Textile-Based Volatile Organic Compound Sensors Using Cellulose Fabrics and Direct Dyes

ABSTRACT In order to fabricate colorimetric textile sensors detecting gas phase of volatile organic compounds (VOCs), 11 kinds of direct dyes were investigated for 13 types of VOCs. Direct Violet 51 was chosen for its best sensing performance both instrumentally and visibly. Direct Violet 51 exhibited the change of color strength and color hue by exposure VOCs. The dyed cotton fabrics detected polar VOCs well, especially N,N-dimethylformamide (DMF). The colorimetric cotton-based textile sensor dyed with Direct Violet 51 exhibited approximately 10 of color difference (ΔE) under 10 ppm DMF gas within 100 min, and the maximum absorption wavelength was shifted from 560 nm to 530 nm. The sensing mechanisms could be summarized in three factors: solvatochromism and aggregative characteristics from the point of view of dye, and adsorption amount of VOCs for fabric. The colorimetric textile sensor retained its sensing performance after 10 cycle tests.


Introduction
A lot of chemicals are being used in almost all industries or research fields, and they are increasing continuously. Although these chemicals are used as basic materials to make life-enriching substances, ironically, most of them are generally recognized as hazardous and toxic to the human body as well as to the environment. As the industry develops, the level of safety awareness in the working environment rises, and harmful chemicals have become a hotly debated social issue. In particular, volatile organic compounds have attracted markable attention, because they are easily vaporized even at room temperature to a dangerous level of concentrations. According to recent reports, VOCs caused many serious symptoms to human body, such as dizziness, confusion, headaches, nausea, dyspnea, hepatotoxicity, and central nervous system depression (Henderson et al. 2011;Matsushima et al. 2015;Prat et al. 2016). Therefore, many efforts are being made to monitor the VOCs. Although they are very carefully controlled, sometimes the leakage of solvents occurs inadvertently in liquid or vapor phase. The leakage of the vapor phase is more serious because it may not be recognized or noticed easily in advance. Therefore, it will be very useful and important to perceive the leakage of the vapor phase of VOCs with textile sensors without any electronic devices before the problem becomes serious.
As the application of dyes extends from conventional to smart coloration, the colorimetric sensors fabricated from chromic dyes have attracted attention. Chromism is a phenomenon in which colors change according to the absorption spectral transition due to changes in the energy level of the π-or d-electrons of the molecules under specific stimuli. There are various categories of chromism for detecting specific hazardous or toxic chemicals, such as corrosive acids, bases, volatile organic compounds (VOCs) (Beatty et al. 2019;Chen et al. 2019;Favaro et al. 2007;Hau et al. 2019;Kim, Wang, and Son 2012;Little and Christie 2016;Ono et al. 2019;Ribeiro et al. 2013;Wang et al. 2019). They include ionochromism, halochromism (De Meyer et al. 2016;Kubota et al. 2021;Kim 2020, 2021;Manjakkal, Dervin, and Dahiya 2020), and solvatochromism. Most chromic dyes can detect chemicals in the liquid or solution phase. However, it is vital to detect chemicals in the vapor phase, even with high sensitivity and visibility. Chromic dyes require a substrate to hold onto. Considering the durability of colorimetric sensors exposed to harsh environments, it is advantageous to confine dyes inside rather than on the surface of substrates. However, the sensitivity of the colorimetric sensors to external stimuli decreases noticeably by being trapped within the substrate. The substrate may cause a delay in contact with the stimuli by trapping the dye or blocking the structural deformation of the chromic dye from the stimuli through steric hindrance.
In order to fabricate a colorimetric textile sensor that can detect the gaseous VOCs, it is necessary for chromic dyes to have special properties. The properties are solvatochromism and linear/planar molecular structures for good intermolecular packing, this was called vapochromism. Herein, we investigated several conventional direct dyes exhibiting vapochromic characteristics when applied to fabrics. Recently, research on various applications using cellulosic fabrics, especially cotton, for example, pH sensors, temperature/humidity sensors, sweat monitoring sensors, biomedical or antistatic applications has been reported, because cotton fabrics have a lot of advantages such as water absorptivity, dyeability, processability, gas-permeability, and eco-friendliness (Abdelrahman et al. 2020;Ahmed et al. 2020;Al-Qahtani et al. 2021;Alaysuy et al. 2022;El-Naggar et al. 2021). Therefore, the cotton fabrics were selected as a substrate. After fabrication, the textile-based VOC sensor, sensing performance upon exposure to 13 kinds of VOCs and sensing mechanism will be studied. Considering practical application, reusability will be also tested.

Materials
Eleven kinds of direct dyes whose chemical structures are known were purchased. The molecular structures of the dyes are shown in Figure 1. To fabricate the VOCs-sensing colorimetric textile sensor, a cotton fabric (KS K 0905, warp; 35 threads/cm, weft; 31 threads/cm, weight; 115 ± 5 g/m 2 ) was used as a substrate.

Application of direct dyes to cotton fabrics
To select dyes exhibiting vapochromic behavior even when applied to fabrics, the 11 direct dyes (at 2% owf) and 2.5 g of sodium sulfate were added in 50 mL deionized water and stirred for 30 min at room temperature. Then, 1.0 g of the cotton fabric was dyed in the solutions at 90°C for 1 h. Thereafter, the dyed fabrics were washed in distilled water 5 times at room temperature to remove the surface deposited dyes.
All dyed fabrics were tested under exposure to 13 gas phases of VOCs, and the best vapochromic direct dye was selected. To determine an optimal concentration showing the maximum color difference before and after exposure to VOCs, pristine cotton was dyed with the selected dye at concentrations of 0.3 ~ 20% owf under the same dyeing conditions.

Color changes in applied dyes to the cotton fabrics in gaseous VOCs
The color strengths and spectra of the dyed fabrics were measured by a color measurement instrument (Spectrophotometer CM-3600d, Konica Minolta) and expressed by K/S values obtained every 10 nm in the range of 360 ~ 740 nm using Equation (1) based on the reflectance (R) of single wavelengths. The measurement was performed using a 10º standard observer under light D 65 (Baumann, Groebel, and Krayer 1987;Koh 2006). (1) To measure the color change upon exposure to gas phase of VOCs, the concept of color difference (ΔE) was employed. Herein, the CIELAB color space was employed, which comprises L*, a*, and b* factors. These three factors are plotted at three dimensions corresponding to the lightness (L*), red to green (a*), yellow to blue (b*) of the color vision. The color difference (ΔE) was calculated as follows (Equation (2)) (Fairchild 2005;Hunt 1991): ΔE ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where Δ indicates the difference before and after exposure to gaseous VOCs.
To test the vapochromism of the samples dyed with various direct dyes, excessive 13 VOCs were added to vials, and the dyed fabrics (15 mm × 20 mm) were hung into each vial with a thread without direct contact with the VOC solution. Finally, the vials were sealed completely and left for 3 h. The dyed samples were measured before and after exposure. After selecting the dye with the best performance and most reactive VOCs, the vapochromic behavior was quantitatively evaluated with respect to the concentrations and time of exposure. Various concentrations (10 ~ 300 ppm) of the selected VOCs could be controlled accurately by injecting the amount of VOCs into the vials.

Vapochromism of the selected direct dye
To investigate the cause of color change in specific gaseous VOCs, we analyzed the properties of the dye. The dye was dissolved in eight solvents of different polarities, including water, glycerin, DMF, methanol, acetone, MEK, pyridine, and THF to examine the solvatochromism. Absorbance was measured by an ultraviolet-visible spectrophotometer (Optizen 2010UV). To verify the crystalline deformation of the dye molecules upon exposure to gaseous DMF, X-ray diffraction analysis was performed using Malvern Panalytical EMPYREAN operating in the reflection mode with Cu-K α radiation (λ; 1.540598 nm, 2° ≤ 2θ ≤ 100°). After recrystallization of Direct Violet 51 in ethanol, the pristine dye and the dye exposed to DMF gas for 12 h were analyzed and compared.

Effect of VOCs adsorption to cotton fabrics on vapochromism
The amount of each VOC adsorbed to the pristine cotton fabrics was investigated by mass spectrometry (GC-2030, Shimadzu, column; dimethylpolysiloxane, 30 m, 0.25 mm I.D., 0.25 μm, PerkinElmer, detector; FID). Various concentrations (1% ~ 50%) of standard solution obtained by mixing the analytical solvent with desorbed solvent were measured to obtain the calibration curve and the relationship between the concentrations of the analytical solvents and the peak area ratio of the analytical solvent. The pristine fabric (0.1 g) was exposed to 300 ppm of the analytical solvent without direct contact in a sealed vial at room temperature for 3 h. The adsorbed cotton fabric with the analytical solvent was soaked in 3 mL of the desorption solvent in the sealed vial and stirred to desorb the analytical solvent. After filtration, the mixed solution was analyzed by the mass spectrometry, and then, the concentration of the solution was calculated as the peak area ratio of the analytical solvent and induction calculation from the calibration curve. To verify the relationship between the color difference before and after exposure to various VOCs, the amounts of solvent adsorbed on the fabrics, n-hexane, benzene, chloroform, methanol, acetone, MEK, THF, pyridine, and DMF were used as the analytical solvents. DMF was used as the desorption solvent. To investigate the adsorbed amount of DMF, methanol was used as the desorption solvent. In addition, to confirm the adsorption behavior of DMF causing the greatest color change, according to the exposure time, the pristine cotton fabric was exposed to 300 ppm of DMF for 1 to 24 h and characterized.

Reusability of the colorimetric sensors
The repeatability of the textile-based VOC sensor was examined. One sensor was exposed to 300 ppm of DMF at 70°C for 200 min, and the color difference was measured before and after exposure, and then it was kept inside a vacuum chamber for 24 h to desorb the adsorbed DMF molecules completely out of the sensor fabric. After that, the process of exposure to DMF, the measurement of color change, and desorption were repeated for 10 cycles with the same sensor.

Selection of the optimum vapochromic direct dye
To select the dyes with the best vapochromic behavior even when penetrated and trapped in a fabric, all the 11 kinds of direct dyes (Figure 1) whose structure was known were applied to cotton fabrics. All the applied fabrics were exposed to excessive 13 gaseous VOCs for 3 h. Thereafter, the color of all samples was measured and the results are shown in Figure 2. Among them, the cotton fabrics applied with Direct Violet 51, Direct Red 28, Direct Yellow 12, and Direct Orange 26 showed noticeable color differences in some VOCs. The color difference in the dyed fabric with Direct Red 28 and Direct Yellow 12 was greater than 10 in the gas phase of methanol. Direct Orange 26 and Direct Violet 51 showed the color difference greater than 10 in gaseous methanol, acetone, MEK, THF, DMF, and pyridine. From the results, these dyes seemed to detect well relatively polar VOCs rather than nonpolar ones. Since the direct dyes and the cotton fabrics used as substrates were polar materials, they did not interact effectively with the relatively nonpolar VOCs such as n-hexane, n-octane, benzene, toluene, chlorobenzene, phenol, and chloroform.
Although the color difference of Direct Yellow 12, Orange 26, Red 28 was high under exposure to certain VOCs, it is not visible to naked eyes. It is because hyperchromic or hypochromic shift occurred, but there was no wavelength shift. In other words, only the color intensity was changed without any color change. However, in Direct Violet 51, there was not only a color change from violet to red (hypsochromic shift) but also an increase in the intensity (hyperchromic shift) upon exposure to gaseous DMF, pyridine, acetone, THF, MEK, and methanol. The color spectra of the samples with the four vapochromic dyes before and after exposure to several gaseous VOCs are shown in Figure 3.
To fabricate a textile-based gaseous VOCs detecting colorimetric sensor, Direct Violet 51 was selected. The color difference of the fabric dyed with Direct Violet 51 before and after exposure to acetone, DMF, and pyridine showed a large color change of 15 or higher. Especially in DMF, the highest color change of 29.4 in this study. Acetone is commonly used in the field, but it is not so harmful to the human body, and pyridine is rarely used. DMF is used in large quantities in various fields, and it is toxic to human health upon overexposure, causing issues as hepatotoxicity. Therefore, the sensor fabricated with Direct Violet 51 was employed in the DMF detection, and the color change was investigated in detail.

Gaseous DMF-sensing properties of direct dyes on cotton fabrics
To determine the optimal concentration of the dye showing the maximum color change upon exposure to gaseous DMF in the fabric, the dye was applied in the range of 0.3 ~ 20% owf. The color strength and difference of the samples dyed with Direct Violet 51 are presented in Figure 4. The color strength increased with increasing dye concentration and arrived equilibrium at a certain concentration. The color change was maximum at 2% owf of the dye, and after reaching the maximum, the color change decreased as the dye concentration increased. We infer that the number of deformable dye molecules at low dye concentrations is very low. Although at high dye concentrations, too many dye molecules made the color too deep and dark. The optimum concentration of the dye was 2% owf, and it was used in subsequent experiments.
The dyed sample with the optimum concentration was investigated quantitatively in gaseous DMF. The color of the dyed fabrics upon exposure to various concentrations (10 ~ 300 ppm) of gaseous DMF was measured with respect to the exposure time of 1 ~ 300 min. The color differences at various concentrations of gaseous DMF with respect to the exposure time are shown in Figure 5(a). The color difference increased with the increase of concentration of gaseous DMF and exposure time, and then it reached equilibrium. A big color difference as high as 10 was obtained within 100 min even at a very low concentration of 10 ppm, which was the standard DMF allowable exposure concentration in many developed countries. Generally, a color difference higher than 1.0 is recognizable to the human naked eyes. It means that the textile-based VOC sensor of this study was sensitive enough to be able to detect a trace amount of DMF gas. Figure 5(b) displays a shift in the K/S spectra of the dyed sample under 300 ppm of gaseous DMF for 24 h. The spectrum exhibited a hypsochromic shift from 560 to 530 nm and a hyperchromic shift over DMF exposure time. Consequently, both color hue and color strength were changed on exposure to DMF. The L*, a*, and b* values for the dyed samples were 23.03, 18.21, and −24.79 before exposure and 26.95, 40.12, and −8.08 after exposure to gaseous 300 ppm DMF for 24 h (Figure 5(c)). The lightness L* was increased by only 3.92, the redness a* was increased by as high as 21.91, and the yellowness b* was increased by 16.71. This means that the color of textile sensor was changed from violet to red. In other words, the redness was increased by increase of a* and blueness was decreased by increase of yellowness b × .

Sensing mechanism (1): solvatochromism of the vapochromic dye on substrates polarity
To evaluate the color change properties of Direct Violet 51 due to the exposure to VOCs, the solvatochromism of the dye was investigated only in polar solvents in which dyes were dissolved, such as glycerin, water, methanol, pyridine, N,N-dimethylformamide, methyl ethyl ketone, acetone, and tetrahydrofuran. Therefore, the following nonpolar solvents were excluded: n-hexane, n-octane, benzene, toluene, chlorobenzene, phenol, and chloroform. The shift in the absorption spectra with the solvent polarity is in Figure 6(a). In this study, the dielectric constant (Ɛ γ ) and relative polarity were referred as a solvent polarity parameters. The dielectric constants and relative polarity of the solvents are as follows: water; 80.4 and 1, glycerin; 43 and 0.812, DMF;36.7 and 0.386,methanol;32.6 and 0.762,acetone;20.7 and 0.355,MEK;18.4 and 0.327,pyridine;12.5 and 0.302,and THF;7.5 and 0.207. The maximum absorption wavelengths versus the solvent polarity are closely correlated (Figure 6(b)). Since the peak shows a bathochromic shift as the polarity of the solvents increases, Direct Violet 51 is a positive solvatochromic dye.
Direct Violet 51 may be affected by the polarity of the substrate to be dyed. Solvent polarity has been commonly reported, whereas solid polarity has been rarely reported. We predicted substrate polarity using the reported polarity for solvents having similar elementary compositions, and we confirmed the effect of optical properties on the substrate polarity. The molecular formula of cellulose is (C 6 H 10 O 5 ) n whose repeat unit is glucose. In molecules, the ratio of the hydroxyl group against carbon largely depends on its polarity. The ratio of the hydroxyl group against carbon in the cellulose repeat unit is 1:2. Therefore, the relative polarity of cellulosic substrates is between that of 1-propanol (0.617; C:OH = 1:3) and glycerin (0.812; C:OH = 1:1). The maximum absorption wavelength of the dye dissolved in DMF with a relatively low polarity (0.386) is 535 nm, and the maximum absorption wavelength of the dye penetrating the cellulosic substrate having a relatively high polarity (0.617 ~ 0.812) is 560 nm. The peak of a DMF-adsorbed dyed cotton fabrics is 540 nm, which is between 535 and 560 nm. Color change is affected by both solid substrate and solvent polarity, and it is called solidsolvatochromism.

Sensing mechanism (2): change of crystal structure of the vapochromic dye by DMF
The colorimetric sensors showed a color change when exposed to gaseous DMF due to solvatochromism ( Figure 5 and Figure 6) and a change in the aggregation state of the dye molecules. To verify the crystalline deformation between the dye molecules due to exposure to DMF gas, XRD analysis was performed. Figure 7 depicts the XRD patterns of Direct Violet 51 before and after exposure to DMF gas. The dye showed diffraction peaks at 27.3°, 31.7°, 45.4°, 56.4°, 66.2°, 75.2°, and 83.9° before exposure and it can be calculated the molecular interplanar distance as 3.26 Å, 2.82 Å, 2.00 Å, 1.63 Å, 1.41 Å, 1.26 Å, and 1.15 Å, respectively. After exposure to DMF gas, the intensity of the peaks decreased, and new peaks appeared at 5.1°, 6.2°, 8.2°, 13.1°, 20.0°, 25.5°, 26.8°, and so on. This indicates that the intermolecular distance was increased by DMF molecules were adsorbed to the dye molecules as 17. 26 Å, 14.30 Å, 10.73 Å, 6.76 Å, 4.43 Å, 3.49 Å, and 3.32 Å. It means that the intermolecular crystal distance changes as VOCs uptake into the crystal lattice, causing a color change when VOCs are exposed (Wenger 2013). Besides, we can infer aggregate of the dye molecules for a π-conjugated molecule (monomer) in relation to the absorption UV-vis spectra. Most organic dye molecules are dissolved as a monomer in good solvents but aggregate in poor solvents or solid-state. The aggregate is theoretically divided into two forms: J-and H-aggregates. The molecules may form end-to-end stacking in J-aggregates. It can be simplified two levels of the excitonic state in the molecule arrangements through the interaction of their transition dipoles, such as unstable aggregation exhibiting higher transition energy by charge repulsion and stable aggregation with lower transition energy by charge attraction compared to the monomeric state. In H-aggregate, the molecules may aggregate in plane-to-plane stacking and it also can be simplified as two arrangements. However, the unstable aggregative form in J-aggregates and the stable aggregative form in H-aggregates are forbidden transitions according to the exciton theory. Therefore, J-aggregates show the red shifted absorption and H-aggregates show the blue shifted absorption based on free molecule absorbance spectrum (Kasha, Rawls, and El-Bayoumi 1965;Wuthner, Kaiser, and Saha-Moller 2011). When the dyed fabrics were exposed to DMF gas, DMF molecules adsorbed to the dye molecules aggregated at the internal substrate, and then, the dye molecules become to the monomer state as increasing the aggregation distance. Direct Violet 51 in the cotton fabrics showed a hypsochromic shift. Based on this, the Direct Violet 51 might be dyed in the fabrics by J-aggregate form.

Sensing mechanism (3): effect of VOCs adsorption to cotton fabrics on vapochromism
Gaseous VOCs must have an affinity for the cotton fabrics to be adsorbed and stimulated directly by the permeated dye. Figure 8 displays the relationship between the amount of VOCs adsorbed on cotton fabric and color difference before and after exposure to gaseous VOCs. It was confirmed that the adsorption amounts of DMF and pyridine, which showed the strongest color change, were the highest (0.176 mL and 0.192 mL, respectively) on 0.1 g of cotton fabrics. Relatively nonpolar solvents such as n-hexane, benzene, and chloroform, which are very weakly detected, hardly adsorbed to the polar material, cotton fabric. The amount of VOCs adsorbed to the fabric was proportional to the intensity of color change. The affinity of VOCs with fabric substrates, in other words, the amount of solvent absorbed by the substrate, affects the sensing performance of a sensor.
To investigate the adsorption behavior of DMF to the cotton fabric, which showed the greatest color change, the pristine cotton fabric (0.1 g) was exposed to DMF at the same concentration as before (300 ppm) for different exposure times (1 ~ 24 h). Figure 9 presents that the amount of DMF gas absorbed by the fabric increased with an increase in DMF exposure time and was saturated within 5 h. The color change of the dyed sample showed a similar pattern. Since DMF has a great affinity for a cotton fabrics, the adsorption rate of DMF to the fabric was superior to that of other VOCs. In conclusion, the sensing mechanism cannot be explained by one factor. The sensing performance was affected by the three factors together. However, in order to change the color, the role of the dye will be the most important, so the sensing mechanism from the dye point of view (solvatochromism and aggregative properties) would be more important. Among them, the major factor is thought to be the solvatochromism first rather than the change in aggregative properties. The structure of all 11 dyes investigated in this study were highly linear and planar, so they would have good aggregative properties; however, the sensing performances of each of dyes were very different as shown in Figure 2. Therefore, the effect of aggregative characteristics and adsorption amount is considered to be the second effect to strengthen the color change.

Reusability of the colorimetric sensor
Since reusability or repeatability is one of the critical properties of sensors, the textile sensor was investigated for this. One sensor was exposed to 300 ppm of DMF at 70°C for 200 min, and the color difference was measured before and after exposure, and then it was kept inside a vacuum chamber for 24 h to desorb the adsorbed DMF molecules completely out of the sensor fabric. After that, the process of exposure to DMF, the measurement of color change, and desorption were repeated for 10 cycles   with the same sensor. According to Figure 10, the textile-based VOC sensor of this study could be reused continuously and repeatedly. The color of the sensor changed upon exposure to DMF could be returned to its original color by desorption of the DMF out of the sensor. The reason was that the solvatochromism disappeared and the aggregates of dye molecules were restored as desorption. As displayed in Figure 10, the color change of the sensor was maintained at almost the same level as the initial sensing performance even after repeating DMF adsorption and desorption 10 cycles.

Conclusion
A textile-based colorimetric sensor was fabricated to detect low concentrations of VOCs in the vapor phase. The 11 kinds of direct dyes having linear/planar structures were investigated in terms of solvatochromism and aggregative characteristics for color change in cotton fabrics upon exposure to VOCs. Direct Violet 51 was selected as the best vapochromic dye, and it showed the strongest color change under DMF. The colorimetric sensor based on the dye exhibited a hypsochromic shift as well as a hyperchromic shift upon exposure to polar VOCs. The reason of color change of the textile-based sensor on VOCs exposure was the solvatochromism and the aggregative property of the dye and the adsorption amount of VOCs on the substrate. The sensing performance was maintained after 10 repeat cycles.