Environmental Monitoring and Assessment

, Volume 186, Issue 2, pp 961–969

Evaluation of diffusive gradients in thin films technique (DGT) for measuring Al, Cd, Co, Cu, Mn, Ni, and Zn in Amazonian rivers

Authors

  • Lauren Nozomi Marques Yabuki
    • Programa de Pós-Graduação em Geociências e Meio Ambiente, Instituto de Geociências e Ciências Exatas - IGCEUniversidade Estadual Paulista - UNESP
    • Centro de Estudos Ambientais - CEAUniversidade Estadual Paulista - UNESP
  • Camila Destro Colaço
    • Programa de Pós-Graduação em Geociências e Meio Ambiente, Instituto de Geociências e Ciências Exatas - IGCEUniversidade Estadual Paulista - UNESP
    • Centro de Estudos Ambientais - CEAUniversidade Estadual Paulista - UNESP
    • Centro de Estudos Ambientais - CEAUniversidade Estadual Paulista - UNESP
  • Roberto Naves Domingos
    • Centro de Estudos Ambientais - CEAUniversidade Estadual Paulista - UNESP
  • Chang Hung Kiang
    • Laboratório de Estudos de Bacias - LEBAC, Instituto de Geociências e Ciências Exatas - IGCEUniversidade Estadual Paulista - UNESP
  • Domitila Pascoaloto
    • Coordenação de Pesquisas em Clima e Recursos HídricosInstituto Nacional de Pesquisas da Amazônia - INPA
Article

DOI: 10.1007/s10661-013-3430-x

Cite this article as:
Yabuki, L.N.M., Colaço, C.D., Menegário, A.A. et al. Environ Monit Assess (2014) 186: 961. doi:10.1007/s10661-013-3430-x

Abstract

Studies concerning the lability and bioavailability of trace metals have played a prominent role in the search for contamination of water resources. This work describes the first application yet of the diffusive gradients in thin films technique (DGT) to the determination of the fraction of free plus labile metals in waters from the Amazon Basin. Due to the complexity of the use of DGT for samples with low ionic strength and high organic matter content (characteristic of Amazonian rivers), a new analytical procedure was developed. The method is based on the determinations of apparent diffusion coefficients (Dap) in the laboratory, by performing deployments in samples collected in the corresponding sites of study. The Dap thereby determined is then used for in situ measurements. The suitability of the proposed approach for determination of labile Al, Cd, Co, Cu, Mn, Ni, and Zn in the Amazon River and Rio Negro (English: Black River) was evaluated. Except for Co, Mn (in a deployment at Rio Negro), Ni and Zn (in a deployment at Amazon River), labile in situ measurements were lower or similar to dissolved concentrations, indicating suitability of the proposed approach.

Keywords

DGTAmazon BasinLabilityTrace metals

Introduction

The Amazon Basin is the largest in the world, with an area of 5,846,100 km2. The area includes portions of those countries that surround the Amazon River. This includes both portions of those countries through which the Amazon River directly passes, namely Peru, Colombia, and Brazil, as well as portions of those countries that lie in the vicinity of the river, namely Bolivia, Ecuador, Venezuela (small stretches only), and Guyana (formerly British Guiana). The Amazon River has 5,825 km of extension and its discharge is by far the most voluminous (216,342 m3 s−1) of all of the world's rivers. It is estimated that this discharge is equivalent to 11 % of the entire mass of continental waters (Cunha and Pascoaloto 2009).

Among the hundreds of tributaries of the Amazon, Rio Negro (English: Black River) is considered as the most important, due to its discharge (154.62 m3 s−1). Rio Negro is an Amazon tributary (left margin) with 1,700 km of total extension (approximately 1,200 km of which lie in Brazil). The waters of Rio Negro are very acidic (pH <5) and contain high concentrations of humic material, which explains the river's black coloration. This river has relatively low concentrations of dissolved salts (the average of cations and anions being 0.72 and 1.62 mg L−1, respectively, predominantly potassium and sulfate) (Cunha and Pascoaloto 2009).

Studies are important on chemical speciation (involving the determination of the species of an element) because of the fact that toxicity, bioavailability, transport, and physical–chemical properties of an element may differ widely depending on its chemical form (Tonello et al. 2011). Research has been confirming the importance of conducting in situ measurements of the labile fraction of metals, because the latter is considered potentially bioavailable (Moore and Ramamoorthy 1984).

The diffusive gradients in thin films technique (DGT) is a promising tool for in situ analysis in aquatic systems. The DGT device uses a diffusion layer, conventionally a polyacrylamide hydrogel, and a binding agent, typically the Chelex-100 resin, impregnated within the polyacrylamide hydrogel. The function of the diffusion layer is to promote a constant level of diffusion of the metallic element to be retained by the binding agent, independently of changes in the flow system where the device is immersed.

Conventional DGT design allows the determination of free metals, labile inorganic and complex organic compounds, excluding large particles and colloids (which are not quantified) (Zhang and Davison 1995). The technique is based on Fick's first law of diffusion, which describes the diffusion of matter in a medium in which, initially, there is no chemical equilibrium. A concentration gradient produces a stream of particles, which tends to homogenize the solution and standardize the concentration (Atkins 1982).

The following equations exposed by Tonello et al. (2011) are used in the determination:

The flow of ions (F) that diffuse from the diffusive gel to the binder resin is expressed by:
$$ F=D\left({C}_{\mathrm{b}}-C\prime \right)/\varDelta g $$
(1)
where D is the diffusion coefficient of the gel, Cb is the labile concentration of metal in the external solution, C' is the concentration of free metal in the resin, and Δg is the thickness of the diffusive gel.
Equation (1) can be simplified when the metal ions are in rapid equilibrium with the resin. In such cases, the concentration C' is effectively zero:
$$ F=D{C}_{\mathrm{b}}/\varDelta g $$
(2)
By definition of flux, F = M/A·t, where M is the mass of the analyte, A is the area of the window device, and t is the time, Equation (2) can, therefore, be rewritten as:
$$ D\;{C}_{\mathrm{b}}=\left(M\;\varDelta g\right)/\left(t\;A\right) $$
(3)
After M has been determined by an analytical technique, the concentration Cb can be calculated by using the following reformulation of Eq. (3):
$$ {C}_{\mathrm{b}}=\left(M\;\varDelta g\right)/\left(t\;D\;A\right) $$
(4)

DGT has the advantage of enabling preconcentration of the analytes, multielemental sampling, and time-integrated measurements, the latter ensuring quantification of short-term events (Zhang and Davison 1995). However, the technique has some limitations as well, such as requiring prior knowledge of the analyte diffusion (diffusions of free ions (M)—or small inorganic complex metals (metal inorganic ligant)—and organic complex metals (metal–humic substance—Me-HS) being considerably different) and of the retention capacity of the Chelex resin. In addition, it cannot be used for systems with ionic strength lower than 0.1 mmol L−1.

Alfaro-De La Torre et al. (2000) performed tests with DGT devices in the waters of Lake Tantaré (Canada). Results showed higher concentrations of Cd and Ni as compared with those detected by other in situ sampling techniques (dialysis samplers). Since the publication of this work, the use of DGT for aquatic systems with low concentration of cations has been seen to have some restrictions.

Warnken et al. (2008), aiming to improve DGT measurements in low ionic strength samples, performed various types of washes and subsequent conditioning (by using solutions with 1.0 and 10 mmol L−1 sodium nitrate) in diffusive gel for determination of Cu and Cd. The results were erratic at ionic strength of 0.1 mmol L−1.

Another approach has been proposed for analyses of organic-rich river waters (Tonello et al. 2007, 2011). A DGT technique using different gel pore sizes was used to determine the concentrations of free, labile inorganic and organic species of Al and Cu in samples of water from two rivers (with high concentrations of dissolved organic carbon (DOC)). Higher concentration values obtained by in situ sampling, in relation to those obtained in the laboratory, indicated considerable changes of the analytes in the samples (Tonello et al. 2007). Alternatively, a procedure based on ultrafiltration data has been proposed to determine apparent diffusion coefficients of the analytes in water samples and model solutions containing both free metal ions (M) and complex metals (Me–H) (Tonello et al. 2011). However, both approaches require previous knowledge of diffusion coefficients of metal-HS complex, which is difficult to obtain. Additionally, these approaches cannot be applied to systems containing low ionic strength, such as the rivers in the Amazon Basin.

The main objective of this study was to develop an effective method to assess the free plus labile fraction of the metals Al, Cd, Co, Cu, Mn, Ni, and Zn in two rivers, the Amazon River and Rio Negro, both of which are aquatic systems with low ionic strength and/or high organic matter content.

Materials and methods

Total, dissolved, and labile-plus free metal concentration levels determinations were performed using an X Series II (Thermo Electron Corporation, Bremen, Germany) inductively coupled plasma mass spectrometer (ICP-MS) with a concentric nebulizer (Glass Expansion – Melbourne, Australia) and a spray chamber in a temperature-controlled environment. DOC determinations were performed using a GE Sievers InnovOx Laboratory Total Organic Carbon (TOC) Analyzer.

DGT holders, a Chelex-100 disk, and a diffusive disk (composed of polyacrylamide gels) were purchased from DGT Research (Lancaster, UK).

Reagents and solutions

Sub-boiling nitric acid (Merck, Darmstadt, Germany) and a purified water deionizer in a Milli-Q system (Millipore, Bedford, MA, USA) with a resistivity of 18.2 mΩ cm were used. The working standard solutions were prepared from a multielement 100 mg L−1 stock solution of the analytes (SpecSol, São Paulo, Brazil).

The Chelex-100 resin (200–400 mesh) obtained from Bio-Rad Labs (Richmond, CA, USA) in sodium form was used.

Ammonium acetate buffer solutions of pH 6.5 were prepared from NaC2H3O2 (QUIMIS, São Paulo, Brazil) and NH3 (Merck, Darmstadt, Germany). In order to avoid contamination of the solutions, they were purified by a batch solid phase extraction (SPE) with Chelex-100.

Samples

Water samples (for laboratory deployment) were collected in polyethylene flasks that had previously been decontaminated, simultaneously with in situ immersion of the DGT devices. Sampling sites and coordinates included the following: Rio Negro I S03.030316 W060.25235, Rio Negro II S03.14505 W060.02042, and the Amazon River S03.14839 W058.44892. For a more comprehensive study, water samples for analysis of total and dissolved metals were also collected. Dissolved samples were immediately filtered through a 0.45-μm Millipore filter into 50-mL Falcon tubes. For analysis of total concentrations, samples were digested in a microwave oven with ultrapure HNO3, following the EPA 3010A method from US Environmental Protection Agency.

Table 1 shows physical chemical characteristics for water samples collected in Rio Negro and in the Amazon River. The pH and DOC values of each river are consistent with studies and classification of waters by Sioli (1984). All water samples were collected at the same season when performing the deployments. Thus, there was no influence of precipitation on studies, as evidenced by studies of Silva et al. (2008).
Table 1

Physical chemical characteristics of water samples collected in the Amazon River and Rio Negro

Parameters

Amazonas

(21 May 2011)

Rio Negro I

(25 May 2011)

Rio Negro II

(05 July 2011)

pH

6.4

5.2

5.0

Temperature (°C)

28.9

28.0

29.1

Conductivity (ms cm−1)

40.7

8.6

7.0

DOC (mg L−1)a

7.80 ± 0.21

15.30 ± 0.24

10.00 ± 0.14

Ionic strength (mmol L−1)b

0.53

0.11

0.08

TDS (mg L−1)c

26.0

5.5

4.5

aDissolved organic carbon (DOC), determined with a GE Sievers Innovox Laboratory Total Organic Carbon (TOC) Analyzer

bIonic strength calculated from the conductivity values, Jurinak and Griffin (1973), cited by Lindsay (1979)

cTotal dissolved solid (TDS) calculated from the conductivity (Water Treatment Solutions Lenntech)

General procedure for assembling, deployment, and elution of DGT devices

The binding layer (Chelex-100 immobilized on agarose–polyacrylamide gel) was placed on the piston of the DGT device, followed by a diffusion layer of agarose–polyacrylamide gel, and finally, a layer of cellulose acetate filter. To avoid contamination, the procedure was performed in a laminar flow hood.

The DGT devices were deployed in situ and in the laboratory. Laboratory deployments were carried out under constant agitation for a period of 24–72 h. The immersion temperature was kept constant at 23 ± 1 °C. In situ deployment varied from 1 to 7 days.

After deployments, the Chelex-100 disks were removed from the device and transferred to Falcon-type tubes (15 mL) containing 2 mL of 1 mol L−1 nitric acid and maintained under constant agitation for 24 h. Subsequently, the disks were removed from the tube and the solution was analyzed by means of ICP-MS. Elution factors (calculated by dividing the eluted mass by the adsorbed mass) of Al, Cd, Co, Cu, Mn, Ni, and Zn were found for 0.83, 0.86, 0.78, 0.75, 0.91, 0.81, and 0.89, respectively.

Procedure for determination of the apparent diffusion coefficient

The method for determination of the apparent diffusion coefficient (Dap) is based on determination of the diffusion coefficient of the analyte in the laboratory from a sample collected from the study site. The diffusion coefficient thereby determined is then used for in situ determinations.

Dap is calculated according to Eq. (5):
$$ {\mathrm{D}}_{\mathrm{ap}}=\left(M\;\varDelta g\right).{\left(A\;C\;t\right)}^{-1} $$
(5)
where Δg = 0.093 cm and A = 3.14 cm2. C is the labile-plus free concentration of the analyte in the sample (determined by solid phase extraction), M (cumulative mass in nanogram), and t (time in seconds). Considering the angular coefficient (a = M/t) of deployment curve in the respective sample (mass × time), then Dap is calculated according to Eq. (6):
$$ {\mathrm{D}}_{\mathrm{ap}}=\left(a\;\varDelta g\right).{\left(C\;A\right)}^{-1} $$
(6)

Solid phase extraction

SPE was performed using Chelex-100 resin followed by a similar procedure previously proposed by Yabutani et al. (1999): Adding 45 ml of standards or samples in 50-mL Falcon tubes; adjusting the pH to 6.5 ± 0.5 with 0.2 mol L−1 of acetic acid/ammonium acetate buffer; adding 1.5 mL of 10 % w/v Chelex-100 suspension (200–400 mesh); agitating the tubes for 2 h in an orbital shaker; filtering the solution through a 0.45-mm syringe filter; washing the resin with 15 ml of Ultrapure water; eluting the metals with 2 ml of 2 mol L−1 nitric acid; and analyzing the metals by means of ICP-MS.

Results and discussion

Solid phase extraction (SPE)

Using the SPE procedure described above, preconcentration factors obtained for Al, Cd, Co, Cu, Mn, Ni, and Zn were 4.2, 4.5, 7.6, 8.2, 7.0, 8.2, and 24.3, respectively. The preconcentration factors were determined from the ratio between the slopes of the calibration curves obtained with and without extraction. Labile-plus free concentrations of the analytes in the samples (Rio Negro I, Rio Negro II, and Amazon River) determined by SPE are presented in Table 2.
Table 2

Linear functions obtained from the ratio of mass per time (deployment curves—ng s−1) for Al, Cd, Co, Cu, Mn, Ni, and Zn

 

Coefficients

Labile fraction (SPE) μg L−1a

Angular (ng s−1)

Linear (ng)

R2

Negro I

 Al

0.002

6.8

0.96

144.4 ± 0.4

 Cd

0.000005

3.3

0.65

<LQ (0.02)

 Co

0.00003

2.6

0.97

0.119 ± 0.003

 Cu

0.0001

4.5

0.50

0.24 ± 0.01

 Mn

0.002

−28.4

0.97

12.1 ± 0.2

 Ni

0.0001

2.7

0.92

0.144 ± 0.008

 Zn

0.0007

3.8

0.75

1.66 ± 0.08

Negro II

 Al

0.004

−94

0.99

176 ± 59

 Cd

<LQ (0.02)

 Co

0.00004

1.7

0.98

0.079 ± 0.001

 Cu

0.00002

21.0

0.15

0.2 ± 0.1

 Mn

0.002

−2.3

0.98

7.8 ± 0.2

 Ni

0.0002

21.8

0.94

0.10 ± 0.02

 Zn

0.001

49.4

0.89

7 ± 5

Amazon

 Al

0.005

75.3

0.78

31 ± 20

 Cd

0.000002

0.5

0.95

0.009 ± 0.006

 Co

0.000005

4.0

0.54

0.009 ± 0.003

 Cu

0.00009

7.4

0.82

0.62 ± 0.02

 Mn

0.0003

11.4

0.45

3.1 ± 0.8

 Ni

0.00015

5.6

0.96

0.24 ± 0.03

 Zn

0.0008

50.5

0.64

4.1 ± 3.5

aMean ± standard deviation, n = 3

Determination of apparent diffusion coefficients (Dap)

Table 2 shows data obtained from deployment curves (in nanograms per second) for Al, Cd, Co, Cu, Mn, Ni, and Zn. For samples from Rio Negro, except for Cd and Cu, the curves show a satisfactory (acceptable to our purpose) linear relationship (R2 from 0.75 to 0.98) thereby showing retention of the element by the resin. An unsatisfactory linear relationship was obtained for Cu. For example, Warnken et al. (2008), performing in situ studies in the Wyre River (UK), with a high content of DOC (15 mg L−1), reported only 5.51 of entirely labile Cu. Detection limits were inadequate for determining Cd in the Rio Negro samples. For samples from the Amazon River, unsatisfactory linear relationships only obtained for Co and Mn (values of R2 0.54 and 0.45, respectively), suggesting strong interaction of these elements with particulate material. Linear coefficients were very close to zero, indicating that there was almost no contamination from the resin or the acid elution.

From the slope of the deployment curve and by applying Eq. (5), the Dap for each of the analytes was determined. Figures 1, 2, and 3 compare the theoretical diffusion coefficients (Zhang and Davison 1999) with the apparent calculated diffusion coefficients for the samples from Rio Negro I, Rio Negro II, and the Amazon River, respectively. To facilitate comparison of the results, all the diffusion coefficients had their values adjusted to a temperature of 25 °C, using the Stokes–Einstein equation, which provides an approximate relation between the diffusion coefficient, viscosity, and temperature (Miller 1924). For all samples, ionic strengths were lower than 1 mmol L−1. Thus, change of diffusion coefficients of the analytes is expected. High levels of DOC suggest the presence of organic complex in the samples.
https://static-content.springer.com/image/art%3A10.1007%2Fs10661-013-3430-x/MediaObjects/10661_2013_3430_Fig1_HTML.gif
Fig. 1

Theoretical (Zhang and Davison 1999) and apparent (25 °C) diffusion coefficients obtained from Rio Negro I sample

https://static-content.springer.com/image/art%3A10.1007%2Fs10661-013-3430-x/MediaObjects/10661_2013_3430_Fig2_HTML.gif
Fig. 2

Theoretical (Zhang and Davison 1999) and apparent (25 °C) diffusion coefficients obtained from Rio Negro II sample

https://static-content.springer.com/image/art%3A10.1007%2Fs10661-013-3430-x/MediaObjects/10661_2013_3430_Fig3_HTML.gif
Fig. 3

Theoretical (Zhang and Davison 1999) and apparent (25 °C) diffusion coefficients obtained from Amazon River sample

Apparent diffusion coefficients for Co, Mn, and Zn found in the Rio Negro samples (Figs. 1 and 2) were similar to or higher than theoretical values. Forsberg et al. (2006) compared the DGT and ultrafiltration techniques for the metals Cd, Mn, and Zn. The results were similar, suggesting a weak tendency of these metals to produce complexes with organic compounds. Thus, for these elements, changes in diffusion coefficients can occur, due to the low ionic strengths of the samples. The value of the Ni diffusion coefficient found by Scally et al. (2006) for an ionic strength of 0.1 mmol L−1 and a temperature of 25 °C was 5.77 10−6 cm2 s−1. This value is much lower than those found for water samples from Rio Negro (Figs. 1 and 2), despite the possibility of Ni interacting with organic material.

Apparent diffusion coefficients for Al were (in the case of the Amazon River sample) similar to or (in the cases of Rio Negro I and II samples) lower than theoretical values, indicating complexation of Al with organic matter only in the samples from Rio Negro. Tonello et al. (2007) reported labile organic Al-HS complex for samples containing high amounts of DOC (from 14 to 47 mg L−1). This fact was also reported by Jansen et al. (2001), where the percentage of free metal plus soluble inorganic complexes to Al in the soil solutions with pH 4 and DOC content (75 mg L−1) was 11.8 %. For pH 7, the free fraction decreased to 3.5 %.

For the Amazon River sample, the mean value of apparent diffusion coefficient of Cd was 36–38 % higher than that reported by Zhang and Davison (1999) and Larner and Seen (2005), who obtained the values of 5.37 10−6 cm2 s−1 for the polyacrylamide gel of ionic strength 0.1 mol L−1 and pH 6.0 (Zhang and Davison 1999) and 5.41 10−6 cm2 s−1, for the polyacrylamide gel of ionic strength, 5 10−2 mol L−1 and pH 6.5 (Larner and Seen 2005). Similar behaviors were observed for Ni and Zn, indicating a low level of formation of Me-HS complexes in this sample, since concentration of DOC is relatively low in the Amazon River.

In situ analysis

Results obtained for in situ labile-plus free (DGT measurements), dissolved, and total measurements in samples from Rio Negro (site I), Rio Negro (site II), and the Amazon River are shown in Tables 3, 4, and 5, respectively. The concentrations were obtained using Dap (Dap were normalized to sampling temperature by using the Stokes–Einstein equation (Miller 1924)). Except for Co, Mn (in deployment at Rio Negro I), Ni, and Zn (in deployment at Amazon River), labile-plus free fractions were similar or lower than dissolved fraction, indicating suitability of the proposed approach. The disagreement between the results for Co, Mn, Ni, and Zn occurred possibly due to contaminations or imprecision from measurement of apparent diffusion coefficients that can introduce addition error on determination of labile concentration in the samples. Also, it should be pointed out that the measurements of DGT represent the concentration mean value during deployment and dissolved fraction are discrete measurements (before and after deployments).
Table 3

Total, dissolved, and in situ labile fractions (μg L−1), Rio Negro I

 

LQa

Totalb

Dissolvedb

DGT labilec

Al

2.3

257 ± 81

158 ± 27

67 ± 17

Co

0.06

<LQ

<LQ

0.14 ± 0.02

Mn

0.02

12.7 ± 2.7

10.3 ± 1.8

17 ± 3

Ni

0.3

0.41 ± 0.09

<LQ

0.3 ± 0.1

Zn

6.0

<LQ

<LQ

<LQ

aLQ (10 σ criterium) considering standard deviation of blank solution (or DGT blank) and sensitivity from direct determination (improvement from DGT preconcentration is not included)

bMean ± confidence limit, n = 2 (measurements before and after deployment)

cMean ± confidence limit, n = 3 (deployment of three DGT sampler)

Table 4

Total, dissolved, and in situ labile fractions (μg L−1), Rio Negro II

 

LQa

Totalb

Dissolvedb

DGT labilec

Al

2.3

239 ± 8.1

237 ± 81

125 ± 10

Co

0.06

<LQ

<LQ

0.051 ± 0.02

Mn

0.02

9.6 ± 0.7

8.6 ± 0.9

12 ± 5

Ni

0.3

<LQ

<LQ

0.012 ± 0.009

Zn

6.0

<LQ

<LQ

1.6 ± 0.7

aLQ (10 σ criterium) considering standard deviation of blank solution (or DGT blank) and sensitivity from direct determination (improvement from DGT preconcentration is not included)

bMean ± confidence limit (0.05 level), n = 2 (measurements before and after deployment)

cMean ± confidence limit (0.05 level), n = 3 (deployment of three DGT sampler)

Table 5

Total, dissolved, and in situ labile fractions (μg L−1), Amazon River

 

LQa

Totalb

Dissolvedb

DGT labilec

Al

2.3

1383 ± 342

32.6 ± 4.5

20.0d

Cd

0.06

<LQ

0.021 ± 0.04

0.021d

Cu

0.02

2.9 ± 0.9

2.20 ± 0.7

2.3 ± 1.0

Ni

0.3

2.0 ± 0.6

0.76 ± 0.09

2.6 ± 0.6

Zn

6.0

<LQ

<LQ

13.7 ± 0.2

aLQ (10 σ criterium) considering standard deviation of blank solution (or DGT blank) and sensitivity from direct determination (improvement from DGT preconcentration is not included)

bMean ± confidence limit (0.05 level), n = 2 (measurements before and after deployment)

cMean ± confidence limit (0.05 level), n = 3 (deployment of three DGT sampler)

dn = 1

Total and dissolved Co and Zn were not detected in Rio Negro (Tables 3 and 4). Horbe and Oliveira (2008) conducted studies of black water in the northeastern portion of the Amazon River, reporting low concentrations of dissolved metals. The high dilution and very low concentrations of trace elements are typical of geochemical intemperism in humid tropical conditions, where there is an intense and rapid percolation of water. On the other hand, labile-plus free Co and Zn concentrations were quantified in these samples, as a consequence of the ability provided by DGT technique to handle preconcentrations.

Relatively high levels of labile Zn measured in the Amazon River sample (Table 5) by using DGT suggests that contamination occurred during the sampling procedure.

The results found for Ni in the Rio Negro samples (Tables 3 and 4) are in disagreement with those suggested by Windermere Humic Aqueous Model 6 form models predictions, according to which it has been estimated that between 40 and 60 % of Ni occurs as inorganic species (Warnken et al. 2008). For the Amazon River samples, the labile-plus free Ni was similar to the total fraction.

For Rio Negro samples (Tables 3 and 4), labile-plus Mn concentrations were similar to the total and dissolved fractions, as expected, based on the low degree of interaction of this element with organic material. Dupré et al. (1999) performed ultrafiltration experiments in natural waters with DOC content between 30 and 40 mg L−1 (content higher than that found in the Rio Negro samples) at pH 3, showing that over 95 % of Al, 25 % of Cu, 10 % of Co, and 5 % of Mn, is complexed with the organic material. The tendency of the mean value of the labile (plus free) fraction to be higher than the dissolved fraction (Tables 3 and 4) can be explained by contamination or losses of soluble Mn during sample preservation.

The difference between the levels of total and dissolved Al was higher in the Rio Negro I sample (Table 3), possibly because the collection was performed in May, when there is a higher precipitation, which can dilute the aquatic system and, consequentially, decrease the dissolved fraction (Pinto et al. 2009). The total Al is not decreased because particulate material is carried to the river. A high concentration of total Al in the Amazon River sample (Table 5) has also been reported by Miranda et al. (2009).

For the Rio Negro and Amazon River samples, labile-plus free Al was about 42–62 % of the soluble fraction, confirming the largest mobilization of particulate Al (Dupré et al. 1999).

Labile-plus free Cu concentrations in the Amazon River sample were similar to the total and dissolved fractions (Table 5). Zhang and Davison (2000) reported a value of total labile concentration of Cu (sum of labile inorganic and organic) which is about 10 % less than the total dissolved concentration measured by ETAAS in filtered samples, in natural water with high concentrations of DOC (14.6 mg L−1).

Denney et al. (1999) performed studies using DGT in samples collected from the Ring and Stitt rivers, reporting that 70 % of dissolved Cd and Cu were in labile or free form. Also, good agreement between DGT and the fraction dissolved in the Ring and Stitt Rivers samples were reported, suggesting that strong complexation with organic matter is not occurring. The result was expected for Cd, which is known to be weakly complexed, but the value was surprising for Cu. The Stitt River has a relatively low concentration of Cu and a high concentration of DOC (7.3 mg L−1), while the Ring River has a high concentration of Cu combined with low concentrations of DOC (4.2 mg L−1) and pH 5, which may have mitigated the complexation of Cu. The Amazon River samples showed a similar DOC concentration and a similar Cu labile (plus free) fraction to those reported for the Stitt River sample.

Conclusions

The proposed approach, based on the determination of apparent diffusion coefficients from deployment curves for the samples, provided adequate values, since labile (plus free) fraction were similar or lower than dissolved fraction, except for Co and Mn (in a deployment at Rio Negro) and Ni and Zn (in a deployment at Amazon River), possibly due to contaminations or imprecision from measurements of apparent diffusion coefficients or dissolved fractions. Additionally, the precision of the results determined by this procedure was confirmed by the good correlation obtained between the deployment curves. In this way, it was possible to obtain the apparent diffusion coefficients for the water samples from the two rivers.

For the first time, in situ data (based on DGT measurements) with respect to lability of Al, Cd, Co, Cu, Mn, Ni, and Zn in Rio Negro and the Amazon River is reported. The proposed method may also be used for other types of water system which have low ionic strength and/or high DOC concentrations.

Acknowledgments

The authors thank the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), and CAPES for financial support.

Copyright information

© Springer Science+Business Media Dordrecht 2013