Facile preparation of nanostructured copper-coated carbon microelectrodes for amperometric sensing of carbohydrates

The procedure for microelectrode fabrication is as follows: the carbon fiber is glued using conductive silver epoxy (EC101, Polytec, Germany) onto a copper wire and the junction is then cured at 150 °C for 10 min. The fiber with copper contact attached is fitted into the glass capillary, about 10 mm of the fiber is left protruding from its contracted end. Both ends of the capillary are sealed using epoxy resin (CHS Epoxy 1200, Sindat Pilsen, Czech Republic). Prior to use, the protruding fiber is cut to a length of about 5 mm by lancet, and the fiber end of the electrode is briefly sonicated in dichloromethane.

Carbohydrates are generally regarded as difficult-to-analyze compounds, since they lack suitable chromophores, uorophores or substituents required for direct detection by spectrophotometric or uorimetric devices. Under specic conditions, carbohydrates can be measured electrochemically aer labelling with electroactive markers. 1,2 Of the direct electrochemical methods, pulsed amperometric detection (PAD) at platinum or gold electrodes in alkaline media has attracted a lot of attention as a method that provides a sensitive response in carbohydrate analysis. The method is based on the application of a complex, multistep potential waveform to noble-metal (predominantly gold) working electrodes. This approach effectively solves the problems associated with electrode surface fouling that occurs rapidly when working with simple directcurrent methods of carbohydrate oxidation. Although PAD is already a rmly established approach for carbohydrate detection at least for HPLC applications, the constant potential detection mode at metal electrodes offers several advantages over PAD. Apart from its inherently simpler instrumentation, it should, in principle, provide higher sensitivity, because all other parameters being equal, noise should be minimized when the applied potential is constant. 3 Therefore, the constant-potential detection of carbohydrates is being intensely studied, with copper as one of the materials of choice due to the specic catalysis of carbohydrate oxidation based on Cu II /Cu III redox cycling in surface copper oxides. In the last decade, electrodes modied with copper-based nanomaterials were introduced as efficient devices in carbohydrate analysis, particularly in glucose sensing. Non-enzymatic glucose sensors, based on the combination of copper/copper oxide nanoparticles and carbon (carbon nanotubes and graphene) nanomaterials are part of current research trends. These devices achieve micromolar to submicromolar detection limits in 0.1 M NaOH (1.0 mmol L À1 for Cu nanocubes/MWCNT 4 and 0.2 mmol L À1 for irregular Cu nanoparticles/graphene 5 ) with response times in seconds. In general, the nanostructuring of catalytic electrode surfaces offers a combination of increased electroactive surface area with a decrease in the potentials of surface redox transitions. 6 When applied as sensors, electrodes modied with nanomaterials frequently allow for higher sensitivity and a broader concentration range of analytes in comparison to their bulk counterparts. 7 Nevertheless, when used it should be kept in mind that some gradual loss of outer material via dissolution at the electrode/solution interface may occur at such electrodes depending on the nanostructured lm thickness, possibly leading to a lower response stability.
Despite the known benets of employing metal electrodes in the electroanalysis of carbohydrates, a limited number of papers cover the detection of carbohydrates by constant potential amperometry in ow systems and only a few papers have been devoted to the high pressure liquid chromatography with electrochemical detection (HPLC-ED) of carbohydrates based on copper electrodes. [8][9][10][11][12] In our previous works, we have reported that carbon ber microelectrodes (CFMEs) can be utilized as highly sensitive electrochemical detectors, especially in HPLC. [13][14][15][16] Thereby, we developed a facile technique enabling CFMEs to be modied with nanostructured copper layers. The dimensions of a carbon ber (3-4 mm in length, 7 mm in diameter) enable the direct insertion of the active electrode tip into a narrow-bore capillary even aer the modication with copper. The advanced mechanical properties of carbon bers grant the resulting sensing element excellent mechanical stability, superior to other electrode materials of similar size and shape, e.g. extremely fragile copper microwires. The copper-modied CFME exhibits high performance in carbohydrate electrooxidation-based sensing, as tested in the amperometric, ow injection analysis (FIA) and HPLC detection of selected mono-, di-and oligosaccharides.

Preparation of copper-coated carbon fiber microelectrode
The procedure for CFME modication by copper is based on a methodology 15 previously described by ourselves for the preparation of nanostructured silver layers, i.e. electrodeposition of the material formed by the dissolution of the anode. The formed anode-derived material is transferred towards the cathode by movement through the interelectrode space originating from the combined action of migration, diffusion and convection induced by thermal and density gradients. Reductive deposition occurs at the cathode. Specically, copper wire was anodized in a two-electrode cell containing ultrapure water as the medium while the CFME was connected as the cathode, the interelectrode distance being 1 cm. Due to the low conductivity of ultrapure water, relatively high potentials (10-30 V) had to be employed to induce copper anodic dissolution. Following optimization experiments, a potential of 15 V was applied to the electrodes for 20 min. These conditions represent a right compromise between the analytical performance of the modied CFME and stability of the modifying layer in FIA and HPLC arrangements (see below). We preclude that copper electrodeposition proceeds by a similar mechanism to that suggested for silver. It is to be expected that the potential is sufficient to induce a two-electron oxidation of copper (eqn (1)). 17 Cu(II) readily reacts with OH À available from the autoprotolysis of water to give Cu(OH) 2 (eqn (2)) The solubility product of Cu(OH) 2 , restricts the maximum concentration of free Cu II ions available for electroreduction to the low value of $1.7 Â 10 À7 mol dm À3 which, together with the absence of the supporting electrolyte during the deposition, represent key factors [18][19][20] in the formation of a high number of nuclei leading to metal-nanostructures rather than to the production of smooth or larger particles containing deposits. In addition to Cu II reduction, the proton is reduced on the cathode (eqn (4)), forming hydrogen gas where hydrogen bubbles released from the cathode are observable with the naked eye during the experiment.
The hydrogen gas formed may contribute towards keeping the copper layer in the reduced state. The copper-coated CFMEs were examined by scanning electron microscopy (SEM), revealing that the deposit is a randomly oriented Cu nanostructured network (Fig. 1A and B). The elemental composition of the deposit was veried by energy dispersive X-ray (EDX) spectroscopy (Fig. 1C). In addition to the carbon peak originating from the carbon substrate, only copper features and a small peak of oxygen, presumably formed as an artefact due to manipulating with the copper-coated CFME in an air atmosphere, were found.
Cyclic voltammetric (CV) record of the copper-coated CFME, for comparison shown along with the CV of a copper wire electrode recorded under identical conditions, contains all copper redox transitions (peaks I and II in Fig. 2) and is identical to the CV records reported in the literature. The copper wire electrode used, having an almost 200-fold higher geometric area than the copper-coated CFME (2.4 Â 10 À5 vs. 1.4 Â 10 À7 m 2 ), provided only ca. 20-fold higher currents for the observed surfacial redox transitions, i.e. approx. 10-fold higher current densities were achieved on the CFME due to electrodeposition (Fig. 2).

The performance of copper coated CFMEs in carbohydrate sensing
The performance of copper electrodes in carbohydrate sensing depends heavily on the pre-oxidation of the copper electrode to form catalytically active surface copper oxides. Various anodisation procedures were suggested in the literature, leading to differently sensitive surfaces; probably the most sensitive electrodes were reported by Kano et al., 8 who oxidised a polished copper electrode by the action of hot air (100 C, 3 h) and employed these electrodes in the detection of a mixture of sugars in 0.15 mol L À1 NaOH at 0.45 V vs. Ag/AgCl aer HPLC separation. The drawback of these electrodes was the requirement of 2 hours of stabilization of the background current. For copper-coated CFMEs, satisfactory results were achieved using a simpler and less time-consuming procedure, i.e. anodisation at +700 mV vs. Ag/AgCl for ve minutes. In this procedure, a complete transformation accompanied by a morphology change to cube-shaped, 50-400 nm sized copper oxide particles occurs (Fig. 3).
Experiments demonstrating the performance of the copper nanostructured CFMEs in the amperometric sensing of glucose were then performed, employing working potentials between 500-800 mV vs. Ag/AgCl. Optimum performance was observed between 600-700 mV, though for the applied potential of 500 mV, both the electrode response and background were lowered proportionally. On the other hand, for the applied potential of 800 mV (trace not shown) the sensitivity of the nanostructured CFMEs towards glucose remained similar to that observed at 700 mV, but the baseline current was markedly elevated. The nanostructured CFMEs therefore provide usable analytical signals when operated between 500-700 mV vs. Ag/AgCl, which is clearly shown in Fig. 4.
Rapid response times, high surface area, radial diffusion and the guaranteed electrical contact with the carbon ber substrate ensured by the electrochemical nature of the nanostructures' formation lead to increased analytical usability of fabricated nanostructured copper-coated CFMEs. Therefore, we incorporated the sensor, treated according to the procedure described above, into a FIA/HPLC detection ow-cell, described in our previous works. 13,14 FIA was tested at various ow rates (15-200 mL min À1 ), nevertheless, using our instrumentation, similar values of LODs were found (Table 1). For illustration, the FIA record of lactose over a broad concentration range is shown in Fig. 5A. Fig. 2 Cyclic voltammograms of (A) copper wire electrode and (B) copper-coated CFME recorded in 0.01 mol L À1 NaOH, scan rate ¼ 100 mV s À1 , E init +0.8 V. I a : Cu 0 / Cu I ; II a : Cu I / Cu II ; I c : Cu II / Cu I ; II c : Cu I / Cu 0 . Fig. 3 SEM of copper-coated CFME after anodisation in 0.01 mol L À1 NaOH at +0.7 V vs. Ag/AgCl for 5 min. Fig. 4 Amperometric traces of glucose at different applied working potentials. 0.01 mol L À1 NaOH was used as supporting electrolyte, additions (series indicated by arrows) led to the following increases in glucose concentration: 20-times 5 Â 10 À6 , 20-times 1 Â 10 À5 and 20-times 2 Â 10 À5 mol L À1 , stirred solution (300 r.p.m.).
The response stability of the nanostructured electrode in FIA mode was tested for lactose. For this purpose, ve consecutive injections of lactose solution were carried out at 0, 0.5, 1, 2 and 3 h from the beginning of the test. Aer overnight break, the test was continued at 12, 13 and 14th hour. For details, see Fig. S3 and corresponding part of the ESI. † The R.S.D. within individual test runs ranged from 1.6-7.2%. The interelectrode reproducibility was 20%, which was deemed satisfactory considering differences in surface areas and state of the individual carbon bers and a difficulty to control their exact lengths in the assembled CFMEs.
Finally, the performance of the detector in an HPLC setup was investigated using a model mixture of three disaccharides (sucrose, lactose and maltose). For the separation we used an HPLC column, which is, unlike the majority of silica-based sorbents, alkali-resistant, and thus no extra post-column addition of concentrated hydroxide solution is needed. This way, the detection of carbohydrates can be accomplished directly in the mobile phase and the necessity of other additional instrumentation (such as a pulseless pump for post-column alkalinisation of the mobile phase) is eliminated. Similarly to the FIA experiment, the sensor exhibited a rapid response, very short reacquisition times and good baseline stability within runs. Highly stable, low-noise backgrounds were achieved, especially when the CFME detector was operated at a low applied potential. The HPLC analysis of disaccharides (292 or 58 pmol per column) is shown in Fig. 5B.

Conclusions
Carbon ber microelectrodes coated with a copper layer were prepared using the above novel electrodeposition method. Following electrochemical transformation of the copper layer to copper oxide, the resulting device was shown to be an efficient non-enzymatic carbohydrate sensor, exhibiting low detection limits and a wide linear concentration range, comparable to the best copper-based sensors, e.g. a sensor utilizing a graphenecopper composite. 5 Short response time, small dimensions and high stability favour the use of the sensor in FIA and HPLC systems. The results presented here could be the starting point for the further development of novel nanostructured microelectrodes for not only the analysis of carbohydrates in solution, but also sensing their binding to proteins, i.e. the trace detection of glycoproteins or monitoring of non-enzymatic glycation processes. a Data for dextran in mg mL À1 . b Sensitivity in A g À1 m À2 ; 0.025 mol L À1 NaOH as mobile phase; ow rate ¼ 200 mL min À1 ; working potential 700 mV.