Introduction

Enzyme electrodes using glucose oxidase (GOx), alcohol dehydrogenase and other enzymes have been of analytical significance and widely employed for the electrochemical sensors, however, have crucial disadvantage of instability owing to the nature of enzyme. GOx based electrochemical sensors have additional drawback of oxygen dependence or necessity of mediator [1]. Thus, it would be desirable to determine bio-molecule concentration without using enzymes. For some recent examples of the electrochemical glucose sensors, GOx-free sensors have been developed using meso/nanoporous Pt films deposited in the presence of suitable surfactants [2–5], three-dimensional (3D)-network electrodes with various bimetallic compositions [6], highly dispersed metallic nanoparticles on composite film of carbon nanotubes (CNTs) [7, 8], nanoporous gold film electrode [9–11], metal-functionalized graphene nanohybrids [12], and metal nanoparticles incorporated into the porous carbon support [13, 14]. Among them, highly porous metallic nanocomposites are of great interest due to the selective/sensitive enhancement of kinetically sluggish heterogeneous faradic reactions including electrochemical glucose oxidation [2–6, 15, 16].

In particular, porous nanostructured materials based on Pt by electrochemical deposition have been intensively investigated with the aid of surfactant as a suitable template, such as hexagonal (H1) lyotropic liquid crystalline (LLC) phase [17, 18], potential-controlled surfactant assembly [19], and reverse micelle (L2) solution of a nonionic surfactant [4, 5, 19]. Pt thin films with hexagonally ordered nanopores (one-dimensional, 1D) on the scale of a few nanometers from both LLC template and the micelle-type aggregation were produced, so-called H1-ePt [3]. Highly desirable 3D-nanoporous Pt films, namely L2-ePt [4], were formed by simply electroplating in a L2 phase solution, where detailed studies regarding morphologies and roughness factors were also investigated [5, 20]. It is notable, however, there have been no reports regarding 3D-nanoporous metal films electroplated from L2 solution except Pt.

Ruthenium oxide has received attention for catalytic applications [21, 22] and pH measurements [23] due to its metallic conductivity and thermal stability. As far as we know, no porous RuO2 electrode has been reported except macroporous Ru oxide electrode for pH and NADH sensing via templates by the controlled evaporation (CE) and Langmuir-Blodgett (LB) technique [24]. In this study, we have applied simple fabrication method for 3D-nanoporous Ru oxide film, L2-eRuO, using a reverse micelle surfactant procedure, where several day evaporation time in CE technique or the LB trough in LB technique are not necessary. The pore diameter and widths of interstitial nanoparticles on an L2-eRuO film have been analyzed through a high-resolution transmission electron microscopy (HR-TEM) and the surface area of the electrodes was estimated by Cu underpotential deposition (UPD) [25, 26]. The electroplated L2-eRuO film possesses considerably high real surface area, and it could enhance the kinetically controlled electro-oxidation of glucose, which is demonstrated in this presentation together with pH sensing.

Experimental

Reagents

Ruthenium chloride hydrate (RuCl3∙xH2O), Trition X-100, sodium dihydrogen phosphate (NaH2PO4∙H2O), sodium hydrogen phosphate (Na2HPO4), citric acid, boric acid, D-(+)-glucose, L-ascorbic acid (AA), 4-acetamidophenol (AP), uric acid (UA), and dopamine hydrochloride (DA) were purchased from Sigma-Aldrich (St. Louis, MO, USA http://www.sigma-aldrich.com). CuSO4 (anhydrous) was supplied by Junsei Chemical Co., Ltd (http://junsei.lookchem.com). All other chemicals used were of analytical grade, all solutions were prepared with deionized water (resistivity ≥18 MΩ cm).

Electrochemical deposition of ruthenium and electrochemical measurements

Thin films of L2-eRuO were formed by electroplating in a similar method described previously for L2-ePt [4]. Briefly, in a solution containing a ruthenium precursor (RuCl3), Triton X-100, and NaCl aqueous solution (5:45:50 in wt%, at 40 °C), the nanoporous metal oxide was deposited on a Au disk electrode (1.6 mm in diameter, Bioanalytical Systems, Inc. http://www.basinc.com) by scanning the potentials from +0.0 to −0.8 V at a scan rate of 50 mV s−1. The numbers of potential cycling were varied from 5 to 70, in order to obtain the L2-eRuO films with different roughness factor (Rf). The Triton X-100 used was extracted by placing the electrode deposited with nanoporous metals in distilled water, which was replaced with fresh water every 2 h for 4–5 times. Then, the electrodes were electrochemically cleaned in 0.1 M H2SO4 by scanning the potentials from +0.0 to +1.5 V at a scan rate of 50 mV s−1 to remove the remained surfactant before performing other experiments. A saturated calomel reference electrode (SCE) and Pt wire counter electrode were used. UPD of Cu on as-prepared L2-eRuO electrode was performed in 2 mM CuSO4/0.1 M H2SO4. The amperometric response of the prepared electrode to varying glucose concentration was measured at +0.5 V (vs. SCE) with the successive addition of a glucose standard solution into a 0.05 M phosphate buffer solution (pH 7.4) with constant stirring. The electrode pH response was obtained by titrating a universal buffer composed of 11.4 mM boric acid, 6.7 mM citric acid, 10.0 mM NaH2PO4 with small aliquots of NaOH and HCl while monitoring the electrode potentials (vs. Ag/AgCl reference electrode). Thin films of L2-eRuO were also formed on a Au wire (0.5 mm in diameter, Aldrich) to check pH response. The solutions were stirred magnetically and the equilibrium potentials were recorded.

Characterization

The electroplated L2-eRuO film structures were examined by field emission scanning electron microscopy (FE-SEM, Jeol JSM-6700F http://www.jeol.com), which was equipped with an energy dispersive X-ray spectroscopy (EDS) system, and HR-TEM (Jeol JEM-2100F, 200 kV). FT Raman spectroscopy (Renishaw InVia Sustem http://www.renishaw.com) was used to characterize the RuO materials. The electrochemical measurements were performed using a CHI 705 workstation (CH Instruments http://www.chinstruments.com). All experiments were carried out in a Faraday cage to increase the signal-to-noise (S/N) ratio. For the potentiometric measurements, the potential differences between the working electrode and Ag/AgCl reference electrode were measured using a PC equipped with a high-impedance input 16-channel analog-to-digital converter (KOSENTECH Inc., Korea http://www.physiolab.co.kr).

Results and discussion

FE-SEM images of the surface morphology of the RuO films grown on polished gold substrate obtained in 0.2 wt% and 5 wt% RuCl3 by repetitive potential cycling in the absence of the surfactant are depicted in Fig. 1a and b, respectively. It is observed that typical surface morphology of Ru film obtained from low concentration of RuCl3 (Fig. 1a) is obviously different from that from high concentration (Fig. 1b). Although Ru film in Fig. 1a was found to have relatively a little rough surface and a lot of protruding spikes compared to that obtained from higher concentration of Ru precursor as shown in Fig. 1b, no obvious porous structure was found even at high concentration of RuCl3. These deposits result from uninhibited and more vigorous metal growth over the electrode substrates in the absence of surfactants which direct the metal structure and increase electroplating solution viscosity.

Fig. 1
figure 1

FE-SEM images of the electroplated Ru oxide film electrodes induced in the absence (a and b) and presence (c and d, 50 wt%) of surfactant: RuO images electroplated in an aqueous solution containing low (a, 0.2 wt%) and high concentration (b, 5 wt%) of RuCl3; L2-eRuO images (obtained in an aqueous solution containing 5 wt% RuCl3 and 50 wt% surfactant) taken before (c) and after (d) electrochemical cleaning by voltammetric cycling in 0.1 M H2SO4

Full size image

Electroplating of Ru with the aid of a surfactant template onto polished gold electrodes was also conducted by potential sweep method rather than constant potential method, which was used widely in previous reports, owing to the improved stability of the films [27]. The ternary plating systems used in our experiments were consisted of a 5 wt% RuCl3, 50 wt% Triton X-100, and 45 wt% NaCl. After electroplating, the electrodes were rinsed to remove the remained surfactant with copious amounts of deionized water followed by further cleaned using a cycling potential between +0.0 V and +1.5 V (vs. SCE) in 0.1 M H2SO4 until reproducible cyclic voltammograms were obtained. The surface morphologies for the L2-eRuO were observed using FE-SEM before and after the electrochemical cleaning process as shown in Fig. 1c and d, respectively. EDS confirmed that no surfactant was present in the washed films (Fig. 1d). The electrochemical cleaning resulted in a deeper porosity with uniform nanoparticle distribution within L2-eRuO film. Indeed, the L2-eRuO film was composed of Ru nanoparticles interconnected with each other. Note that these FE-SEM images of L2-eRuO are contrast to that of L2-ePt [4] where no apparent grain or pore was observed.

TEM studies (Fig. 2a –c) supported that the electroplated L2-eRuO film revealed a highly porous structure consisting of regular holes of 2.0 (±0.2) nm in diameter. The interstitial nanopores among the partially merged Ru nanoparticles are quite evenly distributed, and their width is about 2.3 (±0.2) nm. The resulting morphology is supposed to be related to their catalytic activities of nanostructured Ru toward direct glucose oxidation. More detailed discussion regarding electrocatalytic glucose sensor is described in a later section. In addition, Raman spectra in Fig. 2d clearly shows the three major Raman peaks corresponding to crystalline Ru oxide in the rutile form (Eg, A1g, and B2g, modes are located at 508, 622 and 687 cm−1, respectively). HR-TEM results in Fig. 2c agree with the crystalline structures of the L2-eRuO film.

Fig. 2
figure 2

TEM images (a−c) corresponding to the L2-eRuO showing the detailed actual nanoporous morphology at three-different magnifications. Raman spectrum (d) showing the three major Raman peaks corresponding to crystalline Ru oxide

Full size image

The Cu UPD on the L2-eRuO film for determining the real surface area (RSA) is a relatively effective technique over other methods such as CO stripping, hydrogen adsorption and Brunauer-Emmett-Teller (BET), owing to the similarity of the atomic radii (Cu, 128 pm; and Ru, 134 pm) [26]. Fig. 3a presents that the Cu surface coverage is determined using cyclic voltammetry (CV) in 0.1 M H2SO4/2 mM CuSO4 solution purged by N2 at a potential scan rate of 10 mV s−1 on the L2-eRuO electrode to find out what the potential range of Cu UPD growth is. The CV scans began at 0.4 V vs. SCE, moved gradually in cathodic direction and then reversed to anodic direction at various switching potentials to allow repetitive deposition/desorption of Cu2+ ions to take place on the L2-eRuO electrode surface. Presented L2-eRuO film (as shown in Fig. 1d) was prepared from surfactant-assisted RuCl3 solution by scanning the potentials from +0.0 V to −0.8 V for 30 cycles at a scan rate of 50 mV s−1. It is accepted that UPD of metals starts in a potential region more positive than the Nernst potential by forming a monolayer. Indeed, the calculated equilibrium Nernst potential for Cu2+/Cu is around 0.016 V (vs. SCE) for 2 mM CuSO4 solution, and the first cathodic current increase from +0.2 V (vs. SCE) is assigned as a Cu UPD on the L2-eRuO surface. The deposited Cu atoms were stripped from Ru surface at around +0.15 V (vs. SCE) when the electrode was subjected to the reverse anodic scan. A strong oxidation peak of the deposited Cu overlayers on the anodic scan was not observed readily on 3D-nanoporous Ru electrode surface probably owing to the relatively large density of the oxide overlayers with high surface area to volume ratios, as discussed in previous study [27]. In this experiment, a large charging capacitance current background was observed, indicating that the conductive Ru oxide with large surface area was formed during the electroplating process. The Cu surface coverage on L2-eRuO film surface as displayed in Fig. 3b was calculated from the integrated charge of UPD stripping peak after subtracting its background. The surface coverage of Cu UPD increases with more negative switching potential and reaches a plateau of ca. 1.1 × 10−8 mol·cm−2 or 2.1 mC·cm−2 beyond −0.05 V (vs. SCE). This surface coverage can be calculated as 0.91 monolayer of Cu using the conversion factor in the literature [25].

Fig. 3
figure 3

Cyclic voltammograms (a) and Cu surface coverage vs. cathodic switching potentials (b) obtained from L2-eRuO electrode immersed in a 2 mM CuSO4/0.1 M H2SO4 solution at a scan rate of 10 mV s−1. The Cu surface coverage on nanoporous Ru surface was calculated from the integrated charge of UPD stripping peak (ca. 0.220 V vs. SCE) after subtracting its background

Full size image

As-prepared L2-eRuO electrode was tested for a non-enzymatic amperometric glucose sensor. To determine the appropriate glucose oxidation potential, the amperometric response depending on the applied potentials (varied from +0.3 to +0.7 V with a step of 0.1 V) was measured at an L2-eRuO electrode prepared by 30 CV cycles in 0.05 M phosphate buffer solution (pH 7.4) containing 2.0 mM glucose. The glucose oxidation current gradually increases at more positive potential (data not shown). For example, the amperometric response at +0.6 V or +0.7 V showed three fold higher than that at +0.5 V, however, stability of current response and S/N ratio at +0.5 V seems to be better than that at +0.6 or +0.7 V. Furthermore, it is well-known that the susceptibility to interference species decreases with lowering detection potential of the sensor. Hence the applied potential for glucose sensing was fixed at +0.5 V for all subsequent experiments.

The sensor sensitivity (from 0 to 4 mM) depending on the number of the repetitive CV cycles during the preparation process was also examined to find optimum preparation condition (Fig. 4). For this, the L2-eRuO films were prepared on Au disk electrodes by 5, 10, 20, 30, 50, and 70 CV cycles, respectively, and the geometric surface area (GSA) of each electrode was determined by chronocoulometric analysis as described previously [28]. As aforementioned, the RSA of each electrode was estimated using the Cu UPD on L2-eRuO. The Rf value of each electrode was calculated by dividing RSA by GSA. The (number of CV cycles, Rf) data are (5, 17.4), (10, 90.4), (20, 191), (30, 320), (50, 508), and (70, 731), respectively. Interestingly, Rf value was roughly hundred times of the number of CV cycles except for the initial stage of film growth. The sensor sensitivity (current response vs. glucose concentration) was normalized to the corresponding RSA (j). As seen in Fig. 4, the sensitivity increases with a Rf increase, and the sensitivity enhancement was large at Rf < 200, while increment of the sensor sensitivity became much smaller at Rf >200. For all subsequent experiments, thus the number of repetitive CV cycles in L2-eRuO electrode was fixed at 30 (Rf = 320).

Fig. 4
figure 4

Dependence of glucose sensitivities (normalized to GSA) on Rf values of L2-eRuO, which were varied by changing the number of CV cycles (5, 10, 20, 30, 50, and 70 from the left to the right, respectively) for electroplating. Glucose sensitivity of L2-eRuO was examined by successive adding 0.25 mM glucose in 0.05 M phosphate buffer solution (pH = 7.4) at +0.5 V (vs. SCE)

Full size image

Figure 5a presents typical dynamic steady-state current response curves (i-t curves) of four types of electrodes to consecutive increments in glucose concentration. Four electrodes used are bulk Au, RuO without surfactant, L2-eRuO (not electrochemically cleaned), and L2-eRuO (electrochemically cleaned in H2SO4) for comparison. The applied potential was +0.5 V to minimize interferent oxidation and the measurements were carried out in a deoxygenated 0.05 M phosphate buffer solution. The glucose concentrations were changed from zero to 16 mM by the successive additions of a pre-calculated amount of the glucose stock solution, considering that the normal physiological level of glucose is 3–8 mM [3]. Figure 5b shows the corresponding calibration plots for the amperometric detection of glucose at those four electrodes. The overall current responses in the range of zero to 16 mM glucose were 0.8 and 13.5 μA at RuO without surfactant and L2-eRuO (electrochemically cleaned in H2SO4) electrode, respectively. The electrochemically cleaned L2-eRuO electrode showed 16.9 fold higher electrocatalytic activities than RuO without surfactant. The sensitivity of 40.2 μA mM−1 cm−2 (normalized to GSA) was obtained at the electrochemically cleaned L2-eRuO electrode in a linear range of zero to 4 mM glucose with a detection limit of approximately 21 μM glucose (S/N = 3). This sensitivity value is comparable with those of the reported nanostructured gold electrodes [29]. For example, comparison of analytical performance of our sensor with other published nonenzymatic glucose sensors is summarized in Table 1. The developed method in this work exhibits the characteristics of the good sensitivity and interference resistance under biological conditions, especially at physiological pH.

Fig. 5
figure 5

Amperomeric current responses (a) and calibration curves in terms of glucose concentration (b) at the bulk Au, RuO, L2-eRuO (without electrochemical cleaning) and L2-eRuO (with electrochemical cleaning) in deoxygenated 0.05 M phosphate buffer solution (pH = 7.4) at +0.5 V (vs. SCE)

Full size image
Table 1 Comparison of the reported nonenzymatic glucose sensors recently
Full size table

The other positive characteristics of L2-eRuO electrode is the selectivity against biological interferents. The anionic (AA and UA), neutral (AP) and cationic (DA) molecules in biological samples could be oxidized easily at relatively positive potentials and often interfere with glucose detection: for our glucose sensors, +0.5 V vs. SCE. Figure 6 shows the effects of the interfering species, where the injection of interferents was performed while monitoring the amperometric response in phosphate buffer solution containing 5 mM glucose. The injection of AA (100 μM), AP (100 μM), UA (20 μM), and DA (20 μM) did not cause serious change in the current response. Note that the normal physiological levels of the above four interferents are commonly below the injected amounts. In addition, to exclude the possibility of saturation after the last addition of glucose, we compared the amperometric responses between glucose only and the mixture of analyte and four interfering substances because the real sample contains both analyte and interfering species. As shown in Fig. 7, the amperometric responses between these two samples were virtually the same each other. We also have measured the amperometric responses between the analyte and the mixture of analyte and a specific interfering substance, i.e., AA, AP, AU, or DA, where the differences were hard to be observed (Fig. S1 in Electronic Supplementary Material, ESM).

Fig. 6
figure 6

Amperometric current responses with the injection of glucose and interfering species at the (a) RuO, (b) L2-eRuO (without electrochemical cleaning) and (c) L2-eRuO (with electrochemical cleaning) in deoxygenated 0.05 M phosphate buffer solution (pH = 7.4) at + 0.5 V (vs. SCE). Added concentration is AA (100 μM), AP (100 μM), UA (20 μM), and DA (20 μM)

Full size image
Fig. 7
figure 7

Comparison of amperometric current responses between the injection of glucose (5 mM) only (left) and a mixture of glucose (5 mM) + four interference species (right) at the L2-eRuO (with electrochemical cleaning) in deoxygenated 0.05 M phosphate buffer solution (pH = 7.4) at +0.5 V (vs. SCE) with interfering species mixture of AA (100 μM), AP (100 μM), UA (20 μM), and DA (20 μM)

Full size image

Such selectivity of sensing against interferents stems from the nature of electron-transfer [3]. Briefly, glucose oxidation without enzyme is a sluggish electron-transfer reaction, i.e., kinetic-controlled electrochemical system, which is different from the electrode reaction of other interfering species, i.e., diffusion-controlled electrochemical system. Therefore, the electrode with nanoporous structure could selectively enhance the faradaic current of non-enzymatic glucose oxidation since the interfering species should be depleted inside the diffusion layer.

The potentiometric responses of the L2-eRuO electrode to pH were examined from pH 2 to 11 by adding aliquots of NaOH to a universal buffer solution. Again, pH was changed reversely from 11 to 2 by adding HCl to check the reliability, where no hysteresis of pH response was observed. Figure 8 shows the typical potentiometric response curves for L2-eRuO and glass pH electrode. The L2-eRuO electrode showed reliable potentiometric pH response including near Nernstian behavior (slope = −60.5 mV pH−1, r = 0.9997) and reasonable response time (t 90% = ca. 6.5 ± 2.0 s). The pH response of L2-eRuO on Au wire electrode (0.5 mm in diameter) also revealed a slope of −55.2 mV pH−1, which showed the possibility of miniaturization for local pH sensing [24]. Note that the reported value for L2-ePt showed a slope of −51 mV pH−1 and response time of t 95% = ca. 60 s [20].

Fig. 8
figure 8

Comparison of the pH responses between a glass pH electrode and an L2-eRuO: dynamic potentiometric responses towards varying pH values (a); and calibration curves (b) corresponding to (a). The pH changes were realized by additions of NaOH and HCl solution to a universal buffer solution

Full size image

Conclusions

We have sucessfully synthesized L2-eRuO directly grown on Au substrates using an electrochemical deposition of RuCl3 in the presence of reverse micelles of Triton X-100. FE-SEM and HR-TEM images of L2-eRuO show apparent grain and pores (2.3 ± 0.2 nm) contrast to that of L2-ePt where no apparent grain or pore was observed. The facile and simple approach described in this study not only allows controllable variation of the Rf to achieve optimum performance, but also eliminates complicated experimental procedures and purification steps. The surface area of the electrodes was estimated by Cu UPD, showing high RSA which favors to obtain larger electrochemical response of glucose and better selectivity over AA, AP, UA and DA, due to the enlarged active surface area for the kinetically slow glucose oxidation. The optimized L2-eRuO electrode (Rf = 320) exhibited 17 fold higher total current responses for amperometric glucose detection compared to the RuO electrode, with a sensitivity of 40.2 μA mM−1 cm−2 (normalized to GSA) in a linear range of zero to 4 mM glucose. Compared to the reported value for L2-ePt, the prepared L2-eRuO electrode also showed better potentiometric responses to pH changes, such as a steeper potential shift per pH, and a faster response time as well as facility of miniaturization for possible microsensor due to the increased surface area.