68 and 71% of mineralization happened for the two light sources.

The composition of the water matrix also affects the efficiency of removal. The effect of the surface water matrix has also investigated for the degradation of different β-blockers (Píšt'ková et al. 2015). The surface water samples came from a brook in Slovenia that had pH 7.6 and NPOC 2.26 mg dm⁻³ (non-purgeable organic carbon). The surface water matrix had a detrimental effect on the efficiency of photocatalyst using immobilized P25 TiO₂. It seems that the water constituents affected the stability of the immobilized catalyst; the catalyst surface was damaged during the process.

In order to get a comprehensive picture about the impacts of different water matrix constituents such as bicarbonate ion or humic acid, degradation of ATE was investigated in details in real (Rhône River, France) and synthetic wastewater samples (Ji et al. 2013). Interestingly, HCO₃⁻ enhanced the efficiency of photocatalysis when the ion was present at higher concentration than ATE. As a result of the reaction between HCO₃⁻ and ^•OH, the forming CO₃^•− had a longer lifetime and showed relatively high reactivity towards aniline derivatives (Hu et al. 2007). In a real water sample from the Rhône River (France) ATE degradation was completed in 180 min (Ji et al. 2013), the same value in MilliQ water was 60 min. Anions and DOM may inhibit the active sites on the photocatalyst and may quench the reactive species in this system.

The effect of two humic substances (Aldrich humid acid and Fluka humic acid) was also studied in the photocatalysis of ATE (Ji et al. 2013). Both humic substances showed adverse effects during the degradation. Humic acid may block the active sites on TiO₂, decreasing the light absorption by the target molecules and there is competition between the organic substrates and HA for active species on the surface.

In the UV/chlorine (UV/Cl) combined treatment ^•OH and chlorine-containing radicals such as Cl^•_, ClO^• and Cl₂^•− form in the photolysis of HOCl/OCl⁻. Cl^• and Cl₂^•− possess high reduction potentials (2.43 and 2.1 eV, respectively), ClO^• is a less reactive radical with reduction potential of 1.39 eV (Table 3).

Degradation of PRO, ATE and MET was studied by UV/Cl methods (Gao et al. 2020; Pan et al. 2020; Xiong et al. 2020) applying UV and LED light sources emitting at 254 and 275 nm, respectively. These compounds were removed completely in 20 min under the given conditions at pH 7. The effects of pH, initial substrate concentration, chlorine dosage and the different water constituents were investigated to optimize the operating parameters. The results showed that the degradation rate decreased with increasing initial substrate concentration. The increase of the chlorine dosage resulted in a positive effect; more chlorine-containing reactive species were produced, therefore faster degradation of β-blockers took place.

Variation of the pH showed interesting effects during the degradation of PRO and ATE. Hypochlorous acid is present in two forms (HOCl and OCl⁻) in aqueous solutions; the acidic dissociation constant is pK_a 7.5. HOCl and OCl⁻ have different absorbance spectra. At 254 nm the quantum yields of decompositions are 1.4 and 0.97 for HOCl and OCl⁻, respectively (Feng et al. 2007; Watts & Linden 2007). Moreover, OCl⁻ has much higher molar absorption coefficient than HOCl above 260 nm in aqueous solutions (Wang et al. 2012). The contribution of ^•OH and chlorine-containing species was proven to be different for degradation of ATE, MET and PRO (Pan et al. 2020). The contribution of ^•OH was 35.2%, 26.7% and 19.1% at pH 7, respectively. At lower pH more ^•OH form, higher pH favours the formation of chlorine-containing species. Thus it is not surprising that the rate constants for degradation of PRO increased with the pH (Xiong et al. 2020). In the case of ATE higher pH retarded the efficiency of degradation.

In sulfate radical anion-based oxidation processes (SROP) ^•OH and SO₄^•− coexist simultaneously as major reactive species. The reaction of SO₄^•− with many organic compounds takes place by a one-electron transfer mechanism (O'Neill et al. 1975; Luo et al. 2018) in which SO₄^•− picks up an electron from the ring of aromatic molecules and transforms to SO₄²⁻. SO₄^•− has longer half-life, higher selectivity, mineralization capability and reduction potential (2.43 V) as ^•OH.

SO₄^•− may be generated in several ways: activation of peroxymonosulfate (PMS) and persulfate (PS) by heat, UV irradiation, γ-ray or accelerated electron irradiation, transition metal catalysis and ultrasound treatment (Liang & Su 2009; Ghanbari & Moradi 2017). Another method for the production of SO₄^•− is the application of sulphite ions under transition metal catalysis. Sulphite is less expensive and has lower toxicity compared to PS and PMS (Zhou et al. 2018).

In UV/PS, PMS/synthesized catalyst, ultrasound-assisted heterogeneous activation (nZVI) of persulfate (US/nZVI/PS) and FeS-activated sulfite reactions, predominantly PRO degradation was investigated (Gao et al. 2018a, 2018b; Minhui et al. 2019; Yang et al. 2019; Chen et al. 2020). In these experiments the effects of PS dose, type of catalyst, initial pH and water matrix were studied. More than 90% PRO removal was achieved in 10, 30, 20 and 20 min treatments using UV/PS, PMS/synthesized catalyst, US/nZVI/PS and FeS/sulphite methods under similar conditions, respectively. The degradation was described by pseudo-first-order kinetics. SO₄^•− was more reactive towards PRO than ^•OH (Gao et al. 2018b; Yang et al. 2019; Chen et al. 2020). ^•OH showed similar reactivity with the neutral and anion forms of PRO (Yang et al. 2019). By contrast, the reaction between SO₄^•− and the three β-blockers was pH dependent, the rate constant declined with the increasing pH. Surprisingly, in the PMS/synthesized catalyst system in addition to SO₄^•−and ^•OH ¹O₂ played also an important role in the reactions (Minhui et al. 2019). A quenching reaction with ethanol, tert-butyl alcohol and furfuryl alcohol confirmed the highest contribution of ¹O₂ to the degradation of PRO.

The amounts of PS and the catalyst were important factors in PRO degradation when the sulfate technique was applied. The removal efficiency increased with the increasing PS dose due to the higher amount of reactive radicals (Gao et al. 2018a, 2018b; Minhui et al. 2019). The increase in catalyst dosage resulted in faster degradation.

The pH is a complex parameter in these experiments. pH may affect the ionization state of PRO (pK_a 9.42), the solubility of the applied reagents and the generation of the reactive species. In the UV/PS process, the higher pH resulted in a higher degradation rate (Gao et al. 2018b). Persulfate may also be activated by the basic pH, generating more reactive species. In the PMS/synthesized catalyst system, the following degradation percentages were established: 70.8% (pH₀ 5) < 81.4% (pH₀ 7) < 82.1% (pH₀ 11) < 86.2% (pH₀ 9) (Minhui et al. 2019). The isoelectric point of the applied catalyst (3.68), the ionization state of PRO, and the pK_a of permonosulfuric acid used (9.4) can explain this order. Higher pH was more favourable in PRO degradation. In US/nZVI/PS and FeS/sulphite systems low pH favoured PRO degradation. In both systems, Fe²⁺ ions are present whose solubility is pH dependent (Gao et al. 2018a; Chen et al. 2020). Basic pH may result in precipitation of iron oxides and may passivate the nZVI surface. The best efficiency was gained at pH 3 in US/nZVI/PS-induced PRO degradation (Gao et al. 2018a). In the FeS/sulphite system the effect of pH was studied between 5 and 9. At pH 5 and 6, Fe leaching is greater than at neutral media, inducing faster activation of sulphite (Chen et al. 2020).

The effect of HA and the inorganic ions, HCO₃⁻, NO₃⁻, Cl⁻, SO₄²⁻ and H₂PO₄⁻ was studied to model the efficiency of PRO degradation in natural waters in SO₄^•−-initiated reactions. In the presence of HCO₃⁻, the carbonate radical anion (CO₃^•−) is generated which is less reactive and more selective than ^•OH and SO₄^•−, but the longer lifetime and higher concentration may compensate for the effect of ^•OH and SO₄^•− scavenging (Liu et al. 2016). In UV/PS and PMS/synthesized catalyst systems the presence of HCO₃⁻ favoured PRO degradation (Gao et al. 2018b; Minhui et al. 2019). However, in other systems HCO₃⁻ showed inhibitory effects, the reactivity of the active species and activation of PS decreased (Gao et al. 2018a; Yang et al. 2019; Chen et al. 2020). In the presence of Cl⁻, reactive chlorine-containing species may form in the system. It is not clear that Cl⁻ has a positive or a negative effect on PRO degradation. The effect depended on the applied system and the Cl⁻ concentration. In the lower concentration range, Cl⁻ had a slight inhibitory effect on PRO degradation, in the higher concentration range its presence resulted in faster degradation due to the reactions of chlorine-containing reactants (Gao et al. 2018b; Minhui et al. 2019; Yang et al. 2019). In the FeS/sulphite systems, Cl⁻ had an inhibitory effect, the removal decreased by ∼30%. Cl⁻ raises the ionic strength in the solution that changed the iron leaching for sulphite activation. HA hindered PRO degradation due to its high absorbance in UV/PS, PMS/synthesized catalyst and US/nZVI/PS systems (Gao et al. 2018a, 2018b; Minhui et al. 2019). In the UV/PS system the effect of NO₃⁻ was also investigated. The reaction takes place between NO₃⁻ and SO₄^•−, generating less reactive nitrogen-containing radicals (NO₂^• and NO₃^•) which may decrease the degradation rate (Gao et al. 2018b). In US/nZVI/PS-induced PRO degradation, NO₃⁻ showed a slight inhibitory effect (Gao et al. 2018a).

The basis of the different Fenton methods is the reaction between Fe²⁺ and H₂O₂. In this reaction, reactive ^•OH forms that is responsible for the elimination of organic molecules. Predominantly, the amounts of Fe²⁺ and H₂O₂ are key parameters from the point of view of efficiency.

Only removal of MET was studied by a simple Fenton process at pH 2.8 (Romero et al. 2016). The results showed that 67% conversion of MET can be obtained in 60 minutes in the presence of 10 mg dm⁻³ Fe²⁺ and 150 mg dm⁻³ H₂O₂. Degradation of ATE, MET and PRO was also investigated by the Fenton method using low-intensity interior lighting (Li et al. 2012). All three molecules were eliminated completely by the Fenton method using 20 mg dm⁻³ Fe²⁺ concentration and a 2.5 H₂O₂/Fe²⁺ molar ratio. The degree of degradation rose with the increasing Fe²⁺ concentration, whereas the degradation efficiency changed barely above the molar ratio of 1 H₂O₂/Fe²⁺. Degradation of PRO took place with a higher rate constant than ATE and MET. Regarding the treatment time, more than 90% of PRO can be eliminated in 10 minutes, whereas the degree of degradation was only 80–90% for ATE and MET.

The efficiency of the photo-Fenton reaction was investigated in the cases of ATE and MET (Veloutsou et al. 2014). Similarly to Fenton method, the higher Fe²⁺ concentration increased the efficiency, and more ^•OH forms in the system. However, the excess H₂O₂ may lead to negative results, acting as ^•OH scavengers and forming peroxo compounds. The optimized Fe²⁺ concentrations were 5 mg dm⁻³ and 2.8 mg dm⁻³ for ATE and MET, respectively. In these systems, 100 and 95 mg dm⁻³ H₂O₂ concentrations were used for ATE and MET degradation, respectively. In the first few minutes, ATE and MET disappeared from the solution. After 150 minutes, total mineralization was achieved for both compounds. The time dependence of β-blocker degradation and H₂O₂ consumption was described by pseudo-first order and first-order kinetics, respectively. Photo-Fenton reaction was effective also in real water matrices (river water from the Aksios river, Greece). After 10 min both compounds were eliminated and after 150 min considerable TOC removal was observed (Veloutsou et al. 2014).

Compared to the two Fenton methods, photo-Fenton was proven to be more efficient from the point of view of both treatment time and the amount of Fe²⁺ and H₂O₂ in degradation of β-blockers.

Among electrochemical advanced oxidation processes (EAOP) anodic oxidation (AO) and electro-Fenton (EF) methods can differ. Additionally, sometimes the EF method is combined with photolysis, that is the so-called photo-electro-Fenton (PEF) method. ^•OH forms at the anode in AO and/or in bulk solution through EF reactions (H₂O₂ + Fe²⁺) (Sirés et al. 2010; Isarain-Chávez et al. 2010, 2011; Nsubuga et al. 2019).

The efficiency of AO was investigated with a carbon-felt cathode and Pt or boron-doped diamond (BDD) anode for elimination of the three compounds (Sirés et al. 2010). The BDD anode was more effective than the Pt anode in ATE degradation. At higher applied current (300 mA) a higher mineralization degree was achieved, almost total mineralization took place after 480 min treatment. The electrode area is a key parameter in AO. Higher electrode area resulted in faster degradation.

The removal efficiency of EF reactions was studied in PRO, MET and ATE degradation by BDD, Pt, BDD/ADE, BDD/ADE-Pt/CF combined system and Cu-B-Fe supported graphite electrodes (Isarain-Chávez et al. 2010, 2011; Nsubuga et al. 2019). A Cu-B-Fe supported graphite electrode was applied in PRO and ATE degradation. This method is a so-called droplet flow assisted heterogeneous EF process. Continuous adsorption, electrocatalytic generation of H₂O₂, and diffusion of air occur at the electrode. Copper has a promoter role in the reduction of Fe³⁺ to Fe²⁺. Almost 100% degradation was attained in the cases of PRO and ATE without pH adjustment (Nsubuga et al. 2019).

Degradation of the three β-blockers was studied by EF and photo-EF methods in single (BDD/ADE, Pt/ADE) and combined (BDD/ADE-Pt/CF, Pt/ADE-Pt/CF) cells (Isarain-Chávez et al. 2011). In both EF and PEF methods the combined cells were more efficient than single cells. BDD(^•OH) radicals have higher oxidation capability than Pt(^•OH). Moreover, the application of sunlight and UV light may accelerate the rate of degradation. During EF treatment, Fe(III) complexes may form, which can retard the degradation. The UV light may also induce photolysis of these complexes. When increasing the current intensity, the degradation and mineralization efficiencies were also increasing. Light absorption promoted the mineralization of N-intermediates.

In recent years, there has been a tremendous effort to remove harmful and recalcitrant organic compounds (medicines, personal care products, pesticides, etc.) from wastewaters and purified wastewaters. In AOP, reactive radicals are utilized for this purpose. This paper reviews the elimination of the three most often applied β-blockers, ATE, MET and PRO, using different AOP. This paper summarizes the recent developments in this field. Reactive radicals and agents such as ^•OH, SO₄^•−, O₃, O₂^•−, ¹O₂, O₂^•−/HO₂^• and chlorine-containing species may participate in these degradation reactions. β-Blockers have two susceptible sites: the aromatic ring and the amino moiety. The rate constants measured for the reactive radicals show the following order: SO₄^•− > ^•OH > ClO^• > Cl₂^•− > CO₃^•−.

Ipso attack (i), hydroxylation on the aromatic ring or the amino moiety (ii) and the cleavage of the side chain may be the first steps in the β-blocker degradation. As a result of ipso attack, 3-(isopropylamino)propane-1,2-diol forms with m/z 134 in degradation of β-blockers. Following the hydroxylation on the aromatic ring, ring-opening and rearrangement reactions may take place.

In comparison of several AOP, photocatalysis and sulfate radical-based techniques were the most effective for removal of β-blockers. Using these techniques, it is easy to achieve close to 100% removal of these molecules and close to 100% mineralization.

The effectiveness of different AOP was investigated in detail for degradation of β-blockers, depending on the operating parameters like substrate concentration, photocatalyst loading, pH and the effect of different ions. In order to achieve the best performance, the operating parameters have to be optimized.

The authors thank the International Atomic Energy Agency (IAEA) for support (Coordinated Research Project F23034, Contract no. 23754).

All relevant data are included in the paper or its Supplementary Information.