For Reviewers

Ethanol, methanol, iso-propanol, resazurin (RE), methyltrimethoxysilane (MTMS), hydrotalcite (HT), L-ascorbic acid, tartaric acid, malic acid, lactic acid, acetic acid, succinic acid, glucose, fructose, and saccharose were purchased from Sigma-Aldrich Chemical Co. (Seoul, South Korea). Other chemicals, such as hydrochloric acid, sodium hydroxide, and sodium phosphate (mono- and di-basic), were of analytical grade and used without further purification.

First, 10 mM RA (30 μL) in ethanol was absorbed in 10 mg HT and aged at room temperature for 24 h. Sol-gel of MTMS was prepared by mixing 0.2 mL MTMS with 0.8 mL ethanol and 20 μL concentrated HCl. The sol-gel was aged at room temperature for approximately 24 h for polymerization. Subsequently, the solgel (1 mL) was mixed with the RA/HT combination (~69 μg RA/ 10 mg HT), then immediately coated on PET film in the area of 42.25 cm² and dried at room temperature for several hours before being allowed to dry at 60^oC for 12 h. The membrane was washed by soaking in distilled water for 15 min, and no dye leakage was observed.

The RA/RO*HT membrane was cut and pasted onto the bottom of each well of a 24-well microtiter plate. The membrane was cleaned by soaking it in distilled water for 4 h prior to being used for ethanol determination. Upon cleaning, the membrane was exposed to different concentrations of ethanol (0–20 vol%) for 10 min and the fluorescence intensity was measured using a multifunctional fluorescence microtiter plate reader (Safire², Tecan Austria GmbH, Austria). Data were collected from the fluorescence intensity at an emission wavelength of 570 nm (λ_em=570 nm) and an excitation wavelength of 500 nm (λ_ex=500 nm) for the RA/RO*HT membrane. The limit of detection (LOD) was based on the ratio of the standard deviation (SD) of the blank samples to the slope of the calibration curve (SI) as follows:

LOD = 3.3*SD/SI

The reversibility of the RA/RO*HT membrane was measured in 10% ethanol and distilled water. The membrane was first exposed to distilled water and then to a 10% ethanol solution before being repeated. A microtiter plate reader was used for fluorescence measurement over a period (10 min) with an interval of 30 s during measurement.

Subsequently, the effects of pH and temperature on the ethanol measurements were investigated. The RA/RO*HT membrane was exposed to 10% ethanol solution in the range from pH 2.0 to pH 8.0. Additionally, the RA/RO*HT membrane was tested with different temperatures (20, 25, 30, 33, and 35oC) in the ethanol concentration range from 0 to 10%.

The long-term stability of the RA/RO*HT membrane was evaluated by comparing the sensitivity of the RA/RO*HT membrane at initial and after a certain time of use. The effects of common beverage components on the RA/RO*HT membrane were investigated by adding individual components to a 10% ethanol solution. More specifically, 2 g/L glucose, 2 g/L fructose, 2 g/L saccharose, 0.15 g/L ascorbic acid, 1.8 g/L malic acid, 1.2 g/L lactic acid, 0.6 g/L acetic acid, 2.4 g/L tartaric acid, and 1.2 g/L succinic acid.

Based on the linear detection range of the RA/RO*HT membrane for the ethanol concentration, three types of Korean alcoholic beverages with different ethanol concentrations were chosen for this study, i.e., beer (4.5%), makgeolli (6%), and soju (16.9%). Samples of beer and makgeolli were pretreated before measurement to reduce interference from organic compounds in alcoholic beverages through filtration with a filter of 0.45 μm pore size and pH adjustment to pH 6. Additionally, carbon nanoparticles reduced the color of the beer.

Resazurin (RA) is a phenoxazine dye widely used in biology [17]. It can participate in redox reactions in biological processes as an oxidation reagent and is converted to the resorufin form (RO) with a pink color that differs from the blue color of RA in the initial samples. RO has a high fluorescence intensity, making it suitable for use in optical devices. As shown in Fig. 1(a), RA shows maximum emission wavelengths (λ_em) of 580 nm and 625 nm in distilled water (above, left). These two emission wavelengths can vary approximately ±10 nm based on the solvents used and the measurement conditions, e.g., λ_em=590 m and 635 m in ethanol (Fig. 1(a), above, right). Moreover, they can also be shifted to shorter emission wavelengths (i.e., 570 and 615 nm for RA in distilled water) using another fluorescence spectrophotometer (Fig. 1(a), below, left), and RO has a maximum emission wavelength of 570 nm. The properties of the ethanol-sensing membrane were studied on a 24-well microtiter plate, and emission wavelengths of 570 and 615 nm were chosen to collect the fluorescence data.

Fig. 1. 
(a) (above) 2D-fluorescence spectra of 0.1 mM RA(resazurin) in distilled water (DW) and absolute ethanol (EtOH) measured by a fluorescence spectrometer (F-4500), and (below) 2D-fluorescence spectra of 0.1 mM RA and RO (resorufin) in DW, measured using a fluorescence microtiter plate reader (Safire²), (b) response of 0.1 mM RA in short chain alcohols with different polarities (E_T^N: MeOH = 0.762, EtOH = 0.654, IPA = 0.546) and (c) fluorescence emission spectra of RA and RO, and their photos under sunlight and UV light at 365 nm (power of 8 W)

Based on the emission wavelengths of RA and RO, the ratio of the fluorescence intensities at λ_em =570 nm and 615 nm must be considered to evaluate the conversion of RA to RO in different environments. In addition to its capability in redox reactions, RA is sensitive to the polarity of several short chain alcohols (Fig. 1(b)). In this study, 0.1 mM RA was exposed to various solutions (water, methanol, ethanol, and isopropanol) with different polarities. The relative polarity (E_T^N) of the solutions decreased in the order: 1> 0.762> 0.654> 0.546, corresponding to distilled water (DW)> methanol (MeOH)> ethanol (EtOH)> isopropanol (IPA). As depicted in Fig. 1(b), the ratio of the fluorescence intensities at λ_em=570 and 615 nm (FI570/FI615) increased as the polarity of the solutions decreased. This was attributed to the conversion of RA to RO in solvents with different polarities and the reorientation of solvent molecules by RO [18,19]. Additionally, the pH of the solvents can explain the varying fluorescence intensities of these two bands. The decrease in the intensity of the band at 615 nm and the increase in the intensity of the band at 570 nm are caused by the protonation of the Noxide group of RA at lower pH [17]. Therefore, the sensitivity of the RA/RO complex to the polarity of short chain alcohols and the pH of the solvents contributed to the sensitivity of the ethanolsensing membrane in this study.

As shown in Fig. 1(c), the fluorescence emission wavelength (λ_em = 615 nm) of RA was shifted to λ_em = 570 nm when RA was reduced to RO. The change in the fluorescence emission spectrum of RA before and after reduction is clear under normal sunlight and UV light, as shown in the photos in Fig. 1(c).

The sol-gel matrix of MTMS was selected to support the indicator dye because of its high stability and good capture into the matrix. The polymerization and condensation of MTMS can produce Si-O-Si-CH3 bonds that render the MTMS membrane hydrophobic [20]. Additionally, we used MTMS, which has been shown to be an excellent supporting material for fluorescent dyes, in our previous studies for the fabrication of sensing membranes [21].

HT is a rhombohedral hydrotalcite with the chemical formula Mg₆Al₂(OH)₁₆CO₃.4H₂O. It is well-dispersed in polymer solutions and is commonly used as an anion-exchange clay. Aluminum (Al or Al³⁺) in HT plays an important role in reducing resazurin (RA) to resorufin (RO) and in the oxidation of ethanol by electron donation or acceptance, respectively, as shown in Scheme 1.

Membranes produced from the sol-gel of silanes are generally thin and smooth. The SEM and AFM images in Fig. 2 show that the MTMS membrane was extremely smooth with a surface mean roughness (Ra) of 0.24 nm and a root mean square roughness (Rq) of 0.298 nm. However, the surface of the RA/RO*HT membrane containing the RA/RO complex and HT in MTMS was rougher than that of the MTMS membrane alone, with Ra was 1.81 nm and Rq was 2.85 nm. Fig. 2 shows the strong fluorescence emission of the RA/RO*HT membrane under UV irradiation.

Fig. 2. 
SEM and AFM images of the sol-gel membrane (MTMS) and RA/RO*HT membrane, and photos of the RA/RO*HT membrane under sunlight and UV light.

When RA and HT were captured in the MTMS sol-gel membrane, the reduction of RA occurred immediately. This is because of the acidity of the MTMS sol-gel, which is a good medium for the reduction of RA during HT catalysis. Therefore, the emission wavelength of RO at λ_em = 570 nm appears prominently compared to the emission wavelength of RA at λ_em = 615 nm in terms of the fluorescence intensity (Fig. 3(a) (left)). As depicted in Fig. 3(a) (left), the fluorescence intensity of the RA/RO complex at λ_em = 570 nm increased significantly with the concentrations of ethanol solution. This good response of the RA/ RO*HT membrane to an ethanol solution is presented using a calibration curve with a linear detection range of 1%–20% ethanol (Fig. 3(a) (right)) and a limit of detection (LOD) of 0.178%. Compared to the results of Simon et al. [9], the RA/RO*HT membrane proposed in this study shows a better response than the malachite membrane used in Simon’s research in terms of the limit of detection. In fact, the RA/RO*HT membrane also had high sensitivity to ethanol concentrations of over 20%, but the lifetime of the membrane could be shortened owing to the loss of the RA/RO complex. Therefore, we chose an ethanol measurement range of 1–20% to maintain the long lifetime of the ethanol-sensing membrane. Compared with other ethanol sensors using the enzyme AOX, such as those described by Barsan et al. [11] and Patel et al. [10], this RA/RO*HT membrane showed a wider ethanol detection range than the electrode biosensors. This indicates that the RA/RO*HT membrane can be used to measure the concentration of ethanol in real alcohol samples without dilution.

Fig. 3. 
(a) Response of the RA/RO*HT membrane at different concentrations of ethanol and a calibration curve for ethanol detection at λ_ex = 500 nm and λ_em = 570 nm and (b) the rrepeatability of the RA/RO*HT membrane exposing repeatedly at 0% and 10% ethanol (λ_ex = 500 nm and λ_em= 570 nm)

The RA/RO*HT membrane exhibits high sensitivity and fast reversibility when repeatedly exposed to distilled water (0% ethanol) and 10% ethanol (Fig. 3(b)). The repeatability was stable after each measurement cycle, as the relative standard deviation (RSD) was exceedingly small, that is, 2.1% in 0% ethanol and 0.4% in 10% ethanol. Fig. 3(b) also indicates that 95% of the fluorescence intensity was obtained in a short time (approximately 30 s – 1 min), but approximately 10 min was required to reach saturation at an ethanol concentration of 10%. Meanwhile, the return time from 10% ethanol to the starting point (0% ethanol) was short time (1 2 min).

In this study, alcoholic beverages in the food industry are the main targets for the use of RA/RO*HT membranes in real applications. Therefore, pH and temperature are crucial parameters that must be investigated to adjust the conditions suitable for ethanol measurement. The fluorescence intensity of the RA/RO*HT membrane increased as the temperature increased from 20^oC to 25^oC at all ethanol concentrations in the range of 1 – 10% (Fig. 4(a) (left)). In the temperature range of 30 – 35^oC, the fluorescence intensity of the RA/RO*HT membrane was higher than that in the temperature range of 20 – 25^oC. However, the fluorescence intensity of the membrane at λ_em = 570 nm did not increase significantly with increasing temperature in the ethanol concentration range from 1% to 6% in this temperature range (30 – 35^oC).

Fig. 4. 
(a) (left) Response of the RA/RO*HT membrane at different temperatures of 20^oC – 35^oC in the ethanol concentration range of 1 - 10%; and (right) at 10% EtOH in the pH range of 2.0 – 8.0 and (b) Responses of the RA/RO*HT membrane at 10% ethanol in the presence of various acids such as 0.6 g/L acetic acid, 0.15 g/L ascorbic acid, 1.2 g/L lactic acid, 1.8 g/L malic acid, 1.2 g/L succinic acid, 2.4 g/L tartaric acid, and 2 g/L of different sugars such as glucose, fructose, and saccharose

The fluorescence intensity of the RA/RO*HT membrane increased only at ethanol concentrations greater than 6%. According to Flamigni et al., [18], the fluorescence quenching of RO at low temperatures is owing to solute-solvent interactions between the negative partial charges localized on the two end oxygen atoms of RO and the solvent. Shielding of the oxygen negative charges reduces the interaction between RO and bulk alcohol.

The pH of a solution is often measured during the fermentation of alcoholic drinks. Low pH typically occurs in fermented solutions during the production of alcoholic beverages. The fluorescence intensity of the RA/RO*HT membrane decreased slightly in the ethanol solution at a low pH (from pH 2 to 4) (Fig. 4(a) (right)). Higher fluorescence intensities were observed in ethanol solutions with pH > 4. In low-pH solutions, the low fluorescence intensity is explained by the formation of RO agglomerates [18]. Protonation results in the formation of a neutral form of resorufin, which can lead to aggregation. However, in this study, the RA/RO complex was captured in the MTMS sol-gel matrix, minimizing the possibility of dye agglomeration in acidic solutions. It can be concluded that the effects of pH on the RA/RO*HT membrane were insignificant within the pH range of 2–8, but pH 6 was chosen for the preparation of the standard samples.

The operating mechanism of the RA/RO*HT membrane is based on the interaction between the oxidation of ethanol and the reduction of resazurin (RA) to resorufin (RO) during the catalysis of HT through electron donation and acceptance. Additionally, the sensitivity of the RA/RO*HT membrane to the polarity of short chain alcohols is a crucial reason for its operation. Many oxidizing compounds are present in beverages and can affect the ethanol measurement of ethanol-sensing membranes. This could be attributed to the presence of acids, such as tartaric acid and ascorbic acid, in alcoholic beverages. Therefore, some acids that arose during processing or were produced as final products were tested using the RA/RO*HT membrane.

The acid concentration used in this study was based on the typical amount of each acid in the beverage. The RA/RO*HT membrane was stable in the presence of a given amount of acid (Fig. 4(b)). Compared to the control sample, the recovery of the RA/RO*HT membrane is 110.3% for 0.15 g/L ascorbic acid, 110.2% for 1.2 g/L lactic acid, 102.5% for 1.8 g/L malic acid, 105.1% for 1.2 g/L succinic acid, and 100.9% for 2.4 g/L tartaric acid. However, the fluorescence intensity of the RA/ RO*HT membrane decreased significantly at 0.6 g/L acetic acid (73.1% recovery). This can be attributed to the relationship between acetaldehyde produced from the oxidation of ethanol and acetic acid. Acetaldehyde is commonly used as a precursor to acetic acid; therefore, the presence of acetic acid with amount of 0.6 g/L inhibited the oxidation of ethanol to acetaldehyde, which destroyed the electron flow of the reactions shown in Scheme 1 and resulted in a decrease in the fluorescence intensity of the RA/RO*HT membrane. Thus, with minimal effect from most common interferences, the RA/RO*HT membrane can be used in the food industry and other fields (e.g., the environment).

In addition to the effects of various acids, the amount of sugar contained in the beverage is usually very high (the total sugar amount varies in the range of 1–8 g/L); therefore, the amount of sugar may affect the measurement of the ethanol-sensing membrane. Large amounts of sugar affect the motion of the alcohol in the solution and reduce the contact between ethanol and the RA/RO*HT membrane. The presence of 2 g/L of each type of sugar did not significantly affect the response of the RA/RO*HT membrane, with recovery percentages of 104.7%, 107.1%, and 107.5% for glucose, fructose, and saccharose, respectively (Fig. 4(b)).

The RA/RO*HT membrane was stable and could be used for continuous analysis of approximately 65 samples. Fig. 5 shows that, after 65 consecutive measurements, the slope index (SI = 122.8) of the RA/RO*HT membrane was not significantly different from that of the RA/RO*HT membrane (SI = 129.9). There were no chemical bonds between the sol-gel MTMS and the RA/RO complex, but the RA/RO*HT membrane could be used to measure many samples without dye leakage. This can be attributable to the absorption of the RA/RO complex in the HT layer, as HT is an anion exchange clay and RO exists in the anion form, which leads to a tight bond of RA/RO in the layers of HT. We attempted to conjugate the RA/RO complex and HT with other supporting polymers (e.g., ethyl cellulose) through chemical binding; however, this bonding reduced the sensitivity of the RA/RO complex to the ethanol solution, resulting in a lower response of the RA/RO*HT membrane (data not shown).

Fig. 5. 
Lifetime of RA/RO*HT membrane after 65 consecutive measurements of ethanol samples

Table 1 lists the ethanol sensors used for determining the ethanol in the liquid phase. Some analytical parameters were compared with those of the RA/RO*HT membrane developed in this study. The RA/RO*HT membrane exhibited good analytical performance, specifically in terms of high sensitivity, wide detection range, fast response time, and longterm stability.

To evaluate the accuracy of the ethanol sensor for practical applications, it must be tested using real samples. An RA/RO*HT membrane was used to determine the ethanol content of three Korean alcoholic beverages: beer (4.5%), makgeolli (6%), and soju (16.9%).

According to the equation collected from the linear calibration curve of standard ethanol solutions (y=396.14*x – 623.95 with R²=0.991) with the RA/RO*HT membrane and the data obtained from measuring samples of Korean alcoholic beverages (Fig. 6), the ethanol concentrations of three real samples are presented in the table in Fig. 6. In the table, the accuracy of the RA/RO*HT membrane is compared with the alcohol content provided by the producers. The data in the table in Fig. 6 show that the RA/RO*HT membrane has a high sensitivity and accuracy for real samples.

Fig. 6. 
Response and recovery of RA/RO*HT membranes from several Korean alcoholic beverages