The ultraviolet (UV) range of the electromagnetic spectrum has historically been split into three bands. These three bands — UVA (315 to 400 nm); UVB (280 to 315 nm); and UVC (100 to 280 nm) — corresponded to barium-silex, bariumsilex- Pyrex, and Pyrex filters. All three filter types are widely available. A butterfly is imaged under ultraviolet (left) and visible light. Ultraviolet imaging provides insights that enable the approach’s use in a range of applications — from environmental to industrial. Courtesy of Oxford Instruments — First Light Imaging. As some of the shortest and most energetic wavelengths in the spectrum, UV light at any wavelength exhibits distinct characteristics, which are harnessed across various domains (Figure 1). For example, UVA is valuable for fluorescence imaging, and UVB light can be used for medical treatments, such as psoriasis therapies. The highly energetic UVC band is commonly used for sterilization. Figure 1. A graphical representation of UV CMOS applications by wavelength, with specific examples, as grouped by the distinct UV spectral bands. Courtesy of Oxford Instruments — First Light Imaging. Accordingly, UV-emitting sources are already used in a wide range of applications in both industry and R&D. Now, the prospect of using UV-sensitive sensors to record data over time and store the data for subsequent analysis is increasing in interest. However, most CMOS sensors for visible light imaging are fabricated from silicon. This material exhibits low transmission efficiency in the UV range. To optimize transmittance, all elements within the camera’s optical path, including the microlenses integrated into the CMOS sensor and the protective window, must be meticulously engineered and treated for UV (Figure 2a,b). The shorter UV wavelengths, and significantly lower penetration depth compared with visible light, mean that photons are less prone to absorption by the photodiodes within the sensor. Also, the quantum efficiency is typically lower than for visible light. Figure 2. Design architectures of back-side- and front-side-illuminated UV CMOS image sensors (a, b). A gallium nitride (GaN)-based CMOS image sensor, isolating material layers (c). Courtesy of Oxford Instruments — First Light Imaging. Manufacturers have worked to overcome this constraint through the development of image sensors capable of accommodating UV cameras and UV imaging more broadly. In the early days of working to meet this challenge, CCD sensors became more sought after due to their higher sensitivity and reduced noise. CCD sensors transfer the generated charges from an array of photodiodes after photons hit pixels, leading to lower noise. However, CCD sensors come with limitations and known artifacts, such as blooming. During the last few decades, CMOS technology has undergone significant improvements. Owing to advancements in lithography, these gains are most noticeable in the sensors’ ability to handle noise. The resulting image quality has approached levels comparable to that which CCD sensors can deliver. In addition to image quality, CMOS sensors, now operating in or for the UV range, offer an effective alternative to their CCD counterparts. Among other benefits, they are cost-effective, consume ~100× less power than CCDs, and offer higher operational speeds1. Technical and material innovations Adapting CMOS technology for the UV range requires developers to consider the high energy of UV photons and the corresponding effects of this characteristic on sensor design. With UV photons, penetration depth — or the distance at which light intensity decreases as it moves through the sensor — is low. CMOS manufacturers have explored various approaches to address the low penetration depth. In 2011, Sony released its first back-side-illuminated CMOS sensor. Compared with the front-side-illuminated sensors, which represented the original approach to building sensors, the photosensitive matrix in the Sony design was positioned closer to the surface. This enabled better light collection, including for the shorter UV wavelengths (Figure 2a). In a similar vein, fabricators have turned to thinning silicon substrates to enhance UV photon collection. Further, certain designs apply antireflective coating to maximize the transmission of photons; indeed, UV light interacts strongly with the sensors’ surface, which makes it more prone to noise. Algorithms have been implemented in several recent cases to mitigate this. The effect of correlated double sampling, for example, reduces noise by subtracting the background noise on the pixel level. The latest iteration of CMOS sensors features an on-chip digital signal processing mechanism that reduces noise more rapidly, making these sensors ideal for fast imaging applications. UV-sensitive sensors are available on the market, and research is ongoing to improve their durability and quantum efficiency. As these innovations have taken off, various materials have been under investigation. Materials that have a large bandgap due to the high energy of UV photons and effectively transmit UV light have gained preference for UV imaging and sensing. Photodetectors made of various materials are currently under development as well, though progress must be made before multi-pixel and CMOS integration is achieved. Among the most popular of these options, nitride alloys — particularly of gallium, aluminum, and indium-aluminum — offer large bandgaps. At the same time, oxides, such as zinc and/or magnesium-zinc, also respond well to UV light, but in some applications, customers may need to select a specific range in the UV. For instance, ternary zinc germanium oxide responds well to UVC but not UVA or UVB. For high-energy imaging applications (imaging alpha particles, neutrons, gamma rays, and more) diamond and silicon carbide (SiC)-based sensors are strong candidates due to their wide bandgap and resistance in harsh environments2. Operating in or under UV light is challenging due to its high energy, and UV CMOS sensors, which are usually under prolonged UV radiation, can undergo reduced performance over time, or experience irreversible damage, in the worst cases. Although the fabrication of multi-pixel sensors combining CMOS technology with large bandgap materials is still complex and costly, sensors based on SiC are emerging for such applications. Applications: Industrial and R&D UV CMOS technology currently enables applications that are firmly established — though, in many cases, still improving — as well as those in basic and applied research. Among the most notable applications, UV CMOS sensors help to inspect damaged high-tension electric cables by detecting corona discharges that ionize the air and emit UV photons. UV light also enables the discernment of different types of plastics. And most body fluids strongly absorb in the UV, making UV sensors a powerful resource in disciplines such as forensics. As much as technological advancements have positioned UV CMOS to aid numerous imaging applications, recent improvements in manufacturing are likely to lead to many more advancements. There is an increasing demand in the industrial sector for automation systems to handle repetitive and tedious tasks for high-throughput productions. Humanbased quality checks can be subjective, as well as time-intensive, whereas automated systems support unbiased, reliable, and contactless methods. Under this umbrella of inspection tasks, nondestructive testing refers to techniques that check properties of a material without contact. This method is essential to industries such as automotive, aerospace, and electronics. UV CMOS sensors are excellent candidates for nondestructive testing deployments due to their ease of integration, versatility, and facilitation of high-speed, high-resolution imaging. By combining rapid detection with machine vision algorithms, these systems enable efficient materials sorting, automatic inspection, process control, and countless other possibilities. In the semiconductor industry, electronic components are miniaturized to unprecedented sizes. Identifying fine cracks, defects, and dust contamination on wafers is crucial, because these factors often determine whether a device will function as intended. UV CMOS technology excels in providing detailed and fast imaging, though UV CMOS technology cannot match the detail of electron-based microscopies, due to the diffraction limit of light’s shorter wavelengths, which reveal minute details in the blink of an eye. But, electron-based microscopies can complement UV CMOS to provide deeper insights into specific wafer areas. UV CMOS in the field The application benefits of UV CMOS can apply to areas outside of industrial settings. In agriculture, UV CMOS sensors mounted on drones have the potential to identify fruit flowers for artificial pollination, which can provide a path to improved yields. In surveillance applications, cameras similar to the infrared imagers already in use could be used to detect hazardous or biological agents. And with the growing interest in artificial intelligence, large volumes of images generated by high-speed recording CMOS sensors can be used to train models that could ultimately be used to assist humans, resulting in rapid analysis and decisionmaking in real time. In astrophysics, the potential of UV light has always garnered high interest. But this has remained underexplored, mainly for technological bottlenecks. It is also worth noting that absorption and diffusion of UV light by the ozone layer makes ground-based observations complex, and these necessitate the deployment of instruments in outer space. UV CMOS sensors, which are more cost-effective than CCD options, consume less power and can more effectively withstand harsh environments, such as UV radiations. With recent advancements in UV CMOS technology, astrophysicists are aiming to uncover critical information about high-energy processes, such as stellar formation, hot sources, and interstellar media. This technology allows scientists to trace stellar nurseries and gain a better understanding of the early stages of star formation. Additionally, UV CMOS sensors can be mounted on a satellite and seize the atmosphere composition of exoplanets. Gases that have strong absorption in the UV spectrum, such as helium, hydrogen, and even ozone, can indicate whether life-forms are present. For example, UV CMOS technology was integrated on a satellite called the Ultraviolet Transient Astronomy Satellite (ULTRASAT). This wide-field Schmidt telescope in orbit aims to deliver a comprehensive understanding of the high-energy transient universe. It refers to high-energy events that take place in a brief timescale in the cosmos. ULTRASAT excels in the near-ultraviolet range from 230 to 290 nm. Its observations are expected to contribute to multiple fields, including the study of supernovae and neutron stars, galaxies, and gravitational waves. Scheduled for launch in 2027, ULTRASAT uses UV CMOS technology to capture high-resolution images in real time. Deutsches Elektronen-Synchrotron (DESY) developed the camera, which features a focal plane array with four independent 45- × 45-sq-mm back-sideilluminated CMOS sensors and a pixel size of 9.5 × 9.5 μm. To address the low penetration depth of UV, DESY designed a sensor with a proper antireflective coating, called T2. This coating has a high quantum efficiency in the near-ultraviolet range for the target application. Gas spectroscopy On Earth, gas detection, particularly of noxious gases, is also a significant focus. Scientists in the last few decades have gained a heightened awareness regarding global warming, often highlighting ozone. This gas protects the planet from harmful radiation, and it exhibits a strong absorption in the deep-UV (2), from fossil fuels; and benzene, from residential wood-heating, are gases that each absorb in UV and are therefore detected and quantified using UV CMOS sensors. Looking ahead Despite significant advancements in UV CMOS technology, certain challenges remain, limiting the full potential of these sensors for some applications. The main issue: UV CMOS is relatively low in terms of sensitivity and quantum efficiency compared with visible light sensors. Although thinner material layers enhance UV transmission, there is still room for improvement. Overcoming these obstacles and improving the sensor fabrication processes will expand the scope of UV CMOS sensors and enhance their overall efficiency, durability, and performance in harsh environments. It is fortunate, in this sense, that CMOS technology has been around for decades and continues to benefit from a wellestablished road map, even if only for the visible and infrared ranges. UV CMOS opens its own spectrum of possibilities — but it does not exist in a vacuum. With continuous improvements, UV CMOS sensors are pushing limits and signaling a new era for the CMOS industry. UV CMOS: Application in Action In an application example, attempts were made to detect sulfur dioxide (SO2) emission out of a refinery’s chimney. A CB1 UV camera from Oxford Instruments — First Light Imaging, with an f/2.8 100-mm UV sensitive objective (CERCO) was used for the test. This objective is apochromatic and demonstrates excellent transmission between 240 and 900 nm. A UV (310-nm ± 5-nm) bandpass filter was used to select the wavelength where the absorption of UV by SO2 is optimal. It is well known that the processing of crude oil involves combustion, and the sulfur contained inside the oil is rejected by the chimney after oxidation as SO2. While emission is not detected with visible light, results enabled the observation (black) of the presence of SO2 (Figure 3). This was possible due to the 310-nm (± 5-nm) bandpass filter used in the test. This setup allows for imaging of emitted gases from a certain distance, ~10 s of m. Figure 3. A refinery chimney emitting sulfur dioxide (SO2) imaged in ultraviolet (top) and visible light (bottom). Courtesy of Oxford Instruments — First Light Imaging. Meet the author François Yaya, Ph.D., is a product manager at Oxford Instruments — First Light Imaging, specializing in high-performance cameras for scientific research and industrial applications. He collaborates with scientists from a variety of research areas for solutions development; email: ">francois.yaya@oxinst.com. References 1. M. Bigas et al. (2006). Review of CMOS image sensors. Microelectronics J, Vol. 37, No. 5, pp. 433-451. 2. N.G. Wright et al. (2008). Prospects for SiC electronics and sensors. Mater Today, Vol. 11, No. 1-2, pp. 16-21.