**Impact of Preventive Measures on Subjective Symptoms and Antigen Sensitization against Japanese Cedar, Cypress Pollen and House Dust Mites in Patients with Allergic Rhinitis: A Retrospective Analysis in the COVID-19 Era**

**Takashi Oda 1, Fumiaki Maeda 1, Sachio Takeno 1,\*, Yuri Tsuru 1, Chie Ishikawa 1, Takashi Ishino 1, Kota Takemoto 1, Takao Hamamoto 1, Tsutomu Ueda 1, Tomohiro Kawasumi 1, Hiroshi Iwamoto 2, Kazunori Kubota 3, Yoshio Nakao <sup>4</sup> and Masaru Kunimoto 1,5**


**Abstract:** For >2 years, Japan's government has been urging the populace to take countermeasures to prevent COVID-19, including mask wearing. We examined whether these preventive behaviors have affected the rate and degree of sensitization against pollen and house dust antigens in patients with allergic rhinitis. We retrospectively surveyed 2565 patients who had undergone allergy blood testing during the period 2015–2021. We subdivided this period into eras based on the COVID-19 pandemic: the pre-COVID (2015–2019, n = 1879) and COVID (2020–2021, n = 686) eras. The positive rates for Japanese cedar and cypress in the 40–59-year-olds and those for house dust in the 20–39-year-olds were significantly reduced in the COVID era versus those in the pre-COVID era. Each group's mean antigen-specific CAP scores decreased significantly from the 1st to 2nd era: from 1.98 to 1.57 for cedar (*p* < 0.01), 1.42 to 0.95 for cypress (*p* < 0.05), and 2.86 to 2.07 for house dust (*p* < 0.01). Our survey of the patients' clinical records indicates that 47.5% of the pollinosis patients reported improvement in nasal symptoms after the three seasons of pollen dispersion in the COVID era. Japan's quarantine policies designed to combat the spread of COVID-19 thus coincide with pivotal measures to alleviate allergic reactions.

**Keywords:** allergic rhinitis; face mask; COVID-19; house dust mite; Japanese cedar pollen; Japanese cypress pollen; preventive behavior; subjective symptoms

## **1. Introduction**

Coronavirus disease 2019 (COVID-19) is an ongoing pandemic infection caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [1]. For over 2 years, Japan's government has urged the populace to take countermeasures to prevent COVID-19, with guidance such as "wear a mask", "wash your hands frequently", and "stay home and avoid the 3Cs outside (crowded places, close contact settings, and closed spaces)" [2]. These measures have helped suppress respiratory viral infections and the risk of asthma exacerbations [3]. We hypothesized that these preventive behaviors may also have affected the severity of nasal symptoms and the rate and degree of sensitization against common

**Citation:** Oda, T.; Maeda, F.; Takeno, S.; Tsuru, Y.; Ishikawa, C.; Ishino, T.; Takemoto, K.; Hamamoto, T.; Ueda, T.; Kawasumi, T.; et al. Impact of Preventive Measures on Subjective Symptoms and Antigen Sensitization against Japanese Cedar, Cypress Pollen and House Dust Mites in Patients with Allergic Rhinitis: A Retrospective Analysis in the COVID-19 Era. *Atmosphere* **2022**, *13*, 1000. https://doi.org/10.3390/ atmos13071000

Academic Editors: Ashok Kumar, Amirul I Khan, Alejandro Moreno-Rangel and Michał Piasecki

Received: 1 June 2022 Accepted: 20 June 2022 Published: 21 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

antigens among patients with allergic rhinitis (AR) in Japan. We conducted a retrospective survey of the data of 2565 patients who showed allergic symptoms and had undergone allergy blood testing during the period from 2015 to 2021. We subdivided the period into two eras based on the outbreak of the COVID-19 pandemic in Japan: the pre-COVID-19 era (2015–2019) and the COVID-19 era (2020–2021). The patients' data in the two eras were compared based on four age groups: 0–19, 20–39, 40–59, and ≥60 years old. We also conducted a survey to investigate changes in the patients' social behaviors and subjective nasal symptoms after the start of the COVID-19 pandemic.

Allergic rhinitis (AR) can be classified into seasonal allergic rhinitis (SAR) and perennial allergic rhinitis (PAR). Japanese cedar/cypress and house dust mites are major causative antigens for SAR and PAR, respectively [4], and the numbers of individuals in Japan with SAR or PAR both increased markedly in the pre-COVID-19 era. The results of a nationwide epidemiological survey conducted in 2019 revealed the presence of cedar pollinosis in 38.8% of the respondents compared with 16.2% in 1998 and 26.5% in 2008 [5]. The presence of PAR has also shown gradual increases from 18.7% in 1998 to 23.4% in 2008 and 24.5% in 2019 [4].

The elimination and avoidance of antigens are pivotal measures to alleviate allergic reactions, and these measures coincide with quarantine policies designed to combat the spread of COVID-19 [4,6,7]. We have found no prior study that specifically assessed the impact of anti-COVID-19 measures on AR patients in Japan based on both objective and subjective parameters.

#### **2. Materials and Methods**

#### *2.1. Patient Enrollment and Pollen Counts*

We retrospectively analyzed the data of 2565 patients who showed allergic symptoms and had allergy blood tests against common antigens at Hiroshima University Hospital during the years 2015–2021. The antigens included Japanese cedar pollens (*Cryptomeria japonica*), Japanese cypress pollens (*Chamaecyparis obtusa*), and house dust mites (*Dermatophagoides pteronyssinus)*. The patients' antigen-specific IgE levels were determined by the ImmunoCAP™ Specific IgE system (ThermoFisher Scientific, Waltham, MA, USA). The blood tests are sandwich tests in which the solid phase of specific antigens ensures binding of all relevant antibodies, providing a uniquely high binding capacity. The CAP score system is graded as follows: 0, ≤0.34 UA/mL; 1, 0.35–0.69 UA/mL; 2, 0.70–3.49 UA/mL; 3, 3.50–17.4 UA/mL; 4,17.5–49.9 UA/mL; 5, 50.0–99.9 UA/mL; 6, ≥100 UA/mL. The CAP-RAST scores ≥2 are regarded as positive, 1 as borderline, and 0 as negative, respectively. The diagnosis of AR was based on the combination of the CAP score ≥2 and the presence of nasal symptoms such as sneezing, nasal discharge, and nasal congestion [4].

We divided the study period 2015–2021 into two eras based on the outbreak of the COVID-19 pandemic in Japan: the pre-COVID (2015–2019) and COVID (2020–2021) eras. We compared patients in the two eras classified into four age groups: 0–19, 20–39, 40–59, and ≥60 years old. The proportion and distribution of patients with positive ImmunoCAP scores were determined for each age group, and we compared the differences in these scores between the two eras.

In addition, to assess the possible changes in the patients' social behaviors and subjective allergic symptoms after the start of the COVID-19 pandemic, we analyzed the clinical records of pollinosis patients who visited the above-mentioned hospital or related ENT clinics (led by KK, YN and KM) in Hiroshima City after the onset of the COVID-19 pandemic. The records included whether the patient (1) had engaged in limited outdoor behaviors in accord with social regulations, i.e., wearing a mask, and (2) reported improvement in his or her nasal and ocular symptoms in the pollinosis seasons after the change in social behaviors.

The degree of pollen dispersion was monitored annually by a gravitational pollen sampler on the roof of the Hiroshima University Hospital. The cypress and cedar pollen counts were determined daily by staining with Calberla solution from 15 January to 31 May each year.

This study was performed in accordance with the Declaration of Helsinki, with approval from the Hiroshima University School of Medicine Institutional Review Board (approval no. E-1738; approval date: 4 September 2019).

#### *2.2. Statistical Analyses*

The Kruskal–Wallis and Mann–Whitney U test were used for between-group comparisons. Fisher's exact test was used to compare qualitative data. *p*-values < 0.05 were considered significant. JMP Pro ver. 14 (SAS, Cary, NC, USA) was used for the analyses.

#### **3. Results**

#### *3.1. Changes in the Proportions of SAR and PAR Patients*

A total of 1879 patients in the pre-COVID era and 686 patients in the COVID era were enrolled (Table 1). There were no significant differences in the age or gender distributions between the two eras. The pollen counts and dispersion periods of cedar and cypress fluctuated annually, due mainly to weather conditions in the previous summer. The amounts of mean season cedar pollens and the mean dispersion periods of cypress pollens tended to increase during the COVID era, but no significant between-era differences in these values were observed (Mann–Whitney U test, *p* = 0.6985 for cedar pollen counts and *p* = 0.0506 for cypress dispersion periods) (Table 2). We also collected information on the air pollutant levels in Hiroshima City that exert adverse effects on allergic symptoms [8] (Table 2). The mean levels of particulate matter (PM) 2.5 and related gaseous materials (SO2 and NOx) tended to improve in the COVID era, but a significant between-era difference was not observed (Mann–Whitney U test).

**Table 1.** Demographics of the study population, air pollutants, and pollen dispersion in Hiroshima City.


**Table 2.** Comparison of pollen dispersion counts and periods, and air pollutants in Hiroshima City.


Figure 1 illustrates the time-course changes in the proportion of patients with positive ImmunoCAP scores (≥2) from 2015 to 2021. In the SAR patients, the positive rates for both cedar and cypress in the 40–59-year-olds were significantly reduced in the COVID era compared with those in the pre-COVID era (Figure 1a,b). In this age group, the mean positive rates decreased significantly from 63.2% to 51.4% for cedar (*p* < 0.05) and from 53.7% to 31.0% for cypress (*p* <0.01). The positivity tended to be lower for both antigens in 2021 compared with 2020. No similar significant between-era differences existed in the other age groups. In the PAR patients, the positive rates for house dust in the 20–39-yearolds were significantly reduced in the COVID era compared with those in the pre-COVID era (Figure 1c). In this age group, the mean positive rates decreased significantly from 70.5% to 47.8% (*p* < 0.01). We further compared positive antigen rates of the patients between the two eras stratified by 10-year-olds generation, gender and CAP scores for each antigen (Table 3). The positive rates in the pre-COVID era for the antigens in each generation were compatible with those by previously published data in 2006 and 2016 [9]. A significant decrease in the positive rates for both SAR and PAR was observed in the generations of COVID-19 era that corresponded to the results shown in Figure 1. No significant differences were detected in the gender proportions between the two eras.

**Figure 1.** Changes in the positive antigen rate against (**a**) cedar, (**b**) cypress, and (**c**) house dust mites for each age group before and after the COVID-19 outbreak. \* *p* < 0.05, \*\* *p* < 0.01, Fisher's exact test.

#### *3.2. Changes in the CAP Scores for Pollens and House Dust*

We also examined the CAP score for the responsible antigens in each age group between the pre-COVID and COVID eras. The results are consistent with those of the antigen proportion rates. Seasonal AR patients in the 40–59-year-old group showed a significant reduction in CAP scores in the COVID era compared with those in the pre-COVID era (Figure 2). The mean CAP scores decreased from 1.98 to 1.57 for cedar (*p* < 0.01) and from 1.42 to 0.95 for cypress (*p* < 0.05). No similar significant differences were identified in the other age groups. In the PAR patients, the 20–39-year-old group showed a significant reduction in CAP scores in the COVID era versus those in the pre-COVID era (Figure 3). The mean CAP scores for house dust decreased from 2.86 to 2.07 (*p* < 0.01).

**Table 3.** Comparison of positive antigen rates of the patients between the two eras stratified by 10-year-old generation and CAP scores for each antigen.


† *p* = 0.125, \* *p* < 0.05, \*\* *p* < 0.01 versus the corresponding groups of the pre-COVID era.

#### *3.3. Changes in Subjective Nasal Symptoms after the Start of the COVID-19 Pandemic*

Figure 4 illustrates the changes in the patients' social behaviors and reported subjective allergic symptoms after the start of the COVID-19 pandemic, based on our survey of their clinical records. We collected records from 122 patients with cedar/cypress pollinosis based on a positive CAP score. Their mean age was 42.3 years, with 49 males and 73 females. Regarding outdoor behaviors compared with before the COVID-19 pandemic, almost 90% of patients (111/122) reported that they always wore masks when out in public, in accordance with the Government's request. The proportions of patients who reported improvement in nasal symptoms and ocular symptoms after the start of the COVID era were 47.5% (58/122) and 30.3% (37/122), respectively; the nasal symptom improvement rate is significantly higher than that of ocular symptoms (*p* < 0.01).

**Figure 2.** Comparison of CAP scores for cedar and cypress in each age group between the pre-COVID-19 (2015–2019) and COVID-19 (2020–2021) eras. (**a**) 0–19, (**b**) 20–39, (**c**) 40–59, (**d**) ≥60 years old groups. The data are mean ± SD (error bars). \* *p* < 0.05, \*\* *p* < 0.01.

KRXVHGXVW*'HUPDWRSKDJRLGHVSWHURQ\VVLQXV*

**Figure 3.** Comparison of CAP scores for house dust in each age group between the pre-COVID-19 (2015–2019) and COVID-19 (2020–2021) eras. (**a**) 0–19, (**b**) 20–39, (**c**) 40–59, (**d**) ≥60 years old groups. The data are mean ± SD (error bars). \*\* *p* < 0.01.

**Figure 4.** Changes in the social behaviors and the severity of subjective allergic symptoms after the start of the COVID-19 pandemic. Clinical records were surveyed from 122 cedar/cypress pollinosis patients who visited the hospital/clinics from April to May in 2022. \*\* *p* < 0.01, Fisher's exact test.

#### **4. Discussion**

Environmental variations induced by industrialization and climate change partially explain the increases in the prevalence and severity of allergic disease. The prevalence of cedar/cypress pollinosis and that of house dust mite-induced PAR continue to show increasing trends in Japan, accompanied by environmental and circumstantial increases in antigen exposure [4,8]. During the post-World War II period 1946–1980, approx. 20% of Hiroshima prefecture's land area (1660 km2) was planted with cedar or cypress forests in compliance with a massive national afforestation policy designed to secure timber and maintain land preservation. The resulting increased exposure to pollens among younger individuals has been reported to have led to early sensitization to pollen antigens [10].

Each year during the Japanese cedar pollen-dispersion season (February to April), followed by that of Japanese cypress (April to May), a large number of pollinosis patients experience severe nasal and ocular symptoms. Because Japanese cypress pollens contain several components that cross-react with cedar pollens, about 70% of pollinosis patients suffer the symptom burdens from both types of trees. A severe affliction due to cedar and cypress pollens is also attributable to the nature of the pollens' dispersion, i.e., their large quantities and long (nearly 100-km) airborne distances [11].

As a quantitative test, CAP scoring enables monitoring of the development and severity of allergic diseases [12]. In sensitized pollinosis patients, antigen-specific IgE levels are strongly affected by the amount and periods of cedar or cypress pollen exposure [8,13]. These patients' CAP scores rise after they experience larger amounts and/or longer periods of pollen exposure, and their scores remain high through to the next season. We recently reported the trends in pollen dispersion and their possible relationship with the degree of antigen sensitization against cedar and cypress pollens from 2001 to 2018 in the same singleinstitution setting [14]. Our present analyses revealed that the proportion of patients with positive CAP scores for both cedar and cypress increased continually over the past 18 years, with the increase rate of the cypress CAP scores more prominent at 25%. These results indicate that the levels of pollen dispersion have provided sufficient exposure to maintain antigen sensitization during daily activities before the COVID-19 outbreak in Japan. In contrast, the preventive measures during the COVID-19 pandemic period appear to have attenuated the potential pollen sensitization. Inhaled airborne allergens such as pollen (10–100 μm) and house-dust mite feces (10–40 μm) play a significant role in triggering IgE-mediated immunologic responses in typical allergic rhinitis symptoms [4].

Our present findings indicate that both the positive rates and the mean CAP scores for both cedar and cypress in the 40–59-year-old group of SAR patients decreased significantly in the COVID era, i.e., 2020 and 2021. The improvement of these objective parameters could be attributable to the potential contribution of wearing non-woven surgical masks that can filter particles > 3 μm [15]. The scrupulous nature of Japanese citizens, who have agreed to change their outdoor behaviors as requested by the Government, is likely to have reinforced the protective effects. Our analyses considered environment factors such as air pollutants, which have adverse effects on allergic symptoms; the analyses revealed no significant difference in the mean annual levels of PM2.5 and related gaseous materials (SO2 and NOx) between the pre-COVID and COVID eras.

The present findings are in line with the recent reports in other countries describing a reduction in the subjective burden of screened nasal symptoms that are due to seasonal or perennial allergic rhinitis after the adoption of anti-COVID-19 measures [6,7,16]. Mengi et al. evaluated the use of face masks on AR symptoms in 50 pollen allergy patients who were compulsorily using face masks due to the COVID-19 pandemic in Turkey [6]. They found that the rate of participants with severe-moderate nasal symptoms decreased significantly during the pandemic with the use of face masks, from 92% (46/50) to 56% (28/50), and the corresponding rate of ocular symptoms decreased significantly from 60% (30 patients) to 32% (16 patients). An investigation conducted in Northern Italy examined the effects of quarantine and face-masking policies on nasal and ocular symptoms in a pool of 124 patients suffering from ragweed allergy [16], and the results demonstrated that the overall burden of oculorhinitis decreased significantly during the 2020 ragweed season. Reductions in the use of the common anti-allergic medications, such as oral antihistamines and nasal steroids, were also observed. To assess the impact of face masks on subjective AR symptoms, Dror et al. analyzed the data of a multicenter questionnaire distributed in 2020 for 2 weeks to hospital nurses in Israel, and they reported reductions in AR symptoms [7]. Our present findings also showed that 70% of the SAR patients described an improvement in nasal symptoms after the three seasons of pollen dispersion in the COVID era. Together the above-cited results highlight the potential benefit of face masks for AR patients.

Several reasons have been proposed in relation to the worldwide increase in PAR caused by house dust mites. The exacerbation factors include global warming as well as the increased time spent indoors with air conditioning inside higher and more airtight residences [17,18]. In the present study's PAR patients, a significant decrease occurred in the positive rates and the mean CAP scores for house dust in the 20–39-year-olds after the outbreak of COVID. In this sense, the elimination and avoidance of mite antigens are pivotal measures to alleviate allergic reaction. They coincide with official quarantine policies designed to combat the spread of COVID-19, i.e., mask-wearing and improved or frequent ventilation of indoor spaces [2].

Our patients in the 40- to 59-year-old age group of SAR patients and those in the 20 to 39-year-old PAR group showed significant reductions in positive rates and CAP scores in the COVID-19 era. There are possible explanations for the differences between the age groups. The prevalence of Japanese cedar pollinosis is most common among middle-aged individuals, especially 40–49-year-olds, whose morbidity rate is up to 40%. In contrast, house dust mite allergy is prevalent mainly among younger individuals, i.e., 20–39-yearolds [4,5]. Yonekura et al. reported that SAR induced by cedar pollen took chronic courses

in the majority of middle-aged patients [19]. The prevalence of cedar pollinosis in Chiba, Japan increased over 10 years from 1995 to 2005 probably due to a high level of pollen exposure. It is thus likely that preventive behaviors would manifest their inhibitory effects in these corresponding age groups.

Another possibility is that the individuals in these generations tend to be more amenable to social infection-avoidance behaviors such as remote indoor working and restrictions on outdoor activities as well as mask-wearing and hand-washing [20]. In any case, the importance of determining the reason(s) for the age-group differences should be acknowledged, as the possible relationship between lifestyle habits and associated exposure routes might have non-negligible effects on the present outcomes worldwide.

This study has several limitations. It was cross-sectional and based on a regional sample collection. A causal relationship between the two eras cannot be directly determined. We did not include information on the severity of nasal and ocular symptoms of the subjects that correspond directly to CAP scores. A direct relationship between the subjects' lifestyle and associated exposure routes was not clarified. Possible important factors that determine the age-group differences should be further assessed, including lifestyle habits and associated exposure frequencies. Further studies on a larger scale and with control groups are also necessary to elucidate the effects of personal protective equipment.

At the time of writing the manuscript (18 June 2022), Japan's government still enforces mask-wearing to the citizens and applies other virus-related preventive rules. However, acknowledgement of increased vaccination certificates and changes in genome shifts to omicron variants with mild virulent forms might lift regulation guidelines on how to prevent the spread of the coronavirus. In this sense, our data presented herein would be of value as the reference.

In conclusion, the COVID-19 pandemic has provided an opportunity to assess the effects against allergen exposure to pollens and other air pollutants. Our findings suggest that appropriate COVID-preventive actions are important and effective measures as reflected by the decreased allergic parameters (both objective and subjective) in the Japanese population investigated herein.

**Author Contributions:** Conceptualization, T.O. and S.T.; methodology, T.O., S.T. and H.I.; validation, S.T. and T.H.; formal analysis, T.O. and S.T.; investigation, T.O., F.M. and S.T.; data curation, F.M., Y.T., C.I., T.I., T.K., K.T., K.K., Y.N. and M.K.; writing—original draft preparation, T.O. and F.M.; writing—review and editing, S.T.; visualization, T.O. and F.M.; supervision, S.T. and T.U.; project administration, S.T. and H.I.; funding acquisition, S.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded partly by a grant from the Japan Society for the Promotion of Science KAKENHI: 22K09668 and a Health Labor Sciences Research grant: 21FC1013.

**Institutional Review Board Statement:** This study was performed in accordance with the Declaration of Helsinki, with approval from the Institutional Review Board at the Hiroshima University School of Medicine (approval no. E-1738; approval date: 4 September 2019).

**Informed Consent Statement:** Informed consent was obtained from all participants involved in the study.

**Data Availability Statement:** The data presented in this study are available on reasonable request from the corresponding author.

**Acknowledgments:** We thank Yukako Okamoto for the technical assistance.

**Conflicts of Interest:** The authors declared no potential conflict of interest with respect to the research, authorship, and/or publication of this article.

#### **References**


## *Article* **Evaluation of Typical Volatile Organic Compounds Levels in New Vehicles under Static and Driving Conditions**

**Ruihua Guo 1, Xiaofeng Zhu 1, Zuogang Zhu 1, Jianhai Sun 2, Yongzhen Li 1,\*, Wencheng Hu <sup>1</sup> and Shichuan Tang 1,\***


**Abstract:** In modern societies, the air quality in vehicles has received extensive attention because a lot of time is spent within the indoor air compartment of vehicles. In order to further understand the level of air quality under different conditions in new vehicles, the vehicle interior air quality (VIAQ) in new vehicles with three different brands was investigated under static and driving conditions, respectively. Air sampling and analysis are conducted under the requirement of HJ/T 400-2007. Static vehicle tests demonstrate that with the increasing of vehicle interior air temperature in sunshine conditions, a higher concentration and different types of volatile organic compounds (VOCs) release from the interior materials than that in the environment test chamber, including alkanes, alcohols, ketones, benzenes, alkenes, aldehydes, esters and naphthalene. Driving vehicle tests demonstrate that the concentration of VOCs and total VOCs (TVOC) inside vehicles exposed to high temperatures will be reduced to the same level as that in the environment test chamber after a period of driving. The air pollutants mainly include alkanes and aromatic hydrocarbons. However, the change trends of VOCs and TVOC vary under different conditions according to various kinds of factors, such as vehicle model, driving speed, air exchange rate, temperature, and types of substance with different boiling points inside the vehicles.

**Keywords:** vehicle interior air quality; volatile organic compounds; new vehicles; static conditions; driving conditions

#### **1. Introduction**

Social concerns over indoor-air quality extend not only to the indoor-air environment of newly built apartment houses but also to that of vehicles [1,2]. In modern societies, as a result of urban sprawl, a vehicle cabin has been recognized as a part of the living environment because people are spending increased time in vehicles than ever before during business, shopping, recreation or travel activities [3–6]. Unfortunately, harmful volatile organic compounds (VOCs), such as benzene, toluene, ethylbenzene, xylenes, styrene, butyl acetate, and undecane, etc., exist in vehicular cabins, which deteriorate vehicle interior air quality (VIAQ) and threaten the health of drivers/passengers [7–14].

The vehicle interior pollution results from the emission of interior furnishing materials and the infiltration of engine exhausts and other exterior environmental pollutants [15–17]. In a vehicle cabin, the concentration of VOCs may be higher in comparison to concentrations found in public or private buildings [18–20], which may vary in time and are dependent on the interior temperature, humidity, ventilation, vehicle age and other parameters [21–23]. Wensing [24] proved that the concentrations of in-vehicle total volatile organic compounds (TVOC) decrease exponentially over a 40-day period from 35~120 mg/m3 to 10~30 mg/m3. Yoshida and Matsunaga [25] also demonstrated that the TVOC concentration decreased

**Citation:** Guo, R.; Zhu, X.; Zhu, Z.; Sun, J.; Li, Y.; Hu, W.; Tang, S. Evaluation of Typical Volatile Organic Compounds Levels in New Vehicles under Static and Driving Conditions. *Int. J. Environ. Res. Public Health* **2022**, *19*, 7048. https:// doi.org/10.3390/ijerph19127048

Academic Editors: Ashok Kumar, M Amirul I Khan, Alejandro Moreno Rangel and Michał Piasecki

Received: 25 April 2022 Accepted: 5 June 2022 Published: 9 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

from over 10 mg/m<sup>3</sup> to 200 μg/m<sup>3</sup> during the first three years after delivery. The investigation carried out by Grabbs also found that TVOC levels inside new vehicles decreased by more than 90% during a three-week test period [26]. The concentrations of most VOCs declined over time, but increased with increasing interior temperature [27]. Increasing temperature encourages higher desorption of VOCs from interior materials, and the amount of VOCs in vehicles is higher during the summer than during the winter [28]. During vehicle operation, air pollutants originating from interior materials are reduced as more and more VOCs initially captured from interior materials are eventually removed by ventilation [28]. However, heavy traffic problems result in poor air quality in the city, and subsequently cause more serious in-vehicle air pollution problems [29]. Vehicle interior benzene concentrations range from 10~20 μg/m<sup>3</sup> during freeway travel to 150 μg/m3 in heavy urban traffic [30]. VOCs in new vehicles are usually measured under static conditions, in which the air exchange rate is very low (1~3 h<sup>−</sup>1) [31]. However, under other operating conditions, such as setting the fan to fresh air (closing windows), the air exchange rate in the vehicle will increase (13.3~26.1 h<sup>−</sup>1), thus reducing the level of air pollutants in the vehicle [27].

In this study, in order to further understand the level of air quality under different conditions in new vehicles, the concentrations and types of typical VOCs are measured and identified under the static (parked in environment test chamber and in sunshine, vehicle's engine is off) and driving conditions, respectively. Additionally, the limits specified in Chinese national guidelines for air quality assessment of passenger cars (GB/T 27630-2011) [32] are cited for comparison with the concentrations of typical VOCs.

#### **2. Experimental Methods**

#### *2.1. Vehicicles under Study and Their Pretreatment*

The vehicles under study included three brands (Brand A, B, C) of vehicles from different manufacturers, which were all newly domestically produced vehicles in China (tested less than 28 ± five days from their date of production). There were three different models in vehicles of Brand A (A1, A2, A3) and B (B1, B2, B3) respectively, and one model in vehicles of Brand C (C1). The number (n) of vehicles of each model in Brand A, B and C were one (labeled as A1-1, A2-1, A3-1), two (labeled as B1-1, B1-2, B2-1, B2-2, B3-1, B3-2) and six (labeled as C1-1, C1-2, C1-3, C1-4, C1-5, C1-6) (Table 1). All vehicles were well maintained and in good operating condition. None of the vehicles had fuel leakages or any mechanical problems. The covering films, such as plastic film, on the surfaces of the vehicles' trim materials were ripped off. The passenger compartments were completely free of cigarette smoke and deodorizers. Details of interior materials of all vehicles are presented in Table 1.

**Table 1.** Vehicles under study and corresponding test conditions.



**Table 1.** *Cont.*

#### *2.2. Sampling Process under Static Conditions*

Under static conditions, the air exchange between the interior and exterior environment of vehicles was significantly reduced when all windows and doors of vehicles were closed. In the case that the concentrations of air pollutants outside of vehicles were very low, VIAQ mainly depended on the amount of VOCs released from vehicle interior sources. That was, the amount of harmful substances released from vehicle interior trims could reflect the situation of vehicle interior air pollution. In this study, the static state was first chosen as the test state of the vehicles, and the vehicle interior air samples were collected in the environment test chamber and in sunshine condition respectively. Standard sampling and analytical methods were used in this experiment.

#### 2.2.1. Vehicle Test Protocol in Environment Test Chamber

As many factors could affect concentrations of VOCs in vehicles, an environment test chamber with a volume of 100 m3 was utilized in this experiment, which could provide stable and accurate control of the required temperature, relative humidity (%RH), and airflow velocity, according to set parameters. The air in the chamber was purified by activated carbon filters to reduce the influence of background VOCs on VIAQ. The interior surface of the chamber was constructed with stainless steel, which could minimize adsorption and emission of VOCs [29].

The test protocols utilized with the vehicles (Brand A~C) when sampling the interior air were as follows [33]: (a) The vehicle was moved (the engine was off) to the environment test chamber, and then the chamber's door was closed. (b) The environmental conditions in the chamber were adjusted according to the set parameters and kept for the whole test duration: environment temperature: 25.0 ± 1.0 ◦C; relative humidity: 50% ± 10%; airflow velocity: ≤0.3 m/s; background toluene and formaldehyde concentration: ≤0.02 mg/m3; background TVOC concentration: ≤0.1 mg/m3. The vehicle was aired by opening the windows and doors for 8 h to make a good mixture inside and outside of the vehicle, and then the vehicle was left closed for 16 h, assuming it could reach a steady state pollutant concentration in this period.

The sampling position inside the vehicle was set at head height in the middle of the front two headrests to simulate the height of the driver's breathing zone. Teflon tubing was used as the sampling line, which was led outside the vehicle from the upper corner of the vehicle's door, and the length between the sampling device and the sampling location was 2 m. In this way, the sampling process could be done outside the vehicle, thus eliminating the influence of the operator activities on the testing results. In-chamber air samples as background samples were also collected simultaneously. The sampling position in the chamber was within a range of 0.5 m from the test vehicle and at the same height with that in the vehicle. During the whole test procedure, temperature and relative humidity in the chamber were recorded to satisfy the required parameters [29,33].

#### 2.2.2. Vehicle Test Protocol in Sunshine Condition

After the test in the environment test chamber, the vehicles of Brand A were moved outside of the chamber and parked under direct sunlight between 10:00 a.m. and 3:00 p.m. on a sunny and windless day in summer. The ambient temperature during the process of sun exposure should be in the range of 35~37 ◦C, which could make in-vehicle air temperature increase quickly. In the course of experiments, the Telfon tubing and sensor probe of the thermometer and hygrometer were fixed at the predetermined sampling point (Section 2.2.1), which could auto-monitor and record the temperature and relative humidity inside the vehicles. The tested vehicles were left closed for 4 h (A1-1, A2-1, A3-1) and then the samples of the vehicle's interior air were collected. The background concentrations of toluene, formaldehyde and TVOC in ambient air should meet the same requirements as in the environment test chamber (Section 2.2.1). Air samples outside the vehicles were collected for blank analysis simultaneously, and the sampling position was within a range of 0.5 m from the test vehicle and at the same height with that in the vehicle.

#### *2.3. Sampling Process under Driving Conditions*

After the test in the environment test chamber, the vehicles of Brand B and C were moved to the outside of the chamber and were tested on the vehicle proving ground according to the driving test conditions shown in Table 1. Similarly, a sunny and windless day in summer was chosen as the sampling day, and the background concentrations of toluene, formaldehyde and TVOC in ambient air should meet the same requirements as those in the environment test chamber (Section 2.2.1). The windows and vents of the vehicles were kept closed for the entire test procedure. The air conditioner in the vehicle was turned on and set to internal circulation, and the temperature was adjusted to 25 ◦C. Under driving conditions, since the in-vehicle air samples couldn't be collected outside the vehicle, it was necessary to bring the sampling device into the vehicle for sampling. The sampling points were arranged strictly according to the requirements of that in static conditions. The sampling workers entering the vehicle had to wear masks and foot covers to reduce the influence of the external air pollutants on the air quality of the tested vehicles. The sampling time began when the driver and sampling workers entered the tested vehicle and turned on the air conditioner in the vehicle (recorded as 0 min). The work from entering the vehicle to when sampling started should be completed within 1 min. The in-vehicle air samples of 0~30 min and 60~90 min were collected with a sampling duration of 30 min. The ambient air samples on the vehicle proving ground as background samples were also collected simultaneously, and the sampling position was at the same height as the sampling position in the vehicle during driving.

#### *2.4. Air Sampling and Analysis*

Air sampling and analysis was conducted under the requirement of HJ/T400-2007 [33]. In-vehicle VOC emissions (C6–C16) were taken by active sampling using controlled flow pumps at a rate of 100 mL/min for 30 min onto a stainless steel tube packed with Tenax TA. The samples obtained were stored at 4 ◦C and were protected from light in a refrigerator until analysis. The thermal desorption-gas chromatography/mass spectrometry (TD-GC/MS) was employed for identification and quantification of in-vehicle VOC emissions. The collected VOCs were thermally desorbed at 270 ◦C for 3 min. The desorbed compounds were cryogenically focused in a cold trap at −30 ◦C. After focusing, the trap underwent rapid heating to 280 ◦C to volatilize the compounds into a GC capillary column through a fused-silicaline heated at 250 ◦C. The desorbed compounds were identified from the mass spectral data by using the US National Institute of Standards and Technology (NIST). The standard curves were produced with the mixed standard solutions, which were composed of only 9 VOCs, such as benzene, toluene, ethylbenzene, p-xylene, m-xylene, o-xylene, styrene, butyl acetate, and undecane. The identifications of these 9 VOCs were confirmed by their respective chromatographic retention time, and their quantifications were based

on a multipoint external standard curve. The response factor of toluene was utilized for quantification of other VOCs and TVOC [22,29].

Aldehydes and ketones in the vehicles were sampled by active sampling using controlled flow pumps at a rate of 400 mL/min for 30 min onto an adsorption tube coated with 2,4-dinitrophenylhydrazine (DNPH). The samples obtained were stored at 4 ◦C and protected from light in a refrigerator until analysis. Adsorbed aldehydes and ketones were extracted using acetonitrile (HPLC grade) and analyzed by high performance liquid chromatography (HPLC) using a UV detector (360 nm). The HPLC analysis was performed using acetonitrile/water elution (60%/40%, *v*/*v*) as a mobile phase at a flow rate of 1.0 mL/min, an injection volume of 25 μL, and a column temperature of 40 ◦C. The standard curves were produced with the mixed standard solutions, which were composed of 14 aldehydes and the DNPH derivatives of the ketones, such as formaldehyde-DNPH, acetaldehyde-DNPH, acrolein-DNPH, acetone-DNPH, propionaldehyde-DNPH, butenal-DNPH, butanone-DNPH, methacrolein-DNPH, butyraldehyde-DNPH, benzaldehyde-DNPH, valeraldehyde-DNPH, methylbenzaldehyde-DNPH, cyclohexanone-DNPH, and n-hexanal-DNPH. The identifications of these 14 aldehydes and ketones were confirmed by their respective chromatographic retention time, and their quantifications were based on a multipoint external standard curve.

#### **3. Results and Discussion**

#### *3.1. Temperature and Relative Humidity in Vehicles of Brand A*

The interior temperature and relative humidity in the tested vehicles remained constant in the environment test chamber. however, they were tested in sunny conditions, which would vary with external conditions, such as temperature, whether it was sunny or cloudy out, whether it was windy, etc. Thus, in order to increase the emission of VOCs from vehicle interior trims and reduce the air exchange rate inside and outside the vehicles to the greatest extent, a sunny and windless day in summer, as a "worst case" scenario, was chosen as the sampling day. For a better understanding of the temperature and relative humidity which could be reached inside the vehicles, temperature and relative humidity data inside the tested vehicles were recorded during enclosure and air sampling. As shown in Figure 1, in sunny conditions, the vehicles' interior air temperature ranged from 28.7 ◦C to 61.5 ◦C instead of remaining at 25 ◦C, and the relative humidity ranged from 11.8% to 26.1% instead of remaining at 50%. The temperature in the vehicle was inversely related to the relative humidity. Without the interference of external conditions, the longer the enclosure time, the higher the temperature in the car.

**Figure 1.** The changes in temperature and relative humidity in sunny conditions inside the vehicles of (**a**) A1-1, (**b**) A2-1 and (**c**) A3-1 during enclosure and air sampling.

#### *3.2. Interior Concentration changes in Vehicles of Brand A*

It had been commonly found that interior temperature was an especially important factor influencing the test results [22,29]. As shown in Figure 2, the ratios (CS/CE) of concentrations of the confirmed compounds and TVOC measured in Brand A in sunshine condition (CS) to that in the environment test chamber (CE) were greater than 1, indicating that the VOCs and TVOC pollution concentrations in the three model vehicles increased sharply when the temperature rose from 28.7 ◦C to 61.5 ◦C. In addition, as illustrated in Figure 3, when the in-vehicle temperature was 25 ◦C (in the environment test chamber), the concentrations of the eight in-vehicle confirmed compounds were all lower than their respective limited values in the national standard GB/T 27630-2011. However, when in-vehicle temperature increased, the concentrations of formaldehyde, acetaldehyde, and acrolein in Vehicle A1-1 (Figure 3a) and that of styrene, formaldehyde, acetaldehyde, and acrolein in Vehicle A2-1 (Figure 3b) were 1.89, 1.92, 1.44, 1.40, 7.84, 2.1, and 1.76 times more than their corresponding limited values in the national standard GB/T 27630. Therefore, in-vehicle high temperature was helpful for the evaporation and off-gassing of more VOCs from vehicle interior trims, for the main reason that the release amount of such VOCs as organic solvents, adhesives and additives contained in the interior trim materials could increase more when in-vehicle temperature rose [34,35]. In addition, the reports also showed that the concentrations of VOCs in vehicle interiors increased in concert with the temperature [25,27,36]. Reducing in-vehicle temperature could slow down the VOC emissions from the vehicles' interior materials.

**Figure 2.** The ratios of concentrations of confirmed compounds and TVOC measured inside the (**a**) Vehicle A1-1, (**b**) Vehicle A2-1, (**c**) Vehicle A3-1 in sunshine condition (CS) to that in the environment test chamber (CE). The black short dash corresponds to a value of 1.

**Figure 3.** The ratios of concentrations of eight confirmed compounds measured inside (**a**) Vehicle A1-1, (**b**) Vehicle A2-1, (**c**) Vehicle A3-1 in the environment test chamber (CE) with sunny outdoor conditions (CS) to their corresponding limited values (CV) of the national standard GB/T 27630, respectively. The black short dash corresponds to a value of 1.

#### *3.3. Interior VOC Type changes in Vehicles of Brand A*

Table 2 shows the types of the top 18 VOCs, excluding benzene, toluene, xylene, ethylbenzene, styrene, butyl acetate and undecane, identified in interior air samples of Vehicle A1-1, Vehicle A2-1, Vehicle A3-1, and changes of VOC types under different static conditions. In general, the most alkanes were in the three vehicles under each test condition, but there were more alkanes in the environment test chamber than in the sunshine condition. While in the sunshine condition, there were more other types of compounds, such as alcohols, ketones, benzenes, alkenes, aldehydes, esters, than that in environment test chamber (shown in Table 2). This was because the exposure of the vehicles to direct sunlight in the sunshine condition could lead to interior temperatures up to 61.5 ◦C, therefore the surface temperatures of the interior materials would be higher, which could cause the volatilization of various chemical substances with different boiling points from the interior surfaces. Thus, the chemical composition and types of VOCs were changed under different static conditions. In conjunction with high temperature, the transmission of solar radiation through glass windows could induce photochemical reactions and the production of degradation of byproducts, which could also cause changes in the chemical composition and types of VOCs.

An environment with a high concentration of VOCs could pose a very large health hazard to drivers and passengers. However, considering that few drivers and passengers would stay in such high-temperature vehicles, it was necessary to further study the changes in the concentrations and types of VOCs in the vehicles under driving conditions when they were exposed to direct sunlight.


**Table 2.** Types of the top 18 VOCs identified in interior air samples of Brand A (excluding benzene, toluene, xylene, ethylbenzene, styrene, butyl acetate and undecane).

#### *3.4. Interior Concentration changes in Vehicles of Brand B and C*

To further evaluate the differences in VOC concentrations in the vehicles' interior, air samples were also collected under driving conditions. According to the test results, the concentrations of eight confirmed compounds inside the vehicles of Brand B and C in the environment test chamber were commensurate, and the TVOC concentrations inside Brand C were obviously higher than that of Brand B. In addition, the concentrations of eight confirmed compounds and TVOC inside Brand B and C in the environment test chamber (CE) and under driving conditions (CD(0–30min), CD(60–90min)) were also compared and discussed.

As shown in Figure 4, in general, the concentration changes of eight confirmed compounds inside six vehicles of Brand B were CD(0–30min) > CD(60–90min) ≥ CE. But there were some exceptions. Inside the vehicle B1–1, the concentration change of benzene was CD(0–30min) = CD(60–90min) > CE, that of xylene was CD(60–90min) > CD(0–30min) > CE, that of ethylbenzene and styrene were CD(60–90min) > CD(0–30min) = CE, and that of acrolein was CE > CD(0–30min) = CD(60–90min). Inside the vehicle B1-2, the concentration change of formaldehyde was CD(0–30min) > CD(60–90min)>CE, while that of other compounds were CE > CD(0–30min) ≥ CD(60–90min). Inside the vehicle B2-1, the concentration change of styrene was CE > CD(0–30min) = CD(60–90min), and that of formaldehyde was CD(0–30min) > CE > CD(60–90min). Inside the vehicle B2-2, the concentration change of ethylbenzene was CE = CD(0–30min) > CD(60–90min). Inside the vehicle B3-1 and B3-2, the concentration change of acetaldehyde was CD(0–30min) > CE > CD(60–90min). Whereas, as shown in Figure 5, the change trends of concentrations inside six vehicles of Brand C were clearly different with that inside Brand B, which generally were CE > CD(0–30min) > CD(60–90min). But there were also some exceptions. Inside the vehicle C1-1, the concentration changes of toluene and formaldehyde were CD(0–30min) > CE > CD(60–90min). The concentration changes of acetaldehyde inside vehicle C1-2, that of formaldehyde and acetaldehyde inside C1-3 and that of toluene inside C1-4 were all CE > CD(60–90min) > CD(0–30min). Inside vehicle C1-4, the concentration change of xylene was CE > CD(0–30min) = CD(60–90min), that of acrolein was CD(60–90min)> CE = CD(0–30min). Similarly, as shown in Figure 6, the TVOC concentration changes inside the six vehicles of Brand B were all CD(0–30min) > CD(60–90min) > CE, while that inside Brand C were CE > CD(0–30min) > CD(60–90min), which had the same trend as that of eight confirmed compounds.

**Figure 4.** Concentration changes of eight confirmed compounds inside six vehicles of Brand B under different conditions ((**a**). Vehicle B1-1; (**b**). Vehicle B1-2; (**c**). Vehicle B2-1; (**d**). Vehicle B2-2; (**e**). Vehicle B3-1; (**f**). Vehicle B3-2). *Y*-axis indicates the concentration difference between CD(0–30min) and CE, CD(60–90min) and CE.

**Figure 5.** Concentration changes of eight confirmed compounds inside six vehicles of Brand C under different conditions ((**a**). Vehicle C1-1; (**b**). Vehicle C1-2; (**c**). Vehicle C1-3; (**d**). Vehicle C1-4; (**e**). Vehicle C1-5; (**f**). Vehicle C1-6)). *Y*-axis indicates the concentration difference between CD(0–30min) and CE, CD(60–90min) and CE.

**Figure 6.** Concentration changes of TVOC inside six vehicles of (**a**) Brand B and (**b**) C under different conditions. *Y*-axis indicates the concentration difference between CD(0–30min) and CE, CD(60–90min) and CE.

As the air conditioning mode inside the tested vehicles was adjusted to internal circulation, and the infiltration air flow through joints and leaks in vehicle envelopes was the predominant airflow that could affect pollutant transportation inside vehicle cabins [17,37]. Under driving conditions, the vehicles ran at a certain speed, which could accelerate the air exchange inside and outside vehicles. Therefore, the concentration changes of VOCs and TVOC inside the vehicles should be CE > CD(0–30min) > CD(60–90min) in theory, which was consistent with the results tested inside the vehicles of Brand C. But based on the test results of Brand B, except for some substances inside the vehicle B1-1 and B1-2, CD(0–30min) were generally higher than CE and CD(60–90min). That was because the six vehicles of Brand B were all parked and enclosed under direct sunlight (Table 1) before the experiments were carried out under driving conditions, which could accelerate volatilization of VOCs from vehicle interior trims due to the high temperature inside the vehicles. Therefore, it was reasonable that more pollutants were collected inside the vehicles of Brand B at the first 30 min under driving conditions than that in environment test chamber, even though the temperature was the same under these two conditions. However, although the vehicle C1-5 and C1-6 were also parked and enclosed under direct sunlight for 2 h before running on the vehicle proving ground, the CD(0–30min) of VOCs and TVOC were actually lower than their CE. That was because different vehicle brands have different air exchange rates. And the higher speed of the vehicles of Brand C than that of Brand B could lead to the increase of the air exchange rate inside and outside Brand C on one hand, while on the other hand, there might be (no air sampling) less pollutants emitted from the interior trims of Brand C than that of Brand B under direct sunlight, and then the concentration of pollutants inside Brand C would decrease rapidly at a higher air exchange rate.

As the vehicles continued to run at a certain speed, the concentration of VOCs and TVOC inside the vehicles would decrease with time due to the air exchange inside and outside the vehicles. Thus, CD(60–90min) of VOCs and TVOC inside Brand B were less than their CD(0–30min). But the air exchange rate value determined whether a longer driving process was required to reduce the concentration of airborne pollutants in the vehicles to be consistent with or lower than that in the environment test chamber. According to the test results shown in Table 3, the CD(60–90min) of eight compounds and TVOC were 1.00~3.27 times and 0.99~7.93 times more than their CE, demonstrating that after a period of driving, the concentrations of air pollutants inside the vehicles of Brand B were decreased to the same level with or close to that in the environment test chamber (Figures 4 and 6a). If the air conditioning mode was switched to external circulation (the concentration of ambient air pollutants should meet the requirements in Section 2.2.1), it would take a shorter time to reduce the concentration of air pollutants in the vehicles. The above results further indicated that the concentration values obtained by the standard method (in environment test chamber) were close to the actual exposure level for drivers and passengers.


**Table 3.** The ratios of CD(60–90min) and CE of eight confirmed compounds and TVOC measured inside Brand B.

"/" means that the value of CE was not detected.

#### *3.5. Interior VOC Type changes in Vehicles of Brand B and C*

The difference in the types and boiling points of VOCs inside the vehicles of Brand B and C would also lead to the difference in the trend of TVOC concentration changes inside these vehicles in actual driving conditions. It would take a longer driving time to reduce the TVOC concentration inside these vehicles to a level that was consistent with or lower than that in the environment test chamber.

Tables 4 and 5 showed types of the top 10 VOCs identified inside the vehicles of Brand B and C, and changes of VOCs types under different test conditions. For the vehicles of Brand B, alkanes were the main pollutants. But for the two vehicles of each model in Brand B, there was no difference in the VOC types under different conditions. Whereas inside the vehicles of Brand C, benzenes and alkanes were the main pollutants. But in the environment test chamber, there were more benzenes and less alkanes than that in driving conditions. In addition to the effect of temperature on the air quality inside the vehicles, air exchange rate inside and outside the vehicles was also the factor influencing the VOCs types.


**Table 4.** Types of the top 10 VOCs identified in interior air samples of Brand B.

**Table 5.** Types of the top 10 VOCs identified in interior air samples of Brand C.



#### **Table 5.** *Cont.*

#### **4. Conclusions**

The following conclusions can be drawn through the comparison and discussion of the quantitative and qualitative analysis results of VOCs in the vehicle interior air of Brand A~C under different conditions:

(1) The in-vehicle high temperature in the sunshine condition causes higher concentrations and more different types of VOCs released from the interior materials with different boiling points of VOCs than that in environment test chamber. In the vehicles of Brand A, more alkanes were identified in the environment test chamber, while more other types of compounds, such as alcohols, ketones, benzenes, and esters, were identified in the sunshine condition. In the vehicles of Brand C, there are aromatic hydrocarbons and alkanes, and the number of aromatic hydrocarbons is slightly more than that of alkanes. While in Brand B, there are mainly alkanes without aromatic hydrocarbons. Therefore, the drivers and passengers should ventilate their vehicles and turn on the air conditioner to cool down the interior before using a vehicle exposed to the sunshine.

(2) Because of different in-vehicle temperatures, driving speeds, air exchange rates, and types of substances with different boiling points inside the vehicles, the change trends of eight typical organic pollutants and TVOC concentrations in the air of Brand B and C tested vehicles under different conditions are also different, namely:

The vehicles of Brand B: CD(0–30min) > CD(60–90min) ≥ CE,

The vehicles of Brand C: CE > CD(0–30min) > CD(60–90min).

The existence of individual exceptions is not excluded.

(3) There are no requirements with regard to the limited values and measurement methods for VIAQ under high temperature in the national standard GB/T 27630 and HJ/T 400. However, according to the results in this study, the concentrations of eight typical organic compounds inside the vehicles exposed to high temperature after a period of driving will be reduced to the same level as or tend to that in the environmental chamber test, indicating that the results of standard test methods are close to the actual driving exposure level. However, due to the variety of pollutants, it takes a longer time for the TVOC concentration to reduce to the same level as that in the environmental chamber test. **Author Contributions:** Conceptualization, R.G. and Z.Z.; data curation, R.G. and Y.L.; formal analysis, R.G.; funding acquisition, R.G., X.Z. and Y.L.; investigation, R.G. and J.S.; methodology, R.G.; validation, R.G., X.Z., Z.Z. and Y.L.; visualization, R.G. and X.Z.; writing—original draft preparation, R.G.; resources, Z.Z., W.H. and S.T.; writing—review and editing, Z.Z. and Y.L.; supervision, R.G. and S.T.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the BJAST Budding Talent Program (BGS202101), National Natural Science Foundation of China (No. 61874012) and Beijing Natural Science Foundation (No. 3224063).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **Nomenclature**


#### **References**


## *Article* **Volatile Organic Compounds in Finnish Office Environments in 2010–2019 and Their Relevance to Adverse Health Effects**

**Kaisa Wallenius \*, Hanna Hovi, Jouko Remes, Selma Mahiout and Tuula Liukkonen**

Finnish Institute of Occupational Health, P.O. Box 40, FI-00032 Työterveyslaitos, Finland; hanna.hovi@ttl.fi (H.H.); jouko.remes@ttl.fi (J.R.); selma.mahiout@ttl.fi (S.M.); tuula.liukkonen@ttl.fi (T.L.) **\*** Correspondence: kaisa.wallenius@ttl.fi

**Abstract:** We gathered recent (2010–2019) data on the VOC and formaldehyde levels in Finnish non-industrial indoor work environments. The data comprised 9789 VOC and 1711 formaldehyde samples collected from the indoor air of offices, schools, kindergartens, and healthcare offices. We assessed the health risks by comparing the measured concentrations to the health-based RW I/II and EU-LCI reference values. The concentrations of individual VOCs and formaldehyde in these work environments were generally very low and posed no health risks. Total VOC concentration (TVOC) as well as concentrations of several individual compounds, including aromatic compounds, alkanes, 2-ethyl-1-hexanol, and formaldehyde, showed clearly decreasing trends. In contrast, several aldehydes, acids, and a few other compounds showed increasing trends. However, the increasing trends did not seem to affect the higher ends of the distributions, as the 95th percentile values remained fairly stable or decreased over the years. The VOC patterns in the environments of the offices, schools, kindergartens, and healthcare offices varied, probably reflecting the differences in typical activities and the use of materials. However, we do not expect these differences to be relevant to health outcomes.

**Keywords:** indoor air quality; VOC; formaldehyde; office; health risk; trend

## **1. Introduction**

Volatile organic compounds (VOCs) are a versatile group of chemical compounds that are present both indoors and outdoors. Indoor air has a greater variety of different VOCs than outdoor air, and the concentrations of these compounds are typically higher indoors than outdoors [1–4]. The differences in the composition of indoor and outdoor air derive from greater human impact, a larger surface area to volume ratio, and lower levels of light and oxidising agents indoors [5]. Typical groups of the organic compounds present in indoor air are alkanes, terpenes, aromatic hydrocarbons, and aldehydes [4,6].

The spatial and temporal variation of volatile compounds in indoor environments is great. This variation is governed by both natural and man-made sources of emissions both indoors and outdoors. Indoor emission sources include building and interior materials, technical systems and equipment, buildings' occupants, and a large variety of chemicals, products, and materials used and produced by occupants in different activities. The main outdoor emission sources are traffic and industrial processes. VOC levels are impacted by the seasons [3,7] and the geographic locations of building [6–9]. The long-term chemical composition of indoor environments changes as new technologies and materials are developed [10–12].

Concerns over possible health risks related to VOC emissions in the indoor air are prevalent. Many VOCs and formaldehyde are known to be hazardous and to pose health risks in high concentrations in, for example, industrial work environments. The same does not necessarily apply to non-industrial environments if the concentrations remain below levels that induce health hazards.

**Citation:** Wallenius, K.; Hovi, H.; Remes, J.; Mahiout, S.; Liukkonen, T. Volatile Organic Compounds in Finnish Office Environments in 2010–2019 and Their Relevance to Adverse Health Effects. *Int. J. Environ. Res. Public Health* **2022**, *19*, 4411. https://doi.org/10.3390/ ijerph19074411

Academic Editors: Ashok Kumar, M Amirul I Khan, Alejandro Moreno Rangel and Michał Piasecki

Received: 17 February 2022 Accepted: 24 March 2022 Published: 6 April 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The aim of the study was to (1) compile recent measurement data on the levels of VOCs and formaldehyde in offices and similar non-industrial indoor work environments in Finland and (2) to evaluate the health relevance of these. We also extended the study to analyse trends during the study period (2010–2019) and to inspect the differences between the environments of offices, schools, kindergartens, and healthcare offices.

To analyse the VOCs, we used the definition and analytical procedure described in the ISO 16000-6:2011 standard, which defines VOCs as substances that elute between n-hexane and n-hexadecane under defined chromatographic conditions. To analyse formaldehyde, which belongs to very volatile organic compounds (VVOC), we used an ISO 16000-3:2011-based analytical procedure.

#### **2. Materials and Methods**

#### *2.1. Description of Data*

The data of this study comprised results from indoor air samples of VOCs and formaldehyde from Finnish offices and similar non-industrial work environments analysed at the laboratory of the Finnish Institute of Occupational Health (FIOH) between January 2010 and December 2019. The total number of samples to measure VOCs was 9789, collected from offices (3872 samples), schools (3583 samples), kindergartens (727 samples), and healthcare offices (1607 samples). Healthcare offices were office-like spaces in healthcare premises where patients were not treated. The total number of samples for measuring formaldehyde was 1711, collected from offices (521 samples), schools (938 samples), kindergartens (68 samples), and healthcare offices (184 samples). The yearly distributions of VOC and formaldehyde samples are presented in Table 1.

**Table 1.** The yearly numbers of VOC (**a**) and formaldehyde (**b**) samples collected from office, school, kindergarten, and healthcare office environments.


<sup>(</sup>**a**) VOC samples.

(**b**) Formaldehyde samples.



**Table 1.** *Cont.*

The sampling was performed by experts from a multitude of consultant engineering offices (customers of FIOH's analytical laboratory) as part of indoor air services, without a systematic monitoring programme. The samples were collected from occupied buildings during all seasons and from different parts of Finland. Winter (December, January, February) and spring (March, April, May) seasons accounted for approximately 60% (30% each) of all the VOC samples collected, whereas summer (June, July, August) and autumn (September, October, November) were less active sampling seasons with the shares of 15% summer and 24% autumn. Similar to VOCs, the seasonal distribution of formaldehyde samples was 32% winter, 27% spring, 17% summer, and 25% autumn.

The majority of the samples were collected from buildings with suspected indoor air-quality problems. In the buildings in which VOCs were measured, the suspected problems did not necessarily concern VOC pollution. In contrast, formaldehyde was measured in buildings in which the suspected problems specifically concerned formaldehyde. The laboratory did not record the age, the renovation status, or technical characteristics of the buildings, but we assume that the studied premises represent well the relatively homogenous Finnish building stock of public and office buildings built in 1960–2000. The vast majority of public and office buildings in Finland are equipped with mechanical supply and exhaust ventilation. The average number of samples per one sampling occasion (forming one laboratory assignment) was 3.1 for VOC (3167 sampling occasions) and 3.3 for formaldehyde (520 sampling occasions). The samples from single sampling occasions were mainly collected from one building but from different rooms. Thus, the number of sampling occasions gives a rough estimate of the number of sampled buildings.

#### *2.2. Sampling and Analysing VOCs*

The VOCs were sampled and analysed in accordance with the ISO 16000-6:2011 standard. The samples were collected in Tenax TA or Tenax TA/Carbograph 5 TD tubes from central room locations 1–1.5 m above floor level, mainly (99.5%) by active sampling (pump). Diffusive sampling was used for approximately 0.5% of the samples. We analysed the samples using thermal desorption and GC/MS. The compounds were identified by comparing them to pure reference substances and/or the Wiley or NIST mass spectral databases. FIOH's laboratory has approximately 150 different compounds calibrated with their own response factors. Calibrated compounds cover most of the common compounds in indoor air. Some individual compounds that are not frequently identified or quantified from samples, and some mixtures, such as higher boiling alcohols or hydrocarbon mixtures, are quantitated as toluene equivalents because reference compounds are not available. All the individual compounds whose results are presented in this article were quantified with substance-specific responses. The total concentration of volatile organic compounds (TVOC) was determined for each sample as a toluene-equivalent concentration. TVOC was determined as the area in the chromatogram between the n-hexane and n-hexadecane, including both substances.

FIOH's laboratory has ISO 17025 accreditation. The laboratory provided detailed sampling instructions and calibrated sampling equipment for the experts who performed sampling. The recommended sampling volume in active sampling was 7–12 dm3. Typical sampling volume was 9 dm3, which is equal to 45 or 90 min of sampling time. Diffusive sampling times were 2 to 4 weeks. The laboratory evaluated sampling volumes and times for every sample.

#### *2.3. Sampling and Analysing Formaldehyde*

Formaldehyde was sampled and analysed in accordance with the ISO 16000-3:2011 and the ISO 16000-4:2011 standards. The samples were collected in 2,4-dinitrophenylhydrazine (DNPH)-coated silica gel cartridges (Sep-Pak® DNPH-Silica Plus Short Cartridge) using active sampling (91%) or in DNPH-coated passive samplers (UMEx 100 Passive Sampler for formaldehyde) using diffusive sampling (9%). The DNPH derivate of formaldehyde was desorbed using acetonitrile and then analysed with a high-performance liquid chromatography (HPLC) that contained a UV detector. Formaldehyde was identified and quantitated with reference compounds.

The laboratory provided detailed sampling instructions and calibrated sampling equipment for the experts who performed sampling. Recommended sampling volume in active sampling was 100 dm3, which is equal to 100 min of sampling time. Diffusive sampling time was 24 h. The laboratory evaluated sampling volumes and times for every sample.

#### *2.4. Statistical Methods*

First, we calculated the frequencies of the different analytes. Frequency indicates the percentage of samples whose results exceed the limit of quantitation (LOQ). To describe the distribution of the variables, we calculated the 50th percentile (median, Md), 90th percentile (P90), 95th percentile (P95), and 99th percentile (P99), first covering all the samples and then separately for different indoor environments and different years (2010–2019). For percentiles, we calculated the 95% confidence intervals based on normal distribution.

The association between the frequency of the different measures in the different types of environments was analysed using log-binomial regression analysis. The office was used as a reference group. We dichotomised the values such that <LOQ values were 0, and ≥LOQ values were 1. In log-binomial regression analyses, we compared percentages (value 1) separately school/kindergarten/healthcare offices to offices' percentages. We calculated the crude risk ratios (RR) and their 95% confidence intervals and made no adjustments.

We analysed the frequency and concentration trends of each compound. The frequency trend was tested using Somer's D test, in which the input data were binomial (measurement result either ≥LOQ or <LOQ). The concentration trend was tested using Somer's D test, in which the input data comprised all the measured concentration values. Here, measurement results that were <LOQ were replaced by a concentration value of LOQ/2. Statistically significant trends were evaluated using the Z value. This statistic is used to test the null hypothesis, according to which no trend exists. A positive Z indicates an increasing trend in a time-series, while a negative Z indicates a decreasing trend. We used the SPSS and SAS statistical software packages for our calculations.

#### *2.5. Health Risk Assessment*

Health-based RW I/II and EU-LCI values were used to evaluate the possible health risks of VOCs and formaldehyde at the concentration levels measured. Both reference values have been set by experts using epidemiological and toxicological data. We primarily used the RW I/II values. If the RW values were not available, we used the European EU-LCI values if they were available.

The German RW I and II values [13,14] are reference values for indoor air-quality assessment that are set to cover the entire population and long-term exposure. In Germany, they are used in assessment of indoor air quality in homes, offices, schools, and other public spaces. The RW II values represent concentration levels that may cause adverse health effects for individuals who are sensitive due to their health status, for example, if they are exposed to these concentration levels for extended periods of time. The RW I values represent concentration levels which, in the light of current knowledge, do not cause adverse health effects even in cases of lifetime exposure.

The EU-LCI values (LCI, Lowest Concentration of Interest) are intended for assessing the safety of construction products regarding the potential health risks posed by inhaling emissions from new products [15]. The values aim to prevent health risks and cover the lifetime exposure of the entire population. They therefore represent concentration levels that, in the light of current knowledge, are considered unlikely to cause adverse health effects even in the long term. However, the EU-LCI values are not intended for use as reference values for assessing indoor air quality; they are for assessing building material emissions under experimental conditions. Nonetheless, as they were derived using the principles of toxicological hazard assessments, we considered them adequate for use in our study as indicative reference values when RW values were not available.

Health risk assessment was performed by calculating risk characterisation ratios (RCRs; [16]) for the 42 most frequently detected VOCs. The calculations were made by dividing the 99th percentile (P99) of the measurements by the above-mentioned health-based reference values. The RCRs were considered to indicate a potential health risk if they exceeded 1.

#### **3. Results**

#### *3.1. Statistics of Entire Ten-Year Data*

Around 400 different analytes (single compounds or mixtures that cannot be separated by the measurement protocol) were detected and quantitated at least once in the ten-year VOC data, which consisted of 9789 air samples. Formaldehyde data collected and analysed using a different procedure from that used for VOC during the same study period consisted of 1711 air samples. The LOQ for different compounds ranged between 0.3 and 1 μg/m3. The number of VOCs that had a frequency of ≥10% was 42. Table 2a presents the descriptive statistics of these 42 VOCs and the TVOC sum variable. Table 2b presents the corresponding statistics for the formaldehyde measurements. In addition to the measurement data, Table 2 is supplemented with up-to-date health-based reference values (EU-LCI, RW I, and RW II) for respective compounds if available.

The median values of the 42 most common VOCs were mostly below the respective LOQ: only 12 compounds had a frequency greater than 50% (Table 2a). P90 values were typically only slightly above the respective LOQ: 0.5–3 μg/m3. The largest P90 value was observed for 1,2-propanediol, at 7 μg/m3. P95 values ranged from 0.6 μg/m3 (octane) to 13 μg/m<sup>3</sup> (1,2-propanediol). P99 values ranged from 1 μg/m<sup>3</sup> (octane, nonane) to 45 μg/m3 (1,2-propanediol). The maximum values were tenfold to hundredfold larger than the P99 values for most of the compounds. The maximum values ranged from 12 μg/m3 (octane) to 1110 μg/m3 (xylenes p, m).

None of the RCRs, representing possible health risks, exceeded 1 when the P99 values of the 42 common VOCs were compared with their respective RW I, RW II, and EU-LCI values; all P99 values were well below their respective health-based reference values (Table 2a). The highest RCR compared to the RW I values, 0.75, was found for 1,2-propanediol.

No RW I/II or EU-LCI values exist for the alkanes nonane, octane, and 2,2,4,6,6 pentamethylheptane. All their P99 values (1–6 μg/m3) remained notably below the EU-LCI value of heptane (15,000 μg/m3), which belongs to the same group of alkanes. Based on the similar chemical properties and health hazard profiles of all these four alkanes, we can assume that this reference value would probably be in the same range as their own EU-LCI values. No RW I/II or EU-LCI values are available for benzene, but the UBA derived risk-related guide values for carcinogenic substances in indoor air [13]. The risk-related value for benzene is 4.5 μg/m3 (preliminary value), which is above the P99 value of 3.0 μg/m3 [14].

The maximum values of several of the 42 most common VOCs exceeded their RW I values. The maximum values for xylenes, 1,2-propanediol, 2-(2-ethoxyethoxy)ethanol, and phenoxyethanol also exceeded the RW II and/or EU-LCI values. In addition, the maximum value of benzene exceeded its risk-related value.

The TVOC sum variable has no health-based reference values. The health relevance of TVOC can only be meaningfully evaluated if the components of the total concentration are known.


(**a**) Statistics of 42 most frequently detected VOCs in air samples collected from Finnish office and similar non-industrial indoorenvironments in 2010–2019. Bottom line presents TVOC sum variable statistics. VOCs were sampled and analysed in accordance



**Table 2.** *Cont.*


**Table 2.** *Cont.*


**Table 2.** *Cont.*

The formaldehyde results revealed a shape of distribution that was different from most of the VOCs (Table 2b). For formaldehyde, the gap between P99 and the maximum values was notably smaller, and the proportion of the results that fell below the LOQ was also smaller than that for most VOCs. The formaldehyde data represented buildings/indoor locations that indoor air specialists anticipated to be at risk of elevated formaldehyde concentrations. Therefore, it is likely that in this sample of buildings, formaldehyde occurred more frequently and in higher concentrations than among the normal population of this type of building. Still, the maximum concentration of formaldehyde measured was also below the respective health-based RWI and EU-LCI reference values.

The majority of all the VOCs detected during the ten year-study period were present in less than 10% of the samples. These infrequent VOCs represented a wide variety of different organic compounds. Chlorinated and other VOCs, which are often studied and reported in scientific publications due to their related health concerns, were selected from the data mass and are presented in Table 2c. These chlorinated VOCs were detected very rarely in the ten-year VOC data. The frequency of aromatic styrene was also fairly low. The P99 values of carbon tetrachloride, chloroform, 1,4-dichlorobenzene, and trichloroethene were below the limit of quantitation. The P99 values for tetrachloroethene (0.4 μg/m3) and styrene (1.0 μg/m3) in turn were far below their RW I and EU-LCI values. It should be noted that Tenax TA has a rather low breakthrough volume for halogenated compounds, such as carbon tetrachloride, trichloroethene, and chloroform, so their concentrations are approximate.

#### *3.2. Differences between Indoor Environments*

Table 3a shows the differences between the four studied indoor environment types in terms of the frequency and P95 values of the 42 most common VOCs and the TVOC sum variable. Table 3b presents the corresponding information on formaldehyde. The comparison of environments was based on P95 values instead of medians or other lower percentiles because medians of most individual VOCs were below LOQ in all environments, and many of the P90 values were only slightly above LOQ. Geometric or arithmetic means were also regarded as unfit for this data due to the high percentages of measurement results below LOQ. Median and P90 values are presented as additional information for TVOC and formaldehyde, where they have informative value.

The alkanes heptane, nonane, and octane were detected more frequently in the offices (freq. 17–18%) than in the other studied indoor environments (freq. 10–13%), whereas 2,2,4,6,6-pentamethylheptane was more frequently found in healthcare offices (freq. 16%) than elsewhere (freq. 10%). The P95 values were ≤<sup>2</sup> <sup>μ</sup>g/m3 for all alkanes in all the environments. The differences between the P95 values in the offices and other environments were minor but mainly statistically significant.

Aromatic VOCs, namely benzene, ethylbenzene, 1,2,4-trimethylbenzene, xylenes (p, m, o), toluene, and benzaldehyde (presented under the aldehyde group), were more frequent in the office environments than in the other studied environments. For most of the aromatic compounds presented, the lowest frequencies were measured in the kindergartens. For example, xylenes (p, m) had a frequency of 69% in the offices and 47% in the kindergartens. Despite notable differences in frequency, the differences between the P95 values of the aromatics in the environments were small. The highest P95 values in the aromatic group were observed for toluene: 4–5 μg/m3.

Terpenes 3-carene and α-pinene were more frequent in the kindergartens than in the other environments, whereas limonene was the most frequent in the office environment. The lowest frequencies of terpenes were typically observed in the healthcare office environment. The differences between the P95 values were small although mainly statistically significant between office and other environments. The highest P95 values within the terpene group were observed for α-pinene: 6–9 μg/m3.



**Table 3.**

*Cont.*


*Int. J. Environ. Res. Public Health* **2022**, *19*, 4411

**Table**

**3.**

*Cont.*


**Table 3.** *Cont.*

Alcohols generally had the lowest frequencies in the healthcare offices, but the differences between the environments were small. Alcohol ethers and phenol ethers also had the lowest frequencies in the healthcare offices, but their differences to other environments were more prominent. The only representative of the phenol group, namely phenol, had the lowest frequency in the kindergartens, but the differences between the environments were small. The P95 concentrations of alcohols, phenols, alcohol ethers, and phenol ethers were generally rather similar in all the environments. The biggest difference between the P95 concentrations in the different environments was observed for 2-(2-ethoxyethoxy)ethanol and ranged from 1 μg/m3 (healthcare office) to 7 μg/m<sup>3</sup> (school). The highest P95 concentrations within these VOC groups were observed for 1,2-propanediol: 12–14 μg/m3.

Apart from aromatic benzaldehyde, the VOCs in the aldehyde group, namely decanal, hexanal, heptanal, nonanal, octanal, and pentanal, had notably higher frequencies in the kindergartens than in the other environments. The P95 concentrations of aliphatic aldehydes were also systematically slightly higher in the kindergartens than in the other environments. The greatest difference observed in P95 values was for nonanal, with a P95 of 11 μg/m<sup>3</sup> in the kindergartens and 5–6 μg/m3 in the other environments.

The only representative of the ketone group, namely acetophenone, had rather similar frequencies (range 21–25%) and P95 concentrations (range 0.8–1 μg/m3) in all the studied environments.

The three carboxylic acids presented under the acids group, namely hexanoic acid, pentanoic acid, and propionic acid, had the highest frequencies in the office environment and the lowest frequencies in the healthcare offices. The P95 values of acids were also slightly lower in the healthcare offices than in the other environments. The largest difference between the P95 values was seen in hexanoic acid, with a P95 value of 5 μg/m<sup>3</sup> in the healthcare offices and 7–8 μg/m3 in the other environments.

As regards esters, healthcare office environment displayed generally the lowest frequencies, whereas the environments with the highest frequencies varied. The kindergarten environment stood out with a substantially higher TXIB frequency (57%) than in any other environments (19–21%). n-butyl acetate also displayed a notably higher frequency (20%) in the kindergartens than in the other environments (11–15%). The differences between the frequencies of the other esters in the environments were small. TXIB had the largest difference in P95 values, with a P95 value of 6 μg/m3 in the kindergartens and 2–3 μg/m3 in the other environments.

The only representative of Si-compounds, namely decamethylcyclopentasiloxane, displayed rather high frequencies (65–76%) in all the environments. Its P95 concentration ranged from 9 μg/m3 in the school environments to 12–13 μg/m<sup>3</sup> in the other environments.

The total sum of the VOCs was rather similar in the different environments, as demonstrated by the TVOC median, P90, and P95 values in Table 3a.

The frequency of formaldehyde was over 90% in all the studied environments (Table 3b). The P95 values of formaldehyde ranged from 14 μg/m3 (school) to 25 μg/m<sup>3</sup> (office). Similarly, the lowest median and P90 values of formaldehyde were observed in the schools and the highest in the offices.

#### *3.3. Trends*

The total concentration of VOCs (TVOC) showed a strongly decreasing trend over the ten-year study period, 2010–2019. The shift in TVOC distribution is graphically presented in Figure 1. Table 4a presents the annual statistics of the 42 most common VOCs and the TVOC sum variable. Table 4b shows the corresponding information on formaldehyde. For each compound, we analysed the trends of both frequency (i.e., the share of quantifiable measurement results) and concentration. Statistically significant (*p* < 0.0001) trends are indicated by downward and upward arrows. The Z values shown below the arrow symbols indicate the direction and steepness of the trend.

**Figure 1.** The shift in TVOC distribution in 2010–2019. The four different colour patterns represent the four quarters between the data points of Min, P25, Md, P75, and Max of the ten-year TVOC data (*n* = 9789). It can be seen from the graph that the share of the first quarter (Min–P25) increased and that the share of the top quarter (P75–Max) decreased during the study period.

The frequency and concentration trends of all four alkanes were clearly decreasing. This means that they have become rarer findings and that their concentration levels decreased during the ten-year study period. The most prominent decreases in frequency were observed for nonane and octane, which had 24% frequency in 2010 but only 4–5% frequency in 2019. The P95 values of the alkanes decreased during the study period although they were already minor (≤<sup>1</sup> <sup>μ</sup>g/m3) at the beginning of the study period.

Like those of the alkanes, the frequency and concentration trends of aromatic hydrocarbons were either strongly or moderately decreasing. The most prominent decrease in frequency was observed for benzene, which was present in quantifiable concentrations in nearly all the samples (98%) in 2010 but in only half of the samples (49%) in 2019. The frequency of xylenes (p, m) and ethylbenzene also fell considerably, from 84% to 52% and from 47% to 18%, respectively. These decreasing trends were reflected at the higher ends of the distributions, as demonstrated by the gradual decreases in the P95 concentrations of aromatic compounds.

Both the frequency and concentration of terpenes 3-carene and β-pinene showed slightly increasing trends. However, this was not reflected in their P95 values, which remained fairly stable over the years. The terpenes limonene and α-pinene showed no concentration trends, and α-pinene had only a very slightly increasing frequency trend. Interestingly, the P95 values of limonene decreased gradually from 5–6 μg/m3 to around 2 μg/m3 during the study period although there was no overall concentration or frequency trend.

Apart from 2-ethyl-1-hexanol, the alcohol compounds displayed a slightly increasing frequency trend and either no concentration trend or a slightly increasing concentration trend. In contrast, 2-ethyl-1-hexanol displayed clearly decreasing frequency and concentration trends. The P95 values of alcohols varied annually but no obvious decreases or increases were observed over the ten years.


(**a**) Annual frequency and P95 values of the 42 most common VOCs during the ten-year study period 2010–2019. For TVOC presented on thebottomline,median,andP90valuesarealsoshown.


ང

ང

ང

ང

ང

ང

ང

 ང

 ང

ང

 

 

 ང

 

 ང

 

ང

 

 

 ང

 

 ང

 

ང

 

 

 ང

 

 

 ང

 

 

ང

 

 

 

 

 

 

 ང

 ང

ང

 

 

 

 

 

 

 

 

 

 

 ང

 

 ང

 

ང

 

 

 

 

 

 

 ང

 

 

 

 ང

 

 

 

ང

 

 

 

 

 


**Table 4.**

ང

 ང

 ང

ང

ང

ང

*Cont.*

 ང  ང

 

 

ང

ང

ང

 ང

 ང

ང

 

 

 

 

 

 

   

 

 

 

 ང

 

ང


ང

ང

ང

ང

ང

ང

ང

ང

 ང

 ང

ང

 ང

 ང

ང

 ང

 ང

 ང

ང

ང

ང

ང

ང

 ང

 

ང

ང

 

ང

 ང

 

 ང

 

 

 

 ང

 

 ང

 

 ང

 

ང

ང

 

 ང

 

 


*Int. J. Environ. Res. Public Health* **2022**, *19*, 4411

**Table 4.** *Cont.*

ང

ང

ང

 ང

ང

 ང

ང

 ང

 ང

ང

ང

ང

ང

ང

ང

ང

ང

 ང

ང

 ང

ང

 ང

 ང

ང

ང

ང

ང

 ང

 ང

ང

 

 

   

 

 

 

 ང

 

ང

 

 

 

 ང

 ང

 

ང

 

 

 

 ང

 ང

ང


ང

ང

ང

 ང

 ང

    ང

 

ང

 

ང

 ང

 

 

 

 ང

 

 

 ང

 

 

 

 

 

Phenol displayed moderately increasing frequency and concentration trends. However, its P95 concentration remained rather steady at around 1 μg/m3 over the ten years.

Apart from 2-butoxyethanol, alcohol and phenol ethers displayed slightly decreasing frequency and concentration trends or none at all. The P95 concentrations of alcohol and phenol ethers slightly decreased during the study period. The frequency and concentration trends of 2-butoxyethanol were moderately increasing. However, the P95 concentration of 2-butoxyethanol did not increase over the ten years.

All seven aldehydes displayed increasing frequency trends during the study period, whereas the concentration trends varied. Interestingly, benzaldehyde displayed completely opposite frequency and concentration trends. This means that benzaldehyde is increasingly being detected but that its concentration level is decreasing. Despite some aldehydes showing a slightly increasing concentration trend, the P95 concentrations of aldehydes showed either a slightly decreasing trend or remained fairly stable over the ten years.

The ketone compound acetophenone displayed strongly increasing frequency and concentration trends. However, the P95 concentration of acetophenone stayed at ≤ <sup>1</sup> <sup>μ</sup>g/m<sup>3</sup> throughout the study period.

Hexanoic, pentanoic and propionic acids showed similar trends of increasing frequency and concentration levels. However, the P95 concentrations of acids decreased during the study period.

Esters showed varying trends. 2-(2-Butoxy ethoxy)ethyl acetate displayed slightly decreasing frequency and concentration trends. n-Butyl acetate showed no significant trends. Ethyl acetate, Texanol, and TXIB displayed slightly increasing frequency trends from 11–16% to around 20%. We also found slightly increasing trends in the concentration levels of Texanol and TXIB. At the same time, the P95 values of all esters remained fairly stable over the ten years.

Decamethylcyclopentasiloxane showed a slightly increasing frequency trend, rising from around 60% to over 70%. At the same time, the concentration trend was slightly decreasing, and the P95 values of this Si-compound varied in the range of 7 to 16 μg/m3, with no obvious trend.

Table 4b presents the annual statistics for formaldehyde. The frequency of formaldehyde displayed no trend, as it was high, 89–100%, every year. At the same time, the concentration level of formaldehyde showed a decreasing trend. The decrease in concentration was evident in all parts of the distribution, as demonstrated by the fall in the median, P90, and P95 values from 9 to 3 μg/m3, from 25 to 9 μg/m3, and from 43 to 19 μg/m3, respectively.

#### **4. Discussion**

#### *4.1. Indoor Air Concentrations of VOCs and Formaldehyde Are Generally Low and Pose No Health Risks*

A large variety of volatile organic compounds originating from both indoor and outdoor sources are present in indoor environments. According to our ten-year VOC data collected in 2010–2019 in Finland, the most frequent VOCs in offices and similar non-industrial indoor environments are benzene, xylenes (p, m), toluene, α-pinene, 1-butanol, 2-ethyl-1-hexanol, 1,2-propanediol, benzaldehyde, hexanal, nonanal, and decamethylcyclopentasiloxane. These compounds were present in concentrations exceeding the respective LOQs (LOQ values varied 0.3–1 μg/m3) in more than half of the 9789 samples collected in 2010–2019; i.e., these twelve compounds had a frequency over 50% in the total ten-year VOC data. Typical sources of these common VOCs are motor-vehicle emissions from outdoors (aromatic compounds), wood-based and other interior and construction materials (α-pinene, alcohols, aldehydes), water-dilutable glues and PVC materials (alcohols), detergents, fragrances, cosmetics (aldehydes, siloxanes), and food supplies (aldehydes).

Despite the VOC's multiple sources, the TVOC levels and the concentrations of single VOCs were generally low in comparison to previously published data on similar indoor environments [2,7,8,17]. However, the maximum values were generally bigger in our dataset than in other datasets, which is probably due to the notably bigger sample size of our study compared to the samples of others. The median, P90, P95, and P99 TVOC values in our ten-year dataset were 30, 90, 137, and 290 μg/m3, respectively. This means that in over 99% of cases, TVOC concentration fell well below 400 μg/m3, which is the TVOC reference value set by the Finnish Decree on Housing Health 545/2015. These concentration levels also fall into the stage one (out of five) hygienic guide value of ≤0.3 mg/m<sup>3</sup> set by the UBA for indoor air TVOC [13,14]. The UBA considers this first stage hygienically safe. Furthermore, the TVOC concentration levels displayed a clearly decreasing trend during the ten-year study period. In the last study year, 2019, the median, P90, P95, and P99 TVOC values were 20, 70, 100, and 215 μg/m3, respectively.

Although the majority of VOC sampling was performed in buildings with suspected indoor air problems of some kind (not necessarily VOCs), we believe that the VOC levels measured in our study rather accurately represent those of a normal population of offices and similar non-industrial indoor environments in Finland. Considering the very low levels of VOCs measured in our large dataset, it appears obvious that VOCs are not currently a major concern in Finnish public and office buildings.

Unlike VOC measurements, the formaldehyde measurements (belonging to VVOCs) were targeted to buildings where elevated formaldehyde levels were expected. Nevertheless, the measured formaldehyde levels in our ten-year dataset were low in comparison to previously published data on similar indoor environments [2,7,8,17]. It is also noteworthy that the concentration level of formaldehyde displayed a clear decreasing trend during the ten-year period.

None of the RCRs calculated for the 99th percentile of the VOC and formaldehyde measurements exceeded 1 in comparison to the RW I values; the vast majority were far below this. Therefore, according to the current knowledge on their health hazards, the measured concentrations of these individual substances do not pose health risks. Several of the measured maximum values exceeded their respective RW I value, and a few also exceeded their RW II value. Such measurements originated from individual sampling locations, which should be further investigated.

Many VOCs can be smelled at low concentrations; their odours can be perceived as unpleasant and thus cause nuisance. However, they do not necessarily indicate a health risk, as odour thresholds can be considerably lower than the concentration levels that induce adverse health effects [18].

It should be noted that the occurrence of VOC mixtures was not studied in the dataset, and thus, the health risk assessment does not cover possible mixture effects. These could be relevant particularly for compounds that induce equivalent health hazards via the same action mechanism and would be an interesting topic for further research. However, considering the overall low levels of VOCs in the dataset in comparison to their respective health-based reference values, no immediate concern related to mixture effects is evident from these data.

#### *4.2. Differences between VOC Patterns in Different Types of Indoor Environments Have No Practical Relevance*

We inspected differences between offices, schools, kindergartens, and healthcare offices and found that these environments differed in terms of frequencies of several VOCs, whereas the differences between the P95 values were typically small. The office environment had a higher frequency of alkanes, aromatic hydrocarbons, and acids than the other environments. The kindergarten environment had a higher frequency of aldehydes, certain terpenes (α-pinene, 3-carene) and certain esters (TXIB, n-butyl acetate). The differences between the measured VOC patterns in these environments probably reflect the differences between their typical distances from busy roads, their typical activities, and the use of construction and interior materials, cleaning products, and utility articles.

Unlike VOCs, formaldehyde showed similar frequency, over 90%, in all the studied environments, whereas the concentration of formaldehyde ranged in a relatively large scale. The lowest median, P90, and P95 concentration values of formaldehyde were observed in the schools and the highest in the offices. As formaldehyde has a multitude of possible emission sources, e.g., traffic, construction materials, furniture, textiles, cosmetics, wood burning, cooking and oxidative reactions, it is difficult to try to explain the observed differences.

Despite the differences between VOC and formaldehyde patterns in these environments, all the measured concentrations were low in comparison with their respective health-based reference values, and therefore, we do not expect these differences to have any practical relevance for health outcomes.

#### *4.3. Interpretation of VOC Trends Involves Uncertainties*

The total VOC concentration, TVOC, calculated from each VOC sample in our dataset, showed a clearly decreasing trend during the study period, 2010–2019. Both increasing and decreasing trends were observed in individual compounds. Most notably, aromatic compounds and alkanes displayed systematic decreasing trends, whereas several aldehydes and acids showed increasing trends.

We analysed both frequency and concentration trends. The frequency trend reveals whether the frequency value, i.e., the percentage of results above LOQ, is rising or falling. However, the frequency trend does not prove that the concentration level trend is parallel. We found several examples of divergent frequency and concentration trends in our data. Benzaldehyde and decamethylcyclopentasiloxane showed completely opposite trends of increasing frequency and decreasing concentration. In other words, these two compounds were measured more frequently over the years but at the same time the measured concentrations became generally smaller. Formaldehyde showed no frequency trend but a decreasing concentration trend, whereas some VOCs, such as α-pinene and TXIB, showed slightly increasing frequency trend but no concentration trend. In most cases of increasing frequency trend, also the concentration trend was positive, although less prominent (smaller Z). This was true, for example, for several aldehydes, alcohols, and acids. Interestingly, P95 values tended to decrease or remain stable even in situations in which the concentration trend was increasing. In other words, several compounds were detected and quantified more frequently, but a larger share of the measurements were only slightly above the LOQ. One explanation for this phenomenon might be the increased detection and quantification of small peaks in VOC chromatograms due to decreased total VOC concentrations. The ISO 16000-6 standard requires at least two-thirds of the total TVOC area to be identified. This may lead, at low TVOC level, to quantification of very small concentrations that would be neglected with larger TVOC concentrations. Therefore, it is reasonable to view all increasing trends, especially increasing frequency trends, with caution. At the same time, it is important to keep in mind that increased use of textile carpets and epoxy and acrylic resins as well as novel products used in cleaning and personal hygiene might also partially explain some of the increasing trends observed during the study period.

Compared to increasing trends, decreasing trends have less uncertainties due to the increased detection of small peaks discussed above. Aromatic hydrocarbons showed equally strong decreasing frequency and concentration trends during the ten-year study period, 2010–2019. These compounds are mainly linked to fossil fuel combustion [19]. Therefore, we believe that the decreasing trends of aromatic hydrocarbons are mainly connected to better outdoor air quality due to, for example, an ongoing shift towards low-emission vehicles.

The decreasing trends of alkanes and formaldehyde are most likely connected to the increased use of low-emission building and interior materials. This positive development is probably due to the development and application of emission classification systems, the aim of which are to enhance the development and use of low-emission building materials and furniture. In Finland, M1-classification, developed and published for the first time in 1996, sets limit values for the emission of VOCs, formaldehyde, and ammonia as well as criteria for the acceptability of odour [20]. M1-classification is a voluntary labelling system for manufacturers, importers and exporters of building products. However, public property

developers often demand the use of M1-certified products in their projects in Finland, thereby enhancing the development and usage of low-emission materials in buildings.

The decreasing frequency and concentration of 2-ethyl-1-hexanol is probably due to the development of non-phthalate plasticizers, which are increasingly being used in flooring materials.

#### **5. Conclusions**

This study presents a valuable dataset on VOCs and formaldehyde from an exceptionally large number of samples collected from offices and similar indoor environments over a decade in Finland. The data showed that indoor air concentrations of VOCs and formaldehyde in these environments are generally low and pose no health risks. The differences between VOC patterns in office, school, kindergarten, and healthcare office environments have no practical relevance to health outcomes. TVOC concentration showed a clear decreasing trend over the ten-year study period.

**Author Contributions:** Conceptualization, K.W., H.H., J.R., S.M. and T.L.; data curation, H.H. and J.R.; statistical analyses, J.R.; writing—original draft preparation, K.W.; writing—review and editing, K.W., H.H., J.R., S.M. and T.L.; project administration, T.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by the Finnish Ministry of Social Affairs and Health as part of the Finnish Indoor Air and Health Programme 2018–2028.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The raw data of this study are not publicly available because they are owned by a multitude of laboratory customers.

**Acknowledgments:** We thank Sirpa Rautiala, Pirjo Korenius, and Katri Leino for their expertise and fruitful discussions during the project. Tiina Santonen and Piia Taxell are acknowledged for their valuable comments, and we also thank the personnel of FIOH's analytical laboratory for analysing the VOC and formaldehyde samples. We are grateful to the numerous customers of FIOH's analytical laboratory for the collection and delivery of the VOC and formaldehyde samples of this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Analysis of the Airflow Generated by Human Activity Using a Mobile Slipstream Measuring Device**

**Minkyeong Kim 1, Yongil Lee <sup>2</sup> and Duckshin Park 3,\***


**Abstract:** Human activities, including walking, generate an airflow, commonly known as the slipstream, which can disperse contaminants indoors and transmit infection to other individuals. It is important to understand the characteristics of airflow to prevent the dissemination of contaminants such as viruses. A cylinder of diameter 500 mm, which is the average shoulder width of an adult male, was installed in a motorcar and moved at a velocity of 1.2 m/s, which is the walking speed of an adult male. The velocity profile of the slipstream generated during this movement was measured by locating the sensor support at 0.15–2.0 m behind the cylinder. The wind velocity was set to 1.2 m/s to conduct the numerical analysis. The measurement data revealed the velocity profile of the space behind the cylinder, and a comparison of the numerical analysis and the measurement results indicate very similar *u* (measured velocity)/*U* (moving velocity) results, with a maximum difference of 0.066, confirming that the measured values were correctly estimated from the results of the numerical analysis.

**Keywords:** infectious viruses; indoor contaminants; mobile slipstream measuring instrument; cylinder; airflow

#### **1. Introduction**

The severe acute respiratory syndrome epidemic that began in Hong Kong in 2003 reportedly spread widely through the air [1]. Middle East respiratory syndrome (MERS) was first reported in Saudi Arabia in 2012 and then spread worldwide, including Korea. Severe acute respiratory syndrome coronavirus 2 (COVID-19) was first detected in December 2019, in Wuhan City, Hubei Province, China, and then very quickly spread throughout other countries including Korea. Although the transmission pathway of MERS and other pathogens has not been identified with 100% certainty, it is presumed that it is transmitted via close contact among human beings [2]. In this case, the term "close contact" is defined as being within 2 m of another person, in a room with others, or in the care area of an infected person; it also includes direct contact with infectious secretions while not wearing appropriate personal protective equipment [2]. Airflow generated by human movements can accelerate the spread of various airborne materials, such as COVID-19 virus particles during the current pandemic [3–7].

Numerical analyses are used by many researchers to understand aerodynamic properties because the procedures have no spatial constraints. Regarding the dissemination of substances, airflow by heating, ventilation, and air conditioning systems have been reviewed [3–7]. The main factors in the diffusion of pollutants are the airflow generated by human breathing and body heat [8,9]. Liu et al. [8] numerically analyzed natural convection, and Martinho et al. [9] modeled an actual mannequin using 3D scans to analyze the airflow generated by body heat when the mannequin was placed in a sitting position

**Citation:** Kim, M.; Lee, Y.; Park, D. Analysis of the Airflow Generated by Human Activity Using a Mobile Slipstream Measuring Device. *Environments* **2021**, *8*, 97. https:// doi.org/10.3390/environments8100097

Academic Editor: Ki-Hyun Kim

Received: 25 July 2021 Accepted: 21 September 2021 Published: 23 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

indoors. In both studies, the results were compared with actual measurements. One study conducted a numerical analysis of airflow around the human body, taking into account the layer of air created by clothes [10]. The trajectory of particles exhaled by breathing was numerically analyzed according to the wind speed of the surrounding airflow [11]. Pollutant movement according to the airflow generated while passing through a door between a room and hallway was analyzed [12–15]. Simulations and numerical analyses have been performed using the speed of a moving person and the opening and closing of a door [16].

To evaluate the diffusion of pollutants, a study using a chamber was conducted. Zhang et al. [17] built a cabin with dimensions of 4.9 × 4.23 × 2.1 m in a chamber to evaluate the dissemination of contaminants in aircraft cabins. Experimental measurements and numerical simulations of airflow and contaminant transport were conducted in a "half occupied, twin-aisle cabin mockup". Poussou and Mazumdar [18] simulated an aircraft cabin using a small underwater tank and measured the variables using particle image velocimetry and planar laser-induced fluorescence. Contaminant propagation according to human movement in the aircraft cabin was analyzed [19]. Han et al. [20] built a test bed that could house actual-size mannequins and installed a wind velocity sensor to measure the velocity profile under different conditions. The airflow was measured using a warm mannequin and a mannequin with mobile parts moving at a velocity of 0.5–1.5 m/s. Milanowicz and Kedzior [21] modeled a human body falling from a height. Liu et al. [22] examined the airflow around a human body model in an enclosed space, examined changes in temperature and velocity, and simulated the transmission of infectious respiratory diseases.

Studies using mannequins have compared numerical analysis and measurement results, and confirmed particle diffusion [23,24]. In this study, a mobile slipstream measuring device was constructed to measure the velocity profile behind an object. The velocity profile behind a moving cylinder was measured directly, and a numerical analysis was performed to elucidate the airflow characteristics of the slipstream. In previous studies, no attempts were made to change the speed of the moving cylinder, or to detect changes in airflow by installing a flow rate sensor on the side of the cylinder.

Airflow is generated by human movements, which can affect the acceleration of the spread of various airborne materials, such as the COVID-19 virus, which is currently an important issue [3–7]. In this study, a moving cylinder was used to represent an adult male, and a constant walking pace was maintained. A regularly shaped cylinder of diameter 500 mm (average shoulder width of an adult male) was mounted on a motorcar and set to move at a velocity of 1.2 m/s (average walking velocity of humans). This cylinder approximates the movement of the human body. A flow rate sensor was installed on the back of the moving cylinder at different heights and distances to measure the change in airflow. In previous studies, no attempt was made to change the speed of a moving cylinder or to detect the changes in airflow by installing a flow rate sensor on the side of a cylinder. This would enable the changes in air flow around the cylinder to be determined during movement, and height and distance could be considered as variables. Studies of slipstreams have been conducted mainly in tunnels and laboratories. By performing a numerical analysis of the results of this study, the effects of the diffusion of air currents carrying air pollutants and viruses can be determined.

#### **2. Materials and Methods**

#### *2.1. Experimental Equipment*

A device measuring 3 m (W) × 35 m (L) × 2.5 m (H) was constructed to investigate the impact of the movement of an object on the slipstream in a large chamber. A rail was installed in the center of an indoor space to minimize the impact of the airflow. The mobile slipstream measuring device consisted of a rail, moving part, and sensor support (Patent-10-1517092). The rail consisted of a cylindrical aluminum pipe that helped the

moving part, which was a specially constructed motorcar, to move seamlessly at adjustable velocities (Figure 1).

**Figure 1.** Testbed and sensor support.

A total of 20 wind velocity sensors (Model 0962-00, Kanomax, Andover, NJ, USA) were installed on the sensor support, with horizontal and vertical distances between sensors of 200 mm, as shown in Figure 2. The measuring range of the wind velocity sensors was 0.1–50 m/s, and the measurement error was ±0.1 m/s over the range 0–4.99 m/s. The relative wind speed was measured. The measured speed was subtracted from the human's moving speed to calculate the actual wind speed at each point.

$$\mathcal{U}\_r = \mathcal{U}\_h - \mathcal{U}\_m$$

Data were recorded every 0.1 s using a data logger (Model 1560, Kanomax, Suita, Japan). Table 1 shows the positioning of the sensors installed on the sensor support.


**Table 1.** Locations of wind velocity measurement sensors.

**Figure 2.** Schematic of the mobile wind velocity measurement system.

#### *2.2. Experimental Methods*

A normally shaped cylinder of diameter 500 mm, which is the average shoulder width of an adult male, was mounted on a motorcar and set to move at a velocity of 1.2 m/s. A running test was conducted five times, with a few pieces of corrugated cardboard placed in the empty spaces between the object and sensor support to flatten the bottom. Measurements were conducted at the rear of the cylinder after varying the sensor support position to 0.15, 0.25, 0.5, 0.75, 1, or 2 m. A numerical analysis was performed to evaluate the airflow of the cylinder slipstream. Fluent 17.7 software (Ansys Co., Canonsburg, PA, USA) was used to conduct the numerical analysis, and the test bed was modeled. The test bed dimensions were 3 m (W) × 2.5 m (H) × 10 m (L), and a cylinder of diameter 500 mm (2 m high) was placed at the center. Tetrahedral grids were placed in a tight formation around the cylinder and along the wall. A total of 354,330 nodes were generated in the flow field. The right, left, top, and bottom of the cylinder were specified as walls, and the no-slip condition was set. The front of the cylinder was specified as the inlet, and the wind velocity was set to 1.2 m/s (the walking speed of an adult male). The Reynolds number was 4.0 × <sup>10</sup>4. The rear surface was specified as the outlet and set to atmospheric conditions. A model built with the commercial code K-epsilon was used to solve the numerical analysis. The k-epsilon equation used in this study is as follows.

$$\mu\_{l} = \frac{\mathbb{C}\_{\mu\rho\mathbb{K}^{2}}}{\mathfrak{e}}$$

$$\rho\left(\mathcal{U}\_{l}\frac{\partial k}{\partial \boldsymbol{x}\_{i}}\right) = \frac{\partial}{\partial \boldsymbol{x}\_{i}}\left(\mu + \frac{\mu\_{l}}{\sigma\_{k}}\frac{\partial k}{\partial \boldsymbol{x}\_{i}}\right) + \mu\_{l}\left(\frac{\partial \mathcal{U}\_{i}}{\partial \boldsymbol{x}\_{j}} + \frac{\partial \mathcal{U}\_{l}}{\partial \boldsymbol{x}\_{i}}\right)\frac{\partial \mathcal{U}\_{i}}{\partial \boldsymbol{x}\_{j}} - \rho\epsilon\mathfrak{e}\_{i}$$

$$\rho \left( \mathcal{U}\_{l} \frac{\partial \varepsilon}{\partial \mathbf{x}\_{l}} \right) = \frac{\partial}{\partial \mathbf{x}\_{l}} \left( \mu + \frac{\mu\_{l}}{\sigma\_{\varepsilon}} \frac{\partial \varepsilon}{\partial \mathbf{x}\_{i}} \right) + \mathbb{C}\_{\varepsilon 1} \frac{\varepsilon}{k} \left[ \mu\_{l} \left( \frac{\partial \mathcal{U}\_{l}}{\partial \mathbf{x}\_{j}} + \frac{\partial \mathcal{U}\_{l}}{\partial \mathbf{x}\_{i}} \right) \frac{\partial \mathcal{U}\_{i}}{\partial \mathbf{x}\_{j}} \right] - \mathbb{C}\_{\varepsilon 2 \rho} \frac{\varepsilon^{2}}{k}$$

After a comparison of the results obtained using the cylinder, mannequin measurements were conducted using the same process.

#### **3. Results and Discussion**

#### *3.1. Measurement Results*

Each variable was measured five times using a mobile slipstream measuring device, and the average value was calculated. Figure 3 shows the measurement results at the 4th height position (Y = 1.4 m). In this case, *Z* was the distance between the cylinder and sensor support. The motor reached a speed of 1.2 m/s after 4 s and then continued at this velocity for approximately 8 s. The wind velocity approached the moving velocity of 1.2 m/s as *Z* increased. Although the change in wind velocity for rows B and F was not significant, the wind velocity near the sensor support was approximately 1.5 m/s, which was higher than the moving velocity of 1.2 m/s. The wind speed difference among rows B and F was not statistically significant according to non-parametric test results (*p* > 0.05). On the other hand, although the wind velocity varied greatly for rows C, D, and E according to the distance, the wind velocity near the sensor support was approximately 0.4 m/s, which was lower than the moving velocity of 1.2 m/s (Table 2).

**Figure 3.** Velocity measurements at different locations (average of five measurements taken at the 4th height position).


**Table 2.** Wind velocity measurement results (average of five measurements taken at the 4th height position).

Figure 4 shows the measurement results at various sensor heights taken at a distance of 0.5 m between the cylinder and sensor support. Similar patterns were observed to the right and left of row D. Wind velocity decreased as the sensor height decreased, which appeared to be because there was no significant difference in distance between the second height position and the top of the cylinder, although the sensor at the second height position was positioned behind the cylinder. A similar trend was observed in previous studies as the wind speed varied according to the measurement height [23,24].

**Figure 4.** Velocity measurements at different locations (average of five measurements, Z = 0.5 m).

#### *3.2. Results of Cylinder Measurements and the Numerical Analysis*

The average Reynolds number in the 5–6 s constant velocity section was calculated using the measurement results presented above. Figure 4 shows the variation in the dimensionless velocity ratio calculated by dividing the measured velocity *u* by the moving velocity *U*. Here, *X*/*d* is a dimensionless value calculated by dividing the width of the cylinder by the diameter, and *Z*/*d* is a dimensionless value calculated by dividing the distance between the cylinder and sensor support by the diameter of the cylinder.

Figure 5 shows that *u*/*U* approached 1 as the airflow increased with the increase in the distance from the rear of the cylinder. This was also confirmed by the measurements made in this study. The value of *u*/*U* at *X*/*d* = 0 increased toward 1 as *Z*/*d* increased, and the right and left sections of the graph at *X*/*d* = 0 were symmetrical at all heights. The values of *u*/*U* at *X*/*d* = −0.4, 0, or 0.4 were directly affected by the cylinder because sensor supports were positioned immediately behind the cylinder, and *u*/*U* increased toward 1 as *Z*/*d* increased.

**Figure 5.** Profiles of normalized velocities at different *Z/d* locations.

The results of the numerical analysis and the *u*/*U* measurements are similar, with an approximate difference of only 0.023. This difference increased with *Z*/*d*, with the maximum difference (0.442) observed at the sixth and eighth height positions when *Z*/*d* = 4. However, the maximum difference at the second and fourth height positions was 0.089, which was not greater than that observed at the sixth and eighth height positions when *Z*/*d* = 4. This appeared to have been affected by the trailing vortex generated at the edge of the top of the cylinder [25].

Figure 6 shows the velocity profile according to *Z*/*d* when *X*/*d* = 0. The numerical analysis of *U*<sup>x</sup> showed that the wind direction changed. The value of *u*x/*U*<sup>x</sup> was –0.130 in the opposite direction of the wind at all heights when *Z* = 0.15 m, whereas the value of *u*x/*U*<sup>x</sup> was –0.009 at the sixth and eighth height positions in the opposite direction of the wind for heights up to *Z* = 0.75 m. As shown in Figure 7, the value of *u*x/*U*x was negative because a vortex was generated when the cylinder moved, and the vortex increased as the height decreased. It was presumed that the point where the vortex was generated was the point where the direction of the measured value changed.

**Figure 6.** The normalized velocity profiles obtained by measurements and numerical analysis results.

**Figure 7.** Velocity vector profile of the cylinder in the y–z plane (x = 0 m).

Figure 8 shows the fluid flow at four heights. It also shows that the velocity increased rapidly to the left and right of the cylinder, and that a vortex was created behind it. The strength of the vortex increased toward the bottom.

**Figure 8.** Time-averaged streamlines in different x–z planes.

Figure 9A shows the measurement results for the mannequin at the fourth height position (Y = 1.4 m). Figure 9B shows the measurement results according to various sensor heights taken at a distance of 0.5 m between the mannequin and sensor support. Patterns similar to those for Figure 9A,B of row D were observed. The average Reynolds number in the 5.6–8.0 s constant velocity section was calculated using the measurement results described above. Figure 10 shows the dimensionless velocity ratio calculated by dividing the measured velocity *U* by the moving velocity.

**Figure 9.** Results obtained at the 4th height position; (**A**) results for the mannequin at the fourth height position; (**B**) results according to various sensor heights taken at a distance of 0.5 m.

Figure 10 shows that as *u/U* approached 1, the air flow was more developed as the distance from the rear of the mannequin increased. This was also confirmed by the measurements in the study. The *u/U* at X (m) = 0 increased toward 1 as Z increased, and the right and left values at X (m) = 0 were symmetrical at all heights. The positions X (m) = −0.4, 0, or 0.4 were directly affected by the cylinder because they were positioned immediately behind the cylinder, and *u/U* increased toward 1 as *Z/d* increased.

**Figure 10.** Results obtained for the cylinder and mannequin at the shoulder position (500 mm).

#### **4. Conclusions and Discussion**

In this study, a mobile measuring device was used to directly measure the slipstream generated when a cylinder of diameter 500 mm (the average shoulder width of an adult male) was moved at a velocity of 1.2 m/s (the average walking velocity of humans) to obtain the velocity profile of the rear side of a cylinder. The cylinder was then fixed, and a wind velocity of 1.2 m/s was created at the inlet to provide data for the numerical analysis; the results were compared with the measurements. The results of the measurements and numerical analysis of the area behind the cylinder are very similar, with a maximum *u*/*U* difference of 0.066 at *Z*/*d* = 0.3. The difference was 0.042 and was larger at the 2nd height position when *X*/*d* = 0.0 and *Z*/*d* = 4. The increase in the difference with increasing *Z*/*d* was attributed to the impact of the trailing vortex, and to the fact that the numerical analysis was performed with the cylinder in a fixed position. The direction of the measurement was estimated as per the position of the vortex obtained from the numerical analysis. The wake of the mannequin was measured in the same way as that of the cylinder. Similar measurement results were obtained at the shoulder height, which was the same diameter as the cylinder, and a velocity profile for the mannequin wake was obtained.

Previous experimental studies [20] on wakes were conducted in wind tunnels or as lab-scale experiments, whereas the present study used a mobile cylinder that can reveal the actual characteristic of wakes. In this manner, air flow characteristics were identified. This study presented a method based on a mobile measuring device to obtain the velocity profile of the slipstream of a moving object. If the moving velocity of the sensor support is accounted for, the mobile measuring device could be used to verify the results of simulations of the dispersion of airborne contaminants derived from moving objects.

Recently, the social and economic damage caused by viruses such as COVID-19 has increased. To prevent the spread of such airborne substances, a ventilation system can be used, along with restrictions on movement [26,27]. Droplet dispersion during coughing by people who are walking plays an important role in the transmission of COVID-19. A

simulation showed that droplets in the air in a narrow space, such as a hallway, were transmitted below waist height (of the emitter) [28]. Wake measurements confirmed that the speed and distribution stabilized as the distance from the mobile slipstream at chest and waist height increased. The speed rapidly increased to the left and right of the mobile slipstream, and a vortex was created on the back side. The vortex formed as it moved to the bottom surface. Based on the results of this study, movement restrictions appear necessary to slow the spread of the virus, in accordance with the correlation between the wake velocity distribution and spread of the virus.

Numerical analysis results for a system comprising a moving object could not be compared with the measurement results. Therefore, in future studies, methods need to be devised that consider moving velocity when measuring wind velocity, and numerical analysis of a system including a moving object could be performed for comparison.

**Author Contributions:** D.P. planned the study and contributed the main ideas; M.K. was principally responsible for the writing of the manuscript; Y.L. revised the manuscript and methodology analysis. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant from the Subway Fine Dust Reduction Technology Development Project of the Ministry of Land Infrastructure and Transport (21QPPW-B152306-03).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Window-Windcatcher for Enhanced Thermal Comfort, Natural Ventilation and Reduced COVID-19 Transmission**

**Odi Fawwaz Alrebei 1,\*, Laith M. Obeidat 2, Shouib Nouh Ma'bdeh 2, Katerina Kaouri 3, Tamer Al-Radaideh <sup>2</sup> and Abdulkarem I. Amhamed 1,\***


**Abstract:** We investigate and test the effectiveness of a novel window windcatcher device (WWC), as a means of improving natural ventilation in buildings. Using ANSYS CFX, the performance of the window-windcatcher is compared to a control case (no window-windcatcher), in three different geographic locations (Cardiff, Doha and Amman) which are representative of three different types of atmospheric conditions. The proposed window-windcatcher has been shown to improve both thermal comfort and indoor air quality by increasing the actual-to-required ventilation ratio by up to 9% compared to the control case as per the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) standards. In addition, the locations with minimum velocities have been identified. Those locations correspond to the regions with a lower infection risk of spreading airborne viruses such as SARS-CoV-2, which is responsible for the COVID-19 pandemic.

**Keywords:** natural ventilation; thermal comfort; infection prevention; building cfd analysis; windcatcher

## **1. Introduction**

As one of the most significant threats to our lives, global warming needs to be tackled immediately and effectively. Global warming has been destroying the environment, thus affecting almost all aspects of our lives [1]. The construction industry is one of the main sectors contributing to global warming, through the emission of vast greenhouse gas (GHG), mainly CO2 [2]. In addition, 40% of the CO2 emissions is generated by buildings and around 40% of the global energy generated is consumed by buildings [3]. Energy consumption by buildings is expected to reach 64% of the total energy consumption by 2100, if no action is taken [4].

More than 60% of the energy consumption in buildings is used for heating, cooling, and ventilation [5]. This energy is obtained mainly from fossil resources [5]. In addition to the challenges of high energy consumption and CO2 emissions, maintaining acceptable indoor air quality (IAQ) using mechanical HVAC systems can be challenging. On average, people spend up to 90% of their time working and living indoors. The risk of sick building syndrome (SBS), metabolic diseases and transmission of COVID-19 and other airborne viral diseases are increased in air-conditioned buildings compared to naturally ventilated buildings [6]. It is, thus, critical to ensure good IAQ to maintain health and productivity [5]. In this context, passive strategies implemented in the building's architecture, such as daylighting, natural ventilation, passive cooling and passive heating may provide significant benefits to users and to the environment as they lead to reduced energy consumption and CO2 emissions, mitigating negative impacts on the environment and health. Implementing such passive strategies could also lead to significant cost savings [7].

**Citation:** Fawwaz Alrebei, O.; Obeidat, L.M.; Ma'bdeh, S.N.; Kaouri, K.; Al-Radaideh, T.; Amhamed, A.I. Window-Windcatcher for Enhanced Thermal Comfort, Natural Ventilation and Reduced COVID-19 Transmission. *Buildings* **2022**, *12*, 791. https://doi.org/10.3390/ buildings12060791

Academic Editors: Ashok Kumar, M Amirul I Khan, Alejandro Moreno Rangel and Michał Piasecki

Received: 11 April 2022 Accepted: 3 June 2022 Published: 9 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Passive cooling techniques have been studied extensively [7]. Acceptable thermal comfort and IAQ using passive cooling can be achieved with only a small fraction of the energy consumed by mechanical ventilation systems [6]. Natural ventilation is one of the leading passive cooling strategies that effectively improve the indoor atmosphere by: (1) providing good IAQ and (2) improving thermal comfort, through affecting ventilation rates, air velocity, temperature and humidity [7]. Natural ventilation is air exchange between outdoor (fresh) air and indoor (used) air [1]. It occurs naturally, without mechanical assistance, by establishing a pressure difference or a temperature difference.

Indoor thermal comfort could be directly improved by increasing the cooling sensation through increased airflow or indirectly by night ventilation (night flushing) [7,8]. Night ventilation can be achieved through natural ventilation, especially in the context of adequate fluctuation in air temperature over the day and the night. For example, the temperature during the hot, sunny summer months in Amman, Jordan usually reaches 40 ◦C in the day and drops significantly, to around 20 ◦C in the night. (Average temperatures during the day and night are 36 ◦C and 22 ◦C, respectively [9].) Such a significant fluctuation in temperature provides great potential for night ventilation and improvement of the indoor thermal comfort [10] by providing a cool breeze during the night, which would flush out hot air and cool off the internal thermal masses to effectively delay the thermal gain during daytime [11].

Implementing effective natural ventilation solutions in multi-family residential buildings is challenging and sometimes not easily applicable. The internal spatial organization of the building [11], and to a great extent, its layout and limited shared exterior walls with the outdoor environment contribute to the difficulty. The limited shared exterior walls, especially in the generic architectural typology of the multi-story residential buildings, lead to little or even no condition of the opposite openings (inlet and outlet) necessary to achieve effective cross-ventilation [12]. In addition, even in the single-family detached house typology, cross-ventilation conditions can be difficult to achieve, especially for large houses where some rooms have only a single exterior wall with all window openings on the same wall. Various techniques and systems have been developed in order to exploit these natural phenomena and conditions to achieve adequate natural ventilation. These systems vary in performance, requirements, and settings: they include Trombe wall, double skin façade, solar chimney, solar walls, atrium, wind tower, windcatcher and fenestration (single-sided ventilation and cross ventilation) [7]. These systems can be intertwined and used for other passive design strategies.

A traditional and common natural ventilation system for buildings is the windcatcher [13] as it can provide good air quality and improve thermal comfort in an environmentally friendly manner, using, mainly, renewable wind energy [14]. Windcatchers have a long history with enhancing indoor environmental comfort in arid and semi-arid regions. They achieve harmony between built environments and the surrounding natural environments. A windcatcher or wind tower is generally defined as a tower-like architectural component designed to be mounted on the building roof "to 'catch' the wind at higher elevations and direct it into the inner environment of a building" [2].

Windcatchers operate mainly with wind-driven ventilation and stack (buoyancy) effects [2]. Moreover, windcatchers are low maintenance since they operate without moving parts [2]. However, windcatchers have limitations. A key limitation is their large size and centrality (i.e., located at the building's center). Since several windcatcher elements are generally required to achieve adequate natural ventilation, especially for large-scale buildings, restrictions are imposed on the building geometry. For example, windcatcher systems may limit future expansion and roof space use [3]. Many researchers have extensively studied their effectiveness and performance using different evaluation methods such as computational methods, experimental methods, analytical and empirical methods, or a combination. Each of these methods has its own advantages and limitations in terms of accuracy, cost, complexity of geometry, detail of the results and time to implement the method [15]. CFD is the most used computational method as it offers high accuracy and

low financial cost (only the cost of the software package), and it is suitable for complex geometries, generating detailed results. Detailed information regarding the advantages of CFD compared to other methods can be found in [15].

According to [16,17], Environmental Controls (ECs) are based on methods for reducing concentrations of an infectious agent in the air and on surfaces in indoor environments. World Health Organization (WHO) guidelines recommend a combination of environmental control mechanisms to avoid the spread of viruses to health care workers (HCWs) and patients in health care settings [17]. The environmental regulations rely on the design of the healthcare environment and include (a) building materials, surfaces, and products used [17]; (b) indoor environmental factors (e.g., temperature, humidity, light, and airflow) [17]; and (c) indoor access to the outside. All of the above may influence the survival of infectious agents in the built environment.

Airflow and ventilation systems play a significant role in the airborne transmission of pathogens; improving ventilation decreases the risk of transmission. Identifying how the novel window-windcatcher improves ventilation is the focus of our work here [6,7]. The design of a ventilation system depends on the ability to contain, mitigate, and remove airborne pollutants through air change and inward indoor airflow [17].

Reducing the effect of overheating, and thus, improving thermal comfort in a building, is achieved mainly by two methods: building elements and ventilation [18]. The heat conductivity (k-value) of the building elements (such as walls, windows, etc.) can be reduced through selecting appropriate building elements. Ventilation is enhanced through increased air circulation, using, for example, window-windcatchers. The overheating is given by *Hloss* = *cp* . *<sup>m</sup>*Δ*T*, where *cp* is the air specific capacity, . *m* is the air mass flow rate and Δ*T* is the temperature difference between the internal and external domains. The window-windcatcher can increase *Hloss* by increasing . *m*.

Our work aims to investigate and test, using CFD, the design of a novel window- windcatcher device that can be mounted on exterior walls to capture the prevailing wind and redirect it into indoor spaces. We demonstrate that the ventilation rate in an indoor space increases when we use the novel window-windcatcher. According to the World Health Organization (WHO), viral diseases, such as COVID-19, can be more easily transmitted in poorly ventilated enclosed spaces which have low ventilation rates [19]. The proposed design could replace or complement traditional large-scale windcatchers by small-scale decentralized windcatchers which can be mounted on exterior walls as a window component. The design of the device is suitable for retrofitting existing buildings or for new buildings. We find that the device could enhance the effectiveness of natural ventilation (passive cooling) in buildings, significantly improving IAQ and thermal comfort by increasing the actual-to-required ventilation ratio, as per the ASHRAE standards, by up to 9% compared to the control case without a window-windcatcher.

This can be achieved by increasing the ventilation rate while ensuring minimal turbulence. Increasing the ventilation rate is primarily achieved by increasing air velocity; however, as the turbulence kinetic energy is dependent on the air velocity, improving the ventilation rate could potentially increase the turbulence kinetic energy. Therefore, it is crucial to ensure that the increase in the turbulence kinetic energy is kept sufficiently low as the velocity increases.

Portable air cleaners, also known as air purifiers or air sanitizers, are designed to filter the air in a single room or area [20]. Central furnace or HVAC filters are designed to filter air throughout a home. Portable air cleaners and HVAC filters can reduce indoor air pollutants, including viruses, that are airborne [20]. By themselves, portable air cleaners and HVAC filters are not enough to protect people from the virus that causes COVID-19 [20]. Air purifiers could be placed in regions of high turbulence kinetic energy. Hence, such regions shall be carefully identified in order to formulate recommendations on the use of air purifiers. Placing air purifiers in high turbulence kinetic energy regions will: (1) ensure that the highest air flow rate is entering the air purifier; and (2) prevent the building's residents to occupy those locations.

We outline the structure of the paper. The Introduction provides the rationale of the research, the benefits of implementing passive design techniques such as natural ventilation, the context of using the windcatcher natural ventilation system, our goals and objectives, and an overview of the used methodology. The Material and Methods section provides more detailed information about the proposed window windcatcher, building geometry, and the geographical locations of the three cities we study (Amman, Doha and Cardiff). In addition, this section provides information about the CFD method, software, sensitivity analysis, and inlet and outlet boundary conditions. The Results section consists of a qualitative and a quantitative part. The qualitative part presents visual comparisons of a case without or with a window windcatcher device, respectively. The comparison includes the turbulence kinetic energy, velocity, and velocity streamlines for the three selected cities, on selected planes of interest. The quantitative part shows the simulations with and without a window-windcatcher and the device's effect on the actual-to-required ventilation ratio. In the Discussion section, we present the effect of the windcatcher on the IAQ. In addition, potential locations for deploying air purifiers in indoor spaces are discussed. Lastly, the Conclusions section presents the key results; the window-windcatcher increases the actual-to-required ventilation ratio as per the ASHRAE standards by up to 9%.

#### **2. Materials and Methods**

#### *2.1. Description of the Window-Windcatcher and of the Building*

To examine the efficiency of the proposed window-windcatcher device, a simplified building model representing a typical single-family house is used—see Figure 1. This building model was used in several studies to investigate natural ventilation and passive cooling in Jordan [21]. For example, it was used to examine the efficiency of the Solar-Wall system (combination of Trombe wall and solar chimney) on enhancing indoor thermal comfort [21]. The selected room dimensions are 4 m × 3 m × 2.7 m (length × width × height), so the floor area is 12 m2, and the volume is 32.4 m3. The external window of the room is 2 m in width, 1 m in length, and 1 m in sill height, with a total window-to-wall ratio of 18%. The window has an operable sliding glazing mechanism that allows up to 50% of its total area to be opened.

The proposed window-windcatcher aims to enhance natural ventilation in a particular room and in the whole residential unit—see Figure 1. The device is to be mounted on the envelope of existing or new buildings with minimal changes on the building shape or structure. Hence, the device could be used as a passive retrofitting technique. The design could be further developed to take into account requirements about privacy, daylighting, and aesthetic issues. Generally, as illustrated in Figure 2, the device consists of four vertical supporting elements (joists) that connect two horizontal planes, an upper and a lower plane. The distance between the upper and the lower plane is equal to the height of the window opening, in this case 1 m. The device consists of four fins (1 m height × 0.15 m width × 0.02 m thickness) tilted by 45 degrees located at the outside edge of the device and one curved element that spans from a point near the last fin in the group into the right internal corner of the two planes. The fins and the curved element are mounted between the two planes supported by the four supporting elements at the four corners of the horizontal planes. The aim is that the prevailing wind will be captured and pass through the device opening on the shorter side opposite the curved element and the areas between the tilted fins.

The vertical curved plane aims at enhancing the captured airflow toward the interior spaces. The tilting angle could be optimized by running the simulation for different angles; this is outside our research scope. Moreover, other issues could be investigated in future work, such as daylighting performance, privacy, and the finishing materials used for the various components of the proposed device.

**Figure 2.** The design and dimensions of the proposed window-windcatcher. (**A**) 2D view (**B**) 3D View.

#### *2.2. Case Studies*

The proposed window-windcatcher's performance has been evaluated against a control case without a window-windcatcher, in three geographical locations: Cardiff, Doha and Amman—see Figure 3. The three cities are located within three different climatic zones: Amman in a hot-summer Mediterranean climate, Doha in a hot desert climate and Cardiff is in an Oceanic climate. The Total Wind Velocity (TWV) and temperature in each city have been obtained for 2021 [22]. The average temperature and TWV have been evaluated and plotted in Figure 4. We find, respectively, for Cardiff, Doha and Amman, these values to be [10.7 ◦C, 5.45 m/s], [27.8 ◦C, 4.2 m/s], and [22.3 ◦C, 3.6 m/s].

**Figure 3.** Geographical locations of the three cities we consider (Cardiff, Doha and Amman).

**Figure 4.** (**A**) Temperature and (**B**) Total Wind Velocity (TWV) in 2021 in Amman, Cardiff, and Doha.

As shown in Figure 5, the orientation of the building with respect to the direction of the average TWV is simulated for equal shear (SWV) and normal (NWV) wind velocity components (i.e., 3.85 m/s, 2.97 m/s and 2.55 m/s for Cardiff, Doha and Amman, respectively). The velocity components have been estimated, according to [23], as follows:

#### *2.3. CFD Simulations*

CFD simulations of the airflow passing through the proposed window-windcatcher were performed with the ANSYS 2020 CFX package [24]. We used a virtual machine with computational capabilities of Processor Intel® Core™i7-7700 CPU @ 3.6 Hz, installed memory 32 GB and a 64-bit operating system. The computational time was approximately 12 h for each case study. The geometry of the building was modelled and loaded into ANSYS. The building model was created using the Boolean algorithm (i.e., the solid domain is subtracted from the internal and external fluid domain [25]—see Figure 5. Figure 6 shows the computational mesh for these domains, created with approximately 2.19 × 106 elements.

Performing experiments with the novel windcatcher is not currently possible as the device has not been manufactured. In fact, the focus of the work is generating CFD simulations that are as accurate as possible in order to assess this device before proceeding to manufacture it. We establish the accuracy of the simulations through performing meshsensitivity analysis for the building that ensures that the findings are independent of the mesh size. The WWC CAD model is created using AutoCAD. This approach has been adopted in our previous studies [26,27]. Figure 7 shows how the number of elements was determined following mesh sensitivity analysis. The mesh sensitivity study was done for the Amman case study using an average velocity across an interior plane with an offset of 1.7 m from the floor, which is around the average person's height [28]. At around 9.4 × 105 elements, the results do not change with the mesh size. Nevertheless, a much finer mesh with 2.19 × <sup>10</sup><sup>6</sup> elements was chosen for the simulation to give a high level of confidence and accuracy. In addition, the relative error to the selected mesh size has been estimated and shown to decrease when the mesh size decreases—see Figure 7. The velocity values converged with a relative error margin of 1%.

**Figure 6.** CFX mesh of the CAD model presented in Figures 2 and 3 (building and window-windcatcher).

**Figure 7.** Mesh analysis—average velocity at 1.7 m offset (internal plane).

Furthermore, the mesh quality (Table 1) is consistent with previous studies [27–29]. The mesh is exported to CFX-PRE.

Using the k-ε turbulent flow model, we proceed with a steady-state analysis to determine the average velocities [27–29]. The fluid domain was chosen from the ANSYS library as air at 10.7 ◦C, 27.8 ◦C and 22.3 ◦C corresponding to the Cardiff, Doha and Amman case studies, respectively (as discussed in Section 2.2). The no-slip boundary conditions were implemented in ANSYS as 'No Slip Walls' [27–29].

The goal of this study is to demonstrate that the ventilation rate in an indoor space increases when we use the novel window-windcatcher. According to the World Health Organization (WHO), viral diseases, such as COVID-19, can be more easily transmitted in poorly ventilated enclosed spaces which have low ventilation rates [19].



Finally, the inlet boundary conditions (as specified in Figure 5) were set to velocity inlets with a turbulence intensity of 5% (default setting in ANSYS).

The Turbulence Kinetic Energy (*k*) is a proxy for the flow mixing level [26,29]. TKE has been chosen as a quantity to study in this work since COVID-19 spread increases with mixing. Equations (2)–(5) [26,29] give TKE:

$$k = \frac{3}{2}(III)^2\tag{2}$$

$$
\varepsilon = \sigma\_{\mu}^{\frac{3}{4}} k^{\frac{3}{2}} l^{-1} \tag{3}
$$

$$I = 0.16Re^{-\frac{1}{\xi}}\tag{4}$$

$$l = 0.07L \tag{5}$$

We plot TKE in ANSYS CFX. In addition, the performance of the window-windcatcher in preventing the spread of COVID-19 by increasing the ventilation rate has been evaluated against the required ventilation rate (*Q*required) as per the ASHRAE standards [30]. The acceptable ventilation rate in residential buildings, as defined by the ASHRAE standards, is given in Equation (6) [30]. The convergence of the results is shown in Figure 8.

**Figure 8.** Convergence of results.

The actual ventilation rate (*Q*actual) has been estimated computationally using AN-SYS, and the ventilation performance has been benchmarked by the actual-to-required ventilation ratio (*nQ*), Equation (7) [30]:

$$Q\_{\text{required}} = 0.15 A\_{flovr} + 3.5(N\_{br} + 1) \tag{6}$$

$$n\_{\mathbb{Q}} = Q\_{\text{actual}} / Q\_{\text{required}} \tag{7}$$

The streamlines and contours of the velocity and TKE for the entire fluid domain are plotted in Figures 9–12. In addition, as shown in Figure 5, data have been displayed at two distinct heights (1.7 m and 1 m) for four planes (two planes for the internal domain and two for the external domain): the first height is 1.7 m above the floor (i.e., the breathing level of an average person [26]). The second is at the height of 1 m above the ground, which is about the same as the height of a sitting person.

**Figure 9.** Turbulence kinetic energy profile at the 1 m interior plane (A and B: High-turbulence region, C: Low-turbulence region).

**Figure 10.** Turbulence kinetic energy profile at the 1.7 m interior plane (D: High-turbulence region, E: Low-turbulence region).

**Figure 11.** Velocity profile at the 1 m interior plane (H, F and G: High-velocity regions).

**Figure 12.** Velocity profile at the 1.7 m interior plane (I and J: High-velocity regions).

#### **3. Results**

#### *3.1. Qualitative Analysis*

As discussed in Section 2, we show how the novel window-windcatcher device we propose could increase the IAQ and hence mitigate the spread of COVID-19 indoors. This is essentially achieved by increasing the ventilation rate while ensuring minimal turbulence. Increasing the ventilation rate is primarily achieved by increasing air velocity; however, as the turbulence kinetic energy increases with the air velocity, increasing the ventilation rate would increase the turbulence kinetic energy.

Therefore, it is crucial to ensure that the increase in the turbulence kinetic energy is maintained at a low value. In addition, regions of high turbulence kinetic energy shall be carefully identified to inform other potential mitigations, for example placing air purifiers in relatively high turbulence regions. Therefore, the turbulence kinetic energy of the interior plane has been plotted for the 1 m (the average height for a seated person) and 1.7 m (the average height for a standing person) planes (Figures 9 and 10, respectively) for the three locations of interest (Cardiff, Doha and Amman).

As shown in Figure 10 for Cardiff, the window-windcatcher has slightly increased the TKE inside the building compared to the case without the window-windcatcher. On the other hand, this increase was less significant in Doha and Amman. This is attributed to the lower levels of wind velocity in Amman and Doha compared to Cardiff. Moreover, as shown in Figure 9, Regions A and B, the high-turbulence regions are located at the building corners while the central region (Region C) remains approximately unaffected.

By comparing the turbulence kinetic energy profiles at the 1 m plane (Figure 9) to those at the 1.7 m plane (Figures 9 and 10), it can be seen that the turbulence energy is higher at the 1.7 m plane (i.e., Region A in Figure 9 to Region D in Figure 10). Hence, Regions D and E could be good locations to place air purifiers.

The velocity profile of the interior plane has been plotted for the 1 m and 1.7 m planes (Figures 11 and 12, respectively) for Cardiff, Doha and Amman. As shown in Figure 11, the window-windcatcher has increased the air velocity inside the building compared to the case without the window-windcatcher (i.e., Region F in Figure 12 compared to Region G in Figure 11). In addition, by comparing Regions F and G, it can be noted that the windowwindcatcher has directed the air velocity towards the centre compared to the case without the window-windcatcher, where air velocity tends to be more attached to the building's wall. This essentially indicates the window-windcatcher's capability in vectoring the shear wind towards the building's centre—thus increasing the ventilation rate.

Comparing the velocity profiles of the three case studies, the Cardiff case demonstrates higher velocity levels than Doha and Amman. This is directly related to the fact that the wind velocity in Cardiff is higher than in Doha and in Amman. By comparing the velocity profiles at the 1 m plane to those at the 1.7 m plane (Figures 11 and 12, respectively), while Region F in the 1 m plane (Figure 11) follows approximately the same profile as the velocity at the 1.7 m plane (Region I in Figure 12), Region H in the 1 m plane (Figure 11) differs from that in the 1.7 m plane (Region J in Figure 12). Region H has a more consistent velocity magnitude compared to Region I, where a drop in the magnitude of the velocity has been demonstrated.

The streamlines at the interior plane have been plotted for the 1 m and 1.7 m planes (Figures 13 and 14, respectively) for Cardiff, Doha and Amman. Streamlines are the fluid particle paths. Thus, they provide a visualization of the distribution of the air in the space. The locations with minimal velocities are also identified. These correspond to the regions with lower likelihood of spreading airborne viruses, such as SARS-CoV-2 [16,17]. Those locations correspond to the points where the velocity magnitude gradually decreases to approximately 0 m/s. Those locations are defined here as the near-zero-velocity regions. Correlating these regions to the potential of deploying air purifiers, it is recommended to avoid choosing those locations as those locations correspond to 'safe' regions. On the other hand, air purifiers could be deployed at locations with high velocity to maximise the airflow rate enters the air purifiers. As shown in Figures 13 and 14, the centre of safe near-zerovelocity regions (SNZVR) has been identified for the 1 m and 1.7 m planes, respectively.

In addition, a 70% household limit of the maximum velocity has been used to propose the locations to deploy air purifiers. Those locations have been defined as the air-purifier regions (APR). As shown in Figures 13 and 14, the streamline patterns at the 1 m follow closely the patterns in the 1.7 m plane. However, by comparing the case with the windowwindcatcher to the case without the window-windcatcher, it is noted that the SNZVR in Region K without the window-windcatcher (Figure 12) is split into two smaller SNZVR when the window-windcatcher is deployed (Region L). This is attributed to the fact that the window-windcatcher directs the air velocity towards the center, thus splitting the SNZVR. The interaction of the interior and exterior plane streamlines has been plotted for the 1 m and 1.7 m planes (Figures 15 and 16, respectively) for Cardiff, Doha and Amman. This helps us understand the interaction with the interior plane. As shown in Region M of Figure 15, the window-windcatcher has redirected the shear wind towards the interior plane of the building. However, it can be seen that some high-velocity wind streamlines (Region N) are not directed towards the internal plane of the building. This essentially suggests that the WWC design could be further optimized to enhance its aerodynamic capabilities. The WWC design optimization could include the depth of the WWC, the angles of the slanted fins and the number and thickness of the fins.

By comparing the 1 m plane in Figure 15 to the 1.7 m plane in Figure 16, we see that the streamline patterns are quite similar. Nevertheless, comparing the Cardiff case to Amman and Doha, it is noted that the velocity in the building is higher. This is due to the fact that the average wind velocity in Cardiff is higher than the other two cases of study. Another crucial point to note is that, in Region O, where the window-windcatcher has not been deployed (compare to region M), most of the shear wind streamlines are not directed towards the interior plane of the building. This suggests that the ventilation rate could be further increased by installing more than one WWC in Region O. Optimizing the number and positions of WWCs for a given space could significantly enhance the ventilation rate; this is another direction for future work.

**Figure 13.** Streamlines at the 1 m interior plane (APR: Air purifiers region, SNZVR: safe near-zerovelocity regions).

#### *3.2. Quantitative Analysis*

Figure 16 displays the average turbulence kinetic energy of the two interior planes, 1 m and 1.7 m. It can be noted that the window-windcatcher has slightly increased the average turbulence kinetic energy at both planes, in all three cases of study. However, as discussed in Section 3.1, the turbulence kinetic energy increases in the regions that are less likely to be occupied by residents. The increase in the turbulence kinetic energy is directly caused by the increase in the velocity at those planes, as highlighted in Figure 17.

By correlating the average turbulence kinetic energy results in Figure 17 to the velocity results in Figure 18, in a case study with low average velocity (i.e., Amman), it can be noted that the window-windcatcher effect of increasing the turbulence energy is less significant than in the other two cases. This is due to the lower wind speed compared to the other two cases.

**Figure 15.** Streamlines at the 1 m plane (internal and external domain).

As shown in Figure 19, the ventilation rates using the window-windcatcher have increased for the three case studies compared to those without the window-windcatcher. In all three case studies, the ventilation rate using the window-windcatcher was increased by approximately 9% compared to the case without the window-windcatcher.

As discussed in Section 2, the ventilation rate has been evaluated against the required ventilation rate (*Q*required) as per ASHRAE standards [30]. The acceptable ventilation rate in the studied residential building was estimated using Equation (6) [30] (i.e., approximately 168.5 L/s). The actual ventilation rate (*Q*actual) has been estimated computationally using ANSYS, and the ventilation performance has been benchmarked by the actualto-required ventilation ratio (*nQ*) using Equation (7) [30]. As shown in Figure 19, the window-windcatcher for Cardiff has managed to increase the actual-to-required ventilation ratio by approximately 9% compared to the case without the window-windcatcher (*nQ* increased from 96.7% to 106%). For Amman and Doha, *nQ* has increased by approximately 6% (from 61.5% to 67.5%) and 7% (from 70% to 77.3%), respectively (Figure 20).

**Figure 16.** Streamlines at the 1.7 m full plane.

**Figure 17.** Average turbulence kinetic energy of the interior planes (1 m and 1.7 m) using a Window-WindCatcher (WWC) and Without Window-WindCatcher (WWWC).

**Figure 18.** Average velocity of the interior planes (1 m and 1.7 m) using a Window-WindCatcher (WWC) and Without Window-WindCatcher (WWWC).

**Figure 20.** The actual-to-required ventilation ratio using Window-WindCatcher (WWC) and Without Window-WindCatcher (WWWC).

However, for Amman and Doha, although the actual-to-required ventilation ratio was increased using the window-windcatcher, the actual-to-required ventilation ratio is still less than 100%, that is, the actual ventilation rate is still lower than the required ventilation rate needed to fulfill the ASHRAE standards [30]. This suggests that deploying another windowwindcatcher is necessary (possibly at Region O, as indicated in Figure 14). On the other hand, for Cardiff, the ASHRAE standards are fulfilled using only one window-windcatcher since the actual-to-required ventilation ratio is higher than 100% (i.e., *nQ* = 106%).

To investigate how the performance of the window varies with boundary conditions, the actual ventilation rates of the three cities (which correspond to three different boundary conditions of wind velocities) are plotted against the total wind velocity in Figure 21. As the wind velocity increases, the window-windcatchers performance improves. In Figure 22, we plot the actual-to-required ventilation ratio of each city against the wind velocity (TWV), which also increases as the wind velocity increases.

**Figure 21.** Actual ventilation rate with respect to the total wind velocity.

**Figure 22.** Actual-to-required ventilation with respect to the total wind velocity.

By utilizing the curve-fitting tool of Matlab "cftool", we determine the relationships (8) and (9) which describe, respectively, the actual ventilation rate and the actual-to-required ventilation ratio against the wind velocity:

$$Qact = \text{35.75TVV} - 16.9 \tag{8}$$

$$m\_{\mathbb{Q}} = 0.212 \text{TWV} - 0.009 \tag{9}$$

These relationships could then be applied to other geographical locations. In this study, we focus attention to specific regions and boundary conditions (in Amman, Doha and Cardiff); additional locations and hence boundary conditions could be investigated in future studies.

#### **4. Discussion**

The purpose of this study was to investigate, using CFD, the effectiveness of a novel window-windcatcher device that could be mounted on the exterior walls of new or existing buildings in order to capture and redirect prevailing wind into interior spaces. The proposed design could replace or supplement the typical large-scale windcatchers by utilising small-scale decentralised windcatchers on exterior walls as a window component. The suggested window-windcatcher is among the passive cooling approaches that have been shown in several studies to provide outstanding thermal comfort and indoor air quality while consuming only a fraction of the energy used by mechanical air conditioning systems [6]. Natural ventilation is one of the most popular passive cooling design strategies for improving indoor air quality and thermal comfort. The effectiveness of natural ventilation and passive cooling in buildings could be improved by using this new device, resulting in better indoor air quality and environmental comfort.

Furthermore, ensuring high indoor air quality is essential in mitigating the spread of COVID-19 and other viral diseases. Increasing air velocity is the most common way to increase ventilation rates; however, increasing ventilation rate increases turbulence kinetic energy. As a result, it is critical to keep the increase in turbulence kinetic energy as low as possible. In addition, high turbulence kinetic energy regions have been carefully identified to facilitate potential mitigation measures, such as the use of air purifiers in high turbulence/mixing areas (Figures 9 and 10).

By plotting the streamlines in Figures 15 and 16, the effect of the window-windcatcher on the airflow streamlines has also been evaluated. Streamlines trace fluid paths and, as a result, contribute to the understanding of the indoor air distribution. They also visualise the regions with the lowest velocities, and, therefore, with a lower risk of spreading airborne viruses like SARS-CoV-2 [16,17]. At those locations, the velocity magnitude gradually decreases to approximately 0 m/s; Figures 13 and 14 show the near-zero-velocity regions. By relating these regions to the possibility of deploying air purifiers, it is suggested that these locations should be avoided, as they correspond to 'safe' regions. Air purifiers, on the other hand, should be placed in areas with high velocity to maximise the airflow rate into the purifiers. In Figures 13 and 14, those locations have been labelled as air-purifier regions (APR).

Furthermore, we looked at the interaction of the interior and exterior planes' streamlines. Understanding the fluid interaction with the interior plane requires plotting the streamlines for the exterior planes. We also identified potential locations to place the window-windcatcher in order to increase the ventilation rate. The window-windcatcher redirects the shear wind towards the building's interior plane, as shown in region M of Figure 14. However, some high-velocity wind streamlines (Region N) are not directed towards the building's interior plane. This implies that the window-windcatcher design could be further improved in future work in order to enhance its aerodynamic capabilities.

Another important point is that, unlike Region M, where the window-windcatcher was not deployed, most of the shear wind streamlines are not directed towards the interior plane of the building in Region O (Figure 14). This means that, by deploying another window-windcatcher in Region O, the ventilation rate could be increased. According to the quantitative analysis (Section 3.2), the window-windcatcher has slightly increased the average turbulence kinetic energy at both planes in the three cases studied. The increase in turbulence kinetic energy, on the other hand, occurred in areas that are less likely to be occupied by people. However, the ventilation rate was compared to the required ventilation rate *Q*required) as specified by the ASHRAE standards [30]. Compared to the case without the window-windcatcher (*nQ* from 96.7 percent to 106 percent), the windowwindcatcher managed to increase the actual-to-required ventilation ratio by approximately 9%. The actual-to-required ventilation ratio was increased by approximately 6% (from 61.5 percent to 67.5 percent) and 7% (from 70 percent to 77.3 percent) in the Amman and Doha cases, respectively.

#### **5. Conclusions**

This research aimed to investigate, using ANSYS CFX, the effectiveness of a design of a novel window-windcatcher device to be mounted on the exterior walls to capture the prevailing wind and redirect it into interior spaces. The performance of the proposed window-windcatcher has been evaluated in comparison to a control case (without a window-windcatcher) in three different geographical locations (Cardiff, Doha and Amman). The proposed window-windcatcher has proven to enhance thermal comfort and indoor air quality by increasing the actual-to-required ventilation ratio as per the ASHRAE standards by approximately 9% compared to the case without the window-windcatcher (*nQ* from 96.7% to 106%). For Amman and Doha, the actual-to-required ventilation ratio was increased by approximately 6% (61.5% to 67.5%) and 7% (70% to 77.3%), respectively.

In addition, the location with minimal velocities has been identified by plotting the streamlines. Those locations correspond to the regions with a lower likelihood of spreading viral diseases such as COVID-19. Regarding the potential of deploying air purifiers, airpurifier regions (APR) with high velocity, where the airflow rate entering the air purifiers is increased, have been identified.

For future work, a prototype is recommended to be manufactured and tested in a wind tunnel to further examine the aerodynamic performance of the window-windcatcher.

**Author Contributions:** Conceptualization, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; methodology, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; software, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; validation, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; formal analysis, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; investigation, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; resources, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; data curation, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; writing—original draft preparation, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; writing—review and editing, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; visualization, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; supervision, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; project administration, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A.; funding acquisition, O.F.A., L.M.O., S.N.M., K.K., T.A.-R. and A.I.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This publication was made possible by NPRP 13 Grant No. NPRP13S-0203-200243 from the Qatar National Research Fund (a member of the Qatar Foundation). The findings herein reflect the work and are solely the responsibility of the authors. Open Access funding is provided by the Qatar National Library.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors gratefully acknowledge the support Oxford University through Ian Griffiths.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Experimental Methods of Investigating Airborne Indoor Virus-Transmissions Adapted to Several Ventilation Measures**

**Lukas Siebler \*, Maurizio Calandri, Torben Rathje and Konstantinos Stergiaropoulos**

Institute for Building Energetics, Thermotechnology and Energy Storage (IGTE), University of Stuttgart, Pfaffenwaldring 35, 70569 Stuttgart, Germany

**\*** Correspondence: lukas.siebler@igte.uni-stuttgart.de; Tel.: +49-711-685-60785

**Abstract:** This study introduces a principle that unifies two experimental methods for evaluating airborne indoor virus-transmissions adapted to several ventilation measures. A first-time comparison of mechanical/natural ventilation and air purification with regard to infection risks is enabled. Effortful computational fluid dynamics demand detailed boundary conditions for accurate calculations of indoor airflows, which are often unknown. Hence, a suitable, simple and generalized experimental set up for identifying the spatial and temporal infection risk for different ventilation measures is more qualified even with unknown boundary conditions. A trace gas method is suitable for mechanical and natural ventilation with outdoor air exchange. For an accurate assessment of air purifiers based on filtration, a surrogate particle method is appropriate. The release of a controlled rate of either trace gas or particles simulates an infectious person releasing virus material. Surrounding substance concentration measurements identify the neighborhood exposure. One key aspect of the study is to prove that the requirement of concordant results of both methods is fulfilled. This is the only way to ensure that the comparison of different ventilation measures described above is reliable. Two examples (a two-person office and a classroom) show how practical both methods are and how the principle is applicable for different types and sizes of rooms.

**Keywords:** aerosol infection risk; SARS-CoV-2 transmission; measurement methods; surrogate particles; trace gas; ventilation measures

#### **1. Introduction**

The ongoing pandemic teaches us that the implementation of appropriate ventilation measures can significantly reduce indoor infection risks [1–4].

For fast and simple assessments, many calculation models regarding the infection risk of SARS-CoV-2 in indoor environments exist [5–10]. Often, these are based on idealised assumptions (e.g., ideal mixed ventilation). In reality, the condition of ideal mixed ventilation is impossible to achieve (finite velocity for distributing locally released substances), especially for larger rooms. Even for small to medium scales, the calculation models may not resolve aerosol dispersion appropriately.

In practice, different ventilation principles—natural ventilation and mechanical ventilation (mixing ventilation, displacement ventilation, downward ventilation)—are applied, depending on the type of use and the occupancy density of the room [11,12]. Depending on the meteorological boundary conditions such as wind velocity, natural ventilation can be classified between mixed ventilation and displacement ventilation [13]. These transient effects cannot be represented by a simple calculation model [14]. Furthermore, existing disturbance influences (leaks and opening of windows/doors, downdrafts, etc.) have to be considered for the respective ventilation type.

Displacement ventilation concepts, often applied in large halls, impede estimations of virus transmission to neighbors. In ideal theory, there would occur only unobjectionable vertical buoyancy flows. In reality, however, these are superimposed by disturbance

**Citation:** Siebler, L.; Calandri, M.; Rathje, T.; Stergiaropoulos, K. Experimental Methods of Investigating Airborne Indoor Virus-Transmissions Adapted to Several Ventilation Measures. *Int. J. Environ. Res. Public Health* **2022**, *19*, 11300. https://doi.org/10.3390/ ijerph191811300

Academic Editors: Ashok Kumar, Michał Piasecki, M Amirul I Khan and Alejandro Moreno Rangel

Received: 15 August 2022 Accepted: 3 September 2022 Published: 8 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

effects [15–17], which might be critical regarding neighbor infection risks. If the relevant boundary conditions for these disturbing influences are unknown, the estimation is regarded to be even more critical.

For individual considerations and higher accuracies, experimental studies examine the effects of ventilation measures on airborne infections. Some of them deal with a real virus load of the indoor air with actually present infectious persons, without controlled boundary conditions [18]. In Li et al. [19], trace gas measurements and flow simulations for an actual infection scenario are conducted in order to reproduce transmission paths in detail. Transient and absolute considerations of virus loads are omitted in this context. Another research study examines the effects of ventilation measures, but does not take into account the transfer of substitutes to viruses and thus the actual infection risk [20]. As a result, only relative statements on the risk reduction are possible.

In general, ventilation measures can be divided into filtering recirculation methods (e.g., air purification) and ventilation principles with outside air exchange (e.g., natural/mechanical ventilation). However, there are no published studies on the comparative assessment of both measures with regard to the infection risk.

Previous experimental research is not yet suitable for a holistic individual assessment of the infection risk. Therefore, a comprehensive and standardized measurement procedure for any airborne transmitted disease is required. In this paper, two consistent measurement methods are presented, which allow for determining spatially and temporally resolved infection risks in rooms accurately. For the first time, the two consistent approaches can be used to evaluate both filtering recirculating air operations and ventilation principles with outdoor air exchange.

#### **2. Theory of Airborne Virus Infections**

In 1955, the first approach of assessing general infection risks was developed, which was improved and specified in 1978 on the disease measles [21,22]. The authors introduced the dose of inhaled quanta *D*<sup>q</sup> as an indicator of whether an infection occurs. Taking into account the efficiency of masks, it can be calculated as:

$$D\_{\mathbf{q}}(t) = \dot{V}\_{\text{inh}} \left(1 - \eta\_{\text{inh}}\right) \int\_{0}^{t} c\_{\mathbf{q}}(t) \, \mathbf{d}t + D\_{\mathbf{q}|0} \tag{1}$$

with *V*˙ inh, *η*inh, *c*q(*t*), and *D*q 0 as inhalation volume flow, mask filtration efficiency for inhaling, quanta concentration at time *t* and the initial value of *D*q, respectively.

The approach of Wells et al. and Riley et al. allows the computation of the predicted infection risk via aerosols (PIRA), labeled as *P*<sup>I</sup> [22]:

$$P\_1 = 1 - \mathbf{e}^{-D\_q}.\tag{2}$$

As an alternative, the dose–response model follows a modified principle that is similarly referenced in science and should be roughly presented. Hereby, the number of pathogens that result in infections of a certain proportion of a group is usually determined on the basis of empirical animal experiments. Based on an analysis of how many pathogens a person exhales, the infection risk can be derived. The dose–response as well as a Wells– Riley model are accepted as valuable tools in epidemiological studies. When using these models, the respective advantages, disadvantages and uncertainties must be weighed [23].

In this paper, the Wells–Riley model is followed. However, the relevant equations can be easily modified to a dose–response model, due to substituting quanta emission rates (QER) with pathogen emission rates and an adaption of the risk assessment.

In general, infection risks can be reduced by either ventilation measures (mechanical and natural ventilation) or air purification concepts (without outside air exchange) can be used. For a non-spatially resolved estimate of the quanta concentration, the general differential Equation (3), which includes various ventilation concepts, has to be evaluated:

$$\frac{\mathrm{d}c\_{\mathrm{q}}(t)}{\mathrm{d}t} = \underbrace{\frac{\dot{q}\_{\mathrm{out}}\left(1-\eta\_{\mathrm{exh}}\right)}{V\_{\mathrm{r}}}}\_{\begin{subarray}{c}\mathrm{quanta}\\ \mathrm{of an inflection}\\ \mathrm{perconf}\end{subarray}} - \underbrace{\frac{\dot{V}\_{\mathrm{dev}/\mathrm{w}}c\_{\mathrm{q}}(t)}{V\_{\mathrm{r}}}\underbrace{\eta\_{\mathrm{dav}}}\_{\begin{subarray}{c}\mathrm{quanta}\\ \mathrm{of ventilation}\\ \mathrm{depostion}\end{subarray}} - \underbrace{\Phi c\_{\mathrm{q}}(t)}\_{\begin{subarray}{c}\mathrm{natural}\\ \mathrm{activation}\\ \mathrm{depostion}\end{subarray}}(0) = c\_{\mathrm{q}0}\tag{3}$$

with *q*˙out, *η*exh, *V*r, *V*˙ dev/w, *η*dev, Φ˙ as quanta rate (output), mask filtration efficiency for exhaling, room volume, volume flow of device or window and device efficiency (filtration ratio for air purifiers, exhaust air to outdoor air exchange for ventilation systems, equals 0 for natural ventilation) and combined rate of natural inactivation and deposition of existing viruses, respectively. The last equation term can be described and modelled well in theory, but, in practice, it is challenging to include it in experimental settings. Integrating Equation (3) over time followed by using (1) and (2), the dose of inhaled quanta and finally PIRA can be calculated.

With these assumptions, it is possible to estimate infection risks using ventilation devices, air purifiers and natural ventilation under ideal mixed ventilation conditions. However, experimental methods of substance dispersion concerning the airborne transmission of virus infections enable accurate spatially and temporally resolved results for any ventilation principle even for unknown boundary conditions.

#### **3. Experimental Methods**

Airborne virus transmission is mostly dealing with particles below 10 μm of size [24,25]. These particles are meant to have negligible sink velocities and follow airflows almost exactly [26]. A common method investigating airflows quantitatively is releasing substances (which also follow the airflow) and measuring their concentrations. In order to evaluate the totality of the ventilation measures, suitable substances are trace gas and surrogate particles. For scenarios with outside air exchange, only trace gas is appropriate because of entering particles from the outside, which would falsify particle measurements. For recirculation devices based on filtration, however, trace gas is inapplicable since it is not filtered and therefore no particle removal can be measured. For this reason, two different methods are essential.

#### *3.1. Trace Gas Method*

The focus of previous indoor air investigations using trace gas, also conducted at the University of Stuttgart, has mostly been on evaluating the ventilation effectiveness rather than infection risks [27]. The method is based on the emission of trace gas, which is not present in the natural surrounding and whose concentration can be measured by infrared spectrometers (e.g., N2O or SF6). First, the room needs to be freed from trace gas from eventual previous measurements. By using a mass flow controller with gas-specific adapted properties, a constant and continuous mass flow of the trace gas is possible.

Along with the measured trace gas concentration, the quanta rate (output) and the respiratory rate following theoretical post processing lead to a calculation of the infection risk of a certain disease.

For trace gas, which is transported in a similar way to airborne particles and viruses (below 5 μm) [28], the following assumption without mask filtration efficiency is applied:

$$\frac{\dot{m}\_{\text{in tg}}(t)}{\dot{m}\_{\text{out tg}}} = \frac{\dot{q}\_{\text{in}}(t)}{\dot{q}\_{\text{out}}} \tag{4}$$

with *n*˙ in tg, *n*˙ out tg, *q*˙in, *q*˙out as (fictitious) molar flow for trace gas (input), molar flow for trace gas (output), quanta rate (input) and quanta rate (output) respectively. Note that input and output represent a fictitious inhalation and exhalation.

From the (time dependent) quanta rate (input) *q*˙in, the dose of inhaled quanta *D*<sup>q</sup> is determined by integration over time (assumption: *D*q 0 = 0, *c*q 0 = 0):

$$D\_{\mathbf{q}} = \int\_{0}^{t} \dot{q}\_{\mathrm{in}}(t) \,\mathrm{d}t = \frac{\dot{q}\_{\mathrm{out}}}{\dot{n}\_{\mathrm{out}\,\mathrm{tg}}} \int\_{0}^{t} \dot{n}\_{\mathrm{in}\,\mathrm{tg}}(t) \,\mathrm{d}t. \tag{5}$$

The molar flow for (fictitious) trace gas input is calculated by using *n*˙ = *m*˙ /*M* and the assumption *c*tg(*t*) 1 moltg/molair:

$$
\dot{m}\_{\rm in\,tg}(t) = c\_{\rm lg}(t)\,\dot{n}\_{\rm inh} = c\_{\rm lg}(t)\,\frac{\rho\_{\rm air}\,\dot{V}\_{\rm inh}}{M\_{\rm air}}\tag{6}
$$

with *c*tg, *n*˙ inh, *ρ*air, and *M*air as measured trace gas concentration, inhalation molar flow, density and molar mass of air, respectively.

The resulting equation for the dose of inhaled quanta *D*q, considering mask filtration efficiency (see [29,30]), is:

$$\begin{split} D\_{\mathrm{q}} &= \int\_{0}^{t} \dot{\eta}\_{\mathrm{in}}(t) \, \mathrm{d}t \\ &= \left( 1 - \eta\_{\mathrm{inh}} \right) \left( 1 - \eta\_{\mathrm{exh}} \right) \frac{\dot{\eta}\_{\mathrm{out}} \, M\_{\mathrm{tg}}}{\dot{m}\_{\mathrm{out}} \, \mathrm{tg}} \, \frac{\rho\_{\mathrm{air}} \, \mathrm{V}\_{\mathrm{inh}}}{M\_{\mathrm{air}}} \int\_{0}^{t} c\_{\mathrm{tg}}(t) \, \mathrm{d}t \end{split} \tag{7}$$

with *M*tg, *m*˙ out tg as molar mass and mass flow of trace gas (output), respectively.

A numerical integration of the concentration of trace gas over time *c*tg(*t*) (e.g., via trapezoidal rule) allows the transfer of measurement data via Equation (2) into an infection risk for scenarios with outdoor air exchange. Furthermore, data for the quanta rate (output) are required, which can be obtained from literature or identified by a backward processing of this introduced approach, (see Section 6).

#### *3.2. Surrogate Particles Method*

To evaluate the function of an air purifier based on filtration, an alternative method is required because trace gas can not be filtered. Even though the approaches have analogous equations, they still involve different physical units and more elaborate conditioning, since the room needs to be initially particle-free. From then on, only particles that are actually released by particle generators are measured. In reality, it cannot be avoided that a certain number of particles is still present. Superimposing this concentration with high emissions is a reasonable measure. If the particle emission rate is set very high, particles coming from outside (e.g., via infiltrations) have no influence on the measurement results because their particle rate is several powers of ten lower. In this way, a simple and cost-efficient measuring technique is enabled, which can be widely applied.

A suitable substance for particles is Di-Ethylhexyl-Sebacat (DEHS). Besides the advantage that its particles keep the same size over time because of very low evaporation rates, the measured size distribution is similar to the exhaled particles of humans [25,31].

Any other possible particle sources (even persons) should be removed from the room of investigation or kept emitting as low as possible. Using several optical particle counters (OPC) or scanning mobility particle sizers (SMPS) on different positions allow a spatial and temporal measurement of the number concentration and size distribution of particles.

The measured surrogate particle concentration, the quanta rate (input) and the respiratory rate lead to the calculation of the infection risk for a room equipped with an air purifier. Furthermore, data for the quanta rate (output) are required, which can be obtained from literature.

Analogous to the trace gas method for surrogate particles, the following assumption without mask filtration efficiency is applied:

$$\frac{\dot{m}\_{\text{in }\text{sp}}(t)}{\dot{m}\_{\text{out }\text{sp}}} = \frac{\dot{q}\_{\text{in}}(t)}{\dot{q}\_{\text{out}}} \tag{8}$$

with *m*˙ in sp, *m*˙ out sp as mass flow for surrogate particles input and output, respectively (fictitious inhalation and exhalation of particles).

Similar to the trace gas method, the dose of inhaled quanta *D*<sup>q</sup> is (assumption: *D*q 0 = 0, *c*q 0 = 0):

$$D\_{\mathbf{q}} = \int\_{0}^{t} \dot{q}\_{\mathrm{in}}(t) \, \mathrm{d}t = \frac{\dot{q}\_{\mathrm{out}}}{\dot{m}\_{\mathrm{out}\,\mathrm{sp}}} \int\_{0}^{t} \dot{m}\_{\mathrm{in}\,\mathrm{sp}}(t) \, \mathrm{d}t. \tag{9}$$

Substituting the assumed constant mass flow (output) of surrogate particles (*m*˙ out sp = *c*out sp *V*˙ ag) and a fictitious mass inhalation rate of surrogate particles (*m*˙ in sp(*t*) = *c*in sp(*t*) *V*˙ inh) in (9) (considering mask filtration efficiency see [29,30]) results in:

$$D\_{\rm q} = \int\_{0}^{t} \dot{q}\_{\rm in}(t) \,\mathrm{d}t = \left(1 - \eta\_{\rm inh}\right) \left(1 - \eta\_{\rm exh}\right) \frac{\dot{q}\_{\rm out}}{c\_{\rm outp} \dot{V}\_{\rm ag}} \,\mathrm{\,\dot{V}\_{\rm inh}} \int\_{0}^{t} c\_{\rm in.sp}(t) \,\mathrm{d}t \tag{10}$$

with *V*˙ ag, *c*out sp, *c*in sp(*t*) as volume flow of aerosol generator and mass concentration of surrogate particles (output and input), respectively.

With regard to suspected agglomeration effects, it seems appropriate to extend the detected bandwidth of the emission size distribution upwards compared to the emission bandwidth and to operate via mass-related units. If particles agglomerate, this has an influence on the particle number concentration but not on the particle mass concentration and the infectivity. Therefore, the mass concentration should be applied (in this case, up to an optical diameter of 10 μm, see also Section 6). This can be determined by cumulating mass concentrations of each size fraction by calculating their volumes and their weight with the density of the specific particle substance.

#### **4. Comparison between the Two Methods**

One key aspect of the study is to prove that both methods are concordant. Hence, a comparison between the two measuring methods for substance dispersion is essential. Bivolarova et al. [32] compare trace gases with different particle sizes on the basis of steadystate observations and decay curves of substance concentrations. A temporal resolution of these concentrations from the initial to the steady state—i.e., the dispersion of the substances—as well as a transfer to an infectious event is not focused here.

However, depending on the infectiousness of a disease and the exposure time, the transient dispersion is relevant to an infection process. In addition, the relation of substance release rates (emission) to emerging concentration profiles (emission) is not yet examined and thus a transfer to a fictitious unit such as quanta is not possible.

For these reasons, an experimental comparison is conducted by the authors with respect to transient quanta concentration courses in an airflow visualization laboratory with controlled boundary conditions.

In this laboratory, both a controlled supply and exhaust air volume flow is adjusted. At a specific position (see Figure 1), controlled flows of both a DEHS particle-laden aerosol (with known emission particle concentration) and a SF6 trace gas are released simultaneously. Data of the aerosol generator for the particle release and the mass flow controller for trace gas are given in Table 1. OPCs and sampling tubes for the FTIR gas analyser are placed at six positions in the laboratory. This allows the concentrations of both the particles and the trace gas to be measured simultaneously. Since the authors have six OPCs but

only one FTIR at their disposal, the suction of the six trace gas sampling points is changed automatically every minute. In order to take into account the time required for the gas to pass through these tubes to the FTIR, the measured values of each first 30 s are discarded. Switching the sampling points between six positions results in a 6-fold lower sampling rate compared to the OPCs (see Figure 2). In order to ensure that almost no outdoor particles enter the room, a HEPA 14 filter (corresponds with Minimum Efficiency Reporting Values (MERV) 19) was integrated into the duct of supply air. Otherwise, in addition to the released particles, those from the outside would also be measured and therefore corrupts the experimental data. A simultaneous release and measuring of particles and trace gas at the same position are expected to result in the same dispersion of both substances. In this case, the calculated course of the curves of quanta concentrations for both substances is supposed to be similar in all measuring positions.

**Figure 1.** Set up to compare the two methods (**left**) and top view of the conference room (**right**).

**Figure 2.** Comparison between particle and trace gas measurement as a function of quanta concentration over time exemplified in a conference room (position 3).


**Table 1.** Relevant parameters of the measurement devices.

In Table 1, the parameters of the measurement devices are shown.

The aerosol is generated with a particle number concentration of 1.5 × <sup>10</sup><sup>7</sup> cm<sup>−</sup>3, which are approximately logarithmic normal distributed (median optical diameter ≈ 0.3 μm). The OPCs are able to measure particle size distributions of 0.175... 20 μm over 64 channels. Furthermore, data for the quanta rate (output) are required, which can be obtained from literature, see Peng et al. [33].

The experimental set up for the comparative measurement is presented in Figure 1. For the delta variant of SARS-CoV-2, a medium spreader is assumed by Peng et al. [33]. The corresponding boundary conditions are shown in Table 2.

**Table 2.** Relevant parameters of the comparative measurement.


Deviating from usual recommendations of a certain air exchange rate, reference is given to the personal volume flow, see Kriegel et al. [2]. Depending on the viral load, exposition time, non-pharmaceutical measures, etc., this value may differ. However, this paper will focus on the comparison and concordance of the two measurement methods.

Over the entire measurement period, both quanta concentration courses are similar at all measurement positions. Therefore, it is expected that both methods are suitable and can be applied for different ventilation cases. In Figure 2, the exemplary result at a single measuring position (best fit) in the conference room is shown. In addition, it is worth emphasizing the small discrepancy between the two experimental curves compared to the theoretical one after a quasi-steady state has been reached. This is due to high air exchange rates and the possibility of accurate volume flow measurements (correlation to steady state concentration) in the laboratory. However, during the transient course, a discrepancy is generally observed, since substances in reality need time to disperse to a certain position.

Different measurement techniques can generally lead to different sampling rates. In this measurement, only one infrared spectrometer is used for the trace gas concentrations. By means of a measuring position switch, six positions are taken into account successively over one minute each. On the other hand, six OPCs simultaneously record the particle concentrations of a single measuring position every minute. Therefore, the sampling rate for trace gas is six times lower compared to particles.

#### **5. Experimental Studies**

In the following section, it is shown that both methods are practical. The measured scenarios go from small to medium scale and contain the two of three main ventilation types: a two-person office (air purifier) and a typical classroom (natural ventilation vs. air purifier). These investigated example scenarios are shown in detail.

Besides supply and extract air effects, indoor airflows are influenced by sub- and over-tempered surfaces (e.g., thermal sources). For an appropriate simulation of human heat outputs for a certain activity rate (low activity: 75 W) [34], thermal dummies with a surface area of an average human (1.8 m2) [35] are recommended. Figure 3 shows a professional thermal person dummy and also a low cost variant (cartons with lightning bulbs inside) for substituting many persons in large rooms (e.g., Stuttgart Drama Theatre).

**Figure 3.** Professional thermal person dummy (**left**), low cost variants in Stuttgart Drama Theatre (**right**).

#### *5.1. Spatially Resolved Measurements in a Laboratory*

Under reproducible boundary conditions in an air visualisation laboratory, an air purifier has been investigated by the authors. The focus lies on the particle removal effectiveness of the device in a two-person office.

Thereby, one of both people is assumed to be infectious. Figure 4 shows a photo and the top view of the experimental set up; the relevant parameters of the experiments concerning the effectiveness of air purifiers are described in Table 3; for the corresponding parameters of the measurement devices, see Table 1.

**Figure 4.** Photo (**left**) and the top view (**right**) of the experimental set up of a two-person office with air purifiers in an air visualisation laboratory. \* length unit in metre.


**Table 3.** Relevant parameters of the experiments concerning the effectiveness of air purifiers.

The particle measurement is analogous to the procedure described in Section 4. However, the filtered outdoor air is now replaced by recirculated air of a filtering air purifier. First, the room is freed from particles by the air purifier. Thereafter, the aerosol is subsequently emitted from the marked dummy (see Figure 4) with the properties of Table 1 over 60 min. The particle concentrations are determined by the OPCs at six different positions in the room in order to obtain spatial information about the substance removal.

The temporally-resolved data allow a computation of the infection risk (PIRA) at these positions using Equation (10). An exemplary measurement result is shown in Figure 5.

**Figure 5.** Theoretical (ideal mixed air) and measured quanta concentration (left axis) and their related mass concentration (right axis) of six positions in a two-person office.

Depending on the ventilation principles (mixing ventilation, displacement ventilation, downward ventilation), the spatial deviations from the assumption of ideal mixed ventilation (in relation to the entire room) may vary. The deviations of the individual positions among each other show a significant benefit of a spatial view. Moreover, quanta concentrations in Figure 5 do not only mismatch the ideal mixed ventilation in the transient range but also in the steady state. In this case for a duration of 1 h, the overall infection risk (PIRA) for a medium spreader (delta variant) by Peng et al. [33] using Equation (2) results in a range from 2.2% (position 4) to 4.0% (position 2).

#### *5.2. Comparison of Ventilation Measures in Classrooms*

Further investigations of ventilation measures in classrooms were conducted by the authors, in order to assess the effectiveness of periodic window ventilation, decentralised ventilation systems and air purifiers regarding infection risk and thermal comfort. For this study, it is required that both methods (trace gas and surrogate particles) perform similarly. The set up and the three ventilation measures are shown in Figure 6 exemplarily.

**Figure 6.** Experimental set up (**top**), exemplary window ventilation (**bottom left**), exemplary decentralised ventilation system (**bottom middle**), exemplary air purifier (**bottom right**).

Table 4 describes the relevant parameters of both experiments in an exemplary classroom, and Table 1 shows the corresponding parameters of the measurement devices. The weather data given are mean values of a meteorological station within the measurement period. The weather data are particularly important for window ventilation. They are irrelevant for the operation of the air purifier, since this experiment was carried out separately.


**Table 4.** Boundary conditions of the experiments in the classroom.

<sup>1</sup> mean values. <sup>2</sup> th: top-hung, sh: side-hung. <sup>3</sup> clear opening.

Due to changing boundary conditions during this study (compared to reproducible conditions in the laboratory), only one exemplary classroom is considered for an air purifier and a periodic window ventilation (in this case, 20/5/20: 20 min closed, 5 min opened). For an assumed medium-spreader (SARS-CoV-2, delta variant) [33] and an exposition time of 1.5 h in both scenarios, there is the same position of the infectious person (substance release).

The investigation of air purification and natural ventilation is carried out successively. The experimental method for the air purifier is carried out in the same way as described in Section 5.1 using DEHS-particles. The reliable measurement of natural ventilation is conducted with trace gas. In this case, SF6 is released in a controlled manner analogous

to Section 4, and its concentration is measured at six different positions. The periodic window opening is carried out manually according to a stopwatch. The room geometry and dimensions as well as the detailed set-up of the experiment including the positions of the substance release, the measurement and the air purifier are shown in Figure 7.

**Figure 7.** Draft (**left**) and the top view (**right**) of the experimental set up of a classroom with an air purifier.

At two different measurement positions (1 and 4), the quanta concentrations are illustrated in Figure 8. The varying curves show how valuable both spatial resolved methods are.

**Figure 8.** Comparison of quanta concentrations for a window ventilation and an air purification in a classroom.

Besides the position of an infectious person, the location of exposed persons and the air in-/outlets have a significant influence on the infection risks, especially for a noncontinuous ventilation via windows. In this particular case, position 4 is proximate to the window front and therefore experiences higher local ventilation rates, which results in a lower quanta concentration course. Although this position is also close to the air

purifier, in this situation, it experiences lower local ventilation rates than position 1 and thus a higher quanta concentration course. It is assumed that this anomaly is based on the occurring indoor airflow due to an upwards directed supply air. Besides the possibility of detailed local analysis, this plot emphasizes that the assumption of a theoretical ideal mixed ventilation is inappropriate in this example.

In scenarios with both principles (filtering device and ventilation with outdoor air exchange), like an air purifier combined with a window or mechanical ventilation, a new challenge arises. Outdoor air contains many particles of the same sizes as released by an aerosol generator, which means that an OPC cannot separate between surrogate and outdoor particles. One pre-study in this project shows that a simultaneous outdoor SMPS measurement of particle size distribution provides a solution. There are usually overlaps of the size distributions of outdoor particles and released ones. Consequently, if only particles outside this overlap are considered for both release and detection, the effect of both ventilation measures can be evaluated.

#### **6. Discussion**

There are two different models for estimating infection risks. Both Wells–Riley and dose–response models are accepted in the scientific community as valuable tools. Even though both experimental methods are easily adaptable, the formulas used in this paper are based on Wells–Riley. One of its criticisms is the uncertainty of quanta emission rate (QER) determination, using a backward calculation for the assumption of an ideal mixed ventilation [23]. Agreeing on that criticism, it should be suggested to use the two introduced experimental methods as well for an accurate determination of these values. With an experimental reconstruction of several infection scenarios, and a backward procedure to the one introduced in this paper, QER could be determined more reliably. By iteratively adjusting a still fictitious QER, infection risks could be derived for all exposed persons. Taking into account numerous scenarios and the persons actually infected, a mean resilient value could be identified using appropriate stochastic tools. This procedure enables a consideration of any kind of ventilation principle, instead of being limited to an ideal mixed ventilation.

A closer look at the experimental methods reveals that operating with surrogate particles (compared to trace gas) has higher uncertainties due to agglomeration, sedimentation and deposition effects. The intensity of agglomeration effects is discussed on the basis of measurements related to Figure 2, which is illustrated in a more detailed version (particle number concentration, particle mass concentrations PM1/PM2.5/PM10) in Figure 9.

The assessment on how well the quanta curves determined by surrogate particle method fit relatively to the ones by trace gas method is calculated as follows:

$$\Psi = \frac{\int\_0^t |c\_{\rm qref}(t) - c\_{\rm q}(t)| \,\mathrm{d}t}{\int\_0^t c\_{\rm qref}(t) \,\mathrm{d}t} \tag{11}$$

with Ψ and *c*q ref as curve agreement evaluation and reference quanta concentration (trace gas method), respectively. Table 5 shows the results of the evaluation on the curve agreement between the various approaches.

**Figure 9.** Comparison of quanta concentrations based on particle mass, particle number and trace gas concentrations in a conference room.



Since particle number concentration decreases due to agglomeration effects, this results in Ψ = 48.7%. For mass concentrations considered below 1 μm, agglomeration might cause particles leaving the upper limit of the OPC's detection range, which results in lower calculated quanta concentrations compared to trace gas. For 2.5 μm, this effect becomes smaller because of a lower amount of large particles. It is even negligible for 10 μm, and its curve almost matches the trace gas course (Ψ = 5.6%). To avoid underestimated infection risks, it is highly recommended to consider mass concentrations up to 10 μm even though the median size of particle release is below 1 μm. Several studies suggest that infectious particles, which remain suspended in the air, can be much larger [36] However, the release of particles with a median of less than 1 μm provides an overestimation of the infection risk. Apart from the lower filter removal efficiency for these particle sizes, they are also more likely to be airborne, resulting in fewer deposition effects. Even if in reality larger particles are emitted by humans, these two aspects ensure a conservative assessment of the infection processes.

Particle agglomeration further impacting both methods might be caused by wall effects, whereas trace gas should be reflected, and liquid particles are assumed to be trapped at walls. This might be one possible explanation for a lower curve of particle mass compared to trace gas concentration (Figure 9).

#### **7. Conclusions**

In order to estimate the infection risk of airborne indoor virus-transmissions, either calculation models or measurements can be carried out. The implemented simplifications of these calculation models (e.g., ideal mixed ventilation) might deliver fast but inaccurate results for certain scenarios. Previous experimental studies examine the effects of ventilation measures in more detail. However, instead of transient and absolute considerations of viral loads, they often regard relative statements of the infection risk.

Therefore, this study presents two experimental methods that are capable of determining a temporally and spatially resolved infection risk (absolute values) for different ventilation measures. Since particles entering from the outside would falsify the particle measurements, only the trace gas method is suitable for ventilation systems with air exchange and for natural ventilation. However, an accurate assessment of air purifiers based on filtration is only applicable by the surrogate particle method because trace gas is not filtered and therefore no device effect can be measured. For this reason, two different methods are essential.

Both methods are based on the theory that particles of relevant scales for infection procedures are airborne. The release of a controlled rate of either trace gas with a mass flow controller or particles with an aerosol generator allows a simulation of an infectious person releasing virus material. The measurement equipment includes an infrared spectrometer (trace gas method) or optical particle counters (surrogate particle method). For both approaches, the mathematical transfer of measured concentrations into infection risks is presented. In order to prove that the two methods are concordant, a comparison is essential. In an air visualisation laboratory with filtered outside air exchange, both methods are executed simultaneously. In fact, they provide similar results. This allows a firsttime reliable experimental comparison of ventilation systems, natural ventilation and air purifiers.

Besides the detailed explanations and the comparison of both methods, several aspects that might influence the accuracy are discussed. Two exemplary scenarios show how practical both methods are and how scalable this principle is. Even if the ventilation concept deviates significantly from mixed ventilation, infection risks can be determined. Besides a two-person office, results of measurements performed in an exemplary classroom are presented. Both scenarios highlight the value of experimental investigations with temporal and spatial resolutions for determining infection risks. This allows, even for complex geometries, an assessment of the exposition time in rooms (e.g., workplace, events) and the identification of critical zones. On the one hand, the selection of a device and the related operating parameters such as volume flow can be determined. On the other hand, optimizations can be carried out (e.g., device positioning, orientation of supply air diffusers, permissible occupancy density). Furthermore, an evaluation of air purifiers beyond the quantity (Clean Air Delivery Rate, CADR) could be supplemented with information about locally resolved substance removal even under transient conditions.

**Author Contributions:** Conceptualization, L.S. and K.S.; methodology, L.S.; software, L.S., M.C. and T.R.; validation, L.S., M.C. and T.R.; formal analysis, L.S., M.C., T.R. and K.S.; investigation, L.S., M.C. and T.R.; resources, K.S.; data curation, L.S., M.C. and T.R.; writing—original draft preparation, L.S., M.C. and T.R.; writing—review and editing, K.S.; visualization, M.C. and T.R.; supervision, L.S. and K.S.; project administration, L.S. and K.S.; funding acquisition, K.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** Investigations of Section 5.1 was funded by the Ministry of Science, Research and the Arts of Baden-Württemberg (MWK) as part of the project "Test aerosols and methods for efficacy testing of air purification technologies against Sars-CoV-2" with the grant number: 33-7533.-6-21/15/1.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets generated/analysed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** We would like to thank the Ministry of Science, Research and the Arts of Baden-Württemberg (MWK) for funding the experiment of Section 5.1. Special thanks go to Michel Bauer for his scientific editing services.

**Conflicts of Interest:** The authors declare no conflict of interest.


#### **Nomenclature**

#### **Abbreviations**

The following abbreviations are used in this manuscript:


#### **References**


## *Review* **Indoor Air Quality: A Review of Cleaning Technologies**

**Teresa M. Mata 1,\*, António A. Martins 2,3, Cristina S. C. Calheiros 4, Florentina Villanueva 5, Nuria P. Alonso-Cuevilla 6, Marta Fonseca Gabriel 1,\* and Gabriela Ventura Silva 1,\***


**Abstract:** Aims: Indoor air quality (IAQ) has attracted increased attention with the emergence of COVID-19. Ventilation is perhaps the area in which the most changes have been proposed in response to the emergency caused by this virus. However, other strategies are possible, such as source control and the extraction of pollutants. The latter incorporates clean technologies, an emergent area with respect to IAQ. Method: Various air treatment technologies can be used to control contaminants, which are reviewed and discussed in this work, including physicochemical technologies (e.g., filtration, adsorption, UV-photocatalytic oxidation, ultraviolet disinfection and ionization) and biological technologies (e.g., plant purification methods and microalgae-based methods). Results and interpretation: This work reviews currently available solutions and technologies for "cleaning" indoor air, with a focus on their advantages and disadvantages. One of the most common problems in this area is the emission of pollutants that are sometimes more dangerous to human health than those that the technologies were developed to remove. Another aspect to consider is the limitation of each technology in relation to the type of pollutants that need to be removed. Each of the investigated technologies works well for a family of pollutants with similar characteristics, but it is not applicable to all pollutant types. Thus, the optimal solution may involve the use of a combination of technologies to extend the scope of application, in addition to the development of new materials, for example, through the use of nanotechnology.

**Keywords:** adsorption; activated carbon; filtration; indoor air quality; ionization; microalgae; nature-based solutions; photocatalytic oxidation; UV light disinfection; plants

### **1. Introduction**

Increasing urbanization and modern lifestyles have contributed to humans spending an increasing amount of time inside buildings (e.g., at home, in offices, theaters, restaurants, stores, etc.), where they are exposed to indoor air pollutants [1]. Owing to the associated link between air quality and health, the WHO has recognized air pollution as one of the greatest environmental threats to human health [2]. To date, a considerable emphasis has been placed on reducing individual exposure to indoor air pollutants, making it necessary to analyze indoor sources and the possibility of reducing emissions from such sources. Thus, indoor air quality (IAQ), in all spaces where humans live and work, has become an

**Citation:** Mata, T.M.; Martins, A.A.; Calheiros, C.S.C.; Villanueva, F.; Alonso-Cuevilla, N.P.; Gabriel, M.F.; Silva, G.V. Indoor Air Quality: A Review of Cleaning Technologies. *Environments* **2022**, *9*, 118. https:// doi.org/10.3390/environments9090118

Academic Editors: Alejandro Moreno Rangel, Ashok Kumar, M Amirul I Khan and Michał Piasecki

Received: 29 July 2022 Accepted: 20 August 2022 Published: 7 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

issue of utmost importance and a significant determinant of human health and well-being. Several scientific studies have shown a direct relationship between improved air quality and positive impacts on human health [3–5]. The impact of indoor air pollutants on human health can be experienced in both the short and long term. Poor air quality results in unwanted health conditions and, in the worst-case scenarios, can lead to death [6].

Although the atmospheric composition, in terms of its main constituents (oxygen and nitrogen), is essentially the same indoors and outdoors, the types and amounts of indoor air pollutants differ from those found outdoors. Indoor air may contain a variety of contaminants, including particulate matter, tobacco smoke, radon, biological contaminants (e.g., mold, bacteria, fungi, dust mites, spores and pollen) and more than 400 organic and inorganic chemical compounds, with associated health effects [7,8]. Additionally, indoor air pollutants can reach concentrations of up to 10 times their levels in outdoor air, regardless of the building location [8]. Such pollutants are emitted by indoor activities (e.g., cooking and cleaning), products or materials (e.g., in furnishings and structures), to which other contaminants are added from outdoors that can penetrate indoors [9].

The concentration of pollutants in indoor air depends not only on indoor materials and activities but also on external factors [10]. Regardless of the insulation degree, even in naturally ventilated or mechanically conditioned and ventilated spaces, the internal atmosphere is an extension of the external atmosphere, i.e., the outdoor air quality directly influences the indoor air quality [11]. Non-reactive pollutants, such as carbon monoxide (CO), can penetrate the indoor environment and add to the indoor CO from unvented gas burners, defective cooking and heating devices, fireplaces, tobacco smoke and vehicle gases from attached garages [1,12]. On the other hand, reactive pollutants, such as sulfur dioxide (SO2) and ozone (O3), typically originating outdoors, quickly deplete after entering the indoor environment [11]. Carbon dioxide is considered an indicator of air quality in non-industrial indoor environments, such as homes, schools and offices and it is related to the presence of humans indoors and to human metabolism. It is also an indicator of the presence of other pollutants [13]. Although CO2 at low concentrations has few or no toxicological effects on humans, at higher concentrations, it has direct health consequences. At concentrations higher than 5%, CO2 causes the development of hypercapnia and respiratory acidosis, and at concentrations higher than 10%, it may cause convulsions, coma and death [14].

Humans are mostly exposed to air pollutants indoors via numerous sources, such as outgassing from furniture, floors, wall coverings, paints, glues, waxes, polishes, cleaning products, personal care products, tobacco smoke, heating appliances, cooking activities, etc. The indoor concentration of air pollutants can be affected by outdoor pollutant levels, as well as by other factors, such as door and window openings, air exchange rates, house age and size, and building renovations [15]. Cleaning agents and personal care products are common sources of volatile organic compounds (VOCs) and semi-volatile organic compounds (SVOCs), which are partially oxidized and condensed and therefore turn into fine particles. New furniture commonly emits formaldehyde [16]. Gas appliances such as stoves, boilers, smokers and cookers are important sources of indoor NO, NO2, PM2.5 and polycyclic aromatic hydrocarbons (PAHs) [17]. Ultrafine particles with a diameter between 5.6 and 560 nm are commonly detected in indoor air [18]. However, PM10 and PM2.5 particles with a diameter of 10 and 2.5 μm or less, respectively, are more frequently detected indoors [18]. Laser printers emit ultra-fine particles, siloxanes and long-chained alkanes (C21–C45), and 3D printers are a source of nanoparticles [16]. The use of incense sticks and candles [19], toasting, frying, baking, open chimneys and older wood stoves are responsible for PM2.5-10 emissions [16]. Biological pollutants are essentially composed of mites, hair, bacteria, molds, fungi, spores, endotoxins, mycotoxins and other types of living organisms with highly variable and complex characteristics [8,18].

Ventilation to dilute the indoor air contaminants is among the most important passive methods to improve indoor air quality in most buildings [12,20]. Natural ventilation can be achieved simply by opening windows. In modern and well-insulated buildings, it is also common to find mechanical ventilation systems, such as heat recovery ventilation (HRV) and energy recovery ventilation (ERV) systems. These systems continuously remove stale indoor air and replace it with fresh air from outdoors. However, if the outdoor air is more polluted or in certain situations where ventilation is not possible, other methods and/or air purification systems are required [21].

#### *1.1. Scope and Objectives*

Possible methods for air purification (Figure 1), which are reviewed and discussed in this work, include the physicochemical technologies (e.g., filtration, adsorption, ionization, UV-photocatalytic oxidation, ultraviolet disinfection) and the biological technologies (e.g., plant purification methods and microalgae-based methods).

**Figure 1.** Common sources and pollutants in indoor air and some possible technologies to mitigate them (authors' own creation).

Generally, the conventional physicochemical technologies for capturing air pollutants, although effective in the short term, present some disadvantages in the longer term. Most of these methods cannot simultaneously remove all the major gaseous pollutants, some are unsafe due to the emission of ozone, all require regular and expensive maintenance, may have high energy consumption and generate secondary pollutants and waste, and have high installation costs. For example, filters and adsorbents are rapidly saturated and require regeneration or replacement to maintain the capture efficiency and prevent the growth of microorganisms in the organic matter retained in the filter material [10]. Air ionizers are limited due to low removal efficiency and production of harmful secondary products, such as O3, NOx, and VOC oxidation intermediates [22]. After 3 to 6 months of use, adsorbents suffer from VOC saturation [23]. In addition to the saturation effects, some systems present ozone generation, reducing their practical application for air cleaning [24]. In addition, the high capital and operating costs of some of these technologies make them inaccessible to most consumers.

On the other hand, to maintain an acceptable IAQ, biotechnological solutions can offer a realistic alternative to engineering solutions. Indeed, in many situations, biological processes are the most cost-effective technology for treating VOCs and odors at low concentrations, below about 5 g m−<sup>3</sup> [25]. This is based on the ability of microorganisms to convert VOCs into CO2, water and biomass under ambient conditions of temperature and

pressure. Biological methods typically generate fewer secondary pollutants and are less energy intensive, although some significant challenges remain. Thus, much of the scientific research developed recently is based on biological methods, highlighting nature-based solutions as promising alternatives [7,26]. These are based on existing solutions in nature to respond to human needs, following the principle of biomimetics [24], theorized in 1997 by Janine Benyus [27]. Numerous experimental studies have demonstrated the potential of the passive use of potted plants for a significant improvement in IAQ, with around 200 plant species tested so far for their VOC removal capacity, all with positive results. However, there has been little subsequent investigation to test the effectiveness of these potted plant passive systems in real environments and with extremely low airborne concentrations of VOCs [24].

A combination of different methods is also possible for better air purification efficiency. For example, promising results have been obtained with an air filtration system composed of plants and activated carbon [28]. Additionally, Salama and Zafar [29] determined the effectiveness of using nanotechnology, combined with plants, for the purification of ambient air, by using Saudi myrtle plants (*Myrtus communis)* treated with TiO2 nanoparticles. The results showed a significant reduction in the air pollutants concentration (from 10% to 98%), including formaldehyde, NO2, SO2, CO and total volatile organic compounds (TVOCs).

The review organized by Szczotko et al. [30] focused on reducing microbial air contamination by using selected types of indoor air purifiers, showing that according to a wide range of articles on the topic, the actual effectiveness of the selected air purifiers is significantly lower, when applied to real conditions, than the values declared by manufacturers in their marketing materials and technical specifications. This is undoubtedly a major disadvantage of air purifiers offered to consumers.

Indoor air cleaning technologies can play an important role in improving indoor air quality. However, their effective application requires an adequate technical and environmental analysis of the real conditions for their use, as well as a characterization of the main pollutants existing in a given indoor environment. It may be necessary to use different air cleaning technologies simultaneously to more effectively maintain good indoor air quality. Hence, this work reviews the key indoor air treatment technologies, their target pollutants, main advantages and limitations.

#### *1.2. Method*

This review focuses on various air treatment technologies that can be used to control different types of contaminants, which can be physical (particulate matter), chemical (VOCs, NO2, ammonia, etc.) or biological (bacteria, viruses, fungi, etc.). The technologies have been divided in two major groups: the physicochemical technologies and the biological technologies. Within the physicochemical technologies, this review covers filtration (mechanical and electronic), adsorption, UV photocatalytic oxidation, ultraviolet disinfection and ionization. In the case of biological technologies, plant purification methods and microalgae methods are covered.

This review presents some of the most relevant work in each technology, discussing the pros and cons of each one. Since it would not be possible to include all the studies published in recent years in this subject, a selection was made in order to give an overview of the diversity of research taking place around the world. Studies presenting a mix of technologies were also selected, as there is an increasing tendency to seek synergies between different methods in order to increase the range of pollutants covered. Studies with interesting technologies that can be applied in building materials, with the aim of degrading indoor air pollutants, were also reviewed and included in this article.

#### **2. Physicochemical Technologies**

#### *2.1. Filtration*

Two types of filtration technologies for air cleaning are commonly used to remove particles: mechanical filtration and electronic filtration; the latter also includes electronic air cleaners (e.g., ionizers and electrostatic precipitators).

#### 2.1.1. Mechanical Filtration

Mechanical filtration is the most used air cleaning technology for particulate matter (PM) and it can be used even for removing respiratory droplets [31,32]. Mechanical filters use media with porous structures that contain fibers or stretched membrane material in a variety of fiber sizes, densities and media expansion configurations to remove particles from air streams. Some of the particles in the air that enter a filter bind to the medium and are removed from the air as it passes through the filter. Removal mainly occurs by impaction, interception and Brownian motion/diffusion, depending on the particle size. Some filters have a static electrical charge applied to the medium to increase particulate removal [33].

The main object of filtration is the removal of PM. There is a high variety of filter types, with a classification according to its efficiency to retain PM [34]. There are standards to classify filters, such as ISO 16890 [35], EN 1822 [36] in Europe and ANSI/ASHRAE 52.2 [37] in the USA. The ISO 16890 [35] classifies the filters used in general ventilation in four groups, based on the filter efficiency for a particle size: coarse, ePM10, ePM2.5 and ePM1. To belong to each category, a filter must be capable of capturing at least 50% of the particles in that size range. The filters that capture less than 50% of PM10 (particles with diameters that are generally 10 μm and smaller) will belong to the coarse dust group. Both parameters, percentage of filtration and size, are equally relevant. For example, if the filters ePM1 50% and ePM2.5 50% are compared, the first retains 50% of particles between 0.3 μm and 1 μm and the second 50% of particles between 0.3 μm and 10 μm. There are filters with high filtration efficiencies (EPA—Efficient Particulate Air, HEPA—High Efficient Particulate Air and ULPA—Ultra Low Penetration Air), which are used in environments that require a high degree of air asepsis, being classified by EN 1822 [36]. HEPA and ULPA filters can also be used for cleaning ultrafine or nanoparticles (<0.1 μm), such as bacteria and viruses that can even pass the membrane of our lung cells [38].

A very common particulate filter, mostly used in portable air cleaners, is the HEPA filter, which means that the single-pass efficiency of the filter media is ≥99.75% if the filter is classified as H13 and ≥99.97% for filters classified as H14. These filtration characteristics are set according to the efficiency of the 0.3 μm particles, which is the most penetrating particle size (MMPS).

These filters are normally part of a central heating and ventilation system, or part of air purification equipment, usually portable. In the first case, the filters act by filtering outdoor air, although they can filter indoor air if there is recirculation, and in the second case, the Portable Air Cleaners (PAC) exclusively purify indoor air, without air renovation. Real-time sensing has been applied to these devices to optimize their performance. These sensors monitor the ambient conditions online (temperature, humidity, concentrations of key pollutants) and activate the reduction units according to the needs of the occupants and their activities, with resulting energy savings [10]. However, PACs have different modes of operation in which the clean air flow provided is different (CADR—Clean Air Delivery Rate).

During the COVID-19 pandemic, PACs were recommended as a supplementary measure for ventilation or for those spaces in which natural or mechanical ventilation was not available, or was insufficient, in order to reduce the risk of transmission [39,40]. Experimental studies provided evidence for portable HEPA purifiers' potential to eliminate airborne SARS-CoV-2 [41]. Then, an air cleaner can be installed to remove particles potentially carrying viral particles from indoor air.

The air cleaner's effectiveness in reducing particles is defined by their CADR, typically expressed in m<sup>3</sup> per hour. Ventilation is defined parametrically as Air Change Rate (ACR) with its unit Air Changes per Hour (ACH, h−1). The air change rate equivalent to the air cleaner's CADR is determined as follows:

During the pandemic it was recommended that 5–6 air changes per hour in classrooms should be achieved with ventilation and/or using PAC [42]. Therefore, the clean air flow required for a room, CADR, provided by a PAC (or several) can be calculated from Equation (1).

$$\text{ACR (clearing)} = \text{CADR/Volume of air in the room} \tag{1}$$

Portable air cleaners have been widely used in different studies in order to assess the removal of aerosols in indoor air.

Cox et al. [43] performed a study in a total of 46 homes (43 of the homes completed the entire 3-month study, and an additional 3 homes completed a portion (≤1-month)) to assess the effectiveness of a Portable Air Cleaner (PAC) with a HEPA filter in reducing indoor concentrations of traffic-related and other aerosols, including black carbon (BC), PM2.5, ultraviolet absorbing particulate matter (UVPM) (a marker of tobacco smoke), and fungal spores. The PAC selected was a Whirlpool Whispure (Model AP51030K, Austin, TX) with a HEPA filter designed to capture 99.97% of 0.3 μm particles and with a CADR between 360 and 576 m3/h (minimum to maximum speed). The CADR was selected according to the room size. The results showed that Portable Air Cleaners with HEPA filters could significantly reduce traffic-related and other aerosols in different residential environments, such as tobacco smoke, PM2.5 and fungal spores.

Dubey et al. [44] have studied the effectiveness in reducing the concentration of different sized particulate matter (PM) and ions of two types of air purifiers equipped with HEPA filters in general indoor air and the presence of an external source (candles and incense). The first air purifier (API) comprised an anti-dust filter, activated carbon filter, active HEPA filter, electrostatic filter, vita ions, cold catalyst filter with programmable control panel, sleep mode, timer function, independent air ducts and a CADR of 120 m3/h, while the second air purifier (APII) was equipped with six sense technology, a humidifier and a filter replacement indicator, along with filters viz. pre-dust filter, activated carbon filter, HEPA filter, nanocaptur filter; UV lamps, in addition to an ionizer function and a CADR of 150 m3/h. The results showed that both devices reduced PM levels that varied from 12 to 53% for API and 37–68% for APII, depending on the scenario studied. In addition, both air purifiers reduced ions concentration significantly, while the concentration of some of the ions increased after the application of the air purifier. The increase in the mass concentration of ions after the application of the air purifier may be due to that air purifiers release ions continuously to purify the air. Overall, the study recommends the use of air purifiers with mechanical filters (HEPA) instead of those that release ions for air purification.

Blocken et al. [45] assessed aerosol levels in a gymnasium, and demonstrated that the existing mechanical ventilation systems alone are not efficient to decrease the levels. An air cleaning device (AC) alone with ACH = 1.39 h−<sup>1</sup> had a similar effect as ventilation alone. Simplified mathematical models were engaged to provide further insight into ventilation, AC and deposition. It was shown that combining the above-mentioned ventilation and AC can reduce aerosol particle concentrations with 80 to 90%, depending on aerosol size. It should be stressed that it remains imperative that ventilation should be maintained at (at least) the minimum flow rates required by building codes, because many ACs do not remove gasses such as CO2.

Lee et al. [46] assessed the effectiveness of aerosol filtration by portable air cleaning devices with HEPA filters used in addition to a standard building heating ventilation and air conditioning (HVAC) system. The test rooms, including a single-bed hospital room, were filled with test aerosols to simulate aerosol movement. Aerosol counts were measured over time with various portable air cleaning devices and room ventilation systems to quantify the overall aerosol clearance rate. It was found that PAC devices were very effective for the removal of aerosols. The aerosols were cleared five times faster in a small control room with PAC devices than in the room with HVAC alone. The single-bed hospital room (37 m3) had an excellent ventilation rate (ACH = 14) provided by the HVAC system and cleared the aerosols in 20 min. However, with the addition of two air cleaning devices (CADR = 1458 m3/h, ACH = 39 h−1), the clearance time was three times faster.

Finally, Cheek et al. [47] carried out a systematic literature review to examine the impacts of portable air purification on indoor air quality (PM2.5) and health, focusing on adults and children in indoor environments (homes, schools and offices). These authors [47] report positive long-term impacts with reduced PM2.5 concentrations. The current evidence demonstrates that using a PAC results in short-term reductions in PM2.5 in the indoor environment, which has the potential to offer health benefits.

It is consensual that filters can efficiently remove particles but are not effective for organic and inorganic chemical pollutants. A solution seems to be a combination of particulate filter and other air purification systems, or the development of filters with other properties, such as chemisorption. Regarding gas purification, the most effective and commonly used purification method is adsorption, with activated carbon filtration being the most prevalent approach due to their high surface area and high storage capacity, although other technologies are available [34,48].

Swamy [49] assessed the use of filters prepared with different percentages of NaOH doping on calcium silicate granules in order to reduce CO2 concentration levels. The results showed an over 40% reduction in CO2 concentrations with this purification system. Fresh air intake, to maintain the desired ventilation rates, has been reduced to over 50%, further reducing the heat load. Nanofibers can also be modified with adsorbent nanomaterials effective in the adsorption of VOCs [50]. Buyukada-Kesici et al. [51] have used cellulose nanocrystals and polyamide 6 to develop a porous electrospun material, which can adsorb around 50% of toluene in 45 min of trials, maintaining the same removal efficiency of conventional adsorbents in the form of powder or particles.

Protein-based nanofibers have also gained prominence in the air filtration field. The study performed by Kadam et al. [52] showed that Gelatin/β-CD composite nanofibers presented excellent adsorption of xylene (287 mg/g), benzene (242 mg/g) and formaldehyde (0.75 mg/g). The gelatin/β-CD biomaterial-based nanofibers filtered solid/liquid aerosols and gaseous pollutants simultaneously at a lower base weight with low air resistance.

Another interesting study was presented by Liu et al. [34], who developed a transparent polyacrylonitrile air filter to protect indoor air quality through natural passive ventilation windows. They obtained highly effective air filters, ~90% transparency, with a removal rate greater than 95% for PM2.5, under extremely hazardous PM2.5 concentrations (>250 μg m−3). Its material is an excellent alternative for perspective windows, especially in large urban centers.

The studies presented are a small portion of all studies around new materials and new combinations involving filtration. With the advent of nanotechnology, there is enormous research potential in this area.

Concerning usual filtration, the traditional fibrous filters have numerous advantages, such as high removal efficiency, low initial cost and simple structure [34].

The main disadvantages are high pressure drop, high maintenance costs and filter colonization [34]. Another negative point is that filters are not effective for pollutants other than PM. The filtration efficiency of general fiber filters is directly proportional to the air pressure drop. High pressure drop means additional energy consumption and increased operating costs. Filter maintenance is often neglected during a system operating period. Filters must be replaced or cleaned regularly, otherwise they could become sources of pollution. The longer the filter operating time, the greater the possibility of potential pollution, by re-emission of VOCs or dissemination of fungi or bacteria by aerosols diffusion [53].

#### 2.1.2. Electronic Filtration

There are two types of electronic filters for particles removal: electrostatic precipitators and ion generator or ionizers. Electronic filters include a wide variety of electrically connected air-cleaning devices that are designed to remove particles from airstreams. Removal, typically occurs by electrically charging particles, using corona wires or through the generation of ions (e.g., using pin ionizers), and by collecting the particles on oppositely charged deposition plates (precipitators), or by the particles' enhanced removal to a conventional media filter, or to room surfaces [33]. According to Bliss [54], the efficiency of electrostatic filters for a particle range of 0.3–6 μm is over 90%, while this value oscillates between 75 and 95% for ion generators. Some studies have found adverse health effects when using electrostatic precipitators such as the modification of cardiorespiratory function associated with the production of negative air ions, which may outweigh the potential benefits from PM reductions (e.g., Liu et al. [55]). In addition, the filters can become clogged over time, and so require regular cleaning. Generally speaking, electronic filters can generate hazardous charged particles [56] or new pollutants such as ozone, ultrafine particles and other compounds derived from VOCs ionization [57].

#### *2.2. Adsorption*

Adsorption consists of capturing air pollutants on the surface of an adsorbent material. It has been successfully applied for retaining both volatile organic compounds and inorganic pollutants on adsorbents, such as activated carbon, zeolites, silica gel, activated alumina, mineral clay and some polymers. The most commonly used are activated carbon and hydrophobic zeolites, due to their high surface area and adsorption capacity [18]. Activated carbon has high porosity and is a non-polar adsorbent. It can be produced from agricultural wastes such as sugarcane bagasse, apple pomace or coconut shell for more cost-effective pollutants removal [58]. Due to its microporous structure and large surface area, activated carbon is able to remove up to 100 mg m−<sup>3</sup> of VOCs [59], although medium and high molecular weight volatiles are better adsorbed on activated carbon than low molecular weight volatiles.

Adsorbents can be easily incorporated into building materials and/or integrated into interior surfaces to remove air pollutants with no additional energy input and minimal byproduct formation; for this reason, they are classified as passive removal materials (PRMs). Passive removal materials enable ozone control, for example, in susceptible populations with health benefits, creating healthy indoor environments [60].

Hybrid technologies of adsorption combined with other methods have been proposed for pollutants removal from indoor air. For example, Jo and Yang [61] investigated the technical feasibility of an hybrid system composed of activated carbon and photocatalytic oxidation for controlling indoor air levels of BTEX (benzene, toluene, ethylbenzene, and xylenes) at low concentrations (of 0.1–1 ppmv). These authors [61] concluded that this hybrid system can enhance control efficiency of BTEX in indoor air levels with higher removal efficiencies (close to 100%) compared to using activated carbon alone (with removal efficiencies close to or higher than 90%), with a negligible addition to indoor CO levels from the photocatalytic oxidation process. However, some drawbacks of this technology include the decrease in removal efficiency with increasing relative humidity due to the capillary condensation of water vapor inside the activated carbon that blocks the adsorption sites. Additionally, the adsorption capacity of activated carbon decreases with increasing inlet concentrations. Furthermore, these authors [61] reported that a temperature of 300 ◦C is necessary to obtain significant desorption yields (75–95%) and the adsorbents need to be regularly replaced in order to avoid re-emission of already adsorbed compounds.

Ao and Lee [62] examined the effect of TiO2 immobilized on activated carbon under different humidity levels for the removal of air pollutants from indoor air at parts-perbillion (ppb) levels. In this research [62], NO (200 ppb), BTEX (20 ppb) and SO2 (200 ppb) were used as target pollutants. Different resident times and relative humidity levels were tested to investigate their mutual effect on TiO2 and TiO2 immobilized on activated carbon. The results showed that the effect of TiO2/AC is more significant with decreasing residence time and increasing levels of humidity. At a longer residence time, no significant pollutant removal difference is observed between TiO2 and TiO2 immobilized on activated carbon. At high humidity levels, the inhibition effect of water vapor is more significant compared to the presence of other pollutants, although it is still practically feasible to remove multiple pollutants under high humidity levels [62]. To further evaluate the performance for indoor air purification of a TiO2 immobilized on activated carbon (TiO2/AC) filter, Ao and Lee [63] examined it installed in a commercially available air cleaner. The authors tested it inside an environmental chamber, using NO and toluene as target pollutants. The original commercial air cleaner setting (AC + HEPA) showed no NO and little toluene removal. The TiO2 filter removed 83.2% of NO but generated 12.9% of NO2. Using TiO2/AC, the NO removal efficiency increased to 97% and the generation of NO2 decreased to 1.6%. The authors concluded that the TiO2/AC filter not only increases the pollutants removal efficiency, but also reduces the release of intermediate compounds by the system.

Sidheswaran et al. [64] demonstrated the potential environmental and energy benefits of using activated carbon fiber filters for air cleaning in HVAC (heating, ventilation and air conditioning) systems. These filters are prepared from fabric precursors and have a very high specific BET surface area, typically higher than 1000 m2 g−<sup>1</sup> and low pressure drop, making them ideal for use in HVAC systems for VOC removal. In order to measure the removal efficiency of this activated carbon filter, the authors exposed it to a VOCs mixture of model pollutants, with concentrations in the range 20–30 ppbv (parts per billion by volume), composed of toluene, benzene, o-xylene, 1-butanol, limonene, undecane and formaldehyde at 29 ◦C and 30% relative humidity. The experiments showed the consistent removal (retaining) efficiencies of 70–80% for most VOCs [64].

Cheng et al. [65] evaluated the antibacterial and regenerated characteristics of a zeolite impregnated with metallic silver (Ag-Z) for removing bioaerosols (bacteria and fungi) in indoor environments, showing a 95% removal efficiency after 120 min of operation. These authors considered the 1 wt% Ag-Z to be more cost-effective, with an antibacterial efficiency near 90% in less than 60 min and an excellent repeated use performance up to nine times.

Adsorption materials can not only act as a sink for airborne pollutants, but also for excess moisture through adsorption. For example, a medium density fiberboard modified with walnut shell was investigated to regulate relative humidity, toluene, limonene, dodecane and formaldehyde [66]. A negative feature of the adsorption technology is the possibility of the deposition and development of airborne bacteria on the adsorbent surface due to the high biocompatibility of these materials [18]. Additionally, adsorption technology does not treat or destroy contaminants, but they are simply transferred from one phase to another, producing a hazardous solid waste that must be further treated and/or disposed of correctly.

#### *2.3. UV-Photocatalytic Oxidation*

UV-Photocatalytic Oxidation (PCO) is a very interesting air cleaning technology that has been the subject of much investigation in recent years. PCO is defined [33] as a lightmediated, redox reaction of gases and biological particles adsorbed on the surface of a solid pure or doped metal oxide semiconductor material or photocatalyst. The most common photocatalyst is TiO2 (titanium dioxide), while zinc oxide (ZnO), tungsten trioxide (WO3), zirconium dioxide (ZrO2), cadmium sulfide (CdS), and iron (III) (Fe(III)-doped TiO2), among others, are also used. To improve the efficiency of each elementary step, various strategies have been used to modify the physicochemical properties of nanomaterials. For example, element doping (metal or nonmetal) to induce impurity states in wide bandgap semiconductors has been used to extend the light absorption range, while plasmonic noble metal (Ag and Au) deposition is a popular way to enhance light absorption and inhibit photoinduced charge recombination. Several examples of these nanomaterials can be found in the review organized by Cao et al. [67]. The photocatalyst generates oxygen species (or reactive oxygen species) that remain surface-bound when exposed to light

of particular wavelengths in the ultraviolet (UV) range. The oxygen species are highly reactive with adsorbed gases and biological particles. A variety of UV light sources can be used in PCO, including black lights (UV-A: long-wave; 400 to 315 nm), germicidal lamps (UV-C: short-wave; 280 to 200 nm), and lamps that generate ozone (vacuum UV [UV-V]: under 200 nm). Under reaction conditions allowing for deep oxidation (referred to as mineralization), carbon, hydrogen, and oxygen atoms in the reacting species will be converted completely via chemical reaction to water vapor and carbon dioxide. However, the oxidizing process can be incomplete, and origin reaction byproducts can be more toxic or harmful than the original constituents, as is the case for formaldehyde. Several studies [68–70] showed the generation of formaldehyde and acetaldehyde from the partial oxidation of ubiquitous VOCs such as alcohols. To minimize these problems, mixed technologies have emerged, which, for example, combine the use of PCO with filters. The combination of two techniques [71], or the development of filters with PCO properties, was found [72–74].

The research in this area is vast, and it would be impossible to list all the different nanomaterials developed to increase the PCO efficiency. Notably, in 2020, more than 9000 journal papers were published on the photocatalytic degradation of several pollutants [74]. Several review articles can be found [75] where an overview of the most recent studies for applications in indoor air cleaning is presented. In order to restrict the scope, this article presents some of the technologies ready (or near) to be used by consumers to clean indoor environments.

Kaushik and Dhau [71] presented a new air purifier (Molekule) that can trap pollutants and efficiently destroy harmful bio-active compounds using the photoelectrochemical oxidation (PECO) approach. The PECO technology is designed to drastically lower the atmospheric oxidation energy barrier using a specially designed catalyst coated on an air-cleaning filter. The Molekules technology is based on the synergy of nano-assisted efficient photoelectrochemical oxidation and filter membrane and have shown interesting performances. For example, for airborne viruses such as MS2 bacteriophage (a proxy virus for SARS-CoV-2, which travels in tiny droplets that can linger for hours before settling on surfaces) with 99.99% of removal in 30 min, and for formaldehyde with 81% of removal in 8 h.

Weon et al. [73] developed a new material: a TiO2 Nanotubes Photocatalyst filter for Volatile Organic Compounds Removal. This filter was applied in a Commercial Indoor Air Cleaner, achieving an average VOCs removal efficiency of 72% (in 30 min of operation) in one 8 m3 test chamber, showing promising results.

One of the most interesting applications is the incorporation of well-known and efficient photocatalytic materials into construction materials suitable for indoor applications to degrade priority air pollutants indoors and to inactivate various pathogens. Examples of such applications in confined spaces are TiO2-based paints [76–78], roofing tiles [79], paper sheets [80], or textiles [81]. There is also the so-called indoor passive panel technology (IPPT), which includes as typical materials modified gypsum board, acoustic ceiling tiles, ceramic tiles, wallpaper and other coatings and pre-coated products relying on either sorptive or photocatalytic oxidation (PCO) processes [82]. Shayegan et al. [83] present a review focused on the application of passive removal materials to improve indoor air quality, which have recently gained interest due to their ability to remove pollutants without additional energy consumption and lower amounts of byproducts formation. Two types of materials are the object of this review: photocatalytic oxidation-based materials and sorptive-based materials. These authors [83] concluded that further scientific evaluation is necessary to assess the application of these passive materials in the indoor environment, and to better understand their impact on the IAQ.

Maggos et al. [77] developed an innovative paint material using a Mn-doped TiO2 photocatalyst, which exhibits intense photocatalytic activity under direct and diffused visible light for the degradation of air pollutants, suitable for indoor use. A laboratory and a real scale study were performed using the above innovative photo-paint. The lab test was performed in a special design photo-reactor, while the real scale test was performed in a military's medical building. Nitrogen Oxide (NO) and Toluene concentration was monitored between "reference" rooms (without photo paint) and "green" rooms (with photo-paint) in order to estimate the photocatalytic efficiency of the photo-paint to degrade the above pollutants. The results of the study showed a decrease of up to 60% and 16% for NO and toluene, respectively, under lab scale tests, while an improvement in air quality of up to 19% and 5% under real world conditions was achieved.

In the study of Demeester et al. [79] through TiO2 incorporation in roofing tiles, toluene was removed from air. At ambient conditions (T = 25 ◦C and RH = 47%) and toluene concentrations between 17 and 35 ppbv, toluene removal efficiencies between 23% and 63% were achieved. Other interesting data suggest that washing the TiO2 containing building material with deionized water and simulating rainfall could partially (by a factor 1.3) regenerate photocatalyst activity.

Dong et al. [81] showed that the combination of TiO2-loaded cotton fabrics, as wall cloth or curtains used in house rooms, produced using padding or coating methods with UV irradiation of 365 nm wavelengths can effectively eliminate gaseous ammonia in a photocatalytic reactor. The decomposition efficiency of ammonia was much affected by the dosage of TiO2 aqueous dispersion, initial ammonia concentration, relative humidity and gas flow rate.

Zuraimi et al. [82] evaluated the performance of 3 photocatalytic oxidation (PCO) based materials in controlled test chamber experiments with toluene, in order to determine removal rates, and ozone and carbonyl by-product formations. Toluene removal was found to be dependent on the type of light used. Only two PCO IPPTs are capable of performing under visible light. All PCO-based IPPTs generate ozone and carbonyls as byproducts, releasing up to 1.0 mg/h and 3.2 mg/h of ozone and formaldehyde, respectively. This is a concern, as exposures to ozone and formaldehyde are known to be associated with negative health outcomes.

From the exposed, it can be concluded that PCO is a promising cleaning technology, but studies show that there are aspects that need to be addressed before UV-photocatalytic oxidation can be used safely in buildings. The design of a photocatalytic air purification system is complex, and it is dependent on a wide variety of factors, including the intensity of the light falling on the catalyst, chemical makeup and pollutants concentration, the air flow rate through the device, moisture levels in the air, properties of the specific catalyst used, pollutant residence time, and how the device itself is configured [71].

However, the technology has several advantages:


The main disadvantage is the incomplete oxidation, which produces reaction byproducts that can be more toxic or harmful than the original constituents (e.g., formaldehyde). The catalysts can also be contaminated (poisoned) by airborne reagents and/or products of oxidation, which results in reduced or total efficiency failure of the process. In the cases where a lamp us utilized instead of solar light, the list of disadvantages includes the lamp energy consumption, lamp replacement costs, and the likelihood of ozone generation depending on the lamp source employed (e.g., UV-V lamps ~185 nm produces ozone) [33].

#### *2.4. UV Light Technology-Based Disinfection Systems*

As shown in Figure 2, there are four types of ultraviolet (UV) radiation, defined according to their wavelength range in nm: vacuum UV (100–200), UVC (200–280), UVB (280–320) and UVA (320–400). Although all UV wavelengths are described to cause some photochemical effects, wavelengths within the UVC or UVGI range, specifically at 253.7 nm, are particularly harmful to cells, because it is absorbed by proteins, RNA and DNA. This process induces molecular breaks of the simple covalent bonds C, H, O and N of the nucleotide chain, resulting in irreversible molecular damage that leads to the inactivation of all types of microorganisms [84–86].

During the last century, several authors have studied the disinfection properties of ultraviolet (UVGI) light, at a first stage for water disinfection and, afterwards, for both air and surface disinfection purposes. In addition, throughout the years, efforts have been devoted to identify the configuration of an UVGI system that is most effective for each type of building ventilation system, as well as to disclose data on the effectiveness of UV in decreasing levels of microorganisms in indoor air and, consequently, the risk of developing health problems, such as respiratory infections and allergies [87–90]. The bestknown application of UVGI disinfection in air is in buildings belonging to the healthcare sector [91–94]. However, a growing number of studies have provided robust evidence on the effectiveness of this air cleaning technology in several indoor environments, including offices, commercial areas, schools and universities [90,95,96]. For instance, although UVC systems have been used to disinfect hospital environments since 1936 [97], this approach was only formally recognized as being effective against airborne bio-pollutants by official authorities at the Center for Disease Control and Prevention (CDC) in 2003 [98]. Since then, UV light-based systems have been recommended by the CDC (2003) as an adjunct to routine chemical cleaning, as traditional cleaning and disinfection protocols can be insufficient for insuring proper disinfection in some contaminated areas [99]. Recently, ASHRAE and EPA recognized the importance of employing air cleaning solutions that include both filtration and UVC-based systems to prevent the spread of infectious diseases in indoor environments [100,101].

Air disinfection can be performed using UVC in several modalities, including irradiation in forced air or stand-alone systems, upper room, irradiation of an entire room (when the room is empty) and irradiation of the air circulating through a heating, ventilation, and air conditioning (HVAC) system, both in-duct and in the coils [100]. From the existing studies aiming at investigating the effectiveness of the use of UVGI for air treatment, it was particularly demonstrated that the use of in-duct in central ventilation systems, as well as upper-room and stand-alone systems, results in an effective reduction in the levels of microorganisms and endotoxins [90,101–107]. UVC systems in ducts are known to be cheaper and demand a lower energy consumption, as they are able to disinfect air and surfaces, while upper and stand-alone room devices only disinfect air [108].

Building ventilation systems are typically designed taking into consideration comfort needs. Even when ventilation systems are reconditioned to incorporate UV lamps there may be limitations related to air distribution and the removal of polluted air at the source [109]. Thus, based on the evidence, mainly for the indoor spaces with special needs of disinfection (e.g., hospitals) it is preferable to consider, in addition to filters, an UV disinfection technology, at an early stage of designing the ventilation system to serve the building. A very important criteria related to the choice/designing of the system to be installed is the type of lamp(s) to employ. Typically, low pressure (LP) mercury lamps, which radiate 95% of their energy at the wavelength 253.7 nm and are "monochromatic"; are described to have greater germicidal power than medium pressure (MP) polychromatic lamps, Xenon or UV LED [84]. The radiant energy flow of the systems should be determined, since systems, especially in ducts and stand-alone modalities, must be coated with material of high UV reflectance such as aluminum (not plastic) to reduce the number and size of lamps needed to be installed. Nevertheless, these conditions are difficult to standardize, given the geometry of the ducts and the eventual deterioration of reflectivity when internal surfaces are exposed to an airflow that carries can include variable levels of moisture and particles. This is why it is necessary to install lamps with a high UVC, preferably up to 30% of the rated power (e.g., Sanuvox technologies SL), as well as to correctly calculate and size each HVAC duct system, stand-alone, upper-room considering airflow, humidity, temperature, sizes, materials. In 1954, Harstad et al. [110] demonstrated that despite the installation of UV light in the air conditioner, there could still be airborne pollution from the growth of microorganisms in air conditioner components such as filters, cooling coils and duct surfaces [110]. In fact, one of the most widespread applications are UVC systems in cold batteries to destroy fungi, bacteria and endotoxins, avoiding their proliferation in the HVAC systems, achieving energy savings (average of 5–15%) and improving the efficiency of cooling transfer by reducing static pressure through the coil [90,108].

Each organism has a different sensitivity to UVGI light. Kowalski [84] and Malayeri [111] are some of the researchers who have collected data from several studies on the dose of UV needed to achieve the inactivation of bacteria, fungus, virus, protozoan and microalgae in vegetative forms and spores. Bedford [112] and Gates [113,114] were among the first to establish the UV doses needed for bacterial disinfection. Fulton and Coblentz [115] reported the UV doses for fungal inactivation, while data on the respective doses for viruses removal were first published by Rivers and Gates [116]. It is important to know the target pathogens to calculate the optimal UV dose based on their susceptibility constants, named K or z-value. This susceptibility, z-value, of each microorganism varies according to factors such as the pathogen biological structure, the conditions of environmental exposure and the distribution of the particles size, among others [84,105,117]. In fact, some environmental factors can influence the effectiveness of UV microbial inactivation. In this regard, there is evidence showing that airborne bacteria become more resistant to UV rays as relative humidity increases [84,105,117]. In addition, the use of an increased airflow rate passing through the UV system can result in a lower effectiveness of UV disinfection because microorganisms are exposed to UV rays for less time [84,105,106,118]. A low or a very high airflow temperature is also described to negatively affect the UV output, UV disinfection efficiency and susceptibility constants of microorganisms [84,105,106,117,118].

In addition to the selection of proper design, another aspect of utmost importance to consider is to ensure the proper sizing and operation of the systems. In fact, since the irradiance of the UV dose decreases with distance, to reach the same z-value, it can be necessary to operate the system with more time, in order to ensure effective disinfection [93]. Regarding the maintenance of the UV-based system, it is crucial to consider the nominal output potential of UVC lamps, which decreases over time. UVC lamps are classified into effective hours of UVC emission, and not at the end of the hours of electrical life. Many UVC lamps are designed to emit intensity levels at the end of their service life that are 50 to 85% or more than that measured in initial operation (after 100 h of burning time), although current models continue to emit blue visible light. Specification data from lamp manufacturers can verify depreciation over the useful life [119], the better the longer the lamp life, and the cost of spare parts and maintenance will be reduced (e.g., Sanuvox technologies SL lamps have a useful life of 17.000 h).

Importantly, in order to avoid any collateral health risks to the occupants, the potential release of toxic byproducts (such as ozone) by all kinds of UV systems should be carefully controlled. The devices must be tested by the manufacturer (and preferentially also by an independent third party laboratory) to ensure that the ozone concentration generated during operation respects the recommended maximum limits [100]. Another aspect to keep in mind is that there are limit values established for exposure to UVC light that cannot be exceeded. For systems that are closed (e.g., in ducts, cleaning of surface coils, autonomous systems), it is important to include a security system for opening doors that interrupts the operation of the lamps. In the case of upper room systems, these should be placed to an adequate distance from the ground that does not endanger people, because direct UVC light can induce damages to the eyes and skin [84,100,109].

#### *2.5. Ionization*

Bipolar ionization is generated when an alternating voltage (AC) source is applied to a special tube with two electrodes. This phenomenon can occur in nature, especially in mountain areas and waterfalls, where the production of positive and negative ions are reported to purify the air [120]. In the process of ionization, a neutral atom is given a positive/negative charge through the removal/addition of an electron, respectively. In bipolar ionization, positive (H+) and negative (O2−) ions are generated when water molecules are exposed to high-voltage electrodes. Although the mechanism associated with the biocidal effect of positive and negative ions have not been yet clearly established, the purposed mechanism involves the clustering of these ions around micro-organisms, resulting in the formation of OH radicals, which remove hydrogen, leading to the production of water vapor and to microbial inactivation [121]. Air ionization-based devices include those that generate only negative ions (i.e., unipolar ionizers) and those that generate positive and negative ions (i.e., bipolar ionizers). Bipolar ionization technologies, especially Corona Discharge and Needlepoint (NBPI), generally produce the same types of ions, which have the same theoretical mechanism of action when it comes to fighting pathogens, VOCs, and particles. However, a shared mechanism of action between the various air ionization technologies is not necessarily indicative of a shared method of ion creation [96]. Bipolar ionization technology has been around for decades, but the limited number of peer-reviewed studies in this topic makes it difficult to accurately support the effectiveness of this technology for air and surface disinfection purposes. Nevertheless, a growing body of recent evidence has presented the potential of using air ionization to decrease bacterial deposition on surfaces [122,123], inactivate airborne bacteria, viruses, and fungi [107,123–125] and to remove airborne particles and VOCs, being more effective when used in long-term applications [123,126].

Air ionization modules are often fitted directly into central air handling units to treat entire airflows. Modules can also be fitted into existing ductwork immediately downstream of central HVAC systems. Freestanding devices can also be placed into individual room spaces to meet immediate demands from internal sources. It is common to employ air ionization along with other technologies, such as air filtration [127], and some reports suggest that the efficiency of filtration increases when the ionizer is running [125]. Because the ions have a short service life of milliseconds, the distance from the ion generating equipment and the area to be treated must be estimated [127]. There are authors that recommend that the system should be installed vertically and deployed near both the inlet and return outlet for ensuring a better inactivation efficacy against airborne bacteria, especially in hospitals [128].

Since it has been seen that these systems can produce high levels of ozone and other organic compounds, similarly to UV-based systems, an aspect to consider is that the systems to be installed must have the certificate that accredits the non-generation of ozone [129]. In this regard, it has been described that when the energy potential produced through the ionization process is limited to 12 eV or less, ozone will not be produced because the oxygen has an ionization energy of 12.07 eV [120].

Some of the great advantages of NBPI systems are the elimination of VOCs, particles, odors and pathogens with reduced energy consumption and very low maintenance costs [123]. The main disadvantage is that while there is some evidence showing the efficacy of some of these approaches, the literature is still too scarce to draw robust conclusions-Thus, it is necessary to carry out more studies considering the implementation of air ionization-based technologies in real life situations and not only in controlled environments. Likewise, further work is needed to explore the potential capabilities that this technology has in combination with other systems to improve indoor air quality and reduce risks to human health.

#### **3. Biological Technologies**

#### *3.1. Plant Purification Methods*

Indoor greenery provides several benefits such as producing oxygen, generating humidity, pleasant aesthetical integration, passive acoustic insulation system [130], positive psychological effect on task performance, health, level of stress and comfort [131]. Besides that, the potential of plants for purifying and remediating the indoor atmospheric environment has long been identified (e.g., Wolverton [28]). Nevertheless, in the last twenty years, there has been a rising trend of more in-depth research on air purification mechanisms, technological solutions and methods [7,132,133]. Recently, in the context of the COVID-19 pandemic, the role of indoor plants in air purification and human health was looked at closely and considered as an alternative solution that could be used to reduce the viability of SARS-CoV-2 [134,135].

The purification methods mostly rely on phytoremediation, which is characterized by the use of plants to remove pollutants, or in this case from air, since it can be applied to water and soil, being based on a plant's ability to absorb, catabolize and degrade airborne pollutants, associated with their metabolic activities [130,132,133,136]. Regarding the plant microbiome, both the endobacteria and phyllobacteria may have an influence on the removal process to different extents, having in consideration the type of pollutant.

Interest associated with phytoremediation is high since it is considered an approach with low implementation cost and maintenance, compared with other technologies [137]. Moya et al. [130] provide a detailed description of the phytoremediation techniques. Additionally, Teiri et al. [138] revised different phytoremediation methods and the critical factors for the purification of indoor air, concluding that plants are able to efficiently remove different harmful contaminants from indoor air, including VOCs. This can be achieved through passive filtration using potted plants, or through active filtration using plant filters or green walls.

In general terms, active vegetation systems (in combination with mechanical systems) and passive vegetation systems (without additional energy requirements) can be considered for the present purpose [130,133]. Active systems increase the availability of polluting gases through the incorporation of mechanical ventilation devices; being associated with higher air-cleaning rates than passive systems. The passive ones are dependent on the diffusion of polluting gases, characterized by slower operation and lower concentrations [7].

Indoor plants are generally considered small shrubs and herbs that fit into the selected greenery system [132]. Plants used for indoor air phytoremediation are, in general, ornamental and limited to a small number of model species [133]. Following the recent review by Prigioniero et al. [133], the most widely considered species for this purpose is *Chlorophytum comosum* (Thunb.) Jacques, although other species are mentioned in the literature, often related with the type of pollutant to be removed [132,136,139]. An extensive list of indoor plants and their pollutant removal efficiency are mentioned in the literature [131,136]. Pollutant removal is plant specific and each part of the plant has different removal potential [131].

When using plant-based systems to promote indoor air quality, it is also important to consider the pollen load and allergenicity potential, such as flowering plants with strong fragrances [132]. Nevertheless, the main source of indoor pollen and fungi spores typically come from outdoors [140]. Plants also emit compounds, some of which are biologically active [141]. For example, plant volatile organic compounds (VOCs) are important in the ecosystem for chemical information transfer [132]. Gas exchange intensity and extent vary from night to day, and also depends on several factors such as temperature, humidity, light intensity, photosynthetic system, carbon dioxide concentration, and concentration of air pollutants [137,141].

The use of potted plants to promote air purification has been studied for a long time [28]. Although considered a passive approach, it has been investigated for the removal of harmful pollutants such NO2, with promising results, for example with a combination of species *Spathip-hyllum wallisii* "Verdi", *Dracaena fragrans* "Golden Coast" and *Zamioculcas zamiifoli* [142]. Jung and Awad [143] demonstrated that indoor plants can improve the IAQ in University classrooms, showing that the increase in CO2 concentration in classrooms with plant placement was lower (624 ppm) compared with the case without plant placement (about 1205 ppm). Later, Dela Cruz et al. [144] reviewed the use of potted plants to remove VOCs from indoor air, including the removal mechanisms and the relationship between plant, soil, and associated microorganisms. The literature mentions other pollutants, such formaldehyde and particulate matter, that are also affected by potted plants [132]. The extent of cleaning efficiency by plants can be increased by increasing the number of plants [145]. However, it is not possible to accurately estimate the number and size of plants needed to effectively purify indoor air. This is because different plant species have differences in photosynthesis, such as different optimal light intensity levels; some species grow slowly and others faster. Additionally, the type of plant and pollutant affects the phytoremediation efficiency. Moreover, the type of soil used for the plants to grow is another important factor, demonstrated in a study [138] using as a mixture of coco coir and activated carbon (for adsorption) as growing media. This significantly increased the VOCs removal efficiency (*p*-value > 0.05). Additionally, there is a lack of information concerning most contaminants in indoor air, thus it is very important to determine the concentration of indoor air pollutants.

Another challenge when using plants is the quantity needed to effectively improve the indoor air quality, which is not always practical in housing and/or offices with reduced areas. Thus, instead of using potted plants, several studies [146,147] suggested the use of "green walls" or "biowalls" with different plant species, which can simultaneously remove a mixture of different contaminants with less space requirement.

Biowalls are considered plant-based systems that mainly act to enhance the atmosphere and indoor environment and can be classified as living walls, vertical gardens or green facades, depending on their characteristics (type of plants and structural system) [148]. Figure 3 shows an example of a modular living wall system with planter boxes, implemented in an office building (Porto Office Park, Porto-Portugal). Figure 4 shows an example of a continuous system living wall with felt pockets, implemented in a Portuguese shopping center (Norteshopping, Matosinhos-Portugal).

Irga et al. [136] highlighted the importance of the technological advancements related to active botanical biofilters or functional green walls, which are becoming increasingly efficient and a rapidly growing field of research interest.

Several plant-based systems have emerged to support the improved indoor environment, such as the biofiltration system, where air is drawn through organic material (such as moss, soil and plants), resulting in the removal of organic gases and contaminants involving a mechanical system [130], and the nature-based air filtering system, which was installed at a students' residence, replacing an existing window with a mini-greenhouse, containing upwards of 30 plants connected to an air circuit to treat the indoor air [149]. The combination of systems can often boost and optimize the air purification effect, such as the combination of living wall systems with biofiltration. They are thus emerging technologies providing beneficial effects on the improvement of indoor comfort [130].

**Figure 3.** Modular system living wall with planter boxes in an office building (Porto Office Park, Porto-Portugal). Credits: Cristina Calheiros.

**Figure 4.** Continuous living wall system with felt pockets in a shopping center (Norteshopping, Matosinhos-Portugal). Credits: Cristina Calheiros.

The main recommendations associated with vegetation systems for indoor air quality promotion comprise a detailed selection of plant species, growth medium, irrigation systems, adequate light setting and abiotic conditions (e.g., temperature and humidity). The main research gaps center on the need for phytoremediation operational systems, which are important when considering a wider range of pollutants and plant organisms for indoor purposes [133]. Since the process performance depends on the interactions between pollutant, plant and microorganisms [130], it would be relevant to develop tools to support indoor air phytoremediation in order to further expand its use and allow a wider array of applications. For example, Thomas et al. [150] developed a mathematical model that takes into account the amount of plant material, building air volume, VOC concentrations and air exchange. More tools are needed to allow continuous and long-term monitoring of these systems performance.

#### *3.2. Microalgae-Based Air Purification Systems*

Microalgae are prokaryotic or eukaryotic microorganisms with a unicellular or simple multicellular structure. They are among the most efficient photosynthetic organisms on earth, with high biomass productivity and relatively low nutrient requirements, accumulating metabolites with several applications. They can live in harsh conditions, in seawater or freshwater, and multiply exponentially under favorable environments [151].

Microalgae for carbon capture via photosynthesis has gained increasing attention from the scientific community due to their ability to capture CO2 and other pollutants, which they use to grow while producing O2. In particular, microalgae are among the most efficient photosynthetic organisms, with high biomass productivity and relatively low nutrient requirements. They accumulate metabolites (e.g., pigments, fatty acids, etc.) with important biotechnological applications [152,153]. The biomass can be converted into chemicals and/or biofuels (e.g., biohydrogen, biodiesel, bioethanol, biobutanol, biomethanol and other biohydrocarbons), thus generating environmental and economic value [151,154] (Figure 5).

**Figure 5.** Microalgae as CO2 biofixers, O2 producers, and potential applications of microalgae biomass (adapted with permission from Mata et al. [7]).

The CO2 bio-fixation by microalgae is a complex physicochemical process. Microalgae can tolerate up to a certain level of CO2, after which it becomes detrimental to cell growth. This is due to the lowering of the medium pH, and to the environmental stress caused by

higher CO2 levels, reducing the cells' ability to fix more carbon. In an aqueous environment, inorganic carbon is available in different chemical forms, such as CO2, H2CO3, CO3 <sup>2</sup><sup>−</sup> and HCO3−. Depending on the species and on the physicochemical and hydrodynamic conditions, microalgae can biofix around 1.875 g of CO2 per gram of biomass, [155]. Nonetheless, microalgae harvesting is still an energy intensive process due to the low cell density in the culture medium, typically in the range of 0.3–0.5 g/L, with exceptional cases reaching 5 g/L [156].

With the world's rapid urbanization and demand for renewable energy sources, a recent focus on the potential of microalgae is to contribute to the development of green cities and build a more sustainable future [157]. In particular, microalgae systems can be implemented as nature-based solutions to improve indoor air quality, along with other benefits [7], in alignment with the United Nations Sustainable development Goals (UN SDG) [158]. Moreover, their implementation will promote good health and well-being (Goal 3) and sustainable cities and communities (Goal 11), contributing to the European Commission's plans of Building a Green Infrastructure for Europe [159], allowing for thermal regulation and new architectural features [160]. Recommendations for sustainable urban planning include the implementation and, subsequently, the increase in urban greenery [148] as part of an urban renewal and rehabilitation strategy towards smarter and more sustainable development, and responding to global demands such as mitigation and adaptation to climate change [161].

With regard to the construction and architecture sectors, it is very common to find buildings with integrated photovoltaic energy. Despite the potential of microalgae biomass in the generation of renewable energy, its integration in buildings is still quite modest and at an early stage. In particular, microalgae have a great ability to biofix CO2, produce O2, treat wastewater, and produce biomass that can be used for bioenergy and bioproducts. The few examples of microalgae implementation in buildings and architectural design have demonstrated their significant contribution to energy efficiency, but they have never been tested for indoor air purification [7,162]. Recent studies have highlighted the potential of microalgae for indoor air purification, for example:


is a technical barrier to its utilization. Thus, further development is required for the correct operation of this air purifier, in particular for better control of the moisture content and pH value.

• Thawechai et al. [166] studied the oleaginous microalgae *Nannochloropsis sp*. as a potential strain for CO2 mitigation into lipids and pigments, analyzing the synergistic effects of light intensity and photoperiod. The authors obtained a 0.850 ± 0.16 g L−<sup>1</sup> with a lipid content of 44.7 ± 1.2%. The CO2 fixation rate was 0.729 ± 0.04 g L−<sup>1</sup> <sup>d</sup><sup>−</sup>1. The fatty acids were mainly C16–C18, indicating its potential use as biodiesel feedstock.

In buildings, microalgae can be cultivated in closed photobioreactors (PBRs) of different shapes (e.g., flat panels, multi-tubular, etc.) and dimensions that can act as dynamic shading devices (Figure 6). Depending on the density of microalgae biomass inside the PBRs, and the amount of sunlight absorbed by cells, various shading levels can be provided. On the other hand, the biomass density depends on the microalgae species, their growth cycle, the available carbon dioxide and sunlight, temperature of the culture medium, and frequency of biomass harvesting, among other factors. However, these conditions can be adjusted to the needs of the building's users. '

**Figure 6.** Example of integration of microalgae PBRs and plants in buildings (authors' own creation).

#### **4. Conclusions and Future Trends**

Indoor air quality has gained a new focus with the emergence of COVID-19. Ventilation is perhaps the most used strategy to decrease the concentration levels of indoor air pollutants. However, if outdoor air is more polluted, or in certain situations where ventilation is not possible, other strategies need to be applied, such as source control and pollutants extraction. The latter incorporates air cleaning technologies, one of the emergent areas on IAQ. Various air treatment technologies can be used to control contaminants, which are reviewed and discussed in this work and include physicochemical technologies (e.g., filtration, adsorption, UV-photocatalytic oxidation, ultraviolet disinfection and ionization) and biological technologies (e.g., plant purification methods and microalgae-based methods).

It is consensual that filters can efficiently remove particles but are not effective for organic and inorganic chemical pollutants. A solution seems to be a combination of particulate filter and other systems, or the development of filters with other properties, such as chemisorption. Nanofibers can be modified with nanomaterials to form effective

adsorbent materials that can efficiently adsorb the different VOCs. With the advent of nanotechnology, there is enormous research potential in this area.

On the other hand, we have adsorption capable of capturing both volatile organic compounds and inorganic pollutants. Adsorbents can be easily incorporated into building materials and/or integrated into interior surfaces to remove air pollutants with no additional energy input and minimal byproduct formation; for this reason, they are classified as passive removal materials (PRMs). A negative aspect of the adsorption technology is the possibility of the development of airborne bacteria on the adsorbent surface due to the high biocompatibility of these materials. Additionally, adsorption technology produces a hazardous solid waste that must be further treated and/or disposed of correctly.

UV-Photocatalytic Oxidation (PCO) is a promising cleaning technology, ranging from organic to inorganic compounds, as well as microorganisms, but studies show that there are aspects that need to be addressed before UV-photocatalytic oxidation can be used safely in buildings. The design of a photocatalytic air purification system is complex and is dependent on a wide variety of factors, including intensity of the light falling on the catalyst, chemical makeup and concentration of pollutants, the air flow rate through the device, moisture levels in the air, properties of the specific catalyst used, pollutant residence time, and how the device itself is configured. The main disadvantage is the incomplete oxidation, which produces reaction byproducts that can be more toxic or harmful than the original constituents (e.g., formaldehyde).

Ultraviolet light is used mainly for disinfecting air in buildings belonging to the healthcare sector, focusing on microorganisms such as bacteria, virus, fungi, etc. However, a growing number of studies has provided robust evidence on the effectiveness of this air cleaning technology in several indoor environments, including offices, commercial areas, schools, and universities. Currently, ASHRAE and EPA also recognize the importance of employing air cleaning solutions that include both filtration and UVC-based systems to prevent the spread of infectious diseases in indoor environments. Importantly, in order to avoid any collateral risks to the health of occupant, the potential release of toxicant byproducts (as ozone) by all kinds of UV systems should be carefully controlled. Another aspect to keep in mind is that there are limited values established for exposure to UVC light that cannot be exceeded, and direct exposition should be avoided as direct UVC light can damage to the eyes and skin.

Air ionization-based technologies show potential, as there is some evidence showing their efficacy to improve IAQ, but the literature is still too scarce to draw robust conclusions. Nevertheless, a growing body of recent evidence has presented the potential of using air ionization to decrease bacterial deposition on surfaces, inactivate airborne bacteria, viruses, and fungi and to remove airborne particles and VOCs, being more effective when used in long-term applications. Similarly, to UV-based systems, these systems can produce high levels of ozone and other organic compounds, which requires surveillance.

Phytoremediation is characterized by the use of plants to remove pollutants, in this case from air. Interest associated with phytoremediation is high since it is considered an approach with low implementation cost and maintenance, compared with other technologies Different studies on the purification of indoor air have concluded that plants are able to efficiently remove different harmful contaminants from indoor air, including VOCs. This can be achieved through passive filtration using potted plants, or through active filtration using plant filters or green walls. When using plant-based systems to promote indoor air quality, it is also important to consider the pollen load and allergenicity potential, such as flowering plants with strong fragrances. Plants also emit compounds, some of which are biologically active. The main research gaps focus on the needs of phytoremediation operational systems, making it important to consider a wider range of pollutants and plant organisms for indoor purposes.

Microalgae have a great ability to biofix CO2, produce O2, treat wastewater, and produce biomass, which can be used for bioenergy and bioproducts. The few examples of microalgae implementation in buildings and architectural design have demonstrated their significant contribution to energy efficiency, but they have never been tested for indoor air purification in real situations. However, from the studies performed in lab, it seems to be a promising technology.

To conclude, all the technologies presented show advantages and limitations. One of the common problems is the emission of other pollutants, different and sometimes more dangerous to human health than those they were initially intended to remove. Another aspect is the limited type of pollutants removed. The solution seems be the use of combined technologies to extend the scope and of course the development of new materials, using nanotechnology. From this review, one conclusion can be taken: there is still a need for more research in order to consolidate the existing technologies and assure their safety for use in indoor spaces.

Furthermore, the current reality has shown the importance of good indoor air quality. Thus, these new technologies can contribute to its improvement, despite all their limitations. It is essential that people acknowledge the importance of clean air as equally as pure water. It will be necessary to define minimum parameters for the classification of good IAQ. Additionally, it would be important to see a generalization of IAQ audits, which would enhance source control and ventilation, the main vectors to ensure good IAQ.

**Author Contributions:** Conceptualization, T.M.M., M.F.G. and G.V.S.; methodology, validation, writing, review, and editing, T.M.M., A.A.M., C.S.C.C., F.V., N.P.A.-C., M.F.G. and G.V.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financially supported by base funding of the following projects: LA/P/0045/ 2020 (ALiCE), UIDB/00511/2020 (LEPABE), and UIDB/50022/2020 (LAETA), funded by national funds through FCT/MCTES (PIDDAC). António Martins thanks the Portuguese National Funding Agency for Science, Research, and Technology (FCT) for funding through program DL 57/2016 —Norma transitória. Teresa Mata gratefully acknowledge the funding of Project NORTE-06-3559- FSE-000107, co-financed by Programa Operacional Regional do Norte (NORTE2020), through Fundo Social Europeu (FSE). Calheiros C. is thankful to FCT—Fundação para Ciência e Tecnologia within the scope of UIDB/04423/2020 and UIDP/04423/2020.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**


## MDPI

St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

MDPI Books Editorial Office E-mail: books@mdpi.com www.mdpi.com/books

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com ISBN 978-3-0365-5982-7