Detector Technology

Infrared Detector Arrays for Thermal Imaging
Tutorial "Infrared Detectors"

Load this page as PDF

Detector Technology - Overview Detector types and materials

Photon detectors
The absorption of long-wavelength radiation by photon detectors results directly in some specific quantum event, such as the photoelectric emissionof electrons from a surface, or electronic interband transitions in semiconductor materials. Therefore, the output of photon detectors is governed by the rate of absorption of photons and not directly on the photon energy. They normally require cooling down to cryogenic temperatures in order to get rid of excessive dark current, but in return their general performance is higher, with larger detectivities and smaller response times. QWIP is a type of photon detector.

Photon detectors can be further subdivided into photoconductive and photovoltaic devices. The function of photoconductive detectors are based on the photogeneration of charge carriers (electrons, holes or electron-hole pairs). These charge carriers increase the conductivity of the device material. Detector materials possible to utilize for photoconductive detectors are:

• indium antimonide(InSb)
• quantum well infrared photodetector (QWIP)
• mercury cadmium telluride (mercad, MCT)
• lead sulfide (PbS)
• lead selenide (PbSe)

Photovoltaic devices require an internal potential barrier with a built-inelectric field in order to separate photo-generated electron-hole pair. Such potential barriers can be created by the use of p-n junctions or Schottky barriers. Whereas the current-voltage characteristics of photoconductive devices are symmetric with respect to the polarity of the applied voltage, photovoltaic devices exhibit rectifying behaviour. Examples of photovoltaic infrared detector types are:

• indium antimonide (InSb)
• mercury cadmium telluride (MCT)
• platinum silicide (PtSi) -silicon Schottky barrier

Photon detectors may also be classified on the basis of whether the photo-transitions take place across the fundamental band gap of the infrared sensitive material, or from impurity states to either of the valence or the conduction band. In the first case they are denoted intrinsic, in the latter case extrinsic. The quantum well type of detector discussed below is however not easily classified according to this criterion.

As mentioned above the cooling requirements of a detector are governed by the fact that dark current decreases with lower operating temperature. In addition, dark current increases the larger is the cut-off wavelength of the detector (i. e. the largest wavelength at which the detector have a response). In general, for constant cut-off wavelength and temperature, dark current is lower for intrinsic detectors than for extrinsic ones. Therefore the cooling requirements are less pronounced for the former type.

In most cases photon detectors need to be cooled to cryogenic temperatures, i. e. down to 77 K (liquid nitrogen) or 4 K (liquid helium). In some favorable cases thermoelectric cooling down to 200 K is sufficient (e. g. 3 - 5 um wavelength mercury cadmium telluride). The main workhorse in the field of photon detectors is mercury cadmium telluride or "mercad" (HgCdTe), and to some less extent indium antimonide (InSb). Vigorous work has been done on mercad both in US and Europe since its discovery in 1959 and work is still being done. Mercad is used both for the 3 - 5 um (MWIR) and 8 - 12 um (LWIR) atmospheric transmission windows, whereas indium antimonide is solely for the 3 - 5 um range. Platinum silicide (PtSi) Schottky barrier detectors also work in the MWIR domain. Large (512x512 pixels) PtSi focal plane arrays (FPA -seebelow!) have been fabricated, are compatible with silicon CCD/CMOS technology, and show high performance, due to the extremely good pixel to pixel uniformity, in spite of the very low quantum efficiency. As regards FPAs for the 3 - 5 um window, both mercad, InSb and PtSi materials pose no major technological problems and are considered to be a finished product. In contrast, to date, no photondetector FPAs operating in the 8 - 12 um window exhibits sufficient performance to be operated at 77-80 K.

In the course of only the last five years, detectors based on low-dimensional structures have evolved as viable candidates for FPAs (focal plane arrays) especially in the LWIR region. These so called band-gap engineered detectors operate on account of electronic transitions between electronic states arising as a result of size quantization, i. e. electron energy quantization due to the small layer dimensions in the growth direction. There are threemaincandidates of interest for IR detector arrays: i) the AlGaAs/GaAs quantumwells, ii) the strained SiGe/Si superlattices (SL), and iii) the strained InAs/GaInSb SLs.

The most mature is the AlGaAs/GaAs quantum well (QW) structure, which is a spin-off from GaAs technology. This detector type is generally named Quantum Well Infrared Photoconductor or QWIP. Here special grating structures are necessary in order to achieve a high quantum efficiency of the detector. QWIP FPAs need operating temperatures around 70-75 K in order to work properly, temperatures which are easily achievable by miniature Stirling coolers.

The main advantages of SiGe/Si QWs are the compatibility with silicon technology and that grating structures are not necessary. The cooling requirements seem, however, to be more extensive than for AlGaAs/GaAs quantum wells.

InAs/GaInSb so called type II SLs in theoryoffer the possibility of high sensitivity and operating temperatures of an intrinsic detector. In addition, the materials processing and uniformity are expected to be superior than for III-VI materials such as mercad. However, presently the maturity of the detector technology is far from being comparable to mercad detectors.

Thermal detectors
In contrast to photon detectors, the operation of thermal detectors depends on a two-step process. The absorption of infrared radiation in these detectors raises the temperature of the device, which in turn changes some temperature-dependent parameter such as electrical conductivity. Thermal detectors may be thermopile (Seebeck effect), bolometer, Golay cell detectors, or pyroelectric detectors.

In fact, when Herschel discovered the infrared spectrum in the year 1800 he utilized asimple type of thermal detector, a liquid in a glassthermometer, which is considered the first thermal detector. The infrared radiation was absorbed by blackening the bulb of the thermometer.

The major advantage of thermal detectors is that they can operate at room temperature. However, the sensitivity is lower and the response time longer than for photon detectors. This makes thermal detectors suitable for focal plane array operation, where the latter two properties are less critical.

A thermal detector is conveniently divided into three functional parts.

• Absorber for infrared radiation
• Membrane or other structure for achieving thermal insulation
• Temperature detector

The absorber can be a finely subdivided metal such as platinumblack, or be based on an interferometric structure.

The temperature detector is usually intregrated into the membrane, and utilized to detect the usually minute temperature change resulting from exposure to infrared radiation from a room-temperature scene and subsequent absorption. Thermal detectors are conveniently classified according to their means of detecting this temperature change:

A resistive bolometercontains a resistive material, whose resistivity changes with temperature. To achieve high sensitivity the temperature coefficient of the resistivity should be as large as possible and the noise resulting from contacts and the material itself should be low. Resistive materials could be metals such as platinum, or semiconductors (thermistors). Metals usually have low noise but have low temperature coefficients (about 0.2 %/K), semiconductors have high temperature coefficients (1-4 %/K) but are prone to be more noisy. Semiconductors used for infrared detectors are e. g. amorphous, polycrystalline silicon, or vanadium oxide.

A thermoelectric device (thermocouple or thermopile) is based on the presence of one or several junctions between two materials. The junctions properly arranged and connected develop a thermo-emk that changes with temperature, the so called Seebeck effect. In order for the sensitivity to be high the Seebeck coefficientshouldbeas high as possible. Certain alloys containingantimony and bismuth have very high Seebeck coefficients of 150 uV/K. The CMOS compatible combination aluminum/polycrystalline silicon gives about 65 uV/K.

A pyroelectric detector is based on the fact that certain dielectric materials of low crystal symmetry exhibit spontaneous dielectric polarization. When the electric dipole moment depends on temperature the material becomes pyroelectric. Usually a capacitor is fabricated from the material and the variation of charge on it is sensed. Common pyroelectric materials used for infrared detectors are TGS (tri-glycine sulphate), LiTaO3 (lithium tantalate), PZT (lead zinc titanate) and certain polymers. A dielectric bolometer makes use of pyroelectric materials operated in a way to detect the change of the dielectric constant with temperature. A suitable material for this application is SBT (strontium barium titanate).

The Golay detector is based on the volume or pressure change of an encapsulated gas with temperature. The volume change is measured e. g. by the deflection of lightrays resulting from the motion of properly positioned mirrors fastened to the walls of the gas container.

Infrared Imaging
There are two basic types of infrared imaging systems: mechanical scanning systems and systems based on detector arrays without scanner. It should be mentioned that detector arrays as well are used for scanning systems, but the number of detector elements (picture elements - pixels) generally is smaller in this case.

A mechanical scanner utilizes one or more moving mirrors to sample the object plane sequentially in a row-wise manner and project these onto the detector . The advantage is that only onesingle detector isneeded. The drawbacksare that high precision and thus expensive opto-mechanical parts are needed, and the detector response time has to be short. As mentioned above, detector arrays are also used for this application. For example, a long linear detector array can be used to simultaneously sample one column of the object plane. By using a single moving mirror the whole focal plane can be sampled. In contrast, when a single detector is used, two mirrors moving in two orthogonal directions must be used, one of them moving at high speed, the other one at lower speed.

Detector arrays operated as focal plane arrays (FPA) (or staring arrays) are located in the focal plane of a camera system, and are thus replacing the film of a conventional camera for visible light. The advantage is that no moving mechanical parts are needed and that the detector sensitivity can be low and the detector slow. The latter is a result ofthat theintegration time canbelong. The drawback is that the detector array is more complicated to fabricate. However, with the ascent of rational methods for semiconductor fabrication, economy will be advantageous, provided that production volumes are large. The general trend is that infrared camera systems will be based on FPAs, except for special applications.

The spatial resolution of the image is determined by the number of pixels of the detector array. Common formats for commercial infrared detectors are 320x240 pixels (320 columns, 240 rows), and 640x480. The latter format (or something close to it), which is nearly the resolution obtained by standard TV, will probably become commercially available in the next few years. Today, for example indium antimonide and platinum silicide detectors are commercially available in the 320x240 pixels format. Typical pitches between pixels are in the range 20-50 um.

Detector arrays are more complicated to fabricate, since besides the detector elements with the function of responding to radiation, electronic circuitry is needed to multiplex all the detector signals to one or a few output leads in a serial manner. The output from the array is either in analogue or digital form. In the former case analogue to digital conversion is usually done external to the detector array. The electronic chip used to multiplex or read out the signals from the detector elements are usually called simply readout integrated circuit (ROIC) or (analogue) multiplexer.

The ROIC is usually made using silicon CCD (charge coupled device) or CMOS technology. However, the detector elements must often be fabricated from more exotic materials as discussed above. The exceptions are e. g. platinum silicide or micro-bolometers which can be based on silicon technology. In the former case a hybrid approach is most common, in which case all thedetector pixels are fabricated from a separate detector chip. This detectorchip is then flip-chip bonded to the ROIC chip. Flip-chip bonding involves the processing of metal bumps onto contact holes, one per pixel, of both the detector chip and the ROIC. Using special equipment, the two chips are first aligned to each other. Then the chips are put in contact, while applying heat and/or mechanical force. During this process the two chips become electrically connected to each other via the metal bumps. Usually indium is used for the bumps due to its excellent low temperature properties.

Uniformity of the detector elements across the array is a key issue for obtaining high performance. In fact, individual pixel response characteristics differ considerably across an array in most cases. Therefore so called pixel correction has to be done prior to the presentation of the final image. This amounts to calibrating each individual pixel, by exposing the array to calibrated surfaces of known temperature.

Some definitions

Some definitions generally used in the infrared detector field are:

Responsivity
The electrical output of a detector divided by the power of the radiation striking it. The detector transfer function. Since electrical output can be either voltage or current, one distinguishes between voltage and current responsivity.

Voltage responsivity

where

is the voltage output, and F the optical power.

Similarly:

Current responsivity

Responsivity can be measured for monochromatic radiation, in which case the responsivity is denoted spectral responsivity. Alternatively, a blackbody source kept at a fixed temperature can be used. In this case one talk about blackbody responsivity. Spectral responsivity plotted versus wavelength is often used for presentation of a detector's spectral response properties.

Detectivity, D*
Whereas responsivity takes into account the detector's signal properties only, the detectivity or D* value is a measure of its signal to noise properties. The D* value is normalized with respect to detector area (provided that the signal to noise ratio increases with the square root of the detector area, which is often the case, at least for photon detectors). It is defined as:

where u_N and i_N is the noise voltage and current, respectively,
A_D the detector area and Df the noise bandwidth.

Temperature resolution, NETD
NETD is an abbreviation for Noise Equivalent Temperature Difference and is a measure of the smallest object temperature difference possible to detect by an IR camera.