Effect of Defective Graphene Flake for Catalysts of Supported Pd Nanocubes toward Glucose Oxidation Reaction in Alkaline Medium

The effect of defective graphene flake (DGF) as supports for Pd nanocubes (Pd NCs) in the catalysis of D-glucose oxidation reaction (GOR) was studied. The DGF used to support the Pd NCs were prepared via both a sonoelectrochemical method (DGFSECM) and a chemical reduction method (DGFCM). The comparable data supported by the diffraction electron pattern of transmission electron microscope (TEM) and X-ray diffraction (XRD) spectra evidence that the corners of the Pd NCs faced up toward the both DGF supports, leading to an increase in the XRD intensity of the Pd (111) peak. Additionally, a fair electrochemical comparison of the DGF support effect on GOR catalysis via Pd NCs in a NaOH electrolyte indicated that the DGFSECM-supported Pd NCs (Pd NC/DGFSECM) and DGFCM-supported Pd NCs (Pd NC/DGFCM) demonstrated earlier onset potentials and an overall order of Pd NC/DGFSECM > Pd NC/DGFCM > commercial Pd/C catalyst > Pd NCs for the peak charge accompanying the formation of glucolactone was obtained. Through the use of a DGFSECM substrate, the electrochemical surface area and charge transfer resistance toward GOR were significantly improved. In addition, as supported by the catalytic activity for GOR, Pd NC/DGFSECM showed remarkable sensitivity and tolerance to foreign substances in the application as a non-enzymatic glucose sensor, where Pd NC/DGFSECM showed higher two-period sensitivities of 74 and 45.86 μA · mM−1 · cm−2 for 0.25–5 and 5–24 mM D-glucose, respectively. The high recovery in serum sample analyses further confirmed the potential of Pd NC/DGFSECM as glucose sensors. © The Author(s) 2016. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0141609jes] All rights reserved.

Electrocatalytic glucose oxidation reaction (GOR) is of particular interest, and can be applied to a direct glucose fuel cell (DGFC), [1][2][3] implantable fuel cell (IFC), 4,5 and glucose sensor. [6][7][8] Various noble metal materials such as platinum, 9,10 palladium, 11,12 silver, 13,14 and gold 13,15 have shown activity and been used as electrodes for electrocatalyzing GOR in an alkaline medium. Among them, Pt is frequently employed as a GOR catalyst; [16][17][18] however, there are significant drawbacks for the GOR on its bare surface: 19 the poisonous influence of chemisorbed intermediates and the interference of ascorbic acid (AA) and uric acid (UA) in the human body. Therefore, novel catalysts have been investigated to replace Pt catalysts. Wang and coworkers 20 conducted exquisite work preparing Au rhombic dodecahedra, octahedra, and nanocubes (NCs) to catalyze alkaline GOR and systematically compared their activities. The study showed that the shape-dependent properties of these Au catalysts mainly originated from the different crystal planes exposed on their nanocrystal surfaces and that (100)bounded Au NCs were more active than the Au rhombic dodecahedral and octahedral catalysts. Some original studies 21,22 showed that bulk Pd materials reduced to nanoparticles showed good activity and stability toward alkaline GOR. Recently, based on one growth method using hexadecyltrimethylammonium bromide (CTAB) as a protecting agent, we prepared 27-nm Pd NCs enclosed with (100) facets and observed that these cubic catalysts were more active than Pd polyhedra in D-glucose oxidation. 23 Additionally, utilizing expensive Pd nanocatalysts is an important issue for electrocatalytic devices. Highly dispersed Pd catalysts on a conductive support, commonly carbon black, are used as electrodes for oxidation reactions in fuel cells fuelled with formic acid, [24][25][26] glycerol, 27 and glucose. 21,28 With the helpful carbon powders, the high surface-to-volume ratios of Pd nanoparticles maximize the available surface areas for these electrocatalytic reactions. Graphene flake with a high surface area (ca. 2,600 m 2 · g −129 ) is currently the most investigated carbon support as a substitute for carbon black catalyst supports for fuel cell Pd catalysts. [30][31][32] The high degree of sp 2 -bounded carbon atoms on the graphene flake can lead to effective conductivity and possibly decrease overpotential. In addition, the defect effect of graphene flake as catalyst support is of interesting in the electrocatalytic reaction. [33][34][35] Theoretically, the formation of Pt−carbon bonds * Electrochemical Society Member. z E-mail: cl_lee@kuas.edu.tw; cl_lee@url.com.tw at graphene support defects can influence the average bond length and thus the strain in the metal catalysts to enhance the stability of Pt catalyst. 36 The stronger binding of the metal catalyst to the graphene support leads to increase charge transfer from the cluster to the substrate accompanied by a substantial downshift of the catalyst d-band center, leading to improve CO tolerance of platinum nanoparticles on defective graphene flake (DGF). 36 Theoretical calculations further demonstrated that the binding energies, d-band centers, and adsorption energies showed a linear change with the pore size of porous defective graphene support for Pd catalysts. By choosing a graphene support with suitable pore size, Pd catalysts on defective graphene will have similar CO and O 2 adsorption abilities, thus leading to superior CO tolerance. 37 Recently, for oxygen reduction reactions (ORRs), Liu and coworkers 38 carried out first-principles-based calculations and successfully determined that the defects on a DGF can provide anchoring sites for Pd-based catalysts, increasing their stabilities. This interfacial interaction could further tune the d-band center of these ORR catalysts to weaken O adsorption and promote ORR kinetics. The enhanced activities of defective DGF-supported catalysts can be partly ascribed to the isolated sp 2 hybridized bonds (π electrons) and the strong interaction (charge transfer) between the catalysts and the DGF. 39 Therefore, the use of defective DGFs for ORR catalysts could be an efficient method for improving reaction kinetics. Previously, we reported a sonoelectrochemical method to rapidly prepared defective graphene flakes (DGF SECM ), of which surface was dominated by physical defects while they were compared to defective graphene flakes prepared via a chemical reduction method (DGF CM ). 40 The primary defects on the DGF CM were C-OH groups. 40 In this study, given the goal of developing a highly active catalyst for GOR, we study the support effect of DGF SECM and DGF CM for the catalysis of Pd NCs in GORs. We successfully use electrochemical analyses to perform a comparison between DGF SECM -supported Pd NCs (Pd NC/ DGF SECM ), DGF CM -supported Pd NCs (Pd NC/ DGF CM ), and DGF-free Pd NCs when these catalysts were applied to an enzyme-less glucose sensor.

Experimental
Synthesis and materials analyses of Pd NC/ DGF SECM and Pd NC/ DGF CM .-Before the synthesis of Pd NCs supported on DGFs, the solution containing Pd NCs was prepared. The preparation of the solution containing Pd NCs was based on our previous report 41 and is described in the supplemental material. The obtained Pd NC solution was centrifuged at 12,000 rpm for 10 min. Then, the Pd precipitate was redispersed in a 1 mL H 2 O solution, and the solution was centrifuged again at 10,000 rpm. The precipitate was redispersed in a 1 mL H 2 O solution. The Pd NC solution thus prepared was used for the mixture with DGF CM or DGF SECM , whose preparation methods are described below.
The DGF CM was obtained from graphene oxide (GO) using N 2 H 4 as a reducing agent. Initially, GOs were first prepared using a modified Hummers' method. 42,43 Then, 20 mg of the prepared GO powder was dispersed in 100 mL of an aqueous 0.1 M sodium n-dodecyl sulfate (SDS) solution by agitation using a sonicator for a period of 1 h. To remove unreacted graphite powders, the GO solution was centrifuged at 4,000 rpm, and the precipitate was redispersed using deionized water. A 40 mL solution with 4 mg of dispersed GO was prepared. Next, 200 μL of pure NH 4 OH solution was added to the solution via agitation and sonicated for 30 min. Then, 20 μL of the original N 2 H 4 solution was added to the above solution and stirred for 1 h at a fixed temperature of 95 • C for GO reduction. A solution containing DGF CM was thus prepared. Additionally, to synthesize DGF SECM , a sonoelectrochemical method where SDS was employed as intercalating agent was performed and reported elsewhere. 40 The brief description can be referred to the supplemental material.
To purify this mixture and remove some organic compounds (such as SDS and N 2 H 4 ), the obtained DGF solution was centrifuged at 1,000 rpm and then again at 13,000 rpm. The DGF precipitate was redispersed in a certain H 2 O solution under ultrasonic vibration for 10 min to generate a liquid with a concentration of ∼6 μg/μL. The weight of the DGF powder in the solution was confirmed with a quartz crystalline microbalance (QCM; Seiko QCA927). Then, the solution containing DGFs was mixed with the Pd NC solution as described above under ultrasonic vibration for 10 min. Pd NC/DGF SECM and Pd NC/DGF CM were thus prepared.
The morphologies and structures of the prepared Pd NC/DGF SECM , Pd NC/DGF CM and unsupported Pd NCs individually were examined using transmission electron microscopy (TEM; JEM-2100 and 2100F CS STEM, JEOL, Tokyo, Japan) and X-ray diffraction spectroscopy (XRD; Bruker D8; Cu anode; 1.54184 Å). The zeta potentials of the prepared Pd NC solution, DGF SECM solution, and DGF CM solution for preparing the Pd NC/DGF SECM and the Pd NC/DGF CM were determined by dynamic light scattering (Brookhaven Instruments 90Plus PALS Zeta Potential Analyzer). To estimate precisely the weight percentages of Pd catalysts on the DGF, thermal decomposition analyses of Pd NC-supported DGF (Pd NC/DGF) were carried out using a thermogravimetric analyzer (TGA; TA Instruments; SDT-Q600).

Electrochemical characterization of catalyzed D-glucose oxidation reactions and Cu UPD and stripping.-
The support effect of DGF SECM and DGF CM on the electrochemical performances of the Pd NCs in the catalysis of GOR was studied by performing a computercontrolled potentiostat (CH Instruments; CHI 627C) with a catalystcoated glassy carbon electrode (GCE; 0.07 cm 2 ), which was prepared by a drop-casting method using the prepared Pd NC/DGF solution and then heated to 60 • C to evaporate the H 2 O. The weight density of the Pd NCs, acting as active catalysts on the electrode, was 28 μg · cm −2 . All electrochemical measurements were conducted in a N 2 -saturated 0.1 M NaOH (aq) solution, without or with 5 mM D-glucose, using a three-electrode cell with a catalyst-covered electrode serving as the working electrode, a Pt foil as the counter electrode, and a Ag/AgCl (3 M KCl) reference electrode. To remove any potential interference from any impurities on the Pd NC/DGF SECM , Pd NC/DGF CM and unsupported Pd NCs during the D-glucose oxidation, the catalysts were scanned from −0.9 to 0.25 V using cyclic voltammetry (CV) at a rate of 150 mV · s −1 . Then, the redox properties of the catalysts were measured at a scanning rate of 50 mV · s −1 . The sensitivities and interference studies of the two Pd NC/DGFs and Pd NCs to glucose were tested in a stirred solution using amperometric measurements at a fixed potential of −0.05 V (vs. Ag/AgCl). In the interference    Figure 1E reveals point pattern and (200) diffraction facets for a single crystalline Pd NC as shown in the inset. This result indicates that the main facet on the prepared NCs exposed for catalyzing GOR is (100). Figure 2 shows a comparison of XRD spectra for the Pd NCs before and after being supported with DGF SECM and DGF CM . In the DGF-free Pd NC spectrum, the pattern shows a strong peak at 46.95 • assignable to the (200) diffraction plane of a typical Pd spectrum (JCPDS 65-2867), presenting (100) planes on six facets as confirmed by the electron diffraction result (Fig. 1E). One weaker peak at 40.29 • can be assigned to the (111) Pd plane at the truncated corners of some Pd NCs, as evidenced at the arrow in the HR-TEM image (the inset of Fig. 1E). When the Pd NCs were deposited onto the DGF SECM and DGF CM , the intensities of the both (111) peak were significantly enhanced and much higher than that of the (200) peak. In particular, the morphology of the Pd NCs (Figs. 1A and 1B) persisted after experiencing the deposition onto the DGFs. Therefore, the reason for the enhancement of the (111) peak resulting from the change of crystalline structures for the Pd NCs can be neglected. As shown in Fig. 1A, the corners and edges of some Pd NCs on the both DGFs faced upward so that the many (111) facets at the corners were detected by XRD spectroscopy. The XRD pattern of the Pd/C was also detected and shown in Fig. 2. The (111) facet predominates the structure of Pd catalysts in the Pd/C. The DGF effect on catalytic properties of cubic Pd catalysts toward GOR.-Furthermore, in order to precisely study the support effect of DGF on the GOR-catalysis activities and properties for equivalent weights of solid catalysts, the weight percentages of the Pd NCs on the DGF SECM and DGF CM compounds were estimated to be 61% and 52.7% using a TGA under an air atmosphere, respectively. The DGF effect on the catalytic activity of Pd NCs toward GOR was initially measured by CV. Figure 3A shows a comparison of the CV curves of the DGF SECM electrode and the DGF CM electrode working in the 0.1 M NaOH electrolytes without and with 5 mM D-glucose under a N 2 atmosphere. The curve without D-glucose for the DGF CM shows one reducing peak at −0.37 V and one oxidation shoulder at −0.23 V, whereas the cathodic peak for the DGF SECM are observed at −0.3 V, which can be attributed to the redox of oxygen-containing groups on the DGF. 44 The faster reduction for chemical defects on the DGF SECM is obtained. Apparently, the features of the curves for the both DGFs measured in the glucose electrolyte are similar to those in the glucose-free electrolyte, showing that the DGFs are inactive in the catalysis of GOR. The inactivity of DGF in D-glucose oxidation suggests the elimination of any oxidation signal provided by the DGF and that they only have a supporting effect in the oxidation of D-glucose. Figure 3B shows a CV curve corresponding to Pd NC/DGF SECM and Pd NC/DGF CM as compared to DGF-free Pd NCs and Pd/C in a D-glucose-free NaOH solution in a N 2 atmosphere. The currents by Pd NC/DGF CM between −0.55 and −0.9 V are associated with hydrogen adsorption and desorption on the Pd surface and the peak at −0.27 V can be assigned to the reduction of PdO. In the anodic scan, the adsorption of OH − was initially observed at −0.45 V 45 and the significant oxidation current for Pd started from −0.284 V. Compared to Pd NC/DGF CM , the current by Pd NC/DGF SECM is greater but the currents by Pd NCs and Pd/C are smaller, indicating that Pd NCs can spread out over the DGFs and DGF SECM which can effectively promote electron transfer in an electrochemical reaction. Additionally, the CV curve for Pd NC/DGF CM when working in the solution with 5 mM glucose is shown in Fig. 3C; the occurrence of the A1 peak at −0.61 V can be ascribed to chemisorption and dehydrogenation of D-glucose, 46 which removed the first hydrogen at hemiacetalic carbon 1, similar to the GOR on a Pt catalyst. 47 Note that the suppressed A1 peak for Pd NC/DGF SECM (Fig. 3C) was located at −0.67 V, early than that for Pd NC/DGF CM , −0.573 V for Pd NCs, and −0.53 V for Pd/C (Fig. 3C). Following the A1 peak, there was one strong peak (A2) initially at −0.26 V after the adsorption of OH − (−0.45 V) in the curve of Pd NC/DGF CM , as observed in Fig. 3B. At this state, the accumulated hydroxyls adsorbed on the Pd NCs were catalytic to the D-glucose on the catalytic surface. D-Glucose was oxidized to glucolactone 1 as follows:

Characterizations of Pd NC/DGF SECM and Pd NC/ DGF CM compared to individual Pd
Subsequently, in the cathodic scan, an A3 peak appeared at −0.31 V after the reduction of PdO at −0.21 V, which was earlier than −0.27 V without the occurrence of GOR in the Fig. 3B. 41 This relationship can be suggesting that the active sites suppressed by oxides were available. Previously, the GOR activity of Pd catalysts can be influenced by their preparation process using NaBH 4 , which caused the insertion of the produced hydrogen into Pd lattice. 48 The CV curves (Fig. 1S in the supporting information) for Pd NCs show that the A2 current of the scanning between −0.76 V and 0.25 V was almost equal to that of the scanning between −0.9 V and 0.25 V. These Pd NCs were initially experienced by fast CV scan of 150 mV s −1 from −0.76 V or −0.9 V, as show in Fig. 2S. The unchanged A2 currents indicate that the effect of hydrogen for GOR curves measured from −0.9 V can be neglected. Therefore, the A2 peak can be used as an activity indicator in GOR. The charges for the A2 peaks of Pd NC/DGF SECM , Pd NC/DGF CM , Pd NCs, and Pd/C are summarized in Table I. The A2 charges (Q A2 s) for Pd NC/DGF SECM and Pd NC/DGF CM are 939.1 and 685 μC, respectively, significantly greater than 356.4 μC by Pd NCs and 478 μC by Pd/C. The Q A2 of Pd NC/DGF SECM is 1.37 and 2.64 times greater than those of Pd NC/DGF CM and Pd NCs, respectively. Simultaneously, the Q A2 by Pd NC/DGF SECM is 1.96 times greater than that by commercial Pd/C catalysts. These results suggest that the electrocatalytic reaction on the Pd NCs can be enhanced using the DGF SECM as a catalyst support.
In order to study the dependence on the enhanced current of the DGF effect on the catalysis of Pd NCs, the electrochemical surface areas of the Pd NC/DGF SECM , Pd NC/DGF CM , and Pd NCs can be determined using a stripping experiment of Cu UPD 49,50 and an integrating the charge of the PdO reduction over 405 μC cm −2 , which is a theoretical value for the reduction charges of palladium oxide. 51-53 Figure 4 shows a comparison of the CV curves corresponding to Cu UPD and stripping on the surfaces of Pd NC/DGF SECM , Pd NC/DGF CM , Pd NCs, and Pd/C catalysts. In the CV curve of Pd NCs, Cu UPD was initiated at 0.32 V and a maximum current was observed at 0.19 V. With association with the DGF SECM and the DGF CM , initial Cu UPD potential shifted early to 0.37 V and the current was steeply increased. Interestingly, the increased current for the Cu UPD on the Pd NC/DGF SECM as compared with the Pd NC/DGF CM was more significant and a maximum current was early observed at 0.22 V. Compared with Pd NC/DGF SECM and Pd NC/DGF CM , a smaller current for Cu UPD on the Pd/C catalysts was observed. In the Cu stripping, apparently, the suppressed stripping currents was observed on the Pd NCs and Pd/C, whereas the significant current was for Pd NC/DGF SECM and Pd NC/DGF CM . The stripping current charge, as indicated by the black section subtracted from double-layer current for Pd NC/DGF SECM , was 766.2 μC, greater than 623.08 μC (red section) for the Pd NC/DGF CM , 263.6 μC (blue section) for the DGF-free Pd NCs and 340 μC (green section) for the Pd/C catalysts. Additionally, for electrochemically full coverage of Cu on the Pd (100) substrate and the Pd (111) substrate, the theoretical values are approximately 421 μC · cm −254,55 and 486 μC · cm −2 . 54 Therefore, the ESA UPD s for Pd NC catalysts enclosed with (111) planes and Pd/C catalysts dominated with (111) planes can be calculated from the stripping charges over 421 μC · cm −2 and 486 μC · cm −2 , respectively. The ESA UPD s, as summarized in Table I The effect of the DGF support on the catalytic power of Pd NCs toward D-glucose oxidation was further studied by Tafel measurements. Figure 5 depicts a comparison of the Tafel curves of the Pd NCs with and without DGF and Pd/C, and the corresponding kinetic data are summarized in Table I. In the curves of Pd NC/DGF SECM , the rest potential (E r ) as defined by zero overpotential was observed at −0.878 V prior to the A1 peak. The E r of Pd NC/DGF CM as well as that of Pd NC/DGF SECM shows an early potential (−0.876 V), confirming that the dehydrogenation of D-glucose is the first elementary step in the mechanism. 11 At E r , the current density can be determined to be the exchange current density (j 0 ), which was 4.3 × 10 −2 mA · cm −2 under a Tafel slope of 54 mV/dec for D-glucose electrooxidation. Careful study of Fig  (−0.85 V) for Pd NCs, the E r (−0.851 V) for Pd/C catalyst, showing early occurrence of D-glucose oxidation upon electrocatalysis using Pd NC/DGF SECM . Subsequently, as revealed in Fig. 6 and Table I, the j 0 s for Pd NC/DGF SECM and Pd NCs were 1.774 × 10 −1 mA · cm −2 under a Tafel slope of 51.9 mV/dec and 3.858 × 10 −2 mA · cm −2 under a Tafel slope of 75.3 mV/dec, respectively. The j 0 for Pd/C was 4.085 × 10 −2 mA · cm −2 under a Tafel slope of 63.8 mV/dec. The j 0 s for Pd NC/DGF SECM and Pd NC/DGF CM were higher than those for DGF-free Pd NCs and Pd/C and 6.2 × 10 −4 mA · cm −2 for Pd nanocatalysts supported on a carboxylated carbon nanotube. 28 Typically, the electron transfer number (n) involved in an electrocatalytic reaction can be calculated from the Tafel slope (B). 56 2.303RT βn F = B [2] Here, β is an asymmetric parameter and is 0.5, F is the Faraday constant, R is the gas constant, and T is the temperature (298 K). Therefore, the n values for Pd NC/DGF SECM and Pd NC/DGF CM were 2.28 and 2.19, respectively. The n value for Pd NCs was 1.57 and the n value for Pd/C was 1.85. The n values of Pd NC/DGF CM and Pd NC/DGF SECM were greater than 1, the theoretical electron number in A1 reaction. Additionally, the charge transfer resistance (R e ) can be calculated using the following equation: 56 Therefore, the R e values for Pd NCs and Pd/C were determined to be 4.23 × 10 −2 and 3.4 × 10 −2 · m 2 , respectively. The R e for Pd NC/DGF SECM was 6.38 × 10 −3 · m 2 and the R e for Pd NC/DGF CM was 2.72 × 10 −3 · m 2 . With the support of DGF SECM dominated with physical defects reported in our previous study, 40 the R e for the catalysis of GOR by Pd NCs can effectively decrease. Additionally, it has been concluded that the particle size of Pd catalysts cannot influence their GOR activities in a 0.1 M NaOH solution containing 10 mM glucose when their specific activities in terms of ESA and the loading weight of Pd catalyst are compared. 48 In this case, the specific activities for Pd NC/DGF SECM and Pd NC/DGF CM were 18.4 C · g −1 · cm −2 and 16.52 C · g −1 · cm −2 , respectively. The specific activity was calculated from the Q A2 over the loading weight of Pd NCs and ESA UPD , listed in Table I. The greater specific activity of Pd NC/DGF SECM could be resulted from lower R e .
Next, the stabilities of these supported Pd catalysts and DGF-free Pd NCs for GOR were studied by repeated CV testing. After repeated scans, the original currents (Fig. S3) of Pd NC/DGF CM and also Pd NCs and Pd/C were obtained. The current by Pd NC/DGF SECM was slightly increased. Additionally, as shown in the corresponding TEM images (Fig. S4) for the stability tests, these Pd catalysts didn't be significantly aggregated and the original morphologies of these Pd catalysts were remained. D-glucose and formed product easily bind to and desorb from the surfaces of these Pd catalysts, suggesting high D-glucose-tolerance for all Pd catalysts.
Applications for sensing D-glucose using Pd NCs without and with DGF.-The above electrocatalytic study revealed the effect of the DGF SECM on D-glucose oxidation and the potential of the prepared Pd NC/DGF SECM as glucose sensors. Figure 6 shows the amperometric responses to D-glucose catalyzed with the Pd NC/DGF SECM in comparison to those of the Pd NC/DGF CM , the Pd NCs, and the Pd/C at an applied potential of −0.05 V. To reduce interference from the mass transfer of glucose, the 0.1 M NaOH electrolyte was stirred at 200 rpm. When an aliquot of D-glucose ranging in concentration from 0.25 to 1 mM was dropped into the stirred NaOH solution, the oxidation current rose steeply before reaching a steady-state value. Initially, the responding current for Pd NC/DGF SECM was almost equal to those for Pd NC/DGF CM , Pd NCs and Pd/C after 0.25 mM D-glucose was added to the electrolyte. After sequentially dropping aliquots of  [4] Here, slope cal is the linear slope in Fig. 6B, and A geo is 0.07 cm 2 , the geometric surface area of the GCE. The sensitivities for the Pd NCs were 42.86 and 18.86 μA · mM −1 · cm −2 under linear analysis ranges of 0.25-5 and 5-24 mM with correlation coefficients of 0.992 and 0.991, respectively. When Pd NCs were associated with DGF CM , the sensitivities increased significantly to 67.29 and 35.57 μA · mM −1 · cm −2 at the same linear ranges. The correlation coefficients in the 0.25-5 and 5-24 mM ranges were 0.986 and 0.991, respectively. Interestingly, the sensitivities using DGF SECM as a support for Pd NCs can be further improved to 74 and 45.86 μA · mM −1 · cm −2 under linear analysis ranges of 0.25-5 and 5-24 mM with correlation coefficients of 0.984 and 0.993, respectively. Simultaneously, the sensitivities using Pd NC/DGF SECM are more sensible than 65 and 30.4 μA · mM −1 · cm −2 for Pd/C under linear analysis ranges of 0.25-7 and 7-20 mM with correlation coefficients of 0.9952 and 0.9983, respectively. These linear slopes completely cover the normal physiological level of glucose from 3 to 8 mM in human blood. In addition, the limit of detection (LOD) is the lowest concentration of glucose that can be measured with reasonable statistical certainty. 58 Based on the calibration curves (Fig. 6B) and the function for calculating the LOD, 59 the LOD at a signal-to-noise ratio of 3 for Pd NC/DGF SECM is 20.6 μM, lower than 27.2μM for Pd NC/DGF CM , 31 μM for DGF-free Pd NCs, and 29.4 μM for Pd/C. Table II shows a comparison of the sensitivities, linear ranges, and LODs of Pd, Pt, and Au catalysts on various carbon supports used for non-enzymatic glucose sensing in a NaOH solution. A comparison revealed that the Pd NC without and with DGFs show higher sensitivities 60,61 but a narrower linear range 17,61 when sensing glucose than the Pd or Pt nanoparticles on carbon nanotubes. The higher sensitivity of Pd NC/DGF SECM indicates the potential for the use of this electrode in sensing glucose.
For the sensor to be feasible, the effects of several possible interfering substances on the Pd NCs with and without DGF were studied. It is worth noting that these two Pd catalysts show remarkable resistance to foreign substances in the determination of D-glucose. Figure 7 shows a comparison of the current-time curves for Pd NC/DGF SECM , Pd NC/DGF CM , Pd NCs, and Pd/C at an applied potential of −0.05 V in a NaOH solution with an addition order of 1 mM D-glucose, 20 μM AA, 60 μM UA, and 1 mM D-glucose. The curves of Pd NC/DGF CM , and Pd NCs shows the current were slightly decreased after the addition of UA. The decayed current caused by UA was more obvious under catalysis of commercial Pd/C catalyst. These data are not consistent with our previous result for 27-nm Pd NCs, 23 by which the GOR current did not be interfered by UA. In the curve of Pd NC/DGF SECM , obviously, the addition of AA and UA did not influence the responding signal of D-glucose and block the active sites on the catalyst. The prepared Pd NC/DGF SECM catalysts still showed a stable and significant current for D-glucose despite the presence of AA and UA. Careful comparison with Pd NCs revealed that the Pd NC/DGF SECM showed a higher responding current for D-glucose but was inactive for AA and UA. It also can be concluded that the Pd NC/DGF SECM electrodes show good selectivity in glucose detection. Subsequently, real sample analyses were conducted for a calf serum sample, in which the glucose concentration was determined to be  The detected amounts of glucose were obtained from a mean of three measurements. c Recovery = ((C)/(A + B)) × 100%. d The relative standard deviation (R.S.D) was obtained from the found glucose amounts of three measurements. 5.32 mM using a commercial glucometer (OneTouch Ultra Easy, Johnson & Johnson). The analyses of the sensors using the Pd NC/DGF SECM as catalysts were evaluated by a calibration curve method to determine the recovery of various concentrations of glucose mixed with the 0.1-mL serum sample added to a 4-mL NaOH (0.1 M) solution. In this case, the calibration curves in Fig. 6B were chosen and the corresponding results are listed in Table III; higher recoveries by the Pd NC/DGF SECM demonstrate that the methodology can be applied for the detection of glucose in serum samples. We can thus conclude that Pd NC/DGF SECM display potential as a non-enzymatic glucose sensor.

Conclusions
A fair comparison between DGF SECM and DGF CM regarding their respective support effects on GOR catalysis via Pd NCs in a NaOH electrolyte indicated that the Pd NC/DGF SECM demonstrated earlier onset potentials, starting from −0.878 V (vs. Ag/AgCl), and an overall order of Pd NC/DGF SECM > Pd NC/DGF CM > Pd/C > Pd NCs for the A2 charge accompanying the formation of glucolactone was obtained. Through the use of a DGF SECM substrate, the electrochemical surface area and R e toward GOR were significantly improved. In addition, as supported by the catalytic characterization for GOR, Pd NC/DGF SECM showed remarkable sensitivity and tolerance to foreign substances in the application as a non-enzymatic glucose sensor, where Pd NC/DGF SECM showed higher two-period sensitivities of 74 and 45.86 μA · mM −1 · cm −2 for 0.25-5 and 5-24 mM D-glucose, respectively. The high recovery in serum sample analyses further confirmed the potential of Pd NC/DGF SECM as glucose sensors.