The Science of Far-UVC

1. Germicidal UV: Introduction and History

Invisible Colors

The discovery of ultraviolet (UV) light dates back to the early 19th century when Johann Wilhelm Ritter first detected an invisible form of radiation beyond the violet end of the visible spectrum in 1801. He observed that this radiation caused silver chloride-coated paper to darken faster than visible light, indicating a high-energy, chemically reactive form of light, which he initially called ‘deoxidizing rays’ before the term ‘ultraviolet’ was later adopted.

Much like visible light, UV contains different ‘colors’, and despite being invisible to the human eye, these different colors are just as distinct as red and blue (Figure 1.1). What we are familiar with as ‘blacklights’, widely used for everything from zapping mosquitoes to illuminating nightclubs, are actually part of a set of invisible colors we call ‘UVA’. By contrast, it is the invisible colors we call ‘UVB’ that are believed to be the primary cause of most skin cancers. ‘UVC’ is the name of another set of invisible colors further away from visible light in the UV spectrum, of which far-UVC is a subset.

The Discovery of UV’s Germicidal Properties

By the late 19th century, researchers began exploring both the physical and biological effects of light. In 1877, Arthur Downes and Thomas P. Blunt made a key observation when they demonstrated that sunlight—which contains UVA and UVB—could inhibit bacterial growth. They exposed test tubes containing bacteria to direct sunlight and observed that microbial growth was significantly reduced or entirely prevented. They further determined that the germicidal effect depended on the intensity and duration of exposure, with shorter wavelengths of sunlight proving most effective. In 1890, Robert Koch demonstrated the lethal effect of sunlight on Mycobacterium tuberculosis, hinting at UV’s potential for combating diseases like tuberculosis. These discoveries laid the foundation for our understanding of the bactericidal effects of light, including what would later be recognized as the ultraviolet section of the electromagnetic radiation spectrum.

Building on this knowledge, Niels Ryberg Finsen became one of the first to harness UV for medical applications. In the late 19th century, he pioneered the use of concentrated UV therapy to treat skin tuberculosis (lupus vulgaris), a chronic infection caused by Mycobacterium tuberculosis. Finsen’s therapy, which earned him the 1903 Nobel Prize in Medicine, relied on carbon arc lamps and quartz lenses to generate and focus UV light, leading to documented clinical success.

At the same time, laboratory studies were beginning to show that UV had direct bactericidal properties. Early microbiologists such as Valdemar Bie and Sofus Bang, working under Finsen, reported that wavelengths below 300 nm were particularly effective at inactivating bacteria. By 1903, Bang had demonstrated that concentrated UV from an arc lamp could kill Mycobacterium tuberculosis within minutes, providing some of the earliest direct evidence of UV’s germicidal effects.

The early 20th century saw further refinements in understanding UV’s germicidal potential. Frederick L. Gates published the first precise bactericidal ‘action spectrum’—a graph showing the efficacies of different UV wavelengths at killing bacteria. Gates demonstrated that UV wavelengths around 265 nm (within the UVC section of the UV spectrum) were the most effective for inactivating bacteria. He also tested far-UVC (225 nm) and found it to be bactericidal, but technical limitations at the time, including the low power of UV sources available at that wavelength, made it difficult for him to fully evaluate shorter wavelengths and establish a complete action spectrum.

Advancements in mercury-vapor lamp technology provided a new means of generating UV, and in particular UVC. The first mercury-vapor lamp to achieve widespread success was invented in 1901 by American engineer Peter Cooper Hewitt. His initial design, while effective, produced a bluish-green light that limited its applications. In 1903, Hewitt introduced an improved version with enhanced color qualities, making it more suitable for industrial use. Beyond illumination, the UV emitted by mercury-vapor lamps was soon recognized for its potential in disinfection. By 1910, the technology was being applied to water treatment, marking the beginning of UV-based sterilization methods that would later expand to air and surface disinfection. While Finsen carbon arc lamps and quartz lenses created mostly UVA and visible light, these lamps were able to create UVC, with a peak at 254 nm, much closer to the wavelength that Gates would show to be the most effective at inactivating bacteria.

In the 1930s, physician Mildred Weeks Wells and engineer William Firth Wells began investigating how respiratory diseases spread. The husband-wife research team brought complementary expertise: Mildred’s knowledge of infectious diseases and their transmission paired well with William’s background in air and water quality engineering. Through careful experimentation and mathematical modeling, they showed that when people cough, sneeze, or breathe, they emit droplets across a spectrum of sizes. Larger droplets quickly fell to the ground, while smaller ones could evaporate before settling, leaving behind ‘droplet nuclei’ that could float in the air for extended periods. This size-based behavior, later known as the Wells curve, explained why some diseases could spread through the air over considerable distances—a mechanism that differed from the dominant theory of direct droplet spread.

As the Wells developed their theoretical understanding of airborne disease transmission, hospitals began testing UV for infection control. In 1936, Deryl Hart at Duke University Hospital sought to address persistent post-operative infections that conventional sterilization techniques had failed to prevent. Initially, Hart’s team experimented with therapy UV lamps and carbon arc lamps, but these proved ineffective at reducing bacterial contamination in the operating room.

Seeking a more effective solution, they collaborated with Westinghouse Lamp Company, which supplied specially designed low-pressure mercury-vapor lamps optimized to emit germicidal UVC (commonly referred to as GUV) at 254 nm. The installation of these high-intensity UVC lamps dramatically reduced airborne bacteria, and their early studies showed a complete elimination of infections in UV-treated thoracoplasty cases, compared to a 33% infection rate in untreated cases. However, the use of UVC required strict safety measures—surgical staff had to wear goggles, tightly woven cloth hoods, and even sun helmets to protect their skin and eyes from overexposure.

Over time, the effectiveness of UVC disinfection became even more evident. A 1960 study by Hart, analyzing over 4,500 operations, found that UVC reduced post-operative wound infection rates in clean surgeries from 11.62% to just 0.24%, reinforcing its value as a powerful tool for infection control. The results were so compelling that other hospitals quickly adopted similar systems, and Duke University Hospital continues to use UVC disinfection in its surgical suites to this day.

These results also led hospitals to explore UV applications in other settings. At Boston’s Infants’ and Children’s Hospital, UV barriers created cubicle-like divisions designed to prevent respiratory cross-infections between patients. Del Mundo and McKhann reported infection rates of 12.5% in control wards compared to 2.7% in wards with UV barriers. Other hospitals documented similar reductions using both UV barriers and upper-room installations, where UV lamps irradiated only the air above people’s heads. This design allowed warm air currents, generated by people’s body heat, and natural air mixing to carry exhaled pathogens into the upper UV-irradiated zone, where they were rapidly inactivated before they could be inhaled by others.

The Wells saw an opportunity to combine their understanding of airborne transmission with UV technology. In 1937, they launched a controlled study at Germantown Friends School near Philadelphia, where measles regularly swept through classrooms despite strict surface cleaning and student separation protocols. Their controlled studies compared classrooms with UV lamps installed near the ceiling, irradiating the upper unoccupied section of the room, to identical untreated rooms. In the UV-equipped classrooms, measles cases dropped by 13.3% compared to control rooms, even as cases continued spreading in nearby districts.

By 1943, the US Navy launched a systematic four-year study at its training centers, where respiratory diseases posed a serious problem among recruits. The Navy installed GUV systems in alternate barracks housing over 5,000 recruits, documenting all sick bay admissions while monitoring bacterial levels in the air. In the first year at Sampson Naval Training Center, high-intensity UV systems reduced respiratory infections by 25%, though low-intensity systems showed no effect.

However, the medical and public health community was resistant to the idea that pathogens could spread through the air. This resistance was rooted in a paradigm shift led by epidemiologist Charles Chapin in the early 1900s. For much of history, the dominant belief was miasma theory—the idea that diseases were caused by exposure to ‘bad air’ from decaying matter. This view was gradually overturned in the 19th century by germ theory, which demonstrated that specific diseases were caused by microscopic organisms, not miasma. Chapin, a strong proponent of vaccination, pandemic response, and germ theory, argued that diseases primarily spread through direct contact and large droplets. His work helped advance public health and infection control through hygiene, isolation, and sanitation. However, the shift went too far, and the medical community became skeptical that any airborne transmission of diseases existed, despite emerging evidence to the contrary.

The most rigorous evidence for UV’s effectiveness against airborne infection came from studies designed by William Wells and carried out by Richard Riley at a Veterans Administration Hospital TB ward between 1954 and 1961. The ward’s ventilation system exhausted air from TB patient rooms through chambers housing guinea pigs. Some chambers received UV-treated air, while others received untreated air. Over the two-year period, while many guinea pigs breathing untreated ward air developed TB infections, none of the animals breathing UV-irradiated air were infected. Additional compelling evidence came during the 1957 Asian influenza pandemic, when McLean and colleagues observed infection rates of only 1.9% in UV-irradiated wards compared to 18.9% in non-irradiated wards, demonstrating UV’s effectiveness against viral as well as bacterial pathogens.

This research shifted scientific consensus and ultimately led to the widespread acknowledgment of airborne TB transmission. GUV was adopted as a TB control measure in high-risk settings. However, many experts remained reluctant to generalize these findings to other respiratory diseases. Despite its successes, interest in GUV air disinfection waned. Antibiotics revolutionized TB treatment, and a wave of new vaccines targeted diseases like diphtheria, polio, influenza, and even TB itself. Public health officials saw these advances as the future of disease control and largely dismissed GUV as unnecessary, even for diseases accepted to have airborne transmission. Enthusiasm was further dampened by concerns about potential health effects from UV exposure and the ozone produced by early lamp designs. Additionally, follow-up studies failed to replicate the Wells’ dramatic reductions in disease transmission in classrooms. These studies suggested that continued transmission outside of classrooms, particularly in school buses and other communal spaces that were not protected by UV, limited the effectiveness of local UV interventions.

The technology would not see widespread revival until the late 1980s, when an unexpected rise in TB cases and the emergence of drug-resistant strains renewed interest in environmental controls for airborne infection. Riley and colleagues conducted detailed studies of UV air disinfection in model rooms. These experiments revealed critical factors affecting performance: air mixing between the lower occupied space and upper UV-treated zone proved essential, with temperature gradients and ceiling fans significantly impacting this mixing. Their work also showed that high relative humidity reduced UV’s effectiveness, with sharp declines in pathogen kill rates above 60–70% humidity.

Building on the Wells’ earlier work, Riley developed a model to quantify the probability of airborne infection in an indoor space. The model, commonly called the Wells-Riley model, estimates infection risk based on factors such as the number of infectious individuals, room ventilation, duration of exposure, and the pathogen’s transmission characteristics. It remains a foundational tool for understanding airborne disease spread and evaluating mitigation strategies, including the effectiveness of UV air disinfection.

As the limitations of conventional TB control measures became clear, renewed studies on GUV demonstrated its ability to supplement other infection control strategies. Studies revisiting upper-room UV systems confirmed their efficacy in high-risk settings, such as hospitals, operating rooms, homeless shelters, and correctional facilities, led in large part by Riley and Ed Nardell, a physician at Brigham and Women’s Hospital.

This period saw a shift in institutional support, with the US Centers for Disease Control and Prevention (CDC) incorporating GUV into its guidelines for TB control in healthcare settings, marking a turning point for the technology’s acceptance. Results of Escombe et al., 2009 confirmed that upper-room GUV radically reduced the amount of infectious TB present in the air.

Additional research throughout the 1990s refined GUV application, focusing on optimizing fixture design, air circulation, and safety measures to minimize UV exposure risks. Computational fluid dynamics models allowed researchers to predict airflow patterns and ensure effective disinfection while protecting room occupants. These advancements reinforced GUV as a viable intervention, particularly where mechanical ventilation improvements were cost-prohibitive. But outside of the specific use case of spaces that are high risk for TB, there was historically relatively little interest in the use of upper-room UV. To this day, the official guidance on the use of upper-room UV provided by CDC/NIOSH is explicitly for its use in controlling the spread of tuberculosis.

The COVID-19 Pandemic, and the Emergence of Far-UVC

As described above, for much of the 20th and early 21st century there was skepticism that respiratory illnesses are spread by airborne transmission. During the COVID-19 pandemic, this led to slower adoption of measures widely accepted to be effective against airborne transmission of disease, including upper-room UV, ventilation, and filtration of indoor air.

Before the COVID-19 pandemic, Professor David Brenner, Director of the Center for Radiological Research (CRR) at Columbia University, was investigating shorter wavelengths of UVC for disinfection in surgical settings. The aim was to identify new sources of UVC that could effectively kill pathogens without the risks of overexposure associated with conventional germicidal UV from low-pressure mercury vapor lamps. Brenner and his colleagues, especially Manuela Buonanno and David Welch, began exploring the potential of far-UVC, a subset of the UVC spectrum. They hypothesized that these shorter UVC wavelengths could inactivate microbes without penetrating human skin or eyes, due to their strong absorption by proteins.

Initial studies demonstrated that far-UVC could efficiently inactivate drug-resistant bacteria without damaging more sensitive human tissues. Early studies used krypton-bromine excimer (KrBr*) lamps emitting primarily at 207 nm, which effectively inactivated bacteria while being less harmful to human skin compared to 254-nm germicidal UVC. However, krypton chloride excimer (KrCl*) lamps, emitting primarily at 222 nm, became more widely available and offered a higher output intensity, making them more practical for real-world applications.

In 2018, researchers at the CRR demonstrated the first proof of concept that far-UVC could effectively inactivate viruses that cause respiratory illness, showing over 95% inactivation of aerosolized H1N1 influenza virus with low doses of 222-nm far-UVC. This established the potential for the application of ‘whole-room’ far-UVC irradiation in indoor public locations.

When COVID-19 emerged in 2020, research rapidly expanded to examine both efficacy and safety. Studies at the CRR demonstrated far-UVC’s effectiveness against coronaviruses, while Ewan Eadie and colleagues at the photobiology unit at Ninewells Hospital at the University of Dundee showed that properly filtered far-UVC produced no skin erythema (sunburn) even at a dose more than 500 times higher than the consensus exposure limits at the time. The research conducted at Ninewells confirmed that filtering out longer UV wavelengths from far-UVC emitters was important—unfiltered sources could cause erythema, while properly filtered devices showed no acute effects.

In 2022, the first study of far-UVC at room-sized scale was published, showing remarkable efficacy against airborne bacteria.

However, despite the technology’s promise, widespread adoption of far-UVC during the COVID-19 pandemic faced several significant barriers. Early in the pandemic, the medical community had not yet accepted airborne transmission, which meant that prevention efforts focused primarily on surface disinfection and droplet precautions rather than air disinfection technologies. Even after airborne transmission was widely accepted, implementation guidance remained unclear for far-UVC. Unlike upper-room GUV systems, which had decades of established protocols, there was limited practical experience, no authoritative guidance, and no agreed-upon standard for assessing the safety and efficacy of whole-room far-UVC deployment.

In addition, there are not yet any large-scale clinical trials demonstrating far-UVC’s real-world effectiveness at blocking human-to-human transmission. This hinders acceptance of and advocacy for the technology by many important institutions, who are understandably risk-averse at the prospect of increasing the public’s exposure to any form of radiation.

It is also unreasonable to expect laypeople to establish the veracity of manufacturers’ claims about efficacy and safety. Many air cleaning technologies have been offered to the market that are at best ineffective and at worst actively harmful, and there is a need to produce an evidence base that secures the endorsement of safe and effective technologies from trusted institutions.

This document represents our assessment of the state of knowledge of the use of far-UVC to control the transmission of airborne disease. While there is uncertainty, as there is with any technology at this early stage, the potential for far-UVC is supported by quality research. The amount of available funding and the number of researchers, engineers and policymakers engaged with the technology is not commensurate with its tremendous potential.

Further Reading

• The History of Ultraviolet Germicidal Irradiation for Air Disinfection by Nicholas Reed, 2010
• Air-Borne by Carl Zimmer, 2025

2. Far-UVC Primer

This section provides foundational concepts and terminology used throughout the rest of this document. The section ensures all readers have access to the same background knowledge, but readers need not read it all at once. Readers familiar with these concepts are welcome to skip sections, or proceed directly to later sections if comfortable with UV physics and terminology and return to relevant sections when additional context would be helpful.

The Electromagnetic Spectrum

Radiation is energy transmitted through space as a particle or wave. The electromagnetic (EM) spectrum encompasses all energy that is transmitted in the form of electromagnetic waves. The fundamental unit of electromagnetic radiation is a photon.

Visible light is one section of the EM spectrum, although other sections of the EM spectrum are often colloquially referred to as ‘light’. ‘Ultraviolet light’ is one such example of this, and devices that produce ultraviolet (UV) radiation are often called ‘lamps’. In this document, we reserve the term ‘light’ to mean ‘visible light’ for clarity and consistency, although we do refer to ‘lamps’ that produce UV radiation.

Photons have characteristics of both waves and particles, depending on how they are measured (this is known as ‘wave-particle duality’). The energy of a photon with wavelength λ (in meters) is given by:

where h is the Planck constant (6.626×10^-34 Js), and c is the speed of light in a vacuum (2.99×10⁸ ms^-1). Note that photon energy is inversely proportional to the wavelength, so short wavelength X-ray or gamma ray photons transmit more energy than longer radio waves or microwaves. Listed here in ascending order of wavelength, the EM spectrum includes: gamma ray, X-ray, ultraviolet, visible light, infrared, microwaves, and radio waves. Figure 2.1 shows the ultraviolet and visible portion of the spectrum, which is most relevant for understanding far-UVC applications.

Ultraviolet Radiation

Many of the EM bands are well known to the general public, including visible light, microwaves, X-rays, and the focus of this section, ultraviolet (UV) radiation. UV is further divided into partially overlapping subcategories:

UVA (320–400 nm) has the longest wavelength and lowest photon energy within the UV spectrum. UVA is an important component of the sunlight that reaches the earth’s surface. It penetrates more deeply into the skin than other UV subtypes. Excess UVA exposure is a risk factor for erythema (sunburn), cataracts, some forms of skin cancer, and skin aging.

UVB (280–320 nm) is shorter in wavelength, thus higher energy per photon, than UVA. Similar to UVA, it is a component of sunlight and excess exposure is strongly associated with erythema, skin cancer, cataracts, eye surface pathologies (e.g., pterygium and pinguicula), and skin aging. Moreover, it can damage cellular DNA.

UVC (200–280 nm) has shorter wavelengths and higher energy than UVA and UVB. While it is a component of solar radiation, UVC is almost fully absorbed by the Earth’s stratospheric ozone layer. UVC efficiently damages the DNA of microorganisms, such as bacteria and viruses. UVC lamps, and particularly mercury vapor lamps that primarily emit 254-nm UVC, have been used in various germicidal applications for over a century. 254-nm UVC penetrates skin and eyes less efficiently than UVB, although 254 nm exposure limits are not significantly different to 280–300-nm UVB.

Far-UVC (200–235 nm) is a subset of UVC that penetrates human skin and eyes less as compared to longer UVC wavelengths. This results in lower risk of tissue damage and thereby allows for higher human exposure limits (see Skin and eye safety section).

Vacuum ultraviolet (VUV), definitions range from 10 to ~120 nm at the low end to 180–200 nm at the high end, can be considered a subset of UVC, or considered its own section of the EM spectrum. At wavelengths below 200 nm, there is a marked increase in the absorption of photons by O₂ molecules, resulting in significant ozone (O₃) generation. VUV is so named because it can only propagate effectively in a vacuum or in an environment free of oxygen (such as in a controlled gas like argon): in environments with oxygen the photons are quickly absorbed. VUV is not used for disinfection in occupied spaces due to this absorption by oxygen and generation of ozone, but it has niche applications in industrial and laboratory settings including municipal wastewater treatment.

Extreme ultraviolet (EUV, 10–121 nm) can be considered as a separate section, or it can be seen as overlapping with VUV, or as a subset of VUV. It is used in industrial applications, for example in semiconductor manufacturing.

Ionizing and Non-Ionizing Radiation

Radiation is conventionally categorized into ionizing and non-ionizing radiation. This categorization has significant implications for how the risks of radiation exposure are managed, and how they are understood by the general public.

Ionizing radiation consists of EM waves with much higher photon energies than far-UVC, such as extreme UV, X-rays and gamma rays, as well as alpha particles, beta particles and neutrons emitted by atoms undergoing radioactive decay.

Ionizing radiation is so called because it has sufficient energy to remove tightly bound electrons from atoms, creating positively charged ions. These ions can directly interact with and damage essential structures in cells such as DNA, or they can result in free radical production that indirectly damages DNA or other cellular structures. Ionizing radiation exposure is strongly associated with DNA mutation, which can lead to cancer, or at high doses can cause cell death. Because of these risks, exposure to ionizing radiation is carefully regulated and controlled in medical and industrial settings.

EM waves with lower energy levels, such as radio waves, microwaves, infrared, visible light, UVA, UVB, and UVC are classified as non-ionizing radiation. While non-ionizing EM photons do not have enough energy to ionize most atoms, they can still cause biological effects, for example through heating or photochemical reactions.

While categorized as non-ionizing radiation, far-UVC photons do have sufficient energy to ionize some atoms. For example, a 222-nm far-UVC photon has 5.58 electron volts (eV) of energy, which is sufficient to ionize alkali metals, and photons in the UVB range are capable of ionizing caesium. In this sense, far-UVC radiation can be ‘ionizing’ by a literal definition of the term. However, due to its inability to ionize atoms typically found in biological systems, far-UVC is classified as non-ionizing radiation.

Key Definitions

Irradiance and fluence rate both measure the amount of radiant power over a small area. As such, they are typically used to measure how much far-UVC radiation a person or space is exposed to. The difference is illustrated in Figure 2.2 below—fluence rate encompasses radiant power emitted onto the area from all directions, whereas irradiance includes only radiant power from one direction above a surface (for example, from the sun to the surface of the earth).

For UV disinfection of air and water, fluence rate is generally the applicable concept. For human skin and eye exposure, and for disinfection of surfaces, the applicable concept is generally irradiance.

In the most general terms, energy is the integral of power over time. Thus, fluence is the integral of fluence rate over time. In the simplest case, exposure to a uniform fluence rate of 1 W/m² means that every second the fluence delivered is 1 J/m².

Research and guidance on the use and safety of germicidal UV often refers to the concept of dose. In the context of UV air disinfection, dose is based on fluence rate and has the same definition as fluence or radiant exposure. In the context of assessing UV exposure to human skin and eyes, dose is based on irradiance.

Table 2.1. Key Units and Definitions

Angular Dependence

An important concept in electromagnetic radiation is angular dependence, which we experience in our everyday exposure to solar UV. Radiation has direction, and when the angle of incidence is perpendicular to the surface of an object the radiant power is spread over the smallest possible surface area. As the angle of incidence departs from the perpendicular, the same radiant power is spread over a larger surface area, so each part of that surface receives less radiation.

Angular dependence is mathematically described by Lambert’s cosine law, meaning radiant power is proportional to the cosine of the incident angle on the surface.

Inverse Square Law

The inverse square law describes the relationship between fluence rate (or irradiance) and distance from an ideal point source in a system where no UV-absorbing compounds are present. An ideal point source is one that is assumed to emit photons uniformly in all directions. For this source type, fluence rate will decrease with the square of distance from the source. In other words, when an object’s distance from the radiation source is doubled, the fluence rate reduces to ¼, and so forth as illustrated in Figure 2.4.

Because air is a weak absorber of UVC radiation, the inverse square law provides a rough understanding of the spatial distribution of fluence rate as related to UVC source location, even though real UVC sources are not ideal point sources. The practical import of the inverse square law is that fluence rates are significantly higher closer to the source, so that, for example, the areas directly next to a UVC lamp will have much higher fluence rates than those further away.

Chromophores

A chromophore is any molecule, or part of a molecule, that absorbs a specific wavelength of electromagnetic radiation. For example, O₂ is a chromophore of UV radiation of wavelengths less than 242 nm, whereas food colorings contain chromophores of visible light. An important chromophore for far-UVC is the peptide bonds that link amino acids to form proteins, and there are many UV chromophores in commonly used materials (see UVC impact on materials section).

Action Spectra

An action spectrum describes the relative effectiveness of different wavelengths of radiation at producing a specific biological effect, such as DNA damage, pathogen inactivation, or sunburn. In the context of UV radiation, the action spectrum is crucial, because various wavelengths are differentially absorbed by tissue, cells, and cellular components such as DNA, proteins, and lipids. An action spectrum is determined through experimental studies that measure a biological response to various wavelengths of UV radiation, ideally using tunable, nearly monochromatic sources of radiation with a single wavelength.

Action spectra help us understand how different wavelengths of radiation affect biological systems, including human skin, cell types, or tissues, as well as microbial pathogens. For example, Figures 2.5 and 2.6 show the action spectra for the propensity of proteins and nucleic acids to absorb photons of different wavelengths. Higher relative absorption means that it is more likely that a photon will interact with a molecule and produce a photochemical and/or photobiological effect.

The protein and nucleic acid absorbance spectra show that absorption properties can be highly nonlinear, and are not simply a function of wavelength and photon energy. When we then consider complex phenomena like skin cancer, the action spectra can be non-intuitive. For example, the action spectrum for non-melanoma skin cancer peaks sharply in the UVB range.

The cancer risk of UVB is orders of magnitude higher than that of UVC and UVA. This is due to the interaction of certain properties of UVB (penetration depth, propensity to damage DNA) with properties of the body (skin structure and the photon absorption properties of nucleic acids and proteins) (see Evaluating cancer risk section).

It is therefore not the case that we can use simple heuristics like “higher energy photons are always more dangerous,” or “lower energy photons are less effective at killing pathogens.” It depends on a combination of complex photochemical properties of particular macromolecules in specific cells, tissues, and complex biological systems.

UV Exposure Limits

Exposure limits for UV radiation are based on action spectra. These action spectra are provided by organizations such as the American Conference of Governmental Industrial Hygienists (ACGIH) and the International Commission on Non-Ionizing Radiation Protection (ICNIRP). These action spectra were originally established in the 1970s, although ACGIH updated theirs in 2022 based on recent research into far-UVC skin and eye safety.

The ACGIH exposure limits are called Threshold Limit Values (TLVs). ICNIRP exposure limits are sometimes colloquially referred to as ‘TLVs’, but this is a trademarked term by ACGIH for all of their recommendations across industrial hygiene, not just UV exposure limits.

Both ICNIRP and ACGIH recommend a spectrally weighted UV exposure limit of 3 mJ/cm² over 8 hours in a 24-hour period. Each wavelength is assigned a spectral effectiveness value S(λ) that reflects its relative biological impact. The actual unweighted UV dose at each wavelength is multiplied by its S(λ) value, and these weighted doses are summed to calculate the total spectrally weighted dose. In practice, this means that wavelengths considered to be more hazardous count more towards the exposure limit. This total spectrally weighted dose must remain below 3 mJ/cm² to stay within the exposure limit.

As with other action spectra, S(λ) is conventionally presented in a normalized form such that the maximum value (1) is observed at the wavelength that has the greatest potential to cause the relevant effect. For example, the ACGIH Eye and ICNIRP values of S(λ) are based on the photokeratitis (inflammation of the cornea) action spectrum. A higher S(λ) means that UV at that wavelength contributes relatively more towards the 3 mJ/cm² weighted exposure limit.

As can be seen from Figures 2.7 and 2.8, S(λ) for different UVC wavelengths ranges across more than 1.5 orders of magnitude according to ICNIRP, and more than 2.5 orders of magnitude according to ACGIH. Some UVC wavelengths therefore contribute dramatically more towards the 3 mJ/cm² limit than others, with far-UVC wavelengths contributing the least.

This dynamic can be illustrated in the exposure limits for the most widely-used far-UVC source, a krypton chloride excimer (KrCl*) lamp. These lamps produce predominantly, but not exclusively, UVC with a wavelength of 222 nm. For some optically-filtered KrCl* lamps, the exposure limit may be close to a hypothetical source that emitted only 222 nm. However, for others—especially unfiltered or poorly filtered lamps—the exposure limit can vary significantly, due to the fact that they also emit higher wavelengths with a much larger S(λ).

Though these unweighted emission spectra appear similar on a linear scale, even a small emission at, for example, 259 nm in the case of the unfiltered lamp has a profound effect on the spectrally weighted exposure. This is because relative spectral effectiveness of 259 nm, according to the ACGIH Eye S(λ) as displayed in Figure 2.7, is over 50 times higher than that of 222 nm, and therefore exposure to the same dose of 259-nm radiation counts 50 times more towards an individual’s exposure limit. If a person was subject to the same irradiance from each of these two lamps, they would exceed exposure limits for the unfiltered lamp 3 times faster than for the filtered lamp.

While far-UVC has lower spectral effectiveness than higher wavelength UVC, and therefore higher safe exposure limits, it is nevertheless still crucial to adhere to these exposure limits. It is remaining within exposure limits that makes any form of electromagnetic radiation safe, rather than any wavelength being universally safe or unsafe.

What makes far-UVC promising is that studies have demonstrated significant pathogen inactivation while remaining well within human exposure limits. This means a dose that is safe for human exposure can still effectively reduce airborne infectious particles (explored further in the Efficacy section).

Germicidal UV Applications

There are three primary strategies for deploying germicidal ultraviolet radiation (GUV) which are outlined in Figure 2.11. These are upper-room GUV, in-duct GUV, and whole-room GUV.

Upper-room GUV fixtures are mounted high on walls or ceilings. They direct UVC radiation into the upper portion of the room, away from occupants. Air circulation within the room—generated from natural convection, HVAC, or fans—moves air between the lower occupied space and the irradiated zone in the upper room. Upper-room GUV must be carefully designed and installed by professionals to ensure safety for occupants.

Whole-room GUV systems disperse UV throughout the entire space, providing continuous disinfection of air and surfaces in occupied areas. Far-UVC is the only form of UV thought to be viable for whole-room disinfection of occupied spaces, due to the higher exposure limits recommended by ACGIH and ICNIRP.

In-duct GUV systems are installed within HVAC systems to disinfect recirculating air, thus reducing the concentration of airborne pathogens in a room through dilution. In-duct systems are conceptually similar to outdoor air ventilation and filtration of recirculating air.

Effectiveness, Efficacy, and Susceptibility

Finally, there are three different concepts that need to be distinguished when discussing how well far-UVC works.

Effectiveness refers to how well far-UVC reduces disease transmission in real-world settings. For example, if far-UVC were installed in a school classroom or an urgent care waiting room, how many illnesses would it prevent?

Efficacy refers to how well far-UVC inactivates pathogens under controlled conditions. This is typically measured as a reduction (or log reduction) in pathogen concentration compared to controls. In the context of air disinfection it can be expressed in terms of equivalent air changes per hour (eACH) or clean air delivery rate (CADR).

Susceptibility, typically expressed as a k or z value, describes how sensitive a particular pathogen is to far-UVC exposure. It is measured by the pathogen’s dose-response relationship to UV and can vary significantly between different pathogens and in different experimental conditions.

For more discussion on these topics, see the Efficacy section.

Further Reading

• Far UV-C Radiation: Current State-of Knowledge
• Ultraviolet Germicidal Irradiation Handbook: UVGI for Air and Surface Disinfection
• Far UV-C radiation: An emerging tool for pandemic control

References

1. International Ultraviolet Association. Far UV-C Radiation: Current State of Knowledge.
2. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Solar and Ultraviolet Radiation. In Radiation (International Agency for Research on Cancer, 2012).
3. Görlitz, M. et al. Assessing the safety of new germicidal far-UVC technologies. Photochem. Photobiol. 100, 501–520 (2024).
4. Blatchley, E. R. et al. Far UV-C radiation: An emerging tool for pandemic control. Crit. Rev. Environ. Sci. Technol. 53, 733–753 (2023).
5. Kowalski, W. J. Ultraviolet Germicidal Irradiation Handbook: UVGI for Air and Surface Disinfection. (Springer-Verlag, Heidelberg; New York, 2009).
6. Mavragani, I. V., Nikitaki, Z., Kalospyros, S. A. & Georgakilas, A. G. Ionizing radiation and complex DNA damage: From prediction to detection challenges and biological significance. Cancers (Basel) 11, 1789 (2019).
7. Hall, E. J. & Giaccia, A. J. Radiobiology for the Radiologist. (Wolters Kluwer Health, Baltimore, MD, 2018).
8. CRC Handbook of Chemistry and Physics, 84th Edition (National Institute of Standards and Technology). (American Chemical Society (ACS), CRC Press LLC: Boca Raton, 2004).
9. Gold, V., Loening, K. L., McNaught, A. D., & Shemi, P. The IUPAC Compendium of Chemical Terminology. (International Union of Pure and Applied Chemistry (IUPAC), Research Triangle Park, NC, 2025).
10. Wikipedia contributors. Oblique rays 02 Pengo.svg.
11. Inverse square law.svg — Wikimedia Commons.
12. Goldfarb, A. R. & Saidel, L. J. Ultraviolet absorption spectra of proteins. Science 114, 156–157 (1951).
13. Setlow, R. & Doyle, B. The action of monochromatic ultraviolet light on proteins. Biochimica Et Biophysica Acta 24, 27–41 (1957).
14. D. Voet, W. B. Gratzer, R. A. Cox, Paul Doty. Absorption spectra of nucleotides, polynucleotides, and nucleic acids in the far ultraviolet. Biopolymers 1, 193–208 (1963).
15. Sliney, D. H. & Stuck, B. E. A need to revise human exposure limits for ultraviolet UV-C radiation. Photochem. Photobiol. 97, 485–492 (2021).
16. OSLUV Assays.

3. Efficacy

Summary

Far-UVC can inactivate a very wide range of pathogens. Some important pathogens (such as coronaviruses) appear to be especially susceptible. But the critical question is not whether far-UVC inactivates pathogens, but how well; in other words, how much far-UVC do we need to render an occupied indoor space safe.

This section provides an analytical framework and explains the evidence for far-UVC’s promise as an air cleaning technology. The most widely quoted study is Eadie et al., 2022, where a realistic use of far-UVC reduced airborne bacteria by 98.4% in a room-sized chamber. This section highlights the wider evidence base and the importance of gaining a better understanding of things we’re currently uncertain about, as well as the implications of these uncertainties on the practical application and overall promise of far-UVC for suppressing airborne transmission.

Clinical trials studying how many airborne infections can be prevented by environmental or ‘place-based’ interventions like far-UVC are difficult to adequately power, due to challenges associated with studying places rather than people and transmission rather than illness. This section explains those challenges and introduces an alternative engineering approach that has been developed to quantify how well air cleaning technologies work, placing the evidence for far-UVC’s efficacy in the context of that engineering framework.

We do not believe that an engineering and modeling approach is a complete substitute for demonstrating effectiveness through clinical trials, and the challenges associated with these need to be overcome. However, the engineering approach is necessary to design effective far-UVC installations in a wide range of environments, as would need to be done for a clinical trial, and also to interpret and extrapolate the results of any clinical trial to different environments.

Crucial Considerations

• From experimental data to date, there is order of magnitude variation in susceptibility to far-UVC of different aerosolized pathogens and in different experiments.
• If the higher estimates of pathogen susceptibility are true, far-UVC may compare favorably to existing technological approaches on cost, energy efficiency, and the total amount of achievable disinfection.
• However, if the lower estimates of pathogen susceptibility are correct, then far-UVC compares less favorably to existing technological approaches such as ventilation and filtration, and practical application will be more challenging, as larger doses of far-UVC would be required to be efficacious.
• Most research into far-UVC germicidal efficacy to date has been conducted in pathogens aerosolized in media that are importantly different to human respiratory aerosols. In particular, we’d expect the protein content of human respiratory aerosols to attenuate the efficacy of far-UVC to some degree relative to a water-based medium, but the extent of this is currently unknown.
• Human-to-animal studies are one option with historical precedent for studying efficacy of far-UVC against pathogens contained in respiratory aerosols.
• Assumptions about air mixing and the impact of ventilation regimes are likely to affect how we translate experimental findings to practical applications, and more research is needed in this area.
• While this report is primarily focused on long-range airborne disease transmission, if far-UVC efficacy for short-range transmission proves favorable, the value of far-UVC potentially increases significantly.
• Resolving these uncertainties is critical to producing practical deployment guidance.

Analysis

Key Concepts

Disease Transmission

We can broadly categorize respiratory disease transmission into three modes: long-range airborne, short-range airborne, and droplet transmission. Diseases can be primarily transmitted through one of these modes or through multiple modes, and the precise boundaries between these modes of transmission are an area of active research and debate.

Long-range airborne transmission occurs when pathogens are mixed into and diluted by the air in a space. In this scenario, viable pathogens emitted by an infectious person disperse throughout the space and become relatively evenly distributed due to air currents and mixing. The time between emission and potential inhalation by others is typically minutes or longer. To model transmission risk in these cases, we use the average concentration of pathogen in the room air based on an assumed even distribution.

Short-range airborne transmission occurs when someone inhales a concentrated plume of respiratory aerosols directly from an infectious person’s exhaled breath, cough, or sneeze before it has been substantially diluted by room air. Rather than being characterized by average room concentrations, this form of transmission involves exposure to localized high concentrations of pathogens over very short time periods (seconds). A common example would be being downwind of someone’s cough or in their exhaled breath plume.

Speaking, coughing, or sneezing generates droplets and disperses them into the air. Droplet transmission traditionally refers to larger respiratory particles that travel ballistically (like tiny projectiles) rather than floating in air currents, typically depositing within 1–2 meters. However, the distinction between ‘droplet’ and ‘airborne’ is increasingly recognized as oversimplified, with particle size, environmental conditions, and air flows all affecting how long particles remain airborne.

The dominant mode of transmission in any given scenario depends on multiple environmental and built environment factors including ventilation design, air mixing, room layout, and occupant density. Consider two illustrative examples: in a classroom with ceiling fans providing good air mixing but minimal ventilation (low air changes per hour), one student’s sneeze will initially create a concentrated plume that could cause short-range transmission to nearby students. However, over minutes those particles will become mixed throughout the room air, transitioning to a relatively uniform concentration that poses a long-range transmission risk to all occupants. In contrast, in a space with a strong directional ventilation system that quickly removes air from the space, someone’s sneeze might create a concentrated plume that is carried by the air flow toward anyone downwind before being removed by the ventilation system. In this case, the short-range transmission risk to people in the path of the ventilated plume may dominate over long-range transmission risk from well-mixed room air.

In this section, we focus primarily on analyzing far-UVC’s potential effectiveness against long-range airborne transmission, where we can reasonably model pathogen concentrations and decay rates averaged across a space. The potential effectiveness against short-range transmission, where pathogens may have minimal exposure time to far-UVC before being inhaled, requires different analytical approaches.

Pathogen Decay

The rate of exponential decay is defined in the equation:

where:
N_t = number of viable pathogens remaining at time t
N₀ = initial number of viable pathogens
λ = decay constant
t = time

Or if we express the fraction of a population that survives after time t:

This gives the remaining fraction of a population (such as a pathogen) after time t. There are a number of potential sources of decay of viable airborne pathogens in the absence of deliberate attempts to control it—for example, the deposition rate at which aerosol particles hit a surface or fall to the floor, as well as environmental influences like temperature, pH, and relative humidity.

Where there are two or more sources of decay, these are additive to the total rate of decay. For example:

λ_deposition = decay due to deposition
λ_environment = decay due to environment
λ_n = decay constant due to n
λ_sum = total effect of all sources of decay

λ_deposition + λ_environment + … λ_n = λ_sum

Steady-State Concentration

In a scenario where there is a contaminant that is being emitted into a space at a constant rate, the average concentration of the contaminant in a space will tend towards a steady state. This steady state is simply the rate of pathogen emission divided by the total decay, where both the pathogen generation and the decay constant have the same unit of time.

The steady-state concentration C_ss is calculated as:

with:
E = emission rate (quantity of contaminant emitted hr⁻¹)
λ_sum = starting sum of all sources of decay (hr⁻¹)
V = volume of space

Therefore, if we know the change in C_ss when there is no change in emission rate, we can calculate the relative change in λ_sum. Equation 3.3 shows that every halving of C_ss requires λ_sum to double, and that getting additional decreases in C_ss requires higher and higher λ_sum.

Air Changes per Hour

Air changes per hour (ACH) is a simple calculation of how many times per hour the entire volume of air in a given space is replaced.

with:
Q = volumetric air flow rate per hour
V = room volume

Note that to use this equation correctly, the unit of Q needs to be flow rate per hour and V needs to be a volume in the same unit in which the flow rate is calculated. If for example the volumetric flow rate is given in cubic feet per minute (cfm) and the room volume in cubic feet, Q needs to be multiplied by 60 to get ACH. If instead the unit of Q was liters per second (L/s) and the room volume in cubic meters, it needs to be multiplied by 3.6.

If the air in a room is assumed to be ‘well-mixed’, then the ACH is a decay constant (λ_ACH) with respect to the concentration of a contaminant in the air.

This is not an empirical observation, and ‘well-mixed’ has a somewhat circular definition—a room is well-mixed if the ACH predicts the decay. In the case of a room being well-mixed with respect to the average concentration of a contaminant where the source is in the room, the well-mixed assumption requires only that the air leaving the room has the same average contents as the entire room. If the source of the contaminant was in the air supply, a sufficient definition of well-mixed with respect to the average concentration of contaminant would be that the concentration in the air outlet is the same as the air inlet.

We may also care about whether particular parts of a room are well-mixed, such as the ‘breathing zone’, which is defined in ASHRAE Standard 62.1 as the “region within an occupied space between planes 3 and 72 in. above the floor and more than 2 ft (600 mm) from the walls or fixed air-conditioning equipment.” Depending on the particular ventilation design of the room, the breathing zone could be well-mixed but other parts of the room not, and vice versa. But the important thing to understand with respect to air mixing is whether or not a certain amount of airflow per hour can be treated as a decay constant, or if not, what correction needs to be applied to estimate the rate of decay.

eACH and CADR

An equivalent air change per hour (eACH) can be defined as anything that has the same effect as an air change per hour (ACH). Far-UVC, for example, can be said to have 1 eACH if it reduces pathogen concentration by the same amount as 1 ACH in a well-mixed room. This simply means that, mathematically, an eACH is an exponential decay constant (λ_eACH) where the unit is hr.

Clean air delivery rate (CADR) is equivalent to Q in equation 3.4 that defines ACH:

As for equation 3.4, attention needs to be paid to the units of CADR and volume when making calculations.

There are two important things to note about the difference between ACH and eACH. First, ACH is assumed to be contaminant-invariant (i.e. all contaminants are assumed to be subject to the same air mixing, and thus decay at the same rate). eACH and CADR, however, are not contaminant-invariant.

Consider gaseous versus aerosol contaminants. An in-room mechanical HEPA filter will have an eACH and CADR versus aerosol particles, but in the absence of an additional filter type (such as activated carbon) it will have little or no effect on the gas. And as shall be seen later, in the case of far-UVC the eACH and CADR will be different with respect to different pathogens.

Second, once the ‘well-mixed’ assumption is relaxed, ACH is no longer equivalent to eACH. The well-mixed assumption is necessary for Q/V to be treated as a decay constant with respect to the average concentration of a contaminant.

A common practice among ventilation practitioners is to use a ‘mixing factor’ (sometimes referred to as ‘K’) to discount ACH to estimate an ‘effective’ ACH (i.e. what is the decay per hour, which is the same thing as this definition of an equivalent ACH).

There is very little in the recent published literature on mixing factors for ventilation, and references to it are more typically found in manuals for industrial ventilation practice or in standards such as ASHRAE 62.1 which describes the concept of Zone Air Distribution Effectiveness (E_z). E_z is the degree to which ACH should be discounted specifically with regards to contaminants in the breathing zone.

Further research into accurate modeling of ventilation would greatly improve our understanding of the relative effectiveness of ventilation and other interventions such as far-UVC.

Key Takeaways

• Airborne pathogen viability decays due to many factors including deposition, environmental factors, ventilation, and the use of air cleaning technologies.
• When a contaminant is emitted at a constant rate, its steady-state concentration depends on how quickly it is removed from the air.
• Doubling decay rate cuts steady-state pathogen concentration in half, therefore achieving greater reductions requires greater increases in decay.
• Far-UVC adds to decay, similar to ventilation and filtration, reducing airborne pathogen levels.
• ACH (air changes per hour) measures how often air is replaced in a room; it assumes perfect mixing, and does not account for filtration or inactivation.
• ACH can be treated as a decay constant in well-mixed rooms, but real-world airflow varies, meaning actual effectiveness depends on ventilation design, air distribution, and room occupancy.
• eACH (equivalent ACH) represents the pathogen removal effect of air cleaning technologies like far-UVC and filtration, translating their impact into an air exchange equivalent.
• CADR (clean air delivery rate) quantifies the volume of air effectively cleaned per unit time, allowing comparison across different air purification methods, including filtration and UV.
• Far-UVC’s eACH varies by pathogen because different microbes have different UV susceptibilities, affecting how efficiently they are inactivated.

How Do We Study How Well Far-UVC Reduces Transmission of Airborne Diseases?

Effectiveness

The effectiveness of an intervention at reducing transmission of disease means how many infections it prevents in real-world environments. Or to put it concretely: if far-UVC were installed in a school classroom, or an urgent care waiting room, or a conference center, how many illnesses would be prevented?

This deceptively simple question is very challenging to study. This is not a property of far-UVC in particular, but rather there are general limitations to studying interventions that occur at the level of a place, when the endpoint of the study is how many people get sick.

The Challenges of Cluster-Randomized Trials

Suppose a device was invented that stopped 100% of the transmission of respiratory infections in the room in which it was placed. How would we prove, empirically, that such a device worked?

Cluster-randomized trials are the gold standard for assessing the effectiveness of treatments at a group level. In a cluster-randomized trial, researchers randomly assign an intervention like far-UVC to some places, treat other places as a control, and compare the infections of the people who spend time in the places with devices, and the people who spend time in the control places. However, cluster-randomized trials come with a number of challenges.

The first challenge is that we can’t easily measure how much a disease was transmitted between individuals in a given place. We can only measure who has an infection or was infected, and know what places they have been. For some populations, the risk of acquiring a respiratory infection is concentrated only in one particular place: for example, residents of a long-term care facility or a military barracks. But these are the exception rather than the rule. Most people have multiple possible places where they can acquire respiratory infections, and no one place constitutes the vast majority of their overall infection risk.

Therefore, even if an environmental intervention is 100% effective, it is essentially impossible to measure a 100% reduction in infections using a cluster-randomized control trial. Even if no-one contracted an infection in a far-UVC-treated location, in the vast majority of cases, people who visited that place could get sick in many other places. To put it more technically, if one intervention is tested to prevent transmission in a particular place, the measured effect size that can possibly be observed in a cluster-randomized trial of the people who occupy that space is reduced proportionately to how likely the people are to acquire an infection in that place relative to other places. In many situations, if a cluster-randomized trial detected a 20% reduction in infections among, say, children in a particular classroom, that could represent a far larger reduction in the amount of transmission that is actually occurring inside the classroom.

Furthermore, in some circumstances the intervention itself can displace transmission to other places or times. The latter effect is clearly visible in the study of the effectiveness of upper-room UV against measles in New York State schools published in 1947. The total reduction in infections over the study period in the school that had UV lamps in all classrooms was not statistically significant, but it can clearly be seen from the data that there was a successful ‘flattening of the curve’—a reduction in the rate of transmission—in the school where all classrooms had UV lamps. In the school where half of the classes had UV lamps, the results seem to show infections being partially displaced. One critical factor was that children mixed on school buses, so children who avoided measles in the irradiated classrooms could catch it on the bus.

Another historical example during the 1957–58 influenza pandemic in a TB hospital also demonstrated this displacement of transmission. While the ward equipped with upper-room UV showed dramatically lower infection rates among patients (1.9% versus 18.9% in the control ward), staff who worked in both wards had infection rates similar to the control ward (18%). No airborne pathogen is as infectious as measles, and displacement of infections is especially likely to occur with very infectious diseases.

The second challenge is that even in the same kinds of places—hospitals, classrooms, nursing homes, offices—the number of infections between places can have a very high natural variance, and therefore a relatively small effect size signal can be masked by a lot of noise.

The combination of these two challenges means that in a cluster-randomized trial, the study needs to be very large in order to be adequately powered to detect the effect, even if the intervention causes a large decrease in infections.

Unfortunately, studies that are underpowered and fail to find statistical significance often promote incorrect conclusions that interventions ‘don’t work’ when in fact they may provide meaningful benefits that the study wasn’t designed to detect. Cluster-randomized studies therefore need to be designed very carefully in order to understand what effect size can even be achieved given the infection dynamics of the particular infections, people and spaces, and what sample size is going to be required in order to detect that effect.

Alternative Methods of Measuring Transmission

Enrolling and managing studies that are comparable in scale to Phase III pharmaceutical trials requires significant overhead, but one of the main drivers of expense in cluster-randomized studies is the laboratory testing that is usually required to measure infections among study participants.

It is therefore desirable to gather data that is cheaper and more scalable than laboratory testing, in order to make conducting studies at this scale more practical. However, these alternatives also come with drawbacks.

Wastewater surveillance offers one potential alternative, providing population-level data on infection prevalence at a fraction of the cost of individual testing. By sampling building or community wastewater systems, researchers can monitor viral loads without the need for individual participation. However, wastewater measurements are only loosely correlated with actual infection rates due to variations in viral shedding, environmental degradation, and dilution effects.

Absenteeism tracking represents another low-cost approach that leverages existing administrative systems. These data are particularly attractive because they require minimal additional infrastructure or personnel to collect. The main limitation is that absences have low specificity for infections of interest—students and workers miss school or work for many reasons unrelated to the infections being studied.

Hospital admission rates can be accessed through existing healthcare reporting systems and provide a standardized metric across different locations. While this data is relatively inexpensive to collect, hospitalizations represent only the most severe cases, which are typically a small fraction of total infections.

The key challenge with all these surrogate measures of infection transmission is the tradeoff between cost and statistical power. While surrogate measures enable larger and longer studies by reducing per-participant expenses, their indirect relationship with infection rates means that more sites and participants are needed to achieve adequate statistical power.

The Need for Real-World Studies

We need data on the real-world effectiveness of far-UVC in reducing disease transmission among humans. First, we need to understand its true practical effectiveness, not just its theoretical potential, in order to calibrate its role within a layered pandemic defense strategy. Second, for far-UVC to have a meaningful impact on pandemic prevention, it would need to be widely adopted in public spaces, which requires building public trust through robust evidence. Third, regulatory bodies and standards organizations need empirical data to develop appropriate guidelines and requirements. Finally, policymakers and building operators need comparative effectiveness data to evaluate far-UVC against other options for reducing airborne transmission in the built environment.

Key Takeaways

• Directly studying effectiveness at transmission suppression in real-world environments is challenging. However, cluster-randomized trials remain the gold standard of evidence and there are numerous ongoing and attempted efforts to execute them for interventions to reduce airborne transmission.
• It is difficult to know in advance whether a cluster-randomized trial is adequately powered to detect a given effect.
• There’s a risk that underpowered or poorly designed effectiveness trials are interpreted as showing that an intervention ‘doesn’t work’ when in fact a study may not even have been capable of showing the intended effect.
• The use of alternative endpoints to measure transmission of airborne infection can enable larger and relatively less expensive trials, but this comes at the cost of study power.

Efficacy

Efficacy is the measure of how well far-UVC inactivates pathogens. It is presumed to be linked to effectiveness through reducing the probability that a susceptible person is exposed to an infectious dose. The reported 98.4% reduction in airborne bacteria observed in Eadie et al., 2022 is an efficacy study, with the implication being that a large reduction in the quantity of viable airborne pathogen ought to lead to a large reduction in airborne transmission.

Pathogen Surrogates

Efficacy studies often use surrogate pathogens, enabling experiments to take place at lower biosecurity levels. Until recently, handling live SARS-CoV-2 required a BSL-3 lab, leading researchers to use surrogates such as the common cold coronavirus HCoV-OC43 (BSL-2) or bacteriophages such as MS2 and T1 (BSL-1). While SARS-CoV-2 has now been reclassified to BSL-2, much of our understanding comes from these surrogate studies.

Surrogate pathogens can also have other properties that make them more amenable to study. For example, in order to detect large reductions in pathogen load, we need to be able to culture the pathogen to a high initial concentration. One of the reasons that MS2 bacteriophage is commonly used as a surrogate is that it is easy to culture, relatively robust to environmental conditions, and inexpensive. Ideally, a surrogate’s susceptibility to UV should be as similar to the microbe of interest as possible. But the main requirement of a surrogate is that its dose-response to UV be representative of the pathogen.

Efficacy studies can also be conducted in ‘real’ spaces with naturally occurring pathogens, as well as with non-microbial surrogates such as DNA-tagged aerosols. However, these studies are not straightforward, and the concentration of the pathogen will in many situations not be high enough to detect a signal; ‘real’ spaces may also have high background variation in pathogen levels, for example associated with changing occupancy or activities of occupants.

How Efficacy Is Measured

Efficacy studies are often quoted as a reduction in the amount of pathogen when exposed to far-UVC compared to non-exposed controls. Sometimes this is quoted as a log reduction—for example the 98.4% reduction in Eadie et al., 2022 would be a 1.8 log reduction in the concentration of pathogen.

However, percent reductions and log reductions are not straightforward to interpret. Altering the experimental setup—such as the duration of the experiment—can dramatically change the log reduction observed. Furthermore, the size of the chamber in which the experiment is conducted can also dramatically affect the observed reduction. By themselves, log reductions are not a useful metric for understanding the efficacy of air cleaning technologies.

Calculating eACH and CADR provides a more useful metric for translating experimental results to real-world applications for air cleaning, as they relativize the reduction to the duration of the experiment and the size of the chamber in which the experiment is conducted.

Calculating eACH and CADR: Decay Versus Steady State Tests

Calculating eACH and CADR from an efficacy experiment requires different methods depending on whether the test uses continuous introduction of the challenge agent or not (‘steady-state’ versus ‘decay’ method).

In the decay method, the pathogen is introduced and then is allowed to decay naturally, with far-UVC lamps on and off. The eACH and CADR can then be calculated by comparing the rate of decay with and without the lamps on using equation 3.2.

Alternatively, if a pathogen is continuously introduced at a constant rate, then equation 3.3 can be used to estimate the change in decay that is consistent with the observed change in pathogen concentration. However, in order to use this method to calculate an eACH, the base rate of decay without the intervention must be known.

Steady-state and decay tests have different advantages and disadvantages. Both have been used in studies of far-UVC, but the test methods that have been established as consensus standards all employ the decay method. In a direct comparison between the decay and steady-state method, one study found higher inactivation with the steady-state method. It’s not currently clear which test method is preferable for measuring devices which are intended to be used for practical application.

Efficacy in Controlled Chamber Studies

Data from four studies of efficacy in large chambers is currently in the public domain, three of which have been published in peer-reviewed journals. The fourth study has been reported at an academic conference, and the talk is publicly available.

Table 3.1. Summary of Publicly Available Far-UVC Efficacy Data in Room-Sized Bioaerosol Chambers

*Studies had differences in fixtures, radiant power, and/or average fluence rate.

Across these controlled room-sized chamber studies, far-UVC showed very high eACH and CADR against aerosolized Staphylococcus aureus and Pseudomonas aeruginosa, but minimal eACH and CADR against two bacteriophages.

Until we know how susceptible the surrogate pathogens are to far-UVC, it is difficult to understand or interpret these studies. For example, it could be the case that S. aureus is particularly susceptible to far-UVC, and the bacteriophages particularly resistant. There are many more studies on pathogen susceptibility to far-UVC than efficacy, and this will be covered in the next section. But overall, if the real-world efficacy of far-UVC against pathogens we care about is more accurately represented by Eadie et al., 2022, then it is an extremely promising technology. If it is more accurately represented by the Ratliff and Blatchley studies, it holds far less promise.

Efficacy in Real Spaces

In addition, there is one far-UVC efficacy study that has been conducted in an occupied room against naturally occurring pathogens—specifically, a mouse cage cleaning room which contains high concentrations of aerosolized murine norovirus (MNV). Four far-UVC fixtures were placed on the ceiling of the 95 m³ room, and the concentration of aerosolized MNV estimated with lamps on and lamps off. The researchers observed a 99.8% reduction in concentration when the lamps were on.

This result is made even more spectacular when we consider the base level of pathogen decay in this space. The mouse cage cleaning room ventilation rate is nominally 36 air changes per hour based on the air supply. The ACH, calculated at 7 different points in the room using a CO₂ monitor, ranged from 31.5–72.4 ACH, with the latter reading taken near the pathogen air sampler.

Using equation 3.2 and the nominal ACH of 36, this would suggest an additional 14,800 eACH added from the use of far-UVC, assuming a well-mixed space, if the reduction can be interpreted as a change in the average steady-state pathogen concentration. But this calculation is highly sensitive to the assumption around the base decay, and applying a mixing factor could reduce the estimated eACH by as much as a factor of 10.

The seemingly narrow confidence interval on the measured reduction in MNV (95% CI: 98.2–99.9%) translates to a very large difference in the estimated eACH, ranging from 200 to 3000 assuming a mixing factor of 10.

Key Takeaways

• Efficacy studies measure how well far-UVC reduces the concentration of pathogens.
• For the purpose of air cleaning, the key efficacy metrics are eACH and CADR.
• Among the three lab-based studies available to date in room-sized chambers, one has shown far-UVC to have very high efficacy against bacteria, but two studies have shown much lower efficacy against bacteriophages.
• There has been one far-UVC efficacy study in a real space using naturally occurring virus (murine norovirus), which also showed very high efficacy.
• In order to understand and interpret these efficacy studies, we need to know the susceptibility of the surrogate pathogens, and how these compare to pathogens of concern.

Susceptibility

Susceptibility is the relationship between UV dose—how much UV a pathogen is exposed to—and efficacy. Quantitatively, this relationship is expressed using a modified version of the exponential decay equation:

where:

S = survival fraction
D = UV dose
k = UV susceptibility constant

The unit of k is the inverse of the dose, and therefore usually either m²/J or cm²/mJ, just as in the conventional decay equation the unit of λ is the inverse of time. The higher the k value for a pathogen, the more susceptible the pathogen is to inactivation from UV.

In a handful of papers in the UV literature, k is calculated in base 10 rather than base e. A k in base 10 can be converted to a k in base e by using the change of base rule of logarithms, which results in multiplying a k in log 10 by 2.303.

Experimental Procedures for Measuring k

In order to estimate k, equation 3.6 illustrates the basic requirements. The pathogen is exposed to different radiant exposures and for each, the survival fraction is calculated as a ratio of the pathogen quantity at that dose relative to the pathogen quantity in sham irradiated controls. Susceptibility studies are more practical than efficacy studies, and usually involve survival in water rather than in air.

For explanations of typical experimental methodologies for measuring dose and survival fraction as well as techniques for culturing, aerosolizing and controlling irradiation of the pathogen, the report recommends the methods sections of Welch et al., 2018 and Lu et al., 2024.

k values depend on the pathogen, the UV wavelengths, and the environmental conditions. For example, the effect of increasing relative humidity on a number of pathogens, particularly bacteria and some viruses, is to lower their susceptibility to conventional UVC. A pathogen population may also not have a single k, and in some cases inactivation of the pathogen is better described using a biexponential dose-response relation.

with:
f = resistant fraction of pathogen population
k₁ = susceptibility of sensitive fraction
k₂ = susceptibility of resistant fraction

Welch, 2022 showed that biexponential decay provided a much better fit to the observations of the survival of HCoV-OC43.

In this instance, a single exponential model underpredicts inactivation at doses <1.5 mJ/cm², and overpredicts at higher doses >1.5 mJ/cm².

How much the ‘resistant fraction’ matters to efficacy depends on how large the resistant fraction is and how much reduction in pathogen load you are trying to achieve. The 0.3% of OC43 that is modelled with the lower k will have relatively little impact unless you were trying to achieve 99%+ reduction in pathogen load. At a 50% reduction, the 0.3% resistant fraction will have a negligible impact.

The existence of multi-phase decay does not appear to be due to properties of the pathogen, i.e. that there are individual virions or bacteria that are more resistant to far-UVC. Rather, it is more likely explained by physical ‘self-shielding’ from UV which happens when microorganisms are clustered together. It is an open question how much the physical clustering that occurs in controlled studies in small bioaerosol chambers is representative of clustering of pathogens in human respiratory aerosols.

Variation by Pathogen

k values can vary substantially by pathogen. For example, Lu et al., 2024 compare susceptibility for various bacteria and bacteriophages using different types of UV, with the variations in k across pathogen spanning approximately one order of magnitude.

Table 3.2. Wavelength-Specific Inactivation Rate Constants

In another analysis of published data, susceptibility of 59 animal viruses and 33 bacteria to conventional UVC spans one and a half orders of magnitude, and susceptibility is relatively well-predicted by a genomic model.

Relative Humidity

Relative humidity (RH) can significantly affect the UV susceptibility of airborne pathogens, though these effects are complex and species-dependent. Most bacteria show decreased UV susceptibility at high RH, while viruses show mixed results—some experience increased susceptibility with higher RH, others show decreased susceptibility or no effect.

The mechanisms behind these RH effects are varied and include DNA conformational changes, changes in microbe size due to water absorption, and changes in natural pathogen stability. The medium in which a pathogen is aerosolized may also impact its response to RH.

There is not currently any published data showing the effect of relative humidity on far-UVC inactivation of pathogens.

Inconsistencies in k Across Studies

There is also significant divergence in k values of the same pathogen and pathogen family across different studies. Coronaviruses are the pathogens that we have the most far-UVC data points for in aerosol.

Table 3.3. Published Far-UVC Inactivation Constants for Aerosolized Coronaviruses

*k values originally calculated in base 10 have been multiplied by 2.303. MEM = Modified Eagle’s Medium; PBS = phosphate-buffered saline; DI = deionized.

By contrast, the susceptibility estimates for coronavirus inactivation in liquid show much lower variance, and in absolute terms are more comparable to the lower k estimates for aerosol.

Table 3.4. Published Far-UVC Inactivation Constants for Coronaviruses in Liquid

Table 3.5. Published UVC Inactivation Constants for Aerosolized S. marcescens

Human Aerosols

While far-UVC is generally highly effective at inactivating airborne pathogens, its efficacy can be influenced by the surrounding environment. In particular, when pathogens are contained by protein-rich media, some far-UVC radiation will be absorbed by these surrounding proteins before it can reach the pathogen. Respiratory pathogens emitted by infected humans are transported within the airborne particles that are the normal product of the human respiratory system.

At least one attempt has been made to model the optical properties of human respiratory aerosol particles to estimate the potential attenuation of the far-UVC dose that reaches pathogens due to absorption by proteins in the aerosol. Hill et al. modeled how UV wavelengths, including 222 nm and 254 nm, interact with aerosol particles of varying sizes and compositions, including simulated respiratory media.

Their findings emphasized that smaller aerosols (<2 µm) allow deeper UV penetration with minimal shielding, while larger droplets (>4 µm) exhibit significant attenuation, especially at 222 nm, due to its shorter penetration depth. The study also highlighted how proteins and other components in respiratory media contribute to shielding effects, particularly in larger particles, reducing the UV dose that reaches pathogens.

Aerosols generated when coughing, breathing, or sneezing vary in composition and size depending on the site of generation in the respiratory system. The smaller respiratory aerosols which may be disproportionately responsible for airborne transmission originate from deeper in the respiratory tract. The difference in composition plausibly impacts the susceptibility of pathogens within them to UV, for example because of potential protein shielding.

There are currently two peer-reviewed publications examining the effect of medium on far-UVC inactivation in air. Monika et al., 2025 compared inactivation of MS2 and Phi6 bacteriophage using both a KrCl* lamp and a low-pressure mercury lamp with SM Buffer and artificial saliva as the medium. Phi6 was measured to be more susceptible to both sources of UV when using artificial saliva, and MS2 was more susceptible for the low-pressure mercury lamp, but no significant difference was measured for the KrCl* lamp.

The second study is Ratliff et al., 2025, which compared inactivation of MS2 in simulated saliva and deionized water. In this study the effect of the simulated saliva was directionally the same—the measured clean air delivery rate was higher when using simulated saliva rather than deionized water.

Given all the challenges and variability observed in lab-based efficacy studies, identifying experimental designs to establish the effect of far-UVC against pathogens in actual human respiratory aerosols should be a high priority.

Key Takeaways

• Pathogen susceptibility is measured by its dose response to UV, usually called a k or z value.
• k values can vary significantly for different pathogens, in different exposure conditions, and even across different studies for the same pathogen.
• The impact of proteins in human aerosols on pathogen susceptibility to far-UVC has not yet been quantified.

Modeling Efficacy and Susceptibility

The key question is: does this uncertainty about germicidal susceptibility matter? It is possible that far-UVC, when used within safe exposure limits, could be effective enough anyway even if k constants are lower in real-world, non-laboratory conditions. Conversely, it is possible that far-UVC is not effective enough, even against highly susceptible pathogens in ideal conditions, to be practically employed for whole-room disinfection.

A simple dimensional analysis shows how k constants can be used to estimate the eACH and CADRs that may be achievable from the use of far-UVC. It is very important to note that unlike studies that attempt to measure susceptibility and efficacy directly, the results derived in this section have not been shown to be empirically true. Rather, the mathematical definitions of eACH, CADR and k, along with a critical simplifying assumption, allow for translating between experimental measures of efficacy and susceptibility.

The critical assumption that enables this simple analysis is that we can use an average UV fluence rate, and implicitly a uniform dose rate distribution, to estimate inactivation. The use of averages may lead to poor estimates of efficacy from susceptibility.

Estimating eACH from k and Assumption of Uniform Fluence Rate

Equations 3.2 and 3.6 are structurally similar. In order to translate between them we require a relationship between UV dose and time in hours. In other words, if we can calculate the dose per hour, then we can translate between UV susceptibility k and decay rate per hour λ.

Dose per unit time is the definition of UV fluence rate (I). To translate between UV dose D and time t, we use the equation for UV dose:

We can therefore substitute equation 3.8 into equation 3.6. Note that in equation 3.8, time t is in seconds, but to compare to ACH we want time t to be in hours or 3600 seconds.

Recalling the definitions of ACH and eACH, the estimate for eACH using far-UVC is:

If we know k, I, and all the other sources of decay, we can also estimate the ratio of the average steady-state pathogen concentration with and without UV when there is a continuously emitting source.

with:
C_base = average steady-state concentration without UV
C_UV = average steady-state concentration with UV
λ_sum = other non-UV sources of decay (hr⁻¹)
k = pathogen inactivation constant (cm²/mJ)
I = average fluence rate (μW/cm²)

It is important to reiterate that these equations are only valid assuming a uniform distribution of dose, and the fluence rate being used is the average across the whole room.

If instead we wish to estimate efficacy against short-range transmission that occurs entirely in the breathing zone, eACH is not a valid metric, as we are no longer in the framework of room-average concentrations. Instead, we must know the time elapsed between exhale and inhale, so that a dose can be calculated from fluence rate and equation 3.6 used to estimate the survival fraction before the pathogen is inhaled.

If the time elapsed between inhale and exhale is a matter of seconds, then there is not much time for the pathogen to accumulate a substantial dose. For this reason, we expect far-UVC to be less effective against short-range transmission than long-range transmission.

Inferring Susceptibility from Efficacy Studies

These equations can also be used to estimate what pathogen susceptibility is consistent with the results of an efficacy study, assuming a uniform dose distribution. However, even among susceptibility and efficacy studies that are reported in the same paper, there is significant inconsistency when attempting to translate between susceptibility and efficacy.

Table 3.6. Comparison of Inferred and Direct Susceptibility Constants

There is currently no explanation for these large discrepancies, and the only thing we can say with confidence is that currently available susceptibility and efficacy data appear to be inconsistent. Combined with the variation in susceptibility measurements across multiple studies, this incongruence is an additional reason that standardization and validation of experiment methodologies needs to be a high priority.

Deriving Clean Air Delivery Rates from k and Assumption of Uniform Dose

The assumption of uniform dose allows for an alternative method for estimating the clean air delivery rate of a far-UVC device. It suffers from all the same challenges as deriving eACH but gives a point of comparison at the level of the device.

P = UV power output (Watts)
L = average photon pathlength (m)
V = volume (m³)

This equation for I can be substituted into equation 3.10:

This equation can then be substituted into the definition of CADR, which results in the Volume terms cancelling out.

where:
P = UV power output (W)
L = average photon pathlength (m)
k = pathogen susceptibility constant (cm²/mJ)

Comparing Far-UVC Efficacy to Other Air Cleaning Technologies

UV, including far-UVC, and other air cleaning technologies have been compared in modeling studies, as well as in aerosol chamber experiments. One modeling study estimated that the UV-based methods in a prototypical office provide pathogen inactivation with lower energy consumption and less thermal discomfort than ventilation.

In the lab, Landry et al., 2025 found that while HEPA filtration achieved the highest efficacy when tested against a UV-resistant bacteriophage, far-UVC systems likely outperform HEPA in inactivating more UV-susceptible human pathogens such as SARS-CoV-2 and Influenza A(H3N2).

Table 3.7. Estimated CADR (L/s)—Ray Length 2.5 m

Table 3.8. Estimated CADR (L/s)—Ray Length 4 m

These tables illustrate the significance of the current uncertainty about the susceptibility of pathogens contained within human respiratory aerosols. The relative cost-effectiveness of far-UVC relative to other air cleaning technologies will be strongly dependent on differences in the inactivation constant k and also on the average UV ray length in a particular installation.

Table 3.9. Estimated CADR (L/s)/Watt: 2.5 m Ray Length, 12 W Input

This analysis suggests that current far-UVC devices can be more energy-efficient than even the best available portable HEPA filters per unit of disinfection, provided that the pathogen k is above ~3.5 cm²/mJ when ceiling heights are relatively low.

Table 3.10. Estimated CADR (cfm)/Watt: 4 m Ray Length, 12 W Input

MERV 13 Filter-Based Air Cleaners

DIY PC-fan-based Corsi-Rosenthal boxes using MERV 13 filters are considerably more energy-efficient and quieter than HEPA-based air cleaners. One typical kit available for purchase has been shown in tests by Intertek to deliver 401 cfm (189 L/s) of clean air against smoke using just 14 W of input power, or 13.5 (L/s)/Watt.

In terms of noise, the sound power level of this particular Corsi-Rosenthal box is only 39 dBA, which is an order of magnitude lower than the sound output of HEPA-based portable air cleaners with comparable CADR.

We believe that the promise of portable air cleaners optimized for energy efficiency and sound reduction merits further investigation, although this is outside of the scope of the current document. When deciding whether to prioritize research into far-UVC and whether it is sufficiently promising, we should consider the best possible versions of alternative technologies that could also be worth investing resources into.

Key Takeaways

• Far-UVC devices can achieve significant clean air delivery rates (CADR), with higher effectiveness in rooms with greater ceiling heights.
• Far-UVC devices can be more energy-efficient than portable HEPA filters, especially for pathogens with higher inactivation constants (k).
• Portable HEPA filters can have a CADR of up to 600 cfm at their highest settings but require regular filter replacements and produce considerable noise.
• MERV 13 filter-based air cleaners, such as Corsi-Rosenthal boxes, can be quieter and more energy-efficient than HEPA filters but may still fall short of far-UVC’s effectiveness under optimal conditions.

The Need to Model Distributions and Not Averages

All of the foregoing analysis makes two critical simplifying assumptions. The first is that all pathogens in a population are subject to the same average fluence rate. The second is that there is a single k that captures the dose-response relationship to the pathogen.

Relaxing either of these assumptions introduces substantial modeling complexity. However, this complexity may be unavoidable if we want to make accurate predictions.

The effect of non-uniform dose can be illustrated by a simple example. Suppose that instead of a pathogen receiving a uniform dose, half of the pathogen receives 2x of the average dose, and half 0.5x the average dose. The average dose remains unchanged.

At a dose of 8 mJ/cm², assumption of uniform dose would overpredict inactivation by one and a half orders of magnitude. On the other hand, for the first approximately one log of inactivation it is relatively accurate. At low levels of inactivation the assumption of average dose may in some circumstances suffice, but at higher levels of inactivation what matters is not the average but the tail of the dose distribution.

It is not clear whether we should expect multi-phase decay to occur at room scale with pathogens contained within human respiratory aerosols even if it does occur at the scale of a benchtop aerosol chamber. However, the consequence of relaxing the assumption of uniform dose and single decay constant is that there cannot be a single eACH or CADR for far-UVC, even for a single pathogen.

In order to understand the magnitude of this effect, computational fluid dynamics modeling is likely to be required in order to produce realistic estimates of dose distribution in real-world environments. This will allow for both more sophisticated estimates of efficacy based on susceptibility studies, and for translating between efficacy and susceptibility studies and understanding whether or not they are consistent.

A recent innovation is DNA-tagged aerosol tracers, which can be safely aerosolized in a room and used as a surrogate for quantifying pathogen inactivation. If the susceptibility of the tracer is well-characterized under the relevant environmental conditions, this could help us to empirically validate fluence rates and dose distribution in real-world environments.

Key Takeaways

• The assumption of uniform dose and a single dose-response relationship allows for simple estimation of eACH and CADR from susceptibility data.
• This simple estimation suggests that at realistic susceptibility, far-UVC may have highly favorable efficacy compared to other air cleaning technologies.
• However, when using these assumptions, published susceptibility and efficacy data are currently inconsistent.
• As well as obtaining more efficacy and susceptibility data using standardized experiment methodologies, more sophisticated computational fluid dynamics modeling is necessary in order to relax the assumption of averages and obtain more realistic estimates of efficacy.
• DNA-tagged aerosol tracers represent a promising approach for validating modeling and quantifying dose and dose distribution in the field.

Challenge Studies

Challenge studies—where transmission is deliberately induced in a controlled experimental setting—could advance our understanding of far-UVC in three ways. First, they could explore the dynamics of transmission; second, they could act as a powerful proof of concept; and third, they could assist in the validation of the engineering framework for modeling efficacy.

Animal-to-animal studies provide valuable insights into transmission mechanisms and the environmental factors that influence them. Human-to-animal challenge studies have played an important role in the development of upper-room UV, and their results will be the basis of the next generation of deployment guidance for ensuring effective installations. Human-to-human challenge studies are a potentially powerful model for proving the concept of interventions like far-UVC, but even if deliberately causing mild human infections to aid research is ethically acceptable, there are a number of practical challenges in human-to-human studies of airborne transmission that have not yet been successfully overcome.

Animal-to-Animal Challenge Studies

Animal-to-animal challenge studies are a valuable tool for examining the dynamics of airborne infection in controlled environments. By studying how pathogens such as influenza virus, SARS-CoV-2, or M. tuberculosis transmit between animals such as ferrets, guinea pigs, or hamsters, these studies provide key insights into factors like aerosol particle size, environmental conditions, and infectious dose.

Animal-to-animal models can also help us bound expectations for the effectiveness of interventions. Blocking transmission between two animal cages or enclosures using far-UVC would provide useful preliminary data, but falls short of what is needed to justify deploying far-UVC technology in real-world public spaces. The physiological and behavioral differences between animals and humans, combined with the highly controlled nature of these experiments, limit their direct applicability to human environments.

Human-to-Animal Challenge Studies

Animal challenge studies, particularly those involving guinea pigs, have been widely used to study airborne transmission of respiratory pathogens and to evaluate interventions such as upper-room or in-duct ultraviolet germicidal irradiation. In these experiments, guinea pigs are exposed to air exhausted from spaces occupied by infected humans, effectively serving as a model for human-to-animal transmission.

Notably, Riley et al., 1962 provided pivotal evidence that tuberculosis is transmitted through airborne droplet nuclei using the human-to-guinea pig model, reshaping public health strategies and solidifying the importance of airborne precautions. This work laid the foundation for the development of interventions like upper-room UV and continues to inform the next generation of deployment guidance for ensuring effective installations.

Several studies have leveraged this model to test the efficacy of airborne infection control measures, including upper-room UV, in real-world settings. These studies provide an infection endpoint, allowing researchers to measure the direct impact of interventions on transmission, and have formed the basis for dosing guidelines for upper-room UV.

However, human-to-animal transmission studies also have limitations. They rely on indirect measures of human infectiousness, as the outcomes are observed in animals rather than humans. Furthermore, the physiological and immunological differences between humans and guinea pigs may limit the generalizability of findings. Despite these limitations, the model remains invaluable for studying complex transmission dynamics and testing interventions under realistic conditions.

This framework highlights the potential for using human-to-animal studies to assess the efficacy of far-UVC in reducing airborne transmission of pathogens in real-world scenarios. Such studies could provide critical data on far-UVC’s performance in treating human-generated aerosols.

Human Challenge Studies

Human challenge trials have historically provided insights into infectious diseases and their transmission dynamics. These studies, where volunteers are deliberately exposed to pathogens, allow researchers to control variables that are impossible to isolate in natural settings, yielding high-resolution data on infection progression, immune responses, and the efficacy of interventions.

One of the most notable examples of the utility of human challenge trials was at the UK Common Cold Unit. This facility was instrumental in the first identification and description of coronaviruses in the 1960s. Influenza, norovirus, and more recently SARS-CoV-2 have also been studied using human challenge approaches.

However, while these studies provide valuable data, they primarily rely on direct inoculation of participants via methods such as intranasal drops or inhalation through face masks. This limits their applicability to interventions like far-UVC, which aim to interrupt human-to-human transmission. Achieving this requires a shift to studies that more closely mimic real-world conditions of transmission.

Despite these difficulties, human-to-human transmission trials represent a potentially powerful tool for testing interventions like far-UVC. If perfected, they could be used to directly evaluate far-UVC’s ability to interrupt transmission and allow for direct comparisons against other interventions like ventilation and filtration, or even masking, hand washing, or vaccination.

Key Takeaways

• There is not currently a successful model for human challenge studies of respiratory illness that is relevant to far-UVC.
• Animal study designs are feasible, but may be better characterized as ‘efficacy’ studies.

Mechanism of Action

How Far-UVC Inactivates Pathogens

Far-UVC is believed to operate through a dual mechanism of action. It is absorbed by both the nucleic acids (DNA and RNA) and proteins of pathogens, leading to damage and inability to infect and/or replicate. This dual capability distinguishes far-UVC from traditional UV-C at 254 nm.

The most well-understood mechanism by which far-UVC inactivates pathogens is through DNA and RNA damage. Like 254-nm UV-C, far-UVC photons are absorbed by nucleic acids, leading to the formation of cyclobutane pyrimidine dimers (CPDs) and 6-4 photoproducts (6-4PPs). CPDs are covalent linkages between adjacent pyrimidine bases which prevent proper replication and transcription. This renders microbes non-viable and neutralizes their ability to cause infection.

In addition to CPDs, far-UVC also induces the formation of 6-4 photoproducts, which are cross-links between adjacent pyrimidine bases but involve different chemical bonds than CPDs. The formation of these photoproducts further disrupts the genetic material of pathogens, adding another layer of damage that impairs repair mechanisms and, as a consequence, the overall function and viability of the pathogen.

Beyond nucleic acid damage, far-UVC’s shorter wavelengths allow it to be more readily absorbed by the peptide bonds of proteins, which are abundant in the structural components of pathogens, such as viral capsids and bacterial cell walls. The absorption of far-UVC by these proteins can lead to the disruption of their three-dimensional structure, impairing their function.

As we would expect from the nucleic acid absorption spectrum, 222-nm UV is less efficient than 254-nm at inducing direct genome damage. However, the protein absorption properties of far-UVC are not only qualitatively but also quantitatively important to pathogen susceptibility. For HCoV-OC43, the net effect of both protein and nucleic acid damage was higher susceptibility at 222 nm than 254 nm.

Recent findings suggest that far-UVC may also have an effect on the aerosol droplet itself, which could increase the rate of decay of pathogens within the droplet. While these conclusions remain speculative, they highlight an additional potential mechanism that would enhance far-UVC’s germicidal effects against airborne respiratory pathogens.

Acquiring Microbial Resistance to Far-UVC

Unlike antibiotics, which target specific bacterial functions such as protein synthesis or cell wall formation, far-UVC causes widespread and random damage to microbial DNA, RNA, and proteins. The randomness associated with multiple potential pathways for damage makes it harder for pathogens to develop specific resistance mechanisms.

However, given there are differences in the susceptibility of different pathogens, the widespread use of far-UVC could place selection pressure on all microbes to evolve the properties of those pathogens that are already less susceptible.

Experimental evidence has shown that certain bacteria, such as extended-spectrum β-lactamase-producing Escherichia coli, can develop tolerance after repeated far-UVC exposure. In this study, one bacterial strain exhibited a modest increase in survival after multiple sublethal doses. However, not all bacteria exposed to repetitive far-UVC develop tolerance.

There is currently no research that the authors are aware of into susceptibility of viruses to either far-UVC or conventional UVC after repeat exposures. Compared to bacteria, there are fewer ways that viruses can become less susceptible to far-UVC, for example due to their intrinsic lack of repair mechanisms.

The key question, however, is not whether microbes could evolve under selection pressure to become less susceptible to far-UVC. We should expect this to happen to some degree under sufficient selection pressure. The question is whether the decreased susceptibility would be quantitatively significant.

Despite the potential for reductions in susceptibility under ubiquitous use, far-UVC offers a fundamentally different approach to pathogen control than conventional methods. Its mechanism of random, widespread molecular damage works orthogonally to antibiotics, antiseptics, and traditional cleaning agents, which increasingly face resistance challenges. This independent mechanism of action makes far-UVC a promising complementary tool in a layered approach to infection prevention.

Key Takeaways

• Far-UVC inactivates pathogens through damaging both nucleic acids and proteins.
• There is limited experimental evidence that some bacteria may become less susceptible to far-UVC after repeat exposure, potentially due to selecting for mutations related to DNA replication and repair.
• The impact of any selection pressure that consistent far-UVC exposure would place on viruses is currently unknown. However, viruses are less likely to evolve tolerance to far-UVC than bacteria.

Further Reading

• Far-UVC light (222 nm) efficiently and safely inactivates airborne human coronaviruses
• Far-UVC light: A new tool to control the spread of airborne-mediated microbial diseases
• Far-UVC (222 nm) efficiently inactivates an airborne pathogen in a room-sized chamber
• Factors Affecting Reduction of Infectious Aerosols by Far-UVC and Portable HEPA Air Cleaners

References

1. Eadie, E. et al. Far-UVC (222 nm) efficiently inactivates an airborne pathogen in a room-sized chamber. Sci. Rep. 12, 4373 (2022).
2. Tang, J. W., Li, Y., Eames, I., Chan, P. K. S. & Ridgway, G. L. Factors involved in the aerosol transmission of infection and control of ventilation in healthcare premises. J. Hosp. Infect. 64, 100–114 (2006).
3. Henriques, A. et al. An integrated airborne transmission risk assessment model for respiratory viruses: short- and long-range contributions. J. R. Soc. Interface 22, 20240740 (2025).
4. Longest, A. K., Rockey, N. C., Lakdawala, S. S. & Marr, L. C. Review of factors affecting virus inactivation in aerosols and droplets. J. R. Soc. Interface 21, 20240018 (2024).
5. CDC. How Much Ventilation Is Enough? Ventilation https://www.cdc.gov/niosh/ventilation/prevention/Aim-for-5.html (2024).
6. CDC. Appendix B. Air. Infection Control https://www.cdc.gov/infection-control/hcp/environmental-control/appendix-b-air.html (2024).
7. Standards 62.1 & 62.2. https://www.ashrae.org/technical-resources/bookstore/standards-62-1-62-2.
8. Reda, I. et al. A modified decay method based on a proposed uniformity index for measuring air change rates in non-uniform air mixed spaces. Build. Environ. 245, 110941 (2023).
9. Perkins, J. E., Bahlke, A. M. & Silverman, H. F. Effect of ultra-violet irradiation of classrooms on spread of measles in large rural central schools preliminary report. Am. J. Public Health Nations. Health 37, 529–537 (1947).
10. Guerra, F. M. et al. The basic reproduction number (R0) of measles: a systematic review. Lancet Infect. Dis. 17, e420–e428 (2017).
11. Radonovich, L. J. et al. N95 Respirators vs Medical Masks for Preventing Influenza Among Health Care Personnel: A Randomized Clinical Trial. JAMA 322, 824 (2019).
12. ClinicalTrials.Gov. https://clinicaltrials.gov/study/NCT06247059?term=Luby,.
13. Hay, A. D. et al. England RCT of air filtration to prevent respiratory infections (including covid-19) in care homes. Eur. J. Public Health 34, (2024).
14. Samuels, F. Indoor air monitoring goes to school. Chemical & Engineering News https://cen.acs.org/analytical-chemistry/Indoor-air-monitoring-goes-school/102/i27 (20240830).
15. Hernandez, M. Clean air for schools project. https://www.colorado.edu/faculty/hernandez/clean-air-schools-project.
16. Nix, H. P. et al. Preventing Respiratory Viral Illness Invisibly (PRiVII): protocol for a pragmatic cluster randomized trial evaluating far-UVC light devices in long-term care facilities to reduce infections. Trials 25, 88 (2024).
17. Thottiyil Sultanmuhammed Abdul Khadar, B. et al. Air purifiers and acute respiratory infections in residential aged care: A randomized clinical trial. JAMA Netw. Open 7, e2443769 (2024).
18. Munro, A. How to not be completely wrong about air filters. The Munro Report https://alasdairmunro.substack.com/p/how-to-not-be-completely-wrong-about-43c (2024).
19. Kucharski, A. Stop looking for an NPI miracle. Understanding the unseen https://kucharski.substack.com/p/stop-looking-for-an-npi-miracle (2024).
20. Bundgaard, H. et al. Effectiveness of adding a mask recommendation to other public health measures to prevent SARS-CoV-2 infection in danish mask wearers: A randomized controlled trial: A randomized controlled trial. Ann. Intern. Med. 174, 335–343 (2021).
21. Rutterford, C., Copas, A. & Eldridge, S. Methods for sample size determination in cluster randomized trials. Int. J. Epidemiol. 44, 1051–1067 (2015).
22. Parkins, M. D. et al. Wastewater-based surveillance as a tool for public health action: SARS-CoV-2 and beyond. Clin. Microbiol. Rev. 37, e0010322 (2024).
23. CDC. Laboratory Biosafety Guidelines for working with SARS-CoV-2. COVID-19 https://www.cdc.gov/covid/php/lab/index.html (2024).
24. Kim, M. I. & Lee, C. Human Coronavirus OC43 as a low-risk model to study COVID-19. Viruses 15, 578 (2023).
25. Gallandat, K. & Lantagne, D. Selection of a Biosafety Level 1 (BSL-1) surrogate to evaluate surface disinfection efficacy in Ebola outbreaks: Comparison of four bacteriophages. PLoS One 12, e0177943 (2017).
26. Arvelo, I. et al. Investigation of a DNA tagged aerosol tracer method for In Situ evaluation of germicidal UV air cleaner effectiveness. Build. Environ. 262, 111828 (2024).
27. Blatchley, E. Dynamic responses of viral and DNA challenge agents in an IAQ. (2024).
28. Ratliff, K. M. et al. Factors affecting reduction of infectious aerosols by far-UVC and portable HEPA air cleaners. ACS EST Air (2025) doi:10.1021/acsestair.4c00247.
29. Hiwar, W. et al. Experimental analysis to quantify inactivation of microorganisms by Far-UVC irradiation in indoor environments. Build. Environ. 274, 112734 (2025).
30. Buonanno, M. et al. 222 nm far-UVC light markedly reduces the level of infectious airborne virus in an occupied room. Sci. Rep. 14, 6722 (2024).
31. Ma, B., Gundy, P. M., Gerba, C. P., Sobsey, M. D. & Linden, K. G. UV inactivation of SARS-CoV-2 across the UVC spectrum: KrCl* excimer, mercury-vapor, and light-emitting-diode (LED) sources. Appl. Environ. Microbiol. 87, e0153221 (2021).
32. Welch, D. et al. Far-UVC light: A new tool to control the spread of airborne-mediated microbial diseases. Sci. Rep. 8, 2752 (2018).
33. Lu, Y. H., Wang, R. X., Liu, H. L. & Lai, A. C. K. Evaluating the Performance of UV Disinfection across the 222–365 nm Spectrum against Aerosolized Bacteria and Viruses. Environ. Sci. Technol. (2024) doi:10.1021/acs.est.3c08675.
34. Welch, D. et al. Inactivation Rates for Airborne Human Coronavirus by Low Doses of 222 nm Far-UVC Radiation. Viruses 14, 684 (2022).
35. Kowalski, W. J., Bahnfleth, W. P., Raguse, M. & Moeller, R. The cluster model of ultraviolet disinfection explains tailing kinetics. J. Appl. Microbiol. 128, 1003–1014 (2020).
36. Kowalski, W. J., Bahnfleth, W. & Hernandez, M. T. A genomic model for predicting the ultraviolet susceptibility of viruses. IUVA news 11, 15–28 (2009).
37. Kowalski, W. J. Ultraviolet Germicidal Irradiation Handbook: UVGI for Air and Surface Disinfection. (Springer-Verlag, Heidelberg ; New York, 2009).
38. Peccia, J. & Hernandez, M. Photoreactivation in airborne Mycobacterium parafortuitum. Appl. Environ. Microbiol. 67, 4225–4232 (2001).
39. Peccia, J., Werth, H. M., Miller, S. & Hernandez, M. Effects of Relative Humidity on the Ultraviolet Induced Inactivation of Airborne Bacteria. Aerosol Sci. Technol. 35, 728–740 (2001).
40. Lin, K. & Marr, L. C. Humidity-Dependent Decay of Viruses, but Not Bacteria, in Aerosols and Droplets Follows Disinfection Kinetics. Environ. Sci. Technol. 54, 1024–1032 (2020).
41. Webb, S. J. Bound water in biological integrity. (1965).
42. Harper, G. J. Airborne micro-organisms: survival tests with four viruses. J. Hyg. (Lond.) 59, 479–486 (1961).
43. Buonanno, M., Welch, D., Shuryak, I. & Brenner, D. J. Far-UVC light (222 nm) efficiently and safely inactivates airborne human coronaviruses. Sci. Rep. 10, 10285 (2020).
44. Kitagawa, H. Inactivation of airborne SARS-CoV-2 by 222 nm far-UVC irradiation. https://www.youtube.com/watch?v=d0SKLqRgcWg (2023).
45. Lu, Y. H., Shi, X. R., Li, W. S. & Lai, A. C. K. Wavelength-specific inactivation mechanisms and efficacies of germicidal UVC for airborne human coronavirus. J. Hazard. Mater. 484, 136666 (2025).
46. Robinson, R. T., Mahfooz, N., Rosas-Mejia, O., Liu, Y. & Hull, N. M. UV222 disinfection of SARS-CoV-2 in solution. Sci. Rep. 12, 14545 (2022).
47. Schuit, M. A. et al. SARS-CoV-2 inactivation by ultraviolet radiation and visible light is dependent on wavelength and sample matrix. J. Photochem. Photobiol. B 233, 112503 (2022).
48. Ma, B. et al. Inactivation of coronaviruses and phage Phi6 from irradiation across UVC wavelengths. Environ. Sci. Technol. Lett. 8, 425–430 (2021).
49. Fletcher, L. A., Noakes, C. J., Beggs, C. B., Sleigh, P. A., & Kerr, K. G. The ultraviolet susceptibility of aerosolised microorganisms and the role of photoreactivation. in Proceedings of the 2nd International Congress on Ultraviolet Technologies (2003).
50. Nakamura, H. Sterilization efficacy of ultraviolet irradiation on microbial aerosols under dynamic airflow by experimental air conditioning systems. Bull. Tokyo Med. Dent. Univ. 34, 25–40 (1987).
51. Sharp, D. G. The Effects of Ultraviolet Light on Bacteria Suspended in Air. J. Bacteriol. 39, 535–547 (1940).
52. Ko, G., First, M. W. & Burge, H. A. Influence of relative humidity on particle size and UV sensitivity of Serratia marcescens and Mycobacterium bovis BCG aerosols. Tuber. Lung Dis. 80, 217–228 (2000).
53. Lai, K. M., Burge, H. A. & First, M. W. Size and UV germicidal irradiation susceptibility of Serratia marcescens when aerosolized from different suspending media. Appl. Environ. Microbiol. 70, 2021–2027 (2004).
54. VanOsdell, D. & Foarde, K. Defining the Effectiveness of UV Lamps Installed in Circulating Air Ductwork. https://www.osti.gov/servlets/purl/807360 (2002).
55. Hill, S. C., Doughty, D. C. & Mackowski, D. W. Inactivation of virions in host particles in air using 222- and 254-nm UV: Dependence of shielding on particle size and UV wavelength. Aerosol Sci. Technol. 58, 512–535 (2024).
56. Sutton, T. C. et al. Airborne transmission of highly pathogenic H7N1 influenza virus in ferrets. J. Virol. 88, 6623–6635 (2014).
57. Bagheri, G. et al. Size, concentration, and origin of human exhaled particles and their dependence on human factors with implications on infection transmission. J. Aerosol Sci. 168, 106102 (2023).
58. Monika, Damle, E. A., Kondabagil, K. & Kunwar, A. Comparative study of inactivation efficacy of far-UVC (222 nm) and germicidal UVC (254 nm) radiation against virus-laden aerosols of artificial human saliva. Photochem. Photobiol. (2025) doi:10.1111/php.14062.
59. Haddrell, A. et al. Ambient carbon dioxide concentration correlates with SARS-CoV-2 aerostability and infection risk. Nat. Commun. 15, 3487 (2024).
60. Fukushi, R. et al. Reactive oxygen species are involved in inhibition of photoreactivation of Staphylococcus aureus irradiated with 222-nm Far ultraviolet C. Photochem. Photobiol. (2025) doi:10.1111/php.14065.
61. Narita, K. et al. Reactive oxygen species generated by irradiation with bandpass-filtered 222-nm Far-UVC play an important role in the germicidal mechanism to Escherichia coli. Appl. Environ. Microbiol. e0188624 (2025).
62. Faulkner, C. A. et al. Comparison of effectiveness and energy use of airborne pathogen mitigation measures to meet clean air targets in a prototypical office building. Build. Environ. 257, 111466 (2024).
63. Welch, D. et al. Film dosimetry for occupant exposure monitoring within Far-UVC installations. Photochem. Photobiol. 101, 157–166 (2025).
64. Wood, K. Computer modeling of Far-UVC lights – skin safety, pathogen inactivation, and ozone production. https://www.youtube.com/watch?v=U5MFQe60giE (2023).
65. Rudnick, S. N. & First, M. W. Fundamental factors affecting upper-room ultraviolet germicidal irradiation - part II. Predicting effectiveness. J. Occup. Environ. Hyg. 4, 352–362 (2007).
66. Landry, S. A. et al. Ultraviolet radiation vs air filtration to mitigate virus laden aerosol in an occupied clinical room. J. Hazard. Mater. 487, 137211 (2025).
67. Luggables. CleanAirKits https://www.cleanairkits.com/products/luggables.
68. Blue Pure 211i Max. https://www.blueair.com/us/shop/air-purifiers/blue-pure-211i-max/3541.html?cgid=air-purifiers#prefn1=blair_recommendedRoomSize&prefv1=openspace&start=1.
69. Schirn, A. ANSI/ASA S12.60/Part 1-2010 (R2020): Acoustical Performance. The ANSI Blog https://blog.ansi.org/ansi-asa-s12-60-part-1-2010-r2020-school-acoustics/ (2023).
70. Belser, J. A., Pulit-Penaloza, J. A. & Maines, T. R. Ferreting out influenza virus pathogenicity and transmissibility: Past and future risk assessments in the ferret model. Cold Spring Harb. Perspect. Med. 10, a038323 (2020).
71. Zhou, J. et al. Defining the sizes of airborne particles that mediate influenza transmission in ferrets. Proc. Natl. Acad. Sci. U. S. A. 115, E2386–E2392 (2018).
72. Imai, M. et al. Experimental adaptation of an influenza H5 HA confers respiratory droplet transmission to a reassortant H5 HA/H1N1 virus in ferrets. Nature 486, 420–428 (2012).
73. Rockey, N. C. et al. Ventilation does not affect close-range transmission of influenza virus in a ferret playpen setup. Proc. Natl. Acad. Sci. U. S. A. 121, e2322660121 (2024).
74. Riley, R. L. et al. Infectiousness of air from a tuberculosis ward. Ultraviolet irradiation of infected air: comparative infectiousness of different patients. Am. Rev. Respir. Dis. 85, 511–525 (1962).
75. Escombe, A. R. et al. The infectiousness of tuberculosis patients coinfected with HIV. PLoS Med. 5, e188 (2008).
76. Dharmadhikari, A. S. et al. Natural infection of guinea pigs exposed to patients with highly drug-resistant tuberculosis. Tuberculosis (Edinb.) 91, 329–338 (2011).
77. Mphaphlele, M. et al. Institutional Tuberculosis Transmission. Controlled Trial of Upper Room Ultraviolet Air Disinfection: A Basis for New Dosing Guidelines. Am. J. Respir. Crit. Care Med. 192, 477–484 (2015).
78. Clark, S., Hall, Y. & Williams, A. Animal models of tuberculosis: Guinea pigs. Cold Spring Harb. Perspect. Med. 5, a018572 (2014).
79. Russell-Lodrigue, K. E., Zhang, G. Q., McMurray, D. N. & Samuel, J. E. Clinical and pathologic changes in a guinea pig aerosol challenge model of acute Q fever. Infect. Immun. 74, 6085–6091 (2006).
80. Lowen, A. C., Mubareka, S., Tumpey, T. M., García-Sastre, A. & Palese, P. The guinea pig as a transmission model for human influenza viruses. Proc. Natl. Acad. Sci. U. S. A. 103, 9988–9992 (2006).
81. Tyrrell, D. A. A view from the Common Cold Unit. Antiviral Res. 18, 105–125 (1992).
82. Carrat, F. et al. Time lines of infection and disease in human influenza: a review of volunteer challenge studies. Am. J. Epidemiol. 167, 775–785 (2008).
83. Atmar, R. L. et al. Norwalk virus shedding after experimental human infection. Emerg. Infect. Dis. 14, 1553–1557 (2008).
84. Teunis, P. F. M. et al. Norwalk virus: how infectious is it? J. Med. Virol. 80, 1468–1476 (2008).
85. Kirby, A. E., Streby, A. & Moe, C. L. Vomiting as a symptom and transmission risk in Norovirus illness: Evidence from human challenge studies. PLoS One 11, e0143759 (2016).
86. Bernstein, D. I. et al. Norovirus vaccine against experimental human GII.4 virus illness: a challenge study in healthy adults. J. Infect. Dis. 211, 870–878 (2015).
87. Patin, N. V. et al. The role of the gut microbiome in resisting Norovirus infection as revealed by a human challenge study. MBio 11, (2020).
88. Newman, K. L., Marsh, Z., Kirby, A. E., Moe, C. L. & Leon, J. S. Immunocompetent adults from human norovirus challenge studies do not exhibit norovirus viremia. J. Virol. 89, 6968–6969 (2015).
89. Zhou, J. et al. Viral emissions into the air and environment after SARS-CoV-2 human challenge: a phase 1, open label, first-in-human study. Lancet Microbe 4, e579–e590 (2023).
90. Gwaltney, J. M., Jr, Moskalski, P. B. & Hendley, J. O. Hand-to-hand transmission of rhinovirus colds. Ann. Intern. Med. 88, 463–467 (1978).
91. Evaluating Modes of Influenza Transmission (EMIT-2) using Innovative Technologies and Designs in Controlled Environments. https://reporter.nih.gov/search/cK_8orEFTkiFt2Lg8byE6w/project-details/10260845.
92. Nguyen-Van-Tam, J. S. et al. Minimal transmission in an influenza A (H3N2) human challenge-transmission model within a controlled exposure environment. PLoS Pathog. 16, e1008704 (2020).
93. Huang, J.-R. et al. Far-UVC light (222 nm) efficiently inactivates clinically significant antibiotic-resistant bacteria on diverse material surfaces. Microbiol. Spectr. e0425123 (2024).
94. Ponnaiya, B. et al. Far-UVC light prevents MRSA infection of superficial wounds in vivo. PLoS One 13, e0192053 (2018).
95. Glaab, J. et al. Skin tolerant inactivation of multiresistant pathogens using far-UVC LEDs. Sci. Rep. 11, 14647 (2021).
96. Welch, D. et al. Far-UVC light applications: sterilization of MRSA on a surface and inactivation of aerosolized influenza virus. in Light-Based Diagnosis and Treatment of Infectious Diseases (ed. Dai, T.) (SPIE, 2018). doi:10.1117/12.2309424.
97. Lorenzo-Leal, A. C., Tam, W., Kheyrandish, A., Mohseni, M. & Bach, H. Antimicrobial activity of filtered far-UVC light (222 nm) against different pathogens. Biomed Res. Int. 2023, 2085140 (2023).
98. Sun, Y., Wu, Q., Liu, J. & Wang, Q. Effectiveness of ultraviolet-C disinfection systems for reduction of multi-drug resistant organism infections in healthcare settings: A systematic review and meta-analysis. Epidemiol. Infect. 151, e149 (2023).
99. Sousa, M. et al. Far-UV-C irradiation promotes synergistic bactericidal action against adhered cells of Escherichia coli and Staphylococcus epidermidis. Sci. Total Environ. 917, 170352 (2024).
100. Gardner, A., Soni, A., Cookson, A. & Brightwell, G. Light tolerance of extended spectrum β-lactamase producing Escherichia coli strains after repetitive exposure to far-UVC and blue LED light. J. Appl. Microbiol. 134, lxad124 (2023).
101. Shoults, D. C. & Ashbolt, N. J. Decreased efficacy of UV inactivation of Staphylococcus aureus after multiple exposure and growth cycles. Int. J. Hyg. Environ. Health 222, 111–116 (2019).
102. Alcántara-Díaz, D., Breña-Valle, M. & Serment-Guerrero, J. Divergent adaptation of Escherichia coli to cyclic ultraviolet light exposures. Mutagenesis 19, 349–354 (2004).
103. Selveshwari, S., Lele, K. & Dey, S. Genomic signatures of UV resistance evolution in Escherichia coli depend on the growth phase during exposure. J. Evol. Biol. 34, 953–967 (2021).
104. Goldman, R. P. & Travisano, M. Experimental evolution of ultraviolet radiation resistance in Escherichia coli: Adaptation to ultraviolet radiation. Evolution 65, 3486–3498 (2011).
105. Tom, E. F., Molineux, I. J., Paff, M. L. & Bull, J. J. Experimental evolution of UV resistance in a phage. PeerJ 6, e5190 (2018).

4. Skin and Eye Safety

Summary

The acute harmful effects most associated with exposure to UV radiation are erythema (sunburn), photokeratitis (inflammation of the cornea), and DNA damage such as cyclobutane pyrimidine dimers (CPDs) and 6-4 photoproducts (6-4PPs). There is overwhelming evidence that the risks of all of these are significantly reduced for far-UVC wavelengths, as these wavelengths are absorbed by proteins in the outer layers of the skin and eyes, reducing penetration depth.

In addition to erythema, photokeratitis and CPDs/6-4PPs, there will be other biological effects induced by far-UVC, and our understanding of these effects is currently incomplete. Based on currently available data, there is no evidence suggesting that far-UVC has any effects that cause significant short-term or long-term harm within recommended exposure limits. Nevertheless, completing our understanding of the photobiological effects of far-UVC exposure is a clear research priority.

Crucial Considerations

• A recent comprehensive journal review concluded that the evidence supports using far-UVC within existing exposure limits, but that further studies are warranted.
• As compared to similar doses of UVB or conventional germicidal UVC radiation (i.e., 254 nm), far-UVC exposure to skin and eyes from filtered krypton chloride excimer (KrCl*) lamps, depending on dose, does not result in significant erythema (‘sunburn’), photokeratitis (‘sunburn in the cornea’) or DNA damage. However, many of the recent studies that have been influential in updating our understanding of the effects of far-UVC are from animal models, and there is need for additional human studies.
• Action spectra—relative biological effects of each individual wavelength—are critical to understanding the effects of exposure and avoiding overexposure.
• In situations where deeper layers of skin may be exposed in open wounds (for example, in a hospital’s operating theater), the relatively high concentration of protein-rich wound fluids and blood is thought to absorb the far-UVC, protecting underlying tissues.
• While it has not been directly studied, it is believed unlikely that the skin microbiome will be directly affected by far-UVC.
• The effects of protein absorption of far-UVC in the skin, eye, and tear film remain to be elucidated.
• Acute eye overexposure to far-UVC appears to have different effects than conventional germicidal UVC. Overexposure to conventional germicidal UVC causes a delayed onset of photokeratitis, which can be painful and last for many hours. In contrast, overexposure to far-UVC appears to cause a more immediate discomfort that subsides shortly after the exposure.
• There is some research into differences in the effect of far-UVC exposure depending on age, sex, and skin type, but further research is needed and some expert bodies are likely to want this information when updating exposure limits.

Analysis

This section contains many references of different UV doses to skin and eyes. To orient the reader, we provide below the recommended 8-hour exposure limits for different wavelengths of UV, as provided by expert bodies ICNIRP (the International Commission on Non-Ionizing Radiation Protection) and ACGIH (the American Conference of Governmental Industrial Hygienists).

Table 4.1. Recommended 8-Hour Exposure Limits for Example UVC Wavelengths

Note: ICNIRP makes no distinction between skin and eye, and is based on the photokeratitis action spectrum.

ICNIRP guidelines are intended to be applicable to both working populations and the general public, whereas ACGIH guidance is based on “levels of exposure and conditions under which it is believed that nearly all healthy workers may be repeatedly exposed, day after day, without adverse health effects.”

For more information on exposure limits, see Far-UVC primer section.

Skin Safety

Principles of Photobiological Safety in the Skin

The stratum corneum (SC) is the outermost layer of the skin, and is composed of dead, non-dividing cells filled with proteins and lipids (Figure 4.1). These cells form a tough, protective barrier that is regularly shed and replenished. The absence of nuclei in these cells means that there is no risk of the DNA damage and subsequent mutation normally associated with UV exposure, such as the formation of cyclobutane pyrimidine dimers (CPDs) or 6-4 photoproducts (6-4PPs).

Deeper in the epidermis, we find the nucleated but non-dividing stratum granulosum and stratum spinosum. In these strata, a continuous cycle of shedding and replenishment ensures that any damage incurred by far-UVC exposure is both minimal and short-lived. New cells are produced in the lower layers of the epidermis, move to the stratum corneum, and are removed. This process takes around 4 weeks in humans and 8–10 days in mice.

The depth of UV penetration depends on the wavelength. UV penetration depth into human tissues is substantially driven by protein absorption, which increases significantly at 230–240-nm wavelengths and below. At its simplest level, the reason that far-UVC exposure can be safe for humans while being deadly to pathogens is that pathogens are very small relative to the thickness of the protective protein-rich outer layers of human skin that attenuate far-UVC’s penetration to more sensitive tissues.

UVB radiation (280–320 nm) is particularly dangerous, as it can penetrate deeply into the skin while also containing enough energy to cause significant DNA damage. An estimated 1–10% of UVB can reach the basal cell layer, where keratinocytes actively divide and replicate. Melanocytes are also found in the basal cell layer, although these cells do not replicate as frequently as keratinocytes. DNA damage in these dividing cells can lead to mutations that may accumulate over time. DNA is double-stranded and UV damage can lead to breaks in one or both strands, requiring repair. DNA breaks and mutations can lead to skin cancer such as squamous cell carcinoma (see Evaluating cancer risk section). This contrasts with far-UVC, in which far less radiation penetrates beyond the outermost, non-dividing layers of the skin. This limited penetration is the key to the safety profile of far-UVC when compared to other forms of UV radiation.

UV radiation that does reach nucleated cells does not always cause lasting DNA damage, as biological organisms have evolved mechanisms to mitigate and repair the damage. Photoreactivation reverses CPDs through enzymatic action requiring blue-light. This can be found in certain members of archaea, bacteria, fungi, viruses, plants, invertebrates, and many vertebrates including aplacental mammals. Excision repair, either base excision repair (BER) or nucleotide excision repair (NER) replaces abnormal or damaged DNA bases. Double-stranded DNA breaks are repaired through homologous recombination (HR) or non-homologous end joining (NHEJ). HR is generally considered an error-free pathway since it requires a complementary DNA strand as a template. NHEJ is more error-prone as it joins broken strands independent of sequence homology.

Erythema—colloquially known as ‘sunburn’—is the primary acute response experienced in skin as a result of ultraviolet radiation exposure, typically appearing ≥ 24 hours after exposure. Erythema induction has not been observed in any humans exposed to a filtered KrCl* far-UVC lamp, even at a dose 38 times higher than the ACGIH 8-hour exposure limit at 222 or 233 nm. The inability of far-UVC to produce erythema, even at high doses is due to the protective effect of the outer layer of de-nucleated skin cells. A prior study that did find erythema used an unfiltered krypton chloride lamp, and it is now believed that the erythema was caused by the relatively small but consequential emissions in the non-far-UVC range. The erythema action spectrum previously did not extend below 250 nm, but new studies are expanding into the far-UVC range.

Using 3-D human skin culture models, even at a 222-nm exposure of 500 mJ/cm², only minimal formation of DNA damage immediately below the stratum corneum was observed. No increase in 6-4PPs was detected and only 4% of keratinocytes exhibited CPDs, which is significantly less than what would be expected from sun exposure on a typical day. Furthermore, human skin 3-D models irradiated with filtered 222-nm far-UVC had lower CPD yield than those exposed to unfiltered 222 nm or UVB.

3-D human skin models exposed to narrowband UV at 5-nm increments between 215 and 255 nm (100 mJ/cm²) revealed no DNA damage in the epidermis (stratum granulosum, stratum spinosum, and stratum basale) between 215–235 nm, while significant DNA damage was observed at higher wavelengths (240–255 nm) in all layers, except the stratum basale at 240 nm. Studies at 233 nm on reconstructed models, ex vivo human skin, and healthy volunteers also support the safety of far-UVC.

Acute exposure to far-UVC did not cause significant numbers of CPDs in animal studies. Compared to control, only 10,000 mJ/cm² or higher of 222 nm resulted in higher levels of CPDs in the outermost epidermal layer of hairless albino mice or Xpa-knockout mice. When the UVC source was triple filtered, the damage from 10,000 mJ/cm² was greatly reduced, suggesting that the most damage was induced by traces of longer wavelength UVC (>235 nm). Xpa mice, which have a genetic defect similar to the human DNA repair disease Xeroderma Pigmentosum, are highly susceptible to UV-induced DNA damage and cancer, and would be more likely to show adverse skin damage from far-UVC exposure.

It is uncertain to what extent we can extrapolate findings after acute exposures to those arising from prolonged or chronic irradiation, or extrapolate data from animal experiments to humans. Only a few long-term animal studies have been published, and no long-term human data has been reported. One study exposed mice to 222-nm far-UVC for 8 hours a day, five times a week, for 66 weeks and found no induction of skin cancer or other skin abnormalities.

Taking the in vitro 3-D skin culture and animal studies together, it is thought that a boundary in the 230–240 nm range exists where there is a sudden and steep decrease in UV penetration depth. This decrease would align with an inability to cause CPD or 6-4PP DNA damage in proliferating cells, with the boundary marking the difference between superficial effects and the more serious effects of the more deeply penetrating UVC wavelengths. More refined action spectra inferred from studies with greater discrimination in the far-UVC range and in the transition region between far-UVC and conventional UVC (for example, through the use of a monochromator to produce emissions of specific wavelengths) are critical for providing accurate ‘Lamp Exposure Limits’ and preventing overexposure. This is discussed in more detail in the Guidance, standards, and regulations section.

UV Responses in Different Populations

Epidermal skin thickness depends on multiple factors including age, body location, and demographics. For example, infants aged 3–24 months have a SC 20–30% thinner than adults. The epidermis and SC are thicker on parts of the body exposed to the sun compared to those protected from sunlight. Among normally exposed parts of the body, the palms have the thickest SC while the SC is thinnest on the face and eyelids. The effects of far-UVC exposure on mucosal surfaces, such as the lower lip, is poorly defined. Individuals with defects in the DNA damage/repair responses, or those with polymorphisms that decrease these processes, may be more vulnerable to damaging effects from far-UVC than the general population.

Skin type and melanin concentration may influence the impact of far-UVC, as melanin provides photoprotection by absorbing broadband UV and scavenging radicals. This antioxidant property reduces UVR penetration into keratinocyte nuclei, offering greater protection for individuals with higher melanin concentration. Experiments employing far-UVC are essential to determine any potential impacts, if any, of melanin concentration in exposure to far-UVC.

Studies have examined the relationship between UV-induced DNA damage and melanin content/distribution in ex vivo human skin exposed to 222-nm and 233-nm UVC. Generally, lower levels of DNA damage were observed in darker-skinned donors. In contrast, similar studies using reconstructed human epidermis (RHE) models exposed to 233-nm UVC produced opposite results, with tanned RHE showing higher levels of UV-mediated free radicals and DNA damage compared to lighter RHE. Nevertheless, as the damage caused by 233-nm UVC is consistently lower than that produced by UVB at doses considered safe for skin, skin color is not regarded as a significant risk factor for this exposure. In vivo studies on healthy humans using 233 nm are ongoing, and a similar experiment should be conducted at 222 nm using normal exposure levels to ascertain any potential impact of melanin concentration.

Other Photobiological Effects of UV on Skin

In addition to erythema, there are alternative measurable outcomes that may be relevant for far-UVC given the differences in its properties compared to UVA, UVB and conventional UVC radiation. For this reason, it is a priority to conduct observational and mechanistic studies to better understand and identify outcome biomarkers and endpoints. Inclusion of biomarkers of DNA damage other than CPDs and 6-4PPs, such as ɣh2AX foci for double-strand breaks, is valuable. It’s crucial to include this information in larger, longer-term safety studies in diverse populations to create more informative action spectra.

Skin Yellowing

Visible yellow discoloration was observed on the forearm skin of one investigator after a dose of at least 6,000 mJ/cm² (which is more than 10x higher than the current 8-hr ACGIH exposure limit at 222 nm). However, this appeared to be localized to the stratum corneum, as evidenced by the fact that skin tape stripping reduced the discoloration immediately. Zamudio Díaz et al., 2023 also observed skin yellowing 1 hour after exposure to 60 mJ/cm² 233 nm, with recovery after 24 hours.

The mechanism behind this phenomenon is not completely understood, but UV-induced aldehyde production or protein oxidation is possible. Lipid peroxidation and yellowing are observed in solar skin photoaging, suggesting a potential mechanism. The yellowing is not thought to be a byproduct of erythema or to result from DNA damage. No erythema has been reported in human subjects, even after receiving high doses. The effect should also be observed in more than one subject. Understanding the underlying cause of this change in the stratum corneum and creating an action spectrum for that effect should be a priority.

Immune Suppression

Naturally occurring trans-urocanic acid (t-UCA), highly concentrated in the stratum corneum, acts as an endogenous chromophore. This absorbs UV, leading to conversion to the cis-form (c-UCA), and may have an immunosuppressive effect. The action spectrum for c-UCA production is primarily in the UVB range. c-UCA may also contribute to UVA-induced photoaging, following UVA exposure, through UVA sensitization. However, the direct mechanisms are not known.

While primarily linked to immune suppression, c-UCA has a multitude of effects, including potential links to skin disorders such as atopic dermatitis and urticaria. It remains to be seen if far-UVC will induce c-UCA formation or any of the downstream impacts at all or at the doses used for disinfection. Furthermore, while UVR generally suppresses acquired immunity, there is some evidence that UVR enhances innate immunity by producing antimicrobial peptides and activating immune cells. Whether this is the case for far-UVC remains an open question.

Langerhans cells are distributed throughout the epidermis and recognize pathogens/molecular damage in the skin. UVB damage to these cells induces inflammatory responses, leading to changes in how the skin processes immune responses as well as immunosuppression. While not studied yet for 222 nm, 60 mJ/cm² of 233-nm far-UVC showed no significant difference in langerin, a marker for langerhans cells, compared to controls.

Microbiome

The hands, arms, neck, eyes, and face are the skin areas most exposed under typical deployment scenarios. The stratum corneum is one site of the skin microbiome, and the hair follicles deep in the skin are one site of the reservoirs for the microbiome. Surface disinfection with antimicrobial agents, such as alcohol sanitizers or far-UVC, will most likely only have a transient impact on the microbiome since they don’t penetrate to the follicles.

It is expected that proximity, exposure duration, number of repeated exposures, and wavelength all may have an impact on the level of microbiome transient disinfection. Investigating the cause of the skin yellowing in the stratum corneum is also important for understanding any impacts of far-UVC exposure on the skin microbiome. Whether there are microbiome impacts depends on the molecular cause of the yellowing and whether this can be spread between layers of the skin or to microbes.

Wounds and Burns

One important question is the safety of far-UVC exposure on breaks in the stratum corneum, such as wounds and burns. Hospital-acquired infections remain a serious public health concern, with ~4% of hospitalized patients experiencing a healthcare-associated infection (HAI). With the rise of antibiotic resistance and increased HAI, disinfection of wounds and burns is another area where far-UVC could be effective, although evaluation of this potential contribution is outside the scope of this report.

The original idea for far-UVC use for reducing surgical site infections focused on using the lamps in surgical suites, with the lamps facing downward from the ceiling, potentially irradiating the surgical wound from above. Far-UVC skin exposure limits are higher because of absorption in the stratum corneum, where the cells have no nuclei. In burns and wounds, on the other hand, the stratum corneum is often disrupted, exposing underlying layers of cells that do contain nuclei and therefore could be damaged by far-UVC. However, wounds are often covered with a protein-rich wound fluid which is likely to absorb significant amounts of far-UVC.

The use of 233-nm LEDs as well as 222-nm KrCl* lamps has been evaluated in vitro (outside the body) and in vivo (in animal bodies) for wound disinfection. Initial studies revealed disinfection of superficial wounds on the backs of mice infected with MRSA, with minimal CPD induction, at 40 mJ/cm² 222-nm far-UVC. We’d expect 233-nm far-UVC to be more effective due to penetrating slightly deeper than 222 nm, at the cost of increased CPDs in the upper epidermis. However, in one study, 233 nm was shown to be more effective at disinfection despite producing minimal to no CPDs at the wound site. For superficial surgical site infections and skin infections, 222-nm disinfection appears to be efficacious, with little to no DNA damage.

Key Takeaways

• Cumulatively, current evidence suggests little to no harm to the skin from acute far-UVC exposure at a microbicidal/bactericidal dose.
• The outer layer of the skin is composed of non-dividing cells that are regularly shed and replaced. Far-UVC is absorbed by this outer layer, which means it is likely to be less harmful than e.g. UVA and UVB, which penetrate deeper.
• However, the paucity of human studies, chronic exposure studies, and investigation of alternative damage mechanisms renders an incomplete understanding that calls for further research into these areas.

Eye Safety

Principles of Photobiological Safety in the Eye

The tear film is the first part of the eye exposed to UV. Tears are an extremely important feature of the ocular surface anatomy, providing a uniform optical surface to minimize aberrations, washing away cellular and external debris, and containing a variety of macromolecules that both nourish the underlying epithelial cell layers as well as proteins designed to perform antimicrobial functions.

The tear film is composed of three layers. The outermost layer is very thin (40–160 nm) and mainly composed of lipids that help prevent evaporation from the underlying and much thicker (3–5 µm) middle aqueous layer. This aqueous layer contains hundreds of proteins and small macromolecules, and its composition can change depending on external stimuli and emotional state. The innermost layer, whose function is to affix the tears to the outer layer of corneal and conjunctival epithelium is a 0.3 µm mucin layer.

Beneath the tear film, the cornea and surrounding conjunctiva are the outward-facing eye structures which interact with light (see Figure 4.3 below). Experimental measurements indicate that the human cornea absorbs wavelengths below 295 nm. Longer wavelengths of UV radiation, however, may reach underlying structures such as the iris and lens. The outermost layers of the corneal epithelium are thought to provide protection similar to the skin stratum corneum, with surface cells being replenished roughly every 7 days.

Unlike the SC, corneal epithelial cells are nucleated. Therefore, DNA damage, including CPDs and 6-4PPs, can be observed in the outermost layer of corneal epithelial cells after relatively low levels of far-UVC exposure. The cornea is also heavily innervated, with nerve endings terminating in the corneal epithelium and with a wide range of sensory functions.

However, these corneal epithelial cells are post-mitotic (non-dividing) and quickly replaced; therefore any DNA damage in these cells is unlikely to cause concern. In contrast, damage to stem cell progenitors in the limbus (at the edge of the cornea), or to basal corneal epithelial cells, might prove more problematic. Nevertheless, no CPDs were observed in limbal stem cells following far-UVC irradiation from a KrCl* lamp at 1,500 mJ/cm², but CPDs were observed at 2,500 mJ/cm² or higher.

Most conjunctival stem cells are found in the conjunctival fornix, which is located under the eyelids, suggesting that these cells are mostly protected from UV radiation. It is undetermined whether goblet cells, which lie near the surface of the conjunctival epithelium and produce mucin, are at risk for far-UVC-induced damage.

Excess exposure to either UVA or UVB is associated with various pathologies of the eye. Both are linked to cataract formation, while UVB is more strongly associated with epithelial metaplasia and overgrowth, called pterygium and pinguicula, and with photokeratitis. UVA can penetrate more deeply into the eye, even reaching the retina where it is associated with macular degeneration, whereas most UVB is absorbed by the cornea and lens. For these reasons, it is important to wear UV-blocking eyewear (that blocks both UVA and UVB) in bright sunlight to reduce excess risk of ocular damage.

The potential for the tear film to attenuate far-UVC absorbance has been the subject of debate. More than 20 peer-reviewed far-UVC papers have claimed that far-UVC is absorbed by the tear film, preventing far-UVC-induced damage to underlying corneal cells and tissue. Similar to the stratum corneum of the skin, the tear film contains a relatively high concentration of macromolecules such as proteins, lipids, and sugars, which were thought to attenuate far-UVC penetration to underlying epithelial cells. However, recent studies have shown some CPD formation in the first one or two layers of corneal epithelial cells following far-UVC irradiation in experimental animal studies.

Additionally, a study suggests that the tear film transmits >90% of far-UVC wavelengths between 220–230 nm and that 222-nm doses as low as 3 mJ/cm² lead to reduced viability of cultured human corneal epithelial cells, despite overlay with human tears or tear substitutes of equal protein composition and thickness as in vivo tears. One caveat is that in vitro studies do not always replicate in vivo studies due to the difficulty in replicating an intact tear film containing all three layers.

However, physics (Beer’s Law) suggests that the tear film will absorb little far-UVC, given its known thickness and known average protein concentration. Some therefore posit that the tear film provides little protection and most far-UVC reaches the outermost epithelial cell layers, where it is absorbed primarily by the proteins in those cells but will also cause some amount of DNA damage. In contrast, the stratum corneum in the skin is de-nucleated, making it impossible for DNA damage to occur in those cells where the majority of far-UVC is absorbed.

Photokeratitis and the Causes of Eye Irritation

Pitts reported far-UVC-induced ocular effects in rabbits, primates, and humans, including development of photokeratitis, decreases in visual acuity, and biomicroscopic changes in the cornea. While data from Pitts et al., 1974 is commonly referred to as the photokeratitis action spectrum in humans, implying that the recommended thresholds for human exposure derived from this data are based on avoiding inducing photokeratitis, transient discomfort that presented differently from photokeratitis was observed at the threshold level for wavelengths in the far-UVC region.

For wavelengths above 250 nm, epithelial granules, or small, white, discrete, round spots located in the epithelium were observed. In contrast, wavelengths below 250 nm produced fine, discrete granules, and some subjects reported symptoms including tearing up and a foreign body sensation. Across the studied wavelengths, photophobia (abnormal sensitivity to light) was also variable in presentation and was considered an unreliable indicator of photokeratitis. All subjects who were classified as exhibiting the threshold for photokeratitis demonstrated loss of visual acuity that returned to baseline within six hours. These observations are consistent with those reported in more recent human studies on far-UVC ocular effects.

Research on the acute effects of UV on the eyes usually focuses on the cornea, and photokeratitis specifically refers to inflammation of the cornea. However, the cornea is not the only exposed structure. The conjunctival epithelium is 3–5 cell layers thick. It covers the sclera and lines the inside of the eyelid. Interspersed among conjunctival epithelial cells are goblet cells, whose function is to secrete mucins, an important component of the posterior tear film layer.

In addition to goblet cells, other cell types can be found in the conjunctival epithelium, including numerous immunomodulatory cells that migrate through the conjunctiva and stroma. Direct UV irradiation or indirect inflammatory/oxidative stress may impact distribution and function of these cells. Similar to the corneal epithelium, far-UVC radiation may be absorbed by the initial layers of the conjunctival epithelium with unknown adverse health outcomes.

One measure of conjunctival irritation/damage is hyperemia, when the sclera appears red due to blood vessel dilation in the conjunctiva. Sugihara et al., 2024 reported onset of dry eye, conjunctival hyperemia, and mild pain 2–2.5 hours after exposure to far-UVC in 5 participants which disappeared 4.5–11 hours afterwards, employing doses of 22, 50, or 75 mJ/cm². This observation is similar to that reported by Pitts where exposure to 210–230 nm bandwidth radiation resulted in subjective discomfort and conjunctival hyperemia that lessened by 4 hours and disappeared after 6 hours.

While the underlying mechanism is unclear, suggested hypotheses include tear film insufficiency, stimulation or damage to the corneal epithelial nerve plexus, and photo-oxidative production of free radicals in tears. It is critical that this phenomenon be investigated. Documentation of corneal irritation, the perception or sensation of ‘tingling’ at the ocular surface, geographic loss of sensitivity (tested via esthesiometry), and other hallmarks of corneal nerve damage, should be evaluated endpoints in future eye studies.

In terms of the consequences of eye irritation, Sliney & Stuck, 2021 noted that an eye sensation of discomfort at higher irradiances may actually reduce the risk of more serious eye exposure because the eyes will automatically close as a reflex to protect them from damage (in the same way that our natural aversion prevents us from looking directly at the sun or bright lights). Whether or not this is the case depends on the circumstances under which these sensations arise and the underlying biological mechanisms, both of which are not currently well-understood.

Individuals with dry eye or other ocular surface disease may be more vulnerable to far-UVC-induced damage to the corneal and conjunctival epithelium. Given the close proximity of mucin-producing goblet cells to the conjunctival surface, damage to these cells and further loss of tear film adhesion to the ocular surface may exacerbate dry eye symptoms.

Animal Studies

Multiple animal studies on the ocular effects of exposure to 207-, 222-, 235-, 254-, and 311-nm UV in rats have been conducted in recent years. Animal studies permit deeper investigation into the effects of UV on the eye, as conducting the same assays in humans would be either impossible or extremely invasive.

Rat eyes exposed to doses of filtered far-UVC (207 and 222 nm), that do induce effects in the case of longer wavelength UV, revealed no significant change in corneal surface integrity, as measured by fluorescein biomicroscopy, histology, and imaging. Higher exposures (>3,500 mJ/cm²) resulted in measurable damage to the cornea. At supra-threshold doses (1,000 mJ/cm²), both 207 and 222 nm exposure resulted in CPD-positive cells in the most superficial layers of the corneal epithelium and the superior limbus, but not in the inferior limbus. Corneas exposed to 235 nm exhibited CPDs up to the middle layer of the corneal epithelium. Exposure to 254 nm, emitted by a mercury lamp, and 311-nm UVB resulted in CPD-positive cells throughout the corneal epithelium.

More recently, Arden et al., 2024 reported no evidence for changes in visual acuity or contrast sensitivity, assessed through optokinetic methodologies, intraocular pressure, or slit lamp detectable corneal damage in a SKH-1 hairless, immuno-competent mouse model chronically exposed to three different intensities of far-UVC over a 66-week period. One limitation of this finding is that exposure was limited to daylight hours when mice were least active.

Studies of Actual Installations

In a human exposure study, Kousha et al., 2024 simulated an office environment where subjects received a range of doses from 0 to 50 mJ/cm² to the top of the head, five hours per day, over a three-day period. Exposure intensity varied by participant position, with most receiving less than 20 mJ/cm² to the top of the head. Eye dose was not measured.

Subjects were queried for self-reported eye discomfort including dry eye, soreness, and fatigue, immediately after and up to one week following exposure. Besides tracking CPDs and other markers of damage, Kousha utilized the Standard Patient Evaluation Eye Dryness (SPEED) and the Ocular Surface Disease Index (OSDI) questionnaires before and after exposure to 222 nm and found no significant change, even with head exposures up to 50 mJ/cm².

Most recently, Sugihara reported no changes in corneal erosion, conjunctival hyperemia, pterygium, or cataract following chronic occupational exposure to 222 nm. Some of these findings are what we would expect, as UV radiation would not be expected to reach the lens (increasing the risk of cataracts) and pterygium is most strongly associated with UVB exposure. Furthermore, the authors reported that the subjects spent less than one hour per day under far-UVC illumination and estimated the daily dose as less than 2.8 mJ/cm². The authors calculated a maximum theoretical far-UVC exposure of 6.4 mJ/cm² as the daily dose theoretically received by a 170-cm tall subject staring at the lamp for eight hours.

The Effect of Installation Design on Dose to Eyes

In controlled studies and in the recommended photobiological exposure limits, the ‘dose to eyes’ is intended to represent the actual UV dose that reaches the surface of the eye. However, taking irradiance measurements at eye level without factoring in the particular geometry of the human face can significantly overestimate true eye exposure. For example, in one study using a manikin, only 5.8% of the maximum directly measured dose on the manikin’s head, nose and lip was measured on the eye surface.

Furthermore, natural anatomical features absent in manikins, such as eyebrows, eyelids, and the top of the head, further shield the eyes from direct exposure to overhead lighting fixtures.

Taken together, these considerations support the idea that not mounting far-UVC devices on the wall at eye level, and ideally mounting them on the ceiling, likely provides a significant ‘safety margin’ for eye exposure. The interpretation of photobiological exposure limits incorporated into the UL 8802 standard reflects this fact.

Key Takeaways

• The outer layers of the eye are mainly composed of non-dividing cells that are regularly shed and replaced. Far-UVC is absorbed by this outer layer, which means that it is likely to be less harmful than e.g. UVA and UVB, which penetrate deeper.
• While the cornea is the usual site for studying far-UVC’s effects on the eye, the effects on other ocular structures such as the conjunctiva should be included in future research studies.
• Non-traditional outcome measures are important to develop and assess, as a complete safety profile across different body structures has not been established.
• While initially thought to confer full protection, recent studies suggest that the eye’s tear film layer does not significantly attenuate far-UVC radiation.
• The effects of absorption of far-UVC in the outer layer of the human corneal epithelium, especially with respect to anecdotal reports of corneal discomfort after irradiation, are under active investigation.

Further Reading

• Assessing the safety of new germicidal far-UVC technologies
• Extreme Exposure to Filtered Far-UVC: A Case Study
• One-year Ocular Safety Observation of Workers and Estimations of Microorganism Inactivation Efficacy in the Room Irradiated with 222-nm Far Ultraviolet-C Lamps

References

1. Görlitz, M. et al. Assessing the safety of new germicidal far-UVC technologies. Photochem. Photobiol. 100, 501–520 (2024).
2. TLV Physical Agents Introduction. ACGIH. https://www.acgih.org/science/tlv-bei-guidelines/tlv-physical-agents-introduction/ (2022).
3. Tagami, H. Stratum corneum cell layers. in Textbook of Aging Skin 405–413 (Springer Berlin Heidelberg, Berlin, Heidelberg, 2017).
4. Busch, L. et al. Evaluation of DNA lesions and radicals generated by a 233 nm far-UVC LED in superficial ex vivo skin wounds. J. Photochem. Photobiol. B 245, 112757 (2023).
5. Bruls, W. A., Slaper, H., van der Leun, J. C. & Berrens, L. Transmission of human epidermis and stratum corneum as a function of thickness in the ultraviolet and visible wavelengths. Photochem. Photobiol. 40, 485–494 (1984).
6. Boyce, S. T. et al. Assessment of replication rates of human keratinocytes in engineered skin substitutes grafted to athymic mice. Wound Repair Regen. 20, 544–551 (2012).
7. Rastogi, R. P., Richa, Kumar, A., Tyagi, M. B. & Sinha, R. P. Molecular mechanisms of ultraviolet radiation-induced DNA damage and repair. J. Nucleic Acids 2010, 592980 (2010).
8. Eadie, E., Barnard, I. M. R., Ibbotson, S. H. & Wood, K. Extreme exposure to filtered far-UVC: A case study. Photochem. Photobiol. 97, 527–531 (2021).
9. Zamudio Díaz, D. F. et al. Skin optical properties from 200 to 300 nm support far UV-C skin-safety in vivo. J. Photochem. Photobiol. B 247, 112784 (2023).
10. O’Mahoney, P., Wood, K., Ibbotson, S. H. & Eadie, E. Potential harm to the skin from unfiltered krypton chloride ‘far-ultraviolet-C’ lamps, even below an occupational exposure limit. J Radiol Prot 42, (2022).
11. Eadie, E., O’Mahoney, P., Ibbotson, S. H., Miller, C. C. & Wood, K. Far-UVC: The impact of optical filters on real-world deployment. Photochem. Photobiol. (2024) doi:10.1111/php.14005.
12. Woods, J. A. et al. The effect of 222-nm UVC phototesting on healthy volunteer skin: a pilot study. Photodermatol. Photoimmunol. Photomed. 31, 159–166 (2015).
13. Buonanno, M. et al. Extending the acute skin response spectrum to include the far-UVC. bioRxiv (2024) doi:10.1101/2024.07.19.604306.
14. Buonanno, M., Welch, D. & Brenner, D. J. Exposure of human skin models to KrCl excimer lamps: The impact of optical filtering. Photochem. Photobiol. 97, 517–523 (2021).
15. Ivanova, I. et al. Improved spectral purity of 222-nm irradiation eliminates detectable cyclobutylpyrimidine dimers formation in skin reconstructs even at high and repetitive disinfecting doses. Photochem. Photobiol. 98, 1149–1156 (2022).
16. Welch, D., Aquino de Muro, M., Buonanno, M. & Brenner, D. J. Wavelength-dependent DNA photodamage in a 3-D human skin model over the far-UVC and germicidal UVC wavelength ranges from 215 to 255 nm. Photochem. Photobiol. 98, 1167–1171 (2022).
17. Zwicker, P. et al. Application of 233 nm far-UVC LEDs for eradication of MRSA and MSSA and risk assessment on skin models. Sci. Rep. 12, 2587 (2022).
18. Buonanno, M. et al. 207-nm UV light-A promising tool for safe low-cost reduction of surgical site infections. II: In-vivo safety studies. PLoS One 11, e0138418 (2016).
19. Buonanno, M. et al. Germicidal efficacy and mammalian skin safety of 222-nm UV light. Radiat. Res. 187, 483–491 (2017).
20. Yamano, N. et al. Evaluation of acute reactions on mouse skin irradiated with 222 and 235 nm UV-C. Photochem. Photobiol. 97, 770–777 (2021).
21. Fukui, T. et al. Safety of 222 nm UVC irradiation to the surgical site in a rabbit model. Photochem. Photobiol. 98, 1365–1371 (2022).
22. Narita, K., Asano, K., Morimoto, Y., Igarashi, T. & Nakane, A. Chronic irradiation with 222-nm UVC light induces neither DNA damage nor epidermal lesions in mouse skin, even at high doses. PLoS One 13, e0201259 (2018).
23. Tavares, R. S. N. et al. Different biological effects of exposure to far-UVC (222 nm) and near-UVC (254 nm) irradiation. J. Photochem. Photobiol. B 243, 112713 (2023).
24. Yamano, N. et al. Long-term Effects of 222-nm ultraviolet radiation C Sterilizing Lamps on Mice Susceptible to Ultraviolet Radiation. Photochem. Photobiol. 96, 853–862 (2020).
25. Welch, D. et al. No evidence of induced skin cancer or other skin abnormalities after long-term (66 week) chronic exposure to 222-nm far-UVC radiation. Photochem. Photobiol. 99, 168–175 (2023).
26. Stamatas, G. N., Roux, P.-F., Boireau-Adamezyk, E., Lboukili, I. & Oddos, T. Skin maturation from birth to 10 years of age: Structure, function, composition and microbiome. Exp. Dermatol. 32, 1420–1429 (2023).
27. Czekalla, C., Schönborn, K. H., Lademann, J. & Meinke, M. C. Noninvasive determination of epidermal and stratum corneum thickness in vivo using two-photon microscopy and optical coherence tomography: Impact of body area, age, and gender. Skin Pharmacol. Physiol. 32, 142–150 (2019).
28. Zamudio Díaz, D. F. et al. Significance of melanin distribution in the epidermis for the protective effect against UV light. Sci. Rep. 14, 3488 (2024).
29. Buonanno, M. et al. Wavelength-dependent DNA damage induced by single wavelengths of UVC light (215 to 255 nm) in a human cornea model. Res. Sq. (2024) doi:10.21203/rs.3.rs-5129114/v1.
30. Pizzimenti, S. et al. Interaction of aldehydes derived from lipid peroxidation and membrane proteins. Front. Physiol. 4, 242 (2013).
31. Hickerson, R. P. et al. Minimal, superficial DNA damage in human skin from filtered far-ultraviolet C. Br. J. Dermatol. 184, 1197–1199 (2021).
32. Fukui, T. et al. Exploratory clinical trial on the safety and bactericidal effect of 222-nm ultraviolet C irradiation in healthy humans. PLoS One 15, e0235948 (2020).
33. Zucchi, H. et al. Assessing the role of carbonyl adducts, particularly malondialdehyde adducts, in the development of dermis yellowing occurring during skin photoaging. Life (Basel) 12, 403 (2022).
34. Negre-Salvayre, A. & Salvayre, R. Post-translational modifications evoked by reactive carbonyl species in ultraviolet-A-exposed skin: Implication in fibroblast senescence and skin photoaging. Antioxidants (Basel) 11, 2281 (2022).
35. Garssen, J., Norval, M., Crosby, J., Dortant, P. & Van Loveren, H. The role of urocanic acid in UVB-induced suppression of immunity to Trichinella spiralis infection in the rat. Immunology 96, 298–306 (1999).
36. Young, A. R. Chromophores in human skin. Phys. Med. Biol. 42, 789–802 (1997).
37. Hart, P. H. & Norval, M. The multiple roles of urocanic acid in health and disease. J. Invest. Dermatol. 141, 496–502 (2021).
38. Patra, V., Laoubi, L., Nicolas, J.-F., Vocanson, M. & Wolf, P. A perspective on the interplay of ultraviolet-radiation, skin microbiome and skin resident memory TCRαβ+ cells. Front. Med. (Lausanne) 5, 166 (2018).
39. Hatakeyama, M. et al. Anti-inflammatory role of Langerhans cells and apoptotic keratinocytes in ultraviolet-B-induced cutaneous inflammation. J. Immunol. 199, 2937–2947 (2017).
40. Fukunaga, A. et al. Langerhans cells serve as immunoregulatory cells by activating NKT cells. J. Immunol. 185, 4633–4640 (2010).
41. Grice, E. A. & Segre, J. A. The skin microbiome. Nat. Rev. Microbiol. 9, 244–253 (2011).
42. Magill, S. S. et al. Changes in prevalence of health care-associated infections in U.S. hospitals. N. Engl. J. Med. 379, 1732–1744 (2018).
43. Buonanno, M. et al. 207-nm UV light - a promising tool for safe low-cost reduction of surgical site infections. I: in vitro studies. PLoS One 8, e76968 (2013).
44. Wade, C. et al. Loss of protein, immunoglobulins, and electrolytes in exudates from negative pressure wound therapy. Nutr. Clin. Pract. 25, 510–516 (2010).
45. Ponnaiya, B. et al. Far-UVC light prevents MRSA infection of superficial wounds in vivo. PLoS One 13, e0192053 (2018).
46. Narita, K. et al. Disinfection and healing effects of 222-nm UVC light on methicillin-resistant Staphylococcus aureus infection in mouse wounds. J. Photochem. Photobiol. B 178, 10–18 (2018).
47. Masoudi, S. Biochemistry of human tear film: A review. Exp. Eye Res. 220, 109101 (2022).
48. Werkmeister, R. M. et al. Measurement of tear film thickness using ultrahigh-resolution optical coherence tomography. Invest. Ophthalmol. Vis. Sci. 54, 5578–5583 (2013).
49. Mallet, J. D., Gendron, S. P., Drigeard Desgarnier, M.-C. & Rochette, P. J. Implication of ultraviolet light in the etiology of uveal melanoma: A review. Photochem. Photobiol. 90, 15–21 (2014).
50. Rönkkö, S., Vellonen, K.-S., Järvinen, K., Toropainen, E. & Urtti, A. Human corneal cell culture models for drug toxicity studies. Drug Deliv. Transl. Res. 6, 660–675 (2016).
51. Hanna, C., Bicknell, D. S. & O’brien, J. E. Cell turnover in the adult human eye. Arch. Ophthalmol. 65, 695–698 (1961).
52. Dua, H. S., Gomes, J. A. & Singh, A. Corneal epithelial wound healing. Br. J. Ophthalmol. 78, 401–408 (1994).
53. Al-Aqaba, M. A., Fares, U., Suleman, H., Lowe, J. & Dua, H. S. Architecture and distribution of human corneal nerves. Br. J. Ophthalmol. 94, 784–789 (2010).
54. Kaidzu, S. et al. Safety evaluation of Far-UV-C irradiation to epithelial basal cells in the corneal limbus. Photochem. Photobiol. 99, 1142–1148 (2023).
55. DelMonte, D. W. & Kim, T. Anatomy and physiology of the cornea. J. Cataract Refract. Surg. 37, 588–598 (2011).
56. Shumway, C. L., Motlagh, M. & Wade, M. Anatomy, head and neck, eye conjunctiva. in StatPearls (StatPearls Publishing, Treasure Island (FL), 2024).
57. Kaidzu, S. et al. Evaluation of acute corneal damage induced by 222-nm and 254-nm ultraviolet light in Sprague-Dawley rats. Free Radic. Res. 53, 611–617 (2019).
58. Blatchley, E. R. et al. Far UV-C Radiation: Current State-of Knowledge. https://iuva.org/UV-C-Radiation-White-Paper (2021).
59. Hessling, M., Haag, R., Sieber, N. & Vatter, P. The impact of far-UVC radiation (200-230 nm) on pathogens, cells, skin, and eyes - a collection and analysis of a hundred years of data. GMS Hyg. Infect. Control 16, Doc07 (2021).
60. Kaidzu, S. et al. Re-evaluation of rat corneal damage by short-wavelength UV revealed extremely less hazardous property of far-UV-C. Photochem. Photobiol. 97, 505–516 (2021).
61. Folkerts, I. T., Arden, P. C., Jerome, E. H., Brenner, D. J. & Kleiman, N. J. Far-UVC (222 nm) light is not significantly absorbed by the human tear film. Invest Ophthalmol Vis Sci. under review.
62. Pitts, D. G. The human ultraviolet action spectrum. Am. J. Optom. Physiol. Opt. 51, 946–960 (1974).
63. Pitts, D. G. The ocular ultraviolet action spectrum and protection criteria. Health Phys. 25, 559–566 (1973).
64. Sugihara, K., Kaidzu, S., Sasaki, M. & Tanito, M. Interventional human ocular safety experiments for 222-nm far-ultraviolet-C lamp irradiation. Photochem. Photobiol. (2024) doi:10.1111/php.14016.
65. Sliney, D. H. & Stuck, B. E. A need to revise human exposure limits for ultraviolet UV-C radiation. Photochem. Photobiol. 97, 485–492 (2021).
66. Arden, P. C. et al. Chronic far-UVC (222 nm) light exposure of SKH-1 hairless mice does not cause detectable eye pathology or visual deficits. Curr Eye Res under review.
67. Kousha, O., O’Mahoney, P., Hammond, R., Wood, K. & Eadie, E. 222 nm Far-UVC from filtered Krypton-Chloride excimer lamps does not cause eye irritation when deployed in a simulated office environment. Photochem. Photobiol. 100, 137–145 (2024).
68. Sugihara, K. et al. One-year ocular safety observation of workers and estimations of microorganism inactivation efficacy in the room irradiated with 222-nm far ultraviolet-C lamps. Photochem. Photobiol. 99, 967–974 (2023).
69. Duncan, M. A., Welch, D., Shuryak, I. & Brenner, D. J. Ocular and facial far-UVC doses from ceiling-mounted 222 nm far-UVC fixtures. Photochem. Photobiol. 99, 160–167 (2023).
70. Sliney, D. H. Exposure geometry and spectral environment determine photobiological effects on the human eye. Photochem. Photobiol. 81, 483–489 (2005).

5. Evaluating Cancer Risk from Different UV Wavelengths

Summary

Animal studies commonly used to assess cancer risk from UV exposure have not shown an increase in cancer risk from far-UVC. This observation is consistent with biophysical considerations that limit far-UVC penetration into the epidermis (see Skin and eye safety section).

UV radiation as a whole is classed as a carcinogen by expert bodies such as the US National Toxicology Program (NTP) and the International Agency for Research on Cancer (IARC), and far-UVC currently falls under this general classification. The magnitude of cancer risk is not factored into these classifications. Amassing a greater body of evidence may be necessary in order for far-UVC to be formally classified separately from other forms of UV. Further animal studies to evaluate far-UVC cancer risk are warranted, as well as mechanistic investigations into the range of DNA damage types induced by exposure to far-UVC radiation, which may be different from those induced by exposure to longer-wavelength UV.

There are three major types of skin cancer—melanoma skin cancer (MSC), and basal cell and squamous cell carcinomas (termed non-melanoma skin cancers, NMSC). Sun exposure is a risk factor for the more common NMSC form, with squamous cell carcinoma linked to cumulative, long-term chronic lifetime exposure, and basal cell carcinoma more strongly linked to episodic intermittent sunburn.

UVB rays (280–315 nm) found in sunlight are the primary driver of NMSC due to their ability to directly damage DNA, causing the formation of pyrimidine dimers. In contrast, UVA (315–400 nm) induces NMSC by an indirect mechanism involving oxidative stress and free radical production which can lead to DNA damage. Lifetime sunlight UVA exposure also causes age-related photodamage to the skin, as UVA penetrates deeper and can damage the proteins elastin and collagen.

Both UVA and UVB are also implicated in risk for the much rarer but more deadly melanoma, although through different mechanisms. UVB exposure is linked to direct damage to tumor suppressor genes and subsequent carcinogenesis. UVA exposure is associated with immune suppression and altered cell signaling, in addition to indirect DNA damage to tumor suppressor genes, both of which increase the likelihood that nascent melanoma cells will evade the body’s cancer detection systems and survive.

The rarity of chronic human exposures to UVC limits our ability to gather statistically relevant epidemiological data. The non-UVC emissions of the most common man-made UVC sources (such as from arc-welding equipment) reduce certainty about which specific wavelengths pose the greatest risk for carcinogenesis. Much of the evidence for the mutagenic potential of UVC radiation is therefore based on its ability to induce DNA damage similar to that of UVB. Animal studies have also demonstrated tumor formation after exposure to low-pressure mercury lamps, which primarily emit 254-nm UVC but also produce ~4% UVB radiation.

Finally, if either conventional UVC or far-UVC were more widely deployed as a germicidal tool, it would be prudent to conduct observational studies assessing its long-term effects across a wide variety of health outcomes.

Crucial Considerations

• Non-melanoma skin cancer (NMSC) is well-studied, with known epidemiological and mechanistic links to UVA and UVB exposure.
• A few current mouse studies suggest that far-UVC does not induce cancer formation, even in cancer-prone mouse strains. However, it is important to replicate those findings in other animal models, and under different experimental conditions and designs.
• Epidemiological studies like those conducted for occupational or lifestyle-related UVA/UVB exposure will be warranted if far-UVC becomes widely deployed.
• The risk factors and mechanisms underlying melanoma induction are less clear than those for NMSC. There are few good animal models and no well-defined, standardized, and consistent exposure or outcome measures.
• First-order interactions, such as direct DNA damage, have been the primary consideration when evaluating UVC/far-UVC cancer risk. Second-order interactions and alternative damage mechanisms should be considered and discussed. This might include reactive oxygen species, immunological effects, and potential interactions between UVC/far-UVC and other environmental factors, including sunlight.

Analysis

How Does UV Radiation Exposure Cause Cancer?

We have long known that excess exposure to high-energy, penetrating ionizing radiation such as x-rays and gamma rays increases cancer risk by causing DNA damage and mutation in critical genes governing cell division. We also know from human epidemiological studies, experimental animal studies, and in vitro studies that solar ultraviolet radiation can also cause DNA damage and mutation in human skin, similarly increasing cancer risk.

UV radiation type and corresponding penetration depth (see Figure 4.2 in Skin and eye safety section) are two of the most important factors for UV-induced cancer formation. UVA has the lowest energy, penetrates the farthest into biological tissue, is associated with long-term skin photodamage (e.g., wrinkles), and causes oxidative stress, leading to indirect DNA damage. UVB has higher energy than UVA, lower penetration depth, and can directly damage DNA through photochemical mechanisms. UVB exposure is the leading cause of sunburns and is also thought to cause the most skin cancer. UVC has the highest energy, but the shallowest penetration. While solar UVC is completely blocked by the atmosphere, exposure to artificial UVC radiation can result in DNA damage in the upper levels of the epidermis.

UV radiation exposure can lead to direct or indirect damage to the purine and pyrimidine bases in DNA. Cyclobutane pyrimidine dimers (CPDs) and pyrimidine-(6,4)-pyrimidone photoproducts (6-4PPs) are the most common types of DNA base damage resulting from exposure to UVB or conventional UVC wavelengths.

Cells contain protein complexes that are designed to recognize and repair the vast majority of different kinds of DNA base damage, including the kinds caused by UV radiation, but sometimes a misrepair occurs during the process, and an incorrect DNA base is inserted, leading to mutation. If the damage occurs in genes that regulate the process of cell division, or conversely, in genes whose role is to promote cell death when cells are irreparably damaged, after many cell divisions and accumulation of additional mutations, this can result in uncontrolled cell growth, which can lead to cancer.

Carcinogen Classification

The International Agency for Research on Cancer (IARC), part of the World Health Organization (WHO), and the US National Toxicology Program (NTP), both work to identify factors that contribute to cancer. The NTP is formed from parts of several different US government agencies, including the National Institutes of Health (NIH), Centers for Disease Control and Prevention (CDC), and the Food and Drug Administration (FDA).

There is disagreement over the classification of different parts of the UV spectrum as a potential carcinogen, as shown in Table 5.1. Experimental and epidemiological studies indicate that UVA and UVB both have strong causal links to cancer, but only IARC lists these wavelengths as “carcinogenic,” whilst NTP is less definitive about the causal link for any particular form of UV radiation, stating that UVA/B considered alone is only “reasonably anticipated to be” carcinogenic. Although no direct epidemiological evidence links UVC and cancer, IARC and NTP place UVC into the same classifications as UVA and UVB.

Lastly, the magnitude of cancer risk is not factored into these classifications. An IARC label of “carcinogenic to humans” does not address the likelihood of cancer at low doses.

Table 5.1. Classification Across the UV Spectrum

Artificial Ultraviolet Radiation (UVR) and Biological Damage

Epidemiological and experimental studies examining the role of ultraviolet radiation emitted from artificial sources have strengthened a causal relationship between UVA or UVB exposure and skin cancer. Tanning beds (sunbeds) emit primarily UVA radiation 10–15 times higher than that found in sunlight, as well as some UVB. Epidemiological data has consistently shown a correlation between the amount of sunbed use and melanoma risk.

Artificial UVC sources, such as welding torches and germicidal mercury lamps, can also emit non-UVC wavelengths to various degrees. UVB and UVA emissions from these devices may contribute to cancer formation. Welders, especially those using electric arc devices, are exposed to the full UV spectrum, and are also at risk for erythema (sunburn) in unprotected areas. Cancer risk from welding UVR is understudied and our understanding is incomplete, but biological considerations suggest that repeated UVR exposure to unprotected areas increases cancer risk, similar to solar UVA and UVB exposure. No studies have demonstrated a relationship between occupational UVC exposure and skin cancer.

Cancer, Cancer Treatment, and Infection Risk

Individuals undergoing chemo- or radiotherapeutic cancer treatments are at a higher risk of infection due to immune system suppression caused by the treatment. Surgical interventions, by their very nature, also increase infection risk.

Non-Melanoma Skin Cancer

Epidemiological data indicates that cumulative solar UV exposure is highly correlated with non-melanoma skin cancer (NMSC) development. UV exposure is estimated to initiate 90% of NMSCs. NMSCs mainly occur in skin keratinocytes. There are two major forms of NMSC—basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), accounting for ~80% and ~20% of cases respectively.

An estimated 5.4 million cases of NMSC are diagnosed in the US annually, with 2,000–8,000 deaths each year, primarily from squamous cell skin cancer. In the US, the overall incidence of BCC increased by 145% and SCC by 265% between 1976–1984 and 2000–2010. These increases are at least partially caused by dramatic increases in awareness of skin cancer (detection bias), as well as economic incentives in the US healthcare system, since lucrative healthcare screening practices increase the likelihood of skin cancer being detected and diagnosed even if the health risk is relatively minor.

Experimental Animal Studies of the Link Between UVC Exposure and Cancer

Historical experimental animal studies reported a causal relationship between UVC exposure and cancer. This conclusion must be tempered, however, by the realization that these experiments primarily used low-pressure mercury discharge lamps, which emit 90–95% at 254 nm (UVC), but also emit measurable levels of UVA, UVB, and visible light.

One study irradiated mice with 3×10³ mJ/cm² UVC per week, resulting in 7% of animals displaying tumors after a 52-week exposure. This dose is more than 70 times higher than the maximum permitted human exposure every day. Another study by Sterenborg et al. irradiated hairless albino mice, a cancer-prone model, for 75 min/day, 7 day/week, at 23, 146, or 700 mJ/cm² UVC, observing tumor formation that scaled with dose.

The authors of Sterenborg et al., 1988 argued that the UVB emitted by the low-pressure mercury lamps was insufficient to account for the induction of tumors at the rate found. A summary of findings by the IARC Working Group stated that the UVB content of the source could not be excluded as contributing to the induction of skin tumors, and that the possibility remained that an interaction between UVC and UVB radiation led to tumor induction.

Melanoma

Melanoma develops in melanocytes. Melanocytes produce the pigment melanin, which absorbs UV radiation and protects deeper layers of the skin from some of the harmful effects of the sun. In the US in 2024, it is estimated that ~100,000 new cases of melanoma occurred, with ~8,300 deaths.

While solar UVB exposure is a risk factor for melanoma, the link is more complicated than for NMSCs, with genetics, demographics, and other factors contributing to risk. UV exposure is thought to cause 60–70% of cutaneous malignant melanomas. In contrast, acral lentiginous melanoma is not thought to be caused by exposure to sunlight or UV radiation. This form of melanoma is more common in individuals with darker skin, and often first appears on the palms, soles, under the nails, or in the mouth mucosa. Experimental studies of the relationship between UVC exposure and melanoma are complicated by the lack of a good mouse genetic model, and the variety of other causative factors associated with this disease.

Conventional Germicidal UVC Lamps

Overexposure to conventional germicidal UVC lamps (254 nm) is quite rare. The few case reports describe acute reactions resolving within a week, including erythema and photokeratitis, although one study reported long-term effects from acute overexposure. While studies have linked 254-nm UVC exposure with increased CPD formation in skin, very few human studies have assessed cancer risk from germicidal UVC exposure.

To address this knowledge gap, it is important to perform surveillance and mechanistic studies, similar to those conducted to assess cancer risk from solar and occupational UV exposure. It should be noted that conventional germicidal UV lamps may also emit a small proportion of unfiltered UVB, and this must be taken into account in any future studies.

Sun-Sensitizing Conditions and UVC

A small percentage of individuals have mutations in genes that govern various aspects of the DNA damage/repair response. For example, Xeroderma Pigmentosum (XP) is a disease that is caused by defects in nine different DNA repair genes involved in repair of UV-induced DNA damage, but additional unidentified genes may play a role. XP is characterized by photosensitivity, rapid severe sunburn, and an increased risk for all types of skin cancer at an early age. UVB exposure is most directly associated with skin cancer in XP.

The relationship between UVC exposure and cancer in XP, however, is unknown. However, the limited penetration depth of UVC radiation suggests that the risk, if any, may be minimal. To address this knowledge gap, directed research is needed.

Ocular Melanoma

Melanoma in the eye (ocular melanoma) most commonly develops in pigmented cells in the middle layer of the eye (uvea). The uvea is composed of three distinct anatomical regions: the iris; the choroid; and the ciliary body. Melanin granules in melanocytes found in these tissues absorb stray reflections and absorb UVR. In rare cases, melanoma can also occur in other parts of the eye and its external adnexa, including the conjunctiva, eyelid, and orbit.

Uveal melanoma is the most common primary intraocular malignancy with an average of five cases per million in the US. Around 3,320 new cancers (primarily melanoma) of the eye and the orbit are estimated to have occurred in 2024 with 560 deaths.

Impact of Ultraviolet Radiation on Ocular Melanoma Risk

The eyes spend many hours exposed to environmental UVR. It would be reasonable to suspect that this radiation exposure would increase the risk for melanoma. However, uveal melanoma is currently classified by the WHO as one of the five melanomas triggered by “risk factors other than cumulative damage.”

In part, this is because the cornea absorbs all UVR shorter than 295 nm. Longer wavelength UVB and UVA pass through the cornea to reach the iris, parts of the uvea, and lens. Nevertheless, the opaque iris serves as an additional barrier to UV penetration, protecting underlying structures from harmful radiation. The adult lens is similarly opaque to UVR, absorbing all wavelengths below 370 nm, and letting through less than 2% of UVA between 370–400 nm.

Studies of Far-UVC and Cancer

To date, experimental animal studies suggest that far-UVC does not promote skin cancer of any type due to its limited penetration past the stratum corneum (the outer layer of the skin). One recent hairless-mouse study reported no evidence of skin cancer or abnormal skin growths after chronic 222-nm exposures of varying intensity (66 weeks, 5 days/week, 8 hours/day). In another study, mice with defects in the XP gene family were exposed to far-UVC radiation but did not develop skin cancers, swelling of the ears, or erythema. In contrast, similar mice exposed to UVB developed tumors.

The fact that UVC or far-UVC biophysically cannot penetrate deeply into skin suggests that these UV radiations cannot directly damage melanocytes, which lie within deeper skin layers. Any hypothesis for far-UVC-induced melanoma formation must include an indirect mechanism whereby UVR-induced photochemical reactions in the upper epidermis interact with melanocytes located much deeper in the stratum basale. The same rationale applies to individuals with genetic defects in DNA repair.

As far-UVC is an emerging technology, most studies evaluating the impact of chronic far-UVC exposure in human subjects have, to date, involved only a handful of people and lasted no longer than a few months. One exception to this is the 36-month follow up to the Sugihara study on ocular safety. The precautionary principle suggests that skin cancer screening should be included in empirical research of long-term real-world deployments, such as a cluster-randomized clinical trial, in order to provide epidemiological evidence in humans.

Further Reading

• Assessing the safety of new germicidal far-UVC technologies

References

1. National Research Council, Division on Earth and Life Studies, Board on Radiation Effects Research & Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation. Health Risks from Exposure to Low Levels of Ionizing Radiation. (National Academies Press, Washington, D.C., DC, 2006).
2. Berwick, M. & Hay, J. Sun Exposure and Cancer Risk. in Psycho-Oncology 43–46 (Oxford University Press, 2010).
3. Bernerd, F., Passeron, T., Castiel, I. & Marionnet, C. The damaging effects of long UVA (UVA1) rays: A major challenge to preserve skin health and integrity. Int. J. Mol. Sci. 23, 8243 (2022).
4. Tang, X., Yang, T., Yu, D., Xiong, H. & Zhang, S. Current insights and future perspectives of ultraviolet radiation (UV) exposure: Friends and foes to the skin and beyond the skin. Environ. Int. 185, 108535 (2024).
5. Setlow, R. B. The wavelengths in sunlight effective in producing skin cancer: a theoretical analysis. Proc. Natl. Acad. Sci. U. S. A. 71, 3363–3366 (1974).
6. Ravanat, J. L., Douki, T. & Cadet, J. Direct and indirect effects of UV radiation on DNA and its components. J. Photochem. Photobiol. B 63, 88–102 (2001).
7. Alhmoud, J. F., Woolley, J. F., Al Moustafa, A.-E. & Malki, M. I. DNA damage/repair management in cancers. Cancers (Basel) 12, 1050 (2020).
8. Mercadante, A. A. & Kasi, A. Genetics, cancer cell cycle phases. in StatPearls (StatPearls Publishing, Treasure Island (FL), 2023).
9. National Toxicology Program. 15th report on Carcinogens. Rep. Carcinog. 15 (2021).
10. Dessinioti, C. & Stratigos, A. J. An epidemiological update on indoor tanning and the risk of skin cancers. Curr. Oncol. 29, 8886–8903 (2022).
11. An, S. et al. Indoor tanning and the risk of overall and early-onset melanoma and non-melanoma skin cancer: Systematic review and meta-analysis. Cancers (Basel) 13, 5940 (2021).
12. Burgard, B. et al. Solarium use and risk for malignant melanoma: Meta-analysis and evidence-based Medicine systematic review. Anticancer Res. 38, 1187–1199 (2018).
13. Gallagher, R. P., Spinelli, J. J. & Lee, T. K. Tanning beds, sunlamps, and risk of cutaneous malignant melanoma. Cancer Epidemiol. Biomarkers Prev. 14, 562–566 (2005).
14. Boniol, M., Autier, P., Boyle, P. & Gandini, S. Cutaneous melanoma attributable to sunbed use: systematic review and meta-analysis. BMJ 345, e4757 (2012).
15. Colantonio, S., Bracken, M. B. & Beecker, J. The association of indoor tanning and melanoma in adults: systematic review and meta-analysis. J. Am. Acad. Dermatol. 70, 847–57.e1–18 (2014).
16. Scientific Committee on Health, Environmental and Emerging Risks. Opinion on Biological Effects of UV-C Radiation Relevant to Health with Particular Reference to UV-C Lamps. https://health.ec.europa.eu/system/files/2018-03/scheer_o_002_0.pdf.
17. Falcone, L. M. & Zeidler-Erdely, P. C. Skin cancer and welding. Clin. Exp. Dermatol. 44, 130–134 (2019).
18. Heltoft, K. N., Slagor, R. M., Agner, T. & Bonde, J. P. Metal arc welding and the risk of skin cancer. Int. Arch. Occup. Environ. Health 90, 873–881 (2017).
19. Welding. https://www.cancer.org.au/cancer-information/causes-and-prevention/workplace-cancer/welding.
20. Dixon, A. J. & Dixon, B. F. Ultra violet radiation from welding and possible risk of skin and ocular malignancy. Med. J. Aust. 181, 155–157 (2004).
21. Seo, S. K., Liu, C. & Dadwal, S. S. Infectious disease complications in patients with cancer. Crit. Care Clin. 37, 69–84 (2021).
22. Chen, A. et al. Surgical site infection rates following Mohs micrographic surgery by body site: A systematic review and meta-analysis. J. Am. Acad. Dermatol. 89, 862–864 (2023).
23. Heal, C. F., Buettner, P. G. & Drobetz, H. Risk factors for surgical site infection after dermatological surgery. Int. J. Dermatol. 51, 796–803 (2012).
24. Görlitz, M. et al. Assessing the safety of new germicidal far-UVC technologies. Photochem. Photobiol. 100, 501–520 (2024).
25. Pleasance, E. D. et al. A comprehensive catalogue of somatic mutations from a human cancer genome. Nature 463, 191–196 (2010).
26. Nanz, L. et al. Epidemiology of keratinocyte skin cancer with a focus on cutaneous squamous cell carcinoma. Cancers (Basel) 16, (2024).
27. US Environmental Protection Agency Office of Air and Radiation. Health effects of UV radiation. https://www.epa.gov/sunsafety/health-effects-uv-radiation (2015).
28. Basal & squamous cell skin cancer statistics. https://www.cancer.org/cancer/types/basal-and-squamous-cell-skin-cancer/about/key-statistics.html.
29. Skin cancer. American Academy of Dermatology https://www.aad.org/media/stats-skin-cancer.
30. Muzic, J. G. et al. Incidence and trends of basal cell carcinoma and cutaneous squamous cell carcinoma: A population-based study in Olmsted county, Minnesota, 2000 to 2010. Mayo Clin. Proc. 92, 890–898 (2017).
31. Drucker, A. M. et al. Association between health maintenance practices and skin cancer risk as a possible source of detection bias. JAMA Dermatol. 155, 353 (2018).
32. Lyon (FR): International Agency for Research on Cancer; 1992. (IARC Monographs on the Evaluation of Carcinogenic Risks to Humans). Studies of Cancer in Animals 3.
33. Forbes, P. D. & Urbach, F. Experimental modification of photocarcinogenesis. I. Fluorescent whitening agents and short-wave UVR. Food Cosmet. Toxicol. 13, 335–337 (1975).
34. Lill, P. H. Latent period and antigenicity of murine tumors induced in C3H mice by short-wavelength ultraviolet radiation. J. Invest. Dermatol. 81, 342–346 (1983).
35. Joshi, S. D., Shrikhande, S. S. & Ranade, S. S. Studies on UV-induced skin tumors in experimental animals. Photobiochemistry and Photobiophysics 7, 235–243 (1984).
36. Sterenborg, H. J., van der Putte, S. C. & van der Leun, J. C. The dose-response relationship of tumorigenesis by ultraviolet radiation of 254 nm. Photochem. Photobiol. 47, 245–253 (1988).
37. Some Naturally Occurring and Synthetic Food Components, Furocoumarins and Ultraviolet Radiation — IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans. vol. 40 (1986).
38. D’Orazio, J., Jarrett, S., Amaro-Ortiz, A. & Scott, T. UV radiation and the skin. Int. J. Mol. Sci. 14, 12222–12248 (2013).
39. Melanoma of the Skin – Cancer Stat Facts. SEER https://seer.cancer.gov/statfacts/html/melan.html.
40. Koh, H. K., Geller, A. C., Miller, D. R., Grossbart, T. A. & Lew, R. A. Prevention and early detection strategies for melanoma and skin cancer. Current status. Arch. Dermatol. 132, 436–443 (1996).
41. Sample, A. & He, Y.-Y. Mechanisms and prevention of UV-induced melanoma. Photodermatol. Photoimmunol. Photomed. 34, 13–24 (2018).
42. Hall, K. H. & Rapini, R. P. Acral Lentiginous Melanoma. StatPearls.
43. Vaidya, M. U., Gangakhedkar, G. R., Shetty, A. N. & Waghalkar, P. V. A rare occurrence of accidental exposure to UV radiation among operating theatre personnel. Indian J. Anaesth. 64, 230–232 (2020).
44. Trevisan, A. et al. Unusual high exposure to ultraviolet-C radiation. Photochem. Photobiol. 82, 1077–1079 (2006).
45. Zaffina, S. et al. Accidental exposure to UV radiation produced by germicidal lamp: case report and risk assessment. Photochem. Photobiol. 88, 1001–1004 (2012).
46. Glaab, J. et al. Skin tolerant inactivation of multiresistant pathogens using far-UVC LEDs. Sci. Rep. 11, 14647 (2021).
47. Welch, D. et al. No evidence of induced skin cancer or other skin abnormalities after long-term (66 week) chronic exposure to 222-nm far-UVC radiation. Photochem. Photobiol. 99, 168–175 (2023).
48. Yamano, N. et al. Long-term Effects of 222-nm ultraviolet radiation C Sterilizing Lamps on Mice Susceptible to Ultraviolet Radiation. Photochem. Photobiol. 96, 853–862 (2020).
49. Brambullo, T. et al. Xeroderma pigmentosum: A genetic condition skin cancer correlated-A systematic review. Biomed Res. Int. 2022, 8549532 (2022).
50. Kraemer, K. H. Xeroderma pigmentosum. Cutaneous, ocular, and neurologic abnormalities in 830 published cases. Arch. Dermatol. 123, 241–250 (1987).
51. Cleaver, J. E. Diagnosis of xeroderma pigmentosum and related DNA repair-deficient cutaneous diseases. Curr. Med. Lit. Dermatol. 13, 41–48 (2008).
52. Chalada, M., Ramlogan-Steel, C. A., Dhungel, B. P., Layton, C. J. & Steel, J. C. The impact of ultraviolet radiation on the aetiology and development of uveal melanoma. Cancers (Basel) 13, 1700 (2021).
53. Eye melanoma. Mayo Clinic https://www.mayoclinic.org/diseases-conditions/eye-melanoma/symptoms-causes/syc-20372371.
54. Grech Fonk, L., Ferreira, T. A., Webb, A. G., Luyten, G. P. M. & Beenakker, J.-W. M. The economic value of MR-imaging for uveal melanoma. Clin. Ophthalmol. 14, 1135–1143 (2020).
55. Chattopadhyay, C. et al. Uveal melanoma: From diagnosis to treatment and the science in between. Cancer 122, 2299–2312 (2016).
56. Key statistics for eye cancer. https://www.cancer.org/cancer/types/eye-cancer/about/key-statistics.html.
57. Elder, D. E., Bastian, B. C., Cree, I. A., Massi, D. & Scolyer, R. A. The 2018 World Health Organization classification of cutneous, mucosal, and uveal melanoma: Detailed analysis of 9 distinct subtypes defined by their evolutionary pathway. Arch. Pathol. Lab. Med. 144, 500–522 (2020).
58. Boettner, E. A. & Wolter, J. R. Transmission of the Ocular Media. Investigative Ophthalmology & Visual Science 1, 776–783 (1962).
59. Dillon, J. New trends in photobiology: the photophysics and photobiology of the eye. J. Photochem. Photobiol. B 10, 23–40 (1991).
60. Mallet, J. D., Gendron, S. P., Drigeard Desgarnier, M.-C. & Rochette, P. J. Implication of ultraviolet light in the etiology of uveal melanoma: A review. Photochem. Photobiol. 90, 15–21 (2014).
61. Sugihara, K. et al. One-year ocular safety observation of workers and estimations of microorganism inactivation efficacy in the room irradiated with 222-nm far ultraviolet-C lamps. Photochem. Photobiol. 99, 967–974 (2023).
62. Sugihara, K. et al. Ocular safety of 222-nm far-ultraviolet-c full-room germicidal irradiation: A 36-month clinical observation. Photochem. Photobiol. (2024) doi:10.1111/php.14052.

6. Ozone and Indoor Air Quality

Summary

We know that far-UVC generates ozone through photolysis, and the rate of generation per unit of fluence rate is relatively well quantified. However, how this translates to occupants’ exposure to ozone and ozone reaction products is highly sensitive to conditions—particularly outdoor air ventilation rate and what other chemical precursors are present in the space. The potential health impacts of increases in indoor ozone levels are highly uncertain (see Ozone epidemiology section), but there is an established consensus among indoor air quality experts that any ozone emissions indoors are undesirable. Ozone can be removed with activated carbon filters and catalysts.

The question is therefore not whether far-UVC creates ozone, which is beyond doubt, but how much ozone is created by a given amount of far-UVC, how much of the ozone (and the products of chemical reactions that consume ozone) are mitigated through ventilation or removed by activated carbon filters or catalysts, and how the potential costs of this should be weighed against the infection prevention benefits of far-UVC.

Crucial Considerations

• How much ozone far-UVC generates depends on the particular wavelength. Shorter wavelengths generate more ozone.
• For filtered krypton chloride excimer (KrCl*) lamps that have so far been assessed, experimental measurements and first principles calculations support an estimated ozone generation efficiency of 7–10.5 ppb per hour per μW/cm² of fluence rate.
• The expected increase in steady-state ozone concentrations in a space with 1 air change per hour is low to mid single digits ppb per μW/cm² of average fluence rate. This is a small amount relative to both outdoor ozone and fluctuations that can be observed in background levels of indoor ozone, but relatively large compared to the average levels of indoor ozone, which show a central tendency of ~4–6 ppb. Indoor ozone reacts with other substances, which is also known as ‘ozone loss’. The products of this ozone loss may be more significant for health than ozone itself.
• It is possible that there are photochemical interactions that affect indoor air quality other than those caused by ozone generation that we have not yet characterized, and research is ongoing.
• In addition to adequate ventilation, ozone mitigation options include catalysts and activated carbon filters.

When converting between O₃ in ppb and μg/m³, the report uses a value of 1 ppb = 1.96 μg/m³. This value varies slightly based on temperature and air pressure.

Analysis

Quantifying Ozone Production from Far-UVC

Far-UVC photons will cause ozone (O₃) generation through direct photolysis of O₂ molecules in the air (i.e. the photons break down the O₂ molecules). Each photolyzed O₂ molecule leads to the formation of two ozone molecules.

Where hν is a photon with wavelength <242 nm, and M is any other molecule (such as N₂ or another O₂). When wavelengths are below 200 nm, the O₂ absorption cross section is large enough to promote rapid photolysis under typical tropospheric conditions, and this rate of ozone generation increases rapidly with decreasing wavelength. It is for this reason that this section of the electromagnetic spectrum is referred to as ‘vacuum UV’—since it is rapidly absorbed by oxygen, it can only be used in spaces without oxygen (such as a vacuum).

Above 200 nm, the propensity for an O₂ molecule to be photolyzed decreases at a linear rate from 200 nm to 242 nm. However, O₂ absorption at 222 nm—the peak wavelength produced by the most common far-UVC lamps—is strong enough to generate detectable increases in O₃ at fluence rates relevant for far-UVC applications.

To understand how fast far-UVC produces ozone in a given space, we need to take into account both the intrinsic ozone generation efficiency of the wavelengths emitted from a far-UVC device and the total amount of far-UVC employed in the space.

where:

P = ozone production rate (ppb/hr)
G = ozone generation efficiency (ppb/hr/μW/cm²)
I = room average fluence rate (μW/cm²)

Note that many manufacturers and studies quote the measured irradiance of a far-UVC device at a particular distance from the lamp, which makes sense in the context of photobiological safety. However, the average fluence rate in a room is not only a property of the irradiances at any given distance from the lamp but also the average photon pathlength and the room volume (see Efficacy section). The ozone production rate from a far-UVC device is therefore dependent on not just the device, but the context in which it is used.

There are currently three studies which have directly measured the ozone generation efficiency of KrCl* lamps in controlled experimental chambers. It is important to note that these lamps all have filters which result in the reduction of emissions <200 nm and >235 nm. In these studies, the ozone generation efficiency ranges from 7–10.5 ppb/hr per μW/cm² of average fluence rate. This is a relatively small range given the uncertainties inherent in fluence rate determinations.

The first of these studies (Peng et al., 2023) also estimated the ozone generation efficiency using a first-principles model based on the absorption cross-section of O₂ and the fully characterized emission spectrum of one of the tested emitters. The authors arrived at an identical figure to their experimental finding of 10.5 ppb/hr/(μW/cm²). All further calculations in this document use this figure.

However, it is worth noting that the ozone generation rate of a different model using the same bulb was approximately 25% lower. This is believed to be due to the different emissions spectrum caused by the use of a diffuser. It is important to replicate these ozone generation efficiency calculations using different devices from different manufacturers. This could validate the underlying photochemical model of ozone generation based on O₂ absorption cross-sections and the fully characterized emission spectrum of each device.

Given the underlying photochemical model, we should expect the ozone generation efficiency to be modestly sensitive to the exact emission spectrum of a source within the far-UVC range. For example, an ozone generation efficiency of 6.7 ppb per hour per μW/cm² would be expected for a monochromatic device that emitted only 222 nm, but devices emitting different wavelengths would have different efficiencies.

In order to translate ozone production into expected changes in ozone concentrations, we also need to know the rate at which ozone is removed with ventilation or lost to chemical reactions.

Key Takeaways

• Far-UVC generates ozone from photolysis of O₂.
• Current studies show an ozone generation efficiency ranging from 7–10.5 ppb/hr/(μW/cm²), at least for filtered krypton chloride excimer lamps.
• Theoretical models and controlled experimental data are broadly consistent, but further validation with different devices is recommended.

Modeling Framework for Indoor Ozone Concentrations

Nazaroff and Weschler, 2022 present a framework for understanding the determinants of indoor ozone concentrations. Ozone is brought in from outdoors through natural and mechanical ventilation and also removed through the same means. Once indoors, ozone is lost through chemical reactions with surfaces, human occupants, any NO_x gases present, and with volatile organic compounds (VOCs). The potential importance of the byproducts of these reactions is discussed later in this section.

These factors are modeled as a decay constant k, which is a summary statistic of the average rate at which ozone reacts in an indoor space, and has the unit hr. The greater the k constant, the more ‘ozone per hour’ reacts away in an indoor space.

The single largest cause of ozone decay in most spaces is from reactions with surfaces. Nazaroff and Weschler, 2022 identified two studies that have directly quantified the whole-room average ozone decay rate needed for this modeling framework, and both are of residences. Measurements in 43 southern California residences provided a decay per hour of 2.8 ± 1.3, and a study of 15 bedrooms in 14 Chinese residences estimates 2.8 ± 1.1.

While there is an extensive literature on ozone loss to indoor-relevant surfaces, these two studies are the best available empirical data on whole-room average ozone decay rates from reactions with indoor surfaces.

In the absence of indoor sources and where the only source of indoor ozone is outdoors, indoor/outdoor ozone ratios should approximately conform to equation 6.2.

Where λ_ACH is the air exchange rate with outdoor air—ACH stands for ‘air changes per hour’. k_sum represents all the various decay sources for indoor ozone loss, of which surfaces are typically the most important.

Observed indoor/outdoor ozone ratios and realistic assumptions about ventilation rates therefore put a bound on what the values of k_sum can plausibly be. Nazaroff and Weschler, 2022 collated indoor/outdoor ozone concentrations from dozens of studies, inferring a central estimate for the surface loss coefficient of ~2/hr, rather than 2.8/hr as documented in the studies of residences.

These estimates are sourced from a number of different environments, and may not be representative of the spaces in which germicidal UV would be applied. For calculations in this document, unless otherwise stated, a k_sum value for losses to reactions of 2/hr is used.

In a recent study, Sørensen et al., 2024 measured k_sum in four small offices and found k_sum ranging from 0.83 to 3.93.

Table 6.1. Ozone Decay Measurements in Small Offices

For sensitivity analysis conducted later, we assume a lower bound of k_sum of 0.5.

The Effect of Adding Far-UVC as a Source of Ozone Indoors

One way of thinking about equation 6.2 is that there is a rate at which ozone is added indoors ([O₃]_out × λ_ACH) and a rate at which it is removed (λ_ACH + k_sum). Ozone generated from far-UVC (or any other indoor source) is chemically indistinguishable from outdoor ozone and thus is removed in the same way.

Substituting P from equation 6.1, the increase in ozone concentration indoors from introducing far-UVC as an indoor source of ozone can therefore be calculated following equation 6.3.

Combining this with equation 6.2, the complete equation for indoor ozone concentration in the presence of far-UVC is:

Assuming an ozone generation constant G of 10.5 ppb/hr per µW/cm² and holding outdoor ozone levels constant, we can therefore estimate the ozone increases that we would expect to observe under 1 µW/cm² of room average fluence rate.

Table 6.2. Estimated Change in Average Ozone Concentration from 1 µW/cm² of Far-UVC

For comparison, the median indoor ozone concentration from the literature summarized in Nazaroff and Weschler, 2022 is 4–6 ppb.

The Impact of Changing Ventilation on Indoor Ozone Levels

Ventilation (λ_ACH) rates vary considerably, in commercial environments based on occupancy and the use of central HVAC systems, and in residential environments based on factors such as windows being open or shut, outside climate conditions and the construction of the property.

The variation in residential air change rates generally follows a lognormal distribution. Using equations 6.2 and 6.4, we can compare the increases in indoor ozone concentrations that we would expect to see under far-UVC irradiation with those we would expect from changes in ventilation.

If we assume an outdoor level of ozone of 30 ppb, we can use the equations to estimate the steady-state indoor ozone concentration under different ventilation and k_sum reaction loss assumptions, and therefore estimate the effect on steady-state ozone concentrations of increasing λ (ventilation) by 1.

As per Nazaroff and Weschler, 2022, the report conservatively assumes that ~25% of the outdoor ozone is lost in the mechanical ventilation system before it reaches the indoor space, and hence for modeling purposes [O₃]_out is 22.5 ppb rather than 30 ppb.

Table 6.3. Residential Air-Change Rates Utilizing CO₂ as a Tracer

Variations in ventilation in a single residence over time also follow a lognormal distribution.

Table 6.4. Estimated Increase in Ozone Concentration from Increasing λ_ACH by 1

Table 6.5. Difference Between Increasing Ventilation by 1 ACH and Adding 1 µW/cm² of Far-UVC

This modeling demonstrates the importance of characterizing background levels of, and variations in, ventilation. This helps us to contextualize the relative increase in ozone concentrations from far-UVC and also to remove any confounding in real-world observations.

The Impact of Outdoor Ozone Variations on Indoor Ozone Levels

Background levels of outdoor ozone vary significantly day to day and hour to hour. Outdoor ozone levels follow a diurnal (24-hr) cycle, with ozone levels typically peaking in the afternoon and being lowest in the early morning. The magnitude of the difference can be substantial; in a recent study the average difference between daytime peak and nighttime minimum ozone concentration in Chinese cities in the summertime was 45 ± 13 ppb.

The diurnal ozone cycle is also clearly visible in one of the few studies observing real-world ozone formation from the use of far-UVC. Background variations in outdoor ozone levels can, in some circumstances, introduce a significant confounding effect when researchers attempt to estimate the impact of far-UVC on indoor ozone concentrations.

Mixing

It should be noted that this modeling framework implicitly utilizes the ‘well-mixed room assumption’ (see Efficacy section). According to this assumption, ventilation rates can be treated as a decay constant with respect to the average concentration of a contaminant, and additionally there is no short-circuiting between the air inlet and outlet that would also lessen the residence time of ozone introduced from outdoors.

A well-mixed room is likely a more reasonable assumption for a gaseous contaminant like ozone than for a pathogen, but it would be fruitful for indoor air quality researchers to consider the impact of mixing on this modeling framework.

Key Takeaways

• Ozone decays indoors due to chemical reactions, with surfaces being the most important source of loss in most spaces.
• Background indoor ozone concentrations are a function of outdoor ozone concentrations. They depend on the balance of ventilation (which brings in ozone) and ozone reactive loss (with chemicals on surfaces and in the air).
• Variations in indoor ozone concentrations due to changing ventilation and outdoor ozone levels can be substantial relative to a central tendency of 4–6 ppb in the published literature.
• Increases in ozone from realistic use of far-UVC will therefore often be small relative to observed variations in indoor ozone levels.
• When taking measurements in the field, variations in background ozone levels need to be adjusted for when estimating the impact of far-UVC.

Ozone Generation from Far-UVC Devices in Real-World Spaces

As of the time of writing, there are five studies, three published and two in preprint, documenting ozone increases from far-UVC devices in ‘real’ environments, as opposed to experimental chambers. In all of these studies, the results account for the variation in the background levels of ozone while the experiments were taking place.

For the Link et al., 2024 and Peng et al., 2023 experiments, where the ozone generation efficiency is reported or can be directly imputed from reported data, they are consistent with each other and within the previous estimate of 7–10.5 ppb/hr/(μW/cm²).

For Kalliomäki et al., 2023, an estimated k_sum decay rate can be inferred from their report that a 1 ppb increase in outdoor ozone was associated with a 0.33 ppb increase in indoor ozone, and a reported average air changes per hour (ACH) of 1.4. In order for this indoor/outdoor ozone ratio and ACH to be consistent with equation 6.2, a k_sum of 2.8 is implied.

This k_sum estimate combined with an estimated fluence rate of 1.75 and 1.4 ACH implies an ozone generation rate of 13.7 ppb/hr/(μW/cm²). This is a crude estimate using the average values and regression point estimates, and it is possible that there was measurement error, that <200-nm emissions were less well-filtered on this particular lamp, or that there were electrical discharges on the surface of the excimer lamp that acted as a separate source of ozone.

For Narouei et al., 2024, there isn’t data in the paper on k_sum or data that allows k_sum to be implied. However, we can calculate what the implied ozone generation rate would be with a plausible k_sum. At a k_sum of 2, the implied ozone generation rates are 30.7 for the ‘low power’ test and 6.2 for the ‘high power’ test.

In the ‘low power’ test, it is worth noting that there is only a 2.7 ppb increase observed in average ozone concentration, in the context of background levels varying between 2 and 20 ppb and employing an ozone monitor with a reported precision of ± 1.5 ppb. In the context of the effect size and the instrument precision, 2–20 ppb background variation in ozone is substantial. The report hypothesizes that measurement inaccuracy in the low power test is likely to be the source of the substantially larger than expected ozone generation efficiency.

Table 6.6. Studies of Ozone Increases in Real-World Settings Under Far-UVC Irradiation

The fifth study—not included in Table 6.6—is Sørensen et al., 2024, who conducted multiple experiments in four different offices, including experiments where they changed lamp placement and therefore estimated pathlength. Fluence rates were not calculated, and therefore ozone generation efficiency cannot be directly estimated.

However, the expected correlation between increase in ozone production and UV pathlength occurs for pathlengths of 3 m or less, and the rate of O₃ production of the same lamp corresponds with controlled experimental studies. These observations are therefore consistent with the ozone generation efficiency that was found or can be implied from other studies.

Figure 6.4 shows an apparent flattening of the O₃ production curve at pathlengths greater than 3 m. There is no theoretical reason that this flattening should occur if fluence rate is scaling with estimated pathlength.

Without an independent estimate of fluence rate, we can’t rule out that the fluence rate did not fully scale with the estimated pathlength provided in the Sørensen paper, and therefore ozone production didn’t scale either. Pathlength was measured by distance of the lamp to the wall, and it is a reasonable hypothesis that a greater proportion of photons terminate at the ceiling and floor as the lamp is moved further away from the wall.

If this is the case, the true average photon pathlength was no longer increasing as the horizontal distance was increased beyond 3 m.

The other possibility offered by the authors is that O₃ production is not fully linear but requires a certain minimum UV intensity to occur. Only further studies that vary fluence rate and measure and/or estimate all the relevant variables independently can confirm which explanation is correct.

Researchers conducting studies of far-UVC and ozone generation in real spaces are encouraged to analyze their results in this modeling framework, and to independently estimate k_sum, ventilation rates, and fluence rates.

Key Takeaways

• Studies so far suggest that ozone increases from far-UVC installations in real-world settings are broadly consistent with the theoretical framework discussed above, but more high-quality data is needed.
• Measurements need to have sufficient precision to accurately measure the expected effect. This often requires high-precision, research-grade instrumentation.
• It is critical to separately measure ozone concentration, k_sum, ventilation rates and fluence rates in order to establish robust modeling assumptions.

Ozone Loss

Focusing on increases in indoor ozone concentrations obscures what may be the most important dynamic. The k_sum ozone decay constant represents reactions that destroy ozone but create other products, and the concentrations of these products—and in some cases their chemical precursors—are themselves heavily influenced by ventilation. As discussed in the Ozone epidemiology section, these products also have potential negative health effects.

k_sum is a summary statistic for the rate of ozone loss to all sources, such as reactions with chemicals present on surfaces, VOCs in the air, and chemicals produced by our own bodies, especially the skin oil constituent squalene.

Except in rare cases where ozone is catalytically decomposed, each ozone reaction leading to this “loss” of ozone creates two initial chemical products, which can then go on to react with other chemicals present in the indoor environment. These subsequent reactions and their products are often referred to as “secondary chemistry.”

In contexts where ozone is just coming in from outdoors, with no indoor sources, ozone loss is the same as the difference between the indoor and outdoor ozone level. For example, if the outdoor ozone level is 25 ppb and the indoor level is 5 ppb, the ozone loss is 20 ppb.

In more general terms, [O₃]_loss can be defined as the difference between [O₃]_in and what [O₃]_in would be if the decay constant k_sum = 0.

Suppose an increase in [O₃]_in has been observed. Has ozone loss increased or decreased? These equations show that it depends on the cause of the increase in [O₃]_in.

If it is due to an increase in [O₃]_out, or due to an indoor source like far-UVC, then the increase in [O₃]_in also means an increase in [O₃]_loss. However, if there is an increase in [O₃]_in due to increasing ventilation bringing outdoor ozone indoors at a faster rate, then [O₃]_loss has actually decreased.

At low rates of ventilation, changes in ozone loss from an indoor source can be substantial even from fairly small amounts of ozone generation.

Ozone loss from an indoor source is far more sensitive to changes in ventilation than indoor ozone concentrations are, and the potential health effects of the products of ozone loss drive much of the concern about ozone among indoor air quality experts.

Table 6.7. Estimated Sensitivity of Ozone Concentrations and Ozone Loss to Ventilation

Assumptions: fluence rate 1 μW/cm²; ozone generation constant 10.5 ppb/hr/(μW/cm²); k_sum = 2 hr^-1.

Products of Ozone Loss

The best-understood product of ozone loss is the creation of secondary organic aerosol (SOA), through the reaction of ozone with VOCs. This does not mean that other products of ozone loss are necessarily less important when it comes to health effects, but rather that we do not currently have a framework for quantifying specific products and relating them to hazard ratios from epidemiological studies.

SOA represents a small fraction, approximately 5%, of the total yield of products from ozone loss. The vast majority of the loss is due to reactions with surfaces, of which approximately half then partition to the gas phase with the balance of products remaining on surfaces.

The total mass of “fine” particles—those with a diameter of less than 2.5 microns—is commonly referred to as PM_2.5. PM_2.5 in outdoor air has been regulated by the EPA in the United States since 1997, after numerous studies showed PM_2.5 levels to be more important to health than previously monitored measures that included larger particles.

The literature about harms from outdoor particulate matter is significantly more extensive than that of ozone. The current consensus is that there is not a threshold below which PM_2.5 exposure is benign, and that the exposure-response relationship may have a steeper slope at lower PM_2.5 concentrations.

We can estimate the average yield of PM_2.5 using the ozone loss framework. What is required is a parameter for the quantity of secondary organic aerosol produced for every ppb of ozone loss.

Weschler and Nazaroff, 2023 estimate what they term a Yield_SOA of 0.03–0.07 μg/m³ for every ppb of ozone loss in an indoor space.

Table 6.8. Estimated PM Yield from 1 μW/cm² of Far-UVC from a KrCl* Lamp

In poorly ventilated spaces, it seems reasonable to assume that the generation of secondary organic aerosol is at least ~1 μg/m³ for every ~1 μW/cm² of room average fluence rate from a KrCl* lamp, and could be as high as ~5 μg/m³. At higher levels of ventilation, it will be substantially less.

In the case of Link et al., 2024, ~3.2 μW/cm² of room average fluence rate produced 1.8 μg/m³ ± 0.7 of SOA, or 0.56 μg/m³ ± 0.2 of SOA per μW/cm² of room average fluence rate. For context, the EPA estimates the mean annual average PM_2.5 levels outdoors across over 300 sites in the United States to be 8.5 μg/m³.

It should be noted that this estimate implicitly assumes that the creation of ozone alone explains the SOA formation under far-UVC irradiation. In environments with high levels of ozone-reactive VOCs, such as after using household cleaning products containing limonene, the Yield_SOA will be greater.

Other factors that will affect particle formation and growth include relative humidity and the size and density of pre-existing particles in the air. While 0.03–0.07 μg/m³ for every ppb of ozone loss is a reasonable central estimate to use, we should expect the true variation to potentially be significantly wider than this range.

Does Ozone Loss Alone Explain Secondary Organic Aerosol Generation Under Far-UVC?

Ozone is not the only potential oxidant created by far-UVC. O₃ can also be further photolyzed by far-UVC photons to form highly reactive OH radicals.

OH can also be formed through reactions of O₃ with alkenes already present in the environment.

Barber et al., 2023 found that under exceptionally high fluence rates in a small chamber (45 µW/cm²), OH radicals generated under far-UVC irradiation affected aerosol content separately from what would be expected from ozonolysis reactions alone. Separately, two other studies using lower fluence rates in larger chambers found that modeling based on ozonolysis alone was sufficient to account for the vast majority of secondary organic aerosol due to reactions with the VOC limonene.

Similarly, in the real-world bathroom study of Link et al., 2024, the authors found no need to posit any additional source of SOA other than ozone-initiated chemistry.

New Particle Formation

Goss and Kroll, 2024 found new particle formation from far-UVC in a small chamber at high fluence rates (45 µW/cm²) that does not appear to be explicable by ozonolysis or OH oxidation of easily measurable concentrations of VOCs. This new particle formation was not observed at fluence rates relevant to real-world applications.

Link et al., 2024 also found new particle formation, although without an ozone-only control it is not possible to distinguish between particle formation from ozone-initiated pathways and any other as-yet unelucidated particle formation pathways.

In the experiment in an office space reported in Narouei et al., 2024, with an estimated fluence rate of 5.19 µW/cm², no significant new particle formation or change in the distribution of particle size was observed.

Variation in SOA Formation

We should expect the SOA products of ozone reactions to vary across indoor spaces by orders of magnitude more than the variation in the indoor/outdoor ozone ratio.

Limonene is a chemical in a class known as monoterpenes. Monoterpenes can typically be found in plant extracts and essential oils, and they are used to provide fragrances to consumer products. Limonene, which is found in the peel of citrus fruits and is cheap to manufacture, provides the fresh lemon or citrus smell commonly used in cleaning products.

Limonene is an example of an ozone-reactive VOC that will react efficiently with ozone to form SOA.

Typical limonene concentrations in indoor spaces have been cited as 4–12 ppb. In an extensive literature review of concentrations of chemicals in residences, the difference between the 25th and 95th percentile of limonene concentrations was a factor of 100. In places like spas, limonene levels can spike dramatically after the use of scented oils, reaching up to 3,000 ppb.

High limonene emissions from using scented cleaning products, in combination with indoor ozone, can create air quality issues similar to those found in polluted urban areas.

Formaldehyde

Formaldehyde is another product of ozone loss. It is a major indoor pollutant of concern with well-known health effects under high exposure, and is formed by many oxidation reactions, including between limonene and O₃.

In highly controlled experimental settings, an increase in formaldehyde has been observed under far-UVC irradiation in the presence of normal background levels of limonene, although the magnitude—low single digit ppb in a sealed test chamber—is small relative to ambient levels of formaldehyde in urban and indoor air, and to recommended exposure limits.

Key Takeaways

• Ozone reacts with surfaces, VOCs, and human skin oils, leading to the creation of secondary chemical products.
• Secondary products may be of greater relevance to health than ozone itself.
• Increased ventilation generally reduces the concentration of these secondary products despite increasing steady-state ozone levels.
• Ozone appears to be the primary driver of secondary organic aerosol (SOA) formation under far-UVC, though other mechanisms including OH radicals may also contribute.
• SOA production can vary depending on the indoor environment and the presence of specific VOCs such as limonene.
• Variations in limonene concentrations will also drive differences in other potential chemistry, such as the formation of other VOCs like formaldehyde. It is not currently known whether any of these non-SOA byproducts are quantitatively significant.

Other Potential Photochemical Interactions That Could Affect Indoor Air Quality

Thus far, we have discussed products that are downstream of ozone chemistry. However, we must also ask whether far-UVC creates other significant photochemical interactions that are not mediated by ozone chemistry.

Sørensen et al., 2024 detected increases in acetic acid and butyl fragments that are not explained either by O₃ or OH oxidation chemistry. They hypothesized that these increases were caused by either direct photolysis of another compound or by far-UVC accelerating off-gassing of surfaces that was already occurring.

While these increases were small relative to exposure limits for acetic acid and butanol respectively, the study illustrates the possibility that far-UVC could involve relevant photochemistry that is separate from ozone chemistry.

Our knowledge about the propensity of far-UVC to interact with chemicals in the air is based almost entirely on stratospheric chemistry, as no far-UVC is naturally found below the ozone layer. The absorption cross-sections that we do have are therefore primarily based on molecules that can be found in the stratosphere. There has been no cause to investigate the propensity of far-UVC to interact with the much wider range of molecules found in the built environment.

There are also significant unknowns in understanding what, if any, toxicological significance there could be from chemical changes that take place.

The first priority should be taking careful, non-targeted measurements with sensitive equipment in diverse environments, and looking for chemistry that can’t be explained by ozone generation and secondary chemistry.

The second priority is to consider what chemical factors are most likely to be ubiquitously present in occupied spaces utilizing far-UVC, of which the most obvious is the chemical impacts of the occupants themselves.

The propensity of the skin oil constituent squalene to react with ozone and the secondary chemistry engendered creates a “human oxidation field” that surrounds each of us. These occupant-specific factors are likely significant mediators of how increased ozone levels change the chemical species we are exposed to.

Key Takeaways

• Current understanding of far-UVC interactions is based on stratospheric chemistry, leaving many unknowns about indoor impacts.
• Sensitive, non-targeted measurements in diverse environments are needed to detect unexpected chemical changes under far-UVC exposure.

Current Consensus on Indoor Ozone Emissions

The consensus that generating ozone indoors ought to be avoided wherever possible predates the emergence of far-UVC. The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) has maintained this position for years and reaffirmed it in 2024.

Certain air cleaners produce ozone by design to achieve air-cleaning effects and the removal of contaminants. Ozone can also be produced as a by-product of air-cleaning processes. Any air-cleaning device that uses electricity during the air-cleaning process has the potential to generate ozone.

In practice, ozone generation is associated with air cleaners that use high-voltage coronas or pin ionizers, some precipitators or ionizers, UV light of sufficiently short wavelength, some photocatalytic oxidizers and UV-C air cleaners, and some plasma air cleaners.

Ozone is harmful for health and exposure to ozone creates risk for a variety of symptoms and diseases associated with the respiratory tract. Many products of ozone homogeneous and heterogeneous reaction processes also create risks for health, including formaldehyde, unsaturated aldehydes, and ultrafine particles such as secondary organic aerosols.

Ozone emission is therefore considered undesirable. However, there is no consensus on the safe level of ozone. ASHRAE’s Environmental Health Committee has previously suggested that “safe ozone levels would be lower than 10 ppb” and that the introduction of ozone to indoor spaces should be reduced to as low as reasonably achievable (ALARA) levels.

There is relatively little known about the long-term effects of exposure to low concentrations of ozone. Current understanding strongly suggests that air cleaners that emit ozone by design should not be permitted, and similar advice is provided by the US EPA.

There is more uncertainty about recommendations for air cleaners that do not use ozone by design for air cleaning but produce ozone unintentionally as a by-product of their operation. Some devices may emit ozone while simultaneously reducing concentrations of other harmful contaminants.

The state of the science does not allow highly certain trade-offs to be made between increased exposure to ozone and ozone reaction by-products and reduced exposure to other contaminants.

In the absence of robust information regarding safe levels of ozone, the precautionary principle should be used. Any ozone emission beyond a trivial amount that any electrical device can emit should be seen as a negative, and use of an ozone-emitting air cleaner, even when ozone is an unintentional by-product, may represent a net negative impact on indoor air quality and should be used with caution. If possible, non-ozone-emitting alternatives should be used.

At the beginning of the coronavirus pandemic, CDC Ventilation in Buildings guidance recommended that air-cleaning technologies employed should have UL 2998 certification, a “Zero Ozone Emissions” test that is discussed later in the Guidance, standards, and regulations section.

Given the uncertain but potentially meaningful effect sizes in the epidemiological literature on ozone exposure, and that prior to far-UVC there has been no compelling reason to use ozone-generating devices indoors, it is reasonable that this consensus has emerged.

The potential of far-UVC to have benefits that could outweigh these risks requires the careful tabulation of costs and benefits in a manner that has not previously been required.

Ozone Mitigation

Commercial off-the-shelf technology for removing ozone from the air already exists, although it is not widely employed outside of commercial aviation. The technology works by causing chemical reactions with ozone that create harmless products, as opposed to the potentially harmful products of other oxidation reactions that would otherwise occur.

These technologies likely require filtration of other contaminants in order to maintain their efficiency at removing ozone, and therefore it is possible that their use could synergize with the use of mechanical filtration, such as HEPA, in a combined air-cleaning solution.

If ozone mitigation proves to be necessary for far-UVC to reach its potential, this area requires more research and development.

Given the potential harms of indoor ozone exposure even in the absence of far-UVC, the widespread deployment of ozone removal technologies could potentially be justified on their own terms. However, such an analysis is outside the scope of this document.

Activated Carbon

Activated carbon consumes ozone through oxidation reactions resulting in either carbon dioxide (CO₂) or carbon monoxide (CO) products. While both CO₂ and CO can pose health risks, they only do so in quantities vastly higher than those plausibly produced by these reactions.

This is an instructive case for the dictum “the dose makes the poison.” Taking the National Ambient Air Quality Standard for protection of public health as an example, the maximum recommended average concentration over 8 hours for CO is 9 parts per million, whereas for ozone it is 70 parts per billion. Similarly, ambient levels of outdoor CO₂ are typically in the range of 300–500 ppm. Transforming ozone into CO and/or CO₂ is therefore not a cause for concern.

Readily available activated carbon-based commercial devices are expected to be effective at removing ozone, and these are currently being tested under challenge with far-UVC devices in research labs.

The main drawbacks of activated carbon filters are the potential bulk of an adequately sized filter and that, like any filter, they will need periodic replacement.

Activated carbon filters can also remove other products of concern, such as VOCs. If other photochemical interactions are discovered that produce products of concern, then using activated carbon filters alongside far-UVC could mitigate their impact as well.

However, this feature has a drawback. While it is possible for there to be minimal degradation in the performance of activated carbon filters in removing ozone over several years, high VOC loads will reduce the filters’ ozone removal efficiency over time.

Performance will also degrade if the filter accumulates dust and particulate matter, necessitating the additional use of other forms of filtration, such as HEPA or MERV 13, to protect the performance of the carbon filter.

Performance can also be affected by temperature and relative humidity, and there is a tradeoff between putting more airflow through the filter and the amount of ozone removed every time air passes through the filter. In other words, the single-pass efficiency of ozone removal declines as airflow increases.

Catalysts

The other primary method of ozone removal is through catalytic decomposition using a metal oxide catalyst. Ozone adsorbs to the catalyst to initially form a peroxide or superoxide intermediate species, which then desorbs as O₂.

This is not to be confused with photocatalytic air cleaners, which can potentially create a number of unintended byproducts.

Catalysts will also remove some VOCs and may be particularly effective at removing formaldehyde, which can undergo oxidation and decomposition into carbon dioxide and water.

Catalysts can be either “passive” or “active.” In the case of an active catalyst, air is actively moved with a fan over a catalytic surface, much as air is actively moved through a filter.

Passive catalysts, on the other hand, use exposed building materials such as clay-based paint that will catalytically decompose ozone as the air circulates naturally around a room.

Active catalyst devices can likely achieve removal rates comparable to activated carbon filters, whereas the potential for passive catalysts requires more investigation and will be affected by airflow patterns in the room as well as how much of the interior surface can be coated with the catalyst.

Catalyst performance can degrade over time for a number of reasons, all of which ultimately come down to a reduction in the surface area that is available for catalytic reactions to take place.

For example, contaminants such as lubricating oils, hydraulic fluids, and de-icing agents have been shown to reduce available catalytic surface area and degrade catalyst performance in commercial aviation applications.

In indoor spaces, the most likely source of soiling will be dust and particulate matter. For an active catalyst, this could be mitigated by pre-filtering the air to maintain performance. However, unless the filter contained sufficient activated carbon, this would not prevent semi-volatile organic compounds (SVOCs) from potentially soiling the catalyst.

In the case of passive catalysts, cleaning the catalytic surface may be necessary to maintain performance. Studies of long-term catalyst performance in real-world environments are warranted.

Key Takeaways

• Ozone can be effectively removed from indoor air using activated carbon filters and metal oxide catalysts, but this adds additional costs and requires energy to move the air through the filter or catalyst.
• Mitigating indoor ozone from outdoor sources may have public health benefits independent of mitigating ozone from indoor sources like far-UVC specifically.
• Further development of these technologies is needed to ensure their effectiveness in real-world applications.

Further Reading

• Model Evaluation of Secondary Chemistry due to Disinfection of Indoor Air with Germicidal Ultraviolet Lamps
• Indoor Air Quality Implications of Germicidal 222 nm Light
• Ozone Generation from a Germicidal Ultraviolet Lamp with Peak Emission at 222 nm
• Significant Production of Ozone from Germicidal UV Lights at 222 nm
• Ozone Loss: A Surrogate for the Indoor Concentration of Ozone-Derived Products
• Ozone and Ultra-Fine Particle Concentrations in a Hotel Quarantine Facility During 222 nm Far-UVC Air Disinfection

References

1. Watanabe, K., Inn, E. C. Y. & Zelikoff, M. Absorption coefficients of oxygen in the vacuum ultraviolet. J. Chem. Phys. 21, 1026–1030 (1953).
2. Peng, Z., Miller, S. L. & Jimenez, J. L. Model Evaluation of Secondary Chemistry due to Disinfection of Indoor Air with Germicidal Ultraviolet Lamps. Environ. Sci. Technol. Lett. 10, 6–13 (2022).
3. Barber, V. P. et al. Indoor air quality implications of germicidal 222 nm light. Environ. Sci. Technol. 57, 15990–15998 (2023).
4. Link, M. F., Shore, A., Hamadani, B. H. & Poppendieck, D. Ozone Generation from a Germicidal Ultraviolet Lamp with Peak Emission at 222 nm. Environ. Sci. Technol. Lett. (2023) doi:10.1021/acs.estlett.3c00318.
5. Peng, Z. et al. Significant Production of Ozone from Germicidal UV Lights at 222 nm. Preprint at https://doi.org/10.1101/2023.05.13.23289946 (2023).
6. Nazaroff, W. W. & Weschler, C. J. Indoor ozone: Concentrations and influencing factors. Indoor Air 32, e12942 (2022).
7. Lee, K., Vallarino, J., Dumyahn, T., Ozkaynak, H. & Spengler, J. D. Ozone Decay Rates in Residences. J. Air Waste Manag. Assoc. 49, 1238–1244 (1999).
8. Yao, M. & Zhao, B. Surface removal rate of ozone in residences in China. Build. Environ. 142, 101–106 (2018).
9. Weschler, C. J. & Nazaroff, W. W. Ozone Loss: A Surrogate for the Indoor Concentration of Ozone-Derived Products. Environ. Sci. Technol. 57, 13569–13578 (2023).
10. Sørensen, S. B., Dalby, F. R., Olsen, S. K. & Kristensen, K. Influence of germicidal UV (222 nm) lamps on ozone, ultrafine particles, and volatile organic compounds in indoor office spaces. Environ. Sci. Technol. 58, 20073–20080 (2024).
11. Nazaroff, W. W. Residential air-change rates: A critical review. Indoor Air 31, 282–313 (2021).
12. Wallace, L. A., Emmerich, S. J. & Howard-Reed, C. Continuous measurements of air change rates in an occupied house for 1 year: the effect of temperature, wind, fans, and windows. J. Expo. Anal. Environ. Epidemiol. 12, 296–306 (2002).
13. Xia, N., Du, E., Guo, Z. & de Vries, W. The diurnal cycle of summer tropospheric ozone concentrations across Chinese cities: Spatial patterns and main drivers. Environ. Pollut. 286, 117547 (2021).
14. Kalliomäki, P. et al. Ozone and ultra-fine particle concentrations in a hotel quarantine facility during 222 nm far-UVC air disinfection. medRxiv (2023) doi:10.1101/2023.09.29.23296366.
15. Link, M. F. et al. Ozone generation and chemistry from 222 nm germicidal ultraviolet light in a fragrant restroom. Environ. Sci. Process. Impacts (2024) doi:10.1039/D4EM00144C.
16. Narouei, F. et al. Effects of germicidal far-UVC on indoor air quality in an office setting. ChemRxiv (2024) doi:10.26434/chemrxiv-2024-l6k59.
17. Claus, H. Ozone generation by ultraviolet lamps. Photochem. Photobiol. 97, 471–476 (2021).
18. Weschler, C. J. & Nazaroff, W. W. Human skin oil: a major ozone reactant indoors. Environ. Sci. Atmos. 3, 640–661 (2023).
19. Dockery, D. W. et al. An Association between Air Pollution and Mortality in Six U.S. Cities. N. Engl. J. Med. 329, 1753–1759 (1993).
20. Weichenthal, S. et al. How low can you go? Air pollution affects mortality at very low levels. Sci. Adv. 8, eabo3381 (2022).
21. Sigsgaard, T. & Hoffmann, B. Assessing the health burden from air pollution. Science 384, 33–34 (2024).
22. Link, M. F., Robertson, R. & Poppendieck, D. Quantifying Byproduct Formation from Portable Air Cleaners Using a Proposed Standard Test Method. NIST 58, (2024).
23. US Environmental Protection Agency Office of Air and Radiation. Particulate Matter (PM2.5) Trends. https://www.epa.gov/air-trends/particulate-matter-pm25-trends (2016).
24. Jenks, O. J. et al. Effects of 222 nm Germicidal Ultraviolet Light on Aerosol and VOC Formation from Limonene. ACS ES&T Air 820–828 (2024).
25. Goss, M. & Kroll, J. Organic aerosol formation from 222 nm germicidal light: ozone-initiated vs. non-ozone pathways. ChemRxiv (2024) doi:10.26434/chemrxiv-2024-j0vp9.
26. Link, M. F., Robertson, R., Claflin, M. S. & Poppendieck, D. Quantification of byproduct formation from portable air cleaners using a proposed standard test method. Environ. Sci. Technol. 58, 7916–7923 (2024).
27. Logue, J. M., McKone, T. E., Sherman, M. H. & Singer, B. C. Hazard assessment of chemical air contaminants measured in residences. Indoor Air 21, 92–109 (2011).
28. Pytel, K., Marcinkowska, R. & Zabiegała, B. Investigation on air quality of specific indoor environments—spa salons located in Gdynia, Poland. Environ. Sci. Pollut. Res. Int. 28, 59214–59232 (2021).
29. Rosales, C. M. F. et al. Chemistry and human exposure implications of secondary organic aerosol production from indoor terpene ozonolysis. Sci. Adv. 8, eabj9156 (2022).
30. Hauptmann, M. et al. Mortality from lymphohematopoietic malignancies and brain cancer among embalmers exposed to formaldehyde. J. Natl. Cancer Inst. 101, 1696–1708 (2009).
31. Formaldehyde. https://wwwn.cdc.gov/TSP/PHS/PHS.aspx?phsid=218&toxid=39.
32. ECOTOX. https://cfpub.epa.gov/ecotox/.
33. Zannoni, N. et al. The human oxidation field. Science 377, 1071–1077 (2022).
34. ASHRAE Position Document on Filtration and Air Cleaning. Preprint at https://www.ashrae.org/file%20library/about/position%20documents/filtration-and-air-cleaning-pd-2024.pdf (2015).
35. CDC. About Ventilation and Respiratory Viruses. CDC.gov https://www.cdc.gov/niosh/ventilation/about/index.html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fcoronavirus%2F2019-ncov%2Fcommunity%2Fventilation.html (2024).
36. Weisel, C., Weschler, C. J., Mohan, K., Vallarino, J. & Spengler, J. D. Ozone and ozone byproducts in the cabins of commercial aircraft. Environ. Sci. Technol. 47, 4711–4717 (2013).
37. Deitz, V. R. & Bitner, J. L. Interaction of ozone with adsorbent charcoals. Carbon N. Y. 11, 393–401 (1973).
38. Wang, X. et al. Key factors and primary modification methods of activated carbon and their application in adsorption of carbon-based gases: A review. Chemosphere 287, 131995 (2022).
39. US Environmental Protection Agency. NAAQS Table. EPA.gov. https://www.epa.gov/criteria-air-pollutants/naaqs-table (2014).
40. Weschler, C. J., Shields, H. C. & Naik, D. V. Ozone-Removal Efficiencies of Activated Carbon Filters after More than Three Years of Continuous Service. (1994).
41. Metts, T. A. & Batterman, S. A. Effect of VOC loading on the ozone removal efficiency of activated carbon filters. Chemosphere 62, 34–44 (2006).
42. Tang, M., Siegel, J. A., Corsi, R. L. & Novoselac, A. Impacts of usage conditions on performance of ozone removal devices in ventilation systems. Build. Environ. 241, 110460 (2023).
43. Dhandapani, B. & Oyama, S. T. Gas phase ozone decomposition catalysts. Appl. Catal. B 11, 129–166 (1997).
44. Collins, D. B. & Farmer, D. K. Unintended consequences of air cleaning chemistry. Environ. Sci. Technol. 55, 12172–12179 (2021).
45. Zhu, S., Wang, J. & Nie, L. Progress of catalytic oxidation of formaldehyde over manganese oxides. ChemistrySelect 4, 12085–12098 (2019).
46. Gall, E. T., Corsi, R. L. & Siegel, J. A. Barriers and opportunities for passive removal of indoor ozone. Atmos. Environ. (1994) 45, 3338–3341 (2011).
47. Kunkel, D. A. et al. Passive reduction of human exposure to indoor ozone. Build. Environ. 45, 445–452 (2010).
48. Darling, E. & Corsi, R. L. Field-to-laboratory analysis of clay wall coatings as passive removal materials for ozone in buildings. Indoor Air 27, 658–669 (2017).
49. Darling, E., Morrison, G. C. & Corsi, R. L. Passive removal materials for indoor ozone control. Build. Environ. 106, 33–44 (2016).
50. Catalysts. https://www.basf.com/global/en/products/segments/surface_technologies/catalysts.

7. Ozone Epidemiology

Summary

While far-UVC devices generate a small quantity of ozone relative to prevailing levels of outdoor ozone (see Ozone and indoor air quality section), this does not necessarily mean it is insignificant to human health.

We have far fewer studies on the potential harms of ozone exposure relative to other air pollutants, such as fine particulate matter. However, the evidence we do have suggests that ozone has an exposure-response relationship with all-cause mortality. This needs to be taken into account when using far-UVC, with a particular focus on the potential risks of the products of secondary chemistry in low ventilation environments.

The magnitude of potential harms from indoor ozone generation is uncertain, but potentially large enough that it should be explicitly factored into cost-benefit calculations. However, the potential harms are not so large that they cannot be significantly outweighed by benefits if far-UVC is highly effective at preventing the spread of respiratory disease.

Crucial Considerations

• Indoor ozone and products of ozone loss are correlated with outdoor ozone levels.
• Exposure to any of outdoor ozone, indoor ozone, and the products of ozone loss may contribute to the health associations found in the outdoor ozone epidemiological literature.
• Thresholds below which ozone is benign, if they exist, may be different in different contexts depending on the causes of the association between ozone and adverse health outcomes. It is not necessarily the case that a “threshold effect” based on outdoor ozone epidemiology means that such a threshold can naively be applied indoors.
• Depending on these relative contributions, small long-term increases in exposure to ozone indoors could be associated with an increase in all-cause mortality.
• These effects may be higher in some populations, such as infants or the elderly.
• If the health risk associations found for outdoor ozone are caused by the products of ozone loss, then the potential harms of indoor ozone generation will be more severe in poorly ventilated spaces.
• If the health risk associations for outdoor ozone are caused by direct ozone exposure, then increasing ventilation could exacerbate the problem by bringing in ozone from outside and further increasing indoor ozone concentration.
• Technology that mitigates ozone formation directly ought to reduce potential health risks no matter the causal mechanism.

Analysis

Epidemiology of Outdoor Ozone

Long-term epidemiological studies estimate an “exposure-response relationship” arising from correlations between outdoor ozone levels, mortality, and other health impacts such as respiratory and cardiovascular disease.

In work related to far-UVC, the most commonly cited paper has been Turner et al., 2016, which estimated a hazard ratio (HR) for all-cause mortality of 1.02 for every 10 parts per billion (ppb) increase in average outdoor ozone levels in the United States. The data came from nearly 669,046 participants in a long-term cancer cohort study from 1982–2004.

There is also evidence of shorter-term increases in mortality due to increases in ozone exposure. In a study of 95 US urban communities, Bell et al., 2004 found a 0.52% increase in daily mortality following a 10 ppb increase in the previous week’s ozone level.

Two more recent studies on ozone mortality risk focus on specific subpopulations that could be at higher risk. Xue et al. investigated the exposure response of outdoor ozone levels to mortality in children under 5 in low- and middle-income countries, and estimated a HR for all-cause mortality of 1.064 for every 10 ppb, over three times higher than the estimate of Turner et al. However, this was only observed for outdoor concentrations above 50 ppb.

Zhang et al. estimated a HR of 1.097 for every 10 ppb of ozone in a study population of over-65s in China, where the average age was 87. This is five times higher than Turner et al.’s estimate, and with no threshold effect reported.

The negative health effects are observed across a wide variety of vascular and respiratory diseases.

Table 7.1. All-Cause and Cause-Specific Mortality Hazard Ratios

While the 1.03 HR for diseases of the circulatory system is lower than the 1.12 HR for diseases of the respiratory system, the absolute mortality due to circulatory disease is approximately 30% higher than respiratory disease due to the substantially higher underlying level of risk.

Approximately one quarter of the estimated impact on deaths from respiratory illness are accounted for by pneumonia and influenza, demonstrating that infectious disease and air quality are not separate phenomena, but interact with each other.

Asthma

A recent meta-analysis by Zheng et al. estimated a hazard ratio of 1.008 for the effect of a 10 μg/m³ increase in either peak 8-hour outdoor ozone or average 24-hour outdoor ozone on emergency room visits or hospitalizations for asthma. Assuming 1 ppb of ozone is equal to 1.96 μg/m³, the hazard ratio for these outcomes is therefore 1.016 per 10 ppb.

Key Takeaways

• Studies suggest that higher outdoor ozone levels are associated with increased mortality from cardiovascular and respiratory diseases.
• Children under 5 and the elderly may experience higher risks from ozone.

Translating Outdoor Ozone Epidemiology Indoors

Recalling the equation for indoor ozone concentrations from Weschler and Nazaroff, and the equation for “ozone loss” from chemical reactions, if we assume that λ_ACH (ventilation rate) and k_sum (ozone decay due to chemical reactions) are independent of outdoor ozone levels, then any change in outdoor ozone will correlate to changes in indoor ozone, ozone loss and the products of ozone loss.

Therefore, even if we knew that the hazard ratio was caused by variations in outdoor ozone rather than them just being correlated, it would not be possible using epidemiological studies to determine whether outdoor ozone directly causes these health effects or whether it does so through its impact on indoor ozone or ozone loss, and therefore what the exposure-response relationship is for each factor independently.

There are several possibilities for where the balance of the harm falls:

• There might be a concentration threshold below which the antioxidant capacity of our respiratory system can handle breathing in ozone without harmful long-term effects. If this is the case, we would expect the impact of exposure outdoors, where concentrations are higher, to be greater than that of indoor ozone, where concentrations are lower.
• There might be no such threshold. If this is the case, because we spend much more of our time indoors, even though the absolute level of ozone indoors is lower than outdoors, indoor ozone makes up the majority of our exposure and therefore the harm observed.
• The secondary products of ozone loss may be much worse for human health than ozone itself, and therefore the harm may be associated with ozone loss.

Whatever the health effects are from the estimated ~1–5 μg/m³ of secondary organic aerosol for every ~1 μW/cm² of fluence rate in low ventilation conditions, or from other ozone loss products, if these effects are caused only by ozone-initiated chemistry, they are already contained within the hazard ratio from outdoor ozone epidemiology.

The only effects of far-UVC on indoor air quality that would not be “contained” within this hazard ratio are products that cannot be explained through ozone-initiated chemistry. For this reason, it is important to establish whether modeling ozone-initiated chemistry is sufficient to explain real-world observations of chemical changes in occupied spaces.

Insofar as we have separate toxicology data on the exposure-response relationship for secondary organic aerosol (SOA)—one of the products of ozone loss—it is entirely in the context of outdoor air pollution. As SOA is a type of PM_2.5, epidemiological studies on the health effects of outdoor PM_2.5 are often used to estimate the health effects of SOA production.

Outdoor PM_2.5 will also be present indoors through infiltration and ventilation, and therefore these epidemiological studies factor in both exposure to PM_2.5 outdoors and indoors, as they do with ozone.

But even supposing we had correctly decomposed the epidemiological exposure-response relationship for PM_2.5, there are many reasons for skepticism that this would translate to the exposure-response relationship of SOA produced from ozone loss.

There is evidence that smaller particles, often referred to as Ultra Fine Particles or PM_0.1, could be more harmful than PM_2.5. Toxicology studies in mice suggest that the relative harm of toxic substances is proportional not to mass but surface area, and the surface area to mass ratio will increase exponentially with smaller particle size.

The particle size distribution of outdoor PM_2.5 and particulate matter produced under far-UVC irradiation indoors will likely be different.

The implicit assumption of outdoor PM_2.5 epidemiology is that the particular chemical composition of the particles does not matter. This assumption is less likely to hold when translating to a different chemical context.

For example, hydrophobic and hygroscopic particles deposit at different rates in lungs and there are also differences in the oxidative potential of different aerosols. The specific chemical composition of outdoor PM_2.5 may generally be similar enough over time that it makes sense to assume no effect from the particular chemical composition of the aerosol mass nor any effect from size distribution. However, it is not clear that this is valid for indoor-sourced PM.

While in the United States today outdoor PM_2.5 is now primarily constituted by secondary organic aerosol, especially in the summertime, in many of the studies used to assess the harms of PM exposure the secondary organic aerosol will constitute a minority of the total composition, since this was true in the past.

Furthermore, we should not assume that the harm of ozone loss comes primarily from inhaling aerosol or gas phase products into our respiratory system.

While the hazard ratio from outdoor ozone for mortality due to diseases of the circulatory system is lower than for the respiratory system, the absolute mortality from circulatory diseases is higher due to having a higher base rate.

Roughly half of the ozone loss to surface reactions remain on surfaces, and the potential harms of those reaction products, which could for example be ingested after contact with surfaces, is also not known. For gas phase products, dermal absorption may also be an important pathway to products entering the circulatory system.

The data in Turner et al., 2016 suggests that either SOA from ozone loss does not have the same harms as outdoor PM_2.5, or SOA from ozone is not a material cause of the association between outdoor ozone and all-cause mortality.

This is because while the ozone HR is four times higher for respiratory disease (1.12 per 10 ppb) relative to diseases of the circulatory system (1.03 per 10 ppb), it is the inverse for near-source PM_2.5—the HR of 1.17 per 10 µg/m³ for respiratory disease is 60% lower than the HR of 1.41 for diseases of the circulatory system.

If the ozone hazard ratio was actually caused by SOA from ozone loss, and SOA from ozone loss had the same effect on health as PM_2.5, then we would expect the causes of mortality to be similar.

One potential product of ozone loss on which there is some epidemiological data is formaldehyde. Formation of formaldehyde under far-UVC illumination has been observed in artificial experimental conditions.

The epidemiological data is inconsistent on whether formaldehyde constitutes a risk for leukaemia or other forms of cancer. However, cancer risk is not one of the mortality causes with which ozone exposure is associated in the epidemiological literature, suggesting that whatever the cause of the association with mortality is, it is not through the most concerning potential risks of formaldehyde exposure.

If ozone loss did prove to account for a significant proportion of the harm of outdoor ozone, then this would be particularly challenging to characterize. The exact quantity and chemical composition of the products of ozone loss could vary dramatically in different indoor spaces, far more than the variation in indoor ozone concentrations.

The challenging nature of this research means that it is vital to make assumptions clear and explain calculations and methodologies in detail when making an estimate of the potential health effects of indoor ozone generation.

Until we can decompose the cause of the association between ozone and mortality in outdoor ozone epidemiology, we cannot even be confident whether using ventilation to mitigate ozone production from far-UVC lamps is beneficial.

If it is the products of ozone loss that cause the harm, then moderate levels of ventilation ought to meaningfully reduce potential unintended harms from the use of far-UVC, given the strong dependence of ozone loss on ventilation.

If instead it is the presence of the ozone itself that causes the harm, then additional ventilation can further increase the indoor ozone concentration by bringing more ozone in from outdoors.

If it is ozone itself that causes the harm, it would also raise the question of whether in general the use of ventilation or far-UVC causes the greatest reduction in pathogens for the lowest increase in ozone exposure, particularly if outdoor ozone levels are elevated.

Key Takeaways

• The relationship between the health effects of outdoor ozone exposure, indoor ozone exposure and exposure to the products of ozone loss is unclear.
• More ventilation could either reduce or increase health risks, depending on whether ozone or its byproducts are the main concern.

Is There a Threshold Below Which Ozone Is Benign?

If there was a threshold below which there was limited or no evidence of harm from ozone, then this could function as a de facto exposure limit for indoor air quality under far-UVC irradiation. So long as ozone concentration did not exceed this threshold, then we could be confident that there was no significant harm to human health from far-UVC-generated ozone.

On the other hand, it’s possible that there is no threshold below which ozone is benign, and that the harms increase linearly with exposure at all concentrations.

Threshold Effects from Outdoor Ozone Epidemiology

There is mixed evidence on threshold effects for ozone exposure. Xue et al. found a substantial hazard ratio for all-cause mortality in children under 5—1.064 for every 10 ppb increase—but this was only for levels above 50 ppb, suggesting a threshold effect.

By contrast, Zhang et al. found a linear exposure-response, meaning no threshold, among older adults in China.

Turner et al. tested extensively for potential thresholds. The authors found only limited evidence of a threshold, with some evidence that a threshold model improved model fit for respiratory mortality at 35 ppb. Results were somewhat suggestive of a threshold for circulatory mortality at 35 ppb, but the evidence was weaker.

For the effects of outdoor ozone on asthma emergency room visits and hospital visits, only one paper estimated a threshold, at 40 ppb.

There is precedent in other living organisms for threshold effects of ozone exposure. Ozone is also known to affect crop yields, with different crops being sensitive to ambient ozone concentrations at different thresholds.

We cannot directly apply these plant thresholds to humans, but this demonstrates that organisms plausibly have resistance or repair mechanisms that protect against lower levels of oxidant exposure.

At best, the evidence for a threshold is mixed. If there is such a threshold, it may be somewhere between 35 ppb and 50 ppb of outdoor ozone.

Translating Outdoor Ozone Thresholds Indoors

Suppose for the sake of argument that there was a threshold effect at 35 ppb. What would this mean for increases in indoor ozone under far-UVC irradiation? It depends on the decomposition of ozone harms into [O₃]_out (ozone from outside), [O₃]_in (ozone generated inside) and [O₃]_loss (ozone reacting and changing into other products).

The literature summary in Nazaroff and Weschler finds median indoor ozone levels in residences of 6 ppb (Interquartile Range (IQR) 2–11 ppb), with outdoor ozone levels of 22 ppb (IQR 16–35 ppb) and a median indoor/outdoor ratio of 25% (IQR 11–37%).

If a 35-ppb outdoor threshold was supported by the epidemiological literature, then the only thing we could say with confidence would be that indoor ozone levels and/or ozone loss consistent with outdoor ozone being 35 ppb are benign. This is anywhere around 4–10 ppb of indoor ozone based on the IQR.

It is certainly possible that the entire harm comes from [O₃]_out, but this cannot be deduced one way or the other from the epidemiological evidence.

Therefore, we cannot rule out that single-digit increases starting from single-digit ppb concentrations of indoor ozone—such as could be expected from typical uses of far-UVC—could be harmful, even if there is a threshold effect in outdoor ozone epidemiology that is much higher than typical indoor levels.

Key Takeaways

• Some evidence suggests a possible threshold level of outdoor ozone below which there is no harm, but the data is mixed, and the threshold may vary between different populations and settings.
• Even if there’s a threshold effect in outdoor ozone epidemiological studies, it’s uncertain whether and how this applies to indoor ozone levels.

We Should Be Very Uncertain About the Harm of Ozone Exposure

We should place large uncertainty bounds around the health effects of the unintended air chemistry of far-UVC. None of the estimates quoted above are based on studies from which we can directly infer causality, and a number of the assumptions that go into calculating proportional hazard ratios are unlikely to hold.

Even leaving aside causation, there is significant uncertainty about the scale of the association.

For example, Turner et al., 2016 used information from participants in a large (n=669,046) prospective cancer study that began in 1982. However, this was not the first time that dataset was used to estimate ozone harms.

A previous analysis of that same dataset found a higher hazard ratio for PM_2.5 harms, but ozone was actually negatively associated with all-cause mortality, although this hid a positive association with respiratory illness and a negative association with cardiovascular disease.

The Turner et al. analysis is superior, using newer and more sophisticated estimates of ambient ozone, PM_2.5, and NO₂ concentrations to more accurately estimate individual exposure, as well as 50% more participants and nearly double the number of deaths.

However, this is an enormous update in hazard ratio over a short time period between studies using similar methodologies, and our confidence that this latest estimate is definitively correct should be tempered appropriately.

Ozone epidemiology is an intrinsically challenging subject to study, in no small part due to large numbers of potential confounding variables.

Ambient ozone levels are not randomly distributed across geographic regions—they vary due to a combination of emissions sources, atmospheric chemistry, weather patterns, and altitude. As a result, health outcomes associated with ozone exposure can be influenced by a range of factors, including socioeconomic status, baseline air quality, and local climate conditions that affect both outdoor and indoor exposure levels.

These hidden correlations can make it challenging to determine the precise health risks of ozone exposure, as well as the extent to which behaviors and environmental conditions may amplify or mitigate its effects.

Earlier studies have also shown different results to Turner et al. One meta-analysis identified fifteen studies going back to 1999, of which nine were included. This analysis found no association between average ozone levels and all-cause mortality, though it found a similar association to Turner et al. between mortality and peak ozone exposure.

This could also represent some evidence of a threshold effect—if threshold effects were material then we would potentially expect peak ozone to have stronger effects than average concentrations.

The importance of NO_x emissions in increasing tropospheric ozone levels means that variations in local ozone will be related to variations in local NO_x. Vehicles are the most important source of NO_x emissions in urban areas, and therefore locally NO_x levels will vary spatially with proximity to roads and temporally with traffic conditions.

Local spatiotemporal variations in NO_x and therefore ozone levels mean that regional ozone monitors and models will in many cases not represent the conditions to which individuals are actually exposed.

With all of these caveats in place, we nevertheless think that it is responsible to attempt to bound the potential harms of ozone generation from far-UVC were it to be ubiquitously deployed, using the best available estimates from the epidemiological literature.

Key Takeaways

• The health effects of ozone, especially indoors, are highly uncertain due to many confounding factors.
• Different studies offer varying estimates of ozone’s health risks, with some earlier studies showing no clear association between ozone and mortality.

Bounding the Potential Harm of Far-UVC Ozone Generation

Far-UVC could substantially reduce the burden of respiratory illness and potentially prevent future pandemics. Given this context, it is important to estimate some bounds for the plausible extent of harm from indoor-generated ozone under far-UVC irradiation, so that we can compare these harms against the potential benefits of deploying far-UVC.

This section is not a full cost-benefit analysis, but is intended to gesture towards the importance of conducting such an analysis, and to highlight these calculations’ sensitivity to critical assumptions.

Note: The following should be taken as possible orders of magnitude estimates only in a hypothetical scenario where far-UVC was deployed ubiquitously.

Pivotal assumptions include whether there are threshold effects for ozone exposure, how to attribute the harm of outdoor ozone epidemiology to particular causes, and whether exposure to ozone and ozone loss is ubiquitously and permanently increased.

The best-case scenario for potential harms from indoor-generated ozone under far-UVC irradiation would be if the harm comes from ozone itself rather than products from ozone loss, and if there is a threshold effect, and if most indoor spaces—even under far-UVC irradiation—are below that threshold.

In the vast majority of situations with normal levels of outdoor ozone, realistic far-UVC installations are not going to increase indoor ozone levels above 35–50 ppb.

Assuming instead that there is not such a threshold but that the cause of harm is still direct exposure to ozone itself, one estimate of harm would be to estimate the total quantity of individual exposure to ozone both outdoors and indoors, and decompose the epidemiological harm linearly with total indoor and outdoor exposure.

This is the methodology of Xiang et al., which attempted to model the hypothetical benefits of reducing indoor ozone exposures in urban China.

For this quantitative estimate of hypothetical harm under ubiquitous deployment, we will estimate the proportion of total ozone exposure that occurs indoors, and then rebase that hazard ratio to how indoor ozone changes with outdoor ozone—that is, a 10 ppb increase in outdoor ozone causes a less than 10 ppb increase in indoor ozone.

We estimate the proportion of total ozone exposure from time spent indoors using indoor and outdoor ozone concentrations and the amount of time spent indoors and outdoors.

For C_in, we use the median indoor/outdoor ozone ratio found in the Nazaroff and Weschler literature review: 25%, with C_out normalized to 1.

For T_in, a commonly used estimate for high-income countries is that people spend 90% of their time indoors, and 70% in their homes.

Plugging these parameters into the equation yields an estimate that approximately 69% of total ozone exposure is accounted for by indoor ozone.

On these assumptions, approximately 69% of the hazard ratio would therefore be accounted for by indoor ozone exposure. To calculate the final hazard ratio for an increase of 1 ppb indoor ozone exposure, we would then need to rebase the hazard ratio to indoor ozone concentration rather than outdoor ozone concentration.

Taking an all-cause mortality increase of 2% per 10 ppb outdoors per Turner et al., and assuming 69% of that is caused by—and linear in response to—indoor ozone exposure, a permanent and ubiquitous increase in indoor ozone of 1 ppb would increase all-cause mortality by approximately 0.55%.

It is extremely important when it comes to responsible use of far-UVC that this number is substantially lower if ozone concentrations were only increased in nonresidential indoor public spaces.

Taking the same data but only using the approximately 20% of time on average that people spend indoors in places that are not their home:

Rebasing the hazard ratio to indoor ozone concentrations rather than outdoors, this would represent approximately 4,000 excess deaths per annum in the United States for every 1 ppb increase in ozone across all nonresidential indoor public spaces, assuming approximately 3.27 million baseline deaths per annum.

It is important to put this into context. This is an estimate for if far-UVC was deployed in every square foot of indoor public space in the United States without any attempt to mitigate ozone production and/or ensure adequate ventilation.

CDC estimates up to 50,000 deaths per annum from flu, and it is likely that endemic COVID-19 is adding another 50,000 on top of this. This is before we estimate the value of preventing the transmission of the next pandemic pathogen.

Furthermore, if we target far-UVC deployment in spaces with the highest infection risk, then we can potentially achieve a proportionally greater effect on reducing infectious disease while limiting exposure to ozone. Lamps may not need to be switched on at full power at all times, further limiting ozone exposure.

These simple but flawed estimates of potential mortality risk from small changes in chronic ozone exposure suggest that we should take this concern seriously when deploying far-UVC.

It should be factored into cost-benefit analyses, with appropriate uncertainty and caveats. But if far-UVC proves to be highly effective at suppressing the transmission of respiratory pathogens, such that we could meaningfully reduce the burden of pandemic and endemic respiratory disease, these estimates suggest that there could be a substantial net benefit in terms of lives saved.

Asthma Emergency Room Visits or Hospital Admissions

Zheng et al. estimate a 1.6% increase in emergency room visits or hospitalizations for asthma per 10 ppb of outdoor ozone. If we apply the same linear exposure logic to increases in indoor ozone, the potential increase can be estimated using the same framework described above.

For nonresidential spaces only, the estimated effect is substantially smaller because people spend a smaller proportion of their total time in these environments.

In terms of scale, according to Nurmagambetov et al., the average annual cost of hospitalizations and emergency room visits per person with treated asthma in the United States was $629, with inpatient admissions constituting the majority of the cost.

The study estimates that there are 15.4 million patients with treated asthma in the US, and multiplying these together produces an estimate of $9.7 billion in healthcare costs for treated asthma.

Therefore, we would estimate every 1 ppb ozone increase in all indoor public spaces, on this framework, to translate to approximately $97 million in asthma-related healthcare costs per annum.

What If the Harm Is Due to Ozone Loss?

Another possibility is that the harms detected in epidemiological studies are not due to ozone exposure but to exposure to the products of ozone chemistry indoors.

There is some suggestive evidence that this may be the case. Reanalysis of data from two studies—one of healthy adults and one of children with asthma—of biomarkers of cardiorespiratory pathology found more evidence of these biomarkers being associated with ozone loss rather than ozone exposure.

If this were generally the case, how would we estimate the hazard ratio?

All exposure to ozone loss is indoors, and taking a typical indoor/outdoor ozone ratio of 25%, or a typical ozone loss ratio of 75%, we can rebase the outdoor ozone hazard ratios on the level of ozone loss.

In scenarios with 1 ACH and k_sum (decay rate) of 2, this produces estimates for the total mortality increase of increasing indoor ozone by 1 ppb similar to those produced above using the linear ozone exposure method.

However, recalling the analysis in the Ozone and indoor air quality section, ozone loss is far more sensitive to ventilation than indoor ozone concentrations are.

If it is the case that the hazard ratio from outdoor ozone epidemiology is driven substantially by exposure to the products of ozone loss, this greatly magnifies the importance of ensuring adequate ventilation.

But it also means that with adequate ventilation, the harm from far-UVC could be lower than a linear ozone exposure model would predict.

Key Takeaways

• The potential harms from far-UVC-generated ozone are large enough that they should be considered in cost-benefit calculations, but small enough that they may be significantly outweighed by benefits if far-UVC is highly effective at preventing the spread of respiratory disease.
• Different causes of the harm imply different scales of harm in different circumstances, as well as different harm mitigation strategies.
• Estimates of potential harms are necessary for cost-benefit analysis, but there is no single correct way to make an estimate using the epidemiological data.

Further Reading

• Long-Term Ozone Exposure and Mortality in a Large Prospective Study

References

References

1. Turner, M. C. et al. Long-Term Ozone Exposure and Mortality in a Large Prospective Study. Am. J. Respir. Crit. Care Med. 193, 1134–1142 (2016).
2. Bell, M. L., McDermott, A., Zeger, S. L., Samet, J. M. & Dominici, F. Ozone and short-term mortality in 95 US urban communities, 1987-2000. JAMA 292, 2372–2378 (2004).
3. Xue, T. et al. Estimating the exposure-response function between long-term ozone exposure and under-5 mortality in 55 low-income and middle-income countries: a retrospective, multicentre, epidemiological study. Lancet Planet. Health 7, e736–e746 (2023).
4. Zhang, Y. et al. Ambient PM_2.5, ozone and mortality in Chinese older adults: A nationwide cohort analysis (2005-2018). J. Hazard. Mater. 454, 131539 (2023).
5. Zheng, X.-Y., Orellano, P., Lin, H.-L., Jiang, M. & Guan, W.-J. Short-term exposure to ozone, nitrogen dioxide, and sulphur dioxide and emergency department visits and hospital admissions due to asthma: A systematic review and meta-analysis. Environ. Int. 150, 106435 (2021).
6. Weschler, C. J. & Nazaroff, W. W. Ozone Loss: A Surrogate for the Indoor Concentration of Ozone-Derived Products. Environ. Sci. Technol. 57, 13569–13578 (2023).
7. Peters, A., Wichmann, H. E., Tuch, T., Heinrich, J. & Heyder, J. Respiratory effects are associated with the number of ultrafine particles. Am. J. Respir. Crit. Care Med. 155, 1376–1383 (1997).
8. Schmid, O. & Stoeger, T. Surface area is the biologically most effective dose metric for acute nanoparticle toxicity in the lung. J. Aerosol Sci. 99, 133–143 (2016).
9. Löndahl, J. et al. Size-Resolved Respiratory-Tract Deposition of Fine and Ultrafine Hydrophobic and Hygroscopic Aerosol Particles During Rest and Exercise. Inhal. Toxicol. 19, 109–116 (2007).
10. Daellenbach, K. R. et al. Sources of particulate-matter air pollution and its oxidative potential in Europe. Nature 587, 414–419 (2020).
11. Bates, J. T. et al. Review of Acellular Assays of Ambient Particulate Matter Oxidative Potential: Methods and Relationships with Composition, Sources, and Health Effects. Environ. Sci. Technol. 53, 4003–4019 (2019).
12. Avery, A. M., Waring, M. S. & DeCarlo, P. F. Seasonal variation in aerosol composition and concentration upon transport from the outdoor to indoor environment. Environ. Sci. Process. Impacts 21, 528–547 (2019).
13. Dockery, D. W. et al. An Association between Air Pollution and Mortality in Six U.S. Cities. N. Engl. J. Med. 329, 1753–1759 (1993).
14. US Environmental Protection Agency Office of Air and Radiation. Progress Report – emissions reduction. https://www.epa.gov/power-sector/progress-report-emissions-reductions (2018).
15. Weschler, C. J. & Nazaroff, W. W. Dermal uptake of organic vapors commonly found in indoor air. Environ. Sci. Technol. 48, 1230–1237 (2014).
16. Link, M. F., Robertson, R. & Poppendieck, D. Quantifying Byproduct Formation from Portable Air Cleaners Using a Proposed Standard Test Method. NIST 58, (2024).
17. Beane Freeman, L. E. et al. Mortality from lymphohematopoietic malignancies among workers in formaldehyde industries: the National Cancer Institute Cohort. J. Natl. Cancer Inst. 101, 751–761 (2009).
18. Coggon, D., Harris, E. C., Poole, J. & Palmer, K. T. Extended follow-up of a cohort of British chemical workers exposed to formaldehyde. J. Natl. Cancer Inst. 95, 1608–1615 (2003).
19. Hauptmann, M., Lubin, J. H., Stewart, P. A., Hayes, R. B. & Blair, A. Mortality from solid cancers among workers in formaldehyde industries. Am. J. Epidemiol. 159, 1117–1130 (2004).
20. Pinkerton, L. E., Hein, M. J. & Stayner, L. T. Mortality among a cohort of garment workers exposed to formaldehyde: an update. Occup. Environ. Med. 61, 193–200 (2004).
21. Zu, K. et al. Concentration-response of short-term ozone exposure and hospital admissions for asthma in Texas. Environ. Int. 104, 139–145 (2017).
22. Heagle, A. S. Ozone and Crop Yield*. Annu. Rev. Phytopathol. 27, 397–423 (1989).
23. Tarasick, D. et al. Tropospheric Ozone Assessment Report: Tropospheric ozone from 1877 to 2016, observed levels, trends and uncertainties. Elementa (Wash., DC) 7, 39 (2019).
24. Nazaroff, W. W. & Weschler, C. J. Indoor ozone: Concentrations and influencing factors. Indoor Air 32, e12942 (2022).
25. Cox, L. A., Jr. Improving interventional causal predictions in regulatory risk assessment. Crit. Rev. Toxicol. 53, 311–325 (2023).
26. Jerrett, M. et al. Long-Term Ozone Exposure and Mortality. N. Engl. J. Med. 360, 1085–1095 (2009).
27. Huangfu, P. & Atkinson, R. Long-term exposure to NO₂ and O₃ and all-cause and respiratory mortality: A systematic review and meta-analysis. Environ. Int. 144, 105998 (2020).
28. Munir, S., Mayfield, M., Coca, D., Mihaylova, L. S. & Osammor, O. Analysis of Air Pollution in Urban Areas with Airviro Dispersion Model—A Case Study in the City of Sheffield, United Kingdom. Atmosphere (Basel) 11, 285 (2020).
29. Xiang, J. et al. Ozone in urban China: Impact on mortalities and approaches for establishing indoor guideline concentrations. Indoor Air 29, 604–615 (2019).
30. Klepeis, N. E. et al. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Sci. Environ. Epidemiol. 11, 231–252 (2001).
31. Stein, R. Is COVID endemic yet? Yep, says the CDC. Here's what that means. NPR (2024).
32. Nurmagambetov, T., Kuwahara, R. & Garbe, P. The Ecnomic Burden of Asthma in the United States, 2008–2013. Ann. Am. Thorac. Soc. 15, 348–356 (2018).
33. He, L. et al. Ozone Reaction Products Associated with Biomarkers of Cardiorespiratory Pathophysiology. Am. J. Respir. Crit. Care Med. 207, 1243–1246 (2023).

8. Emitters and Luminaires

Summary

Nature makes it easier to produce certain UVC wavelengths than others. The development of 222-nm krypton chloride excimer (KrCl*) emitters enabled a flurry of far-UVC research, and they are used nearly exclusively in commercial applications today.

KrCl*-based luminaires are currently expensive. To achieve average fluence rates likely to be highly effective for air disinfection, upfront costs today are on the order of thousands of dollars for every thousand square feet of indoor space protected. This is prohibitive for many applications.

However, there is scope to significantly reduce the effective annual cost per milliwatt of far-UVC output from KrCl* sources, without any fundamental technological breakthroughs. Basic improvements in product engineering could make KrCl* sources far more cost-effective.

Key developments would include a high-transmittance, long-lasting optical diffuser, dynamic regulation of optical output, and extension of lifetime. However, these improvements will only come about with additional private investments, which will only be made if there is promise of a substantial market.

Semiconductor-based emitters like light-emitting diodes (LED) and second harmonic generation (SHG) emitters offer potential advantages over KrCl* sources. These technologies can target different far-UVC wavelengths and be produced at chip-scale, enabling a wider variety of luminaire form factors.

SHG emitters in particular have the potential to produce very clean emission spectra, which would not require filtering. However, they are early in development, and the economics of semiconductor manufacturing are especially sensitive to scale.

Crucial Considerations

• Krypton chloride excimer (KrCl*) sources are currently the leading choice for far-UVC whole-room disinfection, and will likely be for the next 5–10 years. KrCl* emitters primarily produce far-UVC at a wavelength of 222 nm, but have some emissions outside the far-UVC range. High-quality filters minimize these emissions while maintaining high output at 222 nm.
• KrCl* luminaires could be significantly more cost-effective without fundamental technological breakthroughs. However, the required investment in product engineering needs to be financially justified by the prospect of a substantial market.
• Increasing the output of diffused luminaires is an important avenue of cost reduction for the safe use of far-UVC in spaces with typical 8–9 ft ceilings.
• Semiconductor-based technologies have some important benefits and are promising, but have technological challenges they need to overcome before they are able to replace KrCl* sources for whole-room disinfection.
• The improvements necessary for widespread deployment will likely occur if there is a market to sustain the industry and attract investment.

Note on Terminology

Elsewhere in the Blueprint, we have referred to far-UVC “lamps” in the colloquial sense, as in a floor lamp or a ceiling lamp. However, in illumination engineering, lamp exclusively refers to the light-producing component—that is, the “bulb”—of the final assembled luminaire or lighting fixture.

In this section of the text, emitter refers to the underlying radiative element, such as a krypton chloride excimer bulb or an LED chip, and luminaire refers to the final assembled lighting product, which includes optics, housing, power supply, etc.

To improve readability, where a generic term is preferable to specifying emitter or luminaire, we use source.

Analysis

What Are the Important Factors for a Good Far-UVC Source? Why Are They Important?

Upfront Cost

Cost is one of the main factors that affects how widely far-UVC sources will be adopted. High costs limit deployment in public spaces where far-UVC could offer significant public health benefits.

Driving down the cost of production through economies of scale, investments in product engineering, and new innovations will be crucial for cost-effective applications.

Power Output

The key metric for a far-UVC source is its total far-UVC output, typically in milliwatts (mW).

Lifetime & Reliability

Though an appropriately powerful far-UVC source is desirable, it is not beneficial if that source is short-lived and requires frequent replacement. Ideally, a far-UVC source should not need replacing for many years.

Spectral Purity

Spectral purity refers to the ability of a far-UVC source to emit at the target wavelength with minimal unwanted emissions at other wavelengths.

High spectral purity is critical for safety, as even small emissions in the 240–300 nm range can penetrate deeper into human tissues with potentially harmful effects. Conversely, more ozone is produced as wavelengths decrease, particularly below 200 nm.

High-quality filters are often needed to achieve spectral purity by blocking non-target wavelengths.

Radiant Intensity Distribution

Alongside its spectral content, a far-UVC source’s maximum safe and effective power will depend on its radiant intensity distribution.

Engineering a more uniform fluence rate helps prevent hotspots, which can either exceed safe exposure limits or force intensity reductions that leave surrounding areas underdosed.

This can be done by effectively diffusing a high-power far-UVC source, or distributing a greater number of low-power far-UVC sources throughout a space.

More uniform distribution may also increase far-UVC’s efficacy against pathogens, especially in spaces with low air mixing or high degrees of shadowing, but this requires further study.

Operational Flexibility

Operational flexibility refers to the ability to adjust the far-UVC source based on changing needs or environmental conditions.

For example, some spaces might require higher UV doses during peak infection seasons, or lower intensities when occupants are closer to the source. This requires that sources be readily dimmable without loss of other critical functions, such as reduction of effective lifetime.

Form Factor

The physical size, shape, and flexibility of a far-UVC source—its form factor—significantly influence where and how it can be deployed.

Another element of form factor is the inherent directionality of semiconductor-based emitters, which in principle allows for their radiation output to be used more efficiently.

Wavelength Selectability

The most widely used far-UVC wavelength is 222 nm, because that is the peak emission of KrCl* emitters. However, there is no specific reason to think that this is the ideal far-UVC wavelength when considering tradeoffs between efficacy, photobiological safety, and ozone generation.

KrCl* Excimer Lamps

Krypton chloride excimer (KrCl*) sources, which primarily emit far-UVC at a wavelength of 222 nm, are the most commonly employed sources in commercial applications today. The majority of far-UVC efficacy and safety research published in recent years also uses KrCl* sources.

Excimer—short for excited dimer—emitters are a type of gas-discharge lamp. A glass tube is filled with a gas, or mixture of gases—in this case, krypton and chlorine—a current is passed through the gas, exciting it, and finally excited particles return to the ground state, emitting a photon of the wavelength proportional to the energy.

Excimers require specific gas compositions—either a noble gas, or a mixture of a noble gas and a halogen—at fairly low pressures, and are driven by a pulsed high-voltage waveform.

While there are other ways to achieve excimer luminescence, the most common and successful type of excimer emitter involves the dielectric barrier discharge (DBD), where the dielectric barrier is most commonly the glass envelope of the bulb.

This approach protects the electrodes from corrosive chlorine and enables long lifetimes.

Finally, KrCl* sources, like all far-UVC sources, require high-quality quartz glass to maximize the transmittance of high-energy far-UVC photons.

The Excimer Reaction and KrCl* Spectra

Excimer source emission spectra are determined by the reactions taking place inside the plasma. The primary excimer reaction that proceeds under low pressure in KrCl* emitters is the “harpoon” reaction:

Kr* + Cl₂ → KrCl* + Cl

The KrCl* dimer rapidly returns to ground state and emits a 222-nm photon, with alternative decay transitions forming a small peak at 200 nm and a broader spectrum out to 240 nm.

However, other reactions compete with the primary excimer reaction. At higher chlorine pressures, chlorine may destroy KrCl* dimers before they can decay; it also forms Cl₂* excimers, which decay to emit 259-nm photons.

At higher pressures, KrCl₂* excimers at 325 nm may also form. Finally, the visible purple glow of a KrCl* source is accounted for by atomic krypton discharges.

Far-UVC Optics and Luminaire Components

Besides the far-UVC source itself, important components of a far-UVC luminaire are filters, diffusers, and reflectors.

Filters are part of almost all commercially available KrCl* luminaires, and all are interference (dichroic) filters.

Interference filters are manufactured by depositing thin alternating layers of HfO₂ and SiO₂ onto the quartz via a process called sputtering. The more layers, the more selective the filter—but also the more expensive, since sputtering is a time-consuming process requiring highly specialized machinery.

Because of this, the cost and performance of commercially available filtered KrCl* luminaires vary substantially.

In addition to cost and selectivity, interference filters also exhibit angle-dependent performance, meaning that their effectiveness at blocking non-222-nm emissions can vary depending on the incident angle.

At off-axis angles, the filter’s cutoff wavelength may shift, potentially allowing more out-of-band emissions to pass through.

The primary difference between filters is the amount of out-of-band emissions they allow through, but also their total output at 222 nm. Although a filter may be designed to only subtract >240-nm wavelengths, in practice, filtered sources’ output at 222 nm is meaningfully reduced as well.

Diffusers change the angular distribution of radiation to be more dispersed. This may be desirable in some circumstances, such as spaces with low ceilings, because a well-diffused source of the same total output will much more easily comply with UV exposure limits.

In many applications, a more uniform dose distribution can be beneficial for both safety and efficacy.

If the dose is uneven, this can create “hotspots”—areas where the UV intensity is higher than the surrounding regions. This requires engineers to design a system with a lower average fluence rate, and thus lower efficacy, to ensure compliance with safety thresholds across the entire space.

Conversely, “dead zones”—areas where UV intensity is lower than needed for effective pathogen inactivation—should be minimized to ensure adequate disinfection.

Well-designed diffusion and fixture placement help balance UV intensity across a space, improving overall performance while maintaining safety limits.

Unfortunately, the PTFE (Teflon) diffusers commonly used also significantly absorb far-UVC, reducing the total output of commercially available diffused luminaires by 60–75%.

Despite this downside, diffused luminaires are generally more suitable for ceiling-mounted installations in lower-ceilinged spaces.

For example, using the UL 8802 photobiological safety standard based on ACGIH Skin TLVs, one commercially available undiffused high-powered luminaire can only be safely mounted flush in a ceiling of at least 2.7 meters (9 feet), whereas the same source with a diffuser could be mounted in a ceiling as low as 2.3 meters (7.5 feet).

Finally, reflectors line the inside of the fixture and redirect radiation to increase effective output. PTFE/Teflon is often used for this, along with polished aluminum.

Determinants of KrCl* Performance

In general, the goal for a high-performance KrCl* source is to output as many 222-nm photons and as few other photons for as long as possible. These goals are sometimes in conflict with one another.

For example, the primary determinant of KrCl* lifetime is the amount of chlorine remaining in the glass. The chlorine depletes over time as it reacts with defects in the glass envelope.

Higher chlorine pressure increases the emitter’s lifetime, but reduces the efficiency of the KrCl excimer reaction, as well as increasing 259-nm emissions, requiring more filtering.

On the other hand, higher-quality filters remove more >240 nm output, but also reduce total output. They also take longer to make, use more materials, and are more expensive.

As another example, optimizing the driving voltage of the luminaire electronics results in both improved radiative efficiency and extended lifetime.

However, this requires careful radio-frequency (RF) engineering, and due to the required high voltages, it may result in a luminaire that gives off substantial electromagnetic interference that can put products out of compliance with regulated product standards.

In other cases, these goals are not in tension. Higher-quality glass has fewer defects, transmits more 222 nm photons, and reacts less with chlorine, resulting in improved output and performance—though it is more expensive.

Similarly, lifetime can be extended by reducing the reactivity of chlorine by operating the system at a lower temperature, or by passivating—coating a material so that it is less reactive—the interior of the glass tube.

Though intrinsic efficiencies for KrCl* emitters are predicted to be as high as 15–20%, and observed in the lab as high as 12.5%, the best commercially available emitters achieve a radiative efficiency of around 4%.

This is primarily due to the lifetime-output tradeoff, and commercial sources at 15–20% efficiencies are relatively unlikely.

Losses in the driving electronics and absorption in the emissive glass and filter bring commercial sources’ actual efficiency down to around ~1%, and with diffusers to ~0.25%.

As a result, a state-of-the-art 11-Watt luminaire produces ~110 mW of far-UVC, or ~40 mW if diffused.

What Should Researchers Know About KrCl* Sources?

KrCl* sources have variable outputs depending on how long they have been switched on and how old they are.

In order to control this behavior, fresh sources used for research should be “burned in,” meaning left on, for at least 100 hours.

Even after this burn-in, there will still be a characteristic “warm up” behavior, where sources take some time to reach approximate thermal equilibrium after they are switched on.

Manufacturers do not generally report the variability in output across units of the same model, but data collected by the OSLUV Project suggests that this variability can be at least 10–20%.

So, when performing research that depends on accurate measurement of far-UVC irradiance at a point:

• Always measure the source output at the point of interest at the time the experiment will take place.
• Burn in a fresh source for at least 100 hours before characterizing.
• Once a source is burned in, it should still be warmed up for 1–2 hours before performing experiments with it.

How Much Do KrCl* Lamps Cost?

Currently, a filtered KrCl* luminaire can be purchased for anywhere between $450 and $2,500. However, to estimate true lifetime cost, there are many other considerations than the sticker price.

Estimating Lifetime Cost of Safe and Effective Installations

Because pathogen inactivation is dependent on the dose of far-UVC rather than the number of fixtures, we believe that the most useful metric is the annualized cost per mW of far-UVC output.

The total lifetime cost of a far-UVC device can be broken down into:

To obtain an annualized cost per mW, we divide luminaire cost per annum by UV output:

With such an estimate, we can further estimate roughly how much it would cost per annum to produce a given room average fluence rate at different costs per mW.

For a ceiling-mounted fixture, this is fairly straightforward to estimate. To a first approximation, the average photon pathlength of a far-UVC fixture mounted on the ceiling, facing down, is the ceiling height.

Therefore, for every m² of floorspace, approximately 10 mW of ceiling-mounted far-UVC output would be required to create 1 μW/cm² of fluence rate.

A US classroom is typically around 900 ft² or 84 m², so to produce 1 μW/cm² of far-UVC fluence rate in such a space would likely require approximately ~840 mW of ceiling-mounted far-UVC output.

This simplified estimate of far-UVC output requirements is 5–10% higher than simulations using the Illuminate software. In the analysis below, we use the more conservative estimate.

Simulations of an 84 m² classroom and a ceiling height of 2.75 m using Illuminate suggest that corner mounting rather than ceiling mounting would reduce the installed power required to produce the same room average fluence rate by as much as 30–40%.

However, this does increase the risk of eye overexposure relative to a ceiling-mounted fixture. To be conservative both for the purpose of reducing eye exposure but also for calculating cost, we assume ceiling mounting.

Installation Cost

For a device to be certified to the voluntary standard ANSI/CAN/UL 8802, it must be permanently mounted. We conservatively assume for the purpose of this analysis that installation by a qualified electrician will cost $200 per luminaire.

Lifetime and Energy Costs

Though energy-inefficient on an input/output basis, a typical far-UVC luminaire consumes only 11 W. At the US average energy cost of 18 cents per kilowatt-hour, it would cost $4.12 to run a single luminaire for a year, assuming usage for eight hours a day, five days a week.

As we shall see below, this makes energy a relatively minor component of the overall cost.

KrCl* manufacturers typically claim that their luminaires have a lifetime of up to 10,000 hours, which would mean that replacement would only be necessary every 5 years assuming this same usage—or approximately once every 1 year if run 24 hours a day, 365 days a year.

Unfortunately, most far-UVC manufacturers do not publish data necessary to validate these claims, but it does allow for at least an approximate comparison of the range of annualized costs of a mW of far-UVC. Validation of lifetime claims through independent testing is warranted.

Annualized Lifetime Cost Estimate

In the table below, we present four example filtered luminaires for which we have third-party characterization data collected on a comparable basis.

We calculate the indicative cost per annum to produce 1 μW/cm² of far-UVC fluence rate in a typical US classroom, assuming 5 year luminaire lifetime and luminaire operation of 8 hours a day, 5 days a week. Luminaire power outputs are based on testing by the OSLUV Project.

These luminaires do not necessarily represent the most cost-effective filtered sources currently available, but rather highlight the significant differences across commercially available products, even if we assume equal lifetime.

Table 8.1. Annualized Cost Estimates for Example Fixture Installations

Note: all total outputs listed are shown after a few hundred hours of “burn-in,” when KrCl* output stabilizes. The table assumes that 840 mW of output is needed to produce 1 μW/cm² of far-UVC fluence rate.

A naive reading of this table would suggest that undiffused sources are more cost-effective. However, diffusion is often necessary in order to comply with consensus standards in lower ceiling height spaces.

The output lost with the use of a diffuser raises the question of whether it would be more cost-effective to comply with consensus standards by dimming an undiffused fixture.

Dimming is not a feature of luminaires that are currently sold, but it is in principle doable. Based on simulations conducted in Illuminate for an undiffused device, the output would have to be dimmed by approximately 65% in order to be mountable face down at a ceiling height of 2.3 m and comply with UL 8802 using the ACGIH Skin exposure limit.

By contrast, a diffused fixture can typically be mounted at this height and comply with the standard with significant headroom to increase power output while remaining in compliance.

Furthermore, diffusers can reduce “hotspots” and improve the distribution of dose in a space, which we believe is beneficial to safety, and likely beneficial to efficacy.

We therefore think that the priority for reduction in effective cost per mW for ceiling-mounted installations in typical 8–9 ft drop-ceiling spaces should be improvement in the diffuser to increase total power output.

As can be seen from Table 8.1, at current fixture cost, producing 1 μW/cm² of fluence rate in a typical-sized classroom is expensive.

Even if the required fluence rate for effective disinfection was in the range of ~0.2 μW/cm²—a reasonable best-case scenario given the available data—costs for ceiling-mounted installations would still be in the range of $600–$2,800 per year for a typical classroom-sized space.

This is likely to be cost-prohibitive for applications outside of high-risk spaces such as healthcare settings. We must therefore ask what are the reasonable prospects for future cost reduction.

Prospects for KrCl* Cost Reduction

One manufacturer projected in 2024 that they would achieve a device cost of $1/mW by 2028, an order of magnitude improvement. There are a number of avenues through which this degree of cost reduction can potentially be achieved in the next 3 years.

The Far-UVC Market Is Currently Subscale

Components that presently cost $200–300 can, we believe, readily cost $50/unit. But the emitter market is currently very small, and gross margins for both component and fixture manufacturers are likely higher than we would expect in a thriving competitive market.

Alternative Applications

Some of this cost-reduction via economies of scale may be achieved by finding additional market niches for far-UVC technology besides whole-room disinfection.

Some promising work has been done on allergen reduction, water disinfection, and mold control, but we do not evaluate these applications in this Blueprint.

Technology Improvement

There is room for cost reduction through improvements to product design. Some of this is what might be termed “catch-up growth.”

There are some techniques for improving KrCl* radiative output and lifetime that were freely published as far back as the 1990s, but few manufacturers use these. A larger market would incentivize the adoption of these methods in more commercial products.

For example, the main consumable in KrCl* emitters is chlorine gas. Manufacturers could improve their lifespan by optimizing gas mixtures or using replaceable chlorine gas vials, which may lower costs.

Even if it proves not to be cost-effective to replace chlorine gas, for most commercially available fixtures, the emitter as a whole is not replaceable either. This means that the entire luminaire must be replaced at the end of lifetime and reduces the overall cost-effectiveness.

Emitters that are easily replaceable by users could significantly reduce effective lifetime cost per mW.

Luminaire costs could also be reduced by developing technologies that lower component costs. For example, the dichroic filter is the single most expensive component of a filtered KrCl* source.

Absorption filters, which might be substantially cheaper if not quite as performant as dichroic filters, are under investigation, but none are yet commercially exploited.

While manufacturers typically claim up to 10,000 hours of lifetime, based on discussions with industry experts we believe up to 30,000 could be achievable.

Table 8.2. Possible Future Cost Estimates for KrCl* Luminaire Installations

Sensors

Accurate, low-cost sensors for monitoring far-UVC intensity could improve safety and efficiency. Current sensors are expensive, but advances in semiconductor-based detectors may provide cheaper, real-time monitoring solutions, ensuring consistent output and regulatory compliance.

While not a primary cost driver, better sensors could enhance system reliability and reduce overexposure risks, and potentially reduce the need for dimming or duty cycling for standards compliance.

A dimmed or duty-cycling luminaire has the same input costs but lower output, so if sensors can be used to remove the need for dimming or duty cycling, this will effectively reduce cost per utilized mW, potentially enough to make up for the increased cost of adding the sensor.

There are a number of commercially available luminaires that employ sensor technology such as LIDAR and mmWave. However, none of these specifically employ sensors to facilitate UL 8802 compliance while significantly increasing the output of luminaires that can be safely installed at a given ceiling height.

Possible Future Costs

Taking all the above together, we can re-evaluate the cost of installing far-UVC in a typical classroom using a possible future scenario of approximately $1/mW, with a 30,000 hour/15 year lifetime.

If fluence rates needed to be effective at preventing airborne transmission are in the order of ~0.2 μW/cm², the cost of installing far-UVC in a classroom could feasibly be less than a few dollars a year per student.

Key Takeaways

• Luminaire lifetime is at least as important as power output for determining effective cost.
• Diffused sources are less cost-effective on a per mW basis, but may be the only option at lower ceiling heights.
• Far-UVC luminaires are currently expensive.
• Our best guess is that in the future, KrCl* luminaires could be cheap enough that wide-scale commercial use will be possible, particularly if fluence rates required for effective disinfection are low.

How Might Far-UVC Be Generated in the Future?

Though KrCl* is the dominant source of far-UVC today, alternative technologies are under development in both academia and commercial startups.

Alternative approaches could potentially allow for:

• Different—ideally smaller—form factors, enabling a wider variety of fixture designs.
• Lower voltages, requiring less complex electronics to manage electromagnetic interference (EMI).
• Ability to target different wavelengths, i.e. other than 222 nm.

Krypton Bromine Excimer Emitter

As well as KrCl*, there are alternative excimer-based emitter designs, and these can generate different far-UVC wavelengths.

Krypton bromine excimer (KrBr*) sources, emitting at 207 nm, were the first far-UVC sources investigated for germicidal effectiveness.

However, the KrBr* reaction has a lower intrinsic efficiency, and filtering the out-of-band emissions was more difficult. This meant that KrBr* sources ultimately proved impractical, and none are commercially available today.

Photoluminescent Emitters

Non-far-UVC excimer sources can be used to create far-UVC emissions through photoluminescence (PL). PL is a process whereby luminescent material re-emits radiation after absorbing photons—as opposed to electrons—from an external source.

For example, UVC phosphors can be excited using a 172-nm xenon excimer (Xe₂*) source, creating broad-spectrum UVC emissions peaking at 230 nm.

Alternatively, magnesium zinc oxide can be excited by a 146-nm krypton excimer (Kr₂*) source, producing UVC emissions peaking at 202 nm.

A number of potential approaches are feasible, and it could be possible to filter out unwanted emissions outside the far-UVC range. However, to our knowledge, no PL device has been demonstrated to be suitable for whole-room disinfection of occupied spaces.

Semiconductor-Based Approaches

There has been substantial interest in developing far-UVC emitters based on semiconductor technology, primarily predicated on the notion that in traditional lighting, visible light LEDs have proved highly scalable and energy-efficient.

However, it should be noted that the widespread adoption of visible light LEDs benefited from the sheer size of the existing general lighting market, which supports the fixed costs of semiconductor manufacturing.

Whether the cost efficiency gains possible through semiconductor technology will be realized for far-UVC generation will depend in large part on whether the market for far-UVC expands sufficiently, as well as whether developers can solve challenges unique to producing higher energy far-UVC photons.

Cathodoluminescent Emitters

Cathodoluminescence (CL) is a phenomenon where photons are generated by electrons exciting a luminescent material. Many familiar technologies use cathodoluminescence, including cathode ray tube televisions (CRT TVs) and scanning electron microscopes.

CL emitters allow the usage of electroluminescent materials that are otherwise unsuitable for LEDs, like magnesium zinc oxide (peak emission 229 nm), hexagonal boron nitride (peak emission 215 nm), or a UV-emitting phosphor (peak emission 230 nm).

Like KrCl* and LEDs, published spectra of CL sources in the far-UVC contain unwanted wavelengths, which likely cannot be changed with any method beyond optical filtration.

Traditional CL devices that pump electrons through a vacuum, like CRT TVs, are not semiconductor-based. Their fundamental radiative efficiency is limited; they require high voltages and vacuum-sealing; chip-scale miniaturization is possible, but challenging.

Purely solid-state emitters that employ CL to produce far-UVC are feasible using impact ionization. Despite utilizing a different underlying technology, these emitters would have a number of similarities to LEDs and require solving similar engineering challenges, such as the need for cooling in order to achieve sufficient output and efficiency.

While not purely solid-state, a semiconductor-based CL device primarily emitting at 215 nm has been designed and is commercially available.

Expert forecasts predict that CL devices will be the first semiconductor-based emitter to meet current KrCl*-tier performance for whole-room disinfection. CL emitters may also find niches outside of the applications that KrCl* sources are currently serving.

UVC LEDs

The blue—and consequently, white—LED was a transformational invention for the general lighting industry, one that secured a Nobel prize for its inventors. As such, there has been much excitement over the prospect of UVC LEDs, which are built on a similar material platform of gallium nitride (GaN).

Some early indicators do suggest that conventional UVC LEDs are on a comparable trajectory to blue LEDs.

Commercial 270-nm LEDs are closing in on wall-plug efficiencies of 10%, and firms are aiming to reach 20% within the next year.

Whether these promises will bear out for common UVC wavelengths 260–280 nm remains to be seen, but regardless, this kind of forecasting has questionable relevance to far-UVC LEDs.

UVC LED performance is strongly dependent on wavelength, and plummets exponentially for shorter UVC wavelengths. The trend is also true of lifetime.

Shorter-wavelength aluminum gallium nitride (AlGaN) LEDs are more challenging because they require a higher relative abundance of aluminum.

To achieve far-UVC LEDs, the aluminum fraction would have to be close to 90%, very close to pure aluminum nitride (AlN).

Many challenges emerge at such large aluminum fractions—the materials grow with more defects, are more opaque to short wavelengths, are more difficult to dope, and are more difficult to make good electrical contacts with, along with many other issues.

Some of these issues are also relevant to 260–280-nm UVC LEDs, and far-UVC LEDs will benefit from advancements in that industry. Others, however, are relatively unique to far-UVC LEDs.

For some of these issues, fundamental research is required, such as understanding the role of point defects in radiative efficiency. For others, various device design approaches are being used to address them, and some are seeing early commercial application.

Even relatively simple innovations in fixture design can materially improve LED performance, for example through efficient passive cooling.

An additional consideration is that LEDs are not monochromatic. Current UVC LEDs have fairly wide peaks, and long, lower-energy tails.

Though they are likely to improve with time, filtering will likely still be necessary to use them in applications where KrCl* is used currently.

Despite these challenges, expert forecasts anticipate seeing far-UVC LEDs that are competitive with KrCl* sources for whole-room disinfection on the ~10 year timescale.

This timeline is likely lower for applications where cost per mW is less of a consideration. Whether this prediction bears out, AlGaN devices for UV radiation are well-established and garner substantial academic interest, substantially more than other approaches.

Second Harmonic Generation Emitters

The final technology of note is second harmonic generation (SHG). SHG is an established approach for laser generation when no suitable diode material exists at the desired wavelength—green laser pointers are the classic example.

The basic principle of SHG is simple. Start with a laser at twice the desired wavelength—the “pump” wavelength. Couple the pump laser to a crystal with the appropriate properties. Output half the number of photons at twice the energy and half the wavelength.

For far-UVC, it is straightforward to start with the ubiquitous 450-nm blue laser diode and frequency double it to achieve monochromatic 225-nm emission.

Advantages of SHG Emitters

• Spectral purity: The generated radiation is highly monochromatic, with the peak wavelength an order of magnitude narrower than that achievable by KrCl*. Blue light will likely remain in the spectrum, but this is far less dangerous and easier to remove than UVC or UVB.
• Wavelength targeting: By changing the pump wavelength, the output wavelength changes as well.
• Pump source availability: Blue lasers are a well-developed technology, with wall-plug efficiencies around 40% and reported lifetimes of 20,000–30,000 hours.
• Efficiency: SHG can theoretically achieve 100% conversion efficiency by integrating a resonant cavity into the device, though in reality this is rarely approached.
• Scalable production: With photonic integration, SHG emitters could be mass-produced at chip scale.

Challenges and Limitations of SHG Devices

As usual, there are challenges and limitations to SHG. Among them is the choice of nonlinear crystal, a material that efficiently halves the wavelength of incoming photons by second harmonic interactions.

In addition to being a good nonlinear converter, the material must be transparent at the target wavelength. For far-UVC, there are two primary options: beta barium borate (BBO) and aluminum nitride (AlN).

BBO is generally the material of choice, with conversion efficiencies reported around 15–35%. However, complex bulk optics are required to achieve such numbers, and without them efficiencies typically plummet to 0.05%, though work with silicon nitride (SiN) waveguides has pushed conversion efficiency as high as 4%.

BBO is also sensitive to humidity, potentially making it challenging for applications outside of research.

AlN is another option for UV SHG. Drawbacks include the relatively lower nonlinear coefficient and the low transparency in the far-UVC range.

However, AlN is easier to grow and more widely used in other semiconductor applications, potentially enabling easier scaling and cost advantages.

More broadly, SHG output will need to be shaped and diffused for general applications, and as a laser source may be subject to different safety standards.

As discussed, suitable materials for this application in the far-UVC are limited, and this requirement will likely reduce total power output even without the need for a filter.

Cost-effective diffusion of SHG-produced far-UVC has not yet been demonstrated, but in principle should be physically possible using known materials.

These issues combine to form a picture quite far from the theoretical ideal performance, and claims about cost-effectiveness at large-scale production remain hypothetical. Nevertheless, this is a technology to watch.

Further Reading

• Solid-State Far-UVC Emission Roadmapping Workshop Report
• Illuminate
• RP Photonics Encyclopedia – Frequency Doubling
• NALMCO Germicidal UV Training and Certification
• The 2020 UV Emitter Roadmap

References

18. Matysik, S., Bohmann, B. & Bächler, L. Reducing Indoor Infections: The Economic Potential A Cost Benefit Analysis of Far-UVC Lamps. (2024).
19. Bruns, R. Cost-Benefit Analysis of ASHRAE Standard 241. ASHRAE Journal (2023).
20. Kleinwaks, G., Fraser-Urquhart, A., Kraprayoon, J. & Morrison, J. Air Safety to Combat Global Catastrophic Biorisk. (2022).
21. Faulkner, C. A. et al. Comparison of effectiveness and energy use of airborne pathogen mitigation measures to meet clean air targets in a prototypical office building. Build. Environ. 257, 111466 (2024).
22. Vivian Belenky, Holger Claus, M Pang, Louis Goessling, David J. Brenner. Illuminate: A Free and Open Source Germicidal UV Modeling Tool. (CORM/CIE-USNC Joint Annual Conference, 2024).
23. Tuenge, J. Preliminary Findings from Rounds 3 and 4 of CALiPER GUV Product Testing. (12–14 November 2024).

9. Guidance, Standards, and Regulations

Summary

A wide variety of organizations are relevant to far-UVC governance. These organizations set guidance, standards, and regulations pertaining to UV safety, ozone exposure, air cleaning efficacy and applications, GUV devices, and industries like healthcare.

Within the United States, very few regulations directly pertain to far-UVC. In other jurisdictions, such as the European Union, consensus standards on UV safety have been incorporated into regulation of consumer products.

Even without legal force, voluntary standards can shape markets. Consumers who purchase portable air cleaners are informed about their efficacy through clean air delivery rates advertised by manufacturers and produced under test standards such as ANSI/AHAM AC-1 Method for Measuring Performance of Portable Household Electric Room Air Cleaners.

Installation of Upper Room UV in hospitals is guided by NIOSH 2009-105 Environmental Control for Tuberculosis: Basic Upper-Room Ultraviolet Germicidal Irradiation Guidelines for Healthcare Settings.

For many buyers, compliance with product standards such as ANSI/CAN/UL 8802: Ultraviolet (UV) Germicidal Equipment and Systems can be a de facto procurement requirement, even if it is not always de jure.

Far-UVC consensus standards need to be iterated to ensure that the guidance and best practices they encode reflect the most recent scientific understanding, and the particular properties of far-UVC lamps.

The standards development process is typically slow, as is the cycle to update any regulations based on those standards. In some cases, further research needs to be done in order to rationally update or create a new standard. In other cases, improvements should be made now based on what we already know.

Standards are based on a balance of interests, which means evolution is to be expected even in the absence of significant changes in scientific understanding.

We expect that if far-UVC is widely used in public spaces, it will be regulated. Consensus standards and recommendations from expert bodies will form the substantive content of those future regulations, as they do today.

If we can create effective standards and guidance in advance of the desire of governments and the public for legal regulations, this will help ensure that far-UVC, and the market for air cleaning technologies as a whole, can grow and thrive.

Crucial Considerations

• Guidelines, standards and regulation of far-UVC are a complex patchwork that differ by jurisdiction and in the degree of voluntary versus mandatory compliance.
• Standards often apply UV exposure limits in a more conservative way than the expert guidance documents they reference, by managing peak irradiances rather than time-weighted average exposure of occupants.
• In some circumstances, standards and guidelines are incorporated into legal regulation.
• Some standards, such as methods of testing for far-UVC efficacy, cannot be created until certain open research questions are resolved.
• There is a particular need to develop new standards related to ozone generation.

Definitions

Guidance is simply a recommendation from an expert body. This can come from a wide variety of organizations including public health agencies, professional associations and other expert bodies.

Guidance is produced in a wide variety of ways for different reasons. For our purposes, anything that recommends but is not a consensus standard or does not have the force of law, we term “guidance.”

Guidance can be highly technical, or it can provide practical advice to practitioners and the wider public. Examples include recommendations on UV exposure limits from ICNIRP and ACGIH and public-facing consumer advice on the use of portable air cleaners provided by the US Environmental Protection Agency (EPA).

A consensus standard is a technical document that has gone through a rigorous development process, incorporating the expertise and interests of different parties. These usually take years to finalize.

Different standards bodies have different procedures, but they all include provisions for openness, transparency, balance of interests and clear processes for review, voting and appeal.

Standards are voluntary: people may choose to comply, but they do not themselves create any legal requirements.

Organizations that have produced or accredited standards relevant to far-UVC include:

• American National Standards Institute (ANSI)
• ASTM International (formerly American Society for Testing and Materials) (ASTM)
• International Organization for Standardization (ISO)
• International Electrotechnical Commission (IEC)
• UL (formerly Underwriters Laboratory)
• American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE)
• Illuminating Engineering Society (IES)
• Association of Home Appliance Manufacturers (AHAM)
• International Commission on Illumination (CIE)

These organizations are a mixture of standards development and standards accrediting bodies. For example, ANSI primarily accredits standards created by other bodies such as UL, ASHRAE, AHAM, and IES, while ISO develops a wide range of standards in-house as well as recognizing standards developed by other bodies such as CIE.

Standards that have been accredited by ANSI generally have the prefix “ANSI” before the name of the body who developed the standard. ANSI standards, among others, are part of a mutual recognition agreement with Canada, and therefore the prefix ANSI/CAN can commonly be found.

Standards accredited by European Standards Organizations have the prefix “EN.”

A regulation is simply anything that has the force of law.

It is not uncommon for consensus standards to be incorporated into legislation. The most ubiquitous mechanism by which far-UVC may be regulated in the future is via building codes, which are in most cases standards or model codes voted into law by the “authority having jurisdiction.”

This authority may be, in the United States, the municipal government, the county authority, the state, the federal government in the case of federal buildings, or the relevant branch of the military on military bases.

As such, many standards that are facially “voluntary” are actually quasi-national, with nationwide authorities voting on the adoption of the new standard into building codes within a few years of a new version of that standard being released.

Analysis

Efficacy

Guidance

There is relatively little guidance on minimum recommended efficacy for GUV devices specifically, and none for far-UVC.

The US Centers for Disease Control and Prevention (CDC) recommends 5 or more air changes per hour (ACH), or equivalent, for infection control in the workplace, but there is no guidance on how to calculate the equivalent air changes per hour (eACH) from the use of a particular GUV system.

For upper-room UV, the CDC points to the 2009 NIOSH guidance for tuberculosis control, but this guidance is not relevant to whole-room far-UVC systems.

Currently under development is ASHRAE GPC 37 Guidelines for the Application of Upper-Air (Upper Room) Ultraviolet Germicidal (UV-C) Devices to Control the Transmission of Airborne Pathogens, which we expect to supplant the 2009 NIOSH tuberculosis guidance for the use of upper-room UV systems. These guidelines cover both safety and efficacy.

For the use of portable air cleaners to achieve 5 eACH, the CDC points at guidance from the EPA, which if followed by the user would add approximately 5 eACH under a well-mixed room assumption.

Some test standards that have been developed for measuring the clean air delivery rate (CADR) of air cleaning technologies against bioaerosols are also applicable to GUV devices.

In principle a consumer could use the CADRs from these test methods and apply them in the same way. However, this comes with some additional caveats for GUV technologies.

Standards

Test Methods for Air Cleaners

There are six test standards that provide testing methods for in-room air cleaners against which far-UVC devices could be assessed.

In addition, we are also aware of a German draft standard still under review, DIN/TS 67506:2022-02 Disinfection of room air with UV radiation - UV-C secondary air unit.

The six standards are:

• ANSI/ASHRAE 185.3 Method of Testing Commercial and Industrial In-Room Air-Cleaning Devices and Systems for Microorganism Bioaerosol Removal or Inactivation in a Test Chamber (2024)
• ANSI/AHAM AC-5 Method for Assessing the Reduction Rate of Key Bioaerosols by Portable Air Cleaners Using an Aerobiology Test Chamber (2023)
• ISO 16000-36 Standard method for assessing the reduction rate of culturable airborne bacteria by air purifiers using a test chamber (2018)
• ASHRAE 241 Control of Infectious Aerosols; Normative Appendix A (2023)
• ASTM E3273-21 Standard Practice to Assess Microbial Decontamination of Indoor Air using an Aerobiology Chamber
• GB/T 18801-2022 Air cleaner

In addition to these test methods for in-room air cleaners, there are also test methods that apply to in-duct devices and surfaces, which are not covered in this report.

The first three standards are ANSI- or ISO-accredited. ASHRAE 241 was developed as part of an accelerated process in response to the COVID-19 pandemic, and has not yet been accredited by ANSI.

ASHRAE 241 also permits that, where filtration devices have been tested by ANSI/AHAM AC-1 for the removal of dust, pollen and smoke, manufacturers can use these results to infer their expected performance against bioaerosols.

ASHRAE 241 also permits the use of other consensus standards to determine effectiveness, including ANSI/AHAM AC-5 and ANSI/ASHRAE 185.3.

TABLE 9.1. Comparison between testing standards for bioaerosol removal or inactivation by in-room air cleaners.

Note: see Efficacy section for definition of the decay method.

In the case of GUV technologies, it is not currently clear whether these test methods will tend to produce pathogen inactivation rates at the high or low end of the wide distribution that is found in the published academic literature.

Given that the source of variation across experiments is not currently known, it is quite possible that orders of magnitude variation could be seen in different facilities that perform the same consensus standard test.

Until the source of this variation is known, it is not possible to make specific recommendations about how test methods should be improved.

One clear issue with these test methods when employing GUV devices is that the average photon pathlength—and therefore the average UV fluence rate—will depend on the chamber geometry and placement of the device.

In contrast, the clean air delivery rate of an air purifier ought not to be affected to the same degree.

This intrinsic property of GUV devices—that the air cleaning that they will deliver in real-world spaces is not just a property of the device but the way the device is used in the space—needs to be accounted for in the interaction between test methodologies and the way they are employed in guidance and standards for real-world application.

ASHRAE 241

ASHRAE 241 Control of Infectious Aerosols is the only current standard of relevance for the application of far-UVC that covers efficacy.

It is a new standard, published in 2023 after an accelerated development process commissioned by the White House, and has not yet gone through the ANSI accreditation process.

ASHRAE 241 is currently under “continuous maintenance,” which means that changes to the standard can be proposed at any time, in contrast with the usual ASHRAE policy which is to revise standards every five years.

ASHRAE 241 aims to provide a technology-neutral framework for increasing the amount of air cleaning in indoor public spaces, on top of minimum ventilation requirements.

It proposes the concept of an Infection Risk Management Mode (IRMM), and the required equivalent clean airflow for infection risk management (ECAi) in different spaces.

In order to provide guidance, ASHRAE 241 employs an infection risk model that is explained in Appendix D of the standard, but the broad intention is that, under the assumptions of the model, application of the IRMM requirements would equalize the infection risk to the same low level in different spaces.

The resulting amount of air cleaning required in different spaces, over and above the requirements of ANSI/ASHRAE 62.1 Ventilation and Acceptable Indoor Air Quality, varies substantially.

Table 9.2. Comparison Between ASHRAE 62.1 Minimum Ventilation Requirements and ASHRAE 241 Infection Risk Management Mode Requirements

Note: occupant density standard assumptions are provided in ASHRAE 62.1. ACH and eACH requirements calculated by Blueprint Biosecurity based on 3.5 m ceiling height and the parameters listed in the table.

For an elementary school classroom with a 3.5 m/11 ft ceiling height, under standard occupancy assumptions, the CDC 5 eACH guidance is almost identical to the ASHRAE 241 standard.

However, in the case of a densely packed college lecture hall at that same ceiling height, the minimum ventilation requirements per ASHRAE 241 would be 4.2 ACH, and the ASHRAE 241 IRMM requirements nearly 40 eACH.

Under standard occupancy assumptions, to comply with IRMM a restaurant would require equivalent clean airflow comparable to a modern hospital operating theater as per ANSI/ASHRAE/ASHE 170 Ventilation of Health Care Facilities.

ASHRAE 241 does permit facilities to meet IRMM requirements by restricting occupancy, and therefore the total ECAi requirement which is calculated on a per person basis.

In the absence of a technological breakthrough, there is no other way IRMM requirements could feasibly be met in many densely occupied indoor public spaces.

One challenge with ASHRAE 241, and for any framework for infection control that is intended to be both technology-neutral and pathogen-agnostic, is the selection of a test organism for the purposes of calculating device efficacy.

The susceptibility of pathogens to far-UVC varies significantly, raising the question of how to “credit” efficacy for the purposes of a standard that is intended to be broadly applicable against all airborne infection risks.

ASHRAE 241 requires the use of MS2 bacteriophage in efficacy testing, which is less susceptible than pathogens that cause endemic respiratory illnesses such as SARS-CoV-2 and influenza.

This means that ASHRAE 241 is conservative with regards to permitting the use of UV technologies to meet its requirements, assuming that the intention of implementing IRMM would be to prevent transmission of these typical respiratory illnesses.

Regulations

Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA)

The Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) empowers EPA to regulate the registration, distribution, sale, and use of pesticides in the United States.

Under FIFRA, anything that prevents, destroys, mitigates, or repels a pest—including any microorganism—is considered a pesticide, including UV which inactivates microorganisms. Therefore, all germicidal UV devices are pesticides regulated under FIFRA.

FIFRA imposes a requirement that a pesticide device be registered with EPA if the device is sold with a substance, typically chemical in nature, such as silver, zinc, or copper.

If it works solely by physical means—that is, just the UV radiation—then it does not have to be registered. However, pesticide devices have to be produced in an EPA-registered location.

Unlike chemical pesticides, EPA does not do premarket review of UV devices. EPA does, however, regulate misbranding and false or misleading claims.

In the surface disinfection market, EPA has specifically defined terms in its guidelines:

• Sanitization — 99.9% reduction in specific tested microorganisms.
• Disinfection — 99.999% reduction in specific tested microorganisms.
• Sterilization — 99.9999% reduction in specific tested microorganisms.

If a manufacturer uses these terms but the device doesn’t meet their definitions as supported by acceptable testing data, that can be considered false or misleading.

The EPA currently does not have a comparable set of guidelines or specified acceptable testing procedures for air disinfection.

Other relevant terms that can potentially be subject to misleading claims regulation under EPA jurisdiction include:

• Germicidal
• Elimination
• Safe
• Purify

Each UV device is considered its own product by the EPA, requiring its own data to support any claims. Therefore a device manufacturer cannot rely on testing done by other manufacturers using the same bulb.

While in principle a manufacturer of a UV device can supply data from the various test methods cited above to back up claims, the EPA does not currently specify what tests are required for air disinfection.

Other Regulations

ASHRAE 241 has yet to be adopted into any regulations. However, the Government of New Brunswick in Canada has announced the intention to achieve ASHRAE 241 compliance in new public buildings.

In November 2024 there was a failed ballot initiative in the city of Berkeley to apply ASHRAE 241 to all city-owned and -leased facilities.

The initiative also would have banned the use of all GUV technologies to meet the standard, including far-UVC and upper-room UV systems.

Besides ventilation-related aspects of building codes, a relevant standard for far-UVC may be the International Energy Conservation Code (IECC), developed by the International Code Council (ICC).

Section 405 of the IECC pertains to lighting fixtures and stipulates the minimum acceptable lighting power density (LPD, units W/ft²) for a light fixture.

With the advent of the white LED, minimum LPDs have increased, and no currently available far-UVC fixture would be compliant. For this reason, it is critical that GUV fixtures not be classified as “light fixtures” in this context.

Photobiological Safety

Guidance

An explanation of the UV exposure limits recommended by the American Conference of Governmental Industrial Hygienists (ACGIH) and the International Commission on Non-Ionizing Radiation (ICNIRP) is provided in the Far-UVC primer section of this report.

It is within the gift of any standards organization, legislature, or public health agency to apply guidelines from either ICNIRP or ACGIH, or indeed come up with their own.

ACGIH is a non-profit organization. Despite having “American” in its name, ACGIH Threshold Limit Values (TLVs) are widely used in industrial hygiene and occupational and environmental health and safety practice across the globe.

The TLVs are intended to be exposure limits for healthy working adults—not children or photosensitive individuals. This does not necessarily mean that ACGIH TLVs are inappropriate for populations other than healthy working adults, only that these populations are not required to be considered as part of the process of recommending exposure limits.

Like ACGIH, ICNIRP is an advisory body and does not issue standards or regulations. Unlike ACGIH, ICNIRP assesses health risks for broad groups of people, not just in occupational contexts.

The fact that there is currently a substantial difference between ACGIH and ICNIRP exposure limits for far-UVC is a reflection of the different nature of the institutions, and it is possible that despite being aligned for many years they will not come fully back into alignment.

What ACGIH TLVs and ICNIRP exposure limits have in common is that exposure limits are based on a time-weighted average (TWA) over an 8-hour period.

They are intended to reflect the fact that in real-world situations, individuals are highly unlikely to be exposed at a constant rate over an 8-hour time period. The dose should not be cumulatively exceeded during that period, accounting for variation in exposure.

Standards

ANSI/IES RP 27.1-22

The Illuminating Engineering Society (IES) publishes standard ANSI/IES RP 27.1-22: Photobiological Hazards From UV Lamps.

RP 27.1-22 describes the process for assessing photobiological safety of different devices and contains recommended practices for taking irradiance measurements in an installation using a radiometer.

Different UV devices are assigned different Risk Groups. The Risk Groups are determined based on spectrally weighted irradiance measurements taken a certain distance from the device.

For “Open room germicidal products for occupied spaces,” the irradiance measurement for determining risk groups must be taken 1.0 m from the source.

Table 9.3. Emissions Limits for Risk Groups per ANSI/IES RP 27.1-22

Based on these emissions limits, we would expect commercially available far-UVC devices to be RG-1 or RG-0, especially when applying the skin TLVs.

Importantly, Section 8.2 of RP 27.1-22 specifically states that whole-room luminaires should be mounted on the ceiling and directed downwards.

The standard also specifies acceptance testing for whole-room GUV systems to ensure compliance with ACGIH TLVs.

Irradiance measurements for assessing skin safety should be performed with a radiometer facing the ceiling, while eye measurements should be performed with the radiometer facing the wall, using an 80-degree field-of-view cone to best approximate the human eye.

When determining whether irradiance measurements conform with the ACGIH TLVs, RP 27.1-22 clearly specifies that time-weighted averages should be used in conjunction with spectral weightings.

The standard does not itself provide specific guidance for how time-weighted averaging should be applied by an installer of a GUV system, although it references time-in-motion studies or dose badges as ways of quantifying occupant exposure.

NALMCO, the interNational Association of Lighting Management Companies, recently launched three levels of certification for GUV practitioners and references ANSI/IES RP-27.1-22 in its materials.

ISO 15858

ISO 15858:2016: UV-C Devices - Safety Information - Permissible Human Exposure also specifies procedures for taking irradiance measurements in installed systems to ensure compliance with exposure limits.

Currently it refers only to conventional 254-nm UVC and recommends taking irradiance measurements at eye level, between 1.83 and 2.13 m, with the detector facing the device.

A new version, ISO/DIS 15858, is currently undergoing development. The updated standard references all UVC wavelengths, including far-UVC.

The draft standard recommends walking around with a radiometer and identifying spots where occupants are standing or seated for extended periods and where irradiance readings appear to be higher.

At these “hot spots,” horizontal and vertical irradiance measurements should be taken at a height of 1.8 m. The spectral weightings in the standard are taken from current ICNIRP guidance.

This standard differs importantly from ANSI/IES RP 27.1-22 and the ACGIH/ICNIRP recommended exposure limits because it does not use time-weighted averages.

Rather than considering realistic average exposures, ISO 15858 requires that a hypothetical person who remains in the worst-case-scenario position in the occupied zone of the room for 8 hours still must not be overexposed.

Regulations

ICNIRP guidelines are incorporated into European law for occupational exposures through EU Directive 2006/25/EC.

It is expected that EU occupational exposure limits for UV will likely continue to be drawn from ICNIRP rather than ACGIH.

In the US, the Occupational Safety and Health Administration (OSHA) has regulatory authority to set occupational exposure limits, termed Permissible Exposure Limits (PELs). However, OSHA has not adopted PELs for UV and refers to the ACGIH limits on a more informal basis.

Products

IEC/EN 62471

The International Electrotechnical Commission (IEC) publishes IEC/EN 62471: Photobiological Safety of Lamps and Lamp Systems.

It is in many respects identical to ANSI/IES RP-27.1-22 with respect to ultraviolet lamps, but there are critical differences for whole-room far-UVC systems because IEC/EN 62471 references ICNIRP spectral weightings rather than ACGIH.

Table 9.4. Emissions Limits for Risk Groups per IEC/EN 62471

Under this classification, some high-powered undiffused lamps will be RG-2 rather than RG-1, and diffused lamps RG-1 rather than RG-0.

When a far-UVC device is categorized as RG-2 by IEC/EN 62471, proximity sensors are required to prevent the lamp turning on if an individual is detected within the area of primary emission.

This likely makes RG-2 devices impractical for use for whole-room disinfection of occupied spaces under this standard.

ANSI/CAN/UL 8802

UL publishes ANSI/UL 8802: Ultraviolet (UV) Germicidal Equipment and Systems as well as ANSI/UL 8803: Portable UV Germicidal Equipment With Uncontained UV Sources.

These standards apply to all UV systems with emissions from 200–400 nm, not just far-UVC devices.

In order for a device to be UL 8802-compliant, it must be permanently mounted in a fixed location and the orientation of the lamp must not be adjustable by the use of ordinary tools.

Lamps intended for use in dwellings, or that are not permanently mounted, or that can have their orientation adjusted by ordinary tools, can only be certified to the stricter requirements of UL 8803.

UL 8802 references ANSI/IES RP 27.1-22 and ICNIRP/ACGIH recommended exposure limits. Although those guidance documents specify time-weighted average limits, UL 8802 bases its recommendations on peak irradiance under worst-case test conditions rather than realistic time-weighted exposure.

For UL 8802 certification, a device must be assigned Risk Group 0 according to either IEC/EN 62471 or ANSI/IES RP 27.1-22.

For devices with a minimum mounting height of at least 2.1 m, UL 8802 compares the Risk Group limits to the highest irradiance measurement that can be found in the room at 1.9 m above the floor under the most severe conditions permitted by the installation instructions.

For ceiling-mounted devices facing down, current consensus is that the ACGIH Skin spectral weighting may be used. For other installation orientations, the Eye weighting applies.

UL 8802 also allows in principle for higher-output devices through sensors that modulate device output when a person or object is exposed, although this creates further reliability requirements.

UL 8802 requires the manufacturer to provide instructions to ensure that installation does not produce ozone concentrations in excess of an unspecified “safe value,” taking into account the number of lamps used, ventilation system, and other relevant factors.

ANSI/CAN/UL 8803

UL 8803 is similar in many respects to UL 8802, but is intended for lamps used in dwellings, lamps that are not permanently mounted, or lamps that can have their orientation adjusted by ordinary tools.

For a product to comply with UL 8803, it must have a motion detector, a specified activation cycle, and an operating time limit.

The standard specifies an “exempt perimeter,” and the motion detector must cover the entire exempt perimeter.

UL 8803-compliant devices have a maximum operating cycle of 60 minutes unless their irradiance is further tested at a distance of 20 cm from the lamp.

Given these limitations, the Blueprint expects UL 8803-compliant devices are unlikely to be suitable for whole-room GUV systems, although they may have other applications.

Lamp Exposure Limits

Researchers have proposed that non-monochromatic sources should be labeled with Lamp Exposure Limits (H_LEL) that provide users with information on spectrally weighted emissions.

Eadie et al., 2024 found that spectrally weighted emissions of different KrCl* lamps can meaningfully diverge depending on the use of filters. Under ACGIH spectral weightings, the difference across KrCl* lamps varied by a factor of four.

TABLE 9.5. Lamp Exposure Limits from Eadie et al., 2024. The measured irradiance of 14 commercially available KrCl* far-UVC sources at a distance of 20 cm.

Note: LEL is determined using the technique described by Buonanno et al. and presented for the three guideline S(λ) curves from ICNIRP and ACGIH. The exposure limit values at 222 nm (H₂₂₂) for each organization are displayed for comparison.

For all lamps measured, the spectrally weighted lamp exposure limit was lower than the ACGIH TLV at 222 nm, as would be expected because these lamps emit some wavelengths in the conventional UVC range.

For some filtered lamps the difference was as little as 5%, whereas for at least one unfiltered lamp the spectrally weighted exposure limit was 75% lower than the nominal ACGIH TLV at 222 nm.

Eadie et al. also highlighted that different radiometers do not necessarily give the same answer, because radiometers are calibrated to an intended source.

There is therefore considerable value in a consensus standard specifying procedures for testing laboratories to measure the full emissions spectrum and calculate lamp exposure limits based on ACGIH and/or ICNIRP spectral weightings.

Regulations

IEC/EN 62471 is incorporated into regulation in a number of jurisdictions. A lamp, including a far-UVC lamp, must have a CE mark in order to be sold in the EU, and lamps must comply with IEC/EN 62471 to obtain a CE mark.

As of 2023, the UK has its own UKCA mark for lamps, which also requires IEC/EN 62471 compliance, while continuing to recognize the CE mark.

In the United States, the Federal Communications Commission (FCC) has a mandate to enforce regulations on products emitting radio frequencies. KrCl* lamps create radio-frequency emissions as a byproduct of operation and therefore require FCC certification.

Ozone

Guidance

A number of bodies issue guidance about occupational ozone exposure that applies indoors.

NIOSH recommends an occupational exposure limit of 100 ppb as a time-weighted average over 8 hours, with 5000 ppb (5 ppm) considered immediately dangerous to life and health.

ACGIH 8-hour TLVs for ozone are 100 ppb for light work, 80 ppb for moderate work, and 50 ppb for heavy work, or 200 ppb over 2 hours.

The WHO makes recommendations that apply to outdoor ozone, including that the average of daily maximum 8-hour mean ozone concentration in the six consecutive months with the highest six-month running-average ozone concentration not exceed 30 ppb, or exceed 50 ppb on more than 3–4 days per year.

ASHRAE has taken the institutional position that ozone emissions from air cleaners should be minimized, with intentional ozone generation generally avoided and unintentional generation treated cautiously because of uncertainty surrounding ozone and ozone-reaction byproducts.

Standards

UL 867 and UL 2998

The relevant UL standards for ozone are UL 867 - Electrostatic Air Cleaners, and UL 2998 - Zero Ozone Emissions Validations.

UL 867 describes the test method for measuring ozone. To meet UL 867, the measurement under the specified procedure should not exceed 50 ppb.

UL 2998 references the same testing methodology but requires that 5 ppb not be exceeded.

The test method was originally designed for electrostatic air cleaners and mandates a 950–1100 cubic foot inert chamber with specified dimensions and an observed ozone decay rate of 1.33.

The Blueprint notes several challenges in applying this procedure directly to far-UVC lamps, especially because lamp placement, monitor placement, photon pathlength, and room-average fluence rate can substantially affect the result.

At least two far-UVC devices have passed UL 2998 under these conditions, although the Blueprint reports that higher-output devices generally pass UL 867 but not UL 2998.

Because the procedure is ambiguous for evaluating far-UVC devices, it is not clear what the result means for how much ozone a device will actually produce in a real installation.

The standard also does not account for how many devices may be installed in one space.

Creating a new test standard that unambiguously measures ozone production from far-UVC devices is therefore described as essential.

ASHRAE

Several relevant ASHRAE standards reference UL 2998. Most importantly, far-UVC devices must have UL 2998 certification to comply with ASHRAE 241.

ASHRAE 62.1 also states that air-cleaning devices shall be listed and labeled in accordance with UL 2998, although UV devices are not currently considered air-cleaning devices for the purposes of ASHRAE 62.1.

ASTM WK81750

A new test standard for assessing the generation and removal of chemicals and particles by air cleaners is currently under development by ASTM International.

The new test method places an air-cleaning device in a sealed chamber and challenges the air cleaner with concentrations of various contaminants typical of indoor environments.

Testers then quantify the rate at which the device either reduces or produces contaminants, including ozone.

The draft protocols have been used to evaluate several air-cleaning technologies, including far-UVC lamps.

The procedure removes a number of ambiguities present in UL 867 and UL 2998 and is capable of a much lower limit of detection.

Unlike UL 867 and UL 2998, the ASTM method is currently a test method for quantifying effects rather than a pass/fail standard.

Regulations

In the US, OSHA requires that occupational exposure to ozone should not exceed 100 ppb as a time-weighted average over 8 hours, or 300 ppb over a short period.

The EPA enforces National Ambient Air Quality Standards under the Clean Air Act. The current standard for ozone is that the annual fourth-highest daily maximum 8-hour concentration, averaged over 3 years, should not exceed 70 ppb.

The California Air Resources Board (CARB) currently enforces a regulation that all air-cleaning products sold in California comply with UL 867.

Healthcare in the United States

Healthcare in the United States operates within a complex web of guidance, standards, and regulations governing infection control, air quality, and disinfection practices.

Some guidelines are advisory, some standards become enforceable through accreditation, and some regulations carry legal consequences.

At the top of the regulatory hierarchy, the Centers for Medicare & Medicaid Services (CMS) sets legally binding Conditions of Participation for hospitals and long-term care facilities receiving federal funding.

CMS does not create its own infection-control guidelines but instead relies heavily on standards set by accrediting organizations such as The Joint Commission, while state health departments also play a major regulatory and inspection role.

Beyond hospitals and long-term care facilities, infection-control guidance and regulatory oversight extend to dialysis centers, ambulatory surgical centers, behavioral health facilities, federally qualified health centers, dental offices, and other healthcare settings.

In these settings, far-UVC and other air-disinfection strategies could potentially supplement existing infection-control measures, particularly in areas with high patient throughput or aerosol-generating procedures.

Guidance

Three major guidance documents that address GUV are:

• CDC Guidelines for Preventing the Transmission of Mycobacterium tuberculosis in Health-Care Settings (2005)
• CDC Guidelines for Environmental Infection Control in Health-Care Facilities (2003, updated 2019)
• NIOSH Environmental Control for Tuberculosis: Basic Upper-Room Ultraviolet Germicidal Irradiation Guidelines for Healthcare Settings (2009)

These documents were all last updated before significant research on far-UVC emerged, and therefore focus on conventional 254-nm GUV systems.

CDC Guidelines for Preventing the Transmission of Mycobacterium tuberculosis in Health-Care Settings

The CDC TB transmission guidelines outline a hierarchical framework for TB infection control, prioritizing administrative controls, environmental controls, and personal protective equipment.

Environmental controls include ventilation, HEPA filtration, and GUV as supplementary strategies.

The CDC acknowledges upper-room 254-nm GUV as a validated environmental control method, particularly in settings where achieving high ventilation rates is impractical.

CDC Guidelines for Environmental Infection Control in Health-Care Facilities

These guidelines provide broad recommendations for preventing the transmission of infectious agents in healthcare settings.

The document recognizes both upper-room GUV and duct-mounted GUV systems as effective but supplementary air-cleaning technologies.

The 2019 update reaffirmed that GUV can be used for airborne infection control, particularly in areas where ventilation capacity is constrained.

NIOSH Environmental Control for Tuberculosis

This guidance is the primary federal document outlining best practices for upper-room 254-nm GUV specifically for tuberculosis control in healthcare settings.

It emphasizes that upper-room GUV should supplement, rather than replace, ventilation and air-exchange strategies.

The guidance recommends achieving 30–50 µW/cm² fluence rates in the upper-room zone and maintaining sufficient air mixing to bring airborne pathogens into the irradiated space.

It identifies air circulation, humidity levels, and fixture placement as major factors affecting effectiveness.

Together, these three guidance documents establish GUV as an acceptable infection-control tool, but they do not account for far-UVC because their last updates occurred before interest in the technology grew.

For far-UVC to be formally integrated into infection-control policies, these guidance documents would need to be revised or new guidance issued to establish best practices, safety guidelines, and efficacy benchmarks for 222-nm UV systems.

Additional Healthcare Guidance

The Guideline for Disinfection and Sterilization in Healthcare Facilities also mentions ultraviolet disinfection as a potential method for environmental decontamination.

Because far-UVC systems are being researched for their surface-disinfection potential, this guidance may be relevant for future far-UVC deployment in healthcare settings.

Facilities Guidelines Institute

The Facilities Guidelines Institute (FGI) establishes design and construction standards for healthcare facilities, including requirements for airborne infection isolation rooms.

Current FGI guidelines do not include GUV as a primary control strategy for airborne infection isolation.

If far-UVC were to be considered as an alternative or supplement to ventilation, FGI standards would need to be updated to specify performance benchmarks, required fluence rates, and installation best practices.

The Joint Commission

The Joint Commission is the largest and most influential healthcare accrediting body in the United States, overseeing more than 22,000 healthcare organizations.

CMS grants The Joint Commission “deemed status,” meaning that hospitals accredited by TJC automatically meet CMS Conditions of Participation and do not require separate federal inspections.

TJC does not currently reference germicidal UV or far-UVC in its infection-control standards, but it plays a critical role in determining how new technologies are integrated into healthcare settings.

ASHRAE 170

ASHRAE Standard 170 is the primary ventilation standard for healthcare facilities in the United States, defining minimum requirements for airflow rates, filtration, pressure relationships, temperature, and humidity control.

While ASHRAE 170 does not explicitly reference GUV or far-UVC, it directly impacts how these technologies could be implemented in healthcare settings.

None of the healthcare standards discussed above currently allow GUV to be used in place of ventilation. As such, GUV use in healthcare settings outside of TB is almost exclusively an optional add-on rather than something required or incentivized by existing standards.

Regulations

FDA

The Food and Drug Administration (FDA) regulates UV devices intended for medical use, such as disinfecting medical instruments or preventing disease in humans.

If a UV device is marketed for use in hospitals or healthcare settings, or on patients, it may be considered a medical device under the Federal Food, Drug, and Cosmetic Act and require FDA clearance or approval before marketing.

The distinction between an EPA-regulated and FDA-regulated UV device depends largely on the claims made.

A UV device marketed to kill airborne bacteria in a home, office, or public space remains under EPA jurisdiction because it is targeting microbes in the environment.

However, if the same device is marketed as reducing hospital-acquired infections, protecting immunocompromised patients, or preventing disease transmission, it can become an FDA-regulated medical device.

For manufacturers, the regulatory pathway they choose affects compliance costs, time to market, and product claims.

EPA regulation under FIFRA is likely faster and less expensive—devices do not require premarket approval, but manufacturers must follow labeling requirements and avoid unverified medical claims.

FDA regulation is significantly more complex and costly, requiring premarket clearance, clinical or performance testing, and adherence to FDA’s Quality System Regulations.

While FDA clearance offers greater credibility, it may add significant delays. Manufacturers seeking to sell UV disinfection devices for healthcare settings must carefully assess whether their claims will trigger FDA medical-device requirements, as failing to comply could result in regulatory action or product recalls.

For healthcare facilities, EPA versus FDA jurisdiction may impact procurement, compliance, and liability.

Hospitals using EPA-regulated UV devices, such as upper-room GUV fixtures, must independently verify their effectiveness, as EPA does not test or certify device efficacy before sale.

In contrast, FDA-cleared UV devices must meet safety and performance standards, offering greater assurance of effectiveness and regulatory compliance.

If a hospital purchases an FDA-cleared far-UVC system, it must be used according to FDA-approved indications and may be subject to routine maintenance and calibration requirements.

Conversely, an EPA-regulated far-UVC system can be installed more freely but must adhere to occupational safety guidelines and FIFRA labeling requirements.

Centers for Medicare and Medicaid Services (CMS)

The Centers for Medicare and Medicaid Services (CMS) has significant power over the healthcare sector, largely through its Conditions of Participation (CoPs) for Medicare and Medicaid reimbursement.

Hospitals and healthcare facilities must meet CMS infection-prevention requirements to remain eligible for federal funding, making CMS one of the most powerful regulatory forces in the healthcare system.

While CMS does not directly approve or regulate specific disinfection technologies like far-UVC, its policies influence which infection-control measures hospitals prioritize by determining what counts as a Hospital-Acquired Infection (HAI) and what infection-prevention strategies are reimbursable.

One of the most important CMS policies related to infection control is its Hospital-Acquired Condition (HAC) Reduction Program, which penalizes hospitals with high rates of preventable infections.

Under this program, hospitals that perform poorly on infection metrics—including central line-associated bloodstream infections, catheter-associated urinary tract infections, and surgical site infections—receive reduced Medicare reimbursements.

This creates a strong financial incentive for hospitals to adopt infection-control measures that demonstrably reduce HAIs.

Notably, the HAC Reduction Program does not currently include pathogens such as SARS-CoV-2, influenza, or measles, which diminishes a significant incentive for hospitals to implement measures specifically targeting airborne transmission.

CMS has implemented measures that incentivize preventive actions against respiratory infections in long-term care facilities and nursing homes, emphasizing the importance of airborne infection control in these settings.

Long-term care facilities will be required to electronically report data on COVID-19, influenza, and respiratory syncytial virus (RSV), as well as vaccination status and hospitalizations of residents due to those pathogens.

Another important consideration is CMS’s reliance on accrediting organizations like The Joint Commission to enforce compliance with infection-prevention standards.

If The Joint Commission or FGI were to adjust their standards to facilitate the adoption of far-UVC, hospitals seeking accreditation could implement it as part of their compliance strategy.

Role of Other Federal Agencies

Several federal agencies—including the Department of Veterans Affairs (VA), Department of Defense (DoD), General Services Administration (GSA), and the United States Department of Agriculture (USDA)—are potential large-scale buyers of far-UVC technology.

These agencies manage hospitals, military bases, federal office buildings, and food-safety facilities, all of which could benefit from enhanced airborne infection control.

Because of their size and purchasing power, their decisions to invest in far-UVC could meaningfully shift the market by driving economies of scale and increasing industry competition.

These agencies may also seek input from expert agencies like the CDC before adopting new disinfection technologies, meaning government-wide coordination could be key to accelerating far-UVC adoption.

The Department of Energy (DOE) has a particularly relevant role in advancing far-UVC technology, both through scientific research and energy-efficiency programs.

DOE’s Solid-State Lighting Program has historically supported the development of UV-C LEDs and other novel lighting technologies, which could extend to far-UVC sources in the future.

DOE also plays a role in building-technology innovation, particularly through its work with the GSA Green Proving Ground program, which tests and evaluates emerging energy-efficient building technologies.

If far-UVC were positioned as an energy-efficient alternative or complement to ventilation-based infection control, DOE and GSA could play a role in demonstrating its viability in federal buildings and providing critical real-world data for further adoption.

National laboratories under DOE, such as Lawrence Berkeley National Laboratory, Pacific Northwest National Laboratory, and the National Renewable Energy Laboratory, also have expertise in indoor air quality, photonics, and lighting technologies.

Their further involvement in testing far-UVC systems for safety, efficacy, and energy efficiency could help establish scientific credibility and create the foundation for future standards.

Further Reading

• ASHRAE 241
• How Much Ventilation Is Enough?
• Illuminate

References

1. US Environmental Protection Agency (Office of Air and Radiation). Guide to Air Cleaners in the Home. (2018).
2. US Centers for Disease Control and Prevention. How Much Ventilation Is Enough? Centers for Disease Control and Prevention https://www.cdc.gov/niosh/ventilation/prevention/Aim-for-5.html (2024).
3. US Centers for Disease Control and Prevention. Environmental Control for Tuberculosis: Basic Upper-Room Ultraviolet Germicidal Irradiation Guidelines for Healthcare Settings. Centers for Disease Control and Prevention https://www.cdc.gov/niosh/docs/2009-105/default.html.
4. ASHRAE. ASHRAE GPC 37 Guidelines for the Application of Upper-Air (Upper Room) Ultraviolet Germicidal (UV-C) Devices to Control the Transmission of Airborne Pathogens.
5. Deutsches Institut für Normung. Entkeimung von Raumluft Mit UV-Strahlung - UV-C-Sekundärluftgeräte (Disinfection of Room Air with UV Radiation - UV-C Secondary Air Units). (2022).
6. ASHRAE. ANSI/ASHRAE 185.3: Method of Testing Commercial and Industrial In-Room Air-Cleaning Devices and Systems for Microorganism Bioaerosol Removal or Inactivation in a Test Chamber. (2024).
7. Association of Home Appliance. ANSI/AHAM AC-5 Method for Assessing the Reduction Rate of Key Bioaerosols by Portable Air Cleaners Using an Aerobiology Test Chamber. (2023).
8. International Organization for Standardization. ISO 16000-36:2018: Indoor Air — Part 36: Standard Method for Assessing the Reduction Rate of Culturable Airborne Bacteria by Air Purifiers Using a Test Chamber. https://www.iso.org/standard/65001.html (2024).
9. ASHRAE. ASHRAE 241 Control of Infectious Aerosols; Normative Appendix A. (2023).
10. ASTM International. ASTM E3273-21 Standard Practice to Assess Microbial Decontamination of Indoor Air Using an Aerobiology Chamber. (2021).
11. State Administration for Market Regulation in China. GB/T 18801-2022 Air Cleaner. (2022).
12. Association of Home Appliance Manufacturers. ANSI/AHAM AC-1-2015 Method for Measuring Performance of Portable Household Electric Room Air Cleaners. (2023).
13. Nukit Lantern 222 nm Far-UVC Light (US Warehouse). Nukit 222 https://www.nukit222.com/products/nukit-lantern-222nm-far-uvc-light-us-warehouse.
14. US Environmental Protection Agency. Federal insecticide, fungicide, and rodenticide act. https://www.epa.gov/sites/default/files/documents/fifra.pdf (2008).
15. e-CFR 40 CFR Part 167: Registration of Pesticide and Active Ingredient Producing Establishments, Submission of Pesticide Reports. https://www.ecfr.gov/current/title-40/chapter-I/subchapter-E/part-167?toc=1.
16. US Environmental Protection Agency. EPA Regulations About UV Lights That Claim to Kill or Be Effective Against Viruses and Bacteria. https://www.epa.gov/sites/default/files/2020-10/documents/uvlight-complianceadvisory.pdf (2020).
17. US Environmental Protection Agency. OCSPP 810.2500-Air Sanitizers. (2013).
18. US Environmental Protection Agency. OCSPP 810.2200, Disinfectants for Use on Environmental Surfaces, Guidance for Efficacy Testing. (2018).
19. US Environmental Protection Agency. OCSPP 810.2100, Sterilants, Sporicides, and Decontaminants, Guidance for Efficacy Testing. (2018).
20. US Environmental Protection Agency. EPA’s Research and Regulation of Pesticidal Air Treatment Devices. https://www.youtube.com/watch?v=BP_-poHzo1k (2023).
21. Honourable Gilles LePage. GNB.ca https://www.gnb.ca/en/gov/contacts/members-executive-council/mandates-letters/mandates-lepage.html (2024).
22. Measure HH - City of Berkeley - Indoor Air Quality. https://www.californiachoices.org/alameda-measure-hh-2024 (2024).
23. Link, M. F. Spectral Characteristics and Indoor Air Quality Effects of Germicidal 254 Nm and 222 Nm Ultraviolet Light. http://dx.doi.org/10.6028/nist.ir.8550 (2024) doi:10.6028/nist.ir.8550.
24. International Code Council. Section C405 Electrical Power and Lighting Systems (International Energy Conservation Code). (2025).
25. TLVs and BEIs: Based on the Documentation of the Threshold Limit Values for Chemical Substances and Physical Agents & Biological Exposure Indices. Ultraviolet Radiation. (American Conference of Governmental Industrial Hygienists, Cincinnati, OH, 2022).
26. International Commission on Non-Ionizing Radiation Protection (ICNIRP). Guidelines on limits of exposure to ultraviolet radiation of wavelengths between 180 nm and 400 nm (incoherent optical radiation). Health Phys. 87, 171–186 (2004).
27. Illuminating Engineering Society. ANSI/IES RP 27.1-22: Photobiological Hazards From UV Lamps. (2022).
28. Thornton, J. GUV Training. https://www.nalmco.org/guv-training.
29. International Organization for Standardization. ISO 15858:2016: UV-C Devices — Safety Information — Permissible Human Exposure. https://www.iso.org/standard/55553.html (2022).
30. International Organization for Standardization. ISO/DIS 15858 (draft Standard): UV-C Devices — Safety Information — Permissible Human Exposure. https://www.iso.org/standard/85040.html.
31. Directive 2006/25/EC of the European Parliament and of the Council of 5 April 2006 on the minimum health and safety requirements regarding the exposure of workers to risks arising from physical agents (artificial optical radiation). EUR-Lex https://eur-lex.europa.eu/eli/dir/2006/25/oj/eng.
32. OSHA 1910 Subpart Z - Toxic and Hazardous Substances. 1910 vol. Subpart Z (1975).
33. Bolton, J. R. Workplace exposure limits for ultra-violet radiation. https://www.osha.gov/laws-regs/standardinterpretations/2003-02-26.
34. UL. ANSI/UL 8802: Ultraviolet (UV) Germicidal Equipment and Systems. (2024).
35. UL. ANSI/CAN/UL 8803:2023. Portable UV Germicidal Equipment With Uncontained UV Sources. (2023).
36. UL/CSA/IEC. UL/CSA/IEC 60730 Automatic Electrical Controls. (2024).
37. Eadie, E., O’Mahoney, P., Ibbotson, S. H., Miller, C. C. & Wood, K. Far-UVC: The impact of optical filters on real-world deployment. Photochem. Photobiol. (2024) doi:10.1111/php.14005.
38. Buonanno, M., Welch, D. & Brenner, D. J. Exposure of human skin models to KrCl excimer lamps: The impact of optical filtering. Photochem. Photobiol. 97, 517–523 (2021).
39. International Electrotechnical Commission. ISO/IEC 17025: Testing and Calibration Laboratories. (2017).
40. 47 CFR Part 15: Radio Frequency Devices. eCFR.gov https://www.ecfr.gov/current/title-47/chapter-I/subchapter-A/part-15.
41. US Centers for Disease Control and Prevention. Ozone. Centers for Disease Control and Prevention https://www.cdc.gov/niosh/npg/npgd0476.html.
42. Dinuoscio, C. Ozone. ACGIH https://www.acgih.org/ozone/ (2022).
43. What are the WHO Air quality guidelines? World Health Organization https://www.who.int/news-room/feature-stories/detail/what-are-the-who-air-quality-guidelines.
44. ASHRAE. ASHRAE Position Document on Filtration and Air Cleaning (Preprint). Preprint at https://www.ashrae.org/file%20library/about/position%20documents/filtration-and-air-cleaning-pd-2024.pdf (2015).
45. UL. UL 867 - Electrostatic Air Cleaners. (2011).
46. UL. UL 2998 - Zero Ozone Emissions Validation. (2020).
47. Model 211-G Ozone Monitor. https://2btech.io/wp-content/uploads/docs/brochures/model_211-G.pdf.
48. Shenzhen Nukit Technology LLC - Test Report. https://cdn.shopify.com/s/files/1/0872/8438/4035/files/Nukit_Lantern_UL-1_UL2998.pdf?v=1736930929 (2025).
49. Visium 1 Product Specifications. https://litthinking.com/wp-content/uploads/2024/09/Visium-Standard-Spec-Sheet-2024-Final-091724.pdf (2024).
50. Product Support. Lit Thinking https://litthinking.com/visium/product-support/ (2024).
51. ASTM International. ASTM WK81750: New Test Method for Chemical Assessment of Air Cleaning Technologies. https://www.astm.org/workitem-wk81750.
52. Link, M. F., Robertson, R., Claflin, M. S. & Poppendieck, D. Quantification of byproduct formation from portable air cleaners using a proposed standard test method. Environ. Sci. Technol. 58, 7916–7923 (2024).
53. Ozone. OSHA.gov https://www.osha.gov/chemicaldata/9.
54. US Environmental Protection Agency. NAAQS Table. EPA.gov https://www.epa.gov/criteria-air-pollutants/naaqs-table (2014).
55. Complying with the Air Cleaner Regulation: Frequently Asked Questions. California Air Resources Board https://ww2.arb.ca.gov/resources/fact-sheets/complying-air-cleaner-regulation.
56. California SB1308: Ozone: indoor air cleaning devices. LegiScan https://legiscan.com/CA/text/SB1308/id/2962681.
57. Centers for Disease Control and Prevention. Guidelines for Preventing the Transmission of Mycobacterium tuberculosis in Health-Care Settings. https://www.cdc.gov/mmwr/pdf/rr/rr5417.pdf (2005).
58. Centers for Disease Control and Prevention. Guidelines for Environmental Infection Control in Health-Care Facilities. https://www.cdc.gov/infection-control/hcp/environmental-control/index.html (2003).
59. National Institute for Occupational Safety and Health. Environmental Control for Tuberculosis: Basic Upper-Room Ultraviolet Germicidal Irradiation Guidelines for Healthcare Settings. (2009).
60. Centers for Disease Control and Prevention. Disinfection and Sterilization Guideline. https://www.cdc.gov/infection-control/hcp/disinfection-and-sterilization/index.html (2008).
61. Uses of the Guidelines. Facility Guidelines Institute https://fgiguidelines.org/guidelines/uses-of-the-guidelines/ (2022).
62. History of The Joint Commission. Jointcommission.org https://www.jointcommission.org/who-we-are/facts-about-the-joint-commission/history-of-the-joint-commission.
63. R3 Report Issue 41: New and Revised Requirements for Infection Prevention and Control for Critical Access Hospitals and Hospitals. https://www.jointcommission.org/standards/r3-report/r3-report-issue-41-new-and-revised-requirements-for-infection-prevention-and-control-for/.
64. ASHRAE. ANSI/ASHRAE/ASHE Standard 170-2017, Ventilation of Health Care Facilities. (2017).
65. Center for Devices & Radiological Health. Getting a Radiation Emitting Product to Market: Frequently Asked Questions. Food and Drug Administration https://www.fda.gov/radiation-emitting-products/electronic-product-radiation-control-program/getting-radiation-emitting-product-market-frequently-asked-questions (2025).
66. Product Classification: Whole room microbial reduction device. Food and Drug Administration https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpcd/classification.cfm?id=2995.
67. US Food and Drug Administration. (2023).
68. De Novo Classification Request. US Food and Drug Administration https://www.fda.gov/medical-devices/premarket-submissions-selecting-and-preparing-correct-submission/de-novo-classification-request (2024).
69. Quality System (QS) Regulation/Medical Device Current Good Manufacturing Practices (CGMP). U.S. Food and Drug Administration https://www.fda.gov/medical-devices/postmarket-requirements-devices/quality-system-qs-regulationmedical-device-current-good-manufacturing-practices-cgmp (2024).
70. 42 CFR § 482.42 - Condition of participation: Infection prevention and control and antibiotic stewardship programs. LII / Legal Information Institute https://www.law.cornell.edu/cfr/text/42/482.42.
71. Hospital-Acquired Condition Reduction Program. Centers for Medicare and Medicaid Services https://www.cms.gov/medicare/payment/prospective-payment-systems/acute-inpatient-pps/hospital-acquired-condition-reduction-program-hacrp.
72. Long-Term Care (LTC) Facility Acute Respiratory Illness Reporting Requirements. Centers for Medicare and Medicaid Services https://www.cms.gov/medicare/health-safety-standards/quality-safety-oversight-general-information/policy-memos/policy-memos-states-and-cms-locations/long-term-care-ltc-facility-acute-respiratory-illness-reporting-requirements.
73. Solid-State Lighting. Energy.gov https://www.energy.gov/eere/ssl/solid-state-lighting.
74. Green Proving Ground Program. Energy.gov https://www.energy.gov/eere/solar/green-proving-ground-program.

10. UVC Impacts on Materials

Summary

The built environment contains a myriad of materials that may be exposed to UVC during whole-room irradiation.

The majority of UV materials degradation research has focused on the effects of the UVA and UVB components of sunlight, as the atmosphere filters out UVC. As such, the impact of far-UVC on materials and resulting second-order effects are largely unknown.

While there is research evaluating the effects of conventional 254-nm UVC irradiation on materials, the protocols for characterizing degradation, including irradiation dose and time, are not standardized, and often the degradation mechanism is complex and material-specific.

Crucial Considerations

• Chromophores absorb light and emit light at specific wavelengths. These chromophores can degrade into radicals that form volatile carbonyl products, including ketones, aldehydes, carbon dioxide, and carbon monoxide.
• Commercially formulated polymers often contain additives, impurities, or residual catalysts, meaning they are not pure polymers, which makes understanding their degradation mechanisms and side products more challenging.
• Wood and paper are vulnerable to UVC radiation due to the degradation of lignin, a complex organic polymer. The complexities of lignin structure result in a variety of secondary products from UVC irradiation, which leads to uncertainty about the secondary effects of UVC degradation.
• Material reflectance can cause greater fluence rates and skin and eye exposures than expected. Metal materials have the highest reflectance of 222-nm far-UVC radiation, though this can be reduced with a protective coating or proper lamp placement.
• Many studies only assess color change when testing far-UVC degradation, while other studies include surface roughness, water contact angle, structural changes, and strength changes. Testing should be comprehensive, but also relevant to the material and the environment in which it will exist.
• We should also assess whether far-UVC causes any off-gassing of harmful compounds when interacting with common materials.

Analysis

The impact of UVC on the vast variety of material types and classes, especially those in the built environment, is profoundly understudied.

Common experience shows us that ultraviolet radiation, including conventional UVC, can degrade and discolor materials like plastics, fabrics, and wood.

While conventional UVC research can be used as a guide, the paucity of data concerning the impact of far-UVC on different material classes suggests a preliminary goal of prioritizing which materials should be tested.

A summary of the known impact on materials of UVC and far-UVC, where applicable, is below.

Wood

Wood rapidly absorbs UV in the first ~80 μm of its surface. Lignin, one of the major constituents of wood, is the primary absorber.

Upon UV exposure, lignin breaks down into water-soluble products and generates other species that may absorb radiation.

Lignin is mainly degraded due to its capacity to absorb radiation of short wavelengths, with an absorbance peak at 280 nm. Wood’s potential absorbance in the far-UVC range depends on the chromophores present.

Photoinduced degradation of lignin results in the discoloration, often yellowing, observed on the surface of wood.

In normal weathering, water assists in removing loosened fibers and particles released during irradiation and also causes leached lignin fragments to move to the surface, leading to discoloration and a rough texture.

While this has not yet been directly shown for far-UVC, it is expected that the same effect will occur following photodegradation.

However, it is not expected that bare wood will usually be present in the initial deployment built environment. Most wood present will be covered with a finish and will not be exposed to natural weathering.

The impact of far-UVC exposure on these wood finishes is unclear.

Paper

Paper is known to undergo discoloration during and after exposure to UV radiation. However, there are limited research studies that have investigated the degradation mechanism.

One study investigated bond, rice, kraft, and amate papers exposed to 254-nm radiation for up to 480 hours. While no chemical decomposition was observed, color loss was observed, especially in the colored papers.

Paper is composed of a mixture of cellulose, hemicellulose, lignin, and different lignin compounds.

As discussed in the Wood section, lignin is degraded by UVC and this leads to discoloration, often yellowing, of the surface.

With far-UVC, there is a strong chance that paper will discolor, but the time and dose required is unclear.

Furthermore, it is unclear whether there is a difference between the mechanisms of degradation under 254-nm irradiation and far-UVC irradiation.

Metals

There is likely no effect of UVC and far-UVC on metal surfaces, as metals have available free electrons which can absorb photon energy without undergoing energy transitions or bond dissociation.

In one study, aluminum and stainless steel were exposed to 254 nm at a dose equivalent to 16 years of exposure.

This showed only discoloration in the stainless steel after the equivalent of 8 years of daily exposure at 36.8 mJ/cm².

Generally, metals do not undergo a color change with UVC exposure unless the metal is oxidized, and the authors attributed the stainless-steel discoloration to surface oxidation.

Neither material showed any changes in structural integrity.

While specific testing of far-UVC on commonly used materials in the built environment would provide conclusive evidence for or against the possibility of degradation or discoloration, the lack of structural changes at 254 nm suggests that metals are less likely to be affected than some other commonly used materials.

Metals will reflect far-UVC, but irradiance decreases with distance.

Operating rooms, and other areas where a large percentage of the surface area is covered in metal, may pose a risk due to increased reflectance.

Therefore, initial measurements should be taken to ensure any reflectance does not increase potential harm or increase exposure beyond daily limits.

Ceramics and Glass

Ceramics and glass have tightly bonded electrons, so they are generally unaffected by UV radiation.

There are no literature studies on the degradation effects of UVC on ceramic materials, further indicating that they are believed to be largely unaffected by UVC radiation.

However, ceramic and glass materials may have synthetic coatings which may be susceptible to UVC degradation.

Textiles

Various textiles are part of a built environment, from fabric on furniture to the clothes that individuals are wearing.

Textiles can be made from natural materials, like wool or cotton, synthetic materials, or a combination. Synthetic fabrics and textiles are expected to have the same degradation mechanisms as those of polymers.

In addition to the composition, the dyes used to achieve different colors must also be considered when investigating any potential UVC degradation mechanisms.

Accelerated UVC testing was conducted on various fabrics including SILVERTEX, acrylic, cotton, various colors of polyester, and leather.

The fabrics, especially green cotton, lost their color intensity with UVC irradiation, turning either more yellow or white depending on the original fabric color. In particular, the cellulose in cotton degraded and turned yellow.

No change in tensile strength was observed.

Honeywell’s evaluation of fabrics used in airplanes also only observed color changes with exposure to 254 nm with a dose corresponding to 10.8 years of daily use. No significant changes in mechanical properties were observed.

One study evaluated 222-nm exposure, corresponding to ~6.2 years of daily exposure, on the velour fabric of public bus seats. There was not a quantitative difference in the color of the velour fabric following exposure.

Exposure to shorter-wavelength UVC may have less of an effect on the color of the fabric. However, the other studies that observed a reduction in color intensity used different textile compositions, which would impact the degradation mechanism of the fabric.

Polymers

Numerous studies have investigated the degradation mechanisms of polymers over the past decade.

Although most of these reports are for UVA (315–399 nm) and UVB (280–314 nm) radiation, some studies explore the effects of UVC radiation.

The American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) performed a literature study of the different degradation mechanisms of various polymers and established a testing procedure for HVAC materials.

Pure Polymers

For a polymer to absorb UVC, chromophores must be present in the polymer’s structure or polymer matrix.

In addition to the structure of the polymer, the UV reactivities of the polymers are influenced by the presence of additives and impurities, which may have UV-absorbing properties formed during polymerization, processing, and/or storage.

There are currently no investigations identified in the Blueprint on how the location of chromophores in a polymer matrix influences degradation of the polymer.

In theory, since polyethylene (PE) and polypropylene (PP) do not have any chromophoric groups in their polymer chain, they do not absorb UV above 220 nm and therefore should be resistant to UVC.

Teska and colleagues determined that PP and ultrahigh-weight PE had the least damage when compared to other plastics used in healthcare that were exposed to UVC.

PE and PP may still be susceptible to UVA and UVB degradation depending on any leftover initiator residue and/or chromophoric groups formed during polymerization of the polymer.

Polymers With Additives

Polymers are often not pure and have additives that can prevent or enhance UV degradation.

Inorganic fillers can form a protective coating on the surface or can act as an energy sink for the UV energy absorbed by the polymer.

Exposure of a polymer to UVC can volatilize its surface and reveal the layer of inorganic filler beneath. The inorganic filler can offer protection to the interior of the polymer material.

Protective coatings such as epoxy resins often incorporate additives to stabilize the polymer from environmental factors including light, heat, water, and oxygen.

Often these additives are metal powders, such as aluminum, due to their excellent UV reflective characteristics.

The addition of spherical aluminum powder to epoxy resin provided some resistance to UV degradation 100 days after exposure, but the analysis did not include the wavelength or dose, making it difficult to understand the mechanism.

For the Drungilas study summarized below, samples were exposed to 222 nm for 50, 100, or 150 hours, corresponding to 96.7, 193.4, and 290 J/cm². Irradiation for 150 hours, equivalent to 290 J/cm², corresponds to approximately 6.2 years of daily disinfection.

For the Harris study, 36,000 doses over 25 years of in-service use were assumed, with each dose being 3 seconds. The table reports color changes, and no adverse changes in mechanical properties of thermoplastics or textiles or in flammability properties were reported.

Table 10.1. Material Studies Using 222 nm

Further Reading

• Evaluating the impact of 222 nm far-UVC radiation on the aesthetic and mechanical properties of materials used in public bus interiors
• Compatibility of Aircraft Interior Surfaces with 222 nm Far-UV Light Exposure

References

1. Sørensen, S. B. & Kristensen, K. Indoor Surface Emissions of Volatile Organic Compounds induced by Germicidal UV (222 & 254 nm) Illumination. ChemRxiv (2024). doi:10.26434/chemrxiv-2024-4w3qp.
2. Gržan, T., Grieco, L., Živković, V. & Mihulja, G. UV irradiation of wood surface: Bonding properties. Polymers (Basel) 15, 2552 (2023).
3. Mishra, S. & Sain, M. Long-term performance of natural fiber composites. In Properties and Performance of Natural-Fibre Composites, 460–502 (Woodhead Publishing, 2008).
4. Hon, D. N.-S. & Chang, S.-T. Surface degradation of wood by ultraviolet light. J. Polym. Sci. Polym. Chem. Ed. 22, 2227–2241 (1984).
5. Sadeghifar, H. & Ragauskas, A. Lignin as a UV light blocker—A review. Polymers (Basel) 12, 1134 (2020).
6. Hon, D. N.-S. & Feist, W. C. Weathering characteristics of hardwood surfaces. Wood Sci. Technol. 20, 169–183 (1986).
7. Hon, D. N. S. & Minemura, N. Color and discoloration. In Wood and Cellulosic Chemistry, 385–442 (New York, NY, USA, 2000).
8. Jiménez-Reyes, M., Tenorio, D., García-Rosales, G., Jiménez-Becerril, J. & Luna-Castro, G. E. Effects of UV Radiation on Paper: A Chromatic Study. J. Anal. Chem. 8 (2021).
9. Mitxelena-Iribarren, O. et al. Evaluation of the degradation of materials by exposure to germicide UV-C light through colorimetry, tensile strength and surface microstructure analyses. Mater. Today Commun. 31, 103690 (2022).
10. Kauffman, R. E. Study of the Degradation of Typical HVAC Materials, Filters and Components Irradiated by UVC Energy—Part I: Literature Search. ASHRAE Trans. (2012).
11. Honeywell. Effect of UV-C on aircraft interior materials. (2020).
12. Drungilas, D. et al. Evaluating the impact of 222 nm far-UVC radiation on the aesthetic and mechanical properties of materials used in public bus interiors. Appl. Sci. (Basel) 13, 4141 (2023).
13. Teska, P., Dayton, R., Li, X., Lamb, J. & Strader, P. Damage to common healthcare polymer surfaces from UV exposure. Nano Life 10, 2050001 (2020).
14. Gogotov, I. N. & Barazov, S. K. The effect of ultraviolet light and temperature on the degradation of composite polypropylene. Int. Polym. Sci. Technol. 41, 55–58 (2014).
15. Chylińska, M., Kaczmarek, H., Moszyński, D., Królikowski, B. & Kowalonek, J. Surface studies of UV irradiated polypropylene films modified with mineral fillers designed as piezoelectric materials. Polymers (Basel) 12, 562 (2020).
16. Zhai, Z., Li, H., Feng, L. & Li, F. Effect of UV light on the barrier performance of aluminum powder/epoxy coating. Polymers (Basel) 14, 2405 (2022).
17. Harris, J., Metting, S., Sharma, A. & Elting, A. Compatibility of Aircraft Interior Surfaces with 222 Nm Far-UV Light Exposure. Boeing (2021).

Appendix A: Acknowledgements

We are particularly grateful to the following individuals who read and provided feedback on early drafts of one or more sections of this report. Their input has been invaluable in improving the quality of the report.

This report does not represent a consensus among these individuals, nor does it represent any of their individual opinions nor those of the institutions they are affiliated with. The resulting report solely represents the view of Blueprint Biosecurity, and any remaining errors are, of course, our own.

Additionally, a profound thank you goes to the experts below who lent us their time and insight as we wrote the report and updated it based on feedback. Without their contributions it would not have been possible for us to produce this report.

Appendix B: Impacts on Non-Human Life

Summary

This section discusses the potential effects of far-UVC on other organisms that can be present in indoor public spaces.

Most mammals are thought to have the same protections as humans, or more, with fur and lower height reducing their effective exposure to far-UVC from a ceiling-mounted lamp. Therefore, this section primarily discusses arthropods and plants.

Initial analysis suggests that arthropods, such as insects and arachnids, will be protected from far-UVC exposure by their chitin- and protein-rich exoskeleton.

The vast difference in chitin concentration and exoskeleton thickness across the arthropoda phylum makes it difficult to study each species individually. Instead, studies should evaluate far-UVC absorption and penetration into exoskeletons of various chitin concentrations.

The impact of far-UVC on the insect eye is also unclear.

Studies suggest that far-UVC in high intensities could have a negative effect on plants, but more research is needed. Similar to arthropods, the absorption and penetration of far-UVC into leaves, flowers, and other plant structures of varying thickness and color should be investigated.

Analysis

Arthropods

Arthropods are characterized, among other criteria, by their cuticle or exoskeleton—a multi-layered structure secreted by the epidermis, comprising chitin and proteins.

Chitin is a fibrous substance composed of polysaccharides and is one main component of the exoskeleton. Chitin concentration in arthropod exoskeletons can vary between 2% and 45%, with the remaining matter mostly being protein, phenolics, and minerals in some cases.

While we have a good understanding of the absorption of far-UVC by the human stratum corneum, we do not have a good understanding of the absorption of the arthropod cuticle.

Cuticle thickness varies dramatically from just hundreds of nanometers in tiny ants to multiple millimeters in large crabs.

Based on studies of ants, cuticle thickness appears to roughly correlate with body size. A typical ant has a thorax cuticle thickness of over 10 µm, with large ant species reaching up to 100 µm, though the cuticle will be thinner over joints or more delicate areas.

Assuming arthropod cuticles have similar far-UVC absorption properties to the human stratum corneum, most larger insects such as bees, flies, and beetles, as well as spiders, should experience significant attenuation of far-UVC in the cuticle. However, this remains to be tested.

Effects on arthropod eyes are harder to assess. Arthropods have compound eyes comprising individual ommatidia capped by a transparent, chitinous corneal lens.

The lens is typically a few µm thick, depending on the species, but chitin content appears low at around 20%, and the UV optical properties are poorly understood.

Mosquito cuticle thickness has been researched in the context of insecticide resistance and was found to be quite thin, around 1.5–3 µm in mosquito legs.

Mosquito procuticles are less sclerotized than those of other arthropods, meaning they have less crosslinking of structural proteins and chitin to increase stiffness and toughness.

One might therefore expect significant amounts of far-UVC penetration into the underlying epidermis and nerves.

As mosquitoes are a known vector of disease, research on the impact of far-UVC on mosquitoes can also help inform other disease-vector mitigation strategies.

Plants

So far, the few studies on the effects of far-UVC on plants have found high sensitivity.

One study of irradiated Arabidopsis seedlings observed leaf curling and bleaching 2–4 days after exposure to either 100 mJ/cm² from a KrCl* lamp with imperfect filtering, or 1 J/cm² from an unfiltered low-pressure mercury lamp.

Plant homeostasis was severely impacted, as the guard cells that help regulate moisture and gas levels were damaged.

Generally, the outer layers of plants, including a wax cuticle and single-layered epidermis, have evolved to be thin and transparent to allow maximum light absorption.

The composition and thickness of the waxy cuticle varies between species, making it difficult to draw broad comparisons.

Two caveats include the lamp used and the dose intensity. The KrCl* lamp used was not perfectly filtered and had some emission above 235 nm. The plants also received the far-UVC doses in a short timespan and had high far-UVC intensities during their seedling stage.

Houses and offices are two common places where far-UVC disinfection and house plants intersect.

Houses and offices will have different duty cycles for disinfection depending on occupancy and how often disinfection occurs. The location of plants, and their distance from the far-UVC lamps, will also impact the effective doses they receive.

Plants are most likely positioned lower to the ground and off-axis from lamps, suggesting that they will receive only a single-digit mJ/cm² of 222-nm far-UVC per day.

However, environments with heavier duty cycles and more uniform distribution may lead to higher exposure and potential plant damage or discoloration. Different parts of plants, including petals, stems, and leaves, may also have different susceptibilities.

It seems as if far-UVC can harm plants, especially early in their life, so it may be wise to keep it out of greenhouses.

Otherwise, individuals can reduce plant irradiation by avoiding placement high in the irradiation zone or by placing plants where they are shaded from the lamp.

The Blueprint does not consider far-UVC’s impact on arthropods a high research priority, but any impact on indoor plants may limit enthusiasm for deployment.

Simple experiments, such as including common house or office plants in real-world deployment studies, could reduce uncertainty or hesitation about deployment.

References

1. Muthukrishnan, S., Mun, S., Noh, M. Y., Geisbrecht, E. R. & Arakane, Y. Insect cuticular chitin contributes to form and function. Curr. Pharm. Des. 26, 3530–3545 (2020).
2. Peeters, C., Molet, M., Lin, C.-C. & Billen, J. Evolution of cheaper workers in ants: a comparative study of exoskeleton thickness. Biol. J. Linn. Soc. Lond. 121, 556–563 (2017).
3. Kaya, M., Sargin, I., Al-Jaf, I., Erdogan, S. & Arslan, G. Characteristics of corneal lens chitin in dragonfly compound eyes. Int. J. Biol. Macromol. 89, 54–61 (2016).
4. Wood, O., Hanrahan, S., Coetzee, M., Koekemoer, L. & Brooke, B. Cuticle thickening associated with pyrethroid resistance in the major malaria vector Anopheles funestus. Parasit. Vectors 3, 67 (2010).
5. Otake, M., Okamoto Yoshiyama, K., Yamaguchi, H. & Hidema, J. 222 nm ultraviolet radiation C causes more severe damage to guard cells and epidermal cells of Arabidopsis plants than does 254 nm ultraviolet radiation. Photochem. Photobiol. Sci. 20, 1675–1683 (2021).