We have designed and numerically simulated a novel spot size converter for coupling standard single mode fibers with 10.4μm mode field diameter to 500nm × 220nm SOI waveguides. Simulations based on the eigenmode expansion method show a coupling loss of 0.4dB at 1550nm for the TE mode at perfect alignment. The alignment tolerance on the plane normal to the fiber axis is evaluated at ±2.2μm for ≤1dB excess loss, which is comparable to the alignment tolerance between two butt-coupled standard single mode fibers. The converter is based on a cross-like arrangement of SiO<sub>x</sub>N<sub>y</sub> waveguides immersed in a 12μm-thick SiO<sub>2</sub> cladding region deposited on top of the SOI chip. The waveguides are designed to collectively support a single degenerate mode for TE and TM polarizations. This guided mode features a large overlap to the LP01 mode of standard telecom fibers. Along the spot size converter length (450μm), the mode is first gradually confined in a single SiO<sub>x</sub>N<sub>y</sub> waveguide by tapering its width. Then, the mode is adiabatically coupled to a SOI waveguide underneath the structure through a SOI inverted taper. The shapes of SiO<sub>x</sub>N<sub>y</sub> and SOI tapers are optimized to minimize coupling loss and structure length, and to ensure adiabatic mode evolution along the structure, thus improving the design robustness to fabrication process errors. A tolerance analysis based on conservative microfabrication capabilities suggests that coupling loss penalty from fabrication errors can be maintained below 0.3dB. The proposed spot size converter is fully compliant to industry standard microfabrication processes available at INO.
INO has developed a hermetic vacuum packaging technology for uncooled bolometric detectors based on ceramic leadless chip carriers (LCC). Cavity pressures less than 3 mTorr are obtained. Processes are performed in a state-of-the art semi-automated vacuum furnace that allows for independent activation of non-evaporable thin film getters. The getter activation temperature is limited by both the anti-reflection coated silicon or germanium window and the MEMS device built on CMOS circuits. Temperature profiles used to achieve getter activation and vacuum sealing were optimized to meet lifetime and reliability requirements of packaged devices. Internal package components were carefully selected with respect to their outgassing behavior so that a good vacuum performance was obtained. In this paper, INO’s packaging process is described. The influence of various package internal components, in particular the CMOS circuits, on vacuum performance is presented. The package cavity pressure was monitored using INO’s pressure microsensors and the gas composition was determined by internal vapor analysis. Lifetime was derived from accelerated testing after storage of packaged detectors at various temperatures from room temperature to 120°C. A hermeticity yield over 80% was obtained for batches of twelve devices packaged simultaneously. Packaged FPAs submitted to standard MIL-STD-810 reliability testing (vibration, shock and temperature cycling) exhibited no change in IR response. Results show that vacuum performance strongly depends on CMOS circuit chips. Detectors packaged using a thin film getter show no change in cavity pressure after storage for more than 30 days at 120°C. Moreover, INO’s vacuum sealing process is such that even without a thin film getter, a base pressure of less than 10 mTorr is obtained and no pressure change is observed after 40 days at 85°C.
We describe the fabrication process of silicon nitride (Si<sub>3</sub>N<sub>4</sub>) based two-dimensional photonic crystals. The fabrication process mainly involves e-beam direct-write lithography and reactive ion etching. The concerned photonic crystal structures consist of a periodic arrangement of sub-micrometric holes transferred into a suspended Si<sub>3</sub>N<sub>4</sub> membrane using a poly-methylmethacrylate resist layer as a mask. Numerical simulations based on a plane wave expansion method for 2D photonic band gap approximation were conducted to determine the design parameters of the photonic crystal membranes. Flat and stress free photonic crystal membranes were achieved with very good control in sidewall profile and feature shape.
Modified thermal sensors have been produced and characterized for fingerprint recording applications. The sensors are derived from the IR imaging technology developed at INO. The sensor array is made of 160x120 pixel VO<sub>x</sub> based micro thermistors that provide an image of a surface area of 8.3 x 6.2 mm<sup>2</sup> with a resolution of 488 dpi. The sensors were reinforced to withstand the mechanical pressure of the finger and the electrical discharges from the human skin. It is shown that despite their low thermal insulation, the sensors provide an image of the fingerprint pattern with relatively high contrast and resolution. With the acquisition electronics of an IR imager, the temprature of the sensor must be controlled. Measurements of the thermistor temperature were performed in order to access the intrinsic properties of the fingerprint sensors. The NETD is on the order of 2 10<sup>-3</sup>°C when the pass band of the filter is 330 kHz. The temporal behavior of the thermistor temperature shows that 10 ms after the finger has been brought into contact, with the sensor, the temperature difference between thermistors in ridge and valley areas of the fingerprint DT<sub>r,v</sub> may reach 80 10<sup>-3</sup>°C, for an initial temperature difference between the finger and the sensor of 1°C. Once the sensor reaches a steady thermal state after a long time, the same difference decreases to 1.9 10<sup>-3</sup>°C. The required temperature difference DT<sub>r,v</sub>, estimated to be 4.8 10<sup>-3</sup>°C to achieve an adequate signal to noise ratio, is relatively easy to reach at short and at long time periods. A modification to the method of acquisition is proposed to cancel the effect of the thermal drift of the sensor and to eliminate the need for the sensor temperature stabilization with a TEC. With this method, the recording of the fingerprint pattern may be achieved in 50 ms after the finger has been brought into contact. This leads to interesting gains in space, time and power consumption. Finally, for applications where the finger must remain in contact with the sensor, the same method may be efficient to reduce the need for thermal control.