1. Introduction

Optical gas sensors have been attracting lots of interests for their broad applications in fields of industry, environment, medical care, homeland security, and battlefield. Sensing techniques applied in those optical sensors include surface plasmon resonance (SPR) sensor [1], ring resonant sensor [2–5], long period grating sensor [6,7], fiber Bragg grating sensor [8], and D-shaped grating sensor [9,10].

FP interferometers are another important technology commonly employed for gas sensing [11], as well as for biological [12], temperature [13], strain [14], and humidity [15] sensing. They are compact, relatively easy to fabricate, and yet offers considerable sensitivity. To date, there are mainly two kinds of fiber based FP cavities - extrinsic and intrinsic. An extrinsic FP cavity is usually made of the cleaved endfaces of two fibers [16–19], which requires stringent alignment of fibers. Therefore, more efforts have been focused on developing the intrinsic FP cavity that is fabricated on a single fiber. In recent years, various types of intrinsic FPs have been fabricated by various methods, including introducing a sealed air gap near a fiber tip [20, 21], femto-second laser micromachining a micro-notch into a fiber [22–25], and depositing a thin layer on a fiber endface [26–30]. However, the sealed air gap is impossible for vapor analytes to access, while the micro-notch in a fiber may suffer from the drawbacks of mechanical fragility, lengthy fabrication processes, difficulty in maintaining the optical alignment, and large signal loss due to the scattering from the rough surface.

In contrast, the thin film deposition method meets the gas sensing requirements quite well. It not only enables easy access to gas analytes, but also provides a robust structure and comparatively smooth cavity surfaces. Various approaches are under investigation to coat a thin layer directly on the fiber endface, for instance, thin film growth [26], electrostatic self-assembled monolayer (ESAM) method [27,28], and dip coating method [29,30]. In those configurations, the FP cavity is formed due to the refractive index (RI) contrast at the fiber-film interface and the film-air interface and the interference spectrum is monitored when sensing events take place.

Since different polymers (such as polar and nonpolar polymers) may have drastically different response to vapor analytes, it is very common for a gas sensor to incorporate a matrix of polymers to enhance the vapor detection specificity. Therefore, it is highly desirable that the FP cavity is able to accommodate various polymers that may have a wide range of RIs. Using the existing direct deposition methods described above, the FP cavity performance degrades tremendously when the polymer RI approaches that of the fiber, resulting in nearly indiscernible interference fringes, thus significantly limiting the selection of the polymers.

In this paper, we develop a versatile, easily-fabricated fiber-based FP gas sensing probe that can accommodate any polymers regardless of their RIs. The sensing probe is illustrated in the enlarged part of Fig. 1. It is composed of two layers: a silver layer and a vapor-sensitive polymer layer. Light propagating in a single mode fiber will be partially reflected at the silver layer and the polymer-air interface. These two reflected beams generate interference spectrum. When the sensing probe is exposed to analyte vapor, the vapor–sensitive polymer layer will interact with analyte, and the change of its RI or thickness will change the light path, which in turn causes the interference spectrum to shift. By artificially introducing this reflective silver layer, we are able to coat a polymer of any RI, thus tailoring the gas sensor for versatile usage. In our experiment, we use polyethylene glycol (PEG) 400 with the RI close to that of glass (RI=1.465-1.469) and Norland Optical Adhesive (NOA) 81 (RI=1.53-1.56) as a model sensing polymer to demonstrate the feasibility of the proposed FP gas sensor. We further analyze the different responses of this FP sensor with different analytes. The sensor response to both continuous flow and pulsed flow is also studied.

 

Fig. 1. Experimental setup

Download Full Size | PDF

2. Experiment

2.1. Sensor probe preparation

The fiber probe is prepared by coating a thin layer of silver on the endface of an optical fiber, followed by depositing a layer of the vapor-sensitive polymer. For the silver coating, we use the electroless plating method [31]. First, a single mode fiber (SMF-28) is cleaved and cleaned by deionized (DI) water. Then, it is sensitized and catalyzed by immersing it into sensitization and catalyst solution for 3 minutes, respectively. Sensitization solution is a mixture of tin chloride (10 g/L) and hydrochloric acid (40 mL/L) in DI water, whereas catalyst solution is a mixture of palladium chloride (0.5 g/L) and hydrochloric acid (40 mL/L) in DI water. After sensitization and catalyzation, the fiber is immersed in the plating bath containing silver nitrate (0.002 g), ammonia in water (20%) (50 μL), hydrazine hydrate (1 μL), ammonium carbonate (0.02 g), and 1.6 mL of DI water. After approximately 2 minutes of plating, the fiber is cleaned with DI water and air dried for the subsequent polymer coating. The silver coating prepared in this way has thickness on the order of tens of nanometers and allows 10-20% light intensity to be reflected. Since silver has a very low RI (complex RI=0.3+11j at 1550 nm), small thickness variations in the silver layer do not significantly change the optical path of the FP probe.

In our experiment, we employ two vapor sensitive polymers, PEG 400 and NOA 81. Both polymers have been used extensively for chemical detection [32,33]. They are chosen to simulate the situation where the polymer RI is nearly the same as or dissimilar to that of the fiber, and to represent vapor-sensitive polymers with high (PEG 400) and low (NOA 81) polarity. The polymer layer is deposited by the dip coating method, where the silver-coated fiber is immersed in polymer solution for 5 minutes. PEG 400-coated probe is dried in a vertical position at room temperature for 2 hours, and NOA 81-coated probe is cured slightly with ultraviolet light (365 nm) for 10 seconds. The resultant thickness of the polymer is usually on the order of tens of micrometers.

2.2. Experimental setup

The experimental setup is presented in Fig. 1. The sensing probe is placed in a capillary of 900 μm in inner diameter that serves as the gas fluidic channel. For continuous flow mode experiments, various concentrations of air/analyte mixture are flowed continuously through the capillary by a syringe pump at a flow rate of 0.5 mL/min. In pulsed mode experiments, the analyte is extracted from the headspace of the sample vials using a solid-phase microextraction (SPME) fiber and then injected through a gas chromatography (GC) injector, thus forming a vapor pulse that travels along the capillary. Helium is used as the carrier gas and purging gas. A three-port valve is used to switch between helium gas and analyte. A tunable diode laser (Optical Sensing Interrogator sm125, Micron Optics) is scanned at 2 Hz within a wavelength range of 1510–1590 nm with a spectral resolution of 1 pm. The light reflected from the sensor probe is acquired by the photo-detector on the Micron Optics laser. The corresponding interference spectrum is recorded for post-analysis.

3. Results

The advantages of using a silver layer are clearly demonstrated by Fig. 2, which shows the interference spectra of both fiber-silver-polymer structure and fiber-polymer structure. With the silver coating, both interference spectra (Fig. 2(A) and (B)) from PEG 400-coated and NOA 81-coated sensing probes are clean, stable, and have large contrast. However, no distinctive interference spectrum is observed with the PEG 400 coating in the absence of the silver layer (Fig. 2(C)). With the NOA 81 coating, the interference spectrum (Fig. 2(D)) is noisy and contrast is low due to the low reflectivity at the fiber-polymer interface, which results in a deteriorated detection limit in the sensor. Based on the free spectral range in Fig. 2(A) and (B), the coating thickness is estimated to be 30.3 μm and 29.2 μm for PEG-400 and NOA 81, respectively.

 

Fig. 2. Reflection spectrum of the FP gas sensor coated with PEG 400 (A) and Norland Optical Adhesive 81 (B). Reflection spectrum of a bare fiber in absence of the silver layer coated with PEG 400 (C) and Norland Optical Adhesive 81 (D).

Download Full Size | PDF

3.1. Continuous flow mode

Before the introduction of the analyte, air is flowed through the capillary and no spectral change in the interference spectrum is observed, indicating that humidity in air does not have any impact on the gas sensor. Two representative sensorgrams of the proposed sensor are shown in Fig. 3 by monitoring the interference spectral peak position in real time. Fig. 3(A) is the sensorgram when PEG 400-coated sensor is exposed to methanol vapor, and Fig. 3(B) is when NOA 81-coated sensor interacts with acetone vapor. The interference spectrum shifts to a longer wavelength upon the interaction between analyte vapor and the polymer, and reaches the maximum equilibrium value in approximately 150 and 50 seconds for PEG 400 and NOA 81, respectively. The spectrum then completely returns to the baseline after purge, indicative of complete removal of vapor molecules from the polymer layer.

 

Fig. 3. Sensorgrams obtained by monitoring the interference spectral shift in real time when sensor probe is exposed to 3410 ppm of methanol (A) and 75500 ppm of acetone (B) vapor. Sensor is coated with PEG 400 (A) and NOA 81 (B).

Download Full Size | PDF

Figure 4 depicts the wavelength shift at equilibrium for different concentrations of hexanol, methanol, and acetone vapors for PEG 400 (Fig. 4(A)) and NOA 81 (Fig. 4(B)) coated sensing probes. The wavelength shift is almost linear to the concentration of analyte vapor for both PEG 400 and NOA 81. The gas sensor is highly sensitive, because light travels through the whole vapor-sensitive polymer layer, and thus has full interaction with analyte in the polymer. According to Fig. 4(A), the sensitivity of methanol vapor is 3.53 pm/ppm. Using Δ(nL)/(nL) = Δλ/λ, this sensitivity corresponds to a fractional change of 2.3×10^-6 in the polymer optical path per ppm methanol, which is higher than that for hexanol and acetone (~ 1.1 pm/ppm), because methanol has a higher polarity than hexanol and acetone, and thus, it is more soluble in highly polar PEG 400 polymer. These sensitivities obtained above are similar to what were reported in earlier studies [1,2,4,5]. In comparison, NOA 81 is a thiolene-based polymer and has less polarity than PEG 400. As consequence, NOA 81 sensitivity to methanol is only 0.1 pm/ppm, 30 times lower than PEG 400. To estimate the detection limit of the FP gas sensor, we assume that the resolution of wavelength shift is 1 pm. Hence, the detection limit for methanol vapor can potentially be on the order of 1 ppm for PEG 400 and 10 ppm for NOA 81.

 

Fig. 4. Sensor response to various concentrations of hexanol, methanol, and acetone vapor. The sensor is coated with PEG 400 (A) and NOA 81 (B).

Download Full Size | PDF

3.2. Pulsed mode

To test the rapid response of our sensor probe, which is important for many applications where rapid gas detection is critical, we use a SPME fiber to pick up methanol vapor and then release it at a GC injector to introduce methanol vapor pulses. The carrier gas (helium) flow rate is maintained at 19 mL/min. Fig. 5(A) presents the quick response of PEG 400-coated sensing probe for methanol pulses of different mass loadings. The time response is around 10 seconds, which is much faster than that in the continuous flow mode. The spectrum shift increases with the increased amount of analyte injected. This shows that the FP sensor probe can potentially be used as GC detector for rapid analysis of the vapor molecules flowing through the GC column.

However, our current design is far from its full potential as a rapid sensor. According to our earlier investigation, the pulse width of the methanol less than one second is expected when a pulsed methanol vapor is injected and travels along the PEG 400 coated capillary [34]. The much slower response strongly suggests that the polymer thickness plays an important role in determining the sensor response. To verify this, we carry out a similar experiment with a much thicker polymer layer. The sensor response is shown in Fig. 5(B). Fig. 6 compares the peak response and the total area, which is proportional to the analyte mass interacting with the polymer, for the two polymer thicknesses, showing that the peak response and the total area for the thicker polymer are consistently lower than those for the thinner polymer. This discrepancy can be accounted for by considering the diffusion of the vapor molecules into the polymer. Within the duration of the molecule pulse, the vapor molecules can diffuse into a certain thickness of polymer. Therefore, only a partial sensitivity of the FP sensor probe can be achieved, in contrast to the continuous flow case where the vapor molecules occupy the whole polymer and the full sensitivity of the FP is realized. For estimation purposes, we assume that pulse width is 1 second and the molecule diffusion constant is 10^-10 cm² s^-1. In order to achieve the full sensitivity, the vapor molecules should diffuse through the whole polymer within the duration of the vapor pulse, which requires that the polymer thickness be less than 100 nm.

 

Fig. 5. Detection of methanol vapor pulses with various SPME sampling times when the analyte is injected in a pulsed mode. Polymer (PEG 400) thickness is 18.6 μm (A) and 31.5 μm (B), respectively.

Download Full Size | PDF

 

Fig. 6. Wavelength shift (A) and peak area (B) vs. SPME extraction time for two different PEG 400 coating thicknesses in Fig. 5.

Download Full Size | PDF

4. Conclusion

We have developed a versatile fiber-based FP gas sensor that is easy to fabricate and highly sensitive. This FP gas sensor can employ vapor-sensitive polymer with any RIs, and hence significantly broadening its applications in various areas. The sensitivity of the sensor for methanol vapor is 3.5 pm/ppm and 0.1 pm/ppm, with the detection limit of 1 ppm and 10 ppm, for PEG 400 and NOA 81, respectively. We also analyze the sensor response to two vapor delivery modes: continuous flow mode and pulsed mode; and the rapid response in the pulsed mode suggests that it has great potential for a GC system.

Future work will be focused on optimizing the performance of the FP gas sensor by increasing the Q-factor of the FP cavity to improve the sensor spectral resolution and hence the detection limit, miniaturizing the sensing probe, and improving the fluidics to achieve faster detection in the continuous flow mode. In addition, a matrix of polymers will be employed on an array of FP probes having different responses towards different analytes to enhance the vapor detection specificity. We will further incorporate the sensor probe coated with a much thinner polymer layer into a micro-GC system [34] for rapid separation and detection of vapor molecules. Finally, detection of actual gas analytes relevant to industry, environmental protection, and homeland security, etc. using our gas sensor will also be carried out, in which we will take full advantages of its flexibility in polymer selection by choosing the optimal polymer for maximal response to the analyte of interest without concern about the polymer RI.