Microfluidic devices — tiny fluid-flow systems with channels less than a millimeter wide — have become increasingly useful in biological and metrological applications. Much of their success stems from the fact that, like microelectronics, micro-fluidics enables the integration of multiple functionalities on a single substrate. One such functionality with great promise is photonics.Researchers at Harvard University in Cambridge, Mass., recently created an optical waveguide inside a microfluidic channel. The waveguide is based on a thermal gradient across the width of the channel of laminar-flow fluid. The cooler, more optically dense fluid at the center of the channel acts as the core of the waveguide, and the warmer fluid on the outside acts as the cladding.The channel has similar optical properties to a graded-index optical fiber, and light suitably injected into the core is trapped inside by refraction rather than by total internal reflection. Because the thermal profile and/or flow parameters can be changed quickly, the waveguide presents a new approach to readily reconfigurable micro-optical systems.Figure 1. The waveguide was formed by a transverse thermal gradient in the 5-mm-long microfluidic channel.The researchers based their demonstration on a 5-mm-long channel, using water and other liquids as the flowing fluid (Figure 1). As an alternative to using water at two temperatures to create the waveguide, they could have used two fluids, but the use of a homogeneous fluid has several advantages. It facilitates long-term operation of the device, avoiding the recycling necessary as two fluids become mixed. Also, because the thermal gradient can be modified easily — with in-line heaters, for example — homogeneous-liquid waveguides can be reconfigured more easily.However, homogeneous-liquid waveguides based on thermal gradients have a significant disadvantage in that thermal diffusion usually is more rapid than the diffusion of one liquid into another. Thus, it is more difficult to maintain the necessary refractive-index profile down the channel of flowing liquid. This problem can be minimized with a high flow rate, and the scientists saw the waveguide in their experiments deteriorate as they reduced the flow rate. Figure 2. With a low flow rate through the channel, light was not guided but spread uniformly across the channel (a). When the flow rate increased, light was strongly guided in the center of the channel (b). The traces in (a) and (b) are based on the inset photographs of the light emerging from the far end of the channel. In both (a) and (b), the core water was 21 °C, and the cladding water was 80 °C. The strength of the waveguiding — that is, the ratio of light at the central peak (Icore) to total light in the channel (Itotal) — rose with an increasing flow rate (c) and with an increasing temperature gradient (d). In (c), the hot- and cold-water temperatures were consistently 72 and 21 °C, respectively. In (d) the flow rate was constant at 60 ml/h, and the cold-water temperature was constant at 21 °C.They injected light from an optical fiber into one end of the channel and placed a CCD camera at the other end. With a relatively slow flow rate, they observed a uniform intensity at the camera, indicating that no waveguiding was taking place (Figure 2a). When they increased the flow rate, they saw a sharp intensity peak at the center of the channel, indicating strong waveguiding (Figure 2b). Figure 3. Simulations of the refractive index of water predict that a waveguide probably will not form at a flow rate of 6 ml/h (a) but that one may form at 60 ml/h (b).Between the minimum and maximum flow rates, the researchers saw a monotonic increase in waveguiding as the flow rate increased (Figure 2c). The waveguiding also rose monotonically, and approximately linearly, with the magnitude of the thermal gradient (Figure 2d). These experimental results confirmed a simulation of the refractive index profile (Figure 3). Applied Physics Letters, Feb. 6, 2006, 061112.