Making sense of IR pixel-pitch reduction

The advancement of infrared (IR) sensor technology during the past few decades has led to significant innovations in pixel-pitch reduction, enabling vast improvements in the size, weight, power, and cost (SWaP-C) of fixed field-of-view (FOV) mid-wave (MW) and long-wave (LW) thermal camera modules, known commonly as thermal cores. While such developments have opened a host of new markets and applications for thermal imaging—including the use of thermal imaging modules embedded into mobile phones or drone payloads—the relationship between pixel size and system optimization is more complex than it might initially appear to be.

After all, pixel-size reduction is commonly thought of as the process of decreasing the size of and the distance between pixels on a display that results in increased image resolution and improved visual quality. However, especially in advanced IR camera systems featuring a continuous zoom (CZ) lens, that isn’t always the case.

System engineers are challenged to optimize component specifications and system performance. They may have to complete a tradeoff analysis at the component and system levels to ensure the most appropriate design for the application, especially where SWaP-C is not the most critical factor compared to specific performance requirements, such as range or thermal detection sensitivity.

Based on Teledyne FLIR’s comprehensive experience in manufacturing everything from read-out integrated circuits and detectors to cryocoolers and complete LW- and MW IR thermal camera modules, this article provides an introduction to the practical implications of pixel-size reduction, its advantages, and its challenges. As the world’s only vertical IR hardware and perception stack provider, Teledyne FLIR offers a unique perspective on the relationship between component specifications and IR system optimization.

The theoretical benefits of smaller pixel-pitch seem straightforward at first glance. When considering a given FOV and pixel resolution requirements, reducing pixel pitch should allow for a proportional reduction in both the focal plane array size and the effective focal length (EFL). However, the simplified view does not account for the many technical challenges and trade-offs that emerge in real-world applications, especially in MWIR systems with CZ capabilities.

The following critical factors complicate the relationship between pixel size and system performance.

• Smaller pixels absorb fewer infrared photons, requiring faster f-numbers to maintain sensitivity. The requirement often results in system diameters similar to or even larger than those of systems with larger pixels. The need for faster f-numbers also affects spatial resolution, as achieving diffraction-limited performance becomes increasingly challenging and expensive with f-numbers below f/3. Complications extend beyond basic optical considerations.
• Smaller pixels face challenges with dynamic range, as the reduced well capacity directly affects system performance.
• Smaller pixels are more difficult to manufacture, often resulting in reduced quantum efficiency, increased dark current, and higher fixed-pattern noise. As a result, they are more difficult to operate with good performance at higher operating temperatures. Higher operating temperatures equate to lower system power, faster time to cooldown, and longer cooler lifetimes.
• Pixel-to-pixel crosstalk becomes more difficult to control as the pitch-to-diffusion-length ratio decreases, leading to degraded system modulation transfer function and overall performance reduction.

A real-world comparison of three system designs with 5-, 8-, and 15-μm pixel pitches reveals surprising insights about the practical implications of pixel-size reduction. When designing a system with a 10´ CZ lens and while maintaining equivalent spatial resolution and instantaneous FOV, the smaller pixel-pitch drives requirements for faster f-numbers, which necessitate more complex optical designs. This increased complexity manifests in several ways: more optical elements, larger system dimensions, and increased weight and cost.

The cost implications of pixel-size reduction are noteworthy, because while smaller pixels might seem to offer cost advantages, the system-level analysis reveals a more nuanced reality. When considering both optics and camera module costs (including die, wafer, cryocooler, and other components), the optimal cost point for a MWIR system with a 10´ CZ lens occurs at 8 μm pixel pitch. This configuration offers 9 percent cost savings compared to a 5 μm system and 19 percent savings compared to a 15 μm system, demonstrating that pursuing ever-smaller pixels eventually leads to diminishing returns.

Not surprisingly, larger pixel-pitch systems can offer superior performance in terms of detection, recognition, and identification capabilities. Using the industry standard night-vision integrated performance model with practical detector, read out, and optics parameters for each configuration, and then considering a target with a 3.1 m critical dimension and 4 K temperature variation, the detection, range, and identification ranges increased with increasing pixel pitch. In short, this experiment gives the 15 μm pixel-pitch system the advantage.

These findings challenge the conventional wisdom that smaller pixels automatically lead to system optimization. A full examination of CZ lens costs and IR system-level considerations reveals that SWaP-C and performance optimization do not necessarily align with minimizing pixel pitch. The increased optical complexity required for smaller pixels, manifesting in faster f-numbers, leads to larger, heavier, and more expensive optical systems.

The push toward ever-smaller pixels reaches a point of diminishing returns across multiple key system requirements. While fixed FOV lens systems might benefit from pixel-size reduction, systems incorporating CZ capabilities face a more complex set of trade-offs. The optimal solution often lies not in pursuing the smallest possible pixel-size, but in finding the sweet spot that balances optical performance, system complexity, cost, and practical implementation considerations.

Our analysis underscores the importance of taking a holistic approach to IR system design. Rather than focusing solely on pixel-size reduction as a path to improvement, designers and integrators must consider the complete system context, including optical requirements, cost constraints, and performance goals. The example of the 8 μm pixel pitch finding the optimal cost point, while the 15 μm system excels in performance metrics, illustrates the complex interplay of factors that successful IR system design must balance.

The relationship between pixel-size and system optimization in IR imaging systems is far more nuanced than simple miniaturization might suggest. While smaller pixels can enable more compact designs in some applications, particularly those with fixed FOV lenses, they often lead to increased complexity and system costs with continuous zoom capabilities.

The most effective approach is to evaluate carefully the specific requirements of each application and select a pixel-size that optimizes the balance between size, weight, power, cost, and performance. This ensures that thermal imaging systems deliver the best possible combination of practical utility and operational effectiveness for its intended use.

Julie Moreira is principal system design engineer and Sooyong Nam is an optical engineer at Teledyne FLIR.

For more information: https://www.flir.com/oem.