Gas sensors are an important branch of sensor technology, which have been widely used in various fields, including air pollution monitoring, exhaled breath tests in clinical diagnosis, and drug and explosive detection. Gas sensors are mainly classified into electrical and optical types, the former also being called electronic noses, including metal oxide semiconductor gas sensors, electrochemical gas sensors, and catalytic combustion gas sensors. Electrical gas sensors are susceptible to electromagnetic interference (EMI), and some of them need to work at high temperatures, resulting in poor safety [1,2]. In contrast, optical gas sensors are inherently immune to EMI and can operate with high safety at room temperature [3,4]. In this sense, it is always desirable to develop advanced optical gas sensors to replace the electrical gas sensors currently used.

Surface plasmon resonance (SPR) sensors use an evanescent wave sensing technique capable of real-time detection of gas, chemical, and biological measurands [5,6]. The first SPR sensor was reported as early as 1982 [7], which was also the first SPR gas sensor. Since then, SPR sensors have been extensively studied, mainly for biosensing applications. Because gas molecules are generally much smaller than biomolecules, conventional SPR sensors are not sensitive enough to detect gas adsorption on the sensor surface like label-free detection of adsorbed biomolecules [8]. The gas-sensing films such as SnO₂, WO₃, and ZnO have been coated on Au-SPR sensors to improve sensitivity to toxic and harmful gases, including CO, CO₂, H₂S, NO₂, and SO₂ [9–11]. The sensitivity of such SPR sensors arises from the change in the refractive index of the gas-sensing film caused by chemisorption of gas molecules. Another approach to gas detection using SPR sensors is modification of the sensor chip with a nanoporous dielectric film [12]. Nanoporous dielectric films with huge internal surface areas can adsorb a large number of gas molecules and thus extend the depth of interaction between the plasmon field and the adsorbed molecules to the entire film thickness to cause significant sensitivity enhancement. This approach is effective but complicates the fabrication of SPR chips.

In this work, a simple and sensitive SPR gas sensor is developed by using nanoporous gold (NPG) films for surface plasmon wave propagation and gas molecule enrichment. NPG films are chemically robust and thermally stable and can be easily fabricated on solid substrates [13]. The huge internal surface of NPG films allows for simple functionalization through the well-established gold-thiol surface chemistry [14]. Furthermore, NPG films have salient propagating SPR effects and enable the plasmon field in the film to interact with the analyte molecules adsorbed on the pore wall of the film. Herein, uniform NPG thin films were prepared on glass substrates by using the sputtering-dealloying combined method, and the film thickness and porosity were controlled by controlling the sputtering time and dealloying time, to optimize the gas-sensing performance of the NPG-SPR sensor. Hydrogen sulfide (H₂S), a colorless harmful gas, is selected as the analyte, and the standard sample of 100 ppm H₂S in nitrogen was purchased from Shanghai MAOTOO Gas Company, Ltd. The H₂S-sensing properties of the NPG film are investigated at room temperature by using a laboratory-made spectral SPR platform containing a high-index prism coupler (n = 1.796 @ 656.3 nm).

NPG is optically homogeneous in the visible near-infrared (NIR) range, and its average complex refractive index can be calculated as a function of porosity (p) based on the Bruggeman effective medium approximation. Here, the porosity is defined as the volume fraction occupied by pores. The propagation constant, β_SP, for the surface plasmon traveling at the interface between air and the NPG film can be obtained based on Eq. (1) [15]:

 

Fig. 1. Dispersions for the surface plasmon at the air/infinite-NPG interface, light in air (black solid line), light in the prism (black dashed line), and the evanescent wave produced via a TIR of φ = 45° (black dotted line).

Download Full Size | PDF

The NPG-SPR sensor chips were easily prepared by first sputtering gold–silver alloy films of ∼60 nm thickness on the cleaned glass substrates and then immersing the glass substrates in nitric acid to dissolve the silver component from the alloy film. The silver dissolution process, called dealloying, is accompanied by the self-assembly of gold atoms into a thermally stable nanoporous structure [17]. The above sputtering-dealloying combined method is described in detail in our previous work [13,14]. As shown in Fig. 2, the as-prepared Au–Ag alloy film is bright white, and the dealloyed film is brown. This brown color arises from localized SPR of nanostructured gold, indicating the NPG formation. Figure 3 displays scanning electron microscope (SEM) images of four NPG films with different dealloying times. By binarizing these SEM images, the porosity of the four NPG films was also estimated to be 0.37, 0.39, 0.42, and 0.40, corresponding to the dealloying times of 60 s, 90 s, 120 s, and 180 s, respectively (see Supplement 1, Fig. S1).

 

Fig. 2. Photographs of the large-area uniform Au–Ag alloy film and NPG film on the glass substrates (area: 25 mm × 25 mm).

Download Full Size | PDF

 

Fig. 3. SEM images of the NPG films prepared at different dealloying times: (a) 60 s, (b) 90 s, (c) 120 s, and (d) 180 s.

Download Full Size | PDF

Figure 4 schematically shows the Kretschmann-type spectral NPG-SPR gas sensor platform that contains a high-index glass prism (45°/45°/90°) and a gas-fluidic chamber. The platform allows the initial resonance wavelength (λ_R) to be adjusted by rotating the prism-immobilized stage to change the angle θ between the broadband collimated incident beam and the prism-surface normal (see Supplement 1, Fig. S2). Note that θ $> $ 0 if the incident light beam is on the side closer to the SPR chip, otherwise θ ≤ 0. The TIR angle φ is related to θ by Eq. (2):

 

Fig. 4. Schematic diagram of the Kretschmann-type spectral NPG-SPR sensor platform for in situ gas detection at room temperature.

Download Full Size | PDF

To investigate the effect of porosity of the NPG film on the sensitivity of the NPG-SPR sensor to H₂S gas, four NPG films with different porosities were prepared by controlling the dealloying time. Figure 5(a) shows the reflected light spectra of these four NPG films recorded before and after the injection of H₂S gas. For better comparison, the initial resonance wavelength of each NPG film was adjusted at λ_R ≈ 700 nm. Exposure of these four NPG films to 100 ppm H₂S gas results in an increase of 7.37 nm, 16.7 nm, 33.59 nm, and 25.96 nm in resonance wavelength, corresponding to the dealloying times of t = 60 s, 90 s, 120 s, and 180 s, respectively. The porosities of the NPG films were determined by fitting the experimental SPR spectra using the Fresnel formula combined with Bruggeman’s approximation equation. The effectiveness of this method has been demonstrated in our previous work [13]. The numerical simulation was fulfilled at θ = –10° using a multilayer structure composed of the prism, the glass substrate, a 10-nm Au layer, a 60-nm NPG film, and the air superstrate. As shown in Fig. 5(b), the best fits to the resonance spectra measured with the four NPG films lead to the porosities of p = 0.38, 0.41, 0.42, and 0.41, which are very close to the corresponding values obtained from the SEM images. It is worth noting that without having an NPG film only the 10-nm Au layer cannot support the SPR mode (see Supplement 1, Fig. S3). This confirms that the NPG film is indeed responsible for the SPR generation and the measured response to H₂S gas. The optimal sensitivity of Δλ_R = 33.59 nm was achieved with the NPG film of p = 0.42 and t =120 s. Using the same NPG film, the SPR response to 50 ppm H₂S was measured to be Δλ_R = 27 nm, being smaller than that to 100 ppm H₂S (see Supplement 1, Fig. S4).

 

Fig. 5. (a) Reflected spectra with different dealloying time (t) before and after injection of H₂S gas. The optimal sensitivity of Δλ_R = 33.59 nm corresponding to t =120 s. (b) Measured light intensity spectra and calculated reflectance spectra at θ = –10° with different t (Insert: magnified view of the reflectivity valleys).

Download Full Size | PDF

To compare the sensitivity between the NPG-SPR sensor and a conventional Au-SPR sensor, the Au-SPR chips were also prepared by successively sputtering a 3-nm Cr layer and a 50-nm Au layer on a glass slide. The SPR spectra for the NPG and dense Au films were measured at different incident angles before and after exposure to 100 ppm H₂S (see Supplement 1, Fig. S5). The values of λ_R for both the NPG-SPR and Au-SPR sensors before and after the H₂S injection were measured at different incident angles. Figure 6 shows that angular dependences of λ_R for the NPG-SPR sensor are similar to those for the Au-SPR sensor. A significant difference between them is that the change of λ_R caused by H₂S exposure of the NPG film is larger than 30 nm at each incident angle of interest while exposure of the Au film to the same concentration of H₂S leads to a small change of Δλ_R < 5 nm. Comparing the responses of two sensors at the same incident angle of θ = –13° in Fig. 6 reveals that the NPG-SPR sensor is at least six times more sensitive than the conventional Au-SPR sensor.

 

Fig. 6. Resonance wavelengths of the NPG film measured as a function of the incident angle before and after exposure to 100 ppm H₂S. The curves obtained with a dense gold film also shown for comparison.

Download Full Size | PDF

To determine the sensor response time, the adsorption process of H₂S molecules on the NPG-SPR chip was monitored in real time. The reflected light intensity spectra were successively recorded at 0.2 s intervals starting before the H₂S injection until the spectra became stable. As shown in Fig. 7(a), each spectrum exhibits an absorption valley. The valley locates at λ_R = 702 nm before the H₂S injection and starts to shift to longer wavelengths from t = 40 s and stabilizes at λ_R =735.26 nm at 184 s at which point the H₂S injection was stopped. Figure 7(a) indicates that by choosing the wavelength change (Δλ) at a fixed intensity of I = 0.33 as the sensitivity parameter, a spectral response larger than Δλ_R can be obtained. Figure 7(b) displays the two time courses of Δλ_R and Δλ @ I = 0.33. At a given time, Δλ_R is smaller than Δλ @ I = 0.33. The maximum values of Δλ_R and Δλ @ I = 0.33 are 33.26 nm and 54.27 nm, respectively. The two curves in Fig. 7(b) are similar to each other, which represents the dynamic process of the adsorption of H₂S molecules within the NPG film. The response time of the NPG-SPR sensor, which is defined as the time for Δλ_R to reach 90% of its maximum, is estimated to be 45 s. As can be seen from Fig. 7(b), injecting N₂ gas in the chamber did not restore λ_R, which indicates that adsorption of H₂S molecules within the NPG film is an irreversible process. Figure 7(c) shows the time-dependent spectral changes relative to the spectrum recorded at 40 s, from which a large intensity variation (ΔI) at a fixed wavelength of λ = 675 nm was observed. This implies that the NPG-SPR sensor can operate at a single wavelength for sensitive H₂S detection at room temperature. Figure 7(d) indicates that the time course of ΔI is similar to those of Δλ_R and Δλ @ I = 0.33.

 

Fig. 7. (a) Reflected light intensity spectra measured at different times after injection of 100 ppm H₂S; (b) two time courses of the wavelength changes at a fixed intensity of I = 0.33 and the resonance wavelength shifts (Δλ_R); (c) spectral changes at different times relative to the resonance spectrum recorded at 40 s; (d) time course of the maximum change of the reflected light intensity at λ= 675 nm.

Download Full Size | PDF

After exposure of the fresh NPG film to air for 24 h, the sensor response to 100 ppm H₂S was observed to become smaller (see Supplement 1, Fig. S6), this indicates a negative influence of the air exposure on the NPG-SPR sensor performance due to surface contamination. To overcome this problem, 5-min ultrasonic cleaning of the NPG-SPR chip in de-ionized water followed by quick drying with nitrogen flow is performed before testing. This surface cleaning process was demonstrated to be very effective in keeping the sensor response fully repeatable [18]. Figure 8 shows the sensor responses to 100 ppm H₂S gas measured at different times after preparation of the NPG film (p = 0.42). For better comparison, the initial resonance wavelength was set to be λ_R ≈ 700 nm by adjusting the angle of incidence before each test. The sensor response obtained with the fresh NPG-SPR chip is Δλ_R = 32.86 nm. This chip was ultrasonically cleaned after being exposed to air for 24 hours and 48 hours. Repeat testing of the cleaned chip showed Δλ_R = 32.17 nm and 31.83 nm, respectively, both of which are almost identical to that obtained with the fresh chip, verifying the complete surface regeneration of the NPG film by ultrasonic cleaning in de-ionized water.

 

Fig. 8. Spectral response of the NPG-SPR sensor to 100 ppm H₂S measured (a) immediately, and (b) 24 hours and (c) 48 hours after the sensor chip preparation (p = 0.42 for the NPG film used).

Download Full Size | PDF

In conclusion, a Kretschmann-type NPG-SPR gas sensor with wavelength interrogation was demonstrated in this work. Large-area uniform NPG films with controllable thickness and porosity were fabricated on glass substrates by a sputtering-dealloying combination, which exhibits salient SPR effects in the visible NIR range when exposed to air. The NPG-SPR sensor is highly sensitive to toxic H₂S gas at room temperature due to the interaction of the plasmon field with the gaseous molecules adsorbed on the inner and outer surfaces of the NPG film. The optimal sensitivity of Δλ_R = 33.59 nm at 100 ppm H₂S gas was achieved using the NPG film of p = 0.42, which is at least six times higher than that of a conventional Au-SPR sensor. Long-time storage of the NPG films in air reduces the sensitivity to H₂S, and ultrasonic cleaning of the NPG-SPR chips in de-ionized water results in a fully reproducible response at room temperature, the response time being typically 45 s. To the best of our knowledge, this is the first application of NPG thin films as a highly sensitive SPR gas sensor.