We present a single-mode, single-polarization, distributed-feedback liquid dye laser, based on a short high-order Bragg grating defined in a single polymer layer between two glass substrates. In this device we obtain single-mode operation in a multimode structure by means of transverse-mode discrimination with antiguiding segments. The laser is fabricated using microfabrication technology, is pumped by a pulsed frequency-doubled Nd:YAG laser, and emits narrow-line-width light in the chip plane at 577 nm. The output from the laser is coupled into integrated planar waveguides defined in the same polymer film. The laser device is thus well suited for integration, for example, into polymer based lab-on-a-chip microsystems.
©2005 Optical Society of America
Integrable light sources are essential for lab-on-a-chip devices, to facilitate use of light-based sensors on-chip , which otherwise requires cumbersome optical alignment of light source and chip at every use. Dye lasers are of particular interest, since they can be applied for any wavelength in the visible and into the infrared. In particular single-mode laser devices are important for interference-based sensors [2, 3].
In this paper we demonstrate a single-mode, single-polarization laser based on a multimode waveguide structure and a high-order Bragg grating. In the device, all feature sizes are much larger than the wavelength of light. We obtain single-mode lasing by exploiting the increased effective-mode distance in a long-period, high-order asymmetric grating, and we eliminate lasing in all but the fundamental transverse mode in the resonator waveguide, by inflicting losses on higher modes with the use of antiguiding segments.
This is demonstrated by forming the laser structure in the photodefinable polymer SU-8 10 (from Microchem Corp.) sandwiched between two glass substrates and using the laser dye Rhodamine 6G dissolved in ethanol as an amplifying medium. Pictures of the device are shown in Fig. 1. The dye is optically pumped with a frequency-doubled Nd:YAG laser.
In fabrication, the laser structure is lithographically defined in an 8-µm-thick SU-8 polymer layer on top of a Borofloat glass substrate. The structure patterned in the polymer is sealed off by another glass substrate using polymethylmethacrylate (PMMA) mediated bonding . The resulting microfluidic channel created between the glass substrates and defined by the pattern in the SU-8 film is 15 mm long and 2 mm wide at the fluid input for the dye, narrowing down to 1 mm at the laser region (Fig. 1). Fluidic connections are facilitated by holes drilled through the glass lid with a diamond drill. The finished device is 10 mm×20 mm and 1 mm high and requires an external fluid pump for dye replenishment to counter the slow bleaching of the dye molecules in the laser structure during optical pumping. The SU-8 laser structure can be directly integrated with waveguides and fluid channels on a microchip for lab-on-a-chip applications [8, 9]. Microfluidic dye lasers have been demonstrated earlier [7, 10, 11], but to our knowledge the device presented in this paper is the first demonstration of a laterally emitting, single-mode microfluidic dye laser, which can readily be integrated on a polymer-based microsystem.
3. The laser resonator
The laser resonator is based on a distributed-feedback structure defined by a high-order Bragg grating, with a central phase shift of π/2 to ensure a single resonance for each Bragg reflection . The structure is embedded in a microfluidic channel, as shown in Fig. 1. Waveguides, defined in the same polymer layer and lithography step as the laser structure, guide the emitted light from the laser to the edge of the chip for ease of measurement. The Bragg grating defining the optical resonator is formed by an array of channels interleaved by sections (bars) of polymer. These grating channels carry a laser dye solution, and the structure exhibits a refractive index modulation resulting from the difference in refractive index of the polymer (n=1.596 at 574 nm) and the ethanolic dye solution (=1.33) of 0.27. In our design we have formed the array in an 8-µm thick SU-8 polymer layer by making 23 channels 17.8-µm wide, interleaved by 26.1 µm of polymer. The channels are 1 mm long. The minimum size of the channels is limited by the UV lithography (5–10 µm), and the fabrication dimensions were chosen to lie well within the lithographic possibilities of our equipment.
Because of the width of the resonator, the optical modes are expected to be uniform along the length of the fluid channels forming a slab waveguide mode. It is safe to assume that losses due to the nonconfining elongated structure are lower than loss contributions from other mechanisms.
The periodic structure of channels filled with fluid and polymer walls forms a distributed-feedback resonator (DFB) . Our structure uses a reflection-mode order in the 130s. A first approach to finding the optical response of the grating can be made by assuming the channels and polymer sections to be infinite planes and solving the wave equation with a transmission matrix formalism . The plane-wave approximation is adjusted later by taking into account the altered propagation constants and losses for the modelled structure.
With the transmission matrix method, the nominal spacing between resonance modes is found to be 2.46 nm, and the minimum of the resonator round-trip power loss factor is found to be 0.14, owing to photons coupled out of the ends of the grating, i.e., neglecting any other losses.
The round-trip loss is calculated by first finding the two-by-two transmission matrices, TA and TB, for each half of the resonator. TA=TB=T(λ), since the resonator is symmetric.
With the middle of the structure as a reference plane (see Fig. 2), a wave travelling to the right at the reference plane will experience a reflection rA(λ)=T 21(λ)/T 11(λ) from the right grating and subsequently an identical reflection rB(λ) from the left grating, determined by the two elements of the transmission matrix. The round-trip loss at each longitudinal mode wavelength, λlm, is found as αrt (λlm)=1-|rA(λlm)rB(λlm)|2. Where λlm are the longitudinal mode wavelengths where the imaginary part of rA(λlm)rB(λlm) equals zero, as required for the resonance to have no phase lag during one resonator round trip. The longitudinal modes are spaced 0.12 nm, and a longitudinal mode coincides with each Bragg-order mode because of the π/2 phase shift in the middle of the structure.
The minimum loss varies from one Bragg-order mode to another as a result of the different channel/polymer bar widths (cf. Fig. 4). The resulting beat pattern is caused by the difference of phase evolution with frequency for the light traversing the channel and the polymer segment. Therefore the effective mode distance for laser modes (modes with near lowest loss) is 3 to 4 times the nominal spacing, although there is still a small chance of having two equally strong modes spaced with the nominal mode spacing if the exact channel/polymer bar width ratio is not controlled. The gain peak of the laser dye (Rhodamine 6G) has a FWHM of up to 50 nm , resulting in the availability of five to six resonator modes for lasing under the gain curve. This scarcity enables single-mode lasing, as demonstrated by the measurements.
The sandwich structure of PMMA, SU-8, and glass (Figs. 1 and 3) forms a slab waveguide. Since the SU-8 is relatively thick (8 µm) and has a high refractive index (n=1.596), the slab waveguide supports 16 transverse modes. However, the fluid channels in the resonator are antiguiding, since the refractive index of the fluid (ethanol, n=1.33) is lower than than the refractive index of PMMA and glass that surrounds it. Thus when light travels across a channel, it experiences diffraction, and not all energy is coupled into the next waveguiding polymer segment. The loss for the modes supported by the slab waveguide that travels through the structure increases for increasing mode number, m=0,…,19, because of increasing diffraction. This loss discrimination between the modes enables lasing in a single mode.
To find the loss for traversing a fluid channel, a horizontally polarized light field (in accordance with measurements) is propagated from the exit of a waveguiding polymer segment with a beam-propagation method . The finite transverse length of the channel (1 mm) is neglected.
The amount, |cm,np|2, of propagated light from a slab waveguide mode number n, coupled into a guided mode, m, in the next polymer segment, is found by
where Um(y) is the normalized field distribution of the guided mode m, Un,p(y) is the distribution of the field propagated across the fluid channel, and y is the vertical position (see Fig. 3).
The calculated loss for the first six modes due to a channel crossing is tabulated in Table 1, where the normal reflection due to the refractive-index difference between the SU-8 polymer and the fluid is neglected, since the reflection is not a loss factor. Coupling between guided modes is negligible at low mode numbers.
The loss has a slight dependence on the polarization of 6×10-5 for the fundamental mode, because of the Fresnel-like polarization-dependent reflection from the interface between the channel fluid and the glass or the PMMA. The polarization dependence is small because the corresponding angle of incidence of the propagating light in a ray model picture is at grazing angle. Our experiments indicate that this difference in loss is sufficient to achieve single-polarization lasing.
The round-trip power loss as seen from the middle of the resonator is calculated by using the losses found by the propagation calculation and the mode-dependent propagation constants in the transmission matrix.
The round-trip loss calculated with the diffraction loss for the zeroth transverse mode is illustrated in Fig. 4. The loss spectrum for the fundamental m=0 and the m=1 mode is illustrated in the inset, the round-trip loss factor for the modes being 0.21 for m=0 and 0.33 for m=1 at resonance. The graphs are composed of solutions for each longitudinal mode λlm, connected with lines. Imperfections in fabrication are expected to increase the mode discrimination.
By taking into account the spatiotemporal evolution of the inversion of the dye molecules in the laser cavity volume during a pump pulse, and using a rate equation model for the dye molecule population where intraband vibrational relaxation rates are neglected, we calculate the pump energy threshold for lasing to be 1.3 µJ mm-2 for a 5-ns Gaussian pump pulse at 532 nm, for the fundamental transverse mode. The quantum efficiency of the Rhodamine molecules is assumed to be 0.5. This is the quantum efficiency for a concentration of 2×10-2 Mol/l according to  due to dimer formation. Even though the quantum efficiency of Rhodamine 6G in our solution is low compared with dilute solutions where it approaches unity, we use a high concentration in our lasers based on a compromise between low quantum efficiency and high molecule concentration to achieve a relatively low lasing threshold.
4. Results and discussion
Measurements on the device were performed by pumping the dye inside the laser structure with a frequency-doubled Nd:YAG laser at 532 nm with 5-ns pulse width and 10-Hz repetition rate. The pump light impinged at normal incidence to the chip plane, and the dye solution was 2×10-2 Mol/l Rhodamine 6G in ethanol. Emission from the dye laser was picked up with a 200-µm-diameter multimode fiber placed close to the SU-8 waveguides exiting the laser chip, collecting light from two on-chip waveguides at a time. The fiber connected to a spectrometer with a resolution of 0.15 nm.
The 532-nm pump light energy scattered into the waveguides amounts to less than 4% of the dye laser light energy. The emitted laser light was horizontally polarized, i.e., in the chip plane, as expected. The emission spectrum for a laser at 0.13 mJ mm-2 pump energy exhibits a clearly dominant peak at 576.64 nm (Fig. 5). The linewidth of the laser is unresolved by the spectrometer and remains undetermined.
Few minor sidemodes with intermodal distances from 2.2 to 2.6 nm can be present in the laser spectra at ~10% intensity of the main peak. The intermodal distance suggests the modes to stem from neighboring Bragg orders. Signs of weak side modes close to the main peak are also observable. These modes are unresolved by our spectrometer and could stem from the m=1 (or higher) transverse mode (cf. Fig. 4).
The dye laser output power as function of pump power follows a standard pump/output curve with two linear segments around the lasing threshold (Fig. 5) at a pump pulse energy fluence of 0.02 mJ mm-2. This threshold is ~15 times higher than the calculated value. This could be due to a lowered quantum efficiency in our dye solution caused by impurities, to imperfections in the laser structure, and to inversion quenching due to amplification of spontaneous emission in the low-feedback resonator.
We have measured the dye laser output power to 1.2 µJ (on one end of the laser) at a pump pulse energy of 105 µJ (impinging the laser area), corresponding to a single-ended efficiency of 0.01.
We have demonstrated single-mode lasing in a new type of microfluidic laser using a high-order Bragg grating with antiguiding segments for transversal-mode discrimination in a multimode polymer structure. The large structure size enables fabrication with a single UV lithography process step. The same lithography step can feature optical waveguides and other optical components along with fluidic components such as mixers and reaction chambers for use with lab-on-a-chip .
The work was supported by the Danish Technical Research Council (STVF, grant 26-02-0064).
References and links
1. E. Verpoorte, “Chip vision—optics for microchips,” Lab on a Chip 3, 42N–52N (2003).
3. L. Lading, L. B. Nielsen, and T. Sevel, “Comparing biosensors,” in Proceedings of the IEEE Sensors 2002 (2002), pp. 229–232.
4. C. Bojarski and E. Grabowska, “Photoluminescence decay and quantum yield studies for rhodamine 6G in ethanol,” Acta Physica Polonica A60, 397–406 (1981).
5. G. P. Agrawal and N. K. Dutta, Semiconductor Lasers, 2nd. ed. (Reinhold, N.Y., 1993)
7. B. Bilenberg, T. Nielsen, B. Clausen, and A. Kristensen, “PMMA to SU-8 bonding for polymer based lab-on-a-chip systems with integrated optics,” J. Micromech. Microeng. 14, 814–818 (2004) [CrossRef]
8. K. B. Mogensen, J. El-Ali, A. Wolff, and J. P. Kutter, “Integration of polymer waveguides for optical detection in microfabricated chemical analysis systems,” Appl. Opt. 89, 4072–4078 (2003). [CrossRef]
9. S. Balslev, B. Bilenberg, O. Geschke, A. M. Jorgensen, A. Kristensen, J. P. Kutter, K. B. Mogensen, and D. Snakenborg, “Fully integrated optical system for lab-on-a-chip applications,” in Proceedings of the 17th IEEE MEMS(IEEE, 2004), pp. 89–92.
10. B. Helbo, A. Kristensen, and A. Menon, “A micro-cavity fluidic dye laser,” J. Micromech. Microeng. 13, 307–311 (2003). [CrossRef]
12. S. L. McCall and P. M. Platzman, “An optimized π/2 distributed feedback laser,” IEEE J. Quantum Electron. 21, 1899–1904 (1985). [CrossRef]
13. L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits, (Wiley, New York, 1995).
14. B. B. Snavely, “Flashlamp-excited organic dye lasers,” Proc. IEEE 57, 1374–1390 (1969). [CrossRef]
15. T. B. Koch, J. B. Davies, and D. Wickramasinghe, “Finite element/finite difference propagation algorithm for integrated optical device,” Electron. Lett. 25, 514–516 (1989). [CrossRef]