Electrocatalytic Glucose Oxidation at Coral-Like Pd/C₃N₄-C Nanocomposites in Alkaline Media

Abstract

Porous coral-like Pd/C₃N₄-C nanocomposites are fabricated by a simple one-pot chemical reduction method. Their electrocatalytic performance is ~50% higher than a carbon-loaded palladium electrocatalyst (Pd/C) in alkaline media. This confirms that the glucose electrooxidation and sensing performance of a Pd/C can be improved by the synergy of graphitic carbon nitride (C₃N₄), though C₃N₄ exhibits poor electrical conductivity. Compared to Pd/C, the size of Pd nanoparticles in Pd/C₃N₄-C decreases. As a result, the activity of Pd/C₃N₄-C is enhanced due to the higher dispersion and the synergistic effect. Pd/C₃N₄-C presents a rapid response and high sensitivity to glucose. The sensitivity for glucose sensing at Pd/C₃N₄-C is 3.3 times that of at Pd/C in the range of 0.001–10 mM. In the lower range of 0.001–1 mM, the sensitivity at Pd/C₃N₄-C is ~10 times greater than Pd/C.

1. Introduction

Glucose electrooxidation reaction (GOR), which has been studied for more than 100 years [1], is the key chemical reaction in nonenzymatic-glucose sensors and direct glucose fuel cells [2,3]. Theoretically, glucose can be completely oxidized to CO₂ and H₂O, releasing 24 electrons:

$${\mathrm{C}}_6{\mathrm{H}}_{12}{\mathrm{O}}_6+24{\mathrm{OH}}^{-}\to 6{\mathrm{CO}}_2+18{\mathrm{H}}_2\mathrm{O} +24{\mathrm{e}}^{-}$$

(1)

however, to date, the complete electrooxidation of glucose is rarely reported. The main product for GOR is gluconate, and only two electrons are released [4]:

$${\mathrm{C}}_6{\mathrm{H}}_{12}{\mathrm{O}}_6+2{\mathrm{OH}}^{-}\to {\mathrm{C}}_6{\mathrm{H}}_{12}{\mathrm{O}}_7+{ \mathrm{H}}_2\mathrm{O} +2{\mathrm{e}}^{-}$$

(2)

In recent years, Pd and Pd compounds have been regarded as state of the art electrocatalytic active components for GOR [3,5,6,7,8]. Many efforts have been made to improve the activity of palladium-based catalysts. The activity of a catalyst depends not only on its active component but also on its support. It is an effective way to develop novel support to improve the activity of catalysts. A graphitic carbon nitride (C₃N₄) has a planar two-dimensional (2D) structure, and its surface is rich in nitrogen atoms with a strong coordination ability. We consider that C₃N₄ may be used as excellent electrocatalytic support in Pd-based catalysts for GOR. The Schottky effect [9] of C₃N₄-Pd facilitates electron migration from metal to the N-type semiconductor C₃N₄ and benefits the electrooxidation reaction at Pd. In this work, porous coral-like Pd/C₃N₄-C nanocomposites were fabricated with the hydrogen bubbles as dynamic templates. GOR on Pd/C₃N₄-C was evaluated by electrochemical measurements in alkaline media. The electrocatalytic performance of Pd/C₃N₄-C is ~50% higher than Pd/C. Besides the activity for GOR, the Pd/C₃N₄-C shows competitive glucose-sensing performance compared to some other electrocatalysts reported recently [10,11,12,13,14,15,16,17,18] (

Table 1. Comparison of glucose oxidation on the Pd/C₃N₄-C electrode and other Pd electrodes.

Electrode	Applied Potential (vs. Hg/HgO)	Linear Range (mM)	LOD (μM)	Reference
Pd@Pt CINP/C/GCE 1	+0.01 V	1–8.5	0.82	[10]
Pd-Ni-P	+0.65 V	0.005–10.24	0.15	[11]
Pd nanocubes	+0.50 V	1–10	N/A	[12]
WO3-Pd AA 2	+0.45 V	5−55, 65−375	N/A N/A	[13]
Pd-Fe/Ti	+0.13 V	0.002–3.0	1.0	[14]
Pd-CuO/rGO/SPE 3	+0.70 V	0.006–22	0.03	[15]
Pt-Pd nanowire arrays	+0.30 V	Up to 10	N/A	[16]
Pd-Au cluster	+0.04 V	0.1–30	50	[17]
RGO-Pd NCs@CuO 4	+0.50 V	0.001–3.1	0.01	[18]
Pd/C3N4-C	+0.50 V	0.1–10	0.55	This work

¹ CINP: core/island nanoparticles; GCE: glassy carbon electrode. ² AA: after annealing. ³ SPE: screen printed electrodes. ⁴ NCs: nanocrystals.

). This confirms that Pd/C₃N₄-C is indeed a good electrocatalyst for GOR.

2. Results

2.1. Physical Characterization

As can be seen from the images (Figure 1a–c) of the scanning electron microscope (SEM), Pd/C₃N₄-C possesses a porous coral-like structure. The mapping analysis in Figure 1d confirms the coexistence of elements Pd, C, and N in the catalyst. The spherical Vulcan carbon and the planar graphic nitrogen carbide are easy to assemble into three-dimensional (3D) structures. The hydrogen bubbles produced in the hydrolysis of NaBH₄ act as dynamic templates of the coral-like Pd/C₃N₄-C.

Pd particle sizes are estimated from images (Figure 2a–c) by transmission electron microscopy (TEM). All nanoparticles (NPs) in the images are considered. The average Pd particle sizes (Figure 2d–g) of Pd/C₃N₄, Pd/C, and Pd/C₃N₄-C are 3.6, 4.6, and 3.3 nm, respectively. The introduction of C₃N₄ dramatically decreases the diameter of palladium nanoparticles; thus, the total surface area is increased. More active sites can be provided with an increased surface. The N atoms in C₃N₄ exhibit a strong complexing ability for Pd atoms/ions [19], which is beneficial for strong metal support interactions. There are typical lattice fringes with 0.14, 0.19, and 0.23 nm, which are according to the (220), (200), and (111) planes (Figure 2g) of Pd nanocrystals, respectively. Some Pd NPs are bridging the C₃N₄ and carbon (Figure 2h), then C₃N₄-Pd-C heterostructures appear in Pd/C₃N₄-C composites.

The X-ray diffraction(XRD) patterns of Pd/C₃N₄, Pd/C, and Pd/C₃N₄-C are shown in Figure 3a. The peak values at about 27° can be ascribed to the (002) interlayer stacking of C₃N₄. The peak values at 39.7°, 45.6°, 67.7°, and 80.5° correspond to the Pd (111), (200), (220), and Pd (311) crystal faces, respectively. The peaks corresponding to the C (002) and C₃N₄ (002) crystal planes are significantly weakened in Pd/C₃N₄-C, compared with the C (002) peak of Pd/C and the C₃N₄ (002) peak of Pd/C₃N₄. It suggests that the assemble of C₃N₄ and carbon may mainly occur at the (002) crystal plane of C₃N₄ and the (002) crystal plane of Vulcan carbon. Similar changes of C₃N₄ (002) peak caused by facet coupling can also be found in other reports [20]. The lattice constants (α) and the size of Pd NPs (d) are calculated using Equation (3) [21,22] and the Scherrer’s equation (Equation (4)) [23,24,25,26], respectively,

$$\mathsf{\alpha } =\frac{\sqrt{2}\mathsf{\lambda }}{{\sin \mathsf{\theta }}_{\max }}$$

(3)

$$\mathrm{d} =\frac{\mathsf{\kappa }\mathsf{\lambda }}{\mathsf{\beta }\cos \mathsf{\theta }}$$

(4)

where λ is the wavelength of X-ray, θmax is the peak position of Pd (220), k is a numerical factor, and β is the width in radians (full-width at half-maximum) of the peak. θ is the Bragg angle. The lattice constants of Pd in Pd/C₃N₄-C and Pd/C are 0.3991 nm and 0.3977 nm, respectively. The average Pd NPs sizes in Pd/C₃N₄-C and Pd/C are 3.1 nm and 4.7 nm, respectively. The sizes estimated with XRD patterns are similar to the diameters of Pd/C₃N₄-C (3.3 nm) and Pd/C (4.6 nm) obtained from the TEM images.

The vibration bands [27,28,29,30] in the region of 2900–3500 cm⁻¹, 1200–1700 cm⁻¹, and ~ 800 cm⁻¹ are shown in the infrared spectra of C₃N₄ (Figure 3b). The peak at 2900–3500 cm⁻¹ corresponds to the stretching vibration of the uncondensed residual of the –NH₂ and the –OH bond from the adsorbed water molecules. The peak at 1200–1700 cm⁻¹ is attributed to the stretching vibration of the carbon–nitrogen heterocycle [C–NH–C or C–N(–C)–C]. The peak at 800 cm⁻¹ is a typical triazine-absorption vibration. The three vibration bands in C₃N₄ are also preserved in Pd/C₃N₄-C.

The XPS spectra are shown in Figure 3c–f. Figure 3c reveals the coexistence of Pd, C, and N (or Pd, C) elements in the catalysts. The Pd 3D peak (Figure 3d) of Pd/C₃N₄-C positively shifted to 0.71 eV compared with Pd/C. It indicates that the synergy between C₃N₄ and carbon results in an electron-deficient state of Pd [31]. As mentioned above, the Schottky effect of C₃N₄-Pd facilitates electron migration from metal to the N-type semiconductor C₃N₄. The C 1s region (Figure 3e) can be fitted into three peaks [32,33,34,35]. The peak at 284.6 eV corresponds to the special sp² bond or carbon pollution in C₃N₄. The peak at 286.4 eV is attributed to the hybrid C of C–NH₂ in C₃N₄, and the C peak at 288.2 eV (Pd/C₃N₄) or 288.8 eV (Pd/C₃N₄-C) is attributed to N–C=N cyclic sp² hybrid C in C₃N₄. The relative proportion of different forms of carbon species can be obtained from C 1s spectra (

Table 2. Percentages of carbon species from spectra of C 1s.

Electrocatalyst	C Specie	eV	Atom %
Pd/C3N4	C–NH2	286.4	8.2
-	N–C=N	288.2	58.5
-	C–C	284.6	33.3
Pd/C3N4-C	C-NH2	285.7	15.5
-	N–C=N	288.8	9.1
-	C–C	284.6	75.4

). The N 1s region can be fitted into three peaks (Figure 3f). The peak at 398.6 eV (Pd/C₃N₄) or 398.8 eV (Pd/C₃N₄-C) corresponds to the C–N=C bond formed by the sp² hybrid aromatic ring N and C atom. The N peak at 400.0 eV (Pd/C₃N₄) or 399.5 eV (Pd/C₃N₄-C) is attributed to the N–C₃ bond. The N peaks at 401.1 eV (Pd/C₃N₄ or Pd/C₃N₄-C) are attributed to the residual H–N functional group derived from melamine. As shown in the N 1s spectra, the proportion of N–C₃ in Pd/C₃N₄-C (50.8 atom%) is nearly three times more than the 17.1 atom% of Pd/C₃N₄ (

Table 3. Percentages of nitrogen species from spectra of N 1s.

Electrocatalyst	N Specie	eV	Atom %
Pd/C3N4	N–C3	400.0	17.1
-	C–N=C	398.6	76.3
-	C–N–H	401.1	6.7
Pd/C3N4-C	N–C3	399.5	50.8
-	C–N=C	398.8	31.7
-	C–N–H	401.1	17.5

). As mentioned above, the XRD analysis suggests that the assembly of C₃N₄ and Vulcan carbon mainly occurs at their (002) crystal planes. The extra N–C₃ bonds mainly form between the (002) crystal plane of Vulcan carbon and the (002) plane of C₃N₄. Then the electron’s transportability of C₃N₄ can be improved in the composites by the new electronic transmission path, such as some N–C₃ bonds between C₃N₄ and carbon black, and metal nanoparticles, which bridge the C₃N₄ and carbon black mentioned above.

2.2. Electrochemical Characterization

Cyclic voltammetry (CV) curves (Figure 4) of Pd/C, Pd/C₃N₄-C, and Pd/C₃N₄ were operated in 0.1 M NaOH solutions with and without the presence of 10 mM glucose, respectively. After the addition of glucose, an obvious oxidation current appeared at ~ 0.1 V. The glucose oxidation current at Pd/C₃N₄-C is significantly higher than that at Pd/C and Pd/C₃N₄. The current density of the glucose electrooxidation peak is used to evaluate the catalytic activity [3]. The current density of the anode peak at Pd/C₃N₄-C (31.10 mA mg⁻¹_Pd) is ~ 50% higher than that at Pd/C (21.15 mA mg⁻¹_Pd). As shown in the analysis of TEM images and XRD data, the particle size of Pd NPs in Pd/C₃N₄-C is smaller than in Pd/C. Accordingly, the electrocatalytic performance of the Pd/C₃N₄-C is better. Besides, the Schottky effect of C₃N₄-Pd [9] facilitates electron migration from metal to the N-type semiconductor C₃N₄, which benefits the electrooxidation reaction at Pd. The peak in (−0.1 ~ −0.7) V is the reduction peak of Pd(II) oxide. The glucose oxidation current at Pd/C₃N₄ is much lower than those at both Pd/C₃N₄-C and Pd/C. The lower current is due to the higher resistance of the semiconductor C₃N₄. Compared to the conductivity of the semiconductor C₃N₄ in Pd/C₃N₄, as mentioned above, the electron transportability of the C₃N₄ in the Pd/C₃N₄-C composites has been improved by adding more electronic transmission paths such as N–C₃ bonds bridging carbon black and C₃N₄, and metal nanoparticles bridging the C₃N₄ and carbon black.

2.3. Glucose Sensing

Figure 5a is the CV curves of Pd/C₃N₄-C at different scan rates (50 ~ 500 mV s⁻¹). Ten mM glucose in a 0.1 M NaOH solution was used as the sample. 0.1 M NaOH has been widely used as the media for the evaluation of the electrocatalytic activity of Pd NPs for GOR [12,36,37,38]. It can be seen that as the scan rate increases, the current also increases (Figure 5a). However, when the scan rate is higher than 50 mV s⁻¹, the peak is not obvious. So 50 mV s⁻¹ was chosen as the scan rate for glucose sensing. The fitting curve (Figure 5b) indicates the relationship between the scanning speed and the peak current. The anode peak current has a linear relationship with the square root of the scanning rate (y = 0.138 x − 0.295, R² = 0.996), which reveals that the glucose oxidation is controlled by the diffusion during the voltammetric measurements.

The linear scanning voltammetry (LSV) curves (Figure 6a) of Pd/C₃N₄-C under different glucose concentrations (1–10 mM) and the corresponding calibration curve (Figure 6b) were obtained in alkaline solutions, and the regression equation is I (mA) = 0.0344·c (mM) + 0.0216 (R² = 0.992). Figure 6c displays the amperometric response to glucose at Pd/C₃N₄-C, and Pd/C at 0.5 V. Pd/C₃N₄-C has a rapid response (less than 7 s) in glucose sensing, implying sensitive and fast responses to GOR. Amperometric responses of Pd/C₃N₄-C are more sensitive than Pd/C. The calibration curve (Figure 6d) shows that the glucose oxidation current increased linearly in the concentration range of 0.001 to 10 mM and 11 to 20 mM, respectively. At glucose concentration of 0.001 to 10 mM, the regression equation of Pd/C₃N₄-C is I (mA) = 0.00877·C (mM) + 0.00763 (R² = 0.993) and the regression equation of Pd/C is I (mA) = 0.00263·C (mM) + 0.00744 (R² = 0.996). At higher concentrations with a range such as 11–20 mM, the regression equation is I (mA) = 0.00498·C (mM) + 0.0389 (R² = 0.997) and the regression equation of Pd/C is I (mA) = 0.00179·C (mM) + 0.0156 (R² = 0.998). The sensitivity for Pd/C₃N₄-C and Pd/C in the concentration range of 0.001 to 10 mM is calculated to be 219 and 65.7 μA mM⁻¹ mg⁻¹_Pd, respectively. The sensitivity of Pd/C₃N₄-C is ~ 3.3 times of Pd/C. The limits of detection (LOD) for Pd/C₃N₄-C and Pd/C (S/N = 3) were calculated to be 0.55 μM and 3.1 μM, respectively. (LOD = 3σ/S [39], where σ is the standard deviation of 5 blanks, and S is the slope of the linear regression equation. S/N = 3). The relative standard deviation (RSD) is 4.5% (five repeated measurements). Figure 6e is an enlarged view of the current response calibration curve at 0.1–1 mM. The sensitivity of Pd/C₃N₄-C in the range of 0.001–1 mM is ~ 10 times greater than Pd/C. The calibration curves of both the voltammetric and the amperometric measurements confirm the sensitive and linear response of Pd/C₃N₄-C to the glucose in alkaline media. The responses of KCl (0.1 mM), NaCl(0.1 mM), lactose(0.1 mM), ascorbic acid (AA, 20 μM), and uric acid (UA,120 μM) were also recorded at 0.5 V (Figure 6f). The responses of 1 mM glucose were used as contrasts. Pd/C₃N₄-C exhibits conspicuous resistance to these interfering substances during the GOR.

The glucose concentration after the electrocatalytic oxidation was determined in order to obtain the percentage of degradation. Figure 7 shows the LSV curves at Pd/C₃N₄-C before and after 200 CV cycles. According to the glucose concentrations obtained from the linear equation in Figure 6b, the percentage of degradation is 32% after 200 CV cycles.

3. Materials and Methods

3.1. Materials and Instruments

Nafion (5 wt%, Dupont, Delaware, DE, USA) and carbon black (XC-72R, Cabot Corp., Waltham, MA, USA) were used as obtained. All other chemicals were of analytical grade and used as received. Trip-distilled water was used during the whole experimental process.

FTIR (Nicolet iS10 spectrometer, Thermo Fisher Scientific, Waltham, MA, USA), XPS (XSAM-800 spectrometer, Kratos Analytical Ltd., Manchester, UK), XRD (D8 Advance instrument, Bruker, Karlsruhe, Germany, Cu Kα radiation at λ = 0.15406 nm), SEM (SU-70 microscope, Hitachi, Tokyo, Japan), and TEM (JEM-1011, JEOL, Beijing, China, acceleration voltage 200 kV) were used for the physical characterization of the electrocatalysts.

3.2. Preparation of C₃N₄ and Pd/C₃N₄-C

The preparation of C₃N₄ and Pd/C₃N₄-C is schematically illustrated in Figure 8. The C₃N₄ was prepared by the pyrolysis of melamine at 550 °C in a tube furnace for 4 h. The obtained C₃N₄ was ground into powders after natural cooling. The powders were filtered through a 500-mesh sieve. Sixteen mg C₃N₄ and 24 mg carbon black were added to 25 mL of triple-distilled water and dispersed ultrasonically. Then, a certain amount of PdCl₂ solution (10 mg Pd) was added to the suspension and stirred. After the pH of the mixture was adjusted to 10, a 10 mL 5 wt % NaBH₄ solution was added. The reaction mixture was stirred for 6 h and left to stand overnight. The obtained Pd/C₃N₄-C products were filtered and washed. Then, the catalysts were dried in a vacuum oven (60 °C, 10 h). Pd/C (or Pd/C₃N₄) was synthesized under identical conditions except for the replacement of C₃N₄-C with 40 mg of C (or C₃N₄). The metal loading was 20 wt % in the as-prepared catalyst.

3.3. Electrochemical Experiments of Glucose Sensors and GOR Analysis

Electrochemical tests were conducted in a standard three-electrode system using the CHI660E electrochemical workstation, including a carbon rod as the counter electrode, a Hg/HgO (saturated 1 M NaOH) electrode as the reference electrode, and a modified 3-mm-diameter GCE as the working electrode. All the potentials are quoted with respect to Hg/HgO reference in this paper.

To fabricate the modified GCE working electrode, the GCE should be polished by 0.05 μm Al₂O₃ nanoparticles before use. The ink of catalyst should be prepared by dispersing 10 mg of the catalyst in a mixture of 0.5 mL ethanol and 0.5 mL triple-distilled water ultrasonically. Twenty μL (4 μL × 5 aliquots) of the ink should be dropped onto the GCE surface. This drop of ink should be dried, then another drop should be added. The deposited ink should be finally covered with 4 μL of a Nafion solution. The Nafion solution should also be dried at room temperature. The as-prepared modified GCE was used as the working electrode. The catalysts in 10 μL of ink were enough to cover the whole surface of the 3-mm diameter GCE after the ink dried [40]. A higher amount of catalysts is beneficial to improve the durability or long-term stability of the modified GCE. For example, in our previous work, we dropped 34 μL of ink onto a GCE [41] for the oxygen-reduction reaction. The as-prepared electrode kept a stable electrocatalytic current density during a long-term (84 h) measurement. In this work, a compromised amount (20 μL of ink) of catalysts was used to prepare the working electrodes.

Voltammetric measurements were obtained in the mixture solution of NaOH and glucose. The electrolyte solutions were bubbled with nitrogen for more than 20 min before use. The bubbled solutions were used under a nitrogen atmosphere during the electrochemical measurements. The sensitivity and interference of Pd/C₃N₄-C were recorded in a stirred glucose/NaOH solution at a fixed potential of 0.5 V.

4. Conclusions

Coral-like porous Pd/C₃N₄-C electrocatalysts were synthesized by a simple room-temperature reduction method. The introduction of C₃N₄ significantly decreased the Pd nanoparticle’s size and increased its dispersion, which is beneficial for the improvement of the electrocatalytic activity of Pd for GOR. Pd/C₃N₄-C also exhibits a highly sensitive, fast, selective, and linear amperometric response to glucose, which makes this electrocatalyst a good candidate for nonenzymatic-glucose sensors. The N–C₃ bonds between C₃N₄ and carbon improved the conductivity of the semiconductor C₃N₄ in the composites, which facilitates the application of C₃N₄ in the field of electrocatalysts.