A.
Santoro
a,
R.
Terzano
*a,
L.
Medici
b,
M.
Beciani
c,
A.
Pagnoni
c and
G.
Blo
c
aDipartimento di Biologia e Chimica Agroforestale ed Ambientale, University of Bari, Via Amendola 165/A, 70126, Bari, Italy. E-mail: r.terzano@agr.uniba.it
bConsiglio Nazionale delle Ricerche (CNR), Istituto di Metodologie per l'Analisi Ambientale (IMAA), Contrada S. Loja, I-85050, Tito Scalo, Potenza, Italy
cDipartimento di Chimica, University of Ferrara, Via L. Borsari 46, 44100, Ferrara, Italy
First published on 16th November 2011
Diverse analytical techniques are available to determine the particle size distribution of potentially toxic elements in matrices of environmental interest such as soil, sediments, freshwater and groundwater. However, a single technique is often not exhaustive enough to determine both particle size distribution and element concentration. In the present work, the investigation of mercury in soil samples collected from a polluted industrial site was performed by using a new analytical approach which makes use of sedimentation field-flow fractionation (SdFFF) coupled to cold vapour generation electrothermal atomic absorption spectroscopy (CV-ETAAS). The Hg concentration in the SdFFF fractions revealed a broad distribution from about 0.1 to 1 μm, roughly following the particle size distributions, presenting a maximum at about 400–700 nm in diameter. A correlation between the concentration of Hg in the colloidal fraction and organic matter (O.M.) content in the soil samples was also found. However, this correlation is less likely to be related to Hg sorption to soil O.M. but rather to the presence of colloidal mercuric sulfide particles whose size is probably controlled by the occurrence of dissolved O.M. The presence of O.M. could have prevented the aggregation of smaller particles, leading to an accumulation of mercuric sulfides in the colloidal fraction. In this respect, particle size distribution of soil samples can help to understand the role played by colloidal particles in mobilising mercury (also as insoluble compounds) and provide a significant contribution in determining the environmental impact of this toxic element.
Environmental impactSoil colloidal particles, if mobilised, can spread potentially toxic elements throughout the environment, thus endangering the health of living organisms and human beings. This issue is particularly relevant for elements in insoluble forms which anyway can migrate in water fluxes as submicronic particles. In this research an analytical method to correlate Hg concentration with particle size for the analysis of the colloidal fraction of soil samples has been developed and applied to soil samples collected in an area polluted by the activity of a chlor-alkali plant. Hg was found mostly concentrated in the fraction between 400 and 700 nm. Despite Hg being mostly speciated as insoluble mercuric sulfides, the small size of the particles could still create serious environmental risks. |
As a recognised global environmental pollutant,5–7mercury attracts special interest in soil science, also because it is often found concentrated in the clay (<2 μm) fraction.8
However, for many decades, determination of particle size distributions involving Hg, in natural heterogeneous samples, has been limited not only by the techniques available but also by modifications in the original sample (e.g. aggregation and disaggregation processes) likely to occur during sampling and sample preparation that can affect the measurements.9 More recently, together with traditional filtration and centrifugation, a wide range of methods (electron microscopy, laser diffraction, field-flow fractionation—FFF—techniques, etc.) for the characterisation of particle size distribution has been developed.9,10 In a detailed overview on fractionation methods, Leadet al.11 have pointed out that conventional techniques such as filtration, ultrafiltration and ultracentrifugation are all affected by numerous factors which can lead to a poor resolution of size distribution; whereas different FFF sub-techniques allow a higher resolution.12,13
The main advantage of FFF techniques attracting many scientists is the relatively low interaction between the sample and the separation system as well as the reduced alteration of the sample (e.g.coagulation) during fractionation steps. Another important advantage is the possibility of off-line or on-line coupling of a FFF technique with specific detection systems for a subsequent chemical particle characterisation.14–19
Among the FFF techniques, sedimentation field-flow fractionation (SdFFF) exploits a centrifugal field to separate the particles according to their buoyant mass, which is in turn proportional to the equivalent spherical diameter provided the particle density is constant.20 In a typical SdFFF instrument, the particles are first injected into the separation channel without the field or flows applied, and are distributed by diffusion across the thickness of the channel (injection point). When the field is applied, the solute zone is compressed by the applied centrifugal field into a narrow layer against one wall (relaxation at the accumulation wall). Then, the liquid flow is started and the solute zone is carried downstream at a rate depending on the particle layer thickness, which is related to particle size and density (separation). In the normal (or Brownian) mode of elution, this results in an increase in elution time for larger particles.
When SdFFF is coupled on-line or off-line to an atomic absorption spectrometer (AAS), the elemental characterisation of the separated particles can be determined.21 The UV-Vis signal versus the particle diameter fractogram obtained from SdFFF separation can be transformed into a graph showing the concentration of the element in particles (determined by AAS) versus diameters, giving the size distribution of the element of interest.22
The technique has already been used for the fractionation of colloids in natural water samples;12,16 however only a few studies, involving colloidal distribution of Mn23 or Cd,24 have been made on soil and sediment samples. In this respect, mercury determination in soils has rarely been considered due to the usual difficulties encountered in its analytical determination.
In this study, soil samples contaminated with Hg have been examined using SdFFF separation coupled to cold vapour generation electrothermal atomic absorption spectrometry (CV-ETAAS) to characterise the colloidal Hg-bearing fractions.
In the frame of a monitoring project aimed at evaluating the state of pollution of the area, numerous samples were collected all around the industrial site at different soil depths and by following different sampling schemes.25 Taking into account the dramatic dilution imposed by the SdFFF technique, only three samples (A1, A2, A3) containing the highest amounts of Hg were used in this study. These samples can be considered representative of the investigated area. A1 and A2 were collected at a depth of 0–10 cm, while A3 at 40–50 cm.
After collection, soil samples were homogenised, air dried, sieved at 2 mm, and analysed for the total Hg content. The soil clay fraction (<2 μm) was separated by using an in-house developed procedure, based on repeated sonication–centrifugation steps. This procedure facilitated the separation of the soil aggregates without modifying Hg speciation.28
An amount of 3.5 g of soil was mixed with 50 mL of ultrapure deionised water and stirred for 10 min. Then, the solution was sonicated for 10 min at 100 W by means of a Sonics Vibra-Cell (Sonics and Materials Inc., USA) to break the soil aggregates. After sonication, the suspension was centrifuged for 2.07 minutes at 600 rpm (according to Stokes' equation and assuming a particle density of 2.65 g cm−3 as these conditions yield the 2 μm diameter of the silt/clay boundary31) by using a Mod. CR15 (B. Braun Biotech International) centrifuge. Then the supernatant containing the clay fraction was recovered. The optimal number of repeated centrifugations, performed in order to quantitatively extract the clay fraction, was experimentally established as five repetitions. The pellets obtained by centrifugation of the supernatant for 30 minutes at 5500 rpm from each repetition were finally combined, weighed, lyophilised by using a Hetosicc lyophiliser (Hetolab equipment, Denmark) and stored at 4 °C for further analysis.
In order to verify the reliability of this procedure, two other methods for clay fraction extraction, i.e. repeated sedimentation–syphonation cycles32 and wet sieving,33 were used. The gravity sedimentation was performed by adding 40 g of soil to 400 mL of water in a 500 mL beaker.34 The solution was stirred overnight and then allowed to settle for 3 hours and 30 minutes. The material in the top 5 cm (<2 μm fraction) was recovered by syphonation, and the entire process was repeated 5 times. The five fractions were then collected together and the clay sized material was recovered by centrifugation at 5500 rpm for 30 minutes.
For the wet sieving, 60 g of air-dried 2 mm sieved soil were shaken in 400 mL of deionised water for 1 h at 50 rpm in an end-over-end stirrer, followed by wet sieving using ultrapure water through 1.18, 0.42 and 0.02 mm sieves; then the last fraction was filtered at 2 μm by using polycarbonate Nuclepore® (Whatman) filters.
The Hg concentration was determined in the <2 μm fraction for one of the samples (A3) extracted with all the three methods and the results were compared by using Student's t-test at 95% confidence interval.
XRD patterns were obtained using a Rigaku D-max Rapid micro-diffractometer operating at 40 kV and 30 mA with CuKα radiation and a flat graphite monochromator. Different mounts with corundum NIST 676 as internal standard were prepared and analysed: air-dried, glycolated at 60 °C, and heated at 375 °C. The XRD data were collected as two-dimensional images and converted into 2θ–I profiles. The quantification of the mineralogical phases was obtained by Rietveld refinement using EXPGUI software.35
Dimensional characterisation of soil colloidal sub-samples was performed by using an SdFFF Model S101 instrument (Postnova Analytics, Salt Lake City, UT, USA). The fractionation channel was made of two Hallestoy C inox bars and a Mylar spacer of nominal thickness 0.0254 cm, 2.0 cm nominal width and ca. 90 cm tip-to-tip in length. The channel void volume was 4.8 mL (w = 0.0282) and the axis-to-channel distance was ca. 15 cm. The flow stream, delivered by a Mod. 422 Master HPLC Pump (Kontron Instruments, Italy) was 1.5 mL min−1. The carrier solution used was 0.001% FL70 (Fisher Scientific, France), commonly used in the FFF experiments, to avoid the accumulation of impurities in the FFF channel.36,37
The characterisation of colloidal soil samples by SdFFF was performed using the normal mode, avoiding interference due to the steric mode (co-elution of smaller and larger particles).38 The soil <1 μm fraction was extracted by following the same procedure adopted for the fraction <2 μm with the exception of using a centrifugation time of 4.13 minutes.31 The soil sub-samples for injection in the SdFFF system were prepared by suspending 50 mg of <1 μm soil fraction in 50 mL of 0.001% FL70 surfactant. The suspended soil sub-samples were vigorously stirred for 1 min by using a vortex vibrator at a maximum frequency of 60 Hz before being injected into the SdFFF channel by using a 10 μL chromatographic syringe.
The outlet tube of the SdFFF system was connected to a UV detector operated at 254 nm (Uvidec 100, Jasco Ltf., Japan) and the signal was fed to a Linseis Mod. L6512 X-Y recorder (Linsel GmbH Selb, Germany). UV absorbance data were collected by an in-house data acquisition program and handled by using the FFFractionation Inc. software.
To achieve a good balance between resolution of the sample profile, speed of the analysis and also in order to collect sufficiently concentrated SdFFF eluted fractions to be used in further analytical determinations, some factors affecting SdFFF fractionation were checked and optimised.20 In general, better resolution is often achieved by increasing the field strength (with potential sample loss and longer time of analysis), while increasing the flow rate can speed up the analysis but at the expense of peak resolution.9 By carefully taking into account these factors, the optimal instrumental elution parameters for the particle separation in the analysed soil samples were determined to be the following: initial field G0 = 400 rpm, hold field = 20 rpm, initial delay time tl = 5 min, waiting time = 5 s and flow rate = 1.5 mL min−1.
Eluted sample fractions were collected every 3.5 min by a Mod. 2110 collector (Bio-Rad Laboratories, Italy). Since SdFFF dilutes the injected sample by a factor of 50, in order to have a detectable amount of Hg in the collected fractions, five FFF runs were repeated and the corresponding five fractions obtained were combined and then concentrated in an oven at 30 °C for 24 hours, producing a final volume of 25 mL.
The determination of mercury in the resulting suspension (10 fractions for each soil sample) was performed by cold vapour electrothermal atomic absorption spectroscopy (CV-ETAAS), by using a Mod. A Analyst™ 800 instrument (Perkin Elmer, USA), equipped with a Transversely Heated Graphite Furnace (THGA) with longitudinal Zeeman-effect background corrector. The reducing agent used for cold vapour generation was NaBH4 (0.2% in NaOH 0.05%), while HCl 3% was used as a carrier.
Samples were introduced into the system by an autosampler Mod. AS800 (Perkin Elmer, USA), equipped with a flow injection device (FIAS 100, Perkin Elmer, USA).
Samples were acidified with concentrated HNO3 to promote Hg solubilization and a known amount of sample (500 μl) was introduced into the FIAS loop and then transported into the system by a flow of 3% HCl. After reduction by NaBH4, Hg cold vapours were produced and transported by an inert gas (Ar) to the graphite tube, previously coated with an iridium pure standard solution (PerkinElmer, 1000 mg L−1) to improve the sensitivity of the instrument in determining Hg.39 The detection was operated by using a mercury electrodeless discharge lamp (EDL) at a wavelength of 253.7 nm.
The graphite furnace heating program is summarised in Table 1.
Step | Temperature/°C | Ramp/s | Hold/s | Ar flow/mL min−1 |
---|---|---|---|---|
Dry | 30 | 1 | 50 | 250 |
Pyrolysis | 50 | 1 | 20 | 250 |
Atomisation | 1200 | 0 | 4 | 0 |
Cleaning | 1400 | 1 | 3 | 250 |
Energy dispersive spectra were collected on a liquid nitrogen cooled Si(Li) INCAx-sight EDS detector (Oxford Instruments, England).
A few drops of the SdFFF fractions were distributed evenly over double-sided polycarbonate adhesive tape, and allowed to evaporate to dryness at room temperature. The samples were then placed on aluminium stubs for microscopic analysis.
In order to collect proper images and to avoid the accumulation of charges on the surface of the sample when bombarded with the electron beam, all the samples were sputter coated with a thin layer of conductive carbon.
Before analysing the soil samples (A1, A2, A3) with the CV-ETAAS system, two suitable calibration experiments (external calibration and standard additions) were also performed and the slopes of the two different calibration curves were evaluated for statistically significant difference at 95% confidence level (Student's t-test). The slopes of the two different calibration curves were not found to be significantly different, suggesting a reduced or inexistent matrix effect on Hg determinations by CV-ETAAS analysis.40
Instrumental Limit of Detection (LoD) and Quantification (LoQ) were evaluated using two approaches, according to Massart et al.41 by using linear regression analyses (95% confidence interval, α = 0.05), and as 3 and 10 times the standard deviation (σ) of 10 repeated measurements of the blank solution.42 According to these procedures, LoD and LoQ were evaluated as 0.44 μg L−1 and 1.5 μg L−1 or resulted 0.08 μg L−1 (3σ) and 0.27 μg L−1 (10σ). The highest values found were considered as the LoD and LoQ of the method: 0.44 μg L−1 and 1.5 μg L−1, respectively.
Sample | pH | TOC (%) | Sand (%) | Silt (%) | Clay (%) | Total Hg/μg g−1 | |
---|---|---|---|---|---|---|---|
<2 mm | <2 μm | ||||||
A1 | 8.0 | 3.0 | 53 | 30 | 16 | 50 ± 5 | 130 ± 10 |
A2 | 8.5 | 1.4 | 48 | 39 | 13 | 230 ± 40 | 200 ± 10 |
A3 | 8.5 | 0.7 | 40 | 39 | 21 | 12 ± 2 | 30 ± 2 |
The extraction of the fraction <2 μm with the sonication–centrifugation method developed in-house was compared to two other methods, a traditional wet sieving and sedimentation–syphonation, for sample A3. The amount of Hg determined by using the latter two methods was 28 ± 3 μg g−1 and 31.5 ± 0.5 μg g−1 respectively, which is not significantly different from that obtained by using the in-house developed method (Table 2). Nevertheless, the adopted sonication–centrifugation method proved to be faster than the two other methods and the soil aggregates were almost completely broken up. It must be pointed out that this type of sample preparation on one side could alter the real size of soil particles (i.e. by destroying natural aggregates), but on the other side allows one to obtain better separated materials. This aspect is relevant to study small primary soil particles which are of high environmental concern, as they can more actively sorb pollutants and easily migrate.
![]() | ||
Fig. 1 Fractograms obtained for the collected soil samples A1 (left), A2 (middle), A3 (right). |
![]() | ||
Fig. 2 SdFFF fractograms for A1, A2 and A3 samples, showing the particle distribution (μm) vs. the UV detector response (a.u.) (black line); and distribution of Hg concentrations (μg kg−1) detected by CV-ETAAS (dashed bars). In A1 and A3 graphs, the seventh fraction is not reported as it was below the limit of quantification (LoQ). |
To check the correspondence of the particle size in the eluted fractions to the particle size calculated by the SdFFF instrumentation, three fractions were collected, concentrated and submitted to SEM-EDS investigation. Fig. 3 shows that colloidal particles were eluted in the expected order of increasing size, confirming the expected theoretical fractionation from ca. 0.05 to 1.1 μm. The particles belonging to the three eluted fractions (Fig. 3A–C) show different characteristics (as for shape and size) suggesting a correct separation of the colloids. However, particle aggregates were observed, especially in fractions B and C, presumably due to self-assembly of the particles after fractionation and sample preparation for SEM analyses. In proximity to the void volume, the concurrent elution of particles with different sizes might have occurred, also due to the presence of non-spherical particles (especially disc shaped particles16,43).
![]() | ||
Fig. 3 SdFFF fractogram for A2 sample and related SEM-EDS micrographs corresponding to the fractions ≤100 nm (A), 500–700 nm (B) and ≥1 μm (C). |
The combined corresponding fractions collected from five repeated SdFFF runs for A1, A2 and A3 were analysed for total mercury content by using CV-ETAAS (Fig. 2, vertical bars).
Regarding sample A1, it was found that the mercury level was higher for the fraction around 0.65 μm, where the maximum of the particle distribution was also observed. Sample A2 presented a similar trend in Hg amount, with a maximum in concentration around 0.65 μm. Interestingly, the Hg distribution levels obtained for sample A2 are ca. one order of magnitude lower than the ones determined for A1, despite the fact that the total amount of Hg for the A2 soil is higher (230 μg g−1) than for A1 (50 μg g−1). As reported in Table 2, Hg concentration in A1 seems to increase going down with particle size from 2 mm to 2 μm, while the opposite happens for A2 (Table 2). Thus, it is likely that Hg concentration increases further in the colloidal fraction (<1 μm) for A1, while for A2 Hg is more concentrated between 1 μm and 2 μm. In the case of A3, the maximum of Hg distribution was observed in particles with a size of around 0.4 μm, where the maximum of particle distribution was also registered. In A3, the overall concentrations of Hg in the different colloidal fractions were found to be lower than for A1 and similar to those recorded for A2.
Qtz | Cal | Pla | Fel | Ill | Ver | I/S | Kao | Dol | Chl | Oth | |
---|---|---|---|---|---|---|---|---|---|---|---|
a Qtz: quartz; Cal: calcite; Pla: plagioclase; Fel: feldspate; Ill: illite; Ver: vermiculite; I/S: interstratified illite/smectite; Kao: kaolinite; Dol: dolomite; Chl: chlorite; Oth: others. | |||||||||||
<2 mm | |||||||||||
A1 | 49 | 21 | 11 | 8 | 4 | n.d | n.d. | 3 | 2 | 1 | 1 |
A2 | 27 | 19 | 11 | 4 | 34 | n.d. | n.d. | 4 | n.d. | 1 | |
A3 | 30 | 22 | 13 | 8 | 11 | n.d. | n.d. | 7 | 4 | 4 | 1 |
<2 μm | |||||||||||
A1 | 9 | 1 | 5 | 2 | 32 | n.d. | 13 | 18 | 10 | 10 | |
A2 | 20 | 13 | 3 | n.d. | 24 | n.d. | 25 | 10 | n.d. | 5 | |
A3 | 11 | 9 | 17 | 2 | 18 | 3 | 10 | 13 | 2 | 14 | 1 |
As a practical consideration, although in the soils investigated here Hg seems to be present in sparingly soluble forms (e.g. HgS), the levels of mercury found in the colloidal fractions need to be taken into account when assessing the environmental risk in the area, since pollutant mobility through soil may be higher than expected because of colloidal transport.
In this respect, a more extensive investigation, on a larger scale, taking into account a broader range of influential factors as well as the element speciation, needs to be carried out.
In the collected samples, a maximum in Hg concentration was also found to correspond to the maximum in the particle size distribution, which was in a size range of a few hundred nanometres (400–700 nm). An apparent relation between Hg concentration in the colloidal fraction and organic matter (O.M.) content in the soil samples was observed but it seems to be determined not by Hg sorption to O.M., but rather to colloidal mercuric sulfide particles whose size is likely to be influenced by dissolved O.M.
Several authors3,24,47 have already demonstrated that small colloidal particles can diffuse more easily through the environmental compartments via transportation with hydrologic fluxes, and turn into a serious threat to the environment. Mobilisation of colloidal particles such as small airborne particles can also be a source of potential risk for humans if inhaled.
Therefore, even if Hg is found in the environment as scarcely soluble forms (e.g. HgS), the colloidal size of the Hg-bearing particles should raise the level of attention of regulators, as in this form it can more easily migrate, dissolve, be chemically modified or ingested.
This journal is © The Royal Society of Chemistry 2012 |