Focused-Ion-Beam-Milled Carbon Nanoelectrodes for Scanning Electrochemical Microscopy

Ru(NH₃)₆Cl₃ was obtained from Strem Chemicals (Newburyport, MA). SiO₂-coated silicon wafers were obtained from Graphene Laboratories (Calverton, NY). Silicon wafers coated with a 5.0 nm-thick titanium adhesion layer and then with a 100 nm-thick gold layer were obtained from Platypus Technologies (Madison, WI). All sample solutions were prepared by using ultrapure water with total organic carbon (TOC) of ≤1 ppb,³⁵ which was obtained by passing the final product of the Milli-Q Advantage A10 system (EMD Millipore, Billerica, MA) through a specific activated-carbon filter (VOC Pak, EMD Millipore). The Milli-Q water purification system was equipped with a Q-Gard T1 pack and a Quantum TEX cartridge (EMD Millipore) in order to produce ultrapure water with 18.2 MΩ·cm and TOC of 3 ppb. The Milli-Q system was fed with purified tap water (15.0 MΩ·cm) as obtained by using the Elix 3 Advantage system (EMD Millipore).

A nanopipet was heat-pulled from a quartz capillary (1.0 mm outer diameter and 0.7 mm inner diameter, Sutter Instrument, Novato, CA) and filled with carbon by CVD to yield a slightly recessed tip.²⁴ Specifically, a nanopipet with a tip diameter of 10–100 nm was pulled by using a program based on HEAT = 800, FIL = 4, VEL = 22, DEL = 128, PUL = 110 and HEAT = 830, FIL = 3, VEL = 17, DEL = 130, PUL = 255. A nanopipet was nearly completely filled with carbon deposited from methane in argon (1:1 ratio) for 1 hour at 900 °C. A copper nickel wire (0.13 mm diameter, Alfa Aesar, Ward Hill, PA) was used to establish a connection with a carbon nanotip for electrochemical measurements as well as its grounding to a sample stage in TEM, scanning electron microscopy (SEM), and FIB experiments and for protections from electrostatic damage. A recessed tip was milled by using an FIB instrument (SMI3050SE FIB-SEM, Seiko Instruments, Chiba, Japan)^28,29 to yield a flat tip as confirmed by TEM (JEM-2100, JEOL USA, Peabody, MA).³⁶ The beam of gallium ion (30 keV and 10 pA) was focused at a carbon nanotip for ∼3 second to mill the tip end, whereas the prior adjustment of the FIB condition took several minutes. A carbon nanotip was also characterized by SEM with the dual-beam FIB instrument before and after milling.

A homebuilt SECM instrument³⁷ was used for electrochemical measurements with CVD carbon nanotips. A patch-clamp amplifier (Chem-Clamp, Dagan, Minneapolis, MN) was used as a two-electrode potentiostat to prevent the electrochemical damage of a nanotip.³⁰ A Ag/AgCl wire was used as a counter/reference electrode. SiO₂- and Au-coated silicon wafers were cleaned in piranha solution (a 1:3 mixture of 30% H₂O₂ and 95.0–98.0% H₂SO₄) for 30 minutes and in Milli-Q water for 15 minutes (3 times), and immediately immersed into the electrolyte solution contained in an SECM cell (Caution: piranha solution reacts violently with organics and should be handled with extreme care!). The electrolyte solution was prepared by dissolving Ru(NH₃)₆Cl₃ in phosphate buffer solution (PBS) at pH 7.4 containing 0.137 M NaCl, 0.0027 M KCl, 0.01 M Na₂HPO₄, and 0.0018 M KH₂PO₄. The SECM cell was cleaned and sealed using a rubber cap and silicon gaskets in order to prevent the contamination of the electrolyte solution with airborne organic impurities.³⁵ When a tip was attached to the SECM stage, the perpendicular alignment of the tip's axis with respect to the substrate surface was confirmed within ±0.5° by using a digital angle gauge.

CVD carbon nanoelectrodes were characterized by TEM, SEM, and FIB imaging before and after FIB milling. We selected a CVD condition to obtain a recessed tip without the formation of a conductive carbon film on the outer tip wall.²⁴ The high-resolution TEM image of an unmilled carbon nanotip confirmed the ∼70 nm-depth recession of CVD carbon from the ∼20 nm-radius orifice of the heat-pulled quartz pipet (Figure 2A). The hemispherical tip of CVD carbon had a base radius of ∼25 nm and was surrounded by a ∼10 nm-thick quartz wall, which corresponds to an RG value of ∼1.4 as expected from the corresponding RG value of quartz capillaries. A recessed carbon nanotip was milled by FIB and imaged by TEM to confirm that the tip was flat and completely filled with CVD carbon to yield an outer tip radius of ∼75 nm (Figure 2B). The excellent smoothness of a FIB-milled tip was achieved by employing a low current gallium ion (Ga⁺) beam (10 pA at 30 keV). Noticeably, TEM images showed that the edge of the milled tips was rounded, which is ascribed to the Gaussian broadening of the focused Ga⁺ beam.³⁸

Zoom In Zoom Out Reset image size

The dual-beam FIB/SEM instrument allowed us to visualize the tip of a carbon nanoelectrode by FIB and SEM imaging before and after milling. The resolution of an SEM image was high enough to ensure the intact tip end prior to milling (Figure 3A) when the tip was protected from electrostatic damage (see below). The tip end of an FIB-milled carbon nanotip was also clearly seen by SEM, which enabled us to estimate the outer radius of each milled tip. This information was useful for the analysis of electrochemical data, which depend not only on the radius of a carbon tip, a, but also on the outer radius of the quartz sheath, r_g ( = ∼50 nm in Figure 3B). The outer radius of the smallest FIB-milled tip was practically limited to ∼50 nm by the resolution of FIB imaging, which was needed to locate the tip position prior to milling (Figure 3C). Accordingly, the Ga⁺ beam was focused slightly above the 50 nm-radius portion of the unmilled quartz nanopipet to yield a milled nanotip with an outer radius of ∼50 nm as checked by FIB and SEM imaging (Figures 3B and 3D, respectively). Noticeably, the blurriness of SEM and FIB images is due to the charging of the insulating quartz surface.

Zoom In Zoom Out Reset image size

FIB-milled carbon nanoelectrodes were characterized by cyclic voltammetry (CV) to demonstrate their high electrochemical reactivity to the reduction of Ru(NH₃)₆³⁺, which yielded nearly reversible CVs with well-defined limiting currents, i_T,∞. The inner radius of a nanotip (a = 49 and 29 nm in Figures 4A and 4B, respectively) was determined from the limiting current by using Eq. 4 with D = 6.7 × 10^–6 cm²/s for Ru(NH₃)₆³⁺.³⁹ In addition, this analysis employed a simulated x value of 1.13 in Eq. 4 for an insulating outer wall with RG and θ values as determined by TEM. Thus, an inner tip radius determined voltammetrically gave an outer radius that is consistent with the outer radius estimated by SEM. Noticeably, we employed SECM to confirm the size and geometry of an FIB-milled carbon nanotip as well as a lack of a conductive carbon film on the outer wall (see below).

Zoom In Zoom Out Reset image size

Nearly reversible CVs of Ru(NH₃)₆³⁺ were analyzed to yield extremely high standard electron-transfer rate constants, k⁰, of ≥10 cm/s by using four carbon nanoelectrodes with tip radii of <50 nm. For instance, the slightly quasi-reversible CV of Ru(NH₃)₆³⁺ at the 49 nm-radius carbon electrode fitted well with a simulated CV with a k⁰ value of 11 cm/s (i.e., k⁰a/D = 8) and a transfer coefficient, α, of 0.5. On the other hand, the reversible CV of Ru(NH₃)₆³⁺ at the 29 nm-radius carbon electrode yielded an even larger k⁰ value of ≥23 cm/s (i.e., k⁰a/D ≥ 10). These k⁰ values for Ru(NH₃)₆³⁺ at FIB-milled carbon nanoelectrodes are close to k⁰ values of 17 cm/s at platinum nanoelectrodes,⁴⁰ 13 cm/s at gold nanoelectrodes,⁴¹ and 4 cm/s at individual single-walled carbon nanotubes.⁴² Similar k⁰ values for Ru(NH₃)₆³⁺ among different electrode materials have been considered as evidence of adiabatic electron-transfer reactions.^43,44 We, however, emphasize that nanoelectrodes can be readily contaminated with adventitious impurities from ultrapure water⁴⁵ and ambient air, which may apparently limit k⁰ values to the similar values.³⁵ Moreover, the electron-transfer kinetics of Ru(NH₃)₆³⁺ at FIB-milled carbon nanotips may be affected by the implantation of gallium ion on the carbon surface.³⁸ In addition to these uncontrollable contaminants, the amorphous form of CVD carbon might have caused differences among the k⁰ values obtained by using different FIB-milled carbon nanoelectrodes (Figure 4).

The well-defined geometry of FIB-milled carbon nanotips was confirmed by SECM approach curves at insulating substrates, where negative feedback responses are sensitive to the thickness of the insulating sheath^46,47 as well as the conductivity of the outer wall (see the Theory section). Experimentally, Ru(NH₃)₆³⁺ was reduced at a carbon nanotip at a diffusion-limited rate while the tip approached a SiO₂-coated silicon wafer as a flat insulating substrate. Figure 5A shows the resultant approach curve as obtained using a relatively large carbon nanoelectrode (a = 119 nm). The normalized experimental approach curve fitted well with a simulated approach curve for a tip with an insulating outer wall, but not with a curve for a tip with a conductive outer wall. The good fit was obtained by using an RG value of 1.4 as estimated by TEM. The theoretical analysis also yielded a very short distance of the closest approach, d_c, (i.e., the closest tip–substrate distance where the experimental curve fits with the theoretical curve) to be 8 nm, which confirms the flat tip end (Figure 2B). By contrast, much larger d_c values of 50–150 nm were reported for pyrolytic carbon nanoelectrodes with a similar size (a = 120–150 nm and RG = 1.5) at insulating substrates.^14,15

Zoom In Zoom Out Reset image size

Remarkably, good fits between experimental and simulated approach curves at an insulating substrate were obtained for FIB-milled nanoelectrodes with carbon radii of <50 nm. Figures 5B and 5C show approach curves for FIB-milled carbon nanotips with a = 44 and 27 nm, respectively, where good fits were obtained at short distances of ∼10 nm for the closest tip–substrate approach. These approach curves are strikingly different from those obtained previously with sub-100 nm-diameter carbon electrodes.^14–17 For instance a radius of 6 nm was determined for a pyrolytic carbon nanoelectrode from a approach curve at an insulating substrate that showed a decrease of only 15% in the tip current and significantly deviated from SECM theory for a disk tip,¹⁵ thereby suggesting that the extremely small value of the tip radius may not be reliable. Noticeably, closest tip–substrate distances in this study were consistently ∼10 nm for carbon nanotips with different radii (Figure 5), whereas the tip end was very flat (see Figure 2B). We speculate that the closest tip–substrate approach was limited by the contamination of tips, e.g., with aerosol nanoparticles during their storage in ambient air. Unfortunately, the cleaning of a carbon nanotip in piranha solution or by a UV cleaner damaged the tip end.

We employed FIB-milled carbon nanotips to study approach curves at gold-coated silicon wafers. Since the surface area of the unbiased gold film was much larger than that of a carbon nanotip, we expected that a positive feedback response would be limited by the diffusion of Ru(NH₃)₆³⁺ between the tip and the substrate.^48,49 A feedback response, however, was much lower than the diffusion-limited response when 5.0 mM Ru(NH₃)₆Cl₃ was employed (Figure 6A). The tip current was enhanced only by a factor of ∼1.6 before the 134 nm-radius carbon tip nearly contacted the gold substrate to give much higher currents. Apparently, this approach curve agreed with a theoretical approach curve⁵⁰ limited by the irreversible oxidation of Ru(NH₃)₆²⁺ at the gold surface to yield an electron-transfer rate constant, k_et, of 1.1 cm/s. The tip current, however, was limited not by the kinetics of Ru(NH₃)₆²⁺ oxidation, but by the subsequent transport of electrons through the thin gold film, because a higher positive feedback response was obtained as the concentration of Ru(NH₃)₆Cl₃ was lowered.⁵¹ Specifically, approach curves with 1 and 0.2 mM Ru(NH₃)₆Cl₃ (Figures 6B and 6C, respectively) yielded the highest normalized tip currents of ∼3.5 and ∼6.0, respectively. The latter approach curve fitted well with the diffusion-limited approach curve at a conductive substrate to yield a short distance of 20 nm for the closet approach of the 139 nm-radius tip to the gold surface. Importantly, the limited conductivity of the gold films was manifested owing to extremely high mass-transport conditions under carbon nanotips. By contrast, the film conductivity was high enough to obtain high positive feedback responses by using 1 μm-diameter Pt tips.³⁵ In addition, the CVs of Ru(NH₃)₆Cl₃ at a gold-coated silicon wafer gave a wider peak separation at a higher concentration of the redox species owing to an uncompensated resistance through a gold film, which was estimated to be 1.8 × 10² Ω according to the method proposed by McCreery and co-workers.⁵²

Zoom In Zoom Out Reset image size

It was reported previously that a pyrolytic carbon nanoelectrode gave a positive feedback response that was much lower than a diffusion-limited response as expected for a platinum substrate electrode.¹⁴ The low feedback response was ascribed to a kinetic effect and was fitted with a kinetically limited theoretical response. The rather poor fit yielded an unrealistically low k_et value of 0.8 cm/s for the ferrocenemethanol couple in comparison with a k⁰ value of 6.8 cm/s for this couple at platinum nanoelectrodes.⁴⁰ Alternatively, the low positive feedback response can be ascribed to the limited substrate conductivity⁵¹ or the tip recession.⁵³

To enable quantitative nanoelectrochemical measurements discussed above, one has to avoid the nanoscale damage of carbon nanotips caused by electrostatic discharge (ESD).³⁰ Specifically, the tip of a carbon nanoelectrode was maintained intact without ESD damage when an operator employed ESD protections³⁰ and handled the nanoelectrode under a high humidity of ∼50%³¹ in a plastic box equipped with a humidifier. In addition, we had to ground the lead wire of a carbon nanoelectrode when its tip was inspected by optical microscopy, where a sufficiently high humidity was not achievable. In fact, we found that ESD damage was caused unknowingly when the ungrounded lead wire of a carbon nanoelectrode contacted an insulated stage of the optical microscope under low humidity (<30%). ESD-damaged carbon nanotips had nanometer-sized pinholes when they were imaged by SEM prior to FIB milling (Figure 7A). In addition to pinholes, cracks were observed in FIB-milled carbon nanotips by SEM (Figure 7B) after they were used for CV measurements in the bulk solution. ESD damage was noticeable during the CV measurements, where the tip current was much higher than expected from the tip size owing to the diffusional access of Ru(NH₃)₆³⁺ to the carbon surface through pinholes and cracks.

Zoom In Zoom Out Reset image size