Introduction

Soil pollution is now recognized as a major global environmental hazard (Karim et al. 2015; Bhuiyan et al. 2021; Proshad et al. 2021). Soils serve as both sources and reservoirs for metals and other contaminants, making them an essential indication of environmental quality (Munyasya et al. 2022). Toxic metal poisoning of soil in emerging nations like Bangladesh has become a major environmental issue as a result of fast urbanization and industrialization (Bai et al. 2022). Many anthropogenic activities in urban areas including industry, power generation, mineral extraction, metal processing, waste spills, coal combustion, and disposal of wastes are responsible for hazardous metal emissions resulting in environmental pollution (Wei and Yang 2010; Li and Feng 2012; Martín et al. 2013; Pan et al. 2021). Although heavy metal emissions are lower in rural areas than in urban areas, brick kilns are a major source of pollution for uncontrolled flue gas emissions and waste disposal (Li et al. 2021). Around 17 billion bricks are being produced annually in Bangladesh by 6000 licensed brickfields and several illicit ones (Proshad et al. 2017; Abedin et al. 2019).

The brick kiln sector is regarded as a poor people’s enterprise in India, Pakistan, and Bangladesh producing 140, 100, and 50 billion bricks annually, respectively (Chowdhury et al. 2022). Fuel combustion of timber, recycled gasoline, coal, fuel oil, diesel, rubber, rubbish, and polymers can produce toxic metals in the kiln region. Toxic metals such as lead (Pb), zinc (Zn), nickel (Ni), mercury (Hg), copper (Cu), cadmium (Cd), arsenic (As), and iron (Fe) can be identified in brick kiln area soils (Sikder et al. 2016a, 2016b). The burning fuels in brick kiln promotes the release of particulates and greenhouse gases (Kamal et al. 2014; Mondal et al. 2017; Xu et al. 2021b). On a 12-h burning needed to manufacture 10,000 bricks, a conventional traditional brick kiln can generate 86 kg of particulates (Tissari et al. 2008). In Bangladesh, around four million tons of timber and coal are being utilized in brick kilns, resulting in lower crop production and increased global warming (Kua and Kamath 2014). Aside from that, the burning of fuels produces a variety of harmful chemicals, such as heavy metals all of which are harmful due to their durability and toxic effects (Zhang et al. 2022). Toxic elements may be spread by the dumping of bottom ash and the dissemination of fly ash surrounding the kiln area that eventually falls on the topsoil and pollutes the surrounding soils (Sikder et al. 2016a).

Brick kilns are typically found alongside arable grounds in Bangladesh’s countryside and suburban areas. In brick kiln areas, crops could be vulnerable due to metal toxicity produced for brick kiln operations, eventually posing major health risks to humans (Kumar et al. 2021). Although certain trace metals are important for plant growth, plants cultivated in soil contaminated with toxic elements affect plant development, productivity, and yields (Chibuike and Obiora 2014). Because the food web is one of the prime sources of metallic exposure to humans, metal poisoning can be caused by eating contaminated foods and can lead to serious health-related problems (He et al. 2021). Human exposure to pollutants has been linked to negative impacts on the hormonal, coronary, genital tract, and immunological systems as well as cognitive expression issues and skin blemishes (Muhammad et al., 2019). Toxic metals like Cd, Cu, As, Pb, and Ni have been associated with cancers of the blood, stomach, bladder, kidney, liver, and neurological system (Muhammad et al. 2019). Toxic metals have an impact on soil health because they prevent typical microbial activity and the breakdown of organic pollutants (Yang et al. 2022c).

There is a paucity of data on the metal pollution in soil and crops around brick kilns in Bangladesh (Sikder et al. 2016b; Sakib and Sultana 2017). It is critical to obtain samples at frequent intervals to determine the influence of metallic exhaust from brick kiln activities. Therefore, the present study was conducted in four brick kiln areas in Gopalpur upazila of Tangail district, Bangladesh. Identifying metal sources is a crucial stage in pollution mitigation and multivariate statistics and factor analysis have been frequently employed (Ravankhah et al. 2017; Yadav et al. 2019). Positive matrix factorization (PMF) based on error estimation diagnostics has shown to be more effective in finding numerous solutions and has been frequently utilized to locate metals sources in soils (Jiang et al. 2017; Cheng et al. 2020; Dash et al. 2020; Li et al. 2020). The health hazard model has been widely used to analyze the risk of a single element present in the soil, but the risk from several sources has been overlooked (Wang et al., 2020a, 2020b). The use of a PMF in conjunction with a health risk model to assess hazards from a variety of sources is an essential step in the management of the priority control source and, as a consequence, in the protection of the environment and human health (Guan et al. 2018; Jiang et al. 2020). Several indicators, such as the pollution load index, Nemerow pollution index, contamination factor, degree of contamination, potential ecological risk index, geo-accumulation index, Nemerrow integrated risk index (NIRI), and toxic units (TUs) have proven to be beneficial in distinguishing anthropogenic and natural causes of metal pollution along with quantifying the degree of metal pollution in soils (Chabbi et al., 2021; Wang et al., 2021a, 2021b). Hence, the primary objectives of the study were to (i) determine metal content in soil and rice samples around brickfields, (ii) to apply several multivariate statistical methods (correlation coefficients, non-metric multidimensional scaling analysis, and euclidean distance-based redundancy analysis), and identify critical sources with PMF model, and (iii) to calculate source based ecological risks and positive matrix factorization-human health risk (PMF-HHR) model to quantify the impact of various sources on human health hazards.

Materials and method

Study area, samples collection, and processing

The study area, Gopalpur upazila (24°33.5′N 89°55′E) is located in the Tangail district of central Bangladesh. The geographical location of the brickfields and sample directions for each brickfield for this study are depicted in Fig. 1. We have selected four brick kilns in Pankata, Konabari, and Norilla (Fig. 1). The study area is located in the rural region of Tangail district and rice is cultivated as the main crop here. About 200 samples (100 for soil and 100 for rice) were collected in April 2020 around the brick kiln. Surface soil (0–15 cm) samples were gathered in a zip-lock plastic bag, labeled appropriately based on their location, and then dried in the sun before being kept in a clean bag. Mature rice grains were collected, oven-dried followed by grinding, and stored in a zip-locked bag.

Fig. 1
figure 1

Location of the study area and the sampling sites of brick kiln area in Tangail district of Bangladesh

Full size image

Physicochemical properties and metal assessment techniques

Soil texture, soil reaction (pH), organic carbon (OC), and electrical conductivity (EC) were assessed as physicochemical properties for the present study. A soil-to-water ratio of 1:2.5 was used to assess the pH in soil samples (Ge et al. 2019). The suspension was kept to rest the whole night before the pH was determined. The pH was determined using a pH meter (Model: PB-10, Sartorius AG, Germany) in a 1:2.5 (w/v) combination of soil and deionized water (Wang et al. 2022a, 2022b, 2022c). To evaluate electrical conductivity, 5 g of soil was deposited in 50-mL polypropylene tubes. Then the tube was filled with 30 mL of distilled water. The lid was securely fastened and shaken for 5 min. After that, the EC was determined by an EC meter (Horiba D-52) (Islam et al. 2019). The USDA approach was utilized to assess the textural groups of soil (Santaga et al. 2021). The organic carbon in the soil was determined by an elemental analyzer (model type:vario EL III, Elen-emtar, Germany) (Tan et al. 2022).

Analytical grade reagent chemicals were employed to evaluate studied toxic metals. Milli-Q purified water was used in the solution formulations. For total metal analysis, in a closed Teflon tube, 0.5 g of soil sample was mixed with 1.5 mL HNO3 and 4.5 mL concentrated HCl. About 0.5 g of dried rice samples was digested in the microwave with 6 mL HNO3 and 2 mL H2O2. The digested soil and crop samples were transferred to a Teflon beaker and filtered through a DISMIC®-25HP PTFE filter using a syringe. Metals were determined by an inductively coupled plasma mass spectrometer (ICP-MS) (Proshad et al. 2021). The system was calibrated using Multi-element Standard XSTC-13 (SpexCertiPrep® USA) solutions. All test batches were examined and verified to see if the established internal quality controls (IQCs) were met. A tuning solution was applied to assess RSD < 5% before commencing the analytical procedure. The accuracy of the technique was confirmed using certified reference materials from NMIJ CRM 7303-Lake Sediment and INCT-CF-3–Corn Flour to rule out any batch-specific errors, and samples were examined twice (Table S1).

Source identification by positive matrix factorization model

The present study identified the sources of toxic elements in soil using the positive matrix factorization model which has been found effective in identifying sources. The PMF explains mathematical strategies for estimating the contribution of various sample sources based on their compositions (Guan et al. 2018; Wu et al. 2021). The EPA PMF 5.0 receptor model, designed by the US Environmental Protection Agency (EPA), is particularly useful because it does not need source profiles and weights all data using uncertainty (Liu et al., 2021a, 2021b; Wang et al., 2021a, 2021b, 2021c). The positive matrix factorization method aids in determining the contributions and profiles of various metal sources. In a nutshell, this approach divides the original dataset into two-factor matrices such as a source profile and source contribution matrix, as well as a residual matrix (Chai et al. 2021). The fundamental formula is as follows:

$$e_{ij}={\textstyle\sum_{k=1}^p}g_{ik}f_{jk}+x_{ij}$$
(1)

where eij is metal sources profile; p is the number of factors; f is metal species profile; g is the amount of mass characterized by an individual factor to sample; xij is the residuals of each metal species that may be determined using the objective function Q.

$$Q={\textstyle\sum_{i=1}^n}{\textstyle\sum_{j=1}^m}\left(\frac{x_{ij}}{u_{ij}}\right)^2$$
(2)

where uij is the uncertainty of the jth toxic metal in the ith soil sample. The method detection limit (MDL) of individual metal, metal content, and the supplied error fraction is used to calculate uncertainty (Wu et al. 2020). The uncertainty may be computed as follows if the concentration of metals exceeds the MDL:

$${u}_{ij}=\sqrt{{\left( Error\ fraction\times concentration\right)}^2+{\left(0.5\times MDL\right)}^2}\left(\textrm{Concentration}>\textrm{MDL}\right)$$
(3)

When the metal content is less than MDL, the degree of uncertainty can be calculated as follows:

$${u}_{ij}=\frac{5}{6}\times MDL\left(\textrm{Metal}\ \textrm{content}\le \textrm{MDL}\right)$$
(4)

Soil pollution status assessment

The estimation of metal toxicity in the soil can be identified by determining the contamination status of the individually studied metal. Ecological risk estimation is one of the valuable approaches to know metal toxicity in soils. For this purpose, several indices like pollution load index (PLI), contamination factor (CF), degree of contamination (Cd), geoaccumulation index (Igeo), potential ecological risk index (PER), Nemerow pollution index (PN), and NIRI and toxic units (TU) were assessed for present study, and details of these indices are shown in Table 1.

Table 1 Mathematical expressions and physical significances of environmental and ecological risk indices due to metal contamination in soils and rice
Full size table

Health risk assessment from soils by positive matrix factorization-human health risks (PMF-HHR) model

Based on the PMF values, the PMF-HRA model can compute HHR from various sources. The PMF-HHR model (source-oriented), which is based on projected source profiles generated from PMF, seeks to effectively determine HHR source contributions (Shen et al. 2021; Sun et al. 2021). There are two phases to using a combined model to estimate health hazards. First, the contribution of metal sources for in soil was assessed using the PMF model, and then health risks were calculated to describe the HHR from heavy metals (Xu et al., 2021a, 2021b).

$${C}_{jn}^l=\ast {C}_{jn}^l\times {C}_j$$
(5)

where \({C}_{jn}^l\) is the mass contribution of studied metal n from source l in sample j; \(\ast {C}_{jn}^l\)is the calculated contributions of individual metal; Cjis the metal concentration of sample j.

The average chronic daily intake (mg/kg/day) of studied metals from source l of element n in soil j may be computed using three exposure routes such as inhalation, dermal, and ingestion. In the present study, the potential carcinogenic risk and non-carcinogenic from studied metals in soil for the three exposure pathways were assessed. The CDIs (mg/kg/day) of metals (for adults and children) were calculated using Eqs. 6–8.

$${CDI}_{jn_{ing}}^l=\frac{C_{jn}^l\times IngR\times EF\times ED}{BW\times AT}\times {10}^{-6}$$
(6)
$${CDI}_{jn_{inh}}^l=\frac{C_{jn}^l\times InhR\times EF\times ED}{PEF\times BW\times AT}$$
(7)
$${CDI}_{jn_{dermal}}^l=\frac{C_{jn}^l\times SA\times AF\times ABS\times EF\times ED}{PEF\times BW\times AT}\times {10}^{-6}$$
(8)

Non-carcinogenic risk and carcinogenic risk are included in the HHR model, which is based on the USEPA (United States Environmental Protection Agency). The HQ is usually applied to identify non-carcinogenic risk for a given metal where RfD is the reference dose of a metal. The HI is the result of the overall potential non-carcinogenic risk for toxic metals and is expressed as the sum of the HQs.

$$HI=\sum {HQ}_{jn}^l=\sum \frac{CDI_{jn i}^l}{RfD_i}$$
(9)
$$THI=\sum HI$$
(10)

The cancer risks were assessed to identify the risk of developing any sort of cancer during one’s lifespan as a result of toxic metal exposure.

$$CR=\sum {CR}_{jni}^l=\sum {CDI}_{jni\kern0.5em }^l\times {SF}_i$$
(11)
$$TCRI=\sum CR$$
(12)

where CR represents carcinogenic risks, SF stands for slope factor, and TCRI is the total cancer risk index due to ingestion, inhalation, and dermal contact of studied metals. The definition and data of variables for the assessment of health risk to soil are presented in Tables S2 and S3.

Statistical analysis

To determine the ancestries of studied metals in soil and rice samples, the data was examined statistically using the SPSS 20.0 statistical software. The means and standard deviations of studied metal concentrations in soil and rice samples were assessed. Non-metric multidimensional scaling (nMDS) and Euclidean distance-based redundancy analysis were applied to identify the significant spatial heterogeneity among the sampling locations and significant variation in soil and rice samples, respectively. The correlation coefficient was characterized to identify the inter-relationship of studied metals. PMF model was applied by EPA PMF version 5.0 software.

Results and discussion

Physicochemical properties and metal concentration in soil and rice samples around brick kiln area

The physicochemical properties (soil texture, electrical conductivity, organic carbon, and pH) of soil samples were determined for the present study (Table 2). The physicochemical qualities of soils are widely recognized to be depending on land use types (Yang et al. 2022a, 2022b, 2022c). The soil texture and chemical property of the soil is helpful to assess the plant’s available soil water (Yang et al. 2022a, 2022b, 2022c). Sand, silt, and clay content in soils ranged from 34.00 to 74.00%, 9.10 to 51.60%, and 9.90 to 22.80%, respectively (Table 2). Soil texture plays an important role in metal bioavailability in soil microorganisms (Yang et al. 2021).

Table 2 Summary statistics of physiochemical properties and studied metal concentration in the soils and rice samples
Full size table

Soil pH affects the metal availability in soils and microorganisms (Nag et al. 2022). Controlling metal transfer and concentration in soil is heavily reliant on the pH of the soil (Xu et al. 2022). The pH value ranged from 5.48 to 8.01 in soils indicating slightly acidic to slightly alkaline (Mengistu et al. 2022). The pH of the soil was acidic owing to the generation of carbonic acid, and the pH fluctuation in the soil was controlled by the effect of brick kiln operation and ash disposal (Wang et al. 2022a, 2022b, 2022c).

The mean electrical conductivity value was 0.36 dS/m (Table 2). Soil electrical conductivity is a useful indicator for agricultural systems management. Based on the soil salinity classification, the soil is considered non-saline (0–2 dS/m) and acceptable for agriculture. Brick kiln-mediated contamination, wastewater discharges, and agricultural runoff may generate EC variations in soils (Kumar et al. 2021).

Soil organic carbon is a crucial component of vegetation development and climate change, as well as the global carbon balance (Tian et al. 2022). The percent mean (range) of OC in study area soils was 1.30 (0.15–4.31). The discharge of brick kiln wastage and application of organic manure result in a wide range of OC in soil (Yam et al. 2021).

Six studied metals Cr, Ni, Cu, As, Cd, and Pb concentrations in the soil around the brick kiln of Gopalpur ranged from 1.57 to 45.65, 3.01 to 52.17, 2.91 to 63.93, 3.99 to 18.44, 0.14 to 1.04 and 1.19 to 41.43 mg/kg, respectively (Table 2 and Fig. 2). The highest mean concentration was observed in Cu (25.77 mg/kg), and the lowest was found in Cd (0.29 mg/kg). Significant coefficients of variation (CV) from the harmful concentrations of metals (Cu = 361.20%, Ni = 350.16%, Pb = 149.80%, Cr = 128.58%, As = 24.11%, and Cd = 4.70%) indicated large variation. This means metal distributions with a greater variation are deemed to have a large variance of studied metals (Dai et al. 2022c). Several manmade factors may have resulted in a greater CV, according to earlier findings (Gu and Gao 2019; Lv 2019). The examined metal content in soils was compared to the threshold effect level (TEL), lowest effect level (LEL), severe effect level (SEL), probable effect level (PEL), average shell value (ASV), and toxicity reference value (TRV) based on soil quality standards (MacDonald et al. 2000). The mean concentration of Ni (19.81 mg/kg) was higher than LEL (16 mg/kg) and TEL (18 mg/kg) whereas mean content of Cu (25.77 mg/kg) was higher than LEL (16 mg/kg) and TRV (9 mg/kg). The average content of As (6.11 mg/kg) was higher than LEL (6.0 mg/kg), TEL (5.9 mg/kg), and TRV (1.9 mg/kg), respectively (Table 2).

Fig. 2
figure 2

Heavy metals concentration in soils and rice around brick kiln area in Tangail district of Bangladesh

Full size image

Heavy metal concentration in soils was compared with other studies (Table S4). Compared to present study, the concentration of Cu (23.82 mg/kg) and Pb (8.03 mg/kg) in Ashulia, Bangladesh (Haque et al. 2022), Cu (21.50 mg/kg) and Cd (0.27 mg/kg) in Jhenaidah, Bangladesh (Kumar et al. 2021), Cu (6.75 mg/kg) and Pb (11.78 mg/kg) in Aran-o-Bidgol City, Iran (Ravankhah et al. 2017), and Cr (0.15 mg/kg), Ni (0.05 mg/kg), Cu (0.008 mg/kg), Cd (0.01 mg/kg), and Pb (0.07 mg/kg) in Peshawar, Pakistan (Ishaq et al. 2010) were lower than the present study. Again compared to the present study, the contents of Cr (24.32 mg/kg), Ni (37.41 mg/kg), As (8.46 mg/kg), and Cd (2.82 mg/kg) in Ashulia, Bangladesh (Haque et al. 2022), Cr (61.43 mg/kg), Ni (25.88 mg/kg), As (6.88 mg/kg), and Pb (22.93 mg/kg) in Jhenaidah, Bangladesh (Kumar et al. 2021), Cd (0.64 mg/kg) and Pb (29.97 mg/kg) in Hathazari, Chittagong, Bangladesh (Chowdhury and Rasid 2021), and Ni (20.14 mg/kg) and Cd (0.79 mg/kg) in Aran-o-Bidgol City, Iran (Ravankhah et al. 2017) were higher than soils of brick kiln vicinity.

Toxic metals in the soils as well as their toxicity have a detrimental impact on plant growth, interacting with some critical metabolic activities (Liu et al. 2020b). Application of uncontrolled agrochemicals, brick kiln discharges, and untreated wastewater are all possible sources of excessive Cr in soil for the present study (Akter et al. 2019). The heightened amount of Ni in the soil could be attributed to the disposal of wastes, as well as the presence of Ni-containing substances (Covre et al. 2022). An elevated Cu level in the soil could be the result of vehicular emission, unregulated brick kiln operations, Cu-containing vehicle bodies, and applications of Cu-containing chemical fertilizers, insecticides, and pesticides (Ballabio et al. 2018). Arsenic, a well-known potential contaminant and irrigation with arsenic-contaminated waters, uncontrolled use of arsenic-containing fertilizers and pesticides, and contamination from brickfields can all contribute to elevated As levels in soils (Raju 2022; Viana et al. 2022). Cadmium is considered the most environmentally hazardous metal that can enter the environment from a variety of sources, including atmospheric emissions, paints, melting industries, and the application of agrichemicals (Ren et al. 2022; Yang et al. 2022f). Lead is regarded as the most poisonous metal and the accumulation of it could pose long-term risks to human health, and soil may be contaminated as a result of waste from the automobile sector, battery processing plants, and fuel combustion (Dai et al. 2022a).

Table 2 shows the concentrations of the studied hazardous metals (Ni, Cu, Cr, Cd, Pb, and As) in rice. Metal concentration in rice showed significant fluctuation, which might be attributed to climate differences, paddy growing period, species variation, and varied metal-collecting capacities of rice (Bhandari et al. 2020). The mean metal concentrations were compared to the FAO/WHO recommended maximum permissible limit (MPL) for foodstuffs (FAO/WHO, 2011) (Table 2). The concentration of Cr, Ni, Cu, As, Cd, and Pb in rice ranged from 0.46 to 16.39, 0.22 to 10.75, 0.43 to 9.15, 0.72 to 4.77, 0.001 to 0.37, and 0.22 to 1.35 mg/kg, respectively. According to the current study’s findings, the mean concentrations of Cr (4.55 mg/kg), As (1.87 mg/kg), Cd (0.07 mg/kg), and Pb (0.68 mg/kg) were found to be 1.96, 18.7, 1.54, and 6.89 times higher than the MPL indicating significant hazardous metal contamination in rice in the study region. The content of Cr in rice was also compared to earlier investigations conducted in Bangladesh. The concentration of Cr was reported to be higher than in other studies (Li et al. 2012; Rahman et al. 2013), but lower than in certain studies (Islam et al. 2015, 2016). The higher Cr content in rice, on the other hand, might be linked to the use of unbalanced pesticide applications and irrigation of agricultural land using untreated or inadequately treated wastewater (Salleh et al. 2020). Cr poisoning causes ulcers and skin disorders in those who are exposed to it regularly. It can also cause immune system malfunctions, lung cancer, and nerve tissue damage (Nawab et al. 2018). The arsenic levels in rice samples from all sampling sites were higher than the acceptable range (0.10 mg/kg) indicating unsuitable for human consumption. Arsenic-rich agrochemicals and arsenic-containing groundwater for irrigation applied to rice fields might be the source of increased As levels (Wang et al. 2022c). Arsenic is acknowledged as a “slow poison” causing nervous system disorders, kidney damage, and blood-related ailments (Roel et al. 2022). Cadmium content in 90% of rice samples was higher than MPL (0.05 mg/kg). Cd contamination in rice can result from the application of Cd-containing chemical fertilizers, pesticides, and insecticides (Sheng et al. 2022). Lead in rice may have accumulated as a result of fuel combustion in brick kiln areas (Yang et al., 2022a, 2022b, 2022c, 2022d, 2022e, 2022f).

ANOSIM tests, distance-based redundancy analyses (dbRDA), and correlation of coefficients in soil and rice

The global R in the ANOSIM tests represents the strength of the separation between samples. In this study, our results showed significant differences in heavy metal concentration between rice and soil samples (R = 0.37, P < 0.001) and also four sampling sites (R = 0.33, P < 0.001) which is further supported by ANOSIM tests (Table S5). We used nMDS to identify the heterogeneity of the metal concentration between soil and rice samples among the four sampling sites (Fig. 3 upper panel). Among the heavy metals, the concentration of chromium, nickel, and lead was found higher both in soil and rice. Furthermore, the concentration of heavy metals showed relatively higher in soil compared to rice samples, and ANOSIM revealed a three-fold higher variation for rice than soil (the global R was 0.12*** and 0.32*** for soil and rice, respectively) (Fig. 3 lower panel).

Fig. 3
figure 3

Non-metric multidimensional scaling (nMDS) analysis of heavy metals in soil and rice showing the significant spatial heterogeneity among the sampling locations. The significant difference between and among the samples were assigned by analysis of similarity (ANOSIM) with 9999 permutations. The variation of heavy metals in soil and rice samples (upper left) and spatial differences of heavy metals among the four locations (upper right), the variations of heavy metals concentration in each soil sample (lower left) and in each rice sample (lower right). The significant differences were considered as P<0.05*, 0.01**, and 0.001***

Full size image

We used distance-based redundancy analyses (dbRDA) of heavy metals to represent the total variation in heavy metals found in soil and rice samples. Our results showed that the first two axes of the dbDRA accounted for 58.7% variation between soil and rice while 55.1% variation accounted for four sampling sites as spatial scale (Fig. 4 upper panel). The vector overlay of correlated heavy metals for the two dbRDA axes. Furthermore, by comparing the results between soil and rice, dbRDA accounted for higher variation for rice than soil, for example, 89.7% variation accounted for rice while 63.1% for soil (Fig. 4 lower panel), demonstrating that the spatial variation of metals is more pronounced in rice than soil.

Fig. 4
figure 4

Euclidean distance-based redundancy analysis of heavy metals showing the significant variation in soil and rice samples (upper left), among the four locations (upper right), in soil samples (lower left), and in rice samples (lower right)

Full size image

A correlation coefficient was applied to analyze the interrelationships among the observed physicochemical parameters as well as studied metals in soil and rice samples. In our study, we used Spearman rank correlation to show the significant existing relationship among variables. Analyses showed that there were significant correlations between heavy metals and physicochemical variables (Fig. 5). Among relation, Cr- Ni, Cr-As, Cu-Ni, Cu-As, Pb-pH, pH-EC, pH-sand, and pH-silt showed a significant positive correlation in soil. Certain findings might be explained by the fact that improving certain soil characteristics increases metal adsorption in the soil (Li et al., 2022). In rice samples, several significant correlations were observed such as Cr-Ni (0.59), Cr-Cu (0.62), Cr-Cd (0.33), Cr-Pb (0.55), Ni-Cu (0.45), Ni-Cd (0.27), Ni-Pb (0.52), Cu-Cd (0.62), Cu-Pb (0.64), and Cd-Pb (0.36) (Table S6). Such substantial positive relationships between the metals investigated suggest that these metals contaminate rice resulting in soil pollution (Mlangeni et al. 2022).

Fig. 5
figure 5

Spearman rank correlation showing the significant relationships between heavy metals and physio-chemical parameters in soils. The significant differences were considered as *P < 0.05, **P < 0.01, and ***P < 0.001

Full size image

Critical sources assessment of metals by PMF model with associated model uncertainty

We applied the PMF model for source identification of Cr, Ni, Cu, As, Cd, and Pb in brick kiln area soils. Three candidate variables were chosen according to the constant and the lowest Q values, as well as the coefficient between projected and observed metal contents. To determine the best solution, the technique was repeated 20 times. The model’s anticipated metal contents were compared to observed quantities, and the performance was assessed using determination coefficients (Su et al. 2022). All of the selected elements had signal-to-noise ratios (S/N) (Cr: 8.0, Ni: 8.3, Cu: 9.8, As: 8.9, Cd: 10.0, and Pb: 8.8) greater than 2 and were classified as strong. The R2 values for Ni, Cu, Cd, and Pb were higher than 0.90 indicating that the model yield was strong and well described, suggesting that the model yield was high and well explained (Table 3). Two error estimation approaches (classical bootstrap (BS) and displacement of factor elements (DISP)) were employed to investigate the uncertainty of the PMF results.

Table 3 Summary of input data statistics of studied metals in PMF model (concentration units in mg/kg) and linear correlation between the measured and predicted values with factor profiles and contributions of three PMF-estimated factors for the contents of toxic metals in the soils of brick kiln area
Full size table

The results of the source contribution of studied metal in brick kiln area soils were presented in Table 3 and Fig. 6. Factor 1 (F1) was explained with Cd (77.1%). In general, the application of chemical fertilizers, particularly phosphoric fertilizer is reflected in Cd enrichment in agricultural soils (Wang et al., 2020a, 2020b; Xia et al., 2011). Organic manure, herbicides, and pesticides contain high quantities of Cd (Zhang et al. 2021). Uncontrolled chemical fertilizers and pesticides are being continuously applied in rice fields for intensive cultivation resulting in Cd contamination in soils in Bangladesh (Proshad et al. 2021). According to contamination factors (CF) of Cd in soils, the mean CF value was 2.08 indicating study area soils were moderately contaminated (1 ≤ CF < 3) (Table S8). Hence, F1 can be considered an agricultural practice.

Fig. 6
figure 6

Profiles and contributions of sources of the studied metals based on the PMF model in brick kiln area soil

Full size image

Cu was the most dominant factor in factor 2 (F2), with a factor loading value of 76.3%. As a result, Cu has been recognized as a cause of air pollution, with metal-containing vehicle bodies, and tire wear being the primary sources (Hou et al. 2019). Copper pollution is also caused by tire abrasion, lubricants, and vehicle component corrosion (Zhang et al. 2016). Studied brick kilns are located near the road and several vehicles are being transported every day. Again, there are many vehicles that are being used for bringing raw materials (especially soils) into the kiln to make bricks and load trucks to deliver bricks to customers. Factor 2 can result in vehicular emission.

Factor 3 was dominated by Pb (85.9%). Brick kilns utilize a lot of coal and rubber tires as fuel to make bricks in a non-scientific fashion, which has a negative impact on the environment (Sikder et al. 2015). Again, brick kiln coal ash or fly ash contains a considerable amount of Pb (Ishaq et al. 2010; Bisht and Neupane 2015). In our study of brick kilns, there used coal, timber, and tyre as the main fuel of the kiln to burn bricks. After burning of these materials, huge amounts of ash are produced which are dumped into the open environment and contaminate soils around brick kilns. Therefore factor 3 could be considered a fuel combustion source.

Factor 4 (F4) yielded high loading of Ni (76.5%). The Ni content (19.81 mg/kg) in the study area soils was lower than the background value (22 mg/kg) whereas CV was higher than 85% indicating the non-point source of contamination and the similar results were reported (Dong et al., 2019; Duan et al., 2020; Wang et al., 2021a, 2021b, 2021c). Several studies have observed that the majority of Ni in agricultural soils comes from parent material (Guan et al., 2019; Wang et al., 2020a, 2020b). Again, organic and chemical fertilizer-derived Ni concentration was frequently lower than background soil values (Lv 2019; Liu et al. 2020a). Based on the above discussion, factor 4 can be identified as a natural source.

The foundation for normalized contributions from four factors of examined metals in soils was distributed differently. The ranges of intensities for F1, F2, F3, and F4 were −0.19 to 4.61, −0.04 to 5.71, −0.18 to 5.44, and −0.19 to 7.17, respectively (Fig. S1). The major four sources of studied metals in brick kiln region soils were fuel combustion in brick kilns (30.91%), vehicular emission (24.44%), agricultural practices (22.59%), and natural sources from soil parent materials (22.06%). The contribution of anthropogenic activities to metal contamination in the soil (77.94%) was higher than that of natural sources (22.06%) implying that the PMF model’s output is significant and the model was well explained (Wang et al., 2021a, 2021b, 2021c).

The evaluation of metal sources in soils by applying the PMF model may be subject to uncertainty (Pan and Chen 2021). To solve this problem, the error estimation to contribute sources offered by the DISP and BS techniques was calculated to explain the validity of the model. DISP was deemed an appropriate first step for screening a solution to assess the robustness and reliability of the model findings. The factor swaps were also inversely related to the model’s rationality. There were no apparent swaps in DISP in this investigation, meaning that the PMF solution was well-specified and had few data errors (Wu et al. 2020). The percentage adjustment in Q (DISP %dQ) in this study was less than 1% suggesting DISP results were acceptable (Qiao et al. 2021). The coefficient (R2) for Ni (0.99), Cu (0.99), Cd (0.99), and Pb (0.99) of measured and forecasted metal content was greater than 0.90, indicating that the model results were significant. The observed values were reported to be greater than the projected values suggesting that predicted values had less dispersion than observed values (Fig. S2). For the bootstraps approach, about 20 runs with 53 Bootstrap random seeds were applied. Based on the BS mapping, the bootstrap factor was allocated (R2 > 0.60) to the base factor from 1 to 4 for 100, 100, 94, and 87%, respectively suggesting that the PMF results are robust. The BS and DISP analyses were used to determine the variability of source contributions. The high uncertainty in the source profile produced from the PMF model is frequently visible in these box plots. The base run values of some studied metals like Cr and As and Pb in factor 2 were outside of the interquartile range originating from the BS analysis (Fig. S3). These findings might be resulted from random errors and also suggest that some observations might have a major influence on the PMF solution (Yuanan et al. 2020). The 5th and 95th percentile values for all BS resamples served as the uncertainty interval endpoints for BS, whereas the minimum and maximum values in all displacements served as the DISP interval endpoints. The interval ratios, as a crucial indication, were derived by dividing the interval range of the uncertainty by the corresponding mid-point in order to compare the error estimation findings of studied metals. For individual metals, the interval ratios of BS and DISP followed a similar pattern, showing that the results of uncertainty were consistent (Fig. 7). Studied metals with large interval ratios (around or equal to 2) indicated that there was a lot of uncertainty. Cr, Ni, and Pb in factor 1; Ni and Cd in factor 2; Cr, Ni, Cu, and As in factor 3; Cu, As, Cd, and Pb in factor 4 showed significant uncertainty (Fig. 6). Despite the model doing a good job of measuring the impact of studied metal pollution sources, there is still a lot of uncertainty, especially for metals with low contribution percentages.

Fig. 7
figure 7

Interval ratios of the uncertainty in the estimated contributions of the four factors derived from PMF based on BS and DISP analyses

Full size image

Bioconcentration factor (BCF) of metals in rice from soil

The goal of the BCF was to determine the number of metals that are transferred from soil to rice. Various edaphic (soil physicochemical properties) and plant-related variables influence the accumulation of metals in rice (Gupta et al. 2019). The mean BCF values were in the decreasing order of Cr (0.47) > As (0.47) > Cu (0.43) > Ni (0.30) > Pb (0.30) > Cd (0.12) (Table S7). The range of BCF for Cr, Ni, Cu, As, Cd, and Pb were 0.04–1.97, 0.02–1.07, 0.04–1.33, 0.04–1.91, 0.00–0.98, and 0.02–2.28, respectively. According to studied metals, Cr and As showed higher BCF values suggesting high mobility of these metals from soil to rice. Chromium might have originated in soils due to weathering of parent materials as the average Cr concentration was lower than the background level. Noteworthy sources of As in agricultural soils have been documented to include the use of enhanced pesticides and fertilizers as well as As-contaminated groundwater for irrigation purposes (Dai et al. 2022b).

Soil pollution status and potential ecological risks of toxic metals

The contamination factor (CF) depicts the distribution of toxic elements and avoids the need to compare them at different magnitudes (Gupta et al. 2021). Based on the contamination factor classification, only Cd (2.08) showed moderate contamination (1 ≤ CF < 3) and other studied metals showed low contamination (CF < 1) (Table S8). The range of degree of contamination (Cd) was 1.49–14.01 with a mean of 5.46 suggesting moderate contamination (5 ≤ Cd < 10) in soil (Proshad et al. 2019) (Table S8 and Fig. 8A).

Fig. 8
figure 8

A Degree of contamination (Cd), B Pollution load index (PLI), C Potential ecological risk (PER), D Nemerow pollution index (PN), and E NIRI in soils of brick kiln soil

Full size image

Geoaccumulation index (Igeo) can be assessed to evaluate the level of pollution status caused by hazardous metals (Ren et al. 2022). This index is frequently used for assessing soil contamination (El Fadili et al. 2022). The Igeo value of Cr, Ni, Cu, As, Cd, and Pb ranged from −2.45 to −0.25, −1.93 to 0.30, −1.83 to 0.27, −1.95 to 0.24, −1.51 to 1.23, and −2.30 to 0.46, respectively. Based on mean Igeo, studied metals were uncontaminated (Igeo < 0) (Table S9).

The pollution load index is an integrated method of the studied metals that were computed to evaluate the quality of soil in terms of the comprehensive contamination level of heavy metals (Luo et al. 2022). The average PLI value was 0.66 assuming baseline level pollution (PLI < 1) (Table S10 and Fig. 8B).

The potential ecological risk is the integration of ecological risk for metals with a toxic response factor (Singh and Chandel 2022). The range of PER was 35.99–375.09 (low–very high risk) with a mean value of 179.79 (Table S10 and Fig. 8C). Based on PER classification, the study area soil showed considerable risk (130 ≤RI < 260).

The Nemerow pollution index (PN) was applied to assess overall contamination in soil (Yang et al., 2022a, 2022b, 2022c, 2022d, 2022e, 2022f). The PN value of the study area soil was 1.85 indicating moderately polluted (2.0 ≤ PN < 3.0) based on PN classification (Table S10 and Fig. 8D).

The NIRI can solve the errors in the Nemerow pollution index (PN) and ecological risk index (RI), such as the impact of the number of studied metals on the RI’s value and the failure to account for differences in toxic response factors of metals in the PN evaluation (Wang et al. 2018). The NIRI can offer a more precise assessment of the combined influence of several heavy metal effects. It also introduces the hazardous response factor to distinguish heavy metal impacts (Men et al. 2020). The difference between the PN and the NIRI confirmed that metal poisoning has a significant influence on risk levels (Shi et al. 2018). The toxicity of studied metals varies widely, impacting the degrees of risk associated with metals. The toxic response factor should be added to highlight the disparities in heavy metal features that are overlooked by indices like PN (Gao et al. 2019). In comparison to the PN, the NIRI takes into account not only heavy metal concentrations but also variances in toxic response factors across studied metals (Cheng et al. 2015). In the present study, 3.33, 46.66, and 43.33% of soil sampling sites showed very high, high, and considerable risks (Fig. 8E). The calculated mean value of NIRI was 166.35 and ranged from 34.17 to 340.97 (Table S10). Based on the NIRI classification study area, soil showed high risk (160 ≤ NIRI < 320).

Quantification of health risk from the soil by PMF-HHR model

The human health hazards due to exposure to studied metals from soils were quantified using a PMF-based HHR model with several sources. The non-carcinogenic risks (HQ) and carcinogenic risks (CR) were calculated with four distinct sources computed by the PMF model with dermal contact, ingestion, and inhalation exposure pathways (Table 4 and Fig. 9A, B, and C).

Table 4 Chronic daily intake (mg/kg), non-carcinogenic risks (HQ), hazard index (HI), and cancer risks (CR) obtained from positive matrix factorization-human health risks (PMF-HHR) model
Full size table
Fig. 9
figure 9

A Chronic daily intake, B non-carcinogenic risks (HQ), and C carcinogenic risks of metals due to ingestion, inhalation, and dermal contact of soils

Full size image

The hazard quotient (HQ) values in adults were 9.40E-02 (factor 1), 1.74E-01(factor 2), 1.81E-02 (factor 3), and 1.42E-02 (factor 4), respectively whereas in children were 1.03E-01 (factor 1), 1.15E-01 (factor 2), 1.18E-01 (factor 3), and 9.29E-02 (factor 4), respectively (Table 4 and Fig. 9B). Based on the PMF-HHR model, the non-carcinogenic risk values (THIs) of adults and children were less than one (Table 4). The recommended standard value of THI is one. When the THI value is lower than 1, there do not exist any non-carcinogenic risks (Xu et al., 2021a, 2021b). There were no evident non-carcinogenic hazards for individuals in the study region. There were similar trends in non-carcinogenic risk for adults and children, according to the findings. The exposure pathways that posed non-carcinogenic risks were in the decreasing order of ingestion > dermal contact > inhalation (Table 4). According to the findings, heavy metals are ingested orally, which is the most priority route to non-carcinogenic risks in soil, which is consistent with other research (Sun et al., 2021; Wang et al., 2020a, 2020b; Xu et al., 2021a, 2021b). According to computed factors, the non-carcinogenic risks in adults were dominated by vehicular emission (58%) whereas in children were influenced by vehicular emission (27%) and fuel combustion (27%), respectively (Table S11). The sources for metal contamination in adults were in order of vehicular emission (58%) > agricultural practices (31%) > fuel combustion (6%) > natural source (5%) and in children were vehicular emission (27%) > fuel combustion (27%) > agricultural practices (24%) > natural source (22%). Based on the above discussion, anthropogenic sources were the major priority for posing non-carcinogenic risks in the study area soils.

The carcinogenic risk (CR) values for adults in the whole research region were within the acceptable range for every source calculated with the PMF model but children posed carcinogenic risks. The CR value for children were 1.83E-04 (factor 1), 1.73E-04 (factor 2), 2.30E-04 (factor 3), and 1.71E-04 (factor 4) higher than the acceptable range (1.00E-04) indicating possible cancer risks in children (Table 4 and Fig. 9C). The carcinogenic risks in children from four individual sources were much greater than those of adults, which can be related to child-specific physiological and behavioral characteristics, like a lot of object-to-mouth and hand-to-mouth contact (Shen et al. 2021). As a result, children should be taught to pay attention to personal cleanliness, particularly regular hand washing to avoid any health hazards. Furthermore, the TCRI values for every factor follow the same pattern, with the greatest values due to ingestion, followed by dermal contact and inhalation (Table 4). The cancer risks in adults were highly contributed to fuel combustion (34.3%) followed by agricultural practices (27.33%), vehicular emission (25.78%), and natural source (12.6%) whereas in children were mostly characterized by fuel combustion (30.38%) followed by agricultural practices (24.17%), vehicular emission (22.85%), and natural sources (22.59%) (Table S11). Consequently, fuel combustion and agricultural practices were the major sources of carcinogenic risk in brick kiln area soils according to the present study. As a result, fuel combustion and agricultural practices must be prioritized in order to safeguard human health from carcinogenic risk. Agricultural activities and fuel combustion category produced much higher health hazards to adults and children than other sources. This suggests that the source of pollution that results in the most health hazards is not always the most common source of metals, which is likely owing to more toxic elements such as Cr and As, which cause high risks (Jiang et al. 2020). Fuel combustion like timber, tyre, and coal should be replaced with safe materials in the study area brick kiln. Agricultural practices, such as the use of fertilizers, pesticides, and herbicides, should be prioritized and managed across the research area to prevent and decrease human health concerns.

Health risks from rice

The chronic daily intake (CDI) of Cr, Ni, Cu, As, Cd, and Pb in adults were 4.34E+00, 2.43E+00, 7.12E+00, 8.52E-01, 3.50E-02, and 1.77E+00 (mg/kg/day), respectively, and in children were 5.71E+00, 3.20E+00, 9.38E+00, 1.12E+00, 4.61E-02, and 2.33E+00 (mg/kg/day), respectively (Table S12). The CDI of Cr, Ni, As, and Pb in adults were 4.43, 4.86, 8.52, and 8.85 times higher than the maximum allowable concentration (MAC) whereas in children were 5.71, 6.4, 11.2, and 11.65 times higher, respectively. The tendency of CDI in rice was as follows: Cu > Cr > Ni > Pb > As > Cd. The non-carcinogenic risk (HQ) of As in adults (1.40E+00) and children (1.84E+00) was higher than 1 suggesting high non-carcinogenic risks due to exposure to As in rice. Again, the hazard index (HI) values for adults and children were 1.83E+00 and 2.41E+00, respectively (Table S12). As the HI values were higher than unity, there was a chance that the consumption of rice posed a health risk and the rice investigated was found to be unfit for human intake (Liu et al. 2021a, 2021b). The carcinogenic risk values for adults were Cr (8.78E-02) and As (6.30E-04), respectively, while those for children were Cr (1.15E-01) and As (8.30E-04), respectively. The cancer risk value of Cr and As reported higher than the standard limit (1.00E-04) indicating potential carcinogenic risks posed due to the consumption of rice.

Conclusions

The levels of contamination and sources of six toxic metals in soil and rice around the brick kiln area were studied in this study. The PMF and PMF-HHR model identified four different sources of studied metals, and health hazards were assessed posed by individual sources. The findings revealed that toxic metals were released into the environment resulting in unregulated operations in brick kilns and agricultural activities causing metal contamination in soil rice. Arsenic and cadmium content soil was higher than the background values suggesting Cd and As contamination in the soil. Multivariate statistical analysis revealed that Cd, As, and Cu in soil contributed 46.05% of the total variance and originated due to several anthropogenic sources. The calculated several metal pollution indices such as degree of contamination and PN showed moderate contamination whereas PER and NIRI showed the considerable and high risk to the soil. Four sources were computed by the PMF model named fuel combustion (30.91%), vehicular emission (24.44%), agricultural practices (22.59%), and soil parent materials (22.06%). Health hazards were assessed with the PMF-HHR model due to inhalation, dermal contact, and ingestion of metals from soils. The non-carcinogenic risks in adults were dominated by vehicular emission (58%) whereas in children were influenced by vehicular emission (27%) and fuel combustion (27%), respectively. The cancer risks were highly contributed to fuel combustion for adults (34.3%) and children (30.38%). The carcinogenic risk value for individual sources computed by the PMF model was higher than the standard value (10−4) for both adults and children suggesting potential cancer risks for ingestion, inhalation, and dermal contact in soil. For rice consumption, As posed serious non-carcinogenic health hazards whereas Cr and As resulted in potential cancer risks. In conclusion, this study presented an effective method for quantifying risk apportionment, which is critical for pollution control and risk reduction.