1. Introduction

Highly sensitive gas sensors are indispensable components of environmental safety monitoring for the detection of volatile harmful gases that can induce critical damage, even with small leakages [1,2]. Recently, nanotechnology has emerged as a key enabling technology for realizing high-sensitivity gas sensors by replacing the thin film field effect transistor channel structures that present in most sensors with a semiconductor nanostructure [3,4]. However, the nanotech gas sensors are not yet ready for field applications because they tend to suffer from poor reliability, they require the delivery of local current and heating power, and their scalability with data communication or remote control technologies has been limited. In order to solve the problems, extensive nanotech-based sensing schemes with the development of multi-sensor array configurations, unique nanochannel designs, and novel semiconducting channel materials and structures have been researched [3–5]. In the field of optical gas sensors, sensor designs based on free-space optics, plasmonic resonance detection, or fiber optics have been described [6–8]. Especially, fiber optic gas sensors that employ interferometric techniques have significant advantages over the currently available semiconductor nanosensors, including (i) a small volume of operation, (ii) facile remote control, (iii) no local power requirements, (iv) insensitivity to field interference, and (v) the capacity for multiplexing designs that enable distributed sensor configurations [8–11]. Most optical gas sensors rely on a change in the effective refractive index (n_eff) that generates different phase-matching conditions in the optical modes of an interferometric configuration. However, n_eff-detecting optical gas sensors usually display insufficient sensitivity and are only useful in certain applications in which ultrahigh sensitivity is not required [8,10,12]. Optical gas sensors could potentially replace popular but unreliable nanosensors, provided a dramatic improvement in sensitivity may be achieved, for example, by using new materials and/or developing novel device designs.

In the present work, we developed an optical gas sensor for the detection of very low concentrations of nitrobenzene (NB), an explosive gas. Because of its low vapor pressure at room temperature, reliable NB gas sensors have not been available despite the need [2,12]. The sensor design described here relies on a fiber Fabry–Pérot interferometer (FFPI) fabricated from poly(4-vinylpyridine) (P4VP), which can convert molecular interactions between the polymer and the explosive gas into a defined volume expansion, which translates into changes in the phase-matching conditions of the FFPI resonance [10,13]. The sensor design ensures an ultrahigh sensitivity, relies on a simple fabrication process, and permits room-temperature operation, while retaining all of the advantages of conventional optical sensors. NB gas was detected at a concentration of 10 ppm by achieving a >40 nm shift in the FFPI spectrum. The maximum sensitivity was estimated at approximately 5 ppb, in view of the optical spectrum analyzer (OSA) resolution, 0.02 nm. The sensor could be refreshed at room temperature without the need for supplemental thermal energy.

2. Formation and characterization of an FFPI prepared using P4VP

Optical interferometric approaches depend on two effects that induce spectral changes: the refractive index (n_eff) and the physical length of the optical path (L). The relationship between the spectral change and the FFPI parameters is denoted Δλ = f (n_eff,0ΔL + Δn_effL₀), where Δλ, n_eff,0, Δn_eff, L₀, and ΔL are the spectral change, initial effective refractive index, index change, initial optical path length in the FFPI cavity, and the length change, respectively. Previous interferometric sensors have relied on n_eff modulation as a result of target gas molecule adsorption, which increases the average material density around the interferometer [10–13]. These designs are not very sensitive to small changes in the refractive index, increasing the challenges associated with fabricating highly sensitive fiber optic gas sensors. The physical length changes in an interferometer provide an alternative basis for high-sensitivity measurements. Such designs may be achieved using materials and/or structures that provide a chemomechanical operation. P4VP can undergo a significant volume expansion upon absorption of a gas, thereby providing a dramatic change in the phase matching conditions of a P4VP-incorporated FFPI and substantially improving the sensitivity of the fiber optic sensors [13]. The selective absorption of NB by P4VP followed from the favorable mixing interactions between NB molecules and the P4VP polymer chains [14]. Polymer film swelling was thermodynamically governed by the intermolecular interactions between the gas and the polymer chain, which could be predicted from the interaction parameter (χ), with consideration for the complementarity of the solubility parameters [15]. The interaction parameters between a polymer and a gas (χ_p-v) are given by (V_m/RT)(δ_p–δ_v)² where V_m, δ_p, δ_v, are the molar volume, and the solubility parameters of the polymer and the solvent vapor, respectively. According to the Flory–Huggins theory, small χ_p-v values induce effective swelling of a polymer by the vapor [16]. Previous studies reported that the average solubility parameter of P4VP (δ_P4VP) was 22 MPa^1/2, and NB had a δ_NB of 20.5 MPa^1/2, much closer to δ_P4VP than to water (δ_H2O is 47 MPa^1/2) [17–19]. These results suggested that the intermolecular interactions between P4VP and NB were more favorable than the corresponding interactions with moisture. These results agreed with previous experimental evaluations by Tenhaeff et al. of the χ_p-v values for P4VP and several solvent vapors. In such studies, P4VP and a nitroaromatic solvent showed smaller values of χ_p-v than were observed in water or ethanol [20]. P4VP could, therefore, provide a high selectivity and detection sensitivity by absorbing and detecting NB under ambient conditions.

A schematic illustration of the operation of a P4VP-based FFPI is presented in Fig. 1 . The silica/P4VP and P4VP/air interfaces formed a FFPI by acting as partial mirrors, m₁ and m₂, from which the incident broadband light was reflected. As illustrated in Fig. 1(a), in the absence of the NB gas, the interference between the reflected beams corresponded to survival of specific selected wavelengths, R₁ and R₂. The spectral singularity depended on the geometric and density factors of the FFPI. As the polymer layer underwent a chemomechanical transition upon absorption of the NB gas, the physical length of the light path increased from L₀ to L₀ + ΔL via conformational changes in the P4VP polymer chains. The surviving wavelengths then shifted to R₁’ and R₂’ with the new set of mirrors m₁’ and m₂’, as illustrated in Fig. 1(b). This spectral shift was sensitive to the presence of NB gas.

 

Fig. 1 Schematic illustration of the FFPI sensing mechanism based on a chemomechanical P4VP layer. (a) The initial FFPI conditions provided a reference reflected spectrum with a fixed periodicity determined by the interference between the light beams reflected from multiple mirrors. (b) The interaction of NB molecules with the P4VP chains expanded the volume of the P4VP layer to alter the phase-matching conditions in the FFPI. The periodic spectrum then shifts to a longer periodicity, resulting in a spectral red-shift in the individual peaks. The shifts indicate the presence of gas molecules.

Download Full Size | PDF

Figure 2 shows that the advantages of this sensor design include the simplicity of the device structure and fabrication process, and the high reliability of the sensing operation. Dimethylformamide (DMF) was used as a solvent to dissolve the P4VP powder to a concentration of 14.5 wt%. The high solubility of P4VP precluded the need for additional chemical or physical treatments, thereby simplifying the process. As shown in the Fig., the FFPI resonance cavity was fabricated by drop-casting 0.2 μl of the P4VP layer onto the end facet of an optical fiber ferrule. Prior to application of the P4VP drop, the surface of the fiber ferrule was cleaned with ethanol and acetone to prepare a physically homogeneous P4VP/SiO₂ interface. After coating, the device was heated at 80°C for 2 h to remove any remaining solvent. The device was subsequently aged at room temperature for 4 h to complete the preparation of an optical mirror surface morphology on the P4VP layer.

 

Fig. 2 Experimental procedure used to prepare the FFPI sensor head, highlighting the simplicity and reliability of the design.

Download Full Size | PDF

The swelling of P4VP in the presence of NB reflects the P4VP chain rearrangements induced upon intermolecular diffusion, which disrupted the weak π-π interactions among the P4VP chains. These effects were consistent with the Fourier Transform Infrared Spectroscopy (FT-IR) analysis. As shown in Fig. 3(a) , the P4VP peaks observed in the absence or presence of NB were preserved, and no new peaks were generated by, for example, the formation of chemical bonds between the P4VP and NB. The FT-IR curve of the NB-absorbed P4VP layer presented major peaks at 2931 cm^–1, 1598 cm^–1, 1416 cm^–1, and 996 cm^–1, corresponding to the P4VP-only peaks. The NB peaks at 1522 cm^–1, 1346 cm^–1, and 680 cm^–1 were attributed to N = O nitro, C-N amine, and C-H bond stretches, respectively. These results indicated that the NB-P4VP interactions were not associated with any critical energy barriers to absorption or evaporation. As a result, the detection and refreshing of the NB gases were fast and reversible processes at room temperature. The thickness change in the P4VP layer resulted from the volume expansion, as verified using field-emission scanning electron microscope (FE-SEM) under low vacuum conditions (0.8 ~1.2 Torr) as shown in Fig. 3(b)–3(d). Figure 3(b) shows the as-coated P4VP layer on a fiber ferrule, the end facet of which was treated with optical polish. A magnified view (see Fig. 3(c)) of the layer revealed an initial thickness of 44.6 μm prior to gas absorption. Exposure of the P4VP layer to the NB gas resulted in an increase in the thickness to 47.3 μm corresponding to the layer swelling process (see Fig. 3(d)). Unfortunately, the precise evaluation of the NB concentration was hindered by the sample loading process under the vacuum conditions of the SEM chamber. Spectroscopic ellipsometry was combined with a home-built NB gas chamber to facilitate the collection of information about the P4VP layer upon NB absorption. The sample was prepared by drop-casting the P4VP solution (14.5 wt%) onto a SiO₂ wafer with the final sample size of 1.5 × 1.5 cm. The sensing measurements were performed in the presence of 680 nm light over 40 min. The properties of the P4VP layer were monitored by tracing the changes in both the refractive index and the optical path with respect to NB gas absorption, as shown in Figs. 3(e) and 3(f), respectively. The insets of Fig. 3(e) show the NB gas chamber and the experimental setup in the chamber. Importantly, the negative refractive index change of the P4VP was not observed to change upon volume expansion. The initial refractive index of the layer was 1.056 with a thickness of 3.00 μm. The changes in the thickness, determined by ellipsometry analysis, were very small due to a very thin initial thickness of the P4VP layer. The P4VP layer displayed compressive stress upon NB absorption, as indicated by the positive refractive index change.

 

Fig. 3 Characteristics of the P4VP layers. (a) FT-IR curves of a pure P4VP layer, an NB gas, and a NB + P4VP layer, demonstrating that the volume expansion arose from the ‘physical mixing’ among molecules. SEM images of (b) a P4VP-coated fiber ferrule, (d) a P4VP layer prior to NB absorption, and (d) a P4VP layer after swelling. The images provide direct support that the P4VP volume expanded in the presence of the NB gas. Ellipsometric analysis characterized the changes in the (e) refractive index and (f) physical thickness (beam path length). The insets in (e) show an image of the NB gas chamber and a schematic illustration of the experimental set-up.

Download Full Size | PDF

3. Highly sensitive optical gas sensors for nitrobenzene

The end surface of an optical fiber ferrule was cleaned with ethanol and acetone, and 0.2 μl of the prepared P4VP solution was drop-casted onto the cleaned surface using a micro pipet. Solvent was eliminated by drying the sample for 4 h at room temperature, followed by heating at 80°C for 2 h in an oven. The sensor head was inserted into a tube-type gas chamber connected to the NB gas supply. The sensing operation was demonstrated using an erbium-doped fiber amplifier (EDFA) as a broadband light source. A three-port optical circulator was connected to the light source, sensor head, and OSA, thereby guiding the incident broadband light into the FFPI-loaded sensor head. Reflected light from the sensor head was redirected to the OSA and used as a reference in determining the FFPI spectral shift. Quantitatively controlled NB gas was supplied to the chamber using an NB gas bubbler to ensure a constant gas concentration. The sensor head was inserted into the chamber, which was then sealed to create isolated conditions. Figure 4(a) shows the wavelength shifts due to the volume expansion of P4VP as a result of NB absorption in the presence of 10 or 13 ppm NB. The response time of the sensor was inversely proportional to the NB concentration because the NB concentration gradient between the surroundings and the P4VP layer determined the conditions of both absorption and diffusion of the NB molecules. Both concentrations of NB gas yielded spectral shifts in the FFPI that exceeded 40 nm. The irregular curves originated from the surface distortions of the P4VP layer that formed during the volume expansion process. These spectral distortions may potentially be reduced by optimizing the device processing conditions. Figures 4(b) and 4(c) show the optical spectral shifts observed within the time window 18–20.5 minutes, as shown in Fig. 4(a), for a gas concentration of 10 or 13 ppm, respectively. Note that a trade-off must be made between sensitivity and the sensing range. This permitted customization of the P4VP layer thickness for a specific sensing frame and with a designed sensitivity. In view of the scanning resolution of the OSA (2 × 10⁻² nm) and the detected spectral shift (>40 nm), the maximum sensing resolution may potentially be less than 5 ppb. Modulation characteristics of the sensor in the time domain, are shown in Fig. 4(d). These results indicated that the sensing and refreshing processes were reversible at room temperature. The measurement recording was commenced as soon as the wavelength shift began, upon exposure of the sensor head to the NB gas. The first wavelength shift during the first ~4 min was 0.8 nm. In the absence of external mechanical, thermal, or chemical energy supplies, replacement of the NB gas with pure N₂ gas refreshed the sensor head completely within ~3 min. The surviving wavelength subsequently returned almost to the initial levels, verifying the reversible operation of the all-fiber gas sensor. Figure 4 shows the results of the modulation test, in which the iteratively modulated sensor spectral shift was higher than that obtained from an on/off experiment, suggesting good sensitivity. The NB absorption process disrupted P4VP stacking during subsequent iterative modulation steps, thereby reducing the ‘motional barrier’ to the NB molecules. The P4VP layer, therefore, further increased the degree to which it absorbed. After six iterations, the spectral shift was almost twice the value of the first iteration, displaying a trend toward saturation. Because NB saturation was not observed within a single on/off modulation curve, the wavelength shift range may potentially be extended to both higher NB concentrations and to a wider temporal window for sensing.

 

Fig. 4 The optical gas sensors detected the degree of NB absorption. (a) Magnified views of the spectral shifts produced by the 10 or 13 ppm NB gases. Magnified views of the spectral shifts obtained upon absorption of (b) 10 ppm or (c) 13 ppm NB gas. (d) Modulation characteristics of the sensor upon absorption of 13 ppm NB gas. The sensing and refreshing were operated at room temperature.

Download Full Size | PDF

The data reported here demonstrates excellent sensor sensitivity, room-temperature sensing operation, and a short sensing time in an all-fiber optical sensing device for detecting the presence of explosive gases.