1. Introduction

Microfluidic dye lasers have been under intense investigation in the past few years as a solution for enhancing integration of laser light sources with micro total analysis systems (µTAS) [1,2]. Various types of microcavities have been developed, including distributed feedback grating cavities [3,4], Fabry-Pérot cavities [5], and microknots [6]. Recently, the opto-fluidic ring resonator (OFRR) was also developed to provide a unique interface for the photonic cavity and microfluidics [7–10]. The OFRR utilizes a thin-walled glass capillary, the circular cross section of which forms a ring resonator that supports whispering gallery modes or circulating waveguide modes (WGMs). The gain medium, typically dye solution, flows through the core and interacts with the WGM, which provides the optical feedback for the laser [7–10]. The OFRR exhibits superior performance in handling fluids because of its capillary nature. Additionally, extremely high Q-factor (~10⁹) can be obtained [9], leading to an unprecedented lasing threshold of only a few tens of nJ/mm², two or three orders of magnitude lower than other types of microfluidic lasers [9]. Lasing from FRET emission has also been demonstrated with the OFRR, opening the door for future biosensing applications [8]. Finally, the laser emission can be efficiently coupled out through an optical coupler in contact with the OFRR, thus providing a convenient means to guide and deliver the laser emission [7,9].

Although significant progress has been made in both experimental and theoretical aspects of the OFRR laser [7–10], to date it has operated in free space with the tapered fiber and OFRR capillary being suspended in air, which is too delicate for manual manipulation and is difficult to mount on a chip. The infrastructure for fluidic connectivity is also too bulky to fit within the footprint of a traditional lab-on-a-chip device. Improvement in the portability and mechanical strength of the OFRR laser and demonstration of its compatibility with soft lithography based microfluidics platforms has now become a crucial step towards eventual OFRR based µTAS development. In this paper, we address these issues by embedding the glass based OFRR in a polymer of low refractive index (RI), as illustrated in Fig. 1. When the high index gain medium (such as dye in a solvent) is passed through the OFRR, the WGM forms at the boundary between the liquid and the glass wall and provides optical feedback for lasing. The laser emission can be directionally out-coupled into a fiber through a fiber prism or a tapered optical fiber in contact with the OFRR. We will show that a Q-factor in excess of 10⁶ can still be obtained, leading to a laser threshold below 1 µJ/mm², on par with or even better than most of the microfluidic lasers mentioned previously. The laser output from the fiber is measured to be 80 nW, corresponding to 50% power extraction efficiency. These designs not only enhance the portability and mechanical strength of the microfluidic laser itself, but also provide the convenience of connecting it to microfluidics built within a polymer-based chip. Eventually, due to the small size of the OFRR (~100 µm in diameter), a large number of the microfluidic dye lasers can be packed on a small chip for multiplexed functionalities.

 

Fig. 1. Conceptual illustrations of the glass OFRR embedded in PDMS. The laser emission can be directionally coupled out through a fiber prism (a) or a tapered optical fiber (b) for easy and efficient light delivery. θ (15°) is the fiber prism angle.

Download Full Size | PDF

2. Experiment and results

The fiber prism used in our work is made out of a 200-µm multimode fiber, which is first anchored in acrylic and then angle polished at 15° at one of its ends [11,12]. The tapered fiber is made by pulling an SMF-28 optical fiber under intense heat until it reaches approximately 3 µm at its narrowest part. The OFRR with an outer diameter (OD) of 75 µm and a wall thickness of 5 µm is drawn out of a fused silica preform (Polymicro) using previously published methods [9]. The fiber prism (or tapered fiber) is then positioned against the OFRR and the whole device is placed in a small module before liquid polydimethylsiloxane (PDMS, Dow Corning Sylgard 184) is poured on the device and cured in air at 70 °C for 2 hours. We choose PDMS because it is a material commonly used in microfluidic devices and it has a relatively low refractive index (n ~1.41) [13]. It is also chemically inert and has low loss in the visible spectrum, both of which are important characteristics in microfluidic laser development.

To demonstrate the microfluidic lasers, dye (Coumarin 504, R6G, and LDS 722) in quinoline (n=1.626) is flowed through the OFRR at a rate of 10 µL/min with a syringe pump and is excited by a 5-ns pulsed tunable OPO laser (Opotek, Vibrant, repetition rate: 10 Hz) from the side of the OFRR through PDMS. The microfluidic laser emission is coupled through the fiber and detected at the distal end of the fiber with a spectrometer (Ocean Optics HR4000 USB, spectral resolution: 3.7 nm).

Figure 2 shows the R6G laser spectra for embedded OFRRs using a fiber prism (Fig. 2(a)) and a tapered fiber (Fig. 2(b)) to deliver the lasing output. The linewidth in Fig. 2(a) is approximately 8 nm, larger than the spectrometer resolution, indicating that the laser is operated at multi-mode lasing. However, the actual laser emission line from each individual WGM is too narrow to resolve. Note that the detected laser emission linewidth is broader when the laser is out-coupled with a fiber prism than with a tapered fiber, which is around 3.5 nm, limited by the spectrometer resolution. This discrepancy can be accounted for by considering the different out-coupling efficiencies between the fiber prism and tapered fiber for various WGMs. Since the fiber prism is made of a large fiber (200 µm core) that supports multi-mode transmission, the WGMs having different propagation constants can always be phase-matched by the modes in the fiber prism. The corresponding light in those WGM can thus be coupled out and subsequently transmitted down the fiber with low loss [11,12]. In contrast, the tapered fiber supports many fewer transmission modes and is therefore much more selective in picking up the WGMs. Only those that phase-match the modes in the tapered fiber can be efficiently coupled out. For the same reason, part of spontaneous emission, i.e., the regular fluorescence, can be coupled out through the fiber prism, whereas in the tapered fiber case, only the laser emission is observed, which is in agreement with the previous observation when the OFRR-taper is in free space [9].

Figure 2 also shows the laser emission peak intensity versus the excitation intensity. The lasing threshold is estimated to be 0.7 µJ/mm² and 0.5 µJ/mm² for the fiber prism configuration and tapered fiber configuration, respectively. Although the lasing threshold is over ten times higher than the best threshold reported for the OFRR laser in free space [9], it is still a few times lower than many other types of microfluidic lasers due to its high Q-factors, as discussed later. Note that the threshold reported here is a conservative estimation, as the light intensity is measured in front of the PDMS. The actual intensity impinged on the OFRR is expected to be lower.

The laser output from the fiber prism for 2 mM R6G is measured to be 80 nW (after the removal of the regular fluorescence signal), when the pump intensity is 2.2 µJ/mm², which corresponds to 8 nJ output per pulse. Considering the pump laser is impinging an area of 200 µm by 75 µm, the overall power extraction efficiency is 25%, much higher than the 1% efficiency reported in the microfluidic laser based on distributed-feedback grating cavity [3]. In reality, since only a layer of R6G molecules near the OFRR wall contribute to the laser emission, the actual power extraction efficiency should be much higher. Assuming that R6G molecules within a circular shell of 3 µm in thickness (see the discussion for Fig. 4 later), 75 µm in diameter, and 200 µm in height participate in lasing and that the R6G absorption cross section is 4.5×10^-16 cm² [14], a power extraction efficiency of 50% can be obtained, attesting to the highly efficient out-coupling [9].

 

Fig. 2. (a). Top: R6G WGM lasing spectrum collected with a fiber prism. The inset shows the picture of a fiber prism. Middle: Emission from the fiber prism vs. excitation intensity. Threshold is estimated to be 0.7 µJ/mm². Bottom: Photograph of three color simultaneous emission out of the fiber ends when three parallel OFRRs are filled with Coumarin 504 (left, λ_peak=480 nm), R6G (middle, λ_peak=600 nm), and LDS 722 (right, λ_peak=730 nm). A long-pass filter is used to remove the excitation light. (b) Top: R6G WGM lasing spectrum collected with a tapered fiber. Middle: Emission from the taper vs. excitation intensity. Threshold is estimated to be 0.5 µJ/mm². Bottom: Photograph of R6G laser emission from the end of a tapered fiber. A long-pass filter is used to remove the excitation light. In all cases, dye concentration is 2 mM.

Download Full Size | PDF

To demonstrate the versatility and multiplexing in our device, three fiber prisms are polished simultaneously and brought into contact with three OFRRs before the whole system is embedded in PDMS, as illustrated in the bottom figure of Fig. 2(a). Three different dyes (Coumarin 504, R6G, and LDS 722) at a concentration of 2 mM in quinoline running through the OFRRs are excited with a 460 nm beam, which overlaps with the absorption spectra of all three dyes. The bottom figure in Fig. 2(a) shows the photograph of the three-color laser emission out-coupled by fiber prisms. Color multiplexing can also be realized with taper fibers in replacement of fiber prisms. The bottom figure in Fig. 2(b) shows the photograph of laser emission from the tapered fiber when only one OFRR channel is excited. Unlike the light travelling in the fiber prism, which is protected by the fiber cladding, the light transmitted through the tapered fiber is exposed to the surrounding medium. However, it has been shown that low loss transmission can still be maintained when the taper is immersed in a low RI polymer [15–17]. Finally, color multiplexing can be achieved by running multiple dyes through the same OFRR channel. Figure 3 shows the laser emission from Coumarin 504 and R6G co-propagates along the same fiber.

 

Fig. 3. Spectrum of the OFRR laser when Coumarin 504 (1 mM) and R6G (1 mM) mixture is flowed through the OFRR. Excitation wavelength: 450 nm. Laser emission is out-coupled by a fiber prism.

Download Full Size | PDF

3. Discussion

To better understand the behavior of the WGMs in the embedded OFRR, we examine the radial intensity profile using a theoretical model based on Mie theory [10]. Figure 4 shows the radial distribution of the WGM of the 1^st, 2^nd, and 3^rd order. All modes have strong electric field concentrated within 2–3 µm near the wall surface, which interacts with the gain medium and provides the optical feedback for lasing. The intrinsic Q-factor for these modes, which is determined by the mode radiation loss, exceeds 10⁷, indicating that our device can potentially support high-Q modes.

 

Fig. 4. Radial distributions for (a) 1st order (b) 2nd order (c) and 3rd order modes. Intrinsic Q-factor for each exceeds 10⁷. OD=75 µm, wall thickness=5 µm, n₁=1.626, n₂=1.45, n₃=1.41. Dashed lines are the OFRR inner and outer surface.

Download Full Size | PDF

The actual Q-factor is measured by coupling the light from a 780 nm tunable diode laser (Toptica) into the OFRR through the tapered fiber using the same OFRR in Fig. 2(b). Based on the resonance linewidth in Fig. 5, the highest empty Q-factor when the OFRR is filled with quinoline without dye is approximately 2.6×10⁶. Similar Q-factors are also obtained in the fiber prism case. Since the intrinsic Q-factors of higher order WGMs are lower than 10⁶, we believe that the laser emission observed in our experiment is from the three lowest order WGMs. While the measured Q-factor is substantially below the predicted intrinsic Q-factor, it is still a few orders of magnitude higher than that in other microfluidic resonators [4]. The Q-factor degradation is attributed to the absorption of quinoline and polymer, the roughness induced scattering loss at the OFRR and polymer boundary, and the fiber out-coupling loss.

 

Fig. 5. WGM resonance displaying a Q-factor of 2.6×10⁶. Tunable diode laser wavelength is 780 nm.

Download Full Size | PDF

Since the RI of PDMS is close to that of glass, the OFRR laser can also be achieved even in the absence of the glass wall. In Fig. 6, we use a structure similar to the one described in Fig. 1(a). A small metal rod of 800 µm in diameter is used to define a circular hole in PDMS, which is removed after the polymer cured. The hole is subsequently filled with 0.5 mM R6G in quinoline. The laser emission can readily be seen from this OFRR. Demonstration of the OFRR laser in the absence of a glass capillary enables the fabrication of the OFRR using soft lithographic technologies. However, directly exposing the polymer to organic solvents may degrade the polymer integrity over time, appropriate selection of the solvent therefore become important in this type of microfluidic laser.

 

Fig. 6. WGM lasing emission from a bare PDMS channel using 0.5 mM R6G. Excitation wavelength is 540 nm.

Download Full Size | PDF

4. Conclusions and future work

To summarize, we have demonstrated an OFRR dye laser embedded in PDMS that has enhanced portability, mechanical stability, and compatibility with soft lithography based microfluidic technology. Even in PDMS, the OFRR still retains Q-factors larger than 106 and achieves a lasing threshold below 1 µJ/mm². The laser emission can be efficiently coupled out through a fiber prism or a tapered fiber, which provides a convenient means for on-chip light delivery. We have also demonstrated simultaneous lasing of multiple dyes in individual capillaries as well as multiple excitation of a mixed dye solution, proving the versatility of this system. Our work opens a door to a new µTAS system that brings together the OFRR technology with the traditional polymer based microfluidic technology.

Future efforts will be focused on several important areas. We will improve the OFRR Q-factors by refining our fabrication method, which will result in an even lower lasing threshold. We will also fabricate microfluidic arrays using a polymer molding method, where output couplers (fiber prism, waveguide, or fiber taper) will be spaced as tightly as a few tens of micrometers apart on the same chip without crosstalk. Additional optical functionalities such as in-line optical filter based on fiber Bragg gratings will be added to further enhance the device performance. Concurrent to these developments, applications of the embedded OFRRs in biological and chemical detection will also be pursued.