Non-enzymatic glucose electrochemical sensor based on silver nanoparticle decorated organic functionalized multiwall carbon nanotubes

Abstract

An efficient, fast and stable non-enzymatic glucose sensor was prepared by decorating silver nanoparticles on organic functionalized multiwall carbon nanotubes (AgNPs/F-MWCNTs). MWCNTs were functionalized with organic amine chains and characterized using energy-dispersive X-ray and FT-IR spectroscopy. Moreover, the decorated AgNPs monitored using transmission electron microscopy showed spherical shapes with a mean size of 9.0 ± 2.8 nm. For further study, a glassy carbon electrode (GCE) was modified using the synthesized composite and evaluation of the modification was conducted using cyclic voltammetry and electrochemical impedance spectroscopy. The electrochemical data revealed that modification of the GCE leads to easier electron transfer compared to the bare unmodified GCE due to the presence of the functionalized MWCNTs accompanied with the electrocatalytic effect of the decorated silver nanoparticles. Furthermore, the fabricated modified electrode was applied as a non-enzymatic glucose sensor using electrochemical techniques including cyclic voltammetry and hydrodynamic chronoamperometry. The results obtained from the amperometric analysis of glucose in a 0.1 M NaOH solution indicated an efficient performance of the electrode with a low detection limit of 0.03 μM and a high sensitivity of 1057.3 μA mM⁻¹, as well as a linear dynamic range of 1.3 to 1000 μM. A practical application of this sensor was also examined by analyzing glucose in the presence of common interfering species that exist in a real sample of human blood serum.

---

1. Introduction

Diabetes and its serious health complications have a main position in medicinal sciences, as a principal humanity challenge. Thus, each influential factor, such as the levels of insulin and glucose in human blood, should be investigated.¹ Insulin has been determined using various analytical methods, especially electrochemistry.² Moreover, since the level of glucose in blood should be critically regulated, usually in the range of 3.0–8.0 mM,³ different approaches have been proposed to measure it, including spectroscopy,⁴ spectrofluorimetry,⁵ electrochemistry,³ and chromatography.⁶ Among these methods, electrochemical sensors have been developed based on enzymatic and non-enzymatic applications.⁷ Although enzyme-based sensors have presented some advantages such as high selectivity and a low detection limit, some drawbacks including poor reproducibility, low thermal and chemical stability, and high cost have been reported for these sensors.^8–10 Hence, enzyme-free electrochemical sensors have been developed as a result of electrode modification. Metal nanomaterials, because of their high surface areas and electrocatalytic properties, have been used as the modifier for glucose detection.^11–14 Moreover, these properties can be improved by stabilizing the nanoparticles on an appropriate support. Multiwall carbon nanotubes (MWCNTs), due to their excellent conductivity, good chemical stability, large surface-volume ratio, and high adsorption capacity, have been utilized as a nanoparticle support.^15,16 However, nanoparticles decorated on a bare MWCNTs support can be removed from it, after repeated usage, because of the absence of strong interactions between them, whereas covalent functionalization of the MWCNTs with organic ligands can result in a more effective loading of the nanoparticles.^16–18

Herein, silver nanoparticles (AgNPs) decorated on newly synthesized and functionalized MWCNTs were employed as a new non-enzymatic sensor to detect glucose. The modifier was synthesized based on step by step organic bonding to form functionalized MWCNTs. The synthesis was verified using Fourier transform infrared (FT-IR) spectroscopy, transmission electron microscopy (TEM), cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Moreover, the sensing performance of the synthesized modifier was studied using hydrodynamic chronoamperometry. The resulting data showed high sensitivity and selectivity accompanied with a fast analysis of glucose. At last, in this work, the reliability of the sensor for real sample analysis was examined using human blood serum samples.

2. Experimental

2.1. Reagents

Silver nitrate (AgNO₃), glucose, sodium hydroxide (NaOH), phosphoric acid (H₃PO₄) and sucrose were purchased from Merck. Ascorbic acid, uric acid, fructose and the multiwall carbon nanotubes (MWCNTs) were supplied by Sigma-Aldrich. Phosphate buffer solutions (PBS, 0.1 M) at various pH values were prepared by the addition of 0.10 M NaOH to a 0.10 M H₃PO₄ solution. All the other chemicals were of analytical reagent grade and used without further purification. All of the water used for preparing the solutions in this work was double distilled water. Three human blood plasma samples were provided by Alzahra Hospital (Isfahan, Iran). The proteins contained in the plasma were removed by filtration as in the pretreatment of real samples.

2.2. Apparatus

The products obtained from each step of the synthesis were monitored by recording the FT-IR spectra, as KBr-mixed pellets, using a FT-IR Spectrometer (Jasco, FT/IR-680 Plus). The morphology and size of the generated AgNPs decorated on the functionalized MWCNTs were analyzed using a transmission electron microscope (TEM, TECNAI, Model F30, USA). Energy dispersive X-ray spectroscopy (EDX) spectra were recorded with a Philips XLC instrument.

All the electrochemical experiments were performed using a common three-electrode system (Autolab, PGSTAT-30 potentiostat/galvanostat, Eco-Chemie, Netherlands). The modified glassy carbon electrode (GCE), a Pt rod and Ag/AgCl/3.0 M KCl were used as the working, auxiliary and reference electrodes, respectively. The recorded data were processed by applying a General Purpose Electrochemical System (GPES, version 4.9) and Frequency Response Analyzer (FRA). All the electrochemical experiments were conducted after stabilizing the surface of the electrode through the operation of 20 cycles in the potential window of −1.00 to +1.00 (V vs. Ag/AgCl). Cyclic voltammograms were recorded using solutions of PBS and 0.1 M NaOH at various pH values with a scan rate of 100 mV s⁻¹, for which the results were optimal in the alkaline medium of NaOH. Thus, the amperometric measurements were conducted at a hydrodynamic electrode by the sequential addition of a glucose solution to a 0.1 M NaOH solution. The electron transfer resistances of the electrodes were measured using electrochemical impedance spectroscopy (EIS). All the EIS studies were conducted with a frequency range of 0.01 Hz to 100 kHz, an amplitude wave potential of 10 mV and with 0.20 V as the applied potential in a 10.0 mM Fe(CN)₆^3−/4− solution. All these electrochemical measurements were carried out at ambient temperature.

2.3. Procedure

The synthesis of the functionalized multiwall carbon nanotubes (F-MWCNTs) is schematically summarized in Fig. 1. The MWCNTs were functionalized by refluxing in a mixture of HNO₃ (6.0 M) and H₂SO₄ (2.0 M) for 12 h. Afterwards, the functionalized MWCNTs (CNT-COOH) were continually and sufficiently rinsed with distilled water, and then dried under vacuum conditions for 12 h. 500 mg of the MWCNTs-COOH was thoroughly dispersed in THF in an ultrasonic bath. After that, 12.0 mL of SOCl₂ was added to the former suspension and stirred at room temperature for 24 h. The product, MWCNT-COCl, was refluxed with 15.0 mL of ethylenediamine at 60 °C for 12 h. The resulting mixture was reacted with 1.0 g of cyanuric chloride and then 5.0 mL of diethylenetriamine in THF under a nitrogen atmosphere to form MWCNT-CO-NH-cyanuric and MWCNT-CO-NH-cyanuric-NH₂, respectively. At last, the F-MWCNTs were prepared by centrifuging and drying MWCNT-CO-NH-cyanuric-NH₂.

Fig. 1 The schematic synthetic steps for the organic chain functionalized multiwall carbon nanotubes (F-MWCNTs).

0.10 g of the F-MWCNTs was dispersed in 200 mL of distilled water using ultrasonic waves for 30 min. To form Ag(I)/F-MWCNTs, 10.0 mL of a AgNO₃ solution (1.0 mM) was added drop by drop to the F-MWCNTs suspension over 24 h. The AgNPs/F-MWCNTs were prepared by reducing the silver ions to silver nanoparticles (AgNPs) using NaBH₄.

3. Results and discussion

3.1. Characterization of the synthesized composite

The synthesis of AgNPs/F-MWCNT was verified using several techniques including energy-dispersive X-ray spectroscopy (EDX), Fourier transform infrared spectroscopy (FT-IR), and transmission electron microscopy (TEM). EDX analysis indicated that the F-MWCNTs were prepared with weight percentages of 79.74, 9.68, and 10.58 corresponding to carbon, oxygen, and nitrogen. It confirmed that the MWCNTs were satisfactorily functionalized with the organic compounds. FT-IR spectra exhibit the products from the steps of the organic compound synthesis, as can be seen in Fig. 2A. The spectrum (a) of the MWCNTs shows C–C stretching bands in the range of 1580–1650 cm⁻¹ and the peak located at around 850 cm⁻¹ is due to the nanotube symmetrical modes.¹⁹ The band at 1640 cm⁻¹ can be associated with –C=O stretching of the carboxyl group. As shown in the spectra (b and c), this characteristic band was moved to 1690 and 1630 cm⁻¹ by conversion of COOH to COCl and an amide group, respectively. As shown in Fig. 2B, the silver nanoparticles (AgNPs) fabricated on the F-MWCNTs presented a size distribution of 3.0 to 17.0 nm with a mean size of around 9.0 ± 2.8 nm.

Fig. 2 (A) FT-IR spectra of the products formed in each step of the F-MWCNTs synthesis (CNT-COOH (a), CNT-COCl (b), and CNT-CO-NH-cyanuric-NH₂ (c)); (B) TEM image of AgNPs/F-MWCNTs and size distribution of the generated nanoparticles.

3.2. Electrochemical characterization

The AgNPs/F-MWCNTs were further characterized using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). In a comparison of the unmodified glassy carbon electrode (GCE), F-MWCNTs-GCE, and AgNPs/F-MWCNTs-GCE, the modified electrodes showed a capacity for electron transfer as indicated in Fig. 3(A and B). Accordingly, the cyclic voltammograms recorded in 0.1 M NaOH with a scan rate of 50 mV s⁻¹, as exhibited in Fig. 3A, indicate a corresponding electrochemical oxidation that generates a surface Ag-oxide layer on the nanoparticles, i.e. Ag to Ag^(I) and Ag^(I) to Ag^(II).²⁰ To further confirm the effect of the modifier, further investigations were conducted using EIS analysis in a 10.0 mM Fe(CN)₆^3−/4− solution. Using the Nyquist plots (Fig. 3B), electrical conductivity between the redox probe and the modified electrode was confirmed and it was confirmed that the electron transfer was increasingly improved by the modification of the GCE, using F-MWCNTs and AgNPs/F-MWCNTs, as the semicircle that corresponds to the electron transfer-limited process was decreased. In other words, the charge transfer resistance (R_ct) at the surface of the electrode was decreased because of the good conductivity of AgNPs/F-MWCNTs, which could make electron transfer easier.

Fig. 3 (A) Cyclic voltammograms of the GCE (a), F-MWCNTs-GCE (b), and AgNPs/F-MWCNTs-GCE (c) in 0.1 M NaOH with a scan rate of 50 mV s⁻¹; (B) Nyquist plots of the GCE (a), F-MWCNTs-GCE (b), and AgNPs/F-MWCNTs-GCE (c) in 10.0 mM Fe(CN)₆^3−/4− solution with a frequency range of 0.01 Hz to 100 kHz and an amplitude wave potential of 10 mV.

3.3. Electroanalysis of glucose at AgNPs/F-MWCNTs-GCE

Analysis of glucose was studied using the AgNPs/MWCNTs-GCE and compared with the non-AgNP decorated F-MWCNTs and bare GCEs, as shown in Fig. 4. The electrocatalytic effect of the AgNPs/MWCNTs resulted in a significant increase in the oxidation currents in the presence of glucose, compared to the results obtained for the two other unmodified electrodes. Thus, the further investigations for glucose sensing were performed with the AgNPs/MWCNTs-GCE. Fig. 5 shows that the CVs in the presence of 5.0 mM glucose at different scan rates (10 to 300 mV s⁻¹) had a linear relationship between the oxidation current (in the range of 0.60 to 0.65 V, depending on the scan rate) and the square root of the scan rate (i.e. v^1/2). This proportional linearity, with a correlation coefficient of 0.995, shows that the mass transfer of glucose at the surface of the electrode was controlled by diffusion, which is perfect for quantitative sensing objectives.

Fig. 4 Cyclic voltammograms of the GCE (A), F-MWCNTs-GCE (B), and AgNPs/F-MWCNTs-GCE (C) in 0.1 M NaOH solution in the absence of glucose (a) and in the presence of a 2.5 mM glucose solution at a scan rate of 50 mV s⁻¹ (b).

Fig. 5 The study of the AgNPs/F-MWCNTs-GCE in a 0.1 M NaOH solution and in the presence of 5.0 mM glucose at a potential scan rate of 10–300 mV s⁻¹. Inset: plot of the oxidation peaks vs. the square root of the scan rates.

3.4. Chronoamperometric study of AgNPs/F-MWCNTs-GCE in the presence of glucose

Chronoamperometric responses of the AgNPs/F-MWCNTs-GCE with a working potential of 0.58 V in a 0.1 M NaOH solution for sequential additions of glucose are displayed in Fig. 6. Further amperometric investigations were performed at 0.58 V due to the lower potential values presenting lower increases in the amperometric responses and as the signal obtained at higher potentials could be affected by interfering species, as well as being less stable to changes in the glucose concentration, so, 0.58 V was selected as the optimum potential. Moreover, the signal obtained from the glucose injection using 0.58 V quickly became stable with a response time of less than 3 s. According to Fig. 6, two linear ranges of calibration were observed: at a lower glucose concentration range from 1.3 to 1000 μM, with a regression of I (μA) = 31.720C (mM) − 0.245 and R² = 0.998; and at a higher glucose concentration range from 1.11 to 4.14 mM, with a regression of I (μA) = 3.144C (mM) + 30.370 and R² = 0.955. The two linearities may result from the adsorption of an intermediate.²¹ In other words, at a low glucose concentration, the amperometric responses resulted from the oxidation of diffused glucose, and with higher amounts of glucose the slope of the calibration curve was decreased due to adsorption of the oxidation products of glucose, which decreased the number of active sites of the AgNPs/F-MWCNTs and hindered the glucose diffusion into the electrode surface.

Fig. 6 Current–time study of the AgNPs/F-MWCNTs-GCE vs. glucose addition in a 0.1 M NaOH solution using a hydrodynamic electrode at an applied potential of 0.58 V vs. Ag/AgCl and 1200 rpm, with the corresponding two linear ranges of the calibration curve shown in the inset.

The limit of detection (LOD) was calculated using the signal to noise ratio of three. The sensitivity of the modified electrode in relation to the glucose concentration was obtained using the ratio of the calibration slope to the standard deviation of the current steps.²² Accordingly, using the AgNPs/F-MWCNTs-GCE, the LOD and sensitivity for glucose determination were obtained as 0.03 μM and 1057.3 μA mM⁻¹, respectively. The analytical figures of merit for this sensor are comparable to and even better than those obtained using other reported non-enzymatic sensors, as tabulated in Table 1.

Table 1 Performance of the various reported functionalized MWCNT nonenzymatic modified electrodes for glucose detection

Sensor	Response time (s)	Applied potential (V)	Linear range (μM)	Detection limit (μM)	Literature
a Cu nanowires.b Poly(2-aminothiophenol).
CuO/MWCNTs	3	0.55	4–5000	4.0	23
CuFe2O4/MWCNTs	5	0.40	0.5–1400	0.2	24
RuO2/MWCNTs	—	0.50	500–50 000	33	25
Pt-PbNPs/MWCNTs	12	0.30	Up to 11 000	1.8	26
Cu/MnO2/MWCNTs	3	0.60	10–1000	0.17	16
CuNWs^a/MWCNTs	1	0.55	Up to 3000	0.26	27
PdNPs/MWCNTs	3	0.025	1000–10 000	—	28
MWCNTs-COOH-P2AT^b-AuNPs	—	—	100–3000	3.7	29
Fe3O4/MWCNTs	12	0.50	500–7000	15.0	30
AgNPs/F-MWCNTs	3	0.58	1.3–1000 and 1100–4140	0.03	This work

---

3.5. Repeatability, stability and selectivity of the glucose sensor

To evaluate the performance of the AgNPs/F-MWCNTs-GCE in glucose sensing, ten continuous additions of a 0.2 mM glucose solution were performed and this provided a relative standard deviation (RSD%) of 2.2% as can be seen in Fig. 7A. The addition of a 0.70 mM glucose solution into a 0.1 M NaOH solution at the surface of the modified electrode provided a relatively stable signal up to 1000 s with only a 6.3% decrease in the amperometric signal as shown in Fig. 7B. The selectivity of the glucose sensor was examined in the presence of different interfering species including sucrose (Su), fructose (Fr), uric acid (UA), ascorbic acid (AA) and dopamine (DA). The physiological level of glucose in human blood plasma is within 3 to 8 mM, while the other interfering species including UA, AA, and DA exist at a concentration of 0.1 mM, i.e. one 30^th of the glucose concentration.³ Moreover, the existence of other carbohydrates, such as sucrose and fructose, can affect the performance of the AgNPs/F-MWCNTs-GCE. As shown in Fig. 7C, the influence of the presence of interfering compounds, including ascorbic acid (AA) (0.07 mM), dopamine (DA) (0.07 mM), uric acid (UA) (0.07 mM), sucrose (Su) (0.70 mM) and fructose (Fr) (0.70 mM), on the amperometric response for a 0.70 mM glucose solution at the AgNPs/F-MWCNTs-GCE was investigated. The interference study showed that the amperometric signals were insignificantly affected by the two carbohydrates sucrose and fructose, while an easy oxidation of the other three compounds in alkaline media had a relatively interfering effect on the signals. As mentioned, since the levels of the three interfering species of ascorbic acid, uric acid and dopamine are normally much lower than the glucose concentration, the low interfering effects of them can be resolved by diluting the sample to the real concentrations of the species in human blood plasma, i.e. one thirtieth.³

Fig. 7 Chronoamperometric study of the glucose oxidation current at the AgNPs/F-MWCNTs-GCE in 0.1 M NaOH solution, based on 10 successive additions of a glucose solution (0.20 mM) (A); stability of the sensor signal up to 1000 s with addition of a glucose solution (0.70 mM) (B); and selectivity of the glucose sensor towards the glucose solution (0.70 mM) in the presence of interfering species including sucrose (Su) (0.70 mM), fructose (Fr) (0.70 mM), uric acid (UA) (0.07 mM), ascorbic acid (AA) (0.07 mM), and dopamine (DA) (0.07 mM) (C).

3.6. Real sample analysis

To verify practical application of the AgNPs/F-MWCNTs-GCE as a non-enzymatic glucose sensor, a series of protein-filtered human blood serum samples was examined. The detection of glucose was conducted using a standard addition method. In this case, the serum samples were diluted with a 0.1 M NaOH solution until the glucose content was in the range of 61.0–103.0 μM, within the linear range of the calibration. The resulting data from amperometric analyses of the real samples are summarized in Table 2. These results were satisfactorily comparable with the ones obtained using the hospital’s glucose analyzer.

Table 2 Determination of the glucose level in real samples using the AgNPs/F-MWCNTs-GCE non-enzymatic sensor

Sample	Reference values^a (μM)	Determined values (μM)	RSD (%)
a The values provided by the hospital with a relative population standard deviation of 5%.
Human serum 1	61.4	62.9	2.4
Human serum 2	112.2	110.6	3.8
Human serum 3	85.7	87.3	3.1

---

4. Conclusion

In summary, the applicability of a non-enzymatic glucose sensor based on the decoration of silver nanoparticles on organic functionalized multiwall carbon nanotubes (AgNPs/F-MWCNTs) was investigated. The sensor was characterized using energy-dispersive X-ray spectroscopy, FT-IR spectroscopy, microscopic images from TEM, cyclic voltammetry, electrochemical impedance spectroscopy, and chronoamperometry. According to the impedance spectra, the performance of a GCE was successfully improved by modifying the electrode with the AgNPs/F-MWCNTs, as a decreasing of the electron transfer resistance is related to a shortening of the semicircle diameter. The electrode was employed as a non-enzymatic sensor to measure the glucose level in biological samples, using hydrodynamic chronoamperometry. As an interference study, five common interfering compounds, ascorbic acid, uric acid, dopamine, sucrose, and fructose, were examined in the presence of glucose. These species at biological concentration levels presented no significant interference effects on the amperometric response of glucose oxidation. At last, the reliability of the sensor was verified by analyzing the glucose level in blood serum samples.

Acknowledgements

The authors wish to thank the Isfahan University of Technology (IUT) Research Council and Center of Excellence in Sensor and Green Chemistry for their support. The authors would like to thank Al-Zahra Hospital (Isfahan, Iran) for the preparation of the real samples.

References

1. C. D. A. Stehouwer, I. Ferreira, M. Kozakova and C. Palombo, Glucose Metabolism, Diabetes, and the Arterial Wall, in Early Vascular Aging (EVA): New Directions in Cardiovascular Protection, ed. P. M. Nilsson, M. H. Olsen and S. Laurent, Academic Press, Boston, 1^st edn, 2015, pp. 147–156.
2. A. Arvinte, A. C. Westermann, A. M. Sesay and V. Virtanen, Sens. Actuators, B, 2010, 150, 756.
3. A. A. Ensafi, M. Mokhtari Abarghoui and B. Rezaei, Electrochim. Acta, 2014, 123, 219.
4. M. Goodarzi and W. Saeys, Talanta, 2016, 146, 155.
5. M. Aloraefy, T. J. Pfefer, J. C. Ramella-Roman and K. E. Sapsford, Sensor, 2014, 14, 12127.
6. M. Grembecka, A. Lebiedzińska and P. Szefer, Microchem. J., 2014, 117, 77.
7. S. A. Zaidi and J. H. Shin, Talanta, 2016, 149, 30.
8. J. Lin, C. He, Y. Zhao and S. Zhang, Sens. Actuators, B, 2009, 137, 768.
9. X. Kang, Z. Mai, X. Zou, P. Cai and J. Mo, Anal. Biochem., 2007, 363, 143.
10. X. Li, Q. Zhu, S. Tong, W. Wang and W. Song, Sens. Actuators, B, 2009, 136, 444.
11. J. Chen, Q. Sheng and J. Zheng, RSC Adv., 2015, 5, 105372.
12. J. Yang, X. Liang, L. Cui, H. Liu, J. Xie and W. Liu, Biosens. Bioelectron., 2016, 80, 171.
13. Y. Liu, X. Zhang, D. He, F. Maa, Q. Fua and Y. Hu, RSC Adv., 2016, 6, 18654.
14. X. Luo, Z. Zhang, Q. Wan, K. Wu and N. Yang, Electrochem. Commun., 2015, 61, 89.
15. U. Yaqoob, A. S. M. I. Uddin and G.-S. Chung, Sens. Actuators, B, 2016, 224, 738.
16. Y. Wang, S. Zhang, W. Bai and J. Zheng, Talanta, 2016, 149, 211.
17. H. Sadegh and R. Shahryari-ghoshekandi, Nanomed. J., 2015, 2, 231.
18. R. Devasenathipathy, C. Karuppiah, S.-M. Chen, S. Palanisamy, B.-S. Lou, M. A. Ali and F. M. A. Al-Hemaid, RSC Adv., 2015, 5, 26762.
19. T. Belin and F. Epron, Mater. Sci. Eng., B, 2005, 119, 105.
20. N. Zandi-Atashbar, B. Hemmateenejad and M. Akhond, Analyst, 2011, 136, 1760.
21. L. Qian, J. Mao, X. Tian, H. Yuan and D. Xiao, Sens. Actuators, B, 2013, 176, 952.
22. A. A. Ensafi, N. Zandi-Atashbar, M. Ghiaci, M. Taghizadeh and B. Rezaei, Mater. Sci. Eng., C, 2015, 47, 290.
23. X.-W. Liu, P. Pan, Z.-M. Zhang, F. Guo, Z.-C. Yang, J. Wei and Z. Wei, J. Electroanal. Chem., 2016, 763, 37.
24. Y. Zhang, E. Zhou, Y. Li and X. He, Anal. Methods, 2015, 7, 2360.
25. R. M. A. Tehrani and S. A. Ghani, Biosens. Bioelectron., 2012, 38, 278.
26. H. F. Cui, J. S. Ye, W. D. Zhang, C. M. Li, J. H. T. Luong and F. S. Sheu, Anal. Chim. Acta, 2007, 594, 175.
27. J. Huang, Z. Dong, Y. Li, J. Li, J. Wang, H. Yang, S. Li, S. Guo, J. Jin and R. Li, Sens. Actuators, B, 2013, 182, 618.
28. Z.-X. Cai, C.-C. Liu, G.-H. Wu, X.-M. Chen and X. I. Chen, Electrochim. Acta, 2013, 112756.
29. R. Sedghi and Z. Pezeshkian, Sens. Actuators, B, 2015, 219, 119.
30. S. Masoomi-Godarzi, A. A. Khodadadi, M. Vesali-Naseh and Y. Mortazavi, J. Electrochem. Soc., 2014, 161, B19.

---