Synergy between Sunlight, Titanium Dioxide, and Microbes Enhances Cellulose Diacetate Degradation in the Ocean.

Sunlight chemically transforms marine plastics into a suite of products, with formulation─the specific mixture of polymers and additives─driving rates and products. However, the effect of light-driven transformations on subsequent microbial lability is poorly understood. Here, we examined the interplay between photochemical and biological degradation of fabrics made from cellulose diacetate (CDA), a biobased polymer used commonly in consumer products. We also examined the influence of ∼1% titanium dioxide (TiO2), a common pigment and photocatalyst. We sequentially exposed CDA to simulated sunlight and native marine microbes to understand how photodegradation influences metabolic rates and pathways. Nuclear magnetic resonance spectroscopy revealed that sunlight initiated chain scission reactions, reducing CDA's average molecular weight. Natural abundance carbon isotope measurements demonstrated that chain scission ultimately yields CO2, a newly identified abiotic loss term of CDA in the environment. Measurements of fabric mass loss and enzymatic activities in seawater implied that photodegradation enhanced biodegradation by performing steps typically facilitated by cellulase. TiO2 accelerated CDA photodegradation, expediting biodegradation. Collectively, these findings (i) underline the importance of formulation in plastic's environmental fate and (ii) suggest that overlooking synergy between photochemical and biological degradation may lead to overestimates of marine plastic persistence.

Introduction

As plastic production scales with increasing global demand,[1] its resulting leakage and persistence in the environment have prompted action by many stakeholders.[2] One promising short-term solution is to shift toward plastic formulations that do not persist in the environment.[3−5] However, robust information on the environmental persistence and impacts of specific formulations remains sparse.[2] One polymer proposed to have a short environmental lifetime is cellulose diacetate (CDA), which is already widely used in consumer goods, including textiles[6] and cigarette filters.[7] Because CDA comprises the latter, it is also a major component of the most prevalent form of marine debris: cigarette butts.[7] We recently reported that CDA-based materials (i.e., films, fabrics, and foams) are susceptible to degradation by marine microbes on time scales of months—orders of magnitude faster than previously reported by international government agencies.[8] While this is a promising step toward constraining the environmental persistence of CDA, our understanding of other factors that may influence persistence in the ocean, including photodegradation and additives, are less complete.

Photodegradation is a primary control of plastic fate in the environment,[9−14] but its impact on subsequent biodegradation of CDA and overall lifetime is unclear. Sunlight and microbes are known to synergistically degrade (“photo-biodegradation”) organic molecules in surface waters,[15,16] but such studies on marine plastics remain sparse.[11,17] Several studies have examined CDA photodegradation under ultraviolet (UV) light and collectively suggest that pure CDA has somewhat limited photoreactivity under natural sunlight.[6,18−23] However, the few existing studies on CDA suggest that prior degradation by UV light can appreciably enhance microbial degradation, although the precise pathway is debated.[21,22] Thus, CDA’s overall susceptibility to photodegradation and the kinetics and pathways of photo-biodegradation in the ocean remain open questions.

Plastic formulation influences many aspects of its fate, including photo-[9,11,13,24] and biodegradation.[11,17] Photocatalytic additives, in particular, have emerged as a promising route for reducing plastic lifetimes.[24−27] Titanium dioxide (TiO2) is a semiconductor photocatalyst already widely used in plastics as a white pigment.[28,29] Even at trace concentrations (<1% by weight), TiO2 accelerates degradation of several plastics under environmental conditions.[6,24,27] The influence of TiO2 on CDA has only been studied under UVC light (which does not reach Earth’s surface), but appears to substantially accelerate degradation.[6,22] Furthermore, breakdown of CDA by UVC and TiO2 enhanced the lability of CDA to enzymatic degradation.[22] Thus, the next step in assessing the role of TiO2 in CDA’s fate in the ocean is to constrain degradation kinetics and pathways during exposure to environmentally relevant wavelengths and microbial communities.

Here, we determined the time scales and pathways for photo-biodegradation of CDA fabrics with and without TiO2. First, we constrained photodegradation rates and pathways during irradiation with simulated sunlight by quantifying photochemical oxygen (O2) consumption and carbon dioxide (CO2) production. We used nuclear magnetic resonance (NMR) to measure photochemical chain scission of the polymer and natural abundance carbon isotope measurements to characterize the source of the photoproduced CO2. Next, we examined the degradation of pristine and photodegraded fabrics by native marine microbes using time-lapse photography, mass loss measurements, and enzymatic assays in a series of flow-through natural seawater mesocosm experiments. We found that chain scission of the polymer backbone by sunlight replaced a key metabolic step in microbial degradation of CDA, shortening its overall lifetime. TiO2 further expedited photo-biodegradation. These findings further support plastic formulation as a control of CDA’s environmental fate and suggest that photo-biodegradation may be an overlooked fate of marine plastics.

Methods

Fabrics

Two CDA fabrics obtained from Eastman Chemical Company (Kingsport, TN) were studied: a CDA fabric with no additives (“pure CDA”) and a CDA fabric containing 1.1% TiO2 (“TiO2–CDA”). Both fabrics were a single jersey knit construction composed of 150 denier, drawn CDA yarn with 38 filaments and a fabric weight of 97 g m−2. Fabric thickness was measured with a NIST-calibrated Mitutoyo micrometer with a precision of 1 μm, with six measurements taken of each sample. The pure and TiO2–CDA were 260 ± 7 and 250 ± <1 μm thick, respectively [± 1 standard deviation (SD)].

Characterization of Fabric Optical Properties and Composition

UV and visible light absorbance were determined for the fabrics using a PerkinElmer LAMBDA 650S spectrophotometer. Each fabric was cut into ∼1 × ∼1 cm squares and measured in duplicate at a 10° angle inside the integrating sphere. Naperian absorption coefficients, a (m–1), were calculated by multiplying measured absorbance, A, by ln(10) and dividing by fabric thickness (m).[31]

The organic contents of the fabrics were confirmed using loss on ignition (LOI) measurements based on a modified version of ASTM D 2974-87 (“Standard Test Methods for Moisture, Ash, and Organic Matter of Peat and Other Organic Soils”) for plastics.[24,31,32] For each fabric, ∼500 mg were placed in a precombusted, preweighed ceramic crucible and combusted at 450 °C for 4 h (n = 3), after which the ash-containing crucible was reweighed. The identities of any mineral additives in the fabrics were determined in triplicate using X-ray diffraction (XRD) on the remaining ash (Appendix 1).[33]

Complete and Partial Photochemical Oxidation of Fabrics

Two parallel time series experiments were conducted using approaches described previously[9] to measure photochemical O2 consumption and CO2 production for each fabric. Triplicate samples were irradiated in an Atlas XLS+ solar simulator equipped with a daylight filter (Appendix 2) over multiple time points alongside triplicate dark controls and measured for O2 consumption and dissolved inorganic carbon (DIC) production. These time points were 96 and 144 h for the pure CDA and 4, 6, and 24 h for the TiO2–CDA. The 24 h time point for the TiO2–CDA was designed to be anoxic to gain insight into potential oxidation pathway differences. Anoxic conditions were achieved by irradiating the sample beyond the point at which it had consumed all O2 present in the seawater. The durations of the dark controls matched the longest time point for their respective fabrics. For each fabric, one set of samples was analyzed for O2 consumption on a membrane inlet mass spectrometer (MIMS; Bay Instruments, Inc.) as the difference between dark control and light-exposed dissolved O2 concentrations. The other set was analyzed for CO2 production on an AS-C3 DIC analyzer (Apollo SciTech, Inc.) as the difference between light-exposed and dark control DIC concentrations. For each time point, the ratio of photochemical O2 consumption to DIC production was calculated as a proxy for the relative importance for complete oxidation to CO2 (photomineralization) and partial oxidation to oxygenated products.[9,31,34]

Tracking Photomineralization of Fabrics with Natural Abundance Carbon Isotopes

Carbon Isotope Measurements of Bulk Materials

Natural abundance carbon (C) isotope measurements were leveraged to track photomineralization of the fabrics to DIC using previously described methods.[9,35] We first measured the carbon isotope compositions of the virgin CDA fabrics, as well as the major precursor materials used in their synthesis, cellulose and acetic acid, which were supplied by Eastman Chemical Company. Because CDA is sourced from modern, cellulosic C from wood pulp (Δ14C = 98 ± 3.0‰; δ13C = −24.4 ± 0.02‰) and fossil, acetyl C (Δ14C = −994‰; δ13C = −21.4‰) from acetic acid, the intermolecular carbon isotope composition is mixed (reported in Mazzotta et al.).[8] All fabrics and precursors were measured at the National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) using elemental analysis or closed tube combustion. The Δ14C measurements of the virgin CDA materials were then compared to predictions based on the formulations provided (e.g., degree of acetyl substitution and composition) and the measured Δ14C of the cellulose and acetic acid. Strong agreement between the predicted and measured Δ14C was observed (R2 = 0.99).[8] A fossil fuel-derived pure polystyrene (PS) film (Goodfellow Corporation; product code GF39518556) was also included for quality control assessment throughout the experiments and was consistent with prior measurements.[9]

Photomineralization Experiments

Photochemical mineralization of the fabrics, including any preferential mineralization of cellulosic and acetyl moieties, was tracked during irradiation of the fabrics in natural seawater, as described before.[9,35] All experiments were conducted in low-DIC, low-dissolved-organic-carbon (DOC) seawater (preparation described in Appendix 3) to reduce the high seawater DIC background concentration and minimize interference from photomineralization of seawater DOC. For each sample, 2.5 cm radius circles of MilliQ-cleaned fabric were placed in a 60 mL quartz bomb flask. Next, the flasks were filled with seawater and sealed with no headspace with a glass stopper using Apiezon grease and Keck clips to ensure the flasks were airtight. Each fabric was irradiated in duplicate over multiple time points selected for optimal DIC production alongside two dark controls. The pure CDA was irradiated for 15 and 30 days, and the TiO2–CDA was irradiated for 3, 6, and 12 h. Additionally, a single 48 h sample was included for the TiO2–CDA to examine DIC production under anoxic conditions, and oxygen consumption was measured using the MIMS at 15 and 48 h to ensure that only this final time point was anoxic. Background DIC production from seawater DOC was not detected during control experiments and therefore omitted from mass balance calculations. Light-exposed samples were placed in the solar simulator inside a water bath, which maintained the flasks at 30 °C during the experiment. Samples were covered with a UVA- and UVB-transmitting acrylic heat shield (Arkema Inc.). In this configuration, simulated sunlight was 3- and 10-fold greater than yearly averaged natural sunlight at 0 and 50° N, respectively (Table S1). Dark controls were refrigerated for the duration of the experiment (i.e., the longest time point for each fabric) to inhibit potential microbial activity. DIC samples were analyzed for δ13C and Δ14C via the water stripping line at NOSAMS.[36]

NMR Characterization of Photoweathered and Unweathered Fabrics

Irradiation of Fabrics

NMR was used to measure changes in the molecular structures of each CDA fabric resulting from irradiation with simulated sunlight. First, three ∼3 × ∼5 cm pieces were cut from each fabric and placed in the solar simulator on a gray-painted aluminum tray on a water-cooled plate. The water-cooled plate was connected to a circulating water bath to maintain the fabrics at 25 °C during irradiations. One piece of each CDA type was removed after 3.5, 7, and 11 days of irradiation. During each irradiation period, the pieces were flipped such that both sides of each piece received identical irradiation, equivalent to half of the total exposure time. One piece of each CDA type was kept in a black pouch in the solar simulator throughout the exposure time to serve as a dark control.

Following irradiation, CDA samples were prepared for NMR experiments. Triplicate cut pieces of ∼10 mg (∼1 × ∼1 cm) were dissolved in 1 mL of deuterated dimethyl sulfoxide (DMSO-d6; ≥99.8% deuterated; Sigma-Aldrich), and 500 μL was transferred to an NMR tube. Six powdered dextran analytical standards with different well-defined molecular weights (weight-averaged molecular weights (Mw) of approximately 12,000–270,000 Da; Sigma-Aldrich) were prepared in the same manner. All experiments were performed on a Bruker Avance NEO spectrometer equipped with an Ascend 400 MHz magnet and 4 mm BBO H&F Cryoprobe. Experiments were performed at 25 °C using only 500 μL of the sample (height of 35 mm) to well-contain the solution in the temperature-controlled portion of the probe. Spectra were analyzed using Bruker TopSpin (version 4.1.0); chemical shifts were set relative to that of residual DMSO protons (2.50 ppm), followed by manual phase correction and integration.

Nuclear Magnetic Resonance Experiments

Diffusion-ordered spectroscopy (DOSY) experiments to determine Mw were performed using a longitudinal eddy current delay pulse sequence with bipolar gradient diffusion pulses (1800 μs each) and two spoil gradient pulses (600 μs each) (Bruker pulse sequence “ledbpgp2s”).[37] The eddy current delay time was 5 ms, and the gradient recovery delay was 0.2 ms. Each DOSY experiment consisted of 16 pulse sequences with linearly increasing gradient strengths from 2 to 98% of the maximum strength. For CDA samples, a diffusion delay of 1000 ms and recycle delay of 5 s were used, with 64 scans collected at each gradient strength. For dextran samples, varying diffusion delays of 400–1,150 ms were used, depending on the Mw, and a recycle delay of 10 s was used, with 16 scans collected at each gradient strength. Diffusion coefficients (D) of samples were derived from DOSY experiments by plotting relative integrated intensities of select spectral regions against gradient strength. For dextran samples, the spectra were integrated between 5.1 and 4.4 ppm, and for CDA samples, between 5.7 and 3.5 ppm. The relationship between log10 D and log10 Mw (from manufacturer-reported values) of the dextran standards was established with a linear model with good linearity and fit (Figure S5) and used to determine Mw of CDA samples from their measured D. These analyses were performed using RStudio (version 1.4.1717) and R (version 4.1.1 “Kick Things”).

1H spectra of CDA samples were obtained using a 90° pulse sequence (12 μs length), collecting 128 scans with a 10 s recycle delay. These spectra were integrated to estimate the degree of substitution (DoS; i.e., the average number of acetyl groups per cellulosic ring) of CDA samples by comparing integrals of acetyl protons (2.30–1.54 ppm) with those of cellulosic-backbone protons (5.73–3.41 ppm). DoS of each sample was calculated by dividing the integrated area ascribed to acetyl protons, Aacetyl, by the integrated area ascribed to cellulosic-backbone protons, Abackbone, while normalizing for the number of protons contributing to each area, according to eq  [8]

Biodegradation of Photoweathered and Unweathered Fabrics by Native Marine Microbes

Microbial degradation of photoweathered and unweathered CDA fabrics was determined throughout a 13-week incubation in a continuous flow-through seawater mesocosm.[8] In addition to the CDA fabrics, cotton fabric was used as a positive control (high degradative capacity), while poly(ethylene terephthalate) (PET) fabric was used as a negative control (low degradative capacity), as described previously.[8] Fabrics were irradiated for 14 days in triplicate using a similar setup as the NMR experiments (Section ). After irradiation, the fabrics were disinfected with 70% ethanol and transferred in triplicate to the natural seawater mesocosm (Appendix 4)[8] for a 13-week incubation experiment. At several time points, the fabrics were assessed for disintegration via time-lapse photography (Appendix 5), mass loss (Appendix 6), and esterase and cellulase enzymatic activities (Appendix 7) using methods developed and described by Mazzotta et al.[8]

Results

Characterization of Fabric Optical Properties and Composition

The CDA fabrics had different UV and visible light absorption due to the presence of TiO2 (Figure S1). The pure CDA showed moderate light absorption in the UVB region that declined gradually to baseline around 330 nm, which is typical for CDA.[19] The TiO2–CDA showed strong absorption across the UVB region that declined sharply to baseline at ∼380 nm, a spectral shape characteristic of TiO2.[29]

LOI (Figure S2) and XRD (Figure S3) measurements confirmed that the pure CDA was free of inorganic additives, while the TiO2–CDA was not. 100.0 ± 0.3% (± 1 SD, n = 3) of the pure CDA was lost on ignition, which indicates that the material was completely organic; 98.9 ± 0.1% (± 1 SD, n = 3) of the TiO2–CDA was lost on ignition, indicating that it contains 1.1% inorganic additives. XRD measurements confirmed that the sole inorganic additive in the TiO2–CDA was TiO2, which was present as anatase.

Complete and Partial Photochemical Oxidation of Fabrics

Both fabrics consumed O2 and produced CO2 during simulated sunlight exposure, but the rates for both processes were over an order of magnitude faster in the TiO2–CDA (Figures and S4; note difference in exposure times). The pure CDA consumed 1.1 ± 0.2 μM O2 per hour, while the TiO2–CDA consumed 26.1 ± 1.1 μM O2 per hour—24 times faster for the TiO2–CDA. Under oxic conditions, pure CDA produced, on average, 0.4 ± 0.1 μM CO2 per hour (n = 6), while the TiO2–CDA was 40 times faster, producing 16.2 ± 0.9 μM CO2 each day (n = 6). Under anoxic conditions, the TiO2–CDA was still mineralized to CO2, but the production rate dropped by about half, to 7.6 ± 2.0 μM CO2 per hour (n = 3) (SI Figure S4d). Overall, photochemical O2 consumption exceeded CO2 production for both fabrics, but the ratio of O2 consumption to CO2 production was significantly (p = 0.0071, unpaired Student’s t-test) higher for the pure CDA (2.8 ± 0.3) compared to the TiO2–CDA (1.6 ± 0.8).

Figure 1

Photochemical O2 consumption (orange) and CO2 production (red) for the pure CDA (left) and the TiO2–CDA during simulated sunlight exposure. Error bars represent ± 1 SD (n = 3).

The pure and TiO2–CDA fabrics had the same 14C- and 13C-signatures: Δ14C and δ13C were −408 ± 1 and −33.54 ± 0.02‰, respectively, for the pure CDA and −410.5 ± 0.4 and −33.70 ± 0.02‰, respectively, for the TiO2–CDA (Figure and Table S2). The mixed Δ14C of the CDA reflects the modern wood pulp (Δ14C = 98 ± 3.0‰; δ13C = −24.4 ± 0.02‰) from which the cellulose backbone is derived and the fossil acetic acid (Δ14C = −994‰; δ13C = −21.4‰) from which the acetyl groups are derived, as reported previously.[8] The seawater in which the pure CDA was irradiated had a DI14C of −39 ± 9.0‰ and DI13C of −12.9 ± 0.3‰; for the TiO2–CDA, DI14C was −135 ± 7.4‰, and DI13C was −9.0 ± 0.7‰. These large isotopic differences between the cellulosic C, acetyl C, and seawater DIC thereby provide an opportunity to determine which C atoms in the CDA are preferentially mineralized to DIC upon sunlight exposure.

Figure 2

Natural abundance carbon isotope measurements of bulk and source materials (diamonds), dark control seawater (open circles), and seawater DIC after irradiation of the (a) pure CDA and (b) TiO2–CDA (shaded circles). Circle shading corresponds to DIC concentration and becomes darker with irradiation time as polymer-derived DIC is added. Error bars represent ± 1 SD (n = 1−2). Dotted lines show expected trajectories of the seawater DIC if only acetyl carbon was mineralized (top), if acetyl and cellulosic carbon were proportionately mineralized (middle), and if only cellulosic carbon was mineralized (bottom).

Measurements of DIC concentration and DI14C and DI13C after sunlight exposure confirmed that the CDA-derived C in fabrics were photomineralized to DIC (Figure and Table S2). Sunlight exposure of each fabric led to increases in DIC concentration, but CO2 production was over an order of magnitude faster for the TiO2–CDA relative to the pure CDA. As DIC concentrations increased with sunlight exposure, DI13C decreased, consistent with the production of CO2 from the more 13C-depleted fabrics. For the pure CDA, DI13C decreased to −17.3 and −20.0‰ after 14 and 28 days of exposure, respectively. For the TiO2–CDA, DI13C decreased to −11.3, −13.0, and −14.3 ± 0.25‰ after 3, 6, and 12 h of exposure, respectively. By 48 h, when the water was anoxic, DI13C had dropped to −19.8‰.

Despite substantial photochemical production of DIC and associated shifts in DI13C to more depleted values, substantial shifts in DI14C were not observed for either fabric (Figure and Table S2). For the pure CDA, DI14C decreased slightly from −39 ± 9.0 to −46‰ and −57 ± 10.9‰ after 14 and 28 days of exposure, respectively. For the TiO2–CDA, DI14C decreased slightly from −135 ± 7.4 to −131‰, −132 ± 8.4, and −126 ± 5.5‰ after 3, 6, and 12 h of exposure, respectively. For the anoxic 48 h time point, DI14C was −112‰.

Isotopic mass balance calculations revealed that cellulosic C was preferentially photomineralized for both fabrics. Photomineralized C from the CDA fabrics had a relatively modern Δ14C, similar to that of the seawater DIC, suggesting that a disproportionately large amount of cellulosic carbon was converted to DIC during irradiation. For the pure CDA, the average Δ14C of photomineralized fabric over all time points was −59 ± 13‰ (n = 2). For the TiO2–CDA, the average Δ14C of photomineralized fabric was −104 ± 18‰ over all oxic time points (n = 3). In other words, 86 and 82% of photomineralized C was derived from cellulose for the pure and TiO2–CDA, respectively. In contrast, the proportions of cellulosic C in the pure and TiO2–CDA bulk fabric were 54 and 53%, respectively. Thus, the Δ14C of the mineralized C reflects preferential conversion of cellulosic C to CO2. The δ13C and Δ14C of the DIC produced by the TiO2–CDA under anoxic conditions were similar to that produced under oxic conditions, although the rate of DIC production decreased.

NMR Characterization of Photoweathered and Unweathered Fabrics

Diffusion-ordered spectroscopy (DOSY) and proton (1H) NMR experiments were used to characterize the average molecular weights (Mw) and degrees of substitution (DoS), respectively, of CDA fabrics over a photoirradiation time series. Sunlight exposure reduced Mw of the pure CDA and, to a greater extent, the TiO2–CDA (Figure ). Both fabric types had the same Mw prior to irradiation (pure CDA: 182 ± 13 kDa; TiO2–CDA 178 ± 11 kDa; n = 3 for each fabric at each time point). The Mw of pure CDA was unchanged after 3.5 days but decreased with an increasing rate up to 11 days of irradiation, at which point it was 150 ± 6 kDa. In contrast, after only 3.5 days of irradiation, the Mw of TiO2–CDA was reduced to 142 ± 6 kDa and continued decreasing at a slower rate, reaching 124 ± 9 kDa after 11 days. Dark controls for both fabric types showed no significant change in Mw from the initial values (188 ± 9 kDa and 168 ± 8 kDa for pure CDA and TiO2–CDA, respectively).

Figure 3

Average molecular weight of the bulk pure CDA (blue) and TiO2–CDA (green) over 11 days of simulated sunlight exposure. Error bars represent ± 1 SD (n = 3).

DoS calculated from 1H NMR spectra did not drastically change for either fabric after sunlight exposure (Figure S6). The pure CDA had a measured DoS of 2.58 ± 0.02 and 2.53 ± 0.03 before and after 11 days of irradiation, respectively, while the TiO2–CDA had a DoS of 2.52 ± 0.10 and 2.57 ± 0.04 before and after 11 days of irradiation, respectively (n = 3 for each fabric at each time point). Additionally, we note a very small peak at 1.9 ppm measured in the 1H NMR spectra of both fabrics at the end of the irradiation experiment (Figure S6 insets). This peak location is consistent with the formation of acetic acid upon deacetylation.[18]

Microbial Activity on Photoweathered and Unweathered Fabrics

A series of natural seawater mesocosm experiments revealed the effects of sunlight exposure on CDA fabric degradation by microbes. The light-exposed and dark control treatments of the pure and TiO2–CDA were incubated in a flow-through seawater mesocosm for up to 13 weeks. At specific time points throughout the incubation, samples were analyzed for disintegration via time-lapse photography and mass loss measurements and degradation via enzymatic activity relevant to CDA degradation.

Time-lapse photographs indicated that the CDA fabrics and positive control cotton fabric disintegrated in seawater on time scales of months (Figure a). The irradiated CDA fabrics disintegrated beyond the point of handling after 10 weeks, with no observable difference between the pure and TiO2–CDA. The dark control CDA fabrics disintegrated after 11 weeks and also did not exhibit notable differences between pure and TiO2–CDA. The positive control cotton fabric generally disintegrated on the time scale of the irradiated CDA fabrics (within 10 weeks), while the negative control PET fabric did not visibly disintegrate during the 13-week study period.

Figure 4

(a) Photographic evidence of disintegration at select time points for (top to bottom) the dark (black) and irradiated (orange) TiO2–CDA, the dark (black) and irradiated (orange) pure CDA, the positive control cotton fabric (green), and the negative control PET fabric (purple). The irradiation time for both CDA fabrics was 14 days. (b) Cumulative mass loss over time in the natural seawater mesocosm for the irradiated (orange) and dark (black) CDA fabrics, as well as the positive (green) and negative (purple) control fabrics. Error bars represent ± 1 SD from the mean mass loss (n = 4). (c) Cellulase activities of biofilm communities on the irradiated and dark CDA fabrics. Error bars represent ± 1 SD from the mean.

Mass loss measurements revealed substantial differences in disintegration rates between the light-exposed and dark control fabrics, as well as between the light-exposed pure and TiO2–CDA (Figure b). While the dark control fabrics disintegrated beyond the point of handling only a week after the light-exposed fabrics, there were substantial differences in mass loss prior to complete disintegration. At week 3, the dark control pure and TiO2–CDA had lost 7 ± 3 and 9 ± 3% of their total mass, respectively. The irradiated pure and TiO2–CDA had lost slightly more, at 11 ± 4 and 15 ± 3% of their total mass, respectively. However, large differences emerged by week 10: the dark control pure and TiO2–CDA had only lost 40 ± 4 and 33 ± 4%, respectively. In contrast, the irradiated pure and TiO2–CDA were nearly twice as fast (74 ± 5 and 88 ± 3%, respectively), with a notably greater mass loss occurring for the TiO2–CDA. At 10 weeks—the final time point before the fabrics had disintegrated beyond handling—the TiO2–CDA had lost 19% more mass than the pure CDA (p = 0.0142, unpaired Student’s t-test). The positive control cotton fabric disintegrated by week 10, whereas disintegration of the negative control PET fabric was insignificant.

Measurements of esterase activity showed differences between light-exposed and dark control samples for the pure CDA, but not the TiO2–CDA (Figure S7). Esterase activity was higher for the light-exposed pure CDA than for the dark control, and the difference increased over time. However, esterase activities remained comparable in the light-exposed and dark control TiO2–CDA for the duration of the experiment. At all time points, the negative control PET fabric had substantially lower esterase activity than the CDA and positive control cotton fabrics.

Cellulase activity also showed notable differences between light-exposed and dark control samples (Figure c). Breakdown of the cellulose backbone, as indicated by cellulase activity, began within 3 weeks of incubation in seawater. Cellulase activity was consistently lower (up to 2-fold) in the irradiated fabrics relative to the dark controls for the duration of the experiment. However, the magnitude of this difference declined substantially by 10 weeks for the TiO2–CDA but continued to increase for the pure CDA. The positive control cotton fabric had the highest cellulase activity throughout the experiment, while the negative control PET fabric had the lowest.

Discussion

Governing Pathways of Photomineralization and Photo-Biodegradation of CDA Fabrics

Using multiple lines of evidence, we demonstrate that sunlight degrades CDA fabrics via chain scission of the cellulose backbone, resulting in lower-molecular-weight products and CO2. Prior studies have suggested that both chain scission and deacetylation are primary photodegradation pathways for CDA.[6,18−23] Our findings demonstrate that the equivalent of weeks of natural sunlight exposure decreases the average molecular weight of the pure CDA fabric by 17% (Figure ), consistent with chain scission reactions. Furthermore, the radiocarbon signature of DIC produced during photodegradation disproportionately reflected that of the cellulose backbone (Figure ). This indicates that chain scission reactions resulted in preferential photomineralization of cellulosic C relative to acetyl C, in contrast to microbial degradation alone, for which cellulosic and acetyl C are proportionally respired.[8] Despite rapid and notable changes in molecular weight, sunlight exposure did not alter DoS, indicating minimal deacetylation (Figure S6). We suspect that the discrepancy between previous reports of photochemical deacetylation[19,21,22] and the minimal deacetylation observed in this study is due to the screening of UVC wavebands that are not present in sunlight at the sea surface. Nevertheless, our findings indicate that under environmental conditions, chain scission reactions drive CDA photochemistry, yielding low-molecular-weight products and, ultimately, CO2.

Photochemical chain scission of CDA facilitated degradation by marine microbial communities. After sunlight exposure, CDA fabrics lost mass nearly twice as fast (i.e., ∼40% for the dark control vs ∼75% for the light-exposed CDA after 10 weeks; Figure b). Concurrently, cellulase activity was diminished on photodegraded CDA (Figure c). These findings imply that sunlight carries out the steps typically facilitated by cellulase, thereby increasing the polymers’ lability to native marine microbes. An analogous synergy between photo- and biodegradation has previously been observed for dissolved organic matter in Arctic surface waters, where sunlight exposure attenuated the expression of enzymes designed for aromatic carbon degradation, oxygenation, and decarboxylation.[16,39] Prior studies found that exposure of CDA to UV light facilitated biodegradation but proposed different mechanisms. One study suggested that deacetylation reactions stimulate enzymatic degradation by cellulase,[22] whereas another concluded that chain scission of the cellulose backbone stimulated degradation by bacteria isolated from landfill leachate.[21] Our findings support the latter hypothesis for photo-biodegradation of CDA in the surface ocean: photochemical chain scission of the polymer backbone increases the lability of the CDA, leading to faster biodegradation.

Influence of Formulation on CDA Fabric Degradation

The findings of this study strongly support the hypothesis[24] that plastic formulation is a primary control of its environmental fate. Specifically, we found that the addition of ∼1% TiO2 influenced the fate of CDA in four ways. First, TiO2 increased the rate of molecular weight reduction during sunlight exposure by a factor of two (Figure ). Second, it accelerated photochemical mineralization to CO2 by over an order of magnitude relative to pure CDA (Figure ). Third, continued mineralization of the CDA under anoxic conditions (Figure ) demonstrated that TiO2 altered the underlying oxidation pathway: typically, photooxidation of organic molecules in surface waters requires dissolved O2,[40−42] but mineralization of the TiO2–CDA dropped by only 50% in the absence of dissolved O2. This is consistent with the typical pathway for TiO2-catalyzed oxidation, which relies on hydroxyl radicals generated from water, in addition to oxidants derived from dissolved O2.[29] Fourth, TiO2 slightly expedited biodegradation of the irradiated CDA. After 10 weeks in the natural seawater mesocosm, the irradiated TiO2–CDA lost 19% more mass than the irradiated pure CDA, while the dark controls were not significantly different (Figure b). However, the overall responses of the microbial communities to the light-exposed fabrics were similar. This similarity is likely due to the convergence of fabric molecular weights after 14 days of irradiation prior to incubation in the seawater mesocosm (Figure ). Others have observed that exposure of TiO2-containing CDA to UV light accelerated photodegradation[6,22] and subsequent enzymatic and/or biodegradation;[22] this work demonstrates that these processes are relevant under marine conditions.

While these findings further highlight the potential for photocatalytic additives to reduce plastic persistence in the environment, these benefits must be balanced against potential risks associated with the additives. In this study, the addition of TiO2 substantially accelerated CDA photomineralization by up to ∼40-fold and enhanced subsequent biodegradation by up to 2-fold. In our previous research, the presence of TiO2 in single-use consumer polyethylene shopping bags was associated with enhanced rates of photochemical DOC production (up to 2-fold) and altered photoproduct compositions.[24] Other studies have observed that trace TiO2 substantially accelerated the weight loss of several different polymers during photodegradation.[6,27,43,44] Thus, TiO2 may be valuable in facilitating degradation of more recalcitrant plastics (e.g., due to polymer type, high surface area-to-volume ratio) or enabling degradation in environments with lower microbial activity (e.g., cold and/or nutrient-limited waters). However, this reduction in environmental persistence must be viewed alongside the environmental impacts of TiO2, which may be liberated as the polymer matrix degrades.[45] While natural TiO2 is abundant in the environment, plastics may transport and release it to the open ocean, where it is relatively scarce.[46−49] Moreover, engineered nanoparticles commonly behave differently from naturally occurring ones due to coatings[50] and differences in morphology, size, and other characteristics.[51] As a result, they may have longer residence times in surface waters.[52−54] This enhances their potential to interact with and cross biological membranes and induce (photo)toxic effects.[55−57] These concerns may be addressed by adjusting the amount and type[57] of added TiO2 to reach an acceptable risk level or using photocatalytic additives known to pose little environmental risk, although further study is necessary to identify such additives. Overall, photocatalytic additives offer an appealing method for reducing the persistence of CDA—and potentially other polymers, for which biodegradation is typically less influential—but these benefits must be balanced with potential impacts across the additive’s lifetime, from production to environmental release.

Implications for the Environmental Fates of CDA Fabrics and Other Plastics

Shifting plastic use toward high-utility, low-persistence materials and products is one promising short-term strategy for alleviating the impacts of plastic pollution, but efforts to identify and characterize such materials are in the early stages. We recently reported that several CDA-based materials were degraded by marine microbes on time scales of months, suggesting that CDA may possess the characteristics desired for some classes of next-generation materials.[8] Our current findings suggest that prior sunlight exposure accelerates biodegradation, further supporting our previous conclusion that CDA is a high-utility, low-persistence material. However, we emphasize the need for additional research to assess how these findings about the photo-biodegradation of CDA may apply to different article types and characteristics (e.g., thicknesses, morphologies, formulations) under varying environmental conditions (e.g., temperature, nutrients, irradiance levels, salinity, sunlight screening by biofilms[58]).

Characterizing and, ultimately, mitigating the impacts of plastic pollution requires a robust understanding of the complex interactions between environmental fate processes and their implications for plastic lifetimes.[2] Both microbial[8,59,60] and photochemical[9−14] degradation have been studied separately in the context of marine plastics, but their combined effects are considered less often. For example, despite longstanding knowledge that sunlight exposure can alter the biodegradation of plastics in soil systems,[61,62] only two studies to date focused on the photo-biodegradation of plastics in the ocean, with mixed results.[11,17] Our findings add data for two CDA-based formulations and lend further support to the hypothesis[11] that sunlight and microbes can work synergistically to degrade plastics in the ocean. The process underlying this synergy—chain scission of the polymer backbone by sunlight—has been observed for many polymers, including those considered more “recalcitrant” to biodegradation (e.g., polyethylene, polypropylene).[12] Consequently, by studying environmental weathering processes in isolation, as is often recommended by governing bodies that certify plastics as marine biodegradable,[63] the research community may be systematically overestimating the environmental lifetimes of plastic materials and products. Considering the interplay between plastic weathering processes—photochemical, microbial, mechanical, thermal, and more—is thus essential to evaluate the environmental persistence of current and next-generation materials.