Role of elemental carbon in the photochemical aging of soot

Significance

Photochemical oxidation is an important aging pathway for soot. The organic carbon (OC) component in soot is believed to be photoactive, while the elemental carbon (EC) part with strong light absorption is photochemically inert. By considering the distinct light absorption properties of OC and EC, we have provided direct experimental evidence that EC also plays an important role in the photochemical aging of soot by absorbing the solar light. Our work reveals that the photochemical aging of soot occurs in an extended active spectrum (up to red light) by the light absorption of EC and has major environmental effects, such as enhancing the hydrophilicity of soot, that would enhance its ability to act as cloud condensation nuclei.

Abstract

Soot, which consists of organic carbon (OC) and elemental carbon (EC), is a significant component of the total aerosol mass in the atmosphere. Photochemical oxidation is an important aging pathway for soot. It is commonly believed that OC is photoactive but EC, albeit its strong light absorption, is photochemically inert. Here, by taking advantage of the different light absorption properties of OC and EC, we provide direct experimental evidence that EC also plays an important role in the photochemical aging of soot by initiating the oxidation of OC, even under red light irradiation. We show that nascent soot, in addition to undergoing photochemical oxidation under blue light with a wavelength of 440 nm, undergoes similar oxidation under red light irradiation of λ = 648 nm (L₆₄₈). However, separated OC (extracted from soot by n-hexane) and EC exhibit little reactivity under L₆₄₈. These observations indicate that EC plays a pivotal role in photoaging of soot by adsorbing light to initiate the oxidation of OC. Comparison of in situ IR spectra and photoelectrochemical behaviors suggests that EC-initiated photooxidation of OC proceeds through an electron transfer pathway, which is distinct from the photoaging induced by light absorption of OC. Since the absorption spectra of EC have a much larger overlap with the solar spectra than those of OC, our results provide insight into the chemical mechanism leading to rapid soot aging by organic species observed from atmospheric field measurements.

Soot particles produced by the incomplete combustion of biomass and fossil fuels are emitted in large quantities to the atmosphere (1, 2). These particles have important impacts on global radiative balance and climate, directly by absorbing solar energy and indirectly by acting as cloud condensation nuclei (2–4). In the atmosphere, soot particles undergo transformations in their structure, hygroscopicity, and optical properties by interacting with other atmospheric chemical constituents, including reactive inorganic and organic gases (5–11). Such aging significantly impacts their atmospheric fate, lifetimes, and effects. For example, the light absorption and direct radiative forcing of soot are markedly enhanced during atmospheric aging (12). Under dark conditions, aging usually rapidly ceases because of the depletion of reactive sites (13–19). However, aging of the soot becomes time-independent upon exposure to light. For example, persistence of NO₂ uptake on soot extends from 1 to 2 min in the dark to almost 70 h under illumination, which is comparable to the lifetime of soot in the atmosphere (20). In addition, irradiation was reported to enable the oxidation of soot by O₂ in the air. The apparent rate constants for loss or formation of species on soot in O₂ under sunlight irradiation were larger by factors of 1.5–3.5 than those in 100 parts per billion O₃ (21). Therefore, photochemical aging of soot has recently attracted increasing attention in the field of atmospheric chemistry (22, 23).

Soot particles are primarily composed of elemental carbon (EC) and organic carbon (OC). EC has a graphite-like microcrystalline structure and is refractory and strongly light-absorptive. The OC on soot is primarily composed of saturated and unsaturated hydrocarbons, polycyclic aromatic hydrocarbons (PAHs), and partially oxidized organics that condense onto soot particles during their emission (21, 24, 25). In the ambient atmosphere, the soot would adsorb other OC species from atmosphere; consequently, plenty of proteinaceous substances, lipopolysaccharidic substances, lignans, cellulose, pollens, bacteria, and humic-like substances can also be included in the OC component (26). Recent studies found that enhancement of the photoaging of soot by sunlight is caused by the promoted oxidation of the extractable OC, especially PAHs (21, 23), which has been proposed to be initiated by the excitation of OC. Such a mechanism is analogous to the self-sensitized photodegradation of PAHs on inert supports such as silica (27–29), as supported by the similar photochemical reactions of OC on soot and the OC extracted by n-hexane, as well as by the inactiveness of EC under irradiation (21, 23).

The light absorption ability of EC is much greater than that of OC, particularly for long-wavelength radiation (24, 25, 30). The strong light absorption of EC was predicted to inhibit the photolysis of PAHs on soot by screening the incoming light, as the photolytic half-lives for PAHs on silica and alumina are much shorter than those of PAHs on carbon black (29, 31, 32). On the other hand, carbon-based materials (e.g., carbon dot) were recently found to exhibit promising photocatalytic activity for the oxidation and reduction of organic species under visible irradiation (33–36). Whether the strong visible light absorption of EC in soot, which has graphite-like microcrystalline structures similar to those of carbon dots, can induce photoreactions is of great environmental significance, since visible light is very abundant in the incident solar light near the Earth’s surface.

In the present work_, we show that irradiation by red light at a wavelength of 648 nm, which cannot excite OC, induces efficient photochemical aging of soot by O₂. By comparing in situ IR spectra taken during the photochemical oxidation of soot and OC under blue light with 440-nm irradiation (L₄₄₀) and under red light with 648-nm irradiation (L₆₄₈), we reveal that the visible light absorption of EC in air initiates the oxidation of OC in soot. The mechanism underlying the photochemical activity of EC is further examined by the photoelectrochemical method and by electron paramagnetic resonance (EPR) spectroscopy, and an electron transfer pathway is identified to explain the photosensitizing activity of EC.

Results and Discussion

The n-hexane soot, which has been widely used to model the adsorption and reaction properties of soot in atmospheric aerosols (21, 23, 37–40), was produced by combustion of n-hexane in a coflow diffusion burner (21, 23). Consistent with the earlier studies (21, 24, 25), measurement using diffuse reflectance UV-visible (UV-vis) spectroscopy indicates that the nascent soot absorbs radiation over a broad wavelength range, from 200 to 800 nm (Fig. 1, black solid line). Soot particles are primarily composed of EC and variable fractions of OC, and the OC was extracted by solvent (21, 23–25). To distinguish the absorption characteristics of OC and EC, we abstracted the soot with n-hexane to separate the extractable OC from EC, during which almost all of the EC was left on the support (SI Appendix, Fig. S1). The absorption spectrum measurements indicate that the absorption characteristics of the residual EC (Fig. 1, red solid line) are similar to those of the nascent soot. However, the extracted OC absorbs light at wavelengths ranging from 200–600 nm but is relatively transparent at 600–800 nm (Fig. 1, blue solid line). The different light absorption properties of OC and EC provide an excellent opportunity to distinguish the roles of these two components in the photochemical aging of soot.

Fig. 1.

Normalized diffuse reflectance UV-vis spectroscopy of nascent soot (black solid line), OC extracted by hexane (blue solid line), residual EC after extraction (red solid line), and the solar spectrum air mass 1.5 (AM1.5) in the range of 200–800 nm (pink dashed line). The data on the solar spectrum are from the International Electrotechnical Commission (59).

Light at the wavelengths of 440 nm, which is absorbed by both components of soot, and 648 nm, which excites only the EC moiety, was employed to excite specific parts of soot. The in situ attenuated total internal reflection IR (ATR-IR) method was used to monitor the photochemical reactions of soot. As shown in SI Appendix, Fig. S2, after irradiation by L₄₄₀ (blue line, denoted Soot₄₄₀) and L₆₄₈ (red line, denoted Soot₆₄₈), the intensities of the IR bands at 3,286 cm⁻¹ and 3,038 cm⁻¹, which are assigned to the stretching vibrations of alkyne C-H (≡C-H) and aromatic C-H (Ar-H), respectively (41–44), significantly decrease, indicative of the loss of the C-H species under irradiation with either wavelength. At the same time, bands in the range from 1,800–1,500 cm⁻¹ prominently increase. The peaks in this region relate to the stretching vibrations of carbonyl C = O species (41, 42, 45, 46). The increase in the intensities of these peaks indicates the formation of carbonyl-containing species by photoinduced reactions of soot. No reactions are observed without irradiation or in the absence of O₂ (in Ar atmosphere). These results suggest that the spectral changes under irradiation are caused by heterogeneous photochemical reactions of soot with O₂.

A closer inspection of the IR spectra, which bear helpful information about the photoaging process of soot, reveals a complex profile for the distribution of the formed carbonyl-containing species. Fig. 2 shows the temporal changes of the IR spectra in the range from 1,800–1,500 cm⁻¹ for soot samples during photooxidation by O₂. Specifically, the peaks at 1,590 cm⁻¹ are related to chelated carbonyl species or aromatic stretch-enhanced by carbonyls conjugated to the aromatic structure (43). The bands at 1,685 cm⁻¹ originate from the C = O stretch of unsaturated ketones/aldehydes, and the peak at 1,716 cm⁻¹ is from C = O groups in saturated ketones/aldehydes (43, 45). The peak at ∼1,762 cm⁻¹ is assigned to the C = O stretch of lactone or anhydride species (43, 45). All these carbonyl-containing species are observed by irradiating nascent soot with both L₄₄₀ (Fig. 2A) and L₆₄₈ (Fig. 2B). For the extracted OC, C = O species are obviously formed upon L₄₄₀ (Fig. 2C). However, extracted OC does not react under L₆₄₈ (Fig. 2D) because of its lack of absorption at this wavelength (Fig. 1). Residual EC exhibits little photoreactivity under L₆₄₈ either (Fig. 2E). Interestingly, when we reloaded the extracted OC onto the EC, the photooxidation of OC under L₆₄₈ is restored (Fig. 2F). Since the separated OC and EC do not exhibit any photoreaction activity under L₆₄₈, the photoactivity of soot that consists of OC and EC under this irradiation (Fig. 2 B and F) is somewhat unexpected. These results clearly indicate the pivotal role of EC in the photochemical oxidation of OC in the soot.

Fig. 2.

Temporal changes in ATR-IR spectra in the range from 1,800 to 1,500 cm⁻¹ for nascent soot under L₄₄₀ (A), nascent soot under L₆₄₈ (B), extracted OC under L₄₄₀ (C), and extracted OC under L₆₄₈ (D). Residual EC under L₆₄₈ (E) and OC reloaded on EC under L₆₄₈ (F) are shown. The IR spectra were collected by using corresponding samples untreated by radiation on ZnSe crystals as references. The black lines represent the IR spectra of samples before irradiation. The other lines represent the IR spectra of samples undergoing different times of irradiation, and the time interval of each line is 1 h.

Although the irradiation leads to a general increase in C = O species content under both L₄₄₀ (Fig. 2A) and L₆₄₈ (Fig. 2B), the relative intensities of different C = O species are notably different in samples treated with these two wavelengths. For example, the relative intensity of the unsaturated ketone/aldehyde (at 1,685 cm⁻¹) in Soot₆₄₈ is more significant than that in Soot₄₄₀. To better quantify this intensity difference, we compared the ratio in intensity of the different C = O peaks under different irradiation. SI Appendix, Fig. S3 shows that the ratio of the unsaturated (at 1,685 cm⁻¹)-to-saturated (at 1,716 cm⁻¹) ketone/aldehyde band intensities (I_unsat/I_sat) for Soot₆₄₈ is much higher than that for Soot₄₄₀. However, the I_unsat/I_sat value of extracted OC that received L₄₄₀ (OC₄₄₀) is lower than that of Soot₄₄₀. These results indicate that a greater number of unsaturated ketones/aldehydes are produced under L₆₄₈, while the formation of saturated ketones/aldehydes is more significant under L₄₄₀ (particularly in OC₄₄₀).

The different intermediate distribution suggests that different reaction mechanisms are responsible for the photooxidation of OC under L₄₄₀ and L₆₄₈. Two mechanisms are usually invoked to explain photochemical oxidation reactions of organic compounds: an electron transfer pathway and an energy transfer pathway (47). In the electron transfer pathway, the radical reactions, which are initiated by the photoinduced electron transfer, dominate the oxidation, as in photocatalytic degradation of organic pollutants on TiO₂ (48). The oxidative cleavage of aromatic rings by radical reactions would lead to the formation of unsaturated ketones/aldehydes, such as unsaturated muconaldehyde (SI Appendix, Scheme S1A) and 2-formylcinnamaldehyde (49–53) (SI Appendix, Scheme S1B). The enrichment of unsaturated ketones/aldehydes in the photoproducts of soot under L₆₄₈ suggests that the oxidation of OC is dominated by radical reactions and initiated by the electron transfer pathway under these conditions. Because of the lack of absorption of OC itself under L₆₄₈, the reaction should be induced by the excitation of EC. Excited EC abstracts electrons from the OC and donates electrons to adsorbed O₂. As a result, an OC radical cation and reactive oxygen species, such as superoxide anion radicals (O₂^•−) and hydroxyl radicals (^•OH), would be formed. This pathway is further verified by the detection of radicals and photoelectrochemical measurements as discussed below.

The excitation of OC, especially PAHs, forms singlet oxygen (¹O₂) by energy transfer (21, 29). Addition reactions between ¹O₂ and unsaturated and aromatic compounds, such as olefins and PAHs, are very rapid. Compared with the radical reactions, these reactions are quite mild and result in the formation of dioxetanes. Further cleavage of unstable dioxetanes produces a substitution-dependent range of ketones/aldehydes (SI Appendix, Scheme S1C). If the substituent groups in the olefins are saturated alkyls, the formed ketones/aldehydes are saturated (54). The energy transfer pathway therefore produces a significant portion of saturated ketones/aldehydes. Accordingly, the significance of saturated ketones/aldehydes in OC oxidation under L₄₄₀ (particularly for OC; Fig. 2C) is an indication of the important role of the energy transfer pathway in photooxidation of OC by exciting OC itself. To confirm the prevalence of the energy transfer pathway in these reactions, we examined the formation of ¹O₂ by using furfuryl alcohol (FFA) as a probe (55, 56). Significant amounts of ¹O₂ are observed in both the soot and OC systems under L₄₄₀, as shown by the rapid degradation of FFA (SI Appendix, Fig. S4). By contrast, the lack of FFA loss under L₆₄₈ indicates that little ¹O₂ is generated under these conditions. These results support the above conclusion that the photochemical oxidation of OC under L₄₄₀ is induced dominantly by exciting OC itself and the following ¹O₂ oxidation. Both the photooxidation of OC itself (energy transfer pathway) and the oxidation of OC induced by EC (electron transfer pathway) should contribute to the photoaging of the nascent soot under L₄₄₀, since the significance of saturated ketones/aldehydes in Soot₄₄₀ falls between that of Soot₆₄₈ and OC₄₄₀. The key processes in the EC-initiated photooxidation of OC and the direct photolysis of OC are summarized in Fig. 3.

Fig. 3.

Reaction scheme for the EC-initiated photooxidation of OC and the direct photolysis of OC.

By loading the sample of interest (soot, EC, and OC) on fluorine-doped tin oxide (FTO) as a working electrode, the light-induced events were further probed by open circuit voltage (V_o) measurements, which allow one to probe a photopotential caused by steady-state accumulation of a photogenerated hole/electron. Fig. 4 shows the typical V_o profile in air-equilibrated electrolytes. Upon L₆₄₈ and L₄₄₀, both the soot and EC electrodes show significant positive V_o (Fig. 4), indicating the accumulation of a positive charge (photoinduced holes) on the electrodes due to efficient charge separation by the capture of photoinduced electrons by dissolved O₂. The V_o decay after irradiation represents charge recombination and/or transfer to available acceptors (OH⁻) in the electrolyte. The V_o of soot during irradiation shows a slight increase probably owing to the photochemical reaction, as shown in Fig. 2, while this is not observed for EC, consistent with the lack of photochemical conversion of EC (despite its efficient charge separation). In contrast to soot and EC, OC electrodes do not show responses to L₆₄₈ (Fig. 4A, black line). Even under 440-nm irradiation, under which condition the photochemical reaction occurs, the V_o of the OC electrode is quite small at the beginning stage of irradiation, indicative of the poor charge separation feature of OC. The gradual V_o increase of OC with irradiation time could be attributed to the enhanced charge separation ability of the photooxidation product of OC, as in the case of irradiated soot. Moreover, it is interesting to note that the V_o of soot is higher than that of EC under L₆₄₈ (Fig. 4A), suggesting that more holes are accumulated in the soot electrodes due to the presence of OC. Since L₆₄₈ cannot excite OC in soot, the higher V_o of soot should not be attributed to the light absorption of OC. Reasonably, because of its electron-donating properties, the OC in soot traps the hole generated by excitation of the EC moiety, which would enhance the accumulation of holes in the soot, and consequently increase the V_o. In other words, the higher V_o of the soot electrode suggests that the photoinduced hole in the EC moiety transfers to the OC part, which leads to the photooxidation of OC.

Fig. 4.

V_o responses of soot (blue line), EC (red line), and OC (black line) under L₆₄₈ (A) and L₄₄₀ (B) in air-equilibrated 1 M KCl solutions.

In Ar-saturated electrolytes (SI Appendix, Fig. S5), the V_o generally changes according to two trends: (i) the V_o of electrodes in air-equilibrated electrolytes is higher than that of electrodes in Ar-saturated electrolytes, which implies that O₂ can trap the photogenerated electron and thus enhances the accumulation of a positive charge on the electrodes, and (ii) the increase in the V_o with irradiation time is largely eliminated in Ar-saturated electrolytes, indicative of depression of the photochemical reaction in the absence of O₂. These observations suggest that the O₂ can trap the photogenerated electron.

The short-circuit photocurrents, which reflect the type and rate of the interfacial photochemical redox reactions, of the soot electrodes under different conditions are shown in SI Appendix, Fig. S6. Pronounced negative photocurrents are observed for the soot electrode under L₆₄₈ and L₄₄₀ in the air and Ar-saturated electrolytes, indicating that the soot electrode abstracts electrons from the external circuit, which is in agreement with the positive V_o. The photocurrent of the OC electrode, compared with the nascent soot, is quite low, even under L₄₄₀. The small short-circuit photocurrent of extracted OC confirms that OC oxidation under L₄₄₀ mainly occurs via the energy transfer pathway. Under identical levels (20 mW⋅cm⁻²) of L₆₄₈ and L₄₄₀, the short-circuit photocurrent of EC is much larger than that of soot in air-saturated electrolytes (Fig. 5), demonstrating the excellent ability of EC to realize photoinduced charge separation. The lower photocurrent of soot suggests that the presence of OC in soot can inhibit the photocurrent. This inhibition might be caused by the consumption of photoinduced holes by OC, since OC can donate electrons to the exited EC and become oxidized, which will competitively depress hole transfer from the excited EC to the FTO electrode.

Fig. 5.

Short-circuit photocurrent responses of soot (black line) and EC (red line) in air-equilibrated electrolytes and soot (blue line) and EC (pink line) in Ar-saturated electrolytes under L₆₄₈ (A) and L₄₄₀ (B).

Fig. 5 shows that the short-circuit photocurrents in air-equilibrated electrolytes are much larger than those in Ar-saturated electrolytes, particularly for the EC electrode, suggesting that dissolved O₂ can enhance the photocurrent. Moreover, changes in the transient photocurrent of the EC electrode after the light is turned on largely depend on the atmosphere. In the presence of O₂, the change in the transient photocurrent is insignificant, whereas it rapidly decays in the absence of O₂. Such a photocurrent decay is attributed to the recombination of photogenerated charges on the surface of the electrode. The electron is thus removed from the electrode by reacting with the O₂, inhibiting the recombination of photogenerated charges and resulting in a more sustainable photocurrent. However, the accumulation of electrons on the electrode surface upon irradiation in the absence of O₂ leads to faster charge recombination, which decreases the photocurrent. As proposed from the product analysis and V_o measurements, the reaction of O₂ with the photoinduced electrons on EC would generate reactive oxygen species such as O₂^•− and ^•OH, which play an important role in the aging of soot.

Environmental Implication

Photooxidation of OC on soot largely determines the environmental effects and fate of soot. EC-initiated photooxidation of OC on soot occurs even under red light irradiation. Such an extension of active spectra caused by EC has important effects on the photoaging of soot since red light is abundant in sunlight (Fig. 1), particularly during a small solar zenith angle or cloudy weather. Under these situations, EC-initiated photooxidation likely dominates the photoaging of soot. Moreover, the photoactivity of EC may have important implications for the persistent uptake of environmental species, such as NO₂, by soot under irradiation (20) since the light absorption and photoactivity of EC should persist even after labile OC sites are consumed.

We show that the photooxidation of soot initiated by light absorption of EC transforms the olefins and PAHs in OC into various carbonyl C = O species. These oxygen-containing species are more polar than the original olefins and PAHs. As a result, the hydrophilicity of soot can be largely increased. To verify this change, the hydrophilicity of soot after irradiation was examined by ATR-IR spectroscopy. The dashed lines in Fig. 6 represent the IR spectra for samples equilibrated with a stream of water-saturated Ar before irradiation; the IR bands at 3,415 and 1,623 cm⁻¹ are attributed to the stretching vibration of O-H and the H-O-H bending vibration of adsorbed water molecules. Thus, the absorbance intensities of IR bands at 3,415 cm⁻¹ represent the amount of adsorbed water, which is related to the water affinity of soot. After the samples were irradiated with either L₆₄₈ (Fig. 6, red solid line) or L₄₄₀ (Fig. 6, blue solid line), the absorbance intensities of these adsorbed water peaks significantly increase, indicating the enhanced hydrophilicity of soot after irradiation. By contrast, the water affinity of soot is barely changed after 12 h in the dark (Fig. 6, black solid line). The increased hydrophilicity of soot would enhance its ability to act as cloud condensation nuclei or ice nuclei (5, 57).

Fig. 6.

ATR-IR spectra for soot equilibrated with a stream of water-saturated Ar before (dashed line) and after (solid line) irradiation with 648 nm and 440 nm and in the dark. The absorbance intensities of IR bands at 3,415 cm⁻¹ represent the amount of adsorbed water, which is related to the water affinity of soot.

Our experimental results also show that the EC-initiated oxidation proceeds mainly via an electron transfer pathway in which many radical species can be formed. Reactive oxygen radicals are important environmental species that can cause diverse reactions in addition to oxidizing OC. To show the formation of reactive oxygen radicals during the photochemical oxidation of soot, we examined radical formation by the spin-trapping EPR technique, using 5-tert-butoxycarbonyl-5-methyl-1-pyrroline-N-oxide (BMPO) as a spin-trapping agent (58). Fig. 7 shows that significant amounts of BMPO-OH adducts are detected in the suspensions of soot and EC under both L₆₄₈ and L₄₄₀, indicative of the formation of ^•OH. When these systems are purged with Ar to remove O₂, no BMPO-OH adduct signals are observed any longer (SI Appendix, Fig. S7H), indicating that the generation of ^•OH is closely related to the presence of O₂. However, we did not detect the superoxide radical with BMPO, probably because of the rapid disproportionation of superoxide in aqueous solution.

Fig. 7.

EPR signals for typical BMPO-OH adducts. Soot under L₆₄₈ (green line), soot under L₄₄₀ (pink line), EC under L₆₄₈ (blue line), EC under L₄₄₀ (red line), and samples in the dark (black line) are shown.

Conclusions

We show that, in addition to the direct photolysis of OC, the absorption of sunlight by the unreactive EC initiates efficient photooxidation of surface OC. This process occurs not only with blue light (λ = 440 nm), which photolyzes OC directly as well, but also with red light (λ = 648 nm). Unlike the photochemical oxidation of OC itself, which is dominated by an energy transfer-based ¹O₂ mechanism, EC-initiated oxidation of OC proceeds through an electron transfer pathway, where reactive radical species such as ^•OH are formed. The identification of EC-induced photoaging of soot may have important environmental and atmospheric implications.

Materials and Methods

Soot Production.

Soot was produced by the combustion of n-hexane (HPLC grade) in a coflow diffusion burner, as described by Han et al. (21, 23). The burner consisted of a diffusion flame maintained by an airflow that could be exactly controlled using mass flowmeters. The airflow was a mixture of high-purity O₂ and N₂, and the O₂ content was 21.3%. The fuel/oxygen ratio was 0.18. Fuel transport was achieved using a cotton wick extending into the liquid fuel reservoir. Soot particles from the diffusion flame were directly deposited on supports, such as quartz plates, ZnSe crystals, FTO, and quartz fiber filters, for further reactions and characterization by several methods, including UV-vis absorption (U-3900; Hitachi), in situ ATR-IR spectroscopy, photoelectrochemical measurements, and EPR.

Preparation of Residual EC and Extracted OC Samples.

Residual EC was obtained by soaking the soot samples (deposited on quartz plates, ZnSe crystals, FTO, or quartz fiber filters) four times in 50 mL of n-hexane for 10 min, and the n-hexane on the residue was subsequently evaporated in air. Care was taken not to disturb the sample to minimize mechanical removal of insoluble particles. Most OC, especially PAHs, can be removed from the soot particles, and the residue is mainly EC.

The n-hexane extractions were first concentrated and then added dropwise to quartz plates, ZnSe crystals, FTO, or quartz fiber filters, as they were used for soot collection. After the solvent was evaporated in air, the extracted OC was left on the supports. To reload the extracted OC onto the EC, the n-hexane extractions were added dropwise to ZnSe crystals with EC. After evaporating n-hexane off in the air, the photoaging was examined by ATR-IR spectroscopy. An inspection using ATR-IR spectroscopy indicates that such a reloading process can recover the OC onto the EC (SI Appendix, Fig. S8).

Aging Experiments.

The in situ ATR-IR spectra were recorded using an IS50 FTIR spectrometer, which is equipped with a high-sensitivity mercury-cadmium-telluride detector cooled by liquid N₂. A Lumencor solid-state light engine with switchable wavelength control and tunable intensity was used as the light source throughout the whole study. Soot EC or OC samples on the ZnSe crystals were put into the ATR-IR cell. The ATR-IR cell was sealed with quartz glass, through which samples were irradiated by light with wavelengths of 440 ± 25 nm and 648 ± 25 nm. The light intensities were adjusted to 20 mW⋅cm⁻². Before the reactions, the cell was purged with 20 mL⋅min⁻¹ Ar until the IR spectrum was constant. A mixture of high-purity O₂ (4 mL⋅min⁻¹) and Ar (16 mL⋅min⁻¹) was then introduced into the ATR-IR cell. The spectra of samples were recorded (100 scans, 4-cm⁻¹ resolution) using the ZnSe or ZnSe loaded with a sample as a reference.

Photoelectrochemical Analysis.

V_o values and short-circuit photocurrents were examined in a conventional three-electrode electrochemical cell with a sample/FTO electrode as the working electrode, a platinum wire as the counterelectrode, and Ag/AgCl (saturated KCl) as the reference electrode in an aqueous solution of 1 M KCl. All of the measurements were carried out on a CHI760e electrochemical workstation. For comparing the photoelectrochemical properties between EC and soot, the EC electrode was prepared by washing the soot/FTO electrode with n-hexane after the photoelectrochemical measurements had been completed on this soot/FTO electrode. Care was taken to not disturb the sample to minimize mechanical removal of EC.

EPR Measurements.

A BMPO (a freshly prepared 0.1 M solution in deionized water) spin trap was used for capturing hydroxyl radicals (^•OH) in aqueous solutions. These samples, including soot, OC, and EC, were loaded on a quartz fiber filter (3 mm × 30 mm) and continuously monitored for free radical signals in an EPR spectrometer (ELEXSYS E500 EPR; Bruker) with a modulation of 100 kHz and a microwave frequency of 9.5 GHz. The typical parameters for EPR measurement were as follows: the sweep width was 100 G, the modulation amplitude was 2.00 G, and the x axis had 1,024 points. The EPR microwave power was set specifically to 13 dB, and the sweep time was 81.92 ms.

Water Affinity Test.

Soot samples were first dehydrated by a pure Ar flow until the IR spectrum was constant, and subsequently flushed by a water-saturated Ar flow (20 mL⋅min⁻¹) for 3 h to ensure that water vapor adsorption to the soot reached equilibrium. The ATR-IR spectra of the water-saturated soot were recorded by using the dehydrated soot surface as the reference background spectrum (100 scans, 4-cm⁻¹ resolution). Soot particles were then flushed with a dry Ar flow (100 mL⋅min⁻¹) for 3 h to remove the water from ATR-IR cell until the spectrum showed no further changes. A mixture of high-purity O₂ and Ar was then introduced into the ATR-IR cell, and the samples were irradiated by light with wavelengths of 440 nm or 648 nm for 12 h. The cell was subsequently flushed with a water-saturated Ar flow again, and the IR spectra were recorded. The control experiment was performed using the same procedure in dark.