Capillaries present a promising structure for microfluidic refractive index sensors. We demonstrate a capillary-type fluorescent core microcavity sensor based on whispering gallery mode (WGM) resonances. The device consists of a microcapillary having a layer of fluorescent silicon quantum dots (QDs) coated on the channel surface. The high effective index of the QD layer confines the electric field near the capillary channel and causes the development of WGM resonances in the fluorescence spectrum. Solutions consisting of sucrose dissolved in water were pumped through the capillary while the fluorescence WGMs were measured with a spectrometer. The device showed a refractometric sensitivity of 9.8 nm/RIU (up to 13.8 nm/RIU for higher solution refractive index) and a maximum detection limit of ~7.2 x 10−3 RIU. Modeling the field inside the capillary structure, which is analogous to a layered hollow ring resonator, shows that sensitivities as high as 100 nm/RIU and detection limits as low as ~10−5 RIU may be achievable by optimizing the QD film thickness.
©2011 Optical Society of America
Microscale fluidic structures offer an intriguing array of potential sensing and diagnostic mechanisms. With the ability to sample small fluid volumes , compatibility with silicon integration , parallel detection capability [1,3] and high throughput potential , microfluidic sensors have been the focus of research development aiming toward lab-on-a-chip applications  First-generation devices are now available commercially, and further improvements are on the horizon . Optical detection may become an important transduction method in the development of microfluidic sensors ; it can be fast, it scales naturally toward small dimensions, it is non-destructive, and such systems can be integrated on a common material platform with the fluidic device.
Several structures are being explored for microfluidic optical sensing, including anti-resonant reflection optical waveguides, interferometers, micro-chromatographs, surface plasmon resonance devices, scattering-enhanced Raman spectroscopy, and optofluidics [8–13]. A common difficulty for many microfluidic sensing methods is that the interaction length is small, making it difficult to detect low analyte concentrations. One method that can overcome this problem makes use of the multiple-pass light path at the whispering gallery mode (WGM) optical resonances of cylindrical or spherical microcavities [14,15–17]. Silica microspheres, for example, can be fabricated with WGM quality (Q) factors as high as 109 . For such high Q-factors, light circulating by total internal reflection (TIR) around the sphere-medium interface can increase the interaction length from microns to kilometers. Optical microcavities are widely investigated in the field of quantum optics, and offer a rich array of structures and fabrication methods.
Liquid-core optical ring resonator (LCORR) structures combine WGM-based sensing with the fluidic environment of a microcapillary . In LCORR devices, the WGMs of a cylindrical capillary interact with analyte that is pumped through the channel. These structures show sensitivities typically on the order of 20 nm/RIU for the first-order radial modes , although values up to 570 nm/RIU have been achieved with the higher orders . LCORRs feature analyte volumes as small as picoliters, detection speeds as fast as a few seconds [21–23] and they are amenable to surface functionalization for detection specificity . While these properties make LCORRs favorable for microfluidic sensing, one difficulty is that the WGMs must propagate close to the capillary inner surface in order to interact with the liquid in the channel. This means that the LCORR capillary walls must be no more than a few microns thick.
To form an LCORR-based sensor, the walls of a standard silica capillary are thinned by heating and pulling, and then etched by pumping hydrofluoric acid through the channel to dissolve the interior walls . This results in a sensitive but very fragile fluidic device. LCORR structures employ an evanescent coupling apparatus (lithographically defined waveguides or fiber tapers placed within the evanescent field of the capillary) and require a stable, narrow-bandwidth tunable laser to measure the WGM shift caused by the interaction of the optical mode with the analyte in the capillary channel. Here, we demonstrate a quantum-dot-based “fluorescent-core microcapillary” (FCM) sensor that does not require thinning procedures and uses fluorescence instead of evanescent coupling as the transduction method. The method takes advantage of optical confinement in a high-effective-index quantum-dot layer that is uniformly coated on the capillary inner surface.
In order to achieve the desired FCM structure illustrated in Fig. 1 , we start with commercially available fused-silica capillary tubing purchased from Polymicro Technologies. Two sizes of capillaries were used in this study. The first (Type-I) had an inner diameter (I.D.) of 25 μm and an outer diameter (O.D.) of 360 μm; the second (Type-II) had an I.D. of 100 μm and an O.D. of 160 μm . A quantum dot (QD) film was formed on the inner surface of the capillaries using a method we previously developed for coating non-planar surfaces with a layer of fluorescent particles . In summary, the capillaries were dipped for about 30 seconds into a solution of hydrogen silsesquioxane  (H12Si8O12) dissolved in methyl isobutyl ketone (MIBK). The dipping process causes the solution to be drawn into the central channel via capillary forces. The capillaries were then annealed in a forming gas mixture of 95% Ar + 5% H2 at 300 °C for 3 hours, then at 1100 °C for 1 hour. During the first annealing phase, the solvent MIBK evaporates and hydrogen silsesquioxane (HSQ) molecules are deposited on the inner surface of the capillary. During the second phase, the HSQ molecules collapse, resulting in the formation of silicon QDs embedded in a SiO2-like matrix that coats the capillary channel.
Several characterization methods were performed on the Si-QD films. A flat film was prepared on a silicon substrate for transmission electron microscopy (TEM) in order to confirm the presence of Si-QDs in the films. A second flat-film sample was prepared for variable angle spectroscopic ellipsometry by sequentially spin-coating and annealing the HSQ solution on a silicon substrate. The films formed through spin-coating are generally too thin to develop the well-defined interference fringes required for extraction of the optical constants, so the coating and annealing process was repeated multiple times on the same sample in order to produces a sufficiently thick film. Scanning electron microscopy (SEM) was used to image the films deposited inside the capillary channels. For SEM, the capillaries were cleaved perpendicularly to their axis to expose a cross-sectional surface.
In order to measure the refractometric sensitivity of the fluorescent QD-coated capillaries, they were placed above a microscope objective (10x magnification, numerical aperture = 0.3) and interfaced to a micro-syringe pump. Solutions consisting of different mixtures of sucrose and water were injected into the capillary, while the fluorescence was collected and analyzed. The refractive index of the solutions ranged from 1.333 (water) to 1.453 (850.9 g/L aqueous sucrose solution). The Si-QDs were excited using either the 488 nm line of an Ar laser that was incident on the capillary through the free space above the objective, or with 405 nm incoherent light from an LED that was focused through the microscope objective. The emitted fluorescence was collected and analyzed with an SGS spectrograph from the Santa Barbara Instruments Group. All fluorescence data was calibrated for wavelength (using an Hg-Ar calibration lamp) and intensity (using the LS1 blackbody light source from Ocean Optics).
The ellipsometric data for the QD films converged to a solution giving n = 1.672 in the spectral range of 700-900 nm, with an extinction coefficient of 0.0006. The QD film has a higher index than the capillary wall (fused silica, n = 1.453) , suggesting that good optical confinement can in principle be maintained in the film. The extinction is probably associated with one or more of a variety of loss mechanisms that have been previously investigated in silicon QD films ; it implies a material-limited Q-factor of ~1500 in the wavelength range of interest. While there is uncertainty in these values due to unavoidable thickness variations in films made from multiple spin coating, we found that both the mode structure and the Q-factor calculated from these optical constants are in good agreement with the experimentally observed values.
SEM images of the film were generally difficult to obtain because there is little contrast in either secondary or backscattered electron imaging modes when comparing the SiO2 capillary wall and the thin SiOx film. Essentially, it was necessary to rely on the few cases where the film did not adhere to the capillary surface during cleaving. Figure 2a shows such an example, where one can observe the deposited film extruding from the capillary channel. Generally, on Type-I capillaries the film thickness was on the order of 0.5-1 μm, but for larger-diameter Type-II capillaries the film tended to be thicker. The internal structure of the film consisted of randomly oriented Si-QDs ranging in diameter from ~2-4 nm, embedded in an oxide matrix (Fig. 2b).
Fluorescence of flat films
Bare films were characterized by a spectrally broad and featureless fluorescence peaked near 800 nm. This type of emission spectrum is typical of silicon QDs embedded in an oxide matrix .
The QD films coating the capillary channel appeared beige in transmitted light (Fig. 3 , left). They generally appeared featureless and smooth, although there were defects or impurities in some places. The capillary channel was well defined in fluorescence imaging, indicating that the emitted light was largely confined within the QD film (Fig. 3, right). End-on imaging of a cleaved capillary showed a ring-like band of fluorescence defining the QD film on the capillary channel, consistent with the confinement of the fluorescence emitted within the higher-index QD layer.
Capillary fluorescence spectrum
The fluorescence spectrum collected perpendicular to the capillary axis showed modulations suggesting the development of the WGMs in the QD layer. By using a linear polarizer under the microscope objective, the TEz and TMz modes could be clearly identified (Fig. 4 ). Here, the z-direction is parallel to the capillary axis; modes with electric field in the z-direction are transverse to the propagation plane of the WGMs and are labeled TEz. These modes had considerably higher intensity than the TMz modes. The fluorescence Q-factor, defined as λ/ΔλFWHM, was ~800 for both TMz and TEz polarized modes in the Type-I capillary, and ~1500 in the Type-II capillary, quite close to the material limit discussed earlier, but probably limited by mode overlap. In the Type-I capillaries, the fluorescence peaks were asymmetric and tended to be slightly skewed toward shorter wavelengths. The modes in the Type-II structures appeared more symmetric, but in this case the spectrometer resolution (manufacturer specified at 0.24 nm) combined with the small free spectral range (FSR) may prevent adequate characterization of the peak shapes. The FSR of the modes was 4.45-5.14 nm in the Type-I capillary and 0.85-1.05 nm in the Type-II capillary (note: the FSR decreases with increasing azimuthal mode number, and also with capillary diameter). These values are concordant with the first-order radial modes propagating along the inner diameter of each capillary.
The mode spectrum of both capillary types shifted to longer wavelengths after injection of the sucrose solutions (Fig. 5 ). In the Type-I capillary, tracking a single TEz mode over a change in refractive index from 1.333 to 1.453 resulted in an average sensitivity of 9.8 nm/RIU. To demonstrate that the shifts can be measured without requiring a laser (i.e., an expensive component of proposed refractometric sensors),19 the experiments were repeated using a 405-nm LED as the pump. This analysis was done at a later time and on a different region of the capillary. WGMs were observed in the LED-pumped fluorescence spectrum (Fig. 5b), also exhibiting a spectral redshift with increasing sucrose concentration. The reason that the average sensitivity was lower in this case (5.7 nm/RIU) was likely due to the effect of local variations in the film thickness in the regions where the analyses were performed. The WGM shifts for the Type-II capillaries were similar, giving a slope of 6.8 nm/RIU.
The mode structure and refractometric sensitivity of these fluorescent-core microcapillaries were calculated using a finite-difference frequency-domain (FDFD) approach for a two-dimensional layered cylinder. The inner layer corresponded to the analyte, the middle layer to the QD film, and the outer layer to the glass capillary (Fig. 6 ). Because the capillaries were thick walled and the light was confined primarily within the QD layer, as shown in Fig. 6, the simulation domain did not include the outer air-boundary of the capillary. The z-component of the vector wave equation was isolated (i.e., TEZ polarization, consistent with the measurements) and forced into harmonic solutions, yielding the Helmholtz equation. Finite difference approximations to the partial derivatives of the Helmholtz equation were used to discretize the solution over the computational domain. This discretization was formulated as a matrix eigenvalue problem, with the eigenvalue equal to the square of the resonant frequency and the eigenvector corresponding to a given eigenmode of the cavity. We used radiating boundary conditions based on the “one-way” wave equations to remove optical energy from the computational domain. A small amount of reflection occurred at the boundaries for non-normal-incidence waves but the effect was found to be minor. This was confirmed by varying the domain size and ensuring solution consistency. The method permitted the resonance wavelength, Q-factor, and the mode numbers r (the radial order) and l (the azimuthal order) to be extracted for any given TEz mode, while also allowing considerable geometrical flexibility compared to analogous analytical methods.
The model illustrated the degree to which the channel diameter and film thickness determine the refractometric sensitivity of these fluorescent core microcavities. For larger capillary diameters, the refractometric sensitivity increases because a greater fraction of the mode extends into the capillary channel. This is in general agreement with previous numerical solutions of the field equations describing the cylindrical symmetry . That this sharp increase in sensitivity was not observed in experiment (Fig. 5) is likely a result of the second key parameter: the QD film thickness. In Fig. 7 , the theoretical sensitivity for both capillary types is shown for different film thicknesses. As the QD film thickness is reduced the sensitivity can increase considerably (inset Fig. 7). However, once the films become too thin the Q-factor degrades due to excessive losses into the capillary walls. Thus, the ideal film thickness represents an optimization problem between sensitivity and the Q-factor required for peak fitting and measurement.
The model also shows that the refractometric sensitivity depends on the refractive index of the analyte, evidenced by the non-linearity of the curves in Fig. 7. For the index range investigated (~1.33-1.45), the sensitivity increased with increasing refractive index. For example, in a Type-I capillary with a 400 nm (1000 nm) film, the sensitivity increases from 15.9 nm/RIU (0.33 nm/RIU) to 26.7 nm/RIU (0.73 nm/RIU) over the specified refractive index range. For a given refractive index and film thickness, the Type-II capillary yielded consistently higher sensitivities than the Type-I capillary (inset Fig. 7).
Peak shapes, sensitivity, and detection limits
The WGMs from the Type-I FCMs were clearly asymmetrical and skewed toward shorter wavelengths (Fig. 4). This effect was present for both TEz and TMz modes, and was observed with or without of the use of a polarizer during collection, as evidenced in Fig. 4b. A good fit to the data was obtained by fitting a smaller-amplitude Lorentzian on the high-frequency side of the main peak, as shown in Fig. 8 . These small-amplitude peaks on the short-wavelength wings of the main WGMs are unlikely to represent higher-order radial modes. Such modes would have greater refractometric sensitivity, inconsistent with the observation that the main peaks and the high-frequency sidebands maintained a constant spectral relationship over the range of fluid refractive index investigated. In contrast, the so-called “spiral modes” have a non-zero kz component to their wavevector and can appear as a family of overlapping peaks on the short-wavelength side of the main WGM maxima . Such modes have been reported in silica microcapillaries , and consistent with the present experimental observations, when the refractive index changes they maintain a constant spectral relation with the main peaks. Thus, we attribute the short-wavelength skewness to the spiraling cylindrical modes. In contrast, the Type-II capillary modes did not show clearly defined skewness (Fig. 4). In this case, the low finesse caused significant spectral mode overlap, making the exact peak shape difficult to resolve experimentally.
In the index range of the sucrose solutions (~1.33-1.45) the resonance shifts appeared approximately linear; however, over a larger index range (1.00-1.45) the shifts were obviously nonlinear with refractive index. These results were consistent with the FDFD model (inset Fig. 8), and are partly associated with the field extension into the fluid core. For simplicity the resonance shifts were fit with a quadratic function, yielding a maximum experimental refractometric sensitivity of 13.8 nm/RIU (Fig. 8, inset). The minimum detectable refractive index change can be estimated as ΔRIUmin = p/(δλ/δn) 16 ≈ 7.2 x 10−3 RIU, given a wavelength pitch, p, of 0.1 nm, corresponding to the properties of our spectrometer. These values are comparable with QD-doped fluorescent microspheres , but are much lower than those obtained for thin-walled LCORRs pumped evanescently with a tuneable laser. However, as discussed below, FCM structures could be made with considerably higher sensitivity.
As shown in Fig. 7, the film thickness is an important parameter that controls the overall sensitivity of the FCM structures. Unfortunately, the films are difficult to image in the SEM because of the low contrast with the glass capillary walls. As shown in Fig. 2, we could only make use of the few cases where a film detached and protruded from the channel. As well, the thickness can be variable along the length of a single capillary (Fig. 3), rendering it virtually impossible to obtain an independent measure of the thickness at the location where the measurements were performed. This variation likely accounts for the differences in the sensitivity for the laser- and LED-pumped FCM (Fig. 5a,b), which were taken at different positions in the capillary channel. Instead, the film thickness can be estimated from the FDFD simulations with the measured sensitivity values. In this way, we obtain film thicknesses of 525 nm and 661 nm for the data taken on Type-I (Fig. 5a) and Type-II (Fig. 5c) capillaries, respectively.
Pump-induced thermal fluctuations
Thermo-optic and thermo-mechanical effects may be induced through laser heating in optical microcavities [33,34]. To investigate the extent of these effects in the present FCM structures, two additional experiments were conducted. First, a Type-I capillary was irradiated using the 488-nm laser line at an incident power of ~120 mW on the sample (output laser power = 400 mW) while the spectrum was continuously monitored. There was no detectable change in the mode over a 3-hour time period. Second, both Type-I and Type-II capillaries were exposed to increasing laser power up to 120 mW. A mode shift of approximately 0.4 pm/mW was observed in the Type-I capillaries and 1.4 pm/mW in Type-II structures (Fig. 9 ). Since the incident laser power used in the experiments was constant near 40 mW, we conclude that thermal effects did not cause the spectral shifts observed after injection of the sucrose solutions.
An independent comparison of the expected heating-induced spectral shifts would be fairly unreliable. Because the thermo-optic coefficient of glass is positive, that of water is negative , and that of the QD film is unknown (although likely also positive), the relative thermo-optic-induced spectral shifts will be complicated functions of the field confinement in the three layers (solution, film, and capillary wall) and of their respective thermo-optic coefficients. Furthermore, the laser focus on the two capillaries is likely to have been slightly different in each experiment. The important point, however, is that the spectral shifts observed with sucrose solutions in the capillary channels is much larger than the very small observed pump-induced thermal fluctuations of the FCMs.
Comparison: LCORRs vs. FCMs
Current-generation FCMs bear a clear set of advantages and drawbacks compared with LCORR devices. Compared to LCORR’s, their relatively low detection limits arise from two main issues: device sensitivity and spectrometer resolution. The FDFD model showed that decreasing the QD film thickness could increase the sensitivity to ~100 nm/RIU in the Type-I FCMs, without degrading the Q-factors below the existing material limit of the quantum dot film. Previous calculations of the refractometric properties of rolled-up tubular fluorescent multilayers showed that sensitivities could be as high as 400 nm/RIU in optimized structures . Thus, the maximum of 13.8 nm/RIU demonstrated here could likely be improved. The second problem is associated with spectrometer resolution. For grating spectrometers, this effect will likely dominate other sources of error in determining the mode peak positions . Our spectrometer has a relatively low manufacturer-specified resolution of 0.24 nm and a large wavelength pitch of 0.10 nm per pixel. Higher-resolution grating spectrometers can achieve considerable improvement of these values, and tapered waveguide spectrometers can potentially provide resolutions in the range of a few picometers [38,39].
Regardless of these potential improvements, FCMs already offer some attractive features: (i) they do not require nano-positioning equipment; (ii) no capillary thinning or HF-etching methods are required; (iii) they can be operated with LEDs as an inexpensive alternative to a tunable laser; and (iv) the capillaries are robust, reusable, and easily handled. While we used an existing setup consisting of a fluorescence microscope to measure the FCMs, one strictly only needs a lens to project the capillary image onto the spectroscopy system. Like LCORRs, FCM surface chemistry is silica based and they can therefore be functionalized. For these reasons, fluorescent core microfluidic resonators provide a potentially attractive alternative to other WGM-based microcavities for sensing applications.
FCMs offer an alternative microcavity-based refractometric sensing method that combines the microfluidic advantages of LCORR systems with a lower cost and simpler fabrication. FCM refractometric sensitivity can approach “traditional” structures like fluorescent microspheres [40,41], while at the same time they allow true microfluidic operation similar to liquid-core ring resonators. In this work, the FCM structure consisted of a layer of fluorescent silicon quantum dots coating the inner channel of a microcapillary. The QDs were embedded in silica glass and were, therefore, protected from fluids passing through the capillary channel. The coating has a higher effective index than the capillary wall, supporting the development of fluorescence WGMs that extend slightly into the channel medium. By measuring the WGM peak positions as a function of sucrose concentration in water, the maximum refractometric sensitivity and detection limit were found to be 13.8 nm/RIU and 7.2 x 10−3 RIU, respectively. These values could be improved by optimizing the QD film thickness or by utilizing a higher resolution spectrometer.
We thank NSERC, Alberta Ingenuity, and CIPI for funding. Nakeeran Ponnampalan, X.Y. Wang, and Ross Lockwood are thanked for providing the spectroscopic ellipsometry and the TEM and SEM images of the QD film, respectively.
References and links
1. A. H. Diercks, A. Ozinsky, C. L. Hansen, J. M. Spotts, D. J. Rodriguez, and A. Aderem, “A microfluidic device for multiplexed protein detection in nano-liter volumes,” Anal. Biochem. 386(1), 30–35 (2009). [CrossRef] [PubMed]
3. H. Zhu, I. M. White, J. D. Suter, M. Zourob, and X. Fan, “Integrated refractive index optical ring resonator detector for capillary electrophoresis,” Anal. Chem. 79(3), 930–937 (2007). [CrossRef] [PubMed]
4. E. Krioukov, J. Greve, and C. Otto, “Performance of integrated optical microcavities for refractive index and fluorescence sensing,” Sens. Actuators B Chem. 90(1-3), 58–67 (2003). [CrossRef]
9. A. Ymeti, J. Greve, P. V. Lambeck, T. Wink, S. W. van Hövell, T. A. M. Beumer, R. R. Wijn, R. G. Heideman, V. Subramaniam, and J. S. Kanger, “Fast, ultrasensitive virus detection using a Young interferometer sensor,” Nano Lett. 7(2), 394–397 (2007). [CrossRef] [PubMed]
10. D. Yin, E. J. Lunt, M. I. Rudenko, D. W. Deamer, A. R. Hawkins, and H. Schmidt, “Planar optofluidic chip for single particle detection, manipulation, and analysis,” Lab Chip 7(9), 1171–1175 (2007). [CrossRef] [PubMed]
13. H. Dirac and P. Gravesen, “Realisation and characterisation of all liquid optical waveguides,” Proc. MEMS. 14th IEEE International Conference on Micro Electro Mechanical systems, 4590462 (2001).
15. A. Francois and M. Himmelhaus, “Optical biosensor based on whispering gallery mode excitations in clusters of microparticles,” Appl. Phys. Lett. 92(14), 141107 (2008). [CrossRef]
18. M. L. Gorodetsky and V. S. Ilchenko, “High-Q optical whispering gallery microresonators: precession approach for spherical mode analysis and emission patterns,” Opt. Commun. 113(1-3), 133–143 (1994). [CrossRef]
19. I. M. White, H. Zhu, J. D. Suter, X. Fan, and M. Zourob, Methods in Molecular Biology: Biosensors and Biodetection, edited by A. Rasooly and K.E. Herold, (Humana Press, 2009). Chap. 7.
20. H. Li and X. Fan, “Characterization of sensing capability of optofluidic ring resonator biosensors,” Appl. Phys. Lett. 97(1), 011105 (2010). [CrossRef]
22. I. M. White, H. Zhu, J. D. Suter, N. M. Hanumegowda, H. Oveys, M. Zourob, and X. Fan, “Refractometric sensors for lab-on-a-chip based on optical ring resonators,” IEEE Sens. J. 7(1), 28–35 (2007). [CrossRef]
24. G. Yang, I. M. White, and X. Fan, “An opto-fluidic ring resonator biosensor for the detection of organophosphorus pesticides,” Sens. Actuators B Chem. 133(1), 105–112 (2008). [CrossRef]
25. Product numbers TSP025375 and TSP100170, respectively, from Polymicro Technologies (www.polymicro.com).
26. C. M. Hessel, M. A. Summers, A. Meldrum, M. Malac, and J. G. C. Veinot, “Direct patterning, conformal coating, and erbium doping of luminescent nc-Si/SiO2 thin films from solution processable hydrogen silsesquioxane,” Adv. Mater. (Deerfield Beach Fla.) 19(21), 3513–3516 (2007). [CrossRef]
27. Known by its standard trade name FoX-15, Dow Corning, Inc. FoX-15 is a negative electron beam resist.
28. I. Malitson, “Interspecimen comparison of the refractive index of fused silica,” J. Opt. Soc. Am. 55(10), 1205–1209 (1965). [CrossRef]
29. R. D. Kekatpure and M. L. Brongersma, “Fundamental photophysics and optical loss processes in Si-nanocrystal-doped microdisk resonators,” Phys. Rev. A 78(2), 023829 (2008). [CrossRef]
30. C. M. Hessel, E. J. Henderson, and J. G. C. Veinot, “An investigation of the formation and growth of oxide-embedded silicon nanocrystals in hydrogen silsesquioxane-derived nanocomposites,” J. Phys. Chem. C 111(19), 6956–6961 (2007). [CrossRef]
33. A. W. Poon, R. K. Chang, and J. A. Lock, “Spiral morphology-dependent resonances in an optical fiber: effects of fiber tilt and focused Gaussian beam illumination,” Opt. Lett. 23(14), 1105–1107 (1998). [CrossRef] [PubMed]
34. V. Zamora, A. Díez, M. V. Andrés, and B. Gimeno, “Interrogation of whispering-gallery modes resonances in cylindrical microcavities by backreflection detection,” Opt. Lett. 34(7), 1039–1041 (2009). [CrossRef] [PubMed]
36. G. Huang, V. A. Bolaños Quiñones, F. Ding, S. Kiravittaya, Y. Mei, and O. G. Schmidt, “Rolled-up optical microcavities with subwavelength wall thicknesses for enhanced liquid sensing applications,” ACS Nano 4(6), 3123–3130 (2010). [CrossRef] [PubMed]
38. R. G. DeCorby, N. Ponnampalam, E. Epp, T. Allen, and J. N. McMullin, “Chip-scale spectrometry based on tapered hollow Bragg waveguides,” Opt. Express 17(19), 16632–16645 (2009). [CrossRef] [PubMed]
39. O. Schmidt, P. Kiesel, S. Mohta, and N. M. Johnson, “Resolving pm wavelength shifts in optical sensing,” Appl. Phys. B 86(4), 593–600 (2007). [CrossRef]
40. A. Weller, F. C. Liu, R. Dahint, and M. Himmelhaus, “Whispering gallery mode biosensors in the low-Q limit,” Appl. Phys. B 90(3-4), 561–567 (2008). [CrossRef]