Potentiometry is commonly used to measure glucose concentrations greater than 10⁻⁵ M, which is in the physiological range in most cases. The potential difference between the reference electrode and the indicator electrode is measured at zero current flow. The ideally nonpolarizable reference electrode provides a constant potential, while the indicator electrode shows an erratic potential depending on the concentration of the analytes. The zero current potentials applied between those two electrodes are recorded as a function of the concentrations of target analytes in a logarithmic manner [48]. The potential difference at the electrode-electrolyte interface arising from unbalanced activities of species i in the electrolyte phase (s) and in the electrode phase (β) is related by the following Nernst equation: (2) E = E o + RT Z i F   ln   a i s a i βwhere E₀ is the standard electrode potential, R the gas constant, T the absolute temperature, F the faraday constant, a_i the activity of species i, and Z_i the number of moles of electron involved.

Potentiometric sensors are divided into the metal-oxide sensitive field effect transistor (MOSFET), the light-addressable potentiometric sensor (LAPS), the ion-sensitive field effect transistors (ISFET) and the ion-selective electrodes (ISE). Ali et al. developed a commercial MOSFET using a GOx modified ZnO nanowire. GOx-functionalized ZnO nanowires were grown on the Ag wire and then directly connected to the MOSFET gate (Figure 1) [54]. They tested the response time and the stability of the MOSFET sensor by using three different GOx/ZnO modified Ag electrodes, i.e., vertically aligned, uniform nonaligned, and nonuniform nonaligned ZnO nanowires. The results showed that well aligned ZnO-modified electrode displayed a good stability, short response time (<100 ms), and improved detection limit. They also further demonstrated that the GOx/ZnO modified MOSFET is able to be used for the immobilization of other biomolecules to make versatile electrodes for biosensing.

ISFETs and LAPS have attracted much attention for biosensing application being especially convenient for construction. The principles of LAPS are based on the activation of semiconductor by a light emitting diode [55]. Seki et al. developed a LAPS based on SiO₂/Al₂O₃ film grown on an n-type Si substrate. The GOx was immobilized on the film at various pH in the range of 3 to 11. Upon exposure to the light emitting diode, the equilibrium potential of the GOx-modified SiO₂/Al₂O₃ film was increased linearly with the increase of glucose concentration up to 4 mM. An increased sensitivity was also observed by introducing hexacyanoferrate (III) as an alternative to oxygen because the reduced form of flavin adenine dinucleotide (FADH₂) positioned within the active site of the GOx is more easily oxidized with hexacyanoferrate (III) [56]. While, the principle of ISFET is based on the local potential generated by surface ions from a solution [55]. Luo et al. built up a glucose sensitive enzymatic field effect transistor (ENFET) by modifying the gate surface of the ISFET with SiO₂ nanoparticles and GOx. The SnO₂ based glucose sensor showed a good stability and reproducibility with a detection limit of ca. 0.025 mM [57]. A few reports are available on nonenzymetic potentiometric glucose sensor based on poly(aniline boronic acid) suggested by Shoji and Freund [58,59]. These sensors showed improved sensitivity for fructose compared to glucose.

Amperometry is a quite sensitive electrochemical technique in which the signal of interest is current that is linearly dependent on the target concentration by applying a constant bias potential. Glucose is oxidized at the working electrode composed of a noble metal formed by either the physical vapor deposition or screen-printing and the biorecognition species such as GOx for glucose sensing [60]. An amperometric biosensor comprises two or three electrodes. The former consists of a reference and a working electrode. Application of the two-electrode system to biosensors is limited, because at high current flow the potential control becomes difficult as a result of sizable IR drop. Instead, the third electrode is commonly introduced as an auxiliary counter electrode having a large surface area to make most of current flows between the counter and the working electrodes, though voltage is still applied between the working and the reference electrodes.

There are three modes of the glucose oxidation referred to as the first, the second and the third generation glucose sensors depending on the electron transfer mechanisms. The nanostructured metal-oxide based glucose sensor belong to the third generation. Figure 2(A) depicts the first generation glucose sensor where oxygen is used as a mediator between the electrode and the GOx. The oxygen is reduced to form hydrogen peroxide in the presence of glucose by flavin adenine dinucleotide (FAD), a prosthetic group of GOx, and FADH₂ redox couple. The reduction rate of the oxygen is proportional to the glucose concentration that is quantified by either measuring the augmentation of hydrogen peroxide or decrement of the oxygen concentration [5,61]. On the other hand, artificial electron mediators (M), e.g., ferro/ferricyanide, hydroquinone, ferrocene, and various redox organic dyes between the electrode and the GOx are employed in the second generation glucose sensor. These mediators make the electron transfer rate between the electrode and the GOx faster and also give a way of getting around for a case when limited oxygen pressure commonly observed from the first generation glucose sensor [62,63]. Figure 2(B) represents the second generation glucose sensor, where M_ox and M_red are the oxidized and reduced forms of mediator, respectively. The reduced form of flavin group (FADH₂) of GOx is reoxidized in presence of glucose by the reduction of M_ox to form M_red. The output current signal generated, when further oxidation of M_red occurs at the electrode surface, is proportional to the glucose concentration [5,62,63].

In the third generation glucose sensor, the GOx is directly coupled to the electrode. The direct electron transfer efficiently generates an amperometric output signal. The improved sensing performance by the direct electron transfer has been realized by incorporating the enzyme with metal nanoparticles [64,65], and semiconductive nanomaterials [66,67]. The nanocrystalline metal-oxide plays a vital role in the enzyme immobilization owing to its highly specific surface area, good biocompatibility and stability [68]. Liu et al. fabricated a novel third generation amperometric glucose sensor based on the aligned ZnO nanorod formed on an ITO electrode (Figure 3). The immobilized GOx on the ZnO nanorod shows still high catalytic activity, effective protection by Nafion (perfluorosulfonate ionomer) membrane cast on the film, a wide linear range with good selectivity and freedom from the interference effects of uric acid and ascorbic acid in real samples [69].

In recent years progressive attempts have been made to determine the glucose without any enzyme for the reliable fast determination [70–72]. Most enzymeless electrochemical glucose sensors rely on the properties of the electrode materials, on which the glucose is oxidized directly. Carbon, platinum, gold and nanostructured CuO-modified CNTs have been widely investigated as candidates for improving the sensing performance of enzymeless sensors [73,74]. However, some problems including poor selectivity and low sensitivity due to the surface poisoning by the adsorbed intermediates or chloride ions still remain [75].

The impedimetric biosensor is less frequently used as compared to the potentiometric and the amperometric ones. Electrochemical impedance spectroscopy (EIS) is a powerful analytical tool which allows us to effectively visualize the actual electrical double layer structure of a modified electrode [76] although recording of an impedance spectrum within a broad frequency range is time consuming. The glucose does not affect the dielectric spectrum in the MHz frequency region [77,78] and direct detection of the glucose is available. The expression of impedance in a simple RC circuit is as follows: (3) Z 2 = R 2 + 1 ( 2 f C ) 2where Z is impedance, R the resistance and C the capacitance.

Conductance is the reciprocal of resistance, so sometimes the impedimetric biosensor is also called a conductometric biosensor [78]. Shervedani et al. developed a quantitative method for the determination of glucose based on the EIS measurements. In this method, the GOx was immobilized on the mercaptopropionic acid (MPA) self-assembled monolayer (SAM) modified gold electrode. Parabenzoquinone (PBQ) was used as an electron mediator which is reduced to hydroquinone (H₂Q). The EIS measurements showed that the charge transfer resistance (R_ct) decreases with the increase of the glucose concentration due to the increase of the diffusion current density by the H₂Q oxidation. The nondestructive and straightforward method showed a dynamic range of glucose determination with linear variation of the sensor response (1/R_ct) as a function of glucose concentration in a solution [79]. Recently, versatile biosensing materials such as semiconducting CNTs and conducting polymers have been introduced [80]. Besteman et al. combined the GOx to the sidewall of the single-wall carbon nanotube (SWNT) by the use of a cross linker and found conductance decrease of the SWNT as well as the change of capacitance. The conductance increased upon exposure to glucose indicating that an enzyme-based single molecular level biosensor is available by the use of the SWNT [78]. Very few reports are available on nanostructured metal-oxide modified glucose sensor based on the conductometry or the EIS [81]. Ansari et al. covalently immobilized GOx on the SnO₂ nanostructured thin film grown on anodized aluminum oxide (AAO) pores by the plasma enhanced chemical vapor deposition (PECVD) leading to the conductivity increase of the film. This film showed higher sensitivity towards glucose as compare to other simple nanostrucutres [82]. More improved sensitivity and wider linear response range could be available by tailoring the material properties, for example active surface area, three-dimensional structure, and electrical conductance of the film.

ZnO nanomaterials have been studied extensively in optics, optoelectronics, sensors, and actuators owing to their semiconducting, piezoelectric, and pyroelectric properties [83]. ZnO has many attractive properties for the fabrication of the metal-oxide based biosensors, for example good biocompatibility, chemical stability, non-toxicity, electrochemical activity and fast electron transfer rate. Most of all, the ZnO substrate provides the GOx having many acidic protons with a better electrode surface for immobilization because the isoelectric point (IEP) of the ZnO is about 9.5, while that of the GOx is 4.2 [84,85]. Moreover, high surface-to-volume ratio of nanostructured ZnO gives the immobilized GOx a better electrical contact to the electrode [86,87]. Among many nanostrucuted ZnO, the ZnO nanorods have been widely studied for the immobilization of biomolecules [69,88,89]. Wei et al. introduced the ZnO nanorod grown on a gold electrode by hydrothermal decomposition followed by the iimobilization of GOx in phosphate buffer solution at pH 7.4. Negatively charged GOx in a neutral or basic solution is electrostatically immobilized on to the positively charged ZnO nanorod in the same solution by applying an anodic potential. The modified electrode showed a high and reproducible sensitivity in short response time and apparent Michaelis-Menten constant towards glucose oxidation was 2.9 mM [90].

Nanoporous ZnO and ZnO nanotubes have improved surface-to-volume ratios and also show highly sensitivity towards glucose oxidation. The porous ZnO:Co nanocluster having an average particle size of 5 nm showed much higher sensitivity (about 13.3 μA mM⁻¹ cm⁻²) due to the high specific active sites and electrocatalytic activity of the ZnO:Co nanoclusters as well as strong affinity to the GOx [91]. Yang et al. synthesized porous ZnO nanotube arrays on ITO by two-step electrochemical and chemical processes (Figure 4). The Nafion/GOx/ZnO nanotube arrays on ITO electrode showed good mechanical contact between the ITO substrate and the ZnO nanotubes which leads to the improved sensitivity [83]. Similarly, improved sensitivity with the porous tetragonal pyramid-shaped ZnO nanomaterials and the ZnO nanocomb structure were also reported by other groups [92–94].

Physically or chemically tailored ZnO nanowires also lead to the high specific surface area and high IEP for efficient immobilization of concentrated GOx and the nanowire structure effectively mediates the electron transfer of the redox reaction [69]. Liu et al. developed a carbon-coated ZnO (C-ZnO) nanowire arrayed electrode by taking advantage of electrical conductivity and chemical stability of the carbon material and the one dimensional channel structure of ZnO nanowires [92]. The Nafion/GOx/C-ZnO nanowired electrode exhibited a pair of well-defined redox peaks at −0.43 and −0.48 V, resulted from the direct electron transfer between the immobilized GOx and the electrode. By contrast, no peaks were observed from both Nafion/GOx and Nafion/C-ZnO nanowired electrodes. Meanwhile, only very weak peaks are detected with a Nafion/GOx/pristine nanowired electrode. The EIS measurements confirmed the fast electron transfer at the C-ZnO nanowired electrode with charge transfer resistance of Fe[(CN)₆]^3−/4− was ∼85 Ω, much smaller than that at the pristine nanowired electrode (∼400 Ω). The C-ZnO nanowire could be an inexpensive high-performing alternative to the conventional CNT-modified gold film in biosensor applications.

The nanostructured ZnO shows high sensitivity but very poor stability, because the ZnO nano-structure is easily removed from the electrode surface during functionalization [66,86]. Indeed, improved stability without loss of sensitivity or selectivity is one of the big challenges for glucose monitoring. Extensive efforts have been made to retain the catalytic activity of the immobilized GOx for a long time by using carbon-coated ZnO nanowire [95], functionalized CNTs [96], nanosized CaCO₃ film [97], NdPO₄ nanoparticle-chitosan composite [98], meldolas Blue mediated screen-printed electrodes [99] and many other materials. Wang et al. prepared ZnO nanoparticles and coated them onto a multiwall carbon nanotube (MWNT)-modified electrode for the GOx immobilization to improve the stability. A cationic polydiallyldimethylammonium chloride (PDDA) layer was further coated on the GOx-contained ZnO layer to prevent enzyme leakage [100]. The PDDA/GOx/ZnO/MWNTs film provided the sensing electrode with enhanced sensitivity, lower detection limit and long term stability more than 160 days. Results obtained from this glucose sensor were compared with one hundred human blood serum samples and agreed with the results of conventional spectrometric assay (correlation coefficient, 0.997).

In recent years, monitoring of intracellular glucose levels has received considerable attention and various methods including electrochemical detection [101], use of hypodermic needles [102], and microdialysis [103] have been proposed for the assessment of glucose in interstitial spaces as an alternative site rather than the bloodstream. Recently, Asif et al. developed a functionalized ZnO-nanorod based electrochemical sensor for the selective detection of glucose in human adipocytes and frog oocytes [104]. Hexagonal ZnO nanorods grown on the tip of a silver-covered borosilicate glass capillary make it possible to microinject specific reagents, which can interrupt or activate signal transmission to glucose, into relatively large adipocytes and oocytes. The prepared nanosensor exhibited a glucose-dependent electrochemical potential difference over the concentration range of 0.5–1000 μM versus a micro-sized Ag/AgCl reference electrode with a fast response time (<1 s). Intracellular and extracellular measurements of glucose showed good sensitivity and selectivity without interferences even though the nanosensor encountered stability problems.

Even if ZnO nanomaterials are highly promising electrode materials for glucose sensing, a relatively high potential is still required for operation meaning that unwanted output signal caused by the oxidation of interfering agents such as ascorbic acid (AA) or uric acid (UA), might be usually coexisted with glucose signal in real samples.

Transition metal oxides and alloys significantly enhance direct oxidation of glucose compared with other metals that attribute to the catalytic effect resulting from the multi-electron oxidation mediated by surface metal oxide layers [105,106]. Transition metals such as Cu and Ni can oxidize carbohydrate easily without surface poisoning. Unlike Cu and Ni, corresponding oxides or hydroxides are relatively stable in air and solutions [107,108]. Natural abundance of copper oxide as well as its low production cost, good electrochemical and catalytic properties make the copper oxide to be one of the best materials for electrical, optical and photovoltaic devices, heterogeneous catalysis, magnetic storage media, gas sensing, field-emission emitters, lithium ion electrode and so forth [109–111].

Recent advances in nanoscience and nanotechnology have revealed the catalytic effect of copper oxide in relation to nonenzymetic glucose oxidation, voltammetric sensing of carbohydrates and hydrogen peroxide detection with ultra-sensitive response and good stability [112]. Wang et al. prepared Pd (IV)-doped CuO nanofibers (PCNFs) and CuO nanofibers via electrospinning on glassy carbon electrodes (GCE) and investigated the amperometric direct responses to glucose [113]. The PCNFs modified electrode showed excellent selectivity, reproducibility and stability. The Pd in the PCNFs, with lower electron occupancy in 3d orbital compared to the Cu, act as a Lewis acid for adsorption of polar glucose molecule a Lewis bases, via nucleophilic attack by non-bonded electron pairs. The longer resident time for the reactant species within the electrode-electrolyte interface, the more enhanced electrocatalytic activity towards the glucose oxidation [105,114]. A similar approach was also demonstrated by using three-dimensional network of CuO nanofibers (CuO-NFs) and reported improved sensitivity, stability and fast response time compared to the PCNFs modified electrode [115]. The CuO-NFs modified electrode also shows good selectivity and resistance to electrode fouling. Zhuang et al. developed a highly stable and sensitive nonenzymatic glucose sensor based on copper oxide nanowire modified Cu electrode in an alkaline medium [116]. The CuO nanowire can greatly increase the electrocatalytic active area and promote electron transfer rate of glucose oxidation. The CuO modified electrodes could be used repeatedly and were not contaminated with by-product of glucose oxidation. Experimental data for the glucose detection are in good agreement with the results from the spectrophotometric method performed in local hospital in real sample, where interference effect is negligible.

Recently, the existence of CuO nanoparticles as impurities in a CNT-based electrode has been claimed to be responsible for electrocatalytic activity of glucose oxidation [117] as the synthetic procedures of pure CuO nanowire or nanofiber are tedious and time consuming. In addition, air-sensitive copper substrate can make a big sensor-to-sensor variation when exposed to open air under most environmental conditions. Combination of CNTs with copper oxide nanoparticles enhances the electron transfer rate of Cu₂O and are successfully applied as an enzyme-free glucose Sensor. Very recently, Zhang et al. and Jiang et al. have developed Cu₂O/MWNTs and CuO/MWNTs modified nonenzymatic electrodes for glucose sensing [118,119]. The Cu₂O/MWNTs nanocomposites were prepared on GCEs by a new fixure-reduction method at low temperature. It showed higher sensitivity and lower detection limit as compared to the CuO/MWNTs that attributes to the stability factor of those two different copper oxide films [120]. Very few reports are available on CuO based enzymatic glucose sensors for example, immobilization of CuO-GOx mixture within a carbon paste matrix [121] and immobilization of GOx on flower-shaped CuO nanostructured electrode [122].

Most enzymatic glucose sensors take advantage of the reducing power of hydrogen peroxide produced by the degradation of glucose via the GOx involved catalytic reaction. The oxidation of hydrogen peroxide is accompanied by the application of a bias potential at which, however, coexisting species such as AA are also electroactive [123]. Basically, two options are available to avoid the electrochemical activity of interfering agents. One is to employ a permselective membrane and the other one is to lower the applied potential by using electron mediators [124–129]. The permselective membrane may decrease the sensitivity and may not completely exclude the interfering effect. As an example of the permselective membrane, MnO₂ a strong oxidant has been tested to get rid of interference signals in glucose sensing by Choi et al. [130]. The IEP of MnO₂ is quite low (4–5) at pH ranging from 2.8 to 4.5 but at higher pH, MnO₂ showed favorable environment for the immobilization of biomolecules [131,132].

Turkusic et al. developed an amperometric glucose sensor based on carbon paste electrodes modified with MnO₂ and GOx that operated at pH 9.5 [133]. They also demonstrated glucose degradation mechanism on the MnO₂/GOx modified screen printed electrode (SPE) based on heterogeneous carbon material, as shown in Figure 5. The two enzymatic oxidation products, gluconolactone and H₂O₂ were produced by applying 0.48 V vs. Ag/AgCl. The H₂O₂ further reacts chemically with MnO₂ to produce manganese species having lower oxidation states, which can be electrochemically reoxidized to form MnO₂ and the oxidative current flow is directly proportional to the glucose concentration. This rapid electrochemical process is accompanied by a kinetically slower chemical reoxidation of MnO/Mn₂O₃ coupled with chemical oxidation of H₂O₂. The MnO₂/GOx modified SPE showed partially decreased interfering signals, along with good reproducibility and long term stability.

Poly(diallyldimethylammonium), PDDA/MnO₂ and chitosan/MnO₂ nanocomposites are excellent electrode materials to minimize interference effect of UA and AA at low potential. Xu et al. fabricated a PDDA/MnO₂ nanocomposite on graphite electrode surface at pH 7.0 a favorable pH for the GOx immobilization. As-prepared PDDA/MnO₂/GOx modified electrode is free from interferants when operated at 0.46 V [134]. They also suggested chitosan film containing MnO₂ nanoparticles where the GOx was entrapped to be immobilized into chitosan hydrogel and showed effective suppression of interfering signals of ascorbates [135].

Chen et al. developed a nonenzymatic electrochemical glucose sensor modified with MnO₂/MWNTs [136]. The MnO₂/MWNTs electrode displayed high electrocatalytic activity towards the glucose oxidation in alkaline solutions at 0.3V and also strongly resisted toward poisoning by chloride ions. In addition, the problematic interference that occurs by the oxidation of common interfering species such as AA, dopamine (DA), and UA was effectively suppressed in real sample analysis compare to the unmodified MWNTs.

TiO₂ nanomaterials are basically biocompatible and environmentally-friendly and have been frequently proposed as a prospective interface for the immobilization of biomolecules [137,138], and widely applied in photochemistry [139–141] and electrochemistry [142,143]. Moreover, titanium forms coordination bonds with the amine and carboxyl groups of enzymes and maintains the enzyme’s biocatalytic activity. Nanostructured TiO₂ also provides the enzyme with better immobilization environment by enlarging the surface area [144,145].

Sol-gel technology has been widely explored in the field of chemical sensors and biosensors. Especially, the low-temperature sol-gel process enables encapsulation of heat sensitive biomolecules such as enzymes, protein molecules and antibodies [146–149]. Various sol-gel derived TiO₂ films associated with binding molecules or polymer networks for crack-free porous structures have been characterized and applied to biosensing [150–155]. Viticoli et al. developed a third generation amperometric glucose sensor based on nanostructured TiO₂ on a porous silicon (p-Si) substrate. The Functionalized TiO₂ layers were prepared by the metal organic chemical vapor deposition or by the sol-gel technique at various pressures (0.1–4 torr) and temperatures (300 °C to 800 °C) using Ti(IV) isopropoxide as a starting precursor [156]. Enzyme was directly dip-coated on the TiO₂ modified p-Si substrate. Results showed a good linear response with a few seconds of response time.

Recently, Bao et al. have hydrothermally synthesized a new slack TiO₂ layer having a uniform porous nanostructure by the use of MWNTs template [157]. The TiO₂ nanostructure displayed a big hysteresis loop at high pressure. Abrupt increase of adsorption at high pressure in the nitrogen adsorption and desorption isotherms confirms the presence of porous structure. The porous TiO₂ nanostructure has a large capacity for enzyme immobilization that was proved by observing a big change in shape of the hysteresis loop and a significant decrease of the specific surface area of the nanostructure. The applied potential was 0.45 V at pH 6.6 under a N₂ atmosphere. Although the porous TiO₂ nanostructure-modified GCE strongly depends on pH, it operates at low enough potential to eliminate interference signals of UA and AA.

Fluorescence based glucose sensors have appeared in the literature as an alternative way of continuous monitoring of glucose levels. These sensors are highly specific towards analytes but require built-in probes [158]. A number of reports on reagentless optical biosensors using pH-sensitive fluorescent probes are available [159–163]. Fluorometric detection of dissolved oxygen is available since the oxygen is able to quench the fluorescent probe. Doong et al. developed a novel optical arrayed TiO₂ biosensor for simultaneous detection of anlaytes containing glucose, urea and glutamate in a serum sample [164]. Carboxyseminaphthorhodamine-1-dextran (SNARF-1-dextran), a single-molecule fluorescent probe, was co-immobilized with glucose dehydrogenase on TiO₂ by the sol-gel process. The glucose dehydrogenase decomposes glucose to produce proton and the pH decreases. An enhanced sensitivity at basic condition is primarily attributed to the fluorescent characteristics of the carboxy SNARF-1-dextran that gives strong emission intensity at 630 nm. As-prepared TiO₂ arrayed optical sensor showed still good sensitivity even after stored at 4 °C for 1 month.

Nanostructured CeO₂ is an excellent electrode material because it is a nontoxic, chemically inert and electrically conductive material. It also shows large surface area like other nanostructured materials and good biocompatibility [165–173]. Additionally, CeO₂ can act as electrochemical redox couple that makes it possible to produce a mediatorless glucose sensor. High IEP (∼9.0) and electron-transfer rate constant (18.3 s⁻¹) with a good surface coverage (1.4 × 10⁻¹¹ mol cm⁻²) of CeO₂ allow easier enzyme immobilization and direct electron transfer between the active sites of the GOx and the electrode surface [174,175].

Saha et al. deposited nanoporous CeO₂ thin film on a Pt coated glass plate using pulsed laser deposition (PLD) [176]. GOx was immobilized on to the CeO₂ by the electrostatic interaction between the positively charged CeO₂ and the negatively charged GOx at pH 7. Prior to the enzyme immobilization the GOx/Pt electrode was pretreated electrochemically at 0.8 V to activate the CeO₂ surface and to remove the oxidizable impurities. The resulting GOx/CeO₂/Pt electrode showed a linear response to glucose oxidation with low Michaelis-Menten constant (1.01 mM) indicating enhanced enzyme affinity to glucose. The mechanisms available for the glucose oxidation on the electrode are depicted in Figure 6.

Path A describes direct electron transfer between the GOx and the electrode via oxidation and reduction of CeO₂ in which the CeO₂ is a better electron acceptor than oxygen. This is clearly confirmed by observing no oxidation peak of H₂O₂ [177], whereas, in path B, the electron generated by the glucose oxidation might be coupled with reduction of molecular oxygen followed by reduction of CeO₂ and then finally transferred to the electrode at acidic condition [178]. Sol-gel derived nanostructured CeO₂ film on Au electrode is also available for the GOx immobilization and suggested by Ansari et al. [179]. The GOx physically adsorbed on the CeO₂/Au electrode showed a linear response in the range of 0.5 g L⁻¹ to 4 g L⁻¹. The detection limit was 12.0 μM with a shelf life of 12 days. Recently, various nanostructured CeO₂ such as nanorod, nanocomb, nanocubes, and nanoflower [180,181] have been synthesized. However, no successful reports about glucose detection are available and the potential application of these nanostructures for glucose sensing is of great interest.

Basically, electrode materials should have good electron transport capacity, bioactivities towards target analytes and provide suitable physical or chemical environments for reliable immobilization of biorecognition molecules. Some metal oxides meet those requirements perfectly, but there are still demands for other nanostructured materials including conducting polymers and CNTs to get more enhanced sensitivity and reliability of a glucose sensor. Indeed, composite materials made from two or more nanostructures are another big issue [182–184].

Silica based organic and inorganic nanocomposites are attractive electrode materials since they provide biorecognition molecules with a better entrapment environment and more enhanced electrochemical stability [185]. A mesocellular silica–carbon nanocomposite foam (MSCF) was designed for the GOx immobilization by Wu et al. [186]. The uniformly ordered MSCF showed good biocompatibility, favorable conductivity and hydrophilicity. The narrow pore-size distribution was suitable for the immobilization of not only the GOx but also other redox proteins. A fast electron transfer rate (14.0 ± 1.7 s⁻¹) of the GOx modified MSFC allowed direct electrochemistry and showed a good electrochemical sensitivity to glucose with a linear response range of 50 μM to 5.0 mM with a detection limit of 34 μM at an applied potential of −0.4 V. The low potential for operation prohibits unwanted interference signal in real sample by the oxidation of AA and UA.

Recently, extensive researches have been done on the immobilization of GOx by the use of silica nanocomposites such as unprotected Pt nanoclusters mixed with the nanoscale SiO₂ particles [187], chitosan/SiO₂ nanocomposites attached on the Pt-MWNTs modified electrode [188], TiO₂/SiO₂ nanocomposite [189], sol–gel silica film on a prussian blue modified electrode [190]. Porous SiO₂ nanofiber is also available for the potentiometric biosensor application [191].

Gopalan et al. made a form of silica network on Nafion and subsequently loaded polyaniline grafted MWNTs (MWNTs-g-PANI) on the Nafion–silica nanocomposite for the GOx immobilization [192]. Nafion, silica, MWNTs, and conducting polymers have individually been known as materials for improving the electrocatalytic activity, electron conduction path, sensitivity, and stability of biosensors, respectively [193–195]. The Nafion–silica/MWNTs-g-PANI electrode exhibited a linear response to glucose in the range of 1 mM to 10 mM with a correlation coefficient of 0.9972. The sensitivity was 5.01 μA mM⁻¹ with a low response time (∼6 s). Although the entire fabrication procedure is somewhat complicated and requires high fabrication cost, the Nafion–silica/MWNT-g-PANI composite electrode showed excellent sensing performance with negligible interference from AA, UA, and acetaminophen (AP). The recovery test of the Nafion–silica/MWNTs-g-PANI electrode in real sample was evaluated by the standard addition method, with three times addition of standard glucose solution. Experimental showed that reproducible current response (R.S.D) for three measurements were in the range of 2.9−4.1% with the recovery range of 98.0–105.5.

Nanostructured ZrO₂ is another example for the direct electron transfer between metal-oxide layer and the immobilized GOx for glucose sensing. Because the IEP of ZrO₂ is of 4.15 [196], it is suitable for the adsorption of high IEP proteins. Therefore, any other nanomaterial, which can drop down the IEP of ZrO₂ when it is mixed with ZrO₂ is needed for the application of ZrO₂-based electrode to glucose sensing since the IEP of glucose is of 4.2 [197]. The first amperometric glucose sensor based on sol-gel derived ZrO₂ was reported by Liu et al. [198]. Since then, a numerous efforts have been reported for the development of nano ZrO₂ based biosensors [199–201]. Yang et al. prepared ZrO₂/Pt-PLL (poly-Lysine) and ZrO₂/Pt-PVA (polyvinyl alcohol) to modify pyrolytic graphite (PG) electrodes for the GOx immobilization [202]. Didodecyldimethylammonium bromide (DDAB), a synthetic lipid commonly used as a bio-membrane was introduced to stabilize and protect inorganic nanoparticles. The resulting GOx/ZrO₂/Pt-PVA electrode showed largest reaction activity towards glucose oxidation in the presence of ferrocenium hexafluorophosphate (FcPF₆) as an electron transfer mediator. On the other hand, no enzymatic activity of the immobilized GOx can be observed on ZrO₂/DMSO and ZrO₂/DDAB film. So, the use of colloidal platinum by replacing DMSO and DDAB plays an important role in transferring electrons between GOx and the electrode.

Table 2 summarizes characteristics of the most frequently-used metal oxides what we have discussed in the section 3.1 to 3.7 in terms of EIP, the availability of enzymatic or nonenzymatic sensors, the compatibility with CNTs, conducting polymers or metal nanoparticles, and the application for other biosensors.

Plenty other metal oxides such as NiO, SnO₂, MgO, IrO₂, WO₃, PbO₂, RhO₂, IrO_x, V₂O₅, Fe₃O₄, and RuO_x have been reported and their use, not only for glucose sensing, but also for many other biosensing applications to improve sensing performances attempted [203–213]. In this section we will briefly introduce some of those metal-oxide based sensors, including their applications and future prospects.

In recent years, NiO nanomaterials have received considerable attention due to their broad application to battery cathodes, gas sensors, electrochromic films, magnetic materials and catalysts [214,215]. However, there are only few biosensor applications. Li et al. developed a novel amperometric glucose sensor based on NiO hollow nanosphere [203]. The NiO is suitable for electrostatic immobilization of proteins having low IEP because the IEP of NiO is about 10.7. The hollow-sphered NiO was good responsible for high loading of GOx and showed fast electron transfer with a sensitivity of 3.43 μA mM⁻¹ and a detection limit of 47 μM (S/N = 3).

MgO as a ceramic material has been typically applied to water purification, additives in refractory, paint, and superconductors. Various nanostructured magnesium oxides have been reported [216–218] but only a few reports are available for biosensor applications [219,205]. Umar et al. first introduced the application of MgO as an electrode material for glucose sensing [205]. Polyhedral nanocages and nanocrystals of MgO were grown on silicon substrates via non-catalytic simple thermal evaporation process or on a gold surface and used as immobilization matrices for GOx. The resulting Au/MgO/GOx/Nafion electrode showed a good stability and sensitivity of 31.6 μA μM⁻¹ cm⁻². The response time was less than 5 seconds.

Kotzian et al. reported an amperometric glucose sensor based on rhodium dioxide (RhO₂) modified screen printed carbon electrode for the first time [209]. The GOx were mixed with equal amounts of nafion and then transferred to the RhO₂ modified SPE. The flow injection analysis of glucose at an operating potential of 0.2 V vs. Ag/AgCl showed an excellent detection limit of 0.2 mg L⁻¹ when the optimized flow rate was 0.4 mL min⁻¹ in a 0.1 M phosphate buffer (pH 7.5) solution and the interference effect was surprisingly reduced because the operational potential was relatively very low.