Development of an in-fiber white-light interferometric distance sensor for absolute measurement of arbitrary small distances

Development of an in-fiber white-light interferometric distance sensor for absolute measurement of arbitrary small distances Ayan Majumdar and Haiying Huang* Mechanical and Aerospace Engineering Department, University of Texas at Arlington, 500 W. First Street, Arlington, Texas 76019, USA *Corresponding author: huang@uta.edu Received 28 February 2008; revised 3 April 2008; accepted 14 April 2008; posted 15 April 2008 (Doc. ID 93174); published 14 May 2008 The fabrication, implementation, and evaluation of an in-fiber white-light interferometric distance sensor that is capable of measuring the absolute value of an arbitrary small distance are presented. Taking advantage of the mode-coupling effect of a long-period fiber grating, an additional cavity distance is added to the optical path difference of the distance sensor; therefore, it can generate a sufficient number of fringes for distance demodulation even if the free-space cavity distance is very small. It is experimentally verified that the distance sensor is capable of measuring small distances that are beyond the capability of a Fabry Perot interferometric distance sensor. 2008 Optical Society of America OCIS codes: 060.2370, 120.3180, 120.4610, 180.4243. 1. Introduction 0003-6935/08/152821-08$15.00/0 2008 Optical Society of America Optical fiber sensors have been widely exploited for displacement measurement [1], temperature sensing [2,3], medical diagnosis [4], and confocal microscopy [5,6], due to their compact size, light weight, remote operation, capability to operate in harsh environments, and immunity to electromagnetic interference. Among different measurement schemes, distance/displacement measurement is one of the most common and widely applied techniques that often serves as the base for the sensing of other physical parameters such as pressure [4], shear stress [7], vibration [8], and acceleration [9], just to name a few. Optical fiber-based white-light interferometric (WLI) distance sensors have attracted a lot of attention recently because they can overcome the shortcomings of coherent interferometric distance sensors such as directional ambiguity and the need of a reference measurement [10]. Since an incoherent light source with a broadband light spectrum, i.e., a white light, is utilized as the illumination source for the WLI distance sensors, the resulting interference fringes can be presented in the wavelength domain, from which the absolute distance measurement can be demodulated without requiring any reference signal. An optical fiber-based extrinsic Fabry Perot interferometric (FPI) distance sensor is probably one of the most popular WLI distance sensors because of its simple construction and ultrahigh distance resolution [11]. Even though WLI distance sensors can achieve absolute measurement from its interference spectrum, they require a data interrogation process to demodulate the distance. Data interrogation algorithms such as fringe counting [1], Fourier transform spectroscopy [9], improved fringe matching technique [12], and least square fitting [13] have been published. In general, the more the fringes are presented in the interference fringe spectrum, the higher the accuracy a data demodulation algorithm can achieve. More importantly, in order for the data demodulation algorithm to be applicable, a minimum number of interference fringes must be presented in a given spectral bandwidth. The narrow spectral bandwidth of a white-light source, therefore, imposes a limit on how small a distance a WLI distance sensor can 20 May 2008 / Vol. 47, No. 15 / APPLIED OPTICS 2821

measure. This limits its dynamic range and its application for near-field surface profiling. Near-field surface profiling using a near-field scanning optical microscope (NSOM) is a very useful technique for the optical investigation of materials in the subwavelength range, particularly in the nano/bio areas [14]. A NSOM positions a tapered optical fiber probe in the near-field surface of the sample and thus overcomes the diffraction limit of conventional optical microscopes. However, the technique s further advancement, especially in the field of biology and surface chemistry, is hindered due to its position feedback mechanism [15]. Moreover, despite a variety of light collecting techniques in NSOM, the basic image construction technique relies solely on measuring the intensity of the sample light using a photomultiplier tube. Therefore, its signal is susceptible to noises induced by laser power fluctuation and optical fiber bending. Since its capability for ultraprecise absolute distance measurement can be exploited for position feedback as well as for surface profiling, a WLI distance sensor that can conduct surface profiling in the near-field of the sample surface has obvious advantages. However, in order for a WLI distance sensor to be applicable for near-field surface profiling, it must have the capability of measuring distances as small as a few tens of nanometers, a requirement that can not be satisfied by existing optical fiber-based WLI distance sensing techniques. Here, we present the development of an optical fiber WLI distance sensor that can measure arbitrarily small distance by introducing a long-period fiber grating (LPFG) at a distance from the fiber end. First, the fabrication of the sensor probe and the mechanism for mechanically inducing LPFG in a single mode fiber (SMF) are discussed. To integrate the mechanically induced LPFG with the specially fabricated sensor probe, an in-fiber WLI distance sensor system was implemented and evaluated. The experimental results have confirmed that the sensor is capable of measuring distances smaller than that can be measured by a white-light FPI distance sensor. 2. Principle of Operation The in-fiber LPFG-based WLI distance sensing system, as shown in Fig. 1, consists of a broadband light source, an optical circulator, a distance sensor, and an optical spectrum analyzer (OSA). The light generated by the broadband light source propagates along the optical fiber and is routed toward the distance sensor by an optical circulator. The distance sensor probe, as shown in Fig. 2, is simply a cladding-coated single mode fiber (SMF) with a LPFG located at a distance from its cleaved end. The LPFG serves as a beam splitter that separates the core propagating light into a core mode and a cladding mode. Passing the LPFG, the core mode continues to travel along the fiber core toward the sample under test, as indicated by the solid arrow in Fig. 2. At the end of the optical fiber, it exits the probe, propagates in the air, then is reflected by the sample, and finally is coupled back to the fiber core and propagates toward the LPFG. The light that is coupled into the cladding by the LPFG propagates along the cladding (dashed arrows in Fig. 2) until it reaches the mirrored end of the probe. Reflecting from the mirror, this cladding mode will trace its path back to the LPFG. Based on the reciprocity principle of light propagation, the reflected cladding mode will be coupled back to the core mode. When the cladding mode is recombined with the core mode, the two light waves interfere constructively and destructively because of the different optical paths they have traveled. By routing these two reflected lights to the OSA using the optical circulator, the interference fringe spectrum can be measured. The OPD between the core mode and the cladding mode is contributed to by the difference between the refractive indices of the cladding and the fiber core, and the cavity between the fiber end and the sample surface, i.e., OPD ¼ 2ðΔnL þ n m dþ; ð1þ where Δn ¼ n clad n core is the difference between the refractive indices of the fiber core and cladding, L is the distance between the LPFG and the fiber end, n m is the refractive index of the medium in the free-space (for air, n m ¼ 1), and d is the cavity distance between the fiber end and the sample. Without the presence of the LPFG and the mirrored fiber end, Fig. 2 is reduced to the FPI distance sensor where the two interfering light waves are contributed to by the light reflected from the fiber/air interface Fig. 1. In-fiber LPFG-based WLI distance sensing system. Fig. 2. White-light interference generated by LPFG and cladding-coated sensor probe. 2822 APPLIED OPTICS / Vol. 47, No. 15 / 20 May 2008

and that which is reflected from the sample surface. Therefore, the LPFG-based distance sensor is similar to the FPI distance sensor except that its OPD is contributed to by both the LPFG cavity distance L and the FPI cavity distance, which is equal to the target distance d. For a white-light interferometric sensor, the general expression for the fringe spectrum as a function of the wavelength λ is p RðλÞ ¼R 1 þ R 2 þ 2 ffiffiffiffiffiffiffiffiffiffiffi R 1 R 2 cosð2πopd=λ þ θþ; ð2þ where R 1 and R 2 are the intensities of the two interfering lights and θ is the phase shift. For a whitelight FPI distance sensor, OPD ¼ 2n m d. Therefore, Eq. (2) becomes p RðλÞ ¼R 1 þ R 2 þ 2 ffiffiffiffiffiffiffiffiffiffiffi R 1 R 2 cosð4πn m d=λ þ θþ: ð3þ Figure 3 illustrates a normalized fringe spectrum of a white-light FPI distance sensor. The spacing and the positions of the peaks and valleys, i.e., the fringes, are solely determined by the cavity distance d, assuming n m ¼ 1. As the cavity distance d decreases, the spacing between the fringes increases. Since the distance d is calculated from the positions of two adjacent fringes, e.g., λ 1 and λ 2, d ¼ λ 1 λ 2 2n m ðλ 2 λ 1 Þ ; ð4þ it requires the presence of at least two fringes within the bandwidth of the white-light source. This imposes a limit on the smallest distance a FPI distance sensor can measure. For a LPFG-based distance, however, the reflectance RðλÞ is contributed to by both ΔnL and d, i.e., p RðλÞ ¼R 1 þ R 2 þ 2 ffiffiffiffiffiffiffiffiffiffiffi R 1 R 2 cos½4πðδnl þ dþ=λ þ θš: ð5þ Therefore, the distance d can be arbitrarily small as long as the additional OPD term ΔnL is large enough so that a minimum number of fringes are presented. 3. Sensor Fabrication Techniques The LPFG-based distance sensor requires the integration of two optical fiber concepts; a mirrored surface that only covers the cladding region of the optical fiber and an LPFG that couples a part of the incoming light into the cladding region. In Section 3.A we describe the microlithographic technique we developed to fabricate the cladding-coated sensor probe and the mechanism for mechanically inducing an LPFG in an optical fiber. A. Fabrication of Cladding-Coated Sensor Probe To fabricate a mirrored surface on the cladding region of the optical fiber end face, an innovative yet simple fabrication technique based on microlithography is invented. The concept of the fabrication process is illustrated in Fig. 4. After depositing a negative photoresist on the fiber end, an ultraviolet (UV) light is coupled into the end of the optical fiber that is not coated with the photoresist [Fig. 4(a)]. Because of the multimode propagation of the UV light in the SMF, the UV light is confined within the fiber core and thus only cures the photoresist covering the fiber core when it exits the optical fiber at the other end. As a result, the fiber core is masked by the cured photoresist while the uncured photoresist can be easily removed [Fig. 4(b)]. Metallic thin film is then deposited on the cladding region and the photomask [Fig. 4(c)]. Finally, the photomask is etched to expose the fiber core again [Fig. 4(d)]. The experimental setup for coupling the UV light to the optical fiber is shown in Fig. 5. An UV lightemitting diode (LED) (Nichia NSHU550A, 375 nm wavelength) was chosen as the UV light source. The cleaved optical fiber was mounted on a three-axis fiber alignment stage and manually aligned with the LED. Different coupling mechanisms were experimented. It was discovered that the maximum coupling from the LED to the optical fiber is achieved by removing the glass cover of the LED and directly butting the optical fiber against the LED chip. An optical power meter was employed to monitor the output power of the optical fiber during the alignment process to provide position feedback for the alignment. With a forward current of 50 ma supplied to the UV LED, an output power of 180 nw was achieved at the end of a 50 cm long SMF. Even though the total output power was low, the power density of the UV light was sufficient to cure the epoxy in a short time. Fig. 3. Normalized fringe pattern from a WLI distance sensor. Fig. 4. Sensor probe fabrication process. 20 May 2008 / Vol. 47, No. 15 / APPLIED OPTICS 2823

position the surface was inspected again under the microscope to make sure that a good quality metallic coating had been achieved. The final fabrication step involves submerging the fiber end in Acetone for 40 50 s to dissolve the cured polymer tip. Micrographic images of the SMF afterwards confirm that the polymer tip has been removed and the fiber core is exposed again, as shown in Figs. 6(c) and 6(d). B. Fig. 5. Experimental setup to fabrication the sensor probe. The intensity of the UV light can be conveniently adjusted by changing the supplied current. It s worth noting that the optical fiber should be long enough so that the UV light coupled into the fiber cladding is dissipated out and only the light propagating in the fiber core remains. After one end of the SMF was aligned with the LED chip, the free end of the SMF was dipped into a pool of UV-curable optical adhesive (Norland Products, NOA61) and quickly removed from it. Because of the surface tension, a small amount of uncured epoxy is deposited, forming a dome at the end of the optical fiber. The UV LED was then switched on, exposing the uncured epoxy to the UV light. After irradiating the NOA61 for 1 to 2 min, a polymer tip was formed on the top of the fiber core, as shown in Fig. 6(a). After the epoxy was fully cured, the fiber end was washed in Acetone in order to remove any uncured epoxy from the surface. Inspecting the fiber-end surface using an optical microscope revealed that a clean surface was obtained, as shown in Fig. 6(b). Metal deposition was then carried out in a sputtering machine (CrC 100) at 100 ma for 3 min. After the metal de- Fig. 6. Micrographic images of the sensor probe at different fabrication steps. 2824 APPLIED OPTICS / Vol. 47, No. 15 / 20 May 2008 Implementation of Mechanically Induced LPFG An LPFG can be either permanently engraved in a SMF using a high power UV laser or mechanically induced by introducing periodic microbending. Mechanically induced LPFG is preferred for our purpose because the LPFG cavity distance can be easily adjusted. A diagram of a mechanically induced LPFG is shown in Fig. 7(a). A conventional SMF is placed between two metal blocks; one with periodic notches and one with a flat surface. As pressure is applied to the metal blocks, microbending is introduced into the fiber, generating a LPFG mode-coupling effect. The properties of the LPFG are determined by the length of the metal blocks, the period of the notches, and the pressure applied. The implementation of the mechanically induced LPFG is shown in Fig. 7(b). Instead of using a grooved block, an aluminum block wound with a cooper wire of 500 μm in diameter was employed to apply periodic pressure to the optical fiber [see Fig. 7(c)]. A 100 turn-per-inch (TPI) lead screw was clamped on top of the metal block so that the applied pressure could be easily controlled by adjusting the lead screw. As shown in Fig. 8, the setup is flexible to achieve LPFG spectra with different center frequencies, bandwidths, and coupling coefficients. 4. Experimental Setup The performance of the LPFG-based WLI distance sensor was evaluated using the experimental setup shown in Fig. 9. The mechanically induced LPFG described in Section 3.B was mounted on an elevated platform. The cladding-coated optical fiber probe was held in front of a mirror with the help of a fiber chuck clamped in a V-groove holder. A mirror, serving as the target, was placed on a rotation stage, which was in turn mounted on top of a horizontal translation stage. The movement of the translation stage was controlled by a motorized actuator. A servo controller (Thorlabs APT-DC) was employed to control the motor and to provide feedback on the mirror position. An er- Fig. 7. Mechanically induced LPFG.

Fig. 8. Difference spectra obtained using mechanically induced LPFG. bium doped fiber amplifier (EDFA) with a wavelength bandwidth of 40 nm (1540 1580 nm) was utilized as the broadband light source. The transmission spectrum of the mechanically induced LPFG was adjusted accordingly to fall within this range. 5. Experimental Results The fringe spectra of the LPFG-based distance sensor, acquired at different cavity distances, were first normalized with the spectrum of the laser source. Fast Fourier transform (FFT) was then applied to the normalized fringe spectra in order to obtain their frequency components, after converting the wavelength to wave number. Based on Eq. (2), the frequency position f p of the FFT peak is directly correlated to the OPD, i.e., f p ¼ OPD. Therefore, for a LPFG-based WLI distance sensor, the free-space cavity distance d can be calculated from the FFT peak frequency directly if ΔnL is known, assuming nm ¼ 1. During the course of the experiment, it was observed that the interference fringes could still be observed even when the mirror was not placed in front of the sensor probe. These fringes are due to the interference between the reflected cladding mode and the core mode reflected at the fiber/air interface,, the Fresnel reflection. Because the frequency of this Fig. 9. Experimental setup for LPFG-based distance sensor. fringe spectrum is only determined by the LPFG product ΔnL, this proves to be advantageous since it enables the precise determination of ΔnL. Based on the FFT spectrum shown in Fig. 10, the ΔnL product for our sensor was calculated to be around 503:14 μm. Since the cavity distance L, measured roughly from the center of the grating to the fiber end, was 11:5 cm, the refractive index difference Δn between the fiber cladding and the fiber core is estimated to be 0.0043, which is close to the typical Δn value of 0.003 for a SMF. When a mirror is placed in front of the sensor probe, the reflectance spectra are contributed to by the following three interferences: I. The interference between the reflected cladding mode and the core mode reflected by the fiber/air interface, i.e., the LPFG effect. II. The interference between the core mode reflected at the fiber/air interface and the core mode reflected by the mirror, i.e., the FPI effect. III. The interference between the reflected cladding mode and the core mode reflected by the mirror, i.e., the combined LPFG and FPI effect (LPFGþ FPI interference). The LPFG fringe spectrum, i.e., from interference I, is independent of the FPI cavity distance d and is Fig. 10. Sensor reflectance spectrum and its FFT without the presence of a mirror. 20 May 2008 / Vol. 47, No. 15 / APPLIED OPTICS 2825

Fig. 11. (a) and (b) Fringe spectrum from three interferences and its FFT, (c) and (d) FPI fringe spectrum obtained using a bandpass filter ½ 0:65 e 3 0:8 e 3 and its FFT, (e) and (f) LPFG fringe spectrum obtained using a bandpass filter ½ 0:8 e 3 1:2 e 3 and its FFT, (g) and (h) LPFG þ FPI fringe spectrum obtained using a bandpass filter ½ 1:25 e 3 2 e 3 and its FFT. 2826 APPLIED OPTICS / Vol. 47, No. 15 / 20 May 2008

identical to the fringe spectrum obtained when the mirror is not presented. The other two interferences, however, are directly controlled by the cavity distance d. The resulting interference fringes from these three interferences are shown in Fig. 11(a). The FFT of the interference fringes clearly demonstrated the existence of these three interferences, as shown in Fig. 11(b). The locations of the three frequency peaks in the FFT spectrum correspond to the OPD of the three interferences. Based on Eq. (1), the OPD of the LPFG þ FPI interference is equal to the OPD of the LPFG interference plus that of the FPI interference. Therefore, the frequency spacing between the FPI peak and the LPFG þ FPI peak in the FFT spectrum should be equal to the frequency of the LPFG peak, a constant value determined by ΔnL. At a large cavity distance d, the FPI frequency peak can be easily identified from the FFTof the reflectance spectrum while the LPFG þ FPI interference has a low visibility (see Fig. 11(b)). Since the frequency location of the LPFG þ FPI interference can be calculated from the frequency locations of the LPFG and FPI interferences, an appropriate bandpass filter can be designed to isolate the fringes contributed by the LPFG þ FPI interference and thus improve its visibility. Figures 11(c), 11(e), and 11(g) show the fringes corresponding to these three interferences, obtained by passing the reflectance fringe spectrum through the respective bandpass filters. The FFT spectra of these fringes are shown in Figs. 11(d), 11(e), and 11(h), from which the OPDs for the three interferences can be obtained. Figure 12 compares the cavity distances measured from the FPI interference with those measured from the LPFG þ FPI interference. The distance between the sensor probe and the mirror was adjusted by traversing the mirror using the motorized translation stage. The mirror was moved towards the sensor in steps of 20 micrometers. At each position, the reflectance spectrum of the LPFG-based WLI distance sensor was recorded, from which the cavity distance was calculated from the positions of the FPI peak and the LPFG þ FPI peak independently. As shown in Fig. 12, Fig. 12. Distances estimated from the FPI and the LPFG þ FPI interference fringes. the distances measured from both interferences demonstrated a linear relationship with the traversed distances, calculated from the motor position feedback. In addition, the two lines are parallel to each other, indicating that the difference between the FPI cavity distance and the LPFG þ FPI cavity distance always remains a constant. As the free-space cavity distance d reduces, the FPI peak will shift toward a lower frequency while the LPFG þ FPI peak will shift toward the LPFG peak. In another word, the number of fringes presented in a given spectral bandwidth reduces. When the cavity distance d is less than 83 μm, the number of fringes generatedbythefpieffectistoosmallforthefpipeak to be visible in the FFT spectrum. Because of the contribution of the LPFG cavity, however, the LPFG þ FPI peak would still be visible. Therefore, the freespace cavity distance d can still be measured from the LPFG þ FPI peak even though it is too small to be measured from the FPI interference. Figure 13(a) shows the filtered reflectance spectrum of the LPFG-based WLI distance sensor at a cavity distance in which the FPI fringes are no longer presented. The shift of the LPFG þ FPI frequency peak in the FFT spectrum, as the cavity distance d decreases, is shown in Fig. 13(b). At 83 μm, two distinct frequency peaks, i.e., the LPFG peak and the LPFG þ FPI peak, can be resolved clearly and they have roughly the same intensity. As the distance reduces to 40 μm, the LPFG þ FPI peak starts to merge with the LPFG peak. In order to identify the peak location of the LPFG þ FPI peak Fig. 13. (a) Filtered reflectance spectrum at a distant smaller than 83 μm; (b) FFTs of the reflectance spectra at small distances. 20 May 2008 / Vol. 47, No. 15 / APPLIED OPTICS 2827

Fig. 14. Linear relationship between the distances moved by the target and the distances measured by the LPFG-based WLI distance sensor. precisely, the FFT spectrum of the LPFG þ FPI interference was calculated by subtracting the LPFG spectrum from the overall FFT spectrum. The peak frequency of the resulting LPFG þ FPI spectrum can then be identified and utilized to deduce the cavity distance d. Figure 14 shows the relationship between the distance traversed by the translation stage as well as the distance measured from the LPFG þ FPI peak. A linear relationship is demonstrated over the cavity distances ranging from 350 to 4:5 μm. This experiment proves that the LPFG-based WLI distance sensor is capable of measuring distances that are much smaller than those can be measured by a white-light FPI distance sensor. 6. Conclusions The fabrication, implementation, and evaluation of an in-fiber LPFG-based WLI distance sensor are presented. It is demonstrated that the distance sensor is capable of measuring small distances that a conventional FPI distance sensor cannot measure. The authors acknowledge the National Science Foundation for its financial support (award # CMS-0528681). The support and encouragement of the program manager, Shih-Chi Liu, is greatly appreciated. The authors also thank the reviewer for his/her insightful comments. References 1. V. Arya, K. A. Murphy, A. Wang, and R. O. Claus, Microbend losses in single-mode optical fibers: theoretical and experimental investigation, J. Lightwave Technol. 13, 1998 2002 (1995). 2. L. Ferreira, A. B. L. Ribeiro, J. L. Santos, and F. Farahi, Simultaneous measurement of displacement and temperature using a low finesse cavity and a fiber Bragg grating, IEEE Photon. Technol. 8, 1519 1521 (1996). 3. P. Swart, Long-period grating Michelson refractometric sensor, Meas. Sci. Technol. 15, 1576 1580 (2004). 4. K. Totsu, Y. Haga, and M. Esashi, Ultra-miniature fiber-optic pressure sensor using white light interferometry, J. Micromech. Microeng. 15, 71 75 (2005). 5. T. Wilson, Coherent methods in confocal microscopy, IEEE Engin. Med. Bio. Mag. 15, 84 91 (1996). 6. L. Yang, G. Wang, J. Wang, and Z. Xu, Surface profilometry with a fiber optical confocal scanning microscope, Meas. Sci. Technol. 11, 1786 1791 (2000). 7. C. Lin and F. Tseng, A micro Fabry Perot sensor for nanolateral displacement sensing with enhanced sensitivity and pressure resistance, Sens. Actuators, A 114, 163 170 (2004). 8. T. Gangopadhyay, Non-contact vibration measurement based on extrinsic Fabry Perot interferometer implemented using arrays of single-mode fibers, Meas. Sci. Technol. 15, 911 917 (2004). 9. G. Schrofer, W. Elflein, M. de Labachelerie, H. Porte, and S. Ballandras Lateral optical accelerometer micromachined in (100) silicon with remote readout based on coherence modulation, Sens. Actuators, A 68, 344 349 (1998). 10. B. T. Meggit, Fiber optic white light interferometric sensors, in Optical Sensor Technology (Chapman and Hall, 2000), pp. 193 238. 11. K. A. Murphy, M. E. Gunther, A. Wang, R. O. Claus, and A. M. Vengsarkar, Extrinsic Fabry Perot optical fiber sensor, in Proceeding of the 8th Optical Fiber Sensor Conference (IEEE, 1992), pp. 193 196. 12. H. Huang, Data Interrogation for Fabry Perot white-light interferometry, Proc. SPIE 6174, 617319 (2006). 13. M. Han, Y. Zhang, F. Shen, G. R. Pickrell, and A. Wang, Signal-Processing algorithm for white-light optical fiber extrinsic Fabry Perot interferometric sensors, Opt. Lett. 29, 1736 1738 (2004). 14. R. Dunn, Near-field scanning optical microscopy, Chem. Rev. 99, 2891 2927 (1999). 15. H. Shiku, J. R. Krogmeier, and R. C. Dunn, Noncontact near-field scanning optical microscopy imaging using an interferometric optical feedback mechanism, Langmuir 15, 2162 2168 (1999). 2828 APPLIED OPTICS / Vol. 47, No. 15 / 20 May 2008