Improving Linear Range Limitation of Non-Enzymatic Glucose Sensor by OH − Concentration

: The linear range of the non-enzymatic glucose sensor is usually much smaller than the glucose level of diabetic patients, calling for an e ﬀ ective solution. Despite many previous attempts, none have solved the problem. Such a challenge has now been conquered by raising the NaOH concentration in the electrolyte, where amperometry, X-ray di ﬀ raction, Fourier-transform infrared spectroscopy, and Nuclear magnetic resonance measurements have been conducted. The linear range has been successfully enhanced to 40 mM in 1000 mM NaOH solution, and it was also found that NaOH a ﬀ ected the degree of glucose oxidation, which inﬂuenced the current response during sensing. It was expected that the alkaline concentration must be 25 times higher than the glucose concentration to enhance the linear range, much contrary to prior understanding. (LOD) were estimated to be 7.15 μ M, 2.44 μ M, and 13.19 μ M, respectively. Besides, it was found that the linear range was successfully increased to 40 mM in 1000 mM NaOH solution. More importantly, the ratio of the highest point of the linear range and OH − concentration was fixed at around 25, which meant the linear range could be enhanced by raising the OH − concentration.


Introduction
As a long-term metabolic disease, diabetes may lead to serious complications including cardiovascular diseases, hearing impairment, nerve damage, and kidney damage, etc. [1]. According to the World Health Organization (WHO), there are more than 420 million diabetic patients globally, which causes over one million deaths every year [2,3]. To monitor the glucose level, patients have to detect their blood glucose constantly by invasive enzyme test-strips, which dominate the market currently [4]. Although the result of enzymatic test-strips is accurate; they have many limitations, the most critical of which is their requirement on humidity and temperature [4][5][6][7]. This is caused by the presence of enzyme, which inspired the researchers to focus on non-enzymatic glucose sensing, due to the advantage of easy operation, fast response and low cost [8][9][10][11].
While a non-enzymatic glucose sensor demonstrates many benefits; their measurement range is limited to up to several mmol/L(mM), which is much smaller than the blood glucose level of diabetic patients (1-30 mM) [12,13]. During glucose sensing, the current response first enters a linear range where a linear relationship between the current response and glucose concentration can be expected [14,15]. Next, a non-linear relationship is observed, dubbed non-linear range, during which the relationship between the current response and glucose concentration is unpredictable. Eventually, the current reaches saturation point, beyond which the current response starts to decrease. During the linear range, the glucose level could easily be extracted by the linear relationship and the saturation point is the highest glucose level that the sensor could detect. Although the current could increase along with the glucose level in the range between the linear range and the saturation, i.e., non-linear range, such a range is not useful as the saturation is not fixed. Therefore, the glucose concentration can only be extracted by the current response in the linear range, which leaves the challenge of raising the linear range [14].
A huge number of techniques have been attempted to raise the linear range. Some researchers believed that the linear range was limited by the working potential [16]. By changing the ratio of Pt, Au and Pd metal elements, the working potential was adjusted from 0.45 V vs. Saturated calomel electrode (SCE) to 0.4 V vs. SCE, but the linear range of the measurement was basically unchanged (from 1.5 µM-8 mM to 1µM-8.5 mM) [17]. On the other hand, the others thought this problem was related to the specific surface area of the electrode [18][19][20][21][22]. Mian Li et al. synthesized an electrode of 3D nanostructure with a large specific surface area; but the linear range (50 µM-8 mM) didn't increase to any appreciable level [23]. This indicated that the reason that limited the linear range had to be uncovered in order to find an effective strategy, as all the previous blind trials could not solve this problem.
It was, therefore, the purposes of the current paper to find out what restricted the linear range and raise this range to the needed level. Employing amperometry, infrared spectroscopy, and nuclear magnetic resonance spectroscopy; the influence of hydroxide ion concentration was found, which had to be 25 times higher than that of the glucose. This high OH − concentration enabled the formation of enediol, which lowered the energy barrier of the subsequent electro-oxidation reaction so that more than four electrons were transferred to the cathode. Meanwhile, the final product was glucuronate, instead of gluconolactone. This increased the current response and raised the linear range to 40 mM. The real cause of the limited linear range revealed by us could not only advance the conceptual understanding of the non-enzymatic glucose sensor working mechanism, but also differentiated various hypotheses and conflict. Hopefully, these discoveries are able to improve the sensor performance, and accelerate its commercialization.
Ni(OH) 2 powders were synthesized by the method proposed in the literature [24]. Concentrated sodium hydroxide solution was dropped into a 1-M Ni(NO 3 ) 2 solution. The precipitation reaction occurred under 700 rpm stirring at 45 • C. The green precipitate of Ni (OH) 2 powder was filtered, dried, and washed with DI water several times to remove any impurities. The precipitate was next dried in air. Subsequently, a homogeneous suspension liquid of Ni(OH) 2 sample was obtained by sonicating the powder in several drops of 95% ethanol. Next, the suspension liquid was dropped on the Ni sheet as an electrode, after which 0.5% Nafion binder solution was dropped on the sample layer to glue the material. Finally, the electrode was air-dried before the measurements.
The electrode was characterized by powder XRD (Bruker D8 DISCOVER diffractometer, Bruker Corporation, Billerica, MA, USA). The three-electrode electrochemical cell was used to monitor glucose levels. Ag/AgCl (saturated KCl solution) was used as a reference electrode, although some reference electrodes would be affected by alkaline electrolyte such as Saturated calomel electrode (SCE); Ag/AgCl reference electrode is stable in alkaline electrolyte [25]. Ni(OH) 2 /Ni sheet was the working electrode, and Au wire was the counter electrode. All the measurements were carried out on a CMS100 electrochemical workstation (purchased from The Illinois Department of Central Management Services, Springfield, IL, USA) at room temperature. Amperometry I-t responses at an applied potential of +0.55 V vs. Ag/AgCl reference electrode (for 0.1 mM NaOH solution) and +0.5 V (for 0.5 M and 1 M NaOH solution) were recorded under stirring at 350 rpm. The area of Ni (OH) 2 on the Ni sheet was 1.2 × 0.15 cm 2 . Nicolet 6700 FT-IR spectrometer (Thermo Fisher, Waltham, MA, USA) was used to acquire the FTIR spectra in the wavenumber range of 400-4000 cm −1 . 13 C NMR was recorded on Bruker DRX-500 MHz spectrometers (Bruker Corporation, Billerica, MA, USA) with a frequency of 125.78 MHz at room temperature, analyzed by Bruker Topspin 3.1 (Bruker Corporation, Billerica, MA, USA).

Results and Discussion
The working electrode was characterized first by XRD, which confirmed the formation of Ni(OH) 2 on the Ni sheet, as demonstrated in Figure 1.

X-ray Diffraction (XRD) Measurement of Ni(OH)2/Ni Sheet
The working electrode was characterized first by XRD, which confirmed the formation of Ni(OH)2 on the Ni sheet, as demonstrated in Figure 1.

Electrooxidation of Glucose on Ni(OH)2/Ni Sheet
To investigate the influence of the alkaline concentration on the linear range, I-t measurements were conducted. In Figure 2a-c, a typical amperometric response of the Ni(OH)2/Ni sheet upon the successive addition of a certain concentration of glucose into 0.1 mM, 500 mM and 1000 mM NaOH solution stirred at 350 rpm is shown. Figure 2d-f shows the calibration curve of the Ni(OH)2/Ni sheet electrode, revealing a good linear detection ranging from 21.67 μM to 4 mM (R 2 = 0.9905) in 100 mM NaOH solution, from 7.40 μM to 20 mM (R 2 = 0.9955) in 500 mM NaOH solution, and from 39.98 μM to 40 mM (R 2 = 0.9949) in 1000 mM NaOH solution. By a signal-to-noise ratio of 3, the detection limits (LOD) were estimated to be 7.15 μM, 2.44 μM, and 13.19 μM, respectively. Besides, it was found that the linear range was successfully increased to 40 mM in 1000 mM NaOH solution. More importantly, the ratio of the highest point of the linear range and OH − concentration was fixed at around 25, which meant the linear range could be enhanced by raising the OH − concentration.

X-ray Diffraction (XRD) Measurement of Ni(OH)2/Ni Sheet
The working electrode was characterized first by XRD, which confirmed the formation of Ni(OH)2 on the Ni sheet, as demonstrated in Figure 1.

Electrooxidation of Glucose on Ni(OH)2/Ni Sheet
To investigate the influence of the alkaline concentration on the linear range, I-t measurements were conducted. In Figure 2a

X-ray Diffraction (XRD) Measurement of Ni(OH)2/Ni Sheet
The working electrode was characterized first by XRD, which confirmed the formation of Ni(OH)2 on the Ni sheet, as demonstrated in Figure 1.

Electrooxidation of Glucose on Ni(OH)2/Ni Sheet
To investigate the influence of the alkaline concentration on the linear range, I-t measurements were conducted. In Figure 2a-c, a typical amperometric response of the Ni(OH)2/Ni sheet upon the successive addition of a certain concentration of glucose into 0.1 mM, 500 mM and 1000 mM NaOH solution stirred at 350 rpm is shown. Figure 2d-f shows the calibration curve of the Ni(OH)2/Ni sheet electrode, revealing a good linear detection ranging from 21.67 μM to 4 mM (R 2 = 0.9905) in 100 mM NaOH solution, from 7.40 μM to 20 mM (R 2 = 0.9955) in 500 mM NaOH solution, and from 39.98 μM to 40 mM (R 2 = 0.9949) in 1000 mM NaOH solution. By a signal-to-noise ratio of 3, the detection limits (LOD) were estimated to be 7.15 μM, 2.44 μM, and 13.19 μM, respectively. Besides, it was found that the linear range was successfully increased to 40 mM in 1000 mM NaOH solution. More importantly, the ratio of the highest point of the linear range and OH − concentration was fixed at around 25, which meant the linear range could be enhanced by raising the OH − concentration.

X-ray Diffraction (XRD) Measurement of Ni(OH)2/Ni Sheet
The working electrode was characterized first by XRD, which confirmed the formation of Ni(OH)2 on the Ni sheet, as demonstrated in Figure 1.

Electrooxidation of Glucose on Ni(OH)2/Ni Sheet
To investigate the influence of the alkaline concentration on the linear range, I-t measurements were conducted. In Figure 2a-c, a typical amperometric response of the Ni(OH)2/Ni sheet upon the successive addition of a certain concentration of glucose into 0.1 mM, 500 mM and 1000 mM NaOH solution stirred at 350 rpm is shown. Figure 2d-f shows the calibration curve of the Ni(OH)2/Ni sheet electrode, revealing a good linear detection ranging from 21.67 μM to 4 mM (R 2 = 0.9905) in 100 mM NaOH solution, from 7.40 μM to 20 mM (R 2 = 0.9955) in 500 mM NaOH solution, and from 39.98 μM to 40 mM (R 2 = 0.9949) in 1000 mM NaOH solution. By a signal-to-noise ratio of 3, the detection limits (LOD) were estimated to be 7.15 μM, 2.44 μM, and 13.19 μM, respectively. Besides, it was found that the linear range was successfully increased to 40 mM in 1000 mM NaOH solution. More importantly, the ratio of the highest point of the linear range and OH − concentration was fixed at around 25, which meant the linear range could be enhanced by raising the OH − concentration.

Electrooxidation of Glucose on Ni(OH) 2 /Ni Sheet
To investigate the influence of the alkaline concentration on the linear range, I-t measurements were conducted. In Figure 2a-c, a typical amperometric response of the Ni(OH) 2 /Ni sheet upon the successive addition of a certain concentration of glucose into 0.1 mM, 500 mM and 1000 mM NaOH solution stirred at 350 rpm is shown. Figure 2d-f shows the calibration curve of the Ni(OH) 2 /Ni sheet electrode, revealing a good linear detection ranging from 21.67 µM to 4 mM (R 2 = 0.9905) in 100 mM NaOH solution, from 7.40 µM to 20 mM (R 2 = 0.9955) in 500 mM NaOH solution, and from 39.98 µM to 40 mM (R 2 = 0.9949) in 1000 mM NaOH solution. By a signal-to-noise ratio of 3, the detection limits (LOD) were estimated to be 7.15 µM, 2.44 µM, and 13.19 µM, respectively. Besides, it was found that the linear range was successfully increased to 40 mM in 1000 mM NaOH solution. More importantly, the ratio of the highest point of the linear range and OH − concentration was fixed at around 25, which meant the linear range could be enhanced by raising the OH − concentration. Clearly, OH − affected the glucose electro-oxidation reaction, which needed further investigation by FTIR measurements. The pure glucose electro-oxidized product, glucose/NaOH mixture, and glucose/NaOH electro-oxidized mixture were measured by FTIR. As shown in Figure 3a, the peak of C=C double bond was detected in the glucose/NaOH mixture at 1670 cm −1 [26], which indicated the formation of enediol [27]. This means glucose reacted with NaOH before the electro-oxidation. In addition, the disappearance of the peak at 1080 cm −1 and 1350cm −1 in glucose/NaOH electro-oxidized mixture in Figure 3b proved that the electro-oxidation product was not gluconate, which indicated that more than four electrons were transferred to the cathode [28][29][30][31]. This was beneficial for the glucose sensing.  Clearly, OH − affected the glucose electro-oxidation reaction, which needed further investigation by FTIR measurements. The pure glucose electro-oxidized product, glucose/NaOH mixture, and glucose/NaOH electro-oxidized mixture were measured by FTIR. As shown in Figure 3a, the peak of C=C double bond was detected in the glucose/NaOH mixture at 1670 cm −1 [26], which indicated the formation of enediol [27]. This means glucose reacted with NaOH before the electro-oxidation. In addition, the disappearance of the peak at 1080 cm −1 and 1350cm −1 in glucose/NaOH electro-oxidized mixture in Figure 3b proved that the electro-oxidation product was not gluconate, which indicated that more than four electrons were transferred to the cathode [28][29][30][31]. This was beneficial for the glucose sensing. Clearly, OH − affected the glucose electro-oxidation reaction, which needed further investigation by FTIR measurements. The pure glucose electro-oxidized product, glucose/NaOH mixture, and glucose/NaOH electro-oxidized mixture were measured by FTIR. As shown in Figure 3a, the peak of C=C double bond was detected in the glucose/NaOH mixture at 1670 cm −1 [26], which indicated the formation of enediol [27]. This means glucose reacted with NaOH before the electro-oxidation. In addition, the disappearance of the peak at 1080 cm −1 and 1350cm −1 in glucose/NaOH electro-oxidized mixture in Figure 3b proved that the electro-oxidation product was not gluconate, which indicated that more than four electrons were transferred to the cathode [28][29][30][31]. This was beneficial for the glucose sensing.  Further evidence was obtained by the UV and NMR measurements. As shown in Figure 4, a new absorption peak was observed in glucose/NaOH mixture solution at 278 nm, which was attributed to enediol [32,33]. Similar results could also be found in NMR data, as shown in Figure 5, where two -CH 2 groups were detected, indicating the transformation of glucose to enediol.
Crystals 2020, 10, x FOR PEER REVIEW 5 of 8 Further evidence was obtained by the UV and NMR measurements. As shown in Figure 4, a new absorption peak was observed in glucose/NaOH mixture solution at 278 nm, which was attributed to enediol [32,33]. Similar results could also be found in NMR data, as shown in Figure 5, where two -CH2 groups were detected, indicating the transformation of glucose to enediol.    Further evidence was obtained by the UV and NMR measurements. As shown in Figure 4, a new absorption peak was observed in glucose/NaOH mixture solution at 278 nm, which was attributed to enediol [32,33]. Similar results could also be found in NMR data, as shown in Figure 5, where two -CH2 groups were detected, indicating the transformation of glucose to enediol.   It is clear now that we discovered the reason behind the limited linear range and enhanced successfully such a range to 40 mM, which was in sharp contrast to the literature, where the linear range was limited, and the reason was neglected [16,20,21,23]. Although a similar relationship between the linear range and OH − concentration can also be extracted in other literature, this relationship was overlooked before. The actual influence of hydroxide ions was found to be on the reaction with glucose. When the OH − concentration is low, the glucose electro-oxidation reaction only transfers two electrons to the cathode, as opposed to the four or more electrons transferred in the electrolyte with high OH − concentration. Subsequently, the current response was reduced, which corresponded to the non-linear range in the I-t curve. Therefore, insufficient OH − concentration in the electrolyte was the real cause that limited the linear range, instead of the electrode passivation or limited electrode potential [9,16].
To summarize, high OH − concentration is beneficial for glucose sensing, glucose could react with hydroxide ions, generating enediol, which lowered the energy barrier of the electro-oxidation reaction and more than four electrons are transferred to the cathode [28]. The final product was glucuronate (Equation (1)), instead of gluconolactone (Equation (2)). More than four electrons were transferred in this case, indicating a higher measured current in the I-t curve, which could not only increase the linear range but also the measurement accuracy.

Conclusions
In conclusion, the root cause of the limited linear range of glucose sensing has been successfully uncovered, attributing to alkaline concentration, which affects the degree of glucose oxidation. By increasing the concentration of hydroxide ions, not only has the linear range increased to 40 mM, but also the sensitivity of glucose sensor has enhanced, indicating better measurement accuracy. These discoveries are hopefully able to offer some impact to the non-enzymatic glucose monitoring community, as they may open a new path for biomedical sensing.