COVID-19 pandemic lesson learned- critical parameters and research needs for UVC inactivation of viral aerosols

3.1. Ultraviolet pathogen inactivation

The ultraviolet light spectrum ranges from 10 to 480 nm, including three regions of importance, UVA (320–400 nm), UVB (290–320 nm), and UVC (200–290 nm) (Hadi et al., 2020). The maximum absorbance of DNA is around 265 nm, so UVC radiation exhibits the highest antimicrobial properties. When DNA/RNA absorbs UV radiation, dimeric lesions occur in the genome, preventing DNA/RNA transcription and translation and thereby inactivating the microorganism. Following the adsorption of UVC by DNA/RNA, the most critical photoproducts are cis-syn cyclobutene dimers between adjacent pyrimidines, like thymine-thymine dimers (Kowalski, 2009; Beck et al., 2016), and the formation of covalent linkage between the 6- and 4- position of two pyrimidine bases, called the ‘6-4-photoproduct’ (Douki and Cadet, 1994). Less significant to overall damage are single-strand breaks (SSB) resulting from DNA-protein cross-links (Kiefer, 2007), hydrolytic deamination which converts one base to another (i.e., cytosine <-> uracil), and hydrolytic depurination/ depyrimidination which removes a base entirely (Rastogi et al., 2010). Meanwhile, the formation of reactive oxidant species (ROS) by UVC radiation can cause damage beyond the DNA and RNA, affecting lipids, proteins and other cellular structures (Valko et al., 2007). UVC-induced damage does not directly kill pathogens but can lead to replication arrest, rendering cells unviable (Brickner et al., 2003; Dunkern and Kaina, 2002). When DNA/RNA polymerase encounters a base lesion, a Y-shaped DNA structure is created. Endonucleases recognize this incorrect DNA/RNA architecture and nick the template strand, creating a double-strand break (DSB) (Batista et al., 2009). Interestingly, single-stranded nucleic acid (ssRNA and ssDNA) viruses are more susceptible to inactivation by UVC than double-stranded nucleic acids (dsRNA and dsDNA) viruses (Thurston-Enriquez et al., 2003; Tseng and Li, 2005a). Tseng and colleagues reported that the UVC dose required for 99% inactivation of dsRNA and dsDNA viruses was twice that of ssRNA and ssDNA viruses (Tseng and Li, 2005a). In case of viruses and phages, it has been shown that UVC damages inhibits the genome injection to the host cells through capsid protein damage (Wigginton et al., 2012).

Different UVC wavelengths may affect the dominant pathogen inactivation mechanism. Eischeid and Linden (2011) reported that polychromatic medium-pressure mercury lamps were more effective at damaging adenoviral proteins than low-pressure mercury lamps, suggesting the additional UVC-spectrum activated more damage mechanisms (Eischeid and Linden, 2011) while Gerchman and colleagues reported that shorter wavelengths (267 nm) were more effective at inactivation of HCoV-OC43 (Fig. 3 ). For longer wavelengths (i.e., UVA and UVB), this means that a higher UV dose is required to achieve the same log inactivation. In fact, increasing UV wavelength to 300 nm requires a 10-fold greater dose to achieve similar inactivation when compared to 280–290 nm for MS2, Qβ, feline calicivirus, H1N1 viruses and human Coronavirus (HCoV-OC43) (Gerchman et al., 2020). Beck et al., studied the action spectra for MS2, T1UV, Qβ, T7, and T7m Coliphages, and observed that the UVC susceptibility decreases by increasing radiation wavelength from 210 to 240 nm with the highest at peak at 210 nm, then increases from 240 to 260/270 nm, and then there is another sharp decline by increasing from 270 to 300 nm (Fig. 4 ) (Beck et al., 2015).

3.2. Viral surrogate selection

Inactivation rate constants (k_uv) describe the susceptibility of a microorganism population to UVC radiation. Brickner et al. (2003) summarized the k _uv of dozens of viruses, bacteria and fungal spores from 0.0034 to 0.96 m²/J, confirming a species-dependent response to UVC-radiation (Jensen, 1964; Hollaender, 1943). Generally, susceptibility to UVC radiation can be ranked as: viruses> vegetative bacteria> mycobacteria> bacterial spores> fungal spores (Fig. 5 A) (ASHRAE, 2019). Selecting a representative viral surrogate is crucial, as working with viral pathogens such as SARS-CoV-2 requires stringent safety lab protocols (Biosafety level 3 (BSL-3)). Also, different species have vastly different log inactivation efficiency when exposed to UVC (Fig. 5B). In UVC air disinfection studies, one of the main criteria for surrogate selection is similarity in UVC susceptibility. Various pathogens show different inactivation dose responses due to the following reasons (1) the presence of a cell wall and its thickness, (2) larger genomes present a larger target for UV damage, (3) single-stranded genomes lack a repair template (compared to double-stranded), and (4) the specific protein composition of the capsid, like UV-absorbing chromophores (Tseng and Li, 2005a; Meng and Gerba, 1996).

A virus repair mechanism impacts their UV susceptibility: dsDNA viruses go through reactivation once in the cell host. The possibility of a pathogen's ability for genomic repair should be considered when selecting a viral surrogate and the applied analysis method (i.e., nebulization) (HARM, 1961; DayIII, 1974). Table S1 provides detailed information about the different types of common viruses and virus surrogates composition (e.g., TT, TC, CT, and CC, etc.), genome size and type, and repair mechanism. This dataset can help researchers to make informed decisions about selecting the virus surrogate (Rockey et al., 2021).

Common surrogates for SARS-CoV-2 in air and surface disinfection include murine hepatitis virus (MHV), human coronavirus 229 E, transmissible gastroenteritis virus (TGEV), feline infectious peritonitis virus (FIPV) (Kumar et al., 2020), human coronavirus OC43 (Buonanno et al., 2020), and influenza H1N1 (Welch et al., 2018).

Another approach to choose a surrogate that has a similar inactivation rate is through using the mathematical models specifically facing new and emerging pathogens. However, the few genomic mathematical models developed thus far are based on the UVC data obtained in water disinfection studies. While the UVC rate constant in water can be used implicitly for viral airborne inactivation, there is still a need to develop a protocol for airborne UVC disinfection, consolidate data, and establish the genomic modelling for airborne pathogens separately. UVC inactivation rate constant in water includes the UV scattering and absorption, which differ from those in the air. The genomic predictive modelling can help select the proper viral surrogate (i.e., similar to that of viral pathogen) based on the inactivation rate and susceptibility similarity and predicts the required irradiation dose.

3.3. UVC susceptibility

Airborne microbes are more sensitive to UVC when compared to microbes in films or suspensions. This effect varies between organism, with bacteria generally ∼5-fold more resistant in water than in air at low humidity. Similarly, viruses were 3-fold more resistant to UVGI inactivation when suspended in water compared to dry air (Kowalski, 2009). This was attributed to (i.) absorbance of UVC radiation by water, (ii.) turbulence in air improving the mixing of airborne organisms, (iii.) physical damage during aerosolization, and (iv.) oxygenation and dehydration at low humidity increasing pathogen vulnerability to UV inactivation.

Table 1 shows the reported log reduction of viruses in air using UVC sources. Table S2 provides the inactivation doses for viral pathogens and viral surrogates in the liquid. The number of studies is limited, and a direct comparison is not possible considering experimental variability (Section 4), operating conditions (Section 5), and measurement techniques (e.g., radiometers). Nonetheless, one can conclude that an average dose of 2 mJ/cm² could provide one log reduction (90% inactivation) of airborne viruses. Further study is needed using a uniform protocol to standardize the experimental conditions and ensure a reliable comparison.

3.4. Outlook

The literature lacks a description of the impact of the viral envelop and nucleocapsids on UVC susceptibility. Future research should seek to understand their implications for UVC efficacy. As the dose-response of each microorganism depends on the UVC wavelength, care should be taken on selecting a surrogate pathogen. However, a lack of data on UVC dose-response and protein damage, means that there is currently no solid mechanistic understanding on viral aerosol inactivation. Therefore, further research is needed to understand this inactivation mechanism across diverse species, and to develop a genomic predictive model to enable researchers to select a surrogate.

Also, few existing genomic mathematical models representing the mechanistic UVC inactivation are based on data obtained in water disinfection studies. While the UVC rate constant in water can be implicitly used for airborne inactivation, there remains a need to develop a protocol for airborne UVC disinfection, consolidate data, and establish a separate genomic model for airborne pathogens. Such a genomic predictive model can help select the proper viral surrogate based on inactivation rate and susceptibility, and predicts the required irradiation dose, and the airborne medium.

4.1. Nebulizers (pathogens aerosolization)

Pathogens can become airborne; when nebulized (i.e., aerosolized) they are referred to as bioaerosols. The probability of its survival depends on its ability to resist the stress of aerosolization (Verreault et al., 2008). Pathogens may be aerosolized by various mechanisms (Verreault et al., 2008; Aller et al., 2005)

• •

  Primary aerosolization involves the spread of microorganisms directly into the surrounding air, i.e., sneezing.

• •

  Secondary aerosolization involves the spread of microorganisms through fluids or surfaces, which later become sources of airborne transmission. For example, liquid splashes can aerosolize pathogens on liquids and surfaces.

Generation of aerosols in an experimental setting tries to mimic the primary or secondary aerosolization via different techniques and devices. Aerosols generation in a controlled environment provides the ability to study the viability, transformability, infectivity and inactivation using different disinfestation techniques more accurately. However, there are different steps prior to aerosolization that may impact the data interpretation including preparation and storage of the viral surrogates or pathogens followed by the aerosolization technique and finally the collection method (Alsved et al., 2020)

Most aerosol particles (greater than 10 µm) will not pass the upper airways of the human respiratory system and pose little concern for disease transmission. However, smaller particles travel more efficiently and may infect the tracheobronchial and alveolar regions of the human respiratory systems (Lauck, 2000). Therefore, future studies examining the efficacy of UVC for the disinfection of aerosolized microorganisms should focus on this smaller particle size.

Airborne pathogens aggregate rapidly (Lauck, 2000). Several factors, including the size distribution of the airborne particles, thermo-/hydro-dynamic conditions, relative humidity and the aerosol concentration, govern the rate of pathogen aggregation (Kowalski, 2009; Verreault et al., 2008). Bioaerosols generated in the natural environment often shrinks rapidly due to the differences in the relative humidity between the respiratory tract and the outside environment (Fig. 6 ). Further, airborne particles in clean environments are more likely to remain smaller than particles in a dirty atmosphere, with more potential to grow by adhering to other airborne particles (Morawska, 2006). Artificially generated aerosols are often tested in controlled environments where the nebulized particles cannot bind (Verreault et al., 2008). However, laboratory-specific factors like relative humidity (RH) and temperature may impact the droplet's solute concentration, resulting in variable disinfection efficacy (Verreault et al., 2008). Thus, even controlled studies from different research facilities are difficult to compare.

4.2. Sampling techniques

Aerosolized particles attach to a surface via van der Waals and electrostatic forces (Lauck, 2000). Thus, most air sampling techniques depend on factors governing the adhesion properties of airborne particles, like the aerodynamic radius, thermal gradients, Brownian motion, and particle inertia. Airborne particles of <100 nm are more susceptible to Brownian motion than larger particles, primarily due to the collision frequency of encounters with gas molecules (Verreault et al., 2008). Smaller particles with higher diffusivity exhibit a greater probability of adhering to a surface that they contact. This is the basis for the efficient removal of small particles by filtration. By contrast, larger airborne particles (µm scale) are dominated by gravitational attraction, causing them to settle on surfaces. The larger particles are also more easily diverted from a gas streamline, leading to impaction on surfaces at high velocity and consequent adhesion.

The sampling of airborne pathogens is challenged by physical collection efficiency, viability losses of the airborne pathogen due to dehydration during the collection, damages because of the impaction and re-aerosolization (particularly in impingers), and losses from bioaerosol retention to the sampler (Pan et al., 2019). Sampling techniques to collect airborne pathogens, particularly for viruses and viral surrogates (Fig. 7 ) are described below in Table 2 .

The Andersen sampler has a very low inlet efficiency and high wall losses. Impinger and Cyclone should be fully characterized for the wall loses and inlet efficiency (Henningson and Ahlberg, 1994). It has been noted that that no single sampling method is suitable for all types of bioaerosol and in choosing the appropriate sampling technique factors such as type of airborne pathogen, sampling time and volume, and suspending medium should be taken into consideration. Each sampling method provide different physical and biological efficiency as well as detection limit (Borges et al., 2021a). It is recommended that researchers should use several sampling techniques simultaneously so that limitation of one sampling technique will be compensated with others. The knowledge of the researcher about the pros and cons and limitation of sampling techniques help interpret results more accurately (Borges et al., 2021a).

Reliable data collection for the development and assessment of UVGI systems necessitates accurate and standardized sampling methods, including bioaerosols sampling. Based on the Center for Disease Control and Prevention (CDC) recommendation, bioaerosols collection (e.g., fungi, bacteria and viruses) need new sampling methods with reduced handling risk and enhanced detection accuracy (Borges et al., 2021b). UVC systems for airborne pathogen inactivation is possible with commercial air samplers described above. Table 3 shows sampling techniques and nebulizers used in the UVC inactivation studies in the last 20 years.

4.3. UVC chamber

Experimental setups for testing UVC inactivation of pathogens typically comprise a flow chamber, a specimen rack, and a UVC radiation source (Fig. 8 ) (Kowalski, 2009). Narrow-band spectral ultraviolet sources are used with mirrors to direct radiation. The radiation pattern and flow velocity must guarantee uniform dose delivery: entrance baffles create a well-mixed and consistent flow stream. The test chamber temperature and humidity are controlled. The distance from ultraviolet source to the flow stream determines a uniform light distribution and dose delivery to bioaerosols.

Rotatory drums or dynamic aerosol toroids are ideal for testing aerosols under controlled atmospheric conditions for prolonged durations, particularly for preventing the loss of airborne particles to gravitational settling (Verreault et al., 2008; Henningson and Ahlberg, 1994). Most setups for bioaerosol testing are tailor-made and comparing even controlled studies is often challenging. Some studies use additional components such as neutralizers, humidifiers, and driers to control the aerosol size distribution and humidity. The neutralizer primarily inhibits the particles with opposite charges from attaching and forming larger particles. A desiccator acts as the dryer, and while not common in combination with a humidifier, it assists in adjusting the humidity effectively. Setup variations in different studies is one of the factors contributing to the diverse dose data for even similar aerosolized surrogates. Therefore, applying a standardize test setup can lead to more conclusive data and benefit further the final dose determination for the airborne pathogens.

4.4. Outlook

While several sampling collection techniques exist, the primary challenge is to standardize the viral sampling and detection techniques. Future efforts should also determine the detection limit for each sampling procedure, so that researchers may select an appropriate protocol based on their research needs.

An optimal nebulizing and collecting medium for the targeted bioaerosols is missing: a systematic assessment should determine the optimal nebulizing and collecting medium and assess viral aerosols' collection and storage temperatures. Few studies have reported the impact of relative humidity on the collection efficiency and viability of the viral aerosols. Thus, a thorough investigation of RH and carrier media such as mucus and salt on the viability of viral aerosols could inform future UVGI applications.

5.1. Temperature

Temperature impacts the survivability, infectivity, and the UVC susceptibility of airborne microorganisms (Ijaz et al., 1985; Lowen and Steel, 2014). Operating temperature should be considered when designing a set-up and comparing the results in the literature as an effective variable on the required UVC dose.

The survivability of bioaerosols, including viruses, vegetative bacteria, spores, and fungi, have been studied over a temperature range from sub-zero to 50°C (Haddrell and Thomas, 2017, Ehrlich and Miller, 1973, Tang, 2009, Wathes et al., 1986). Extreme temperatures disrupt cellular protein, decreasing survival (Tang, 2009). However, B. subtilis spores decay rate showed no significant difference from -40 to 49°C (Ehrlich et al., 1970). Airborne Influenza survived longer at low (7–8°C) compared to moderate (20.5–24°C) and high (> 30°C) temperatures (Harper, 1961). More information on survivability changes based on the temperature are tabulated in Table S3.

At lower temperatures, the UVC susceptibility of the porcine reproductive and respiratory syndrome (PRRS) virus increased (Cutler et al., 2012). However, Zhang et al. (2020) reported a maximum UVC disinfection efficacy at 20–21°C when considering a range between 15 and 26°C on aerosolized bacteria Staphylococcus epidermidis, Pseudomonas alcaligenes, and Escherichia coli (Zhang et al., 2020). However, the mechanism of temperature's effect on bioaerosols survivability has yet to be described. Table S4 provides the UVC susceptibility changes as a response to temperature and RH humidity.

Raising temperature changes the output of ultraviolet lamp and 254 nm conversion efficiency. There is an optimal vapor pressure at which mercury emits the highest 254 nm and with increasing temperature the optimum pressure deviates and the 254 nm conversion efficiency decreases (He, 2012). Therefore, understanding and optimizing the temperature effects on airborne microorganism survivability and UV lamp output are crucial considerations toward commercial applications.

5.2. Relative humidity

Despite a report by Walker and Ko (2007) that increasing relative humidity (RH, reported in percentage) did not decrease the inactivation rate of aerosolized MS2 (Walker and Ko, 2007), it has been documented that most viruses (e.g. Influenza A, Vaccinia virus, PRRSV) and viral surrogates (e.g., MS2, phiX174, phi6, T7) exhibit a decrease in UV susceptibility at higher RH (Thornton et al., 2022). Riley and Kaufman (1972) reported a decrease in UVC inactivation for Serratia marcescens at RH above 60% (Riley and Kaufman, 1972). Koller (1939) and Whiser (1940) claimed that airborne bacteria are ten times more resistant to ultraviolet light in high- than low-RH conditions (Rentschler et al., 1941). Peccia et al. (2001) reported a decrease in the inactivation rate of three bacteria species at RH above 50% (Jordan et al., 2001). Cutler et al. (2012) assessed the temperature and RH effects on UVC inactivation of airborne PRRS virus. The statistical analysis showed the significant impact of UVC dose, temperature, RH, and the interactions between UV dose and temperature and the product of UV dose and RH. The highest PRRS virus inactivation rate happened at RH between 25% and 79% (Cutler et al., 2012). More information on UVC susceptibility due to (Ma et al., 2020)the RH and temperature are tabulated in Table S4.

The reduction in the inactivation rate at higher RH may result from the protective layer of water coating around bioaerosols. Also, larger particle sizes may better protect against the destructive impact of UV radiation (Tang, 2009). Non-enveloped viruses like respiratory adenoviruses and rhinoviruses tend to survive longer at higher humidity levels (70–90%). Whereas lipid-enveloped viruses survive longer at low RH (20–30%), including Influenza, coronaviruses (SARS-CoVs), respiratory syncytial virus, parainfluenza viruses, measles virus, rubella, and varicella-zoster virus (Tang, 2009). The survival rate disparities in various RHs are associated with cross-linking between the virus surface protein moieties (Cox, 1989). The nature of reactions determined the extent of the damage to the viruses, which occurred through different mechanisms like hydration/ dehydration, UV radiation, ozonation and other stressors (Cox, 1989). it has been reported that lower temperature and humidity are linked to higher transmissibility of SARS-CoV-2 which translate to higher survivability (Ma et al., 2020). Therefore, it is critical for researchers to consider RH and temperature in their studies on SARS-CoV-2 inactivation

5.3. Particle size distribution and suspending media

Particles may change size due to (i) changing RH, (ii) suspending medium composition, and (iii) cell aggregation. Although studies report the survivability of bacteria and viruses in different suspending media in air, few studies consider the interaction between these factors and ultraviolet radiation.

Increasing RH in turn increases particle sizes and bioaerosol survivability (Jordan et al., 2001; Ko et al., 2000; Riley and Kaufman, 1972). An increase in count median diameter (CMD) of Serratia marcescens was reported from 6% to 16% following an increase in RH, regardless of the suspending medium (Lai et al., 2004). Ko et al. (2000) reported an increase in CMD from 1.9 µm to 2.6 µm when increasing the RH from ∼30% to ∼90% for Serratia marcescens. Meanwhile, the UV resistance of both Serratia marcescens and BCG increased at higher RH (Ko et al., 2000).

The suspending medium composition influences the bioaerosol droplet size and its UVC susceptibility (Schaffer et al., 1976, Benbough, 1969, Benbough, 1971, Dubovi and Akers, 1970, Karim et al., 1985). Lai et al. (2004) studied the impact of suspending media on bioaerosol droplet size and formation, concluding that more organic matter in the medium results in lower UV susceptibility. Interestingly, the protective impact of the suspending medium and humidity shows a compounding effect (Lai et al., 2004; VanOsdell and Foarde, 2002). A suspending medium's proteinaceous composition can create a protective layer around the cells (Di Noto and Mecozzi, 1997). The inactivation of ssRNA viruses, hepatitis A virus, and norovirus surrogates by pulsed light required a higher dose when suspended in a medium containing both phosphate-buffered saline (PBS) and fetal bovine serum (5% v/v; 0.2% v/v protein) than in PBS alone (Jean et al., 2011).

In general, aggregates have greater survival than single microorganisms, where aggregation increases resistance to UV radiation (Haddrell and Thomas, 2017). As the droplets dry out, non-volatile components of the suspending medium may adhere to cells, increasing cell aggregation (Lai et al., 2004; Haddrell and Thomas, 2017). Some surrogate Gram-negative bacteria (Pantoea agglomerans, Serratia marcescens, Escherichia coli and Xanthomonas arboricola) had higher survivability when in clusters (∼3.5 µm) compared to a single cell (∼1 µm). Among them, E. coli showed the greatest increase in stability (D₉₀= 287 min) when in cluster form compared to in a single-cell state (one-log reduction, i.e., D₉₀= 14 min) (Dybwad and Skogan, 2017). Schaffer et al. (1976) showed that the protein content of the suspending medium impacts airborne stability of the Influenza virus below 0.1 mg/ml at high (80%) and low (50%) RH (Schaffer et al., 1976). Fig. 9 shows a typical bacterium in a cluster and single cell format, representing the size changes in two forms.

It has been hypothesized that organic matter creates a protective layer around the cell(s) and causes less susceptibility to desiccation and other stressors. No reported mechanism exists to explain the interaction between bioaerosols like viruses (enveloped and non-enveloped), bacteria, fungi and various suspending media. There exists a knowledge gap on the mechanisms of the protective effect of humidity and organic/ non-organic compounds of bioaerosols; this is quite important for understanding disinfection of human-borne pathogens.Finally, it should be noted that the surrounding medium may absorb UVC radiation, affecting the pathogen's decay rate. This varies greatly with the composition of the surrounding medium, such that generalizations are not useful. Suffice to explain that UVC at 254 nm may penetrate 40 cm into distilled water before its intensity is reduced by 30%, while only 10 cm in sea water and 5 cm in a sucrose (10%) solution for the same intensity reductions (Snowball and Hornsey, 1988). The impacts are difficult to quantify due to the changing size of bioaerosol droplets also changing the concentration of salts, proteins and others, which is affected by the environmental RH. Further, it has been well-reported that clumps pathogens offer protection to the innermost individuals, reducing the apparent disinfection potential (Vitzilaiou et al., 2021). Hence, researchers and practitioners should consider these impacts, with a possible a safety-factor on disinfection dosage calculations to account for these effects. Future work could better understand the reduction in UV dosage with changing bioaerosol particle size.

5.4. Outlook

Literature describes the roles of operational and environmental factors on bioaerosol survivability. However, many studies report data for different operating conditions, viruses and applications. The field suffers from a lack of mechanistic studies to describe the impact of operational and environmental factors in combination with the UVC radiation intensity on bioaerosol survivability. As a result, it is challenging to compare these studies and even more challenging to use the diverse literature to create a model for predicting the disinfection potential of different ultraviolet applications.

The range of findings presented suggests the need for a standardized laboratory model with a repeatable and reliable methodology to explain the fundamental differences based on the characteristics of a target viral aerosol and larger scale bioaerosols. For example, a standard procedure to test and compare aerosolized viruses (enveloped vs. non-enveloped), bacteria, and fungi would be of value for developing recommendations on operating conditions for UV devices.

Thankfully, the plethora of studies shows a keen interest in the field by both academia and industry. With improved coordination, it seems likely that UVC air disinfection can be understood for a range of operating conditions and target microorganisms and will prove a valuable tool to control airborne disease transmission. The effort needs interdisciplinary skills of aerobiology to understand the survivability, engineering for appropriate setup design and photonics to unify the UVC measurement techniques under different conditions. Since this matter is of such great importance to all, it would be desirable if such coordination was spearheaded by an international authority.

6.1. UV-LEDs

Recently, UVC-LED lamps gained popularity for their small size, irradiance control, low operating voltage (Gerchman et al., 2020), fast warmup, long lifetime despite repeated on/off cycles (Song et al., 2016), and containing no hazardous mercury. The wavelength is selected by changing the semiconductor composition (Chen et al., 2017); aluminum nitride-LEDs emit at 210 nm, while aluminum gallium nitride-LEDs emit between 222 and 351 nm (Hirayama et al., 2015). This tunability presents opportunities to tailor for specific industry objectives. In a study using five pathogens (including Legionella pneumophila and Bacteriophage Qβ), UVC-LEDs showed the highest inactivation at 265 nm, but the most energy efficient disinfection was achieved at 280 nm. However, the study reported that LP mercury UV lamps had a higher energy efficiency (Rattanakul and Oguma, 2018). Despite low energy consumption during operation. However, the price per watt for UVC-LEDs can be 1000 times higher than LP mercury lamps, and LEDs struggle to convert electrical energy to germicidal output when compared with conventional UVC lamps (Hadi et al., 2020; Sabino et al., 2020). Notwithstanding the many advances in UVC-LED technologies, further developments in efficiency and cost are needed.

6.2. Industrial installation

Early UVC air-disinfection comprised a UV lamp in an upwards-facing aluminum pan hanging below the ceiling (Fig. 1A) (Wells et al., 1942). Today, there are two installation strategies: 1) upper-room systems to radiate the upper airspace of a room and 2) systems installed in a building's air-handling unit to disinfect passing air (ASHRAE, 2019). The typical in-duct UVC provides ∼0.02 W/cfm, requiring an irradiance around 1000 to 10,000 μW/cm² (ASHRAE, 2019) for large buildings, to target an ∼85% single-pass inactivation (Bahnfleth, 2020). Upper-room UVC comprises many smaller lamps of 30–50 μW/cm² fluence, for a dose around 1.87 W/m² floor area or 6 W/m³ upper zone volume (NIOSH, 2009), and typically resulting in a higher total power requirement than comparable in-duct UVC designs (Bahnfleth, 2020).

In many cases, UVC is combined with conventional particulate filtration and proper dilution ventilation to optimize cost and energy (ASHRAE, 2019; Ko et al., 1998). Each well-mixed air exchange inactivates about 63% of pathogens, producing a logarithmic decay of airborne pathogens (Brickner et al., 2003). HEPA air filters can provide 99.97% removal of airborne particles 0.3 μm in diameter (Miller-Leiden et al., 1996).

In buildings with sufficient air exchange and appropriately sized HVAC systems, in-duct UVC may provide the energy and cost-efficient disinfection option. However, in older buildings with insufficient air exchange or low vertical air velocities, upper-room UVC may offer a superior design.

7.1. Safety concerns

UVC radiation can cause erythema of the skin and inflammation of the cornea (eye) (Parrish, 1979). Chromophores mainly absorb UV radiation at 254 nm in the outermost layer of dead skin, and only 5% of 254 nm ultraviolet radiation penetrates to the top layer of viable cells. Unlike the protective outer skin layer, the human cornea (eye) is unprotected and more susceptible to injury from ultraviolet radiation (Nardell et al., 2008). Inflammation of the cornea (photokeratitis) usually precedes more harmful inflammation of the conjunctiva (the ocular lining, a condition known as photo keratoconjunctivitis) (ASHRAE, 2016; Talbot et al., 2002).

Despite these rare injuries, UVC disinfection would seem a safe process. Proper UVC system installations can reduce human exposure. In the case of accidental exposure, UVC radiation poses only a small risk to the skin, and while cornea injuries can be painful, the effects were reported to be temporary (ASHRAE, 2016).

In 1972, the US Centre for Disease Control and Prevention (CDC) and the National Institute for Occupational Safety and Health (NIOSH) released a recommended exposure limit (REL) for limiting worker exposure to UV radiation which varies based on the exposure duration (Table S5 (National Institute for Occupational Safety and Health, 1972; Chapter 11-Ultraviolet Radiation, 2019).

Another consideration with UVC is organic materials degradation, such as electrical insulation, sealants, filter media, gaskets and pipe insulation and furnishing and finishes (Kauffman and Wolf, 2012). As a result, UV-resistant materials should be used, especially in the upper-room portion of UVC-irradiated facilities, although this may challenge older buildings' retrofits. While adequately installed UVC systems are unlikely to create unintended exposure, they can further protect occupants by enclosing the radiation source and targeting the bioaerosols.

7.2. Far-UVC as a safe alternative

The shorter wavelengths in UVC range are the so-called 'Far-UVC' spectrum (200–230 nm). Far-UVC is equally or more damaging to pathogens as compared to 254 nm (Riley and Kaufman, 1972), but with far less hazard to humans (Buonanno et al., 2017; Garciá De Abajo et al., 2020). Far-UVC has a much shorter penetration depth than longer wavelength UVC radiation (Buonanno et al., 2013; Goldfarb and Saidel, 1951). It is absorbed by the peptide bond in amino acid residues (Rosenheck and Doty, 1961) and so is blocked by the stratum corneum layer of skin (i.e., outermost skin layer) and, importantly, also blocked by the cornea of the eye (Sabino et al., 2020). Therefore, very little irradiation reaches the superficial layer of the epidermis and even less reaches to the deeper basal layer, where it may alter DNA and cause skin cancer (Cadet, 2020; Barnard et al., 2020). There are not many studies on the impact of far-UVC on the skin and eye tissues due to the low availability of commercialized far-UVC devices. Kaidzu et al. (2021) assessed the photokeratitis threshold in a rat and reported 5000 and 15,000 mJ/cm² for UV radiation at 207 and 222 nm. This is well within current safety guidelines.

Far-UVC is commonly generated through barrier discharge excimer lamps which emit quasi-monochromatic radiation (Sosnin et al., 2006). A recent krypton-chloride excimer (Kr-Cl*) lamp emits mostly at ∼222 nm (Kang et al., 2019). Far-UVC generation has also been demonstrated using solid-state emitters and conventional UV-LED construction, including thin layers of AlGaN to form the active region, photon-generating (Kneissl, 2016). Current far-UVC LED models suffer from short lifetimes and low power output (∼micro-watt range). However, similar challenges are being resolved in UVC-LEDs.

Using a fully coupled radiation and fluid dynamic model, the disinfection rate for a typically ventilated room was predicted to increase 50–85% when using the far-UVC radiation compared to just ventilation (Buchan et al., 2020). Exposure to only 2 mJ/cm² of 222 nm UVC inactivated > 95% of aerosolized H1N1 influenza, while only 1.7 and 1.2 mJ/cm² were needed for 99.9% inactivation of coronavirus 229E and OC43 in air, respectively (Buonanno et al., 2020). In fact, at the current REL (3 mJ/cm²/h), far-UVC could inactivate ∼90% of viruses in 8 min, 95% in 11 min, and 99.9% in 25 min (Buonanno et al., 2020). This shows great promise for application in high-use public areas such as hospitals, intensive care units and public transit.

While far-UVC presents an opportunity for the safe application of upper- and perhaps even lower-room UVGI systems, the technology is not broadly available on the market, and the cost is expected to be substantially higher than conventional LP-mercury lamps (Sabino et al., 2020). Some far-UVC lamps generate ozone which negatively impacts the respiratory, cardiovascular, and central nervous systems. Based on the WHO guideline, the mean exposure limit to ozone for 8 h is 0.2 mg/m³ (WHO, 2021). While there is no standard protocol to measure the produced ozone, it is the manufacturer's responsibility to ensure the safety and potential hazard.