Field Tests of Indoor Air Cleaners for Removal of PM 2.5 and PM 10 in Elementary School Classrooms in Seoul, Korea

Control of fine particles in childcare centers, elementary schools, elderly care facilities and other locations where vulnerable people spend most of their time is essential. This study investigated PM 2.5 and PM 10 concentrations in two classrooms equipped with one and two air cleaners, respectively, compared with concentrations outdoors and in a classroom with no air cleaner. Eight models of air cleaners with clean air delivery rates ranging from 9.9 to 21.3 m 3 min –1 were tested in the classrooms of two elementary schools in Seoul. Average PM 2.5 and PM 10 levels were 7.3 ± 0.7 and 45.5 ± 4.1 µ g m –3 in classrooms equipped with one air cleaner and 4.2 ± 0.6 and 24.6 ± 2.5 µ g m –3 in classrooms with two air cleaners, whereas they were 22.1 ± 2.6 and 109.1 ± 9.6 µ g m –3 in classrooms with no air cleaner and 36.9 ± 5.1 and 74.1 ± 10.6 µ g m –3 outdoors, respectively. The PM 2.5 level in classrooms was effectively reduced by one or two air cleaners because PM 2.5 infiltrated mainly from outdoors; meanwhile, PM 10 could not be reduced as PM 10 originated mainly from indoors from student activities. Air cleaners were more effective at removing PM 2.5 and PM 10 in classrooms with high airtightness than in classrooms with low airtightness. The average CO 2 concentration in classrooms was approximately 1500–2000 ppm during class hours and depended on the number of students per classroom; this level was approximately 1.5–2 times higher than the standard, regardless of the use of air cleaners.


INTRODUCTION
The average annual PM2.5 concentration in Korea has reached very high levels due to recent inflows from abroad and local sources, and it is causing widespread public health concern (Choi et al., 2019).Fine particles have significant effects on respiratory and cardiovascular diseases (Pope et al., 2008;Anderson et al., 2012;Lippmann, 2013).In particular, the reported risk is much higher for vulnerable groups such as infants, children and the elderly (Hajat et al., 2015).
Most people spend almost 90% of their time indoors (Klepeis et al., 2001), and indoor air pollution carries a much greater health risk than outdoor air pollution (Bonjour et al., 2013).Therefore, control of indoor air pollution is urgently needed.In particular, it is essential to manage air quality in school classrooms, where students spend approximately 25-30% of their time, as well as daycare centers and elderly care facilities, where vulnerable individuals spend significant time (Chithra and Shiva, 2018).
However, pollutant levels are higher in school classrooms than in general residential buildings or commercial facilities (Lee et al., 2002;Oeder et al., 2012).The concentration of fine particles increases significantly when students are active, based on surveys in the classrooms, libraries, administrative offices, and laboratories of schools (Gaidajis and Angelakoglou, 2009;Diapouli et al., 2008).Fine particle levels in schools are also affected by the floor on which the classroom is located, with the highest levels observed on the ground floor and lower levels on higher floors (Agarwal and Nagendra, 2016).
In Korean studies, crustal elements and chalk components made up large proportions of the dust in high school classrooms, and indoor pollutant concentrations increased even when the air conditioner was operating (Jang et al., 2006).The PM10 concentration in elementary school classrooms was low before school but increased sharply after school due to student activities (Jeong and Lee, 2010).Therefore, particle suspension via student activities has a great effect on fine particle levels in schools.However, air cleaners have rarely been employed to remove fine particles in schools.
In Korea, management of air quality in schools, where students vulnerable to fine particles spend a lot of time, is essential.Therefore, the Korean government is implementing projects to install air cleaners and mechanical ventilation devices in elementary and high schools.However, 20-30 students occupy 66 m 2 of a classroom in schools, and therefore high levels of fine particles can be generated internally.School buildings have lower airtightness compared with other buildings, allowing entry of high levels of fine particles from outside.Therefore, development of an operation and management plan for air cleaners or mechanical ventilation devices in schools is needed.However, basic installation and operation guidelines for air cleaners or mechanical ventilation devices to manage fine particles in schools are currently lacking.The Ministry of Education in Korea recently created a standard that requires the use of an air cleaner with a clean air delivery rate (CADR) of 13 m 3 min -1 or higher.However, no field data exist on the extent to which such an air cleaner can reduce fine particle levels in classrooms.
In this study, eight models of air cleaners with CADR values ranging from 9.9 to 21.3 m 3 min -1 were operated in the actual learning environments of two elementary school classrooms in Seoul constructed during different years.The effect of using one or two air cleaners on the reduction of fine particles in classrooms was investigated.In addition, the size distribution of fine particles in the classroom was assessed in various classroom situations, and the effect of the air cleaner on reducing fine particles was investigated according to particle size.Furthermore, the issue of poor ventilation in classrooms due to the use of air cleaners was investigated by observation of the change in CO2 concentration after using the air cleaners.

METHODS
Table 1 shows the specifications of the eight air cleaners used in this study.CADR values were obtained by measuring the rate of decrease over time for 0.3-µm-sized potassium chloride test particles in a 30 m 3 test chamber according to the SPS-KACA002-132 (KACA, 2018) test standard of the Korea Air Cleaning Association (KACA).Eight filter-type air cleaners were tested at CADR values ranging from 9.9 to 21.3 m 3 min -1 , which is more than twice the capacity of residential air cleaners, typically 4-6 m 3 min -1 , and all but one air cleaner (at 9.9 m 3 min -1 ) satisfied the criterion of 13 m 3 min -1 or more of the Ministry of Education in Korea.
Table 2 shows the temperature, humidity, and wind speed conditions during the experiment.The field experiment was conducted in four experimental target groups, each containing three  Light scattering instruments were used to measure the size distribution and concentration of fine particles.Three optical particle counters (Model 1.109, Grimm, Germany) were used for the indoor classroom measurements, and one (Model 180, Grimm, Germany) with a moisture pretreatment device was used for atmospheric measurements.To increase accuracy, all instruments were matched by calibration prior to the experiments.As ventilation may be reduced during air cleaner use, the CO2 concentration was also measured four times in each classroom.
The installation locations of the air cleaner and measurement instrument are shown in Fig. 1.In classrooms with one air cleaner, the air cleaner was placed in the center of the window side in the classroom, while in those with two air cleaners, one was installed near the doorway at the front of the classroom, and another was placed on the window side at the back of the classroom.The measurement instrument may be affected by the location or number of air cleaners.Therefore, it was installed on a table at a height of approximately 95 cm near the teacher's desk in the front of the classroom.All air cleaners were operated at their maximum air flow rate.In most cases, the noise of the air cleaners was not problematic for students taking lessons in classrooms.In addition, the airtightness of the classroom, which can affect the experimental results, was measured in each school.for PM10 based on an average over the 6 h of class time.The PM2.5/PM10 ratio on this day was approximately 0.79.This day had a relatively high PM2.5 concentration, as the PM2.5/PM10 ratio in urban areas is generally around 0.5 (Souza et al., 2014).However, the concentrations of fine particles in the classroom were 50.6 µg m -3 for PM2.5 and 176.9 µg m -3 for PM10, with a PM2.5/PM10 ratio of 0.29.The proportion of large particles greater than 2.5 µm in PM10 was much higher for classroom particles than atmospheric particles.In particular, whereas PM10 concentrations increased sharply during playtime or lunchtime when students' activity levels increased, no significant change in the PM2.5 concentration occurred.

PM2.5 and PM10 Concentration Changes and Size Distributions during Class Time
Fig. 2(b) shows the size distribution of fine particles in the classroom when no students were present before class, when students were in class, and during lunchtime.Small fine particles of less than 1 µm showed little change in the size distribution between times when no students were present and during lunchtime when students were actively moving.No correlation between the generation of small fine particles of 1 µm or less and the presence or absence of students in the classroom was found, as small fine particles are generated mostly outside the classroom, mainly from fuel combustion in automobiles or industrial facilities (Janssen et al., 2001).On the other hand, large particles greater than 1 µm were at low levels before students arrived at the school, and then particle generation increased significantly due to the students' activities.Thus, these large particles are generated mostly from student activities, arising from dust on their clothes or shoes, as well as dust on the floor being suspended due to the movement of students (Agarwal and Nagendra, 2016;Chithra and Nagendra, 2012;Diapouli, 2008).Therefore, PM2.5 occurs mainly outside the classroom, highlighting the importance of improving classroom airtightness to prevent inflow of PM2.5 from the outside environment.In the case of PM10, suppressing the generation of suspended particles is essential and may involve the wearing of indoor shoes by students and regular floor cleaning with water.

Comparison of PM2.5 and PM10 Changes in Classrooms with versus without Air Cleaners
Fig. 3 shows the PM2.5 and PM10 concentrations in a classroom with no air cleaner, a classroom with one air cleaner, and a classroom with two air cleaners on the same day as the test shown in Fig. 2. Here, three F-type air cleaners with a CADR of 14.5 m 3 min -1 , as described in Table 1, were used.As shown in Fig. 3(a), the average concentration of PM2.5 in the classroom with no air cleaner was 50.6 µg m -3 , which was only 64.5% of the level in outdoor air.During class time without ventilation, the concentration gradually decreased due to natural processes, but the concentration increased during playtime and lunchtime, when students often came and went; in particular, the concentration increased to a level similar to that in outdoor air during lunchtime.When one air cleaner with a CADR of 14.5 m 3 min -1 was used, the average PM2.5 level was 12.5 µg m -3 , representing 15.9% of outdoor PM2.5.During class time in an airtight environment, the concentration decreased to an average of 6.3 µg m -3 , but during playtime and lunchtime, when airtightness was low, the average levels were approximately 30.8 and 38.3 µg m -3 , respectively.Thus, the effectiveness of the air cleaner is greatly reduced during playtime and lunchtime.On the other hand, when two air cleaners were used, it was possible to reduce the average PM2.5 level to 12.3% of that in outdoor air.Of particular importance, the average concentration was reduced to 11.6 µg m -3 during playtime and 18.8 µg m -3 during lunchtime.
As shown in Fig. 3(b), PM10 showed an average concentration of 176.9 µg m -3 in the classroom with no air cleaner, which was approximately 1.8 times higher than the outdoor PM10 concentration of 99.5 µg m -3 .During playtime or lunchtime, when student activity levels are high, PM10 increased by approximately 2.4 and 3.8 times, respectively, compared with the outdoor PM10 concentration.When one air cleaner with a CADR of 14.5 m 3 min -1 was used, the average PM10 was 60.8 µg m -3 , which was 61.1% of the outdoor level.However, the concentration was 2.1 times higher indoors than outdoors during playtime and lunchtime.This finding indicates that the PM10 concentration is not easily controlled using one air cleaner during playtime or lunchtime, when students have high activity levels.On the other hand, when two air cleaners with a CADR of 14.5 m 3 min -1 were used, the average PM10 concentration was 48.6 µg m -3 , which was 48.9% of the outdoor concentration.The concentration could be reduced to 43.0% and 69.6% of the outdoor level during playtime and lunchtime, respectively.

Comparison of Average PM2.5 and PM10 Concentrations in Classrooms with versus without Air Cleaners
Fig. 4 shows the average PM2.5 and PM10 concentrations in outdoor air, a classroom with no air cleaner, a classroom with one air cleaner, and a classroom with two air cleaners for the eight air cleaner models listed in Table 1.In the classroom using one air cleaner, average PM2.5 and PM10 levels were 7.3 ± 0.7 and 45.5 ± 4.1 µg m -3 , respectively.In the classroom with two air cleaners, the levels were 4.2 ± 0.6 and 24.6 ± 2.5 µg m -3 , respectively.On the other hand, in the classroom with no air cleaner, PM2.5 and PM10 concentrations were 22.1 ± 2.6 and 109.1 ± 9.6 µg m -3 , respectively, and the corresponding values in outdoor air were 36.9 ± 5.1 and 74.1 ± 10.6 µg m -3 .For PM2.5, even in the classroom with no air cleaner, the concentration was 59.9% of that in outdoor air.These results show good agreement with previous studies, in which indoor air no internal particle source had a lower concentration than that of outdoor air (Guo et al., 2010;Madureira et al., 2012).On the other hand, the use of one air cleaner reduced the PM2.5 concentration by 67.0% compared with the classroom with no air cleaner, and the use of two air cleaners e reduced the PM2.5 concentration by approximately 81.0%.Compared with outdoor PM2.5, the use of one air cleaner and two air cleaners reduced the PM2.5 concentration by 80.2%, and 88.6%, respectively.Therefore, operation of one air cleaner in an indoor environment can be very effective at reducing fine particle exposure on a day with a high outdoor PM2.5 concentration.PM10 originates mainly from student activities, and its concentration was 1.47 times higher than the outdoor concentration in a classroom with no air cleaner.The PM10 concentration could be reduced by 58.3% using one air cleaner and by 77.5% using two air cleaners relative to classrooms with no air cleaner.However, compared with the outdoor PM10 level, indoor levels were reduced by only 38.6% using one air cleaner and by 66.8% using two air cleaners.Therefore, an air cleaner alone is not an effective solution for reducing the PM10 concentration in classrooms.

Average PM2.5 and PM10 Concentrations in Classrooms with One or Two Air Cleaners at Different Outdoor Concentrations
Fig. 5 shows the average PM2.5 and PM10 concentrations over 6 h of class time at different outdoor air concentrations when one or two air cleaners were used for five air cleaner types with similar CADR values ranging from 13.5 to 14.4 m 3 min -1 .For PM2.5, higher concentrations in outdoor air increased the average concentration in the classroom during class time.This result indicates that PM2.5 is generated mainly from outdoor sources.When one air cleaner in the range of CADR 13.5 to 14.4 m 3 min -1 is used, the PM2.5 concentration in the classroom can generally be brought to 10 µg m -3 , the annual PM2.5 guideline of the World Health Organization (WHO), except when outdoor PM2.5 is very high.On the other hand, for PM10, the concentration in classrooms has little correlation with the outdoor level.As PM10 is generated mainly through students' activities, outdoor PM10 has little effect on indoor PM10.Even when two air cleaners were used, each with a CADR of 13.5 to 14.4 m 3 min -1 , the annual WHO PM10 guideline of 20 µg m -3 was not easily met.The relatively new school classrooms in W school had lower PM2.5 and PM10 concentrations when an air cleaner was used compared to the old classrooms in S school.The airtightness of each classroom in the two schools was measured.The rate of air changes per hour at 50 Pa (ACH50) was 9.5 h -1 for a classroom in W school established in 2016, and 15.8 h -1 for a classroom in S school established in 2011.Here, ACH50 refers to a value representing the number of air changes per hour when the pressure inside of the building differs from that outside by 50 Pa.airtightness, showed relatively low levels of PM2.5 compared to classrooms in S school.Overall, classrooms in W school, with high airtightness, had greater reductions of PM2.5 and PM10 than S school, with low airtightness, when one or two air cleaners were used.The influence of infiltration may be greater in W school, as the wind speed at W school was higher than at S school on the day of measurement.Nevertheless, particle concentrations in classrooms were lower at W school, which has higher airtightness, than at S school.Therefore, use of an air cleaner with appropriately high CADR and improvement of classroom airtightness are essential to the management of fine particles in classrooms.

CO2 Concentration Changes in Classrooms with versus without Air Cleaners
Fig. 7 shows an example of CO2 concentration changes over time in three classrooms, one each with no air cleaner, one air cleaner and two air cleaners, used for 27-28 students in the second grade at S school.Regardless of the presence of an air cleaner, the initial CO2 concentration in all three classrooms was approximately 500 ppm, and it increased to 2,500-3,000 ppm during class, decreased by approximately 100-700 ppm during playtime, and then increased again to 3,000 ppm.During lunchtime, the initial CO2 concentration decreased to 1,500-2,000 ppm and then increased again to 3,000 ppm during afternoon class.Most of the classrooms were closed environments because measurements were taken during early winter, and thus CO2 levels continued to increase during class time due to lack of ventilation.In this case, ventilation occurs naturally as students enter and exit the door during playtime and lunchtime.

Comparison of the Average CO2 Concentration in Classrooms with versus without Air Cleaners
Fig. 8 shows the average CO2 concentration during class time according to the number of students and the presence of air cleaners in the second grade classrooms of two schools.For the same grade, the CO2 concentration was higher in the classroom of S school with 27-28 students than in the classroom of W school with 24 students.In both schools, average CO2 concentrations were no higher in classrooms using one or two air cleaners than in the classroom using no air cleaner.The maximum concentration of CO2 was 2,691-3,225 ppm in a classroom with no air cleaner, 2,592-3,579 ppm in a classroom with one air cleaner, and 2,312-3,176 ppm in a classroom with two air cleaners.Therefore, the use of air cleaners in schools does not specifically reduce the amount of ventilation and increase the CO2 concentration.In general, the concentration of CO2 in classrooms exceeds the standard of 1000 ppm by 1.5-2 times, and this exceedance of the carbon dioxide standard is frequently observed not only in Korea but also in educational environments in other countries (Fromme et al., 2007;Godwin and Batterman, 2007;Pegas et al., 2012;Buonanno et al., 2013).Therefore, development of classroom CO2 management measures, such as installation of ventilation devices separate from the air cleaner or allowing periodic natural ventilation, is essential.

Limitations of this Study and Required Additional Studies in Schools
The present study involved empirical measurements in the field, making it difficult to control variables such as infiltration or ventilation, which are affected by changes in atmospheric conditions.Therefore, interpretation of the concentration of fine particles in classrooms based on the airtightness of the classroom alone has limitations.Analysis of the effect of air cleaners according to changes in the amount of ventilation or infiltration, rather than the airtightness of the classroom, is also needed.Furthermore, this study reports experimental results obtained under the condition of the maximum air flow rate of an air cleaner with a new filter installed and thus provides information about the applicability of air cleaners only when their performance is at its maximum.Observation of performance changes when the air cleaner is operated automatically or used for a long time is necessary.If the air cleaner is used for a long time without maintenance, the air flow rate may decrease due to contamination of the pre-treatment filter, or the performance may deteriorate due to loss of electrostatic force in the high-efficiency particulate air filter.Therefore, further research of the cleaning cycle of the pre-treatment filter and the replacement period of the high-efficiency particulate air filter needed to maintain performance should be conducted.Finally, the actual fine particle reduction that occurs when the air cleaner is used for a long time in a school classroom must be clarified.

CONCLUSIONS
In this study, using eight models of air cleaners with a CADR of 9.9 m 3 min -1 or higher in two elementary schools located in Seoul, PM2.5 and PM10 concentrations in classrooms with one or two air cleaners were compared with those in a classroom with no air cleaner.Classrooms using one and two air cleaners showed 67.0% and 81.0%reductions in PM2.5 and 58.3% and 77.5% reductions in PM10, respectively, compared with classrooms with no air cleaner.
The PM2.5 concentration can be managed to meet the WHO guideline by improving airtightness and using an air cleaner with a CADR of 13 m 3 min -1 or greater, as the sources of PM2.5 generation are primarily outside.However, improving the airtightness of the classroom or using an air cleaner alone had limited effectiveness for reducing the PM10 concentration in classrooms, as PM10 is generated mainly from internal sources, such as suspension of particles due to student activities.
The concentration of CO2 in the classroom exceeded the standard of 1000 ppm by 1.5 to 2 times, regardless of the use of air cleaners.This finding confirms the urgent need for development of a ventilation plan for CO2 management in schools.If fresh outdoor air is introduced into the classroom through natural ventilation, the effectiveness of the air cleaner may be reduced.Therefore, introduction of a ventilation device equipped with a filter along with an air cleaner should be considered.
the first and second grade classroom groups of school A established in 2011 (3 classrooms × 2 grade groups = 6 total classrooms) and the second and third grade classroom groups of school B established in 2016 (3 classrooms × 2 grade groups = 6 total classrooms).Two air cleaner models were employed in each classroom group of three classrooms, and the experiment was repeated twice for each model.To reduce the differences in particle concentration among classrooms according to grade level and location, three classrooms of the same grade, located on the same floor and adjacent to each other were selected as the classroom group.The three classrooms of each group comprised a classroom with no air cleaner, a classroom with one air cleaner, and a classroom with two air cleaners.All experiments were conducted simultaneously across the classroom groups.In addition, during measurement in the three classrooms, the atmospheric particle concentration outside the classroom was also measured for comparison.

Fig. 1 .
Fig. 1.Locations of the air cleaners and measurement instruments in classrooms with (a) no air cleaner, (b) one air cleaner and (c) two air cleaners.

Fig. 2
Fig. 2(a) shows the PM2.5 and PM10 concentrations generated during class time in the second grade classroom of S elementary school in Seoul, along with the concentrations in outdoor air.The atmospheric fine particle concentrations on this day were 78.4 µg m -3 for PM2.5 and 99.5 µg m -3

Fig. 3 .
Fig. 3. (a) PM2.5 and (b) PM10 changes over time in classrooms with no air cleaner, one air cleaner or two air cleaners (F model).

Fig. 4 .
Fig. 4. Average (a) PM2.5 and (b) PM10 concentrations in classrooms with no air cleaner, one air cleaner or two air cleaners for eight models of air cleaners.

Fig. 5 .
Fig. 5. Average (a) PM2.5 and (b) PM10 concentrations in classrooms with one or two air cleaners at different outdoor PM2.5 and PM10 concentrations for five models of air cleaners with CADRs between 13.5 and 14.4 m 3 min -1 .

Fig. 6
Fig. 6 shows the PM2.5 and PM10 in/out ratio (I/O), which is the ratio of indoor to outdoor particle concentrations, in classrooms with no air cleaner, classrooms with one air cleaner, and classrooms with two air cleaners in two schools.The PM2.5 I/O in classrooms with no air cleaner was 0.78 ± 0.07 in S school, and 0.58 ± 0.06 in W school. Classrooms in W school, which had high

Fig. 6 .
Fig. 6.(a) PM2.5 and (b) PM10 in/out concentration ratios for classrooms with no air cleaner, one air cleaner or two air cleaners.

Fig. 7 .
Fig. 7. CO2 concentrations in classrooms with no air cleaner, one air cleaner or two air cleaners (Model F).

Fig. 8 .
Fig. 8. Average CO2 concentrations in classrooms of second grade students with no air cleaner, one air cleaner or two air cleaners in the (a) S school and (b) W school.

Table 1 .
Specifications of the tested air cleaners.

Table 2 .
Experimental data and weather information based on the average values during class hours.