Scanning Electrochemical Microscopy of Carbon Nanomaterials and

Sep 7, 2016 - Published as part of the Accounts of Chemical Research special issue “Nanoelectrochemistry”. Biography. Shigeru Amemiya attended the...
8 downloads 17 Views 4MB Size
Download PDF
Article pubs.acs.org/accounts

Scanning Electrochemical Microscopy of Carbon Nanomaterials and Graphite Published as part of the Accounts of Chemical Research special issue “Nanoelectrochemistry”. Shigeru Amemiya,* Ran Chen, Nikoloz Nioradze, and Jiyeon Kim Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States CONSPECTUS: Carbon materials are tremendously important as electrode materials in both fundamental and applied electrochemistry. Recently, significant attention has been given not only to traditional carbon materials, but also to carbon nanomaterials for various electrochemical applications in energy conversion and storage as well as sensing. Importantly, many of these applications require fast electron-transfer (ET) reactions between a carbon surface and a redox-active molecule in solution. It, however, has not been well understood how heterogeneous ET kinetics at a carbon/solution interface is affected by the electronic structure, defect, and contamination of the carbon surface. Problematically, it is highly challenging to measure the intrinsic electrochemical reactivity of a carbon surface, which is readily passivated by adventitious organic contaminants. This Account summarizes our recent studies of carbon nanomaterials and graphite by scanning electrochemical microscopy (SECM) not only to reveal the fast ET kinetics of simple ferrocene derivatives at their graphitic surfaces, but also to obtain mechanistic insights into their extraordinary electrochemical reactivity. Specifically, we implemented new principles and technologies to reliably and reproducibly enable nanoscale SECM measurements. We took advantage of a new SECM imaging principle to resolve the high reactivity of the sidewall of individual single walled carbon nanotubes. In addition, we developed SECM-based nanogap voltammetry to find that monolayer graphene grown by chemical vapor deposition yields an unprecedentedly high standard ET rate constant, k0, of ≥25 cm/s, which was >1000 times higher than that reported in the literature. Remarkably, the nonideal asymmetry of paired nanogap voltammograms revealed that the high reactivity of graphitic surfaces is compromised by their contamination with airborne hydrocarbons. Most recently, we protected the clean surface of highly oriented pyrolytic graphite from the airborne contaminants during its exfoliation and handling by forming a water adlayer to obtain a reliable k0 value of ≥12 cm/s from symmetric pairs of nanogap voltammograms. We envision that SECM of clean graphitic surfaces will enable us to reliably address not only effects of their electronic structures on their electrochemical reactivity, but also the activity of carbon-based or carbon-supported electrocatalysts for fuel cells and batteries.

■

INTRODUCTION A greater mechanistic understanding of the electrochemical reactivity of carbon nanomaterials as well as graphite as their model is important not only for electrochemical energy and sensing technologies, but also for fundamental electrochemistry.1 Practically, the adsorbing graphitic surfaces are highly attractive as electrocatalysts or supports of electrocatalysts for fuel cells and batteries to accelerate inner-sphere electron-transfer (ET) reactions2 (e.g., the oxygen reduction reaction), which require a direct interaction of a reactant (or a product) with an electrode surface, i.e., specific adsorption3 (Figure 1A). The electrochemical reactivity of graphitic surfaces, however, is not well

understood mechanistically even for simple outer-sphere ET reactions,2 where both reactant and product are located at the outer Helmholtz plane to electronically interact with an electrode surface through a solvent layer3 (Figure 1B). In the framework of ET theories by Marcus4 and Hush,5 the mechanism of a heterogeneous outer-sphere ET reaction varies from the nonadiabatic regime to the adiabatic regime as electronic coupling between an electrode surface and a reactant becomes stronger.6 On one hand, a nonadiabatic ET process is kinetically controlled by the probability of electron tunneling between the electrode surface and the reactant, thereby depending on their density of state (DOS). Current theories assume nonadiabatic ET processes for single-walled carbon nanotubes (SWCNTs),7 graphene,7 and highly oriented pyrolytic graphite (HOPG)8 to imply that the outer-sphere ET kinetics of the graphitic materials is strongly dependent on their unique electronic structures and, subsequently, is slower than that of metals with higher DOS (e.g., platinum and gold). On the other hand, recent experimental studies demonstrated that the

Figure 1. (A) Inner-sphere and (B) outer-sphere reactions of adsorbed and nonadsorbed redox-active molecules, respectively. © XXXX American Chemical Society

Received: June 28, 2016

A

DOI: 10.1021/acs.accounts.6b00323 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research reactivity of these graphitic materials to outer-sphere ET reactions is similar to that of the metals as discussed in this Account, thereby suggesting the adiabatic ET mechanism that is kinetically limited by the reorganization of solvated reactant prior to electron tunneling. It, however, has been argued whether the “metal-like” reactivity of the graphitic surfaces is intrinsic or due to defects.9 Alternatively, the reactants of the “outer-sphere” ET reactions may be specifically adsorbed on the electrode surface to undergo a faster inner-sphere pathway with a shorter tunneling distance.3 Herein, we summarize our experimental studies of ET mechanisms at carbon nanomaterials and graphite by scanning electrochemical microscopy10 (SECM), which advantageously enabled us to assess not only the high reactivity of the graphitic surfaces, but also their cleanliness. Specifically, we quantitatively demonstrate the fast ET kinetics of simple ferrocene derivatives at the sidewall of individual SWCNTs,11 monolayer graphene grown by chemical vapor deposition (CVD),12 and HOPG,13 whereas the airborne contamination of the graphitic surfaces14 compromises their reactivity.15 We ensure that the more reliable reactivity of the cleaner HOPG surfaces protected from airborne hydrocarbons16 is higher than not only the reactivity of airexposed HOPG surfaces as studied by scanning electrochemical cell microscopy,17 but also the reactivity of platinum and gold nanoelectrodes,18,19 which are readily contaminated even in ultrapure water.20 Importantly, we experimentally confirm the outer-sphere character of ferrocene-based redox couples at graphene and HOPG surfaces to assess the adiabaticity of the corresponding ET reactions.16 We envision that future nanoscale SECM studies of clean graphite and metal surfaces are required not only to unambiguously answer fundamental questions about outer-sphere ET mechanisms, but also to reproducibly and quantitatively study electrocatalytic inner-sphere ET reactions, which are highly amenable to surface contamination.21

Figure 2. (A) Setup of nanoscale SECM, which is detailed in refs 25 and 28. Parts (B) and (C) show SEM images of a 100 nm diameter platinum tip before and after ESD damage, respectively. Adapted with permission from ref 29. Copyright 2013 American Chemical Society.

■

evaporation of the solvent (water), which causes a temperature drift, but also to prevent the airborne contamination of the solution and, subsequently, tip and substrate surfaces. In addition, we found that submicrometer- and nanometer-sized platinum29 and carbon31 tips were readily damaged by electrostatic discharge (ESD) upon their contact with an operator (Figure 2B and C before and after ESD damage) as well as by a transient current flow from the saturated amplifier of a bipotentiostat. ESD damage was prevented when a tip electrode was handled by an operator with ESD protections29 under humidified conditions (>50%).32 The prevention of electrochemical tip damage required the substantial modification of both hardware and software of a commercial bipotentiostat.16,25 Traditionally, SECM has been operated in the feedback mode,33 where a constant potential is applied to a tip for the diffusion-limited electrolysis of the redox species that is originally present in the solution, e.g., R → O + e at the tip in Figure 3A.

NANOSCALE SECM Our studies of carbon nanomaterials and graphite were enabled by nanoscale SECM,22 which has been developed by the Bard group23−25 and was recently reinforced by our group through the development of new operation principles26,27 and new enabling technologies.28,29 Noticeably, micrometer-scale SECM has been also used for studies of various carbon nanomaterials and graphite as reviewed in a recent chapter.30 Nanoscale SECM is a scanning probe microscopy technique, where the micrometer- or nanometer-sized tip of a small electrode (ultramicroelectrode) is moved over a target substrate (Figure 2A) to investigate an ET reaction at the tip or the local substrate surface under the tip. In SECM-based imaging, the tip current is measured to study the local substrate reactivity while the tip is scanned laterally with nanometer resolution by using piezoelectric positioners. Alternatively, the tip position is fixed at a nanometer-distance from a substrate to voltammetrically investigate the fast ET kinetics of the tip or the substrate under high mass-transport conditions across the tip−substrate nanogap. Technologically, extra cares are needed to reproducibly and reliably enable nanoscale SECM measurements. Specifically, we developed an isothermal chamber based on thermally insulated plates and aluminum heat sinks (Figure 2A) to suppress the nanoscale drift of the tip position owing to the thermal expansion or contraction of the SECM stage.28 Moreover, the electrochemical cell was sealed with a flexible rubber cap without restricting the tip movement not only to suppress the

Figure 3. (A) Feedback and (B) SG/TC modes of SECM. B

DOI: 10.1021/acs.accounts.6b00323 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Accounts of Chemical Research

■

SECM IMAGING OF INDIVIDUAL SWCNTs We found the high sensitivity of SECM to the electrochemical reactivity of individual one-dimensional nanostructures,26,34 which enabled us to image the high reactivity of the sidewall of individual SWCNTs.11 Significantly, this result unambiguously disproved the presumption that the electrochemical activity of a carbon nanotube is ascribed to defected open ends, not to the basal-plane-graphite-like sidewall.35 Specifically, ultralong SWCNTs (∼2 mm in length) were grown on the insulating SiO2 surface by CVD and imaged by employing the SECM feedback mode, where (ferrocenylmethyl)trimethylammonium (FcTMA+) was initially present in the solution and was oxidized at the tip (Figure 4A). When the tip was positioned above a

When the tip is far away from the substrate, the corresponding steady-state tip current is given by

i T, ∞ = 4xnFDR c0a

(1)

where x is a function of glass geometry including RG (= rg/a; rg and a are outer and inner tip radii, respectively; see Figure 3A), n is the number of electrons transferred for the oxidation of one molecule of R, and DR and c0 are its diffusion coefficient and concentration in the bulk solution, respectively. Importantly, the amperometric tip current is significantly enhanced in the feedback mode when the tip is positioned within the tip diameter from the substrate (i.e., d < 2a in Figure 3A), where a significant fraction of the tip generated species, O, is transported across the gap and reduced at the substrate surface to regenerate species R. Quantitatively, the enhancement of the tip current is a measure of the reduction rate constant at the substrate, kred, which is given by the Butler−Volmer model as ⎡ αF(E − E 0 ′) ⎤ ⎥ k red = k0 exp⎢ − RT ⎣ ⎦

Article

(2)

where k0 is the standard ET rate constant, α is transfer coefficient, E is the substrate potential, and E0′ is the formal potential of the redox couple. While k0 serves as an intrinsic measure of the ET kinetics, an α value of 0.5 is expected for a simple outer-sphere ET reaction.4 Recently, we developed nanogap voltammetry as the new operation principle based on the combination of nanoscale SECM with cyclic voltammetry to investigate the ultrafast ET kinetics of a substrate under extremely high mass transport conditions across a tip−substrate nanogap.27 Specifically, a pair of nanogap voltammograms is obtained by studying the reduction rate at the substrate surface in the feedback mode (Figure 3A) and the oxidation rate in the substrate generation/ tip collection (SG/TC) mode (Figure 3B) to determine kinetic, thermodynamic, and transport parameters with minimum uncertainties. In the feedback mode of nanogap voltammetry, the substrate potential is cycled to investigate the dependence of the amperometric tip current on the substrate potential within a wide range across the formal potential of a target redox couple. The resultant nanogap voltammogram based on the plot of the tip current against the substrate potential is analyzed to

Figure 4. (A) SECM imaging of an individual SWCNT and (B) an actual image as obtained with a 10 μm-diameter platinum tip in 0.3 mM FcTMA+ and 0.1 M KNO3. Adapted with permission from ref 11. Copyright 2010 American Chemical Society.

separately determine the E0′, k0, and α values included in the reduction rate constant (eq 2). By contrast, the oxidation rate at the substrate surface is investigated in the SG/TC mode by setting the tip potential negative enough to detect species O, which is oxidatively formed from the original species R at the substrate surface (Figure 3B). The resultant plot of the tip current against the substrate potential enables the assessment of the potential dependence of the oxidation rate constant, kox, at the substrate as given by ⎡ (1 − α)F(E − E 0 ′) ⎤ ⎥ kox = k0 exp⎢ RT ⎣ ⎦

SWCNT, the tip current was enhanced by the regeneration of FcTMA+ from tip-generated FcTMA2+ at the sidewall surface. By contrast, no FcTMA+ was regenerated at the surrounding SiO2 surface, thereby yielding the contrast of an image when the tip current was plotted against the lateral tip position (Figure 4B). Importantly, SWCNTs were extremely thin (∼1.6 nm in diameter), but were very long in comparison with the diameter of an SECM tip (10 μm). Subsequently, a detectable amount of FcTMA+ was regenerated at the sidewall of an individual SWCNT, which covered only ∼0.05% of the substrate surface under the SECM tip. In fact, the SECM image of the SWCNT was much broader than its actual diameter, because the image resolution was limited by the tip diameter. Conveniently, the SECM image was obtained without externally biasing the

(3)

Ideally, the same E0′, k0, and α values in eqs 2 and 3 yield a pair of symmetric nanogap voltammograms in feedback and SG/TC modes. Practically, the symmetry of paired nanogap voltammograms depends on the cleanliness of a substrate surface.13,16 C

DOI: 10.1021/acs.accounts.6b00323 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research SWCNT, where FcTMA2+ was continuously reduced under the tip by providing electrons based on the FcTMA+ oxidation at the sidewall of the SWCNT far away from the tip (Figure 4A), thereby eliminating the need of an electrode attached to SWCNTs for external bias to simplify sample preparation. This unbiased approach, however, did not allow us to control the SWCNT potential, which was required to investigate the potential dependence of ET rates (see eqs 2 and 3). A readily detectable feedback response from individual SWCNTs (Figure 4B) indicates the high reactivity of their sidewall, which corresponds to a high ET rate constant of ≥4.6 cm/s for the FcTMA2+ reduction on the sidewall surface.11 This result was consistent with the result from voltammetry of ultramicroelectrodes based on the sidewalls of individual SWCNTs,36 which mediated the fast ET reaction of the FcTMA2+/+ couple without significant contributions from open tube ends or catalytic nanoparticles attached to tube ends for CVD growth. Our kred value, however, was significantly smaller than a kred value expected from eq 2 with k0 = 7.62 cm/s as determined for the sidewall of individual SWCNT ultramicroelectrodes.36 We attributed the low kred value of our SWCNT samples to the airborne contamination of their surfaces during long-term storage in ambient air after their growth. In fact, it was pointed out that surface contamination caused variations of voltammograms at individual SWCNT ultramicroelectrodes from those with high k0 values of 7.62 cm/s to those with no or poor response.36 Recently, the sidewalls of individual SWCNTs were also imaged by employing scanning electrochemical cell microscopy to demonstrate high and diminished reactivity of conductive and semiconductive nanotubes, respectively, to the Ru(NH3)3+/2+ couple.37 These SWCNTs, however, were exposed to airborne contaminants during the whole imaging experiments without ensuring the cleanliness of their sidewall surfaces, which is required to unambiguously assess the effects of their electronic structure on their reactivity.

Figure 5. Fabrication of a polymer-supported CVD-grown graphene electrode. Adapted with permission from ref 12. Copyright 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

■

ULTRAFAST ET KINETICS OF CVD-GROWN GRAPHENE We employed SECM-based nanogap voltammetry to find that the electrochemical reactivity of graphene grown by CVD was at least ∼1000 times higher than that reported in the literature.12 Our study indicated that the low reactivity of previous CVDgrown graphene electrodes was due to their coverage with the residue of poly(methyl methacrylate) (PMMA), which was coated on the graphene surface when graphene was transferred from a metal growth substrate (copper or nickel) to a target substrate for electrode fabrication.38−40 In fact, it has been recognized as a serious problem by the graphene community that the conventional lift-off procedure of PMMA from the graphene surface with such solvents as acetone38 leaves a continuous film of polymer residues, through which electron tunneling is prevented.41 In contrast to the PMMA-covered electrodes, our CVD-grown graphene electrodes were only supported with PMMA, because graphene was grown on a copper foil by CVD and coated with a thin PMMA film before the PMMA-free graphene surface was exposed as an active electrode surface by etching the copper foil (Figure 5). Nanogap voltammetry of PMMA-supported graphene electrodes in the SG/TC mode ensured their high reactivity for the oxidation of ferrocenemethanol (FcMeOH) (Figure 6A). The anodic voltammograms were fitted very well with theoretical voltammograms to yield k0 values of 1.6 cm/s and normal α

Figure 6. Nanogap voltammograms of 0.5 mM FcMeOH in 1 M KCl at (A) PMMA- and (B) PS-supported CVD-grown graphene as obtained with 0.98 μm-diameter platinum tips. Forward and reverse waves are represented by solid and dashed lines, respectively. Dots represent theoretical voltammograms for (A) kinetically limited and (B) reversible cases. The graphene potential is defined against the formal potential of the FcMeOH+/0 couple. Adapted with permission from ref 12. Copyright 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. D

DOI: 10.1021/acs.accounts.6b00323 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

the substrate surface. In our hypothesis, the tip current of the FcMeOH+/0 couple at the PS-supported graphene surface was significantly lower in the feedback mode (Figure 6B), where charged FcMeOH+ must be transported through the hydrophobic contaminant layer to be reduced at the underling graphene surface (Figure 7A). By contrast, less hydrophilic

values of 0.5 with various widths of 49−364 nm for the tip− graphene gap. These k0 values were 2 orders of magnitude higher than k0 values of 0.02 cm/s for the FcMeOH+/0 couple at PMMA-covered graphene electrodes.39 By contrast, cathodic nanogap voltammograms in the feedback mode yielded not only lower k0 values of 0.5 cm/s, but also lower α values of 0.29 for the reduction of FcMeOH+ at PMMA-supported graphene electrodes (Figure 6A). Different k0 and α values for oxidation and reduction of the same redox couple are anomalous (see eqs 2 and 3). We ascribed this anomalous behavior to a double layer effect from the supporting PMMA film on the FcMeOH+/0 couple at the opposite side of graphene (insets of Figure 6A). Specifically, the surface electrons of PMMA (i.e., crypotoelectrons42) are oxidatively removed by the graphene electrode at potentials around the formal potential of the FcMeOH+/0 couple to generate excess positive charges. Subsequently, the access of positively charged FcMeOH+ to the graphene surface is prevented, thereby broadening nanogap voltammograms based on the FcMeOH+ reduction in the feedback mode. By contrast, the underlying positive charges of the PMMA support does not affect the access of neutral FcMeOH, thereby yielding theoretically expected α values of 0.5 for the FcMeOH oxidation in the SG/TC mode. These double layer effects are exerted from the PMMA film through graphene, which is atomically thin. Interestingly, we found that the ET kinetics of atomically thin graphene strongly depends on the supporting polymer material, which we call the electrochemical transparency of graphene.12 Specifically, we studied the FcMeOH+/0 couple at graphene supported by a polystyrene (PS) film to yield reversible nanogap voltammograms at various gap widths of 31−432 nm (Figure 6B) in contrast to those limited by ET kinetics at PMMA-supported graphene electrodes (Figure 6A). The slower ET kinetics of the latter is ascribed to the excess positive charges of the underlying PMMA surface (insets of Figure 6A), which affect the electronic coupling of redox species with atomically thin graphene as well as the reorganization of the redox species near the graphene surface. By contrast, no excess charge is expected on the underlying PS surface, where the oxidative removal of cryptoelectrons requires much more positive potentials (∼0.7 V43) than applied during nanogap voltammetry. In fact, Raman spectroscopy and conductivity measurement revealed p-type doping of graphene coated with PMMA, whereas no noticeable doping effect was exerted from PS.44 Remarkably, reversible nanogap voltammograms at PS-supported graphene electrodes in the SG/TC mode yielded an extremely high k0 value of ≥25 cm/s, which is the diffusion-limited minimum given by k0 ≥ 10DR/d. This k0 value is 3 orders of magnitude higher than reported for the FcMeOH+/0 couple at PMMA-covered graphene electrodes.39 Moreover, nanogap voltammograms in the feedback mode were not broadened with PS support (Figure 6B) in contrast to those with PMMA support (Figure 6A), which ensures that the broadening was due to a double layer effect from the surface charges of PMMA support. The tip current at PS-supported graphene, however, was much lower in the feedback mode than in the SG/TC mode, which was ascribed to the airborne contamination of the graphene surface14 during electrode fabrication as discussed in the next section.

Figure 7. Scheme of SECM-based nanogap voltammetry at the contaminated graphene surface in (A) feedback and (B) SG/TC modes. Fc and Fc+ represent FcMeOH and FcMeOH+, respectively. Adapted with permission from ref 16. Copyright 2016 American Chemical Society.

FcMeOH can be more readily transported through the contaminant layer and oxidized at the graphene surface to enhance the tip current in the SG/TC mode (Figure 7B). In fact, contact angle and ellipsometry measurements indicated that the whole graphene surface can be covered with airborne hydrocarbons within 20 min14 when the surface is exposed to ambient air during electrode fabrication (Figure 5). Pairs of more symmetric nanogap voltammograms were obtained for the FcTMA2+/+ couple at HOPG, whereas the tip current was higher by up to 25% in the SG/TC mode than in the feedback mode at various gap widths of 36−271 nm as estimated from anodic voltammograms (Figure 8A).13 We ascribe the more ideal behavior of HOPG to its shorter exposition to ambient air before the exfoliated surface was protected from airborne contamination in the electrolyte solution of a sealed electrochemical cell. We were able to obtain pairs of symmetric nanogap voltammograms of the FcTMA2+/+ couple by using cleaner HOPG surfaces,16 which were protected from airborne hydrocarbon contaminants by a nanometer-thick water adlayer.45 The protecting water layer was formed by exfoliating HOPG in humidified air with dry ice. Contact angle and ellipsometry measurements confirmed that airborne contamination of waterprotected HOPG was reduced to