Orthogonal-Transfer Arrays

For ground-based telescopes with a wide field of view, the atmospheric distortion of the wave front will vary over the focal plane, so that a single large tip-tilt mirror or orthogonal-transfer charge-coupled device (OTCCD) may not be effective. Several years ago Lincoln Laboratory began development of the orthogonal-transfer array (OTA). This device consists of an array of individual OTCCDs on a single silicon substrate with independent control circuitry.

Figure 1. An OTA is composed of an 8 x 8 array of individually controlled OTCCDs.Figure 1. An OTA is composed of an 8 x 8 array of individually controlled OTCCDs.


Figure 1 shows the layout of the OTA. Each OTA consists of an 8 × 8 array of OTCCDs, each of which is approximately 600 x 600 pixels in size, resulting in ~23 million pixels for a single device. The control circuitry is located in the lanes between the closely spaced OTCCDs and allows one row (8 OTCCD cells) to be read out while all other rows are imaging.  Also, the integrating charge in each OTCCD cell may be moved independently of all other cells during the integration period to accommodate the different atmospheric distortions across the OTA.

Figure 2. A 100 mm silicon wafer contains four orthogonal-transfer-array devices.Figure 2. A 150 mm silicon wafer contains four OTA devices.

Figure 2 is a picture of a processed 150 mm silicon wafer containing four of the OTA devices. Each OTA is designed to allow for close abutting of neighbor OTA devices, therefore allowing focal planes of arbitrarily large size. We designed and fabricated these devices for the PanSTARRS program, which has assembled a focal plane containing 60 of these devices with approximately 1.4 billion pixels.

Though the OTA was developed for astronomical use, it has a number of features that are important for large focal-plane arrays (FPAs) for surveillance. For example, the partitioning of the imager into 64 smaller blocks enables more rapid readout than conventional CCDs with one or a few readouts. The current readout time is about four seconds for all 23 million pixels at a 1 Mpixel read rate and is dictated by noise considerations (a row of eight devices is read out simultaneously). We are at present experimenting with a next-generation OTA with JFET-based charge-sensing amplifiers, currently in fabrication, which will offer faster read rates for a given noise level.

We have also developed a substrate-bias feature which enables us to make relatively thick (75 µm) very-deep-depletion devices for high near-IR quantum efficiency (QE) while retaining a charge point-spread function that is small compared to the 10-µm pixel of this device. As a result, the QE reaches nearly 100% at 800 nm; this enhances the sensitivity of the device for night-vision use.




top of page