1. Introduction

Refractive index (RI) sensing is highly important in biological and chemical applications since a number of substances or bio-chemical reaction can be detected via the measurements of RI, such as the detection of cells, proteins, DNA and the growth rate of bacteria [1–4]. Fiber-optic RI sensors are attractive owing to their intrinsic features of high sensitivity, immunity to electromagnetic interference, and compact size. To design a desired fiber biosensor, high sensitivity within the expected RI range, good accuracy and small size are critically important. Various types of fiber RI sensors have been proposed and developed. For instance, fiber Bragg grating (FBG) sensors with side-polished or thinned-cladding fiber are used to measure the RI of different liquids [5, 6]. However, removing the fiber cladding makes the sensor fragile. Long-period fiber gratings (LPFGs) have also high RI sensitivity, but the measurement accuracy is limited by the broad resonance in the transmission spectrum [7, 8]. A number of fiber in-line interferometers, such as Mach-Zehnder interferometers (MZIs) constructed by use of thinned-fiber [9–12], photonic crystal fiber [13–15] or multimode interferometer based on no-core fiber [16–18], Sagnac interferometers [19, 20] and Fabry-Perot interferometers (FPIs) [21–24], have also been proposed, such interferometer sensors exhibit ultra-high RI sensitivities however, due to the limited modal interference between core and cladding modes, multi-modes or polarization modes, the interferometer output spectrum has relatively broad resonance which limits the measurement accuracy. Moreover, interferometer sensors usually exhibit high RI sensitivities in the high RI region, close to the RI of fiber cladding, and low sensitivities in the low RI region, near the RI of water [9, 10, 16–18, 23]. This imposes a limit for their use in direct bio/chemical sensing application unless special film coating is utilized on the sensor head.

In this paper, we propose and demonstrate a fiber RI tip sensor by splicing a tiny segment of capillary tube between single-mode fibers (SMFs). Such a device possesses a FPI with a narrow air-cavity and a FPI with a fiber cavity, and can be used to measure liquid RI and temperature simultaneously via the detection of the interference fringe visibility and the resonant wavelength dip of the reflection spectrum. Since the fringe visibility and resonant wavelength change independently, the temperature cross-sensitivity problem can be overcome. The sensor exhibits the features of ultra-high RI sensitivity under surrounding RI close to 1.333, extremely sharp resonant wavelength dip, and high spatial resolution owing to its tip structure, which make it promising for high precision bio/chemical sensing applications.

2. Device fabrication and operation principle

Figure 1 shows the schematic diagram of the hybrid extrinsic and intrinsic Fabry-Perot interferometer sensor tip. In the fabrication of the device, a quartz capillary tube (TSP050150) is firstly fusion spliced to a section of SMF before being cut with a small length of d, and then it is spliced to another segment of SMF with length of L. The tiny section of capillary tube together with the SMF end faces (M1 and M2), form an air-cavity FPI. The two reflection mirrors M1 and M2 are not in contact with the external liquids, thus the air-cavity FPI can be considered as an intrinsic FPI. The two end faces of the SMF with length of L, M2 and M3 form another FPI. As M3 is in contact with the surrounding liquid, the FPI can be considered as an extrinsic one. The capillary tube has inner and outer diameters of 50 and 150 μm, respectively. To avoid the collapse of capillary tube and the deformation of the air-cavity during the process of fusion splicing, the splicing parameters listed in Table 1 were used for our fusion splicer (FSM-80S).

 

Fig. 1 Schematic of the proposed device.

Download Full Size | PDF

Table 1. Main fusion splicing parameters used in the experiments.

View Table | View all tables in this article

The light launched from a light source is coupled into the SMF and reflected from the surfaces M1, M2 and M3 respectively, before being collected by an optic spectrum analyzer (OSA). The whole reflection spectrum can be expressed as [24]

Figure 2(a) shows the measured reflection spectrum of device S1 fabricated with d = ~5.5 µm and L = ~248.7 µm, and the inset shows its microscope image. It can be seen from the figure that within the scan wavelength range of OSA, the reflection spectrum consists of a series of dips with a short period and different amplitudes, along the envelope with a long period. Moreover, the full-width at half-maximum (FWHM) is observed to be ~0.193 nm for the lowest dip. Figure 2(b) displays the corresponding fast Fourier transform (FFT) results. There are two adjacent peaks, A and B, with relatively large amplitudes in the spatial frequency domain, and their corresponding frequencies are ~0.2993 nm⁻¹ and ~0.3376 nm⁻¹, respectively. There are also other small peaks at the frequency positions of ~0.5923, ~0.8853, ~1.1783, ~1.4712 and ~1.7514 nm⁻¹, respectively, which can be considered as the second, third or even higher harmonics or mixing of the fundamental frequencies of 0.2993 nm⁻¹ and ~0.3376 nm⁻¹. For an FPI, the spatial frequency in FFT spectrum is $\xi =2\text{nL}/\left({\text{λ}}_{\text{1}}{\text{λ}}_2\right)\approx 2\text{nL}/\left({\text{λ}}^2\right)$ [21], where n and L are the effective RI and length of FP cavity, ${\text{λ}}_{\text{1}}$and ${\text{λ}}_2$ represent the wavelength of two adjacent resonate dips, and $\text{λ}$denotes the wavelength of certain resonance dip. According to the relationship and the measured lengths of air-cavity and fiber cavity, peak A and B are considered to be correspondent to the cavities of M1-M3 and M2-M3, respectively. The small spatial frequency difference between them can be explained by the narrow air-cavity. Thus, we can say that the fringe patterns shown in Fig. 1(a) are mainly caused by cavity M1-M3 and M2-M3, and the envelope of the fringes is caused by the difference between the two cavities (i.e. existence of the narrow air-cavity). The whole reflection spectrum in Eq. (1) can then be simplified as

 

Fig. 2 (a) Measured reflection spectrum of the sensor S1 in pure water with RI of 1.333 and (b) the corresponding FFT results.

Download Full Size | PDF

For Eq. (1) or (2), the phase variations,$2\left({\varphi }_1+{\varphi }_2\right)$,$2{\varphi }_1$ and $2{\varphi }_2$, depend on the lengths of air-cavity or fiber cavity (i.e. d and L) and the effective RI of the media in these cavities (i.e.${\text{n}}_{\text{air}}$and${\text{n}}_{\text{SMF}}$). By ignoring the dependence of ${\text{n}}_{\text{air}}$and${\text{n}}_{\text{SMF}}$ on the wavelength within the bandwidth of the light source, d, L,${\text{n}}_{\text{air}}$and${\text{n}}_{\text{SMF}}$can be considered as constants;${\text{α}}_1$and${\text{α}}_2$ are also constant for a certain device; R₁ and R₂ can be calculated from ${\text{R}}_1={\text{R}}_2={\left({\text{n}}_{\text{air}}-{\text{n}}_{\text{SMF}}\right)}^2/{\left({\text{n}}_{\text{air}}{\text{+n}}_{\text{SMF}}\right)}^2$and are also constants. Thus, when the device is immersed into liquids with RI,${\text{n}}_{\text{liq}}$, there is only one variant, R₃ $\left(={\left({\text{n}}_{\text{SMF}}-{\text{n}}_{\text{liq}}\right)}^2/{\left({\text{n}}_{\text{SMF}}{\text{+n}}_{\text{liq}}\right)}^2\right)$, caused by mirror M3, and only the reflection intensity of the device is modulated without any phase variation. That is to say, the proposed device can be used for surrounding RI sensing by measuring its reflection intensity.

To explore the RI response of the proposed device and the influence of narrow air-cavity, simulations were carried out according to Eq. (1). Assuming d = 6 μm, L = 260 μm, ${\text{n}}_{\text{air}}=1$,${\text{n}}_{\text{SMF}}=1.44$, R₁ = R₂ = 0.04, ${\text{α}}_1=0.12$,${\text{α}}_2=0.2$, and ignoring the dependence of ${\text{n}}_{\text{air}}$and${\text{n}}_{\text{SMF}}$ on the wavelength, the normalized reflection intensities of such a device for liquids with different RIs were obtained, as shown in Fig. 3(a). The high-frequency fringes and low-frequency modulation can be observed within the wavelength range of 1450-1650 nm, and the calculated reflection spectrum is very similar to the measured spectrum shown in Fig. 2(a). The inset of Fig. 3(a) shows the resonant dip intensity at different surrounding RIs. Obviously, the intensity of the resonant dips increases with the increase of surrounding RI. Figure 3(b) demonstrates the dip intensity versus surrounding RI at different transmission losses in the air-cavity by tracing the dip with the lowest intensity. It can be seen from the figure that the dip intensity is sensitive to the surrounding RI and depends on the transmission loss in the air-cavity,${\text{α}}_1$, and the device exhibits non-linear RI response with a higher RI sensitivity for a lower surrounding RI. Here, ${\text{α}}_1$depends on the cavity length due to the divergence of light beam propagating in the air-cavity, and increases sharply with the increase of air-cavity length. For instance,${\text{α}}_1\approx 0.06$for d = 25 µm,${\text{α}}_1\approx 0.1$for d = 30 µm and ${\text{α}}_1\approx 0.15$for d = 40 µm by simulations. In practice, the real transmission loss is larger due to the uneven reflection surface. An effective way to reduce ${\text{α}}_1$ and obtain a high RI sensitivity is to use a short air-cavity, especially for a low surrounding RI measurement (e.g. biosensing requires a high RI sensitivity under surrounding RI near 1.333).

 

Fig. 3 (a) Calculated reflection spectra when n_liq is within the range of 1.30 to 1.432 and (b) lowest dip intensity vs surrounding RI with different losses of the narrow air-cavity.

Download Full Size | PDF

3. Liquid RI and temperature sensing experiments

Figure 4 shows the experimental setup for the liquid RI testing. The output from a broadband light source (BBS, Amonics, ALS-CL-20-B-FA) is directed to the sensor head via a circulator, and then the reflected light is collected by an optic spectrum analyzer (OSA, Yokogawa, AQ6370D) to record the interference spectrum. The wavelength range of the BBS covers the wavelength range from 1450 to 1650 nm. The OSA has a scan wavelength range of 700-1700 nm with a highest resolution of 0.02 nm. Owing to the compact size of the sensing device, a tiny drop of liquid is enough to immerse the sensor head in the RI test. The liquid is dropped onto the surface of a copper block which is attached on a thermoelectric cooler (TEC) to allow an accurate temperature control with a resolution of 0.1°C.

 

Fig. 4 Experimental setup for liquid RI sensing.

Download Full Size | PDF

Figure 5 shows the responses of the sensor S1 to the surrounding liquids with different RIs while keeping a constant temperature of 25°C. A series of Cargille oil with nominal RIs of from 1.300 to 1.400 (@ 589.3 nm) in step of 0.01 and from 1.400 to 1.432 in step of 0.004 were tested. It can be observed from Fig. 5(a) that, a dip with the largest extinction ratio appears near 1590 nm in the reflection spectrum for all the tested liquids, which agrees with the calculated results shown in Fig. 3. Similar to the simulation results obtained previously, the dip wavelength intensities of in the reflection spectra are reduced with the increase of surrounding liquid RI, without wavelength shift. For the lowest dip, a total change of ~14.6 dB in reflection is obtained, as shown in the inset of Fig. 5(a). The relationship between the reflection of the lowest dip and the surrounding RI is displayed in Fig. 5(b). By defining the sensitivity as the reflection variation per RI unit (RIU), it can be seen form Fig. 5(b) that the lowest dip exhibits a higher sensitivity for a lower surrounding RI and a lower sensitivity for a higher surrounding RI. For instance, the sensitivity reaches 216.37 dB/RIU when the RI is 1.30. Moreover, it is found that the reflection of lowest dip is varied polynomially with the surrounding RI and a fitting coefficient of R² = 0.9930 is obtained. The RI response of our sensor to the surrounding RI has different characteristics compared with the sensors such as fiber interferometers based on various hybrid fibers [9, 10, 16–18], fiber tapers [25–27] or tips [23, 28, 29] which have high sensitivities for higher surrounding RI (near the RI of fiber cladding) but insensitive to lower surrounding RI (near the RI of 1.33 or even lower). The feature of our device is thus well suitable for biosensing due to the fact that the RI of body fluid is generally close to that of water. To further evaluate the performance of our sensor to the surrounding RI near 1.33, a set of salt solutions with concentrations of 1.64, 9.19, 13.85, 20.58, 31.14, 50.19, 61.90, 69.45 and 78.59‰ (by weight), respectively, corresponding to the RI value of 1.3333, 1.3347, 1.3355, 1.3368, 1.3387, 1.3422, 1.3443, 1.3457 and 1.3474, respectively, were tested. Figure 5(c) demonstrates the response of S1 to salt solutions of different concentrations, and the inset shows part of reflection spectra near the lowest dip. The variation of reflection of the resonant dip can be considered to be approximately linear to the surrounding RI within a narrow RI range according to the linear fit result (with R² of 0.9939). The sensitivity obtained is ~133.52 dB/RIU, which is much higher than that of intensity-based sensors [23, 29–33], as listed in Table 2, thus showing the advantages of our device. An intensity resolution of 0.01 dB can be expected for a commercial OSA, which indicates that the expected RI resolution can reach up to 7.49 × 10⁻⁵ by using our device within the range of 1.3333-1.3474.

 

Fig. 5 Response of the sensor to surrounding RI for S1 at 25°C. (a) Reflection spectra of the sensor under different surrounding RIs. Reflection of lowest dip vs RI within the range of (b) 1.30-1.432 and (c) 1.333-1.347.

Download Full Size | PDF

Table 2. Various intensity-modulated optical fiber RI sensors and their sensitivities.

View Table | View all tables in this article

The temperature response of sensor S1 was also investigated. During the experiment, S1 is immersed into pure water, and the temperature of water was increased from 10 to 60°C in a step of 5°C, with the resolution of 0.1°C. Figure 6(a) shows the reflection spectra of sensor S1 in the heating process. With the increase of ambient temperature, the reflection spectrum experiences a red shift as shown in the inset of Fig. 6(a), which can be explained by the variation of length and RI of the glass cavity between M2 and M3 due to the thermal expansion and thermo-optical effect of the fiber. Figure 6(b) shows the relationship between the resonant dip wavelength and temperature, where a good linear relationship with R² of 0.9923 can be found. The temperature sensitivity (defined as the dip wavelength shift per degree) of 10.85 pm/°C is obtained. The small reflection variation of the resonant dip in the inset of Fig. 6(a) can be explained by the RI variation of water due to its thermo-optic effect, which has a small coefficient of ~2 × 10⁻⁶ RIU/°C.

 

Fig. 6 Response of the sensor to temperature in water with RI of 1.333.

Download Full Size | PDF

A new sensor S2 with similar structure as S1 was fabricated, and the air-cavity lengths and the corresponding lengths of SMF are ~17 and ~244 µm, respectively. Figure 7 gives the RI responses of S2 and S1. The reflection variations of tracing dip for S1 and S2 are ~12.59 and ~8.02 dB, respectively. The average RI sensitivity of S2 is lightly smaller than that of S1 due to its longer air-cavity, which is consistent with the numerical simulation results obtained in Fig. 3. For the temperature sensitivity, the same value as ~10 pm/°C is obtained for the two devices, which is understandable as the temperature sensitivity of the device is mainly dependent on the section of SMF cascaded to the capillary tube, not on the air-cavity.

 

Fig. 7 Response of the sensor to surrounding RI for S1 and S2.

Download Full Size | PDF

The narrow air-cavity plays another role of enhancing the dip wavelength intensity change in the reflection spectrum and narrowing the corresponding full-width-at half-maximum (FWHM). For instance, the FWHM of S1 is only ~0.193 nm for liquid RI of ~1.333, which is tens of times less than those of fiber interferometers (several nanometers or tens of nanometers) due to the limited interference effect between the core mode and cladding modes [9–15, 26, 27, 29, 34], and is even comparable to that of some fiber laser sensors [17]. Thus, the RI and temperature of the liquid can be simultaneously determined by the dip intensity variation and wavelength shift of the device in an independent manner.

4. Solute diffusion testing

The RI of body fluid of human or animals is close to that of pure water, to demonstrate the potential of the proposed device for biosensing, we carried out the test on NaCl diffusion as salt water with RI in the range of 1.333-1.375 can be readily obtained.

A beaker was firstly filled with pure water with depth of ~30 mm, NaCl was subsequently added until superfluous NaCl particles can be seen on the bottom of the beaker. During the adding process, stirring liquid was prohibited. Next, sensor S2 was immersed into the salt water with immersed length of ~3 mm. Figure 9(a) shows the reflection spectra of S2 during the diffusion process of solute, where the inset clearly shows the dip intensity variation. NaCl diffuses with time and the RI of liquid exhibits a gradually varied concentration. The local liquid RI nearby S2 increases slightly during the diffusion process, which is proved by the increase of dip intensity as shown in the inset. As no stirring was implemented, the diffusion process went very slowly. Figure 8(b) shows the dip intensity at different time, and the total variation reaches ~2 dB for 160 min corresponding to a RI variation of ~0.0166, which indicates that the sensor is sensitive enough and well qualified for the task although the diffusion is a rather slow process.

 

Fig. 8 Response of the sensor to NaCl diffusion. (a) Reflection spectra of the sensor versus time at a fixed immersion length and (b) Reflection dip intensity vs time.

Download Full Size | PDF

Then, S2 was picked up and immersed into the salt water again. This time, the immersion length was increased from 0 to 10 mm with a step of 500 µm. Figure 9 demonstrates the reflection spectra of S2 at each immersion length. Obviously, with the increase of immersion length, the dip intensity increases due to the increase of liquid RI. The local RI close to the water surface is lower than that near the bottom of the beaker. Thus, the distribution of liquid RI in the vertical direction with a spatial resolution of 500 µm can be obtained in a convenient way. Since the sensing part is just the end face of the fiber, even smaller spatial resolution can be expected, which is unachievable for the bulk type refractometers with large dimension [5,7,9–14,16–18,20]. Such a feature makes the proposed device particularly suitable for biosensing with high spatial resolution and with tiny sample.

 

Fig. 9 Response of the sensor to salt water with gradually varied concentration. (a) Reflection spectra of the sensor with different immersion lengths and (b) Relationship between the dip intensity and immersion length.

Download Full Size | PDF

5. Protein detecting

To demonstrate the application of the proposed device for direct protein detection, solutions of bovine serum albumin (BSA) with different concentrations were tested. Figure 10(a) illustrates the change of reflection spectrum and the tracing dip reflections with respect to different concentrations of BSA samples for S2. It is found that the device is sensitive to BSA solution and dip wavelength intensity increases obviously with the increase of BSA concentration, and the slope of linear fitting function reaches ~0.4 dB/(mg/ml) within the range of 3-6 mg/ml, which indicates that the device has a capability of distinguishing BSA solutions of different concentrations with high sensitivity. By introducing bacteria sensitive materials to the end face of the device, the detection of the growth rate of any bacteria is also expected, which will be our future work. Figure 10(b) shows the output power of the used light source in 90 min. It can be observed from the figure that the power fluctuation is only ± 0.01 dB, which is comparable to the amplitude resolution of OSA. Thus, the accuracy of the sensor is limited by the stability of the light source.

 

Fig. 10 (a) Response of the sensor to BSA solution with different concentrations. (b) Power stability of the light source.

Download Full Size | PDF

6. Conclusion

A simple and compact FP fiber tip sensor is proposed and demonstrated for simultaneously liquid RI and temperature sensing. The device is based on a narrow inner air-cavity and a short segment of SMF to form an intrinsic FPI and an extrinsic FPI, respectively. Both the theoretical and the experimental results demonstrate that the ambient liquid RI and temperature can be simultaneously determined by the intensity and shift of the resonant dip wavelength in the reflection spectrum. Here, the narrow air-cavity plays an important role to obtain a high RI sensitivity and a sharp resonant wavelength dip. The high RI sensitivity near 1.33 and high spatial resolution of the device makes it promising for high precision bio/chemical sensing applications.