Highly sensitive miniature photonic crystal fiber ... - OSA Publishing

2 downloads 0 Views 365KB Size Report
Highly sensitive miniature photonic crystal fiber refractive index sensor based on mode field excitation. Wei Chang Wong,1 Chi Chiu Chan,1,* Li Han Chen,1 Zhi ...
May 1, 2011 / Vol. 36, No. 9 / OPTICS LETTERS

1731

Highly sensitive miniature photonic crystal fiber refractive index sensor based on mode field excitation Wei Chang Wong,1 Chi Chiu Chan,1,* Li Han Chen,1 Zhi Qiang Tou,1 and Kam Chew Leong2 1

Division of Bioengineering, School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637457, Singapore 2 Globalfoundries Singapore Pte, Ltd., 60 Woodlands Industrial Park D Street 2 Singapore 738406, Singapore *Corresponding author: [email protected] Received February 3, 2011; revised March 29, 2011; accepted April 10, 2011; posted April 11, 2011 (Doc. ID 142194); published April 29, 2011

A highly sensitive miniature photonic crystal fiber refractive index sensor based on field mode excitation is presented. The sensor is fabricated by melting one end of a photonic crystal fiber into a rounded tip and splicing and collapsing the other end with a single-mode fiber. The rounded tip is able to induce cladding mode excitation, which resulted in an additional phase delay. Linear response of 262.28 nm/refractive index unit in the refractive index range of 1.337 to 1.395 is obtained for the physical length of a 953 μm sensor. The sensor is also shown to be insensitive to environmental temperature. © 2011 Optical Society of America OCIS codes: 060.2370, 060.5295, 120.3180, 280.4788.

In situ monitoring of physical, chemical, and biological substances is of great interest in the manufacturing industries and environment control. The refractive index (RI), an indication of the amount of substance in the environment, is one of the most important parameters used in biosensing and quality control in food industries. Traditionally, refractometers, such as the Abbe refractometer and surface plasmon resonance sensor, have been used for RI measurement and are commercially available. However, these systems usually comprise of a bulky prism, which are limited in certain practical applications. An ideal refractometer, especially one for the biochemical system, should be cost effective, have high sensitivity, be mechanically and chemically stable, and small in size. In recent years, fiber-optic sensors have received significant attention as they have shown potential in meeting the required conditions in addition to being immune to electromagnetic interference and remote sensing. Optical fiber refractometers are commonly based on gratings, tapering, and etching to either enable light to interact with the external environment or forming an interferometer to encode external RI information into the spectra fringe. For a grating-based RI sensor, long period gratings (LPGs) and fiber Bragg gratings (FBGs) have been utilized [1,2]. However, these sensors have various disadvantages such as high cross sensitivity to ambient temperatures for LPG based RI sensors, low RI sensitivity, and complicated fabrication steps requirement for FBG-based RI sensors. The use of tapering and dry etching to micromachine the fiber structurally for closer interaction between light and environment has lead to high RI sensitivity and reduction in sensor size [3–5], but these sensors are mechanically unstable. Most recently, there has been development in a photonic crystal fiber (PCF)-based RI sensor. The holes in the PCF represents microfluidic channels, which analyzes can flow into, affecting the way light is guided, which leads to high sensitivity [6]. The nature of this system means it is difficult for analyze to flow out or perform selective 0146-9592/11/091731-03$15.00/0

surface coating in the channels [7]. These problems have been avoided by collapsing air holes in the PCF or having core size mismatch to guide light into the cladding, forming an interferometer [8–10]. However, due to a small difference between the effective indices of the core and cladding, longer sensing fiber is needed to induce enough phase difference for observing interference fringes within the range of the 1500 to 1600 nm spectral window. In this Letter, a highly sensitive miniature PCF modal interferometer sensor formed by the excitation of the cladding mode at the tip is presented. The sensor comprises a collapsed region between a single-mode fiber (SMF) and a short piece of pure silica PCF (LMA10), which has its end melted into a rounded tip. The phase difference leading to spatial interference is caused by the excitation of the cladding mode due to the rounded end. This mechanism offers promise for improved sensitivity in RI sensing by evanescent wave or surface plasmon resonance [11,12] and also reduces the size of the sensor. The core diameter and cladding diameter of the PCF are 10.4 and 124 μm, respectively. The sensor was fabricated by first splicing and collapsing the PCF to the SMF with a commercial fusion splicer (Sumitomo Electric Type 39). Strong electric arc discharges caused localized heating on the PCF, which led to the collapse of the air holes in the cladding region, forming the first collapsed region. Then, the spliced PCF was cleaved to a desired length, and the air holes at the distal end were collapsed by electrical arcing to form the second collapsed region. A schematic diagram to explain the mechanism of the sensor is shown in Fig. 1(a). The first collapsed region is well-known to induce strong mode coupling from the fundamental core mode to the cladding modes of the PCF [8]. The rounded PCF’s end is used to induce higher order mode excitation in the second collapsed region before propagating back in the cladding. Finally, the cladding mode recouples back into the core at the first collapsed region, forming a Michelson interferometer (MI). The recombination of the cladding and fundamental core modes after undergoing reflection and mode © 2011 Optical Society of America

1732

OPTICS LETTERS / Vol. 36, No. 9 / May 1, 2011

Fig. 1. (Color online) (a) Schematic diagram for excitation mode mechanism of the sensor. (b) Experimental setup of the PCF sensor system. (c) Microscope image of the sensor.

excitation follows the standard interference equation for MI: R ¼ ½E 1 #2 þ ½E 2 #2 þ 2E 1 E 2 cosð2πlΔn=λ þ ϕÞ;

ð1Þ

where E 1 and E 2 are the magnitudes of electric field of the core and cladding modes, respectively; Δn ¼ n1 − n2 , with n1 and n2 , the effective indices of the core and cladding modes, respectively; l is twice the length of the second collapsed region; λ is the wavelength of the propagating light; and ϕ is the phase difference obtained along L1 , the remaining length of PCF left after collapsing, and is similar to the phase term described in [8]. The variation of the external RI can change the effective RI of the cladding mode leading to phase shift. A Fabry–Perot (FP) interferometer could also be present due to the RI difference between both fiber’s cores, which follows Eq (1) with Δn ¼ n2 and l ¼ 2Lo where Lo is the original length of the PCF used. By selecting Lo to ensure the FP fringes will have distinctly different spatial frequency from those by the MI, it allows for easy removal of FP fringes through digital filtering. The PCF interferometry sensor system consists of a broadband light source, a circulator, and an optical spectrum analyzer (OSA, Yokogawa AQ 6370). The schematic diagram of the sensor setup is shown in Fig. 1(b). A digital low pass filter is applied to remove the FP fringes for improving the signal-to-noise ratio of theMI fringes. Figure 1(c) shows the microscope image of the sensor. To investigate the performance of the PCF sensor on RI sensing, the PCF sensor was immersed into solutions of different RI values, made by dissolving different amount of glucose in deionized water. The glucose solutions had their RIs measured by a ABBE refractometer (2WAJ ABBE Refractometer) with an RI resolution of 0.0005. Figure 2(a) shows the normalized reflection of the sensor with a varying RI (1 to 1.3950) at room temperature of 23 °C before passing through the digital low pass filter. Lo was measured at about 953 μm, the first collapsed region was about 87 μm, and the second collapsed region, l=2, was about 160 μm. From Fig. 2(a), the period of the small fringes is found to be about 0:85 nm, while the theoretical spectral period defined as λ2 =nz ¼ 0:8503 nm

Fig. 2. (Color online) (a) Normalized reflection of the sensor with varying RI from 1.337 to 1.395 for l ¼ 953 μm, before passing the spectra to the digital low pass filter. (b) Wavelength shift of reflection peaks with varying RI values and temperatures obtained after passing the spectra through the digital low pass filter.

where n ¼ 1:45, z ¼ 1906 μm, and λ ¼ 1533 nm, confirms the theoretical analysis. Figure 2(b) shows the shift of the wavelength peaks obtained after processing the spectra with the digital low pass filter, and with a varying external RI from 1.337 to 1.395. Maximum sensitivity, dðΔλÞ=dn of 262:38 nm= refractiveindexunit (RIU) is obtained with the correlation value of 0.9944. This corresponds to a minimum measurement resolution of 3:811 × 10−5 RIU for the OSA’s resolution of 0:01 nm. The temperature response of the sensor was also investigated by placing the sensor into a breaker of water on a hot plate (Corning PC-4200) for heating. The shift of wavelength peaks of the MI fringes with temperature measured by a digital thermometer (JinMing instrument JM624) is shown in Fig 2(b). The average sensitivity of the sensor in water with respect to temperature dðΔλÞ=dT is 0:019 nm=°C. The thermal optics coefficient of the sensor in water can thus be calculated as dðΔλÞ= dT × dn=dðΔλÞ ¼ dn=dT ¼ 7:24 × 10−5 RIU=°C. Given that the thermal optics coefficient of water is 8 × 10−5 RIU=°C [13], the sensitivity of the sensor with temperature is the difference between the two coefficients, found to be 0:76 × 10−5 RIU=°C. This implies that the sensor has low sensitivity to temperature change. Besides that, it was also found experimentally that the sensitivity of the sensor increases slightly with Lo .

May 1, 2011 / Vol. 36, No. 9 / OPTICS LETTERS

The visibility of the MI fringes decreases with increasing the RI due to the increasing losses of the cladding modes. However, the visibility of the FP fringes increases with increasing the RI, different from the results in [14]. One possible explanation is the curved surface tip, where the core and cladding modes are reflected, is able to refocus back the increasing fraction of the modes into the PCF core after it expands through the collapsed region, increasing the amount of light reflected. This can be seen from the Gaussian optics (paraxial theory) equation [15], n=so þ ne =si ¼ ðne − nÞ=R, where so ¼ 159:26 μm is the displacement from tip where the light starts to diverge, si is the displacement where light converge from the curved surface, R ¼ −25 μm is the radius of curvature of the curved surface estimated from Fig. 1(c), ne is the external RI, and n ¼ 1:45. si increases from −290 to −200 μm when ne increases from 1.337 to 1.395. The decreasing of the negative value of si to be closer to so means more light is able to focus back into the core. To illustrate the excitation of thereflected cladding modes, sensors with three different configurations were fabricated. The three configurations are absent of the first collapsed region to eliminate the cladding modes, a different PCF’s length without collapsing of the fiber’s ends to verify the additional phase delay in the second collapsed region, and, finally, a flat end to eliminate any refocusing effect of the tip. For the first configuration, reflection spectra before and after collapsing the second region of the 1 mm long PCF sensor are shown in Fig. 3(a). Absence of the MI fringes indicates that the cladding mode is needed for the occurrence of the MI interference. The increase of the FP fringes’ visibility within a second collapsed region validates the earlier explanation regarding refocusing of reflected modes. For the second configuration, sensors with PCF’s length of 50 and 1 mm were fabricated, and the reflection spectra are shown in Fig. 3(b). The presence of MI fringes from the longer PCF indicates that ϕ alone is not large enough for MI fringes from the 1 mm sensor to be observed in the same spectral window. Hence, the MI fringes in Fig. 2(a) must be due to the excitation of higher order modes. For the final configuration, a 2 mm long PCF sensor was fabricated by first splicing and collapsing between two SMFs. A cut was made by a mechanical cleaver on one of the collapsed region to produce a sensor with two collapsed regions and a flat end. No MI fringe is observed in Fig. 3(c). An explanation is the cladding mode is reflected back directly near 0° from the flat surface. With these configurations, it can be confirmed that the rounded tip increases the cladding mode incidence angles leading to excitation of higher order modes, which lead to additional phase delay. In conclusion, a miniature PCF RI sensor based on a rounded tip inducing mode excitation is proposed. This increase of effective cladding RI has enabled reduction of the sensor’s area and increased the RI measuring sensitivity. In fact, it could be 10 times shorter when

1733

Fig. 3. (Color online) Reflection spectra of the sensors with different configurations: (a) in the absence of the first collapsed region, with and without the second collapsed region, (b) in the absence of the second collapsed region, with 1 and 50 mm long PCF, and (c) with a collapsed flat end.

compared to [8,9], yet exhibit much higher sensitivity (RI of 1.337 to 1.395). The proposed scheme can be coated easily with biofilms for sensing, is easily reuseable, has low temperature sensitivity, has a simple fabrication process, and is mechanically stable since cladding is kept intact. References 1. L. Rindorf and O. Bang, Opt. Lett. 33, 563 (2008). 2. D. Paladino, A. Iadicicco, S. Campopiano, and A. Cusano, Opt. Express 17, 1042 (2009). 3. T. Wei, Y. K. Han, Y. J. Li, H. L. Tsai, and H. Xiao, Opt. Express 16, 5764 (2008). 4. Z. L. Ran, Y. J. Rao, W. J. Liu, X. Liao, and K. S. Chiang, Opt. Express 16, 2252 (2008). 5. J. L. Kou, J. Feng, Q. J. Wang, F. Xu, and Y. Q. Lu, Opt. Lett. 35, 2308 (2010). 6. D. K. C. Wu, B. T. Kuhlmey, and B. J. Eggleton, Opt. Lett. 34, 322 (2009). 7. J. B. Jensen, L. H. Pedersen, P. E. Hoiby, L. B. Nielsen, T. P. Hansen, J. R. Folkenberg, J. Riishede, D. Noordegraaf, K. Nielsen, A. Carlsen, and A. Bjarklev, Opt. Lett. 29, 1974 (2004). 8. R. Jha, J. Villatoro, G. Badenes, and V. Pruneri, Opt. Lett. 34, 617 (2009). 9. K. S. Park, H. Y. Choi, S. J. Park, U. C. Paek, and B. H. Lee, IEEE Sensors J. 10, 1147 (2010). 10. Y. Jung, S. Kim, D. Lee, and K. Oh, Meas. Sci. Technol. 17, 1129 (2006). 11. A. Messica, A. Greenstein, and A. Katzir, Appl. Opt. 35, 2274 (1996). 12. J. Homola, Sensor Actuators. B Chem. 41, 207 (1997). 13. R. C. Kamikawachi, I. Abe, A. S. Paterno, H. J. Kalinowski, M. Muller, J. L. Pinto, and J. L. Fabris, Opt. Commun. 281, 621 (2008). 14. Y. J. Rao, M. Deng, D. W. Duan, and T. Zhu, Sensor Actuators. A Phys. 148, 33 (2008). 15. C. A. Bennett, Principles of Physical Optics (Wiley, 2008).