In this paper, we proposed a new type high sensitive volatile organic compounds (VOCs) gas sensor array that is based on the pulse width modulation technique. Four different types of solvatochromic dyes and two different types of polymers, were used to make the five different types of sensing membranes. These were deposited on the five side-polished optical fibers by a spin coater to make the five different sensing elements of the array. In order to ascertain the effectiveness of the sensors, five VOC gases were tested. Finally, principal component analysis (PCA) has been used to discriminates different types of VOCs.
© 2013 OSA
Volatile organic compounds (VOCs), which are well-known air pollutants, are a group of chemical compounds that easily evaporate at room temperature. We use thousands of different VOCs in our daily lives. However, we may not be aware of how they affect our air quality. VOCs are widely used as ingredients in household products, and some may have short- and long-term adverse health effects. The concentrations of many VOCs are consistently higher indoors (up to ten times higher) than outdoors . The health risks from inhaling any chemical depend on its concentration in the air, the exposure time, and the breathing rate of the people inhaling it. Over the last few decades, many sensors have been designed to detect VOCs, and considerable research has been focused on the development of both electronic and optical VOCs sensors, in addition to the development of VOC-sensitive materials . Several techniques have been used for gas-sensor arrays, such as surface acoustic wave , polymer composite sensor array [4,5], carbon nanotube , CMOS , and capacitive  techniques.
An optical array sensor is an instrument that comprises an array of optical chemical sensors with partial electivity and an appropriate pattern recognition system, capable of recognizing simple and complex odors. This type of array-sensing system consists of appropriate, chemically sensitive materials on the sensor surface, which is interfaced with a transducer. The interaction of the analyte with the chemically sensitive material generates some physical changes that are interfaced with a transducer.
Information about the VOCs may be provided by colorimetric sensor-array technology. Suslick et al. described the application of the colorimetric sensor array [9,10]. An improved version of the colorimetric sensor-array system to detect VOC gases was reported by Hengwei Lin et al. by applying a disposable preoxidation technique . Their sensors were low cost, easy to fabricate, needed no light source or subsidiary circuits, and were able to monitor different VOC gases in real-time. The main problem was that the system was not able to detect the concentration of VOCs. Localized surface plasmon resonance (LSPR) sensors have become popular for VOC gas detection [12,13]. One example of the LSPR sensor is the silver triangular nanoprism described by Wenying Ma et al. .
Recently, fiber-optic sensors have been shown to be excellent candidates for monitoring environmental changes, and they offer specific advantages . A side-polished optical fiber sensor is a combination of the optical fiber technology with optical waveguide technology to increase the sensor sensitivity to detect chemical, physical, or environmental changes [16–21].
In our study, we propose a low-cost, high-sensitivity, easy-to-fabricate, fiber-optic VOC gas sensor array that is based on the pulse width modulation (PWM) principle. The pulse width of the signal received from a fiber-optic waveguide depends on the absorption of the evanescent field penetrating the fluid or gas on the polished cladding region. A side-polished fiber-optic sensor based on the amplitude modulation method cannot detect a very small change in light. However, for the proposed array sensing system, we observe the change in the pulse width, which depends on the light pulse amplitude and fall time. The effect on the pulse width amplitude and the fall time is caused by a small absorption change. As a result, the width of the pulse changes, which corresponds to a change in the average value of the received signal amplitude. The proposed sensor array has some other advantages such as real time monitoring capabilities, good reproducibility of the sensing system, linear sensing response over a dynamic range, remote sensing capabilities, compactness, and low cost, given that the circuitry consists of easily available and inexpensive opto-electronic components. The designed VOC sensor array has five side-polished optical fiber sensing elements. The optical VOC sensing elements were prepared by incorporating the four types of solvatochromic dyes, individually, into polyvinylpyrrolidone (PVP) and polyvinyl chloride (PVC). These were then used as sensing membranes on a side-polished single-mode fiber to obtain five different types of sensing elements. When the sensing membrane is exposed to the VOC gas, the refractive index of the planar waveguide (PWG) changes due to the charge transfer character of the solvatochromic dye in the PWG of the VOC sensing element. Therefore, the change in the effective refractive index of the PWG is the specific change in the received signal pulse width, which corresponds to the change in the amplitude of the received sensing signal. The proposed fiber-optic VOC gas-sensor array system can effectively detect low concentrations of VOCs. Principal component analysis (PCA) [22,23] has been used to explore the data distribution and classify the VOCs. A good classification success has been achieved for the different tested gas types. The obtained data confirmed our experimental investigation.
2. Theory and operation principle
A schematic diagram of the side-polished optical fiber with a sensing membrane is shown in Fig. 1. The optical fiber was polished down to the core on one side. Therefore, after the deposition of the sensing membrane, two optic waveguides were formed. When light propagates in an optical fiber, a fraction of the radiation extends a short distance from the guiding region into the medium of lower refractive index that surrounds it. This is the evanescent field and can be represented in the following form :24]:
A pulse width modulating system has a pulse input, a control input, and an output. The control input is used to change the pulse width of the output signal. In our proposed system, we send a light pulse through a fiber-optic based waveguide made using VOC-sensitive materials. If the overlay waveguide is exposed to VOCs, its properties, such as refractive index, change. As a result, the properties of the light passing through the waveguide also change. The time interval between the leading and trailing edges (defined at 50% of the maximum pulse amplitude) is called the pulse width. Therefore, when a light pulse is transmitted through a fiber-optic based waveguide, both the peak value of the pulse and the fall time change. This is due to the absorption of light inside the waveguide. The amount of absorption depends on the refractive index change. As a result, the pulse width of the received signal also changes. Because the pulse width depends on the absorption of light inside the waveguide, we can consider it with the pulse control input. This is because the light absorption pulse occurs owing to the change in refractive index. The width of a light pulse, , through a fiber-optic based waveguide can be written as [25–27]:
3. Experimental details
3.1 Fabrication of the side-polished optical fiber device
To fabricate a side-polished optical fiber device, we need to make a curved groove with a certain radius of curvature on a quartz block. A 160-µm-wide V-groove was fabricated on a quartz block (25 × 10 × 5 mm) using a mechanical slicer. Then, a single-mode optical fiber of core radius 3 µm and cladding radius 125 µm was placed in the V-groove. The bending radius of the fiber optic inside the quartz block was about 60 cm. Then, the fiber was glued into this groove and polished. The SEM image of the V-groove in the quartz block is shown in Fig. 2.
3.2 Fabrication of the sensing membrane
The phenomenon whereby a compound changes color, by a change in either the absorption or emission spectra of the molecule when dissolved in different solvents, is called solvatochromism. Solvatochromism is caused by differential solvation of the ground and first excited state of the light-absorbing molecule . When the excited state is more polar than the ground state, its stabilization is favored by more polar solvents. There is a decrease in the transition energy and a bathochromic shift in the spectrum (positive solvatochromism). Conversely, a more polar ground state leads to the opposite effect and a hypsochromic shift in the spectrum (negative solvatochromism) [29,30]. A change from bathochromic to hypsochromic, or vice versa, with an increase in solvent polarity, is called reverse solvatochromism. Solvatochromism can be described by both optical and electrical characteristics.
Four types of solvatochromic dyes, such as Nile red [31,32], Reichardt’s dye (R-dye) , 4-Amino-N-Methylphthalimide (4-ANMP) [33,34], and 4-Aminophthalimide (AP) [35,36] were used to fabricate the fiber-optic VOC-sensing elements. The molecular structures of the solvatochromic dyes are shown in Fig. 3. These dyes can be used to determine the polaritiesof biological objects and polymeric and organic-inorganic systems that include proteins and ormosils, in addition to being used to synthesize optical sensor materials for the detection of the vapors of various solvents.
Nile red, R-dye, 4-ANMP, AP, N,N-dimethylacetamide (DMAC) (99%), PVC, and PVP were used to prepare the sensing solution for the five sensing elements of the array. All reagents were purchased from the Sigma-Aldrich Chemical Corporation and used without further purification. The composition of the sensing solution for the five fiber-optic VOC-sensing elements of the sensor array is given in Table 1. We prepared five types of sensing solutions for the sensor elements, S1 to S5, by the following procedure. The combinations of 0.5 wt% Nile red and 46 wt% PVP; 2 wt% R-dye and 46 wt% PVP; 2 wt% 4-ANMP and 46 wt% PVP; 2 wt% AP and 46 wt% PVP; and 0.5 wt% Nile red and 35 wt% PVC were individually dissolved in 2 ml DMAC at a 1:1 volume. Then the mixtures were sonicated for 2 min to obtain the sensing solutions, for the fiber-optic VOC-sensing elements. The side-polished fiber block was washed with acetone, methanol, and distilled water and dried with N2 gas. Then, the sensing solution was deposited on the side-polished portion of the fiber-optic block by spin coating. After deposition of the sensing solution, the fiber block was dried overnight at room temperature. The fabricated fiber-optic VOC-sensing elements before and after deposition the sensing membrane are shown in Fig. 4. The thickness of the sensing membrane was several micrometers in order to avoid any unwanted environmental noiseinteraction with the evanescent field. A scanning electron microscope (Hitachi S-4800) was used to measure the thickness of the VOC sensing membrane and its value was approximately 2.65 µm which was larger than the penetration depth (approximately 1365 nm) of the evanescent field. D. S. Ballantine and H. Wohltjen proposed a fiber-optic humidity sensor, where they used a cobalt chloride containing PVP polymer sensing membrane and showed that the films of PVP alone exhibited no significant response to water vapor . Therefore, according to the references  we expect that the proposed sensing array, whose optical sensing element was made by a side-polished optical fiber covered by a solvatochormic dye containing polymer sensing membrane has no significant effect of its sensing performance as the changing of humidity.
3.3 VOC detection system
The experimental setup for the characterization of the VOC gas-sensor array is shown in Fig. 5. The VOC-sensing system consists of a flow control system, gas tanks (VOCs and N2), a test gas chamber (the dimension of the inner side 200 × 100 × 50 mm), five side-polished optical-fiber VOC-sensing elements, five laser diodes (LDs) (850 nm), a laser diode driver unit, a signal-processing unit, a multifunction data acquisition (DAQ) module (NI USB-6216 BNC), and a PC. We have designed the LD driver and the signal-processing unit by easily available and inexpensive opto-electronic components. The signal-processing unit could be divided into five parts: light detector, current to voltage converter, amplifier, pulse shaping circuit, and peak detector.
The sensor array was formed by five side-polished fiber-optic VOC-sensing elements, which contained the solvatochromic dye mixed-polymer membrane and one reference element without a polymer membrane. This reference element is used to compensate for common error sources such as environmental temperature and pressure. Therefore, the voltage measured (ΔV) for each of the five sensors was the difference between the voltage of the sensor and that of the reference device. Each sensor had its own LD switching circuit.
The square-wave modulated light of frequency 1 kHz from six identical laser sources emitting at a peak wavelength of 850 nm were coupled to the reference arm, and the five sensing arms, through the reference and VOC-sensing elements. The square wave modulator for the LDs was based on a standard NE555 timer, which operated at 2 kHz. Its output was passed through a D-flip-flop (CD4027) in order to achieve a perfect 1 kHz signal with 50:50duty cycle. The light signals from the reference and sensing arms were detected by the photodiode, and its output was connected to an operational amplifier in the current follower configuration for sufficient amplification. Then, the output of the amplified signal was given as the input to the pulse shaping circuit. The peak values of the signals were obtained from the peak detectors. The outputs of the peak detector circuits were connected to the DAQ module, which was interfaced with a computer.
The concentration of the VOC gas was controlled using a computerized mass flow controller (MFC). Five different types of VOCs, namely, dimethylamine, ethanol, benzene, toluene, and acetic acid, were used to observe the sensing performance of the sensor array. The target gas was organic gas mixed with N2 gas (99.999%) to get the desired concentration. N2 is an inert gas which does not undergo chemical reactions with the VOCs as well as the chemical substances of the sensing membrane. Most of the components of the flow control system were composed of stainless steel and Teflon, which was to prevent target gases from being absorbed by the tubes and valves. The side-polished fiber-optic VOC-sensing elements were fixed inside the test gas chamber. Under the idle condition (no VOC gas in the gas chamber), the system was adjusted in such a way that almost equal amounts of light passed through the fibers of the sensing and reference arm, and the output was set to 0 V. To test the response of the sensor, we allowed N2 gas flow to the test gas chamber to evacuate the existing air in the system and to obtain a stable baseline. Then, the target gas was flowed slowly into the chamber to obtain the response baseline. On exposure to the target gas, the sensing membranes of the sensor array came into contact with the VOC gas; consequently, the refractive index of the PWG changed, and as a results, the output voltage of the received sensing signal changed for a given concentration of gas. The signal-processing unit was connected to the PC via the DAQ module. The LabVIEW program was used to monitor the real time response of the sensor array and to record the results.
4. Results and discussion
The waveforms of the sensing and reference signals under ideal conditions (no VOC gas in the gas chamber) are shown in Fig. 6(a); there was no positive pulse width variation in the sensing signal with respect to the reference-signal pulse width. On the other hand, when the VOC gases flowed into the gas chamber, the positive pulse width of the sensing signal changed with respect to the reference-signal pulse width. The corresponding changes in the output voltages, which were taken from the output of the peak detector for the reference and the sensing signal, were measured by an oscilloscope, Tektronix TDS3032B. The pulse width and output voltages are shown in Fig. 6(b) and Fig. 6(c), respectively. Figure 6(b) shows that the sensing signal’s pulse width variation with respect to the reference pulse width(500.006415 µs) at an acetic acid gas concentration of 20 ppb is 411.23 ns for the sensing element S1. Figure 6(c) shows that the corresponding output voltages of the peak detector for the reference and sensing signals were 3.07619 and 3.07361V, respectively. The change in sensing voltage with respect to the reference voltage at 20 ppb of acetic acid was 2.58 mV. These waveform results indicated that the performance of the designed signal-processing unit was good, and that it could detect small differences in the pulse width and output voltages.
The chemical compounds that were used for the sensing membrane of the sensor array are summarized in Table 1. As can be seen, sensing elements 1 and 5 have the same dye (Nile red), but different polymers, i.e., PVP and PVC, in order to observe different sensor responses.
To observe the performance of each sensing element of the sensor array with respect to different VOCs, we injected the different VOCs individually at different concentrations into the gas chamber. The VOCs were detected at room temperature. Figure 7 shows the response of the five sensors from 0 to 50 ppb in increments of 10 ppb for dimethylamine, ethanol, benzene, toluene, and acetic acid, respectively. The response of a sensor is the difference between the obtained voltage of the sensor and that of the reference sensor. As can be seen, the relative sensing voltage increases as the concentration of the VOCs increases. According to the experimental observation, the sensing element of the array offers a linear sensing performance over the dynamic range. From the sensing response of the array, we see that the lowest detection rate corresponds to benzene gas and the highest detection corresponds to acetic acid.
The sensitivities of the five fiber-optic sensor elements (S1-S5) using a radar chart under different VOCs are shown in Fig. 8. According to the results, we have found that the first sensor, S1, which contains Nile red, presents a higher sensitivity for most of the VOCs, i.e., dimethylamine, ethanol, benzene, and acetic acid, than the other sensor elements of the array. The fourth optical sensing element, S4, which contains the AP dye sensing membrane, shows higher response to toluene than other sensing elements. The R-dye-containing sensing elements, i.e., S2, show lower sensitivities than other sensing elements.
The thickness of the sensing membrane was approximately 2.65 μm and we obtained the better sensing performance at this thickness. According to our experimental observation, we found that the sensitivity of the sensing system decreases as the thickness of the sensing membrane increases. As for example: when the thickness of the sensing membrane was 2.65 μm then the sensitivity of the sensing element S1 for the acetic acid gas was approximately 0.105 mV/ppb but when the thickness of the sensing membrane was increased to 7.78 μm then its sensitivity of the sensing element was decreased and its value was approximately 0.087 mV/ppb. The response and the recovery time of the proposed sensing array system was less than 25 s and 30 s, respectively at the room temperature. We were able to test the performance of our proposed sensing system with other sensing system which was based on the principle of wavelength shift as the sensing membrane interact with the VOCs [38,39]. We found that the proposed sensing system offers a large dynamic range, more linear sensing performance (correlation coefficient, R2 = 0.9936, approximately), short response and recovery time and better sensing stability in comparison with the wavelength shift method. For example, the response and the recovery times in  were less than 60 s, while in our proposed sensing system the response and the recovery times were found to be less than ~30 s.
To determine the power of discrimination and separation of the proposed fibre-optic VOC gas-sensor array system, a statistical technique, namely, principal component analysis (PCA) was used for pattern recognition. The PCA is a simple and effective multi-variate data analysis technique that reduces the dimensions of experimental data by translating the data into a new set of principal coordinates in the order of their variance values. The first few principal coordinates containing most of the data variance can be taken as representatives of the data set without much loss of information and despite some weaknesses, this technique allows the mapping of multi-dimensional data onto 2 or 3 axes.
First, we made a PCA of the data by collecting the maximum relative amplitudes of each sensor exposed to the sight of the VOCs recorded during 5 cycles and converted into an m × n matrix. In this case, m = 15 is the number of measurements and n = 5 the number of sensors, respectively. The average gas-sensing response of the sensor array is shown in Fig. 9(a), while Fig. 9(b) shows the two-dimensional PCA (PC1–PC2) score plot for the sensing discrimination of three VOCs in the case of the proposed fiber-optic sensor array. This result demonstrates that, the sensor-array distinguishes different types of VOCs from the response of five different sensing elements. We have grouped the three different gases in the same ellipse in order to make the separation power of the sensor array clear. PC1 can explain 70% of the variance and PC2 can explain 27% of the variance. The total accumulative variance contribution from PC1 and PC2 is 97%. This result indicates that our proposed fiber-optic VOC gas-sensor array is capable to successfully separate different types of VOCs at low gas concentrations.
In this study, we designed and developed a new type of highly sensitive VOC gas-sensor array, using the evanescent field coupling between a side-polished optical fiber and a polymer planar waveguide. Four different types of solvatochromic dyes, Nile red, Reichardt’s dye, 4-Amino-N-Methylphthalimide, and 4-Aminophthalimide, and two different types of polymers, were used to make the five different types of sensing membranes. These were deposited on the five side-polished optical fibers by a spin coater to make the five different sensing elements of the array. The sensing system was based on the principle of the pulse-width modulation technique, according to which the pulse width of the received signal, which corresponds to the average value of the received signal amplitude, changes according to the concentration of VOCs. The proposed VOC gas-sensor array has many advantages, including high sensitivity, a linear response over a dynamic range, easy fabrication, remote sensing capabilities, compactness, and low cost, on account of the circuitry consisting of easily available and inexpensive opto-electronic components. The designed gas-sensor array had a short response time and high sensitivity. According to the results, the dynamic range of the VOCs was varied from 0 ppb to 50 ppb. The response property of the proposed VOC gas-sensor array was linear and reproducibility of the sensing system was good. A statistical technique, namely, principal component analysis, was used to determine the power of discrimination and separation of the sensor array. The proposed fiber-optic sensor array successfully classified different types of VOCs from the data collected by the five VOC-sensing elements. In future work, we plan to compact the optical probe design and lower the production cost.
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea Government (MSIP) (No. 2008-0062437).
References and links
1. Indoor air quality, sources of indoor air pollution-organic gases (volatile organic compounds-VOCs) (2007): http://www.epa.gov/iaq/voc.html.
2. B. D. Gupta, Fiber Optic Sensors: Principles and Applications (New India Publishing Agency, 2006).
3. M. J. Fernández, J. L. Fontecha, I. Sayago, M. Aleixandre, J. Lozano, J. Gutiérrez, I. Gŕacia, C. Cané, and M. C. Horrillo, “Discrimination of volatile compounds through an electronic nose based on ZnO SAW sensors,” Sens. Actuators B Chem. 127(1), 277–283 (2007). [CrossRef]
4. M. A. Ryan, A. V. Shevade, H. Zhou, and M. L. Homer, “Polymer-carbon black composite sensors in an electronic nose for air-quality monitoring,” MRS Bull. 29(10), 714–719 (2004). [CrossRef]
5. F. K. C. Harun, A. M. Jumadi, and M. N. Humaimi, “Carbon black polymer composite gas sensor for electronic nose,” IJSER 2(11), 1–7 (2011).
6. C. Wongchoosuk, A. Wisitsoraat, A. Tuantranont, and T. Kerdcharoen, “Portable electronic nose based on carbon nanotube-SnO2 gas sensors and its application for detection of methanol contamination in whiskeys,” Sens. Actuators B Chem. 147(2), 392–399 (2010). [CrossRef]
7. L. J. Matthew, H. Edrees, I. Kymissis, and K. L. Shepard, “Integrated VOC vapor sensing on FBAR-CMOS array,” in Proceedings of IEEE Conference on Micro Electro Mechanical Systems (Paris, France, 2012), 846–849.
8. N. A. Pantazis, G. P. Patsis, E. Valamontes, I. Raptis, D. Goustouridis, and M. Sanopoulou, “Capacitive sensor arrays for the real time detection of volatile organic compounds,” in Proceedings of Fifth International Conference on Sensing Technology (Palmerston North, New Zealand, 2011), 422–425. [CrossRef]
9. K. S. Suslick, N. A. Rakow, and A. Sen, “Colorimetric sensor arrays for molecular recognition,” Tetrahedron 60(49), 11133–11138 (2004). [CrossRef]
10. N. A. Rakow, A. Sen, M. C. Janzen, J. B. Ponder, and K. S. Suslick, “Molecular recognition and discrimination of amines with a colorimetric array,” Angew. Chem. Int. Ed. Engl. 44(29), 4528–4532 (2005). [CrossRef]
12. Y. Q. Chen and C. J. Lu, “Surface modification on silver nanoparticles for enhancing vapor selectivity of localized surface plasmon resonance sensors,” Sens. Actuators B Chem. 135(2), 492–498 (2009). [CrossRef]
13. D. Mombello, N. L. Pira, L. Belforte, P. Perlo, G. Innocenti, S. Bossi, and M. E. Maffei, “Porous anodic alumina for the adsorption of volatile organic compounds,” Sens. Actuators B Chem. 137(1), 76–82 (2009). [CrossRef]
14. W. Ma, H. Yang, W. Wang, P. Gao, and J. Yao, “Ethanol vapor sensing properties of triangular silver nanostructures based on localized surface plasmon resonance,” Sensors (Basel) 11(12), 8643–8653 (2011). [CrossRef]
16. A. Leunga, P. M. Shankar, and R. Mutharasan, “A review of fiber-optic biosensors,” Sens. Actuators B Chem. 125(2), 688–703 (2007). [CrossRef]
18. W. G. Jung, S. W. Kim, K. T. Kim, and S. W. Kang, “High-sensitivity temperature sensor using a side-polished single-mode fiber covered with the polymer planar waveguide,” IEEE Photon. Technol. Lett. 13(11), 1209–1211 (2001). [CrossRef]
20. J. K. Yoon, G. W. Seo, K. M. Cho, E. S. Kim, S. H. Kim, and S. W. Kang, “Controllable in-line UV sensor using a side-polished fiber coupler with photo-functional polymer,” IEEE Photon. Technol. Lett. 15(6), 837–839 (2003). [CrossRef]
21. S. W. Jang, Y. H. Yun, D. E. Kim, S. J. Lim, S. Y. Park, Y. H. Lee, and S. W. Kang, “Refractive index change by photoinduction of a UV-sensitive SMF-to-PWG coupler,” IEEE Photon. Technol. Lett. 18(1), 220–222 (2006). [CrossRef]
23. Principal Components Analysis, www.stat.cmu.edu/~cshalizi/490/pca/pca-handout.pdf.
24. W. Cao and Y. Duan, “Optical fiber-based evanescent ammonia sensor,” Sens. Actuators B Chem. 110(2), 252–259 (2005). [CrossRef]
25. T. L. S, N. A. George, P. Sureshkumar, P. Radhakrishnan, C. P. Vallabhan, and V. P. Nampoori, “Chemical sensing with microbent optical fiber,” Opt. Lett. 26(20), 1541–1543 (2001). [CrossRef]
26. G. P. Agarwal, Nonlinear Fiber Optics (Springer-Verlag, 2000).
27. G. P. Agarwal, Fiber-Optic Communication Systems (John Wiley and Sons, Inc., 1992).
28. C. Reichardt, “Solvatochromic dyes as solvent polarity indicators,” Chem. Rev. 94(8), 2319–2358 (1994). [CrossRef]
29. J. F. Deye, T. A. Berger, and A. G. Anderson, “Nile Red as a solvatochromic dye for measuring solvent strength in normal liquids and mixtures of normal liquids with supercritical and near critical fluids,” Anal. Chem. 62(6), 615–622 (1990). [CrossRef]
30. E. Buncel and S. Rajagopal, “Solvatochromism and solvent polarity scales,” Acc. Chem. Res. 23(7), 226–231 (1990). [CrossRef]
31. N. A. Murugan, Z. Rinkevicius, and H. Ågren, “Modeling solvatochromism of nile red in water,” Int. J. Quantum Chem. 111(7-8), 1521–1530 (2011). [CrossRef]
32. K. A. Fletcher, I. A. Storey, A. E. Hendricks, S. Pandey, and S. Pandey, “Behavior of the solvatochromic probes Reichardt’s dye pyrene, dansylamide, nile red and 1-pyrenecarbaldehyde within the room-temperature ionic liquid bmimPF6,” Green Chem. 3(5), 210–215 (2001). [CrossRef]
33. E. W. Diana, C. Carlos, F. P. Roberto, and F. A. Pedro, “Dynamic solvatochromism in solvent mixtures,” Pure Appl. Chem. 73(3), 405–409 (2001). [CrossRef]
34. S. K. Gorbatsevich and O. Y. Smirnova, “Solvatochromic and thermochromic shifts of electronic spectra of polar solute molecules in a mixture of polar and nonpolar solvent; the role of solvent-solvent interactions,” J. Chem. Phys. 120(3), 1369–1374 (2004). [CrossRef]
36. D. Noukakis and P. Suppan, “Photophysics of aminophthalimides in solution I. steady-state spectroscopy,” J. Lumin. 47(6), 285–295 (1991). [CrossRef]
37. D. S. Ballantine and H. Wohltjen, “Optical waveguide humidity detector,” Anal. Chem. 58(13), 2883–2885 (1986). [CrossRef]
38. S. H. Yeom, H. Yuan, B. H. Kang, K. J. Kim, D. H. Kwon, S. H. Kim, and S. W. Kang, “VOC gas detection using solvatochromic dye coated sidepolished optical fiber,” in Proceedings of the 14th International Meeting on Chemical Sensors (Nuremberg, Germany, 2012), 1454–1457.
39. H. Yuan, S. H. Yeom, J. W. Lim, and S. W. Kang, “Side-polished optical fiber odor sensor for VOC detection based on solvatochromism,” Sens. Lett. 9(1), 87–91 (2011). [CrossRef]