Direct Observation of Single Layer Graphene Oxide Reduction

May 15, 2014 - Department of Chemistry and Biochemistry, University of Notre Dame, Notre ... Taras Shevchenko National University of Kiev, Kiev, Ukrai...
0 downloads 0 Views 514KB Size
Download PDF
Letter pubs.acs.org/NanoLett

Direct Observation of Single Layer Graphene Oxide Reduction through Spatially Resolved, Single Sheet Absorption/Emission Microscopy Denis A. Sokolov,†,§ Yurii V. Morozov,†,⊥,§ Matthew P. McDonald,† Felix Vietmeyer,† Jose H. Hodak,‡ and Masaru Kuno*,† †

Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556, United States INQUIMAE-Departamento de Química Inorgánica, Analítica y Química Física, Facultad de Ciencias Exactas y Naturales, University of Buenos Aires, Buenos Aires, Argentina ⊥ Department of Physics, Taras Shevchenko National University of Kiev, Kiev, Ukraine ‡

S Supporting Information *

ABSTRACT: Laser reduction of graphene oxide (GO) offers unique opportunities for the rapid, nonchemical production of graphene. By tuning relevant reduction parameters, the band gap and conductivity of reduced GO can be precisely controlled. In situ monitoring of single layer GO reduction is therefore essential. In this report, we show the direct observation of laser-induced, single layer GO reduction through correlated changes to its absorption and emission. Absorption/emission movies illustrate the initial stages of single layer GO reduction, its transition to reduced-GO (rGO) as well as its subsequent decomposition upon prolonged laser illumination. These studies reveal GO’s photoreduction life cycle and through it native GO/rGO absorption coefficients, their intrasheet distributions as well as their spatial heterogeneities. Extracted absorption coefficients for unreduced GO are α405 nm ≈ 6.5 ± 1.1 × 104 cm−1, α520 nm ≈ 2.1 ± 0.4 × 104 cm−1, and α640 nm ≈ 1.1 ± 0.3 × 104 cm−1 while corresponding rGO α-values are α405 nm ≈ 21.6 ± 0.6 × 104 cm−1, α520 nm ≈ 16.9 ± 0.4 × 104 cm−1, and α640 nm ≈ 14.5 ± 0.4 × 104 cm−1. More importantly, the correlated absorption/emission imaging provides us with unprecedented insight into GO’s underlying photoreduction mechanism, given our ability to spatially resolve its kinetics and to connect local rate constants to activation energies. On a broader level, the developed absorption imaging is general and can be applied toward investigating the optical properties of other two-dimensional materials, especially those that are nonemissive and are invisible to current single molecule optical techniques. KEYWORDS: Graphene oxide, reduced graphene oxide, photolysis, absorption, emission, absorption coefficient

S

fact that single layer GO reduction has only been monitored postfactum by observing resulting changes to a sheet’s conductivity,3 resulting changes to its Raman spectrum,1 or physical changes to its friction coefficient.11 These characterization approaches have yielded limited mechanistic information about GO’s photoreduction due to their inability to capture single sheet reduction dynamics. A need therefore exists for a nondestructive, in situ characterization approach, which allows for the direct observation of single layer GO reduction and which provides detailed information about its chemical kinetics. In a previous study,12 we have observed the photolytic reduction of single layer GO to rGO through changes to its

ingle layer graphene oxide (GO) is a two-dimensional (2D) material that offers exciting opportunities as a graphene precursor. GO contains oxygen bearing functionalities. The removal of these groups reduces GO, yielding reduced-GO (rGO) − a chemical analogue of graphene. During reduction, GO’s chemical, optical, and electrical properties evolve. Specifically, the loss of oxygen-containing functional groups renders rGO hydrophobic,1 more absorptive,2 and significantly more conductive.3 Reduction also induces substantial changes to GO’s bandgap, transitioning it from a wide bandgap semiconductor to a semimetal.4 This transformative nature makes GO a fascinating yet challenging system to study. Among the various approaches for reducing GO, photoinduced deoxygenation5,6 has emerged as a facile alternative to harsh chemical7,8 and high-temperature9−11 treatments. While several reports exist on the laser reduction of GO and graphite oxide, the governing mechanisms behind these processes have not been conclusively established. This has stemmed from the © XXXX American Chemical Society

Received: February 6, 2014 Revised: May 2, 2014

A

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

emission intensity upon prolonged 405 nm (3.06 eV) laser irradiation. Specifically, we have observed the evolution of single layer GO emission during reduction as characterized by (1) an initial decrease in emission intensity along with a corresponding blueshift (associated with the reduction of GO to rGO), (2) a subsequent increase in emission intensity accompanied by a corresponding redshift (associated with the continued reduction as well as photolytic fragmentation of rGO into smaller graphenic fragments), and (3) irreversible photobleaching. Using emission movies that capture these changes, we have extracted rate constants and associated rate constant maps that illustrate the kinetic heterogeneity as well as energetics of photolytic GO reduction. In this study, we develop a spatially resolved absorption imaging capability that provides us with a unique experimental toolset, enabling correlated absorption and emission imaging of single layer GO during reduction. This has allowed us to unambiguously establish GO’s photoreduction life cycle and has led to the first direct estimates of GO and rGO absorption coefficients, absorption-based estimates of GO reduction rate constants, and additional information about intrasheet optical heterogeneities. In tandem, the developed absorption imaging enables the rapid identification of individual GO layers, which complements existing approaches for identifying single layer graphene and GO.13,14 More broadly, the technique opens up exciting possibilities for optically characterizing other 2D materials, especially those that are nonemissive. We have acquired spatially resolved single layer GO absorption data by probing individual GO sheets with 405 nm (3.06 eV), 520 nm (2.38 eV), and 640 nm (1.94 eV) lasers in a transmission experiment. Prior to entering a home-built single molecule imaging microscope, the laser is split into signal and reference beams. The latter is focused directly onto a reference photodiode. The former is focused onto samples using a high numerical aperture (N.A.) microscope objective. At a given point on a sheet, the laser is pulsed once through analogue modulation (1 ms pulse duration and 8 μW average power per pulse at the sample). This low-power pulsing scheme suppresses any light-induced GO photoreduction during measurements. Transmitted light is collected with a second high N.A. objective, arranged in a collinear, transmission geometry. Both detected signal and reference pulses are digitized, fast Fourier transformed (FFT), and low pass filtered to remove any high frequency noise. The signal is then compared with the reference to determine the amount of light attenuated by the sample. To construct an image, specimens are raster scanned under the excitation. Additional details about the direct absorption experiment, including an estimate of its limitof-detection can be found in Methods. Figure 1a−c shows acquired absorption images, which reveal spatial nonuniformities in GO’s % absorption values at all three probing wavelengths (405, 520, and 640 nm). This can be seen through brighter and darker regions in the color mapped images. Such intrasheet (absorption) inhomogeneities are consistent with spatially resolved, emission heterogeneities previously observed in single layer GO.12 Both are attributed to the existence of sizable sp2 domain size distributions within GO’s basal plane.12 This conclusion is supported by pronounced structural and chemical heterogeneities found in transmission electron microscopy (TEM),15,16 scanning tunneling microscopy (STM),17 and nuclear magnetic resonance (NMR)18 experiments of unreduced GO. Additional absorption images of different single layer GO sheets at all three

Figure 1. Absorption images of single layer GO acquired at (a) 405 nm (contour interval, 0.2% absorption), (b) 520 nm (contour interval, 0.05% absorption), and (c) 640 nm (contour interval, 0.05% absorption). Scale bars depict the percent absorption at a given wavelength. Histograms above each image show the distribution of percent absorption values and associated absorption coefficients for a GO ensemble and for two separate single layer sheets. (d) Single layer and ensemble GO absorption spectra overlaid with measured absorption coefficients.

wavelengths can be found in the Supporting Information (Figure S1). We determine that unreduced single layer GO attenuates 405 nm light by ∼0.65 ± 0.11%, 520 nm light by ∼0.21 ± 0.04%, and 640 nm light by ∼0.11 ± 0.03%. Corresponding absorption coefficients (α) are α405 nm ≈ 6.5 ± 1.1 × 104 cm−1, α520 nm ≈ 2.1 ± 0.4 × 104 cm−1, and α640 nm ≈ 1.1 ± 0.3 × 104 cm−1, based on an estimated GO thickness of ∼1 nm.17 To complement this data, Figure 1d shows the extinction spectrum of a single GO sheet acquired using a tunable supercontinuum source employed in conjunction with the direct absorption technique. Also shown is the absorption spectrum of the parent GO ensemble (dashed line). When both are combined with the above three color α measurements, single layer GO α-values across the entire visible range can be estimated (Supporting Information, Table S1). These results represent the first direct measurements of GO’s light attenuation parameters and are consistent with prior literature estimates.19 To put these percent absorption values into perspective, pristine monolayer graphene absorbs ∼2.3% of incident light across the visible.20 Above each image in Figure 1a−c is a histogram that illustrates the sizable inter- and intrasheet α-value distributions that exist. Specifically, ±17% to ±30% α-variations are found across all colors in an ensemble consisting of 12 individual GO sheets (gray histograms). The histograms of two individual sheets (blue and green histograms) are shown for comparison purposes. These latter single sheet distributions rationalize the B

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

breadth of the ensemble distribution and emphasize not only the prevalence of inherent intrasheet optical heterogeneities but also the existence of sheet-to-sheet variations, stemming from differences in GO’s degree of oxidation. Absorption ratio maps, constructed by dividing the single color absorption images in Figure 1a−c [i.e., 640 nm/405 nm (Figure 2a), 520 nm/405 nm (Figure 2b), and 640 nm/520 nm

Figure 2. (a−c) Absorption ratio maps obtained by dividing the single color absorption images in Figures 1a−c. (d,e) Correlated absorption/ emission images of unreduced single layer GO (λabs = 520 nm). Brighter colors indicate higher percent absorption and emission counts, respectively. White regions represent bilayer sections of the sheet. Absorption contour interval, 0.04%; emission contour interval, 500 counts. Figure 3. (a) Correlated absorption/emission images acquired during single layer GO reduction (λabs = 520 nm). Scale bar: 3 μm. (b) Corresponding absorption/emission intensity trajectories taken from the circled location on the sheet (λabs = 520 nm). (c) Representative emission spectrum waterfall plot obtained from a ∼2 μm wide region of a different GO sheet.

(Figure 2c)], highlight the existence of intrasheet spectral heterogeneities. These ratio maps reveal regions where single layer GO absorbs more red light than green light. To illustrate, Figure 2c shows a large (∼1 μm2) area in the middle of the sheet which preferentially absorbs red light. As with corresponding emission ratio maps that reveal prominent red and green emitting regions in single layer GO,12 the observed intrasheet absorption heterogeneity likely stems from different domain sizes as well as chemical disorder present within GO’s basal plane. Interestingly, a pixel-by-pixel analysis of correlated absorption/emission images reveals no indication that either strongly or weakly absorbing regions are more emissive (Figure 2d,e). Having established a baseline optical response for unreduced GO, we now probe its laser-induced reduction using correlated, single sheet, absorption/emission imaging. GO sheets are first identified through absorption imaging. They are then exposed to high power 405 nm illumination (Iexc ∼ 220 W cm−2) to initiate reduction. In tandem, an emission image is acquired with an EMCCD camera. Once complete, the absorption/ emission image sequence is repeated. This process continues until the sheets are fully reduced. In what follows, this method of reduction is referred to as “stepwise reduction”, as opposed to “continuous reduction”, where exposure to the high power 405 nm laser remains uninterrupted once reduction is initiated (Supporting Information, Figure S2). Stop motion absorption/ emission movies are subsequently created from these sequential images. Additional details about the study can be found in Methods with representative movies provided in Supporting Information (Movies S1, S2). Figure 3a shows images acquired from one such correlated absorption/emission movie (Supporting Information, Movie S1). Color bars represent percent absorption values and emission counts, respectively. Analysis of their intensities at a

given point (circled area of the sheet, ∼1 μm2) across all frames of both movies yields the traces shown in Figure 3b. Analogous trajectories obtained from other points of the same sheet are shown in the Supporting Information (Figure S3) along with absorption/emission trajectories from different sheets (Supporting Information Figure S4). These traces collectively depict the photoreduction life cycle of single layer GO through the evolution of its absorption and emission. We find three distinct regions of GO reduction. They are labeled regions 1, 2, and 3 in Figure 3b and are color coded. In region 1, during the first 55 s of stepwise exposure to high power 405 nm excitation (Iexc = 220 W cm−2), a sharp rise in percent absorption occurs. Values increase from ∼0.24 to ∼1.3% at 520 nm. Similar behavior is observed at 405 nm (∼0.8 to ∼1.5%) and at 640 nm (0.14 to 0.8%) [Supporting Information, Figure S5]. Identical behavior is observed with continuous reduction although we find that it produces slightly more absorptive sheets. Specifically, in the case of continuous reduction, percent absorption values increase to ∼2.1 ± 0.06% (405 nm), ∼1.7% ± 0.04% (520 nm), and ∼1.4 ± 0.04% (640 nm). This suggests that reoxidation potentially occurs during stepwise reduction, leading to lower (final) percent absorption values. In either case, whether continuous or stepwise, an increase in absorption is seen along with a concomitant decrease in emission QY (∼0.65% to ∼0.075%) [Supporting Information, Figure S6] accompanied by a spectral blueshift (Figure 3c).12 C

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

We12 and others21−23 have previously shown through ensemble and single sheet measurements that this absorption/emission behavior is consistent with the photoreduction of GO. In particular, ensemble studies show an increase of GO’s absorption across the visible, accompanied by a decrease of its total emission intensity during extended 405 nm illumination.12 Furthermore, X-ray photoelectron spectroscopy measurements reveal an increase of GO’s carbon/oxygen (C/O) ratio, associated with the loss of oxygen containing functional groups. This is further supported by ensemble Fourier transform infrared spectroscopy (FTIR) measurements, which reveal sizable losses of hydroxyl and carbonyl/carboxyl groups during extended 405 nm illumination. In parallel, single sheet Raman measurements indicate a narrowing of the G band full-width at half-maximum (fwhm), suggestive of GO reduction.24 The above assignment additionally agrees with optical Kerr gate measurements conducted by Kikkawa and co-workers, which show a significant blueshift and quenching of rGO’s emission.25 Taken as a whole, we posit that region 1 represents the reduction of single layer GO to rGO, given the large absorption and negligible emission of samples, which resemble the optical response of monolayer graphene.26 To better understand the underlying GO reduction mechanism, absorption and emission movies (Supporting Information, Movie S1) have been analyzed to obtain kinetic information about observed time-dependent changes in region 1. Absorption/emission trajectories extracted from three different positions of a single GO sheet are shown in Figure 4a,b. All absorption traces grow irrespective of λabs as a function

constants at other wavelengths have been provided in the Supporting Information and take similar magnitudes (Supporting Information, Figure S7). Corresponding emission rate constants for positions 1−3 are ⟨k1a1⟩ = 0.229 s−1 ⟨k1a2⟩ = 0.017 s−1, ⟨k1a3⟩ = 0.339 s−1. A comparison of ⟨k1a ⟩ and ⟨k1e ⟩ values shows that they possess the same order of magnitude and further suggests that both the absorption and emission imaging probe the same process. Moreover, these latter rate constants agree with previously obtained emission quenching rate constants of ke = 0.097, 0.128, and 0.169 s−1.12 Intrasheet ⟨k1a ⟩/⟨k1e ⟩ variations are highlighted through rate constant maps shown in Figure 5a,b. These maps have been

Figure 5. Absorption (λabs = 520 nm) and emission rate constant maps obtained for (a,b) region 1 and (c,d) region 2. Histograms illustrate absorption/emission rate constant distributions present within the basal plane.

obtained using SigmaPlot macros (Supporting Information, Appendix S1) that extract ⟨k1a/e⟩ values for individual pixels in both absorption and emission movies. A subsequent comparison of ⟨k1a ⟩ (⟨k1e ⟩) maps shows ±18% (±13%) variations across individual sheets as well as apparent spatial correlations between ⟨k1a ⟩ and ⟨k1e ⟩. To illustrate, a large ∼3 μm long region near the top right edge of the sheet in Figure 5 exhibits distinctly peaked rate constants both in ⟨k1a ⟩ (∼0.16 s−1) and ⟨k1e ⟩ (∼0.07 s−1) maps. Similar trends are observed within the middle and bottom right portions of the same sheet. To convey the dynamic nature of GO photoreduction, absorption/emission rate constant maps are overlaid with corresponding vector plots in the Supporting Information (Figure S8). The comparison highlights the reduction’s progression in region 1 (Java applet source code Supporting Information, Appendix S2) and shows that reduction in both the absorption/emission propagate across GO’s basal plane with the same direction. This illustrates that the two experiments monitor the same process.

Figure 4. Extracted absorption (a,c) (λabs = 520 nm) and emission (b,d) trajectories from three different positions (P1, P2, and P3) of a single layer GO sheet. The analyzed positions are indicated by dashed circles superimposed onto the sheet’s absorption image (inset, a). Corresponding fits (dashed lines) are shown in a−d along with extracted rate constants. Traces offset for clarity.

of time and are accompanied by a corresponding drop of the emission (Supporting Information, Figure S5). These absorption/emission traces are subsequently fit with biexponential rise and decay functions from where weighted (average) rate constants are calculated.27 In either case, region 1 absorption/ emission fits are dominated by their slow components (79 and 82% relative contribution, respectively). Obtained absorption (λabs = 520 nm) rate constants are therefore ⟨k1a1⟩ = 0.168 s−1, ⟨k1a2⟩ = 0.049 s−1, and ⟨k1a3⟩ = 0.099 s−1. Absorption rate D

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

induced loss of oxygen containing functional groups, leading to carbon abstraction.28,29 Resulting graphenic fragments are highly emissive and account for the increased emission quantum yields observed. Fragmentation from a larger network of interacting sp2 domains also removes fast nonradiative recombination pathways that induce apparent blueshifts of rGO’s emission.25 This enables the native (red) emission of individual sp2 domains to be seen and explains the spectral redshifts seen in Figure 3c.12,21 The above hypothesis is supported by a gradual drop of rGO’s absorption from a maximum of ∼1.3 to ∼0.1% at 520 nm during stepwise reduction. Similar behavior is observed at 640 nm (0.8 to 0.1%) and 405 nm (1.5 to 0.1%) (Supporting Information, Figure S5). In all cases, loss of absorption suggests the loss of absorbing species as would be characteristic of the fragmentation/breakup of rGO’s extended sp2 network. Subsequent kinetic analyses, identical to that conducted in region 1, show that region 2 k2a and k2e values are correlated in terms of their magnitudes and spatial distributions. Specifically, absorption and emission traces (Figure 4c,d), extracted from the same positions studied in region 1 are fit with single exponential decay and rise functions. These fits yield the following absorption rate constants (λabs = 520 nm): k2a1 = 2.26 × 10−3 s−1, k2a2 = 1.61 × 10−3 s−1, k2a3 = 2.92 × 10−3 s−1; kavalues for other absorption wavelengths are in agreement and have been provided in the Supporting Information (Figure S7). Corresponding emission rate constants are k2e1 = 4.18 × 10−3 s−1, k2e2 = 3.92 × 10−3 s−1 k2e3 = 4.52 × 10−3 s−1 and possess the same magnitude and trend as counterpart k2a -values. These latter k2e rate constants also agree with previously obtained emission rise rate constants of k2e = 2.26 × 10−3 s−1, 2.69 × 10−3 s−1, 3.61 × 10−3 s−1.12 Resulting absorption and emission rate constant maps (Figure 5c,d) show these spatial correlations more clearly (SigmaPlot macros, Supporting Information Appendix S1) and illustrate that both experiments probe the same process in region 2. To further illustrate this connection, absorption/emission rate constant maps are overlaid with corresponding vector plots in the Supporting Information (Figure S8), revealing that both absorption/emission fronts evolve with the same direction. As with region 1, temperature-dependent emission studies yield Arrhenius plots from where an activation energy can be estimated. We find that Ea = 0.52 eV characterizes photodegradation processes in region 2.12 This activation energy is larger than that found in region 1 and suggests a dissimilar set of photolytic reactions dominating region 2. On the basis of known activation energies, we suggest that region 2 processes involve direct OH photolysis (Ea = 0.7 eV),35 C−O−C migration (Ea = 0.9 eV),28 and C−O−C/CO/COOH photolysis (Ea > 1 eV).28,29,36 In several cases, these processes lead to carbon abstraction through CO and CO2 emission and result in the photolytic degradation of rGO. This would lead to a decrease of its absorption, consistent with Figure 3b. Furthermore, the conclusion is corroborated by mass spectroscopic observations showing CO/CO2 emission during the later stages of GO photoreduction.33 A final illustration of GO’s heterogeneous reduction chemistry involves a comparison of the spatial distribution of its relative region 1-to-region 2 emission magnitudes with the spatial distribution of its corresponding time-averaged absorption. Specifically, relative emission ratio maps have been constructed by taking the ratio of region 1 to region 2 emission maxima while corresponding (average) absorption

We attribute region 1 GO photoreduction to the photolysis12,28−30 of oxygen containing functional groups; dominant species within GO’s basal plane are hydroxyl (OH) and epoxy (C−O−C) groups.31 We exclude a more commonly invoked photothermal reduction mechanism5,19,23,25 because estimated temperature changes within GO following the absorption of light are on the order of a Kelvin.12 Supporting this, we have conducted temperature-dependent emission studies, which have enabled us to construct Arrhenius plots of k1e values. In turn, we find that region 1 reduction processes are characterized by an activation energy of Ea = 0.24 eV (upper and lower limits of 0.13 and 0.39 eV).12 This Ea value is in good agreement with the estimated activation energy for the photoinduced migration of OH functional groups (Ea = 0.32 eV)32 whose accumulation at defect edges leads to sp2 domain growth. In tandem, multiple (oxygen containing) functional group chemistries, also leading to the growth of sp2 domains as well as the release of H2O, are possible given their modest activation energies of Ea = 0.28−0.46 eV. The latter rationalizes mass spectroscopy observations of water evolved during the early stages of GO photoreduction.33 Having established single layer photolytic GO-to-rGO interconversion in region 1, we now employ the absorption technique to characterize rGO’s absorption parameters. By analyzing absorption images acquired at 405, 520, and 640 nm, we find that continuously reduced rGO attenuates 405 nm light by ∼2.1 ± 0.06%, 520 nm light by ∼1.7 ± 0.04%, and 640 nm light by ∼1.4 ± 0.04% (Supporting Information, Figure S9).34 Corresponding α-values are ≈21.6 ± 0.6 × 104 cm−1, α520 nm ≈ 16.9 ± 0.4 × 104 cm−1, and α640 nm ≈ 14.5 ± 0.4 × 104 cm−1. Absorption images, associated absorption coefficient histograms, and corresponding absorption ratio maps reveal the presence of substantial intrasheet absorption heterogeneities (Supporting Information, Figure S10). This is expected given the inherent disorder first seen in the absorption of unreduced GO (Figures 1a−c and 2a−c). It is also consistent with analogous rGO emission ratio maps constructed using emission images taken at 560 ± 20 and 730 ± 25 nm.12 In principle, analyzing the observed absorption heterogeneity enables us to estimate rGO sp2 domain sizes and their size distributions. However, given the diffraction-limited nature of the absorption measurement, as well as chemical variations in domain functionalization,23 it is not possible to immediately establish a quantitative link between α and domain size. The above light attenuation parameters represent the first direct measurements of rGO’s optical constants. In addition, Figure S11 of the Supporting Information shows the absorption spectrum of single layer rGO acquired using a tunable supercontinuum source coupled to the direct absorption experiment. Utilizing an ensemble extinction spectrum and the above α-values, rGO absorption coefficients across the entire visible range can be estimated (Supporting Information, Table S2). Region 2 of GO photoreduction (Figure 3b) is characterized by a dramatic rise of the emission intensity with QYs increasing from ∼0.075 to ∼14% (Supporting Information, Figure S6). A spectral redshift also occurs as illustrated through the waterfall plot shown in Figure 3c. The fact that this behavior occurs after rGO has formed suggests that photoreduction is succeeded by a second light-induced process. We have previously suggested that this behavior arises from the fragmentation of rGO’s extended sp2 domains into smaller molecular-like species.12 This occurs through the continued photolysis and photolysisE

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

reduction to rGO, (2) continued GO reduction as well as rGO fragmentation into emissive molecular-like species, and (3) photobleaching of resulting graphenic fragments. Detailed kinetic analyses of intrasheet rate constants extracted through both direct absorption and emission studies support the photolysis of oxygen containing species in GO as responsible for its photoreduction. The developed absorption technique simultaneously enables us to measure the absorption coefficients of single layer GO and rGO and has revealed sizable intrasheet optical heterogeneities for either system. Beyond GO/rGO the developed absorption/emission technique is general and opens the door to a number of exciting optical studies of 2D materials, not possible today using existing emission-based, single molecule techniques. Methods. Preparation of Graphene Oxide. Graphite oxide was produced using a modified Hummers’ synthesis.41 This entailed mixing 32 mL of concentrated H2SO4, 8 mL of concentrated H3PO4, 1.8 g of KMnO4, and 300 mg of 10 mesh high quality graphite. The mixture was then stirred in an ice bath for 1 h followed by 48 h of continuous room-temperature stirring. Following this, approximately 100 mL of deionized water was added to the mixture and the suspension was sonicated for 60 min using a 20 kHz corrosion resistant glass sonotrode operating at 35% amplitude power. When complete, hydrogen peroxide solution (3% v/v) was added to the graphite oxide mixture dropwise until the suspension turned yellow. Graphite oxide flakes were then recovered by centrifuging the suspension whereupon the recovered precipitate was washed with 1 M hydrochloric acid to dissolve any remaining salts. The resulting material was repeatedly washed with deionized water followed by centrifugation until a neutral pH was obtained. The final suspension was freeze-dried for 24 h to obtain a browncolored (graphite oxide) powder. Dilute solutions, containing single layer graphene oxide sheets, were prepared by exfoliating small quantities of this material in water with the aid of low power sonication. GO specimens were subsequently prepared by drop casting this suspension onto a flame cleaned fused silica coverslip (Esco). Excess liquid was then removed with a pipet and the remaining liquid was allowed to evaporate. AFM measurements reveal that this process yields predominantly single layer GO sheets in close contact with the substrate (Supporting Information Figure S13). Single Sheet Absorption and Emission Microscopy. An absorption instrument was built around a commercial inverted microscope (Nikon). It employs continuous wave laser excitation at 405, 520, and 640 nm (Coherent, Obis). To minimize sample photolysis a pulsed mode of laser operation was employed. Specifically, single GO sheets were exposed to a 1 ms pulse (8 μW average power per pulse at the sample) at each point of an imaging scan. Each laser was spatially filtered, expanded to 3 mm, and linearly polarized with a GlanThompson polarizer. Their outputs were subsequently split into signal and reference beams using a nonpolarizing 50%/ 50% beam splitter. The excitation beam was then focused to a diffraction limited spot on the sample with a high N.A. microscope objective (0.95 N.A., Nikon). Transmitted light was collected with a second high N.A. objective (0.75 N.A., Zeiss), collinear to the first and oriented in a transmission geometry. Signal and reference beams were detected with a pair of near shot noise limited amplified photodiodes (PH200, Highland Technology) whereupon spatially resolved absorption images were obtained by raster scanning samples with a closed loop piezo stage (Mad City Laboratories). A schematic of the

images have been made by averaging absorption values over GO’s entire photoreduction life cycle. In the former case, large emission ratios indicate regions that are more emissive as GO than as rGO while, in the latter case, large values emphasize regions of the sheet resistant to photodegradation. Figure 6

Figure 6. (a) Single sheet false color image showing the ratio of emission maxima between regions 1 and 2 [(max region 1)/(max region 2)]. (b) Average absorption image of the same sheet computed over its entire photoreduction life cycle. (c,d) Analogous data for a different GO sheet.

panels a,b and c,d as well as Figure S12 in the Supporting Information illustrate the apparent correlation that results, showing that highly emissive regions of unreduced GO are linked to photostable portions of the sheet. A possible explanation for this correlation lies in the local abundance of specific oxygen containing functionalities that enable unreduced regions of GO to be strongly emissive while at the same time difficult to photodegrade. Carboxyl, carbonyl, and epoxy groups fit these criteria as they possess theoretical dissociation energies of >1 eV.32,33 While it is possible that areas with high COOH/CO/C−O−C concentrations account for regions of initially intense emission and subsequent photostability, future studies that specifically probe local chemical heterogeneities are necessary to fully understand this observation as well as GO’s overall heterogeneous photochemistry. At this point, to explain the overall emission quenching in region 3 as well as observed periods of fluorescence intermittency37−39 (Figure 3b) we suggest that the underlying transformation here involves molecular photobleaching.40 Namely, excited state reactions of individual graphenic fragments with oxygen lead to the creation of nonemissive species, which permanently quench the emission. In parallel, emissive graphenic fragments undergo fluorescence intermittency prior to photobleaching,12 as characteristic of single molecule behavior. These hypotheses are supported by a steadily decreasing absorption throughout region 3. To conclude, we describe for the first time an in situ correlated absorption/emission study of single layer GO that has allowed us to fully map out its photoreduction life cycle. We find three distinct regions of GO photoreduction: (1) GO F

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

absorption/emission trajectories of a folded GO sheet; quantum yield as a function of irradiation time; absorption rate constants acquired from 520 and 640 nm absorption trajectories; absorption/emission rate constant maps overlaid with vector plots; rGO single layer absorption images overlaid with absorption distribution histograms; rGO absorption ratio maps; tabulated rGO absorption coefficients for the visible range; single layer rGO absorption spectrum overlaid with experimentally measured single layer rGO α-values; emission ratio map; AFM measurement of a graphene oxide sample; schematic of the experimental setup; absorption coefficient calculations; emission quantum yield calculations; and SigmaPlot macro for producing rate constant maps and Java source code for producing vector plots. This material is available free of charge via the Internet at http://pubs.acs.org.

instrument can be found in the Supporting Information (Figure S14). Single sheet absorption spectra were produced with a similar pulse modulation approach using a supercontinuum white light source (Fianium) equipped with an acousto-optic tunable filter (Fianium). Spectra were acquired in the 725 to 430 nm wavelength range using 5 nm steps. To quantify acquired data, the percent absorption at a given point i of an image was calculated using Ai = [1 − (aSi /aRi )/(bS/ bR)] × 100% with aSi (aRi ) as the measured signal (reference) photodetector voltage and ⟨bS/bR⟩ as an average background value from a nonabsorbing region. Detected analogue pulses (signal and reference) were digitized using a 16 bit data acquisition board (National Instruments, NI-6229). Fourier filtering was applied by setting to zero all frequencies larger than 1.25 kHz. The system’s limit-of-detection (LOD) [LOD = average background % absorption +3σ (standard deviation)] is estimated to be LOD640 nm = 0.047% (corresponding α-value 4.7 × 103), LOD520 nm = 0.024% (corresponding α-value 2.4 × 103), and LOD405 nm = 0.060% (corresponding α-value 6.0 × 103). Widefield emission images using 405 nm excitation (S06J Blu-ray laser diode) were obtained by placing a 300 mm focal length achromat prior to the excitation objective’s back aperture. The intensity at the sample was ∼220 W cm−2. Any resulting emission was collected with the same objective and was passed through a 425 nm long pass filter (Chroma) prior to being imaged with a thermoelectrically cooled, electron multiplying camera (EMCCD) (Andor). Correlated absorption/emission imaging during reduction was carried out by first acquiring 640 nm absorption images, followed by analogous 520 and 405 nm images (in this order). This sequence minimized any laser-induced reduction of GO during absorption experiments. In all cases, 8 μW average power per pulse was used. Widefield emission images were subsequently collected by exciting samples with a high power 405 nm laser (S06J Blu-ray laser diode, Iexc = 220 W cm−2) and imaging the resulting emission with an EMCCD camera (Andor). This absorption/emission imaging sequence was then repeated until complete absorption/emission trajectories were obtained. Individual absorption and emission images were combined into corresponding stop motion movies using ImageJ (http://rsbweb.nih.gov/ij). Absorption and emission rate constant maps were generated using SigmaPlot macros. Trajectories from individual pixels in both absorption and emission movies were fit to specified functions (biexponential for region 1 and single exponential for region 2). Spatially resolved rate constants were then extracted from resulting fits using SigmaPlot macros. Rate constant maps were constructed from these values, where each map’s size was increased by a factor of 2 (bilinear interpolation, ImageJ). Corresponding absorption/emission vector plots were produced using a home-written Java applet. The source code for these programs has been provided in the Supporting Information Appendix.

■

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions §

D.A.S. and Y.V.M. contributed equally. D.A.S., Y.M., and J.H.H performed the experiments. D.A.S., Y.M., M.P.M., F.V., J.H.H., and M.K. analyzed and interpreted the results. D.A.S., Y.M., J.H.H., and M.K. cowrote the paper. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Acknowledgment is made to the donors of the American Chemical Society Petroleum Research Fund for support of this research (grant 51657). We also thank the Army Research Office (W911NF-12-1-0578) and the Center for Sustainable Energy at Notre Dame (cSEND) for financial support. J.H.H. thanks CONICET for the international cooperation funds [D979(25-03-2013)] and FONCyT for research grant P.BID2009 PICT-PRH107). We thank Kizhanipuram Vinodgopal for providing us with the graphite oxide sample.

■

REFERENCES

(1) Moon, I. K.; Lee, J.; Ruoff, R. S.; Lee, H. Nat. Commun. 2010, 1, 73−1−73−6. (2) Wang, S. J.; Geng, Y.; Zheng, Q.; Kim, J.-K. Carbon 2010, 48, 1815−1823. (3) Jung, I.; Dikin, D. A.; Piner, R. D.; Ruoff, R. S. Nano Lett. 2008, 8, 4283−4287. (4) Eda, G.; Mattevi, C.; Yamaguchi, H.; Kim, H.; Chhowalla, M. J. Phys. Chem. C 2009, 113, 15768−15771. (5) Cote, L. J.; Cruz-Silva, R.; Huang, J. X. J. Am. Chem. Soc. 2009, 131, 11027−11032. (6) Eswaraiah, V.; Aravind, S. S. J.; Ramaprabhu, S. J. Mater. Chem. 2011, 21, 6800−6803. (7) Gilje, S.; Han, S.; Wang, M.; Wang, K. L.; Kaner, R. B. Nano Lett. 2007, 7, 3394−3398. (8) Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S. Carbon 2007, 45, 1558−1565. (9) Becerril, H. A.; Mao, J.; Liu, Z.; Stoltenberg, R. M.; Bao, Z.; Chen, Y. ACS Nano 2008, 2, 463−470. (10) Ghosh, T.; Biswas, C.; Oh, J.; Arabale, G.; Hwang, T.; Luong, N. D.; Jin, M.; Lee, Y. H.; Nam, J.-D. Chem. Mater. 2012, 24, 594−599. (11) Wei, Z.; Wang, D.; Kim, S.; Kim, S.-Y.; Hu, Y.; Yakes, M. K.; Laracuente, A. R.; Dai, Z.; Marder, S. R.; Berger, C.; King, W. P.; de Heer, W. A.; Sheehan, P. E.; Riedo, E. Science 2010, 328, 1373−1376.

ASSOCIATED CONTENT

S Supporting Information *

Absorption and emission movies; unreduced GO absorption images at three different wavelengths; tabulated GO absorption coefficients for the visible range; experimental GO reduction acquisition sequences; absorption/emission trajectories from GO sheet; absorption/emission trajectories for two different GO sheets; absorption/emission images and corresponding G

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

(12) McDonald, M. P.; Eltom, A.; Vietmeyer, F.; Thapa, J.; Morozov, Y. V.; Sokolov, D. A.; Hodak, J. H.; Vinodgopal, K.; Kamat, P. V.; Kuno, M. Nano Lett. 2013, 13, 5777−5784. (13) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306, 666−669. (14) Kim, J.; Cote, L. J.; Kim, F.; Huang, J. J. Am. Chem. Soc. 2010, 132, 260−267. (15) Erickson, K.; Erni, R.; Lee, Z.; Alem, N.; Gannett, W.; Zettl, A. Adv. Mater. 2010, 22, 4467−4472. (16) Pacilé, D.; Meyer, J. C.; Fraile Rodríguez, A.; Papagno, M.; Gómez-Navarro, C.; Sundaram, R. S.; Burghard, M.; Kern, K.; Carbone, C.; Kaiser, U. Carbon 2011, 49, 966−972. (17) Paredes, J. I.; Villar-Rodil, S.; Solís-Fernández, P.; MartínezAlonso, A.; Tascón, J. M. D. Langmuir 2009, 25, 5957−5968. (18) Cai, W.; Piner, R. D.; Stadermann, F. J.; Park, S.; Shaibat, M. A.; Ishii, Y.; Yang, D.; Velamakanni, A.; An, S. J.; Stoller, M.; An, J.; Chen, D.; Ruoff, R. S. Science 2008, 321, 1815−1817. (19) Zhou, Y.; Bao, Q.; Varghese, B.; Tang, L. A. L.; Tan, C. K.; Sow, C.; Loh, K. P. Adv. Mater. 2010, 22, 67−71. (20) Nair, R. R.; Blake, P.; Grigorenko, A. N.; Novoselov, K. S.; Booth, T. J.; Stauber, T.; Peres, N. M. R.; Geim, A. K. Science 2008, 320, 1308. (21) Luo, Z.; Vora, P. M.; Mele, E. J.; Johnson, A. T. C.; Kikkawa, J. M. Appl. Phys. Lett. 2009, 94, 111909−1−111909−3. (22) Krishnamoorthy, K.; Veerapandian, M.; Mohan, R.; Kim, S.-J. Appl. Phys. A: Mater. Sci. Process. 2012, 106, 501−506. (23) Chien, C.-T.; Li, S.-S.; Lai, W.-J.; Yeh, Y.-C.; Chen, H.-A.; Chen, I.; Chen, L.-C.; Chen, K.-H.; Nemoto, T.; Isoda, S.; Chen, M.; Fujita, T.; Eda, G.; Yamaguchi, H.; Chhowalla, M.; Chen, C.-W. Angew. Chem., Int. Ed. 2012, 51, 6662−6666. (24) Petridis, C.; Lin, Y.-H.; Savva, K.; Eda, G.; Kymakis, E.; Anthopoulos, T. D.; Stratakis, E. Appl. Phys. Lett. 2013, 102, 093115− 1−093115−5. (25) Exarhos, A. L.; Turk, M. E.; Kikkawa, J. M. Nano Lett. 2013, 13, 344−349. (26) Mak, K. F.; Ju, L.; Wang, F.; Heinz, T. F. Solid State Commun. 2012, 152, 1341−1349. (27) Lakowicz, J. R. Principles of fluorescence spectroscopy, 2nd ed.; Kluwer Academic/Plenum Publishers: New York, 1999; pp 129−131. (28) Plotnikov, V. G.; Smirnov, V. A.; Alfimov, M. V.; Shul’ga, Y. M. High Energy Chem. 2011, 45, 411−415. (29) Andryushina, N. S.; Stroyuk, O. L.; Dudarenko, G. V.; Kuchmiy, S. Y.; Pokhodenko, V. D. J. Photochem. Photobiol., A 2013, 256, 1−6. (30) Matsumoto, Y.; Koinuma, M.; Ida, S.; Hayami, S.; Taniguchi, T.; Hatakeyama, K.; Tateishi, H.; Watanabe, Y.; Amano, S. J. Phys. Chem. C 2011, 115, 19280−19286. (31) Lerf, A.; He, H.; Forster, M.; Klinowski, J. J. Phys. Chem. B 1998, 102, 4477−4482. (32) Ghaderi, N.; Peressi, M. J. Phys. Chem. C 2010, 114, 21625− 21630. (33) Shulga, Y. M.; Martynenko, V. M.; Muradyan, V. E.; Baskakov, S. A.; Smirnov, V. A.; Gutsev, G. L. Chem. Phys. Lett. 2010, 498, 287− 291. (34) rGO, used for analysis, was produced via the continuous reduction approach. Specifically, GO sheets were first identified via absorption imaging. Following this, they were continuusly illuminated with a 405 nm widefield laser, while their absorption values were monitored. Once the maximum absorption value was achieved, the 405 nm illumination was terminated and rGO sheet was rescanned with absorption imaging. The same process was repeated for another 8 sheets. (35) Lahaye, R. J. W. E.; Jeong, H. K.; Park, C. Y.; Lee, Y. H. Phys. Rev. B 2009, 79, 125435−1−125435−8. (36) Smirnov, V. A.; Shul’ga, Y. M.; Denisov, N. N.; Kresova, E. I.; Shul’ga, N. Y. Nanotechnol. Russ. 2012, 7, 156−163. (37) Kuno, M.; Fromm, D. P.; Gallagher, A.; Nesbitt, D. J.; Micic, O. I.; Nozik, A. J. Nano Lett. 2001, 1, 557−564.

(38) Kuno, M.; Fromm, D. P.; Hamann, H. F.; Gallagher, A.; Nesbitt, D. J. J. Chem. Phys. 2001, 115, 1028−1040. (39) Wustholz, K. L.; Bott, E. D.; Kahr, B.; Reid, P. J. J. Phys. Chem. C 2008, 112, 7877−7885. (40) Gensch, T.; Böhmer, M.; Aramendía, P. F. J. Phys. Chem. A 2005, 109, 6652−6658. (41) Hummers, W. S.; Offeman, R. E. J. Am. Chem. Soc. 1958, 80, 1339−1339.

H

dx.doi.org/10.1021/nl500485n | Nano Lett. XXXX, XXX, XXX−XXX