Indoor Air Quality Implications of Germicidal 222 nm Light

Abstract

A known strategy for mitigating the indoor transmission of airborne pathogens, including the SARS-CoV-2 virus, is irradiation by germicidal UV light (GUV). A particularly promising approach is 222 nm light from KrCl excimer lamps (GUV222); this inactivates airborne pathogens, but is thought to be far less harmful to human skin and eyes than longer-wavelength GUV (e.g., 254 nm). However, the potential for GUV222 to affect the composition of indoor air has received little experimental study. Here, we conduct a series of controlled laboratory experiments, carried out in a 150 L Teflon chamber, to examine formation of oxidants and other secondary species by GUV222. We show that GUV222 generates ozone (O3) and hydroxyl radicals (OH), both of which can react with volatile organic compounds to form oxidized volatile organic compounds and secondary organic aerosol particles. Results are consistent with predictions from a simple box model based on known photochemistry. We use this experimentally-validated model to simulate the effect of GUV222 irradiation under more realistic indoor air scenarios, spanning a range of light and ventilation conditions. We demonstrate that under some conditions, GUV222 irradiation can lead to levels of O3, OH, and secondary organic products that are substantially elevated relative to normal indoor conditions, especially when ventilation is low and GUV222 intensity is high. Thus, GUV222 should be used at the lowest intensities possible and in concert with ventilation, decreasing levels of airborne pathogens while mitigating the formation of air pollutants in indoor environments.

Full text

1
Indoor Air Quality Implications of Germicidal 222 nm Light 1
Victoria Barber1,2*, Matthew B. Goss1,2, Lesly J. Franco Deloya3, Lexy N. LeMar4, Yaowei Li5, 2
Erik Helstrom1, Manjula Canagaratna6, Frank N. Keutsch5,7,8, Jesse H. Kroll1,4*
3
1Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States 4
3Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States 5
4Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States 6
5John A. Paulson School of Engineering and Applied Sciences, Harvard University, Ca mbridge, Massachusetts 02138, United States 7
6Center for Aerosol and Cloud Chemistry, Aerodyne Research Incorporated, Billerica, Massachusetts 01821, United States 8
7Depart ment of C hemist ry and Chemica l Biolo gy, Har vard Un iversit y, Cambridge, Massachusetts 02138, United States 9
8Depart ment of Earth and Planeta ry Sciences , Harvard University, Cambridge, Massachusetts 02138, United States 10
2Equal contribution, *Corresponding Authors 11
12
vbarber@mit.edu, 13
Massachusetts Institute of Technology 14
Department of Civil and Environmental Engineering 15
77 Massachusetts Avenue, 48-330 16
Cambridge, MA 02139 17
18
jhkroll@mit.edu 19
Massachusetts Institute of Technology 20
Department of Civil and Environmental Engineering 21
77 Massachusetts Avenue, 48-331 22
Cambridge, MA 02139 23
617-253-2409 24
25
Keywords: Ultraviolet Germicidal Irradiation, Indoor Air Quality, Ozone, Photochemistry, 26
Ventilation, Volatile Organic Compounds, Secondary Organic Aerosol 27
28
29

2
30
Abstract 31
A known strategy for mitigating the indoor transmission of airborne pathogens, including the 32
SARS-CoV-2 virus, is irradiation by germicidal UV light (GUV). A particularly promising 33
approach is 222 nm light from KrCl excimer lamps (GUV222); this inactivates airborne 34
pathogens, but is thought to be far less harmful to human skin and eyes than longer-wavelength 35
GUV (e.g., 254 nm). However, the potential for GUV222 to affect the composition of indoor air 36
has received little experimental study. Here, we conduct a series of controlled laboratory 37
experiments, carried out in a 150 L Teflon chamber, to examine formation of oxidants and other 38
secondary species by GUV222. We show that GUV222 generates ozone (O3) and hydroxyl radicals 39
(OH), both of which can react with volatile organic compounds to form oxidized volatile organic 40
compounds and secondary organic aerosol particles. Results are consistent with predictions from 41
a simple box model based on known photochemistry. We use this experimentally-validated 42
model to simulate the effect of GUV222 irradiation under more realistic indoor air scenarios, 43
spanning a range of light and ventilation conditions. We demonstrate that under some conditions, 44
GUV222 irradiation can lead to levels of O3, OH, and secondary organic products that are 45
substantially elevated relative to normal indoor conditions, especially when ventilation is low 46
and GUV222 intensity is high. Thus, GUV222 should be used at the lowest intensities possible and 47
in concert with ventilation, decreasing levels of airborne pathogens while mitigating the 48
formation of air pollutants in indoor environments. 49
Significance Statement 50
Many respiratory pathogens, including SARS-CoV-2, are spread via airborne transmission. This 51
is particularly problematic in indoor environments, due to limited ventilation. One technique that 52
can reduce levels of indoor airborne pathogens is irradiation by short-wavelength (222 nm) 53
germicidal ultraviolet light (GUV222), which inactivates pathogens while being relatively skin- 54
and eye-safe. However, GUV222 implications for indoor air quality have not been investigated in 55
detail. We carry out laboratory studies showing that GUV222 forms ozone (an oxidant and 56
respiratory irritant), the hydroxyl radical (a stronger oxidant), and a range of oxidation 57
byproducts, including fine particulate matter. We extrapolate results to more realistic indoor 58
spaces, and show that to minimize negative health impacts, GUV222 should be used alongside 59
(rather than instead of) ventilation. 60

3
1. Introduction 61
The COVID-19 pandemic has highlighted the critical need to develop and implement strategies 62
to decrease the transmission of airborne pathogens. Approaches include both source control 63
(isolation, masking), and remediation (ventilation, air cleaning). One approach that has received 64
substantial attention is the use of germicidal ultraviolet (GUV) light, which inactivates airborne 65
pathogens. This approach goes back decades (1), traditionally using 254 nm light from mercury 66
lamps. Since light of this wavelength can cause damage to skin and eyes, care must be taken to 67
minimize occupants’ direct exposure to the GUV light (2, 3). 68
A promising new approach to GUV-based air cleaning is the use of KrCl excimer lamps, which 69
emit at 222 nm (GUV222)(4). In contrast to 254 nm GUV, GUV222 does not penetrate deeply into 70
biological materials. Therefore, while GUV222 is effective at inactivating airborne viruses and 71
bacteria, it is unable to penetrate the outer layer of dead skin cells or the ocular tear layer (5). 222 72
nm light is hence less likely to reach and damage living human tissues, offering the potential for 73
air disinfection throughout an entire, occupied indoor space. 74
A risk with GUV222-based air cleaning, as with all types of air cleaning that rely on chemical 75
and/or photolytic processes, is the potential formation of unwanted secondary byproducts (6, 7). 76
A particular concern with GUV222 is the formation of ozone (O3), a harmful air pollutant that acts 77
as a strong oxidant and can lead to respiratory distress when inhaled (8). O3 is formed by the UV 78
photodissociation of oxygen (R1-2) 79
O2 + hv λ<242 nm → O + O (R1) 80
O + O2 + M → O3 + M (R2) 81
Since absorption of UV by O2, and hence O3 production, is strongest at short wavelengths (9), 82
manufacturers of KrCl lamps have added filters to block wavelengths shorter than 222 nm. But 83
since O2 absorbs weakly even at 222 nm (σ = 4.09x1024 cm2 (9)), all KrCl lamps have the 84
potential to generate ozone, possibly in concentrations higher than is typically found indoors 85
(roughly 5 ppb (10)). 86
Ozone generated indoors, in addition to posing a direct health hazard, can set off a cascade of 87
chemical reactions that can also affect indoor air quality. Ozone reacts directly with alkenes, 88
present both in the air and on indoor surfaces, forming a range of oxidized volatile organic 89
compounds (OVOCs)(11, 12) and secondary organic aerosol (SOA)(13), which may negatively 90
impact human health (14–17). O3 chemistry can also lead to the formation of the hydroxyl 91
radical (OH), an even stronger oxidant. This occurs either through reactions with alkenes, which 92
are known to form OH (R3)(11, 18), or through O3 photolysis (R4-5) (19): 93
Alkene + O3 → OH + other products (R3) 94
O3 + hv λ <411 nm → O2 + O(1D) (R4) 95
O(1D) + H2O → 2OH (R5) 96

4
Any increased levels of indoor O3 from GUV222 would likely enhance the importance of these 97
reactions, leading to higher levels of indoor OH. This includes O3 photolysis (R4-5), which is the 98
main source of OH in the troposphere but under normal conditions is negligible in indoor 99
environments, due to the lack of low-wavelength UV. Any OH radicals formed from R3-5 may 100
then oxidize a wide range of organic species and lead to the formation of OVOCs and SOA. 101
GUV222 therefore has the potential to dramatically affect the chemical composition of indoor air, 102
and may lead to the formation of chemical species that are hazardous to human health. However, 103
the extent and nature of this impact remains quite uncertain, even as GUV222 is being deployed in 104
indoor spaces (20). Two very recent experimental studies (21, 22) demonstrate O3 production 105
from GUV222, but these do not examine the overall effects on indoor air quality (including the 106
production of OH, OVOCs, and SOA) by GUV222. To our knowledge the only work that has is 107
a box-modeling study by Peng et al. (23). That work predicted that 222 nm irradiation could lead 108
to elevated levels of O3 and other secondary species relative to non-illuminated conditions, 109
especially under low-ventilation conditions. To date, such modeling results have yet to be tested 110
experimentally. 111
Here we describe a series of laboratory experiments aimed at better understanding the effects of 112
222 nm irradiation on indoor air quality. The goal of this work is to gain process-based insight 113
into how such irradiation affects the chemical composition of the air; we do not examine the 114
effects of GUV222 light on pathogens, indoor surfaces, or human health. These experiments, 115
which use a flow-through Teflon chamber coupled to a range of real-time analytical instruments, 116
explore the effects of several parameters relevant to indoor air processes (VOC level, ventilation, 117
222 nm light intensity, and humidity) on the generation of oxidants and secondary products. 118
Results are then used to validate a simple chemical model of GUV222 irradiation of indoor air, 119
which in turn is used to examine the interplay between GUV222 and ventilation in controlling the 120
levels of ozone and other chemical species in the indoor environment. 121
2. Results and Discussion 122
2.1. Ozone production. The production of ozone by 222 nm light is examined via the irradiation 123
of clean chamber air. Figure 1 shows results from four representative irradiation experiments 124
(average GUV222 irradiance = 45 µW/cm2), run at different ventilation rates (1.3 to 3.1 air 125
changes per hour (ACH)) and relative humidities (25%-45%). O3 production is observed to occur 126
immediately when the lights are turned on. O3 levels increase quickly at first, eventually leveling 127
off to a steady-state value, in which photolytic production is balanced by removal by outflow. 128
The O3 production rate is measured at 324 ± 18 ppb hr-1), in reasonably good agreement with 129
previous measurements (21) when differences in average GUV222 irradiance are considered (see 130
Section S1.1). The steady-state O3 concentration is independent of relative humidity, and 131
inversely proportional to ventilation rate (Figure S1). 132
Dashed lines in Figure 1 denote O3 concentrations predicted from a simple box model, which 133
includes O2 photolysis (R1-2), Ox-HOx chemistry, and dilution (model details are given in the 134
Methods and SI). The model accurately predicts measured O3 levels, indicating that the 135

5
processes describing ozone levels (formation from O3 photolysis at 222 nm, loss by outflow) are 136
well-captured by the simple model. 137
138
Figure 1: Observed ozone production for clean-chamber irradiation experiments. Measurements 139
agree well with the predictions from the simple box model (dashed lines) across a range of 140
ventilation rates and relative humidities. Measurements shown in red are taken at 25% RH. 141
142
2.2. Decay of VOCs upon 222 nm irradiation. 143
In a second set of experiments (listed in Table S1), VOCs are added to the irradiated chamber 144
after O3 levels reach steady state. Experiments center on two VOCs: hexanal (C6H12O), a C6 145
compound that reacts only with OH, and cyclohexene (C6H10), a C6 compound that reacts with 146
both OH and O3. VOC decays are shown in Figure 2. 147
148

6
149
Figure 2: Normalized decays of two VOCs (hexanal and cyclohexene) after introduction to the 150
GUV222-irradiated chamber (see also Figure S2). Time = 0 refers to when the VOC was injected 151
into the chamber. Traces are background- and dilution-corrected, so observed decays are from 152
oxidative loss only. Details of each experimental condition (base, O3 only, low light) are given in 153
the text and Table S1. Solid black lines denote single-exponential fits to the observed decays; 154
dashed black lines show the expected decay of cyclohexene from reaction with O3 only (24).
155
156
Under “base conditions” (10 ppb VOC precursor, 222 nm light, ~25% RH) (Figure 2AB), the 157
concentrations of both hexanal and cyclohexene decrease after being introduced to the irradiated 158
chamber. Concentrations are corrected for dilution; losses by direct photolysis and uptake to 159
surfaces are expected to be minimal (see Section S1.2). Therefore, decays indicate oxidative loss 160
only. This oxidation cannot be explained by O3 alone: hexanal does not react with O3, and while 161
cyclohexene does, its decay is far faster than what can be attributed to the O3 reaction (dashed 162
line). Indeed, for experiments in which the GUV222 light is off and VOCs are exposed to the 163
same levels of O3 as in the irradiated case (Figure 2CD), the hexanal does not decrease at all, and 164
cyclohexene decays far less than in the irradiation case, at a rate consistent with reaction with O3 165
(plus a small contribution from OH generated by the ozonolysis reaction, reaction R3). This 166
observed “excess reactivity” (the difference in observed decays and decays expected from O3 167
reaction alone) indicates that GUV222 irradiation generates not only O3 but other oxidants as 168
well. 169
Additional experiments carried out under a range of reaction conditions provide evidence that 170
these additional oxidants are OH radicals, formed from reactions 3-5. For example, experiments 171

7
with the 222 nm light intensity attenuated substantially (~ 9 μW cm-2) exhibit VOC decay rates 172
that are much slower compared to those under base conditions (Figure 2EF). Attenuating light 173
intensity is assumed to decrease steady-state O3 concentrations proportionally (see methods). 174
However, the observed excess reactivity disproportionately decreases, by approximately an order 175
of magnitude. This is consistent with OH formation, which depends on the photolysis of both O2 176
and O3, as well as (in the case of cyclohexene) the ozonolysis reactions. The dependence of 177
decays on other experimental parameters, such as VOC concentration and relative humidity, are 178
also consistent with OH production from GUV222 lights; this is discussed in detail in Section 179
S1.3. 180
We estimate average OH levels in all experiments, using the excess reactivity and known OH 181
rate constants (25, 26) (see Section S1.4). We also calculate OH levels using our simple box 182
model (see Methods). Measured and modeled average [OH] agree well (Figure 3), providing 183
strong evidence that GUV222 produces not only O3 (R1-2) but also OH (R3-5), and that oxidation 184
by both O3 and OH can take place upon irradiation with 222 nm light. 185
186
Figure 3: Experimentally-derived average OH concentration vs. average OH concentration 187
predicted by the box model, for all cyclohexene and hexanal experiments (see Section S1.4). 188
Note the break in the x-axis. 189
2.3 Formation of gas-phase oxidation products. The formation of oxidized gas-phase products is 190
observed in all experiments in which VOC oxidation occurs. Product distributions for three 191
cyclohexene experiments (base conditions, O3 only, and low light) are shown in Figure 4. 192
Additional product distributions and time-series results (including for the hexanal experiments) 193
are provided in Figures S3 and S4. 194

8
195
196
Figure 4: Gas-phase products from cyclohexene experiments. Panel A: Normalized mass 197
spectrometric signal of products formed for the GUV222 irradiation (base conditions), O3-only, 198
and low-light experiments (see Section S1.5 for calculations and Figure S3 for other 199
experimental conditions). Signals are integrated from t = 250 s to 2500 s, normalized to total 200
integrated ion signal and grouped by carbon number (nC). In all cases products are dominated by 201
C6H10O2 (the major cyclohexene + OH reaction product) and C6H10O3, (the major cyclohexene + 202
O3 product). Panel B: The ratio of the C6H10O2-to-C6H10O3 signals vs. the ratio of the rates of 203
OH and O3 oxidation, for all cyclohexene experiments. Concentrations of OH are determined 204
from the fits in Figure 2, while concentrations of O3 are measured directly. The dashed line is a 205
linear fit to the data; since the two products have differing sensitivities in the instrument, this 206
differs from the 1:1 line. Error bars represent the range of values observed throughout the 207
experiment. 208
209
Measured products are dominated by C6 and C5 compounds, as expected given that cyclohexene 210
is a C6 species. The two products with the largest mass spectrometric signals, C6H10O2 and 211
C6H10O3, are the major products of the OH and O3 initiated oxidation of cyclohexene, 212
respectively (27, 28) (see Scheme S1) (Products are detected as the analyte-NH4+ adduct, and 213
reported as the analyte formula.) The ratios of the signals from the two products varies among 214
experiments, indicating differences in the relative concentrations of OH and O3. In Figure 4b, the 215
ratio of the mass spectrometric signals of these two products is shown vs. the relative OH-to-O3 216
oxidation rate ratios (calculated from the experimentally-determined values of [OH] and [O3]) 217
for each cyclohexene experiment. A strong correlation (R2 = 0.98) is found between the two 218
ratios, providing further support for OH-initiated oxidation, and more generally for OH radical 219

9
production from irradiation by 222 nm light. The products formed in the 222 nm irradiation of 220
hexanal are also broadly consistent with OH-initiated oxidation (see Scheme S2) (29). 221
2.4 Secondary organic aerosol formation. In all experiments, dry ammonium sulfate seed 222
particles are added to the chamber, providing surface area onto which low-volatility species may 223
condense, and enabling the assessment of potential SOA formation. SOA formation is observed 224
in a number of experiments (Table S1 and Figure S5). SOA formation is generally modest for 225
most hexanal and cyclohexene experiments, likely due to the relatively small size (C6) and low 226
concentrations (10 ppb) of those species. Higher concentrations of SOA are observed for 227
experiments with high initial concentrations (100 ppb) of hexanal or cyclohexane, and for those 228
using limonene (C10H16, a monoterpene commonly found in fragrances and cleaning products). 229
In fact, the GUV222 irradiation of 100 ppb limonene (a level that can be found in indoor 230
environments immediately after cleaning events (30, 31)) results in exceedingly high SOA 231
loadings, on the order of 400 ± 80 µg m-3. Additionally, the formation of new particles is 232
observed upon 222 nm irradiation under some conditions (Section S1.6 and Figure S6). This 233
occurs even when no VOCs are added, and so may result from the oxidation of trace organic 234
species on the chamber walls. Whether this is a general feature of the irradiation of organics on 235
indoor surfaces is unclear from the present experiments, but it does suggest that 222 nm 236
irradiation may induce new particle formation in some environments. 237
238

10
239
2.5 Extrapolation to indoor environments 240
241
Figure 5: Effects of ventilation and GUV222 irradiation on modeled GUV efficacy and indoor air 242
quality (see also Figures S7 and S8). Panel A: effective air changes per hour (eACH) for indoor 243
pathogens, based on the previously reported inactivation rate of SARS-CoV-2 at 222 nm (32) 244
(Section S2.1). Panels B-D: steady-state concentrations of (B) O3, (C) OH, and (D) organic 245
oxidation products, respectively, as predicted by the photochemical box model. Panel D 246
calculations assume unit yields, and do not account for VOC production from surfaces, so likely represent 247
lower limits. Lighter colors represent larger values; note that the logarithmic color scaling is 248
different for each panel. 249

11
250
The above laboratory experiments demonstrate that GUV222 irradiation forms ozone, OH, and a 251
range of oxidation products; measured ozone and inferred OH agree broadly with predictions by 252
a photochemical box model. However real-world indoor environments are more complex than 253
our simple laboratory system, as they involve a large number of organic compounds, 254
depositional loss of ozone and other species, infiltration of outdoor pollutants, and a wide range 255
of possible ventilation rates. Here we extend our photochemical model to a more realistic indoor 256
air scenario, with the goal of understanding how GUV222 may impact indoor air quality under a 257
range of ventilation and irradiation conditions. The expanded model is described in detail in the 258
Methods and in SI. Briefly, reactive VOCs are not simulated individually but rather as “lumped” 259
species, defined by their OH and O3 reactivities, which are chosen based on previous 260
measurements of indoor air (33, 34). The model is run at 298 K, 1 atm, and 30% RH. We also 261
include a background concentration of O3 in the ventilation air (40 ppb, consistent with typical 262
outdoor O3 concentrations), a 25% loss of O3 to the ventilation system, and an O3 deposition 263
constant of 3 hr-1 (10, 35). 264
The range of light intensities chosen covers US and international guidelines on 222 nm exposure 265
limits (ranging from 0.8 to 16 µW/cm2 over 8 hours (36, 37)) as well as the values in previous 266
studies used for pathogen deactivation (average intensities of 0.09 to 14.4 µW/cm2 at 1.7 m 267
above the ground from Eadie et al. (38) and 3.5 µW/cm2 from Peng et al. (23)). Ventilation rates 268
span a range of typical indoor values, and include the minimum American Society of Heating, 269
Refrigerating and Air-Conditioning Engineers (ASHRAE) recommendations for homes (0.35 270
ACH), offices (~2-3 ACH), and health care settings (10 ACH)(39). 271
Key model results are provided in Figure 5. Figure 5A shows the effective air change rate 272
(eACH) across a wide range of 222 nm light intensities and ventilation rates; even modest 273
irradiation levels lead to substantial increases in eACH (see also Figure S7A). Figures 5B, C, and 274
D show the steady-state indoor concentrations of O3, OH, and total oxidation products (assuming 275
unit yield), respectively. 276
Steady-state ozone levels (Figure 5B) are higher with 222 nm irradiation than without. Sources 277
of O3 include photochemistry (R1-2) and infiltration of outdoor air, while sinks include 278
deposition, ventilation, and chemical reaction (rates and contributions of individual processes are 279
given in Figures S7B-E). With low irradiation, O3 levels are governed mainly by infiltration of 280
outdoor air, and O3 increases are modest. Under the highest irradiation levels (>25 µW/cm2), and 281
especially under low ventilation rates (<1 ACH), indoor O3 can reach levels exceeding that of the 282
outdoors, and can even exceed the OSHA indoor limit of 100 ppb. However, even a small 283
change in indoor O3 levels can have a dramatic effect on people’s total ozone exposure, given the 284
large fraction of time people spend indoors (40). In most cases, deposition represents the 285
dominant sink of ozone (Figure S7D). While product formation due to these surface losses (e.g., 286
paint, textiles, skin) is not included in the model, volatile secondary organic products stemming 287
from these reactions (12, 41) could represent an additional secondary effect of GUV222 on indoor 288
air quality. 289

12
Figure 5C shows steady-state levels of OH as a function of ventilation and 222 nm light 290
intensity. Sources of OH include O3-alkene reactions (R3) and photochemistry (R4-5), while 291
sinks are dominated by reactive losses (see also Figures S7F-G). In the absence of GUV222 292
irradiation, modeled OH is from alkene ozonolysis only, with predicted levels (~105 molec cm-3) 293
overlapping but falling on the low end of measured and modeled OH in unperturbed indoor 294
spaces (which range from 6x104-1.6x106 molec cm-3) (42–50); this underestimate may arise from 295
the omission of photolysis of trace species such as nitrous acid (HONO) or aldehydes, which 296
may be important in some environments (51). As with O3, GUV222 irradiation leads to increases 297
in indoor levels of OH. At low to moderate light intensities, this increase in OH is mostly due to 298
the alkene ozonolysis reaction, while at higher light intensities, ozone photolysis plays a larger 299
role (Figure S7G). OH increases with increasing photochemistry (e.g., light intensity and ozone 300
concentrations), but is substantially modulated by reactive losses with VOCs. VOC 301
concentrations are higher at low ventilation rates (see Figure S7H), due to the buildup of emitted 302
VOCs, which suppress OH concentrations. At high light intensities, steady-state OH levels can 303
approach outdoor levels, matching or exceeding indoor OH measurements during transient 304
events such as cleaning or cooking activities (13, 52). We do not examine the role of HONO, 305
which can be present in high (ppb) levels indoors (50) and absorbs strongly at 222 nm (σ = 1.35 306
x 10-18 cm2 (9)); HONO photolysis may lead to even higher OH levels than predicted here. 307
The production of O3 and OH by GUV222-driven chemistry and their subsequent reactions with 308
VOCs leads to an increase in organic oxidation products (OVOCs and SOA). Steady-state levels 309
and production rates of such products (assuming unit yields) are shown in Figures 5D and S7I. 310
Concentrations increase with increased light intensity, and are especially high at low ventilation 311
rates. Since more than one product molecule may be formed per oxidation reaction, and OVOCs 312
may also be formed by surface reactions of O3 or OH, these numbers likely represent lower 313
limits. Of particular concern is the production of hazardous air pollutants (HAPs, such as CH2O) 314
and secondary organic aerosol, both of which may represent health hazards in the indoor 315
environment. Concentrations of SOA are challenging to predict, as SOA production depends on 316
the amounts and identity of the indoor VOCs, as well as on a host of reaction conditions. 317
However, SOA levels on the order of a few µg/m3 are reasonable (Figure S8); the production of 318
SOA from 222 nm irradiation in realistic indoor settings is an important area of future research. 319
320
Conclusions 321
In this study, we have demonstrated that GUV222 light leads to the production of (1) ozone, (2) 322
OH radicals, and (3) secondary organic species (OVOCs and SOA); these are in broad agreement 323
with prior model predictions (23). The resulting concentrations of such secondary species can be 324
substantially higher than are normally found in indoor environments; in extreme cases, these 325
increases can be dramatic, leading to oxidation conditions more similar to those found in outdoor 326
environments. The negative health impacts associated with the unavoidable generation of these 327
secondary species – most importantly O3, fine particular matter, and HAPs – thus need to be 328
taken into account (and ideally mitigated) when considering the use of 222 nm disinfection in 329
indoor spaces. 330

13
While a detailed analysis of the health impacts of GUV222 use (both the benefits from 331
inactivation of airborne pathogens and the drawbacks from secondary pollutant formation) is 332
beyond the scope of this work, our results offer some broad guidance as to the optimal use of 333
GUV222 in indoor environments. Most importantly, GUV222 disinfection alone is not a safe 334
substitute for ventilation as a means to control levels of indoor airborne pathogens, as it can lead 335
to the buildup of indoor ozone and other pollutants to dangerous levels (Figure 5). However, 336
GUV222 may be effectively used in conjunction with ventilation: relatively modest irradiation 337
levels combined with carefully chosen ventilation conditions can greatly enhance the effective 338
air change rate (Figure 5A), while limiting the levels of secondary pollutants (Figures 5B-D). 339
Moreover, due to the unavoidable formation of secondary pollutants, GUV222 lights should be 340
run at the lowest effective levels whenever possible. Further, the combination of GUV222 341
irradiation with air-cleaning technologies (e.g., sorbents for ozone and OVOCs, filters for 342
particulate matter) may serve to minimize indoor secondary pollutant levels, potentially enabling 343
safer use of GUV222 under poorly-ventilated environments. Quantifying the benefits and 344
tradeoffs of these combined approaches (ventilation, GUV222 irradiation, and/or air cleaning) in 345
terms of pathogen transmission, air pollutant levels, human health, and cost-effectiveness, is a 346
critical next step toward ensuring healthier indoor environments. 347
348
Materials and Methods 349
a. Experimental Methods 350
Experiments are carried out in a 150 L Teflon chamber, outfitted with inlet ports (for 351
introduction of clean air and trace species) and outlet ports (for sampling by analytical 352
instrumentation). Clean air from a zero-air generator (Aadco Model 737) is sent into the chamber 353
directly and through a water bubbler. Dilution rates are measured using acetonitrile, an inert 354
dilution tracer (8.0 x 10-4 – 9.7 x 10-4 s-1, 2.9 – 3.5 ACH). Most experiments are conducted at 355
22°C and ~25% RH; “higher RH” experiments are carried out at ~45% RH. 356
GUV222 light is provided by a single filtered KrCl excimer lamp (Ushio, Care222 B1 Illuminator, 357
peak emission at 222 nm), centered directly above the Teflon chamber. Average irradiance 358
within the chamber is ~45 µW/cm2; light intensities are calculated from the lamp intensity profile 359
provided by the manufacturer (53) (see Section S4.2). Most experiments are carried out at the 360
full light intensity; for “low light” experiments, the lamp emission is attenuated by several layers 361
of plastic, achieving a factor of ~5 reduction in intensity (determined by the reduction in the 362
steady-state O3 concentration). For the “O3-only” experiments, the light is left off, and O3 is 363
introduced via a Pen-Ray ozone generator, with a steady-state O3 concentration matching that of 364
the GUV222 experiments (~100 ppb). Reaction conditions for each experiment are described in 365
detail in Table S1. 366
For all VOC oxidation experiments, the chamber is first allowed to reach a steady-state 367
concentration of O3, either via 222 nm irradiation or direct addition. This is followed by the 368
addition of 5.3 ppb of acetonitrile (the dilution tracer), 1.2 ppb of 1-butan-d9-ol (intended as an 369
OH tracer, but not used here due to the relatively low OH levels), and 120 ± 11 µg m-3 of 370

14
ammonium sulfate particles (to act as seed particles for any SOA production). Finally, the VOC 371
(hexanal, cyclohexene, or (R)-(+)-limonene, 10 or 100 ppb) is added to chamber (see Section 372
S3.1). Because the oxidants are already present in the chamber, oxidation begins immediately, so 373
VOC injection is taken as t = 0. 374
Real-time measurements of gas- and particle-phase composition in the chamber are conducted 375
using a suite of analytical instruments, summarized in Table S2. Ozone is measured by a UV 376
absorption monitor (2BTech). Reactant VOC and OVOC products are monitored using a Vocus 377
proton transfer-reaction mass spectrometer (PTR-MS, Tofwerk, Aerodyne Research, Inc. (54)), 378
and an ammonium chemical ionization mass spectrometer (NH4+ CIMS, modified PTR3, see 379
Zaytsev et al. (55)). Particle-phase products are quantified using a scanning mobility particle 380
sizer (SMPS, TSI) and an aerosol mass spectrometer (Aerodyne Research, Inc. (56)). Data 381
analysis and quantification approaches are described in Section S3.2. 382
b. Modeling Methods 383
The photooxidation chemistry in both the chamber (Figures 1-3) and in more realistic indoor 384
environments (Figure 5) is described using a simple photochemical box model. The model uses 385
rate constants and photochemical parameters from the literature (9–11, 24–26, 35, 57), and 386
includes Ox and HOx chemistry; the full set of reactions used is listed in Table S3. The only 387
photolysis reactions included are of O2 and O3 (R1 and R3). 388
For simulations of chamber chemistry, the model includes a highly simplified oxidation scheme 389
of the injected VOC. Model parameters (e.g. VOC starting concentration, light intensity, air-390
exchange rate, and RH) are matched to the experiment in question. O3 deposition (which is likely 391
small on Teflon surfaces) is not included. 392
For simulations of chemistry in a more realistic indoor environment (Figure 5), two “lumped” 393
VOCs are included in the model: one (VOC1) that reacts with OH but not with O3, and another 394
(VOC2) that reacts with both OH and O3. Rate constants for VOC1 are chosen based on typical 395
values for indoor VOCs (Section S4.1 and Table S4); rate constants for VOC2 are assumed to be 396
equal to those of limonene. OH yields from O3 + VOC2 are assumed to by 0.86, equal to that of 397
limonene (11). All oxidation reactions form lumped organic products that also can react with 398
OH. VOC emission rates (84 ppb hr-1 and 4.2 ppb hr-1 for VOC1 and VOC2, respectively) are 399
determined from previous measurements of OH and O3 reactivities in indoor environments (33, 400
34); details of these calculations are given in Section S4.1. 401
3. Acknowledgements and Funding Sources 402
This work is supported by the U.S. National Science Foundation under grants ECS-2108811 and 403
AGS-2129835 and the Harvard Global Institute. The authors thank Bella Nesti (Harvard 404
University) for assisting with the initial phases of data analysis, and Jose Jimenez (University of 405
Colorado Boulder) for helpful discussions. 406
407
408

15
4. References 409
1. W. F. Wells, M. W. Wells, T. S. Wilder, The environmental control of epidemic contagion: 410
I. An epidemiologic study of radiant disinfection of air in day schools. American Journal of 411
Epidemiology 35, 97–121 (1942). 412
2. E. A. Nardell, Air Disinfection for Airborne Infection Control with a Focus on COVID-19: 413
Why Germicidal UV is Essential†. Photochemistry and Photobiology 97, 493–497 (2021). 414
3. Centers for Disease Control and Prevention, Upper Room Ultraviolet Germicidal 415
Irradiation (UVGI) (2021) (May 17, 2023). 416
4. E. R. Blatchley, et al., “Far UV-C Radiation: Current State-of Knowledge” (International 417
Ultraviolet Association Task Force). 418
5. M. Buonanno, D. Welch, I. Shuryak, D. J. Brenner, Far-UVC light (222 nm) efficiently and 419
safely inactivates airborne human coronaviruses. Sci Rep 10, 10285 (2020). 420
6. D. B. Collins, D. K. Farmer, Unintended Consequences of Air Cleaning Chemistry. 421
Environ. Sci. Technol. 55, 12172–12179 (2021). 422
7. Q. Ye, et al., Real-Time Laboratory Measurements of VOC Emissions, Removal Rates, and 423
Byproduct Formation from Consumer-Grade Oxidation-Based Air Cleaners. Environ. Sci. 424
Technol. Lett. 8, 1020–1025 (2021). 425
8. M. C. Turner, et al., Long-Term Ozone Exposure and Mortality in a Large Prospective 426
Study. Am J Respir Crit Care Med 193, 1134–1142 (2016). 427
9. J. Burkholder, et al., “Chemical Kinetics and Photochemical Data for Use in Atmospheric 428
Studies” (NASA Jet Propulsion Laboratory, 2020). 429
10. W. W. Nazaroff, C. J. Weschler, Indoor ozone: Concentrations and influencing factors. 430
Indoor Air 32 (2022). 431
11. J. G. Calvert, The Mechanisms of Atmospheric Oxidation of the Alkenes (Oxford University 432
Press, 2000). 433
12. A. Wisthaler, C. J. Weschler, Reactions of ozone with human skin lipids: Sources of 434
carbonyls, dicarbonyls, and hydroxycarbonyls in indoor air. Proceedings of the National 435
Academy of Sciences 107, 6568–6575 (2010). 436
13. C. M. F. Rosales, et al., Chemistry and human exposure implications of secondary organic 437
aerosol production from indoor terpene ozonolysis. Science Advances 8, eabj9156 (2022). 438
14. US EPA, Dose-Response Assessment for Assessing Health Risks Associated with Exposure 439
to Hazardous Air Pollutants (2021) (May 2, 2023). 440
15. D. W. Dockery, et al., An Association between Air Pollution and Mortality in Six U.S. 441
Cities. New England Journal of Medicine 329, 1753–1759 (1993). 442

16
16. R. Burnett, et al., Global estimates of mortality associated with long-term exposure to 443
outdoor fine particulate matter. Proceedings of the National Academy of Sciences 115, 444
9592–9597 (2018). 445
17. H. O. T. Pye, C. K. Ward-Caviness, B. N. Murphy, K. W. Appel, K. M. Seltzer, Secondary 446
organic aerosol association with cardiorespiratory disease mortality in the United States. 447
Nat Commun 12, 7215 (2021). 448
18. N. M. Donahue, G. T. Drozd, S. A. Epstein, A. A. Presto, J. H. Kroll, Adventures in 449
ozoneland: down the rabbit-hole. Phys. Chem. Chem. Phys. 13, 10848–10857 (2011). 450
19. H. Levy, Normal Atmosphere: Large Radical and Formaldehyde Concentrations Predicted. 451
Science 173, 141–143 (1971). 452
20. E. Bender, Disinfecting the air with far-ultraviolet light. Nature 610, S46–S47 (2022). 453
21. Z. Peng, et al., Significant Production of Ozone from Germicidal UV Lights at 222 nm. 454
2023.05.13.23289946 (2023). 455
22. M. F. Link, A. Shore, B. H. Hamdani, D. Poppendiek, Ozone Generation from a Germicidal 456
Ultraviolet Lamp with Peak Emission at 222 nm (2023). 457
23. Z. Peng, S. L. Miller, J. L. Jimenez, Model Evaluation of Secondary Chemistry due to 458
Disinfection of Indoor Air with Germicidal Ultraviolet Lamps. Environ. Sci. Technol. Lett. 459
10, 6–13 (2023). 460
24. D. J. Stewart, et al., The kinetics of the gas-phase reactions of selected monoterpenes and 461
cyclo-alkenes with ozone and the NO3 radical. Atmospheric Environment 70, 227–235 462
(2013). 463
25. S. M. Aschmann, J. Arey, R. Atkinson, Kinetics and Products of the Reactions of OH 464
Radicals with Cyclohexene, 1-Methyl-1-cyclohexene, cis -Cyclooctene, and cis -465
Cyclodecene. J. Phys. Chem. A 116, 9507–9515 (2012). 466
26. B. D’Anna, Ø. Andresen, Z. Gefen, C. J. Nielsen, Kinetic study of OH and NO 3 radical 467
reactions with 14 aliphatic aldehydes. Physical Chemistry Chemical Physics 3, 3057–3063 468
(2001). 469
27. A. Hansel, W. Scholz, B. Mentler, L. Fischer, T. Berndt, Detection of RO2 radicals and 470
other products from cyclohexene ozonolysis with NH4+ and acetate chemical ionization 471
mass spectrometry. Atmospheric Environment 186, 248–255 (2018). 472
28. S. M. Aschmann, E. C. Tuazon, J. Arey, R. Atkinson, Products of the Gas-Phase Reaction 473
of O3 with Cyclohexene. J. Phys. Chem. A 107, 2247–2255 (2003). 474
29. S. Barua, S. Iyer, A. Kumar, P. Seal, M. Rissanen, An aldehyde as a rapid source of 475
secondary aerosol precursors: Theoretical and experimental study of hexanal autoxidation. 476
EGUsphere, 1–24 (2023). 477

17
30. B. C. Singer, et al., Indoor secondary pollutants from cleaning product and air freshener use 478
in the presence of ozone. Atmospheric Environment 40, 6696–6710 (2006). 479
31. T. Wainman, J. Zhang, C. J. Weschler, P. J. Lioy, Ozone and limonene in indoor air: a 480
source of submicron particle exposure. Environmental Health Perspectives 108, 1139–1145 481
(2000). 482
32. B. Ma, et al., Inactivation of Coronaviruses and Phage Phi6 from Irradiation across UVC 483
Wavelengths. Environ. Sci. Technol. Lett. 8, 425–430 (2021). 484
33. D. J. Price, et al., Budgets of Organic Carbon Composition and Oxidation in Indoor Air. 485
Environ. Sci. Technol. 53, 13053–13063 (2019). 486
34. J. M. Mattila, et al., Contrasting Chemical Complexity and the Reactive Organic Carbon 487
Budget of Indoor and Outdoor Air. Environ. Sci. Technol. 56, 109–118 (2022). 488
35. T. Grøntoft, M. R. Raychaudhuri, Compilation of tables of surface deposition velocities for 489
O3, NO2 and SO2 to a range of indoor surfaces. Atmospheric Environment 38, 533–544 490
(2004). 491
36. International Commission on Non-Ionizing Radiation Protection (ICNIRP), Guidelines on 492
limits of exposure to ultraviolet radiation of wavelengths between 180 nm and 400 nm 493
(incoherent optical radiation). Health Phys 87, 171–186 (2004). 494
37. American Conference of Governmental Industrial Hygienists, 2023 Threshold Limit Values 495
(TLVs) and Biological Exposure Indices (BEIs) (2023) (May 3, 2023). 496
38. E. Eadie, et al., Far-UVC (222 nm) efficiently inactivates an airborne pathogen in a room-497
sized chamber. Sci Rep 12, 4373 (2022). 498
39. , ASHRAE/ANSI Standard 62.1-2022. Ventilation and Acceptable Indoor Air Quality 499
(2022). 500
40. C. J. Weschler, Ozone’s Impact on Public Health: Contributions from Indoor Exposures to 501
Ozone and Products of Ozone-Initiated Chemistry. Environ Health Perspect 114, 1489–502
1496 (2006). 503
41. H. Wang, G. Morrison, Ozone-surface reactions in five homes: surface reaction 504
probabilities, aldehyde yields, and trends. Indoor Air 20, 224–234 (2010). 505
42. C. J. Weschler, H. C. Shields, Production of the Hydroxyl Radical in Indoor Air. Environ. 506
Sci. Technol. 30, 3250–3258 (1996). 507
43. C. J. Weschler, H. C. Shields, Measurements of the Hydroxyl Radical in a Manipulated but 508
Realistic Indoor Environment. Environ. Sci. Technol. 31, 3719–3722 (1997). 509
44. N. Carslaw, A new detailed chemical model for indoor air pollution. Atmospheric 510
Environment 41, 1164–1179 (2007). 511

18
45. M. S. Waring, J. R. Wells, Volatile organic compound conversion by ozone, hydroxyl 512
radicals, and nitrate radicals in residential indoor air: Magnitudes and impacts of oxidant 513
sources. Atmospheric Environment 106, 382–391 (2015). 514
46. N. Carslaw, L. Fletcher, D. Heard, T. Ingham, H. Walker, Significant OH production under 515
surface cleaning and air cleaning conditions: Impact on indoor air quality. Indoor Air 27, 516
1091–1100 (2017). 517
47. S. Gligorovski, C. J. Weschler, The Oxidative Capacity of Indoor Atmospheres. Environ. 518
Sci. Technol. 47, 13905–13906 (2013). 519
48. E.-A. Fiorentino, et al., Measurements and Modelling of OH and Peroxy Radicals in an 520
Indoor Environment Under Different Light Conditions and VOC Levels. Atmospheric 521
Environment 292, 119398 (2023). 522
49. M. Mendez, et al., Identification of the major HOx radical pathways in an indoor air 523
environment. Indoor Air 27, 434–442 (2017). 524
50. E. Gómez Alvarez, et al., Unexpectedly high indoor hydroxyl radical concentrations 525
associated with nitrous acid. Proceedings of the National Academy of Sciences 110, 13294–526
13299 (2013). 527
51. S. F. Kowal, S. R. Allen, T. F. Kahan, Wavelength-Resolved Photon Fluxes of Indoor Light 528
Sources: Implications for HOx Production. Environ. Sci. Technol. 51, 10423–10430 (2017). 529
52. E. Reidy, et al., Measurements of Hydroxyl Radical Concentrations during Indoor Cooking 530
Events: Evidence of an Unmeasured Photolytic Source of Radicals. Environ. Sci. Technol. 531
57, 896–908 (2023). 532
53. Ushio, Inc., Care222 Technical Specification Sheet (May 3, 2023). 533
54. J. Krechmer, et al., Evaluation of a New Reagent-Ion Source and Focusing Ion–Molecule 534
Reactor for Use in Proton-Transfer-Reaction Mass Spectrometry. Anal. Chem. 90, 12011–535
12018 (2018). 536
55. A. Zaytsev, et al., Using collision-induced dissociation to constrain sensitivity of ammonia 537
chemical ionization mass spectrometry (NH4+ CIMS) to oxygenated volatile organic 538
compounds. Atmos. Meas. Tech. 12, 1861–1870 (2019). 539
56. P. F. DeCarlo, et al., Field-Deployable, High-Resolution, Time-of-Flight Aerosol Mass 540
Spectrometer. Anal. Chem. 78, 8281–8289 (2006). 541
57. S. M. Saunders, M. E. Jenkin, R. G. Derwent, M. J. Pilling, Protocol for the development of 542
the Master Chemical Mechanism, MCM v3 (Part A): tropospheric degradation of non-543
aromatic volatile organic compounds. Atmospheric Chemistry and Physics 3, 161–180 544
(2003). 545
546