1. Introduction

Motivated by the increasing environment and health concerns, there are intense efforts to develop a desirable sensor with capability of rapid, selective and sensitive detecting of chemical vapor. Current detection mechanisms [1–3] rely mainly on the interaction between target vapor and permeable material. Among them, the optical mechanism seems to be the most promising one by virtue of electromagnetic interference immunity, simple architecture and compact size. To date, various optical sensing structures, including fiber Bragg grating [4], micro-resonator [5], Fabry-Pérot interferometer [6] and planar waveguides [7,8], have been reported. In these structures, upon the diffusion of chemical vapor, the polymer layer will undergo a combined change of refractive index (RI) and thickness, which leads to a detectable optical signal, i.e., spectral shift [4,5] or light intensity [3, 6–8]. For instance, by monitoring the whispering-gallery mode spectral shift, the detection limit (DL) of about 200 ppm ethanol in the optofluidic ring resonator was achieved [5]. However, the precise determination of spectral shift requires an ultrahigh-resolution spectrometer and the explicit distinction of light intensity variation needs a subtraction of the light source fluctuation.

The phase of the light wave is another kind of optical signal responding to the vapor-polymer interaction. Moreover, the Goos-Hänchen (GH) shift [9], which refers to a lateral discrepancy of the reflected light point from its incident counterpart, is found to be proportional to the first derivative of light phase. The GH shift is usually of the wavelength scale, and several models have been proposed to enhance the magnitude with the assistance of dispersive materials [10] or structural resonances [11]. Since the material and structural parameters can exert a significant influence on the GH shift, it is an excellent approach to carry out the sensing performance [12,13] by monitoring the GH shift signal. Our previous work has demonstrated that a millimeter-scale GH shift in a symmetrical metal-cladding waveguide (SMCW) can provide a temperature resolution of approximately $5\times {10}^{-3}$°C [14]. Here, we present an optical chemical vapor sensor based on the SMCW structure, in which an amorphous Teflon AF polymer layer is added into the guiding layer, to detect toluene and benzene with high sensitivity by monitoring the enhanced GH shift. The obtained relative GH shift signal is position encoded, making fluctuations in input power irrelevant and any complicated optical equipment unnecessary. Experiment results imply that the relative GH shift has a linear relation with respect to the toluene and benzene vapor concentration. In addition, our sensor is a versatile platform for detecting other chemical vapors because any polymer can be employed in the SMCW structure regardless of the polymers’ RI.

2. Structure and principle

The schematic layout of the SMCW based chemical vapor sensor is shown in Fig. 1. A 0.2 mm thin glass slab is successively coated with a 36 nm gold film and a 1.5 $\mu m$amorphous Teflon AF 1601 polymer layer. The gold film is prepared by the sputtering technology in a vacuum chamber. The polymer layer is fabricated by the spin-coating method and then dried in an oven with 70 °C for 5 h. Teflon AF 1601 polymer is selected due to its excellent characteristics of easy synthesis, optical transparency and specific interaction with toluene or benzene vapor. A 300 nm thick gold film is sputtered on a thick glass substrate. Two identical C-shaped glass plates are set a certain distance apart from each other to form two 0.2 mm channels serving as the vapor inlet/outlet. The vapor room with 5 mm radius and 0.6 mm depth is manufactured by sandwiching the C-shaped glass plates between the two former metal-cladding components. All elements are attached together by the optical cement technology to guarantee a good parallelism and thus perform a perfect detection [15].

 

Fig. 1 Schematic layout of the optical chemical vapor sensor based on the enhanced GH shift in the SMCW, where an amorphous Teflon AF polymer layer is added into the guiding layer.

Download Full Size | PDF

Actually, this SMCW structure consists of three key components: the thin gold film acts as the coupling layer and the thick gold film works as the substrate, while the polymer layer and the vapor room jointly function as the guiding layer. Symbols $n_j$ (${\epsilon }_j$) and$d_j$ represent the RI (dielectric coefficient) and thickness, respectively, and $j=1,2,3,4,5$ refer to the thin glass slab, thin gold film, polymer layer, chemical vapor, and thick gold film, respectively. As shown in Fig. 1, the incident light can be conveniently coupled into the guiding layer through the free-space coupling technology [16] regardless of the polymer’s RI. This striking feature significantly extends the sensor versatility for detecting other chemical vapors and the reason is that the real part of ${\epsilon }_2$ is negative in the visible and near-infrared regions. Once the phase-mating condition is satisfied, the so-called ultrahigh-order mode [17] will be excited in the guiding layer. The characteristic equation for a four-layer waveguide can be expressed as [18]

According to the stationary-phase approach [9], the GH shift is proportional to the first derivative of reflected phase $\varphi$ with respect to the effective RI $N$, the expression is given by

 

Fig. 2 (a) The simulated reflectivity (solid curve) and the reflected phase (dashed curve) with respect to the effective RI with $n_3=1.31$and $n_3=1.31+4e-4$. (b) Profiles of the Gaussian incident beam and the corresponding reflected beam. The incident angle is $\theta ={4.32}^{\text{o}}$, the waist radius is 800 $\mu m$. Perpendicular dashed lines represent the magnitudes of the GH shift.

Download Full Size | PDF

In the detection procedure, once the chemical vapor is injected into the vapor room, the RI of both the vapor room and the polymer layer will change according to the Lorentz-Lorenz equation [19], meanwhile, the thickness of polymer layer will increase and that of the vapor room will accordingly decrease due to the swelling effect of the polymer layer. These above vapor-polymer interaction induced changes lead to a variation in the effective RI [8]

3. Experimental results

The experimental arrange for chemical vapor detection based on the SMCW-enhanced GH shift is drawn in Fig. 3(a). An aperture with a diameter of 0.5 mm, a polarizer, a beam splitter and another same aperture are subsequently placed about 0.3 m apart. The incident light from a tunable laser passes through them to be TE polarized and further collimated. One part of the incident beam is fed to a wavemeter for the wavelength measurement, and the other part is incident onto the top surface of the SMCW, which is encased in an airtight container and then firmly mounted on a computer controlled $\theta /2\theta$ goniometer. As shown in Fig. 3(b), the airtight container composes of two 3 cm thick aluminum plates and a rubber O-ring, and all of them are rigidly tightened together by four screws. There are a rectangular light window in the top aluminum plate for facilitating the light coupling and two circular holes in the bottom aluminum plate for aiding the vapor flowing. The polymer-coated SMCW is sticked onto the aluminum plates by the UV-curable glue to ensure that the sensing system has excellent air tightness. While the angular scan is carried out, the intensity of the reflected beam is detected by a photodiode (PD) and a series of resonance dips corresponding to the excited guide modes are recorded. The incident angle is eventually settled at the midpoint on the falling edge of one selected ultrahigh-order mode ($\theta ={4.32}^{\text{o}}$), since this position can offer a quasi-linear detection of chemical vapor concentration. During our experiment, the GH shift of the reflected beam is measured by a position sensitive detector (PSD), which is replaced the PD location in the optical path. Moreover, the room temperature is held constant (at 11.2 °C), because any minute variation of the RI and thickness in the guiding layer induced by the thermo-optic and thermal expansion effects will easily give raise to a dramatic change of the GH shift [14].

 

Fig. 3 (a) Experimental arrangement for detecting of chemical vapor. PSD: position sensitive detector, PD: photodiode; (b) schematic of the airtight container; (c) device for generating, mixing and delivering of chemical vapor.

Download Full Size | PDF

A two-arm gas flow device for generating, mixing and delivering chemical vapor is illustrated in Fig. 3(c). The carries gas used in our experiment, i.e., the dry air, is obtained by filtering the compressed air through a PALL Dominator system. The saturated chemical vapor is generalized in arm b by passing the dry air through the liquid adsorbate contained in a thermostated bubbler. The resulted vapor pressure in the bubbler can be evaluated by using the Antoine equation [21]. The chemical vapor concentration is adjusted by diluting the saturated vapor (arm b) with the dry air (arm a) at different flow rates. The vent of the tube mixer is connected to the vapor inlet of the aluminum container by a plastic pipe. A switching valve in arm b can be used to quickly switch between the vapor tube (arm b) and the dry air tube (arm a). Before and after each detecting of chemical vapor, dry air is applied to establish the sensing baseline and purge the chemical vapor, respectively.

Figure 4(a) plots the relative GH shift obtained at equilibrium for various concentrations of toluene and benzene vapors. As expected, the relative GH shift increases monotonically toward higher values with an increase of the chemical vapor concentration and the results show a good linear relationship between them. It indicates that the experiment agrees qualitatively with the above theoretical prediction. The slope of the fitting line for toluene detection is steeper than that of benzene. A PSD with a resolution of 1.5 $\mu m$ is employed [13], therefore the DL of 9.5 ppm for toluene and 28.5 ppm for benzene have been achieved. The insets of Fig. 4(a) are two representative real-time records of the relative GH shift when 867 ppm toluene and 5015 ppm benzene were imported into the vapor room of the SMCW. Upon diffusing of chemical vapor, the relative GH shift increases immediately and the parameter $t_{90}$, defined as the time taken during the response for the GH shift signal to reach the 90% of its final equilibrium value, are about 5 s for toluene and 9 s for benzene, respectively. The relative GH shift eventually returns to the initial value ($t_{10}$are 22 s for toluene and 35 s for benzene, respectively) after switching the gas valve in arm b. These temporal results indicate that the vapor-polymer interaction is prompt and recoverable. The longer time of purge process than diffusion process could be explained that the interaction between vapor molecules and polymer is stronger than that of dry air between vapor molecules. To investigate the reproducibility of the sensor, the vapor room of SMCW was repeatedly exposed to 867 ppm toluene vapor, Fig. 4(b) illustrates that the recovery abilities are not reduced after several sensing cycles. In other words, our SMCW structure is not a disposable chemical vapor sensor but can be continuously reused.

 

Fig. 4 (a) Experimental results of the GH shift vs. the concentrations of toluene and benzene. Insets are two representative sensorgrams via monitoring the relative GH shift in real-time. (b) Three-cycle responses of the GH shift to 867 ppm toluene vapor.

Download Full Size | PDF

4. Conclusion

In conclusion, a high-sensitivity chemical vapor sensor based on the enhanced GH shift in a polymer-coated symmetrical metal-cladding waveguide is theoretically proposed and experimentally demonstrated. Theoretical analysis shows that the phase-matching condition of the ultrahigh-order modes can be changed by the swelling effect and refractive index variation in the polymer layer, which is caused by the vapor-polymer interaction, and that the relative GH shift responds linearly to the chemical vapor concentration. The experiment results agree quantitatively with the theoretical prediction, yielding a DL of 9.5 ppm for toluene and 28.5 ppm for benzene. There is no need to employ any complicated optical equipment and servo techniques, since our transduction scheme is irrelevant to the light source fluctuation. Moreover, any polymer can be employed regardless of the polymers’ RI, therefore our sensor structure could have potential applications in environmental monitoring for detecting other chemical vapors.