Assessing the photodegradation potential of compounds derived from the photoinduced weathering of polystyrene in water

(cid:129)

 Both benzoate and acetophenone undergo important reactions with  OH  Acetophenone is also degraded by CO3  , while its direct photolysis would be minor  Volatilisation would be an important removal pathway for aqueous acetophenone  Both compounds show limited reactivity with Br2   Both compounds have higher stability in high-DOC freshwaters and in seawater

Introduction
Pollution by plastics is a growing environmental concern, especially for natural waters where plastic inputs have reached alarming proportions.It has been calculated that approximately 6300 Mt of plastic waste had been produced until 2015 and around 79% (~ 4977 Mt) was accumulated in landfills or the natural environment (Geyer et al., 2017).The majority of plastic debris reaches the sea through rivers, and different estimates have been performed to quantify the amount of plastic transported by rivers to the sea.Such estimates range from 0.47 up to 12.7 million metric tons of plastic per year (Jambeck et al., 2015;Schmidt et al., 2017).
Once they reach the natural environment, plastics undergo weathering largely due to the action of sunlight that operates through two main processes.The former is sunlight-induced fragmentation of plastic objects (macroplastics) into smaller particles, i.e., micro-and finally nanoplastics (Andrady et al., 2022).Adding to the primary environmental sources of small plastic particles (which include their use in fertilizers and release upon runoff from artificial turf pitches; Kumar et al., 2020;Salthammer, 2022), fragmentation highly affects environmental behaviour because of the reactivity of small-sized plastics (Bianco et al., 2020).These can impact living organisms to a different degree than macroplastics, for instance by reaching tissues and crossing membranes (Liu et al., 2021).The second weathering process by sunlight is photodissolution, which consists in photoinduced release to the water phase of additives or polymer fragmentation products (Gewert et al., 2015;Zhu et al., 2020).The chemical compounds arising from plastic weathering can have adverse effects on living organisms, in particular if they have endocrine-disruption properties (Burgos-Aceves et al., 2021).
These compounds might also play potential role as markers of plastic pollution, especially in cases where plastics are difficult to detect, e.g., because of their small size: nanoplastics are particularly challenging as far as detection and polymer identification are concerned (Jakubowicz et al., 2021).
The environmental role of plastic weathering products is deeply affected by environmental persistence.For instance, a persistent compound can undergo accumulation up to relatively high concentration values, which enhances its effects and also makes quantification much easier.
However, it can also migrate quite far from the pollution source.Sunlight, in addition to being a major driver of plastic weathering, would also play a role in the transformation of plastic degradation products in natural surface waters.In particular, solar radiation would operate through direct photolysis and indirect photochemistry (Guo et al., 2022).Direct photolysis occurs when a compound absorbs sunlight, and the absorption process triggers transformation by, for instance, ionisation, bond breaking, and excited-state reactivity (Katagi, 2018).In the case of indirect photochemistry, sunlight is absorbed by naturally occurring photosensitisers such as nitrate, nitrite, and chromophoric dissolved organic matter (CDOM).These species absorb sunlight and yield the so-called photochemically produced reactive intermediates (PPRIs), of which the main ones are the hydroxyl (  OH) and carbonate (CO3  ) radicals, singlet oxygen ( 1 O2), and CDOM triplet states ( 3 CDOM*) (Yan and Song, 2014;Vione and Scozzaro, 2019).The photolysis of nitrate and nitrite yields  OH, while CDOM irradiation yields 3 CDOM*, 1 O2, as well as  OH.CO3  is produced by oxidation of HCO3  /CO3 2 by  OH, or CO3 2 oxidation by 3 CDOM* (Rosario-Ortiz and Canonica, 2016;McNeill and Canonica, 2016;Yan et al., 2019).The PPRIs have very low steady-state concentrations in sunlit natural waters (10 18 -10 14 M), because they are quickly scavenged/quenched by several processes.In particular:  OH is mainly scavenged by dissolved organic matter (DOM, either chromophoric or not) and, usually to a lesser extent, by HCO3  /CO3 2 ; CO3  is mostly scavenged by DOM; 3 CDOM* is mainly quenched by O2 to produce 1 O2, with 50% yield; finally, 1 O2 is mostly quenched by collision with water (Gligorovski et al., 2015;Yan et al., 2019;Ossola et al., 2021).In the case of saltwater and especially seawater, the main scavenging process for  OH is the reaction with Br  to eventually yield Br2  , which is also a PPRI (Parker and Mitch, 2016).The budget between photoproduction and scavenging/quenching controls PPRI occurrence in natural waters under sunlight.In particular,  OH and CO3  tend to be more concentrated at low values of the dissolved organic carbon (DOC), while the opposite happens in the case of 3 CDOM* and 1 O2 (Vione and Scozzaro, 2019).The higher is the steady-state concentration of a PPRI, the higher is its ability to trigger transformation of dissolved compounds, including those released by plastic weathering.
Previous studies have shown that weathering of polystyrene particles under sunlight yields a number of dissolved compounds, among which benzoate (Bz  ) and acetophenone (AcPh) have been identified as important photoproducts (Bianco et al., 2020).For these compounds, reactivity with  OH and CO3  can be assessed on the basis of the literature (Buxton et al., 1988;Neta et al., 1988;Wols and Hofman-Caris, 2021), but no data are currently available for the other photoinduced reactions.Therefore, the goal of this work is to assess the potential for abiotic degradation of both compounds by direct and indirect photochemistry, with the purpose of better defining their fate in the environment.In the case of semivolatile AcPh, we also consider its potential partitioning between the liquid and the gas phase and, finally, its possible photodegradation in both surface waters and the atmosphere.The reactivity of Bz  and AcPh with Br2  was also investigated, to tentatively assess photodegradation of the two compounds in seawater vs. freshwater.

Irradiation experiments
Solutions to be irradiated (5 mL total volume) were placed in cylindrical Pyrex glass cells (height 2.5 cm, diameter 4.0 cm, cut-off wavelength 280 nm), equipped with a lateral neck for liquid insertion and withdrawal.The neck was tightly closed with a screw cap during the irradiation experiments, in which solutions underwent magnetic stirring.Irradiation sources in different experiments were a UVB lamp (Philips Narrowband TL20W/01 RS, emission maximum at 313 nm), a UVA lamp (black lamp TL-D 16W BLB, emission maximum at 369 nm), or a yellow lamp (TL-D 18W/16, emission maximum at 580 nm).The UVB lamp was used to study the direct photolysis of Bz  and AcPh.The UVA lamp was used to achieve selective excitation of 4carboxybenzophenone (CBBP), which was employed as CDOM proxy to study the reactions with 3 CDOM*.Because previous studies have shown that 3 CBBP* has comparable reactivity as average 3 CDOM* (Carena et al., 2019), CBBP irradiation can be used to assess the second-order reaction rate constants between dissolved compounds and 3 CDOM*, by means of steady-state irradiation experiments.Finally, the yellow lamp was used for selective excitation of Rose Bengal, to produce The spectral photon flux density in the solutions subject to irradiation was determined by a combination of lamp spectral measurements (Ocean Optics USB2000 CCD spectrophotometer) and chemical actinometry with 2-nitrobenzaldehyde (Bacilieri et al., 2022).Details are described in Text S1 of the Supplementary Material (hereinafter SM), which also reports the lamp spectra (Figure S1).

Instrumental analysis
Absorption spectra were measured with a Varian Cary 100 Scan UV-Vis double-beam spectrophotometer, using Hellma quartz cuvettes (1 cm optical path length).
After scheduled irradiation times, the cells were withdrawn from the lamp and their contents was analysed by liquid chromatography (HPLC-UV).The instrument used was a Merck-Hitachi chromatograph, equipped with AS2000A autosampler (60 µL injection volume), L-6200 and L-6000 pumps for high-pressure binary gradients, and L-4200 UV-Vis detector.The instrument mounted a Merck LiChroCART® 125-4 column, packed with LiChrospher® 100 RP-18 (5 µm).
Elution (1 mL min 1 flow rate) was carried out with a mixture of 40% acetonitrile and 60% water, acidified with H3PO4 to pH 3. The combinations of detection wavelengths and retention times were 225 nm and 2.75 min for Bz  , and 245 nm and 4.60 min for AcPh.2-Nitrobenzaldehyde, used for actinometry measurements, was determined with the same column eluting (1 mL min 1 ) with a mixture of 35% methanol and 65% acidified water (pH 3, H3PO4), with 7.3 min retention time and detection at 258 nm.

Laser flash photolysis experiments
The fourth harmonic (266 nm) of a Quanta Ray GCR 130-01 Nd:YAG laser system was used to generate and follow the reactivity of Br2  in aqueous solution, in the presence of Bz  and AcPh.
The excitation energy was 36 mJ/pulse and an individual cuvette sample (3 mL volume) was used for a maximum of four consecutive laser shots.The spectroscopic system has been described before (Brigante et al., 2014).
Br2  was generated through photolysis of 10 mM H2O2 in the presence of 10 mM Br  .Under such conditions, the presence of a long-lived transient species absorbing from 270 to 550 nm with a maximum at 360 nm (ε ~ 9.9 × 10 3 M −1 cm −1 ) was assigned to the formation of Br2  (Hug, 1981).
The second-order reaction rate constants of Br2  with Bz  and AcPh were determined from the slopes of the regression lines of the logarithmic decays of the Br2  transient (monitored at 360 nm), as a function of the concentration of each selected aromatic compound (Stern-Volmer method).The error was estimated as ±σ, obtained from the scattering of the experimental data around the linear fit line.

Environmental fate modelling
Photodegradation kinetics of Bz  and AcPh were modelled with the APEX software (Aqueous Photochemistry of Environmentally-occurring Xenobiotics) (Vione, 2020).This software predicts pseudo-first order photodegradation rate constants by direct photolysis and indirect photochemistry, as a function of sunlight irradiance, water depth, and water chemistry (concentration values of photochemically relevant compounds: nitrate, nitrite, bicarbonate, carbonate, and dissolved organic carbon, DOC) (Silva et al., 2015;Silva et al, 2019;Vione, 2020).APEX assumes well-mixed water bodies under clear-sky conditions, and in these circumstances it has been shown to predict environmental photodegradation kinetics with good accuracy (Avetta et al., 2016).
The volatilisation rate constant of AcPh from aqueous environments was predicted with the EPISUITE TM package by US-EPA (US-EPA, 2021), based on molecular structure (quantitative structure-activity relationship).Water depth was varied in the range of 1 to 5 m, while it was assumed 1 m s 1 wind speed.EPISUITE TM (AOPWIN sub-package) was also used for the prediction of the reaction rate constant between AcPh and gas-phase  OH, assuming average [  OH(g)] = 1.510 6 cm 3 .

Results and Discussion
To assess the photochemical degradation potential of Bz  and AcPh, all the reaction rate constants with photochemically produced reactive intermediates (PPRIs) and the direct photolysis quantum yields need to be retrieved from literature, or measured through laboratory experiments.
The second-order reaction rate constants of Bz  and AcPh with  OH are known from the literature, and they are both equal to 5.910 9 M 1 s 1 (Buxton et al., 1988;Wolf and Hofman-Caris, 2012).
Irradiation experiments, carried out in the framework of this work, allowed for excluding significant reactions of the studied compounds with either 3 CBBP* (and, therefore, 3 CDOM*) or 1 O2.

Photochemical persistence of Bz  and AcPh in natural surface waters
Based on the above photoreaction parameters, the photochemical degradation kinetics of Bz  and AcPh in sunlit natural waters was assessed by means of the APEX software (see Figure 1).The assumed conditions of sunlight irradiation ('day') correspond to the 24-h-round spring equinox at 45°N latitude.As such, they also represent reasonable year averages at mid latitude under fairweather conditions (Vione, 2020).Calculation results suggest that AcPh direct photolysis would play minor role, thus Bz  would be photodegraded by  OH and AcPh would react with both  OH and CO3  .Predicted photodegradation lifetimes range from 1-2 weeks to several months.Photodegradation kinetics would be highly dependent on the environmental conditions.Kinetics would be the fastest in shallow waters with low DOC, because: (i) shallow waters are better illuminated by sunlight than deep waters (Loiselle et al., 2008), the bottom layer of which is usually in the dark, and (ii) organic matter is a major scavenger of both  OH and CO3  (Yan et al., 2019) and it acts as a light-screening agent that prevents the direct photolysis of other compounds, including the photosensitisers nitrate and nitrite.Therefore, kinetics of photodegradation by  OH and CO3  are slower at high DOC values.
The important role played by  OH and/or CO3  in the photodegradation of Bz  and AcPh suggests that nitrate concentration might affect phototransformation kinetics considerably.Nitrate is in fact a direct  OH and indirect CO3  source (Vione and Scozzaro, 2019) and, as shown in Figure S2 (SM), the pseudo-first order rate constants of both Bz  and AcPh increase with increasing nitrate, all other conditions being equal.The nitrate effect is substantial at low DOC (increase by 4-5 times in degradation kinetics when passing from 10 6 M to 10 4 M NO3  , for DOC = 1 mgC L 1 ; Figure S2), while the effect is much less important at higher DOC values.The rationale is that high DOC entails the scavenging by DOM of the vast majority of  OH and CO3  photogenerated by nitrate.

Possible environmental scenarios
Different scenarios of photoinduced degradation in the environment were taken into account for the studied compounds.In the case of non-volatile Bz  we have considered photodegradation in surface waters only.Bz  formation kinetics from polystyrene exposed to sunlight has been studied previously (Bianco et al., 2020) and the available data allow for describing a possible scenario where polystyrene degradation yields Bz  , which is then photodegraded by  OH.By so doing, it is possible to assess a potential for Bz  accumulation in natural waters.In the case of semivolatile AcPh, we simulated a scenario of degradation in both surface waters and the atmosphere, combined with AcPh volatilisation from the aqueous phase to the gas phase.

Benzoate (Bz  )
In the case of Bz  , formation kinetics from 0.2% v/v polystyrene in water under environmental irradiation is known (Bianco et al., 2020), which would be representative of strong water pollution by polystyrene particles.The scenario considered here assumes that polystyrene particles occur near the water surface (i.e., they float), while released Bz  undergoes diffusion in the whole water volume.In addition, it is assumed that Bz  is degraded by reaction with  OH (see Figure 1a), following first-order kinetics.Details of kinetic calculations are provided in Text S2 of SM.The time trend of Bz  that results from the formation-photodegradation budget is the following: where RBz-= 3.110 7 M day 1 and do = 0.023 m are derived from the literature (Bianco et al., 2020).They are, respectively, the experimental formation rate of Bz  from polystyrene and the experimental water depth, see also Text S2.Furthermore, d [m] is the assumed depth of the water body, kBz-[day 1 ] the pseudo-first order degradation rate constant of Bz  (modelled with APEX), and t is time in days.
To simplify calculations, as an approximation we kept kBz-constant with time at its equinox value (yearly average).In other words, the value of kBz-we used varied in different conditions of, e.g., water depth (d) and DOC as per Figure 1a, but its seasonal fluctuations were neglected.Seasonal variations were neglected also in the case of RBz-, which is a year average as well.Considering that polystyrene releases Bz  under irradiation and that Bz  would react with photogenerated  OH, it is likely that RBz-and kBz-vary in parallel and that their seasonal variations cancel out at least partially in Eq. ( 1).With these approximations, the Bz  time trends in different conditions are shown in Eq. (1) predicts that [Bz  ] reaches a plateau after quite long time (high t values).The plateau concentration results from the budget between generation from polystyrene and degradation by  OH.Furthermore, the initial increase in [Bz  ] (low t values, t < 50 days) mostly depends on water depth d: if water is deep, generated Bz  is diluted in a higher volume, and its concentration increase is slower (and vice versa, in the case of shallow water).At later times, water depth has two opposite impacts on the Bz  time trend, which partially offset each other (Text S2): the former is the dilution effect, already mentioned; the latter is the fact that the deeper is water, the slower is Bz  degradation upon reaction with  OH (see Figure 1a).
The plateau concentration of Bz  strongly depends on the DOC: if DOC is high, Bz  degradation is slow and [Bz  ] can reach higher values with potential accumulation; the opposite happens at low DOC.As shown in Figure 2, Bz  would eventually reach sub-µM levels.By comparison, in the literature experiment that yielded RBz-= 3.110 7 M day 1 (Bianco et al., 2020) the volume was quite small (d = 0.023 m, thereby ensuring limited dilution) and the aqueous phase did not contain  OH sources.In such circumstances, Bz  could accumulate with little dilution or degradation and reach 0.13 mM levels in 14 months.In the current simulations, for water depths of some metres and in the presence of Bz  degradation by  OH, we predict [Bz  ] to reach values that are lower by 2-3 orders of magnitude (sub-µM vs. sub-mM), which looks reasonable.

Acetophenone (AcPh)
As mentioned before, AcPh could be degraded in water by (mainly)  OH and CO3  , or volatilise to the gas phase where it would react with  OH(g), similarly to other semivolatile compounds (Arsene et al., 2022).The overall scenario is depicted in Scheme 1a, and two specific scenarios are illustrated in Scheme 1b.
Starting from AcPh in water (AcPhw) at initial concentration Co, the kinetic system depicted in Scheme 1 gives the following time trends for [AcPhw] and [AcPhg] (w = water phase, g = gas phase).Equations (2,3) are solutions of differential equations and were derived in similar way as Arsene et al. (2022) (see Scheme 1 for the meaning of the different parameters): The values of kAcPh as a function of water depth and DOC were modelled with APEX (see Figure 1b); kg was assessed as 0.12 day 1 by (Figure 3a,c).Furthermore, removal kinetics of AcPh from the aqueous phase would be considerably faster at d = 1 m (where kAcPh and especially kv are considerably higher, see Figure 3a and Scheme 1b) than at d = 5 m (Figure 3c).The initial increase of [AcPhg] is controlled by kv and, therefore, it does not depend on the DOC (Figure 3b,d).The gas-phase degradation of AcPh is also DOC-independent, because kg = 0.12 day 1 is only linked to the value of [  OH(g)].In spite of these issues, there are some differences in the [AcPhg] time trends (Figure 3b,d), which depend on the DOC value of the aqueous phase.The reason is that low DOC entails slightly faster decay of AcPhw (Figure 3a,c), because of higher values of [  OH] and [CO3  ] in the aqueous phase that lead to higher kAcPh.As AcPhw undergoes slightly faster consumption at low DOC, a lesser amount is available for partitioning to the gas phase and, therefore, AcPhg reaches slightly lower concentration values if the water DOC is low (Figure 3b,d).
Overall, model results suggest that volatilisation and reaction with gas-phase  OH have potential to play key roles in the environmental fate of AcPh.

Reactivity with Br2  , and implications for photodegradation in seawater
The radicals  OH and CO3  are expected to play important roles in the photodegradation of Bz  (only  OH) and AcPh (both species) in sunlit freshwaters.However, in the case of brackish waters and saltwater,  OH would be very effectively scavenged by bromide (Buxton et al., 1988;Parker and Mitch, 2016).This phenomenon would strongly inhibit the  OH-mediated processes, and it would also block the main CO3  formation pathway (HCO3  /CO3 2 oxidation by  OH itself; Canonica et al., 2005).At the same time, the reactive radical Br2  would be generated by  OH in the presence of bromide (Parker and Mitch, 2016): the main processes connected with the occurrence of Br2  in environmental waters are depicted in Scheme 2.
Based on Scheme 2, and applying the steady-state approximation to both Br  and Br2  , one gets the following equations for, respectively, Br2  formation rate and its steady-state concentration (De where the reaction rate constants k i -k v are those shown in Scheme 2. (2.00.5)10 7M 1 s 1 as the second-order reaction rate constants of Br2  with, respectively, Bz  and AcPh.
On the basis of the Br2  second-order reaction rate constants obtained by laser flash photolysis, it is possible to assess the first-order rate constants of degradation by Br2  as kSubstrate = First of all, 1 mM bromide (relevant to seawater conditions; Fukushi et al., 2000) would slow down photodegradation of both compounds by over ten times.The  OH pathway would be particularly affected, due to  OH scavenging by Br  , followed by CO3  that would also be inhibited considerably.Although both Bz  and AcPh would be degraded by Br2  to a significant extent in the presence of 1 mM Br  , this process would be unable to offset inhibition of the  OH and CO3  pathways.Therefore, Bz  and AcPh are potentially more photostable in saline waters than in freshwater.
where p°() is the spectral photon flux density (measured by chemical actinometry) that passes through the irradiated solutions (FigureS1, SM), AcPh() the molar absorption coefficient of AcPh (Figure S1, SM), b = 0.4 cm the optical path length of the irradiated solutions, and CAcPh = 5 µM.

Figure 2 .
Figure 2. Time trends of Bz  in different surface-water scenarios, as predicted by Eq. (1).The value of kBz-

Figure 3
Figure 3 for different values of the water depth d (1 or 5 m) and of the DOC (1, 3, or 10 mgC L 1 ).

Scheme 1 .
Scheme 1.In a) overall reaction scheme of AcPh in the aqueous phase (w) and in the gas phase (g).The

Scheme 2 .
Scheme 2. Main reactions accounting for the production (in the light blue triangle) and scavenging/quenching (in the dark blue rectangle) of Br2  in bromide-containing natural waters.The  ], where 'Substrate' = Bz  or AcPh.The effect of bromide on the overall kinetics (pseudo-first order rate constants) and on the different pathways of Bz  and AcPh degradation is shown in Figure 4.

Figure 4 .
Figure 4. Modelled pseudo-first order photodegradation rate constants of Bz  and AcPh, in the absence of