Hydrazine Electrooxidation with PdNPs and Its Application for a Hybrid Self-Powered Sensor and N₂H₄ Decontamination

Polyvinylpyrrolidone (PVP, MW ∼ 55,000 g mol⁻¹), L-ascorbic acid, potassium bromide, potassium tetrachloropalladate(II), 1-pyrenemethylamine hydrochloride (PMA), 5,5'-dithio-bis(2-nitrobenzoic acid) (DTNB), N-hydroxysuccinimide (NHS), hydrazine solution (N₂H₄; 35 wt% in H₂O) and 1-methyl-2-pyrrolidone (NMP), disodium citrate, and sodium phosphate were obtained from Sigma Aldrich and used as received. N-(3-Dimethylaminopropyl)-N^'-ethylcarbodiimide hydrochloride (EDC) was purchased from Fluka. Bilirubin oxidase (BOx) from Myrothecium verrucaria (E.C. 1.3.3.5) was purchased from Sigma Aldrich. Reagent grade acetonitrile (CH₃CN) was received from VWR Chemicals Commercial grade thin multi-walled carbon nanotubes (MWCNTs) (NC3100, 9.5 nm diameter, 1.5 μm length, purity > 95%,) were obtained from Nanocyl (Belgium). Gas diffusion layers (GDL) were non-treated carbon cloth electrodes and were purchased from PAXITECH (France). Distilled deionized water (15 MΩ) was obtained from a PURELAB ELGA purification system. Oxygen and argon were purchased from Air Liquide (France).

The anodes used for N₂H₄ oxidation were made by layer-by-layer deposition of MWCNTs with cubic PdNPs. A 2.5 mg mL⁻¹ MWCNT dispersion in NMP was prepared by 30 min sonication to achieve a homogeneous black suspension. The black suspension was deposited onto the glassy carbon electrode (GCE) to give an electrode disc with a geometric surface area of 0.126 or 0.196 cm² for 10 or 20 μL droplet deposits, respectively, followed by deposition of various loadings of PdNPs (4 mg mL⁻¹ in NMP). The load of PdNPs was adjusted by depositing different solution volumes onto the electrode surface. The distribution of the deposited cubic PdNPs on the MWCNT electrode surface was verified by scanning electron microscopy (SEM) with a QUANTA FEG 250 (FEI) without coating the substrates with metal. Secondary electron and backscattered electron detection modes were used to assess topography and chemical composition of the samples, respectively. The morphologies of the cubic PdNPs were previously characterized by transmission electron microscopy (TEM) using a JEM-1011 TEM transmission electron microscope operated at 100 kV. For the analysis, one aliquot of the dispersed samples in ethanol was deposited onto a 400-square mesh copper grid with carbon film and dried under vacuum.

MWCNT enzyme-modified electrodes were prepared from a 2.5 mg mL⁻¹ suspension in NMP after 20 min sonication and vigorous shaking immediately before use. The black suspension was deposited onto the GCE to give an electrode disc with a geometric surface area of 0.196 cm² for a 20 μL droplet. The solvent was then removed over several hours under vacuum. Next, the electrodes were incubated in a 10 mmol L⁻¹ PMA solution in CH₃CN for 1 h and thoroughly rinsed with CH₃CN then deionized water. Subsequently, electrodes were left to incubate in a 4 mmol L⁻¹ DTNB, 4 mmol L⁻¹ NHS and 2 mmol L⁻¹ EDC solution (0.2 mol L⁻¹ phosphate buffer (PB), pH 7.0) for 5 h. The electrodes were rinsed with deionized water to remove any loosely bound species then soaked in McIlvaine buffer (pH 7.0). Modified-(DTNB-Pyr)/MWCNT electrodes were then incubated overnight at 4°C in 40 μL of BOx solution (2.5 mg mL⁻¹) prepared in McIlvaine buffer (pH 7.0). Before testing, electrodes were thoroughly rinsed with deionized water and soaked in fresh McIlvaine buffer solution. For preparation of GDL-based air-breathing biocathodes for sensing, the same preparation methodology was applied as that used for GC electrodes. To ensure the biocathode was not the limiting electrode, three droplets of 20 μL each were deposited onto the GDL surface. The geometric surface area of the GDL-based biocathodes was S_GDL = 0.378 cm².

A Biologic VMP3 Multi Potentiostat controlled with EC-lab software was used to perform electrochemical experiments. A three-electrode setup was used for half-cell characterization comprising a glassy carbon electrode (Ø = 3 mm, ALS Co., Japan) as the working electrode, platinum wire as the counter electrode, and a saturated calomel electrode (SCE) as the reference electrode. Prior to modification, the GCE working electrodes were polished with a 2 μm diamond paste purchased from Presi (France), rinsed successively with deionized water, acetone, and ethanol, then left to sonicate for 5 min in ethanol. McIlvaine buffer solutions (pH range: 3 ≤ pH ≤ 9) were used as supporting electrolytes. All solutions for N₂H₄ oxidation experiments at the half-cell anode were purged with nitrogen gas and a nitrogen flow was maintained on top of the solution during analysis. All experiments were conducted at room temperature unless stated otherwise. Catalytic onset potentials were determined from the foot of the electrocatalytic wave.

The fuel cell/sensor set-up comprised of a rounded poly(methyl methacrylate) tube and a flat poly(methyl methacrylate) piece assembled over a rubber joint to form the fuel cell reservoir (see Figure S1). For the air-breathing cell, a 5 mm diameter hole was made into the flat piece to expose the BOx-modified air-breathing biocathode to air. The complete hybrid biofuel cell and sensor system consisted of a PdNPs/ MWCNT anode and a BOx/(DTNB-Pyr)/MWCNT biocathode or BOx/(DTNB-Pyr)/MWCNT/GDL biocathode. The distance between the electrodes was approximatively 1 cm. Oxygen flow was qualitatively maintained but not monitored to obtain reproducible steady state currents between different experiments. When the air-breathing GDL biocathode was used, no additional O₂ was provided to the electrolyte. Testing was performed with a relatively slow scan rate of 2 mV s⁻¹ using linear sweep voltammetry. All experiments were performed in a 5 mL volume of McIlvaine buffer (pH 7.0) solution.

The cubic PdNPs used here with average size of 22.5 ± 2.6 nm were synthesized and characterized as described in our recent work.²⁰ Figure 1A shows a TEM micrograph and the corresponding correlated size distribution for the PdNPs, which confirms good dimension homogeneity and a high yield of cubic shapes. SEM imaging was subsequently performed to assess the presence and distribution of the PdNPs on the MWCNT-based electrodes prepared by deposition from the nanoparticle solution as described in the Experimental section. Figure 1B and 1C show SEM micrographs recorded at an unmodified MWCNT electrode and a PdNP-modified MWCNT electrode, respectively. A well dispersed and homogeneous three-dimensional CNT network is observed for the unmodified MWCNT electrodes. For the PdNPs-modified MWCNT electrodes, the CNT network was maintained and the appearance of bright spots associated with individual and aggregated nanoparticles were observed at high magnification. The relatively uniform distribution of bright spots was observed at multiple regions of the surface indicating an even coverage of PdNPs across the MWCNT electrode surface. The increased brightness observed across parts of the carbon nanotube network for the PdNP-modified MWCNT electrodes is attributed to polymeric material associated with the as-prepared nanoparticle solution. Overall, the electron microscopy analysis confirms the presence and good homogeneous distribution of PdNPs on the surface of the modified electrodes.

Zoom In Zoom Out Reset image size

Figure 2A shows typical cyclic voltammograms (CVs) recorded at unmodified MWCNTs and PdNPs/MWCNT electrodes with cubic PdNPs (22.5 nm) in deaerated McIlvaine buffer (pH 7.0) solution. The CV for bare MWCNT electrodes did not present any faradaic signal when scanning in the range from −0.85 V to 0.3 V vs. SCE at 10 mv s⁻¹. For the PdNP-modified MWCNT electrodes, the voltammograms in the same range reveal the appearance of electroactivity between −0.7 to −0.4 V. The redox peaks observed at negative potentials are ascribed to the characteristic peaks of hydrogen adsorption and desorption on palladium, as reported in different literature reports.^21,22 Upon scanning of the PdNPs/MWCNT electrode at more positive potentials, a broad oxidation wave was observed with an onset potential of around 0.5 V and current maximum at 0.6 V, corresponding to hydroxide (PdOH) and oxide (PdO) formation on the Pd surface, as reported previously.²¹ After formation of the oxide, a new reduction peak appeared on the reverse sweep near 0.0 V, corresponding to palladium oxide reduction. The reduction peak currents were found to be linearly dependent on the scan rate, sustaining the presence of an adsorbed species on the MWCNT electrodes and, furthermore, confirming the good adhesion of the nanoparticles on the MWCNT surface (Figure 2B).

Zoom In Zoom Out Reset image size

The hydrogen evolution region for MWCNT electrodes modified with cubic PdNPs was investigated in McIlvaine buffer over a pH range from 3.0 to 9.0. Figure 3 shows second scan CVs recorded in N₂ saturated solution at different pHs. The oxidation peak potentials shifted in a linear manner toward more negative values with increasing pH, giving an average slope of −55 ± 2 mV per pH unit.

Zoom In Zoom Out Reset image size

In order to maximize the surface coverage of PdNPs we investigated the effect of PdNP loading on the MWCNT electrode by monitoring the increase in peak current corresponding to dihydrogen oxidation. Figure 4 depicts the anodic current peak height recorded at −0.45 V as a function of the volume of PdNPs drop-casted for MWCNT electrodes prepared from two different volumes of 2.5 mg mL⁻¹. The palladium loading on the PdNPs/MWCNT-GC modified electrodes investigated therefore varied. The plots in Figure 4 reveal that the dihydrogen oxidation current obtained for electrodes prepared by addition of between 1–5 μL of PdNPs solution (4 mg mL⁻¹) reaches a limiting current after addition of 1 μL of nanoparticle solution for the MWCNT electrodes prepared with a smaller CNT solution volume (10 μL) of CNTs (ie 10% v_PdNPs/v_CNT). However, for the MWCNT electrodes prepared with a larger volume (20 μL) of CNTs (ie 12.5% v_PdNPs/v_CNT), larger oxidation currents were obtained after addition of between 2.5 to 5 μL of nanoparticle solution (4 mg mL⁻¹), with a limiting current after addition of 2.5 μL of solution. These experiments clearly show that that the amount of hydrogen oxidation current and therefore PdNP loading can be controlled by finely adjusting the volumes of PdNP and CNT solutions used during electrode fabrication. Based on these results, nanocomposite PdNP electrodes with high PdNPs loadings were prepared for electrocatalytic oxidation of hydrazine. The electrodes used to study hydrazine oxidation were prepared using 10% v_PdNPs/v_CNT and 12.5% v_PdNPs/v_CNT ratios.

Zoom In Zoom Out Reset image size

Since we previously showed that oxygen can be effectively reduced by PdNPs (22.5 nm diameter) at potentials below 0.25 V vs. SCE in aqueous solution,²⁰ we investigated these PdNPs for the oxidation of hydrazine on MWNT-modified electrodes at neutral pH and in deaerated solution. Figure 5A shows typical CVs recorded at unmodified MWCNTs and PdNPs/MWCNT electrodes with cubic PdNPs in deaerated McIlvaine buffer (pH 7.0) solution in the absence and presence of 8 mmol L⁻¹ N₂H₄. At unmodified MWCNT electrodes, the onset potential for N₂H₄ oxidation was observed at 0 V. In contrast, the PdNP-modified MWCNT electrodes exhibited a drastic change in the electrocatalytic properties. The presence of the metallic nanoparticles caused a shift in the oxidation onset potential toward a more negative potential at around −0.4 V and with the hydrazine oxidation peak at E_p = −0.2 V. Comparison of the two different electrodes prepared using "small" (10% v_PdNPs/v_CNT) or "large" (12.5% v_PdNPs/v_CNT) amounts of PdNPs revealed, surprisingly, that the electrodes with the smaller amount of PdNPs gave the highest electrocatalytic current, which reached 2.80 ± 0.13 mA cm⁻² compared to 2.04 ± 0.34 mA cm⁻². During electrocatalytic oxidation, a larger amount of gas bubble formation was observed for the electrodes prepared with the larger amount of PdNPs. The lower catalytic current observed is attributed to impeded hydrazine diffusion and reduced electroactive surface area due to the gas bubbles at the surface. Next the catalytic performance of the PdNPs was assessed over a wide concentration range for PdNPs/MWCNT-modified electrodes prepared with the smaller amount of PdNPs. A linear calibration plot was obtained from CVs for the catalytic oxidation current at 0.1 V as a function of increasing hydrazine concentration. This experiment shows that in the μmol L⁻¹ to mmol L⁻¹ concentration range there are no significant limitations due to substrate diffusion, catalytic surface area or catalyst degradation with repeat electrochemical cycling over a wide concentration range and at very high mmol L⁻¹ concentrations of hydrazine. It is noted that for these experiments the catalytic current was measured from the reverse sweep at 0.1 V as this provides a more stable and reliable picture of the current achieved under continuous operation of the electrode e.g. repeat cycling. High electrooxidation currents were recorded which reached a current maximum of 4.59 ± 0.20 mA cm⁻² at 100 mmol L⁻¹ N₂H₄ solution.

Zoom In Zoom Out Reset image size

In order to develop the self-powered sensor, an enzymatic biocathode was employed as the cathode to pair with the optimized PdNPs/MWCNT-modified electrode as the anode. The standard biocathode used here for the enzymatic four-electron oxygen reduction reaction was described in our recent work.²³ The biocathode exploits the use of negatively charged (5,5^'-dithiobis(2-nitrobenzoic acid)) groups to provoke preferential orientation of the BOx enzyme toward the surface and thus facilitate efficient direct electron transfer for the oxygen reduction reaction. Figure 6 shows the CVs recorded at a GC-based BOx/(DTNB-Pyr)/MWCNT in quiescent solution and under oxygen purging. Firstly, for both oxygen conditions, the onset potential for the oxygen reduction reaction is E_onset = 0.52 ± 0.01 V. This is close to the redox potential of the laccase T1 copper center responsible for the intramolecular electron transfer to the T2/T3 site at which O₂ is reduced to H₂O. Comparison of the CVs recorded in quiescent solution and under oxygen purging reveal the drastic difference in catalytic current possible as a function of oxygen content. In buffer solution the dissolved oxygen is typically [O₂] ≈ 0.2 mmol L⁻¹ whereas under oxygen bubbling saturated oxygen levels of [O₂]_sat ≈ 1.2 mmol L⁻¹ are possible. Under continuous oxygen saturation conditions, the maximum achievable current densities were 1.34 ± 0.15 mA cm⁻² at 0.1 V, which compares to 0.06 ± 0.01 mA cm⁻² at 0.1 V with only dissolved oxygen. After the addition of 8 mmol L⁻¹ of hydrazine under oxygen saturation conditions with bubbling, a decrease of the observed resting potential from 0.52 V to 0.34 V was observed. Moreover, the voltammograms show that the O₂ reduction onset potential was lowered to more negative potentials (E_onset = 0.35 ± 0.01 V) and with a noticeable decrease in cathodic current. This observation reveals that there is a competition between oxygen reduction from the BOx enzyme immobilized on the electrode surface and the electrooxidation of N₂H₄ in solution at the electrode surface (see Figure 5). Nevertheless, at high overpotential with E_app = 0 V to 0.15 V, the maximum steady-state current densities obtained were identical to that recorded in the absence of hydrazine.

Zoom In Zoom Out Reset image size

The complete hybrid N₂H₄ self-powered sensor was built by combining a PdNPs/MWCNT modified electrode as the anode and a BOx/(DTNB-Pyr)/MWCNT biocathode supported either on GC or GDL. In initial experiments, the GC-based biocathode was employed and oxygen was continuously purged to keep the oxygen solution concentration constant and to maximize power output (see cell configuration in Figure S1). The bioanode was prepared using the optimized anode prepared using the 10% v_PdNPs/v_CNT ratio which gave the highest catalytic activity for hydrazine oxidation. In the absence of hydrazine, the open circuit voltage (OCV) of the fuel cell / sensor was 147 ± 7 mV. Upon polarization of the electrodes between OCV and 0 V a maximum power density of P_max = 0.2 μW cm⁻² was obtained (Figure S2). Next, polarization curves were recorded for 10 different concentrations in the range of 0.2 to 15 mmol L⁻¹ of hydrazine and the three parameters of OCV, P_max and current density (J_max) were monitored, and importantly, all three parameters increased with increasing hydrazine concentration. For hydrazine sensing, the maximum power density was measured from linear sweep voltammograms after each individual hydrazine injection into the solution, then a calibration curve was recorded presenting P_max = f ([N₂H₄]) (Figure 7A). It can be clearly observed that the P_max varies linearly with N₂H₄ concentration in the range of 0 to 4 mmol L⁻¹ with a slope of 53.4 ± 3.4 mW L mmol⁻¹ cm⁻² (R² = 0.994) before saturation is reached at a hydrazine concentration ≥ 6 mmol L⁻¹ where J_max = 937 ± 23 μA cm⁻² and P_max = 211 ± 30 μW cm⁻², respectively. Finally, with the aim of alleviating the need for an oxygen source to operate the sensor device, for a more realistic sensor which could be applied at contaminated hydrazine sites, a second configuration of biocathode was employed. The biocathode was prepared as before but on a GDL substrate rather than GC (see cell in Figure S1) and in complete quiescent solution rather than under continuous oxygen bubbling. The advantage of the GDL is that oxygen can be brought to the cell system naturally via diffusion from the air giving increased currents compared to simple quiescent solution. Furthermore, the GDL electrodes further benefit from being lighter and more portable than typical GCE working electrodes. The device with the GDL-based cathode was operated and the data analyzed as described for the first self-powered sensor, and the results obtained were in agreement with those obtained previously (see Figure 7A). The only significant difference was regarding the limiting current obtained for this type of device which was slightly higher for the cell setup using the GDL-based biocathode. Additionally, the shapes of the power curves clearly indicate a diffusion limitation of the system at low concentration of N₂H₄ due to the absence of mechanic convection from the O₂ purging. For GCE-based biocathodes this type of diffusion limitation was not observed due to very high rates of mass transport (see Figure S2).

Zoom In Zoom Out Reset image size