1
The dissociation kinetics of Cu-dissolved organic
matter complexes from soil and soil amendments
F. Amery, F. Degryse, C. Van Moorleghem, M. Duyck and E. Smolders
Division Soil and Water Management, Department Earth and Environmental Sciences, K.U.Leuven,
Kasteelpark Arenberg 20, 3001 Heverlee, Belgium
Corresponding author: F. Amery. Tel.: + 32 (0)16 32 14 12. Fax.: + 32 (0)16 32 19 97. Email:
Fien.Amery@ees.kuleuven.be
2
1
Abstract
2
Complexes between dissolved organic matter (DOM) and copper (Cu) that dissociate
3
very slowly can theoretically facilitate Cu leaching to the groundwater. Data on
4
dissociation kinetics of Cu-DOM complexes present in soil and in soil amendments
5
are limited. The dissociation kinetics of different Cu-DOM complexes from soil,
6
wastewater, pig manure and sewage sludge was measured with the Competitive
7
Ligand Exchange Method (CLEM) and Diffusive Gradient in Thin films (DGT)
8
technique. The solutions were set at constant pH, Ca concentration and free Cu2+
9
activity to allow comparison between the different samples. The average dissociation
10
rate constant kd of the complexes, as measured by CLEM, was about 10-3 s-1 and the
11
fractions of dissolved Cu that were undissociated after 8 hours ranged from <1 to
12
25%. These fractions determined by CLEM were significantly correlated with the
13
non-labile fractions (0–82%) determined in the DGT tests and data analysis show that
14
DGT data can be predicted from CLEM data. The dissociation rates decreased when
15
Cu-DOM complexes had been equilibrated at lower Cu2+ activities. Increasing the Cu-
16
DOM contact time (7–297 days) decreased the dissociation rate. The non-labile
17
fractions were positively correlated with the specific UV absorbance suggesting that
18
aromatic moieties in DOM hold non-labile Cu. All natural Cu-DOM complexes
19
contained a detectable fraction with a dissociation rate constant kd lower than 10-5 s-1
20
which can theoretically lead to non-equilibrium conditions and leaching risks in soil.
21
22
Keywords
23
Copper, Dissolved organic matter, complexation, dissociation, kinetics, soil
3
24
1. Introduction
25
The application of wastewater, sludge and manure are the main sources of copper (Cu)
26
enrichment in soils [1,2]. Leaching of Cu from soil to deeper groundwater can affect
27
groundwater quality. Column and field experiments have shown that Cu mobility
28
increases with increasing dissolved organic carbon (DOC) concentration [3-6].
29
Wastewater, sludge and manure are not only sources of Cu but also temporarily enhance
30
DOM concentrations. Therefore, it has been suggested that transport of Cu by Cu-DOM
31
complexes below the zone of incorporation might occur directly after application [2,7].
32
Several field studies with sewage sludge have shown that the mass balance of Cu in the
33
topsoil is not closed, suggesting losses of Cu by leaching [2,8-11]. Increased Cu
34
concentrations in leachates have been observed after sewage sludge application in several
35
[1,2,12,13] but not all field studies [14,15]. Sidle and Kardos [13] noticed decreasing Cu
36
concentrations in the lysimeter leachates with increasing depth, suggesting Cu
37
readsorption at deeper soil layers. However, the increase in soil total Cu under the
38
incorporation zone is commonly not detectable [8,9,13,16-18] or is small and/or
39
inconsistent [1,2,10,11,19]. Speir et al. [18] measured elevated Cu concentrations in the
40
groundwater after sludge application. These results suggest that Cu may be transported
41
down the soil without readsorption in deeper soil layers. This unretarded transport of Cu
42
may be related to preferential water flow paths [2], or non-equilibrium conditions for the
43
Cu-DOM complex dissociation or Cu2+ sorption to soil particles [20]. If dissociation of
44
the Cu-DOM complex is slow compared to the soil pore-water velocity, Cu can be
45
transported by the percolating soil water instead of being readsorbed by the solid phase.
46
4
47
Complexes can be operationally divided in labile and non-labile (or inert) fractions
48
depending on whether or not they dissociate within a defined reaction time. Alternatively,
49
dissociation kinetics can be expressed by the dissociation rate constant (kd) of the Cu-
50
DOM complexes. The dissociation kinetics of DOM complexes is strictly not
51
characterized by one single rate constant since DOM is a heterogeneous mixture of
52
binding sites. Therefore, DOM is sometimes divided in fractions with different lability as
53
expressed by their reaction rate constant [21].
54
55
There are various methodologies to characterize dissociation kinetics, which inherently
56
leads to differences in the outcome. In isotopic exchange experiments, an isotope is
57
added as a free metal ion, equilibrated for a given exchange time and the solution is
58
subsequently fractionated by ultrafiltration, resin purification or Donnan dialysis. Isotopic
59
shifts between the free ion or the smallest size fraction and the total solution allow the
60
detection of non-isotopically exchangeable Cu in the colloids [22]. The Diffusive
61
Gradients in Thin films (DGT) technique [23] can be used for dissociation kinetics
62
measurements as well. Performing DGT measurements with different gel thicknesses
63
allows detection of dissociation rate constants under the experimental conditions [24]. In
64
the Competitive Ligand Exchange Method (CLEM), a competing ligand for metal
65
binding is added to the metal-DOM solution and the rate of metal concentration decrease
66
in the solution is monitored [21]. Often Chelex-100 cation exchange resin is used as zero
67
sink for metals. The type and quantity of competing ligand used for CLEM has an
68
influence on the dissociation kinetics [25].
69
5
70
There are a few data on dissociation kinetics of natural Cu-DOM complexes in soils [26-
71
28] and soil amendments such as manure [29] and wastewater [30]. The conclusions
72
about the lability of these complexes differ widely and are difficult to compare because of
73
methodological differences. In addition, the Cu dissociation kinetics of complexes with
74
humic acids extracted from soil is affected by Cu loading, contact time, pH and solution
75
composition [31]. These factors were uncontrolled in the above-mentioned studies with
76
natural Cu-DOM complexes in soil and soil amendments.
77
78
The objectives of this study are to characterize the dissociation kinetics of a suite of Cu-
79
DOM complexes present in soil and soil amendments and to experimentally assess
80
factors explaining differences in lability of Cu-DOM complexes. These factors include
81
pCu (=-log (Cu2+)), source of DOM and reaction time between Cu and DOM (‘ageing’).
82
The pH and Ca concentration were identical for all solutions. The kinetics was detected
83
by CLEM and DGT. The usual time scales of CLEM and DGT tests (0.1–4 hours) were
84
extended to study also very slowly dissociating Cu-DOM complexes. Most Cu-DOM
85
complexes were extracted from soil or soil amendments. It is assumed that the majority
86
of Cu is associated with DOM and not with inorganic colloids. In soil solutions Cu is
87
mainly organically bound [32] and Cu and DOC concentrations are highly correlated
88
[33,34].
89
90
2. Theory
6
91
Dissolved organic matter has a heterogeneous composition and can be assumed to contain
92
n Cu binding sites Li. The formation and dissociation reaction of Cu and binding site Li
93
of DOM in the soil solution is given by:
kf,i
Cu +
Li
CuLi
(1)
kd,i
94
where kd,i and kf,i are the dissociation and formation rate constants of the complex CuLi.
95
Charges have been omitted for simplicity. The dissociation rate constant is equal to:
k d,i 
k f, i
Ki
(2)
96
with Ki the equilibrium constant of reaction (1). This complexation constant (Ki) is a
97
measure of the affinity of the ligand for Cu. The formation rate constant kf is determined
98
by Cu metal properties rather than by characteristics of the Cu-ligand complex. The
99
Eigen-Wilkins mechanism states that the rate-limiting step of the formation reaction is
100
the exchange of an inner-sphere water molecule for the outer-sphere ligand, which
101
follows after the relatively fast outer-sphere complex formation with the ligand [35].
102
When complexes with the same metal ion are compared, the dissociation rate of a
103
complex is usually inversely proportional to the stability of the complex (equation 2),
104
since the formation rate constant only depends on the water exchange rate constant (k-w),
105
which is a constant for a given metal ion, and the outersphere complexation constant
106
(KOS), which only depends on the charge of the metal and ligand.
107
108
It is common to describe dissociation kinetics of heterogeneous complexes as a parallel
109
set of first order reactions in which n different components are distinguished, each with a
110
dissociation rate constant kd,i and an initial concentration [CuLi]°. In other words, the
7
111
decrease of the CuL concentration is commonly described as a sum of exponential decay
112
terms:
CuL t    CuL i O  exp - k d,i  t 
n
(3)
i 1
113
Mostly only 2 components (n = 2) are distinguished. This distinction is arbitrary,
114
although statistical rules can be defined to identify the number of significant components,
115
and has the drawback that the parameters have a strong covariance. The dissociation rate
116
constants of the defined fractions differ for every fitted dissociation kinetics which
117
complicates comparison between different complexes [21,35,36]. The discrete
118
component kinetic model can be replaced by the assumption that the parameter kd. has a
119
continuous distribution as proposed by Rate et al. [37]. These authors studied the
120
dissociation kinetics of Cu-HA complexes by CLEM and applied different methods for
121
kinetic data analysis. Application of discrete multicomponent regression, discrete kinetic
122
spectra and Laplace spectra resulted all in a simple and usually unimodal distribution of
123
log kd. Kinetic analysis by the log-normal distribution of kd was preferred because of the
124
very good agreement with the observations and the restricted number of parameters. This
125
approach was also successfully applied on Ni-FA dissociation data of Cabaniss [38].
126
127
The description of a continuous distribution of parameter kd is given by:
CuL t   CuL O  f log k d   exp - k d  t  d log k d 
(4)
128
with f (log kd) the normal distribution of log kd that has a mean µ and standard deviation
129
σ:
8
f log k d  
 log k d  μ 2
exp  
2σ 2
σ 2π

1




(5)
130
131
A continuous distribution of kd is not often assumed in literature, in contrast to the
132
approach for the complexation constant K for humic substances and metals. For instance,
133
the complexation constants log K in the NIC(C)A-Donnan model have a Sips distribution
134
function [39]. If K has a continuous distribution, it follows from Equation 2 that kd has a
135
similar distribution.
136
137
138
3. Materials and methods
3.1. Cu-DOM complexes
139
This study compared dissociation kinetics of Cu-DOM complexes from soil, wastewater,
140
pig manure and sewage sludge and included a commercially humic acid and 3 synthetic
141
ligands as references. All solutions with Cu-DOM complexes were equilibrated at
142
constant pH 7.0 and dissolved Ca concentration. This was necessary as the native Cu-
143
DOM samples largely vary in these properties which may otherwise explain the
144
differences in dissociation kinetics. Here, we wanted to assess the effect of the source and
145
properties of the DOM, ageing time and free Cu2+ activity on the dissociation kinetics.
146
The effect of the source and properties of the DOM was studied on several DOM samples
147
at same pCu, the effect of Cu2+ activity on the same sample at different pCu (11.3, 10.3
148
and 9.9) and the effect of ageing time on the same DOM sample that was equilibrated
149
with Cu for various times.
150
9
151
The Cu loading or unloading (in case of samples with high Cu concentrations) of the
152
DOM was performed by equilibrating the DOM solutions with a Chelex-100 cation
153
exchange resin with a set Cu and Ca loading. Resins with three different loadings were
154
prepared by shaking 25 g of Chelex-100 resin (Na-form, Bio-rad) with 100 mL solution
155
containing CuCl2 and Ca(NO3)2. The capacity of the resin, estimated from preliminary
156
experiments, was 0.8 meq g-1. The Cu addition was selected to obtain a resin coverage of
157
20%, 47% and 70%, and Ca addition was selected to saturate the remaining binding sites
158
with a little excess (1.5 mM in solution) to exchange almost all Na. The resin solutions
159
were shaken end-over-end for 7 days to reach equilibrium [40]. The pH was checked
160
daily and adjusted if necessary to 7.0 with 2 M NaOH. After 7 days the resins were
161
washed extensively with ultrapure water and filtered.
162
163
The test solutions were equilibrated with the Cu/Ca resin in a solid/liquid ratio of 1/100
164
kg wet weight L-1. The Ca concentrations in the original solutions were first analyzed by
165
ICP-OES (Perkin Elmer, Optima 3300 DV) and were adjusted with 1 M Ca(NO3)2 to
166
obtain a final Ca concentration of 1.5 to 2 mM. The test solutions were shaken end-over-
167
end with the Cu/Ca resin for 7 days. Solution pH was checked and adjusted if necessary
168
to pH 7.0 (± 0.1) with little amounts of 0.1 M HCl or 0.1 M NaOH. After 7 days the
169
solutions were filtered (0.45 µm). Before and after shaking with the resin, the solutions
170
were analyzed for cation concentrations (ICP-OES, Perkin Elmer, Optima 3300 DV),
171
DOC concentrations (AnalytikJena, Multi N/C 2100S) and UV absorbance at 254 nm
172
(A254, dimensionless) (Perkin Elmer, Lambda 20, 1 cm path length, quartz cells). The
10
173
specific UV absorbance at 254 nm (SUVA, L g-1 cm-1), a measure for the aromaticity of
174
DOM, was calculated by:
A 254
SUVA 
b  DOC
(6)
175
where b is the path length of the spectrophotometer (1 cm) and [DOC] the DOC
176
concentration in solution (g L-1) [41].
177
178
The free Cu2+ activity in equilibrium with the resin was verified by shaking the resin with
179
a synthetic ligand with known complexation constants (NTA or EDTA) using the same
180
procedure as with the test solutions. Addition of the natural DOM or synthetic ligand was
181
assumed not to influence Cu2+ activity because the quantity of Cu on the resin is much
182
larger than that in solution. The concentration of Cu on the resin maximally changed
183
0.36% before and after equilibration with the ligands, supporting our assumption that the
184
buffering pool was large enough to maintain constant Cu2+ activity in solution. The added
185
NO3- and synthetic ligand concentrations and the measured Cu, Ca and Na concentrations
186
in the equilibrium solution were entered in GEOCHEM-PC [42] to calculate free Cu2+
187
activity and concentration. The relevant infinite dilution complexation constants (log K)
188
are CuNTA- = 14.4; CuEDTA2- = 20.0; CaNTA- = 7.7; CaEDTA2- = 11.9. The pCu in
189
equilibrium with the Cu/Ca resins was 11.3 (resin with 20% Cu), 10.3 (47% Cu resin)
190
and 9.9 (70% Cu resin).
191
192
Three solutions were equilibrated at pCu 11.3: PM1, SE1 and WW1. Pig manure was
193
centrifuged for 30 minutes (2500 g), filtered (0.45 µm), acidified to pH 7.0 with 2 M
194
HCl, filtered again (0.45 µm) and 20 times diluted (= PM1). Topsoil of a historically Cu
11
195
contaminated fallow soil (Waltham Abbey, UK) was sampled, sieved (4 mm) and stored
196
at 4 °C. Soil solution was extracted (S/L = 1/5 kg L-1) with 1 mM CaCl2. After shaking
197
end-over-end for 24 hours, the suspension was centrifuged for 30 minutes (2500 g),
198
filtered (0.45 µm) and 8/5 times diluted (= SE1). Influent of a wastewater treatment plant
199
in Leuven (Belgium) was filtered (0.45 µm), acidified to pH 7.0 with 2 M HCl, filtered
200
again (0.45 µm) and 10/7 times diluted (= WW1). Seven solutions were equilibrated at
201
pCu 10.3: NTA, EDTA, HA, SS, PM2, SE2 and WW2. Two solutions with synthetic
202
ligands were made: 10-5 M NTA with 3 µM CuCl2 (= NTA) and 10-5 M EDTA with 3
203
µM CuCl2 (= EDTA). A solution with 90 mg L-1 Aldrich humic acid (in Na-form, Sigma-
204
Aldrich) and 0.9 µM CuCl2 was prepared 60 days before equilibration with the Cu/Ca
205
resin (= HA). Sewage sludge of a wastewater treatment plant in Leuven (Belgium) was
206
centrifuged for 30 minutes (2500 g), filtered (0.45 µm), acidified to pH 7.0 with 2 M
207
HCl, filtered again (0.45 µm) and 4/3 times diluted (= SS). Aside from the different pCu
208
at which the solution was equilibrated, PM2 had an identical preparation as PM1. Also
209
for WW2, the preparation was identical as for WW1 except for a difference in dilution
210
factor (4/3 instead of 10/7). SE2 was prepared slightly different from SE1: the S/L ratio
211
for the extraction was 1/4 kg L-1 instead of 1/5 kg L-1 and the dilution factor was 5/3
212
instead of 8/5. Four solutions were equilibrated at pCu 9.9: PM3, SE3, WW3 and CI. The
213
PM3, SE3 and WW3 had the same preparation as PM2, SE2 and WW2 except for the
214
dilution factors: 50/3 instead of 20 for PM3, 1 instead of 5/3 for SE3, and 11/7 instead of
215
4/3 for WW3. The fourth solution consisted of 2 mM citrate (= CI).
216
12
217
Dissociation kinetics experiments (DGT and CLEM) were started within one hour after
218
equilibration with the Cu/Ca resin. CLEM experiments with solutions PM3 and SE3 were
219
performed for a second time 290 days after the resin equilibration (PM3a and SE3a) to
220
study the effect of ageing on the dissociation kinetics.
221
222
3.2. CLEM tests
223
The CLEM experiments were performed in duplicate based on the principles described
224
before [21]. Chelex-100 cation exchange resin was used as competitive ligand for Cu
225
binding. It was first converted to the Ca-form by shaking Chelex-100 in the Na-form
226
(Bio-rad) for 24 hours in a large volume (excess) of 1 M CaCl2, followed by washing
227
with ultrapure water and filtering. In a glass beaker of 100 mL, 2.5 g of chelex in the Ca-
228
form was stirred in 50 mL of the test solution. The maximum Cu amount in the solution
229
(in PM3, see 4.1) is only 0.1% of the maximum Chelex capacity. The beaker was covered
230
to prevent evaporation. Five mL of the stirred solution (solution + Chelex) was sampled
231
and filtered (0.45 µm) at 5, 30, 60, 120, 240, 480, 1440, 2880 and/or 4320 to 8700
232
minutes after addition of the Chelex to the solution. The Cu concentration in the
233
subsample was measured by ICP-OES after acidification to pH 1 with 5 M HNO3. The
234
dissociation kinetics of the Cu ligands (CuL) was described by assuming a normal
235
distribution of log kd (equations 4 and 5). Equation (4) was numerically integrated and
236
was fitted by minimizing the sum of squared differences of modeled Cu concentrations
237
with each measured duplicate set of dissociation curves. Fitting was made by the Solver
238
function in Microsoft Excel® [43].
239
13
240
3.3. DGT tests
241
The DGT tests were performed in triplicate on all solutions except the aged ones (PM3a
242
and SE3a) and CI. The method and used materials are different from those normally
243
applied in DGT experiments due to a limited volume of available solution. Results are,
244
therefore, difficult to compare with other studies but internal comparisons between values
245
obtained by this method are possible. The DGT devices were custom made in 6 mL
246
scintillation vials (13 mm diameter, 55 mm height). An agar diffusion layer of 3 mm
247
thickness was poured on top of a 2 mm agar layer containing Chelex-100 in the Na-form
248
(Bio-rad) on the bottom of the vial. Five mL of the test solution was added to the DGT
249
vial. The closed vial was shaken horizontally at 20 °C during 16 hours. After this
250
deployment time, the solution was decanted off. The remaining Cu concentration was
251
measured by ICP-OES after acidification to pH 1 by 5 M HNO3. The labile Cu
252
concentration in solution [Cu]DGT can be calculated using the mass of copper mCu bound
253
by the chelex which includes also the Cu of the complexes that dissociated within the
254
diffusion zone [23]:
CuDGT  mCu  g
D  At
(7)
255
where ∆g is the thickness of the diffusive layer (3 × 10-3 m), A is the exposed surface area
256
(1.3 × 10-4 m2), t is the deployment time (5.76 × 104 s) and D is the diffusion coefficient
257
of Cu species through the agar. Several studies found smaller diffusion coefficients for
258
Cu complexed by organic ligands than for free Cu [44-46]. Warnken et al. [47] concluded
259
that diffusion coefficients of Cu complexed by natural organic matter from freshwater
260
may be larger than the values commonly assumed, which are based on laboratory
261
experiments with extracted fulvic acids. Since diffusion coefficients for the different
14
262
complexes in the test solutions are unknown, a constant diffusion coefficient D is
263
assumed, equal to the diffusion coefficient of Cu through agarose gel (6.20 × 10-10 m2 s-1
264
[46]). It is hypothesized that variations in DGT results are mainly determined by kinetic
265
processes and less to diffusion limitations (see Discussion 5.3). In contrast to other
266
studies where mCu is measured by Chelex desorption, mCu is calculated in this method by
267
the difference in total Cu concentration in the solution before and after the deployment
268
time, multiplied by the volume of the solution (5 mL). An excess volume of solution is
269
normally used in DGT studies. It can, therefore, be assumed that solution metal
270
concentrations do not change during deployment time. This is not the case in this study:
271
Cu concentration decreases varied between 6% and 32%, with an average of 16%. As the
272
Cu amount was determined by difference, copper that entered the agar layer but was not
273
sorbed by the Chelex is also taken into account. Preliminary experiments with several
274
Cu-DOM solutions were set up to quantify this residual Cu in the diffusive gel by
275
washing the vial with ultrapure water for 24 hours after the deployment time. The tests
276
showed no significant Cu diffusing out of the agar, suggesting that the amount of this
277
residual Cu is small. The %CuDGT is the percentage [Cu]DGT of the initial dissolved Cu
278
concentration in the test solution. The %Cun-DGT is the percentage ‘non-labile’ Cu
279
concentration in solution (= 100% - %CuDGT).
280
281
282
4. Results
4.1. Cu-DOM solutions
283
The properties of the original Cu-DOM solutions changed upon equilibration with the
284
Cu/Ca resin (Table 1). The DOC concentrations decreased in all cases, likely due to
15
285
sorption on the resin, coagulation of DOM as a result of higher Ca concentration in the
286
equilibration solution than in the original sample or due to mineralization during the 7-
287
days equilibration time with the resin. The SUVA of the DOM decreased slightly in most
288
cases. The [Cu]/[DOC] ratio increased after equilibration with the resin in most cases
289
illustrating that the Cu2+ activity in the initial solutions was smaller than the Cu2+ activity
290
imposed by the resin. The reverse was true for the extracts of the Cu contaminated soil
291
(SE1-SE3) that clearly had a higher Cu2+ activity than those in the resin systems. The
292
[Cu]/[DOC] ratio of PM, SE and WW upon equilibration with the resin logically
293
increased as Cu2+ activity increased.
294
295
4.2. CLEM and DGT experiments
296
The CLEM and DGT experiments were first performed with a 2.67 µM CuCl2 solution
297
without organic ligands as a blank. The Cu concentration in solution in the CLEM
298
experiment was decreased to 0.03 µM (1%) after 5 minutes. The %CuDGT for the Cu2+
299
salt was 89%. However, the copper present in this solution should be fully labile
300
(%CuDGT = 100%). The difference can be caused by small deviations of the thickness of
301
the diffusive layer ∆g and/or the diffusion coefficient D from the assumed values, or by
302
the fact that the Cu concentration in the solution was not constant during the deployment
303
time (see 3.3). The ratio of ∆g/D was adjusted by 12% to obtain 100% for %CuDGT of the
304
Cu2+ salt solution. These adapted values were then used for the DGT calculations of all
305
Cu-DOM solutions.
306
16
307
After an initial fast reduction, the Cu concentration of the Cu-DOM solutions in the
308
CLEM experiments decreased more slowly in time (Figure 1) with detectable Cu in some
309
solutions after 48 hours. The fitted µ of the normal distribution of log kd varied between
310
-3.7 (PM1) and 0.1 (WW3), corresponding to kd values of 2 × 10-4 s-1 and 1.2 s-1 (Table
311
1). No values are given for NTA because the initial Cu concentration decrease in the Cu-
312
NTA solution was as fast as that in the Cu2+ salt solution. The standard deviation σ of log
313
kd was clearly lower for the three synthetic ligands (0.1–0.2) than for the natural DOM (σ:
314
1.1–2.2) which logically follows from the larger heterogeneity in binding sites of the
315
natural DOM. The combination of µ and σ allows estimating the fractions of Cu-DOM
316
that have a kd lower than a given threshold. For example, the percentage with kd < 10-5 s-1
317
is 17.2% for PM1 and 5.1% for PM3 (Figure 2). Alternatively, the CLEM data can be
318
summarized as the fraction remaining in solution after a fixed time, e.g. 8 hours. The
319
percentage of the Cu concentration in solution after 8 hours (%Cu8h) varied from <1%
320
(CI) to 25% (HA) (Table 1). The non-labile fractions determined in the DGT tests (%Cun-
321
DGT)
322
the %Cun-DGT because of different processes involved, i.e. mainly dissociation (CLEM) or
323
a combination of diffusion and dissociation (DGT), and the different time scale of the
324
methods (see section 5.3). Both fractions correlate (r = 0.56, P < 0.05), however, even
325
more so when only the natural DOM samples are considered (r = 0.74).
varied between 0 and 82% (Table 1). Absolute values of %Cu8h differ from that of
326
327
4.3. Influence of pCu and equilibration time on the Cu-DOM dissociation kinetics
328
The influence of pCu on the Cu-DOM dissociation kinetics was studied in solutions of
329
PM, SE and WW equilibrated at different pCu. Recall that the initial solutions of the PM,
17
330
SE and WW equilibrated with the different Cu/Ca resins were not exactly the same, but
331
have high similarities given the identical source and similar preparation method. Higher
332
Cu loadings of DOM generally increased the Cu-DOM dissociation kinetics. The %Cu8h
333
in the PM, SE and WW solutions decreased with increasing free Cu2+ activity and
334
increasing Cu loading ([Cu]/[DOC] ratio) (Table 1). Similarly, the mean dissociation rate
335
constant (kd) increased as Cu2+ activity increased (Figure 1(a), Figure 2(a) and Table 1).
336
There was no clear effect of pCu on the standard deviation σ of log kd. In general, the
337
%Cun-DGT decreased with increasing Cu2+ activity, but the %Cun-DGT was smaller at pCu
338
11.3 than at pCu 10.3 for the SE and WW samples, for unknown reasons.
339
340
Increasing ageing time generally decreased the Cu dissociation rate. An additional 290
341
days Cu-DOM contact time after uploading the DOM with Cu in the resin equilibration
342
significantly decreased mean log kd from -2.9 to -3.4 for the pig manure sample (PM)
343
(Figure 1(b), Figure 2(b) and Table 1). At the same time the σ of the normal distribution
344
of log kd increased from 1.3 to 1.5 which contributes to a larger fraction of non-labile
345
complexes in the aged samples. The combined effect of the change in these two
346
parameters increased the estimated fraction with kd < 10-5 s-1 from 5.1% to 14.4% after
347
290 days extra contact time (Figure 2(b)). Likewise, %Cu8h increased from 9% to 21%
348
with increased ageing. In contrast, no significant ageing effects were found for Cu-DOM
349
complexes present in the soil extract (SE), likely because the resin equilibration step
350
desorbed indigenous Cu and the native Cu-DOM complexes in the soil had already been
351
aged (see Discussion 5.1).
352
18
353
4.4. Cu lability in different DOM samples
354
Differences in Cu lability among DOM samples should be compared at equal pCu and
355
ageing time due to the effects noted above. The pH (7.0) and Ca concentrations (circa 1.5
356
mM) were similar for all solutions. All natural DOM samples were analyzed at pCu 10.3
357
(Table 1). The value of µ (average of log kd) increased and the %Cu8h values decreased in
358
the order: HA > PM2 > SE2 > SS > WW2. The %Cun-DGT values followed the same order
359
as the %Cu8h. The CLEM as well as the DGT experiment showed that the Cu-HA
360
solutions had the slowest dissociating Cu complexes. Both methods suggest that Cu from
361
pig manure has a higher leaching risk in soils compared Cu from waste water and sewage
362
sludge. For the PM, SE and WW samples, measurements were also made at pCu 11.3 and
363
9.9. Similarly as for the solutions equilibrated at pCu 10.3, the µ values increased and the
364
%Cu8h and %Cun-DGT values decreased in the order PM > SE > WW, except that %Cu8h
365
was slightly (not significant) larger for SE than for PM at pCu 9.9. A strong significant
366
positive correlation was found between %Cu8h and the SUVA of all samples at pCu 10.3
367
(r = 0.96; P < 0.05; Figure 3). There was also a significant positive correlation between
368
%Cun-DGT and the SUVA of the DOM (r=0.78). The dissociation kinetics were unrelated
369
to the [Cu]/[DOC] ratio among different Cu-DOM at same Cu2+ activity (P > 0.05 for any
370
kinetic parameter). This contrasts the analysis on the same Cu-DOM tested at different
371
pCu values (section 4.3), i.e. the conclusion of slower dissociation kinetics at lower Cu
372
loadings is only valid when comparing the same DOM.
373
374
5. Discussion
19
375
376
5.1. Influence of pCu and Cu-DOM contact time on the Cu-DOM dissociation
kinetics
377
Relatively faster dissociation kinetics were observed at higher Cu2+ activity (lower pCu)
378
for the same DOM. The same trend was already observed for isolated fulvic acids [21,25]
379
and humic acids [31,48]. The highest affinity sites are first occupied by Cu at low Cu2+
380
activity followed by low affinity sites at larger Cu2+ activity. This means that the
381
dissociation rate is slower for the high affinity sites (i.e. large K) than for low affinity
382
sites (small K) within the same DOM, as is theoretically expected (Equation 2). For the
383
SE solutions, there was a small increase of σ at increasing Cu2+ activity and [Cu]/[DOC]
384
ratio. This trend suggests an increase in kinetic diversity of the Cu-DOM ligands when
385
more binding sites are occupied, but this was not observed for PM and WW.
386
387
Slower and more heterogeneous dissociation kinetics for PM was observed when the Cu-
388
DOM complexes had an additional 290 days contact time after the resin equilibration.
389
Rate et al. [31] observed similar trends for Cu and a pedogenic humic acid. For example,
390
increasing the pre-equilibration time from 24 hours to 168 hours of 7.87 µM Cu and 25
391
mg L-1 peat extract at pH 6.5, decreased µ (log kd) from -1.8 to -2.2 and increased σ (log
392
kd) from 0.9 to 1.1. They attributed this effect to slow formation of stable complexes by
393
complexation-induced conformational changes. No significant effects of ageing were
394
found here for Cu-DOM complexes in the soil extract SE. The SE sample was partly
395
unloaded during the 7 days resin contact, i.e. the [Cu]/[DOC] ratio decreased (Table 1) in
396
contrast to the PM sample that was uploaded with Cu by the resin. In other words, the Cu
397
of SE that was still associated with the DOM after the resin equilibration had already
20
398
been aged for a long time in the native environment and an additional ageing of 290 days
399
has, therefore, little effects.
400
401
5.2. Variation in dissociation kinetics among samples
402
The denominator in Equation 2 shows that the dissociation rate constant is inversely
403
proportional to the stability of the Cu-ligand complex, i.e. stronger bonds imply
404
kinetically more inert bonds. This relationship between complex stability and dissociation
405
kinetics was also stated by Rate et al. [31] on the basis of CLEM experiments with Cu-
406
HA complexes by varying pH, ionic strength and [Cu]/[HA] ratio. They formulated two
407
possible processes determining dissociation kinetics: intraparticle diffusion in the coiled
408
HA molecule or thermodynamic stability of the complex. If dissociation kinetics is
409
controlled by intraparticle diffusion, actions causing some unfolding of the HA will
410
increase the dissociation rate. Unfolding HA can be achieved by increasing the pH, or
411
decreasing the ionic strength or metal loading of HA. The authors observed, however,
412
decreasing dissociation rates by these actions. They concluded that the dissociation rate is
413
controlled by the thermodynamic stability, because these three actions increase overall
414
complexation stability. Similar to the observations of Rate et al. [31], an increase in Cu-
415
DOM dissociation rate at higher Cu loadings was noticed in this study. In addition a
416
positive correlation is found between the SUVA of DOM and %Cu8h and %Cun-DGT of the
417
solutions equilibrated at pCu 10.3. It has been shown that Cu affinity at low Cu2+ activity
418
is linked to SUVA and DOM aromaticity [40,49]. The correlation suggests that the Cu
419
complexes with more aromatic DOM components hold Cu with slower dissociation rate
420
at the same free Cu2+ activity. The dissociation of Cu-HA complexes is clearly slower
21
421
than that of natural DOM. This difference can also be related to the SUVA and
422
differences in Cu-DOM stability constants. Humic and fulvic acids are assumed to have
423
larger Cu affinity compared to the natural, more heterogeneous DOM [50]. Beside humic
424
and fulvic acids, DOM contains less humified components, such as small organic acids,
425
carbohydrates and amino acids. The smaller Cu affinity of these components compared to
426
humic compounds results in faster overall Cu-DOM dissociation kinetics. The large
427
heterogeneity is reflected in the larger σ for SS, SE and WW (but not for PM) than for
428
HA.
429
430
5.3. Comparison of studies and methods for dissociation kinetics measurements
431
Similar to this study, Degryse et al. [51] calculated a single dissociation rate constant kd
432
of 10-3.7 s-1 for Cu-EDTA complexes and >10-2.0 s-1 for Cu-NTA by CLEM experiments
433
at pH 6. Data on Cu-DOM dissociation kinetics are mainly available for artificial
434
mixtures of Cu with isolated fulvic acids (FA) [21] or humic acids (HA) and rarely for
435
native Cu-DOM complexes. Comparison with results from other Cu-HA dissociation
436
kinetics studies is difficult because of methodological differences. Generally a large
437
range in lability of Cu-HA complexes is found, from more inert [35] to more labile
438
[31,37,52] than the Cu-HA complexes in our study. Sivry et al. [22] found higher lability
439
in a one-hour isotopic exchange experiments for Cu-HA complexes compared to Cu
440
added to natural organic-rich waters. This is in contrast to our study since the Cu
441
complexes with HA showed the slowest dissociation kinetics of all analyzed Cu-DOM
442
complexes. The limited data on Cu dissociation kinetics of pedogenic DOM broadly
443
correspond with our data, however methodological differences are a major obstacle for a
22
444
detailed comparison. A one-hour isotopic exchange experiment showed 100% lability for
445
four natural DOM samples (aquogenic and pedogenic) pre-equilibrated with Cu for two
446
days [26]. Del Castilho et al. [28] found 30%–70% labile Cu in water saturation extracts
447
of manure contaminated soil using CLEM with only 30 seconds residence time. In our
448
study, a similar range of 32% to 69% of total Cu had disappeared in the CLEM
449
experiments with the contaminated soil extracts (SE) at the first sampling time (5
450
minutes). At 24 hours, this fraction was increased to 88%–95%, which is in the range of
451
60%–96% labile Cu found by Ma et al. [27] for water extracts of 18 Cu contaminated
452
soils in a 24-hours isotopic exchange experiment.
453
454
Very little data are available on dissociation kinetics of Cu-DOM complexes present in
455
soil amendments such as wastewater, pig manure or sewage sludge. Del Castilho et al.
456
[29] observed that 50% (low molecular weight fractions) to 100% (high molecular
457
weight) of Cu in pig slurry did not dissociate after 4 hours CLEM contact time. This is in
458
contrast with the pig manure in our study (PM). After four hours in the CLEM
459
experiment, only 13% to 27% of total Cu was left. The smaller non-labile Cu fraction can
460
be explained by the presence of recently complexed Cu due to equilibration with the
461
Cu/Ca resin in this study. This can also be one of the reasons why Buzier et al. [30] found
462
larger %Cun-DGT values in wastewater (58% –71%) compared to WW in our study (0% –
463
16%). However, the difference can also be due to other methodological differences,
464
especially the thickness of the diffusive layer of the DGT device (0.8 mm versus 3 mm).
465
23
466
It is well known that dissociation rate data are affected by methodological differences.
467
Here, a significant correlation was found between the data of CLEM and DGT method,
468
i.e. between %Cu8h and %Cun-DGT although absolute values differ. The absolute values
469
can be converted under the assumption of a continuous distribution of the first order
470
dissociation rate constants, as is illustrated in Figure 4. This distribution can be used to
471
estimate kinetic parameters of different methods with a different time scale τ. First order
472
dissociation kinetics predicts that complexes in the solution with kd = τ-1 are 37% (= e-1)
473
undissociated at t = τ. This τ-1 is chosen as threshold kd; complexes with larger kd than
474
this threshold are assigned labile (mostly dissociated at t = τ) while complexes with kd <
475
τ-1 are assigned inert (largely undissociated at t = τ). The fraction of inert complexes in a
476
mixture (%Cuτ) can be calculated from the cumulative distribution function at a given
477
threshold kd (shaded area in Figure 4). For the CLEM experiment a time scale τ of 8
478
hours was chosen here, giving a threshold kd = τ-1 = 3 × 10-5 s-1 (or log kd = -4.5). The
479
%Cuτ for τ = 8 hours are indeed close to the percentages of undissociated complexes after
480
8 hours CLEM experiment %Cu8h (Figure 5). This is logical since the parameters for the
481
distribution of log kd (µ and σ) were fitted on the same CLEM data. The time scale of the
482
DGT method is determined by the thickness of the diffusion layer ∆g (= 3 mm) and the
483
diffusion coefficient D (= 6.2 × 10-6 m2 s-1) [23]:
τ
g 2
πD
(8)
484
A timescale of 77 minutes for the DGT method is calculated by Equation (8), giving a
485
threshold kd of 2 × 10-4 s-1 (or log kd = -3.7). The %Cuτ for τ = 77 minutes of the different
486
Cu-DOM solutions is generally lower than %CuDGT but the values do not deviate strongly
487
from the 1:1 line and follow the same trend (Figure 5). An exception is WW1, which has
24
488
an exceptionally low %CuDGT value (0%), for reasons unknown, and may have been a
489
analytical problem, since a larger %Cun-DGT would have been expected for WW1 than
490
WW2, given its slower kinetics (cf. lower µ in Table 1). This shows that kinetic
491
parameters of the DGT method can be estimated from the data of another dissociation
492
kinetics experiment (CLEM). Of course, no extrapolations beyond the kinetic window of
493
the method used should be made. For instance, the kinetic fingerprint derived from the
494
CLEM data may not give an accurate estimate of how complexes would react at time
495
scales < 100 s, which is for instance the time scale at which voltammetric measurements
496
occur [53].
497
498
Chakraborty [54] also applied both CLEM and DGT for Cu complexes in mine effluents.
499
The percentage of labile Cu according to CLEM (with kd = 1.79 × 10-1 s-1) was 53.5%,
500
larger than the percentage of labile Cu according to DGT (33.9%). The authors attributed
501
this difference to the smaller diffusion coefficient in DGT gel of large Cu-DOM
502
complexes compared to free Cu. It is hypothesized that this slower diffusion of Cu-DOM
503
complexes is limited in our experiments. Warnken et al. [47] have already suggested that
504
diffusion coefficients of natural Cu-DOM complexes are larger than values determined
505
by experiments with extracted fulvic acids. Two wastewater solutions had a %Cun-DGT
506
value of 0%, which demonstrates that diffusion limitations for this DOM did not occur.
507
The large variation in %Cun-DGT for PM (56–82%) and SE (26–56%) at different Cu
508
loadings would not have been observed if diffusion was a dominant process in DGT
509
experiments with these solutions. The significant positive correlation between %Cu8h and
510
%Cun-DGT, and the reasonably good prediction of the non-labile Cu fraction of DGT
25
511
based on CLEM experiments (Figure 5) also suggests that dissociation kinetics is the
512
most important process in DGT. However, probably some diffusion limitations occur
513
given the general underestimations of %Cun-DGT.
514
515
5.4. Environmental relevance of Cu-DOM dissociation kinetics
516
The question remains if the observed slow Cu-DOM dissociation kinetics contributes to
517
Cu leaching in soils. The Cu concentrations in the non-synthetic Cu-DOM solutions
518
decreased already to 1–25% after only 8 hours contact with the Chelex during the CLEM
519
experiment. Degryse et al. [20] calculated that, at relevant soil conditions, non-
520
equilibrium conditions can be observed if dissociation rate constants (kd) of Cu-DOM
521
complexes are smaller than about 10-5 s-1. Larger pore-water velocities, e.g. 30 cm day-1
522
instead of 0.3 cm day-1, can already result in kinetic dissociation constraints if kd is
523
smaller than about 10-3 s-1. Although µ of log kd was > -5 for all the solutions, all non-
524
synthetic solutions had fitted Cu-DOM fractions with log kd < -5 (e.g. 7.0% for PM2; see
525
Figure 2(a)). The percentage of complexed Cu with log kd < -5 is rather low for the
526
samples studied: at most 17.2% (PM1). Solutions at high pCu had the largest fractions
527
with log kd < -5, but their total Cu concentration in solution was smaller. The risk for
528
substantial Cu leaching as a result of non-equilibrium conditions in soils seems therefore
529
rather low: relatively non-labile Cu complexes have only small Cu concentrations and
530
soil solutions with larger Cu concentrations show relatively high lability. Marx &
531
Heuman [26] even found 100% lability for Cu saturated pedogenic DOM. Isotopic
532
exchange experiments for 24 hours showed 0.003–0.085 mg L-1 non-labile Cu
533
concentration in water extracts of 18 Cu contaminated soils, corresponding to 4–40% of
26
534
total Cu concentration in solution [27]. Similar to our study, the highest percentage non-
535
labile Cu (40%) was associated with the lowest total Cu concentration in the extract of
536
the historically contaminated soils (0.14 mg L-1). Trivalent metals are expected to exhibit
537
more non-equilibrium behavior. Schmitt et al. [55] monitored the Al(III), Fe(III), Pb(II)
538
and Zn(II) decrease in CLEM experiments with metal-DOM solutions and fitted a
539
dissociation rate constant of ~10-3 s-1 for the divalent metals and ~10-5 s-1 for the trivalent
540
metals. Of the trivalent metals, 19% to 55% was eluted one bed volume after addition to a
541
quartz column, indicating unretardated leaching of metal-DOM complexes, compared to
542
at maximum 0.02% for the divalent ions. Migration experiments with the trivalent 241Am
543
also showed leaching of DOM-complexed Am through sand columns [56]. Clearly,
544
dedicated studies, e.g. using variable pore water velocity, are required to validate
545
predicted transport of non-labile Cu complexes in porous media such as soils.
546
547
6. Conclusions
548
The dissociation kinetics of Cu-DOM complexes showed fast and first order dissociation
549
kinetics for Cu complexes with NTA and citrate while natural Cu-DOM complexes
550
exhibited a more complex and, generally, slower dissociation that was fitted by a model
551
assuming a frequency distribution of first order rate constants. Equilibration of the Cu
552
complexes at lower free Cu2+ activity, larger Cu-DOM contact time and increasing DOM
553
aromaticity were associated with slower dissociation kinetics. A fraction ranging between
554
<1–17% of dissolved Cu has a dissociation rate constant < 10-5 s-1 , a threshold value
555
below which the local equilibrium assumptions at most relevant soil conditions is invalid
556
[20].
27
557
558
Acknowledgements
559
This research was funded by the Onderzoeksfonds K.U.Leuven under the project number
560
GOA/2006/07-TBA, and was supported by the fund for Scientific Research-Flanders
561
(F.W.O.) through a doctoral fellowship awarded to F. Amery and a post-doctoral
562
fellowship awarded to F. Degryse.
563
564
References
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
[1]
L. L'Herroux, S. Le Roux, P. Appriou, J. Martinez, Environ. Pollut. 97 (1997)
119-130.
[2]
B.K. Richards, T.S. Steenhuis, J.H. Peverly, M.B. McBride, Environ. Pollut. 99
(1998) 365-377.
[3]
V.J. Camobreco, B.K. Richards, T.S. Steenhuis, J.H. Peverly, M.B. McBride, Soil
Sci. 161 (1996) 740-750.
[4]
E.J.M. Temminghoff, S.E.A.T.M. Van Der Zee, F.A.M. De Haan, Environ. Sci.
Technol. 31 (1997) 1109-1115.
[5]
E.J.M. Temminghoff, S.E.A.T.M. Van Der Zee, F.A.M. De Haan, Eur. J. Soil Sci.
49 (1998) 617-628.
[6]
E.D. Burton, I.R. Phillips, D.W. Hawker, D.T. Lamb, Aust. J. Soil Res. 43 (2005)
503-513.
[7]
L.X. Zhou, J.W.C. Wong, J. Environ. Qual. 30 (2001) 878-883.
[8]
P. Baveye, M.B. McBride, D. Bouldin, T.D. Hinesly, M.S.A. Dahdoh, M.F.
Abdel-sabour, Sci. Total Environ. 227 (1999) 13-28.
[9]
W.R. Berti, L.W. Jacobs, J. Environ. Qual. 27 (1998) 1280-1286.
[10] S.P. McGrath, P.W. Lane, Environ. Pollut. 60 (1989) 235-256.
[11] A.C. Chang, J.E. Warneke, A.L. Page, L.J. Lund, J. Environ. Qual. 13 (1984) 8791.
[12] C. Keller, S.P. McGrath, S.J. Dunham, J. Environ. Qual. 31 (2002) 1550-1560.
[13] R.C. Sidle, L.T. Kardos, J. Environ. Qual. 6 (1977) 431-437.
[14] D. Gavalda, J.D. Scheiner, J.C. Revel, G. Merlina, M. Kaemmerer, E. Pinelli, M.
Guiresse, Sci. Total Environ. 343 (2005) 97-109.
[15] R.G. McLaren, L.M. Clucas, M.D. Taylor, T. Hendry, Aust. J. Soil Res. 42 (2004)
459-471.
[16] R.H. Dowdy, J.J. Latterell, T.D. Hinesly, R.B. Grossman, D.L. Sullivan, J.
Environ. Qual. 20 (1991) 119-123.
[17] J.J. Sloan, R.H. Dowdy, M.S. Dolan, J. Environ. Qual. 27 (1998) 1312-1317.
[18] T.W. Speir, A.P. Van Schaik, H.J. Percival, M.E. Close, L.P. Pang, Water Air
Soil Pollut. 150 (2003) 319-358.
28
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
[19] Y.M. Liang, R.B. Corey, J. Environ. Qual. 22 (1993) 1-8.
[20] F. Degryse, E. Smolders, D.R. Parker, Eur. J. Soil Sci. 60 (2009) 590-612.
[21] A.L.R. Sekaly, R. Mandal, N.M. Hassan, J. Murimboh, C.L. Chakrabarti, M.H.
Back, D.C. Gregoire, W.H. Schroeder, Anal. Chim. Acta 402 (1999) 211-221.
[22] Y. Sivry, J. Riotte, B. Dupré, J. Geochem. Explor. 88 (2006) 144-147.
[23] H. Zhang, W. Davison, Anal. Chem. 67 (1995) 3391-3400.
[24] K.W. Warnken, W. Davison, H. Zhang, J. Galceran, J. Puy, Environ. Sci.
Technol. 41 (2007) 3179-3185.
[25] T. Yapici, I.I. Fasfous, J.J. Zhao, C.L. Chakrabarti, Anal. Chim. Acta 636 (2009)
6-12.
[26] G. Marx, K.G. Heumann, Fresenius Journal of Analytical Chemistry 364 (1999)
489-494.
[27] Y.B. Ma, E. Lombi, A.L. Nolan, M.J. McLaughlin, Eur. J. Soil Sci. 57 (2006)
147-153.
[28] P. del Castilho, W.J. Chardon, W. Salomons, J. Environ. Qual. 22 (1993) 689697.
[29] P. del Castilho, J.W. Dalenberg, K. Brunt, A.P. Bruins, Int. J. Environ. Anal.
Chem. 50 (1993) 91-107.
[30] R. Buzier, M.H. Tusseau-Vuillemin, J.M. Mouchel, Sci. Total Environ. 358
(2006) 277-285.
[31] A.W. Rate, R.G. McLaren, R.S. Swift, Environ. Sci. Technol. 27 (1993) 14081414.
[32] D. Berggren, Water Air Soil Pollut. 62 (1992) 111-123.
[33] K. Kalbitz, R. Wennrich, Sci. Total Environ. 209 (1998) 27-39.
[34] P.F.A.M. Römkens, W. Salomons, Soil Sci. 163 (1998) 859-871.
[35] I. Gaabass, J.D. Murimboh, N.M. Hassan, Water Air Soil Pollut. 202 (2009) 131140.
[36] A.L.R. Sekaly, J. Murimboh, N.M. Hassan, R. Mandal, M.E. Ben Younes, C.L.
Chakrabarti, M.H. Back, D.C. Gregoire, Environ. Sci. Technol. 37 (2003) 68-74.
[37] A.W. Rate, R.G. McLaren, R.S. Swift, Environ. Sci. Technol. 26 (1992) 24772483.
[38] S.E. Cabaniss, Environ. Sci. Technol. 24 (1990) 583-588.
[39] D.G. Kinniburgh, W.H. van Riemsdijk, L.K. Koopal, M. Borkovec, M.F.
Benedetti, M.J. Avena, Colloids and Surfaces A: Physicochemical and Engineering
Aspects 151 (1999) 147-166.
[40] F. Amery, F. Degryse, W. Degeling, E. Smolders, R. Merckx, Environ. Sci.
Technol. 41 (2007) 2277-2281.
[41] J.L. Weishaar, G.R. Aiken, B.A. Bergamaschi, M.S. Fram, R. Fujii, K. Mopper,
Environ. Sci. Technol. 37 (2003) 4702-4708.
[42] D.R. Parker, W.A. Norvell, R.L. Chaney, in: R.H. Loepert (Ed.), Chemical
equilibrium and reaction models, Soil Sci. Soc. Amer., 1995, pp. 253-269.
[43] R. de Levie, J. Chem. Educ. 76 (1999) 1594.
[44] A.J. Downard, J. Panther, Y.C. Kim, K.J. Powell, Anal. Chim. Acta 499 (2003)
17-28.
[45] S. Scally, W. Davison, H. Zhang, Environ. Sci. Technol. 37 (2003) 1379-1384.
[46] H. Zhang, W. Davison, Anal. Chim. Acta 398 (1999) 329-340.
29
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
[47] K.W. Warnken, W. Davison, H. Zhang, Environ. Sci. Technol. 42 (2008) 69036909.
[48] J.G. Hering, F.M.M. Morel, Environ. Sci. Technol. 24 (1990) 242-252.
[49] M.L. Brooks, J.S. Meyer, D.M. McKnight, Hydrobiologia 579 (2007) 95-113.
[50] F. Amery, F. Degryse, K. Cheyns, I. De Troyer, J. Mertens, R. Merckx, E.
Smolders, Eur. J. Soil Sci. 59 (2008) 1087-1095.
[51] F. Degryse, E. Smolders, D.R. Parker, Plant Soil 289 (2006) 171-185.
[52] D.L. Olson, M.S. Shuman, Geochim. Cosmochim. Acta 49 (1985) 1371-1375.
[53] H.P. van Leeuwen, R.M. Town, J. Buffle, R.F.M.J. Cleven, W. Davison, J. Puy,
W.H. van Riemsdijk, L. Sigg, Environ. Sci. Technol. 39 (2005) 8545-8556.
[54] P. Chakraborty, J.J. Zhao, C.L. Chakrabarti, Anal. Chim. Acta 636 (2009) 70-76.
[55] D. Schmitt, F. Saravia, F.H. Frimmel, W. Schuessler, Water Res. 37 (2003) 35413550.
[56] R. Artinger, B. Kienzler, W. Schüssler, J.I. Kim, J. Contam. Hydrol. 35 (1998)
261-275.
658
Figure captions
659
Figure 1. Dissociation kinetics of Cu-DOM complexes as affected by Cu2+ activity (pCu)
660
and contact time between Cu and DOM. The dynamics of the fraction of the initial Cu
661
concentration in solution ([Cu]/[Cu]°) in the CLEM-experiments for (a) pig manure
662
derived Cu-DOM equilibrated for 7 days at different pCu values and (b) pig manure
663
derived Cu-DOM immediately after or 290 days after the 7-days equilibration at pCu =
664
9.9. Replicate values are given by symbols and full lines are the fits of Equation (4).
665
666
Figure 2. Frequency distribution of the logarithm of the dissociation rate constant kd (s-1)
667
obtained when fitting the CLEM data (see Figure 1) with Equation 4 for (a) pig manure
668
derived Cu-DOM equilibrated for 7 days at different pCu values and (b) pig manure
669
derived Cu-DOM immediately after or 290 days after the 7-days equilibration at pCu =
670
9.9. Values left from the pointed line are below the threshold value of 10-5 s-1. Log kd
671
values > -2 (shaded area) are not reliable because the half life for the decrease in Cu
30
672
concentration of Cu2+ salt is about 1 minute, and complexes with log kd > -2 can therefore
673
not be kinetically distinguished from the copper-aquo complex.
674
675
Figure 3. Significant positive correlation (r = 0.96) between the SUVA of DOM and the
676
percentage ‘non labile Cu’, i.e. Cu present in solution after 8 hours in the CLEM
677
experiment (%Cu8h) for the Cu-DOM solutions at pCu = 10.3.
678
679
Figure 4. The kinetic fingerprinting approach: CLEM data are used to fit µ and σ of the
680
normal distribution of log kd. With this information, the percentage of complexes that are
681
‘non-labile’ at a given time scale τ (%Cuτ), i.e. the complexes that have a dissociation
682
constant kd < τ-1, can be calculated This also allows an estimation of the fraction non-
683
labile metal according to a given method (CLEM at a given time, DGT, etc.), and a
684
comparison of methods. The time scale τ depends on the experimental conditions, e.g. in
685
the case of DGT on the thickness of the diffusion layer (g).
686
687
Figure 5. Conversion of CLEM data to DGT data. The percentages of inert Cu complexes
688
as measured by the CLEM and DGT method (%Cu8h and %Cun-DGT) versus the
689
percentages of inert Cu complexes as measured by the kinetic fingerprinting approach
690
(%Cuτ) , i.e. using the lognormal kd distribution that is characterized by  and  and
691
which was derived from the CLEM data (Figure 4) . The open diamond represents an
692
outlier for the DGT measurements as discussed in the text (WW1).
693
694
31
695
Table legends
696
Table 1. Dissociation characteristics of Cu-DOM complexes determined by the CLEM
697
and DGT assays: µ and σ: fitted mean and standard deviation of the lognormal
698
distribution of the dissociation rate constant (kd in s-1); %Cu8h: the fraction of Cu that is
699
undissociated after 8 h CLEM; %Cun-DGT: the fraction of Cu that is not DGT-labile. The
700
Cu and DOC concentrations and SUVA of DOM (specific UV absorbance at 254 nm) of
701
the test solutions are measured before (‘initial’) and after the Chelex equilibration (‘resin-
702
eq.’). Standard deviations in brackets.