Electrocatalytic and Electroanalytic Investigation of Carbohydrates Oxidation on Gold-Based Nanocatalysts in Alkaline and Neutral pHs

Potassium tetrachloropalladate (II) (K₂PdCl₄, 99%) was provided by Alfa Aesar. Hexachloroplatinic (IV) acid hexahydrate (H₂PtCl₆·6H₂O, ≥37.50% Pt basis), hydrogen tetrachloroaurate trihydrate (HAuCl₄·3H₂O, ≥99.9%), sodium borohydride (NaBH₄, 99%), L-ascorbic acid (AA, ≥99%), potassium bromide (KBr, ≥99%), Nafion suspension (5 wt% in aliphatic alcohols), sodium hydroxide (NaOH, 97%), sodium phosphate dibasic dihydrate (Na₂HPO₄·2H₂O), sodium phosphate monobasic monohydrate (NaH₂PO₄·H₂O), D-(+)-glucose (99.5%), D-(+)-galactose (≥99%), were purchased from Sigma-Aldrich. Lactose monohydrate (analytical reagent grade) was purchased from Fisher. These chemicals were used without further purification. Prior its use, Vulcan XC 72R carbon black (from Cabot) was thermally pretreated at 400°C under nitrogen (N₂) atmosphere as described elsewhere.⁴¹ The anion exchange membrane (AEM) was purchased from Fumatech (Fumasep FAA) and activated by soaking in 0.1 M NaOH overnight (2 h in 1 M NaOH is fine). Carbon Paper (Spectracarb 2050L-1050) was purchased from Fuel Cell Store. The phosphate buffer solution, PBS (0.2 M, pH 7.4), was prepared from Na₂HPO₄·2H₂O and NaH₂PO₄·H₂O solutions (0.2 M). All the solutions were prepared with ultrapure water (Milli-Q Millipore, 18.2 MΩ cm at 20°C) and outgassed by bubbling N₂ prior to electrochemical measurements.

Au-Pt-Pd/C nanomaterials were synthesized according to the BAE procedure for the nanocatalysts preparation, described in previous reports.^31,32 The method consists of the reduction of the metallic precursor(s) in aqueous solution in the presence of bromide ion as a size and morphology regulator. In a typically synthesis of Au_xPt_yPd_z nanoparticles (where x, y and z are the corresponding atomic percentages) supported on Vulcan XC 72R, HAuCl₄·3H₂O, H₂PtCl₆·6H₂O and K₂PdCl₄ as metal precursors and taken in the proportions corresponding to the desired atomic composition of the nanoparticles (total metal weight of 20 mg) are dissolved in MQ water (100 mL, 25°C). Then, KBr was added to the solution (ϕ = n(KBr)/n(metals) = 1.5) under vigorous stirring. After complete solution homogenization, 80 mg Vulcan was added under constant ultrasonic homogenization for 45 min. Then, a freshly prepared reducing agent aqueous solution was added dropwise to reduce AuCl₄⁻, PtCl₆²⁻ and PdCl₄²⁻ species to Au, Pt and Pd. Specifically, NaBH₄ (0.1 mol L⁻¹, n(NaBH₄)/n(metals) = 15) was used for Au-Pt/C and AA (0.1 mol L⁻¹, n(AA)/n(metals) = 7) for Au-Pd/C in order to avoid hydrogen (present in NaBH₄) insertion into the palladium lattice. After 2 hours of operation at 40°C under vigorous stirring, the materials are filtered, rinsed several times with ultrapure water and dried in an oven at 40°C for 12 h. The extensive physicochemical characterization of these Au-Pd and Au-Pt binaries are reported in Ref. 32.

The catalyst ink was prepared by mixing 375 μL MQ water and 50 μL Nafion suspension in a water ultrasonic bath. Then, 4 mg of catalyst powder was added to obtain a homogeneous ink to be coated on different supports for electrochemical tests.

The tests were performed in a conventional three-electrode cell using a potentiostat EG&G PARC Model 362 (Princeton Applied Research). The reference electrode was a daily-made RHE, separated from the solution by a Luggin capillary and a slab of GC (6.48 cm²) acts as a counter electrode. The working electrode consisted of either 3 μL catalyst ink deposited onto a well-polished glassy carbon disk (GC, 0.071 cm², 80 μg_metal cm⁻²)³² or an optimized 2 × 20 μL ink onto carbon fiber (CF^42,43, 2 × 1 cm², 38 μg_metal cm⁻²). The entire tests were conducted at controlled temperature, from 5 to 40°C.

In situ Fourier transform infrared spectroscopy (FTIRS) measurements were performed with a homemade three-electrode electrochemical cell where a slab of GC and RHE act as counter and reference electrodes, respectively. The working electrode consisted of 3 μL of catalyst ink deposited onto a GC disk of 8 mm diameter. The amount of Nafion in the ink was halved and a small amount of ink was deposited to reduce the interference with the IR beam. Excellent reflectivity was obtained by pressing the working electrode against a CaF₂ window to obtain a thin layer of electrolytic solution. The coupling of CV to FTIRS (CV-FTIRS) known as "single potential alteration infrared spectra, SPAIRS" consists in recording the electrode reflectivity R_Ei at different potentials E_i in steps of 0.05 V at a quasi-steady-state scan rate of 1 mV s⁻¹. The coupling of chronoamperometry (CA) to FTIRS (CA-FTIRS) relies on a spectrum acquisition every Δt = 3 min at a fixed electrode potential. A setup of the μ-AUTOLAB Type III (Metrohm Autolab BV, Netherlands) potentiostat was used for electrochemistry and OPUS software (Bruker) was used for IR spectroscopy by using a Bruker IFS 66v spectrometer as described elsewhere.^44,45

Bulk solution electrolysis was performed at different electrode potentials (from 0.2 to 0.8 V vs. RHE, EG&G PARC Model 362 potentiostat) by using a Pyrex H-type cell separated with the AEM to prevent the contents mixing and to provide the current relay through ion exchange between both compartments. The auxiliary electrode (GC, 22 cm²) compartment contained the electrolyte and the working electrode (Carbon Paper coated with the catalytic ink: 2 × 4 cm², 24 μg_metal cm⁻²) compartment is filled with 43 mL of the substrate (glucose, galactose, lactose) solution, which is gently stirred during electrolysis by a bar magnet. The sampling was obtained by high-performance liquid ionic chromatography (HPLIC, Dionex) in gradient elution with a conductivity detector (elution by conductivity strength) and an amperometric detector (allows in situ electrolysis for qualitative and quantitative analyses). The HPLIC included an AS50 autosampler and a 2 × 250 mm column (IonPac AS15), operating at 30°C. A constant flow of eluent (0.3 mL min⁻¹ at pressure of ∼1500 psi) is provided by a pump (ICS-5000 P).

Since the CFs support cannot be used for spectroelectrochemical analysis, the catalytic ink was coated onto a vitreous GC for in situ FTIRS experiments, firstly in 0.1 M NaOH and secondly in PBS (0.2 M, pH 7.4). Figure 1a displays the chemical structures of the predominant forms in an alkaline solution (25°C) of glucose, galactose, lactose and the potential reaction products. Please, note that the structure of galactonate is similar to glucose, except the spatial position of –OH at C4-position. Before running spectroelectrochemical analysis, i.e., coupling in situ FTIRS to electrochemistry to probe the electrooxidation reaction and subsequently provide relevant qualitative data about the nature of intermediates/products, we first recorded FTIR spectra of the substrates in both media as references for band assignment. Figure 1b depicts the corresponding ex situ FTIR spectra in alkaline media showing the infrared specific bands for glucose and products assuming an exchanged number of electrons n_ex = 2 (gluconolactone, gluconate), n_ex = 4 (glucuronate) and n_ex = 6 (glucarate). Gluconolactone is well evidenced by its νC=O specific band at 1742 cm⁻¹ (stretching vibration mode). The specific COO⁻ function of an oxidized carbohydrate is characterized by the two intense bands at 1584 and 1413 cm⁻¹ that correspond to the asymmetric stretching vibration ν_as(O–C–O) and symmetric ν_s(O–C–O), respectively. However, a thorough observation of spectra shows a dilemma for assigning these bands. Specifically, depending on the oxidized carbon in the glucose molecule (Figure 1a), different products having these bands are obtained, gluconate (carbon C1-position oxidation), glucuronate (carbon C6-position oxidation) and glucarate (both C1-position and C6-position oxidation). Interestingly, only glucose, gluconolactone and gluconate have the –CH₂- function in C6-position. Thus, its band (deformation vibration mode) is a powerful tool for a precise and unambiguous identification since the position of δ(CH₂) is ∼1368, 1365 and 1352 cm⁻¹ (±5 cm⁻¹) for glucose, gluconolactone and gluconate, respectively. It was observed (not reported herein) that in comparison to gluconate, galactonate shows identical spectra and in the other side, the O–C–O vibration of the osidic bridge of the lactobionate⁴⁶ (from lactose oxidation) is situated in 1635–1660 cm⁻¹ range. In PBS (Figure 1c), the IR bands of the phosphates are situated at 990 cm⁻¹ for ν_s(PO^{3 −}₄), 1080 cm⁻¹ for ν_as(PO^{3 −}₄) and 1180–1200 cm⁻¹ for δ(HPO^{2 −}₄).

Zoom In Zoom Out Reset image size

Figure 1d shows the CV profiles of Pd/C and Pt/C electrodes highlighting that the voltammograms recorded in the IR cell are typical of those obtained in a conventional three-electrode electrochemical cell. Thus, it can be argued that all processes are typical of the involved reactions. The shape of Pd/C corresponds to Pd nanoparticles wherein the oxidation of the palladium surface occurs at E ≥ 0.63 V vs. RHE in the positive-going sweep (peak A4) followed by the reduction of this PdO at ca. 0.65 V vs. RHE (peak C4) and the hydrogen adsorption on the released Pd active sites takes place at E < 0.42 V vs. RHE (peak C3) during the negative-going sweep. The shoulder A3 centered at 0.53 V vs. RHE is attributed to the hydrogen desorption process. For Pt/C electrode, the hydrogen desorption from Pt sites occurs during the positive potential-going scan between 0.05 and 0.4 V vs. RHE (peaks A1 & A2), followed by the double layer charging current up to 0.7 V vs. RHE and finally the Pt surface oxidation at higher potentials (region A5). In the reverse scan, after the Pt oxide reduction – corresponding to the cathodic peak centered at 0.7 V vs. RHE, peak C5 – the Pt free surface becomes covered by adsorbed hydrogen in the lower potential region from 0.4 to 0.05 V vs. RHE (peaks C2 & C1). For the electrochemical hydrogen adsorption (or desorption) process, the number and position of different peaks depend not only on the crystallographic orientation (hkl) of the surface, for instance (100), (110), and (111) for the low indexes but also on the nature of the metal. For instance, the hydrogen desorption on Pd⁴⁷ surfaces occurs in the same potential range leading to a broad peak, while separate peaks are routinely observed for Pt⁴⁸ in alkaline media.

It is generally accepted as a fait accompli that CV profiles for organic molecules oxidation on metal catalysts show at least one peak during the positive and one peak for the negative scanning of the electrode. To robustly and qualitatively elucidate here the exact nature of the backward peak, we designed specific experiments to reveal whether it belongs to "strongly adsorbed reaction intermediates" or to new reactant molecules coming from the solution. On the basis of our previous screenings on monometallic Au/C¹³ and bimetallic Au₉₀Pd₁₀/C^12,49 and Au₆₀Pd₄₀/C³⁵ materials, we examine herein other Pt/C, Pd/C and Au_100-xM_x/C (M = Pt, Pd) electrodes in order to rationalize the trends. The forward and reverse configurations were performed by starting at either the lower (0.05 V vs. RHE) or upper (1.40 V vs. RHE) potential limit at a quasi-steady-state scan rate of 1 mV s⁻¹ (Figure 2a) and by recording the electrode reflectivity R_i at different potentials E_i, each separated by 50 mV intervals (Figures 2b–2e). The first procedure starts at 0.05 V vs. RHE and spectra are recorded during the forward going scan from 0.05 to 1.4 V vs. RHE (Figure 2b) and backward going scan from 1.40 to 0.05 V vs. RHE (Figure 2c). The second procedure starts, with a new electrode and solution, at 1.40 V vs. RHE and spectra are recorded during the forward going scan from 1.40 to 0.05 V vs. RHE (Figure 2d) and backward going scan from 0.05 to 1.40 V vs. RHE (Figure 2e).

Zoom In Zoom Out Reset image size

Figures 2a to 2c display the screening results with Au₈₀Pt₂₀/C. Figure 2a shows that the glucose oxidation reaction (GOR) starts at E ≤ 0.2 V vs. RHE substantiated by the synchronous appearance of intense bands at 1588 and 1411 cm⁻¹ assigned to the specific COO⁻ function of an oxidized carbohydrate. While the positive band at 1080 cm⁻¹ indicates glucose consumption through its dehydrogenated species, the water amount in the thin-electrolyte during the reaction is demonstrated by its δ(H₂O) at ∼1660 cm⁻¹. Remarkably, no C–C bond cleavage compounds (CO: 1900–2100 cm⁻¹, carbonate: 1396 cm⁻¹, CO₂: 2340–2350 cm⁻¹) were observed. It should be noted that the weak signal around 2345 cm⁻¹ results from the oxidation of the support, i.e. Vulcan (this was evaluated during the control and blank tests with Vulcan and NaOH only). Additionally, gluconolactone (νC=O specific band at 1742 cm⁻¹) is not visible from the spectra. Since the same bands are obtained from the two approaches, the oxidation peak during the reverse scan corresponds to fresh molecules oxidation. To validate and extend the reaction selectivity on the entire range of bimetallic materials, we recorded CV-FTIRS. It was remarkably observed that the spectra present the same bands. Specifically, the obtained SPAIRS results at an electrode potential of 0.7 V vs. RHE and collected on different electrodes during the forward-going scan (Figure 2f) provide the direct evidence of quasi-identically nature of reaction product(s). It is worth mentioning that the same type of spectra was obtained for galactose electrooxidation, which would suggest that the reactivity of this glucose's isomer did not depend strongly on the nature of catalysts.

The aforementioned set of data indicates a highly selective oxidation at the whole electrode potential range without C–C bond breaking and a fast gluconolactone hydrolysis, since it is an unavoidable intermediate. Conclusively, the band at ∼1350 cm⁻¹ is rationally assigned to the deformation vibration mode δ(CH₂) of gluconate. Taken together, these clues underpin a drastic change of the reaction kinetics with the revisited BAE protocol since C–C bond cleavage compounds and gluconolactone were observed on other electrodes based on nanoscale materials.^52,53 In summary, it can be qualitatively concluded that glucose is selectively oxidized to gluconate according to a 2-electron pathway.

The voltammetry plays a pivotal exploration role wherein series of CV experiments help to determine the oxidation/reduction potential of the substrate, thereby allowing a precise quantitative study under either galvanostatic or potentiostatic conditions. Thus, having successfully examined the spectra as a function of the electrode potential by SPAIRS during our forays toward the electrooxidation, we next spectroelectrochemically probed the reaction at different fixed electrode potentials during the chronoamperometry (CA). Yet, the thin-layer electrolyte configuration does not allow a good mass transport, but it is scientifically sufficient enough for product(s) identification purposes (qualitatively). Figure 3a displays the typical behavior of I vs. t curves at an applied potential of E_appl = 0.8 V vs. RHE along with the recorded spectra in Figures 3b–3d. The fast current delay during the first minutes is rationally assigned to the thin-film configuration and the obtained histograms of current density distribution in the inset do not represent the real short-term stability trend. In order to establish universal selectivity-electrocatalyst-potential relationships, different electrode potentials from 0.2 to 1.2 V vs. RHE were examined (two potentials are displayed in Figure 3) and the same current and spectra profiles were obtained. Specifically, the same set of bands, 1583, 1412, 1350, and 1080 cm⁻¹ demonstrate that the nature of the product(s) is identical to the previous methodology.

Zoom In Zoom Out Reset image size

The previous findings show only, at a qualitative level, the remarkable selectivity in an alkaline medium. Otherwise, the nature of the reaction product can no longer be extended a priori to other aqueous media. Given the great interest of metallic nanoparticles at pHs close to biological conditions for sensing or other "abiotic" purposes,^{33,34,52–58} we subsequently turned our spectroelectrochemical apparatus out to elucidate the nature of the reaction product. Compared to the investigations in a "relatively clean electrolyte" that is basic NaOH solution, the translation of spectroelectrochemistry in pH close to 7 is prohibitively challenging with added factors such as the low activity of metals toward organics electrooxidation and the presence of species such as phosphates which can interfere since they have vibration bands in the considered spectral domain. The acquisition of the FTIR data of the buffer solution and subsequent reagents, as well as the supposed reaction products, thus preceded the initiation of the reactions at a quasi-steady state scan rate of 1 mV s⁻¹. The obtained results are gathered in Figures 4a–4c for CV-FTIRS (SPAIRS) and Figures 4d–4f for CA-FTIRS. The first observation is that the activity is lowered at pH 7.4 compared to the data recorded at pH 13 even if with a tenfold substrate concentration, which underpins the crucial role of hydroxyls.^59,60 The characteristic bands of gluconate (Figure 4b) and galactonate (Figure 4c) are observed at 1410 and 1588 cm⁻¹. Interestingly, the characteristic vibration band of ν(C=O) for gluconolactone at ca. 1744 cm⁻¹ can be kinetically translated as a slow lactone hydrolysis at pH 7.4 in comparison to pH 13. The second remarkable observation is the presence of a second vibration band ν(C=O) at 1780–1785 cm⁻¹, which is typical of carboxylic acids. This is therefore indicative of a local decrease of pH that would result in a rapid protonation of gluconate to gluconic acid given its pK_a of 3.75.⁶¹

Zoom In Zoom Out Reset image size

To support such explanations, we performed CA-FTIRS experiments (Figure 4d) wherein an increase in the intensity of the band at 1780–1785 cm⁻¹ for –COOH over time is clearly evident (Figure 4e). The set of spectra after 30 min are shown in Figure 4f where the intensity of the band due to gluconolactone at 1744 cm⁻¹ decreases gradually while those at 1588 cm⁻¹ (gluconate) and 1780 cm⁻¹ (gluconic acid) increase over time. The second characteristic substantiating the formation of –COOH acid form in lieu of a –COO⁻ carboxylate is a very broad band around 3000 cm⁻¹ (not shown here) assigned to the carboxylic acid's –OH elongation. This first spectroelectrochemical work in buffer at pH 7.4 with nanomaterials enlightens us of the exact nature of the final product in neutral pHs.

It should be noted that a prominent local pH depletion has been firstly reported in 2010 by Camara's group⁶² before confirmation by several groups including us and now widespread in alkaline media as a fait accompli for glycerol electrooxidation enabling the detection of CO₂ (pKa = 6.4, 10.3 for the dissolved CO₂ as H₂CO₃).^44,63–65 To explain the origin of this surprising phenomenon, it is mainly argued that the thin-layer configuration (about 1 and 10 μm)^66–68 of the solution during the spectroelectrochemical study induces a significant consumption of OH⁻ during the organic molecule oxidation.

Having demonstrated the ability to use spectroelectrochemistry as a powerful tool to probe the electrochemical reaction, we sought to study the reaction itself more carefully at different temperatures. Such investigation is crucial for the assessment of the electrochemical activation energy (E_a), an important reaction kinetic parameter indicating the activation energy barrier to overcome before the reaction occurs. A qualitative examination of the polarization curves at model electrodes of Pt/C (Figures 5a–5c) and Au/C (Figures 5d–5f) for different mono- and di-saccharides, glucose (Figures 5a and 5d), galactose (Figures 5b and 5e) and lactose (Figures 5c and 5f) shows that I increases with T for E > 0.4 V vs. RHE, indicating that the electrochemical reaction at the surface of these catalysts is exothermic (E_a > 0). For the particular case of Pt/C, the peak current between 0.2–0.4 V vs. RHE decreases as the temperature increases. This therefore reflects a thermodynamically unfavorable process (endothermic, E_a < 0) that has been firstly reported by Essis-Yei et al.³⁶ in 1988 on bulk Pt electrode. In order to access the value of E_a, the electrochemical formulation of the Arrhenius relationship was used (Equation 1).⁶⁹

j: current density. E: electrode potential. Δ_rG: Gibbs' free energy (J mol⁻¹). Δ_rH°: standard enthalpy or chemical activation energy (J mol⁻¹).  Δ_rS°: standard entropy (J K⁻¹ mol⁻¹). A or A': pre-exponential factor. α': effective symmetry factor (related to properties of the electric double layer: α' = αn is used because the number of electrons n of the rate limiting step, rds, is unknown). R: universal gas constant (8.314 J K⁻¹ mol⁻¹). T: temperature in Kelvin (T = 273.15 + θ, where θ is the temperature in °C). F: Faraday's constant (96485 C mol⁻¹). : electrochemical activation energy (J mol⁻¹).

Zoom In Zoom Out Reset image size

The value of E_a = can be determined from the Arrhenius' plots (Equation 2). Figures 5g to 5i depict the trends of E_a as a function of the oxidation potential (E_ox). Overall, E_a < 40–50 kJ mol⁻¹, suggesting that the reaction kinetics is not controlled by the reactant adsorption but by the reactants/products diffusion processes.⁷⁰ Since , it was theoretically expected that E_a vs. E_ox follows a linear distribution. However, the fitting result shows a polynomial distribution, supported by the theoretical calculations of Protsenko and Daminov.^71,72 To date, the dependency of E_a to E_ox has been observed for methanol electrooxidation at Pt bulk,⁷³ and polynomial trend has been recently observed for glucose electrooxidation at pHs 7.4⁷⁴ and 13.³³ Herein, we hypothesize that the effective symmetry factor α' ( = αn) is a E_ox-dependent parameter, which explained the polynomial distribution of E_a.

Based on the qualitative data obtained by spectroelectrochemistry, we next used a tandem of electrolysis and chromatography to electro-oxidize, and identify the main stable reaction products from these carbohydrates. Figure 6a compares the results in terms of oxidation current (I_ox, left y-axis) and quantity of electricity (Q_ox, right y-axis), wherein the progressive decrease in I_ox is ascribed to the reactant consumption and/or to catalyst deactivation. The deactivation results from the strongly adsorbed reaction intermediates/products that block the catalytic surface.^12,44 The trends in I_ox and Q_ox indicate that Au/C electrode material behaves as the best stable monometallic materials compared to Pt/C that is rapidly deactivated. The bulk electrolysis was carried out at different applied potentials of E_appl = 0.2, 0.4, 0.6, and 0.8 V vs. RHE. Figure 6b shows the amperometric (and coulometric) I_ox (and Q_ox) vs. t traces at E_appl = 0.8 V vs. RHE with the overall goal of comparing the efficiency at a given electrode potential rather than enabling individual catalysts to operate at their specific peak current. The observed trend is explained by the fact that at this potential, for 50 mM substrate, the polarization curves I vs. E of Au/C exhibits nearly its maximum peak current, while those of bimetallic electrodes are situated at E < E_appl = 0.8 V vs. RHE. HPLIC analysis of samples after the bulk electrolysis (Figure 6c) highlights the presence of two unresolved peaks at a retention time of t_R = 5.5 and 6.1 min, nearly coinciding with gluconate (galactonate), however very far from glucuronate (t_R = 10 min) and glucarate (t_R = 21 min) as previously reported.^13,35 After a thorough study on a monometallic Au/C electrocatalyst, it was argued that the peaks A and B belong both to the same product, gluconate.¹³ Indeed, the LC-MS analysis (Figure 6d) after acidification and lyophilization show that the obtained spectra belong to the lactobionic, galactonic and gluconic acids. Basically, LC-MS analysis in negative ionization mode allows to distinguish the different hypothesized acids based on the mass of their pseudo-molecular ions (M–H)⁻: m/z = 195 (gluconic), 195 (galactonic), and 357 (lactobionic). The other peaks at m/z = 129 and 391 are attributed to the fragmentation and simple dimerization processes, thus confirming the feature of a unique reaction product for which n_ex = 2. Our findings conclusively provide direct evidence that the oxidation happens exclusively at carbon in the C1-position and involves two electrons, thus underpinning that electrochemistry can be elegantly used for a selective saccharides oxidation in aqueous media.

Zoom In Zoom Out Reset image size

Next, we investigated the electrochemical kinetics to compare the exchange current (I₀) or exchange current density (j₀), normalized to the geometric surface area of the electrode (Equation 3). Prior to that, the short-term stability tests performed by CA at different electrode potentials from E_appl = 0.5 to 0.9 V vs. RHE in 0.1 M NaOH with 10 mmol L⁻¹ substrate (Figure 7a: glucose), (Figure 7b: galactose) and (Figure 7c: lactose) show that the bimetallic materials exhibit an enhanced electrocatalytic performance for mono-saccharides.

E: electrode potential. E_eq: equilibrium potential. η: overpotential. j: current density. j₀: exchange current density. α': effective symmetry factor (related to properties of the electric double layer: α' = αn is used because the number of electrons n of the rate determining step, rds, is unknown). R: universal gas constant (8.314 J K⁻¹ mol⁻¹). T: temperature in Kelvin (T = 273.15 + θ, θ(°C)). F: Faraday's constant (96 485 C mol⁻¹). b: Tafel slope (mV dec⁻¹).

Zoom In Zoom Out Reset image size

Figure 7d displays a typical Tafel plot from the polarization curve (no iR correction) recorded at a quasi-steady state scan rate of 2 mV s⁻¹ from which the effective symmetry factor α', the Tafel slope b and j₀ can be determined. The metric of j₀ (function of b and E_eq) depends only on the catalyst material and electrolyte. Thus, it reflects the ability of electron transfer from the substrate to the electrode for an oxidation reaction. Technically, a catalytic material having a high j₀ and a small b is desirable since a smaller value of b means that increasing the same current density (dec⁻¹) requires smaller overpotential (mV), implying faster charge transfer kinetics. Depending on the investigated materials, the determined values of α' range from 0.1 to 0.5 and b from 100 to 300 mV dec⁻¹ therefore confirming the complexity of the electrooxidation reactions (the three substrates were probed), which might involve the reactant(s) adsorption, electron transfer and reactant(s)/product(s) diffusion as rds. Typically, α' is ∼0.5 and leads to: (i) b = 120 mV dec⁻¹ after the potential drop (iR) correction and n = 1 as transferred number of electrons during the rds and (ii) b = 60 mV dec⁻¹ for a pseudo two-electron (n = 2) reaction as the rds.⁷⁵ Since no iR compensation (may contribute up to tens of mV)⁷⁶ was applied to the present data, the results could suggest the first-electron transfer during the oxidation of saccharides as rds or a pseudo two-electron reaction as the rds. However, according to International Union of Pure and Applied Chemistry (IUPAC, Technical Report 2014), the simultaneous transfer of more than one electron is considered to be highly improbable.⁷⁷ The determined value of j₀ at low η (E ≤ 0.55 V vs. RHE) and high η (E > 0.60 V vs. RHE) for glucose electrooxidation in 0.1 M alkaline media support the good electrochemical kinetics of the bimetallic materials. Specifically, Au₉₀Pd₁₀/C is the most active with j₀ > 0.5 mA cm⁻². However, no data is available in the literature for comparison under alkaline environments. We finally evaluated the kinetics at pH 7.4 on GC (0.07 cm²) and CF (2 cm²) supports by using the metrics of j₀ and I₀ because of different geometric surface areas. For Au₉₀Pd₁₀/C, the obtained exchange current at CF support is larger in comparison to Au/C, which indicates a substantial "support effect" from CF on the catalytic activity. Besides, no rational comparison can be made between GC and CF. It should be noted that GC (very expensive, very small electrode area) is only interesting for examining fundamental aspects while CF mats can be fabricated at large scale for real applications.

We finally combined the in/ex situ investigations on the monometallic and bimetallic materials to revise and generalize the previously established catalytic reaction cycle for glucose electrooxidation in alkaline medium on Au/C¹³, as displayed in Scheme 1. This scheme is valid for galactose and lactose since only the carbon at the C1-position is involved in the catalytic cycle. Owing to its good spatial arrangement,⁷⁸ the adsorption of the β anomeric form (1) leads to the intermediate 2 (rate constant k_1→2) before yielding the adsorbed lactone (3) after a second electron transfer with a rate constant k_2→3/3'. Then, the lactone undergoes either desorption (5, rate constant k_3/3'→5) and then hydrolysis to gluconate (7, rate constant k_5→7) or a self-rearrangement to yield an adsorbed gluconate (4 or 4') at k_3/3'→4/4' as a rate constant before being desorbed as a final product (7: rate constant k_4/4'→7). It has been argued that (4') leads to an enhanced antisymmetric O–C–O stretching vibration (ν_as = 1588 cm⁻¹), while (4) favors the symmetric motion (ν_s = 1410 cm⁻¹).⁷⁹ Based on the collected spectroelectrochemistry data in alkaline medium, herein, it can be concluded that gluconate is weakly adsorbed on the surface by either only one oxygen (4) or two oxygen (4'). However, no rational conclusion can be made in neutral pH given the weak intensity of the spectra. Furthermore, the intermediate (6) is likely obtained from (7) due to a local pH decrease in neutral pH than a direct lactone hydrolysis (rate constant k_5→6). A significant consumption of OH⁻ during the electrolysis with high glucose concentration (> [NaOH]) may lead to a mixture of (6) and (7) as final and stable products only if the pH at the end is above their pK_a that is 3.75. It should be noted that spectroelectrochemistry enables the detection of the reagents, intermediates (stable enough) and products, while basic analytic methods (HPLIC, LC-MS, FTIRS) allow detecting only the very stable products. This reaction scheme could be likely extended to other monosaccharides as well as polysaccharides.

Scheme 1. Proposed scheme based on the collected electrochemical, spectroscopic and spectrometric data during the electrochemical oxidation of glucose (also valid for galactose and lactose) in aqueous media. Cat. refers to a catalytic species involved in the reaction, which can be a pure metal M, M(OH)_x or MO_x from Au-Pt or Au-Pd nanomaterials.