Particulate Matter and Total Volatile Organic Compound Emissions Following Surface Cleaning: Comparison of Cleaning Agents and Locations

Abstract

Cleaning activities are essential for maintaining hygiene in indoor environments but can significantly influence indoor air quality (IAQ). We investigated emissions of volatile organic compounds (VOCs) and particulate matter (PM) during cleaning events across various indoor settings including two laboratories, an office, and a residential bathroom, with room volumes ranging from 22 to 206 m³ and air changes rates (ACR) of 0.85–9.14 h^-1. Four cleaning solutions with different active ingredients were evaluated: quaternary ammonium compounds (quats), hydrogen peroxide (H₂O₂), sodium hypochlorite (bleach), and thymol. Cleaning increased PM_2.5 by 0.7–14.5 μg m^–3, depending on location and cleaning solution, with quats generally yielding the greatest increases. Measured total volatile organic compound (TVOC) mixing ratios also increased following cleaning by 10–104 ppbv, with the exception of experiments performed using thymol. We note that sensors such as the photoionization detector (PID) used in this work do not provide quantitative TVOC measurements. In general, greater emissions of PM_2.5 and TVOCs were observed in locations with lower ACR. We also measured PM_2.5 in a lobby, elevator, and public bathroom in a hotel with a number of COVID-positive occupants during routine surface disinfection using a quats-based disinfectant: increases of 5.5–14.2 μg m^–3 were observed. This study demonstrates that emissions other than active ingredients can affect IAQ during surface cleaning, and provides information that may help mitigate harmful effects. It also provides insight into the use and limitations of low-cost sensors (LCS) in determining IAQ impacts from cleaning.

1. Introduction

Cleaning is a crucial activity in indoor environments, where people spend most of their time. It plays an important role in enhancing aesthetic appeal and reducing the risk of exposure to harmful pathogens that may reside on surfaces. However, cleaning also introduces a range of chemical species into the air, with the identities and concentrations of individual chemicals depending on the composition of the cleaner, the cleaning methods employed, and the physical properties of the environment in which the cleaning is performed. ,

Commercial cleaning solutions often have an extensive list of ingredients that typically includes water, fragrances, preservatives, and surfactants in addition to the active ingredient, which targets and eliminates dirt, odors, or harmful microorganisms. Different active ingredients have different toxicological profiles and established exposure limits. For example, chlorine gas (Cl₂) released from sodium hypochlorite bleach (bleach) has a recommended 15 min exposure limit of 0.5 ppmv (parts per million by volume), as established by the National Institute for Occupational Safety and Health (NIOSH) due to its highly corrosive effects on moist tissues, including eyes, skin, and the epithelial lining of the upper respiratory tract. Similarly, hydrogen peroxide (H₂O₂) is an irritant to the respiratory system and eyes, with NIOSH recommending a 15 min exposure limit of 1 ppmv. Quaternary ammonium compounds (quats) are the active compounds in many “all-purpose” disinfectants. They have become a concern in recent years due to their widespread use and potential to cause skin irritation and respiratory issues, although exposure limits have not been established. − Botanical disinfectants, commonly identified as “green/natural” alternatives, have also gained popularity in recent years. Thymol-based cleaners, derived from thyme oil, are among the most popular. They exhibit potent antimicrobial properties by disrupting microbial cell walls, effectively inactivating pathogens. , Specific exposure risks from thymol have not been identified, to our knowledge, though aerosol formation from these cleaners has been noted as a concern.

Several recent studies have reported time-resolved concentrations of active ingredientsand in some cases their reaction productsin indoor spaces following the use of cleaning products. Bleach, for example, emits hypochlorous acid (HOCl) and Cl₂ with peak levels reaching hundreds of parts per billion by volume mixing ratios (ppbv) during cleaning, , and up to tens of ppmv near the cleaned surface. These species can further react indoors to produce chlorinated organics, such as chloroform, carbon tetrachloride, and chloramines, − and even ultrafine particles under some conditions. Hydrogen peroxide-based cleaners emit H₂O₂, often at tens to hundreds of ppbv, with concentrations depending on the H₂O₂ concentration, product application method, ventilation rates, and surface-area-to-volume ratios of the environment. , Hydrogen peroxide, Cl₂, HOCl, and other chlorinated species emitted from cleaners can also undergo photolysis indoors, producing OH and Cl radicals that can lead to further reactions and changes to indoor air quality (IAQ). ,,−

Nonactive ingredients are also emitted during cleaning, which may affect IAQ. For instance, laboratory and field measurements show that many cleaning agents temporarily increase VOC levels during and following cleaning. ,, Once airborne, VOCs can undergo gas-phase and multiphase reactions, producing harmful byproducts such as formaldehyde, radicals, and particulate matter (PM). − Particulate matter can also be directly emitted from spray bottles during the application of cleaning solutions. Exposure to PM is linked to adverse health outcomes, including cardiovascular disease, chronic obstructive pulmonary disease, and asthma, and is considered a major contributor to global mortality. It is worth noting that most of the current knowledge on the effects of PM on human health is based on typical outdoor pollution. Given the different formation and growth mechanisms particles undergo in indoor environments (e.g., direct and indirect emissions from cleaning and cooking activities), especially during cleaning events, the toxicity of particles formed indoors remains poorly understood.

In addition to their role in PM formation, VOCs can have direct adverse health effects, and concentrations of many VOCs are higher indoors that outdoors, especially during the use of personal care products, paint, and cleaners. ,− The effects of cleaners on indoor VOC levels depend on the concentrations and identities of additives, such as perfumes. In the case of botanical-based cleaners and quats, the active ingredients are themselves organic molecules. Botanicals have been reported to emit large amounts of VOCs, such as monoterpenes and monoterpenoids (e.g., limonene, α-pinene, thymol), which readily react with indoor oxidants (e.g., ozone) to form PM and gas-phase byproducts. ,,− Quats-based cleaners have also been reported to emit large quantities of VOCs, though these were primarily linked to additives such as ethanol and byproducts of the quats synthesis process. − Due to their permanent ionization, quats are nonvolatile, leading to their adsorption onto indoor surfaces and solid particles, and can be reaerosolized through mechanical disturbances such as sweeping.

While many studies have characterized the chemical makeup of cleaners’ emissions, less attention has been paid to the influence of environmental factors, such as ventilation rates, in shaping indoor air quality during cleaning events. Low ventilation rates have been linked to the accumulation of pollutants, whereas higher ventilation can dilute emissions but may also introduce outdoor pollutants indoors. These interactions became especially significant during the COVID-19 pandemic, when rigorous cleaning routines and large disinfectant application volumes were widely adopted to minimize viral transmission. The intensified cleaning practices significantly increased the release of chemical pollutants into indoor environments, and subsequently outdoors when ventilated. ,,

A growing public interest is emerging for monitoring IAQ using low-cost air quality monitors. Many of these are marketed as tools for measuring key IAQ parameters such as total VOC concentrations (TVOC) and PM. While these devices often lack the precision and accuracy of research-grade instruments, they can provide valuable real-time insights into changes in IAQ that consumers can relate to their behaviors. For example, low-cost monitors can identify temporal changes in TVOC or PM concentrations during cleaning activities, even though the quantitative measurements may not be entirely reliable. This qualitative information is valuable for identifying trends and encouraging behavioral changes, such as improving ventilation during cleaning or switching to less polluting products. Further, these monitors can be deployed in indoor locations such as residences that are inaccessible to research-grade instruments, and their low costs allow for wide proliferation. In this study, we used primarily low-cost monitors to investigate trends in TVOC and PM levels during cleaning events using a range of cleaning solutions in different environments, including during real-life surface disinfection in a hotel. The primary goal of this work is to provide insight into general effects of cleaning solution composition and environmental variables such as ventilation rates on IAQ during and following the use of common cleaners and disinfectants.

2. Methods

2.1. Locations

Experiments were performed in five locations in Saskatoon, Saskatchewan, and Toronto, Ontario (Canada). Four of the locations (two chemistry laboratories, a residential bathroom, and a university office) were used for controlled cleaning experiments with a range of cleaning solutions. Pictures of these locations, apart from the Toronto lab, can be found in Figure S1. These locations were selected for their range of air change rates (ACRs), with means of 0.85 to 9.14 h^–1. Field measurements were performed in the fifth location; a hotel in Saskatoon that housed a large number of COVID-positive people. Measurements at this location were performed on a single day (December 17, 2020) while a professional cleaning company disinfected surfaces in the lobby, a public bathroom, and an elevator, using a quats-based cleaning solution. Further information on each location is provided below.

2.2. Cleaning Solutions

The cleaning experiments were carried out using four types of spray cleaners: botanical disinfectant (0.05% thymol), general purpose disinfectant (quaternary ammonium compounds, 0.105% dimethyl benzyl ammonium chloride and 0.105% dimethyl ethyl benzyl ammonium chloride), bleach (2.0% sodium hypochlorite), and H₂O₂ (1.0% hydrogen peroxide). Hereafter, they will be referred as thymol, quats, bleach, and H₂O₂, respectively. Ingredient lists for each cleaning agent are provided in Table S1.

2.3. Instrumentation

Table lists the instruments used in this study along with the locations at which they were employed. The experimental setup consisted of a combination of seven low-cost sensors and two research-grade instruments. The low-cost sensors included the AQMesh multisensor pod (Ambilabs, United States), AirBeam (versions 2 and 3, HabitatMap, United States), Plantower PMS5003 (Plantower Technology, China), Ambi-VOC photoionization detector (PID) equipped with a 10.6 eV UV lamp (Ambilabs, United States), Atmocube multisensor (Atmotech Inc., United States), and Senseair K30 (Senseair, Sweden). The research-grade instrumentation included a 2-wavelength integrating nephelometer (2WIN, Ambilabs, United States) and a Picarro G2401 CO₂ analyzer (Picarro Inc., United States). The readings provided by all analyzers, apart from the AQMesh, were averaged to 1 min resolution prior to data analysis. The AQMesh data set was kept at 15 min resolution.

1. Instrumentation Used in Each Indoor Setting Assessed in This Work, Their Operating Principle, and Time Resolution.

PM	TVOC	CO2
Nephelometer	AmbiLab AmbiVOC PID , , , (photoionization; 2 s resolution)	Picarro , (cavity ring-down spectroscopy; 2 s resolution)
(light scattering; 0.1–10 μm; 30 s resolution)
AQMesh		Atmocube (nondispersive infrared, 1 min resolution)
(optical particle counter; PM2.5; 15 min resolution)
AirBeam ,		Rasberry Pi with K30 CO2 sensor (nondispersive infrared, 1 min resolution)
(optical scattering; PM2.5; 1 s resolution)
Raspberry Pi with Plantower
(optical particle counter; PM1, PM2.5, PM10; 1 s resolution)

^aSaskatoon lab.

^bSaskatoon office.

^cSaskatoon residential bathroom.

^dSaskatoon hotel.

^eToronto lab.

All instruments were factory calibrated. No corrections were applied to the data, which was obtained in units of aerosol concentration (μg m^–3) or mixing ratios (gases, ppbv or ppmv), with the exception of the 2-WIN nephelometer, which reports scattering coefficients (σ_sp) rather than concentration. We followed the manufacturer’s recommended method for converting σ_sp to PM_2.5 concentration. Full details of the conversion method and validation of the PM_2.5 measurements are provided in Section S3 of the SI.

We also used σ_sp at two wavelengths (450 and 635 nm) to calculate time-resolved scattering Ångström exponents (SAE) using eq . This parameter characterizes the wavelength dependence of light scattering by PM; its value is strongly influenced by the particle size distribution. In general, larger, coarse-mode particles scatter light with less wavelength dependence, resulting in lower SAE values, whereas smaller, fine-mode particles produce a steeper wavelength dependence and thus higher SAE values. , Consequently, SAE provides additional information about the dominant size mode: an increase in PM concentrations accompanied by an increase in SAE often indicates secondary formation (by nucleation, growth, and coagulation) of small, fine particles, while a decrease in SAE implies that the mass increase is due mainly to direct emission of larger particles.

$$\mathrm{SAE}=-\frac{\ln \left(\frac{{\sigma }_{\mathrm{sp}}450}{{\sigma }_{\mathrm{sp}}635}\right)}{\ln \left(\frac{450}{635}\right)}$$ 1

We note that TVOC concentrations are by their nature not quantitative, due to varying detector response factors to different molecules, and a lack of sensitivity to some classes of molecules. The Ambilabs PID used in this work is calibrated using isobutylene; all signal is converted to mixing ratio equivalents of isobutylene. VOCs we might expect to observe following cleaning have response factors that can vary by up to an order of magnitude compared to that of isobutylene. Further, this PID uses 10.6 eV light to effect photoionization; it will be insensitive to molecules with higher ionization energies. Most VOCs common indoors have ionization energies below this threshold. Exceptions include several halogenated species which could be formed following reactions of VOCs with HOCl, and some small alkanes and organic acids such as acetic acid. We therefore do not consider the TVOC concentrations we report to be quantitative; they are used for comparative purposes. We use measured TVOC values in alignment with recommendations from Salthammer: nonhealth related screening and statistical evaluation of IAQ, and demonstration of temporal trends.

2.4. Air Change Rate Measurements

Air change rate (ACR) was determined by distributing several dry ice pellets across the floor of a room, vacating the room, and recording CO₂ mixing ratios at 3 s intervals until CO₂ levels returned to background levels (0.4–4 h, depending on the location). An exponential regression (eq ) was applied to the data to determine ACR.

$$C_t=C_0{\mathrm{e}}^{-\mathit{kt}}+C_{\mathrm{b}}$$ 2

where C _t is the CO₂ mixing ratio after time t, C₀ is the initial mixing ratio prior to the observed decay, C _b is the background mixing ratio, and k is ACR in units of h^–1.

We also developed an alternative method to estimate ACR (full details are provided in Figure S3 and Table S2). This alternative method was employed in the hotel because dry ice could not be safely used in enclosed occupied spaces. A full description of the method is provided in the SI. In short, a bottle of nail polish remover with acetone as the active ingredient was opened in the room for 1 min, then both the bottle and the door were closed, and TVOC mixing ratios were monitored until they returned to baseline. We applied a correction factor to convert the measured TVOC decay rate constant k _TVOC to ACR as described in the SI. We refer to this as the NPR procedure.

2.5. Hotel Cleaning

Real-life COVID disinfection procedures were observed at a hotel in Saskatoon, Saskatchewan, Canada, that housed a number of residents with COVID. Professional cleaners, equipped in full personal protective equipment, including respirators and full-face masks, performed intensive disinfection of public surfaces in the hotel multiple times a day using a quats-based disinfectant. Surface disinfection was performed by spraying a cloth with the disinfectant solution, then wiping the surface. The applied solution was not removed from the surface. This procedure differs from the manufacturer’s recommendation of spraying the disinfectant on a surface and wiping it off after 5–10 min. The cleaning company used this alternative method to minimize exposure to emissions to the employees performing the cleaning. Emissions from surface disinfection of common-touch surfaces in the hotel lobby, an elevator, and a bathroom were monitored with the AirBeam2 (Table ) on one occasion in December 2020. Room volume and surface area cleaned can both affect observed concentrations of emitted species; Table provides information relevant to these variables.

2. General Information on the Hotel Cleaning Events.

Location	Surfaces cleaned	Cleaning time (min)	Room volume (m3)
Lobby	door handles, tables, hand sanitizer pumps	2	411
Elevator	button pad, railings, walls	3	7
Bathroom	floor, countertop, door handles, and sink and toilet surfaces including faucets and flush levers	15	44

2.6. Controlled Cleaning Experiments

Cleaning was performed in a method consistent with manufacturer suggestions. It consisted of spraying enough cleaning solution to fully wet a target area of the floor (Table ), followed by immediately wiping off excess solution. The entire process took 2–3 min, consistent with the cleaning procedures used in two of three locations in the hotel (Table ). For a select set of experiments in the bathroom, quat disinfection was performed as per the manufacturer’s directions, with a 10 min dwell time rather than immediate removal.

3. Description of Controlled Cleaning Experiments in the Different Indoor Settings.

Location	Target area (m2)	Type of surface cleaned	Distance from instruments inlet (cm)	Volume of cleaning solution applied (mL)	ACR (h–1)	Room volume (m3)	Dates of experiments
Saskatoon lab	3.6	Linoleum	130–250	24–44	9.14 ± 2.72	206.5	Mar.–Aug. 2023
Saskatoon office	3.6	Linoleum	210–270	25–49	5.85 ± 0.82	42.5	Sep. 2021
Toronto lab	4.1	Tile	100–130	50–60	0.85	58.9	May–July 2021
Bathroom	1.9	Tile	120–170	∼30	0.92 ± 0.62	22.1	Jan.–Aug. 2021

^aTwo adjacent separate target areas (each 4.1 m²) were alternately cleaned.

3. Results and Discussion

3.1. Hotel Cleaning Changes Particulate Matter Loadings

Cleaning events were carried out in different environments within a hotel in Saskatoon, Canada, during the COVID-19 pandemic to investigate exposure of cleaning personnel to PM_2.5 during surface disinfection. TVOCs were also measured, but the PID sensor’s sensitivity to vibrations made the results unusable due to constant foot traffic. PM_2.5 levels in the hotel were highly variable, even in the absence of cleaning. Despite this, concentrations increased following each monitored disinfection event (Figure ), with observed increases of 5.5 μg m^–3 (lobby), 14.2 μg m^–3 (elevator), and 13.9 μg m^–3 (bathroom) respectively. PM concentrations decreased rapidly in the lobby and elevator following cleaning, while a much slower decay was observed in the bathroom. These observations can be explained by several factors. First, larger surface areas were washed in the elevator and bathroom than in the lobby. Second, the volumes of the elevator and bathroom were much smaller than that of the lobby (Table ). We note that ACR was quite high in the bathroom (∼9.6 h^–1, as estimated using the NPR procedure). We were unable to measure k _TVOC in the lobby and elevator, so we do not know ACR in those locations.

1.

Fine particle mass loadings observed following the use of quats for surface cleaning in different rooms in the hotel. Shaded light gray regions represent time periods when cleaning was taking place. Data shown are 10 s averages (from data collected with 1 s time resolution). PM levels during cleaning periods showed statistically significant differences compared to unperturbed periods (p < 0.05).

Following the measurements in the hotel, we performed a series of controlled experiments in four locations to test the effects of specific environmental conditions (specifically ACR; Section ) and to investigate differences in PM and VOC emissions between different common disinfectants.

3.2. Changes in TVOC and PM Concentrations during Controlled Cleaning Events

In the controlled experiments, we measured TVOC and PM_2.5 emissions from the four selected cleaning solutions in four test environments. Figure shows average changes in PM_2.5 and TVOC concentrations following cleaning events in the Saskatoon laboratory. High variability was observed between replicates using the same cleaner, with coefficients of variation ranging 48%–95% for peak PM_2.5 and 42%–71% for peak TVOC across all tested cleaners. This variability, which is common with experiments in indoor environments, may be attributed to the spray bottles delivering inconsistent amounts of solution to the surface, as well as to the accumulation of chemical species on the surface of the target cleaning area. Differing sensitivity of the PID to different VOCs may also have contributed to the uncertainty in observed TVOC levels. To overcome potential biases arising from the high heterogeneity in our data set and to improve our understanding of the trends in both particle and VOC emission processes, we performed multiple experimental replicates when possible.

2.

Average change in PM_2.5 and TVOC concentration (solid-colored traces) relative to background levels (dashed lines) following cleaning events in the Saskatoon lab measured by nephelometry (PM) and PID (TVOC). Shaded regions along the traces represent the standard error of the mean, and shaded light gray regions represent time periods when cleaning was taking place. The number of replicate experiments is provided in parentheses.

Increases in PM_2.5 were observed immediately following the use of each cleaner, with maximum PM_2.5 concentrations observed 2–4 min, on average, after cleaning started (Figure a). TVOC mixing ratios measured by our sensing packages increased following the use of all cleaners except for thymol (Figure b). A small but reproducible decrease in observed TVOC mixing ratios followed the use of thymol in these experiments. This decrease was not observed in the office or bathroom; mixing ratios did not deviate significantly from background levels in those environments. We do not know what is responsible for this observation. PID sensors have poor sensitivity toward some VOCs, as discussed in the Methods section. , However, this does not explain the observed decrease in the lab. It is possible that background VOCs partitioned to the floor following thymol application, or to particles and droplets emitted during cleaning. There is currently insufficient data to speculate further.

Further, the TVOC time series following bleach cleaning was bimodal. This is consistent with formation of oxidation products of monoterpenes and, more likely, isoprene, as observed by Mattila et al. In that work, the authors attributed the formation of oxidized VOCs (OVOCs) to heterogeneous oxidation chemistry, with OVOCs production occurring before the decrease of total VOCs to background levels.

TVOC levels did not increase immediately following application of the cleaners, unlike PM_2.5; mixing ratios remained at background levels for 1–3 min following application. This lag is consistent with TVOCs partitioning from the floor (or from the aqueous solution on the floor) to the air following cleaning, as we have previously observed for H₂O₂ and HOCl following the application of H₂O₂- and bleach-based cleaners. ,, The immediate increase of PM_2.5, conversely, may point to direct emission of aerosols from the spray bottles.

To elucidate the mechanisms driving increases in observed PM concentrations (specifically, direct emissions from spray bottles versus chemical formation), we analyzed the SAE during cleaning events with the dual wavelength nephelometer. Figure shows PM_2.5 concentration and SAE following thymol cleaning in the Saskatoon laboratory.

3.

Temporal variations of PM_2.5 and scattering Angstrom exponent following thymol cleaning in the Saskatoon lab (measured by nephelometry). Shaded regions in light gray represent time of cleaning.

Our results showed a mean decrease of 0.88 in SAE following cleaning, which, when considered together with the concurrent increase in PM_2.5 concentration, suggests that the particles are primarily generated by mechanical processes rather than through secondary chemical formation. Additionally, the fact that the mean decrease in SAE following cleaning was statistically indistinguishable among the different cleaning solutions (within experimental uncertainty at the 95% confidence level, Table ) supports the interpretation that the PM_2.5 increase is driven predominantly by direct emissions. In support of this, measurements in the Toronto lab provided information on PM₁ and PM₁₀ as well as PM_2.5; the ratio of PM₁₀ to PM₁ concentrations increased compared to background levels following the use of each cleaner (Table S3). Our conclusion aligns with a recent study that reported that PM emitted during cleaning activities in a commercial kitchen was primarily composed of mechanically generated particles, as opposed to new particle formation and growth.

4. Average Change in Scattering Ångström Exponent (SAE) Values for Cleaning Events in the Saskatoon Lab.

Cleaning solution	ΔSAE
Quats (N = 5)	–0.67 ± 0.21
Bleach (N = 2)	–1.1 ± 0.5
H2O2 (N = 4)	–0.79 ± 0.32
Thymol (N = 3)	–1.1 ± 0.3

Figure and Tables S4 and S5 detail the maximum increases of PM_2.5 and TVOCs following cleaning. As shown in Figure a, the increase in PM_2.5 was generally highest in the bathroom and in the Toronto lab, both of which had lower ACR (0.92 and 0.85 h^–1, respectively) compared to the Saskatoon lab and office (ACR = 9.14 and 5.85 h^–1, respectively). On average, the sites with limited ventilation exhibited 146%–346% higher fine particle loadings when compared to environments with higher ACR. Similarly, increases in TVOC levels following cleaning were also generally highest in the bathroom (Figure b), with mixing ratios ∼7-fold greater than those in higher ACR locations; TVOC concentrations were not recorded in the Toronto lab. In general, increasing ACR leads to more rapid removal of species from indoors, consistent with our observations in this work. However, complex effects have been reported. For example, we demonstrated that increasing ACR by opening patio doors in a residence decreased indoor NO levels, but increased indoor ozone (O₃) and NO₂* (the sum of NO₂ and HONO) mixing ratios. Increased indoor O₃ mixing ratios were easily explained by infiltration from outdoors, where O₃ levels are often much higher than indoors. The increase in NO₂* was not anticipated, as indoor concentrations exceeded outdoor concentrations in that study. The increase was determined to be due to NO₂ formation from reactions between NO and O₃. While we did not observe evidence of chemistry following the application of cleaners in this study, it is a possibility that should be considered, even in buildings with high ACR, due to the positive correlation between ACR and O₃ mixing ratios. , We note that the area cleaned in the residential bathroom was approximately half the areas cleaned in the other locations, although a similar volume of cleaning solution was applied. We have previously reported that H₂O₂ mixing ratios depended on surface area cleaned; in that work, decreasing the surface area by 50% while applying the same volume reduced measured mixing ratios by 20%. This observation highlights the importance of ACR to emissions from cleaning.

4.

Box-and-whiskers plot highlighting average increases in (a) PM_2.5 and (b) TVOC levels after cleaning activities using all evaluated cleaners in the bathroom, office and Saskatoon and Toronto laboratories. Numbers in parentheses in the legends provide ACR and room volume. Peak concentrations were determined as the maximum 1 min averaged maximum concentration observed following cleaning events. Box plots display quartile values and medians. Whiskers extend to the minimum and maximum values within 1.5 times the interquartile range. Black diamonds represent the mean value for each condition, and gray circles represent outliers.

Surface area-to-volume ratio (S/V) can influence indoor mixing ratios, with higher ratios often resulting in faster loss and lower observed mixing ratios due to surface deposition and reactions. , Calculated lower limits for S/V (i.e., not considering rooms’ contents) ranged from 1.3 to 1.7 m^–1 for the university spaces, and 2.2 m^–1 for the bathroom. This range is typical for empty indoor spaces. Manuja et al. reported that room contents increase S/V by 78%, on average. We may expect the actual S/V in the laboratories to average ∼2.7 m^–1 and the bathroom to be ∼3.8 m^–1. We do not believe surface uptake was a major contributor to the observed TVOC or PM_2.5 concentrations and decay kinetics in the lab and office, given that k _TVOC in both locations (as shown in Table ) was lower than ACR (though this difference was only statistically significant in the office). In the bathroom, k _TVOC was on average 90% larger than ACR. This difference was not statistically significant, but it may suggest a VOC sink other than air exchange, such as surface partitioning. This speculation is supported by the larger S/V in the bathroom compared to the lab and office.

5. Air Change Rates and TVOC Decay Rate Constants at Three Experimental Sites.

Cleaner	Bathroom (h–1)	Saskatoon office (h–1)	Saskatoon lab (h–1)
Quats	1.78 (N = 1)	2.39 ± 2.61 (N = 5)	7.85 ± 2.79 (N = 6)
Bleach	2.05 ± 1.18 (N = 2)	2.92 ± 0.47 (N = 5)	5.60 ± 4.62 (N = 5)
H2O2	1.35 ± 0.42 (N = 5)	3.73 ± 0.61 (N = 2)	11.2 ± 3.4 (N = 4)
ACR	0.92 ± 0.62 (N = 9)	5.85 ± 0.82 (N = 4)	9.14 ± 2.72 (N = 11)

^aStatistically significant difference (p-value <0.05) between the TVOC decay rate constant mean and ACR.

Another factor that we have shown to be important in chemical emissions is the composition of the surfaces cleaned. We demonstrated that H₂O₂ peak emissions and decay kinetics were different on polished wood and stone surfaces, and Poppendieck et al. reported variable H₂O₂ uptake coefficients to different common indoor surfaces. However, the effects of room volume and ACR were large enough to dominate any additional effects we might expect from factors such as S/V and surface composition. This is reflected in the trends in Figure , with observed increases in PM and TVOC concentrations inversely correlated to ACR.

With respect to emissions from individual cleaners, PM_2.5 emissions were generally highest from quats, though large increases were also observed from H₂O₂ in the Toronto lab (up to 13.4 times higher than in the other locations). In the bathroom, increases in TVOC following bleach and H₂O₂ use were 2.2–4.3 larger than increases following quats use. While the same trend was observed in the Saskatoon lab and office, the differences in TVOC levels between bleach and H₂O₂ compared to quats was not statistically significant.

In experiments performed in the bathroom with 10 min dwell time of quats, mean PM_2.5 and TVOC levels increased by 22.2 ± 8.4 μg m^–3 and 198 ± 84 ppbv, respectively. This corresponded to a 68% increase in PM_2.5 levels compared to shorter dwell times, and a factor of ∼8 increase in TVOC concentrations. To our knowledge, there are no reports in the literature describing the influence of solution dwell times on both PM and VOCs levels, but previous works from our groups suggest that longer dwell times can increase levels of active cleaning ingredients (H₂O₂ and HOCl). ,,

3.3. Ventilation Effects

Table shows rate constants for the decrease of TVOC concentrations (k _TVOC) following the observed peaks in the Saskatoon office, lab, and residential bathroom. The TVOC decay rate constant for thymol use was not determined.

Measured k _TVOC for quats, bleach, and H₂O₂ were largest in the lab and smallest in the bathroom, consistent with loss due to air exchange. Measured k _TVOC was on average 90% greater in the bathroom than ACR, but the difference was not statistically significant. Measured k _TVOC in the office and lab were ∼50% and ∼90% of ACR in those locations. This difference was statistically significant in the office but not the lab, due to much lower uncertainty in rate constants measured in the office compared to in the lab and bathroom. Due to the high experimental uncertainty, we cannot make definitive statements regarding the importance of factors other than air exchange in these locations. It is possible that surface loss played a role in the bathroom, given the high k _TVOC relative to ACR in that location. This is supported by the estimated S/V being higher in the bathroom than in the office and lab (by approximately a factor of 1.4). However, this is highly speculative, as k_TVOC in the bathroom was not statistically different from ACR.

3.4. Qualitative Comparison of Low-Cost PM and VOC Sensors

This study employed a range of low-cost air quality sensors to monitor temporal changes in particulate matter and total volatile organics, allowing us to draw several conclusions regarding their performances. These sensors have been extensively used for monitoring PM_2.5 levels in indoor environments and were demonstrated to provide reliable measurements. − However, the AQMesh multisensor pod, designed for outdoor pollution monitoring, was limited by its poor time resolution (15 min), which hindered its ability to capture rapid changes in IAQ during cleaning events.

We stress that TVOC sensors such as the PID used do not provide quantitative concentrations. With that said, the PID captured consistent increases in TVOC levels following the use of all cleaners other than thymol, including capturing a bimodal increase in TVOC mixing ratios following the use of bleach, as has been reported by others. Its reliability for capturing trends in TVOC levels following the use of thymol (and likely other cleaning agents not investigated here) is in question, given the observed repeatable decrease in TVOC levels during those experiments. This observation may point to an issue with the PID sensor, or it may be real, for example due to increased partitioning of background VOCs to the thymol-covered floor or to particles emitted during cleaning.

We also note that the PID’s performance was significantly affected by vibrations, such as those caused by nearby foot traffic, which introduced noise into the measurements. The AQMesh was insensitive to changes in TVOC mixing ratios following cleaning.

4. Implications

The primary goal of this work was to provide insight into general effects of cleaning solution composition and environmental variables such as ventilation rates on IAQ during and following the use of common cleaners and disinfectants. Our results indicate that emissions beyond the molecular active ingredients from commercial cleaners can affect IAQ. This is intuitive, and has been previously reported, but bears repeating, as in some cases additives may contribute to exposure risk. Significant increases in PM_2.5 were observed following the use of each cleaner. Across all investigated indoor settings, average fine particles levels presented the smallest increase for bleach (1.48 μg m^–3) and the highest for quats (6.35 μg m^–3). Similarly, TVOC levels rose by an average of 27 ppbv across all locations and cleaners following cleaning events, with the exception of thymol, which did not increase TVOC levels, though this may be due to instrumental limitations.

The observed increases in PM_2.5 and TVOC concentrations were relatively small in these experiments. However, we note that our experiments differed from common cleaning practices, especially in commercial buildings, in two important ways. First, in our controlled experiments we primarily followed manufacturer instructions for cleaning as opposed to disinfection, which requires minimum dwell times of 5–10 min. In experiments where quat solution was left on the bathroom floor for 10 min following application, in accordance with disinfection protocols, large increases in PM_2.5 and especially TVOC concentrations were observed. Second, we cleaned one ∼4 m² surface (1.9 m in the bathroom) in each experiment. It is more common to clean multiple surfaces in succession, as occurred in the hotel. In this case, people doing the cleaning will be exposed to elevated PM_2.5 and TVOC levels repeatedly throughout the duration of the cleaning. Cleaning practices in the hotel also differed from manufacturer recommendations, as disinfectant was sprayed into a cloth that was wiped on surfaces, and the solution was not removed from surfaces following application. This approach was taken to minimize exposure to cleaning personnel.

To contextualize the observed changes in this work, typical indoor daily activities such as cooking can increase PM_2.5 by 0.3–480 μg m^–3, and TVOC by 10–560 ppbv, concentrations depending on numerous factors (e.g., frying vs steaming, food composition, and amount of oil used). ,,− Other activities such as smoking can elevate PM_2.5 up to 253 μg m^–3 and TVOC up to 600 ppbv, depending on the number of cigarettes smoked. , Increases in PM_2.5 and TVOC concentrations observed in this work fall within the lower end of reported ranges. Given that cleaning often takes place at times when cooking and cigarette smoking are not occurring, we suggest that cleaning will influence IAQ. Additionally, the levels observed in this study are likely lower than we would expect under real-life circumstances due to the small surface areas cleaned. Further, it is important to emphasize that the composition of both VOC and particulate loadings differ depending on their source, so overall concentrations may not provide accurate estimates of exposure risk.

Temporal behavior, optical properties, and changes in size distributions of PM during cleaning events were consistent with direct emission from the cleaning solution due to the nebulizing action of the spray bottles. This suggests that exposure may be decreased by using a different application technique, though we did not investigate this possibility. Prior work that observed cleaning in a commercial kitchen where bulk cleaners were applied to tiled floors, followed by mopping, also observed strong primary aerosol production. Secondary aerosol formation resulting from the use of cleaning products has been reported in the literature following application of bleach and botanical disinfectants, and pine-oil cleaner, ,,,, though we did not observe evidence of this. Contrary to PM_2.5, changing the cleaning method will not likely change TVOC emissions, as most VOCs originating from cleaning solutions will eventually partition to the air, whether they are released in aerosols during the cleaning process or evaporate gradually from the cleaned surface. In this work and previous work we report that wiping cleaned surfaces to remove excess cleaning agent immediately following application can reduce emissions compared to extended dwell times. , However, this is only an option if the purpose is cleaning rather than disinfection, which requires dwell times of minutes.

Our results also highlight ventilation as a critical parameter in shaping IAQ, as it can help dilute and remove pollutants, thereby mitigating potential exposure. ,, However, the relationship between ventilation and pollutant decay is not always straightforward. In this work, we showed that ventilation was an important loss route for VOCs and PM_2.5, but other factors, such as surface partitioning, are also known to contribute to temporal trends.

The controlled experiments in this work help explain the observed field measurements performed in the hotel. Rapid dilution in the large hotel lobby, combined with a small surface area cleaned, resulted in low PM emissions that dissipated quickly. A greater surface area cleaned in the elevator, combined with a smaller air volume, increased both the peak PM_2.5 concentration and the residence time. The bathroom, where the largest surface area was cleaned, showed the highest emission of PM_2.5. ACR was high in the hotel (∼9.6 h^–1 in the bathroom); in locations with lower ACR, both PM_2.5 and VOC levels will likely be higher.

A secondary outcome of this study is that it extends the current state of knowledge on the use, and limitations, of measuring IAQ using low-cost air quality sensors. While these sensors may lack the selectivity and sensitivity to identify specific chemical species, they can effectively detect changes in pollutant levels characterized by a broad property such as light scattering for respirable PM and interactions with UV light for TVOC. This capability is useful for individual consumers to adopt simple yet impactful interventions in their behavior in residences to mitigate pollutant exposure, but also to aid building operators in providing safe workplaces for cleaning staff, such as ventilating a room/building during cleaning in addition to selecting cleaners that produce low emissions. , Consumers may be resistant to simple interventions without an indication of the impacts they yield (e.g., the simplicity of opening a window). Low-cost sensors also have limitations, such as in cases where the greatest cleaning exposure risk may be from a chemical not detected due to method selectively or sensitively, and as such may provide a false sense of security. We recommend caution when using low-cost sensors as absolute indicators of IAQ.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsearthspacechem.5c00046.

• Cleaning solutions ingredients list; nephelometer and low-cost sensor correlation; alternative ACR determination; PM size distribution; PM and TVOC concentrations (PDF)

§.

Atmospheric Services, WSP Australia, Brisbane, QLD, 4006, Australia

The authors declare no competing financial interest.

Published as part of ACS Earth and Space Chemistry special issue “Hartmut Herrmann Festschrift”.