Monitoring Indoor Air Quality in Classrooms Using Low-Cost Sensors: Does the Perception of Teachers Match Reality?

Abstract

This study intended to understand whether teachers’ perceptions of indoor air quality (IAQ) during classes aligned with the real levels of air pollutants and comfort parameters. For this purpose, an IAQ monitoring survey based on low-cost sensors using a multi-parameter approach was carried out in nine classrooms (a total of 171 monitored classes) in a Portuguese school. In each monitored class, the perception of IAQ reported by the teacher was assessed using a scale from 1 (very bad IAQ) to 10 (very good IAQ). Several exceedances regarding national legislation were found, with temperature being the parameter with a higher percentage of exceedance in all the studied classrooms (46%), followed by PM₁₀ (32%), and then CO₂ (27%). Temperature was found to be the only environmental parameter that was significantly associated with lower IAQ perception reported by the teachers, highlighting that typical pollutants such as CO₂ (which can be identified as stuffy air) did not contribute to the teachers’ perceptions.

1. Introduction

Indoor air quality (IAQ) is critically important in educational settings to provide students and teachers with a healthy indoor environment that maximises their productivity [1] and well-being [2,3]. With students typically spending about 6.5 h per weekday in classrooms [4], the quality of indoor air plays a significant role in their daily integrated exposure to environmental pollutants, which can have long-term health implications [5].

Over the past decades, numerous studies have highlighted various IAQ issues in classrooms, focusing on different types of pollutants and their sources [6,7], such as carbon dioxide (CO₂) [8], volatile organic compounds (VOCs) [9], and particulate matter [10]. Among these, ventilation has been consistently identified as a key factor in controlling pollutant levels [11], ensuring that air quality remains within acceptable limits as defined by national and international guidelines [3]. Poor IAQ has been linked to a range of negative effects [12,13], including cognitive decline, reduced focus, and decreased attendance. For example, studies have shown that lowering CO₂ concentrations from 2100 ppm to 900 ppm can improve task performance by up to 12%, while reducing CO₂ levels to 900 ppm can enhance the learning process by 5%, and daily attendance can be improved by 2.5% when CO₂ reduces from 4100 ppm to 1000 [12].

Despite the critical importance of IAQ in educational settings, several challenges persist in maintaining optimal air quality. These challenges include inadequate ventilation [11], pollution from building materials, infiltration of outdoor air pollutants, and specific indoor pollution sources promoted by classroom activities [14,15]. These challenges make it essential to implement effective strategies for monitoring and mitigating IAQ problems in classrooms. A comprehensive approach to improving IAQ begins with regular monitoring of air quality indicators and comfort parameters, which allows for accurate assessments and targeted interventions if needed.

In recent years, advances in low-cost sensor technology and Internet-of-Things (IoT)-based solutions have allowed the implementation of new multi-pollutant monitoring systems, which have allowed us to overcome shortcomings of traditional monitoring methods such as price, interference with local activities (due to noise), and specialised human resources [16,17]. The use of these tools has already been applied in different environments and has made it possible to gather relevant information that allows enhancing automatic controls and human decision-making [18] to improve air quality [19]. However, to ensure the accuracy and reliability of low-cost sensors, proper calibration against reference equipment is crucial to avoid measurement discrepancies [20,21].

In the absence of this type of IAQ monitoring technology system to inform the decision-making process, the management of IAQ in classrooms relies on the subjective perception of the occupants (teachers and students) during classes [22]. Teachers, particularly in naturally ventilated classrooms, often take action based on their sense of air quality, such as opening windows or doors to improve ventilation. However, their ability to assess IAQ accurately can be influenced by several external factors, such as weather conditions, noise pollution, and the presence of outdoor pollutants, which may compromise the reliability of their assessments [23]. Understanding the accuracy of teachers’ perceptions of IAQ (which drives their actions) is essential for establishing best practices and guidelines to ensure a healthy classroom environment.

The present study aims to contribute to this topic and to provide insights into how the IAQ perceptions of teachers align with real IAQ levels in classrooms. To this end, this study aimed to perform an IAQ assessment in classrooms during teaching hours using a multi-pollutant approach involving low-cost sensors and identifying the key environmental factors that shape teachers’ perceptions of air quality.

2. Materials and Methods

2.1. Study Site

The present study was conducted in a high school in the municipality of Ponte de Sor, Portalegre district, inland Portugal. Yearly, it has around 400 students attending classes. The school was remodelled in 2010, and a central air conditioning and ventilation system was installed. Additionally, new windows were installed in all the school buildings. However, due to the inability of the occupants to manually operate the windows (such as opening and closing), along with the inoperability of the central air conditioning and ventilation system due to budget restrictions, several complaints have been reported by the occupants of the classrooms (teachers and students), especially regarding thermal comfort during winter and summer.

Ponte de Sor municipality experiences a temperate climate characterised by dry, hot summers, as described in the Iberian Climate Atlas [24]. The mean monthly air temperature in Ponte de Sor in May 2023 was 20 °C; this month was considered “very hot” based on air temperature and very dry based on precipitation. Overall, the air temperature in Portugal in this month was 18.2 °C; this was considered the eighth hottest May since 1931 [25].

2.2. IAQ Monitoring Equipment Based on Low-Cost Sensors

Three monitoring boxes (MBs) based on low-cost sensors were assembled with four different low-cost sensors to provide a comprehensive characterisation of indoor environments regarding CO₂, VOCs, fine (PM_2.5), and inhalable (PM₁₀) particulate matter, and the comfort parameters temperature (T) and relative humidity (RH).

The low-cost sensors selected for this study included (1) an SCD30 for measuring CO₂, with a range between 400 and 10,000 ppm and an accuracy of ±30 ppm; (2) a MiCS-VZ-89TE for detecting VOCs, with a range of 0–1000 ppb in isobutylene; (3) an HPMA115S0 for monitoring PM_2.5 and PM₁₀, with a range of 0–1000 μg.m⁻³ and an accuracy of 15%; and (4) an SHT31 for monitoring temperature and relative humidity, with a precision of ±0.3 °C for T and of ±2% for RH. These MBs monitored real-time data for the selected parameters with a 1-min monitoring frequency and the data were sent instantaneously to an online platform via wireless communication (for cloud storage). Additional information about the MBs’ hardware platform and internal layout is described in previously published studies where these MBs were successfully used to assess IAQ in different micro-environments such as grocery stores [26] and bedrooms [27].

To assure data reliability, correction factors were identified for each parameter and for each MB in an inter-comparison study with calibrated real-time instruments. The reference real-time instruments were factory-calibrated devices from TSI (TSI, Shoreview, MN, USA), namely a DustTrak DRX monitor model 8533 to assess PM_2.5 and PM₁₀, and two IAQ-Calc Indoor Air Quality Meters: model 7545 to assess CO₂, T, and RH, and model 9565 to assess VOCs.

Table 1. Intercomparison results between MBs and the calibrated instruments for all the studied parameters. RMSE stands for root mean square error and Range REF stands for the range of the parameter as measured by the calibrated equipment (CAL).

Monitoring Unit	Parameter	Unit	m	b	R2	RMSE	Frequency	Range CAL	n Points	Ratio MB/CAL
AirQ1	PM2.5	µg.m−3	2.193	1.006	0.673	4.0	5 min	1.3–43.1	1005	0.37
	PM10	µg.m−3	4.202	−5.552	0.616	6.6	5 min	1.7–102.9	1005	0.43
	CO2	ppm	1.014	−27.310	0.997	32	5 min	400–2327	1530	1.02
	VOCs	ppb	0.320	−67.074	0.443	462.4	5 min	20–200	1290	5.05
	T	°C	1.284	−10.431	0.861	3.1	5 min	15.3–27.5	1493	1.13
	RH	%	1.164	−3.232	0.908	4.4	5 min	26.4–61.9	1530	1.09
AirQ2	PM2.5	µg.m−3	2.082	0.728	0.715	4.5	5 min	1.3–43.1	1306	0.33
	PM10	µg.m−3	3.527	−4.685	0.975	25.6	5 min	1.7–262.7	1102	0.42
	CO2	ppm	0.992	−23.907	0.997	39	5 min	400–2327	1530	1.04
	VOCs	ppb	0.370	−36.339	0.525	294.6	5 min	63–183	1202	3.64
	T	°C	1.254	−9.389	0.893	2.9	5 min	15.3–27.5	1481	1.12
	RH	%	1.159	−2.478	0.916	4.8	5 min	26.4–61.9	1530	1.10
AirQ3	PM2.5	µg.m−3	1.755	0.698	0.719	3.9	5 min	1.3–43.1	1110	0.50
	PM10	µg.m−3	2.653	−3.267	0.981	22.2	5 min	1.7–262.7	1293	0.45
	CO2	ppm	1.015	−74.847	0.997	69	5 min	400–2327	1528	1.09
	VOCs	ppb	0.315	−97.282	0.417	562.5	5 min	63–208	1174	5.91
	T	°C	1.260	−9.611	0.902	3.0	5 min	15.3–27.5	1416	1.12
	RH	%	1.234	−6.860	0.895	4.2	5 min	26.4–61.9	1528	1.07

presents the intercomparison results of each MB, including correction factors, such as the slope (m) and intercept (b) of the regression line between calibrated and low-cost sensor equipment.

Very strong correlations were obtained (R² ranging from 0.86 to 1.00) for CO₂, T, and RH for all the studied MBs. For PM₁₀, strong to very strong correlations were found (R² ranging from 0.62 to 0.98), while only strong correlations were found for PM_2.5 (expressed as R² ranging from 0.67 for AirQ1 to 0.72 for both AirQ2 and AirQ3). For VOCs, only moderate correlations were found (with R² ranging from 0.42 for AirQ3 to 0.53 for AirQ2).

The raw monitoring data were downloaded from the online platform and, afterwards, the correction factors for each sensor and each MB were applied to the raw data.

2.3. IAQ and Perception Survey During Classes

A total of 9 classrooms were selected to be monitored during class periods for one full week in May of 2023. The IAQ monitoring equipment was located on the teacher’s desk.

The teachers were asked to complete a brief questionnaire after each class to provide some details (day, start and end time of the class, and the number of students present and their average age) and their IAQ perceptions during the classes, ranging from 1 (very bad IAQ) to 10 (very good IAQ). The complete questionnaire that the teachers were requested to fill is presented in Table S1 of the Supplementary Information.

2.4. Data and Statistical Analysis

A Python script was built to aggregate the raw data, apply the identified correction factors, and evaluate the environmental parameters for each class, considering the start and end times. Further data treatment was performed in Microsoft Excel. Data analysis was conducted by applying non-parametric statistics, such as Spearman’s correlation (with a significance level of 0.050), using SPSS (Statistical Package for the Social Sciences, IBM SPSS Statistics, version 27).

2.5. Exposure to PM and the Doses Inhaled by Students

The exposure of students to particulate matter (E, μg.m⁻³.h) was determined by both the concentration and duration of exposure. It was calculated by multiplying the daily mean concentration (Cj) in the classroom by the time spent in it by the student, as expressed by the formula E = Cj.tj. The potential inhaled dose (D, μg), which represents the amount of a contaminant absorbed by the body of an exposed individual over a specific period (tj), was computed by multiplying the exposure by the Inhalation Rate (IR, m³.h⁻¹) [26,27,28].

It is well established that IR varies across different age groups and activities, ranging from 0.3 m³.h⁻¹ for children and young people at rest to 2 m³.h⁻¹ for adults engaged in physical activity. In the classroom setting, where activities are generally sedentary, an IR of 0.45 m³.h⁻¹ was used to estimate the potential inhaled dose by students aged between 11 and 18 years old [28].

3. Results

Table 2. Summary of the total monitored classes per classroom, including the mean number of students attending and the mean IAQ perception reported by the teachers.

Classroom	Number of Evaluated Classes	Mean Number of Students per Studied Class	Mean Age of Students Attending Classes	Mean IAQ Perception of Teachers (1–10)
B2	17	16	16	8.1
B11	14	23	16	5.4
B14	12	16	16	5.7
C2	25	13	17	7.2
C3	22	19	16	6.6
C11	23	14	14	5.3
C12	23	15	16	7.7
C14	12	19	18	7.3
C16	23	19	16	6.1

provides an overview of the number of classes evaluated per classroom, along with information on the mean number of students attending and the mean IAQ perception reported by the teachers. Overall, a total of 171 classes were monitored.

3.1. Indoor Air Quality Parameters

3.1.1. Temperature

The ISO 7730 guideline defines the optimal temperature range for indoor environments as being from 23 to 26 °C during the summer [29]. Figure 1 shows the mean, minimum, and maximum temperature levels assessed per classroom, as well as its compliance with the guideline, considering all the monitored classes. Overall, the mean temperature in all 171 monitored classes was 25.5 ± 1.8 °C, ranging from a minimum value of 19.9 ± 0.4 °C (classroom C12, twelfth monitored class) to 29.5 ± 0.2 °C (classroom C11, third monitored class). Only one classroom had mean temperature levels outside the optimal range, namely, classroom 11, with a mean temperature of 27.9 ± 1.4 °C.

Out of the 171 classes monitored, 76 (46%) had mean temperature levels outside the acceptable range. It is important to note that the monitoring period was in May, which was characterised by unusually high temperatures that may have contributed to the high levels recorded in the classrooms, as no mechanical ventilation or air conditioning was used.

3.1.2. Relative Humidity

Figure 2 presents the mean RH levels in all the monitored classrooms. A mean value of 43.1 ± 9.9% was registered, ranging from 24.6 ± 0.6% (classroom B14, ninth monitored class) to 65.1 ± 0.6% (classroom C12, eighth monitored class). All classrooms had mean RH levels within the acceptable range of 30 to 70% as defined by ISO 7730 [29]. Overall, only 15% (25 out of 171) of the classes had mean RH levels outside the acceptable range.

3.1.3. Carbon Dioxide

Figure 3 presents the mean CO₂ levels measured in each studied classroom, considering all the monitored classes. The overall mean CO₂ level found in the 171 monitored classes was 1042 ± 449 ppm, ranging from 405 ± 17 ppm (classroom B14) to 2476 ± 433 ppm (classroom C12). Since CO₂ is produced exclusively by occupants, this parameter often serves as an indicator of both ventilation quality and occupancy level [30].

The Portuguese legislation (Portaria n.º 138-G/2021, de 1 de julho) [31] establishes a CO₂ limit value (LV) of 1250 ppm. A total of 46 of the 171 monitored classes (27%) had mean CO₂ levels above this threshold. Fourteen classes were found to have CO₂ levels consistently above the LV (from the start to the end of the class), showing the lack of ventilation and accumulation effect from previous classes, which shows that the break between classes was not taken as an opportunity to ventilate the classrooms effectively.

The classrooms that had higher mean CO₂ levels during all their classes were B11 (1245 ± 430 ppm) and C14 (1255 ± 498 ppm), with 36% and 42% of their time above the established LV, respectively.

It is important to highlight that all classrooms had a central ventilation system; however, this was not working due to financial constraints. Additionally, the windows in the classrooms were not easy to open. If natural ventilation had been easy for the teachers to operate, the CO₂ levels in the classrooms would probably have been lower, and a lower number of exceedances would have been identified.

High levels of CO₂ in classrooms are a known fact, especially in countries where natural ventilation is usual. For instance, a study conducted in 50 classrooms in France identified a weekly mean of 1400 ± 400 ppm for naturally ventilated classrooms during the occupied period, while a weekly mean of 1000 ± 200 ppm was found for mechanically ventilated classrooms [32]. A study in the north of Portugal (Porto city, winter season) found that 86% of the 73 studied naturally ventilated classrooms had median CO₂ levels above 1000 ppm [33]. In naturally ventilated Albanian schools, weekly average CO₂ levels during classes ranged from 1286 to 5546 ppm (median 2776 ppm) during the cold season [30].

3.1.4. Total Volatile Organic Compounds

Figure 4 shows the mean VOC levels registered in all classrooms. All of them were below the limit value of 262 ppb (corresponding to 600 µg.m⁻³) established by Portuguese legislation [31]. The overall VOC mean, considering the 162 monitored classes, was 144 ± 79 ppb, with only 4% of the monitored classes having values above the established limit.

3.1.5. Particulate Matter

Figure 5 and Figure 6 indicate the mean levels of PM_2.5 and PM₁₀ in the studied classrooms, respectively.

Regarding PM_2.5, three classrooms had mean PM_2.5 levels above the limit value of 25 μg.m⁻³ established by the Portuguese legislation [31], namely, B14, C3, and C16.

The overall mean PM_2.5 concentration was 19.6 ± 11.6 μg.m⁻³, ranging from 4.2 ± 0.0 μg.m⁻³ (classroom B2, seventh monitored class) to 78.0 ± 14.5 μg.m⁻³ (classroom C2, twenty-fourth monitored class). Considering the Portuguese legislation limit value, only 22% of the classes had mean PM_2.5 levels above the threshold; however, if the air quality guideline value for PM_2.5 defined by the World Health Organization (WHO) is considered (5 μg.m⁻³) [34], then only 2% of the monitored classes were below this value.

For PM₁₀, all the studied classrooms had mean levels below the limit value of 50 μg.m⁻³ established by the Portuguese legislation [31], with an overall mean of 30.0 ± 17.4 μg.m⁻³, ranging from 3.5 ± 1.3 μg.m⁻³ (classroom B2, seventh monitored class) to 96.1 ± 2.9 μg.m⁻³ (classroom C3, twentieth monitored class).

Of the 170 monitored classes, only 12% had mean levels of PM₁₀ above the limit value established by the Portuguese legislation; however, 82% of the classes had mean PM₁₀ levels above the air quality guideline threshold of 15 μg.m⁻³ established by WHO [34].

The exposure and potential inhaled doses of PM in the classrooms, considering 8 h of classes per day, are described in

Table 3. Exposure to PM₁₀ and PM_2.5 and corresponding potential inhaled doses during classes in the studied classrooms.

	PM Concentration, Cj (µg.m−3)		Time, tj (h)	Exposure to PM (µg.m−3.h)		Potential Inhaled Dose (µg)
Classroom	PM2.5	PM10		PM2.5	PM10	PM2.5	PM10
B2	8.2	10.5	8.0	66	84	29.5	37.9
B11	17.1	24.1	8.0	137	193	61.5	86.8
B14	25.9	37.6	8.0	207	301	93.2	135.4
C2	20.3	30.8	8.0	162	247	72.9	111.0
C3	29.2	46.1	8.0	234	369	105.2	166.1
C11	15.0	24.3	8.0	120	194	54.1	87.4
C12	14.2	21.2	8.0	113	170	51.0	76.4
C14	18.8	30.5	8.0	151	244	67.8	109.7
C16	26.6	42.0	8.0	212	336	95.6	151.3
Mean ± SD	19.5 ± 6.8	29.7 ± 11.1		156 ± 54	238 ± 89	70.1 ± 24.4	107 ± 40

. Mean exposures to PM₁₀ and PM_2.5 were estimated to be 156 ± 54 μg.m⁻³.h and 238 ± 89 μg.m⁻³.h, respectively.

The exposure levels found in the present study were slightly lower than those found in classrooms in Lisbon primary schools (Portugal), namely, 218 μg.m⁻³.h for PM_2.5 and 404 μg.m⁻³.h for PM₁₀ [4]. This difference may be due to the age of the students and the typical activities in primary schools (which include more physical activities that may cause resuspension of particles [4]) compared with classrooms where older students have classes, such as the ones in the present study. Additionally, local sources of PM can also explain higher exposure (especially PM₁₀). This may be the reason for the higher exposure in Lisbon schools, where the use of chalk in the classrooms (contributing to PM₁₀ exposure [4]) was identified, along with the impact of heavy local traffic (also contributing to PM_2.5 exposure) that is known to be much higher in Lisbon (the capital of the country) than in inland Portuguese towns (with rural surroundings), such as the site where the studied school is located [35].

3.2. Indoor Air Quality Perception

Figure 7 provides an overview of the IAQ perception reported by the teachers in the studied classes (a total of 170). It can be seen that, in general, most teachers rated IAQ as being medium to good (76% of the attributed IAQ perception values were between 6 and 10).

Table 4. Monitored mean levels of IAQ parameters for each perception category (as perceived by teachers in the studied classes).

IAQ Perception	Temperature (°C)	Relative Humidity (%)	CO2 (ppm)	VOCs (ppb)	PM2.5 (μg.m−3)	PM10 (μg.m−3)
1	-	-	-	-	-	-
2	-	-	-	-	-	-
3	27.7 ± 2.0	40.7 ± 10.1	1070 ± 350	167 ± 72	29.0 ± 3.2	45.1 ± 3.9
4	26.8 ± 2.6	46.4 ± 6.0	1030 ± 450	116 ± 60	13.0 ± 5.6	19.1 ± 8.8
5	26.0 ± 1.7	42.8 ± 10.3	1090 ± 410	136 ± 84	22.8 ± 14.5	32.0 ± 14.5
6	25.5 ± 1.7	41.2 ± 10.8	960 ± 300	139 ± 81	22.1 ± 12.2	34.2 ± 19.1
7	25.1 ± 1.7	45.1 ± 9.8	1180 ± 580	161 ± 68	21.1 ± 12.5	33.5 ± 19.5
8	25.3 ± 1.6	41.8 ± 9.6	970 ± 440	126 ± 88	17.3 ± 9.6	27.3 ± 17.2
9	25.4 ± 2.0	48.0 ± 6.7	1010 ± 430	178 ± 95	11.6 ± 5.1	17.0 ± 8.9
10	24.2 ± 0.7	36.5 ± 15.7	1000 ± 460	87 ± 60	17.6 ± 2.4	30.5 ± 4.5

summarises the mean levels of the IAQ parameters monitored for each perception category.

3.3. Spearman’s Correlation Between IAQ Perception and Environmental Parameters

Table 5. Spearman’s correlations between IAQ perception and the environmental parameters monitored in the studied classrooms. Bold values represent significant correlations, with p-value < 0.05.

		Environmental Parameters
Spearman Parameters		T	RH	CO2	VOCs	PM2.5	PM10
IAQ Perception	ρ	−0.225	−0.001	−0.073	0.031	−0.146	0.119
	p-value	0.003	0.991	0.346	0.689	0.057	0.122
	n	170	170	170	169	170	169

presents Spearman’s correlations between the IAQ perception indexes provided by the teachers for the 170 studied classes and the mean levels of the different environmental parameters. Only temperature had a significant negative impact on IAQ perception (Spearman’s rho of −0.225, p-value = 0.003), while the remaining parameters did not show any significant correlation. Overall, an increase in mean temperature during the class period caused a decrease in the teachers’ IAQ perception levels at the end of the class.

4. Discussion

The present study confirms that the use of monitoring tools based on low-cost sensors provides an extensive and successful survey in classrooms without having an impact on teaching activities.

Overall, as shown in

Table 6. Percentage of exceedances of the different monitored environmental parameters in the studied classrooms (relative to Portuguese legislation).

	Environmental Parameters
Classrooms	T	RH	VOCs	CO2	PM2.5	PM10
B2	35%	6%	0%	29%	0%	0%
B11	43%	0%	0%	36%	7%	0%
B14	50%	67%	0%	8%	42%	33%
C2	40%	28%	0%	8%	16%	63%
C3	36%	41%	0%	14%	59%	41%
C11	87%	0%	0%	26%	13%	43%
C12	43%	0%	13%	39%	0%	0%
C14	42%	0%	0%	42%	8%	83%
C16	35%	0%	17%	43%	48%	22%
Mean exceedances	46%	16%	3%	27%	21%	32%

, temperature was the parameter with the highest percentage of exceedances in all the studied classrooms (46%), followed by PM₁₀ (32%) and then CO₂ (27%).

Additionally, temperature was the only parameter that had a significant correlation with the teachers’ perception of IAQ during the classes. On the contrary, the remaining environmental pollutants, even when their levels were above the thresholds established by Portuguese legislation, did not cause a lower ranking of IAQ perception.

The findings of the present study, targeting the IAQ perceptions of teachers during classes, agree partially with the results of a study conducted in the UK [36], where children’s perception was assessed over a full year. In that study, 28 naturally ventilated classrooms in eight UK primary schools were studied, and 805 children were surveyed throughout the year regarding their perceptions of IAQ. It was found that their perceptions were affected by CO₂ levels (during the non-heating season) and by the operative temperature (during the heating season); the latter result is similar to that of the present study.

A study conducted in office buildings in southern Brazil also concluded that the air quality perception of the occupants was influenced by factors other than CO₂ [37], with a significant influence of thermal and humidity comfort on air quality satisfaction being identified.

A study conducted in Portugal revealed that people with more knowledge about air quality have a higher level of perception about it [38]. Therefore, awareness campaigns about IAQ targeting teachers may allow them to improve the factors affecting IAQ during classes, decreasing both their exposure and their students’ exposure to high levels of pollutants that may compromise their health and performance.

For instance, high CO₂ levels are known to cause a reduction in cognitive performance in pupils and to lower students’ and teachers’ productivity [3]. Additionally, high CO₂ levels are identified by the commonly called stuffy air; however, from the perspective of the occupants of the classrooms, it may go undetected, as was confirmed by the results of the present study, where high levels of CO₂ were not associated with lower IAQ perception.

It is important to highlight some limitations of the present study and potential further work:

(i)

IAQ perception focused on the teachers’ perceptions and not on the students’ perspectives, which should also be evaluated. Considering that students tend to feel comfortable in indoor climates that are cooler than environments where adults feel thermally neutral [3], it would be relevant to assess their perceptions and identify potential associations with environmental parameters.

(ii)

This type of study should be performed throughout the academic year to understand the influence of seasons on IAQ perception and IAQ parameters (including ventilation).

(iii)

The impact of different ventilation typologies in the classrooms (natural and mechanical) on IAQ perception and IAQ levels should be studied.

(iv)

To evaluate the effective impact of an IAQ awareness campaign for teachers and students on the improvement or not of IAQ levels during classes and their perception.

Additionally, when mechanical ventilation systems are not properly working or when in naturally ventilated classrooms, the adoption of good ventilation practices would result in lower accumulation of air pollutants in classrooms, minimising the degradation of IAQ. The following measures can be included:

(i)

Ventilate the classrooms between classes by opening doors and windows during breaks (to avoid pollutant accumulation from one class to the next);

(ii)

Classrooms should be cleaned at the end of the day instead of in the morning before classes so that the decay of pollutants (particles and VOCs) can occur during nighttime;

(iii)

If outdoor meteorological conditions allow, always have some ventilation during classes by opening windows and/or doors.

Given the budget constraints faced by many schools, the recommendations provided in this study are designed as cost-effective solutions that can be practically implemented to improve IAQ without straining limited financial resources.

Nevertheless, if during classes teachers are aware that IAQ is getting degraded, they would actively act to improve ventilation in the classrooms by opening windows and/or doors; therefore, real-time monitoring of IAQ with indication of IAQ levels to teachers should be available in the classrooms. Moreover, specific training on IAQ, its impact on students’ well-being and performance, and strategies to improve it should also be provided to teachers so that they are conscious of IAQ and what should be done to improve it.

5. Conclusions

This study allowed the successful application of an IAQ monitoring strategy based on low-cost sensors using a multi-parameter approach to provide an understanding of IAQ levels in nine classrooms (totalling 171 classes). Additionally, using a simple IAQ perception index (1 to 10), it was possible to obtain the IAQ perceptions of the teachers.

Temperature was the only environmental parameter that was significantly associated with lower IAQ perception, which highlights that typical pollutants such as CO₂ (which can be identified as stuffy air) do not contribute to teachers’ perceptions of IAQ. It is important to implement monitoring strategies that provide real-time information on pollutant levels, especially in naturally ventilated classrooms such as the ones evaluated in this study, considering the negative impacts that such air pollutants may have on health and performance. With this valuable information, teachers can make decisions based on reality and can act immediately to mitigate situations when levels are above established thresholds.