An Enzymatic Glucose Sensor Composed of Carbon-Coated Nano Tin Sulfide

In this study, a biosensor, based on a glucose oxidase (GOx) immobilized, carbon-coated tin sulfide (SnS) assembled on a glass carbon electrode (GCE) was developed, and its direct electrochemistry was investigated. The carbon coated SnS (C-SnS) nanoparticle was prepared through a simple two-step process, using hydrothermal and chemical vapor deposition methods. The large reactive surface area and unique electrical potential of C-SnS could offer a favorable microenvironment for facilitating electron transfer between enzymes and the electrode surface. The structure and sensor ability of the proposed GOx/C-SnS electrode were characterized using scanning electron microscopy (SEM), X-ray diffraction (XRD), Raman spectroscopy, UV–vis spectroscopy, Fourier transform infrared spectroscopy (FTIR), and cyclic voltammetry study (CV).


Results and Discussion
The structures of the as-synthesized SnS 2 and carbon-coated powders were observed from SEM images and XRD patterns, as depicted in Figures 1 and 2, respectively. The SnS 2 nanoflake powder size, observed from the SEM image, shows an average size of 500 nm in width with a thickness of 75 nm ( Figure 1a); after carbon coating, the particle size increases dramatically and formed a laminar microstructure with a thickness of 200 nm (Figure 1b). From the XRD patterns in Figure 2a,b, corresponding to the as-synthesized SnS 2 and the carbon-coated powders, both structures fit well with the JCPDS cards corresponding to hexagonal SnS 2 and orthorhombic SnS crystalline structures; in addition, both possessed excellent crystallinity. The average domain sizes were 36 nm and 17 nm for SnS 2 and SnS, respectively, calculated by the full-width-half-maximum (FWHM) at each diffraction peak using the Scherrer equation [18], and the results indicate an improvement in crystallinity from SnS 2 to SnS. A small diamond carbon peak around 44 • was found in Figure 2b and the existence of carbon can be further confirmed from the Raman spectrum (Figure 3), in which the D and G bands of carbon with I(D)/I(G)~1 appeared after carbon coating, while the Raman signal of SnS 2 disappeared. The results of the Raman spectrum are consistent with those of the XRD patterns ( Figure 2). Nanomaterials 2017, 7, 39 3 of 9

Results and Discussion
The structures of the as-synthesized SnS2 and carbon-coated powders were observed from SEM images and XRD patterns, as depicted in Figures 1 and 2, respectively. The SnS2 nanoflake powder size, observed from the SEM image, shows an average size of 500 nm in width with a thickness of 75 nm ( Figure 1a); after carbon coating, the particle size increases dramatically and formed a laminar microstructure with a thickness of 200 nm (Figure 1b). From the XRD patterns in Figure 2a,b, corresponding to the as-synthesized SnS2 and the carbon-coated powders, both structures fit well with the JCPDS cards corresponding to hexagonal SnS2 and orthorhombic SnS crystalline structures; in addition, both possessed excellent crystallinity. The average domain sizes were 36 nm and 17 nm for SnS2 and SnS, respectively, calculated by the full-width-half-maximum (FWHM) at each diffraction peak using the Scherrer equation [18], and the results indicate an improvement in crystallinity from SnS2 to SnS. A small diamond carbon peak around 44° was found in Figure 2b

Results and Discussion
The structures of the as-synthesized SnS2 and carbon-coated powders were observed from SEM images and XRD patterns, as depicted in Figures 1 and 2, respectively. The SnS2 nanoflake powder size, observed from the SEM image, shows an average size of 500 nm in width with a thickness of 75 nm ( Figure 1a); after carbon coating, the particle size increases dramatically and formed a laminar microstructure with a thickness of 200 nm (Figure 1b). From the XRD patterns in Figure 2a,b, corresponding to the as-synthesized SnS2 and the carbon-coated powders, both structures fit well with the JCPDS cards corresponding to hexagonal SnS2 and orthorhombic SnS crystalline structures; in addition, both possessed excellent crystallinity. The average domain sizes were 36 nm and 17 nm for SnS2 and SnS, respectively, calculated by the full-width-half-maximum (FWHM) at each diffraction peak using the Scherrer equation [18], and the results indicate an improvement in crystallinity from SnS2 to SnS. A small diamond carbon peak around 44° was found in Figure 2b and the existence of carbon can be further confirmed from the Raman spectrum (Figure 3), in which the D and G bands of carbon with I(D)/I(G)~1 appeared after carbon coating, while the Raman signal of SnS2 disappeared. The results of the Raman spectrum are consistent with those of the XRD patterns ( Figure 2).      The transformation from SnS2 to SnS could be due to the phase stability of SnSx during the CVD process at 400 °C and also the reactive carbon gas environment [10,19]. According to the ab initio calculations by Burton et al. [19] and Vidal et al. [10], the SnS Pnma structure should be a preferable phase at 400 °C, which is consistent with our XRD and Raman observations. The excessive S during the phase transformation may assist the growth of amorphous hydrogenated carbon sulfur (a-C:H:S), and, thus, a clear D band was observed [16]. The enlarged laminar particle size and finer crystalline domain of SnS suggest that the SnS crystals were embedded in a preferentially-grown amorphous carbon, where the carbon layer grows faster along the surface (1-100) of the SnS2 nanoflake. The edge surface (1-100) of hexagonal SnS2 consisted of layered S and Sn atoms, which was able to promote the growth of a-C:H:S [16]. In sum, an obvious structural and compositional transformation, from hexagonal SnS2 to orthorhombic SnS, embedded in amorphous carbon with an improvement of crystallinity, could be achieved using a simple two-step synthesis process.
Prior to the glucose detection test, FTIR and UV-vis spectroscopies were used in order to examine the adhesion of GOx onto C-SnS, and the results are shown in Figures 4 and 5. Without the GOx layer, the bare C-SnS did not react to FTIR and UV-vis due to the large wavelength compared to the lattice parameters of carbon and SnS. With respect to GOx, the Amide I and II bands can be observed from the FTIR pattern at 1537 cm −1 and 1608 cm −1 ( Figure 4); additionally, the polypeptide chains at 276 nm and the oxidized form of the flavin groups in the proteins at 379 and 452 nm are also shown in the UV-vis pattern ( Figure 5). From the FTIR and UV-vis patterns, the combination GOx on C-SnS shows the same characteristic signals as those mentioned above, indicating that the structure and activity of GOx remain the same in C-SnS.  The transformation from SnS 2 to SnS could be due to the phase stability of SnS x during the CVD process at 400 • C and also the reactive carbon gas environment [10,19]. According to the ab initio calculations by Burton et al. [19] and Vidal et al. [10], the SnS Pnma structure should be a preferable phase at 400 • C, which is consistent with our XRD and Raman observations. The excessive S during the phase transformation may assist the growth of amorphous hydrogenated carbon sulfur (a-C:H:S), and, thus, a clear D band was observed [16]. The enlarged laminar particle size and finer crystalline domain of SnS suggest that the SnS crystals were embedded in a preferentially-grown amorphous carbon, where the carbon layer grows faster along the surface (1-100) of the SnS 2 nanoflake. The edge surface (1-100) of hexagonal SnS 2 consisted of layered S and Sn atoms, which was able to promote the growth of a-C:H:S [16]. In sum, an obvious structural and compositional transformation, from hexagonal SnS 2 to orthorhombic SnS, embedded in amorphous carbon with an improvement of crystallinity, could be achieved using a simple two-step synthesis process.
Prior to the glucose detection test, FTIR and UV-vis spectroscopies were used in order to examine the adhesion of GO x onto C-SnS, and the results are shown in Figures 4 and 5. Without the GO x layer, the bare C-SnS did not react to FTIR and UV-vis due to the large wavelength compared to the lattice parameters of carbon and SnS. With respect to GO x , the Amide I and II bands can be observed from the FTIR pattern at 1537 cm −1 and 1608 cm −1 ( Figure 4); additionally, the polypeptide chains at 276 nm and the oxidized form of the flavin groups in the proteins at 379 and 452 nm are also shown in the UV-vis pattern ( Figure 5). From the FTIR and UV-vis patterns, the combination GO x on C-SnS shows the same characteristic signals as those mentioned above, indicating that the structure and activity of GO x remain the same in C-SnS.  The transformation from SnS2 to SnS could be due to the phase stability of SnSx during the CVD process at 400 °C and also the reactive carbon gas environment [10,19]. According to the ab initio calculations by Burton et al. [19] and Vidal et al. [10], the SnS Pnma structure should be a preferable phase at 400 °C, which is consistent with our XRD and Raman observations. The excessive S during the phase transformation may assist the growth of amorphous hydrogenated carbon sulfur (a-C:H:S), and, thus, a clear D band was observed [16]. The enlarged laminar particle size and finer crystalline domain of SnS suggest that the SnS crystals were embedded in a preferentially-grown amorphous carbon, where the carbon layer grows faster along the surface (1-100) of the SnS2 nanoflake. The edge surface (1-100) of hexagonal SnS2 consisted of layered S and Sn atoms, which was able to promote the growth of a-C:H:S [16]. In sum, an obvious structural and compositional transformation, from hexagonal SnS2 to orthorhombic SnS, embedded in amorphous carbon with an improvement of crystallinity, could be achieved using a simple two-step synthesis process.
Prior to the glucose detection test, FTIR and UV-vis spectroscopies were used in order to examine the adhesion of GOx onto C-SnS, and the results are shown in Figures 4 and 5. Without the GOx layer, the bare C-SnS did not react to FTIR and UV-vis due to the large wavelength compared to the lattice parameters of carbon and SnS. With respect to GOx, the Amide I and II bands can be observed from the FTIR pattern at 1537 cm −1 and 1608 cm −1 ( Figure 4); additionally, the polypeptide chains at 276 nm and the oxidized form of the flavin groups in the proteins at 379 and 452 nm are also shown in the UV-vis pattern ( Figure 5). From the FTIR and UV-vis patterns, the combination GOx on C-SnS shows the same characteristic signals as those mentioned above, indicating that the structure and activity of GOx remain the same in C-SnS.   The results of the CV scans, using four different working electrodes, are shown in Figure 6. By adding C-SnS to GCE, the overall current range increases, but there is no obvious redox behavior. On the other hand, GOx shows a symmetric reduction and oxidation path, indicating that the redox behavior of GOx is a reversible reaction. With the assistance of C-SnS, the peak current of both the reduction and oxidation parts increased two times, and the applied potential was -0.41 V, which is comparable to results from the literature [20]. Hence, the GOx/C-SnS/GCE electrode shows a good redox behavior and, therefore, the stacking can serve as a glucose sensor. In order to calculate the electron transfer rate constant (Ks), the potential (ΔEp) difference between the reduction and oxidation was recorded using different scanning speeds (Figure 7), and calculated using Laviron's equation [21]. The Ks of our proposed sensor stacking is 7.461 s −1 , which is two times higher than those of similar glucose sensors (SnS2/GOx/GCE), proposed by Yang et al., which were measured at 3.68 s −1 [20]. Furthermore, the scanning rate showed a linear correlation with the redox current. We also utilized CV scans at different pH values (Figure 8) in order to optimize the sensing conditions. As the pH value decreased from 7 to 4, which is the working pH value of GOx, the reduction peak moved forward to a lower potential, and the corresponding current also decreased. The optimized response current occurred at pH = 7, which is the same pH value as that of human blood; hence, the proposed GOx/C-SnS/GCE sensor shows an excellent advantage in practical usage. The results of the CV scans, using four different working electrodes, are shown in Figure 6. By adding C-SnS to GCE, the overall current range increases, but there is no obvious redox behavior. On the other hand, GO x shows a symmetric reduction and oxidation path, indicating that the redox behavior of GO x is a reversible reaction. With the assistance of C-SnS, the peak current of both the reduction and oxidation parts increased two times, and the applied potential was -0.41 V, which is comparable to results from the literature [20]. Hence, the GO x /C-SnS/GCE electrode shows a good redox behavior and, therefore, the stacking can serve as a glucose sensor. The results of the CV scans, using four different working electrodes, are shown in Figure 6. By adding C-SnS to GCE, the overall current range increases, but there is no obvious redox behavior. On the other hand, GOx shows a symmetric reduction and oxidation path, indicating that the redox behavior of GOx is a reversible reaction. With the assistance of C-SnS, the peak current of both the reduction and oxidation parts increased two times, and the applied potential was -0.41 V, which is comparable to results from the literature [20]. Hence, the GOx/C-SnS/GCE electrode shows a good redox behavior and, therefore, the stacking can serve as a glucose sensor. In order to calculate the electron transfer rate constant (Ks), the potential (ΔEp) difference between the reduction and oxidation was recorded using different scanning speeds (Figure 7), and calculated using Laviron's equation [21]. The Ks of our proposed sensor stacking is 7.461 s −1 , which is two times higher than those of similar glucose sensors (SnS2/GOx/GCE), proposed by Yang et al., which were measured at 3.68 s −1 [20]. Furthermore, the scanning rate showed a linear correlation with the redox current. We also utilized CV scans at different pH values (Figure 8) in order to optimize the sensing conditions. As the pH value decreased from 7 to 4, which is the working pH value of GOx, the reduction peak moved forward to a lower potential, and the corresponding current also decreased. The optimized response current occurred at pH = 7, which is the same pH value as that of human blood; hence, the proposed GOx/C-SnS/GCE sensor shows an excellent advantage in practical usage. In order to calculate the electron transfer rate constant (Ks), the potential (∆E p ) difference between the reduction and oxidation was recorded using different scanning speeds (Figure 7), and calculated using Laviron's equation [21]. The Ks of our proposed sensor stacking is 7.461 s −1 , which is two times higher than those of similar glucose sensors (SnS 2 /GO x /GCE), proposed by Yang et al., which were measured at 3.68 s −1 [20]. Furthermore, the scanning rate showed a linear correlation with the redox current. We also utilized CV scans at different pH values (Figure 8) in order to optimize the sensing conditions. As the pH value decreased from 7 to 4, which is the working pH value of GO x , the reduction peak moved forward to a lower potential, and the corresponding current also decreased. The optimized response current occurred at pH = 7, which is the same pH value as that of human blood; hence, the proposed GO x /C-SnS/GCE sensor shows an excellent advantage in practical usage.  Figure 9 shows the sensitivity of the GOx/C-SnS/GCE sensor with respect to different glucose concentrations. According to the results of triplicates, the relative standard deviation (RSD) was less than 5.2% calculated from the current response of freshly prepared electrodes. The results convinced that the fabrication method was highly reproducible comparing with reported enzymatic and nonenzymatic glucose sensors [22,23]. A linear correlation, from 0.03 to 0.7 mM, can be found in glucose concentrations, and the sensitivity was 43.9 mA·M −1 ·cm −2 , roughly six times that found in Yang's work, using SnS2 [8]. Since human blood always contains different hormones and chemicals, which can interfere glucose sensor detection, an amperometric test was carried out using CA and UA, as depicted in Figure 10. Although the reactions are not clear, only a minor reaction occurred with the two interferents, which are in an acceptable and distinguishable response range. The amperometric response of the GOx/SnS/GCE electrode demonstrates an acceptable selectivity to glucose. For enzymatic glucose sensor, it is reported that the retention of the original response may drop to less than 76% within seven days [24]. Furthermore, long-term stability was studied, and the results are provided in Figure 11. A stable and reproducible current over seven days was observed in our investigation. The presently commercialized glucose sensors are designed for blood samples. The users have to use lancet or syringe to collect blood from finger pricking or phlebotomizing. Noninvasive routes for glucose monitoring are highly expected to prevent these disadvantages. Thus, succedaneous body fluid samples such as urine, tear or saliva become prevailing targets for novel glucose sensor design. Proportional to blood glucose, however, they have relatively low glucose content [25]. A more sensitive glucose sensor with lower detection limit is required, and the developed sensor in this study will benefit. As to human blood, nevertheless, lower detection range needs fewer samples with proper dilute design on the system.  Figure 9 shows the sensitivity of the GOx/C-SnS/GCE sensor with respect to different glucose concentrations. According to the results of triplicates, the relative standard deviation (RSD) was less than 5.2% calculated from the current response of freshly prepared electrodes. The results convinced that the fabrication method was highly reproducible comparing with reported enzymatic and nonenzymatic glucose sensors [22,23]. A linear correlation, from 0.03 to 0.7 mM, can be found in glucose concentrations, and the sensitivity was 43.9 mA·M −1 ·cm −2 , roughly six times that found in Yang's work, using SnS2 [8]. Since human blood always contains different hormones and chemicals, which can interfere glucose sensor detection, an amperometric test was carried out using CA and UA, as depicted in Figure 10. Although the reactions are not clear, only a minor reaction occurred with the two interferents, which are in an acceptable and distinguishable response range. The amperometric response of the GOx/SnS/GCE electrode demonstrates an acceptable selectivity to glucose. For enzymatic glucose sensor, it is reported that the retention of the original response may drop to less than 76% within seven days [24]. Furthermore, long-term stability was studied, and the results are provided in Figure 11. A stable and reproducible current over seven days was observed in our investigation. The presently commercialized glucose sensors are designed for blood samples. The users have to use lancet or syringe to collect blood from finger pricking or phlebotomizing. Noninvasive routes for glucose monitoring are highly expected to prevent these disadvantages. Thus, succedaneous body fluid samples such as urine, tear or saliva become prevailing targets for novel glucose sensor design. Proportional to blood glucose, however, they have relatively low glucose content [25]. A more sensitive glucose sensor with lower detection limit is required, and the developed sensor in this study will benefit. As to human blood, nevertheless, lower detection range needs fewer samples with proper dilute design on the system.  Figure 9 shows the sensitivity of the GO x /C-SnS/GCE sensor with respect to different glucose concentrations. According to the results of triplicates, the relative standard deviation (RSD) was less than 5.2% calculated from the current response of freshly prepared electrodes. The results convinced that the fabrication method was highly reproducible comparing with reported enzymatic and non-enzymatic glucose sensors [22,23]. A linear correlation, from 0.03 to 0.7 mM, can be found in glucose concentrations, and the sensitivity was 43.9 mA·M −1 ·cm −2 , roughly six times that found in Yang's work, using SnS 2 [8]. Since human blood always contains different hormones and chemicals, which can interfere glucose sensor detection, an amperometric test was carried out using CA and UA, as depicted in Figure 10. Although the reactions are not clear, only a minor reaction occurred with the two interferents, which are in an acceptable and distinguishable response range. The amperometric response of the GO x /SnS/GCE electrode demonstrates an acceptable selectivity to glucose. For enzymatic glucose sensor, it is reported that the retention of the original response may drop to less than 76% within seven days [24]. Furthermore, long-term stability was studied, and the results are provided in Figure 11. A stable and reproducible current over seven days was observed in our investigation. The presently commercialized glucose sensors are designed for blood samples. The users have to use lancet or syringe to collect blood from finger pricking or phlebotomizing. Noninvasive routes for glucose monitoring are highly expected to prevent these disadvantages. Thus, succedaneous body fluid samples such as urine, tear or saliva become prevailing targets for novel glucose sensor design. Proportional to blood glucose, however, they have relatively low glucose content [25]. A more sensitive glucose sensor with lower detection limit is required, and the developed sensor in this study will benefit. As to human blood, nevertheless, lower detection range needs fewer samples with proper dilute design on the system.

Conclusions
In this work, we reported a C-SnS nanoparticle powder with a laminar structure for use in enzymatic glucose sensor applications via a simple two-step synthesis process, using hydrothermal and chemical vapor deposition methods. The preparation of the proposed sensor is easy and costefficient. The fast electron transformation rate (Ks = 7.46 s −1 ), high sensitivity (43.9 mA·M −1 ·cm −2 ), linear range from 0.03 to 0.7 mM glucose, and acceptable selectivity show promising development potentials for glucose sensing.

Conclusions
In this work, we reported a C-SnS nanoparticle powder with a laminar structure for use in enzymatic glucose sensor applications via a simple two-step synthesis process, using hydrothermal and chemical vapor deposition methods. The preparation of the proposed sensor is easy and costefficient. The fast electron transformation rate (Ks = 7.46 s −1 ), high sensitivity (43.9 mA·M −1 ·cm −2 ), linear range from 0.03 to 0.7 mM glucose, and acceptable selectivity show promising development potentials for glucose sensing.

Conclusions
In this work, we reported a C-SnS nanoparticle powder with a laminar structure for use in enzymatic glucose sensor applications via a simple two-step synthesis process, using hydrothermal and chemical vapor deposition methods. The preparation of the proposed sensor is easy and costefficient. The fast electron transformation rate (Ks = 7.46 s −1 ), high sensitivity (43.9 mA·M −1 ·cm −2 ), linear range from 0.03 to 0.7 mM glucose, and acceptable selectivity show promising development potentials for glucose sensing.

Conclusions
In this work, we reported a C-SnS nanoparticle powder with a laminar structure for use in enzymatic glucose sensor applications via a simple two-step synthesis process, using hydrothermal and chemical vapor deposition methods. The preparation of the proposed sensor is easy and cost-efficient. The fast electron transformation rate (Ks = 7.46 s −1 ), high sensitivity (43.9 mA·M −1 ·cm −2 ), linear range from 0.03 to 0.7 mM glucose, and acceptable selectivity show promising development potentials for glucose sensing.