Defect Surface Engineering of Hollow NiCo2S4 Nanoprisms towards Performance-Enhanced Non-Enzymatic Glucose Oxidation

Transition metal sulfides have been explored as electrode materials for non-enzymatic detection. In this work, we investigated the effects of phosphorus doping on the electrochemical performances of NiCo2S4 electrodes (P-NiCo2S4) towards glucose oxidation. The fabricated non-enzymatic biosensor displayed better sensing performances than pristine NiCo2S4, with a good sensitivity of 250 µA mM−1 cm−2, a low detection limit (LOD) of 0.46 µM (S/N = 3), a wide linear range of 0.001 to 5.2 mM, and high selectivity. Moreover, P-NiCo2S4 demonstrated its feasibility for glucose determination for practical sample testing. This is due to the fact that the synergetic effects between Ni and Co species, and the partial substitution of S vacancies with P can help to increase electronic conductivity, enrich binary electroactive sites, and facilitate surface electroactivity. Thus, it is found that the incorporation of dopants into NiCo2S4 is an effective strategy to improve the electrochemical activity of host materials.


Introduction
The accurate measurement of glucose concentration from blood or other sources is of great significance to diagnose diabetes and monitor food quality in the field of pharmaceuticals and foods [1]. Recently, electrochemical sensors provide such simple operation, rapid response, and high sensitivity that they have aroused wide attention for use in glucose detection. Commercial glucose biosensors are generally based on the glucose oxidase enzyme (GO x ), where glucose is converted into gluconolactone in the presence of saturated oxygen. The enzymatic sensors exhibit outstanding sensitivity and selectivity but suffer from several limitations caused by environmental effects, high cost, and tedious enzyme immobilization process [2]. In order to break the above-mentioned drawbacks, it is essential to develop and explore the non-enzymatic sensors for direct electrooxidation of glucose. Over the past decades, thanks to their highly active area, excellent stability, and relatively cheap price, various transition metal chalcogenides, including oxides, sulfides, and selenides, have been of great promise in electrochemical sensors. Among these, transition metal sulfides with high electrochemical activity have drawn extensive attention [3][4][5][6]. Compared with oxides, sulfides have a more striking electrochemical performance with outstanding electronic conductivity and abundant redox chemical properties, because of their more flexible structure with an elongation of chemical bonds constructed by replacing O with S, which benefits electron transport [7,8]. In particular, NiCo 2 S 4 usually exhibits high electrochemical activities as a kind of single-phase binary metal sulfide, rather than the simple mixture of NiS x and CoS x , due to rich chemical redox, low cost, complex chemical compositions, abundant resources, environmental friendliness, and the synergetic effect of both individual components [9]. In spite of promising results, NiCo 2 S 4 still suffers from a reduced density of electrochemically active sites and severe polarization on account of lower conductivity and volume changes during electrochemical reactions. It still cannot meet the demands of high capacitance, which brings a challenge to its practical utilization [10,11]. Therefore, enormous research efforts have been devoted to enhancing the electrochemical performance of NiCo 2 S 4 through regulating nanostructures or combinations with conductive materials.
Typical nanostructures of nanowires, nanotubes, or nanorods could remarkably enlarge the accessible areas between the electrolyte and electrode and shorten the ionic diffusion distance, while introducing conductive materials may optimize the electrical conductivity and extend the electroactive surface [3,12,13]. For example, Huang's group fabricated a non-enzymatic glucose sensor based on the 3D flower-like NiCo 2 S 4 deposited Ni-modified cellulose filter paper, exhibiting a wide linear range of 0.5 µM-6 mM, high sensitivity of 283 µA mM −1 cm −2 , and a low detection limit (LOD) of 50 nM [14]. Guo et al. reported a NiCo 2 S 4 nanowire array with a unique core-shell structure grown on electrospun graphitic nanofiber film, endowed with a wide linear range and a low LOD for glucose sensing [15]. Although NiCo 2 S 4 could achieve outstanding electrochemical properties through the above two methods, it is not always feasible to modulate the inherent electronic properties to ideal levels. Recent reports have demonstrated that an additional surface-modified coating can change surface charge, microenvironment, and active site exposure, to achieve the purpose of adjusting activity [16]. Especially, anionic doping is verified as an effective method to introduce the surface defects for altering electron density, thereby improving the redox reactivity of the NiCo 2 S 4 complex [13,17,18]. Among them, the phosphorus (P) element with fully vacant 3D orbitals and lone-pair electrons has drawn enormous attention for promoting the surface electroactivity of host materials. Thereby, the surface P-anion doping of NiCo 2 S 4 might accommodate the surface charge state by tuning the partial charge density of neighboring bonded Ni and Co atoms. This method would enhance the redox activity of NiCo 2 S 4 by optimizing the adsorption energy of electrochemical analytes and reducing the strain during redox reactions [13,[19][20][21].
Inspired by these analyses, P-doping NiCo 2 S 4 with a hollow nanoprism-like structure (P-NiCo 2 S 4 HNPs) is successfully synthesized in this work through a simple phosphatization reaction. Benefiting from the unique hollow structure and surface engineering of NiCo 2 S 4 , the P-NiCo 2 S 4 HNPs have offered a better electrochemical sensing performance than that of pristine NiCo 2 S 4 , such as higher sensitivity, better selectivity, and reproducibility, as well as excellent feasibility toward glucose determination in practical samples.

Chemicals and Materials
All reagents are of analytical grade and used without further purification.

Apparatus
The morphologies and structures of the as-prepared samples were characterized by using scanning electron microscopy (SEM, Zelss Sigma300) at 3 kV and transmission electron microscopy (TEM, FEI Talos-F200s) at a voltage of 200 kV. The crystalline structure Biosensors 2022, 12, 823 3 of 12 and chemical states of the samples were verified by powder X-ray diffraction (XRD, D/max-2500 diffractometer, Rigaku, Japan), energy-dispersive spectroscopy (EDS), and X-ray photoelectron spectroscopy (XPS, Kratos Axis Ultra DLD spectrometer). Surface area allowing Brunauer-Emmett-Teller (BET) isotherms was carried out by monitoring N 2 adsorption/desorption using a NOVA 2000 surface area analyzer (Quantachrome) at 77 K.

Fabrication of the P-NiCo 2 S 4 HNPs
According to the previous literature, NiCo 2 S 4 NPs were firstly synthesized [22]. Briefly, Co(OAc) 2 ·4H 2 O (0.25 g), Ni(OAc) 2 ·4H 2 O (0.12 g), and urea (0.56 g) were successively added into EtOH (80 mL) under stirring. Then, the resulting mixture was heated to 65 • C for 4 h, and Co/Ni precursors were obtained after processing. Then, the Co/Ni precursors were re-dispersed into EtOH, followed by the addition of TAA. Subsequently, the resulting solution was transferred into a Teflon-lined stainless-steel autoclave and heated to 160 • C for 6 h. NiCo 2 S 4 NPs were obtained after processing and drying in the air. Finally, NiCo 2 S 4 (0.05 g) and NaH 2 PO 2 (0.2 g) were placed in tandem and annealed at 350 • C for 2 h under a flowing N 2 atmosphere to produce the P-NiCo 2 S 4 HNPs.

Electrochemical Measurements
Before preparing modified electrodes, the indium tin oxide (ITO, diameter of 3 mm) electrodes were successively sonicated and cleaned with acetone, EtOH, and water for use. After sonication for an adequate time, 6 µL of a uniform dispersion of P-NiCo 2 S 4 HNPs in EtOH (1 mg mL −1 ) was drop-casted onto the prepared ITO and dried at room temperature for next use. Cyclic voltammogram (CV) and chronoamperometry were performed by a CHI 660D instrument (Chenhua Instrument Co., Shanghai, China) to evaluate the electrochemical performance of the samples. In addition, the glucose detection was carried out in 0.2 M NaOH solution.

Results and Discussion
As illustrated in Figure 1A, the NiCo 2 S 4 was synthesized using our previous method [22], and the synthetic procedure for P-NiCo 2 S 4 was described through a facile P-doping. Utilizing CO 3 2− and OH − derived from the hydrolysis of urea in the presence of Co 2+ and Ni 2+ , the Co/Ni precursors were obtained by a solution reaction and then employed as the self-engaged templates. The XRD pattern confirmed that the crystalline Co/Ni precursors possessed a tetragonal cobalt/nickel acetate hydroxide phase ( Figure S1, Supplementary Materials) [22]. Moreover, The EDS spectrum substantiated the successful synthesis of Co/Ni precursors with the mole ratio of Co to Ni of 2:1 ( Figure S2). In addition, the morphologies and structures were further characterized by SEM. As shown in Figure 1B,C, the uniform prism-like Co/Ni precursors possessed a length of~1.3 µm and a width of~280 nm with a smooth surface. After sulfidation and the follow-up phosphating process, the length and width of the resulting P-NiCo 2 S 4 were~1.5 µm and~250 nm, respectively ( Figure 1D,E). It could be seen that the P-NiCo 2 S 4 HNPs still maintained the original prism-like appearance with a hollow structure and rough surface, which was of benefit to enhancing the electron transfer between interfaces. As shown in Figure 1F of the TEM image of the P-NiCo 2 S 4 HNPs, this further demonstrated the detailed geometrical morphology and the hollow interior structure. The XRD pattern of pristine P-NiCo 2 S 4 was presented in Figure 2A. Obviously, all the diffraction peaks can well correspond to the cubic NiCo 2 S 4 phase (JCPDS Card. No. 20-0782) [13,22,23]. The result indicated that the NiCo 2 S 4 crystal structure was not distorted significantly by the introduction of P [13,23]. In accordance with the XRD analysis, the high-resolution TEM (HRTEM) depicted that the inter-planar distances of P-NiCo 2 S 4 crystals were~0.28 nm and~0.56 nm, corresponding to the (311) and (111) planes of P-NiCo 2 S 4 , respectively ( Figure 1G). Additionally, the corresponding EDS mapping further confirmed the uniform distribution of Co, Ni, S, and P elements within the hollow prisms ( Figure 1H). Such results stayed in step with the EDS spectrum ( Figure S3). the inter-planar distances of P-NiCo2S4 crystals were ~0.28 nm and ~0.56 nm, corresponding to the (311) and (111) planes of P-NiCo2S4, respectively ( Figure 1G). Additionally, the corresponding EDS mapping further confirmed the uniform distribution of Co, Ni, S, and P elements within the hollow prisms ( Figure 1H). Such results stayed in step with the EDS spectrum ( Figure S3). As shown in Figure 2, XPS was performed to further identify the chemical environment and surface element state of the as-prepared P-NiCo2S4 HNPs. The XPS survey spectrum confirmed the presence of P, S, O, Co, and Ni elements on the surface of P-NiCo2S4 HNPs ( Figure 2B). For the 2p spectrum of Co ( Figure 2C), it could be noticed that two main peaks at 782.4 and 798.7 eV were consistent with Co 2p3/2 and Co 2p1/2 of Co 2+ . Two other peaks at 778.8 and 793.9 eV could be indexed to Co 2p3/2 and Co 2p1/2 of Co 3+ , respectively, together with two satellite peaks at 786.2 and 803.5 eV [5,24]. The Ni 2p spectrum exhibited similar characteristics, as shown in Figure 2D. Two peaks at 853.3 and 870.2 eV were attributed to Ni 2p3/2 and Ni 2p1/2 of Ni 2+ , while two additional peaks at 857.4 and 871.3 eV resulted from Ni 2p3/2 and Ni 2p1/2 of Ni 3+ , respectively [25]. For the 2p spectrum of the S element ( Figure 2E), the peak at 161.7 eV (S 2p3/2) was ascribed to the metal-sulfur bonds, while the binding energy of 162.8 eV (S 2p1/2) was linked to S 2-with low coordination at the surface. In addition, the two signals at 129.6 and 130.2 eV were assigned to P 2p3/2 and P 2p1/2, relating to metal phosphides. Moreover, the peak at 133.4 eV belonged to the oxidized phosphorus species (POx), which originated from the exposure of its surface to air for ages [26]. Finally, the XPS result suggested a P content of about 3.8 at%. Therefore, the XPS results demonstrated the doping of P in NiCo2S4, which would increase the number of active sites, improve its conductivity, and thus enhance its electrochemical performance. The CV was conducted to investigate the mechanism and electrochemical sensor activity of the P-NiCo2S4 HNPs modified electrode for glucose oxidation via a standard three-electrode system in a 0.2 M NaOH solution at a scan rate of 50 mV s −1 . As shown in Figure 3A, although the CVs of bare ITO in the absence (black line) and presence (red line) of 1 mM glucose in an applied potential range of 0-0.6 V were active for glucose electrooxidation, the response was very weak. Figure 3B displays the CV curves of P-NiCo2S4/ITO with cathodic and anodic peaks at about 0.4 and 0.45 V, respectively. It was noticeable that a higher glucose oxidation peak was observed when adding 1 mM glucose, suggesting the excellent electrocatalytic activity of P-NiCo2S4/ITO towards glucose oxidation. The As shown in Figure 2, XPS was performed to further identify the chemical environment and surface element state of the as-prepared P-NiCo 2 S 4 HNPs. The XPS survey spectrum confirmed the presence of P, S, O, Co, and Ni elements on the surface of P-NiCo 2 S 4 HNPs ( Figure 2B). For the 2p spectrum of Co ( Figure 2C), it could be noticed that two main peaks at 782.4 and 798.7 eV were consistent with Co 2p 3/2 and Co 2p 1/2 of Co 2+ . Two other peaks at 778.8 and 793.9 eV could be indexed to Co 2p 3/2 and Co 2p 1/2 of Co 3+ , respectively, together with two satellite peaks at 786.2 and 803.5 eV [5,24]. The Ni 2p spectrum exhibited similar characteristics, as shown in Figure 2D. Two peaks at 853.3 and 870.2 eV were attributed to Ni 2p 3/2 and Ni 2p 1/2 of Ni 2+ , while two additional peaks at 857.4 and 871.3 eV resulted from Ni 2p 3/2 and Ni 2p 1/2 of Ni 3+ , respectively [25]. For the 2p spectrum of the S element ( Figure 2E), the peak at 161.7 eV (S 2p 3/2 ) was ascribed to the metal-sulfur bonds, while the binding energy of 162.8 eV (S 2p 1/2 ) was linked to S 2− with low coordination at the surface. In addition, the two signals at 129.6 and 130.2 eV were assigned to P 2p 3/2 and P 2p 1/2 , relating to metal phosphides. Moreover, the peak at 133.4 eV belonged to the oxidized phosphorus species (PO x ), which originated from the exposure of its surface to air for ages [26]. Finally, the XPS result suggested a P content of about 3.8 at%. Therefore, the XPS results demonstrated the doping of P in NiCo 2 S 4 , which would increase the number of active sites, improve its conductivity, and thus enhance its electrochemical performance.
The CV was conducted to investigate the mechanism and electrochemical sensor activity of the P-NiCo 2 S 4 HNPs modified electrode for glucose oxidation via a standard three-electrode system in a 0.2 M NaOH solution at a scan rate of 50 mV s −1 . As shown in Figure 3A, although the CVs of bare ITO in the absence (black line) and presence (red line) of 1 mM glucose in an applied potential range of 0-0.6 V were active for glucose electrooxidation, the response was very weak. Figure 3B displays the CV curves of P-NiCo 2 S 4 /ITO with cathodic and anodic peaks at about 0.4 and 0.45 V, respectively. It was noticeable that a higher glucose oxidation peak was observed when adding 1 mM glucose, suggesting the excellent electrocatalytic activity of P-NiCo 2 S 4 /ITO towards glucose oxidation. The corresponding reaction mechanism for the glucose oxidation on the P-NiCo 2 S 4 HNPs modified electrode can be briefly described as follows: NiS Biosensors 2022, 12, x FOR PEER REVIEW 6 of 12 The identification of the optimal concentration of NaOH solution for an effective nonenzymatic glucose sensor was monitored at P-NiCo2S4/ITO for a range of 0.01~0.5 M (Figure 4A). Upon the increment in concentration from 0.01 to 0.2 M, the electrooxidation kinetics of glucose was dramatically enhanced, and the utmost response towards glucose was observed in 0.2 M NaOH solution. High-alkaline conditions will generate higher oxidation state species (Ni 2+ /Ni 3+ and Co 3+ /Co 4+ ), which provide the maximal glucose oxidation responses at P-NiCo2S4/ITO. However, stronger alkaline conditions tend to corrode In addition, it can be easily seen from a comparison of current increments (∆I) for bare ITO and P-NiCo 2 S 4 /ITO in Figure 3C that the electrochemical performance of the electrode modified with P-NiCo 2 S 4 was greatly improved. Figure 3D presents the CVs of P-NiCo 2 S 4 /ITO with different glucose concentrations to evaluate the feasibility of glucose sensing, and the anodic peak current significantly increased with increasing concentration from 0 to 10 mM, representing typical catalytic oxidation of glucose. As shown in Figure 3E, the CVs of P-NiCo 2 S 4 /ITO at a series of scan rates with 1 mM glucose suggested that increasing the scan rate will lead to a rise in currents and a shift in potentials, which mainly comes from the increasing internal diffusion resistance within P-NiCo 2 S 4 HNPs as the scan rate increases [2,27]. Furthermore, Figure 3F exhibits a good linear dependence relation between the anodic peak current and the square root of the scan rate with a linear regression coefficient (R 2 ) of 0.9838, implying that the glucose oxidation on the P-NiCo 2 S 4 /ITO is a reversible and typical diffusion-controlled electrochemical process [28].
The identification of the optimal concentration of NaOH solution for an effective non-enzymatic glucose sensor was monitored at P-NiCo 2 S 4 /ITO for a range of 0.01~0.5 M ( Figure 4A). Upon the increment in concentration from 0.01 to 0.2 M, the electrooxidation kinetics of glucose was dramatically enhanced, and the utmost response towards glucose was observed in 0.2 M NaOH solution. High-alkaline conditions will generate higher oxidation state species (Ni 2+ /Ni 3+ and Co 3+ /Co 4+ ), which provide the maximal glucose oxidation responses at P-NiCo 2 S 4 /ITO. However, stronger alkaline conditions tend to corrode electrodes resulting in limiting the electrochemical reaction [14]. Next, the optimal value of the applied potential was achieved in order to investigate the effect of detection potential on the amperometric response of P-NiCo 2 S 4 /ITO to glucose. As shown in Figure 4B, the oxidation current increased with increasing the detection potential and reached the highest value at 0.4 V. Therefore, 0.4 V was chosen as the optimal working potential for the follow-up study. Subsequently, for a systematic comparison, the P-NiCo 2 S 4 nanoprisms with different amounts of NaH 2 PO 2 were also prepared under the same synthetic conditions. In addition, the amperometric current responses of P-NiCo 2 S 4 /ITO with 0.1 g, 0.2 g, and 0.4 g NaH 2 PO 2 at 0.4 V were represented in Figure 4C, which exhibited an almost obvious current difference. It was found that 0.2 g NaH 2 PO 2 was the optimal admixing quantity in the following experiments. To analyze the interfacial behaviors of P-NiCo 2 S 4 with 0.1 g, 0.2 g, and 0.4 g NaH 2 PO 2 , the electrochemical impedance spectroscopy (EIS) was performed in 0.1 M KCl solution containing 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ]. Obviously, the P-NiCo 2 S 4 with 0.2 g NaH 2 PO 2 showed a lower charge-transfer resistance (diameter of the semicircle) value than those of the other two, revealing the highest electrical conductivity of P-NiCo 2 S 4 with 0.2 g NaH 2 PO 2 , and the result was consistent with Figure 4C. P-doping content has an effect on the electrical conductivity and electrocatalytic activity of materials. The high content of P-doping can significantly improve the fixation site and catalytic site of polysulfide, and thus promote the catalytic activity of materials. However, excessive P-doping will seriously affect the structural integrity of the materials, which significantly reduces the electrical conductivity [19,29,30]. The appropriate amount of P-doping can accelerate electron transport, and thus it would be a promising electrode to fabricate sensitive sensors. As shown in Figure S4, the BET surface areas of P-NiCo 2 S 4 with 0.2 g and 0.4 g NaH 2 PO 2 were 53.679 and 56.725 m 2 g −1 , respectively, which were slightly higher than that of P-NiCo 2 S 4 with 0.1 g NaH 2 PO 2 (43.558 m 2 g −1 ). However, compared with pure NiCo 2 S 4 , P-doping can improve the specific surface area and electrochemical activity. Besides, the effective surface area of working electrodes will directly influence their sensitivity, thus playing a large role in developing electrochemical sensors. The CVs of P-NiCo 2 S 4 /ITO were measured in 5 mM K 3 Fe(CN) 6 containing 0.1 M KCl solution, shown in Figure S5A. Additionally, Figure S5B indicates that the I p values of P-NiCo 2 S 4 /ITO increase linearly with the square root of scan rates. Therefore, the effective surface area of the P-NiCo 2 S 4 was assessed by the Randles-Sevcik equation: I p = (2.69 × 10 5 )n 3/2 AD 1/2 v 1/2 C, where n was the number of electrons transferred, D was the diffusion coefficient of Fe(CN) 6 3− (7.6 × 10 −6 cm 2 s −1 ), C was the reactant concentration (mol cm −3 ), v is the scan rate (V s −1 ), A was the effective Biosensors 2022, 12, 823 7 of 12 area of the electrode, and I p was the peak current [31,32]. The slope was the ratio of I p to v 1/2 ( Figure S5B); thus, the effective surface area of the P-NiCo 2 S 4 was calculated as 0.134 cm 2 .
CVs of P-NiCo2S4/ITO were measured in 5 mM K3Fe(CN)6 containing 0.1 M KCl solutio shown in Figure S5A. Additionally, Figure S5B indicates that the Ip values of P NiCo2S4/ITO increase linearly with the square root of scan rates. Therefore, the effectiv surface area of the P-NiCo2S4 was assessed by the Randles-Sevcik equation: Ip = (2.69 10 5 )n 3/2 AD 1/2 v 1/2 C, where n was the number of electrons transferred, D was the diffusio coefficient of Fe(CN)6 3-(7.6 × 10 -6 cm 2 s -1 ), C was the reactant concentration (mol cm −3 ), v the scan rate (V s -1 ), A was the effective area of the electrode, and Ip was the peak curren [31,32]. The slope was the ratio of Ip to v 1/2 ( Figure S5B); thus, the effective surface area the P-NiCo2S4 was calculated as 0.134 cm 2 . The amperometric analysis was used as an extremely attractive electrochemical tech nique to study the sensitivity, selectivity, and detection limit of the proposed glucose sen sor. Under the above-optimized conditions, Figure 5A depicts a typical current-time pl of the P-NiCo2S4/ITO on the successive step-wise addition of glucose. The inset in Figu  5A represents the magnification of glucose concentration ranging from 1 to 4 µM. A wid linear response of P-NiCo2S4/ITO for glucose sensing from 0.001 to 5.2 mM was shown Figure 5B. The linear regression equation for P-NiCo2S4/ITO was I/µA = 17.66C/mM + 1.7 (R 2 = 0.9903) with a high sensitivity of 250.0 µA mM -1 cm -2 . As a consequence, the LOD the sensor was calculated to be 0.46 µM (S/N = 3). In addition, after adding glucose to th NaOH solution, the P-NiCo2S4 sensor produced steady-state signals less than 0.1s (Figu The amperometric analysis was used as an extremely attractive electrochemical technique to study the sensitivity, selectivity, and detection limit of the proposed glucose sensor. Under the above-optimized conditions, Figure 5A depicts a typical currenttime plot of the P-NiCo 2 S 4 /ITO on the successive step-wise addition of glucose. The inset in Figure 5A represents the magnification of glucose concentration ranging from 1 to 4 µM. A wide linear response of P-NiCo 2 S 4 /ITO for glucose sensing from 0.001 to 5.2 mM was shown in Figure 5B. The linear regression equation for P-NiCo 2 S 4 /ITO was I/µA = 17.66C/mM + 1.72 (R 2 = 0.9903) with a high sensitivity of 250.0 µA mM −1 cm −2 . As a consequence, the LOD of the sensor was calculated to be 0.46 µM (S/N = 3). In addition, after adding glucose to the NaOH solution, the P-NiCo 2 S 4 sensor produced steady-state signals less than 0.1s ( Figure 5C). Table 1 lists the detection performance comparison of P-NiCo 2 S 4 /ITO with some similar non-enzymatic glucose sensors. It is clear that the performance of P-NiCo 2 S 4 /ITO is comparable or superior to some reports, for example, Co 3 O 4 /NiCo 2 O 4 , NiCo 2 S 4 /Pt, and CoNi 2 S 4 @NCF [33][34][35]. The remarkable electrochemical performance of P-NiCo 2 S 4 /ITO could stem from the hollow structures, synergetic effects between Ni and Co species, and partial substitution of S vacancies with P. This can provide more electrochemical active sites and enhance electrical conductivity, implying the potential value of P-NiCo 2 S 4 in many systems for glucose detection. ilar non-enzymatic glucose sensors. It is clear that the performance of P-NiCo2S4/ITO is comparable or superior to some reports, for example, Co3O4/NiCo2O4, NiCo2S4/Pt, and CoNi2S4@NCF [33][34][35]. The remarkable electrochemical performance of P-NiCo2S4/ITO could stem from the hollow structures, synergetic effects between Ni and Co species, and partial substitution of S vacancies with P. This can provide more electrochemical active sites and enhance electrical conductivity, implying the potential value of P-NiCo2S4 in many systems for glucose detection.  Selectivity and stability are the major factors to evaluate the performance of the nonenzymatic glucose sensor. It is clear that the P-NiCo2S4/ITO displays an anti-interference  Selectivity and stability are the major factors to evaluate the performance of the nonenzymatic glucose sensor. It is clear that the P-NiCo 2 S 4 /ITO displays an anti-interference advantage after the addition of glucose (1 mM) and a series of interfering species (0.1 mM each) at a working potential of 0.4 V, shown in Figure 5D, indicating the good ability of anti-interference of P-NiCo 2 S 4 with negligible interference from urea, NaCl, KCl, AA, UA, Fru, Lcy, and AP. The long-term stability of the P-NiCo 2 S 4 /ITO electrode was evaluated by measuring the current peak towards 1 mM glucose over a week at room temperature. The electrode basically kept the same value as the original current after 7 days, showing good stability of the P-NiCo 2 S 4 electrode at room temperature ( Figure 6). The enhanced electrochemical behavior may be attributed to: (i) more redox reactions from multiple oxide states of the Ni and Co species, (ii) a much lower optical band gap energy and higher electric conductivity of binary metal sulfides, (iii) its unique hollow structure that enlarges surface area and shortens electronic transmission, and (iv) the anionic phosphorus doping that provides some new active sites and introduces surface defects.
anti-interference of P-NiCo2S4 with negligible interference from urea, NaCl, KCl, AA, UA, Fru, Lcy, and AP. The long-term stability of the P-NiCo2S4/ITO electrode was evaluated by measuring the current peak towards 1 mM glucose over a week at room temperature. The electrode basically kept the same value as the original current after 7 days, showing good stability of the P-NiCo2S4 electrode at room temperature ( Figure 6). The enhanced electrochemical behavior may be attributed to: (i) more redox reactions from multiple oxide states of the Ni and Co species, (ii) a much lower optical band gap energy and higher electric conductivity of binary metal sulfides, (iii) its unique hollow structure that enlarges surface area and shortens electronic transmission, and (iv) the anionic phosphorus doping that provides some new active sites and introduces surface defects. In addition, the feasibility of the proposed electrode was evaluated by monitoring glucose in human serum. The serum samples were donated by Jiangsu Province Hospital. The serum sample was firstly separated by centrifugation at 8000 rpm for 15 min, then the collected supernatant was obtained for further electrochemical tests. The standard addition method was employed to detect the glucose concentration in serum under optimal experimental conditions. The successive addition of glucose was continuously added into the mixture of 9.5 mL of 0.2 M NaOH and 0.5 mL of the treated serum supernatant ( Figure  S6A). The corresponding calibration curve of glucose ranging from 0.005 to 5.2 mM on the P-NiCo2S4/ITO was illustrated in Figure S6B, and the linear regression equation was ΔI/µA = 4.62C/mM + 36.51 (R 2 = 0.9891), disclosing good stability and anti-interference of the proposed glucose sensor for real blood samples. Moreover, we also measured the recoveries in serum samples by adding glucose solution with different concentrations. As presented in Table 2, the recoveries were calculated to be in the range of 96.4-101.8%. The above results demonstrated that the P-NiCo2S4/ITO had favorable feasibility of glucose detection in real serum samples.  In addition, the feasibility of the proposed electrode was evaluated by monitoring glucose in human serum. The serum samples were donated by Jiangsu Province Hospital. The serum sample was firstly separated by centrifugation at 8000 rpm for 15 min, then the collected supernatant was obtained for further electrochemical tests. The standard addition method was employed to detect the glucose concentration in serum under optimal experimental conditions. The successive addition of glucose was continuously added into the mixture of 9.5 mL of 0.2 M NaOH and 0.5 mL of the treated serum supernatant ( Figure S6A). The corresponding calibration curve of glucose ranging from 0.005 to 5.2 mM on the P-NiCo 2 S 4 /ITO was illustrated in Figure S6B, and the linear regression equation was ∆I/µA = 4.62C/mM + 36.51 (R 2 = 0.9891), disclosing good stability and anti-interference of the proposed glucose sensor for real blood samples. Moreover, we also measured the recoveries in serum samples by adding glucose solution with different concentrations. As presented in Table 2, the recoveries were calculated to be in the range of 96.4-101.8%. The above results demonstrated that the P-NiCo 2 S 4 /ITO had favorable feasibility of glucose detection in real serum samples.

Conclusions
In summary, we have rationally synthesized hollow NiCo 2 S 4 nanoprisms in situ using a urea precursor as a sacrifice template by an anion exchange reaction, followed by constructing P-NiCo 2 S 4 through simple P element doping for enhancing the electrochemical performance towards glucose oxidation. Compared with pristine NiCo 2 S 4 , a non-enzymatic sensor with good sensitivity, low detection limit, wide linear range, and high selectivity was established. It is confirmed that the incorporation of P can induce the surface defect at the atomic level to improve the electrochemical activity of host materials, which often suffer from slow ion diffusion and charge transfer kinetics, and the deficiency of electrochemically active sites. A series of electrodes with different amounts of P suggest that an appropriate amount of P-doping can accelerate electron transport. Moreover, P-NiCo 2 S 4 demonstrates its feasibility for glucose determination for practical sample testing. These results reveal that the synergetic effects between Ni and Co species and the partial substitution of S vacancies with P can help to increase electronic conductivity, enrich binary electroactive sites, and boost surface electroactivity. Thus, our proposed non-enzymatic sensor has great prospects in the application of detecting glucose rapidly and effectively. In addition, the reported nanomaterials-based electrodes have been mostly applied in strongly alkaline conditions for non-enzymatic glucose sensors so far. However, the electrochemical characteristics in weakly alkaline or even neutral environments close to human body fluid are rarely reported. It is possible that, before the test, a negative potential is first employed to reduce the water in situ, and the hydrogen gets released resulting in the formation of OH − ; thus, it produces an alkaline microenvironment, so as to realize glucose detection in a neutral environment. This will be further studied in the future.