CFF2 2005 Conference on Lasers & Electro-Optics (CLEO) Fluidic Photonic Integrated Circuit (FPIC) for Cytometric Detection Victor Lien, Kai Zhao and Yu-Hwa Lo Electrical and Computer Engineering Department, University ofCalifornia, San Diego 9500 Gilman Drive, La Jolla, California 92093 Tel: 1-858-822-2777, Fax: 1-858-534-0556, Email: vlien@ucsd.edu Abstract: We present a fluidic photonic integrated circuit (FPIC) performning the detection function for flow cytometry. An array waveguide design was chosen to achieve superb sensitivity and the time-of-flight measurement for each particle flowing by. 02005 Optical Society of America OCIS codes: (130.0130) Integrated optics; (170.0170) Medical optics and biotechnology Since the first cell counter was invented in 1950s and commercialized in 1970s, flow cytometry has become one of the most needed and widely used tools for clinical diagnosis and biomedical research. Its applications include white blood cell analysis for AIDS diagnosis, cancer diagnosis, and stem cell sorting, just to name a few [1,2]. Developing a compact, low cost, and easy to operate flow cytometer system has been a goal for researchers in this area [3-8]. However, it is still a farfetched goal to turn a flow cytometry system into a laptop or hand-held unit and major technology breakthrough is required. In this paper, we report a fluidic photonic integrated circuit (FPIC) being able to perform one of two core functions of flow cytometry: cell detection with superb sensitivity. The biggest challenge for flow-cytometer-on-a-chip is detection sensitivity, which needs to be significantly enhanced to compensate for the low excitation intensity as a low cost, weak excitation source (e.g. LED lamp) that will replace the mainframe laser used in today's system. To solve this problem, we have invented an array-waveguide-architecture with multiple interrogation zones so that any object in the flow can be detected many times on the flight. Figure 1 shows the schematic design and the microscopic picture of (part of) the device. At the opposite ends of the fluidic channel are two waveguides for the excitation light. Perpendicular to the flow direction are two waveguide arrays, 8 waveguides in each array, to detect the fluorescence light (and/or side scattered light) produced by the objects. The waveguide dimension is 50x50 Vm2, made of polymer of higher index than the cladding layer. The center-to-center spacing between waveguides is 100 pm. When a particle (micro-bead or cell) travels through the region of array waveguides, the fluorescent signal is detected 8 times by the detector at each end of the waveguide. Because the measured signal is correlated with a time delay, one can use the technique of cross-correlation to achieve great enhancement of the signal-to-noise ratio. To characterize the device, we first used fluorescent micro-beads with a lOPm diameter. When such beads travel through the interrogation zones, the relatively bright fluorescent signal was detected by detectors (CCD) connected to the 8 waveguides. The sequential images were recorded by the CCD camera and the intensity of each channel was measured. Fig. 2a shows the cross section of the 8- channel waveguide array. The striations in Fig. 2a were caused by waveguide cutting and can be removed by polishing. We did not do the end facet polishing because the striations did not disturb the measurements. Figure 2b shows the sequential images where each waveguide was lightened up when the fluorescent particle flew by. To demonstrate high-sensitivity detection, we then used ltm fluorescent particles with fluorescent intensity 1000 times weaker than that in previous experiment. Fig. 3a shows the superimposed data detected at the outputs of 8 waveguide channels. Apparently, the fluorescent signal was too noisy to be recognized. This signal also represents the signal detected by a conventional flow cytometer provided that the excitation laser is replaced with a low power light source (i.e. a lamp). However, in our device with eight detection zones in sequence, the true signal detected at each zone was time correlated while the noise was not. Therefore, after performing the cross-correlation analysis that can be done by software or hardware, we can restore the signal with an exceedingly high signal-to-noise ratio, as shown in 2148