Environ. Sci. Technol. 43, 9155–9160
Light-Initiated Transformations of Fullerenol in Aqueous Media L I N G J U N K O N G , * ,† O ’ N I E L L T E D R O W , ‡ YAU FONG (KYLE) CHAN,‡ AND R I C H A R D G . Z E P P * ,§ National Research Council Associate, U.S. Environmental Protection Agency, NERL/ERD, Athens Georgia 30605, Student Services, Ecosystems Research Division, U.S. Environmental Protection Agency, Athens Georgia 30605, and U.S. Environmental Protection Agency, Athens Georgia 30605
Received June 22, 2009. Revised manuscript received October 22, 2009. Accepted October 23, 2009.
We provide the first evidence that a fullerene derivative can be extensively mineralized under environmental conditions by direct photolysis. Dissolved inorganic carbon (DIC) was identified as a major photoproduct of fullerenol, a hydroxylated C60 molecule and the ratio of moles DIC produced to moles of fullerenol reacted reached 28 or approximately 47% of complete mineralization on extensive irradiation by simulated solar radiation. The direct photoreaction kinetics of fullerenol in dilute aqueous solution can be described by pH-dependent biexponential rate expressions. This photoreaction slowed by a factor of 2 in nitrogen-saturated water. The oxygen dependence is attributed to photoinduced electron or hydrogen atom transfer from fullerenol to oxygen to produce superoxide ions with a quantum yield of 6.2 × 10-4. Fullerenol can photosensitize the production of singlet oxygen (1O2) in dilute aqueous solution with quantum yields ranging from 0.10 in acidic water to 0.05 in neutral and basic solution. However our results indicate that chemical reactions involving diffusive encounters between 1 O2 or superoxide and fullerenol are too slow to significantly contribute to the fast component of fullerenol photoreaction in sunlight. The pH dependence of the direct and sensitized photoreactions is attributed to changes in intramolecular hemiketal formation in fullerenol.
Introduction Fullerenes and fullerene derivatives have gained attention due to their unique properties, which have applications in optical, electronic, biological, and medicinal industries. Fullerenol (or fullerol), a hydroxylated derivative of Buckminsterfullerene (C60), is significantly more hydrophilic than C60. Depending on the number of hydroxyl groups, which varies from 12 to 40, the aqueous solubility may achieve 58.9 mg mL-1 (1), nearly 10 orders of magnitude greater than the aqueous solubility of C60 (2). Although most research on fullerenol has been conducted with its potential biological and medical applications in mind, fullerenol also has been the focus of environmental studies (3-7). Published cytotoxicity and phototoxicity data of fullerenol show a wide variety of responses from living cells or bacteria (3, 8, 9). * Corresponding author phone: (706)355-8117; e-mail:
[email protected]. † U.S. Environmental Protection Agency, NERL/ERD. ‡ Ecosystems Research Division, U.S. Environmental Protection Agency. § U.S. Environmental Protection Agency. 10.1021/es901839q
Not subject to U.S. Copyright. Publ. 2009 Am. Chem. Soc.
Published on Web 11/12/2009
Past studies indicate that fullerenol is reactive under certain conditions in aqueous media. Investigations using laser flash photolysis and pulse radiolysis have revealed that fullerenol can efficiently scavenge reactive radical species such as hydroxyl radicals (•OH), superoxide (O2- · ), hydrogen atoms (•H) and hydrated electrons (eaq-) to form fullerenol radical adducts and radical anions (10-13). Light-induced production of 1O2 and O2- · by fullerenol has also been observed (14-16). Fullerenol photosensitized the oxygenation of furfuryl alcohol (FFA) by 1O2 under both ultraviolet (UV) and visible irradiation (5). Fullerenol also forms clusters in aqueous media and cluster formation potentially could influence its reactivity (17). Recent studies have shown that aqueous clusters of C60 undergo direct photolysis in sunlight (18). The clusters photosensitized 1O2 production and it was suggested that their photoreaction was mediated by this transient (19). However, studies of the environmental transformations of fullerenol are sparse (20), and its direct photoreactions have not been reported. The objectives of this research were to provide data that can be used to evaluate the rates and products of fullerenol transformations in aqueous media under conditions that are commonly found in the environment. One key objective was to provide new data regarding fullerenol phototransformation kinetics, including absorption spectra, pH- and wavelengtheffects, and quantum yields for its photodegradation. New evidence is presented that fullerenol can be extensively mineralized by sunlight-initiated reactions. A final objective was to provide data on possible mechanisms for direct photoreaction of fullerenol, including the possible involvement of singlet molecular oxygen and superoxide ions.
Experimental Section Chemicals. Fullerenols used in these experiments, the salt form (C60 (OH)x(ONa)y, with x + y ) 24, y ) 6-10, purity >99%) and the acid-precipitated form (C60(OH)24), were purchased from Materials and Electrochemical Research (MER) Corporation (Tucson, AZ), and used as received. Molecular weights were estimated to be 1300 and 1128 for the salt form and the acid-precipitated form, respectively. Acetic acid, acetonitrile (HPLC grade), hydrochloric acid, sodium hydroxide, sodium nitrate, and sodium phosphate (mono- and di- basic) were purchased from Fisher Scientific. Quinine sulfate · dihydrate was obtained from Fluka. Potassium ferrioxalate was purchased from Alfa Aesar Products. Rose bengal and methylene blue were obtained from ACROS Organics. The source of acridine orange, ammonium acetate, ferrozine, furfuryl alcohol (FFA), hydrogen peroxide (30%, w/w), p-nitroanisole and superoxide dismutase (SOD, 2680 units mg-1) was from Sigma-Aldrich. Sodium tetraborate was purchased from Brand-Nu Laboratories, Inc. Ammonium chloride was from J. T. Baker. Horseradish peroxidase (HRP, 268 units mg-1) was purchased from Thermo Scientific. Reagent grade p-hydroxyphenylacetic acid from Aldrich was purified by repeated recrystallization from ethanol. Water was purified using a Barnstead Nanopure D-7331 system (g18.0 MΩ cm) was used in the experiments. The fullerenol solutions used in this research were prepared by magnetically stirring the fullerenol salt or acidprecipitated form with Nanopure water over a minimum of a 24 h period. Irradiation Kinetic Procedures. An Atlas SunTest CPS solar simulator equipped with a 1 kW xenon arc lamp was used for simulated solar irradiations, and a rotating turntable reactor (a merry-go-round reactor (MGRR) (21)) was used for monochromatic irradiations. The temperature was mainVOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9155
tained at 22 ( 1 °C using a NESLAB recirculating water bath. Glass filters were used to isolate the 366 nm wavelength from a Hanovia mercury vapor lamp in the MGRR. The MGRR provided 0.00759 einsteins L-1 hr-1 of 366 nm light within the irradiated tubes. Simulated solar irradiation experiments were conducted using completely filled quartz tubes (∼24 mL) sealed with Teflon-lined septum caps that were submersed in a thermostatted water bath. The incident irradiance at the tube surface, summed from 290 to 700 nm, was 0.0065 W cm-2. Monochromatic irradiation experiments were conducted in 8 mL Pyrex glass tubes that each contained 5.00 mL of solution, including ferrioxalate actinometers (see Supporting Information (SI)). All the samples were irradiated in triplicate. Dark controls were contained in corresponding quartz or glass tubes wrapped with aluminum foil, and kept at room temperature (22 ( 1 °C). The reaction quantum yield of fullerenol was calculated as described elsewhere (21) (SI).To investigate the effect of oxygen on fullerenol photolysis, fullerenol solutions were purged with N2 for 1 h, then transferred to 24 mL quartz test tubes inside of an anaerobic glovebox prior to simulated solar irradiation. Photooxygenation kinetic studies of furfuryl alcohol (FFA) in the presence of three separate photosensitizers (fullerenol, rose bengal, and methylene blue) were conducted using monochromatic irradiation (366 nm). Acetic acid, sodium phosphate (monoand di- basic) and sodium tetraborate/sodium hydroxide were used to buffer fullerenol solutions at pH values of 4.6, 7.0, and 9.7, respectively. Hydrogen peroxide was measured by its peroxidase-catalyzed reaction with p-hydroxyphenylacetic acid using a modified fluorometric procedure that was originally described by Lazrus et al. (22). Fullerenol concentrations were measured using a Dionex Ultimate 3000 high performance liquid chromatography (HPLC) system and/or a PerkinElmer Lambda 35 UV/Visible spectrometer. The HPLC was equipped with an Ultimate 3000 photodiode array detector and a RF-10AXL fluorescence detector. A mixture of acetonitrile and degassed Nanopure water was used as the mobile phase with an acetonitrile:water ratio of 40:60. The FFA was also measured using the Dionex HPLC system with a mobile phase of acetonitrile:water (30:70) at a wavelength of 218 nm. Spectroscopic and Particle Size Measurements. Electronic absorption spectra were measured using the PerkinElmer Lambda 35 UV/visible spectrophotometer. Particle size distributions and ξ potentials were determined using a Malvern Zetasizer Nano-ZS. Dissolved Inorganic Carbon (DIC) Measurements. Aqueous solutions of fullerenol (5.6 µM) were irradiated in the solar simulator as described in the previous section. Following irradiation, all tubes were immediately wrapped with aluminum foil and stored at room temperature until the analysis for dissolved inorganic carbon (DIC) was performed. A Shimadzu TOC-5050A was used to measure DIC in the irradiated and nonirradiated fullerenol solutions. Each solution was measured in triplicate. Solutions of sodium bicarbonate and sodium carbonate were used as standards to calibrate the instrument.
Results and Discussion Ultraviolet-Visible Spectra. Ultraviolet (UV)-visible absorption spectra of aqueous fullerenol were measured at various concentrations ranging from 0.66 to 6.65 µM. A linear correlation was observed between aqueous fullerenol concentrations and absorption coefficients (Figure 1, inset) (r2 ) 1.00). Light absorption by fullerenol increases as wavelength decreases (Figure 1), and the fullerenol molar absorption coefficients calculated using the Beer-Lambert law ranged from 1.01 × 104 to 3.40 × 104 L mol-1 cm-1 in the 290-400 nm (UVR) spectral region, consistent with previous studies (8). Comparison of the fullerenol absorption spectrum with 9156
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009
FIGURE 1. Spectroscopic properties of aqueous fullerenol and C60 in toluene. UV-visible absorption spectrum of fullerenol (in water) is represented by the solid line, and C60 (in toluene) is represented by the dashed line. Inset represents the linearity between absorption coefficients measured at 300 nm and fullerenol concentrations. that of C60 indicated that both absorb UVR and visible (400-700 nm) radiation from sunlight (Figure 1). The molar absorption coefficients of fullerenol are lower than those of C60 at wavelengths below 380 nm. In addition, C60 in toluene exhibits a characteristic absorption peak centered at 335 nm which was not observed in aqueous fullerenol (23). Kinetic Studies of Fullerenol Phototransformation. Dilute aqueous fullerenol (in the range of 0.8-5.6 µM) was irradiated using simulated solar radiation and monochromatic radiation (366 nm). Following simulated solar and monochromatic (366 nm) irradiation, the UV-visible absorption of fullerenol solutions decreased, indicating a reduction in fullerenol concentration by direct photoreaction. With two weeks of irradiation in the solar simulator, the absorption coefficients at 300 nm decreased by 90%. No measurable changes in fullerenol absorption coefficients were observed in dark controls. Although the absorption coefficient decreased 42% with 48 h irradiation (366 nm), the fluorescence intensity increased with the excitation emission matrix (EEM) peaks shifting to shorter wavelengths (SI Figures S2a and S2b), indicating the formation of fluorescent photoproducts. However, fluorescence decreased back to below detection following two weeks of irradiation by simulated solar radiation. The fluorescent photoproducts were not identified. Under dilute conditions in which the absorbance is 3 µM) could not have accounted for the observed decrease in rate with increasing conversion because product quenching could not have successfully competed with quenching by dissolved oxygen (∼200 µM). Previous studies have shown that fullerenol surface charge is pH dependent with the charge becoming more negative with increasing pH (17). These studies were particularly relevant to our studies because the same source of fullerenol was used. Our studies of changes in surface charge conducted at a lower fullerenol concentration showed that at pH 3, the surface charge of fullerenol aggregates was -35 mV, less
negative than that at pH 10 (-54 mV). The changes in surface charge observed in either of these studies could not account for the much larger changes in photoreaction rates that were observed in this study. Our studies of pH effects on fullerenol particle size using dynamic light scattering (DLS) agreed with earlier studies that showed a surprising insensitivity of mean cluster size to pH change (17). Thus, changes in cluster size do not appear to have a major effect on the photoreaction kinetics. Changes in the hemiketal constituents of fullerenol may help explain the pH-dependency of the photoreaction. Fullerenol contains hemiketal components that are subject to acid-catalyzed reactions (24, 25). Equation 2 illustrates a representative hemiketal moiety in a fullerenol molecule which can reversibly convert to a ketone moiety (25). It has also been proposed that conversion to the ketone moiety (CdO), which occurs under acidic conditions, results in a cage-opened structure that is less stable (more reactive) than the hydroxylated hemiketal carbon (CsOH) in fullerenol molecules (26, 27). Carbonyl compounds can participate in a variety of excited state reactions that could be involved in photoreaction of fullerenol, including triplet energy transfer, hydrogen atom abstraction, singlet oxygen, and superoxide production (3, 21). Therefore, the pH dependent ratio of CdO to CsOH in fullerenol may play an important role in its reactivity. We hypothesize that the fast component was dominated by photoreaction of reactive components of the fullerenol (i.e., higher quantitative ratio of CdO to CsOH) and that less photoreactive constituents (i.e., lower quantitative ratio of CdO to CsOH) dominated the slow component.
The UV-visible spectral data and reaction quantum yields can be used to compute photolysis rate constants of fullerenes as a function of time-of-day, season, location, and depth in water bodies (21, 28). The seasonal changes in simulated half-lives for fullerenol photoreaction near the surface of a water body at latitude 40°N were estimated (SI Figure S3a.). Minimum half-lives for both fast and slow components occur during summer, and maximum values occur during winter. The computed wavelength dependence of the specific sunlight absorption rate of fullerenol (SI Figure S3b) indicates that absorption of both UV and visible wavelengths initiate photolysis with peak photoreactivity in the 330-430 nm range. Under simulated solar radiation the average half-life t1/2 calculated from eq 1 was 73 h of simulated solar irradiation in good agreement with the computer simulations. This corresponds to about 9.6 cloudless days of average sunlight exposure near the surface of water bodies during summer at latitude 40°N. Photomineralization of Fullerenol. Although a complete identification of fullerenol photoproducts is not available, results of this study provided the first evidence that a fullerene derivative can be extensively mineralized under environmental conditions. Dissolved inorganic carbon (DIC), (i.e., CO2 and its hydrolysis products, bicarbonate and carbonate), was identified as a photoproduct in the pH 5.0-7.0 range. The measured DIC production rate decreased with increasing irradiation duration and after three weeks of simulated solar irradiation, the rate decreased to
