A modeling study of the impact of photolysis on indoor air quality

Abstract The importance of photolysis as an initiator of air chemistry outdoors is widely recognized, but its role in chemical processing indoors is often ignored. This paper uses recent experimental data to modify a detailed chemical model, using it to investigate the impacts of glass type, artificial indoor lighting, cloudiness, time of year and latitude on indoor photolysis rates and hence indoor air chemistry. Switching from an LED to an uncovered fluorescent tube light increased predicted indoor hydroxyl radical concentrations by ~13%. However, moving from glass that transmitted outdoor light at wavelengths above 380 nm to one that transmitted sunlight above 315 nm led to an increase in predicted hydroxyl radicals of more than 400%. For our studied species, including ozone, nitrogen oxides, nitrous acid, formaldehyde, and hydroxyl radicals, the latter were most sensitive to changes in indoor photolysis rates. Concentrations of nitrogen dioxide and formaldehyde were largely invariant, with exchange with outdoors and internal deposition controlling their indoor concentrations. Modern lights such as LEDs, together with low transmission glasses, will likely reduce the effects of photolysis indoors and the production of potentially harmful species. Research is needed on the health effects of different indoor air mixtures to confirm this conclusion.


| INTRODUC TI ON
In developed countries, people spend most of their time (~90%) indoors, 1,2 where they consequently receive most of their exposure to air pollution. The ongoing COVID-19 pandemic has heightened awareness of the importance of good indoor air quality. Many governments have asked their citizens to work or study at home, and restricted travel to prevent the spread of disease. 3 Therefore, air quality in the indoor environment and especially in our homes has become more important than ever.
The role of photolysis as a mediator of atmospheric chemistry has long been recognized for the ambient atmosphere, but there has been far less focus on the role that photolysis can play indoors. 4 Indoor light includes artificial lighting indoors and attenuated sunlight that can move into indoor environments through windows and skylights. Nazaroff and Cass 5 were the first to recognize the importance of indoor photolysis, using a simple model to show that increased photolysis rates enhanced the rate of chemical reactions, producing higher concentrations of reactive species. Carslaw 6 investigated the indoor air chemistry of a typical urban residence in the UK with a detailed chemical box model, showing that light intensity level indoors were a key determinant of model uncertainty when simulating OH concentrations. The simulated indoor OH concentration was ~4 × 10 5 molecule/cm 3 assuming that 3% and 10% of outdoor UV and visible light were transmitted indoors, respectively, but this concentration increased by 281% when UV and visible transmission increased to 27.5% and 75%, respectively. These higher simulated OH concentrations were confirmed by the measurements of Gomez Alvarez et al. 7 , who measured up to 1.8 × 10 6 molecule/cm 3 of OH in a school classroom in Marseille, when light shone directly through a window and photolysed nitrous acid (HONO) to produce OH.
The contribution of artificial light to overall photolysis indoors depends on the location of the light within the room, the geometry of the room, and the type of light, with different artificial lights having unique spectral 8 and spatial 9 characteristics. The amount of light that can penetrate indoors from outdoors is influenced by the type of window, time of year and day, the building orientation and location and meteorological conditions (e.g., cloudiness). For instance, Crawford 10 found that an unoccluded solar disk with slightly overcast conditions enhanced spectral actinic flux by 20% compared to clear sky conditions, while an 80% reduction was noted for more overcast conditions. Blocquet et al. 11 used both modeling and measurements to investigate the spatial and spectral distribution of sunlight which passed from outdoors through windows, finding that 0.15% to 30% of outdoor UV light (300-400 nm) and 0.7% to 80% of outdoor visible light (400-750 nm) were observed indoors depending on the glass type and time of day. 11 Similar reductions were reported by Zhou et al. 4 for 77 windows and glass samples.
It is worth considering how lighting and glazing has changed in recent years. The long lifetimes and high efficiency characteristics of fluorescent tubes (used mainly in office blocks and industrial settings) led to their being a dominant indoor lighting source for many years in such locations. 12 However, LED lights are becoming more popular, owing to much higher energy efficiency compared to more traditional lighting. For instance, they are estimated to provide 56%-62% energy savings and an increase in lifetime by a factor of 9 compared to the use of fluorescent tubes. 13,14 In residential settings, incandescent lighting was a dominant lighting source for many years. 15 However, this type of lighting is also being replaced by LED lights. Relative to incandescent lights, LEDs use ~85% less energy and have 50 times longer lifetimes 16 and are likely to remain as the dominant source of illumination in the future. 17 Glass composition has also become increasingly sophisticated in recent years, such as through multipane glazing, 18 tinting, 19 lowemissivity coatings, 20 anti-reflective coatings, 21 and vacuum glazing, 22 compared to the single pane and compositionally simple glass types that used to be more common. 23 These changes will undoubtedly affect levels of indoor lighting and hence indoor air chemistry.
There have been a few papers that have focused on the impacts of different drivers of indoor air chemistry to date. For instance, Zhou and Kahan 24 undertook a thorough photochemical characterization of a test house in Texas, including the determination of spatial and temporal photochemical rate constants and quantification of the effects of cloud cover and diffuse light. In addition, Zhou et al. 4 investigated the impacts of different window materials and outdoor meteorological conditions on indoor photolysis rates. However, detailed chemical models can provide deeper insight and consider a wider range of conditions than experimental data alone. This paper describes an improved representation of photolysis in a detailed chemical model for indoor air. The improved model is then used to investigate the impacts of different controlling factors on indoor photolysis rates and hence indoor air chemistry. york.ac.uk/), that includes around 20,000 reactions and 5000 species, and represents the near-explicit degradation of ~143 VOCs in the gas-phase. [26][27][28][29] The chemical degradation of each VOC is initiated by reaction with hydroxyl (OH) radicals, nitrate radicals (NO 3  are formed. 27 The model also includes terms that represent deposition to and emission from surfaces, exchange with outdoors and gas-to-particle partitioning reactions for limonene. 25

| Representation of artificial lighting
The INDCM considers 35 photolysis processes for either individual or groups of species based on the Master Chemical Mechanism protocol. 27,29 For these 35 processes, 27 occur in the UV region only, 5 in the UV and visible regions and 1 in the visible only (see Table S1).
In the previous version of the model, 6 flat transmission of light in the UV and visible wavelength ranges was assumed from outdoors, with only one type of indoor lighting, based on the methods described in Nazaroff and Cass. 5 The photolysis coefficient (j) for each species i, was calculated for each individual photolysis process.
For the modifications made for this study, the UV wavelength region from 300 to 400 nm was split into ten different 10 nm subregions (300-310 nm; 310-320 nm, etc.). In the 400-800 nm wavelength region, fewer species absorb and transmission is much flatter than in the 300-400 nm wavelength range, 8 so it was considered as one further wavelength interval.
The overall photolysis rate coefficient (j) was calculated for each species using a modified form of Equation 1 as Kowal et al. 8

| Representation of attenuated outdoor sunlight
The previous model assumed that 3% of UV and 10% of visible light from outdoors passed through the windows and ended up indoors. 6 However, in reality, transmission of outdoor light indoors will vary depending on the window material (glass) composition. Blocquet

| Model simulations
The model location was set to York, UK, and the date was set to June 21. The indoor temperature was assumed to be 300 K, rela- mutagenicity, 38,39 and eye irritation. 40,41 Organic nitrates were also found to have adverse health effects. 42 These two groups of species therefore act as a proxy for the potentially harmful species that can be formed through secondary chemistry indoors under the different lighting conditions.

| Spectral radiometer measurements
Photolysis values were measured indoors using a spectral radiometer, which provided a direct measurement of solar actinic UV flux and permitted the determination of photolysis frequencies. 43 The The results focus on 25th, which was the sunniest day with internal lights off.

| Impact of model improvements on predicted concentrations
The first test was to compare the impacts of these changes to the previous model output.     concentrations produce more PANs and organic nitrates for these lighting conditions compared to the others.

| Impacts of different indoor artificial lights on indoor air chemistry
In summary, the differences between indoor lights are relatively modest, although there clearly are some differences, with some of the more intense fluorescent lights generating higher OH concentrations than the other lighting we simulated.  the concentration under Glass C and LE, respectively, so less NO is available to react with HO 2 and RO 2 . In fact, increased photolysis rates will increase OH concentrations and hence its ability to form HO 2 and RO 2 , but also form more NO from NO 2 photolysis under the same conditions, which can then react with HO 2 and RO 2 . There are subtle differences in the balance between these processes which determine the predicted peroxy radical concentrations.

| Impact of cloudiness on indoor air chemistry
The level of cloudiness outdoors can have a large impact on pre-   For future studies, more measurements of indoor photolysis rates are essential, as well as measurements of the key species fo-