A spectroscopic study to assess the photogeneration of singlet oxygen by graphene oxide

Carles Felip-León, Marta Puche, Juan F. Miravet, Francisco Galindo,* ,a Marta Feliz** ,b a Universitat Jaume I, Departamento de Química Inorgánica y Orgánica, Avda.Sos Baynat s/n, 12071, Castellón, Spain. *Corresponding author. E-mail: francisco.galindo@uji.es (Francisco Galindo) b Instituto de Tecnología Química (Universitat Politècnica de València – Consejo Superior de Investigaciones Científicas), Avd. de los Naranjos s/n, 46022 Valencia, Spain. **Corresponding author. E-mail: mfeliz@itq.upv.es (Marta Feliz)

Despite the efforts devoted to understand the chemistry of graphene and its derivatives still important questions remain to be answered. Some of those questions are concerns about their potential environmental toxicity [16][17][18][19][20][21][22][23][24][25], and to the in vivo toxicity of graphene, GO and derived hybrid nanocomposites used for biological imaging and therapy [15,26,27,28,29]. In this last realm, the use of shot exposure times to light (minutes) is frequent, in contrast to environmental studies (hours). Reactive oxygen species (ROS) such as singlet oxygen ( 1 O2), superoxide radical anion (O2 -. ) and hydroxyl radical (·OH) have been detected upon irradiation of fullerenes [30], carbon nanotubes [31,32], and graphene quantum dots [33]. However, the investigation of ROS production by GO and reduced graphene oxide (rGO) is much recent. Krishnamoorthy et al. reported on the behaviour of GO as a photoreductant but not informed about the generation of ROS [34]. More recently the group of Jafvert detected O2 -. but not 1 O2 or ·OH after irradiation of GO in water [35].
Later, the group of Sarkar reported that rGO was responsible for the production of ROS in aged samples of GO (suggesting the involvement of 1 O2, although not directly detected) [36]. More recently, Li, Keller et al. have studied the photochemistry GO and rGO, finding that O2 -. was the most abundant ROS generated upon irradiation during long irradiation times (>24h) and using high-power light sources (800 W Xe lamp) [37,38]. In this case, the concentration of 1 O2 was found <3·10 -14 M. 4 The most frequently used probe for determining de 1 O2 production is FFA [39]. But the use of this molecule requires a centrifugation step and HPLC analysis for monitoring the reaction. In the last years, very sensitive spectroscopic probes based on fluorescence are gaining acceptance since they allow an in situ measurement of the reaction progress, especially for short irradiation times. In the course of our research on graphene composites [40], we were interested in the analysis of ROS produced by GO, and we hypothesized that highly-sensitive spectroscopic methods to detect 1 O2 could be used as a complement to chromatographic ones. Hence we turned our attention to fluorescent probes based on anthracene as 1 O2 indicators. Additionally, we combined this approach with the direct measurement of 1 O2 phosphorescence at ca. 1275 nm. Our investigation confirms that 1 O2 production upon irradiation of GO is minimal, as reported with the traditional FFA method [35,37]. This finding is applicable specifically to the sample of GO prepared by us; samples with a different degree of oxidation or aging after preparation could have a different behaviour. This fact highlights the utility of having a variety of detection methods to assess the production of ROS in graphene derived materials. The need of a toolbox of probes for ROS is strongly recommended in research on GO, especially to evaluate the interaction of GO based nanomaterials with biological media [41,42].

Instrumentation
Combustion chemical analysis of the samples were carried out using a Fisons EA 1108-CHNS-O analyzer. Fourier-transform infrared spectroscopy (FT-IR) spectra were measured on KBr pellets with a Nicolet 8700 Thermo spectrometer. The Raman spectra were obtained from solid samples previously deposited onto aluminium or quartz wafers, indistinctively, using a "Reflex" Renishaw spectrometer, equipped with an Olympus microscope. The exciting wavelength was 514 nm of an Ar + ion laser. The laser power on the sample was ~10-25 mW and a total of 20 acquisitions were taken for each spectra. UV-Vis spectra were recorded in solution on JASCO V-630 spectrophotometer. Fluorescence measurements were recorded using a JASCO FP-8300 apparatus. Solid-state 13 C magic angle spinning nuclear magnetic resonance (MAS-NMR) spectra were recorded at RT by using a Bruker AV400WB spectrometer. The samples were spinning at the magic angle at 10 kHz, choosing /2 pulses of 5 s and a recycle delay of 5 s. Powder X-ray diffraction (XRD) patterns were obtained by using a Philips X′Pert diffractometer and copper radiation (CuKα = 1.541178 Å). X-ray photoelectron (XPS) spectra were collected using a SPECS spectrometer with a 150MCD-9 detector and using a non monochromatic AlKα (1486.6 eV) X-Ray source. Spectra were recorded using analyzer pass energy of 30 eV, an X-ray power of 50W and under an operating pressure of 10 -9 mbar. During data processing of the XPS spectra, BE values were referenced to C1s peak (284.5 eV). Spectra treatment has been performed using the CASA software. Atomic force microscopy (AFM) images were recorded by using a Multimode Nanoscope 3A instrument operating in tapping mode and with a Si wafer as the substrate. Samples for high resolution electron microscopy (HR-TEM) were ultrasonically dispersed in Milli-Q water and transferred into carbon coated copper grids. HR-TEM images were recorded by using a JEOL JEM2100F microscope operating at 200 kV. 6

Synthesis and characterization of GO
GO has been prepared by following the improved Hummer's synthetic method, and by optimization of a previous reported procedure [43,44]. A mixture of concentrated H2SO4/H3PO4 (360:40 mL) was added to a mixture of graphite (3 g) and KMnO4 (18 g) to produce an exothermic reaction to 35-40 ºC. The reaction was then heated to 50 ºC and stirred for 13 h, then cooled to room temperature and poured onto 400 mL of ice with 30% H2O2 (3 mL). After air cooling, the suspension was filtered and washed first with an aqueous HCl (1:10, 37%) solution, and finally with water until pH 7. The resulting solid was suspended in water (400 mL) and sonicated for 30 min; then the suspension was centrifuged at 4000 rpm for 4 h and the solid was removed. The liquid suspension was newly centrifuged at 15000 rpm for 1 h, and the resulting solid was dried at 60 ºC to afford ca. 1 g of a dark brown material identified as GO. This material has been characterized by FT-IR, Raman, XRD, XPS, solid-state 13 C MAS-NMR, AFM, HR-TEM, and combustion chemical analyses (C 40.07%, H 2.19%, N 0.0%).

Chemical trapping of 1 O2
Photo-oxidation reactions reactions were performed under air inside 3 mL fluorescence quartz cuvettes (1 cm light path) containing aerated aqueous solutions of the singlet oxygen trap ABDA (3 mL, 50 µM) and RB (4 µM) or hetereogeneous GO (0.05 mg/mL) photosensitizers. Prior to irradiation, GO was dispersed in distilled water (1mg/mL), sonicated for 60 min, and 150 L of the resulting dispersion was diluted to 3 mL and sonicated again for 1 min. Irradiations were carried out, with continuous stirring, using a cylindrical reactor (equipped with LED lamps, 400-700 nm emission output, 15.6 mW/m 2 ) placed 12 cm away from the cuvette. The evolution of the photoreactions was monitored over time (maximum 16 min.) by means of fluorescence spectroscopy (decrease of fluorescence emission at λem = 430 nm, λexc = 375 nm). The initial points of the kinetic traces were fitted to a pseudo-first order model (ln C/C0 = -kobs · t, where C is the concentration of ABDA at a certain time t and C0 is the initial concentration of ABDA (for low concentrations is can be assumed that fluorescence intensity is proportional to concentration). The same methodology was followed by using fullerene-C60 (0.05 mg/mL) as photosensitizer and DMA (3 mL, 50 µM) as a probe. Control experiments were performed by using ABDA in the absence of the photosensitizer, in the dark or under N2 atmosphere, using GO as photosensitizer. Quenching experiments were done in the presence of NaN3 (10 mM).

Direct singlet oxygen detection
The singlet oxygen phosphorescence decay traces after the laser pulse were registered at ca. 1275 nm employing a Peltier-cooled (-62.8 ºC) A pulsed Nd:YAG L52137 V LOTIS TII was used at the excitation wavelength of 355 nm. The single pulses were ca.
10 ns duration, and the energy was lower than 5 mJ per pulse. The system consisted of a pulsed laser, a 77250 Oriel monochromator coupled to a Hamamatsu NIR detector and an oscilloscope connected to the computer. The output signal was transferred from the oscilloscope to a personal computer. All measurements were made at room temperature, under air atmosphere, and using the chosen solvent (MilliQ water for previously ultrasonically dispersed GO, and toluene for fullerene-C60) in 10 × 10 mm 2 quartz cells with a capacity of 4 mL. The absorbance of the freshly prepared samples was adjusted to 0.28 for the singlet oxygen measurements at the laser excitation wavelength.

Results and Discussion
GO was synthesized by an optimization of the improved Hummers method [43,44].
The characterization of the material was done using spectroscopic, X-ray, morphological, and combustion chemical analyses. [13] The identification of hydroxyl, epoxy and carboxylic groups in the GO structure has been confirmed by infrared, solid-state 13 C MAS-NMR and XPS analyses (Fig. 1, A-C). [45,46] The Raman spectrum of GO (Fig 1D) shows the characteristic D (1346 cm -1 ) and G bands (1605 cm -1 ), where the G band is associated to the carbon-carbon vibrations of the aromatic rings, and the D band to the presence of defects and to the grade of disorder introduced by the oxygen functionalities. [47][48][49][50] The XRD pattern of GO nanosheets (Fig. 1E) reveals the most prominent (001) diffraction peak centered at 2 11.5º, which corresponds to an interlamellar spacing (7.64º) associated to the grade of oxidation of the graphene sheets, and a turbostratic stacking arrangement of the structure. [51,52] The presence of single and 2-4 layers of GO have been confirmed by HR-TEM and AFM techniques (Fig. 2). [53,54]   In order to assess the production of 1 O2, the probe ABDA (Fig. 3) was used since it is widely employed as a trap for this species in the biomedical realm and can be monitored easily by UV-vis absorption or fluorescence spectroscopies [55,56]. The underlying operational principle for ABDA involves the disappearance of its main absorption centred at 375 nm (and hence its fluorescence emission at λ = 430 nm) after reaction with 1 O2 (ABDA·O2 endoperoxide depicted in Fig. 3 is formed). In Fig. 3 it is shown a representative series of fluorescence spectra, specifically monitoring the reaction of ABDA with 1 O2 generated by the well-known photosensitizer Rose Bengal (RB).  Table 1).
An aqueous suspension of GO (0.05 mg/mL) was prepared, in the presence of ABDA (50 M), and irradiated with a photo-reactor containing white light emitting diodes (LED 400-700 nm). The fluorescence emission of ABDA was recorded and the intensity plotted against the irradiation time (Fig. 4). The data were fitted to a pseudo-first order model, and kobs are compiled in Table 1. The reaction was repeated in deuterated water since it is reported that the rates of singlet oxygen mediated reactions are enhanced in this medium (about ten-fold) due to the longer lifetime of 1 O2 as compared to water (67.9 s vs 3.45 s in D2O and H2O, respectively) [57]. The isotopic effect can be expressed as the rate of the constants in water and D2O (kD/kH) resulting a value of ca. 3. Finally the use of a specific quencher of singlet oxygen, like NaN3, was assayed [58]. The azide quencher did not cause any relevant decrease in the rate, resulting a value for this effect of kH/kH(N3 -) = 1.1. A well-known photosensitizer like RB was tested as a control. In this case the isotopic effect was found according to the expected value (kD/kH is ca. 7) and more importantly the azide effect was as pronounced as described in the literature (kD/kD(N3 -) = 47.6). See all the kinetic traces in Fig. 4 and a compilation of the values in Table 1. Overall, considering the weak isotopic effect and absence of quenching by azide, in the case of GO, it must be concluded that the drop in the emission of ABDA would not be attributable to 1 O2 but to another side reaction.  notably faster than the rate measured for GO. As a matter of fact, RB depletes the emission of ABDA in seconds (Fig. 3), whereas the photoirradiation of GO leads to minor detectable changes in ABDA fluorescence only after minutes. This supports the view that the amount of generated 1 O2 by GO must be minimal. Just for qualitative assessment another wellknown photosensitizer, fullerene-C60, was used to certify that GO generates very low amounts of 1 O2. The purpose of this assay was to use an excellent photosensitizer (= 0.96) with a comparable absorption to GO [59]. Hence, a sample of 0.05 mg/mL of fullerene-C60 was irradiated in toluene, since this photosensitizer generates highly efficiently 1 O2 in organic apolar medium (but not in water). In this case another anthracene derivative was employed (dimethylanthracene, DMA) since ABDA is not soluble in toluene. The measured reaction rate was 0.6920 min -1 . Considering that the lifetime of singlet oxygen in toluene (30.3 µs) is about half of the value in D2O [57], this result supports the idea of minimal amount of 1 O2 generated by GO upon irradiation.

14
A question that can be raised is whether the slight decrease of the ABDA fluorescence could be attributed to other species different from 1 O2. It is described that anthracene and its derivatives (An) are very prone to oxidation leading to the corresponding radical cation (An +. ). This species react with superoxide to give an adduct (AnO2) identical to the one obtained by reaction with singlet oxygen. This reaction has been described many times for anthracene and electron transfer photosensitizers [60]. As a matter of fact, the specificity of anthracenic probes for the detection of 1 O2 has been studied in detail recently: the genetically encodable fluorescent tag miniSOG has been reported to produce 1 O2 with a notable quantum yield (0.47) using the probe ADPA [61]. However, Nonell and Flors have measured a significantly smaller value (0.03) [62]. This discrepancy was attributed to the ability of miniSOG to photooxidize the probe and hence lead to reaction with superoxide.
Hence, ADPA would be actually measuring both ROS produced by type I (electron transfer) and type II (energy transfer) mechanisms. In another example, the endoperoxide of DMA was formed upon irradiation of alkyloxo(methoxo) tetraphenylporphyrinatoantimony via photoinduced electron transfer mechanism involving DMA +. and O2 -.
exclusively [63]. Accordingly, the mechanism involving O2 -. described in eqs.1-4 can be proposed to account for the minimal bleaching of ABDA upon irradiation of GO. In this case, apart from 1 O2, the superoxide anion would be involved in the reaction with ABDA.
Unfortunately the slopes of the kinetics recorded in this study are too close to the control irradiations to allow any conclusion on this regard. What it can be affirmed, however, is It is not ruled out that ageing of the GO samples would led to the formation of byproducts similar to rGO and low molecular-weight species, as recently demonstrated by other groups [37,38,64]. If this is the case, the origin of O2 -. would be this phototransformed fraction of the sample since it has been demonstrated the reducing capacity of rGO.
However this possibility is very unlikely in our case since photoreduction of GO to rGO has been described using high energetic UV light during several hours, whereas we are using a source of visible light and very short irradiation periods (minutes), precisely to avoid photoageing of our GO. The ageing of GO upon irradiation is a matter of debate that falls out of the scope of this research, which is focused on the use of sensitive spectroscopic probes to study GO, as a complementary tool to FFA and chromatography.
Finally, a direct measurement of the phosphorescence of 1 O2 at ca. 1275 nm was also attempted [57], since this is an irrefutable probe of the existence of this ROS. Hence, GO dispersed in water and irradiated at 355 nm yielded only a noisy signal (red line, Fig. 5), in contrast with the strong intensity showed by irradiated fullerene-C60 in toluene (black line,

Conclusions
We have found that generation of singlet oxygen upon short periods of visible light irradiation of samples of freshly prepared GO is almost negligible, using a high sensitivity spectroscopic fluorescent probe like ABDA. Exact quantification of this species remained elusive due to the dispersion of GO in water (light scattering), however there is enough qualitative evidence to suggest that the irradiation of GO with visible light during short periods of time gives rise to negligible amounts of this ROS. These results confirm recent measurements reported by other groups using FFA and chromatographic analyses as tools.
We hope, firstly, that this study will contribute to answer some questions regarding the potential toxicity of nanomaterials derived from graphene and, secondly, that it will add ABDA, or other fluorescence probes, along with direct detection via phosphorescence emission, to the toolbox that environmental scientist use in their researches.