Multi-Elemental Characterization of Soils in the Vicinity of Siderurgical Industry: Levels, Depth Migration and Toxic Risk

Abstract

The assessment of soil contamination in the vicinity of integrated siderurgical plants is of outmost importance for agroecosystems and human health, and sensitive techniques should be employed for accurate assessment of chemical elements (metals, potential toxic elements, rare earths, radioelements) in soil and further evaluation of potential ecological and safety risk. In this paper a total of 45 major, minor and trace elements (Al, As, Au, Ba, Br, Ca, Cd, Ce, Co, Cr, Cs, Cu, Dy, Eu, Fe, Hf, Hg, I, K, La, Mg, Mn, Mo, Na, Nd, Ni, Pb, Rb, Sb, Sc, Sm, Sn, Sr, Ta, Tb, Th, Ti, Tm, U, V, W, Y, Yb, Zn and Zr) were quantified in soils located around a large siderurgical works (Galati, SE Romania) using instrumental neutron activation analysis (INAA) in combination with X-ray fluorescence (XRF) and inductively coupled plasma mass spectrometry (ICP–MS). The statistical analysis results and vertical distribution patterns for three depths (0–5 cm, 5–20 cm, 20–30 cm) indicate inputs of toxic elements in the sites close to the ironmaking and steelmaking facilities and industrial wastes dumping site. For selected elements, a comparison with historical, legislated and world reported concentration values in soil was performed and depth migration, contamination and toxic risk indices were assessed. The distribution of major, rock forming elements was closer to the Upper Continental Crust (UCC), and to the Dobrogea loess, a finding confirmed by the ternary diagram of the incompatible trace elements Sc, La and Th, as well as by the La to Th rate. At the same time, the La/Th vs. Sc and Th/Sc vs. Zr/Sc bi-plots suggested a felsic origin and a weak recycling of soils’ mineral components.

1. Introduction

Soil is a vital resource which provides essential support to humans and ecosystems. Industrial activities, such as metal smelting, steel production, mining, the chemical industry and coal-fired power production, may cause soil quality degradation and serious pollution of the environment with toxic trace elements which could negatively impact human health due to their accumulation, persistence and transfer in the compartments of the food chain [1,2,3,4,5,6,7]. The elemental analysis at trace level is an important analytical task in various laboratories and high precision techniques, such as Instrumental Neutron Activation Analysis (INAA) [6,7,8,9,10], X-ray Fluorescence (XRF) [4,8,10,11] and inductively coupled plasma mass spectrometry (ICP–MS) [2,8,12,13,14], have shown to be powerful tools for multi-elemental analysis of a variety of environmental matrices, including soil and rock material, as well as for the assessment of mineralogical and geochemical features and spatio-temporal contamination trends of a certain region.

This work aimed to apply INAA, XRF and ICP–MS in order to examine the soil multi-element load, level of soil pollution with Potential Contaminant Elements (PCEs) and their ecological impact around the largest integrated ferrous (iron and steel) metallurgical plant (ISP) in Romania located at Galati town, SE Romania, which is one of the most important integrated complexes in the South-East of Europe (Figure 1) [11,15,16], with a great potential for environmental contamination and health damage (neurological disorders, asthma, cancer, cardiopulmonary disease, allergies, organ damage, hormonal imbalance, anemia, etc.) due to specific emissions from iron smelting, processing of ores, coal and auxiliary materials, agglomeration and sintering, minerals and scrap transportation, casting, molding, steelmaking, etc. [3,5,17]. The activity of this large ferrous metallurgical facility started in 1965, is continuing today, but in the last decade some of the units (blast furnaces, steelworks and mills) were closed and slag dumping has stopped [18].

The aim of this paper was two-fold: (1) to study the soil mineralogy, spatial and depth distribution pattern of 45 major and trace elements in soils adjacent to the siderurgical industry; and (2) to assess the ecological risk of selected PCEs, which might produce serious health disorders.

The work was carried out as a continuation and extension of pollution and risk assessment in previous studies limited to soil contamination with heavy metals [6,11,15] and persistent organic pollutants (POPs) [16] in the Galati industrial area and SE Romania in the period 2005–2009, and in connection with other international projects focused on the determination of toxic compounds and assessment of ecological and human health risk in the Black Sea Basin [3,14,19,20].

2. Materials and Methods

2.1. Sampling and Sample Processing

Composite soil samples of about 1.5 kg were collected at nine sites belonging to several territorial-administrative units (TAUs) of Galati (GL) and Braila (BR) counties in the South–East Development Region of Romania, around a large steel plant in SE Romania (Figure 1), from three different layers at depth intervals of 0–5 cm, 5–20 cm and 20–30 cm. The samples labeled as G1 were taken from the village of Vadeni (TAU Vadeni), Braila County, and the other samples (G2, G3, G4) from Galati County, from the localities of Sendreni and Smardan TAUs and Galati town. A control sample (GC) was collected from Vanatori village, in the northern part of Galati town (Figure 1). The site description is presented in

Table 1. Site description.

Code	Longitude	Latitude	Description
G1.1	45°23′00.4″	27°57′46.2″	TAU Vadeni, BR, rural, S of ISP
G1.2	45°22′22.0″	27°56′40.6″	TAU Vadeni, BR, rural
G1.3	45°23′03.7″	27°54′56.8″	TAU Vadeni, BR, agricultural
G2.1	45°25′53.8″	27°55′28.4″	TAU Sendreni, GL, agricultural, W of slag dump
G2.2	45°25′02.0″	27°56′31.6″	TAU Sendreni, Movileni, GL, rural, SW of slag dump
G2.3	45°24′18.2″	27°58′26.1″	GL town limit, peri-urban, S of ISP
G3.1	45°29′39.6″	27°55′17.0″	TAU Smardan, Mihail Kogalniceanu, GL, rural, N of ISP
G3.2	45°28′27.0″	27°57′13.7″	TAU Smardan, GL, agricultural, N of ISP
G4.1	45°25′14.1″	28°01′31.3″	GL town, urban, E of ISP
GC	45°31′20.5″	27°59′41.6″	TAU Vanatori, GL, control rural site

. The importance of the targeted region also derives from its closeness to the Danube River, the confluence of Danube with its main tributaries the Siret and Prut Rivers, and proximity to the ecologically valuable natural sites and special protected areas in Romania, Republic of Moldova and Ukraine in the Lower Danube River and Black Sea basins [16,19,21].

The soil samples were taken from homogeneous, undisturbed soils, belonging to the classes: (a) Cernisols, Chernozems type, calcareous subtype—developed on parental materials of aeolian origin (loess and loess deposits), and (b) Protisols, Alluviosol type, calcareous and gleic mollic limestone subtypes—developed on carbonate fluvial parent materials. The general characteristics of soils in the target region are given in

Table A1. General characteristics of soils in the investigated region.

Location	pH	OC (%) *	Clay (%)	Soil Type (WRB and SRTS) [23,24]	Soil Class (SRTS) [23]
TAU Smardan [21]	8.00–8.37	0.86–2.74	12.26–19.97	Calcaro-calcic Chernozem	Cernisols
TAU Sendreni [21,22]	7.88–8.44	0.91–1.66	1.15–27.98	Calcaro-calcic Chernozem	Cernisols
TAU Vadeni [21,22]	8.18–8.21	0.69–1.45	19.20–33.87	Calcaric and Calcaric Mollic Gleic Fluvisols	Protisols
Galati town [11,16]	8.40–8.84	1.37–2.47	13.00–20.00 **	Calcaro-calcic Chernozem	Cernisols

* Considering OC (%) = OM (%)/1.7241 (OC—organic carbon, OM—organic matter); ** References [25,26].

, Appendix A [11,16,21,22,23,24,25,26]. In the studied area, the soils do not show variability within very wide limits, regarding their delimitation at the level of type and subtype. On the territories of Smardan, Sendreni and Galati, located in the Covurlui Plain, the soil samples were taken from agricultural lands located in the interfluvial area, without considerable slope variations. On the Vadeni territory, geographically located in the Lower Siret Plain, the samples were collected from the alluvial plain. From the climatic point of view, the territories mentioned are located in the temperate-continental climate zone, with excessive nuances and variations during the year regarding the distribution of temperatures and precipitation. The sampling was performed from three layers in the first 30 cm of soil in order to observe the migration of elements in deeper layers and compare the actual contamination status with historical data.

The collected soil samples were prepared for various analyses at INPOLDE research center, Dunarea de Jos University, Galati, Romania, cleaned from stones, vegetal material and other debris, dried at room temperature, then grounded and sieved to a granulation of 0.01 mm.

2.2. Analytical Techniques

Two non-destructive multi-elemental nuclear analytical techniques commonly applied in geochemistry studies and materials certification programs were employed in this research: epithermal neutron activation analysis (INAA) and X-ray fluorescence (XRF) analysis. The two analysis techniques were used to determine with high accuracy and precision the total concentrations of 44 major, minor and trace elements (INAA—Na, Mg, Al, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Co, Zn, As, Br, Rb, Sr, Zr, Nb, Mo, Sb, I, Cs, Ba, La, Ce, Nd, Sm, Eu, Tb, Dy, Tm, Yb, Hf, Ta, W, Au, Hg, Th and U; XRF—Cu, Pb, Sn, Y) in the soil samples taken around a large steel plant in SE Romania. QA/QC was performed using certified reference materials having a similar matrix as the samples. Due to unsuitability of INAA and XRF for quantification of Cd at very low concentrations in soil, the ICP–MS technique was employed in this work for Cd analysis.

The instrumental neutron activation analysis (INAA) analytical technique was applied at the REGATA installation of the IBR–2M reactor within the Frank Laboratory of Neutron Physics (FLNP), of the Joint Institute for Nuclear Research (JINR) Dubna, Russian Federation. Several channels of the IBR–2M reactor were used, which allowed irradiation of the samples with thermal and epithermal neutrons. Soil samples weighing around 100 mg, together with certified reference materials, were wrapped in polyethylene bags and aluminium foils for short and long-term irradiation, respectively, according to the analytical scheme described in other work [27,28,29]. The induced radioactivity in the samples was measured with the aid of gamma spectrometric chains equipped with an automatic switch system for soil and standard material samples [27]. The gamma spectra were processed using the software developed at JINR [28] and the analysis was performed based on the radionuclides formed in the samples during activation with neutrons with various energies, listed in [27].

For the application of XRF spectrometric analysis with energy dispersion (EDXRF), the prepared samples were put in specific capsules coated with a Myler foil. The encapsulated samples were irradiated for 120 s in SOIL mode using a portable Genius XRF spectrometer manufactured by Skyray Instruments Inc., equipped with a large area and detector with Be window and an excitation source of 40 kV/100 µA miniature X-ray tube with an Ag-target [14,18].

ICP–MS was applied for cadmium determination with the aid of an Agilent 7700Χ ICP–MS (Agilent, Santa Clara, CA, USA) spectrometer from the International Hellenic University (IHU), Kavala, Greece, a partner laboratory of the INPOLDE center within the framework of the MONITOX network [20], as described in a previous study [14]. The solid homogenized soil samples were weighed via the analytical balance (≈3 g), and 30 mL of strong HCl and HNO₃ mixture at a ratio of 1:3 (aqua regia) was added. The samples were put in a Berghof MWS-2 microwave oven (P = 2000 W) using the following ramping parameters: T = 200 °C, P = 90%, t = 25 min. After digestion, the obtained solutions were filtered, transferred into graduated flasks (100 mL) and diluted with 5% HNO₃ solution to the filling mark. A calibration curve with five points (R² = 0.997) was built with the aid of stock standard solutions of different concentrations. All samples were analyzed in triplicate. Using an independent certified standard, a 105% recovery of Cd was obtained.

2.3. Depth Migration Index

In order to qualitatively assess the degree of migration/mobility in soil of the determined chemical elements, the depth migration index (DMI) was calculated for each element using the Equation (1), adapted by us after [30], for all the depths from which samples were taken (from 0 to 30 cm):

$$\mathrm{DMI}=\sum _{\mathrm{i}=1}^{\mathrm{n}}\left(\frac{{\mathrm{c}}_{\mathrm{i}}}{{\mathrm{c}}_{\mathrm{T}}}\right){\mathrm{D}}_{\mathrm{i}}$$

(1)

where n = 3 represents the number of layers from which the samples were taken, c_i is the value of the chemical element concentration in each sampled layer i, $c_T=\sum _{i=1}^n{c}_i$ is the sum of the values of the chemical element concentrations from all the depths of the sampled layers, and D_i is the depth (in cm) of the lower limit of each sampled layer (5, 20 or 30 cm). The value of this index ranges from 1 (if the element accumulated entirely in the first cm) to 30 (if the element accumulated entirely between 29 and 30 cm). The migration/mobility potential is classified into four categories, labelled as: A (DMI < 5), very low; B (5 < DMI < 10), moderate; C (10 < DMI < 20), high; D (DMI > 20), very high [15].

2.4. Contamination and Toxic Risk Indices

In order to describe the pollution extent and risk in a region impacted by multi-pollutants, several single and complex risk indices have been developed [2,12,31,32,33,34,35].

In this paper, we only used single and multi-element contamination indices, as well as the ecological risk index to highlight the impact of selected PCEs on the industrial soil quality and surrounding ecosystem state. Nevertheless, to assess the toxicity of PCEs in environmental compartments and for better understanding the ecotoxicological danger, these evaluations could be completed with bioassays [4,35] and health risk assessments [5].

Contamination factor (CF) for an individual chemical element is defined by Equation (2) as the ratio of each element concentration in the soil sample to the element concentration in background soil [33]:

$$\mathrm{CF}=\frac{{\mathrm{c}}_{\mathrm{sample}}}{{\mathrm{c}}_{\mathrm{background}}}$$

(2)

where c_sample and c_background are the measured concentration and the background concentration value of metal i, respectively. The contamination classification of soils using this index is: CF < 1, low; 1 ≤ CF < 3, moderate; 3 ≤ CF < 6, considerable; CF ≥ 6, very high [2,31,32,33].

To evaluate the pollution levels of various contaminants specific to a site, the site Pollution Load Index (PLI) was employed, utilizing Equation (3) as introduced by Tomlinson et al. (1980) [34]. This equation involves computing the nth root of the product of the highest individual contamination factors ${\mathrm{CF}}^{\mathrm{i}}$, calculated for specific metals (PCEs) which might result from siderurgical industry activity (n = number of elements/contamination factors). In this work we considered n = 13 PCEs, chosen from those specified in Romanian legislation [36] and probably emitted during the iron and steel making and steel products manufacturing processes, e.g., Hg, Cd, As, Pb, Cu, Co, Ni, Cr, Mn, Zn, Fe, V and Sb. The site PLI categories are the following: PLI ≤ 1, unpolluted (Background Pollution); PLI ≥ 1, polluted.

$${\mathrm{P}\mathrm{L}\mathrm{I}=\left({\mathrm{CF}}^1{\times \mathrm{CF}}^2{\times \dots \dots \dots \times \mathrm{CF}}^{\mathrm{n}}\right)}^{1/n}$$

(3)

Similar to the definition of the site PLI and classification, a Regional Pollution Load Index (RPLI) is defined in this work with the aid of Equation (4), as the mth root of the product of the individual site average PLIs [34], calculated for the m = 9 industrial sites (except for the control site GC) in the Galati ISP area,

$${\mathrm{R}\mathrm{P}\mathrm{L}\mathrm{I}=\left({\mathrm{PLI}}^1{\times \mathrm{PLI}}^2{\times \dots \dots \dots \times \mathrm{PLI}}^{\mathrm{m}}\right)}^{1/\mathrm{m}}$$

(4)

Ecological risk factor ($E_r^i$) for an element i is defined with the aid of Equation (5):

$${\mathrm{E}}_{\mathrm{r}}^{\mathrm{i}}={\mathrm{T}}_{\mathrm{r}}^{\mathrm{i}}{\times \mathrm{CF}}^{\mathrm{i}}$$

(5)

where the toxic response factors ($T_r^i$) for the selected toxic trace elements, indicated in the literature studies, are the following: Hg–40 [31,32,33,37], Cd–30 [31,32,33,37], As–10 [2,31,32,33,37], Pb–5 [2,31,32,33,37], Cu–5 [2,31,32,33,37], Co–5 [32], Ni–5 [2,31,32,37], Cr–2 [2,31,32,33,37], Mn–1 [2,32], Zn–1 [2,31,32,33,37]. The risk classification using this individual ecotoxicological index is: $E_r^i$ < 40, low potential ecological risk; 40 ≤ $E_r^i$ < 80, moderate potential ecological risk; 80 ≤ $E_r^i$ < 160, considerable potential ecological risk; 160 ≤ $E_r^i$ < 320, high potential ecological risk; $E_r^i$ ≥ 320, very high potential ecological risk [31,37].

The risk index RI is the sum of calculated ecological risk factors corresponding to all n = 10 hazardous elements analyzed, according to Equation (6):

$$\mathrm{R}\mathrm{I}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{E}}_{\mathrm{r}}^{\mathrm{i}}$$

(6)

The risk classification using this complex ecotoxicological index is: RI < 90, low; 90 ≤ RI < 180, moderate; 180 ≤ RI < 360, strong; 360 ≤ RI < 720, very strong; RI ≥ 720, highly strong [31].

2.5. Mapping and Statistical Data Analysis

Principal Component Analysis (PCA), Discriminant Analysis (DA), and Cluster Analysis (CA), i.e., Joining or Tree Clustering, were applied to examine the relationship among the sampling sites and elemental load for each soil layer. The inputs consisted of 45 variables representing the concentrations for all 45 chemical elements, which were determined in soils collected from nine industrial sites surrounding the enterprise and one control site (GC).

Given the high number of elements of which mass fractions were determined by INAA, we have used Principal Component Analysis (PCA) [38] due to its capacity to perform a linear dimensionality reduction. This technique permits the organizing and plotting of a great volume of independent data in more two-dimensional scatterplots, thus visualizing the clusters of closely related data. Consequently, all data were classified into cases, in our case the sampling points, and variables which, in the present study, represent the mass fractions of the considered elements distributed in each case, i.e., sampling point.

PCA can be performed in Q or R mode, depending on what needs to be grouped, variables, or cases in clusters. In the case of R mode, samples, in our case sampling points, are classified using the mass fractions of all 45 elements as variables.

Closer to PCA, DA also represents a classification technique [39] used when a priori the number of clusters or populations in our case the soil levels, are known. In this way, the variables that consist of independent variables, i.e., the PCE mass fraction values and the dependent variables, in this case, the soil layers, generate a bi-plot that illustrates several clusters, whose number is equal to the number of dependent variables, more or less distinctly separated.

The experimental data concerning the spatial distribution of RI as well as PLI values for selected PCEs in all three investigated soil layers were organized in a geospatial data base, further processed by ArcMap 10.4 Spatial Analyst [40]. This permitted the generating of six maps representing the spatial distribution of both RI and site PLI indices.

Due to the fact that the sampling points did not follow a regular network, the entire considered area was decomposed in Thiessen–Voronoi polygons [41,42]. In this case, an inverse distance weighting algorithm [43,44] gave the best results. Due to a reduced number of sampling points and in the absence of confident data concerning the distribution function of PLI values, the nonparametric ANOVA Tukey, Mann–Whitney, Dunnett post hoc and Spearman correlation were used to calculate the probabilities that the PLI distributions corresponding to the considered three layers are closer.

Further, Discriminant Analysis (DA) [39] was used to evidence to which extent the PLI values of each layer are correlated.

PAST 4.1 [45], TIBCO^® Data Science/Statistica™ 10 [46], and OriginLab^® OriginPro^TM 2021 [47] were used for statistical data analysis, and MS Excel 2019 was employed for experimental data processing and graphing.

3. Results and Discussion

3.1. Concentrations of Elements in Industrial Soils

The results obtained by the combined analytical methods INAA, XRF and ICP–MS highlighted the existence of 45 elements in the soils. Figure 2 shows the spatial and depth wise distribution of the element concentrations and

Table 2. Elemental levels (range, average) in investigated industrial soil samples (except for the control site), norms and Upper Continental Crust (UCC) values, expressed in mg·kg⁻¹ (except for the major elements, whose concentrations are given in g·kg⁻¹, and Au, in µg·kg⁻¹).

Element	Literature Data				This Work
	Normal [36]	Low; High-Alert [36]	Low; High-Intervention [36]	Upper Continental Crust Mean [48]	Galati ISP, min–max	Galati ISP, Average
Al, g·kg−1				79.24	38–55.6	44.68
As	5	15; 25	25; 50	4.8	3.72–11.1	8.55
Au, µg·kg−1				1.5	3.53–69.4	10
Ba	200	400; 1000	625; 2000	628	143–430	355.6
Br		50; 100	100; 300	1.6	3.19–13.2	6.87
Ca, g·kg−1				24.93	10.23–47.82	29.26
Cd	1	3; 5	5; 10	0.09	0.22–0.56	0.33
Ce				63	28.6–80.8	66.65
Co	15	30; 100	50; 250	17.3	4.23–15.4	11.08
Cr	30	100; 300	300; 600	92	40–120	92.11
Cs				4.9	1.88–7.89	4.69
Cu	20	100;250	200;500	28	13.79–54.93	23.13
Dy				3.9	n.d.–6.91	4.20
Eu				1	n.d.–1.62	0.377
Fe, g·kg−1				38.06	21.5–42.8	31.89
Hf				5.3	3.43–11.6	7.85
Hg	0.1	1; 4	2; 10	0.05	n.d.–0.9	0.13
I				1.4	4.82–10.6	6.62
K, g·kg−1				25.14	16.73–24.42	16.8
La				31	13.6–40.9	33.85
Mg, g·kg−1				14.56	2.04–3.6	2.7
Mn	900	1500; 2000	2500; 4000	753	554–1130	749.5
Mo	2	5; 15	10; 40	1.1	0.37–2.04	1.06
Na, g·kg−1				23.57	5.19–8.27	6.88
Nd				27	8.9–36.4	25.46
Ni	20	75; 200	150; 500	47	13.5–63.7	37.99
Pb	20	50; 250	100; 1000	17	10.84–52.90	29.40
Rb				84	32.4–119	82.53
Sb	5	12.5; 20	20; 40	0.4	0.51–1.62	0.95
Sc				14	7.31–13.9	10.29
Sm				4.7	2.54–7.67	6.23
Sn	20	35; 100	50; 300	2.1	n.d.–9.16	0.89
Sr				320	43–173	117.6
Ta				0.9	0.42–1.14	0.925
Tb				0.7	0.31–0.91	0.725
Tm				0.3	0.28–3.43	0.88
Th				10.5	4.65–13.2	10.17
Ti, g·kg−1				3.69	3.19–4.76	3.84
U				2.7	1.05–3.34	2.651
V	50	100; 200	200; 400	97	51.2–103	67.25
W				1.9	0.84–2.74	2.14
Y				21	3.04–6.58	4.47
Yb				1.96	1.14–3.77	2.76
Zn	100	300; 700	600; 1500	67	52.6–242	102.26
Zr				193	105–482	300.63

n.d.—not detected.

presents the concentration ranges and average of values obtained in this paper, along with the limits allowed by Romanian norms for trace elements in soil (normal values, low alert and intervention for sensitive use of soil, and high alert and intervention for less sensitive use of soil) [36] and literature data for Earth’s crust composition [48].

From Figure 2, it can be seen that the highest concentrations of most of the trace elements occur in sites G.2.2. G2.3 and G3.2 which are closest to the industrial complex, as well as in location G1.1, situated within the predominant wind direction, a trend similar to previous findings for the same industrial area [11,15].

Compared with norms stipulated by Romanian law [36], the concentration values obtained in this work for several PCEs (Table 2)—As, Ba, Co, Cr, Cu, Hg, Mn, Mo, Ni, Pb and Zn—exceed the normal levels, while in the case of Ba, Cr, Pb and V the alert levels are surpassed. With the exception of Al, Ba, Co, K, Mg, Na Sc, Sr and Y, all other major and trace elements are present in the investigated soils in concentrations higher than the Upper Continental Crust (UCC) values [48].

The variation of the depth migration index DMI of identified chemical elements for each sampling location is illustrated in Figure 3.

DMI showed very high values (DMI > 20 cm) for the following elements at specified locations: Ca at G2.1 and G4.1; Co and Ni at G2.1 and G4.1; Cr and Cu at G4.1; As and Sb at G4.1; Mo at G1.1, G1.3 and G4.1; Rb, Sr and Zn at G4.1; Br at G2.1 and G4.1; Cs at G4.1; Th, Tm, U, and Yb at G4.1; Sc at G2.1; Ce and Zn at G4.1; Ba at G2.1; Dy at G2.3; La and Sm at G4.1; Tb at G2.1 and G4.1; Nd at G1.2, G2.1 and G4.1; Au at G2.1, G3.1, G3.2, G4.1; Hf, W, Y and Ta at G4.1. For the other chemical elements, the DMI values point to an accumulation in the second soil layer (5–20 cm). Thus, a tendency can be noticed for element mobilization in the deeper layers of soil, most of the sites being characterized by high (class C) and very high (class D) vertical migration potential. The urban site G4.1. is different from other locations, being classified with very high migration potential of trace elements; this trend is also visible in Figure 2, as the elemental levels in the first soil layer are lower than in the second and third layers of this sublocation.

The results for PCA and CA of elemental concentrations are presented in Figures S1 and S2, Supplementary Material, respectively, for 10 cases (soil sampling sites—9 industrial sites around Galati integrated siderurgical plant and 1 control site) and 45 variables (concentrations of chemical elements).

PCA results modeled on the basis of the first two most significant principal components (PC1, PC2) specified in Figure S1 are depicted in Figure 4.

The PCA shows that as the depth increases the sites are better grouped, with the exception of sites G4.1 for the depth 0–5 cm, and G1.1 for the depths 5–20 cm and 20–30 cm, respectively. In all cases, it is noticed that the control site GC is distant from the main group, as expected. Similar results were found by applying CA for site clustering (Figure S2, Supplementary Material).

3.2. Soil Mineralogy

In assessing the origin of the soil mineral component, the incompatible trace elements, such as Sc, Co, Zr, rare earth elements (REE), Hf, or Th, are very useful [7,10]. In this regard, felsic rocks are depleted in Sc but enriched in Th and light REE, while the mafic rocks appear enriched in Sc, Cr, Co, Ni and heavy REE [49]. On this subject, the Sc mass fraction in felsic rocks is less than 20 mg/kg but exceeds 20–40 mg/kg in mafic rocks so that Sc is one of the most appropriate proxies for this kind of study [50]. In the case of the investigated soils, Sc mass fraction was 9.8 ± 2 mg/kg. In this regard, the discriminating bi-plot Th/Co vs. La/Sc (Figure 5a) [51] points with clarity towards a felsic origin of the mineralogy of the investigated soil.

The presence and distribution of Sc, Zr and Th in the investigated material permitted inference of the past processes that shaped the current situation. Indeed, the mineral zircon, due to an extreme resilience to abrasion explained by its hardness on Mohs scale as greater than 7.5, showed extreme resilience during recycling, so that the higher the Zr mass fraction, the larger the sedimentary material for sorting and recycling [52].

In this case, the biplot Th/Sc vs. Zr/Sc showed a relatively reduced recycling when compared with the Dobrogea loess (Figure 5b) [53], a fact explained by the predominance of clastic material in the considered soils. More information concerning the affinities of the mineral constituents of the soil with the other mineral systems can be furnished by reciprocal distribution of Sc, La and Th of La and Th [54]. In this regard, both discriminating La-Sc-Th ternary (Figure 5c) and La vs. Th biplot (Figure 5d) showed a good correlation between inorganic soil, minerals, components and the Upper Continental Crust (UCC) [48,54] and the North American Shale Composite (NASC) [55].

In the case of the Sc-La-Th ternary diagram (Figure 5c), all the experimental points corresponding to the Galati soils form a cluster around UCC [48,54] and NASC [55], suggesting a composition consisting of clay, silt or gravels from mixed sources. In its turn, the the La vs. Th bi-plot (Figure 5d) shows a good correlation between these incompatible elements, of which the ratio of 3.3 ± 0.35 is close to the UCC ratio of 2.95 [54] or the NASC value of 2.65 [55]. This approach to the UCC [48,54] and NASC [55] also demonstrates an obtained La/Sm mass fractions ratio of 5.47 ± 0.17, close to the UCC value of 6.6 [48] and NASC of 5.7 [55].

3.3. Soil Contamination and Ecological Risk

The second aim of this study consisted of a detailed investigation of 13 PCEs’ distribution around the Galati integrated steel plant (ISP) given its status as a potential contamination source. Accordingly, based on the mass fractions of V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Cd, Hg, Sb and Pb determined in soil cores covering the 0–5, 5–20 and 20–30 depth layers, the corresponding PLI (series PLI-13 in Figure 6) was calculated for each sampling point and each layer as one of the most representative contamination indices (Figure 6). In the calculations for Contamination Factors (CF) (

Table A2. Contamination Factors (CF) for 13 PCEs in investigated soils for three layers.

Site	D (cm)	As	Cr	Cu	Cd	Co	Hg	Mn	Ni	Pb	Zn	Fe	V	Sb
G1.1	0–5	1.49	1.06	0.86	3.56	0.67	2.14	1.26	0.91	2.11	1.96	0.88	1.06	2.03
	5–20	1.65	1.03	0.68	3.22	0.89	0.69	1.18	1.36	2.77	1.76	1.11	1.05	2.45
	20–30	2.21	0.84	1.07	2.78	0.65	0.24	0.97	0.85	1.53	1.02	0.71	0.57	2.60
G1.2	0–5	2.01	1.28	0.71	3.11	0.82	0.50	0.98	1.03	1.73	2.33	1.12	0.76	2.83
	5–20	1.44	1.00	0.77	2.89	0.65	4.11	0.99	0.99	1.51	1.78	0.87	0.77	2.60
	20–30	1.65	0.94	0.75	2.56	0.71	0.69	0.86	0.98	1.41	1.26	0.78	0.63	1.95
G1.3	0–5	1.54	1.09	0.79	2.67	0.53	3.04	0.92	0.65	1.34	1.41	0.76	0.65	3.33
	5–20	1.66	0.96	1.96	2.56	0.55	2.14	1.09	0.64	1.28	1.75	1.10	0.65	2.33
	20–30	1.45	0.99	0.66	2.44	0.69	0.85	0.91	0.87	1.47	1.21	0.75	0.71	1.80
G2.1	0–5	1.68	0.80	0.49	3.78	0.40	3.75	0.91	0.35	1.44	1.12	0.56	0.68	3.23
	5–20	1.90	0.94	0.57	3.11	0.60	2.32	0.78	0.65	1.46	1.32	0.77	0.60	2.50
	20–30	1.82	1.10	0.58	2.89	0.73	0.57	0.74	0.92	0.64	1.07	0.81	0.53	2.25
G2.2	0–5	1.71	1.20	0.62	4.67	0.60	16.07	1.37	0.76	1.80	1.69	0.93	0.72	2.48
	5–20	1.76	1.28	0.76	4.33	0.62	7.50	1.05	0.71	1.62	1.41	0.89	0.72	2.33
	20–30	1.68	1.05	0.69	3.89	0.70	0.60	0.97	1.02	1.60	1.07	0.81	0.59	2.08
G2.3	0–5	2.00	1.30	0.84	5.44	0.62	0.50	1.50	0.71	3.11	3.61	1.10	0.64	4.18
	5–20	1.76	0.79	0.84	5.89	0.58	4.16	1.08	0.75	2.19	2.22	0.66	0.85	2.38
	20–30	1.92	0.97	1.04	5.56	0.64	0.49	1.05	0.82	1.97	1.38	0.74	0.56	2.78
G3.1	0–5	2.03	1.08	0.79	2.89	0.68	0.50	0.86	0.71	1.60	1.52	0.81	0.69	2.63
	5–20	1.73	0.98	0.78	2.78	0.66	1.13	0.83	0.85	1.37	1.07	0.71	0.58	2.30
	20–30	2.04	1.07	0.82	2.56	0.73	0.71	0.81	0.87	1.41	1.00	0.77	0.59	2.13
G3.2	0–5	2.31	1.10	0.83	6.22	0.72	2.68	1.14	0.86	2.02	1.91	1.01	0.75	3.05
	5–20	2.04	1.10	0.93	5.67	0.72	2.14	1.10	0.86	1.70	1.58	0.92	0.78	2.53
	20–30	2.10	0.93	0.86	5.22	0.68	0.65	0.90	1.00	1.77	1.06	0.79	0.58	2.10
G4.1	0–5	0.78	0.43	0.80	3.00	0.24	0.50	0.94	0.29	2.48	0.79	0.84	0.70	1.40
	5–20	1.90	0.88	0.53	2.78	0.57	2.32	0.95	0.61	1.87	1.58	0.72	0.75	3.33
	20–30	1.85	0.85	1.28	2.56	0.62	0.53	0.76	0.81	1.52	1.35	0.69	0.56	2.50
GC	0–5	1.03	0.60	0.27	1.89	0.45	0.50	0.56	0.45	0.64	0.89	0.45	0.45	1.60
	5–20	0.98	0.48	0.31	1.78	0.42	2.50	0.61	0.36	1.12	0.75	0.43	0.54	1.53
	20–30	0.83	0.46	0.09	1.56	0.39	0.35	0.45	0.26	0.53	0.45	0.35	0.32	1.38

, Appendix A) with the aid of Equation (2), the UCC element concentrations [48] were considered as background values.

Based on the average values of 9 site PLIs, a regional index RPLI of 1.20 was calculated for the metallurgical industrial area of Galati, SE Romania, emphasizing a polluted zone (RPLI >1).

Comparing the PLI values for 13 PCEs with the PLI calculated for all 45 identified elements (series PLI-45 in Figure 6), it can be stated that the use of pollutant elements gives rise to a relative higher PLI than in the case of considering the elements present at near baseline levels [34], suggesting a real deterioration in soil quality in the region, which is the correct approach.

Final results showing the PLI distribution for each sampling point, including the remote one GC, are illustrated by the violin diagram reproduced in Figure 7a. Here, it can be remarked that, for all three soil layers, the reference example appears less contaminated, a fact sustained by PLI values varying between 0.45 and 0.78, i.e., lower than the threshold equal to one, which is necessary to prove local contamination.

To deepen the study of local contamination, we have performed a DA of the PCE values by considering the sampling points as cases and the mass fractions of each of the 13 contaminating elements as independent variables, while the dependent variables are designated as soil layers. The result is reproduced in Figure 7b by a Root 2 vs. Root 1 biplot. This graph shows with clarity the presence of two distinct clusters, one consisting of a partial superposition of the 0–5 and 5–20 cm points and the other containing only the 20–30 cm data. In our opinion, this graph suggests that the contamination process is spread onto the 0-20 cm layer while the third layer, between 20 and 30 cm seems less affected.

This observation is confirmed also by several samples in the ANOVA, Tukey, Mann–Whitney (

Table 3. The probabilities that the distribution of the 13 considered PCE are closer according to Tukey’s test (a, lower diagonal), Mann–Whitney (a, upper diagonal), Dunnette post hoc (b, lower diagonal), and Spearman correlation coefficient (b, upper diagonal).

a				b
Depth (cm)	0–5	5–20	20–30	depth (cm)	0–5	5–20	20–30
0–5		0.791	0.063	0–5		0.883	0.672
5–20	0.871		0.004	5–20	0.573		0.672
20–30	0.065	0.022		20–30	0.030	0.006

a) and Dunnett post hoc tests, as well as Spearman rank correlation (Table 3b), which showed a certain degree of similitude only between the 0–5 and 5–20 cm layers.

Another peculiarity can be remarked by analyzing the spatial distribution of PLI corresponding to each of the considered soil layers and illustrated by the maps in Figure 8a–c. Here, it can be observed that the maximum contamination corresponds to two well evidenced maxima in the sites G2.2 (close to the slag dump) and G3.2 (at the north gate of the enterprise, close to the iron scrap deposit, lime factory, steel plant and plate product mills), for which the PLI reaches closer values of 1.55 and 1.51, respectively, for the 0–5 cm layer, closely followed by the G2.3 site (at the south gate of the enterprise, close to agglomeration, sintering and coking plants, blast furnaces and ironmaking facilities) with a PLI value of 1.46.

With the depth, these areas enlarge in the second layer (5–20 cm) until they merge in the third layer (20–30 cm), while the maximum PLI values decrease to 1.41 in all three places in the second layer, and 1.10, 1.15, and 1.18, respectively, in the third layer. In our opinion, this finding illustrates a 3D diffusion of contaminating elements into soil starting from the two hotspots better evidenced in Figure 8a.

Three maps were generated representing the spatial distribution of the risk index RI for each soil layer (Figure 8d–f), calculated in

Table 4. The ecological risk factors and risk index of the 10 considered PCEs.

Site Code	Depth (cm)	${\mathit{E}}_{\mathit{r}}^{\mathit{i}}$										RI
		Hg	As	Pb	Cd	Cu	Cr	Zn	Co	Ni	Mn
G1.1	0–5	85.71	14.90	10.54	106.67	4.32	2.11	1.96	3.35	4.53	1.26	235.35
	5–20	27.71	16.54	13.84	96.67	3.41	2.05	1.76	4.45	6.78	1.18	174.39
	20–30	9.43	22.08	7.63	83.33	5.35	1.69	1.02	3.27	4.27	0.97	139.03
G1.2	0–5	20.00	20.08	8.63	93.33	3.57	2.57	2.33	4.10	5.14	0.98	160.73
	5–20	164.29	14.42	7.55	86.67	3.85	2.00	1.78	3.27	4.96	0.99	289.75
	20–30	27.57	16.48	7.06	76.67	3.73	1.89	1.26	3.53	4.89	0.86	143.93
G1.3	0–5	121.43	15.40	6.68	80.00	3.95	2.17	1.41	2.66	3.26	0.92	237.88
	5–20	85.71	16.65	6.41	76.67	9.81	1.93	1.75	2.75	3.20	1.09	205.96
	20–30	33.93	14.48	7.33	73.33	3.29	1.98	1.21	3.47	4.34	0.91	144.27
G2.1	0–5	150.00	16.79	7.19	113.33	2.46	1.60	1.12	2.01	1.74	0.91	297.15
	5–20	92.86	18.98	7.29	93.33	2.86	1.88	1.32	2.98	3.26	0.78	225.54
	20–30	22.64	18.23	3.19	86.67	2.88	2.20	1.07	3.67	4.61	0.74	145.88
G2.2	0–5	642.86	17.06	9.02	140.00	3.09	2.39	1.69	3.01	3.81	1.37	824.28
	5–20	300.00	17.56	8.08	130.00	3.79	2.57	1.41	3.09	3.54	1.05	471.09
	20–30	24.00	16.81	8.01	116.67	3.43	2.09	1.07	3.50	5.10	0.97	181.64
G2.3	0–5	20.00	20.04	15.56	163.33	4.21	2.61	3.61	3.12	3.57	1.50	237.56
	5–20	166.43	17.60	10.97	176.67	4.22	1.59	2.22	2.92	3.73	1.08	387.44
	20–30	19.79	19.19	9.86	166.67	5.19	1.93	1.38	3.21	4.09	1.05	232.35
G3.1	0–5	20.00	20.27	7.99	86.67	3.96	2.16	1.52	3.38	3.57	0.86	150.39
	5–20	45.21	17.33	6.83	83.33	3.91	1.97	1.07	3.32	4.27	0.83	168.08
	20–30	28.36	20.40	7.03	76.67	4.08	2.13	1.00	3.67	4.34	0.81	148.48
G3.2	0–5	107.14	23.13	10.08	186.67	4.14	2.20	1.91	3.58	4.30	1.14	344.29
	5–20	85.71	20.44	8.52	170.00	4.66	2.20	1.58	3.58	4.31	1.10	302.09
	20–30	26.07	21.04	8.87	156.67	4.31	1.86	1.06	3.38	5.02	0.90	229.19
G4.1	0–5	20.00	7.75	12.38	90.00	3.98	0.86	0.79	1.22	1.44	0.94	139.35
	5–20	92.86	19.00	9.34	83.33	2.66	1.75	1.58	2.84	3.03	0.95	217.34
	20–30	21.36	18.48	7.59	76.67	6.39	1.71	1.35	3.12	4.04	0.76	141.47
G.C.	0–5	20.00	10.29	3.21	56.67	1.33	1.19	0.89	2.23	2.26	0.56	98.63
	5–20	100.00	9.77	5.60	53.33	1.53	0.95	0.75	2.09	1.82	0.61	176.46
	20–30	14.07	8.31	2.67	46.67	0.46	0.91	0.45	1.95	1.29	0.45	77.22

with the aid of Equation (6) for 10 selected PCEs based on the individual ecological risk factors $E_r^i$ computed with Equation (5).

The RI distribution maps highlight the depth diffusion trend of hazardous elements shown by the PLI maps (Figure 8a–c).

The obtained values for the individual ecological risk factors $E_r^i$ and the risk index RI (Table 4) show that highly ecological risk in the identified industrial hot spots is due to Cd and Hg.

The percentage contribution of each of the ten PCEs at the total value of RI depicted in Figure 9 shows the considerable risk due to highly toxic elements Hg and Cd, followed by As, Pb, Cu, Ni, Co, Cr, Zn and Mn.

Due to the negative impact of PCEs emitted as a result of steel production on soil quality, terrestrial and aquatic ecosystems’ state and people’s health [13,17,56], it is very important to regularly monitor these elements in soils neighboring industrial factories and waste deposits and assess the environmental damage.

At the most contaminated sites, it is recommended to apply soil depollution strategies and metallogenic area remediation techniques, such as phytoremediation, microbe-assisted remediation, constructed wetlands or extraction of polluted soil material [32,57,58,59], depending on the local weather conditions, contamination degree and depth, and soil physico-chemical properties, which could influence the mobility of elements in soil [58].

4. Conclusions

The current study aimed to bring new insights into the quality of soils in the neighborhood of a large siderurgical complex located in Galati, SE region of Romania, regarding the compositional scheme, spatial and depth migration of 45 chemical elements (metals, radioelements, rare earth elements and other trace elements), contamination trends, mineral assemblage, ecological risk of selected potential contaminant elements and regional contamination level due to industrial activity related to ferrous metallurgy.

The results obtained for the concentrations and environmental and safety risk indices of elements of concern demonstrate a present pollution of surface layer of soils adjacent to the siderurgical industry, although the industrial production and activity had been restrained in the last decade, along with a vertical migration of most of the analyzed elements in deeper soil layers.

The geochemical and contamination features suggest a similarity with results obtained in previous surveys performed in the Galati industrial area, regarding the location of the impacted sites with regard to the distance from the enterprise and dominant direction of wind. Our results can constitute a useful database for further investigations in the area and elaboration of land management strategies and exploitation of terrains, as well as decommissioning of the slag dump territory and valorization of metallic materials in various purposes.