The second part of this three part feature discusses the technology behind uncooled detectors technology.
By: Christopher Alicandro, FLIR Systems Inc.
Stabilization and isolation
Advantages and drawbacks
System designers face engineering tradeoffs in deciding whether to choose cooled or uncooled detectors for their particular application. Part I (see Cooled or Uncooled IR Imagers: Which system is right for me?) of this feature discussed quantum effect, or cooled, detector technology. This installment covers the technology, advantages, and drawbacks of uncooled detectors; the final installment will discuss how to determine what technology is best for your system.
While the most common type of IR detector has traditionally been a quantum effect system, by no means is that the only choice. Thermal detector technology has been available for several years. Limitations in both a technical and a manufacturable standpoint, however, prevented their use in commercial based systems.
The fundamental operating principles of thermal detectors are intrinsically different than those of quantum effect detectors. The sensor material in thermal detectors does not count photons of energy as their quantum effect cousins do. Rather, their material physically heats and cools in response to thermal variations in the temperature scene. When the detector physically changes temperature, the resistive properties of the material change. By injecting a bias current into the detector, these changes in exit voltage can be readily measured (see Figure 4).
To better understand how a thermal detector operates, consider a standard mercury thermometer. When the bulb heats up, the mercury (or other fluid) rises. The rise or fall of the mercury is directly related to the temperature to which the thermometer bulb is subjected.
Historically, thermal detectors have been characterized by slow thermal response times that would affect the sensitivity as well as image frame rate. However, modern micro-machining technology has allowed manufacturers to produce thermal detectors made up of arrays of microscopic elements (pixels) which have response times similar to those of quantum effect detectors.
Thermal detectors enable the production of systems that do not require cryogenic cooling. Eliminating the need for coolers allows manufacturers to significantly reduce not only the weight and power consumption of the systems, but the cost. However, as this technology is still in the infancy stages, technical challenges still arise which inevitably delay the cost savings to the end user.
Stabilization and isolation
The relationship of scene energy to detector temperature is obviously not one-to-one. In fact, for each 1°C change in scene temperature, thermal detectors changes temperature only 1/300th of a degree. Because the systems today require sensitivity to 0.1°C, this means that the detector must be able to discern better than 1/3000th of a degree. Thus temperature control of the detector is vital.
The term "uncooled" as applied to detectors is actually a misnomer. Because thermal detectors base their performance on the physical heating and cooling of the material physical heating and cooling, thermal isolation is critical to performance. Thus, although systems do not incorporate traditional cryogenic coolers, they do require some form of thermal stabilization. Designers invest substantial effort in the thermal isolation of the detector elements to ensure that parasitic radiation from the system electronics does not compromise operation.
To enhance thermal stability, "uncooled" designs incorporate thermo-electric (TE) controllers that keep the detector at a fixed temperature independent of the ambient temperature. Typically the detector is maintained at approximately 23-25°C.
To ensure that the thermal detector is isolated from the parasitic energy present from the camera electronics, most sensors incorporate the use of a radiation shield. The shield can be thought of as a tunnel that does not permit stray radiation from impinging onto the detector.
The two most common types of thermal detectors are pyroelectric and microbolometer-based. Pyroelectric detectors are made from a temperature-sensitive pyroelectric ceramics such as barium strontium titanate (BST) or doped zirconate titanate (PZT). The materials produce a transient voltage when they undergo physical temperature changes. Conversely, if there is no temperature change, they produce no voltage.
As a result, pyroelectric detectors require the use of a chopper to modulate the incoming radiation. This chopper can be a mechanical wheel that has slits to allow radiation to pass at timed intervals. Alternately, work is being done on electronic chopping, similar to thermal subtraction. Ferroelectric detectors are a subset of the pyroelectric family.
The pyroelectric detectors have been made manufacturable with respectable yields and at moderate prices. There are technical challenges that need to be overcome to allow this technology to be suitable for temperature measuring devices, however. The sensors typically have very limited dynamic range, which severely limits the temperature spans that the camera is capable of viewing. In addition, the requirement for chopping does not lend itself for repeatable calibration results.
Bolometric detectors were actually introduced in 1880 by American physicist Samuel Langely. Each detector element changes its resistance when it heats and cools as a result of the incident radiation impinging upon it. The detector carries a bias current, and readout electronics measure the minute changes in voltage.
Although the technology was available for more than 100 years, bolometer-based thermal detectors were not feasible because of the slow response time and limited sensitivity of the devices. To overcome these challenges, the bolometer had to be substantially reduced in physical size. This feat was not possible until recently when micromachining technology was perfected (see Figure 4, at top).
The bolometers, which are made from a vanadium oxide compound (VOx), use a two-layer design in which the IR detector and readout electronics are on separate levels. This hybrid style design more efficiently utilizes detector area, and increases the fill factor, or ratio of active detector material to inactive area, called the fill factor.
The fabrication process requires a readout structure to be layered on a silicon wafer. On top of this layer of readout electronics, a sacrificial layer is added. After this step, the bolometric material is formed and layered along with silicone nitride. Using micromachining technology, the sacrificial layer is removed to leave a miniature silicone nitride bridge supporting a VOx microbolometer.
A paradox arises in the design process: the bridge "legs" must be thick enough to properly support the microbolometer. However they must be small enough to minimize heat transfer. If the legs are too thick, the structure will be very rugged but there will be excessive heat sinking which will affect the performance. If the legs are too thin, there will be minimal heat sinking however the integral structure will be very fragile.
Advantages and drawbacks
One of the benefits of the microbolometer, besides the elimination of the cryogenic cooler, is that the detectors are responsive to energy in the longer wavelengths, typically 7.5 to 12µm (see Figure 5). This thermal responsivity without the need for cryogenic coolers means that manufacturers can produce long-wave, focal-plane-array devices at a reasonable price.
Ironically, the inherent lower cost as a result of the elimination of the cooling system is counteracted by the cost of system optics, however. Because of the inherent sensitivity issues associated with the microbolometers, the detectors require "fast" optics. A optical component is considered fast when the f-number the ratio of effective focal length to clear aperture is small. Faster optics will require larger lenses and in the case of longwave germanium, more expense. Inevitably, the costs will be reduced.
Because microbolometers are thermally sensitive, there is flexibility in the spectral response. There is currently work being performed to "tune" the microbolometer to be responsive in the shorter wavelengths. The tuning is executed by lowering the height of the microbolometer bridge closer to the readout array. Typically, certain wavelengths pass through the microbolometer. If the microbolometer is lowered closer to the readout electronics, some of the transmitted energy is reflected back to the detector
This material first appeared in a technical paper presented at Infraspection Institute's IR/Info '99, Las Vegas, NV, January 17-20, 1999.
About the author…
Chris Alicandro is new business development manager at FLIR Systems Inc., 16 Esquire Rd., North Billerica, MA 01862. Phone: 978-901-8227; fax: 978-901-8887; e-mail: mailto:firstname.lastname@example.org.