What You Need to Know About IR Detectors
There are as many types of infrared—or thermal imaging—cameras as there are scientific applications. So, which type of camera is better suited to your R&D application?
Infrared cameras are often the missing piece needed to truly understand what’s happening in an R&D environment. Because they detect and visualize heat, these cameras provide insight into everything from engineering flaws in printed circuit board designs to the combustion of individual coal particles in a larger flame.
Within the families of science-specific cameras, researchers have a range of choices from entry-level, uncooled thermal cameras to high-performance cameras with a cooled detector. Choosing the correct model depends heavily on your intended application, so consider the variables below.
Cryocooled indium antimonide (InSb) camera core
Defining Cooled and Uncooled Detectors
Cooled infrared (IR) detectors use materials that must operate at cryogenic temperatures. These materials respond to individual photons of light, which makes them very sensitive and very fast—a good choice for high-performance imaging applications. A variety of material choices makes it possible for cooled cameras to operate in the near-infrared, mid-wave, long-wave bands. Their main drawbacks are cost and complexity: the detectors must be packed in a vacuum Dewar, while the cryocoolers are a mechanical device that requires periodic service and consumes significant power.
Uncooled infrared detectors, on the other hand, work at ambient temperature and do not require specialized cooling systems, making them simpler, more affordable, and more durable. These detectors primarily operate in the long-wave infrared (LWIR) range and are commonly used in security cameras, industrial monitoring, and general thermal imaging. While they have lower sensitivity and resolution than cooled detectors, their practicality and cost-effectiveness make them widely adopted across commercial and consumer applications.
Infrared Wavelength Regions
Looking at the electromagnetic spectrum, the infrared region spans wavelengths from approximately 780 nanometers to one millimeter, bridging visible light and microwaves. It is commonly divided into three subcategories: near infrared (0.78–3 µm), mid-wave infrared (3–8 µm), and longwave infrared (8–15 µm). Each infrared wavelength range serves a unique purpose, making infrared technology indispensable across multiple industries and applications.
Near infrared (NIR) is commonly used in industrial inspections to detect defects or differences in materials that might not be visible to the naked eye, but can also be used environmental monitoring, agriculture, scientific research. Its ability to see through layers of paint even allows NIR to be used in art restoration applications.
The true thermal band begins at 3 µm with the mid-wave band. This is a standard wavelength range for radiometry and high-performance thermography applications due to its excellent thermal contrast and sensitivity. It makes MWIR ideal for defense research and electronics design and testing applications.
Electromagnetic Spectrum
The longwave band starts at around 8 µm, a region where both uncooled and cooled infrared cameras operate. Better transmission properties through atmospheric conditions such as fog, smoke, and dust allow LWIR to be used very harsh industrial environments and long-range security operations.
Detector Speed
A camera’s speed depends heavily on the detector materials. For SWIR cameras and those imaging up to 1.7 µm, the most common sensor material is indium gallium arsenide (InGaAs). For applications in the MWIR band, cooled MW mercury cadmium telluride (MCT), indium antimonide (InSb), or MW-tuned Type II Strained Layer Superlattice (SLS) sensors provide both fast frame rates and high sensitivity.
For LWIR applications, there are silicon or metallic-based microbolometers that don’t require cooling, or there are cooled, high-performance MCT or SLS sensors, which provide fast integration speeds and a wide temperature range.
Microbolometers:
Cameras with microbolometers are most often sensitive from 7.5 µm to 14 µm wavelength, primarily because these detectors need the abundant supply of photons typical to the longwave band. They’re tuned for peak sensitivity at 10 µm for objects around 30°C or ambient temperature. But this tuning can have a trade-off, providing good sensitivity at the expense of speed. Because speed and sensitivity are inversely proportional to thermal resistance, a highly-sensitive bolometer will often be a slow bolometer.
Another challenge for bolometer speed is rolling shutter readout. Unlike global shutter detectors that capture an entire frame simultaneously, bolometers read row by row, creating distortions when capturing fast-moving objects or transient temperature variations. A general rule of thumb for good imaging with a microbolometer is events should be slower than 60 milliseconds—sufficient time for the sensor to stabilize between frames and accurately capture the scene.
Cryocooled detectors:
Cryocooled cameras incorporate a photon-sensitive detector with adjustable integration times to capture infrared energy; a readout integrated circuit (ROIC) to collect incoming signals, a cold filter to help limit the detector's sensitivity to specified wavebands, and a cooling system to reduce sensor noise.
Each pixel in the detector array functions as a unit cell, containing an integration capacitor that governs how much energy is accumulated during exposure time. Unlike bolometers, which rely on thermal detection, cooled detectors allow users to define integration time, offering precise control over image quality and timing, as well as faster frame rates and windowing capabilities.
A cooled sensor allows users to capture motion without blur, such as rifle suppressors undergoing intense heat transfer, while also capturing accurate temperature readings.
Detector Sensitivity
Thermal sensitivity, also called NETD or noise equivalent temperature difference, is a signal-to-noise figure that represents the temperature difference needed to produce a signal equal to the camera's temporal noise.
Commonly confused with camera specifications such as accuracy, the easiest way to understand NETD is to consider it the temporal noise floor of the camera expressed as an equivalent temperature value.
Uncooled, low-sensitivity image of a VXI board
Cooled, high-sensitivity image of a VXI board
Camera makers calculate NETD by dividing the standard deviation of the temporal noise by the response per degree or responsivity; they express it in units of millikelvins. When looking at camera specifications, remember that as NETD value decreases, the sensitivity of a camera will increase. So, while bolometers typically have an NETD of 30 mK to 50 mK, a cryocooled camera would have an NETD of 20 mK.
Both cooled and uncooled detectors experience temporal and spatial noise. Temporal noise happens randomly per pixel over time and is the most noticeable noise in a system that has a good uniformity—or at least good non-uniformity correction. Spatial noise does not change over time and can appear as darker corners from cold shield shading or as vertical bars caused by differences in the “ROIC” or readout integrated circuit. Performing a good non-uniformity correction, or “NUC,” with the camera will clean up the spatial or fixed pattern noise.
Where NETD comes into play for applications, is in the image detail. A cooled detector will image a target with more detail because each pixel is reading smaller steps in temperature compared to noise in the image. A microbolometer camera, experiencing more noise, can’t compare. So, when working on applications where you need to resolve low energy events or even transient thermal events, a bolometer-based detector won’t provide the same clear, sharp image as a photon-sensitive detector at the same pixel resolution.
When imaging subjects at the same temperature, cooled and uncooled cameras show only a degree of difference. But as the subjects cool and emit less energy, the advantage of a cooled camera's higher NETD becomes clear.
Initial image: cooled camera
Initial image: uncooled camera
Two min. later: cooled camera
Two min later: uncooled camera
Detector Spatial Resolution
Spatial resolution, often referred to as instantaneous field of view (IFOV) or spot size, defines the area that a single detector cell or pixel captures in an imaging system.
This concept applies universally—whether it is an uncooled or cooled detector or even a standard digital camera. The fundamental principle remains the same: each pixel covers a specific area as it extends outward into space.
To understand spatial resolution, it is helpful to first explore field of view (FOV). Think of FOV like a flashlight you’re shining on a wall: as you step backwards, increasing your distance to the wall, the area the beam covers (field) gets larger.
IFOV describes the area covered by a single pixel at a given distance. Determine IFOV using the following formula:
Spatial Resolution = 61 × Wavelength × Numerical Aperture
Beyond basic visualization, scientific and engineering applications rely on high spatial resolution for accurate measurements. For example, when imaging a printed circuit board (PCB), it’s important to obtain thermal data on very small targets. If a single pixel covers multiple components on the PCB, it will average the thermal data, making it impossible to obtain an accurate reading.
Synchronization and Triggering
The term "trigger" commonly describes the mechanism that initiates data capture, either based on an action happening or a set threshold. It can be deterministic, such as the rising edge of an electrical signal, or it can be as simple as pressing a record button.
Synchronization controls the generation of individual frames and happens continuously throughout the acquisition of data. This is only possible with a photon detector, meaning you need a cooled IR camera to perform trigger and sync functions. Uncooled thermal cameras can provide synchronization and triggering but they're limited to record-start and master-slave sync, often with less precision due to slower frame rates and rolling shutters.
To illustrate the importance of triggering over record-start, consider a high-speed experiment in which researchers analyze how a vehicle engine distributes, consumes, and expels new fuel sources. Since this is a rapid event, there may be only one opportunity to capture accurate data before the test window ends.
A trigger mechanism is essential for initiating recording, ideally in sync with other data acquisition systems and test instrumentation. To ensure precise recording, a control station monitors the vehicle and reads signals to identify the exact moment to trigger the recording. When the system detects the threshold condition—such as a fuel burn event inside the piston chamber—it sends a trigger signal to the camera and other acquisition devices, ensuring synchronized data capture.
Gain a deeper understanding of synchronization and triggering with our recorded webinar, What You Need to Know About IR Detectors - Synchronization.
SPECTRAL FILTERING
The spectral response of a detector is highly specific to its design and application. For example, a midwave IR camera exhibits a relative flat and stable spectral response, making it well suited to a wide range of thermal imaging applications. An SLS LWIR camera shows peaks and troughs in its photon response, which can influence spectral filtering options.
Selecting the right spectral filtering setup depends on the specific measurement goal and signal strength.
Warm vs. Cold Filters
Cameras can be spectrally filtered with warm or cold filters. Warm filters are external, often placed behind the lens or inside a motorized wheel. While they’re easy to use and can be changed out frequently, they can generate thermal radiation that adds noise to the image. Warm filters do require careful calibration, especially when used with a narrow band-pass to isolate weak signals.
Cold filters are integrated with the camera itself. They are cooled along with the sensor, ensuring there’s no radiation interference. While they offer maximum sensitivity and accuracy, they’re also a permanent part of the detector.
Typically, cameras with spectral filters will feature cooled detectors. An exception would be the FLIR GF77: an uncooled optical gas imaging camera with removable lenses that are spectrally filtered to detect specific gases.
So, where could spectral filtering come into play? An example would be an application where it is necessary to measure the temperature of a material such as thin film plastic, which is transparent to most MWIR wavelengths but opaque to 3.4–3.5 µm wavelengths. By filtering the camera to only the opaque wavelengths it is possible to measure the temperature of the plastic.
Learn more about spectral filtering by watching the final recorded webinar in our series, What You Need to Know About IR Detectors - Spectral Filtering.
Spinning car wheels imaged with a cooled, high speed IR camera (left) and uncooled IR camera (right).
Infrared imaging technology has evolved to provide unparalleled precision in thermal detection, with both cooled and uncooled cameras offering distinct advantages depending on the application. Cooled detectors excel in high-speed, high-resolution imaging, enabling precise thermal measurements and motion-free clarity, while uncooled microbolometers are more accessible for general thermal imaging applications.
Factors such as spectral response, NETD, spatial resolution, triggering, and synchronization significantly impact imaging performance. Choosing the right detector type, wavelength range, and filtering method ensures that researchers and engineers can obtain accurate, high-quality infrared data for their specific needs.
