Article pubs.acs.org/ac
Whispering Gallery Mode Dip Sensor for Aqueous Sensing Monica Agarwal† and Iwao Teraoka* Department of Chemical and Biomolecular Engineering, Polytechnic School of Engineering, New York University, 6 MetroTech Center, Brooklyn, New York 11201, United States S Supporting Information *
ABSTRACT: We report fabrication of a 4 mm thick, preassembled whispering gallery mode (WGM) sensor that can be repeatedly dipped into aqueous solutions and lifted. We built the viable photonic sensor assembly by bending an optical fiber by 90° and molding its tip into a sphere, thus, a long stem holding a submillimeter sensor at the end of a short arm of fiber, and positioning a pair of parallel cone-tipped tapers along the long stem so that the tips touch the sensor. Our sensor head is an optical fiber device just a few millimeters thick and yet has a sensitivity of the resonance wavelength shift comparable to the one obtained with conventional WGM sensors in a planar arrangement. Since dipping and lifting from the solution changes the temperature of the sensor, affecting the resonance wavelength, we enclosed a thermistor within the sensor head to monitor the temperature. We demonstrate that the resonance shift in repeated transfer of the sensor head between water and a solution of sucrose, after correction by the temperature change, is reproducible and agrees with a theoretical estimate of the shift for different concentrations.
P
waveguide that feeds and picks up the light is thin. None of the parts move during measurement. The resonator is typically made of silica, and therefore, well-established methods of ligand immobilization can be applied.13 These inherent advantages are ideal for a dip sensor that facilitates interfacing with widely used well plates. One of the most difficult problems of WGM sensors, in any geometry, is mechanical robustness of the sensor head: Coupling of light between the thin waveguide and the resonator is achieved by precarious positioning of the resonator in three dimensions with micrometer precision and therefore is extremely vulnerable to the environment. Mechanical disturbance and noises commonly encountered on lab benches are sufficient to change the profile of resonance spectrum or, worse, knock the resonator out of resonance. This problem will become serious when developing a preassembled dip sensor. The other problem that we encounter when trying to construct a dip sensor is that the stem and the coupling fiber are perpendicular to each other and, therefore, are in planar arrangement.5,6 A preassembled dip sensor must have collinear arrangement of the stem and the coupling fiber, and their coupling must withstand mechanical shocks once they are assembled. To satisfy the two requirements, we developed a novel design of the dip sensor, which is briefly described below: We bent the stem by ∼90° near the neck of the resonator and, thus, could place the coupling fibers along the stem. We used two separate fiber tapers for light feed and pick up. Each taper was not a
hotonic sensors based on resonance wavelength shift of whispering gallery mode (WGM) were introduced more than a decade ago.1,2 Unparalleled sensitivity to detect the binding of a receptor to surface-immobilized ligand was demonstrated using a resonator with smooth curved surface of circular symmetry such as a sphere, a cylinder,3 and a toroid.4 However, a need for frequent optical alignment,5,6 vulnerability to environmental noises,7 and short lifetime of the sensor8 has limited its use primarily to laboratories of optical scientists. Here we show that the WGM sensor can be made into a preassembled dip sensor housed in a cylindrical body less than a few mm thick. Our sensor assembly requires no further alignment, withstands shocks of dipping into water and lifting from it, lasts for months, and yet has a sensitivity comparable to the conventional bulky arrangement of the mechanically delicate WGM sensor. Usually, a WGM sensor is prepared as follows.5,6 A submillimeter resonator of circular symmetry is fabricated at the tip of a mechanical support (stem), for example, a sphere is molded by melting the tip of an optical fiber. The stem is mounted on a three-dimensional translation stage, and the position of the resonator is adjusted relative to a waveguide (typically, a core-exposed single-mode fiber) to optimize the evanescent coupling. One of the ends of the waveguide leads to a tunable laser and the other end to a photodetector. The sensor, together with a part of the waveguide, is enclosed in a fluidic device in a way similar to those found in other resonance-based biosensor instruments such as surface plasmon resonance (SPR),9 quartz crystal microbalance,10 and interferometer-based instruments.11,12 Unlike sensors in these other methods, WGM’s resonator is tiny, typically less than 1 mm in any direction, and the © XXXX American Chemical Society
Received: August 10, 2015 Accepted: September 21, 2015
A
DOI: 10.1021/acs.analchem.5b03066 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry thinned core-exposed fiber of constant thickness, but rather a cone formed at the end of a fiber. Light leaks from the tip of the cone to transfer to the resonator. The coupling between the resonator and the cones withstood the disturbance when dipped into water. However, the resonance was lost when lifted from water. Obviously, drainage dislodged the fiber tips. We then enclosed the resonator and fiber tips in a glass tubing which retained water or solution by capillary effect, thereby minimizing the drainage shock. We could see resonance and track resonance peaks throughout dipping and lifting. It is often the case that a change in the resonator’s temperature, ΔT, makes a non-negligible contribution to the total resonance wavelength shift Δλ: The fractional shift is as large as 10 ppm/K.14 To monitor ΔT and estimate its contribution to Δλ in water, we installed a small thermistor next to the resonator and within the outer glass tubing. A recent study15 employed a mixture of water and glycerol surrounding a silica resonator to bring the coefficient near zero. In the past, fiber dip sensor-based SPRs were designed and fabricated.16−18 A recent report shows that an SPR dip sensor that had a long-period fiber grating near the fiber tip measured the refractive index (RI) of a liquid with an accuracy of 1 × 10−5.19 Microcantilever dip sensor was also considered.20 An interferometry-based fiber optic dip sensor was developed that has a photonic crystal at the end. The sensor measured the RI of a liquid with an accuracy of 2.6 × 10−5.21 Our dip sensor produced a comparable resolution.
Figure 1. (a) Spherical resonator formed at the tip of a 90° bent stem. The stem diameter is 425 μm. (b) A pair of nearly parallel single-mode fibers honed into cones, glued to each other. The straight portion has a diameter of 125 μm. (c) Spherical resonator coupled to the two cones in a preassembly. (d) Resonator−fiber assembly and a thermistor encased in an outer tubing. The photograph was taken after being lifted from water. Images (a) and (c) are tilted for a better view.
assembly was fixed by glue to the outer wall of the capillary. Figure 1c shows the head of the assembly. (4) We placed the assembly in an ∼5 mm long outer tubing (i.d. = 3.5 mm, o.d. = 4.0 mm) together with a thermistor (NTC 10K) and glued them, see Figure 1d. The distance between the tip of the fiberresonator assembly and the edge of the outer tubing was ∼1 mm. When dipped into water, the outer tubing was filled with water. Most of the water was retained when lifted. We used a fiber of 400 μm diameter to fabricate a resonator. This diameter gave sufficient rigidity to keep the resonator always coupled to the tapers. The distance between the two cone tips was around 0.6 mm, determined by our gluing procedure. With this spacing, the cones snugly fit the resonator. Resonance Spectrum Measurement. We used a spectrometer developed earlier23 except that we used a different light source emitting at around 1070 nm (QDL1061-AE, QD Laser, Yokohama), unless otherwise mentioned. We also added a polarizer, followed by a mating sleeve on a rotation stage, to select transverse magnetic (TM) modes. Figure 2 shows a diagram of the measurement system that involves the dip sensor. The current of a distributed feedback (DFB) laser was changed in a triangular wave (∼24.40 Hz) to scan the wavelength, and the light was fed into one of the fibers in the assembly. The relationship between the laser current and the wavelength was found using a wavelength meter (Exfo-WA1650). The cone-shaped taper made the coupling weak; thus, the apparent Q value estimated from the resonance spectrum was high, ∼1 × 107 in air and ∼5 × 106 in water at 1070 nm. At 1310 nm, the Q value was ∼1 × 107 in air and ∼1 × 106 in water. The light picked up by the other fiber was connected to a photodetector (Thorlabs, PD10CS, 1.5 × 105 V/A for a resonator in water), and its output was digitized at 50 kHz. Each of the up and down current scans of the triangular wave and the photodetector signal consisted of 1024 points (channels) of data. We analyze the down scan data only, since the linearity between the laser current and the wavelength is far better in the down scan compared with the up scan.24 As the laser current is changed in a triangular wave, the wavelength changes nearly in a triangular wave. Concomitantly,
■
EXPERIMENTAL AND METHODS SECTION Fabrication of a Dip Sensor. We used two types of optical fibers for the present study. A multimode fiber (FT400EMT, ThorLabs) that has a pure silica core (400 μm) and TECS cladding (425 μm), coated with Tefzel (730 μm), and a singlemode fiber (smf28e+, Corning) that has a germanium-doped silica core (8.2 μm) and pure silica cladding (125 μm), coated with acrylic (245 μm). We assembled the sensor in four steps. (1) The resonator was fabricated on a stem by first bending the multimode fiber at ∼90° using an electric arc (bend radius ∼0.5 mm), cutting the bent arm to ∼1 mm, and then using the arc to ball up the tip to a diameter of ∼0.6 mm, see Figure 1a. The stem was then glued onto the interior wall of a slant-cut glass capillary (i.d. = 0.81 mm, o.d. = 1.57 mm). (2) We honed the single-mode fiber into a cone by etching it in hydrofluoric acid solution covered with silicone oil.22 The etching was done as follows. A pair of fibers were cleaned with ethanol after removing the acrylic coating, exposing the glass for ∼10 mm. The glass part was dipped into a hydrofluoric acid (HF) solution (50%), topped with silicone oil. The HF solution forms a raised meniscus along the glass fibers within the layer of the oil, which causes the fibers to etch into a cone shape. After rinsing the honed fibers in ethanol, we glued them close to parallel to each other, see Figure 1b. They were slightly converging to facilitate contact between the cone tips and the resonator. (3) We mounted the resonator−capillary assembly on a fixed horizontal stage with the resonator pointing upward. The parallel fiber assembly was mounted on another horizontal stage that could be moved in XYZ directions. The parts of the fibers near the cone ends ran over the outer wall of the capillary. The fiber assembly was moved to let the tips of the cones touch the resonator’s equator, first by observing them by microscope objectives and then looking at the resonance spectrum. However, it was sufficient to just casually position the fiber assembly by the 3D positioner. Subsequently, the fiber B
DOI: 10.1021/acs.analchem.5b03066 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
around 2 mK. Since the spectrum collection was more frequent, the temperature for a specific spectrum was estimated by interpolating the temperature data.
■
RESULTS AND DISCUSSION For the first test of the dip sensor, we dipped the sensor assembly into water and lifted from it, which was repeated several times. The spectrum was collected and the temperature was recorded all the while. Figure 3a shows some of the spectra for the scan range of 112.2 pm (104.9 ppm). A zoomed view of the spectra is shown in Figure S1 in the Supporting Information. There is a small shift, but the profile of the spectrum remains unchanged. In Figure 3b, the temperature dropped as much as 2 K when lifted, but returned to nearly the same value when dipped again. The instances when the temperature started to rise indicate the moments when the sensor head was dipped again. The period when the sensor was in air was not controlled, and therefore, the dip in the temperature is different for each lifting. The cooling is caused by vaporization of water from the side and bottom of the outer tubing. Figure 3c shows a time trace of the resonance shift, estimated by analyzing all the spectra. The resonance shows a sharp blue shift immediately after being lifted from water, caused by the cooling. The red shift after redipping reaches a steady value with a time constant of ∼0.091 s−1. The steadystate value gradually increased with time, which is echoed by the temperature plot. We are concerned whether all of the resonance shift observed in Figure 3c can be accounted for by the temperature change shown in Figure 3b. We evaluated the resonance shift due to a temperature change by dipping the sensor head into warm water and recording the resonance spectrum and temperature as the water cooled. Results are shown in Figure S2 in the Supporting Information. Good linearity was obtained for the relationship between the temperature change ΔT and the fractional resonance wavelength shift Δλ/λ with a coefficient of 5.92 ppm/K.
Figure 2. Schematic of the measurement system with a dip sensor. A pair of tapers are connected to a light source (distributed feedback (DFB) laser) on one end and to a photodetector on the other end. A thermistor, placed in close proximity to the resonator, is connected to a bridge circuit. The cone tips of the tapers, the resonator, and the thermistor are encased in a glass tubing of 4.0 mm diameter.
the sensor is brought into resonance, one after another, by different modes that may satisfy the resonance condition. The photodetector signal as a function of the laser current looks like a spectrum that holds many resonance lines. Multiplicity of WGM (number of waves around the circular orbit, radial orders, meridional numbers, and polarization) makes the spectrum busy. For a resonator of 0.6 mm diameter, we see more than a hundred lines in a 30 pm scan in air; the number is smaller in water, since a weaker RI contrast between the resonator and the surroundings kicks out WGM of high radial orders. Temperature Measurement. The thermistor (Amphenol, MA100BF103A) was connected to a house-made bridge circuit, and the differential voltage was digitized by a data logger (Omega, OM-CP-Volt101A) at 4 Hz. The resolution was
Figure 3. (a) Resonance spectra collected during repeated dipping of the sensor assembly into water and lifting. The laser current increases from left to right. The concomitant wavelength scan range is shown below the abscissa. The number above each spectrum indicates the instance of the spectrum capture, shown by the arrow in (c). The lightly shaded zone is an eye guide to track a group of peaks. (b) Temperature monitored by the thermistor in the sensor head. The head was in air during the five periods with a decreasing temperature. (c) Resonance shift through the five-time lifting and dipping. (d) Resonance shift, compensated for the temperature change. C
DOI: 10.1021/acs.analchem.5b03066 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 4. (a) Raw resonance shift, plotted as a function of time, during repeated transfer of the sensor head between water and a 0.10 wt % solution of sucrose in water. (b) Resonance shift, compensated for the temperature change. (c) Steady-state value of temperature-compensated, fractional shift of resonance wavelength, plotted as a function of sucrose concentration. The error bar indicates the range of shift in repeated transfer into the solution. The solid line is a theoretical estimate for a TM mode. See the text for the equation and parameter values.
observed resonance shift. We will find whether the temperature-compensated shift is proportional to c and what is the smallest detectable RI change. The sensor head was transferred between water and a 0.10 wt % solution of sucrose a few times. Figure 4a shows the fractional resonance shift as a function of time, not corrected for the temperature change. Again, there is a sharp dip whenever the sensor head was in air. The steady-state value was lower when immersed into the solution compared with the time when the head was in water. The blue shift occurred despite the positive contribution from the ∂ ln λ/∂n2, indicating that the temperature of the solution was lower compared with water. When transferred back into water, the resonance shifted to a longer wavelength compared with the early state before being immersed into the solution. It may be due to heat of mixing. Over repeated transfers between water and the solution, the steady-state value of the shift continued to increase, and the difference between the solution and water continued to decrease. The temperature-compensated shift is shown in Figure 4b. Each time the head was immersed into the solution, the resonance shift approached a steady value in several minutes. When transferred back into water, the shift approached zero. There is a transient of overshoot and undershoot in both transfers, caused by the slower response of the thermistor compared with the resonator. We also find that it takes several minutes for the temperature-compensated shift to reach a steady value, due to the hold-up volume of the outer tubing. The experiment was also done for other concentrations of sucrose. Temperature-compensated fractional shifts are plotted in Figure S4 in the Supporting Information. Figure 4c shows the steady value of the fractional shift as a function of the concentration. The straight line represents eq 2 with neff = 1.41, λ = 1069.72 nm, a = 303.2 μm, n1 = 1.4496, n2 = 1.3259, d n2/dc = 0.144 mL/g,27 that is, Δλ/λ (ppm) = 7.92 × c (wt %). The agreement is good between experimental shifts and theoretical estimates. We also estimate the smallest detectable RI change to be 3 × 10−5. All of these experiments were performed by using a single dip sensor. Unless fouling occurs, the sensor lasts for months.
Thus, obtained coefficient was used to subtract the effect of the temperature from the observed resonance shift, as shown in Figure 3d. The large dips were mostly removed. Also notice a near-zero level when dipped in water, in all of the five repetitions. See, also, Figure S3 in the Supporting Information. In the above experiment, dipping the sensor head again into the same water brought the temperature of the sensor head to a value close to the one before being lifted. The situation may be different, when we use the dip sensor for sensing. Typically, the sensor head resting in water or a buffer in an open vial is lifted from the solvent, transferred to a solution containing target in another open vial, and dipped into the solution. During the transit, the temperature of the head will decrease, as we saw in Figure 3. On top of the temperature change during the transit, the temperature of the solution may differ from that of the solvent, since evaporation lowers the temperature of the solution when the cap is opened, even if the vial has been sitting in the lab for a long time. Now we examine the sensing capability of our dip sensor using dilute solutions of sucrose in water at different concentrations. The sensing will rely on a small difference of RI of the solution from that of water. When the sensor head is transferred from water to a solution of concentration c, the resonance wavelength changes by two mechanisms: Δλ ∂ ln λ dn2 ∂ ln λ c+ ΔT = λ ∂n2 dc ∂T
(1)
where n2 is the RI of water (or the solution) and dn2/dc is called the specific refractive index increment of the solution. It is known that 2 (neff − n22)1/2 n ∂ ln λ λ = 2 2 2 2 2 2 2 ∂n2 n1 − n2 (1 + n2 /n1 )neff − n2 2πa
(2)
for the TM mode,25,26 where n1 is the RI of the resonator, neff is the effective RI of WGM, which is slightly smaller than n1, and a is the radius of the resonator. By monitoring ΔT throughout multiple transfers of the sensor head between water and one of the solutions, we can subtract the temperature effect, (∂ ln λ/∂T)ΔT, from the D
DOI: 10.1021/acs.analchem.5b03066 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
■
(13) de Smet, L. C. P. M.; Ullien, D.; Mescher, M.; Sudho, E. J. R. Organic Surface Modification of Silicon Nanowire-Based Sensor Devices. In Nanowires: Implementations and Applications; Hashim, A., Ed.; InTech: Rijeka, Croatia, 2011; pp 267−288. (14) Luo, H. N.; Kim, H. S.; Agarwal, M.; Teraoka, I. Appl. Opt. 2013, 52, 2834−2840. (15) Kim, E.; Foreman, M. R.; Baaske, M. D.; Vollmer, F. Appl. Phys. Lett. 2015, 106, 161101. (16) De Maria, L.; Martinelli, M.; Vegetti, G. Sens. Actuators, B 1993, 12, 221−223. (17) Fontana, E. IEEE Trans. Microwave Theory Tech. 2002, 50, 82− 87. (18) Kurihara, K.; Ohkawa, H.; Iwasaki, Y.; Niwa, O.; Tobita, T.; Suzuki, K. Anal. Chim. Acta 2004, 523, 165−170. (19) Schuster, T.; Herschel, R.; Neumann, N.; Schäffer, C. G. J. Lightwave Technol. 2012, 30, 1003−1008. (20) Morshed, S.; Prorok, B. C. Exp. Mech. 2007, 47, 405−415. (21) Mudhana, G.; Park, K. S.; Ryu, S. Y.; Lee, B. H. IEEE Sens. J. 2011, 11, 1178−1183. (22) Hoffmann, P.; Dutoit, B.; Salathé, R.-P. Ultramicroscopy 1995, 61, 165−170. (23) Agarwal, M.; Teraoka, I. Appl. Phys. Lett. 2012, 101, 251105. (24) Agarwal, M. Whispering Gallery Mode Biosensor: Dip Sensing and Power Effect. Ph.D. Dissertation, New York University, 2014. (25) Teraoka, I.; Arnold, S.; Vollmer, F. J. Opt. Soc. Am. B 2003, 20, 1937−1946. (26) Teraoka, I.; Arnold, S. J. Opt. Soc. Am. B 2006, 23, 1381−1389. (27) Weast, R. C.; Astle, M. J., Eds., CRC Handbook of Chemistry and Physics, 61st ed., CRC Press, Boca Raton, FL, 1980; pp D229−D274.
CONCLUSIONS We have shown that it is possible to construct a dip sensor for whispering gallery mode and demonstrated its sensing capability for aqueous solutions. Compensation for a temperature change is necessary, as the resonance wavelength shifts with the temperature, and solutions may have different temperatures. Since the surface of the resonator made of silica is intact, it will allow a user to customize the sensor post assembly and modify the silica surface for his/her own sensing applications. The outer tubing helped stabilize the resonator−taper coupling and allowed us to track the resonance all the time, but the volume of the fluid held in the tubing slowed exchange of the solution surrounding the resonator. To cut the time for equilibration, we need to make this volume as small as possible while retaining the advantages imparted by the tubing.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b03066. Additional information as noted in text (PDF).
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: 646 997-3466. E-mail:
[email protected]. Present Address †
Department of Biomedical Engineering, Rensselaer Polytechnic Institute, 110 Eighth Street, Troy, NY 12180, U.S.A. Notes
The authors declare the following competing financial interest(s): An associated patent has been filed.
■
ACKNOWLEDGMENTS We acknowledge partial support by NSF through Grant No. 1040015.
■
REFERENCES
(1) Krioukov, E.; Klunder, D. J. W.; Driessen, A.; Greve, J.; Otto, C. Opt. Lett. 2002, 27, 512−514. (2) Vollmer, F.; Braun, D.; Libchaber, A.; Khoshsima, M.; Teraoka, I.; Arnold, S. Appl. Phys. Lett. 2002, 80, 4057−4059. (3) Shopova, S. I.; White, I. M.; Sun, Y.; Zhu, H.; Fan, X.; FryeMason, G.; Thompson, A.; Ja, S. Anal. Chem. 2008, 80, 2232−2238. (4) Lu, T.; Lee, H.; Chen, T.; Herchak, S.; Kim, J.-H.; Fraser, S. E.; Flagan, R. C.; Vahala, K. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 5976−5979. (5) Ward, J.; Benson, O. Laser Photonics Rev. 2011, 5, 553−570. (6) Luchansky, M. S.; Bailey, R. C. Anal. Chem. 2012, 84, 793−821. (7) Savchenkov, A. A.; Matsko, A. B.; Ilchenko, V. S.; Yu, N.; Maleki, L. J. Opt. Soc. Am. B 2007, 24, 2988−2997. (8) Vernooy, D. W.; Ilchenko, V. S.; Mabuchi, H.; Streed, E. W.; Kimble, H. J. Opt. Lett. 1998, 23, 247−249. (9) Chen, Y.; Ming, H. Photonic Sens. 2012, 2, 37−49. (10) Cheng, C. I.; Chang, Y.-P.; Chu, Y.-H. Chem. Soc. Rev. 2012, 41, 1947−1971. (11) Ymeti, A.; Greve, J.; Lambeck, P. V.; Wink, T.; van Hovell, S. W. F. M.; Beumer, T. A. M.; Wijn, R. R.; Heideman, R. G.; Subramaniam, V.; Kanger, J. S. Nano Lett. 2007, 7, 394−397. (12) Swann, M. J.; Peel, L. L.; Carrington, S.; Freeman, N. J. Anal. Biochem. 2004, 329, 190−198. E
DOI: 10.1021/acs.analchem.5b03066 Anal. Chem. XXXX, XXX, XXX−XXX
