Comprehensive target analysis of micropollutants in soil at debris storage sites of the Kumamoto earthquake

ABSTRACT On April 16, 2016, a series of earthquakes with a maximum magnitude of 7.0 struck Kumamoto Prefecture, Japan, and about 1,300,000 metric tons of land-derived debris (building materials, household appliances, furniture, and wood) required disposal. Because public lands were used as emergency debris storage sites, environmental contamination by organic micropollutants from the debris has been a public concern. To identify chemicals of concern in the soil, our research group used a novel scheme in the combination of comprehensive target analysis (CTA) with an automated identification and quantification system (AIQS) that made use of gas chromatography–mass spectrometry (GC-MS). CTA-AIQS identified a total of 120 organic micropollutants in the soil samples from the debris storage sites. The predominant chemical groups in debris soils were PAHs, plasticizers, and pesticides. Higher concentrations of pesticides (insecticides and fungicides) were detected in soil under waste wood because those chemicals were used as wood preservation and termiticides. Estimated hazard quotients of the detected chemicals via soil ingestion by children (a high-risk group) suggested that the health risk was probably negligible. It was concluded that CTA-AIQS would be a useful tool for emergencies such as natural disasters.

Introduction facilities suited for waste storage and treatment. In such an emerging situation, it requires an analytical technique for monitoring environmental contamination which can detect a wide range of micropollutants and semiquantify them rapidly.
Analytical techniques for the identification and quantification of hazardous chemicals in environmental media can generally be categorized as "target analysis," "non-target screening analysis," and "comprehensive target analysis (CTA)" (Díaz et al. 2012;Krauss, Singer, and Hollender 2010;Matsuo et al. 2019;Zushi et al. 2014). Target analysis has the advantage of being accurate and precise because it focuses on only a limited number of similar chemicals as target compounds. Other substances, however, are removed from the sample via cleanup processes during analytical pre-treatment. Target analysis is generally used for routine environmental monitoring; it is the method specified in governmental regulations, even if it is laborious and time-consuming. However, in emergencies (e.g., earthquakes, tsunamis, and typhoon), information about environmental pollutants likely to be present may be unavailable. Choosing a suitable target analytical method and rapidly analyzing the large number of samples collected in a disaster area is therefore likely to be difficult. In contrast, no target substances are defined in non-target screening analysis, and environmental samples are comprehensively screened for the identification of unknown chemicals. However, non-target screening analysis is not quantitative (or quantification is relative to an internal standard) and is resource intensive; it requires high-performance equipment (e.g., high-resolution mass spectrometer), technicians (data interpretation with high-performance software), and time. In emergencies, quickly processing large numbers of samples is difficult. Because of the above-mentioned disadvantages of target analysis and non-target screening analysis, CTA can be a useful tool in emergencies. An Automated Identification and Quantification System (AIQS) that involves the use of full-scan-mode gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) has been developed for CTA (Kadokami, Jinya, and Iwamura 2009;Kadokami et al. 2012Kadokami et al. , 2004Kadokami et al. , 2005Kadokami and Ueno 2019). AIQS analysis can be used to identify and semiquantify chemicals in environmental samples by using a combination of information, including retention times, mass spectra, and internal standard calibration curves registered in GC-MS and LC-MS databases. With AIQS, it is possible to identify and semiquantify the chemicals registered in the database without measuring authentic standards for each analysis. Although sensitivity and accuracy could be relatively lower than target analysis, AIQS enables the rapid identification and semiquantification of approximately 1500 micropollutants in environmental media by means of GC-MS and LC-MS. Because of these advantages, AIQS analysis has been authorized for use in Japanese Industrial Standard (JIS) methods developed by the Japanese Standards Association (JIS 2018).
The comprehensive target analysis with AIQS (CTA-AIQS) has been used as a useful tool to screen micropollutants in emergencies, eg. earthquake and flood. It was reported that various hazardous chemicals were detected using CTA-AIQS in sediment debris samples collected near coastal area which are brought by Tsunami attack in the Great East Japan Earthquake, 2011. Within this research, it has established a novel scheme for environmental monitoring using CTA-AIQS in emergency situations, which named Rapid screening for Environmental micropollutants in Emergency situations (REPE) (Matsuo et al. 2019). Water samples collected near emergency debris storage sites made after the Great East Japan Earthquake were also analyzed by CTA-AIQS and various chemicals which might reached from the debris were detected (Nakajima et al. 2019). CTA-AIQS was applied to the accident of paddy soil contamination by industrial oil which was spilled from steel mill factory by flood with heavy rain in Japan (Nishimuta et al. 2020).
In this study, investigation of micropollutants in soil samples collected from debris storage sites of Kumamoto Earthquake using the CTA-AIQS, and evaluation of health risks by ingestion of soil dust were conducted. The objective of this study was to demonstrate the usefulness of CTA-AIQS as a scheme for rapid screening in emergencies.

Sample collection
On May 19, 2016 (1 month after the earthquake), soil samples were collected at three debris storage sites in Kumamoto Prefecture that were separated from each other by~10 km. These debris storage sites were located on a parking lot, vacant ground, and athletic field at locations A, B, and C, respectively, in Table 1. Those debris storage sites had accumulated various types of wastes, including building materials (wood, cement, insulation, etc.), electric appliances (televisions, refrigerators, washing machines, etc.), and household furniture (beds, couches, other furniture, etc.) after the earthquake in April 2016. Table 1 provides details about the samples that were collected in this study, and Figure S1a, b, and c shows photographs of the sampling sites. The soils collected from the debris storage sites fell into two categories: soil from under the piles of waste (debris soil) and soil from within the debris storage sites but not beneath a pile of waste (reference soil). Samples were collected in accord with the previously reported methods (MOE 2009). Five portions of surface soil were collected and composited as one sample (nearly 1 kg of each sample collected from an area of 100 m 2 at each site). Those samples were transferred to a laboratory, air-dried for several days at room temperature, sieved (1-mm mesh), and stored at −20°C until chemical analysis.

Microwave-assisted extraction (MAE)
In this study, the method of sample preparation was Microwave-assisted extraction (MAE) which included extraction and cleanup with a biphasic solvent (hexane/water) (Miyawaki et al. 2018). MAE is a high-throughput method for preparing soil and sediment samples for GC-MS analysis because it does not require filtration, dewatering, solvent evaporation, or solvent replacement. The use of MAE stabilizes the GC-MS chromatograms of soil and sediment samples better than conventional extraction methods (Matsuo et al. 2019;Miyawaki et al. 2018). Briefly, each soil sample (1-3 g) was placed in a quartz glass extraction cell and then extracted with 10 mL of 3:2 (v/v) hexane/water for 30 min in a microwave digestion system (ETHOS TC, Milestone, Italy). A 5-mL aliquot of the hexane layer were applied to a silica gel column (1.5 g) and was eluted with 10 mL of 7:3 (v/v) hexane/acetone. The eluate was concentrated under flowing nitrogen to a final volume of 100 µL after the addition of the internal standards listed in Table S1 (NAGINATA Internal Standard Mix II, Hayashi Pure Chemical, Japan).

Instrument for CTA-AIQS
For identification and semiquantification of organic micropollutants, we used an AIQS that made use of GC-MS information (Kadokami et al. 2005). The information included retention times, mass spectrum, and calibration curves for 937 organic micropollutants (Table S1). AIQS analysis does not require the use of authentic standards for identification and semiquantification. Each sample was injected into a GC-MS system (6890/5973 N; Agilent Technologies, USA) that was operated in scan mode (m/z 50-500) and was equipped with an HP-5 MS column (30-m length, 0.25-mm i.d., 0.25-mm film thickness; Agilent Technologies). The organic micropollutants in the chromatogram were identified and semiquantified via NAGINATA2 add-in software (Nishikawa Keisoku, Japan) for ChemStation (Agilent Technologies); the add-in contained an AIQS database that included 937 organic compounds (Table S1).

CTA-AIQS quality control
Stable GC-MS performance is important for the optimization of AIQS analysis. We monitored GC-MS performance by using a performance check standard that consisted of 25 indicator compounds (Table S1; NAGINATA Criteria Sample MixII, Hayashi Pure Chemical, Japan). The results obtained with this standard were evaluated by means of three criteria: spectrum validity, inertness of column and inlet liner, and stability of response (Kadokami et al. 2004(Kadokami et al. , 2005. When these criteria had been satisfied, the environmental samples were injected into the GC-MS system. Recovery tests for the entire CTA-AIQS procedure (MAE, cleanup, and AIQS analysis) have been investigated for various chemicals in previous studies, and the results have demonstrated the advantages and limitations of this system when used for comprehensive screening of environmental samples (Matsuo et al. 2019;Miyawaki et al. 2013Miyawaki et al. , 2018. Table S1 lists the CTA-AIQS method detection limits (MDLs) for all the chemicals. The MDL estimated from instrument detection limits in the AIQS database for 1 g of soil samples was 1 ng g −1 dry wt. Because the standard deviations of the procedural blanks of some chemicals (e.g., phthalates) were high, these chemicals were not quantified (shown as NQ in Table S1). The sensitivity of the CTA-AIQS method was lower than that of the target analytical method because the former involved the use of the GC-MS scan mode to screen for many chemicals rather than the use of the GC-MS to accurately quantify a limited number of chemicals. Validation of CTA-AIQS has been conducted by analysis of two certified reference materials (NIST 1941b and NMIJ CRM7302a), and the estimated concentrations have equaled 70-120% of the certified values .

Statistical analysis
Concentrations of soil samples were expressed as μg g −1 dry weight. Values under the MDL were considered as zero in the calculation of total concentration. Principal component analysis (PCA) and group comparisons (Wilcoxon rank-sum test) were conducted with the JMP software suite (version 12, SAS Institute Inc.). In this study, a type I error rate (p value) less than 0.05 was considered to indicate statistical significance.

Detection of organic micropollutants by means of CTA-AIQS
In this study, CTA-AIQS was used for comprehensive target analysis of debris soil and reference soils collected during 2016 from debris storage sites in Kumamoto. One hundred twenty organic micropollutants were detected in the debris soil and reference soil samples collected on March 19, 2016 (Table 2; individual data are in Table S2). The median and range (minimum-maximum) of chemical numbers detected in debris soils and reference soils were 51 (27-74) and 30 (24-37), respectively. The chemicals detected were classified into the following five chemical groups: PAHs, alkanes, plasticizers, pesticides (including insecticides, fungicides, and herbicides), and others (Table 2 and Figure 1). The predominant chemical groups in debris soils were PAHs and plasticizer, alkanes, and pesticides; those in reference soils included plasticizer and alkanes. The median and range of the total concentrations in debris soils and reference soils were 2.3 (0.20-5.5) and 0.37 (0.22-0.55) µg g −1 dry wt, respectively. Larger numbers of chemicals and approximately 10 fold higher concentrations were detected in debris soils than reference soils.

Characteristic chemicals in debris soils
We conducted a statistical analysis to identify the characteristic chemicals among those detected by the CTA-AIQS. In this study, we used principal component analysis (PCA) to identify chemicals that were characteristic of the debris soils. Prior to PCA, we calculated the percentage contribution of each chemical to the total concentration detected by CTA-AIQS, because the aim of this initial approach was not only to find chemicals present at higher concentrations in the debris soils but also to search for characteristic chemicals that were present only at low concentrations. Figure 2 shows the PCA score plot. Two major independent principal components (PCs) were extracted by the PCA analysis. PC1 and PC2 explained 41% and 31%, respectively, of the total variance between samples and together accounted for 72% of the variance. Analysis of the projections of the variables onto PC1 and PC2 confirmed that the most important parameter was the projection onto PC1 (Figure 2a). In the score plot for PC1, the scores were positive for the sites that included reference soils (Aref, Bref1, Bref2, and Cref) and debris soils from beneath household furniture (beds, couches, and other furniture: A3 in Table 1) and building materials (cement: C1 in Table 1). The scores were negative on PC1 for the sites that included soils from beneath electric equipment waste  (televisions, refrigerators, washing machines: A2 and C3 in Table 1) and building materials waste (wood: A1, B1, B2, and C2 in Table 1) (Figure 2a). The score plot for PC2 showed that the scores of soils beneath waste from building materials (wood: A1, B1, B2, and C2) were positive, and the scores of soils beneath electric equipment waste (A2 and C3) were negative. The scores of reference soil and the soils from beneath household waste and building material (cement) were clustered around zero. The results showed that three clusters were separated by PC1 and PC2. Examination of the characteristic vectors (Figure 2b) revealed positive scores along PC1 for the concentrations of alkanes and other chemicals. The implication is that high concentrations of alkanes and other chemicals (including 1,4-dichlorobenzene and acetophenone) were present in reference soil and the soils beneath waste from households and building materials (cement). A negative score along PC1 and positive score along PC2 for the concentrations of PAHs and pesticides suggested that higher concentrations of those chemicals were present in soils beneath building materials waste (wood). A negative score along PC1 and negative score along PC2 for the concentrations of plasticizers suggested that higher concentrations of those chemicals were present in soils below electrical equipment waste (television, refrigerator, washing machine). Taken together, the PCA results indicated that the characteristic chemicals in soils below building materials waste were pesticides and PAHs. The characteristic chemicals in soils below electric equipment waste were plasticizers. Further research was then conducted to more clearly define the characteristic chemicals detected in each location.

Source estimation of characteristic chemicals
The initial CTA-AIQS survey suggested that pesticides and PAHs from waste wood, and plasticizers from waste electric equipment were the characteristic organic micropollutants in soils below stored debris (Figure 2a). It was obvious that plasticizers detected at A2 and C3 (Table 1) were derived from plastics used in waste electric equipment. Source of PAHs and related chemicals (biphenyl, carbazole, dibenzofuran, dibenzothiophene, etc.) detected in soil under waste wood in A1, B1, and C2 could be pyrogenic and petrogenic (including coal-tar paint) origins. In contrast, it was unclear why the group of pesticides (insecticides, fungicides, and herbicides) was detected in those samples of soil under waste wood. To try to identify the source of those chemicals, we summarized in Figure 3 the profiles of the pesticides that were detected (Table S3 summarizes the individual concentrations and uses of those pesticides). The major categories of agricultural chemicals were insecticides (chlordane, chlorpyrifos, and fipronil) and fungicides (iprobenfos, isoprothiolane, and propiconazole). The herbicides thiobencarb, bromobutide, and mefenacet were also detected in the soil under waste wood. Among the pesticides, several insecticides and fungicides have been used as wood preservatives and termiticides in Japan (JWPA 2006). Table S4 summarizes the chemicals used as wood preservatives and termiticides in Japan. It was reported that pentachlorophenol (PCP), chlordanes, and polychloronaphthalenes (PCNs) which used as wood preservatives were detected in waste timber (woods) collected in Japan (Koyano et al. 2019). Soil contamination by dieldrin and pentachlorophenol (PCP), which leach from timbers treated with wood preservative, has also been reported (McNeili 1990). Those results suggested that the waste wood debris included timbers used as building materials that had been treated with wood preservatives and termiticides; those chemicals then leached from the wood after it was piled together at debris storage sites, and they contaminated the soil at those locations.

Initial risk assessment
We conducted an initial risk assessment using the CTA-AIQS data. We estimated the oral intake of those chemicals via soil ingestion by children (a high-risk group) by using hazard quotient (HQ) values, as described in previous reports (Jones-Otazo et al. 2005;Matsuo et al. 2019;Mizouchi et al. 2015;Wilford et al. 2005). The body weight and amount ingested per child have been reported to be 20 kg and 0.2 g day −1 , respectively (USEPA 1997). Because a variety of agricultural chemicals (insecticides, fungicides, and herbicides) were detected from beneath waste wood, for estimation purposes we used the highest concentrations of each chemical detected among samples at sites A1, B1, B2, and C2. The acceptable daily intake (ADI) for pesticides and tolerable daily intake (TDI) for other chemicals were used for calculation purposes (WHO 2017). Figure S2 summarizes the HQs for the various chemicals. Although the calculated HQs were 1 to 3 orders of magnitude higher in debris soils than in reference soils, the HQs in debris soils were 2 to 3 orders of magnitude lower than their tolerable intake. Our calculations suggest that the health risk associated with exposure to those chemicals due to soil ingestion is likely to be negligible. It is also important to evaluate the risk for organisms in soil environment. It has been reported that OECD provides the test guidelines of acute toxicity test 207 (TG 207) using earthworm (OECD 1984). Although only human health risk was evaluated in this research as an emerging situation, toxicity of organisms in soil should be monitored for long-term effects on ecological risk.

Conclusions
In this study, we used CTA-AIQS scheme to analyze soil samples at sites where debris from the Kumamoto Earthquake of April 2016 was stored. The risk assessment was carried out for 120 chemicals detected by CTA-AIQS and our calculations suggest that the health risk associated with exposure to those chemicals due to soil ingestion is likely to be negligible. It was concluded that CTA-AIQS would be a useful tool for emergencies such as natural disasters.