Contactless Surface Conductivity Mapping of Graphene Oxide Thin

Dec 22, 2012 - Atomic force microscopy for two-dimensional materials: A tutorial review ... J. Molina , J. Bonastre , J. Fernández , A.I. del Río , ...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Contactless Surface Conductivity Mapping of Graphene Oxide Thin Films Deposited on Glass with Scanning Electrochemical Microscopy Joel Azevedo,† Céline Bourdillon,§ Vincent Derycke,† Stéphane Campidelli,† Christine Lefrou,‡ and Renaud Cornut*,§ †

CEA, IRAMIS, Service de Physique de l’Etat Condensé, Laboratoire d’Electronique Moléculaire, F-91191 Gif sur Yvette, France Laboratoire d’Electrochimie et de Physico-chimie des Matériaux et des Interfaces, UMR 5279, CNRS-Grenoble-INP-UdS-UJF, 1130 rue de la piscine, B.P. 75, Domaine Universitaire, 38402 Saint Martin d’Hères Cedex, France § CEA, IRAMIS, Service de Physique et de Chimie des Surfaces et Interfaces, Laboratoire de Chimie des Surfaces et Interfaces, F-91191 Gif sur Yvette, France ‡

S Supporting Information *

ABSTRACT: The present article introduces a rapid, very sensitive, contactless method to measure the local surface conductivity with Scanning Electrochemical Microscopy (SECM) and obtain conductivity maps of heterogeneous substrates. It is demonstrated through the study of Graphene Oxide (GO) thin films deposited on glass. The adopted substrate preparation method leads to conductivity disparities randomly distributed over approximately 100 μm large zones. Data interpretation is based on an equation system with the dimensionless conductivity as the only unknown parameter. A detailed prospection provides a consistent theoretical framework for the reliable quantification of the conductivity of GO with SECM. Finally, an analytical approximation of the conductivity as a function of the feedback current is proposed, making any further interpretation procedure straightforward, as it does not require iterative numerical simulations any more. The present work thus provides not only valuable information on the kinetics of GO reduction in mild conditions but also a general and simplified interpretation framework that can be extended to the quantitative conductivity mapping of other types of substrates. The conductivity of various thin films is generally evaluated using a four-point probe technique (see for example ref 5). This requires putting metallic tips in contact with the substrate which unavoidably damages the film, all the more when it is only few nanometers thick. In addition, this procedure provides a conductivity evaluation that is averaged over large areas of typically few millimeters. This is acceptable for highly homogeneous films but not suitable for very thin GO films from which it cannot reveal the local variability of conductivity at the micrometric scale. Measurements by conducting-AFM could be a credible alternative as it is commonly used for the electrical characterization of molecular materials,6 but the complex interactions between the probe and the substrate make difficult the quantitative determination of the local conductivity. Most importantly, conventional conducting-AFM can only be performed on a conductive underlying substrate and would not be a simple alternative for the present study, where the graphene oxide is deposited on an insulator. Scanning Electrochemical Microscopy (SECM) is an efficient analytical method for the local characterization of interfaces at a

F

ullerene and carbon nanotubes have been the subject of intense research in the past decade as promising carbon nanomaterials. During the past few years, researchers have also increased their attention to graphene, this one atom thick honeycomb lattice of carbon atoms. Grabbing the scientific curiosity and revealing its outstanding electrical and mechanical properties were not enough, and graphene is now promised to revolutionize a large panel of applications. However, material availability is still a key challenge to efficiently introduce this material in practical devices. The oxidation and exfoliation of graphite into Graphene Oxide (GO) is commonly accepted as one of the most promising scalable route to obtain large graphene flakes.1 However, this approach faces two crucial requirements: finding not only an effective reduction step to convert GO into a conductive material2 but also an effective, reliable, and nondestructive method to measure its electrical properties. In the literature, among the numerous studies devoted to the reduction of GO (see ref 3 and references therein), very few concern the use of alkaline conditions.4 In the present study, the reduction of GO in such alkaline conditions but at room temperature is highlighted, and the reaction kinetics is investigated by measurement of the substrate’s conductivity after different exposition times. © 2012 American Chemical Society

Received: October 31, 2012 Accepted: December 22, 2012 Published: December 22, 2012 1812

dx.doi.org/10.1021/ac303173d | Anal. Chem. 2013, 85, 1812−1818

Analytical Chemistry

Article

micrometric scale.7 The method is based on the contactless electrochemical interaction of a microelectrode with the interface to be analyzed, generally a solid−liquid interface. Until now, very few SECM studies have involved graphene. The pioneer works have investigated charge transfer8 and quantified the surface diffusion coefficient of a tripodal adsorbate on graphene,9 but, to the best of our knowledge, graphene oxide has never been investigated with SECM. In principle, SECM can be used to evaluate the local conductivity of the substrate without any solid−solid contact, as it is a solution containing a redox species that transports the current between the probe and the substrate. A model based on a first order kinetics of redox mediator regeneration at the electrolyte/substrate interface combined with electron transport in the substrate material has been previously introduced.10 Such a model has been used in a few studies, for example for the conductivity measurement of composite Langmuir-Shaefer films,11 polyaniline monolayers,12 gold nanocrystals ensembles,10c or porphyrin films.13 These studies are focused on the analysis of homogeneous substrates, through the record of the probe current during its approach to the substrate. On the contrary, the present study deals with heterogeneous substrates that are analyzed by mapping the feedback current at a given distance from the substrate. Conductive spots of finite size are thus investigated theoretically in order to determine the most appropriate conditions for quantitative conductivity mapping. In the present study, quantitative mapping of the electronic conductivity of reduced GO layers is performed with SECM. As verified, the charge transfer on reduced GO is not limiting when the usual redox mediator that is ferrocenedimethanol (FcMeOH) is used, so that it gets possible to simplify the original equation system, with a feedback response that depends on only one parameter, i.e. the dimensionless substrate conductivity. A deep investigation of the model results is performed as well as a discussion on the most appropriate conditions (microelectrode geometry, active spot size, redox mediator concentrations) for the evaluation of the local conductivity from feedback current measurements. Additionally, an analytical approximation of the conductivity as a function of the relevant parameters of the study is proposed. The present study thus provides a general and simplified interpretation framework that can be extended to the quantitative conductivity mapping of other types of materials than GO.

formation through the bubble deposition method has already been described in a previous report.14 It leads to extremely thin films of about 3−5 nm thickness composed of percolating GO flakes. In a typical procedure, GO films are heated (near 100 °C, during 1 min) and rinsed with ethanol and water in order to remove surfactant residues. The reduced GO substrates are obtained by exposition to a pH 13 aqueous solution (KOH 0.1M) at room temperature, for durations ranging from 30 s to 2 min. A thermal annealing (1 h at 150 °C, under vacuum) was performed after each exposition. Electrochemical Setup. SECM experiments were all performed on the Princeton Applied Research 370 SECM Workstation. A conventional three-electrode setup was used for the voltammetry and SECM experiments. It involved a platinum (Pt) microdisk working electrode, an Ag/AgCl electrode as reference, and a 0.5 mm diameter gold wire auxiliary electrode. 0.1 M KCl was used as supporting electrolyte. A VSP Biologic potentiostat was used for the electrolysis of the FcMeOH solutions in order to perform SECM experiments with a mixture of reduced (Red) and oxidized (Ox) mediator forms in solution. A Pt grid was used as working electrode, polarized at 0.4 V vs Ag/AgCl. It took about 1 h to transform half of the FcMeOH initially present in solution, as evidenced by cyclic voltamograms of the solution before and after the electrolysis (see section 8 of the Supporting Information). Numerical Calculations. All the numerical calculations have been performed using Comsol Multiphysics 4.2, installed on an Optiplex 720 Dell workstation, with 8 GB RAM, i7-2600CPU. See section 4 of the Supporting Information for more details on the calculation method.

■

RESULTS AND DISCUSSION Model. In SECM, the most widespread operating mode is based on the evaluation of the substrate regeneration rate of a redox mediator present in solution and consumed at the microelectrode. This situation, generally called the feedback mode,15 was considered in the present study. During an experiment, if the electrons are not brought by an external source (i.e., in an unbiased situation), the regeneration of the mediator underneath the tip is necessarily associated with a counter reaction at the same substrate. It occurs at a different place than that of the regeneration reaction, so that the electrons need to be carried through the substrate.10c In this way, this is similar to the galvanic coupling in corrosion,16 with a unique substrate that has anodic and cathodic zones. In the present study, the charge transfer at the film−solution interface is assumed to be fast and has thus a reversible and Nernstian behavior at any point of the interface

■

EXPERIMENTAL AND TECHNICAL SECTION Materials. Electrochemical measurements were performed in aqueous media (MiliQ water 18.2 MΩ),) and potassium chloride (KCl) as supporting electrolyte. Electrode fabrication required 25 μm diameter Pt-wire (Goodfellow, purity 99.9%), glass Pasteur pipettes, silver conducting paint (Electrolube), and standard copper connection wires (diameter 5 (this validity range strongly depends on Rg and to a lower extent on L). This surprisingly large range of validity is related to the fact that the counter reaction current density is almost entirely concentrated close to the edge of the active spot, so that the presence of the probe in its center is of little consequence (see more details in section 5 of the Supporting Information). In addition, Figure 2b illustrates the specificity of unbiased measurements: the feedback current depends significantly on the size of the active spot. The figure shows that in the case of a biased substrate it is not the case anymore, except for particularly small active spots, in accordance with the literature.20 In fact, the situation of an unbiased substrate is particularly problematic for conductivity imaging. Figure 2c shows simulated lateral scans of unbiased conductive spots of different sizes having a limited conductivity (σ* = 5). In this case again, the current depends on the spot size, which is in accordance with literature data.21 More importantly, it remains constant during a scan. As a consequence, contrary to reactivity heterogeneities, in the case of conductivity heterogeneities, a current that does not depend on the position of the tip is not a reliable indication that the response of the spot is the same as the one of an homogeneous substrate. This shows that the interpretation of SECM experiments for conductivity mapping is particularly problematic since in this case, the link between the current and the local conductivity strongly depends on the surrounding substrate properties, without any experimental warning such as a current variation during the scan. It is thus very important to clarify when it is possible to rely on the current measurements for the evaluation of the local conductivity. This is detailed in the next part. Optimization of the Experimental Conditions for an Accurate Conductivity Mapping. In this section, the most appropriate experimental conditions for the evaluation of the conductivity are reviewed in detail. We show how a judicious choice of the experimental conditions allows avoiding an inaccurate evaluation of the conductivity from the feedback current. For this, a current that depends as little as possible on the size of the active spot is clearly preferable, as in this case the link between the measured current and the deduced conductivity does not necessitate the precise consideration of the surrounding conductivity distribution. First of all, by changing N, the fraction of species reacting at the tip versus the total concentration in redox mediator, it is possible to decrease the impact of the spot size. Figure 3a presents the current as a function of the spot size for different N values. It shows that in the absence of Ox in solution, i.e. for N = 1, the feedback current strongly depends on Rs even for very large Rs values (Rs = 20): see for example the difference between the value for Rs = 20 and the corresponding dotted line for Rs = 100, representative of an homogeneous substrate. This particular situation comes from the fact that when the concentration at

Figure 2. (a): Schematic representation of the bipolar nature of the feedback process above an unbiased substrate. (b): Normalized current as a function of the active spot size, for a biased and unbiased substrate with infinite conductivity. The current as predicted by eq 5 is also presented (Rg = 2, L = 0.5, N = 1). (c): Normalized current during a lateral scan above unbiased substrates of different sizes, for L = 0.5, Rg = 2, N = 1, and σ* = 5.

situation bears some resemblances to the one investigated by Oleinick et al.18 In that study, the authors have specifically designed a substrate in order to separate the location of the two reactions, so that the diffusion processes do not interact with each other. The present part deals with a more conventional situation, where a circular conducting substrate is deposited on an insulator, as presented in Figure 2a. In this case, it is possible to derive an analytical approximate expression of the current. The detail of the calculations, valid when one can consider that the probe hardly disturbs the counter reaction process, is presented in section 5 of the Supporting Information. It leads to 1815

dx.doi.org/10.1021/ac303173d | Anal. Chem. 2013, 85, 1812−1818

Analytical Chemistry

Article

in real situations because the contribution of the surrounding substrate may lead to an overestimation of the local conductivity. In fact, the most favorable situation arises when the measured conductivity is approximately between σ* = 0.2 and σ* = 2. It is then advised to adjust the redox mediator concentration and eventually the microelectrode radius (see eq 3) in order to avoid extreme values of conductivities. In addition to σ* and N, the surrounding glass thickness of the probe (Rg) impacts the applicability of the infinite substrate model. Figure 3c shows the normalized current as a function of Rs for different Rg and highlights that the current is different from the one that is obtained at large Rs as soon as Rs is smaller than Rg + 4. For smaller conductive spots, the transport of reactive species used for the counter reaction at the substrate is hindered by the insulating part of the probe. Moreover it must be mentioned that the locations of the reactions occurring at the substrate (the feedback reaction and the counter reaction) are much more distant if Rg is large, as the separation distance of the reactions is typically equal to Rg. The distance over which the substrate is solicited for the electronic transport is thus larger for large Rg. This leads to the important variations of the current obtained at large Rs when different Rg are used (dotted lines of Figure 3c). These two effects, i.e. hindering of counter reaction and increase of the electron transport length, combine to give the very important effect of Rg as observed in Figure 3c. Experimentally, working with small Rg is then very preferable. Finally, L has also an impact on the results: the closer the probe, the higher the contrast between the zones. On the other hand, a too small value of L increases the risk of damaging the substrate by a contact between the probe and the substrate as well as the sensitivity of the measurements to any eventual substrate’s topological features. One can mention that L has a very limited impact on the conditions for which Rs impacts the results. To sum up, evaluation of the microscopic scale conductivity without having to consider the explicit size and geometry of the active spot is guaranteed for Rs > Rg + 4 and 0.2 < σ* < 2 if one uses a mix of Ox and Red species for the measurements. In an experimental procedure, it is then advised to adjust the microelectrode geometry and the redox mediator concentration in order to fulfill these conditions. To obtain a conductivity map from SECM measurements, an analytical expression of the conductivity as a function of the measured normalized current can be very helpful, as it suppresses the necessity of iterative numerical simulations in order to find the σ* value that leads to the appropriate probe current. For this purpose, using a step by step procedure similar to the one detailed in the literature,22 eq 6 has been constructed from the observation of numerous numerical simulation results (for N = 0.5).

Figure 3. Normalized current as a function of Rs for L = 0.5 and (a) different N, Rg = 2, σ* = 1. (b) N = 0.5, Rg = 2, different σ*. (c) N = 0.5, different Rg, σ* = 1. For each curve, the dotted line is the corresponding normalized current at Rs = 100 (representative of an homogeneous substrate).

the substrate is close to 1, a very small variation of the concentration induces a very large change in the electrical potential, as predicted by eq 1 (see section 6 of the Supporting Information for more details). Importantly, the impact of the spot size is much less important when equal amounts of Red and Ox are present in solution (N = 0.5): in this case, even for Rs = 5, the current is hardly different from the one obtained with larger spots. In fact, when N is close to 0.5, the electrical potential depends much less on the concentration variation, and the driving force for the regeneration process is then much less sensitive to a local change in the concentration at the substrate. As a matter of fact, experimentally, it is much more preferable to work with a solution that has a significant amount of both Red and Ox species, ideally in equal concentration. In addition to N, the impact of the active spot size on the feedback current depends on the conductivity of the spot, as presented in Figure 3b. It shows that when conductivity is large, the feedback current is importantly depending on Rs, even if this latter is large. This comes from the bipolar effect limitation, as discussed in the previous section. In addition, it must be mentioned that extreme values (large or low) of the conductivity lead to a feedback current that hardly depends on the conductivity, implying that the evaluation of this latter from the measurement will be very imprecise (see section 7 of the Supporting Information for more details). One should also add that measuring an exact small conductivity may be difficult

σ* = (0.138L + 0.0858) (Ni T 0.6 + 2.27L /Rg

0.146

+ 0.22ln(Rg) + 0.4)3.28 + 2.3L /Rg

− 0.1152Rg 0.2L2 − 0.165ln(Rg) − 0.0087Rg − 0.03

1.54

(6)

It is valid with 7% accuracy (when compared to the numerical simulation results) for 0.3 < L< 1.5, 2 < Rg < 20, and 0.1 < σ* < 5. Application to the Study of the Reduction of GO by Exposition to an Alkaline Solution. The previously introduced theoretical framework has been applied to the investigation of the reduction kinetics of GO layers deposited 1816

dx.doi.org/10.1021/ac303173d | Anal. Chem. 2013, 85, 1812−1818

Analytical Chemistry

Article

Figure 4. Area scans with a 12 μm radius probe (Rg = 3) and a probe-substrate distance during the scans of 12 μm (L = 1); before and after exposure of GO substrates to a pH = 13 solution. (a) and (b): before exposure, (a) tip current and (b) corresponding σe according to eq 6 and 3 (iT,sol = 0.43 nA). (c) and (d): after 30 s exposure, (c) tip current and (d) corresponding σe (iT,sol = 1.21 nA). The scanned zones are not the same in the two data sets.

four-point probe measurements on similar substrates lead to a value of 0.003 μS with 50% uncertainty, which is close to the average value presently obtained with SECM. The difference between the two methods probably comes from the inhomogeneity of the films that is unavoidable with the adopted substrate preparation procedure. This conductivity values for substrate that have not been exposed yet may come from a very slight reduction that occurred during the treatment of the films after GO deposition. As a matter of fact, this illustrates the sensitivity of the SECM for surface conductivity measurements. Figures 4c and 4d show that very mild conditions (pH 13, room temperature, followed by thermal annealing at 150 °C under vacuum) are sufficient to significantly increase the conductivity of the substrate, as the average σe after 30 s exposition is 0.03 μS, with about 50% variation over the substrate. The same substrate after 2 min exposure and annealing has a similar conductivity map (see section 9 of the Supporting Information), with an average value of 0.05 μS, showing that an important part of the reaction occurred within the first step. One can mention that in the present study, the measured conductivities (more precisely the product σe) were never close to the extreme values achievable with the technique. On the one hand, lower σe values could be measured, as the concentration of redox mediator could be further decreased without reaching the sensitivity limit of the current measurement. On the other hand, higher value of σe would have been accessible by an increase of the redox mediator concentration up to the solubility limit of the FcMeOH. The precise range depends on the experimental set up and conditions.

on glass, when exposed to an alkaline solution (KOH 0.1 M) and then annealed at 150 °C under vacuum. In the literature, heated (50−90 °C) alkaline solutions are already known to reduce GO, leading to an increase of its conductivity.4 Conversely, we performed the experiment at room temperature: Figure 4a and 4c show area scans before and after 30 s of exposition to a KOH solution (0.1 M) at room temperature. For each scan, the chosen experimental conditions lead to an optimal accuracy, in accordance with the previous theoretical prospection. In particular, the solutions used for the measurements contained both FcMeOH and FcMeOH+ (FcMeOH was produced by electrolysis of FcMeOH), the microelectrode had a small Rg (Rg = 3), and a more concentrated solution was used to map the more conductive situation in order to have σ* values that are mostly within the 0.2−2 range. Figure 4b and 4d present respectively the same data as Figure 4a and 4c, after conversion of the measured current into conductivity using eq 6 and 3. In eq 3, the value of the experimental current when the electrode is far from the substrate (iTsol) is used instead of the value of each individual parameter (i.e., σe = σ*iTsolnF/4βRpT). From the measurements, without the precise knowledge of the film thicknesses e, it is only possible to deduce the product of the conductivity with the thickness, σe. First, the figure shows that the surface is not perfectly homogeneous. This probably comes from an inhomogeneous distribution of the GO flakes over the surface during the processing, leading to areas where there are more overlapping flakes. Figures 4a and 4b show that the studied GO films present a small conductivity, even before exposition. An average value of 0.005 μS (or 2 × 108 Ohm/square) is obtained, with about 50% variations. One has to mention that 1817

dx.doi.org/10.1021/ac303173d | Anal. Chem. 2013, 85, 1812−1818

Analytical Chemistry

■

Article

(7) (a) Amemiya, S.; Bard, A. J.; Fan, F. R. F.; Mirkin, M. V.; Unwin, P. R. Annu. Rev. Anal. Chem. 2008, 1, 95. (b) Mirkin, M. V.; Nogala, W.; Velmurugan, J.; Wang, Y. X. Phys. Chem. Chem. Phys. 2011, 13, 21196. (8) (a) Xie, X. A.; Zhao, K. K.; Xu, X. D.; Zhao, W. B.; Liu, S. J.; Zhu, Z. W.; Li, M. X.; Shi, Z. J.; Shao, Y. H. J. Phys. Chem. C 2010, 114, 14243. (b) Güell, A. G.; Ebejer, N.; Snowden, M. E.; Macpherson, J. V.; Unwin, P. R. J. Am. Chem. Soc. 2012, 134, 7258. (c) Tan, C.; Rodriguez-Lopez, J.; Parks, J. J.; Ritzert, N. L.; Ralph, D. C.; Abruna, H. D. ACS Nano 2012, 6, 3070. (9) Rodríguez-López, J.; Ritzert, N. L.; Mann, J. A.; Tan, C.; Dichtel, W. R.; Abruña, H. D. J. Am. Chem. Soc. 2012, 134, 6224. (10) (a) Ruiz, V.; Liljeroth, P.; Quinn, B. M.; Kontturi, K. Nano Lett. 2003, 3, 1459. (b) Liljeroth, P.; Quinn, B. M.; Ruiz, V.; Kontturi, K. Chem. Commun. 2003, 1570. (c) Liljeroth, P.; Vanmaekelbergh, D.; Ruiz, V.; Kontturi, K.; Jiang, H.; Kauppinen, E.; Quinn, B. M. J. Am. Chem. Soc. 2004, 126, 7126. (11) Nicholson, P. G.; Ruiz, V.; Macpherson, J. V.; Unwin, P. R. Phys. Chem. Chem. Phys. 2006, 8, 5096. (12) Zhang, J.; Barker, A. L.; Mandler, D.; Unwin, P. R. J. Am. Chem. Soc. 2003, 125, 9312. (13) Leroux, Y.; Schaming, D.; Ruhlmann, L.; Hapiot, P. Langmuir 2010, 26, 14983. (14) Azevedo, J. l.; Costa-Coquelard, C.; Jegou, P.; Yu, T.; Benattar, J.-J. J. Phys. Chem. C 2011, 115, 14678. (15) Wipf, D. O.; Bard, A. J. J. Electrochem. Soc. 1991, 138, 469. (16) Lefrou, C.; Fabry, P.; Poignet, J.-C. Electrochemistry -The basics, with examples; Springer: 2012. (17) Sun, P.; Laforge, F. O.; Mirkin, M. V. Phys. Chem. Chem. Phys. 2007, 9, 802. (18) Oleinick, A. I.; Battistel, D.; Daniele, S.; Svir, I.; Amatore, C. Anal. Chem. 2011, 83, 4887. (19) Lefrou, C.; Cornut, R. ChemPhysChem 2010, 11, 547. (20) Fulian, Q.; Fisher, A. C.; Denuault, G. J. Phys. Chem. B 1999, 103, 4387. (21) Xiong, H.; Guo, J.; Amemiya, S. Anal. Chem. 2007, 79, 2735. (22) Cornut, R.; Lefrou, C. J. Electroanal. Chem. 2008, 621, 178.

CONCLUSION The present article introduces a rapid, very sensitive, contactless method to measure the conductivity with SECM. The method is particularly appropriate for low conductivity materials as proven by the investigation of graphene oxide films deposited on glass and reduced under very mild conditions (pH 13, room temperature, followed by thermal annealing at 150 °C under vacuum). In addition, the present study provides a consistent theoretical framework for the reliable quantification of the local conductivity with SECM. An analytical approximation of the conductivity as a function of the feedback current is proposed (eq 6), making any further interpretation step very straightforward, by avoiding the use of iterative numerical simulations. The method can be applied to a prospective analysis of the different GO reduction procedures, as well as to the conductivity characterization of other substrates. Contrary to typical four-point probe conductivity measurements that only provide average conductivity values at a millimeter scale, SECM proves to be very powerful to reveal micrometric conductivity inhomogeneities, the impact of which can be of critical importance for an effective integration of reduced GO films in practical devices.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: + 33 169 088332. Fax: + 33 169 086462. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS C. Zobrist is acknowledged for technical assistance with the construction of the TOC. J. Azevedo acknowledges funding from the DGA. C. Bourdillon and R. Cornut acknowledge funding from ANR-PDOC20011-Copel.

■

REFERENCES

(1) (a) He, Q.; Sudibya, H. G.; Yin, Z.; Wu, S.; Li, H.; Boey, F.; Huang, W.; Chen, P.; Zhang, H. ACS Nano 2010, 4, 3201. (b) Park, S.; Ruoff, R. S. Nat. Nanotechnol. 2009, 4, 217. (2) (a) Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Alfred, K.; Yuanyuan, J.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S. Carbon 2007, 45, 1558. (b) Shin, H.-J.; Kim, K. K.; Benayad, A.; Yoon, S.-M.; Park, H. K.; Jung, I.-S.; Jin, M. H.; Jeong, H.-K.; Kim, J. M.; Choi, J.-Y.; Lee, Y. H. Adv. Funct. Mater. 2009, 19, 1987. (c) Moon, I. K.; Lee, J.; Ruoff, R. S.; Lee, H. Nat. Commun. 2010, DOI: doi:10.1038/ ncomms1067. (d) Pei, S.; Cheng, H.-M. Carbon 2012, 50, 3210. (3) Mao, S.; Pu, H.; Chen, J. RSC Adv. 2012, 2, 2643. (4) Fan, X.; Peng, W.; Li, Y.; Li, X.; Wang, S.; Zhang, G.; Zhang, F. Adv. Mater. 2008, 20, 4490. (5) Zheng, Q.; Ip, W. H.; Lin, X.; Yousefi, N.; Yeung, K. K.; Li, Z.; Kim, J.-K. ACS Nano 2011, 5, 6039. (6) (a) Loiacono, M. J.; Granstrom, E. L.; Frisbie, C. D. J. Phys. Chem. B 1998, 102, 1679. (b) Reid, O. G.; Munechika, K.; Ginger, D. S. Nano Lett. 2008, 8, 1602. (c) Pingree, L. S. C.; Reid, O. G.; Ginger, D. S. Nano Lett. 2009, 9, 2946. (d) MacDonald, G. A.; Veneman, P. A.; Placencia, D.; Armstrong, N. R. ACS Nano 2012, 6, 9623−9636. (e) Hauquier, F.; Alamarguy, D.; Viel, P.; Noel, S.; Filoramo, A.; Huc, V.; Houze, F.; Palacin, S. Appl. Surf. Sci. 2012, 258, 2920. 1818

dx.doi.org/10.1021/ac303173d | Anal. Chem. 2013, 85, 1812−1818