Comparative performance and environmental evaluation of 222 and 254 nm UV lamps for in-duct air disinfection in building systems

Abstract

UV germicidal irradiance has gained significant popularity as a disinfection technology since the COVID-19 pandemic. Challenges remain in accurately characterizing the in-duct irradiance distribution of UVC lamps under varying air parameters, introducing uncertainty in exposure and efficiency estimations. This study experimentally characterizes irradiance of 222 and 254 nm UV lamps in a conduit using a 9-point average method. Irradiance measurements are recorded from five sides at each point, and a correction factor is proposed to account for spectrometer detection angle overlaps. Results indicate that the 254 nm lamp output increases by 18% with an air temperature rise from 25°C to 35°C but decreases by 21% as air velocity rises from 0.5 to 2 m/s. In contrast, the 222 nm lamp shows negligible sensitivity to changes in air temperature or velocity. Relative humidity variations (25%–60%) did not significantly affect either lamp’s output. In addition, the 222 nm lamp generates minimal ozone and total volatile organic compounds (TVOCs) during operation in stagnant air. The particle tracking simulation results revealed higher velocity fluctuations in proximity to the installed 222 nm lamp than the 254 nm lamp due to the larger lamp geometries. Experimental UVGI efficacy tests targeting airborne E. coli inactivation, combined with comprehensive life cycle analysis, demonstrated that the 254 nm lamp offers superior cost-effectiveness and sustainability for continuous in-duct air disinfection. These findings provide practical insights for optimizing UVC systems for air disinfection applications.

Keywords

UVGI in-duct air disinfection 254 nm UVC 222 nm Far-UVC irradiance life cycle analysis

Introduction

The COVID-19 pandemic has underscored the critical need to focus on preventing the transmission of airborne pathogens within indoor environments, where individuals often spend prolonged periods. While conventional approaches such as increasing ventilation rates and utilizing high-efficiency filters have shown some effectiveness, they present challenges in terms of energy consumption and economic viability (Azimi and Stephens, 2013; Huang et al., 2008). Historically employed for water disinfection, Ultraviolet Germicidal Irradiation (UVGI) systems offer an alternative solution to the limitations of traditional methods for mitigating airborne pathogens. UV light, which covers the electromagnetic spectrum of 100–400 nm, is classified as UVA (320–400 nm), UVB (280–320 nm), and UVC (200–280 nm) for disinfection purposes (Guerrero-Beltrán and Barbosa-Cánovas, 2004; Lu et al., 2021). Among them, UVC has the highest genetic damage capability to microorganisms such as viruses, bacteria, fungi, and spores (Yaun et al., 2004). Upon interaction with UVC photons, DNA molecules undergo photochemical reactions, forming pyrimidine dimers within the DNA strands. This process inhibits DNA replication and transcription, ultimately resulting in cell death of microorganisms (Franz et al., 2009).

Based on their application, the UVGI air disinfection system can be characterized as an upper-room system, a Whole-room system, and an In-duct system (Nunayon et al., 2020b). This study specifically focuses on UVGI in-duct air disinfection applications, the performance of which primarily depends on the type of UV wavelength, UV irradiation distribution, UV exposure time to bioaerosols, and susceptibility of microorganisms to UV wavelength (Cattai et al., 2023; Luo and Zhong, 2021). UV exposure time is mainly dependent on the air velocity, while the susceptibility of microorganisms mostly relies on the inherent biological properties of microorganisms, which cannot be changed. Sometimes, air parameters like relative humidity (RH) can also affect the susceptibility of the microorganism. However, the most important parameter from an engineering perspective is UV irradiation distribution, coupled with exposure time. While the UV exposure time can be easily calculated based on traditional flow field calculations for uniform fluid flow or using the computational fluid dynamics for non-uniform flow, characterizing UV irradiance poses a lot of uncertainty and challenges. The type of UV lamp used, UV lamp arrangement, duct dimensions, and duct material can potentially affect the UV irradiation distribution trend in addition to air operating parameters like air temperature and air velocity (Cattai et al., 2023; Luo and Zhong, 2021). Therefore, understanding and accurate mapping of the actual irradiation field present inside the duct is essential for avoiding the poor in-duct UVGI system design.

Characterizing detailed irradiation trends often involves simulations using trigonometric functions, with the lamp assumed as a line or point source, or employing a view factor approach, both commonly utilized by researchers (Luo and Zhong, 2022). However, these methods necessitate complex engineering calculations and costly computational resources, which may not be feasible in practical applications. An alternative approach involves experimental irradiation measurement using instruments such as radiometers, spectrometers, photometers, or chemical actinometers at designated locations. Although disinfection within the duct or chamber is a volumetric phenomenon, most of the available sensors measure irradiance in a planner way. In the duct, where reflections from the side walls are significantly high, it becomes necessary to account the irradiation from all six sides (Luo and Zhong, 2021). Some studies only measured irradiations from single side leading to underestimation, while others simply added the irradiation from all six sides, without considering the angle detection limit of the sensor used, potentially resulting in overestimation of irradiation (Nunayon et al., 2020a; Wang et al., 2023; Zhang and Lai, 2022).

Traditionally, the 254 nm low-pressure mercury lamp (LP) is widely used and studied. The irradiation of a 254 nm lamp has been observed to vary with the operating conditions, primarily with air temperature and velocity (Lau et al., 2009; Luo and Zhong, 2021). While one study noted a change in the output of a 254 nm lamp with different RH levels (Zhang et al., 2020), another study found no such correlation (Lau et al., 2009; Luo and Zhong, 2021). Moreover, the mercury used inside the LP lamp is hazardous to human health in case the lamp breaks, and it may not be environmentally friendly since it requires special attention to handle the mercury at the end of the lamp’s life. Since the UVC light at 254 nm degrades human health (skin and eye damage) upon exposure to it, a safer alternative, the use of 222 nm far UVC-filtered excimer lamp, is currently gaining popularity in the UVC lamp market. Studies have found that 222 nm does not penetrate inside the human skin, making it safe for exposure (Buonanno et al., 2017; Yamano et al., 2020). However, some studies showed concerns regarding the ozone level and total volatile organic compounds (TVOC) generated by these excimer lamps (Graeffe et al., 2023). Moreover, detailed life cycle assessments comparing the cost-effectiveness, energy efficiency, and sustainability of both lamps for disinfecting specific microorganisms are not widely available. Therefore, E. coli, the most commonly found bacteria in various environments, has been selected in this study to compare the disinfection performance of both lamps in the duct.

Overall, this paper aims to characterize and compare 222 and 254 nm UVC lamps by (1) experimentally assessing irradiance distribution using a novel correction factor, (2) analyzing the impact of air temperature, velocity, and RH on the lamp output, and their operational impact on ozone and TVOC generation levels, (3) evaluating exposure time using computational fluid dynamics (CFD) and calculating the UV dose for each lamp, and (4) compare cost-effectiveness, energy efficiency and sustainability for E. coli air disinfection in a duct.

Materials and methods

Pilot scale in-duct UVGI system

A pilot-scale stainless steel duct system with cross-sectional dimensions of 12.7 cm × 12.7 cm was utilized for in-duct UVGI disinfection experiments. The system comprises essential components like fans, UV lamps, mixing baffle, filters, injection, and sampling ports, as depicted in Figure 1. A feedback control system (Opto 22) was installed to regulate air parameters (velocity, temperature, and RH) within the duct system. Additionally, for uniform injection and sampling of aerosols, cross-shaped probes were used, with each probe consisting of four branches and five sampling holes per branch. Pre-qualification tests, including leakage, velocity, and aerosol uniformity, were conducted to align with the ASHRAE Standard for testing UVC lights in ducts (ASHRAE 185.1-2020), the comprehensive results of which are provided previously (Wu et al., 2022).

Figure 1.

Pilot scale HVAC duct setup.

UVC lamps: Installation and characterization

The UVC low-pressure mercury lamp (OSRAM GCF5DS, 5 W × 1) and far-UVC filtered excimer lamp (Cure-UV 510100-20, 20 W × 2) with peak wavelengths of 254 and 222 nm, respectively, were installed as depicted in Figure 2. Both UVC lamps are placed in the same x-y plane, which distributes the total UV exposure length (116.3 cm) in the duct into two sections: upstream (47.5 cm) and downstream (68.8 cm). A single 254 nm lamp (Ø12 mm) is installed at the center of the cross-sectional plane, while two 222 nm cylindrical lamps (Ø35 mm) are installed symmetrically as the output efficiency of 222 nm lamps is very low.

Figure 2.

Installation of lamps; (a) 254 nm, (b) 222 nm. All dimensions are in mm.

Both lamps’ UVC spectra (Supplemental Figure S1) and their irradiance (mW/cm²) were characterized using a recently calibrated spectrometer (ILT 960; International Light Technologies, Peabody, MA, USA) with input optics RAA4. The device had ±15% and ±10% uncertainty in measuring irradiance of 222 and 254 nm lamps, respectively. Irradiance measurements were conducted on five vertical planes (P1–P5) that are parallel to the lamps and perpendicular to the directed airflow as depicted in Figure 3(a). The irradiance measurements were taken downstream the installed lamps, and the principle of symmetricity was applied to predict the irradiance of the upstream side. Each vertical plane was divided into 3 × 3 grids, as depicted in Figure 3(b), and a 9-point average method was used to characterize the irradiance of each plane. While measuring irradiance from all six sides is essential due to potential reflections from the duct walls, the preliminary test revealed that irradiance contribution from the side facing opposite (back side) of the lamp is negligible. Thus, during irradiance measurements, the light detector was positioned at the center of each grid, and irradiance from only five sides (front, left, right, top, and bottom) was measured in this study. Lamp start-up time is also observed for both lamps using the same spectrometer device until the output of each lamp is stabilized.

Figure 3.

(a) Irradiation measurement planes at downstream; (b) Dimensions of measured 3 × 3 grids for each plane. All dimensions are in mm.

Initially, both the lamps underwent a burnt-in period of a minimum of 100 hours before characterizing them under static conditions at room temperature (20°C ± 1°C). Subsequently, air velocity was varied from 0.5 to 2 m/s (T = 25°C and RH = 40%) and changes in irradiance were measured from all the sides at downstream planes for both lamps. Similarly, air temperature is varied from 25°C to 35°C (v = 0.5 m/s, RH = 40%) and relative humidity from 25% to 60% (v = 0.5 m/s, T = 25°C) to evaluate their effects on lamp output. Irradiance has been measured a minimum of thrice under all the mentioned air parameter conditions to ensure the reliability and accuracy of the data over the average fluctuations in the supplied voltage and current. It should be noted whenever the velocity is mentioned in this study, it represents the flow conditions in the duct, which results in an average uniform velocity downstream of the lamps and closer to the region of sampling.

Finally, both lamps are tested for the generation of ozone and TVOCs. Since the air velocity is likely to dilute their concentration and render them non-detectable, static room air conditions (T = 20°C, RH = 60%) were used to measure them. Initially, background ozone and TVOC levels were recorded. Then, the lamps were turned on for 30 minutes and average ozone and TVOC levels were continuously monitored for a minimum of three repetitive sessions. A daily calibrated photoionization detector (PID, HYPERESENSE 1002, HENGAN, China) with a resolution of 5 ppb, response time of less than 5 seconds, and error of ±10% was used to record TVOC levels. The ozone concentration was measured using an ozone monitor (Teledyne API’s Model 465L) with a measuring range of 1–1000 ppm, response time less than 30 seconds, sampling flow rate at 0.8 LPM, and error of ±1%.

Irradiance calculation and correction factor

In enclosed spaces like air ducts, the reflections of light from the walls are significant compared to an open environment (Luo and Zhong, 2022). The majority of sensors available in the market measure irradiance in a 2-D plane, suggesting that they do not measure irradiation from all 360° angles, thus necessitating irradiance measurement from multiple sides. However, it is crucial to consider the angle detection limit of the sensors used. Any sensor that measures properties of light follows the ideal cosine response, indicating the need for manufacturers of spectrometers to employ a detector that automatically applies the cosine correction based on the standard calibration to give a reading closer to the actual value of the irradiance. The deviation of the actual sensor’s response in relation to the standard cosine curve decides the detection angle range for each light-measuring sensor. For instance, the actual cosine response curve of ILT960 provided by the manufacturer depicts more than ±10% fluctuations from the ideal cosine values beyond ±60° and quickly approaches zero, making the measured readings negligible and unreliable beyond those angles. Therefore, the total angle detection limit of the sensor employed in this study is recommended to be 120° (−60° to +60°). Similarly, some of the sensors available in the market have an angle detection limit of up to 160°, which may differ the irradiance readings at a point, especially in a region closer to the lamp, where irradiance levels are high. In addition, when the irradiance is measured from two adjacent sides, each side’s total 120° detection limit results in a 30° overlap of irradiance, as illustrated in Figure 4, potentially leading to an overestimation of irradiance.

Figure 4.

Illustration of irradiance measurement and its overlap with adjacent sides.

To address the irradiation overestimation issues due to the measurement from multiple sides and unify the irradiance readings obtained from two spectrometer devices having different detection angles limit range, the correction factor $(C)$ is introduced in this study. Although the actual irradiance will be non-uniform over the whole 120° measuring range, for the simplicity of calculating the correction factor, it is assumed that the total measured irradiance from a side will be uniformly distributed over the whole range of measured angles. Thereby, the correction factor introduced in this study aims to reduce the measured irradiance from each side in proportion to the overlap angle values as expressed in equation (1).

C_{i} = (1 - \frac{\sum_{j = 1}^{2} \emptyset_{j}}{(\sum_{j = 1}^{2} j) * 2 * θ})

(1)

where, $C_{i}$ = Correction factor for irradiance for side, $i$ ; $\emptyset_{j}$ = Total overlapped angle in measured direction, $j$ and $θ$ = $Total angle detection limit$ = 120°.

Finally, the irradiance at a point can be calculated from the following equation.

I = C_{f} I_{f} + C_{l} I_{l} + C_{r} I_{r} + C_{t} I_{t} + C_{b} I_{b}

(2)

where, $I$ = Total average irradiance at a specific point (mW/cm²); $I_{i}$ = Irradiance (mW/cm²) measured from side, i; $f$ = front side; l = left side; r = right side; t = top side; and b = bottom side; $C_{f} = 0.75$ ; $C_{l}$ = $C_{r}$ = $C_{t} = C_{b}$ = 0.8125.

Based on the approach mentioned above, the resulting measured UV irradiance with correction factor at each measured point for 254 and 222 nm lamps are mentioned in Supplemental Tables S1 and S2, which was further analyzed to obtain the irradiance trend and total UV dose in the duct.

UV exposure time and UV dose

The inactivation efficiency of the UVGI system primarily relies on the UV dose: the product of UV irradiance and its exposure time. While the UVC irradiance was characterized for both lamps experimentally in this study, the non-uniform velocity distribution within the duct caused by lamp geometries and flow disturbances necessitated a computational fluid dynamics (CFD) approach to estimate the UV exposure time for airborne particles.

This CFD-based methodology is consistent with the approach employed by Luo and Zhong (2022) who validated their simulation results with experimental measurements for the same duct setup. Since the present study utilizes an identical duct configuration and builds upon the same numerical modeling framework, the CFD results here are considered valid without requiring additional experimental validation. However, key methodological enhancements were incorporated. Compared to Luo and Zhong (2022) where mesh refinement was applied only near the UV lamp region, the current study additionally refined the mesh near the inlet region, where significant velocity gradients and turbulence are expected due to abrupt flow development. This ensures better capture of entrance effects and overall particle dynamics throughout the duct. Furthermore, the UV dose estimation in the current study was improved by integrating experimentally measured irradiance values with CFD derived exposure times. In contrast, Luo and Zhong (2022) relied solely on computational irradiance fields to calculate dose received by particles in each voxel. This hybrid approach enhances the reliability of UV dose predictions, particularly under varied lamp configurations where accurate irradiance modeling becomes challenging.

To accurately resolve flow dynamics and particle trajectories within the UVGI duct system, a high-fidelity unstructured tetrahedral mesh was generated using ANSYS ICEM 19.2 for 254 nm lamp configurations. The overall duct geometry (0.127 m × 0.127 m × 1.16 m) was discretized with localized mesh refinement applied in regions with expected velocity gradients and turbulence, particularly near the duct inlet and around the UV lamp surfaces, to capture flow development and lamp-induced disturbances more accurately. Three mesh sizes were generated to assess grid sensitivity: 390,663 elements with a characteristic element size of 0.04 m, 694,861 elements at 0.02 m, and 2,087,682 elements at 0.01 m. Mesh quality was monitored using several standard metrics to ensure numerical stability. The average element quality was 0.835 ± 0.095 while the orthogonal quality ranged from 0.202 to 0.99 (0.766 ± 0.118). Skewness varied between 0.00006 and 0.797, averaging 0.232 ± 0.12 and aspect ratio values spanned from 1.15 to 10, with a mean of 1.86 ± 0.46. These values are well within accepted limits for tetrahedral meshes in duct flow simulations, indicating good mesh regularity and ensuring reliable CFD performance.

Average velocities were recorded at three cross-sectional planes located at 0.2 m (inlet), 0.5 m (lamp region), and 1.16 m (outlet) from the duct entrance to evaluate numerical grid convergence. The Grid Convergence Index (GCI) and observed order of convergence (p) were calculated using the Richardson extrapolation method. As shown in Supplemental Table S3, all GCI values remained within 5% with safety factor (Fs = 1.25), and the refinement was in the asymptotic range of convergence (≈1), indicating satisfactory grid independence. Based on this analysis, the medium-resolution mesh (694,861 elements, 0.02 m size) was selected for all simulations to maintain an optimal balance between computational efficiency and solution accuracy. Similarly, for the 222 nm lamp, the same grid resolution (0.02 m size) was used for mesh generation, and the optimal numerical grids for both lamps are shown in Figure 5.

Figure 5.

Duct geometry with non-structured mesh and UVC lamps: (a) 222 nm lamp (755 k mesh elements); (b) 254 nm lamp (694 k mesh elements).

Following mesh convergence, the airflow field within the UVGI duct was resolved by numerically solving the continuity, momentum, and turbulence equations (see Supplementary Material: Appendix A). Once the solution achieved convergence, the steady-state velocity field was extracted for further analysis. The airflow was assumed to be isothermal and turbulent, modeled using the realizable k–ε turbulence model with standard wall functions. Subsequently, Lagrangian particle tracking simulations were conducted by integrating the force balance equations over the computed velocity field to determine the trajectories of airborne E. coli bioaerosols. These aerosols were modeled as liquid droplets with a diameter distribution ranging from 0.65 to 2.1 μm and were injected at the duct inlet. A minimum of 11,250 particles were introduced in each simulation, following the methodology of Luo and Zhong (2022). A steady particle tracking approach was employed, assuming one-way coupling wherein the particles were treated as massless and did not influence the airflow.

The boundary conditions applied in both the airflow and particle tracking simulations are summarized in Supplemental Table S4. The duct inlet was prescribed with a uniform velocity corresponding to the flow rate of 29.05 ± 0.4 m³/hour, while the outlet was defined as an outflow boundary. All duct and UV lamp surfaces were treated as stationary no-slip walls. For particle tracking, the inlet and outlet were set to “escape,” allowing particles to enter and exit the domain, while all internal solid surfaces, including duct walls and UV lamps, were designated as “trap” boundaries, such that particle collisions with these surfaces resulted in deposition.

The particle trajectory data obtained from CFD were post-processed in MATLAB R2022b to extract average particle velocities between adjacent cross-sectional planes spaced 1 cm apart. Particles exhibiting negative Z-direction velocities (backflow) were excluded. The average velocity between each plane pair was used to calculate the exposure time (sec), which was multiplied by the average UV irradiance in the same region to compute the local UV dose. This dose was then integrated throughout the duct across all planes to determine the total UV dose received by airborne particles, which was subsequently used to estimate the UV inactivation efficiency.

UVGI in-duct air disinfection of E. coli

The Escherichia coli (E. coli C3000, ATCC 15597) preparation and sampling procedure is followed those outlined by Luo and Zhong (2022). In brief, 2% lysogeny broth solution (pH = 7.0) was prepared to add the test organism and the suspension culture was then inoculated in a constant temperature shaking incubator (Corning 6750) at 37°C and 170 rpm for 16–18 hours. Thereafter, the culture was centrifuged at 1750 relative centrifugal force for 15 minutes and washed three times with 10% phosphate-buffered saline (PBS) solution to harvest E. coli stock. The final pellet was resuspended in 10 mL PBS.

For sampling, aerosols were initially generated from the 50 mL diluted liquid PBS suspension using a 1-jet nebulizer (Collision Nebulizer, BGI) with 138 kPa compressed air in an atomizer. These bioaerosols were then injected into the duct through an injection tube after passing through the dryer. Six-stage Andersen cascade impactor with Glass nutrient agar (NA) Petri dishes between each stage were employed to collect the samples at downstream with a vacuum pump (flow rate: 28.3 L/min). Following sample collection, the plates were sealed and incubated overnight at 37°C for almost 16–18 hours. Finally, the number of colonies formed on the plates was counted and converted to the corresponding corrected particle counts using the positive-hole conversion table (Macher, 1989).

The pilot HVAC system was maintained at the desired operating condition (T = 25°C, Q = 29.05 m³/hour, and RH = 40%), after which the compressed air was turned on for the nebulization process. After 20 minutes of stabilization, the UVC-off samples are collected at downstream. Then, the UVC light is turned on for 15 minutes, and again, samples downstream are collected in the six-stage impactor. The duct was disinfected with 75% ethanol before and after each experiment. Each test is replicated thrice for reliability.

Preliminary tests revealed that most of the cultivable bioaerosol particles are recovered in the size range of 0.65–2.1 μm, which refers to stages 5 and 6 of the cascade impactor. Therefore, only data from those two stages are utilized to evaluate the overall disinfection efficiency of both lamps. It can be formulated as shown in equation (3).

η_{UVGI} = (1 - \frac{\sum_{s = 5}^{s = 6} C_{s, UV - ON}}{\sum_{s = 5}^{s = 6} C_{s, UV - OFF}}) \times 100 %

(3)

where $η_{UVGI}$ = UVGI disinfection efficiency; $C_{s, UV - ON}$ = Average cultivable bioaerosol concentrations (CFU/m³) with UV light is on; $C_{s, UV - OFF}$ = Average cultivable bioaerosol concentrations (CFU/m³) with UVC light off; and s = impactor stage numbers.

Economic and sustainability analysis

Both 254 and 222 nm UVC technologies exhibit distinct advantages relevant to air disinfection applications. The 254 nm lamp is widely available and exhibits relatively higher DNA damage potential, making it effective for microbial inactivation. However, it requires continuous operation to maintain optimal performance and lamp longevity. In contrast, 222 nm excimer lamps offer a higher protein-damaging capability and have been demonstrated to be safe for human exposure due to their limited skin penetration, though they generally operate with lower output efficiency and higher initial costs (Lu et al., 2024). In the context of sustainable HVAC design, selecting a cost-effective and environmentally responsible UVC solution is critical. To assist HVAC engineers in making informed decisions, a comparative analysis of the economic, energy, and environmental implications of both UVC lamp types was conducted. The analysis targeted E. coli inactivation to achieve 99.99% disinfection efficiency within an in-duct UVGI system.

There are two main costs associated with UVC lamps’ life spans: Initial and operating costs. This study does not consider the cost associated with recycling lamps at the end of their life. Each cost is further simplified for comparison per unit output irradiance. The initial cost can be given as follows.

Initial cost = Lamp buying cost + Cost of accessories

(4)

Lamp buying cost depends on the manufacturer, type of UVC lamp, rating of UVC lamps, requirement of any fixtures to install the lamp, and other associated costs. Here, it is assumed that there is no additional cost other than buying the UVC lamp. The operational cost is based on annual energy consumption, which is a function of rated power and operational lifetime. Normally, when the output irradiance of a UV lamp declines to 70% of the initial value, the lamp is considered to have failed. Lamp manufacturers refer to this as L70 lifetime hours, which is usually 8000 hours for 254 nm (Lu et al., 2024) and 9000–10,000 hours for 222 nm lamp (Care222). For simplification and alignment with manufacturer recommendations, both lamp types are considered to operate continuously for 1 year (8760 hours) in this assessment.

Annual Electricity Consumption, E (kWh) = Lifetime (L 70) hours * Rated Wattage (kW)

(5)

Annual Operational Cost = E (kWh) * Price of Electricity (CAD / kWh)

(6)

Annual eC O_{2} generation = E (kWh) * Electricity consumption intensity ({eCO}_{2} / kWH)

(7)

The total cost can be calculated as the sum of the initial and annual operational costs. Since both the price of electricity and the equivalent CO₂ emissions (eCO₂) generated per kilowatt-hour vary by region, a comparative analysis was conducted across five major Canadian provinces, including Alberta (AB), British Columbia (BC), Ontario (ON), Quebec (QC), and Manitoba (MB) (Wu and Zhong, 2023). These provinces were selected based on their population size and energy grid diversity. This provincial-level breakdown provides HVAC designers and policymakers with a more regionally tailored perspective when selecting the most economically and environmentally sustainable UVGI lamp system for in-duct air disinfection.

Results and discussion

UV lamps: Irradiance distribution

The measured irradiance spectra for both lamps suggested that the mercury lamp had a peak wavelength of 254 nm, and the filtered excimer lamp had a peak wavelength of 222 nm, as illustrated in Supplemental Figure S1. The irradiance of both lamps was exclusively measured in the duct for the region downstream of the lamps, and the irradiation of the upstream region was derived from the application of symmetricity. The average total irradiation obtained from equation (2) at each point was used to characterize the irradiation field of UV lamps. Experimental verification in this study confirmed that irradiation varies linearly within the 3 × 3 measured grid on a vertical plane with a relative error of less than 10% throughout the duct. Therefore, linear interpolation was used to fully characterize the irradiation level within each measured plane, as seen in Figure 6 (a) and (b).

Figure 6.

Plane-wise measured irradiation distribution (a) 222 nm; (b) 254 nm.

It is evident that the plane closest to the lamp (P1) has the highest irradiation in both types of lamps, with a non-uniform distribution trend. However, as we moved far from the lamp (P4 and P5 in Figure 6), irradiation uniformity became more apparent, albeit with a drastic reduction in irradiation levels. In comparison to 254 nm, it was evident that the 222 nm lamp arrangement has more uniform irradiance, suggesting the installation of two lamps in one row for uniform irradiation distribution closer to the lamp. It’s apparent that even after using two 20 W lamps for 222 nm, the obtained irradiance is still roughly 10 times lower than what is achieved by only one 5 W of 254 nm lamp, showcasing lower efficiency of 222 nm lamp output and the need for more lamps to achieve the similar irradiance level as 254 nm. The irradiance fluctuations at the duct walls or corners may demonstrate the limitation of experimental data and underscore the need for more data points to be measured in those regions.

The linear average of irradiance from 9 measured points accurately represents the overall total average irradiance for the vertical plane. Thus, the 1-D plot illustrating irradiance in relation to the distance from the lamp surface is depicted for both lamps in Figure 7. Notably, reflections from the duct walls significantly contribute to the irradiance, resulting in values higher than those estimated by the inverse square law as claimed by lamp manufacturers, and aligning more closely with the inverse law for the duct applications. The experimental modeling of data is further validated with additionally measured vertical planes, demonstrating that the power law accurately predicts the irradiance trend throughout the duct and that the relative error between measured and predicted values is less than 15%.

Figure 7.

Average irradiation trend along the duct length (a) 222 nm; (b) 254 nm.

Furthermore, the irradiation values obtained through the new correction factor method have been compared with the CFD results obtained by Luo and Zhong (2022) for one 254 nm lamp in a same stainless-steel setup. As shown in Figure 7(b), the average irradiance trends obtained by the new correction method closely align with those simulation results, with a maximum relative error of 18%, affirming the reliability and efficacy of the correction ratio used in this study.

UV lamps: Effect of operating condition

The influence of air parameters on lamp output is illustrated in Figure 8 for both UV lamps. For 254 nm lamps, it is evident that an increase in temperature from 25°C to 35°C results in a significant 18% rise in lamp output. Nonetheless, the rate of increase diminishes when the temperature is raised from 30°C to 35°C, suggesting that beyond 35°C, 254 nm lamps may reach an optimal temperature for maximum output, as confirmed by Lau et al. (2009)

Figure 8.

Effect of air parameters (a) varying temperature; (b) varying velocity.

Conversely, velocity exhibits an opposite trend: higher duct velocities correspond to lower lamp output. This phenomenon arises from the wind chill effect generated by higher velocities on the lamp surface temperature, reducing the temperature required for the mercury vaporization process. Consequently, as velocity increases, so does the wind-chill effect, leading to a decrement in lamp surface temperature and output (Lau et al., 2009; Luo and Zhong, 2021). This explains the significant drop in mercury lamp output to approximately 80% when the velocity for the 254 nm lamp increases from 0.5 to 2 m/s at a constant air temperature of 25°C. RH levels were not found to affect the output of any lamp.

The 222 nm lamp exhibited no sensitivity to air parameters, attributable to its operational principle. Upon receiving electricity, the excimer lamp promptly forms excited dimers, transitioning spontaneously from an excited state to a ground state and thus releasing the far-UVC photons in the air, which has nothing to do with air parameters. While the 254 nm lamp took approximately 4.5 minutes to stabilize its output, the 222 nm lamp initiated within 3 seconds. This variance stems from the differing operational principles of the lamps. The presence of liquid mercury in the 254 nm lamp necessitates vaporization to generate UV light, requiring a warm-up period to reach the vaporization temperature post-startup. Conversely, the 222 nm lamp rapidly forms dimers upon electricity supply, requiring only a few seconds to emit UV light. Regarding ozone generation, the 254 nm lamps did not produce any ozone within a 30-minute sampling period, nor did they generate any levels of TVOCs. However, the 222 nm far-UVC lamps were found to generate approximately 2.8 ± 1.1 parts per billion (ppb) of ozone over a 30-minute average sampling period due to its operation. Although this level of ozone is considered negligible, it implies that it’s concentration might elevate over time in an enclosed room with a low air change per hour, suggesting careful operational considerations. The fluctuation in TVOC levels, ranging from 1 to 10 ppb for 222 nm lamps, demonstrates their insignificance in by-product generation.

UV exposure time and UV dose

Based on the CFD approach employed in this study, the particle path lines, and air velocity profiles at the middle of duct viewed from the top plane are shown in Figure 9 for both 254 and 222 nm lamps.

Figure 9.

Particle path lines and velocity distribution: (a) 254 nm lamp; (b) 222 nm lamp.

The velocity distribution near the inlet region is highly non-uniform, and velocity disturbances caused by the lamps being significantly lower in comparison to those in the inlet section. In addition, the path line diagram suggests the potential for backflow in the inlet section due to high velocity gradients. Therefore, it is essential to exclude velocities of backflow particles from this steady-state simulated results, when calculating the average velocity between the two planes for obtaining an exposure time. The plot of average tracked particle velocities, shown in Figure 10, indicates that velocity uniformity is achieved only in the downstream region of the lamps (duct length >0.55 m). Notably, the 222 nm lamp shows greater velocity fluctuations compared to the 254 nm lamp, attributed to its larger geometry and the installation of multiple lamps. The UV dose calculations along the duct length, depicted in Figure 11, reveal very high dose values in close proximity to the lamp, as expected. The 254 nm achieved much higher dosage compared to the 222 nm, attributed to its higher irradiance output. This suggests that the majority of disinfection occurs near the lamp. Consequently, efforts to improve UV dose should primarily focus on optimizing conditions in the regions closest to the lamps.

Figure 10.

Effect of lamp geometry on velocity.

Figure 11.

UV dose distribution for both lamps.

E. coli UVGI disinfection efficiency

The airborne E. coli inactivation efficiency was determined at air operating parameters of 29.05 ± 0.4 m³/hour, 25° ± 0.3°C, and 40 ± 3% RH for both 254 and 222 nm lamps. The average total irradiance obtained was 1.795 mW/cm² for 254 nm and 0.15 mW/cm² for 222 nm, respectively (Figure 7), throughout the duct under a static environment. However, since the output of 254 nm changes with air parameters (Figure 8), the average irradiance at 0.5 m/s, 25°C, and 40% RH should be corrected, as mentioned by other work (Luo and Zhong 2022), which is almost 94% of the irradiance measured at static conditions. Therefore, the corrected value of average irradiance for 254 nm lamp at the experimental air parameters (0.5 m/s, 25°C, and 40% RH) was 1.687 mW/cm². Coupling these irradiance values with the exposure time results in the total UV dose values (Figure 11) of 0.207 mJ/cm² for 222 nm lamp and 2.568 mJ/cm² for 254 nm lamp throughout the duct.

The average disinfection efficiency achieved for the 254 nm was 99.97% ± 0.03% (Figure 12), which is in close agreement with Luo and Zhong (2022), where similar average UVC lamp set-up and air parameters were used for E. coli inactivation. For the 222 nm lamps, 77.22% ± 4.58% average inactivation efficiency was obtained (Figure 12). These results are compared with Zhang and Lai (2022), where they have obtained a UV rate constant of 4.9 cm²/mJ for far UVC lamps operated in single pass in-duct application. Based on that, to achieve 77.22% inactivation efficiency with 222 nm for E. coli, they required UV dose was 0.3 mJ/cm², which is 30% more than the UV dose applied in this study. However, their setup did not include a drier, and the suspension media used was deionized water. Moreover, their average RH level was 53% during the experiments, and their irradiance was not corrected after measuring it from all six sides. These factors may lead to significant differences in the value of the UV dose requirement for achieving the same efficiency. Therefore, future experiments are targeted to find the UV rate constant for both lamps, enabling us to compare both technologies under the same UV dose, although their output irradiance is different.

Figure 12.

E. Coli inactivation efficiency for 222 and 254 nm lamps.

Comparative life cycle analysis of 254 nm versus 222 nm

A life cycle analysis has been performed to attain the target of 4-log reduction (99.99% inactivation efficiency) for in-duct E. coli single-pass air disinfection at 0.5 m/s and 25°C. This study uses the UV rate constant of E. coli from similar most recent studies for the life cycle analysis. While the Luo and Zhong (2022) reported the UV rate constant as 5.245 cm²/mJ (40% RH) for 254 nm lamp with the same operating conditions, Zhang and Lai (2022) reported 4.9 cm²/mJ (53% RH) UV rate constant for 222 nm lamps for duct applications. Based on these reported values, the required dose to achieve 99.99% efficiency is 1.88 and 1.76 mJ/cm² for 222 and 254 nm, respectively. Analyzing the UV dose and average irradiance mentioned in the previous sections, the average exposure time of 1.5 seconds can be approximated, and UV irradiance of 1.16 and 1.25 mW/cm² is required for 254 and 222 nm respectively to achieve the 4-log reduction target. Table 1 demonstrates the initial cost and kWh required to achieve this target irradiance levels for both lamps. The energy-efficiency and cost-effectiveness of 254 nm is eye-opening since the buying cost of 222 nm is 1000 times more and consumes almost 75 times the electricity than the 254 nm lamp to achieve the same disinfection target. This suggests that 222 nm excimer lamps need to be developed further for their higher lamp output and lower costs for disinfection applications.

Table 1.

Initial cost and kWh requirement for both lamps.

Lamp type	222 nm	254 nm
Initial cost, CAD (Supplemental Tables S3 and S4)	900	11
Annual kWh	289	44
Output, mW/cm²	0.15	1.68
Initial cost (CAD) to achieve target irradiance	7500	7.5
Annual kWh to achieve target irradiance	2409	31

To further examine the environmental and economic impacts across Canada, province-specific analyses were conducted (Supplemental Table S7) based on local electricity prices and grid emission intensities reported by Wu and Zhong (2023). Figure 13 illustrates the total cost and equivalent CO₂ (eCO₂) emissions associated with UV lamp use in five major provinces: Alberta (AB), British Columbia (BC), Ontario (ON), Quebec (QC), and Manitoba (MB). The comparison reveals that the 222 nm excimer lamp consistently incurs substantially higher total costs and carbon emissions than the 254 nm mercury lamp across all regions. Notably, although the 222 nm lamp has significantly higher energy consumption, its environmental impact is mitigated in provinces with cleaner electricity grids, such as BC, QC, and MB. For example, Alberta’s grid emits approximately 640 g eCO₂/kWh, while Manitoba’s grid emits only 1.2 g eCO₂/kWh. Consequently, the same 2409 kWh energy use by the 222 nm lamp results in 1541.76 kg eCO₂ emissions in Alberta, but only 2.89 kg in Manitoba. Interestingly, the emissions from 254 nm lamps in Alberta (19.36 kg) are still higher than those of 222 nm lamps in BC (18.79 kg), QC (4.6 kg), and MB (2.9 kg), despite 254 nm lamp’s greater energy efficiency. Furthermore, 254 nm lamps generate almost negligible emissions in Quebec (0.057 kg) and Manitoba (0.036 kg), demonstrating that in provinces with clean electricity, UVGI systems integrated with efficient lamp technologies can offer effective air disinfection with minimal environmental impact. These results underscore that the environmental feasibility of any UV-based disinfection approach must be considered in the context of regional energy sources. Additionally, the total cost differences across provinces were minimal, indicating that the cost is primarily driven by the initial capital investment for both lamps, while energy costs contribute relatively little due to price variability.

Figure 13.

Province-wise comparison for sustainability: (a) total cost, (b) total eCO₂ generated.

While this study presents a life cycle analysis (LCA) for continuous 24/7 in-duct air disinfection applications, it is important to acknowledge that the suitability of a UV lamp also depends on its operational context. The 222 nm far-UVC excimer lamp is inherently more compatible with applications involving frequent on/off cycles, as its lifespan is not significantly affected by switching frequency. In contrast, the 254 nm lamp is susceptible to early failure under such conditions. Therefore, the selection of UV technology must align with the intended application mode, continuous versus intermittent operation. Additionally, although the 254 nm lamp was found to be more cost-effective and environmentally sustainable for continuous in-duct disinfection in this study, certain limitations should be noted. The 254 nm lamp contains mercury, which may introduce additional environmental impact and disposal costs after its operational life, unlike the mercury free 222 nm excimer lamp, which does not require special post-use handling.

Moreover, the LCA calculations in this study do not account for variations in average UV dose that may arise from differences in lamp and duct configurations, including geometry, installation positions, and lamp arrangements. Additionally, the analysis does not consider real-world discrepancies between the required UV output and the commercially available lamp wattages. Typically, the price per watt decreases with increasing lamp wattage, which may influence overall cost assessments. Similarly, the cost and complexity of additional components required to accommodate different lamp types (e.g. housing or mounting accessories) are not considered. Despite these considerations, the impact of such unaccounted factors may not be sufficient to alter the core conclusion of this analysis: for continuous air disinfection applications, the 254 nm technology remains the more economically viable and environmentally favorable solution under current market conditions.

Conclusions

A novel correction factor method was introduced to experimentally characterize the irradiance distribution of 222 and 254 nm lamps inside a duct. The irradiance was first evaluated under static conditions, and the resulting average corrected irradiance values were verified against simulation data from the prior study, ensuring the accuracy and reliability of the proposed approach. These findings highlight the critical need to consider spectrometer angular detection limits to avoid irradiance overestimation during experimental measurements.

The 254 nm mercury lamp exhibited notable sensitivity to temperature and air velocity, with its output decreasing at higher velocities due to the wind-chill effect on the lamp surface. In contrast, the 222 nm excimer lamp’s output remained stable across varying environmental conditions. Relative humidity had minimal impact on the irradiance of both lamps. The 222 nm excimer lamp also demonstrated operational advantages, including rapid startup, initiating UV emission within seconds, compared to the 254 nm mercury lamp, which requires several minutes to stabilize due to the mercury vaporization process. Additionally, the 222 nm lamp generated only negligible ozone concentrations during operation; however, the potential for ozone accumulation in poorly ventilated environments necessitates careful operational management.

The in-duct E. coli disinfection experiments demonstrated a markedly higher inactivation efficiency for the 254 nm lamp (99.97% ± 0.03%) compared to the 222 nm lamp (77.22% ± 4.8%), consistent with the observed higher UV dose delivered by the 254 nm system. Life cycle analysis revealed that, despite the 222 nm lamp’s potential safety advantages and suitability for intermittent operation, it incurs substantially higher initial costs and energy consumption than the 254 nm lamp for achieving equivalent disinfection targets under continuous operation. Province-specific LCA results further underscored that the carbon footprint of the 222 nm lamp can be mitigated in regions with cleaner electricity grids; however, the 254 nm lamp remains the more cost-effective and environmentally sustainable choice overall for continuous air disinfection applications.

Future research should focus on experimentally determining standardized UV rate constants for both lamp types to enable direct comparison at equivalent UV doses. Furthermore, expanding this framework to include emerging technologies such as UV LEDs will be essential for advancing efficient, safe, and sustainable air disinfection solutions in HVAC systems.

Supplemental Material

sj-docx-1-jen-10.1177_17442591251382166 – Supplemental material for Comparative performance and environmental evaluation of 222 and 254 nm UV lamps for in-duct air disinfection in building systems

Supplemental material, sj-docx-1-jen-10.1177_17442591251382166 for Comparative performance and environmental evaluation of 222 and 254 nm UV lamps for in-duct air disinfection in building systems by Jay Patel and Lexuan Zhong in Journal of Building Physics

Footnotes

ORCID iDs

Jay Patel

Lexuan Zhong

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) Alliance Program (ALLRP 560993–20). We sincerely appreciate Engineered Air for their financial support and technical contributions to this study.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets generated and analyzed during the study are available from the corresponding author upon reasonable request.

Supplemental material

Supplemental material for this article is available online.

References

ASHRAE 185.1-2020. (n.d.) Method of Testing UV-C Lights for Use in Air-Handling Units or Air Ducts to Inactivate Airborne Microorganisms (ANSI Approved). Available at: https://store.accuristech.com/ashrae/standards/ashrae-185-1-2020?product_id=2185612 (accessed 24 February 2025).

Azimi

Stephens

(2013) HVAC filtration for controlling infectious airborne disease transmission in indoor environments: Predicting risk reductions and operational costs. Building and Environment 70: 150–160.

Buonanno

Ponnaiya

Welch

, et al. (2017) Germicidal efficacy and mammalian skin safety of 222-nm UV light. Radiation Research 187: 493–501.

Care222® Filtered Far UV-C Excimer Lamp Module | Ushio America, Inc. (n.d.). Available at: https://www.ushio.com/product/care222-filtered-far-uv-c-excimer-lamp-module/ (acc-essed 28 March 2024).

Cattai

D’Orazio

Sbardella

(2023) A systematic review on the application of ultraviolet germicidal irradiation to HVAC Systems. Energies 16(22): 7569.

Franz

CMAP

Specht

Cho

, et al. (2009) UV-C-inactivation of microorganisms in naturally cloudy apple juice using novel inactivation equipment based on Dean vortex technology. Food Control 20(12): 1103–1107.

Graeffe

Luo

Guo

, et al. (2023) Unwanted indoor air quality effects from using ultraviolet C lamps for disinfection. Environmental Science and Technology Letters 10(2): 172–178.

Guerrero-Beltrán

Barbosa-Cánovas

(2004) Advantages and limitations on processing foods by UV light. Food Science and Technology International 10(3): 137–147.

Huang

Agranovski

Pyankov

, et al. (2008) Removal of viable bioaerosol particles with a low-efficiency HVAC filter enhanced by continuous emission of unipolar air ions. Indoor Air 18(2): 106–112.

10.

Lau

Bahnfleth

Freihaut

(2009) Estimating the effects of ambient conditions on the performance of UVGI air cleaners. Building and Environment 44(7): 1362–1370.

11.

Yang

Zhang

, et al. (2021) Inactivation of foodborne pathogenic and spoilage bacteria by single and dual wavelength UV-LEDs: Synergistic effect and pulsed operation. Food Control 125: 107999.

12.

Wang

Liu

, et al. (2024) Evaluating the performance of UV disinfection across the 222–365 nm spectrum against aerosolized bacteria and viruses. Environmental Science and Technology 58(16): 6868–6877.

13.

Luo

Zhong

(2021) Ultraviolet germicidal irradiation (UVGI) for in-duct airborne bioaerosol disinfection: Review and analysis of design factors. Building and Environment 197: 107852.

14.

Luo

Zhong

(2022) Development and experimental validation of an improved mathematical irradiance model for in-duct ultraviolet germicidal irradiation (UVGI) applications. Building and Environment 226: 109699.

15.

Macher

(1989) Positive-hole correction of multiple-jet impactors for collecting viable microorganisms. American Industrial Hygiene Association Journal 50(11): 561–568.

16.

Nunayon

Zhang

Lai

ACK

(2020a) A novel upper-room UVC-LED irradiation system for disinfection of indoor bioaerosols under different operating and airflow conditions. Journal of Hazardous Materials 396: 122715.

17.

Nunayon

Zhang

Lai

ACK

(2020b) Comparison of disinfection performance of UVC-LED and conventional upper-room UVGI systems. Indoor Air 30(1): 180–191.

18.

Wang

Zhang

Chan

, et al. (2023) Experimental study of the disinfection performance of a 222-nm Far-UVC upper-room system on airborne microorganisms in a full-scale chamber. Building and Environment 236: 110260.

19.

Luo

Zhong

(2022) Development of a multi-functional HVAC facility for the evaluation of air cleaning technologies and ventilation performance. In: Proceedings of the Canadian Society for Mechanical Engineering International Congress 2022, CSME Congress 2022, Edmonton, AB, Canada https://doi.org/10.7939/r3-wgys-f764

20.

Zhong

(2023) An integrated energy analysis framework for evaluating the application of hydrogen-based energy storage systems in achieving net zero energy buildings and cities in Canada. Energy Conversion and Management 286: 117066.

21.

Yamano

Kunisada

Kaidzu

, et al. (2020) Long-term effects of 222-nm ultraviolet radiation C sterilizing lamps on mice susceptible to ultraviolet radiation. Photochemistry and Photobiology 96(4): 853–862.

22.

Yaun

Sumner

Eifert

, et al. (2004) Inhibition of pathogens on fresh produce by ultraviolet energy. International Journal of Food Microbiology 90(1): 1–8.

23.

Zhang

Lai

ACK

(2022) Evaluation of single-pass disinfection performance of far-UVC Light on Airborne Microorganisms in Duct Flows. Environmental science & technology 56(24): 17849–17857.

24.

Zhang

Jin

Nunayon

, et al. (2020) Disinfection by in-duct ultraviolet lamps under different environmental conditions in turbulent airflows. Indoor Air 30(3): 500–511.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.14 MB

0.00 MB