656 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 16, NO. 3, MAY/JUNE 2010 (1) and (2) can be rewritten as follows: ����� = ��m ,i neff ��sense ���nsense (3) where ��m ,i /neff ,i can be regarded as a constant reflecting the initial conditions at the ring surface just before the deposition of a film in the sensing region. This allows a simple thin-film- based model to be constructed, relating cavity shift to the mean thickness of the resulting film. This model is valid where the quantity of bound molecules on the sensor surface is sufficient to represent a thin film, as discussed later. In order to couple light into the ring, a straight waveguide is fabricated and brought in close proximity to the ring waveg- uide, such that the respective optical fields of the waveguide and the ring are coupled, allowing the ring to interact with the waveguide that is typically terminated by a photodetector at the other end. This configuration is shown in Fig. 1(b). In order to achieve a high extinction ratio of the cavity resonance, a con- dition called critical coupling must be satisfied [19]. At critical coupling, little or no light travels past the ring toward the pho- todetector. This occurs when the coupling loss between the feed waveguide and the ring equals the round-trip propagation loss inside the ring, and the wavelength of the laser is tuned to a cavity mode of the ring. This means that the gap between the waveguide and the ring must be set based on the waveguide loss of the manufacturing process being utilized. In practice, gaps ranging between 180 and 240 nm have resulted in an extinction ratio of better than 10 dB. This type of process control is well within the bounds of a modern lithographic process, which typ- ically has tolerances well below 10 nm. A typical ring transmis- sion spectrum is shown in Fig. 1(c), where cavity Q = 43000, ER ��� ���15 dB, and FSR = 5.98 nm. III. BIOSENSING PLATFORM ARCHITECTURE In order to realize the full potential of a ring-resonator-based sensing platform, a system for efficient and expedient interroga- tion of sensor arrays was developed. Fig. 2 shows the high-level system architecture where light from an external cavity tunable laser is sourced via a fiber optic system to a free-space opti- cal scanner that shapes, guides, and focuses the optical beam onto the sensor array chip. With each wavelength sweep by the laser, the location of the ring resonance is recorded by the data acquisition and processing systems. A. Sensor Array Chip The sensor array chip comprises 32 individually address- able rings accessed via waveguides. The chips are fabricated in collaboration with the European network of excellence on pho- tonic integrated components and circuits (ePIXnet) [20], [21]. As shown in Fig. 2, light is coupled in and out of the chip via linear grating couplers [22], [23] etched into multimode waveg- uide expansions at the ends of the silicon waveguides. The chip is encased in a cassette that comprises fluidic ports, channels, reservoirs, and a mechanism to interface the fluidics with the sensor chip. All 32 sensors can be accessed using two or more microfluidic channels formed by placing and aligning a gasket over the chip that contains the imprint of the channels. Flow Fig. 2. High-level architecture of the biosensing platform. channel dimensions are 175 ��m �� 500 ��m. Additionally, the entire chip is spin coated with a perflouropolymer cladding. Us- ing lithography and etching, annular windows are opened over the 24 rings used as sensors such that their surface is exposed to samples flowing through the fluidics channels [see Fig. 1(b)]. The other eight rings, left under the cladding are used as con- trols to remove temperature-induced drift. All the input/output grating couplers are placed at the chip edges, which are left optically accessible to the free-space optical scanner. B. Optical Scanner The optical mode profile of a grating coupler is matched to that of a single-mode optical fiber with a coupling incidence an- gle of approximately 12��� and a typical insertion loss of 6 dB [22]. In order to achieve maximum coupling efficiency, the mode overlap integral requires the beam emerging from the optical scanner to meet strict specifications. These include spot diam- eter, encircled energy distribution, wavefront error, numerical aperture, and the chief ray angle of the incident beam. Optics inside the scanner are carefully designed to meet these require- ments. Since grating couplers are highly polarization sensitive, the scanner was designed to provide linearly polarized light to the chip. Our system repeatedly achieves an average grating coupler insertion loss of ���5.5 dB. The optical scanner employs two tip-tilt beam steering mirrors to direct the free-space beam spot from one sensor to the next. When the chip is first placed in the instrument, the mirrors are used to perform a raster scan, creating an image of the chip surface by looking at the reflection from the laser spot. This Authorized licensed use limited to: University of Washington Libraries. Downloaded on June 14,2010 at 00:22:22 UTC from IEEE Xplore. Restrictions apply.