1. Introduction

Gas detection using high-sensitivity sensors is mandatory in many fields such as environmental monitoring, chemical processing, energy supervision, and medical testing, which is done with the aim of controlling environmental pollution, production safety, public health, and so on. In this regard, as safe and reliable sensors, optical fiber gas sensors demonstrate great potential in practical applications, thereby attracting considerable interest among researchers. In the past, various optical fiber sensors based on fiber Bragg gratings [1–3], long-period gratings [4–6], photonic crystal fibers [7–9], and interferometers [10–12] have been reported and they all revealed different peculiarities in sensing. But it still has a problem that they leave much to be desired when used for gas detection. That is because gas detection with an external refractive index of around ∼1.0 using optical fiber sensors should overcome the mismatch between the allowable guided mode and the external gaseous medium. It leads to the optical fiber sensor performance, which mostly determined by the sensitivity, is hard to be satisfactory by the conventional devices. Being confined to their low sensitivity, which is mostly just several thousands of nanometers per refractive index unit (RIU) or even lower, detecting the tiny variations of the gas environment in sensing is difficult. Thus, designing an optical fiber sensor that has a high sensitivity in a gaseous medium is challenging. In nowadays, a useful and prevailing measure is the coating method. Researchers came up with this that we can use the coating as an intermediate to indirectly magnify the influence on the sensors induced by the tiny variations of the gas environment. In this manner, one can take complete advantage of different coating materials and make the sensors more flexible. Traditionally, palladium coating is commonly used to improve the performance of optical fiber sensors in H₂ sensing, as the palladium coating expands considerably after absorbing H₂ [13,14]. Moreover, several organic or inorganic materials such as metal–organic frameworks (MOFs) [15–17], graphene [18,19], and zinc oxide [20–22] also can be used as coating on optical fiber sensors. The sensitivity of the sensors will have different degrees of enhancement after coating, and some of these materials can even make the sensors be selective to specific gas. For instance, an optical fiber sensor coated with ZIF-8 shows high sensitivity and selectivity to CO₂ gas relative to other small gases (H₂, N₂, O₂, and CO) [23]. However, it is important to note that the structural design of sensors still directly influences sensitivity in gas detection.

From the viewpoint of sensor structural design, many realized measures have successfully achieved high sensitivity; these include optical fiber spectral combs [24], fiber-optic acoustic sensors [25], among others. In addition to these measures, there is a method that can significantly improve the sensitivity, which is by means of the dispersion turning point. It is a critical point that can be observed in microfiber modal interferometers [26–28], long-period grating [29], highly birefringent microfiber-based Sagnac interferometers [30,31] and microfiber couplers [32,33]. These sensors working at the turning point would have ultrahigh sensitivity and some special characteristics, compared to that without the turning point. Most of these sensors have already been studied or experimentally applied in sensing. For example, recent research demonstrated that an ultrasensitive gas RI sensor based on optical nanofiber couplers can achieve an RI sensitivity of 46,470 nm/RIU [34]. Previously, a reported highly birefringent (HiBi) elliptic microfibers demonstrated an RI sensitivity of ∼195,348 nm/RIU around RI = 1.35887 [35]. However, the practical application of this highly birefringent microfiber based Sagnac interferometer operating at the turning point in a gaseous medium (RI ∼1.0) remains unexplored, as the requirements for the sensor to work well in a gaseous medium are quite demanding.

In the subsequent sections, we focus on the elliptic microfiber based Sagnac interferometer operating at the group birefringence turning point and apply it in a gaseous medium. Through theoretical analysis and simulated calculation, we obtain the specific parameters of the elliptic microfiber for the group birefringence turning point. With these specific parameters, the elliptic microfiber will have high birefringence and its group birefringence is close to zero at the working wavelength, indicating that it would have high sensitivity. Practical application in a gaseous medium such as relative pressure detection and CO₂ concentration detection is done to verify the theory as well as the technical feasibility. Furthermore, the temperature response is discussed for a deeper understanding of this device.

2. Configuration, theory, and simulation

Figure 1 shows the schematic configuration of the proposed elliptic microfiber; it contains two transition regions and a uniform waist region with an elliptic cross-section. The method of fabricating elliptic microfibers has already been studied in a previous report [35]. At first, we produce the preprocessed optical fibers with an elliptic cross-section through laser cutting. A standard single-mode fiber (Corning G.652.D) is exposed to a CO₂ laser (C-50W) on two opposite sides in turn, with a machining length of 5 mm. Then, we taper the preprocessed optical fibers via the flame brushing technique, and we finally get the elliptic microfibers. When light passes through the elliptic microfiber, owing to its non-circular-symmetrical cross-section, two orthogonal polarization fundamental modes that have a smaller or a larger effective refractive index (ERI) occur. The electric field distribution of these two orthogonal polarization fundamental modes are illustrated in Fig. 1. Compared to the first one, the second one has a larger evanescent field. We splice the elliptic microfiber in a Sagnac loop that has a polarization controller in it to form the elliptic microfiber based Sagnac interferometer. Light from the broadband light source (Golight) pass the 3-dB coupler and is split into two counter propagating beams in the Sagnac loop. Furthermore, the elliptic microfiber produces two orthogonal polarization fundamental modes with different ERIs, and there is a phase difference between them, which would generate interference after recombination of the two beams via the 3-dB coupler again. The phase difference can be expressed as φ=2πBL/λ, where L is the effective interaction length of the elliptic microfiber, birefringence B represents the ERI difference between the two orthogonal polarization fundamental modes, and λ denotes the wavelength. Through derivation, we can obtain the RI sensitivity

 

Fig. 1. The schematic of elliptic microfiber and electric field distribution of the two orthogonal polarization fundamental modes.

Download Full Size | PDF

Hence, we undertake a simulated calculation to obtain the specific parameters and study its sensing characteristics. We define a as the long axis length and ellipticity e as the ratio of the short axis to the long axis. The simulated result shows that the elliptic microfiber has a birefringence of 10⁻² and the group birefringence G will increase from negative to positive with an increase in wavelength, which means that the value of G will pass through zero, indicating the group birefringence turning point. Through data processing, the RI sensitivity can be obtained using the value of B and G. Figure 2(a) plots the RI sensitivity of the elliptic microfibers with e = 0.5 and different values of a. It has │Sn│→∞ when approaching the turning point, and the Sn is positive at shorter wavelengths and negative at longer wavelengths. This is because the birefringence B decreases with an increase in the external RI and the group birefringence G is negative to the left side of the turning point and positive to the right side. Within ± 30 nm of the turning point, the RI sensitivity increases fast and is calculated to be over 5×10⁴ nm/RIU. This high sensitivity will decrease rapidly as the wavelength drifts away from the turning point. However, the region with │Sn│ > 10⁴ nm/RIU still has a range of approximately ± 200 nm around the turning point. Beyond that, with a decrease in a, the turning point will move to a shorter wavelength. Figure 2(b) shows the simulated data of a and e when the turning point occurs at 1550 nm (a from 1.4 µm to 1.9 µm and e from 0.4 to 0.6). We can use these parameters to control the position of the turning point. Figure 2(c) illustrates the simulated transmission spectra of the elliptic microfiber with e = 0.5, a = 1.66 µm, and L = 3 mm, as the external refractive index changes from 1 (in vacuum) to 1.00027330 (in air), which can be seen from the device’s response across different relative pressures. We also observe that the free spectral range (FSR) will be larger when the wavelength is closer to the turning point. Moreover, with the external RI increasing, the dips come toward the turning point, as shown in Fig. 2(d). The calculated average RI sensitivity can reach 104,281 nm/RIU at λ∼1530 nm and −107,940 nm/RIU at λ∼1570 nm.

 

Fig. 2. (a) The RI sensitivity of the elliptic microfibers with e∼0.5 and different diameters. (b) The simulated data of a and e when the turning point occurs at the wavelength 1550nm. (c) The simulated transmission spectra as the external refractive index changes from 1 to 1.00027330 (e = 0.5, a = 1.66µm, L = 3mm). (d) The wavelengths shift of peak A and peak B.

Download Full Size | PDF

3. Experiments and discussions

According to the simulated result, we conducted experiments to verify the theory mentioned above. Several preprocessed single-mode fibers with ellipticity e∼0.5 are fabricated via laser cutting. And we splice the preprocessed single-mode fiber into the Sagnac loop. We then taper the preprocessed fibers by the flame brushing technique. Meanwhile, real-time transmission spectra are observed by the optical spectrum analyzer (OSA, Yokogawa AQ6370D). When the desirable group birefringence turning point occurs, we stop tapering. With the help of the electronic control system [13], we can fabricate the elliptic microfiber with good reproducibility. Finally, we obtain elliptic microfibers whose group birefringence turning point wavelength is at ∼1551 nm (with a∼1.57 µm and L∼3 mm, its spectrum is shown in Fig. 3(b)), ∼1510 nm (with a∼1.53 µm and L∼4.4 mm, its spectrum is shown in Fig. 4(b)), and so on. To get its RI sensitivity, we first perform a relative pressure sensing experiment. The equipment we used in this detection is shown in Fig. 3(a). An elliptic microfiber that has a group birefringence turning point wavelength at 1551 nm is put in a vacuum chamber at room temperature (∼20°C). First, we use a vacuum pump to suction out the air and make the negative gauge pressure in the vacuum chamber to be −80 KPa. The negative gauge pressure can be observed using a barometer in the vacuum chamber. Then, via a switchable air hole, let the air get into the vacuum chamber to achieve different vacuum degrees and record the spectra simultaneously. Figure 3(b) plots the transmission spectra for different relative pressures. With an increase in negative gauge pressure, the dips move toward the turning point, standing for negative or positive sensitivity at the two sides of the turning point. Further, the FSR reduces as the wavelength gets away from the turning point. Figure 3(c) illustrates the dips wavelength shift of peak A and peak B as a function of negative gauge pressure. Peak A shifts 23.6 nm and Peak B shifts −17.52 nm in total as the negative gauge pressure changes from −80 KPa to 0 KPa. The measured average sensitivity is 0.295 nm/KPa at ∼1518 nm and −0.219 nm/KPa at ∼1587 nm. As the external air refractive index is related to the atmospheric pressure [36], it can be express as

 

Fig. 3. (a) The schematic of detection equipment in gas pressure and RI sensing. (b) The transmission spectra in different relative pressure. (c) The dip wavelengths shift as the function of negative gauge pressure. (d) The RI sensitivity as the function of wavelength.

Download Full Size | PDF

 

Fig. 4. (a) Schematic of the detection system. (b) Transmission spectra in different CO₂ concentrations. (c) Dip wavelength shift as the CO₂ concentration detection experiment continues. (d) Dip wavelength shift as a function of CO₂ concentration. (e) The RI sensitivity as the function of wavelength. (f) The wavelengths shift in the test of repeatability (with an elliptic microfiber having a group birefringence turning point at ∼1470 nm).

Download Full Size | PDF

To demonstrate its practical applications in gas sensing, we apply this device to have a CO₂ concentration detection, which is necessary for breath detection, environmental monitor and industrial production and storage [37]. Figure 4(a) plots the schematic of the detection system. The elliptic microfiber is put in a gas chamber that has a gas outlet and two gas inlets. We delivered CO₂ gas and N₂ gas through the two gas inlets and they are mixed in the gas chamber. By controlling the flow rate of the CO₂ gas and the N₂ gas through the flow control mechanisms, we can obtain different CO₂ concentrations in the gas chamber. The CO₂ concentration is changed per 500 s, and the spectra are recorded per 10 s by the OSA. Figure 4(b) illustrates the spectra for different CO₂ concentrations with an elliptic microfiber having a turning point wavelength at 1510 nm. Compared to the elliptic microfiber used in gas pressure and RI sensing, the FSR of this elliptic microfiber is narrower. That is because this elliptic microfiber has a longer effective interaction length, which can be controlled in the tapering process. Figure 4(c) shows the dip wavelength shift throughout the experiment. When the CO₂ concentration is changed, it appears a hopping and then becomes steady quickly, revealing good stability. The dips near the turning point shift more than the dips farther away, as shown in Fig. 4(d). Its sensitivity can reach 0.1544 nm/% CO₂ at ∼1490 nm and −0.1272 nm/% CO₂ at ∼1530 nm. And the wavelength fluctuation of this sensor is measured to be 0.06 nm, standing for a limit of detection of ∼0.3886% (3886ppm) CO₂. Considering the RI of pure CO₂ gas is ∼1.00043837 and pure N₂ gas is ∼1.00027907, we can get its RI sensitivity to be as high as 107,125.5 nm/RIU at ∼1490 nm, as shown in Fig. 4(e). As for other gases such as HCl/N₂, N₂O/N₂, NH₃/N₂ [38], the maximum refractive index change is ∼2.2×10⁻⁴ and it produces a wavelength shift of about 35.4 nm, but it still in the dynamic range of the sensor. Simultaneously, the sensor may combine with the coating method [23] or photothermal spectroscopy [8] to make it selective.

Figure 4(f) plots the wavelength shift as we delivered only CO₂ gas or N₂ gas to the gas chamber repeatedly, with an elliptic microfiber having a turning point wavelength at 1470 nm. The dip shifts 9.4 nm at ∼1442 nm and −9 nm at ∼1498 nm in total as the gas in the gas chamber changes from N₂ to CO₂. Each time, the dip wavelengths are observed to be similar in CO₂ (N₂), demonstrating good repeatability. From the figure, we can observe that the dip wavelength need ∼10 s to be steady. However, this is not the real response time of this sensor. It mostly denotes the stabilization time of the gas in the gas chamber, which is determined by the mixing speed of the gas.

In some situations, temperature is an undesirable interference factor that exists during detection. As thermal-expansion and thermo-optic effects exist in optical fibers, when the temperature changes, these will also influence the sensor. Thus, we place an elliptic microfiber that has a turning point wavelength at 1551 nm in a resistance furnace to survey its temperature response. Figure 5(a) plots the transmission spectra for temperature range of 30 °C to 90 °C, rising in intervals of 10 °C. As the temperature changes, the behavior of the sensor is similar to that in RI sensing. The dips drift away from the turning point when the temperature rises. Hence, the temperature sensitivity is negative to the left side of the turning point and positive to the right side. Figure 5(b) illustrates the dips wavelength shift. In the measurement, the temperature sensitivity can reach 486.7 pm/ °C at 1590 nm. And its temperature sensitivity also will reduce quickly as the wavelength shifts away from the turning point. The sensitivity can be range from −48 pm/ °C to 760 pm/ °C as the wavelength changes. We also calculate the temperature cross-sensitivity, which is a small value of around 3×10⁻⁶ RIU/ °C and it is various at different wavelength. We should select an appropriate working wavelength base on its temperature cross-sensitivity value to balance the influence of temperature.

 

Fig. 5. (a) Transmission spectra for the temperature range of 30 °C to 90 °C. (b) Dip wavelength shift as a function of temperature.

Download Full Size | PDF

4. Conclusion

We study an ultrasensitive elliptic microfiber based Sagnac interferometer operating at group birefringence turning point using in a gaseous medium. Through theoretical analysis and simulated calculation, we discuss its characteristics with the same ellipticity and different long axis, knowing that it can perform an ultrahigh sensitivity. In the experiment, we achieve an RI sensitivity of 160938.9 nm/RIU in relative pressure detection and 107,125.5 nm/RIU in CO₂ gas concentration sensing. These results show its potential for use in acoustic sensing and biochemical detection.