phân bố của các dạng kim loại nặng trong môi trường đất
Trang 1PROOF
Distribution and mobility of metals in contaminated sites.
Chemometric investigation of pollutant profiles Ornella Abollinoa, Maurizio Acetob, Mery Malandrinoa, Edoardo Mentastia,*,
Corrado Sarzaninia, Renzo Barberisc
a Department of Analytical Chemistry, University of Torino, Via P Giuria 5, 10125 Torino, Italy
b Department of Science and Advanced Technologies, University of East Piedmont, Corso Borsalino 54, 15100 Alessandria, Italy
c Environmental Protection Agency of the Regional Government of Piedmont (ARPA Piemonte), Via della Rocca 49, 10123 Torino, Italy
Received 10 July 2001; accepted 9 November 2001
‘‘Capsule’’: Chemometrics allowed identification of groups of samples with similar characteristics
Abstract
The distribution and mobility of heavy metals in the soils of two contaminated sites in Piedmont (Italy) was investigated, evalu-ating the horizontal and vertical profiles of 15 metals, namely Al, Cd, Cu, Cr, Fe, La, Mn, Ni, Pb, Sc, Ti, V, Y, Zn and Zr The concentrations in the most polluted areas of the sites were higher than the acceptable limits reported in Italian and Dutch legisla-tions for soil reclamation Chemometric elaboration of the results by pattern recognition techniques allowed us to identify groups of samples with similar characteristics and to find correlations among the variables The pollutant mobility was studied by extraction with water, dilute acetic acid and EDTA and by applying Tessier’s procedure The fraction of mobile species, which potentially is the most harmful for the environment, was found to be higher than the one normally present in unpolluted soils, where heavy metals are, to a higher extent, strongly bound to the matrix # 2001 Published by Elsevier Science Ltd All rights reserved
Keywords: Heavy metals; Contaminants; Soils; Mobility; Speciation
1 Aim of investigation
The problem of contaminated soils is becoming of
increasing concern for the environment because of the
large number of polluted sites in existence (Ferguson
and Kasamas, 1999) The main sources of soil pollution
are improper waste dumping, abandoned industrial
activities, incidental accumulation (e.g leakage,
corro-sion), atmospheric fallout, agricultural chemicals
(Allo-way, 1994) Many contaminated sites date back to two
or three decades ago, when environmental legislation on
solid and liquid waste disposal was not as strict as
nowadays
Before starting the reclamation of a site, the extent
and distribution of contamination must be investigated,
in order to identify the area to be treated and choose the
proper clean-up strategy A few examples of such
investigations, with regard to heavy metal pollution, are the determination of arsenic, chromium and copper in Danish soils after spill of chemicals (Lund and Fobian, 1991), the evaluation of the heavy metal content around
a disused mine in Korea (Jeong et al., 1997) and the assessment of arsenic contamination in Germany due to ore and industrial sources (Bombach et al., 1994) The present paper describes the characterisation of heavy metal pollution in the soils of two sites formerly used for industrial waste disposal The horizontal and ver-tical distribution of contaminants was investigated and the concentrations were compared with the acceptable limits imposed by Italian and Dutch legislation (Minis-try of Housing, 1994; Ministerial Decree, 1999b) for soil reclamation A chemometric treatment of the data was performed
The toxicity of metals depends not only on their total concentration, but also on their mobility and reactivity with other components of the ecosystem The most com-mon way to study element mobility in soils is by treat-ment with extractants of different chemical properties 0269-7491/01/$ - see front matter # 2001 Published by Elsevier Science Ltd All rights reserved.
P I I : S 0 2 6 9 - 7 4 9 1 ( 0 1 ) 0 0 3 3 3 - 5
Environmental Pollution & (&&&&) &–&
www.elsevier.com/locate/envpol
* Corresponding author Tel.: 6707625; fax:
+39-011-6707615.
E-mail address: mentasti@ch.unito.it (E Mentasti).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
Trang 2PROOF
(Nowak, 1995; Szulczewski et al., 1997; Rauret, 1998)
In this work the release of metals into water, dilute
acetic acid and EDTA was investigated, and the
Tes-sier’s partitioning scheme (Tessier et al., 1979) was
applied to selected samples
The results obtained can be of use for the local
authorities to decide about the necessity of reclamation
of the two sites and the level of priority of the
interven-tion, with respect to the situation of other polluted areas
Moreover, the data can be of interest to the European
Environment Agency for its activities of soil monitoring
2 Description of the experimental procedures
2.1 Site description
The two investigated sites are located in northeast
Piedmont, Italy The first one (hereafter called site A) is
in a flat area near the small town of Pieve Vergonte
(3000 inhabitants), in the province of Verbania, located
about 100 m from a small river The zone stands on
alluvial deposits of the river The top layers of the soil
are made of sand and silt Groundwater flows at a depth
of about 5 m The contaminated area, whose estimated
respectively, is made up of a mixture of industrial
wastes and soil The original soil is almost absent in the
area and in its surroundings, because of repeated
exca-vations; the soil covering the area, probably carried
from nearby zones, is mainly made of gravel and sand
The presence of debris, probably coming from copper
and brass foundries, can also be visually detected owing
to the presence of coloured (mainly blue-green) spots
due to metal salts and of small plastic strips, deriving
from wire coatings The material was not placed in a
previously excavated area, but it forms an artificial relief
with respect to the surroundings The zone where the
relief lies consists of three levels: the relief itself, an area
at ground level, at least twice as wide, and an excavated
basin about 7 m deep
The other site (hereafter called site B) is located near
the town of Borgomanero (19,400 inhabitants), in the
province of Novara The contamination occurred
because of the repeated floods of a small stream, which
today has a new course, caused by the insufficient size of
the stream bed with respect to the flow in rainy periods
The stream collected wastewaters of local industries,
some of which operating in the electroplating field, and
its floods caused an accumulation of contaminants,
mainly of inorganic nature, in the soil The extension of
the polluted area is estimated between 20,000 and
3000 m2wide: it is a flat, uncultivated area, covered by a
layer of black sludge about 1.50 m deep carried by
the floods, where a scant vegetation grows The rest of
the area is covered by trees and spontaneous plants The land in the zone is made of alluvial deposits The top layer of the soil, down to a depth of from 0.6 to 2 m, is composed of sand with silt and clay, with a low gravel content This layer gives a discrete impermeability to the soil Below there is an alluvial layer with sand and gravel, down to groundwater which flows at 4–5 m depth
Table 1 reports a brief description of the location of the single sampling points, which were chosen in a ran-dom fashion in order to cover the whole areas A total
of 33 samples was collected at site A, both at different points of the presumably most contaminated zone and
in the surroundings Some were sampled from the sur-face and others immediately below, at a depth of 10 cm One specimen was obtained in a hole (1 m deep) dug on the relief Two pieces of blue-green material were also collected For comparison, a sample from a park in the city centre was considered Fifteen samples were col-lected at different depths on one side of the relief, down
to 330 cm At site B 28 samples were collected from the core of the contaminated zone and its surroundings, both at the surface and 10 cm below Also in this case, one soil specimen from the nearby town nearby was collected Eleven samples were obtained from different depths, down to 160 cm, from one point in the central area of the core
The collected samples, referred to as ‘‘soil’’ in this paper, when coming from the most polluted areas of the sites, were not strictly ‘‘soil’’ but rather a material cov-ering the original soil, with the characteristics described above (mixture of soil and debris at site A, black sludge
at site B)
2.2 Apparatus and reagents Most metal determinations were performed with a Varian Liberty 100 (Varian Australia, Mullgrave, Aus-tralia) inductively coupled plasma–atomic emission spectrometer (ICP–AES) The spectral interference of
Fe and V, which have an emission line close to that one
of La (379.478 nm), was taken into account by selecting background correction positions outside the interfering peaks Alternatively, La can be determined at 407.672
nm Standards for calibrations were prepared in ali-quots of sample blanks
Cadmium and lead, when present below the ICP–AES detection limits, were determined with a Perkin Elmer
5100 (Perkin Elmer, Norwalk, Connecticut, USA) elec-trothermal atomic absorption spectrometer (ETAAS) equipped with Zeeman-effect background correction Sample dissolutions for the determination of total concentrations were performed in tetrafluormethoxyl (TFM) bombs, with a Milestone MLS-1200 Mega (Milestone, Sorisole, Italy) microwave laboratory unit Analytical grade reagents were used throughout Stan-dard metal solutions were prepared from concentrated
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 3PROOF
Merck Titrisol stock solutions (Merck, Darmstadt,
Germany)
2.3 Procedures
All experiments were performed in triplicate and
blanks were run simultaneously The relative standard
deviations of the results were typically below 10%
Higher deviations were observed, for some data, for
extractions in water and acetate, owing to the low
con-centrations measured; in some cases also the total
concentrations in the polluted area at A site showed a
variability higher than 10%, especially for copper,
because of the heterogeneity of the samples
The evaluation of pH and EDTA-extractable
frac-tions was performed according to the official methods of
soil analysis envisaged by the Italian legislation
(Minis-terial Decree, 1992) After completion of the
experi-mental work, a new revision of official methods was
issued (Ministerial Decree, 1999a), which in any case
only slightly differs from the previous one The leaching
test with acetic acid was performed according to the
Italian official methods for sludge analysis (Water Research Institute, 1985)
2.3.1 Sampling and pretreatment Surface samples were obtained with a trowel (after removing the top layer in contact with the atmosphere) and stored in plastic bags In-depth samples at site B were collected with the aid of a motor-driven corer The samples were air-dried and, after breaking the agglom-erates with a plastic hammer, sieved through a 2-mm sieve and ground with a ball mill
2.3.2 pH Sample pH was determined in sample-water suspen-sions (8 g of sample in 20 ml of water) The suspensuspen-sions were shaken and left standing overnight before the measurement (Ministerial Decree, 1992)
2.3.3 Sample digestion for total metal determination Aqua regia (5 ml) and 2 ml of hydrofluoric acid were added to 100 mg of sample in TFM bombs and heated
in a microwave oven following the sequence: three steps
Table 1
Description of sample collection points
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 4PROOF
of 5 min each (at a power of 250, 400, 500 W,
respec-tively) followed by a final 3 min step at 600 W Then 0.7
g of boric acid were added and the bombs were further
heated for 10 min at 250 W Finally the samples were
filtered and diluted to 100 ml (Bettinelli et al., 1989;
Aceto et al., 1994; Gulmini et al., 1994)
2.3.4 Available metal fraction
acid and brought to pH 4.65 0.05 Twenty-five
milli-litres of the extractant were added to aliquots of 5.0 g of
sample The suspension was shaken for 30 min, filtered
and the extract was analysed (Ministerial Decree, 1992)
2.3.5 Leaching tests
Leaching tests were performed with HPW and
and treated as described for pH measurement was
cen-trifuged and the supernatant was separated and analysed
The leaching test with acetic acid was performed on a
suspension of 1.0 g of sample in 16 ml of HPW brought
The suspension was shaken for 24 h; the pH was
peri-odically checked and maintained at the original value
Afterwards the suspension was centrifuged and the
Research Institute, 1985)
2.3.6 Sequential extractions
The sample (1.0 g) was sequentially extracted with
different reagents according to the following procedure
(Tessier et al., 1979): (1) 8 ml of 1 mol dm 3MgCl2, for
HNO3, for 5 h at 85 2C, followed by addition (after
min at room temperature
After each step the suspension was centrifuged, the
supernatant was separated and the solid phase was
added with the reagents for the subsequent extraction
The extracts were diluted to 25 (first fraction), 50
(sec-ond fraction) and 100 (next two fractions) ml, stabilised
by the addition of concentrated nitric acid (25, 50 and
100 ml respectively) and analysed in order to calculate
the element percentages extracted in each fraction The
residual element percentages (fifth fraction) were
com-puted from the total concentrations by subtraction The
mass balance was evaluated for a few samples by
com-paring the total metal content with the sum of the metal
percentages extracted in the five fractions after digestion
and analysis of the fifth fraction The recovery was high
(i.e > 90%) for Cd, Cr, Cu, Ni, Pb, Zn (i.e the heavy metals of greatest interest from the environmental point
of view) and Al, whereas Fe, Mn, Ti, V, Zr were par-tially lost (recoveries ranged between 67 and 82%) Zir-conium was mostly lost in the first extraction step, manganese in the fourth one, whereas losses of the other elements took place in all the first four steps, probably during filtration of the surnatant
2.3.7 Chemometric data treatment Two unsupervised methods (Hierarchical Cluster Analysis, HCA, and Principal Component Analysis, PCA) and a supervised one (Discriminant Analysis, DA) were applied to the data The treatment was performed with XlStat, an add-in package of Microsoft Excel
HCA was run applying Ward’s method of agglom-eration and squared Euclidean distance as similarity measure All variables were standardised by
for the average) Dendrograms were obtained
As to DA, two classes were defined a priori, con-sidering samples from sites A and B respectively Uni-variate ANOVA was used to calculate F-ratios and find out variables with higher discriminating power Prob-abilities of class membership were calculated for all samples
3 Results and discussion 3.1 Total metal concentrations Fifteen metals, namely Al, Cd, Cu, Cr, Fe, La, Mn,
Ni, Pb, Sc, Ti, V, Y, Zn and Zr, were determined Tables 2–3 report their concentrations and the pH values, in samples collected at sites A and B respec-tively The corresponding ranges, averages and medians are reported in Table 4; to allow an easier interpretation
of the results, calculations were performed for three groups of data: (1) all samples except A15 and A16, which consist of coloured material, and the vertical profile; (2) all samples except A15, A16, the vertical pro-file and the ones outside the most polluted area; (3) vertical profile Of course a detailed mapping of the contamination cannot be achieved from the relatively small number of sample points, but the results obtained allow anyway to make some considerations about the distribution and extent of the pollution in the areas
It must be recalled that the considered elements are present in unpolluted soils at what can be defined
‘‘background level’’, both as a result of natural phe-nomena, such as the contribution of the parent material, and of common anthropogenic activities (e.g agri-culture, traffic, etc.) We can suspect or confirm the pres-ence of pollution when the concentrations are higher than the typical values for soils found in literature and
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 5PROOF
exceed the levels present in the nearby areas: in fact in
some (albeit uncommon) cases high concentrations of
one or more elements have a natural origin, as in some
soils in California rich in selenium (Halloway, 1990)
In this study, in order to assess the presence and
extension of contamination, the concentrations of some
elements measured at the sites were compared with the
normal ranges and the most common values typically
present in soils (Halloway, 1990; Merian, 1991) and
with the maximum admissible levels in soils according
to Italian (Ministerial Decree, 1999a,b) and Dutch
(Ministry of Housing, 1994) legislations These data are
collected in Table 5, which also reports, for comparison,
the mean content in the earth’s crust (Weast, 1974)
3.1.1 Site A
Abnormally high levels of Cd, Cu, Pb and Zn were
found at site A In particular, the presence of copper is
related to the disposal of electric cables The most
pol-luted zone is the relief, in which an overall increase of
these four elements and, to a lesser extent, of chromium,
manganese, nickel and zirconium can be observed The
concentrations of these metals are smaller in the basin, even if a contamination of cadmium, copper, lead and zinc is present Also the neighbouring zone under the vegetation has relatively high levels of Cu, Pb and Zn The concentrations at the base of the relief usually fall between the ones in the relief and under the vegetation The contents of Cr and Ni do not exceed the typical ranges, but in many samples are above the common values reported in Table 5 and, especially in the vertical profile (whose behaviour will be discussed below), are higher than in the surroundings: therefore an input of these elements with the waste can be supposed The same hypothesis is valid for manganese, whose level in the vertical profile, moreover, is higher than typical values Some samples on the relief (A9, A10, A12, A13) are also rich in zirconium, which might have been con-tained in the wastes as well
The concentrations found a few hundred meters from the site are not higher than the ones present in the sample collected in the city centre, which can be assumed to be unaffected by the waste disposal which caused the contamination of the site; in both cases the
Table 2
Total metal concentrations (mg/kg) and pH at site A
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 6PROOF
contents of Cr, Cu, Pb, Ni, Mn and Zn are inside the
typical ranges
Al, Fe, La, Sc Ti, V and Y concentrations do not
show a definite trend as a function of sampling position
The level of iron on the relief is higher than at the base
and in the basin, but lower than in the city center: it is
likely that the disposed wastes contained iron, causing a
local increase in its concentration, even if the levels
reached are not abnormal An input of lanthanum with
the wastes may have occurred, since its concentrations
are slightly higher at the site than in the city centre and
the surrounding area
The pH is lower in the relief than in the surrounding
area (except under vegetation)
The concentrations of Cd, Cu, Mn, Ni, Pb, Zn and Zr
are lower in surface samples than 10 cm below, with a
few exceptions In most cases the levels of La, Sc, Ti, V
and Y show the opposite behaviour There is not a
reg-ular trend for Al, Cr or Fe
The concentrations of Cd, Cr, Cu, Mn, Ni, Pb, Zn
and Zr in the vertical profile are similar or even higher
than the ones found on the relief surface, confirming the
presence of a bulk mass of disposed wastes There is not
a regular trend in the concentrations as a function of
depth, with the exception of copper which tends to
decrease with increasing depth; in any case for many
elements (Cd, Cr, Cu, Ni, Pb, Zn, Fe, Zr and V) larger fluctuations and generally higher concentrations are observed in top layers than in deeper ones The highest value for Al, Cd, Cr, Cu, Fe, Mn, Pb, Zn and Zr is between 30 and 135 cm, whereas the lowest in many cases is below
One of the pieces of material analysed (A15) has con-centrations similar to ones of the relief, whereas the other (A16) has a very high copper level and low con-tent of Fe and of La, Sc, Ti, V and Y
In general the metal distribution is heterogeneous, owing to the heterogeneous mixing of the soil with par-ticles coming from the waste
From the above observations the metals can be divi-ded into four groups: (1) Cd, Cu, Pb and Zn, which are present at very high levels at the site; (2) Cr, Mn and Ni,
by which the site is supposed to be contaminated (see also the discussion on legislation below), but to a lesser extent; (3) Fe, La and Zr, in which an input from wastes
is supposed but whose level is not a sign of pollution; (4)
Al, Sc, Ti, V and Y whose concentrations in the pol-luted area and in the surroundings are similar There-fore it can be presumed that the elements of the first three groups have both geochemical and (to different extents) anthropogenic sources, whereas the ones of group four have mainly a geochemical origin
Table 3
Total metal concentrations (mg/kg) and pH at site B
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 7PROOF
3.1.2 Site B
High concentrations of Cd, Cu, Cr, Ni, Pb and Zn were
found at site B In particular, the presence of Cr and Ni
could be due to an input from factories manufacturing
taps and fittings The contamination is spread all over the
core of the site, and below the vegetation growing just
outside it At a short distance from the core (points B23,
24, 25, 26) there is still a certain level of contamination,
but less pronounced than in the core; there are some
exceptions, such as the high levels of Cr, Pb and Zn in
sample B23 Metal concentrations are within normal
ranges about 200 m from the site and in the city centre
No enrichment of Mn, Al, Fe, Sc, Ti, V or Zr in the
site is observed A few samples (e.g B2, B6) have a
relatively high concentration of lanthanum, but it is not
possible to say whether it is due to contamination
A decrease in metal concentration from the surface to
the layer underneath was observed in many cases
As to the vertical profile, the concentrations of Cd,
Cr, Cu, Ni and Pb, are higher in the top layers, and tend
to decrease below 40 cm Many data in the lower layers
are above the common values (Table 5) but still within the typical ranges, even if some local maxima are pres-ent, such as for Ni (sample B17), Cd (B15–B16) and for
Cr, Cu and Pb (B19) The concentrations of zinc increases down to 80 cm
A general trend to higher values at depth than on the surface is observed for Al, Fe, Mn, Sc, V and Y The concentrations of La and, partially, of Ti tend to decrease with depth, whereas those of zirconium do not show any trend The pH value is lower than 4 at a depth
of between 15 and 80 cm: it is likely that this low value
is due to an input of acidic wastewater
In general, the metals can be divided into two groups: (1) Cd, Cr, Cu, Ni, Pb and Zn, whose concentrations are heavily affected by anthropogenic inputs, and (2) Al,
Fe, Mn, Sc, Ti, V and Zr, which are mainly of geo-chemical origin
3.1.3 Legislation limits The results were compared with the maximum acceptable concentrations in soils reported in the
Table 4
Mean, median, ranges of total concentrations (mg/kg) at sites A and B
a All samples excluding A15, A16 and the vertical profile (A1–A14 and A32–A33; B1–B11 and B23–B28).
b All samples excluding A15, A16, vertical profile and the ones outside the most polluted area (A1–A14; B1–B11).
c Vertical profile samples (A17–A31; B12–B22).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 8PROOF
sites and with Dutch intervention values (Ministry of
Housing, 1994), formerly known as C values (Table 5)
Italian limits depend on land use, and are lower for
public and private green areas and residential sites (‘‘A’’
limits) and higher for industrial areas (‘‘B’’ limits) At
site A, the levels of Cd, Cu, Pb and Zn exceed ‘‘A’’ and
‘‘B’’ limits in most samples; the concentrations of Cr
and Ni are higher than ‘‘A’’ limits, but below ’’B’’ ones
Copper, chromium and nickel contents at site B are
above both limits in many samples, whereas lead and
zinc, and in some cases cadmium, are between ‘‘A’’
and ‘‘B’’ values
All samples exceeding ‘‘A’’ and some of the samples
exceeding ‘‘B’’ levels have concentrations above Dutch
intervention values, which are intermediate between the
two Italian sets of limits, and are to be considered,
according to the official terminology, ‘‘seriously
pol-luted’’
3.1.4 Chemometric processing
The results were processed by PCA and HCA, in
order to obtain a visual representation of the data set
and gain insight into the distribution of the pollutants
by detecting similarities or differences which would be
more difficult to identify only by looking at the tables
As to PCA, both scores, which allow us to recognise
groups of samples with similar behaviour, and loadings,
which show the correlation among variables, were
evaluated and reported as biplots The first three PCs
were computed, but only PC1 and PC2 gave useful
information The data for the horizontal (including 10
cm depth) and vertical profiles were processed both
together and separately: the results of the separate treatment will be described hereafter, and hints on the joint processes, which provided little further informa-tion, will be given This paper reports some PC and dendrogram plots, as an example, for this and the fol-lowing sections All other PCA and HCA plots are available on request from the authors
As to site A, the following observations can be made: for the data set relative to the horizontal profile, the variance explained by the first two PCs is 22 and 49% respectively (71% in all) The plot of PC1 vs PC2 is reported in Fig 1a; in this figure, as well as in the other PCA plots shown in the paper, the position of the loadings is marked with a squared frame Fig 1a shows a certain degree of similarity for samples A1– A8, which were collected outside the relief, but still at the site or very close to it Samples A32 and A33, collected outside the polluted area, are somewhat apart but not very far from them The specimens from the relief (A9–A13, with the exception of A14) are in other zones of the plot They are distanced from each other, owing to the heterogeneity of the wastes The combined plot shows that they are mainly characterized by high concentrations of the polluting elements One of the pieces of material (A16) is com-pletely isolated from the other samples, confirming its different characteristics, and is strongly characterised
by its copper content;
the metals belonging to the first two above identified groups, together with zirconium, are correlated, with the exception of copper which stands alone They have opposite values of PC1 with respect to the other
Table 5
Typical concentration ranges and most common values present in soils, average abundance in the earth’s crust, acceptable concentrations in soils for Italian legislation (A: limits for public and private green areas and residential use; B: limits for commercial and industrial use of soil), target and intervention values for Dutch legislation (values in mg/kg unless otherwise stated)
30–100 (urban)
a Values for agricultural soils
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 9PROOF
elements: this component may therefore be connected
to their origin The pH is anticorrelated to most of
the pollutants;
two main groups can be identified from the
dendro-gram reported in figure 1b: one is made by most
samples from the relief and by the two pieces of
material; the second group contains the other samples
and it is possible to distinguish: (1) sample A33 (city
centre) which stands on its own; (2) group
A3-A4-A7-A8, coming from the ground level;
for the data regarding the vertical profile, the variance
explained by the first two PCs is 34 and 24
respec-tively (58% in all) In the plot of PC1 vs PC2 (not
shown) samples A17 and A18, corresponding to the
first two layers, stand out because of the high
con-centrations of Cd, Cu, Mn, Fe (for A17) and Zn (for
A18) A21 and A23 form a separate group due to the
high content of Al, Fe, Pb, La, Zn and Zr (for A21),
and Al, Fe, Cd, Ni, Ti, V and Zn (for A23) No
sig-nificant distribution can be identified for the other
samples, apart from the close resemblance of A20
and A24;
as to the loadings, the polluting elements are less
strictly correlated than in the horizontal profile, even
if they are in the same area of the plot and load
posi-tively on PC1, with the exception of Zr and Cr pH is
anticorrelated to this groups of variables V and Fe behave like the pollutants, whereas Al, Sc, La, Ti and
Y are in other areas of the plot It can be supposed that PC1 is connected to the elements of mainly anthropogenic origin and PC2 to the ones of a mainly geochemical source;
HCA confirms the different characteristics of the first two layers;
when the data for site A are processed all together, the variance explained by the first two PCs is 50 and 19% respectively (69% in all) The samples for ver-tical and horizontal profiles form two groups in the plot of PC1 vs PC2, the former being characterised
by their content in polluting elements; exceptions to this distribution are A10 and A13, which show a stronger similarity with the vertical samples, and A16, the piece of material, which is isolated from the other specimen Two clusters corresponding to horizontal and vertical (plus A10, A13, A16) profile samples are also present in the dendrogram
Data processing for site B gave the following results:
as to the horizontal profile, the variance explained by the first two PCs is 38 and 19% respectively (57% in all) According to the plot of PC1 vs PC2 (Fig 2a) samples B1–B11, collected in the core or just outside, and B23–B27, from the surroundings, form two groups; samples B8 and B9, collected under the vege-tation grown just outside the site core, have a stron-ger similarity to the second group Sample B28 from the city centre is clearly differentiated from all the others Group B1–B11 is characterized by the pre-sence of the polluting elements Sample B6 stands out because of its high content of Cr, Cu, La, Pb and Zn; the correlation among the elements identified as pol-lutants (Cd, Cr, Cu, Ni, Pb, Zn) is evident Such ele-ments are anticorrelated to Mn and Fe A weak correlation is also present among Al, La, Sc, Ti and
Y The pollutants have high positive loading on PC1: this component therefore takes account of the pollu-tion of the site On the other hand, PC2 is influenced
by the elements of mainly geochemical origin Sur-prisingly, pH is unrelated to the pollutants: a high pH value would be expected to be connected to high metal concentrations because it stabilises metal oxide and hydroxide forms and reduces their mobility;
the dendrogram in Fig 2b confirms the different characteristics of specimen B28 and, apart from sam-ples B23 and B26, the division between groups B1– B11 and B23–B27 The closeness of most samples coming from the core of the site (B1–B4), is also apparent;
as to the vertical profile, the variance explained by the first two PCs is 51 and 23% respectively (74% in all) The scores on PC1 of the first two layers (B12 and
Fig 1 Combined plot of scores and loadings obtained by (a) PCA
and (b) dendrogram for horizontal profile samples at site A (total
metal concentrations).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
Trang 10PROOF
B13) are very different from those of the layers below
A differentiation between groups B14–B17 and B18–
B22 (with sample B19 as an outlier) is also present;
the group of the polluting elements is more scattered
than in the horizontal profile, but it still has loadings
on PC1 of the opposite sign with respect to many
elements of a mainly geochemical source PC1 is
almost unaffected by pH, which on the other hand
heavily loads on PC2 pH is anticorrelated to Cd and
Zn and, at a lower level, to Ni, La, Zr, Al;
HCA confirms the observations made above: the first
two layers from a separate cluster, and groups
B14-B17 and B18-B22 are again present, with sample B19
being closer to the former;
when all data for site B are considered together, the
variance explained by the first three PCs is 38 and
18% respectively (56% in all) The sampling locations
can be divided into three main groups: (1) the
hor-izontal profile in the site core or just outside it (B1–
B11), together with the first two layers of the vertical
profile (B12–B13), characterized by a high content of
contaminants; (2) the deeper layers of the vertical
profile (B14–B22); (3) the samples collected in the
surroundings of the site (B24–B27), excluding B23,
and in the city centre (B28) The corresponding
den-drogram showed a similar clustering
The data for sites A and B were also treated together The variance explained by the first two PCs is 40 and 17% respectively (57% in all) Three groups are present, corresponding to (1) most A samples, (2) B samples from the horizontal profile, (3) vertical B samples and some A ones A samples are characterized by their con-tent in Cu or Cd, Pb, Zn, Mn, and B ones by Cr and Ni Specimens from A and B sites are also differentiated in the dendrogram
Data processing with DA allowed us to identify all samples as belonging to the expected classes ( > 99.9% probability), i.e to the site of collection, except B20, which was classified as ‘‘A’’ type, and B21, which was assigned to the correct class but with 74.5% probability The two pieces of materials (A15 and A16) were exclu-ded from the data set because they were not actually
‘‘soil’’ samples The variables with the highest dis-criminating power were Cd (F=31.79), Cu (F=36.50),
La (F=51.72), Pb (F=35.92), Ti (F=37.02), Zn (F=50.28) The tabulated F value for a confidence level
of 95% is 4.00
3.2 Mobility Extraction studies were carried out in order to inves-tigate the mobility of the metals and therefore their possible release into the environment and their toxicity Experiments were carried out by extraction with reagents of different chemical properties, in order to identify fractions of analytes with different labilities Extractions with water and EDTA solutions were per-formed on samples from the depth profiles The first layers at site A were mixed together (three by three) in order to have a sufficient amount of specimen for all experiments The leaching test with acetic acid, per-formed at pH 5.0 according to Italian official methods
of sludge analysis (Water Research Institute, 1985), was applied only to site A, because the pH of the water sus-pensions of most site B samples was already lower than 5.0 Tessier’s protocol was applied to two samples for each site
PCA and HCA were performed on the percentages extracted in water, acetic acid and EDTA
3.2.1 Leaching with water The leaching test with pure water was performed in order to evaluate the fraction of metals weakly bound to the matrix, e.g present as inorganic soluble salts The results can also give a preliminary indication on the possible release of pollutants by rains, although of course the laboratory experimental conditions are dif-ferent from the on-site situation Moreover, it is likely that most of the very labile metal fraction has already been leached over the years The percentages of metals solubilised by water, their median and ranges are reported in Table 6 As can be seen, the extracted
Fig 2 Combined plot of scores and loadings obtained by (a) PCA
and (b) dendrogram for horizontal profile samples at site B (total
metal concentrations).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111