Evaluation of air quality in indoor and outdoor environments: Impact of anti-COVID-19 measures

This study monitors the presence of 88 volatile organic compounds (VOCs) and semi-volatile organic compounds (semi-VOCs) at the gas phase of seven indoor settings in a school in the city of Tarragona, Spain, and five outdoor locations around the city. The VOCs and semi-VOCs monitored were solvents (∑Solvents), aldehydes (∑Aldehydes), emerging organic compounds (∑EOCs), and other VOCs and semi-VOCs (∑Others). Passive sampling campaigns were performed using Carbopack X tubes followed by thermal desorption coupled to gas chromatography with mass spectrometry (TD-GC-MS). Overall, 70 of the target compounds included in the method were determined in the indoor air samples analysed, and 42 VOCs and semi-VOCs in the outdoor air samples. Our results showed that solvents were ubiquitous throughout the school at concentrations ranging from 272 μg m−3 to 423 μg m−3 and representing 68%–83% of total target compounds (∑Total). The values of ∑Total in 2021 were three times as high as those observed at the same indoor settings in 2019, with solvents experiencing the greatest increase. A plausible explanation for these observations is the implementation of anti-COVID-19 measures in the indoor settings, such as the intensification of cleaning activities and the use of hydroalcoholic gels as personal hygiene. The ∑Total values observed in the indoor settings evaluated were twenty times higher than those found outdoors. ∑Solvents were the most representative compounds found indoors (74% of the ∑Total). The concentrations of VOCs and semi-VOCs observed in the outdoors were strictly related to combustion processes from automobile traffic and industrial activities, with ∑Others contributing 58%, ∑Solvents 31%, and ∑Aldehydes 11% of the ∑Total. EOCs, on the other hand, were not detected in any outdoor sample.


Introduction
Numerous studies in recent years have focused on monitoring the presence of volatile organic compounds (VOCs) and semi-volatile organic compounds (semi-VOCs) in outdoor environments such as industrial and urban atmospheres (Gallego et al., 2016;Maceira et al., 2017;Wania and Shunthirasingham, 2020;Xuan et al., 2021). This is because of the negative impact on human health and the environment associated with their accumulation in the atmosphere and the contribution of these compounds to the formation of photochemical smog via chemical reaction in the presence of light. Major anthropogenic sources of VOCs are combustion processes associated with road traffic (Nagpure et al., 2016;Raysoni et al., 2017) and industrial areas, especially petrochemical facilities where VOCs are produced, stored, handled and present in petroleum fuels (An et al., 2014;Vallecillos et al., 2019). Semi-VOCs comprise combustion products such as polycyclic aromatic hydrocarbons (PAHs) and a wide range of commercially available substances, including additives to consumer products, industrial chemicals or pesticides (Ma et al., 2021;Wania and Shunthirasingham, 2020).
It is known that VOCs and semi-VOCs are also found in indoor environments at even higher concentrations (Ma et al., 2021;Madureira et al., 2016;Yang et al., 2018). Indoor air quality has become an important focus area for the scientific community in order to evaluate the presence of VOCs and semi-VOCs in places where humans spend most of their time (PHE, 2019;WHO, 2000). The indoor environments evaluated include workplaces (Dugheri et al., 2018), schools (Becerra et al., 2020;Vallecillos et al., 2020a) and homes (Lizana et al., 2020;Villanueva et al., 2015b). Special attention is paid to monitoring VOCs and semi-VOCs in schools because, detector (de Lima et al., 2018;Dugheri et al., 2018;Rosenberger et al., 2016). Active sampling is commonly used for a short period of time and provides information about episodic emissions of VOCs and semi-VOCs in the atmosphere (Gallego et al., 2018;Maceira et al., 2017). Passive sampling, in both axial and radial diffusion mode, involves monitoring for seven or fourteen days and yields average concentrations data, but underestimates episodic emissions of VOCs and semi-VOCs (Poulhet et al., 2014;Vallecillos et al., 2019;Villanueva et al., 2018). Although active sampling is believed to be more versatile, passive sampling is the preferred sampling technique for many authors in the field due to its simplicity, low price and easy implementation (Hoang et al., 2017;Vallecillos et al., 2020b;Walgraeve et al., 2011;Wania and Shunthirasingham, 2020).
In this study, passive sampling was used to monitor VOCs and semi-VOCs, such as aldehydes, solvents and EOCs in the gas phase of various indoor and outdoor environments. All the target compounds except aldehydes were determined by means of TD-GC-MS. Due to their thermal instability, aldehydes were determined using liquid desorption followed by LC-DAD.
To evaluate the effect of anti-COVID-19 measures on indoor air quality, we compared the concentrations of solvents at indoor settings in a school in Tarragona (Spain) with those reported before the COVID-19 pandemic (Vallecillos et al., 2020a(Vallecillos et al., , 2020b).
Successive dilutions of 1,3-butadiene, supplied by Sigma-Aldrich as a 15 % wt. solution in nhexane, were applied to obtain the corresponding working solutions. Table 1S shows the complete list of target VOCs included in this study along with the compounds classified as semi-VOCs according to the definitions of VOCs stated in the European Directives 1999/13/EC (EU, 1999) and 2004/42/CE (EU, 2004).
of the school building, activity or internal covering) at a school in Tarragona and three living rooms in different houses. The fourteen-day passive sampling measurements were conducted at outdoor sites in Tarragona city centre and four towns near Tarragona's northern industrial complex (Constantí, Perafort, Sant Salvador and Els Pallaresos). Further information about sampling sites is found in Table 1. As recommended in USEPA methods 325A and 325B for determining VOCs from Fugitive and Area sources (EPA, 2019a(EPA, , 2019b, all sampling points were installed at a height of 2 m. VOCs and semi-VOCs passive sampling devices consisted of a previously conditioned CX tube with a diffusion cap (Supelco) and a stainless-steel protective hood. TH0160 data loggers (Perfect Prime, London, UK) were used to monitor temperature and pressure during the sampling period. The Radiello samplers used to monitor aldehydes included an adsorbent tube of florisil and 2,4-DNPH, a diffusive body (Supelco) intended to prevent the entrance of particulate matter, and a Triangular Support (Supelco).

GC-MS analytical method for VOCs and semi-VOCs
A thermal-desorption gas-chromatography mass-spectrometry-based method previously developed by Maceira et al. (2017) and Vallecillos et al. (2020b) for determining VOCs and semi-VOCs was applied. The target compounds trapped in the CX adsorbent tubes were thermally desorbed using a Unity 2 Thermal Desorption system connected to an Ultra A automatic sampler, both of which were supplied by Markes International Limited (Llantrisant, UK). Thermal desorption was performed in two steps known as primary and secondary desorption. For the primary desorption, CX tubes were heated to 330 °C for 10 min while a Chromatographic analyses were performed by gas chromatographic separation on a 7890 gas chromatograph (GC) with a quadrupole mass spectrometric detection system using a 5975 inert MS, both of which were supplied by Agilent Technologies (Palo Alto, CA, USA). GC separation was conducted on a 60 m x 0.32 mm i.d. x 1 µm film thickness capillary column with a film of 5% phenyl-95% dimethylpolysiloxane obtained as ZB-5 (Phenomenex, Torrance, CA, USA). The oven-temperature programme was as follows: an initial temperature 40 ºC (5 min), which increased at 6 ºC min -1 to 140 ºC and at 15 ºC min -1 to 220 ºC (8 min) with helium as carrier gas at a flow rate of 1.2 mL min -1 . The transfer line, ion source and quadrupole temperatures were set at 280 ºC, 230 ºC and 150 ºC, respectively. Identification of the target compounds was based on the coincidence of the retention times and ratios of the quantifier and qualifier ions when working in full scan mode (35-280 m/z) at an electron impact energy of 70 eV. Quantification was performed via integration of the quantifier ion peak areas of each compound. Table 1S shows the retention times and the qualifier and quantifier ions of each compound.

LC-DAD analytical method for aldehydes
Radiello tubes were desorbed with 2 mL of ACN in 20 mL glass vials. The extraction was performed in an ultrasound bath for 30 min. The extracts were then filtered using a paper filter to remove the granules from the adsorbent, and ACN was added to a volume of 2 mL. The extracts were evaporated to 1 mL under a gentle stream of nitrogen and reconstituted to 2 mL with Milli-Q water to obtain a 50:50 (v/v) water/ACN solution (initial conditions of the mobile phase). Before analysis by LC-DAD, the extracts were filtered with a 0.45 m PTFE syringe filter to avoid small particles damaging the analytical column.
(ACN) as the mobile phase. The gradient began at 38% ACN and increased to 70% at minute 2, then increased to 80% at minute 8 and held for 2 min. Finally, it was increased to 100% at minute 11 before returning to the initial conditions at minute 15, where it was held for a couple of minutes. The flow rate was 0.8 mL min -1 and the injection volume was 20 L. The column oven was set at 35 ºC and UV detection was performed at 365 nm (Andreini et al., 2000;Rosenberger et al., 2016). To identify the aldehydes, it was established that the retention time should coincide or have a variation of less than 0.1 min. The retention times are shown in Table   1S.

Results and discussion
In this section we report the validation parameters obtained by each analytical method and the concentrations of the target compounds observed in the indoor and outdoor environments evaluated. To obtain these concentrations in µg m -3 , we applied the equation based on Fick's first law described by Andrietta et al. (2010) and the experimental diffusive uptake rate previously calculated by Vallecillos et al. (2020bVallecillos et al. ( , 2020a corrected according to the average temperature and pressure of the sampling period (Q K , Table 1S). For the aldehydes we used the theoretical diffusive uptake rate provided by the supplier of Radiello tubes corrected for temperature and pressure (Supelco, 2021).

Validation
The validation parameters evaluated for each compound were linear range, instrumental limit of detection (LOD), instrumental limit of quantification (LOQ), method detection limit (MDL), method quantification limit (MQL), repeatability and reproducibility. Table 1S shows the instrumental validation parameters obtained with TD-GC-MS and LC-DAD used to determine the target compounds in indoor and outdoor air samples.
As specified in UNE-EN 14662-1 (UNE-EN, 2006), multistep external standard calibration curves for the target compounds were constructed by connecting sampling tubes to a Calibration Solution Loading Rig (Markes International Limited) and enriching the tubes with 1µL of the J o u r n a l P r e -p r o o f Journal Pre-proof corresponding standard solution. The tubes were purged for 5 min under a helium flow rate of 50 mL min -1 to ensure the evaporation of the solvent as well as the repeatability of the spiking procedure. To achieve the quality assurance conditions for determining VOCs and semi-VOCs in ambient air by USEPA Method TO-17 (EPA, 1999), the determination coefficients (r 2 ) of all calibration curves were above 0.999. LODs and LOQs ranged from 0.005 to 0.25 ppm and from 0.01 to 0.50 ppm, respectively. Repeatability and reproducibility values (n = 5, 1 ppm) expressed as relative standard deviations (RSD, %), ranged from 3% to 12% and from 4% to 18%, respectively. MDLs and MQLs were calculated from the LOQs and LODs and the equation described by Andrietta et al. (2010). MDLs and MQLs for seven-day sampling periods ranged from 0.002 to 0.049 μg m -3 and from 0.006 to 0.097 μg m -3 , respectively. The MDLs and MQLs obtained for fourteen-day passive sampling dropped to roughly half depending on the average temperature and pressure during the sampling period. Repeatability of the method was tested in the real working conditions used in this study. Specially, repeatability of the seven-day passive sampling was conducted at one of the indoor sites while that of the fourteen-day passive sampling was carried out at one of the outdoor sites. The results show that, regardless of the sampling site, % RSD (n =5) were between 6% and 21% for the target compounds detected in the samples.
The TD trap and analytical column were conditioned before and after the air samples were analysed to avoid background contamination. Blanks of the conditioned tubes were performed before use, and control samples were carried out during the sampling periods. More information about the method validation and quality assurance procedures are available from Vallecillos et al. (2019).

Indoor air quality
In this section we report the concentrations of the target compounds observed at the seven indoor environments of a school in Tarragona evaluated between April and June 2021. Table 2 J o u r n a l P r e -p r o o f Journal Pre-proof shows the maximum, minimum and average gas phase concentrations of the 70 target compounds detected in at least one of the samples analysed. Target compounds are listed by families, solvents, EOCs, aldehydes and others, and the semi-VOCs are in bold. The median concentrations and detection rates found for the detected compounds are in Table 2S.
As Table 2 shows, the sum of the average concentrations of the target compounds (∑Total) observed at the indoor settings of the school ranged from 343 μg m -3 (S7) to 563 μg m -3 (S2) with detection rates percentages higher than 93.6% in all cases. In general, the average and the median values of ∑Total obtained were similar, except for S2, indicating that they followed a normal distribution.
The ∑Solvents were ubiquitous throughout the school and had the highest concentrations, which ranged from 277 µg m -3 (S1) to 458 µg m -3 (S4), or between 71% and 85% of the ∑Total.
Overall, S7 showed the lowest values for each family of compounds except ∑Solvents, while the highest values were observed at S6 for ∑Others and S2 for ∑Solvents and ∑EOCs.
∑Solvents. In fact, the percentage of ∑Solvents in ∑Total increased from 29%-71% in 2019 to 68%-83% in 2021. Ethanol and isopropyl alcohol contributed the most to this increase in ∑Solvents, as their concentrations increased 2-8 fold and 4-18 fold, respectively. Although the concentrations of ∑Others found in 2021 were generally higher than those reported by and S6), while in other cases they decreased significantly (i.e. S1 and S2).
Regardless of the indoor setting evaluated (see Figure 1), the percentages of each family of target compounds followed a similar pattern, probably because they were of the same origin.
Since most indoor settings were normal classrooms, the main sources for the target compounds were probably the chemical formulations of cleaning products, building materials, and the pupils and teachers themselves (Even et al., 2020;Madureira et al., 2016;Tang et al., 2015;Yang et al., 2018). However, it is also plausible to think that the increase in the presence of solvents detected in 2021 is strictly related to the anti-COVID-19 measures implemented in schools, while other indoor sources of the target compounds are relegated to the background.
More specifically, these measures include more intensive cleaning activities conducted in the classrooms and other indoor settings and the use of hydroalcoholic gels for personal hygiene.
Although the classrooms were periodically ventilated, this was not sufficient to counteract the increase in the use of cleaning products and avoid the increase in solvent concentrations.

J o u r n a l P r e -p r o o f
Due to the high concentrations of ∑Solvents found in the school in 2021, the concentrations of aldehydes were monitored since they are known to be degradation products of primary alcohols (Carey et al., 2007). As Table 2 summarizes, aldehydes were ubiquitous throughout the school and the sum of their average concentrations (∑Aldehydes) ranged from 23.6 µg m -3 (S7) to 46.7 µg m -3 (S6). The aldehydes that contributed the most to ∑Aldehydes were formaldehyde, acrolein and hexanal, with concentrations ranging from 4.09 µg m -3 (S7) to 23.2 µg m -. 3 (S6).
The remaining aldehydes were found at average concentrations ranging from 0.21 µg m -3 (S4 and S6) to 6.92 µg m -3 (S6) while isopentanal was not detected in any of the samples analysed.
The range of aldehyde concentrations in the kindergarten (S1) and primary school (S2-S4) were generally comparable to those found in urban schools from Central-Southern Spain, 0.5 µg m -3 -35.7 µg m -3 for kindergartens and 0.5 µg m -3 -31.1 µg m -3 for primary schools, using the same doubling the values found in this study. The concentrations of aldehydes found in the school settings were also compared with those found in the living rooms of three houses (H1-H3). Table 3 summarizes the average and the median concentrations of the target aldehydes found in the houses under study, as well as the concentration ranges of each aldehyde. As Figure 2 shows, houses H1 and H2 presented similar values for ∑Aldehydes, i.e. 35.1 μg m -3 and 49.3 μg m -3 , respectively. As Table 3 shows, slightly lower concentrations of aldehydes were found in H1 because the weekly cleaning was done with bleach, known by-products of which are carbon tetrachloride and chloroform (Odabasi et al., 2014). H3, on the other hand, showed an average value for ∑Aldehydes of 164 μg m -3 , with acrolein as the most prevalent aldehyde (135 μg m -3 ).
A probable explanation for that is the intensive cleaning carried out with chemical products. In fact, the presence of acrolein in the composition labels of the cleaning products used was confirmed. The concentrations of aldehydes found in H1 and H2 were generally comparable to those found in the school in terms of average ∑Aldehydes and the most prevalent compounds.

J o u r n a l P r e -p r o o f
The values of ∑Aldehydes found in H1 and H2 were slightly lower than those reported by Diodiu et al. (2016) in Romanian houses in which the same sampling methodology was used (5 µg m -3 -150 µg m -3 ). Even so, the average concentrations of formaldehyde (50 µg m -3 ) and acrolein (25 µg m -3 ) found in the Romanian houses were five and two times higher, respectively, than those observed in H1 and H2. The average levels of formaldehyde (55.0 µg m -3 ) and acetaldehyde (28.8 µg m -3 ) found in different houses of the Central-Southern Spain using radial passive sampling were also much higher than the values found in H1 and H2 (Villanueva et al., 2015b).

Outdoor air quality
The target compounds were monitored at five outdoor sites in Tarragona city centre and four towns close to Tarragona's northern industrial complex between April and June 2021. Table 4 shows the maximum, minimum and average gas phase concentrations found for the 42 target compounds detected in at least one of the outdoor air samples analysed. Information on median concentrations and detection rates can be found in Table 3S. The concentrations of the target compounds observed at the indoor sites, i.e. the school and houses, were much higher than those found at the outdoor sites due to the limited dispersion of the target compounds even with periodic ventilation. The average ∑Total obtained at the indoor sites was 442 μg m -3 , whereas outdoors this figure dropped to 21.5 μg m -3 . The average ∑Total outdoor figure was twenty times lower than the indoor figure. Moreover, as the pie charts in Figure 3 show, the part played by each family of target compounds at the indoor sites was different from the part they played outdoor. Indoors, ∑Solvents contributed the most (79%) to ∑Total. The contributions of ∑EOCs, ∑Others and ∑Aldehydes ranged from 5% to 10%. On the other hand, the outdoor environments were characterized by an average ∑Others contribution of 58%, while ∑Solvents represented a 31% and ∑Aldehydes an 11%. EOCs were not detected in any outdoor sample analysed. Although the concentrations of ∑Total were much higher indoors than outdoors, the target compounds with the greatest contribution (∑Solvents) often did not present a health risk or their reference values for risk-assessment estimations for inhalation of these compounds is much lower than for the most common VOCs and semi-VOCs in outdoor atmospheres. The USEPA Risk Assessment Information System (RAIS, 2021), does not provide risk-assessment data on the EOCs evaluated in this study, while the compounds that contribute most to risk assessment are aldehydes and VOCs and semi-VOCs linked to industrial processes.
The high solvent content in the chemical formulations of cleaning products and the intensive use of these compounds during COVID-19 pandemic is a likely explanation for the high contribution of solvents to the ∑Total observed indoors. Other important sources of VOCs and semi-VOCs at the school, such as building materials and the students and teachers themselves (Even et al., 2020;Trocquet et al., 2021;Yang et al., 2018), have been relegated to the background. Our analyses of outdoor air samples provide clear evidence of the contribution to air quality of incomplete combustion processes, especially traffic emissions, in urban zones J o u r n a l P r e -p r o o f Journal Pre-proof (Pandit et al., 2011). The towns located near Tarragona's northern industrial complex also showed a higher presence of the VOCs produced and handled in the area.

Conclusions
In this study passive sampling in CX tubes was successfully used to determine 70 of the target compounds (including solvents, EOCs, aldehydes and others) in the gas phase of indoor settings at a school in Tarragona  The ∑Total values found at the outdoor sites (ranging from 14.8 µg m -3 to 33.0 µg m -3 ) were twenty times lower than those found indoors, while EOCs were undetected. The ∑Others, whose percentage of ∑Total was 16%-78%, were the most representative compounds, which shows that road traffic is one of the main sources of air pollution in urban areas. Although some peak concentrations of compounds linked to chemical industries were detected in towns close to Tarragona's northern industrial complex, none of the outdoor settings exceeded the maximum J o u r n a l P r e -p r o o f annual average concentrations. Therefore, the concentrations of the target compounds observed do not present a potential health risk. Although the concentrations of VOCs and semi-VOCs in indoor environments were much higher than those outdoors, the most prevalent target compounds, i.e. ∑Solvents, either do not present a health risk or their reference values for preforming risk assessment are much lower than those of the most common VOCs and semi-VOCs in outdoor atmospheres.  Xuan, L., Ma, Y., Xing, Y., Meng, Q., Song, J., Chen, T., Wang, H., Wang, P., Zhang, Y., Gao, P., 2021. Yang, X., Cheng, S., Wang, G., Xu, R., Wang, X., Zhang, H., Chen, G., 2018. Characterization of volatile organic compounds and the impacts on the regional ozone at an international airport. Environ. Pollut

Table captions
J o u r n a l P r e -p r o o f Table 1. Indoor and outdoor sampling sites where the target compounds were monitored plus a brief description of the characteristics of each site. Table 2. Concentrations of the target compounds (µg m -3 ) found in the gas phase of indoor air samples from a school in Tarragona (sites S1-S7). N = 5 seven-day passive sampling. Table 3. Concentrations of the target aldehydes (µg m -3 ) found in the gas phase of indoor air samples from the three houses evaluated. N = 5 seven-day passive sampling. Kindergarten (old floor with windows, handicrafts)

S2
First year of primary school (old building, first floor with windows, main classroom)

S3
Fourth year of primary school (new building, second floor with windows, main classroom)

S4
Last year of primary school (old building, second floor with windows, main classroom)

S5
High school (old building, second floor with windows, main classroom)

S6
Workshop (new building, second floor with windows, art activities, robotics and technological activities)

S7
Main entrance* J o u r n a l P r e -p r o o f Table 2. Concentrations of the target compounds (µg m -3 ) found in the gas phase of indoor air samples from a school in Tarragona (sites S1-S7). N = 5 seven-day passive sampling.  Table 3. Concentrations of the target aldehydes (µg m -3 ) found in the gas phase of indoor air samples from the three houses evaluated. N = 5 seven-day passive sampling.  -The indoor presence of VOCs and semi-VOCs is currently three times higher than it was in 2019.
-Solvents were the most prevalent target compounds in the indoor settings evaluated.
-Cleaning products and hydroalcoholic gel used against COVID-19 are the main solvent source.
-In outdoor air samples, solvent concentrations were twenty times lower than in indoor samples.
J o u r n a l P r e -p r o o f