Tracing Ni, Cu, and Zn kinetics and equilibrium ... › ~kbruland › ... · time and location....

PII S0016-7037(02)00907-9

Tracing Ni, Cu, and Zn kinetics and equilibrium partitioning between dissolved andparticulate phases in South San Francisco Bay, California, using stable isotopes

and high-resolution inductively coupled plasma mass spectrometry

ALISON K. GEE1,* and KENNETH W. BRULAND

2

1Departments of Earth Sciences and2Ocean Sciences, University of California at Santa Cruz, Santa Cruz, CA 95064, USA

(Received June 20, 2001;accepted in revised form April 2, 2002)

Abstract—Additions of the low occurrence stable isotopes61Ni, 65Cu, and68Zn were used as tracers todetermine the exchange kinetics of metals between dissolved and particulate forms in laboratory studies ofnatural water and suspended sediments from South San Francisco Bay, CA. Dissolved metal isotope additionswere made so that the isotope ratios (rather than total metal partitioning) were significantly altered from initialambient conditions. Dissolved metal concentrations were determined using an organic ligand sequentialextraction technique followed by analysis with high-resolution inductively coupled plasma mass spectrometry(HR-ICPMS). Exchangeable particulate concentrations were extracted using a 20% acetic acid leach followedby determination using HR-ICPMS. Equilibrium and kinetic sorption parameters were quantified according toa general model for trace metal partitioning assuming pseudo-first-order kinetics. Partition coefficients (KD)were tracked as a function of time over the fortnight experiment. For Ni, Cu, and Zn the initial ambient KD

values were found to be 103.65, 103.88, and 104.52 L kg�1, respectively. As a result of the dissolved metalisotope additions, the partition coefficients for all three metals dropped and then increased back to nearambient KD values after 14 days. Curve-fitting concentration versus time profiles from both dissolved andexchangeable particulate data sets allowed determination of kinetic rate constants. The best estimates offorward and backward kinetic rate constants for Ni, Cu, and Zn respectively are k'f � 0.03, 0.07, 0.12 d�1 andkb � 0.13, 0.12, 0.15 d�1. These results predict that sorption equilibria in South Bay should be reached onthe order of a month for Ni, on the order of 3 weeks for Cu, and on the order of 2 weeks for Zn. Together,the dissolved and exchangeable particulate data indicate more sluggish sorption kinetics for Ni than for Cu andZn and suggest that different chemical forms control the speciation of these three metals in South Bay. Orderof magnitude metal sorption exchange rates were estimated using these kinetic results. These calculationsindicate that sorption exchange between dissolved and suspended particulate phases can cause dynamicinternal cycling of these metals in South San Francisco Bay.Copyright © 2002 Elsevier Science Ltd

1. INTRODUCTION

The exchange of dissolved trace metals such as Ni, Cu, andZn between solution species and particle surfaces is an impor-tant topic for estuarine waters. To date, the majority of studiesconducted on suspended particles have focused on the observedfractionation between dissolved and particulate metal concen-trations (Kuwabara et al., 1989; Smith and Flegal, 1993; Benoitet al., 1994; Rivera-Duarte and Flegal, 1994; Sanudo-Wilhelmyet al., 1996). This fractionation, quantified as a distributioncoefficient, measures the distribution of different metals be-tween the particulate phase and dissolved species at a giventime and location. However, studies seldom address kinetics ofthe exchange process, such as the rate at which metal–particleassociations occur, information relevant for metal fate andtransport assessments. For example, if there is an input per-turbing either the dissolved or suspended particulate phase of acertain metal, at what rate will the system come to a newequilibrium? We would also like to know if there is dynamicexchange occurring between the dissolved and exchangeableparticulate metal pools, or if suspended particulate metal isessentially inert with respect to sorption processes on the time

scales of sediment resuspension events and particle residencetimes in the water column. Thus, constraining the kinetics ofmetal–particle adsorption/desorption reactions in estuarine wa-ters can substantially enhance current understanding of bothscavenging and desorption from contaminated particles as im-portant controls on dissolved metal concentrations.

To assess the exchange rates of metals between dissolvedand particulate forms, one needs methods to trace the process insuch a manner that it adequately simulates natural “ambient”conditions. Most previous studies in the literature of sorptionkinetics have involved laboratory experiments using pure phaseparticles with well characterized surface characteristics andinorganic solution species, the results of which report rapidadsorption rates on the order of minutes to hours (Atkinson etal., 1972; Anderson and Rubin, 1981; Hayes and Leckie, 1986).Due to the heterogeneous nature of most natural suspendedmaterial, however, rate constants determined for pure phaseparticles may have little relevance to the kinetics of metalsorption in natural environments. In addition, natural waterscan have complex solution speciation with much of the Ni, Cu,and Zn in estuaries existing bound to organic ligands (Donat etal., 1994; Sedlak et al., 1997; Kozelka and Bruland, 1998).From the work of Li et al. (1984), Nyffeler et al. (1984),Balistrieri and Murray (1986), and Jannasch et al. (1988), thereis experimental evidence that under natural conditions, dis-

* Author to whom correspondence should be addressed([email protected]).

Pergamon

Geochimica et Cosmochimica Acta, Vol. 66, No. 17, pp. 3063–3083, 2002Copyright © 2002 Elsevier Science LtdPrinted in the USA. All rights reserved

0016-7037/02 $22.00� .00

3063

solved/particulate metal partitioning can occur over longer timescales on the order of days to months. In these studies, metalradioisotopes were used to trace adsorption from dissolved toparticulate forms in laboratory experiments using natural fieldsamples.

Because natural aquatic systems are complex, one wouldideally employ a method that could document this processdirectly and in situ, with minimal disturbance from “ambient”conditions. This requires tracers that have little or no environ-mental consequences, ruling out radioactive isotopes com-monly used to study natural processes within the confines of thelaboratory. However, with the development of high-resolution,inductively coupled plasma mass spectrometry (HR-ICPMS),naturally occurring, low abundance, stable isotopes can be usedas tracers for such experiments. In this study, such isotopes ofthe metals Ni, Cu, and Zn were added to estuarine waters andsuspended sediments in laboratory bottle experiments. Theselow abundance isotope additions shift the isotopic ratios foreach metal away from natural abundance ratios but still allowthe sorption process to be studied close to “ambient” metalconcentrations. This study verifies the potential for low abun-dance stable isotopes to be used as powerful environmentaltracers, while assessing sorption equilibrium and kinetic pa-rameters between the estuarine waters and suspended particlesof South San Francisco Bay. The study is valuable in partbecause of the novel approach employed to examine metalsorption partitioning, with its potential application to future insitu studies using limno-corrals in natural aquatic systems. Thisstudy is also important because it uses estuarine seawater as amedium, while more of the existing literature on the subject hasfocused on sorption processes in fresh waters.

1.1. The Study Site: South San Francisco Bay Estuary

The San Francisco Bay estuary is a contaminated coastalembayment with respect to Ni, Cu, and Zn concentrations inboth sediments and dissolved phases. In South San FranciscoBay, where this study is focused, both dissolved Ni and Cuconcentrations can exceed Environmental Protection Agencymarine water quality criteria (Flegal et al., 1991) (Fig. 1).“Over three billion dollars have been invested on waste watertreatment facilities to improve the water quality of the estuaryduring the past three decades” (Flegal et al., 1996). Despitethese efforts, high dissolved metal concentrations persist in theestuary, suggesting possible causal relationships between his-torically contaminated sediments and contemporary aqueousconcentrations (e.g., South Bay concentrations of dissolved Niand Cu as high as 70 nM in summer, with dissolved Zn as highas 20 nM in summer; Flegal et al., 1991).

In addition, San Francisco Bay is a shallow estuary, with anaverage depth of only 6 m (Conomos, 1979) (Fig. 1). Largeportions of the Bay are as shallow as 2 m, particularly in theSouth Bay, allowing wind-induced resuspension of contami-nated bottom sediments to be a common, if not a daily occur-rence, during summer (Krone, 1979; Conomos et al., 1985).High suspended solids concentrations resulting from these re-suspension events tend to coincide with peak dissolved metalconcentrations in South Bay, although there have been nostudies to date investigating any causal link between the two

(data compared from Flegal et al., 1991, Flegal, 1993, andSchoellhamer, 1993).

To address the possibility of resuspended sediments acting asa source (and/or sink) for dissolved metal concentrations in theBay, the kinetic time scales of sorption reactions must first beassessed. Knowing the kinetic time scales of metal sorptionprocesses allows evaluation of the exchangeable particulatemetal pool as an essentially inert or labile chemical form in theoverall speciation of Ni Cu, and Zn in the South Bay. Thedynamics of sorption exchange between these chemical formscan then be evaluated within the context of other metal cyclingprocesses occurring in the Bay. These time scales can becompared with particle residence times in the water columnduring typical resuspension events, as well as with water massresidence times in the Bay. Overall, site-specific kinetic rateconstants estimated here can also be used in numerical metalpartitioning models to provide more rigorous evaluation of thefate and transport of metals injected into the Bay via waste andurban runoff processes.

2. BACKGROUND

Most existing environmental data on metal–particle associ-ations in natural waters are expressed as the extent to whichmetals have different particle binding affinities, and are re-ported as a distribution coefficient, KD. The term KD is anobservational parameter based upon “snap-shot” fi eld measure-ments and is defined as the measured ratio of metal concentra-tion on particles (Cs in moles of exchangeable metal per kg drysediment), to the concentration in water (Cw in moles metal perL water):

KD �Cs

Ch�L kg�1� (1)

Another common way to write KD is:

KD ��Mp�

�Md��Cp��1 �L kg�1� (2)

based on observed measurements of the exchangeable particu-late metal concentration [Mp] (moles per L, operationally de-fined here as that metal retained by a 0.4-�m membrane filterand subsequently leachable by 20% acetic acid), the dissolvedmetal concentration [Md] (moles per L), and the particle con-centration (Cp) (kg particles per L). Note that KD is usuallyreported as an observed parameter. It is often unknown if asystem is at equilibrium with respect to sorption processes, yeta true equilibrium KD value is only given if equilibrium dis-solved and exchangeable particulate metal concentrations areused in the calculation.

Results of a South Bay, January 1994 sampling effort at thesame sampling location as this study (Dumbarton Bridge sta-tion) report KD values for Ni, Cu, and Zn of 104.6, 104.2, and104.8 L kg�1, respectively (Sanudo-Wilhelmy et al., 1996).Mean observed KD values of 104.7, 104.3, and 105.5 L kg�1, forNi, Cu, and Zn in the South Bay were also calculated here usingFlegal (1993, 1995) Dumbarton Bridge station data (publishedin the San Francisco Estuary Institute Regional MonitoringProgram for Trace Substances Annual Reports, SFEI-RMPTS).These data came from six sampling events taken in different

3064 A. K. Gee and K. W. Bruland

seasons (three in each year), with measured particle concentra-tions that ranged from 3 mg/L to 72 mg/L. These mean KD

values correspond to 57% of total Ni, 74% of total Cu, and 30%of total Zn being in the dissolved phase in the South Bay in1993 and 1995.

A simple sorption reaction reflecting this metal partitioningcan relate kinetic rate constants to the distribution coefficient:

Md � particles

kf

34kb

Mp (3)

Assuming the principle of microscopic reversibility and usingEqn. 2 to calculate an equilibrium KD, we now have

kf �Cp) [Md]eq � kb[Mp]eq (4)

where kf is the forward adsorption rate constant for the reaction(Eqn. 3) proceeding to the right, and kb is the backward

desorption rate constant for the reaction mechanism proceedingto the left. kf is a second-order rate constant with units of L (kgparticles)�1 (day)�1 while kb is a first-order rate constant withunits of (day)�1. The particle concentration (Cp) is measurable.For a given experiment dealing with an isolated samplingevent, (Cp) can be treated as a constant. By incorporating (Cp)into the second-order forward rate constant kf, a new pseudo-first-order forward rate constant k'f with units of (day)�1 can bedefined, and Eqn. 4 can be simplified to:

k�f[Md]eq � kb[Mp]eq where k�f � kf(Cp) (5)

The ratio of the kinetic rate constants in Eqn. 5 is thus equiv-alent to the product of the equilibrium KD and the particleconcentration, according to Eqn. 2.

Perturbations to the system, such as inputs of metals indissolved or particulate form, will cause the variables [Md] and[Mp] to change with time depending on whether net adsorption

Fig. 1. Study location map of the San Francisco Bay Estuary. The water collection location is indicated by the solid circle,located near the Dumbarton Bridge in the South Bay.

3065Tracing Ni, Cu, and Zn partitioning in South San Francisco Bay

or net desorption is occurring. Given enough time, the systemwill come to a new equilibrium. We can formulate a newtransient nonequilibrium KD that will change over time suchthat

KD�t� ��Mp�t

�Md�t�cp��1 where KD (t3) � KDeq (6)

Figure 2a depicts such a scenario, in which the dissolved metalconcentration [Md] is suddenly increased at time zero. Netadsorption begins to occur, causing [Md] to drop and [Mp] toincrease. This process is countered by desorption and steadystate is eventually reached where the two concentrations againbecome constant. Figure 2b represents graphically the expectedevolution of KD (t), with (Cp) held constant. For a reversiblefirst-order reaction such as the sorption process assumed here,the characteristic reaction time “�” is equal to (k'f � kb)�1. Fortimes �, the reaction has not progressed significantly,whereas for times � �4� the reaction approaches completionand the system is near equilibrium (Morel and Hering, 1993).

The change in dissolved metal concentration with time canbe expressed as:

d�Md�

dt� �k�f �Md� � kb�Mp� (7)

While both dissolved and particulate metal concentrations canbe measured in the laboratory, for studies using tracers that areadded in dissolved form it is useful to rewrite the particulatemetal isotope concentration as a function of the dissolved metalisotope concentration at time t:

�Mp� � �Mp�0 � �Md�0 � �Md� (8)

Here [Mp]0 and [Md]0 are measured values (the initial partic-ulate and dissolved metal isotope concentrations when thesystem perturbation occurs). [Md]0 includes both the initialambient dissolved concentration as well as the added amount(spike) of that metal isotope. The difference ([Md]0 � [Md])represents the lowering of the dissolved isotope concentrationby adsorption onto the particulate phase at time t. SubstitutingEqn. 8 into Eqn. 7 gives:

d�Md�

dt� �k�f �Md� � kb��Mp�0 � �Md�0 � �Md�� (9)

Integrating Eqn. 9 yields a solution where only k'f and kb areleft unknown:

�Md� �1

�k�f�kb)�kb�Mp�0�1 � exp�(k�f�kb)t�

� �Md�0�kb � k�f exp��k�f � kb�� (10)

Equation 10 can be used to fit experimentally generated timeseries of changing dissolved isotope concentrations to deter-mine k'f and kb. Note that from Eqn. 10 the new equilibriumdissolved metal concentration following a perturbation can beexpressed as:

�Md�eq �kb��Mp�0 � �Md�0�

�k�f � kb�(11)

This is the essential background summarizing our approachto the quantification of sorption kinetics. It is adapted fromNyffeler et al. (1984), in which time series of dissolved con-centrations of metal radioisotopes were used to assess thekinetics of adsorption with natural particles. In our study,however, we also tracked the evolution of particulate metalconcentrations. This allows both an assessment of mass balanceand provides a second and independent data source for esti-mating kinetic constants. Substituting Eqn. 10 into Eqn. 8 givesan expression for the particulate metal concentration as a func-tion of time that can be used to curve fit experimentally deter-mined particulate time series data:

�Mp� � �Mp�0 � �Md�0 � � 1

�k�f � kb��kb�Mp�0�1 � exp��k�f�kb�t�

� �Md�0�kb � k�f exp��k�f�kb�t�� (12)

In analogy with Eqn. 11, the new equilibrium exchangeableparticulate metal isotope concentration from Eqn. 12 can beexpressed as:

�Mp�eq �k�f��Mp�0 � �Md�0�

�k�f � kb�(13)

Fig. 2. (a) The dissolved metal concentration [Md], the exchangeableparticulate metal concentration [Mp], and (b) the distribution ratio KD

as a function of time with the particle concentration held constant. Att � 0 an addition of dissolved metal is made. The system reachesequilibrium at approximately 4(k'f � kb)�1.


3. EXPERIMENTAL

3.1. Sampling Procedures

In conjunction with the summer 1997 sampling effort for theRegional Monitoring Program for Trace Substances (RMPTS)in San Francisco Bay (A. R. Flegal), a 20-L acid-washedcarboy was filled with unfiltered water from the South Bay, justnorth of the Dumbarton Bridge (monitoring station BA30).Simultaneously, three acid-washed 2.5-L polycarbonate bottleswere filled with 2 L 0.45-�m cartridge-filtered water to be usedas particle-free controls in the adsorption experiment. An ad-ditional L of 0.45-�m cartridge-filtered water was also takenand acidified, and 0.5 L of unfiltered water was taken andacidified. These acidified samples were used to determine am-bient metal concentrations prior to spike additions in the ad-sorption experiments. All unacidified samples were stored in acooler and immediately brought back to the laboratory.

The same day, the 20-L carboy was homogenized using aTeflon-coated propeller blade and shaft attached to a motor onthe outside of the carboy cap. Eight 2.5-L acid-washed poly-carbonate bottles were rinsed with a portion of this well-mixedwater-sediment suspension, and then filled to premeasured 2-Lvolume lines. Four of the eight 2.5-L unfiltered bottles werethen placed in black plastic bags to protect from light exposureand put on a shaker table to keep the sediment in suspension.The four remaining 2.5-L unfiltered bottles were then individ-ually microwaved using the high setting on a conventionalQuasar microwave oven. They were microwaved in 2-minincrements for a total of 14 min per bottle to sterilize thesamples in an attempt to control for microbial growth effects onadsorption (Keller et al., 1988). Once the microwaved bottlescooled they were also placed in black plastic bags and then puton the shaker table. The three control bottles containing0.45-�m cartridge-filtered Bay water were also placed in blackplastic bags on the shaker table. The total of eleven 2.5-Lexperimental bottles were then kept at constant temperature(24°C) in bags on the oscillating shaker table at all times duringthe 2-week experiment, except for when removal was necessaryfor filtration sampling.

3.2. Low Abundance Stable Isotope Spike Additions

Before spike additions, a trace metal clean, in-line duo-filtration system was set up in a clean hood. This system allowsdirect filtration of 100-mL subsamples out of the larger exper-imental bottles, providing “snap-shot” samples where “dis-solved” metals are separated from operationally defined “par-ticulate” metals at set times during the experiment. 0.4-�mpore size filter membranes (Poretics brand) that had beenpreviously soaked in 3 N HCl acid overnight and then rinsedwith MilliQ water were used in the filtration apparatus. Within24 h of sampling from the South Bay, additions of selected lownatural abundance, stable isotopes of Ni, Cu, and Zn were madeto each experiment bottle, marking the “ time zero” point in the2-week adsorption experiment. To each of the eleven 2.5-Lbottles on the shaker table 500 �L of 85.2 �mol 61Ni/kg, 500�L of 144.4 �mol 65Cu/kg, and 250 �L of 95.3 �mol 68Zn/kgwere added. These stock solutions were separately analyzedusing HR-ICPMS, to check for isotope impurities. This corre-sponds to concentration additions of 21 nM 61Ni, 2 nM 60Ni, 2

nM 58Ni; 36 nM 65Cu, 0.11 nM 63Cu; 12 nM 68Zn and 0.25 nM64Zn. Spikes were added to two bottles at a time, which werethen immediately shaken and placed on the duo-filtration rig toobtain the first set of postspike filtration samples. This processtook roughly 20 min between each set of bottles. 100-mLfiltration samples were then taken after 8 h and then at 1, 3, 6,10, and 14-day intervals. Experiment bottles were sampledfollowing the same sequence each time so that the filtrationsample times were correct within 20 min. For each sampletaken, the filtrate was filtered directly into 125-mL acid-cleanedpolyethylene bottles, acidified with 400 �L of 6 N QHCl, andstored double bagged. The filter membranes were placed inacid-cleaned 7-mL high-density polyethylene vials, doublebagged and frozen for later analyses.

3.3. Determination of Trace Metal Concentrations

Trace metal concentrations for both dissolved and particulatesamples were determined using HR-ICPMS. A method wasdeveloped that employs a combination of medium- and low-resolution mass spectrometry, counting and analog mode de-tection, and uses In, Ga, and Rh as internal standards. A suiteof metal isotopes were analyzed, including the stable isotopesof Cu, Ni, Zn, Fe, Mn, Co, Pb, Ag, and Cd. Mixed calibrationstandards were used, having the appropriate matching acidmatrix as the final samples analyzed from both dissolved andparticulate procedures.

Before analysis on HR-ICPMS, the dissolved samples re-quired a preconcentration step, which was accomplished usinga multiple-ligand sequential extraction technique. This precon-centration step also served to isolate the metals of interest fromthe original seawater matrix to an acid matrix ideal for subse-quent analysis with HR-ICPMS. The technique was developedfor this study from a synthesis of various procedures outlined inDanielsson et al. (1978), Bruland et al. (1979), Landing andBruland (1980), Bruland et al. (1985), and Batterham and Parry(1996). Here, a 100-mL filtered, acidified sample of Bay wateris transferred to a Teflon separatory funnel. The sample isbuffered at pH 4.3 with 400 �L of QNH4-Acetate. 500 �L ofa 1� % solution of each of the ligands 1-pyrrolidinedithiocar-bamate (PDC) and diethyldithiocarbamate (DDC) are thenadded to chelate dissolved Cu, Ni, Zn, Fe, Co, Pb, Ag, and Cdin the Bay water sample. The metal–dithiocarbamate com-plexes that form are neutrally charged and hydrophobic andthus, extract into chloroform. To extract the metals, 7 mL ofchloroform is added and the funnel shaken vigorously for 5min. The two phases are then allowed to separate for 5 min. Asthe chloroform solution containing the metal–ligand complexesis denser than the aqueous phase, it can be drained off into atrace metal clean polyethylene bottle suitable for use in theHR-ICPMS auto-sampler.

After the first extraction, the sample pH is adjusted to 9.3with 500 �L of QNH4-OH. Then a second extraction is per-formed by adding the ligand 8-hydroxyquinoline (oxine) al-ready dissolved in a chloroform solution to bind dissolved Mnand residuals of the other metals not extracted in the first step.Here 6 mL of 0.1� % oxine in chloroform is added to the same100-mL sample of Bay water. The funnel is again vigorouslyshaken for 5 min and the phases allowed to separate for another5 min. Then the second chloroform solution containing the


oxine–metal complexes is drained from the funnel and com-bined with the first chloroform extraction. Together they areacidified with 200 �L of concentrated nitric acid, evaporatedovernight in a laminar flow trace metal clean hood, and thentaken up in 5 mL of a stock 0.5 N nitric acid solution containingthe internal standards In, Ga, and Rh. This procedure results ina 20-fold concentration of the original sample metal levels intoa matrix suitable for ICPMS. It also simultaneously removesthe bulk of the major seawater salts that can build up on thetorch of the ICPMS as well as produce potential isotope inter-ferences upon analysis.

This method was verified to be quantitative and reproducibleusing metal radioisotope additions to South Bay water samples.Means and standard deviations for the total extraction percentrecoveries of each metal tested are: 95.3% 1.8% (n � 4) for109Cd, 94.3% 2.1% (n � 3) for 57Co, 93.7% 1.5% (n �4) for 59Fe, 94.1% 1.8% (n � 4) for 65Zn, and 91.3% 0.8% (n � 4) for 54Mn. These total extraction percent recov-eries derived in the radioisotope laboratory were taken intoaccount for each stable metal isotope analyzed using HR-ICPMS. The accuracy of the method was tested by extractingsamples of Canadian National Research Council certified ref-erence nearshore seawater (CASS-3) at the same time as fil-trates from the adsorption experiment. Results from theseCASS-3 organic extractions are compared with the certifiedconcentrations in Table 1. Organic extraction results for Co, Fe,Mn, Ni, and Zn are within the reported 95% confidence limitsof the certified concentrations. The value we obtained for Cuwas 18% higher than the certified value and falls just outsidethe reported 95% confidence limit.

The exchangeable particulate metal pool was assessed usinga 20% acetic acid leach, adapted from Landing and Bruland(1987). The frozen filter membranes were transferred to acid-cleaned centrifuge tubes and allowed to defrost. 7 mL of 20%HAc containing 50 ng/g of the internal standards In, Ga, and Rhwere added to each tube and shaken. The samples were kept atroom temperature for 2 h and then centrifuged at an effectiveg-force of 1600 for 5 min. The supernate was decanted backinto the acid-cleaned sample vials and 25 �L of concentratednitric acid was added to each acetic acid leachate. Duringearlier method development investigations, considerable signaldrift on the ICPMS was observed when running straight 20%acetic acid standards. However, this drift was removed if asmall amount of nitric acid was added to the standards, stabi-

lizing the signal and allowing direct analysis of metals in thecarbon-rich acetic acid matrix without requiring further wetchemistry manipulations.

3.4. Particle Concentrations

Particle concentrations for each of the eight unfiltered ex-perimental bottles were determined by preweighing 47-mm-diameter, 0.4 �m pore size Poretics polycarbonate filter mem-branes, filtering known volumes from each well-mixed bottle,and then drying the membranes in a Precision Thelco oven at68°C overnight. Membranes were placed in a dessicator for halfan hour to cool before reweighing. The mean particle concen-tration for the eight unfiltered bottles was 34.5 1.4 mg/L Baywater (n � 8). The particle concentration determined with the0.4 �m pore size Poretics filters on the 0.45-�m cartridge-filtered control bottles was found to be 0.9 0.1 mg/L water (n� 4), suggesting that there was a small amount of particles inthe size class between 0.4 and 0.45 �m.

4. RESULTS

For the metals Ni, Cu, and Zn, three sets of data weregenerated. These included dissolved and exchangeable partic-ulate concentrations as a function of time, graphs of changingisotopic ratios as a function of time, and curve-fits of meandissolved and exchangeable particulate isotopic concentrationsas a function of time, which were used to extrapolate kineticrate constants. Of the three elements studied, the Ni data setdisplays all of the following: mass balance between the ob-served loss of 61Ni from the dissolved pool and the resultingincrease in exchangeable particulate 61Ni over time, the dataconverge to a predicted isotopic equilibrium ratio, and thecurve-fits of the separately generated dissolved and exchange-able particulate 61Ni data give matching results for both for-ward and backward modeled kinetic rate constants.

4.1. Nickel Concentration and Isotopic Ratio Results

Before spike addition, “ initial ambient” dissolved metal con-centrations were measured from replicate subsamples of the0.45-�m cartridge-filtered and acidified South Bay water. ForNi, the mean total dissolved concentration was 56.0 1.5 nM(n � 6). The initial ambient acetic acid leachable (exchange-able) particulate Ni had an average value of 8.7 nM. Thus, atthe start of the experiment the Ni was predominantly dissolved,with only 13.4% occurring as exchangeable particulate Ni.Using the mean particle concentration from the experimentbottles of 34.5 1.4 mg L�1 (n � 8), an initial ambient Ni KD

of 103.65 L kg�1 can be calculated (Table 2).This initial ambient KD value is nearly an order of magnitude

lower than the mean Ni KD value calculated using the SouthSan Francisco Bay monitoring data from 1993 and 1995, pre-sented in the background section (Flegal, 1993, Flegal, 1995).Both seasonal and particle concentration effects are likely to beinvolved in this difference, but it should also be noted thatdifferent methods were used to estimate the exchangeableparticulate metal fraction. Here, filter membranes containingBay particles were exposed to a 20% acetic acid leach (pH �2) for 2 h to measure the exchangeable particulate metal pool.Under the regional monitoring program protocols, “ total dis-

Table 1. Comparison of certified concentrations and organic extrac-tion results.

CASS-3 certifiedconcentrationsa

(nM)

Organic extractionresultsb

(nM)

Mn 45.7 6.6 48.0 1.2Fe 22.6 3.0 23.1 0.2Co 0.70 0.15 0.75 0.04Ni 6.58 1.1 7.13 0.3Cu 8.14 1.0 9.58 0.1Zn 19.0 3.8 18.0 1.1

(n � 3)

a 95% confidence limits.b 1 standard deviation.


solvable metal concentrations” (unfiltered samples acidified topH �1.8 with hydrochloric acid and stored for a minimum ofa month before analysis) and “dissolved metal concentrations”(filtered samples acidified to pH �1.8) are measured. Theleachable particulate metal is then estimated as the differencebetween the measured dissolved and total dissolvable concen-trations. (Flegal, 1993).

The dissolved Ni data from the time series sorption experi-ment for the unmicrowaved samples are presented in Figure 3a.Each data point in the time series represents an isotope con-centration determined from separate bottle experiment repli-cates. Mean “ initial ambient” concentrations for the individualNi isotopes measured before spike addition are indicated abovethe asterisk at the left side of Figure 3a. These measured isotopeconcentrations were compared (as percentages of the measuredtotal ambient Ni concentration) with expected natural abun-dance percentages and agreed to within 1%. The isotope chosento trace Ni adsorption in the experiment was 61Ni, which undernatural abundance occurs at 1.13% (the measured 61Ni initialambient abundance percentage determined for South Bay was1.2%).

A 21 nM addition of dissolved 61Ni was made just beforefiltering the first set of samples. 2 nM of both 58Ni and 60Niwere also added along with the 21 nM 61Ni due to a slight Niisotope impurity in the 61Ni spike, giving a predicted “ timezero” total dissolved Ni concentration of 81 nM. The 61Niaddition altered natural abundance causing 61Ni to become thesecond most abundant Ni isotope in the bottle experiments. Inthe first filtration samples taken just after spiking, the 61Niaccounts for 28.4% of the total dissolved Ni concentration (upfrom 1.2%). Dissolved 61Ni concentrations are highest in thefirst filtration samples (average of 23 nM) and decrease withtime, reaching a concentration of 18.5 nM after 14 d. Resultsfrom the microwaved 14-d sample are included in Fig. 3a toshow that there was good agreement between the dissolvedisotope concentrations determined for microwaved and unmi-crowaved experimental bottles.

The �4.5 nM decrease in 61Ni dissolved concentrations seenover the 14-d period was reflected in a 3.6 nM increase inexchangeable particulate 61Ni concentrations in both unmicro-waved and microwaved bottle experiment replicates (Fig. 3b–c). 61Ni in the exchangeable particulate fraction already startedto increase in the first samples taken after the spike additionwas made, while natural abundance would predict 61Ni to beabout a third of what is observed. There was a 20-min lagbetween metal spike additions and completion of the first set offiltration samples. Together these suggest that a small amountof the added dissolved 61Ni had already become adsorbed toparticle surfaces by the time the first “ t � 20 min” sampleswere filtered. Results from microwaved replicates indicate asimilar pattern (Fig. 3c).

The perturbation to the system occurred by increasing thedissolved Ni fraction from 86.6% to 90.3% by t � 20 min justfollowing the spike addition (Table 2). By day 14 the dissolvedNi fraction dropped back to 80.4% of the day 14 total (Fig.3a–c). These results are reported as distribution coefficients(KD) in Table 2. As expected, due to the dissolved addition, thet � 20 min KD fell from the ambient value of 103.65 to 103.49

L kg�1. By day 14 it had increased to slightly higher than theoriginal ambient partitioning with a KD of 103.85 L kg�1.

These data can also be presented as changing isotopic ratioswith time. There was a modest decrease in the dissolved 61Ni/60Ni concentration ratio and a dramatic increase in the ex-changeable particulate 61Ni/60Ni concentration ratio (Figs. 3dand 3e). Due to the predominance of the dissolved pool inSouth Bay Ni partitioning (see Table 2), the small exchange-able particulate 61Ni/60Ni ratio was much more sensitive to the61Ni spike addition (Fig. 3e). The 4.5 nM 61Ni loss seen in thedissolved data was only 5.5% of the total dissolved Ni mea-sured just after the spike addition, whereas the 3.6 nM 61Niincrease seen in the exchangeable particulate data raised thetotal particulate Ni concentration measured just after the spikeaddition by 40%.

Predicted equilibrium ratios were also calculated for eachmetal isotope addition made. Tallying a budget that includesthe initial dissolved, exchangeable particulate and added iso-tope concentrations for each isotope in the ratio and thendividing appropriately gives a theoretical predicted equilibriumratio for the Ni sorption experiment:

��Diss 61Ni]i � [Partic 61Ni]i � [Added 61Ni])/([Diss 60Ni]i

� [Partic 60Ni]i � [Added 60Ni]) � Predicted Equilibrium Ratio

(14)

For 61Ni/60Ni, this resulted in a predicted equilibrium ratio of1.1, which both the dissolved and exchangeable particulateratio data converge to from different directions (Fig. 3d–e). Byday 14, the mean exchangeable particulate 61Ni/60Ni ratio datahad climbed 89% of the way from the 0.04 natural abundanceratio to the 1.1 predicted equilibrium ratio.

4.2. Nickel Kinetic Rate Constant Determinations

With the assumption that the sorption processes are revers-ible, rate constants were estimated by curve-fitting the experi-mentally derived data as outlined in the “Background” section.The dissolved 61Ni concentration data were modeled usingEqn. 10, and are presented in Figure 4a. A forward rate constantk'f of 0.038 d�1 and a backward rate constant kb of 0.13 d�1

were estimated. A close match was found for the backward rateconstant when the exchangeable particulate 61Ni data wereseparately modeled using Eqn. 12 (Fig. 4b). The forward rateconstant estimate that resulted from the exchangeable particu-late curve-fit was a k'f of 0.026 d�1. Both the k'f and kb valuesfrom the two independent data sets match within the curve-fiterror, with each curve representing 96 to 97% of the variancein each data set. The characteristic time of the sorption reactionbased on these results is 6 d, and at 4� or on the order of 25 d,the reaction should have approached equilibrium (Table 3a).These results concur with the previously discussed laboratoryfindings that the exchangeable particulate 61Ni/60Ni concentra-tion ratios were within 89% of the predicted equilibrium ratioafter 2 weeks.

4.3. Copper Concentration and Isotopic Ratio Results

The mean total initial ambient dissolved Cu concentrationwas 58.3 1.2 nM (n � 6). The initial ambient total acetic acidleachable particulate Cu concentration had an average value of15.3 nM. Thus, at the start of the experiment the Cu was


predominantly dissolved, with 20.8% occurring as exchange-able particulate Cu. Using these initial concentrations and themean particle concentration, an initial ambient Cu KD of 103.88

L kg�1 can be calculated (Table 2). This initial ambient KD is

within the range of KD values calculated from the South SanFrancisco Bay monitoring data (Flegal, 1993, Flegal, 1995),although on the lower end (mean KD for the six samplings inthe 2 yr of 104.3 L kg�1 for Cu).

Fig. 3. Nickel isotope data as a function of time. (a) Dissolved and (b–c) particulate isotope concentration data of Ni61

(closed triangles), Ni58 (open circles), Ni60 (open diamonds), and Ni62 (open squares). The arrow in (a) symbolizes themagnitude of the Ni61 spike addition made up from initial ambient concentrations. (d) Dissolved and (e) particulate isotopeconcentration ratio data of Ni61/Ni60 (closed triangles) and Ni62/Ni60 (open squares). The asterisk below the abscissaindicates the initial ambient isotope concentrations and ratios while the � indicates the natural abundance isotopeconcentration ratios. The dashed horizontal lines in (d) and (e) indicate the predicted equilibrium ratio.


The dissolved Cu isotope concentration data for the unmi-crowaved samples are presented in Figure 5a. 65Cu was used asthe Cu spike isotope, which at natural abundance accounts for30.8% of total Cu (initial ambient abundance measured forSouth Bay was 31.3%). For this element, only two stable

isotopes exist, neither one occurring at very low abundancepercentages under natural conditions. Thus 36 nM 65Cu wasadded to bring the dissolved 65Cu concentration within therange of ambient dissolved 63Cu. As a result of this spikeaddition, total dissolved Cu concentrations were increased by

Fig. 4. Best curve-fit models of (a) dissolved and (b) exchangeable particulate Ni61 concentration data. The results of theindependent model solutions are presented in the embedded tables. Eqns. 10 and 12 were used to model the data as presentedin section 2.


62%, with dissolved 65Cu concentrations remaining slightlyhigher than 63Cu concentrations throughout the 2-week exper-iment. Dissolved 65Cu in the t � 20 min samples accounts for53.9% of the total dissolved Cu measured (up from the 31.3%measured before spiking).

The addition of 36 nM dissolved 65Cu to the ambient 65Cuconcentrations measured gives a predicted concentration of54.3 nM 65Cu for the beginning of the experiment. However,the first postspike dissolved 65Cu concentrations measuredwere lower by 5.5 nM. This suggests a rapid removal process

Fig. 5. Copper isotope data as a function of time. (a) Dissolved and (b) particulate isotope concentration data of Cu65

(closed triangles) and Cu63 (open diamonds). The arrow in (a) symbolizes the magnitude of the Cu65 spike addition madeup from initial ambient concentrations. (c) Dissolved and (d–e) particulate isotope concentration ratio data of Cu65/Cu63

(closed triangles). (The *, �, and dashed horizontal lines are as in Fig. 3)


active in the 20-min time interval between adding the isotopespikes and completing the first set of filtration samples. Thedissolved 65Cu concentrations continued to show a rapid de-crease of �8 nM in the first 24 h of the experiment, followedby a steady decline over the remaining time examined, sum-ming to a total of 15 nM loss of 65Cu from the predicted 54.3nM dissolved pool immediately after the spike addition. Thedissolved 63Cu concentration data showed a similar althoughless dramatic pattern, as only 0.11 nM 63Cu was added alongwith the 36 nM 65Cu due to a slight isotope impurity in thespike used. There was good agreement between samples takenfrom different replicate and microwaved control bottle exper-iments (only one data point appearing per sample time repre-sents two overlapping and matching results).

The exchangeable particulate Cu data are presented in Figure5b. In the first filtration sample in the time series, 63Cu was stillthe more abundant Cu isotope in the exchangeable particulatefraction. By 8 h, however, the 65Cu exchangeable particulateconcentrations were higher than for 63Cu. Both Cu exchange-able particulate isotope concentrations continued to climb astime proceeded, although the increase again was less dramaticfor 63Cu. There was a 16 nM increase in exchangeable partic-ulate 65Cu concentrations over the course of the experiment, ingood agreement with the 15 nM decrease seen in the dissolved65Cu data. The exchangeable particulate 63Cu increased by 6.5nM over the 2 weeks, matching closely the 6.3 nM decreaseseen in the dissolved 63Cu concentrations over this time. Thusthere is a mass balance for both Cu isotopes in the two poolsstudied. The increase in exchangeable particulate concentra-tions and decrease in dissolved concentrations for both stableCu isotopes indicate that net adsorption proceeded for total Cusince the time of the 65Cu spike addition throughout the 2-weekexperiment (Fig. 5a–b).

The 36 nM dissolved 65Cu addition shifted the ambient Cupartitioning from 79.2% to 82.4% dissolved at t � 20 min. Bythe end of the 2-week period, the total dissolved Cu fractionhad fallen below ambient levels to 64% dissolved (Table 2). Asexpected due to the dissolved addition, the t � 20 min KD

dropped from the ambient value of 103.88 to 103.79 L kg�1. Byday 14 it was higher than ambient partitioning with a KD of104.21 L kg�1. The exchangeable particulate 65Cu concentra-tion measured in the t � 20 min samples represents �15% ofthe total 65Cu measured at that time in the bottle experiments,whereas by the end of the 2 weeks, the particulate 65Cu poolhad increased to �39% of the total 65Cu measured.

The dissolved Cu data are presented as isotopic concentra-tion ratios in Figure 5c. We expected the dissolved ratios todecline and approach the predicted equilibrium ratio fromabove, as was seen in the dissolved Ni data (Fig. 3d). However,most of the postspike ratio data hover at or below the predictedequilibrium ratio of 1.16 for 65Cu/63Cu (calculated according toEqn. 14). There is only a slight decline with time in the earliesttime series data. When the predicted isotopic ratio of 1.35calculated for conditions immediately after the spike addition isincluded in Figure 5c (closed square), the decline is moreevident, indicating that a significant part of the dissolved 65Curemoval process had to have occurred in the first 20 min of theexperiment.

The exchangeable particulate Cu isotopic ratio data alsoindicate that rapid adsorption occurred early on in the experi-

ment, especially of the 65Cu spike (Fig. 5d–e). At t � 20 minthe ratio was already halfway between the natural abundanceratio of 0.45 and the predicted equilibrium ratio of 1.16. Themicrowaved data showed a similar pattern (Fig. 5e). For thebulk of the experiment, the 65Cu/63Cu isotopic ratio in theparticulate pool was at or above the predicted equilibrium ratio.Together Figures 5a–e indicate that the 65Cu spike additiondrove net Cu adsorption, with a mechanism whereby the added65Cu was preferentially removed to the particulate phase overambient 63Cu.

4.4. Copper Kinetic Rate Constant Determinations

Curve-fits and corresponding kinetic rate constants extrapo-lated from both the dissolved and exchangeable particulate dataare presented in Figure 6a–b (solid curves) and the first twocolumns of the tables embedded in Figure 6. For both sets ofdata, these initial fits were not as close-fitting as those found forthe 61Ni data. Although the best fits still represent between 89%and 90% of the variance seen in each data set, the two pairs ofrate constant estimates do not match within the error of thecurve-fits. As the rate constants are not well constrained, asecond set of curve-fits were generated from the Cu dataexcluding the t � 20 min data and beginning with t � 8 h(dashed curves). Reasoning for this reinterpretation of the datais developed in section 5.2. The results of this second set of fits(given in the last two columns of the tables in Fig. 6) showsmaller rate constant solutions with higher R2 values for boththe dissolved and particulate Cu data sets. An estimate for k'fand kb of 0.07 d�1 and 0.12 d�1 respectively can be made forCu by averaging the solutions found for the dissolved andparticulate data, excluding the t � 20 min data. These rateconstant estimates suggest a characteristic reaction time of 5 dfor Cu with respect to metal sorption exchange in the watercolumn of South Bay (Table 3b).

4.5. Zinc Concentration and Isotopic Ratio Results

The mean total dissolved Zn concentration was 13.5 0.3nM (n � 6). The initial ambient total acetic acid leachableparticulate Zn had an average value of 15.5 nM. Thus, at thestart of the experiment the ambient Zn was roughly equallysplit between the dissolved and particulate fractions. Usingthese initial concentrations and the mean particle concentration,an initial ambient Zn KD of 104.52 L kg�1 can be calculated(Table 2). This initial ambient KD value is within the range ofKD values calculated from the San Francisco Bay monitoringdata (Flegal, 1993, Flegal, 1995), although on the lower end(mean KD of 105.55 L kg�1 for Zn based on six samplings in2 yr).

Initial ambient dissolved concentrations before spiking weredetermined for the four major stable isotopes of Zn (Fig. 7a).The natural isotopic abundance of 68Zn is 18.8%, comparedwith 17.7% measured in the ambient South Bay water. 12 nMof 68Zn was added as the dissolved Zn spike, which included a0.25 nM addition of 64Zn as well. As a result of this spikeaddition, 68Zn became the dominant Zn isotope throughout thecourse of the bottle experiments. In the t � 20 min samples,68Zn accounted for 54.4% of the total dissolved Zn (up from themeasured ambient abundance of 17.7%). 68Zn concentrations


dropped rapidly in the first 8 h of the experiment and thendecreased steadily, approaching dissolved 64Zn concentrations.A total dissolved 68Zn loss of 5.3 nM occurred through day 10of the experiment. There was good agreement between samplesfrom different bottle experiment replicates, with the exception

of samples from day 14. These day 14 data were omitted fromFigures 7a and 7c due to obvious Zn contamination in thesedissolved samples (obvious as the elevated concentrations oc-curred according to natural abundance and not according to theadjusted experiment isotopic ratios).

Fig. 6. Best curve-fit models of (a) dissolved and (b) exchangeable particulate Cu65 concentration data. The solid curvesare the best fits found using all of the time series data. The dashed curves are the best fits found, excluding the t � 20 mindata and beginning with the t � 8 h data. The results of the independent model solutions are presented in the embeddedtables.


The unmicrowaved exchangeable particulate Zn data arepresented in Figure 7b. Initially the exchangeable particulate68Zn concentration was below that of 66Zn but by 8 h hadexceeded it and continued to increase with time. Over thecourse of the experiment there was an observed 3.7 nM in-

crease in the exchangeable particulate 68Zn from ambient lev-els, low compared with the 5.3 nM drop seen in the dissolved68Zn concentrations (Fig. 7a). However, when mass balance fortotal Zn is considered, a closer match was found. The total Zndissolved data showed a 6.3 nM decrease with a corresponding

Fig. 7. Zinc isotope data as a function of time. (a) Dissolved and (b) particulate isotope concentration data of Zn68 (closedtriangles), Zn67 (open circles), Zn66 (open diamonds), and Zn64 (open squares). The arrow in (a) symbolizes the magnitudeof the Zn68 spike addition made up from initial ambient concentrations. (c) Dissolved and (d–e) particulate isotopeconcentration ratio data of Zn68/Zn64 (closed triangles), Zn67/Zn64 (open circles), and Zn66/Zn64 (open diamonds). (The *,�, and dashed horizontal lines are as in Fig. 3)


6.1 nM increase in the total exchangeable particulate Zn upfrom ambient levels.

The 12 nM dissolved 68Zn addition caused the total Znpartitioning to shift from 46.7% to 59.2% dissolved in the t �20 min samples. The total dissolved Zn fraction dropped backto 44.9% by the end of the experiment (Table 2). Due to the68Zn addition occurring in the dissolved form, the t � 20 minKD fell from the ambient KD of 104.52 to 104.30 L kg�1. By theend of the experiment the KD returned to close to the ambientvalue (104.55 L kg�1).

The dissolved concentration data are plotted as isotope ratiosin Figure 7c. The dissolved 66Zn/64Zn and 67Zn/64Zn ratiosshowed little change with time compared with 68Zn/64Zn. Thechange in slope of dissolved 68Zn/64Zn with time showed abreak at 8 h, became much more gradual by day 1 in theexperiment, and converged towards the predicted equilibriumratio of 1.14 (calculated according to Eqn. 14). As in thedissolved Zn ratio data (Fig. 7c), the exchangeable particulate66Zn/64Zn and 67Zn/64Zn ratios changed only slightly with time(Fig. 7d). Although the exchangeable particulate 68Zn/64Znclimbed slightly with time, it did not reach the 1.14 predictedequilibrium ratio by day 14.

4.6. Zinc Kinetic Rate Constant Determinations

The curve-fit for all the exchangeable particulate 68Zn data ispresented in Figure 8b (solid curve). This solution accounts for95% of the variance seen in the data and estimates k'f as 0.22d�1 and kb as 0.39 d�1. The curve-fit for all the dissolved 68Zndata showed more variance, with an R2 value of 0.86 (solidcurve, Fig. 8a). The kb estimates from both data sets matchedalmost exactly whereas the k'f estimates matched within themargins of error associated with both curve-fits. Curve-fitsexcluding the t � 20 min data and beginning with the t � 8 hdata were also generated for Zn (dashed curves with results inthe last two columns of tables in Fig. 8), the interpretations ofwhich are discussed in section 5.2. This second set of curve-fitsfinds smaller rate constant solutions, giving a slower charac-teristic reaction time of 4 d calculated using an average of thesenew rate constants (Table 3b).

5. DISCUSSION

5.1. Isotope Ratio Results and Dissolved SpeciationConsiderations

Examining the isotopic ratio data of the three metals inves-tigated reveals different responses to the sorption experiment.The Ni experiment proved to be the most successful, with boththe dissolved and exchangeable particulate Ni isotopic ratiodata converging towards the predicted equilibrium ratio withmass balance (Fig. 3d–e). The isotopic ratio data of both Cuand Zn deviated from this expected pattern in different ways,and can be interpreted in light of available dissolved metalspeciation data for South Bay.

For Cu, the nearly constant isotopic ratios seen in the dis-solved data at the predicted equilibrium value from day 1 of theexperiment suggest rapid removal of the added 65Cu from thedissolved phase (Fig. 5c). This is reflected in the roughly fourtimes faster kinetic rate constants determined from the initialCu curve-fits (solid curves, Fig. 6), compared with Ni. The high

65Cu/63Cu exchangeable particulate ratios suggest that theadded 65Cu was more readily adsorbed to particle surfaces thanthe ambient Cu, particularly at the very beginning of the ex-periment (Fig. 5d–e). These results can be explained by con-sidering the different chemical forms in which the ambient andadded dissolved Cu exists at the time the experiment begins.

The Cu isotope spike is added in a matrix of Milli-Q wateracidified with nitric acid, and so is present as dissolved inor-ganic species before being mixed with the Bay water samples.Upon addition to the estuarine water, the inorganic 65Cu spikeshould equilibrate essentially instantaneously with the inor-ganic, dissolved ambient Cu, here defined as [Cu'] (the sumtotal of hydrated free Cu2� and inorganically complexed Cu).Given the salinity and pH conditions of South Bay, 90 to 95%of ambient [Cu'] should be inorganically complexed Cu,(CuCO3

0 being the dominant species), with 5 to 10% as free Cu(Donat et al., 1994). However, studies present evidence thatnaturally occurring Cu-chelating organic ligands exist in SouthBay, dominating dissolved Cu speciation (80 to 92% of totaldissolved Cu being strongly complexed with organic ligands,here defined as CuL1, Donat et al., 1994). More recent dataindicate that �99% of Cu in South Bay is organically com-plexed, with a slight excess of strong Cu-binding ligands (Becket al., 2002). The increase of total dissolved Cu from �58 nMto �94 nM in this study would have titrated this excess ligand,leaving some portion of the 65Cu addition as Cu' rather than asCuL1. Under “ambient Bay conditions,” both the ambient 65Cuand the ambient 63Cu are strongly complexed with organicligands. We suspect that the rapid decrease in the first 24 h ofthe experiment of �7 to 8 nM dissolved Cu65 (Fig. 5a) was dueto the added Cu65 not being equilibrated with the Cu bound toorganic ligands, resulting in the preferential removal of Cu65 asCu' from the dissolved phase.

The exchangeable particulate Cu isotopic data (Fig. 5d–e)also suggest that a portion of the added inorganic 65Cu equil-ibrated with the suspended particle surface sites more quicklythan with the naturally occurring CuL complexes present in theBay water samples. If the kinetics of Cu' sorption to particleswere significantly more rapid than the kinetics of exchangewith the Cu bound to various Cu-binding organic ligands, suchan observed favored adsorption of the added inorganic 65Cuwould result. To date, the kinetics of Cu binding by suchnaturally occurring organic ligands remains unquantified. How-ever, from the speciation work of Donat et al. (1994), thebinding strengths of these South Bay Cu binding organic li-gands have been measured. Conditional stability constants(with respect to Cu2�) were found to be on the order of 1013.5

M�1 for the class of strong Cu-binding organic ligands. Suchstrong binding constants infer slow dissociation rates and sug-gest that the added 65Cu did not adequately equilibrate with theorganically chelated Cu, allowing the Cu65 tracer to becomepreferentially sequestered into particulate form. To test thisinterpretation, in future experiments it would be ideal to add theCu tracer in a form identical to ambient Cu, by somehowpreequilibrating it with filtered South Bay water containing anexcess of natural organic ligands.

For Zn, the dissolved 68Zn/64Zn isotope ratios fall towardsthe calculated predicted equilibrium ratio by day 10 of theexperiment, as expected (Fig. 7c). However, the exchangeableparticulate 68Zn/64Zn isotope data climb just 40% of the way to


the predicted equilibrium value by day 10 (Fig. 7d–e). Of themetals examined here, Zn was the only one where the ambientparticulate fraction was larger than the dissolved fraction (53%particulate, Table 2). Consider that if the 20% acetic acid used

here leaches more than the readily exchangeable fraction fromthe particles, then that portion of metal removed beyond theexchangeable particulate fraction should be comprised of iso-tope concentrations according to natural abundance isotope

Fig. 8. Best curve-fit models of (a) dissolved and (b) exchangeable particulate Zn68 concentration data. The solid curvesare the best fits found using all of the time series data. The dashed curves are the best fits found, excluding the t � 20 mindata and beginning with the t � 8 h data. The results of the independent model solutions are presented in the embeddedtables.


percentages. Here, this would mean more of the high abun-dance isotope 64Zn and less of the low abundance isotope 68Znbeing extracted from the particulate samples than expectedaccording to the predicted equilibrium ratio following thespike addition. This would dampen the increase in particu-late 68Zn/64Zn concentration ratios over time, as seen inFigure 7d–e. It is possible that this process is controlling thedata for Zn without having a significant effect on the Cu orNi particulate isotope ratio results because these latter twometals occur predominantly as dissolved species in SouthBay. These results call for further investigations that rigor-ously define and verify the exchangeable particulate fractionexperimentally. However, these results also illustrate thatthe low occurrence stable isotope tracing technique usedhere would lend itself well to such an investigation. Predictedequilibrium isotope ratios could be used as criterion for eval-uating the effectiveness of a suite of different leaching tech-niques at stripping off only the readily exchangeable particulatemetal fraction.

5.2. Alternate Interpretation of Copper and Zinc KineticResults

The rapid draw down of dissolved 65Cu in the first 8 to 24 hof the experiment may be the result of the added 65Cu initiallyoccurring as Cu' and being preferentially removed to the ex-changeable particulate phase relative to the ambient chelatedCu. As Cu has no truly low abundance stable isotope, a largeaddition of 36 nM of 65Cu was used to bring the dissolved 65Cuconcentration in the range of the ambient dissolved 63Cu con-centration. Only the Cu dissolved isotopic ratio data overshootthe predicted equilibrium ratio in this study. This result, com-bined with the observation that �99% of the Cu in South Bayhas been found to be organically chelated in recent years (Becket al., 2002), suggests that the 65Cu spike may have over-whelmed the small excess of naturally occurring Cu-bindingorganic ligands present in the samples. If this was the case, thenthe modeled kinetic rate constants generated by the initialcurve-fit solutions (solid curves) presented in Fig. 6a–b for Cumay be skewed towards faster sorption rates by the first timeseries data points.

Based on this interpretation of the data, we also modeled thedissolved and particulate Cu data excluding the t � 20 min datapoints and beginning with the t � 8 h concentrations (dashedcurves, Fig. 6a–b). These modeling results exhibited slowerforward and backward sorption rate constants with higher R2

values for both dissolved and particulate curve-fits (Table 3).Comparison of these results for Cu with the original Ni kineticresults shows more similar trends between metals, in terms ofboth sorption rate constant estimates and calculated character-istic sorption reaction times (Table 3). This second set of Cukinetic rate constants suggests that Cu sorption in South Bayshould approach equilibrium on the order of 3 weeks. Ni stillexhibits more sluggish sorption kinetics but the difference withCu is less dramatic.

Table 2. Time series results of two-week sorption experiment:Distribution coefficients and metal fractionations.

Nickel

(Ambient) (20 min) (14 days)KD � 103.65 L kg�1 KD � 103.49 L kg�1 KD � 103.85 L kg�1

86.6% Dissolved 90.3% Dissolved 80.4% Dissolved

Copper



Zinc



Table 3. Sorption reaction time scale and exchange rate estimates for South San Francisco Bay (per L per day).

aModeled forward rate

constant k�f

Modeled backward rateconstant kb

Characteristic reaction time� � [k�f � kb]�1

Ni 0.032 d�1 0.13 d�1 R2 � 0.96 6.3 dCua 0.20 d�1 0.43 d�1 R2 � 0.90 1.6 dZna 0.30 d�1 0.39 d�1 R2 � 0.90 1.5 d

bModeled forward rate

constant k�f

Modeled backward rateconstant kb

Characteristic reaction time� � [k�f � kb]�1

Ni 0.032 d�1 0.13 d�1 R2 � 0.96 6.3 dCub 0.068 d�1 0.12 d�1 R2 � 0.96 5.3 dZnb 0.121 d�1 0.15 d�1 R2 � 0.96 3.7 d

c

Ambientconcentrations

[Md] [Mp]

Adsorption rateestimatek�f � [Md]

Modeled desorptionrate estimate

kb � [Mp]

Ni 56 nM 8 nM 1.8 nM d�1 1.0 nM d�1

Cub 58 nM 15 nM 3.9 nM d�1 1.8 nM d�1

Znb 14 nM 16 nM 1.7 nM d�1 2.4 nM d�1

a Kinetic rate constants from initial Cu and Zn data models (Figs. 6 and 8, solid curves).b Kinetic rate constants from modeling of dissolved and particulate Cu and Zn data, excluding the t � 20 min data points and beginning with t �

8 h (Figs. 6 and 8, dashed curves).


If the sorption process is truly first-order or pseudo-first-order, then the reaction should follow the same kinetics duringall time segments as it approaches equilibrium. To test thisassumption, results were also generated for Ni and Zn wherekinetic rate constant estimates were made excluding the first t� 20 min data. The results for Ni excluding the first time seriesdata matched the original Ni rate constant estimates almostexactly. (Compare k'f � 0.038 d�1 and k b � 0.125 d�1 fordissolved Ni starting at 8 h with k'f � 0.038 d�1 and k b �0.128 d�1 for dissolved Ni using all of the time series data, Fig.4a). While recent studies have found much of the dissolved Niin South Bay to be complexed with ethylenediaminetetraacetate(EDTA) coming from sewage treatment plants (Sedlack et al.,1997; Bedsworth and Sedlack, 1999), we do not believe that theNi-binding organic ligands in our samples were titrated by the61Ni addition. This is based on the observation that both thedissolved and exchangeable particulate Ni isotopic ratio dataconverge on the predicted equilibrium ratio as expected, andthat there is no difference in the kinetic rate constant estimatesfound whether we include the first time series data or not.

In contrast, these results for Zn were more similar to Cu,showing higher R2 values and slower sorption kinetics thanoriginally estimated using all of the Zn time series data (dashedcurves, Fig. 8). These results suggest that for Cu and Zn but notfor Ni, more than one process may be controlling the dissolvedand particulate concentration data in the bottle experiments. ForCu we have considered that the small excess Cu-binding or-ganic ligand may have been titrated by the relatively large 65Cuaddition, leaving some of the added Cu to adsorb as Cu'. Suchdissolved speciation data do not exist for Zn in South Bay.Studies in Narragansett Bay, Rhode Island, have found dis-solved Zn species to range from 51 to 97% organically che-lated, with less organic complexation occurring in parts of theBay where total dissolved Zn exceeded the concentration ofZn-binding organic ligands (Kozelka and Bruland, 1998). It islikely that the case for Zn in South Bay is similar. However, thedifferent behavior seen in the Cu and Zn isotopic ratio datarelative to their predicted equilibrium ratios does not suggestthat the ambient Zn-binding organic ligands were titrated inexcess by the 12 nM 68Zn addition made here (compare Figs.5c–d with 7c–d). The Zn dissolved isotopic ratio data doproceed as expected to the predicted equilibrium value (Fig.7c). These data also show a sharp break in slope at 8 h,resulting in the large difference in the modeled rate constantsfor Zn excluding the t � 20 min data (Table 3).

The break in slope in the dissolved Zn isotopic ratio data andthe change in estimated sorption rate constants when the firsttime series data are excluded could suggest that Zn sorptionmay possibly be more of a second-order process compared withNi (Fig. 7c and Table 3). A number of processes could beinvoked to explain this pattern. Sorption and coagulation stud-ies investigating the role of colloids on sorption kinetics havefound rapid initial sorption followed by a slower reaction in thetransfer of metals from dissolved to particulate pools (Honey-man and Santschi, 1989; Stordal et al., 1996; Wen et al., 1997).In San Francisco Bay, however, colloidal (10 kDa–0.2 �m)concentrations of the metals considered here have not beenfound to be significant in high-salinity regions such as theSouth Bay (10% for Cu, 3% for Zn, and undetectable forNi, Sanudo-Wilhelmy et al., 1996). These results do not sug-

gest that colloids are influencing the kinetic data presented herefor these metals in South Bay. More experiment replication isrequired to address this issue conclusively, including the use ofpreequilibrated spike additions, more time series measurementsmade in the first 24 h of the experiment, and measurement ofthe colloidal metal fraction. Regardless of these considerations,the data find Zn to have the most rapid kinetics with respect tosorption of the three metals considered, even when the initialrapid sorption portion of the time series data is excluded (Table3b).

5.3. Kinetic Results and System Response Considerations

The kinetic results show Ni to have nearly two times slowersorption kinetics than Zn (excluding the t � 20 min data), withthe Cu results being intermediary (Table 3b). These slowerkinetics translate into different equilibration times in the SouthBay, taking roughly a month to approach equilibrium for Ni,about 3 weeks for Cu, and 2 weeks for Zn (estimates based on4�, Table 3). These results suggest that dissolved and particu-late Ni concentrations could show continued fluctuations for aslong as a month following inputs of dissolved or particulate Nito the Bay. In addition, if these perturbations occur in frequen-cies of less than a month, dissolved and particulate Ni concen-trations may continue to change, never reaching equilibriumvalues. In contrast, given the kinetic rate constants estimatedfor Zn, changing fractionations of this metal should come to anew equilibrium within about half the time, following systemperturbations (Table 3b).

The conceptual diagram presented in the background sectiongraphically illustrates this point (Fig. 2). Only at times greaterthan 4[k'f � k b]�1 following a system perturbation shoulddissolved and particulate concentrations as well as KD valuesbecome constant. This means that sampling events after asystem perturbation (at times less than 4�) may detect dissolvedand particulate concentrations that are quite different than equi-librium concentrations and that are likely to change if sampledagain at a later time. In addition, if sampling happens to occurimmediately after a system perturbation, metal fractionationresults may be reversed from what the dominant system ten-dency may be for that metal. Depending on the magnitude andtype of disturbance to the system, kinetic controls on sorptionreactions may be critically relevant to interpretations of “snap-shot” samples monitoring dissolved metal concentrations andKD values made by environmental scientists.

While there are differences between metals in the timepredicted to reach sorption equilibrium, there is a smaller rangeof values in the backward kinetic rate constant estimates than inthe forward rate constant estimates (Table 3). A similar trend isseen in the data set of Nyffeler et al. (1984). The forwardkinetic rate constants for the 12 elements investigated thereranged over five orders of magnitude whereas the backwardrate constants ranged over only one. Of the three metals studiedhere in South Bay, Nyffeler et al determined kinetic rate con-stants for Zn only. Using radioactive 65Zn and sediments fromNarragansett Bay, they found a backward rate constant of 0.32d�1, of the same magnitude as our 0.15 d�1 using stable 68Znand suspended sediments from South Bay. To compare forwardrate constants, we multiplied their reported 65Zn forward rateconstant by their reported particle concentration giving a k'f of


0.039 d�1 for Narragansett Bay. This is an order of magnitudesmaller than the k'f of 0.12 d�1 reported here for Zn in SouthBay. This result is surprising because the particle concentrationreported in Nyffeler et al. (1984) was an order of magnitudehigher than that measured here and there is evidence thatincreasing the particle concentration can cause an increase inmetal sorption forward rate constants and partition coefficients(Honeyman and Santschi, 1988).

Metal adsorption and desorption exchange rate estimates inSouth Bay can be calculated from the product of these kineticconstants and ambient dissolved and particulate metal concen-trations (Table 3c). For the sampling event of this study, theestimated adsorption rates for both Ni and Cu are nearly 2 timeslarger than the estimated desorption rates. For Zn, the estimateddesorption rate is larger. These exchange rate estimates, calcu-lated on a 1-L, daily basis, illustrate that the internal cycling ofmetals between dissolved and particulate forms in South Bay isdynamic and can result in significant concentration fluctuationson daily time scales. This appears to be most significant forexchangeable particulate Zn, as the estimated daily desorptionrate is roughly a fifth of the ambient exchangeable particulateconcentration measured in South Bay (Table 3c).

Although this study considers samples taken from only onesite at one time (under late summer conditions), we can usethese results to make a crude estimate of annual metal sorptionexchange rates for each metal. For these preliminary estimates,a volume of 48 � 107 m3 water for shallow areas in South SanFrancisco Bay (depth of 0 to 4 m) was selected (taken fromsurface area estimates by Fuller, 1982; cited in Rivera-Duarteand Flegal, 1997). Suspended particulate matter (SPM) con-centrations in these shallow areas can be high due to wind-induced resuspension. However, SPM concentrations similar tothose measured in this study are also seen in the channels ofSouth Bay (Schoellhamer, 1993, Schoellhamer, 1996; Schoell-hamer, USGS Water Resources of CA Web Site on Suspended-Solids Concentrations in San Francisco Bay, CA). This volumeestimate, with the modeled kinetic rate constants and ambientmetal concentrations (Table 3), can be used to estimate prelim-inary metal adsorption and desorption exchange rates.

Order of magnitude sorption exchange rate estimates havebeen summarized and compared with internal and externalmetal fluxes reported for South Bay (Table 4). These compar-isons are made to illustrate that particulate metal in the watercolumn of South Bay is by no means inert with respect tosorption processes. The adsorption and desorption exchangerate estimates in Table 4 estimate the magnitude of each metalexchanging back and forth between dissolved and particulateforms. Ideally we would like to know the net exchange fluxbetween dissolved and particulate metal pools by directly mea-suring both adsorption and desorption rates, rather than mod-eling the latter from adsorption data. To calculate net sorptionfluxes, one would also need more spatial and temporal parti-tioning and kinetic data obtained over a range of particleconcentrations. Without such measurements, we are cautiousnot to calculate a net sorption exchange flux here. Comparisonsof measured sorption exchange estimates with the internal andexternal metal fluxes considered in Table 4 should also be madeon time scales relevant to sediment resuspension events. How-ever, addressing these intentions requires more extensive fieldand laboratory investigations on benthic fluxes and sorptionphenomena, numerical models incorporating sorption kineticparameters, and measurement of external sources on short timescales throughout different seasons, the data of which do notcurrently exist for South Bay. Given this, it is still valuable tocompare the exchange estimates made here with reported fluxesin the South Bay on an order of magnitude basis.

Table 4 indicates that the sorption exchange estimates givenat the bottom of the table are of the same order of magnitude orlarger than both external and internal fluxes reported in theliterature for these metals in South Bay. While the combinedpoint source loads for all three metals are the most significantexternal inputs to South Bay, they are of the same magnitude asour internal sorption exchange estimates. For Ni and Cu, oursorption exchange estimates are at least one order of magnitudelarger than measured diffusive fluxes reported for South Bay(Kuwabara et al., 1996; Rivera-Duarte and Flegal, 1997). ForZn the diffusive flux is of the same magnitude as the estimatedZn sorption exchange estimates. These preliminary calculations

Table 4. Comparisons of metal cycling processes in South San Francisco Bay.

Nickel Copper Zinc

External fluxes (inputs) fromAtmosphere 700 kg/y 3,700 kg/y 19,000 kg/ya

Tributaries 5,000 kg/y 5,000 kg/y 7,200 kg/yb

Combined point sources 15,000 kg/y 16,800 kg/y 36,000 kg/yc

Internal fluxes to the water columnDiffusive benthic flux 5,000 kg/y 860 kg/y 27,000 kg/yd

Sorption exchange rates in the water columnAdsorption exchange rate estimate 18,000 kg/y 44,000 kg/y 19,000 kg/ye

Desorption exchange rate estimate 12,000 kg/y 20,000 kg/y 27,000 kg/ye

References refer to all data reported in individual rows of the table.a Estimated atmospheric inputs based on deposition rates measured in other parts of the U.S. (Gunther et al., 1987; cited in Rivera-Duarte and Flegal,

1997).b Average river concentration times average river discharge (Rivera-Duarte and Flegal, 1997).c Estimated loads entering South Bay from point sources (Davis et al., 1991 cited in Rivera-Duarte and Flegal, 1997).d Average diffusive benthic flux calculated using data from two contaminated sample locations in South Bay (Rivera-Duarte and Flegal, 1997) and

the volume of water in South Bay “shallows” (0–4 m depth).e This study, calculated using the volume of water in South Bay “shallows” (0–4 m depth), roughly an eighth of the total volume in South Bay.

Cu and Zn sorption flux estimates were calculated using kinetic rate constants from modeling of dissolved and particulate data excluding t � 20 min,data points and beginning with t � 8 h data (Table 3b).


clearly indicate that dynamic internal cycling is occurring be-tween dissolved and particulate forms on scales comparable tothe loading of these metals from external and internal fluxes toSouth Bay.

Keen interest exists in the predictive capacity of mixingmodels to address metal fate and transport in contaminatedestuarine systems such as the San Francisco Bay. However, theusefulness of numerical models in addressing metal partition-ing in natural systems is often more limited by the availabilityof field data such as sorption rate constants, rather than by anability to develop models that reflect the complexity of estua-rine environments. Wood and Baptista (1993) and Wood et al.(1995) have developed a model of trace metal partitioningspecifically for the physicochemical environment of South SanFrancisco Bay. The model is diagnostic in nature with thecapacity to assess kinetic controls on partition coefficients as afunction of proximity to system equilibrium. The authors con-clude that sorption kinetics can be responsible for variability inmetal partitioning at the basin-scale in South Bay. However,they express concern that the kinetic parameters used in theirmodel may not be representative of South Bay and emphasizethe critical importance of experimentally determined site-spe-cific rate constants to future modeling efforts and to increasethe understanding of metal cycling in South Bay.

5.4. Considerations for Future Studies

Although these observations are important, especially giventhe paucity of kinetic data available in the oceanographicliterature on metal sorption exchange in natural aquatic sys-tems, it is also critical to keep in mind the model and geo-chemical assumptions used to generate these results. The ex-perimentally derived data were modeled using equations thatassume pseudo-first-order kinetics for adsorption, and first-order kinetics for desorption. The irreversibility and hysteresisof metal sorption was not investigated as the desorption param-eters were simply modeled and not directly measured. Studiesthat have measured desorption rates have observed sorptionirreversibility for some metals, including Zn (Duursma andBosch, 1970; Li et al., 1984). As this experiment was con-ducted under a constant particle concentration, we cannot ad-dress particle concentration effects on sorption, an importantvariable shown to affect sorption kinetics in laboratory exper-iments (Li et al., 1984; Honeyman and Santschi, 1988, Hon-eyman and Santschi, 1989).

In addition, the method used here is based on an underlyingassumption that particle mass (kg) is the important particleparameter controlling metal adsorption. Especially for simpleion exchange processes, it is likely that a metal’s partitioncoefficient KD will vary more as a function of the total particlesurface area, rather than with the weight of particles per Lwater. As sorption to suspended matter can be more than simpleion exchange, concentrations of binding sites on particle sur-faces is another important parameter to investigate with respectto metal partitioning. Particle organic carbon content is alsolikely to play a significant role in controlling KD values inaddition to particle surface area (Santschi et al., 1997). Futureinvestigations of sorption kinetics in South Bay need to includesuch particle characterization studies on both South Bay sus-

pended sediments and standard reference material sedimentsfor comparison.

To address metal sorption kinetics using a surface complex-ation approach, one ideally needs to quantify the sum of allavailable exchangeable surface sites available to bind metal ina sample of water, defined as Si' (with units of moles availablesites per L water). This could be accomplished by performinga rigorous set of metal titrations using natural particles asbinding ligands. Si' can be summed with the sites alreadyoccupied by adsorbed metal to estimate ST, the sum total ofsurface sites per L water. Due to the heterogeneous nature ofnatural particles, Si' represents the sum of all available sites,whether on surface Fe and Mn oxide particle coatings, organicsurface ligands, or cation exchange sites on mineral particlesurfaces. Although these analyses were not included in thisstudy, the results presented here are an important first step inestablishing an overall characterization of the time scales anddynamics of metal exchange with heterogeneous suspendedparticles, reflective of what is happening in South Bay. To ourknowledge, the kinetic rate constants reported here are cur-rently the only ones available in the literature for South Bay,and are also the only ones available for Ni with natural particlesfrom any bay.

6. CONCLUSIONS

A first important outcome of this study is that the Ni data setillustrates the successful application of low abundance stableisotopes and HR-ICPMS to be used as a powerful tracer tech-nique to address complicated environmental chemistry ques-tions. Secondly, using this application, preliminary overall,site-specific sorption kinetic rate constants have been estimatedfor Ni, Cu, and Zn in South San Francisco Bay. A second setof sorption kinetic rate constants have been provided for Cuand Zn, based on consideration and interpretation of the Cu andZn concentration and isotope ratio time series data. Togetherthese results find Ni to have the most sluggish sorption kinetics,with predicted times to system equilibration with respect tosorption on the order of a month. Predicted system equilibrationtimes are on the order of 3 weeks for Cu and on the order of 2weeks for Zn.

Estimates of exchange rates between dissolved and particu-late forms calculated using these rate constants indicate thatsorption processes can cause dynamic internal cycling of thesemetals in South Bay, resulting in concentration fluctuations thatcan be significant, even on daily time scales. Preliminary esti-mates of the dynamic exchange of metal between dissolved andparticulate forms are found to be of the same order of magni-tude or larger than reported benthic fluxes and external sourcesof these metals to South Bay. Although future modeling effortsare required to address net sorption fluxes of metals on differenttime scales of significance in the Bay, these conclusions are asignificant first step in providing the site-specific sorption ki-netic rate constant estimates essential for such models. Inaddition, up until now, the role of sorption exchange in theinternal cycling of metals in the water column of South SanFrancisco Bay has remained largely unaddressed, despite itscrucial importance, due to the challenges associated with quan-tifying sorption processes on natural, heterogeneous particles.


Acknowledgments—We wish to thank R. Franks of the UCSC MarineAnalytical Laboratory for assistance with ICPMS analyses and G.Smith of the UCSC Institute of Marine Sciences for support with wetchemistry procedures. Our sincere thanks go to A. R. Flegal and I.Rivera-Duarte of the UCSC Environmental Toxicology Department forsampling support in South Bay. We thank R. S. Anderson of the UCSCEarth Sciences Department for his assistance with the data modelingprocedure. We are grateful to the Harbor Processes Program of theOffice of Naval Research Grant #N00014-99-1-0035 for funding thisresearch. Finally we would sincerely like to thank two anonymousreviewers of the originally submitted manuscript for their detailed andthoughtful comments.

Associate editor: W. H. Casey

REFERENCES

Anderson M. A. and Rubin A. J. (1981) Adsorption of Inorganics atSolid–Liquid Interfaces. Ann Arbor Science, Ann Arbor, MI.

Atkinson R. J., Posner A. M., and Quirk J. P. (1972) Kinetics ofisotopic exchange of phosphate at the �-FeOOH-aqueous solutioninterface. J. Inorg. Nucl. Chem. 34, 2201–2211.

Balistrieri L. S. and Murray J. W. (1986) The surface chemistry ofsediments from the Panama Basin: The influence of Mn oxides onmetal adsorption. Geochim. Cosmochim. Acta 50, 2235–2243.

Batterham G. J. and Parry D. L. (1996) Improved dithiocarbamate/oxine solvent extraction method for the preconcentration of tracemetals from seawater using metal exchange back-extraction. Mar.Chem. 55, 381–388.

Beck N. G., Rue E. L., Bruland K. W. (2002) Short-term biogeochemi-cal influence of a diatom bloom on the nutrient and trace metalconcentrations in a South San Francisco Bay microcosm experiment.Estuaries (submitted).

Bedsworth W. W. and Sedlack D. L. (1999) Sources and environmentalfate of strongly complexed nickel in estuarine waters: The role ofethylenediaminetetraacetate. Environ. Sci. Technol. 33, 926–931.

Benoit G., Oktay-Marshall S. D., Cantu II A., Hood E. M., ColemanC. H., Corapcioglu M. O., and Santschi P. H. (1994) Partitioning ofCu, Pb, Ag, An, Fe, Al, and Mn between filter-retained particles,colloids, and solution in six Texan estuaries. Mar. Chem. 45, 307–336.

Bruland K. W., Coale K. H., and Mart L. (1985) Analysis of seawaterfor dissolved cadmium, copper and lead: An intercomparison ofvoltametric and atomic absorption methods. Mar. Chem. 17, 285–300.

Bruland K. W., Franks R. P., Knauer G. A., and Martin J. H. (1979)Sampling and analytical methods for the determination of copper,cadmium, zinc and nickel at the nanogram per liter level in seawater.Anal. Chim. Acta 105, 223–245.

Conomos J. T. (1979) Properties and circulation of San Francisco Baywaters. San Francisco Bay: The Urbanized Estuary. In: (eds. J. T.Conomos, A. E. Leviton, and M. Berson). Pacific Division of theAmerican Association for the Advancement of Science, San Fran-cisco, CA, pp. 47–84.

Conomos J. T., Smith R. E., and Gartner J. W. (1985) Environmentalsetting of San Francisco Bay. Hydrobiologia 129, 1–12.

Danielsson L. G., Magnusson B., and Westerlund S. (1978) An im-proved metal extraction procedure for the determination of tracemetals in seawater by atomic absorption spectrometry with electro-thermal atomization. Anal. Chim. Acta 98, 47–57.

Davis J. A., Gunther A. J., Richardson B. J., O’Connor J. M., SpiesR. B., Wyatt E., Larson E., and Meiorin E. M. (1991) Status andTrends Report on Pollutants in the San Francisco Bay Estuary.Aquatic Habitat Institute, Richmond, CA 240 p.

Donat J. R., Lao K. A., and Bruland K. W. (1994) Speciation ofdissolved copper and nickel in South San Francisco Bay: A multi-method approach. Anal. Chim. Acta 284, 547–571.

Duursma E. K. and Bosch C. J. (1970) Theoretical, experimental andfield studies concerning diffusion of radioisotopes with sedimentsand suspended particles of the sea: Part II, methods and experiments.Neth. J. Sea Res. 4, 395–469.

Flegal A. R. (1993) Dissolved and total trace element data tables. InSan Francisco Estuary Institute Regional Monitoring Program for

Trace Substances 1993 Ann. Report. U.S. Geological Survey, Sac-ramento, CA. Appendix 2, pp. 159–161.

Flegal A. R. (1995) Dissolved and total trace element data tables. InSan Francisco Estuary Institute Regional Monitoring Program forTrace Substances 1993 Ann. Report. U.S. Geological Survey, Sac-ramento, CA. Appendix C, pp. A15–A20.

Flegal A. R., Rivera-Duarte I., Ritson P. I., Scelfo G. M., Smith G. J.,Gordon M. R., and Sanudo-Wilhelmy S. (1996) Metal contaminationin San Francisco Bay waters: Historic perturbations, contemporaryconcentrations, and future considerations. San Francisco Bay: TheEcosystem. In: (ed. J. T. Hollibaugh). Pacific Division of the Amer-ican Association for the Advancement of Science, San Francisco,CA, pp. 173–188.

Flegal A. R., Smith G. J., Gill G. A., Sanudo-Wilhelmy S., andAnderson L. D. C. (1991) Dissolved trace element cycles in the SanFrancisco Bay estuary. Mar. Chem. 36, 329–363.

Fuller C. C. (1982) The use of Pb-210, Th-234 and Cs-137 as tracersof sedimentary processes in San Francisco Bay, California. M.S.thesis, Univ. of Southern California.

Gunther A. J., Davis J. A., and Phillips D. J. H. (1987) An Assessmentof the Loading of Toxic Contaminants to the San Francisco BayDelta. Aquatic Habitat Institute, Richmond, CA 330 p.

Hayes K. F. Leckie J. O. (1986) Kinetic studies of lead ion adsorptionat the goethite/water interface using the pressure jump relaxationtechnique. In Surface Processes in Aqueous Geochemistry (eds. J. A.Davis and K. F. Hayes). Am. Chem. Soc. Symp. Series.

Honeyman B. D. and Santschi P. H. (1988) Metals in aquatic systems.Environ. Sci. Technol. 22, 862–871.

Honeyman B. D. and Santschi P. H. (1989) A Brownian-pumpingmodel for oceanic trace metal scavenging: Evidence from Th iso-topes. J. Mar. Res. 47, 951–992.

Jannasch H. W., Honeyman B. D., Balistrieri L. S., and Murray J. W.(1988) Kinetics of trace element uptake by marine particles.Geochim. Cosmochim. Acta 52, 567–577.

Keller M. D., Bellows W. K., and Guillard R. L. (1988) Microwavetreatment for sterilization of phytoplankton culture media. J. Exp.Mar. Biol. Ecol. 117, 279–283.

Kozelka P. B. and Bruland K. W. (1998) Chemical speciation ofdissolved Cu, Zn, Cd, Pb in Narragansett Bay, Rhode Island. Mar.Chem. 60, 267–282.

Krone R. B. (1979) Sedimentation in the San Francisco Bay system.San Francisco Bay: The Urbanized Estuary. In: (eds. J. T. Conomos,A. E. Leviton, and M. Berson). Pacific Division of the AmericanAssociation for the Advancement of Science, San Francisco, CA, pp.85–96.

Kuwabara J. S., Chang C. C., Cloern J. E., Fries T. L., Davis J. A., andLuoma S. N. (1989) Trace metal associations in the water column ofSouth San Francisco Bay, California. Estuarine Coastal Shelf Sci.28, 307–325.

Kuwabara J. S., Chang C. C., Khechfe A. I., and Hunter Y. R. (1996)Importance of dissolved sulfides and organic substances in control-ling the chemical speciation of heavy metals in San Francisco Bay.San Francisco Bay: The Ecosystem. In: (ed. J. T. Hollibaugh).Pacific Division of the American Association for the Advancementof Science, San Francisco, CA, pp. 157–172.

Landing W. M. and Bruland K. W. (1987) The contrasting biogeo-chemistry of iron and manganese in the Pacific Ocean. Geochim.Cosmochim. Acta 51, 29–43.

Landing W. M. and Bruland K. W. (1980) Manganese in the NorthPacific. Earth Planet. Sci. Lett. 49, 45–56.

Li Y. H., Burkhardt L., Buchholtz M., O’Hara P., and Santschi P. H.(1984) Partition of radiotracers between suspended particles andseawater. Geochim. Cosmochim. Acta 48, 2011–2019.

Morel F. M. M., Hering J. G. (1993) Kinetics. In Principles andApplications of Aquatic Chemistry. pp. 98–155. John Wiley & Sons,New York.

Nyffeler U. P., Li Y. H., and Santschi P. H. (1984) A kinetic approachto describe trace-element distribution between particles and solutionin natural aquatic systems. Geochim. Cosmochim. Acta 48, 1513–1522.

Rivera-Duarte I. and Flegal A. R. (1994) Benthic lead fluxes in SanFrancisco Bay, CA, USA. Geochim. Cosmochim. Acta 58, 3307–3313.


Rivera-Duarte I. and Flegal A. R. (1997) Porewater gradients anddiffusive benthic fluxes of Co, Ni, Cu, Zn and Cd in San FranciscoBay. Croat. Chemica Acta 70, 389–417.

Santschi P. H., Lenhart J. J., and Honeyman B. D. (1997) Heteroge-neous processes affecting trace contaminant distribution in estuaries:The role of natural organic matter. Mar. Chem. 58, 99–125.

Sanudo-Wilhelmy S. A., Rivera-Duarte I., and Flegal A. R. (1996)Distribution of colloidal trace metals in the San Francisco Bayestuary. Geochim. Cosmochim. Acta 60, 4933–4944.

Schoellhamer D. H. (1993) San Francisco Bay suspended-sedimenttransport processes study and comparison of continuous and discretemeasurements of suspended-solids concentrations. In San FranciscoEstuary Institute Regional Monitoring Program for Trace Sub-stances 1993 Ann. Report. U.S. Geological Survey, Sacramento, CA.pp. 129–134.

Schoellhamer D. H. (1996) Factors affecting suspended-solids concen-trations in South San Francisco Bay, California. J. Geophys. Res.101, 12087–12095.

Schoellhamer D. H. USGS Water Resources of CA Web Site onSuspended-Solids Concentrations in San Francisco Bay, CA. http://water.wr.usgs.gov/abstract/sfbay/index.html.

Sedlack D. L., Phinney J. T., and Bedsworth W. W. (1997) Stronglycomplexed Cu and Ni in wastewater effluents and surface runoff.Environ. Sci. Technol. 31, 3010–3016.

Smith G. J. and Flegal A. R. (1993) Silver in San Francisco Bayestuarine waters. Estuaries 16, 547–558.

Stordal M. C., Santschi P. H., and Gill G. A. (1996) Colloidal pumping:Evidence for the coagulation process using natural colloids taggedwith 203Hg. Environ. Sci. Technol. 30, 3335–3340.

Wen L. S., Santschi P. H., and Tang D. (1997) Interactions betweenradioactively labeled colloids and natural particles: Evidence forcolloidal pumping. Geochim. Cosmochim. Acta 61, 2867–2878.

Wood T. M. and Baptista A. M. (1993) A model for diagnostic analysisof estuarine geochemistry. Water Resources Res 29, 51–71.

Wood T. M., Baptista A. M., Kuwabara J. S., and Flegal A. R. (1995)Diagnostic modeling of trace metal partitioning in south San Fran-cisco Bay. Limnol. Oceanogr. 40, 345–358.


Tracing Ni, Cu, and Zn kinetics and equilibrium ... › ~kbruland › ... · time and location....

Documents

Transcript of Tracing Ni, Cu, and Zn kinetics and equilibrium ... › ~kbruland › ... · time and location....