Estimation of the fraction of soil-borne particulates in indoor air by PMF and its impact on health risk assessment of soil contamination in Guangzhou, China

Highlights

• Dominant contributors to indoor PM₁₀ were combustion (50.53%) and traffic (28.17%).

• The local fraction of soil-borne particulates in indoor air (fspi) is 19.96%.

• PMF model was used to estimate the local fspi in Guangzhou main urban areas.

• Indicator values of inhaled risk of some metals reduced by 62% for localized fspi.

• Localized fspi caused a marked growth in screening levels of Cr (VI), Ni, Be and Cd.

Abstract

The fraction of soil-borne particulates in indoor air (fspi), a principal exposure factor in health risk assessment of soil, is used to calculate the inhaled dose of contaminants in air particulates (PM₁₀) from soil. To investigate the fspi, consecutive 24-h PM₁₀ samples (n = 180) of indoor ambient were collected from September 2019 to January 2020 in Guangzhou main urban areas, China. The concentrations of twenty-six metal elements, five anions, organic carbon (OC) and elemental carbon (EC) in samples were measured. The sources of indoor ambient PM₁₀ and the value of fspi were identified by the method of Positive Matrix Factor analysis (PMF). Results showed that the main sources contributing to indoor PM₁₀ content were combustion sources (50.53%) and vehicular sources (28.17%). The soil sources (the local fspi) were 19.96%. The soil contents of indoor PM₁₀ in Guangzhou main urban areas were in accordance with those in similar monsoon climate regions, such as Malaysia. The health risks of the inhalation route were dropped by about 62% for some common brownfield contaminants (Cr (VI), Ni, Be and Cd) with the investigated local fspi in Guangzhou main urban areas, compared with using the fspi (0.8) recommended by the C-RAG model in China. The results supplied a new effective methodology for estimation of the local fspi value in health risk assessment of soil contamination in urban areas.

Introduction

Health risk assessment of soil contamination of land is a necessary tool of environmental management. The main exposure routes for non-volatile contaminants in soil of land are oral ingestion of soil (Ois), dermal contact with soil (Dcs) and inhalation of soil particles (Pis). Among them, Pis is the human respiratory system exposure to soil contaminants through inhalation of air particulates, and might cause potential damage (Kim et al., 2015). Even if the windows and doors are closed, air particulates carrying soil pollutants can enter the room as long as the wind speed exceeds 0.45 m s⁻¹ (Zhang et al., 2018). Although modern concerns are more focused on fine particulate matter (PM_2.5) than coarse particulate matter (PM₁₀) (Brown et al., 2013), the PM₁₀ dominate the portion of soil-borne particulates and the PM_2.5 just has the low proportion (Harrison et al., 2000; Hwang et al., 2008; Mooibroek et al., 2011). Therefore, PM_10, rather than PM_2.5, was addressed in all the health risk assessment models of soil contamination of land. The fraction of soil-borne particulates in indoor air PM₁₀ (fspi) is one of the main exposure parameters to estimate the inhalation average daily dose (ADD).

Various environmental exposure models have been developed to assess the health risk of soil contaminants in land, including the Risk-Based Corrective Action (RBCA) model (ASTM, 2020), the Integrated Exposure Uptake Biokinetic (IEUBK) model (U.S.EPA, 1998), the Contaminated Land Exposure Assessment (CLEA) model (Defra, 2002), the CSOIL model (Brand et al., 2007), and the C-RAG model (China MEE, 2019). Among these models, IEUBK, CSOIL and C-RAG models used fspi to calculate the total dose of indoor PM₁₀ from soil inhaled by humans. Admittedly, the accuracy of models is susceptible to toxicity parameters, physical and chemical parameters of contaminant, and exposure parameters. Compared to the other parameters, the exposure parameters such as fspi might result in more uncertainty for the assessment of health risk as fspi may vary depending on the climate, seasonal surface, soil characteristic and housing designs in different regions (U.S.EPA, 2011; Wang et al., 2012). Thus, only with local fspi can ADD be accurately estimated.

In the CSOIL model, the recommended value of fspi was 80% obtained by the method of indoor dust Pb isotopes (Hawley, 1985; Van den Berg, 1994; Swartjes, 2003). The U.S. Environmental Protection Agency (EPA) estimated fspi at 70% in the IEUBK model by analyzing Pb in indoor dust and outdoor adjacent soil (U.S.EPA, 1998). China national environmental protection standards (HJ 25.3–2019) “Technical guidelines for risk assessment of soil contamination of land for construction” also recommended fspi of 80% in C-RAG model. The methods of estimation fspi by Pb isotopes in the CSOIL model and the IEUBK model need high requirements on experimental instruments and cost of sample treatment. Except for soil sources, coal burning and vehicle exhaust also have significant contribution to Pb-210 content in particulate matter (Church et al., 2008; Mattsson et al., 1988; M. Baskaran et al., 2011). All the relevant isotope characteristic values of samples from various emission sources need to be investigated (Zheng et al., 2004). Thus, the methods of fspi estimation by Pb isotopes are very complicated. Looking for a more convenient method to estimate the local fspi in different areas would be helpful to the health risk assessment of soil contaminants.

Local fspi might be estimated by source apportionment which quantifies factor (source) contributions with chemical and statistical methods. Positive matrix factor analysis (PMF) and principal component analysis (PCA) are extensively applied in the source apportionment of particulate matter (PM). PMF could quantify the main factors and contributions with the weight among chemical components of particulate matter and the least square method based on assuming mass conservation and chemical mass balance between emission sources and receptors (Wu et al., 2020). Suryawanshi et al. (2016) illustrated by PMF that the contribution of soil to PM_0.6 amounted to 17.5% in India classrooms. Bari et al. (2015) obtained a 16% fraction of soil to residential PM₁ by PMF in Canada. Taghvaee et al. (2018) indicated that the fractions of soil to PM_2.5 were 2.5% and 8% for some nursing homes and dormitories respectively in Iran. In Xi'an, China, some indoor factories witnessed the 5.4% soil of PM_2.5 by PMF (Dai et al., 2019). By PMF some campus indoors had 4.4%, 6% and 18% PM_2.5 originated from soil in the northeastern Nanjing China, United States and Seoul Korea respectively (Heo et al., 2021; Carrion-Matta et al., 2019; Niu et al., 2021). The contribution of soil to residential PM_2.5 with PMF in Pittsburgh, Massachusetts, Hubei and Shanghai, reached 5%, 17%, 17% and 25%, respectively (Tunno et al., (2016); Matthaios et al., (2021); Fang et al., (2021); Brehmer et al., (2019)). By PCA, Wang et al. (2018) investigated that the hall of residence had 11.26% natural source to PM_2.5 in Nanjing China. However, all the above studies just concerned on indoor or outdoor air fine particulates (≤PM_2.5). For PM₁₀, the contribution of soil to outdoor PM₁₀ was estimated as 30%–40% by Cheung et al. (2011) and Ho et al. (2003) using outdoor air receptor model. Very few studies investigated the fraction of soil-borne to indoor PM₁₀ by source apportionment except for Ali et al. (2017), who reported a total of 20% of indoor PM₁₀ from soil in campuses, Malaysia by PCA. And evaluating the contribution of soil to indoor air quality would be different from conventional outdoor air receptor model in methodology. Meanwhile, very few studies used PMF to localize fspi for health risk assessment of soil.

In Guangzhou, a megacity with a population of 18.67 million, about 75% of the total population reside in the main urban areas. 36.56% brownfields in the total ninety-three lands investigated from 2013 to 2018 were found soil contaminated over there (Tan et al., 2021). The accuracy of health risk assessment for soil contamination of land is strongly concerned. Therefore, this study will try to apply PMF methodology to estimate the fraction of soil-borne to indoor PM₁₀ in Guangzhou main urban areas, and supply a new method for investigation of local fspi value in the health risk assessment of soil contamination of land. The specific objectives of this study were: (1) to determine the concentrations of chemical components such as metal elements, anions, organic carbon (OC) and elemental carbon (EC) in indoor ambient PM₁₀ samples collected from Guangzhou main urban areas, China; (2) to characterize the sources of indoor PM₁₀ and the fraction of soil-borne to indoor PM₁₀ by source apportionment; (3) to survey the differences between the health risk results calculated with local fspi value and the recommended value by the Chinese C-RAG model.