Kinetic studies of metal ion speciation

Analyitca Chrmrca Acta, 256 (1992) 183-201 Elsevler Science Pubhshers B V , Amsterdam

183

Kinetic studies of metal ion speciation

Cooper H Langford * and Donald W Gutzman Department of Chemrstry and Bmchemtstty, Concordra Unrversrty, 1455 deh4a~onneuue W, Montreal, Quebec, H3G IU8 (Canada)

(Recemed 29th March 1991, revised manuscript received 3rd July 1991)

Abstract

The use of kmetics of metal ion transfer from naturally occurrmg colloidal hgand systems to complexes with spectrophotometncally favourable hgands as a means of study of speciation of metal ions m natural systems IS described The systems have a species dlstnbution which IS determined by the conditions of sample “eqmhbratlon”, but kinetic analysis is conducted under condltlons (pH, ionic strength) defined by the reagent solutron The resulting species dlstrtbution is that of the eqmhbratmn condmon with relative lability characterized Apphcatlon of the approach to complexes of simple metal ions, Cuzc and Ni*+, with humlc collolds and hydrous fernc oxides and to ternary systems of iron(U) or alum~mum(III) with humlcs and OH- are revIewed The latter studies have been extended to application to natural waters The conditions for extension of the study of simpler systems to actual natural samples are considered A method based on laser thermal lensmg IS proposed

Keywords Kmetic analysis, Spectrophotometry, Iron, Laser thermal lensmg, Metal Ions, Speaatlon, Waters

In environmental toxicology, an idea that has often been discussed IS the so-called “free metal ion” hypothesis [l] This IS the proposal that the toxlclty of a metal ion 1s related to the aqua ion concentration It 1s Introduced to account for reduced metal ion toxlclty observed m natural waters m the presence of complexmg agents, es- pecially humlc substances It 1s rarely noted that there are, m fact, three possible orlgms of the protective effect of complexmg agents The obvl- ous one is a sunple “free metal ion” effect related to preferred uptake of the aqua metal Ion A related thermodynamic effect IS uptake gov- erned by the metal ion activity These are both related to equlhbrmm speclatlon A third aspect involves the kmetlcs of uptake As hfe processes are dynamic, this cannot be overlooked The rele- vance of eqmhbrmm speclatlon will, m fact, be limited to those cases where the equlhbrla are estabhshed rapidly with respect to the time scale of uptake This Issue has been analyzed by Buffle [2] Unfortunately, almost all of the large htera-

ture on metal Ion speciation m natural systems is based on equdlbrmm studies There 1s good reason for this Methods for exammatlon of kmetlc speaatlon with high resolution and high senntlv- lty are extremely hard to design This paper describes an approach that provides a compromise between equlllbrmm and kmetlc speclatlon and that 1s now bemg extended from laboratory models to “real” samples

In order to understand the basis of the method, consider a classical colorlmetrlc (or fluonmetnc) determination A sample containing a metal ion 1s treated with excess of color-formmg reagent Ideally, 100% of the metal ion enters the detection species and 1s determined All that 1s now done 1s to add momtormg of the time course of the formation of the detection species If the tnne course can be resolved into contrlbutlons from the competing reaction of the various species present m the sample, the various species can be identified and determined In order to maintain reproducible and analyzable kmetlcs, It 1s neces-

0003-2670/92/$05 00 0 1992 - Elsevler Science Pubhshers B V All rtghts reserved

184 C H LANGFGRD AND D W CiVTZMAN

sary to control the kmetlcally crucial parameters of temperature, pH and iomc strength Hence the reagent solution must be designed to buffer pH and Ionic strength This means that the speaatlon IS controlled by the history of the sample pnor to the mltlatlon of the metal Ion transfer reaction, but that the rate constants are specific to the condztlons of the run Thus experimental hmlta- tlon is not as serious for metal complexes as it might be m other circumstances, as sur-coordinate metal complexes tend to have reaction rates controlled by the rate of hgand dlssoclatlon, and extrapolation from one condmon to another 1s fairly simple and r&able In particular, relative rates tend to remam the same

It should be evident that the same factors which govern equillb~um colonmetrIc analysis are important m choosing a chromophore for use m kmetlc speclatlon The equlhbrmm of analyte with chromophore should he far to the product side Sensltwlty IS maxlmlzed by use of ar absorbmg complex with a large absorptwlty There is one factor which 1s more crrtlcal m kmetlc study than m equlhbrmm analysis as the samples are mom- tored for an extended time period (sometimes m the regzon of hours), it is necessary that the chromophore remains stable over this period

?mpanv 2 d&rent s&es S 1 stngle site s

zwo k4-lc~ effects non *cm m-62) effects -+ 2 sites L -t large number of

Sates L

TypIcal mciture of sample homopolvsacchandes axamplw Ilgsnds (e g CO;- (ate = -COOH)

ilmmo acads) or ample ltgand wrth A0 vanntwm 4s mostly

2 mdapmdent ba?s due to &acts&l and (4

In order to be able to develop theones of these methods, It IS clearly useful to clarify the nature of the complexmg agents involved Figure 1 outlines several possible dlstrlbutlons of metal complexmg s&es S-sites refer to the functional group of the lrgand directly mvolved m bmdmg L-sites encompass both the coordmatmg group and Its environment, the effects of whch can influence complex stab&y Such secondary effects have been divided by Buffle [3] mto three types, denoted (a>-k$ m Fig 1 Pol~nctional (a-effects) are related to the multlfunctlonahty of the complexmg agent These arlse from the drver- s~ty of S&es and then- electromc and steric envi- ronments Conformation (b-effects) of the overall complexant is largely dependent on the degree of hydra&on and the extent of hydrogen bondmg and brldgmg metal groups These factors can cause the material to appear as dlstmctly dlffer- ent phases such as simple solutes, hydrated surface layers on particles or aggregated gels The thu-d effect (c) IS that due to the polyelectrolyte properties of the complexant The importance of any of these effects IS dependent on the extent of occupation of bmdmg sites

Scheme A m Fig 1 shows bmdmg of a metal by two simple hgands A readdy quantifiable

Blmodal dlstnbutmn

2SlteaS

pepttdes te g sites a*

-COOH and -NH11

PROM AND AROM (sites = -CooH “9”

-NH* -94 ate AG vanat!on IS mostly

due to effects f8l and fcl

Fig 1 Possible chstnbutlons of sites forming metal-hgand complexes p(AG c, ) IS the probablhty of a site being present which will complex a metal wth a free energy of formation of AG ’ PROM and AROM refer to pedogemc and aquagemc refractory organtc matter, respectwely See text for defmztrons of S-sites, L-sates and (a)-(c) effects (ReprInted from [31 by permzsszon of the copyright holders Ellis Horwood, ChIchester, UK )

KINETIC STUDIES OF METAL ION SPECIATION 185

complexatlon constant for each 1s evtdent In scheme B bmdmg by a single S-site 1s distributed over a range of AGo values as a result of secondary effects More complex dlstrlbutlons of sta- bilities are seen m schemes C and D

THEORY

Consider the complexes of Ni’+ with the acid-soluble sod humlc fraction called fulvlc acid (FA) as they react with a reagent R to form a spectrophotometrlcally detectable product NIR There is a set of competing reactions

Ni’+(aq ) + R -+ NiR

FA,-NI+R+FA,+NIR

FA,-Nl + R --) FA, + NIR

FA,-NI + R + FA, + NlR

The subscripted FA, represent the different bmdmg sites for Nl(II) on the FA The first reaction with free Ni2+ IS the fastest Provided that a large excess of R IS used, all reactions will reduce to pseudo-first-order kmetlcs More fun- damentally, the rates of all reactions after the first will often but not always a be described mechanistically as

FA,_Nl (s’ow) --+FA, + Nl*+(aq )

Nl*+(aq ) + R- (fast) NIR

In this favorable case, the rate constant measured for each component will be the dlssoclatlon rate constant of the correspondmg complexmg site With all reactions reduced to pseudo-first order,

a In some Instances hgand-exchange reactlons are of the type where the rate-hmltmg step 1s dlssoclatlon of an mtermedl- ate ternary complex [4] In circumstances such as the condo- tlons used m experiments described here, pseudo-first-order reactlons are guaranteed Species ldentlfled are species of an equlhbrmm dlstrlbutlon However, extrapolation of the rate constants based on a mechanrstlc assumption IS uncer- tam

the experimental rate law with excess of R takes the form

d[NlR]/dt = &$,,[ 1 - exp( -k,t)] +X (1)

where A, I , represents the mmal (time zero) concentration of the z th species expressed m umts consistent with the product NIR, k, IS the rate constant for reaction with the I th species and X IS a time-independent term which contains the spectrophotometrlc blank plus contrlbutlons made by any species which react fast on the time scale of the measurements and therefore appear to be independent of time The task 1s to fit the observed time course to a set of rate constant and concentration parameters, k, and C(0, z) (related stolchlometrlcally to A,,) As the equation 1s non-linear, this 1s a difficult task

The approach Just described was stimulated by an interesting series of papers on kinetic analysis by Margerum and co-workers [51 In a typical study m this series, a mixture of ammopolycarboxylates was analyzed by formmg then complexes with Nl(II), then reacting the mtiure with excess of cyanide, momtormg the formation of [Ni(CN>z-] The number of components is known m this instance and a rate constant for each can be independently estimated by study of the ammopolycarboxylates mdwldually The fitting problem IS only the linear problem of fmdmg the concentrations Unfortunately, m the case of natural mixture colloidal polymeric hgands such as humlc substances (e g , FA), there IS no way to study the mdlvldual bmdmg sites separately, or even to know a priori how many distinct sites to assume-the actual situation may be a nearly continuous dlstrlbutlon Hence the first task is to develop an appropriately ObJective procedure to assign a number of components Then, non-linear fitting must be used to yield the parameters

Two procedures have been used to establish an mitral approxlmatlon to the number of components and the rate constants In the first, the concentration-tnne data can be plotted using the Guggenheun method for obtammg first-order rate constants when the fmal concentration 1s un- known This plots differences m log(concentration) for a fared time interval vs time In the

186 C H LANGFORD AND D W GlJlZhtAN

present context, the plot 1s searched for linear regions where each component dommates and the slopes of those linear regions provtde the mittal estimates of rate constants [61 Intercepts provide approxlmatlons of component amounts

A better procedure was introduced by Shuman and co-workers [7,81 This IS a kmetlc analog of the affinity spectrum method of analysis [71 of the dlstrlbutlon of equlhbrmm constants for a multi- site complexant It uses a numerical approxlmatlon of a Laplace transform The startmg point 1s analogous to an mtegrated form of Eqn 1

C(t) = CC(O, E) exp( -Q) (4

where C(t) represents the decrease over tnne of the overall concentration of the complexes (e g , FA complexes), which IS mirrored by the formation of the product (e g , NIR) The summation m Eqn 2 may be replaced by an mtegral either as an approxlmatlon to a fmlte sum or as the repre- sentatlon of the essentially contmuous dlstrlbu- tron of bmdmg sites which may arlse with the heterogeneous mtiure polymeric hgand systems

C(f) = S/H(k) exp( -kt) d[ln k] (3)

where the quantity H(k) d[ln k] 1s the probabd- lty of fmdmg a site with a rate constant m the range from In k to In k + d[ln kl m molar concentratlon umts Second-order approxunate solutions can be obtamed usmg the first and second derlvatlve of C(t) and the approximate mverslon of a Laplace transform expression equivalent to Eqn 3 where the Integral is taken with respect to In k m place of k The resultmg expresslons are

H(k) = d2C( t)/d(ln t)* - dC( t)/d(ln t) (4)

s=4C(t)/CC(O, 1) (5)

H(k) and S ongmate from the approxnnate Laplace transform and the value of S depends on whether the first, second or nth derlvatwe 1s taken m the approxlmatlon When H(k) IS plotted vs In k [ = ln(2/t)] it gives the dmtrlbutlon of rate constants about a peak value of In k with an area under the peak proportional to the concentration of complexes, since the constraint applies

000’ 3 ’ 0 50 100 150 200

TIME (seconds)

Fig 2 Kmetlc results for the reactIon of excess of chromophore with mckel(II) The metal was pre-eqmhbrated at pH 6 4 at several ratios of fulvlc acid to Nl(II) Background X components have been removed (Results of Lavlgne et al [lOI )

that the Integral over the entire range of H(k) adds to the sum of all C(0, 1)s

Typlcal absorbance versus time results obtamed under pseudo-first-order condltlons are shown m Fig 2 The data refer to mckel(I1) bound to an organic polymer (FA) at different hgand-to-metal ratios Although the curves appear rather featureless, they consist of at least four kmetlcally distinguishable components Re- sults for the Iaplace analysis of the curve at a ratio of 1 1 are shown m Fig 3 This system wdl be fully described below

4

IT 3

0 - 2

3 y 1

OL \ I 2 3 4

In(t) 5

6 7

hg 3 Laplace spectrum of the 1 1 fulwc acid-Nl(II) curve shown m Fig 2 The three slowest components are seen Data corresponchng to the fastest component were truncated prior to Laplace analysis Concentrations are consistent with umts of absorbance change Rate constants are m s-l (Results of Lavlgne et al [lo] )

KINETIC STUDIES OF METAL ION SPECIATION

This Laplace transform approach appears at first sight to provide a solution to the whole problem The Integrated H(k) and peak values of ln(2/t) provide concentrations and rate constants However, the numerlcal dlfferentlatlon re- qmred by Eqn 4 makes the treatment extremely sensitive to artifacts There are a number of ex- amples m the literature which show the dangers In consequence, it IS wise to go beyond the Laplace transform treatment before acceptmg a model of the kmetlc data There IS a further reason for such caution

The second-order approxlmatlon to the La- place transform Introduces broadenmg to a peak m the H(k) vs ln(2/t) Thus, two cases cannot be directly dlstmgulshed In one case, a set of well defined and dlstmct components ~11 gwe broad peaks because of mathematical approxlmatlon In the other case, a dlstnbutlon of dlfferen- tlated sites groups about an average value to @ve broad peaks m the In k spectrum It IS not a sunple matter to dlstmgulsh the two cases Below, a method to address this complex issue 1s out- lined At present, the unmedlate pomt IS that the Laplace transform is used to obtam mitral ObJec- tlve estimates of the number of components and the rate constants which can be used as mput parameters to a non-lmear regresslon (NLR) fitting of Eqn 1 The first condltlon which must be satisfied 1s that the regresslons converge and that the output parameters are sensibly related to the mput parameters

Cabamss [9] evaluated transform and statlstlcal methods for the analysis of kmetlc data based on Eqn 1 These Included continuous dlstrlbutlon (Laplace transform), non-lmear regresslon of k, and C, and lmear regresslon on either k, or C, He concluded that methods based on regression of both parameters were the best suited of these for kmetlc modellmg

Mak and Langford [6] used artlficlal data to evaluate the effectiveness of the NLR method m resolvmg kmetlc parameters m two- and three- component systems The Laplace transform method has also been tested m detail with synthetic data plus added noise [lo]

The approach Just sketched describes the

transform techmques which have been explolted

187

so far m this area Other related mathematical techmques may well merit evaluation (see, e g ,

[ill ad WI) A serious question must now be faced If a

polymeric colloidal complexant murture may have a dlstnbutlon of bmdmg sites, how may we be assured that flttmg of a small number of kmetlc components has chemical agmficance’ Are the rate constants and concentrations which emerge from the combmatlon of the Laplace transform and non&near regression more than a way of archlvmg the concentration time function? There 1s a direct way to answer this key questlon A chemically slgmflcant partmon mto a few kmetl- tally defined components ~111 have two character- lstlcs First, the rate constants ~111 be fairly stable and only the concentration of components ~11 vary slgmflcantly as pH and concentration ratios vary Second, the changes m concentrations of components will follow trends required by mass actlon law conslderatlons As ul7111 be seen m the cases dlscussed below, these two tests are com- monly satlsfled The reported partltlonmg mto two to four kmetlc components does appear to reflect defmlte features of the chemistry of the systems Thus does not mean that the two to four rate constants given refer to specific molecular speaes There may still be some dlverslty of the mdlvldual sites, but they group mto recogmzable sets which may be adequately approximated by a small set of stable rate constants This behavlour 1s found for Cu(I1) and Nl(II) complexes of humlcs, Cu(I1) complexes of hydrous ferric oxide a and for the Fe010 humlc and hydrous oxide (box) species That such a simple speclatlon IS by no means required 1s shown by data for Al(II1) hydrous oxides and humlcs Here the rate constants are not stable and the kmetlc analysis recommended here shows that there is a very complex species dlstrlbutlon for which no small set of rate constants provides chemical Insight

The term “hydrous ferric oxide” IS chosen to Imply the matenal which IS formed on polymerlzatlon of Fe(II1) m aerated systems It is not designated “hydrous lron(III) oxide” because there IS evidence for the presence of some Fe(H) m several cmxmstances To the best of our knowl- edge, there IS no satisfactory nomenclature

188 C H LANGFORD AND D W GUlZMAN

One more prehmmary problem deserves com- ment The test of slgmflcance just proposed re- qmres that a study be conducted over a range of pH and total concentration Such a range can be studied m laboratory modelhng, but 1s not possl- ble if analysis of a specific natural sample 1s desired This problem has recently been ad- dressed by SOJO and De Haan [13] In order to validate a study of n-on speclatlon m a lake, they extracted the humlcs from soils surrounding the lake and prepared models of the humlc-hydrous oxide system over pH and concentration ranges Rate constants obtained from and validated by the modellmg effort could be associated with the rate constants found for the components m the lake samples

STUDIES OF SIMPLER SYSTEMS

Nickel (II) -jidv~ acid The Nl(II)-fuhc acid (FA) system [lo] 1s con-

sidered first The reason 1s that the lablhty of Nl(I1) complexes 1s at a favorable intermediate value This allows the most complete analysis Nlckel(I1) complexes have been important paradigm systems m the study of complex formation kmetics in general

The hgand system was a well characterized FA [141 extracted from a Bh horizon sol1 from Ar- madale, Prince Edward Island This organic polyelectrolyte has 7 7 meq g-l carboxylate groups, of which 3 3 meq g-’ are sahcylate like Total bldentate chelating sites number at least 5 5 meq g -1

The nickel concentration was held constant (1 X lOA M) and kinetics were measured at FA Nl ratios between 1 1 and 9 1 The solution pH was adjusted to 4 0, 5 0 or 6 4 followed by an equlhbratlon period of at least 24 h at room temperature m the dark The low pH is slightly above the unadjusted pH of the FA solutions while 6 4 is Just beyond the pK, values of the FA carboxylates

The colorlmetnc reagent solution included the chromophore 4-(2-pyrldylazo)resorcmol (PAR), NaNO, for ionic strength (Z, mol 1-l) control and NaHCO, for pH buffermg After munng equal

volumes of sample and reagent solution, the mixture had Z = 0 125 and pH = 7 8 A general-pur- pose least-squares polynomial routine was applied to the data Conversion to In A and In t optunlzed the data for polynomial fitting The choice of the lowest degree polynomial which would faithfully follow the curves prevented re- production of noise or the creation of u-regular artifacts m the Laplace analysis A smoothed discrete data set was generated with equal spac- mgs of In t The latter procedure was an essential prerequisite to successful Laplace calculations

Laplace transform analysis of kmetlc curves revealed the presence of four kmetlcally dlstmgulshable components The two fastest components have rate constants separated by only a factor of about 4, making them difficult to dlstmgulsh This comphcatlon led to a standard procedure as follows

Step I Nl(II) has sufflclently low lability that reaction of the “free” hexaaqua species 1s readily measured Nl(OH,)g+ (component 1) 1s certainly the most labile of the Nl(II) species present Kmetlc analysis m the absence of FA yielded a rate constant of 0 62 s-l This rate constant for Ni(aq ) indicates that almost 99% of this component has reacted within the first 7 s and its contrlbutlon to absorbance changes beyond this time are negligible The second fastest component had a rate constant of about 0 14 s-l Therefore, about 40% of this component reacts at times longer than 7 s For these reasons, the first 7 s of data were truncated and handled separately As the time dependence of hexa- aquamckel may be determmed, it does not con- tribute to the time independent X component (Eqn 1) This 1s “background” due entirely to absorbance by FA and PAR The value of X was estimated by fitting the first few data points to a low-degree polynomial and extrapolating to the absorbance at time zero These results were ven- fled by comparison with measurements of blank solutions

Step 2 At sufficiently long tnnes, only the slowest component contributes to absorbance changes (component 4) The linear section m a plot of l&A, -A,) versus time reliably estimates k, and C, from the slope and intercept Based on


TABLE 1

Mean and standard devlatlons of rate constants determmed by Lavlgne et al [lo] for mckel(II) bmdmg to a fulvlc acid at several pH values

pH Mean k, (SC’) k, (s-l) k, (s-l) k, (s-l) and S D

4 Mean 0 61 0 1.53 SD 0 17 0 036

5 Mean 0 70 0 139 SD 0 13 0 031

6 4 Mean 0 533 0 132 SD 0096 0 026

Pooled results n= 40 39 Mean 0 67 0 147 SD 0 16 0 038

00208 0 0021 0 0080 00009 0 0203 0 0032 0 0078 0 001.5 0 0197 0 0018 0 0064 00009

40 24 0 0205 0 0026 0 0082 0 0010

a Number of kmetlc runs for which the rate constant was resolved

these estimates, component 4 1s numencally stnpped from the data The resultmg absorbance-time data contam only contnbutlons from components 2 and 3 whose parameters are estimated by Laplace analysis

Step 3 The estimates and raw data (saved prior to the strlppmg of component 4 and smoothed to equal spacmg of time) are passed through a non-hnear regresslon routme written expressly for sums of exponentlals with posmve values of parameters NLR results for components 2, 3 and 4 are sequentially stripped from the data at less than 7 s Component 1 parameters are determined by NLR and k, verified for agreement with that measured m the absence of FA

Table 1 shows rate constants obtamed at the three different pH values used The standard devlatrons are somewhat larger than those obtamed using simulated data with white noise added shghtly m excess of that estimated to apply to the experimental data This suggests that m the Ni-FA system the kmetlcally Identified components are not discrete Whereas component 1 1s the “free” nickel, each of the other three ks probably represent the mean of a dlstrlbutlon of slmlar kmetlc behavlour The fact that observed rate constants do not vary systematically with eqmhbratlon pH, FA concentration or FA Ni ratio is Important

60

50

40

30

20

10

” i

“0 1 2 3 4 5 6 7 8 9 10

FA NI RATIO Fig 4 Changes m relative contrlbutlon of mmal components CC,) as a functlon of hgand-to-metal ratlo followmg equlhbratlon at pH 4 0 o, C,, 0, C,, v, C,, o, C4 (Results of Lavlgne et al [lo])

It 1s Important to reahze that the four components found represent the mmlmum number needed to model the Nt(II)-FA eqmhbnum Component dlstrlbutlons as a function of FA Nl ratio are shown m Figs 4 and 5 for equlhbratlon pH values of 4 0 and 6 4, respectively Results srmdar to those at pH 4 0 were found at pH 5 0 It may be seen that the decreased proton compe- tltlon at pH 6 4 results m a reduction of the aqua-Nl(II) and an mcrease m the relative contnbutlons of the bound species Further, varlatlon of component contrlbutlons with FA Ni ratlo follow expected mass actlon laws It must be noted that whereas complete N101) recovery was obtained at pH 4 0 and 5 0, only 60% was recovered wlthm 24 h at pH 6 4 The remammg 40% required up to 10 days to be completely recov-

70 1

k 60 1

0 1 2 3 4 5 6 7 6 9 10

FA NI RATIO Frg 5 Imtlal component contnbutlons m the NI(II)-FA system after equtibratlon at pH 64 Symbols as III Fig 4 Note that the sum of components IS only about 60%, showmg the presence of a very slowly labde component (Recalculated from Fig 6 of Lawgne et al [lo] )

190

ered, correspondmg to a rate constant of the order 10e5 s- ’ Even though this very slowly labile component became apparent at pH 6 4, the rate parameters ldentlfymg the other three bound species were unaffected Only theu relative amounts were reduced

More recently, Cabamss [91 studied speclatlon of NGI) complexes of water-derrved FA from an m part different perspective He observed vana- tlons m component k, and C, parameters with changes m kinetic reaction condltlons Ionic strength, pH and FA N1 ratios were adJusted followed by the equlhbratlon period Kmetlcs were determined by the mtroductlon of only the chromophore, PAR It was observed that dlssoa- atlon rate constants mcreased with increasing I, decreasing pH and decreasing FA Nl ratio The effects of pH and I were marked

Rate constants of the more labile species showed greater sensltlvlty to pH change Ionic strength effects were comparable m magnitude to those of pH As expected, Increased lomc strength caused C, to increase The rate constant of the kmetically slowest component was relatively m- sensitive to a change m I from 0 002 to 0 100 whereas k, for the fastest of the bound components Increased fivefold These extremes suggest a difference m the relative importance of electro- static contrlbutlon to the bmdmg, although the author did not consider the effect of lomc strength on aggregation of the fulvlc acid

An interesting observation of envu-onmental Importance is that at the FA Nl ratios found m nature, metal bmdmg 1s expected to be predoml- nantly to the least labile sites As these sites are least sensitive to pH and ionic strength changes, a “buffermg” effect exists It was pomted out, however, that measurements are needed at envlron- mentally significant metal and collold concentra- tlons to confirm this effect Of course, the mter- pretatlon of these experiments IS uncertain because the analysis depends on speclflc mechams- tic assumptions which may not be vahd

Copper@)-humlc acid A study of the bmdmg of copper to humlc acid

(HA) has recently been completed by Bomfazl et al 1151 The HA was extracted from a podzol sod

C H LANGFORD AND D W CZUTZMAN

of the Laurentian Forest Preserve of Lava1 Um- verslty and was characterued by Wang and co- workers [16,17] Samples were prepared at either pH 6 0 or 7 0 and had HA Cu ratios between 2 and 5 The copper concentration was 8 3 x lo-’ M Samples were equlhbrated m the dark for 24 h prior to kmetlc analysis Reaction was with a chromophore specific to copper present at a 30- fold equivalent excess The combined sample-reagent mixture had a pH buffered at 6 0, an ionic strength of 0 100 and was maintained at 24 5 o C

The water exchange rate of Cu(OH&+ (1O84 s-‘) 1s more than four orders of magmtude greater than that of hexaaquaruckel(I1) [18] Free Cu(II) and labile components react on time scales maccesslble by even stop-flow techniques As a result, these components appear as part of the X time-independent component and must be determmed by blank subtraction In this system over 80% of the total copper appeared as this fast “component 1” Laplace/NLR analysis revealed the presence of two less labile components havmg rate constants of 0 093 + 0 013 s-l (component 2) and 0 0078 f 0 0008 s ~ ’ (component 3)

Figure 6 shows component wntrlbutlons as functions of pH and HA Cu ratio A decrease m component 1 and an increase m components 2 and 3 are observed with mcreasmg HA Cu ratlo

90

85 r

‘i w 5 80 : 75

5 15.

8 x lo

5

I

OO- 23456,

HA Cu RATIO

Fig 6 Relatwe mitml contrlbutmns of components m the Cu(ID-hurmc acid system 0, X component, 0, component 2, v, component 3 Note that the malonty of WII) appears m the lnghly latnle X component Results are for solutions eqmhbrated at pH 6 0 (open symbols) and at pH 7 0 (sold symbols)


As expected, there IS httle difference between results obtained at eqmhbrmm pH 6 0 or 7 0 These pH values are m a relatively “qmet” region between the pK, of carboxyl groups and that of the phenols (pK, near 9) Copper speciation could not be studied at higher pH owing to the preclpltatlon of Cu(OH), Much below pH 6 too little copper IS bound by the humlc acid m smt- ably non-labile forms

Again, the stability of the rate constants and the rational behavlour of the component contri- butions indicate that a chemically significant model of speclatlon 1s obtamed As m the case of NI-FA, the components should probably be considered as averages of a range of related sites of slmdar labdlty

Typically, simple complexes of copper have stablhty constants of the order K = lo3 If formation rate constants he near lo* 1 mol-’ s-l, pseudo-first-order dlssoclatlon rate constants are about lo5 s-l It is unportant to note that the lablhtles of some hunuc-bound copper species were measured to be many orders of magnitude below such values In fact, the range of lablhtles from simple complexes to those with complex polyelectrolyte hgands spans more than eight orders of magnitude1

Shuman et al [7] applied Laplace analysis to the kinetic speclatlon of copper Their hgand consisted of an organic material extracted from estuarme waters on the basis of collold size usmg dlaflltration/ultrafdtration Reaction with the chromophore PAR was carried out at pH 7 5 and ionic strength 0 100 using stop-flow mwng For 7 1 X 10T6 M copper equlhbrated with 21 mg 1-l dissolved organic carbon, three components were identified Measurable rate constants ranged from 10-l * to lOI * s-l Some mstablhty of rate constants with changes m total copper concentration was seen at the two extremes Again, over 75% of the copper was too labile to be measured Al- though a different organic fraction was used, these values are comparable to those found by Bomfazl et al [15] for fulv~ acid-associated copper

Shuman et al [7] also made an eqmhbrmm study Potentiometnc titration with copper mdl- cated stab&ties distributed over three regions

with centres at log K = 4 5, 5 3 and 6 2 These values are close to those expected for simple complexes Slmllarly narrow ranges of stability constants for Cu(II)-FA complexes were determined by Turner et al [191 and summarized by Buffle [20] If the range of 3-4 orders of magmtude of stab&y constants encompasses both slm- ple species and polymer-bound complexes, it 1s surprlsmgly narrow m comparison with that of the lablhtles Slmllar “ranges” might be expected

The discrepancy between ranges of complex stability and lab&y suggests the consideration of additional factors Humlc substances are large porous gels Dlffuslon through the gel 1s a slow process which can be kmetlcally limiting Dlffu- slon has a comparable effect on both rates of association and dlssoclatlon Olson and Shuman [21] studied three operationally defined sue ranges of organic matter extracted from the river estuary They observed a shift toward less labile species with increasing size fraction but offered no explanation of its ongms Cabamss [9] started with FA extracted from two different rivers Al- though the materials differed m carbon content (53 8% vs 44 5%), the two displayed very snmlar eqmhbnum bmdmg of copper Cabamss dlscon- tmued the use of the higher carbon sample when It was noticed that rates of Ni(II) dlssoclatlon from it were faster A difference m particle size was postulated to be the cause

Copper@) -hydrous fernc oxide To study the effects of mass transport Imnta-

tions, Gutzman and Langford [22] investigated C&I) bmdmg to hydrous ferric oxide Studies of this type of colloidal adsorbate have indicated specific surface areas greater than 700 m2 g-l [23] Its monofunctlonahty and highly porous structure make It a good candidate for the study of dlffuslon limited processes

Collolds were prepared by base hydrolysis of FeUII) m the presence of Cu(I1) A variety of iron and copper concentrations were used, corre- sponding to Fe Cu molar ratios from 2 1 to 35 1 Collolds were equlhbrated for about 24 h at pH values rangmg from 5 5 to 7 0

Snnple equdlbrlum analysis of the Cu(II)-con- taming collolds involved filtration followed by

192 C H LANGFORD AND D W GUTZMAN

copper determmatlon by atomic absorption spec- trometry (AAS) Analysis of the filtrate for iron showed complete retention of the collold by the membrane Loss of Cu(II) on the falter was also less than the detection lout An example of a binding curve 1s shown m Fig 7 Fitting of data to standard isotherms shows that bmdmg closely obeys the Langmulr formalism Thus goodness of fit (r > 0 996) does not Justify fitting the data to more complex adsorption models The Langmulr equation 1s based on the assumption of unrform adsorption sites with no mteractlon between them and adherence to this model fatls to demonstrate any heterogeneity of binding sites

Kmetlc study of speclatlon of copper was carried out at 25 0 ‘C with a pH of 5 8 and lomc strength 0 100 controlled by the chromophore solution The color-forming reagent was the same as that used 111 the Cu0IM3A study Data were analyzed using the Laplace transform method for estimates and non-linear regression analysts for refinement of parameters Kmetic analysis revealed the presence of at least four components These are shown for pH 6 0 as a function of Fe Cu ratio m Fig 8 The major species at low ratio 1s the X component This 1s expected to consist mostly of “free” copper (largely CuOH+ under these pH condltlons), although copper bound to weak surface sites of the collold 1s probably also very labile This component(s) becomes less significant at larger Fe Cu ratios

TOTAL Cu2+ (1 o-6 M)

Fig 7 Example of Fe-hox bmdmg of Cu(I1) determmed at pH 68 Rttmg of these data to conventional Isotherms shows good adherence to LangmuIr theory (Results of Gutzman and Langford 1221)

0 5 10 15 20 25 30 35 40

Fe Cu RATIO Fig 8 Results of applymg kmetkc speclatlon analysis to the equlhbrmm bmdmg of Cu(II) by Fe-hox at pH 6 0 0, X component 1, 0, component 2, 0, component 3, A, not recovered A mmlmum of four kmetically dlstmgmshable components are necessary to describe the system

Two components which reacted on conslderably longer time scales were ldentlfled kmetlcally These had speclflc rates of 0 029 f 0 008 and 0 0012 f 0 0002 s-l As seen m Fig 8, they account for very similar proportions of bound copper and appear to reach a limit at high ratios The final component 1s one necessitated by mass balance conslderatlons It was not recovered wlthm the tnne scale of the analysis (typlcally 1 h) This component could be bound at strong sites m the hydrous oxide The permeable structure and hnuted functlonahty of the gel make this unlikely A stronger posslbdlty is that the unre- covered copper has become Incorporated in the Fe-hox “lattice” Such a phenomenon has been noted previously 1241

Of greatest Importance 1s the difference m the conclusions which might be drawn from the ther- modynamlc and kmetlc data Equlhbruun results m this system mdrcated the presence of equlva- lent, independent bmdmg sites Kmetlc evidence supports a mmlmum of three site types How can these kinetically dlstmgulshable sites have slmdar bmdmg constants? If rates of binding and dlssoci- atlon are equally reduced by dlffuslon hmltatlons, then the ratio of the two (stablhty constant) remains unchanged If mass transport 1s rate hmltmg for one of the species then it could appear kmetlcally less labile but thermodynamically equivalent to the other species For slow reactlons m natural systems this “lunetlc heterogene-


I@ may be an important factor affecting metal bloavadablhty

STUDIES OF MORE COMPLEX SYSTEMS

Systems mvolvmg colloidal metal oxides mter- acting with colloidal orgamc matter are more complex The metal ions Al(W), Fe(III), Cr(II1) and M&V) have Inherently lower lablhty and can exist as colloidal hydrous oxides m the pH range 4-9 Recovery of the former two metals from collolds has been studied Iron(III) was ml- tlally chosen because simple complex formation rates for Fe(OH,>z’ and Fe(OH,),OH’+ are convenient for stopped-flow analysis

Hydrous femc oxide The kmetlcally simplest of the systems to be

discussed 1s the acid hydrolysis of hydrous ferric oxide [251 Acldlc solutions of Fe(III) were adjusted to pH 4 00 with sodium acetate Collolds formed by this base hydrolysis were eqmhbrated for 24 h The total iron concentration of the eqmhbrmm solutions was 1 X 1O-4 M The reactlon was initiated by the addition of a solution of sulfosahcyhc acid (SSA) m perchlorlc acid Both conventional and stopped-flow techniques were used to momtor the Fe-SSA complex formation followmg acid hydrolysis at pH 1 Components were ldentlfled by the Guggenheim method and fitted by non-linear regresslon using the Mar- quardt algorithm

Three kmetic components were observed The fastest had a rate constant near 0 89 s-l, Identify- ing it as monomeric Fe(III) [26] A rate constant of 7 0 x 1O-4 s-l was found for component 2 This value is very close to that of 10 x 10d4 SKI determined by Sommer et al 1271 with the same agmg period for a collold prepared at higher n-on concentration and having about 90 non atoms per polymer The amount of the third and slowest component correlated with that amount of iron which could be removed from solution by a 0 45- pm filter These larger particles may represent maplent crystalline goethlte formation [28] How-

ever, no particle settling occurred even after several months

Hydrous femc oxrde-fihrc aczd The same procedure was applied to hydrous

ferric oxide equthbrated with Armadale fulvlc acid Two amounts of FA were used, correspondmg to molar ratios of FA-phenohc sites to iron of 1 1 and 2 1 Rate constants extracted were k, = 0 13 s-l and k, = 5 7 x 10e3 s-l at a ratlo of 1 1 and k,=O26 s-l and k,=64X10e3 s-’ at a ratio of 2 1 The sums of the component quantl- ties account for 72% and 61% of the total iron present

The values of k, are very close to that of 7 X 10m3 s-l estimated from the rate of eqmh- bration of Fe(III) with strong complexmg sites of this fulvlc acid at pH 165 [26] The rate constants of the more labile components may vary slightly with FA Fe ratlo They are, however, clearly dlstmgulshable from the well estabhshed constant for reaction of hexaaqualron(III) with SSA The most strdcmg fact 1s that reactions m the presence of FA are faster than acid hydrolysis of Fe-hox polymers Thus 1s not simple solublhzatlon of Fe(III) by an organic complexmg agent The particle size 1s larger m the presence of FA as will be described below

The mcomplete recovery of non IS not due to the presence of an unreactive Fe(II1) component The “component” was found to be available to the Fe(II)-specific chromophore l,lO-phenanthro- hne (analog of the nickel-fulvlc acid system)

The evolution of the experiments led to a new approach m the analytical procedure Hydroxyl- amme hydrochloride was introduced to the acldlc chromophore solution as a reducing agent allow- ing detection by the sensltlve reagent for Fe(II) ferrozme 1291 Kmetlc studies were performed at pH 2 and 25” C on samples which had been equilibrated at pH 6 The reactions were pseudo-first order m both ferrozme and hydroxyl- amme hydrochloride

A 1 1 murture of Fe(U) and Fe(W) was analyzed kmetlcally to determine the dlscrlmmatmg ablhty of the chromophore solution Two rate constants were observed The first, a fairly rapid component, was attributed to the reactlon of free


Fe(I1) The second had a value of 4 3 ( f0 5) X 10e3 s-l and was attributed to reduction of hydrous ferric oxide

Analysis of a hydrous ferric oxide solution showed that 96% of the u-on was present in colloidal form The correspondmg rate constant of 4 8 (*O 2) x 10e3 s-’ is m good agreement with the value above The remammg 4% of non had a constant of 0 15 s-l for reduction of free Fe0111

A fulvlc acid iron mixture of 1 1 FA to Fe010 was analyzed A fast component (k > 5 s- ‘) ac- counted for 10% of Iron A second rate constant of 1 49 s-l was also extracted Note that both of these values are larger than those observed for reactlon of Fe-hox m the absence of FA This greater lab&y suggests that at least some of the non 1s present at eqmhbrmm m the + 2 oxldatlon state In this context, reaction of the Fe(III)-FA solution with ferrozme m the absence of hydrox- ylamme hydrochloride gave complete recovery with rate constants of 0 113 and 8 6 x 10m3 s-’ There is a neghglble recovery of Fe(I1) from hydrous ferric oxides prepared wlthout FA Al- though these rates are conslderably slower than m the presence of the reducmg agent, they show that fulvlc acid can reduce and lablhze Fe(III)

During these experunents, it was observed that more Iron from Fe-hox equilibrated m the presence of FA was retamed on micropore filters than from Fe-hox alone Raylelgh scattering at 90 ’ was measured as a function of the amount of Fe(IlI) added to solutions of constant FA concentration A large mcrease m scattering 1s seen m the region of a 1 1 ratio of FA phenol carboxylate sites to Fe(II1) Although the iron FA complexes are larger, the fulvate-bound n-on 1s more labile

Hydrous alummum (III) oxide and alumrnum fulvates

A major concern m the development of these methods 1s to unprove the sensltlvlty so as to permlt studies at concentrations near natural water levels A gain m sensttlvlty was achieved by use of fluorescence detectlon The fluorophore calcem blue (CB) was applied to kmetlc study of the hydrous oxides of alummum (Al-box) and

Al-hox with FA [30] For fluorescence a Qamag- netlc metal was needed

Hydrolysis of Al(II1) was carrred out in the presence of FA by slow addltlon of NaHCO, untd the desired pH was attamed The “equl- hbrlum period” was 24 h The lomc strength and pH of the kmetlc mixture were estabhshed with a 0 100 M acetate buffer and measurements taken at 19 5 ’ C The pH was 5 0 m kmetlc runs Higher acid&y, favourmg hydrolysis of the colloidal complexes, would have been detrnnental to the mtenslty of the fluorescence Reactions were routmely followed for 24 h Beyond this period, instrument mstablhty became slgmficant Detectlon of components present at levels as low as 5 x lo-’ M shows the potential for apphcatlon to natural waters

In the Introduction it was insisted that vana- tlons of concentrations and pH values should leave the rate constants fairly stable m order to engender confidence that the kmetlc speclatlon model had Identified meaningful components (either unique or a cluster of closely related species) This test was sattsfled for Nl(II), C&I) and Fe(III) systems It 1s well known that a large range of distinct polycations arise m the hydro- lytic polymerlzatlon of Al(II1) Hence the Al(II1) results nught be expected to be the exception that “proves the rule” and to show that m some instances the kinetic model is httle more than a numerical method for constructing a convement archive of the signal versus tnne data All of the curves obtained m this study were fitted well by one or two components Under any speclflc set of chemical condltlons, the rate constants were reproducible to wlthm 10% and the sum of component concentrations achieved mass balance to wlthm 5% The good quahty of the fits does not objectively Justify the use of more parameters

Rows A and F m Table 2 correspond to Al-hox analyzed m the absence of FA at two different concentrations of total alummum Both were equlhbrated at pH 5 The rate constant measured for 2 X low5 M alummum 1s slmllar to that determmed for reaction of unhydrolyzed Al(III) mth CB at pH 3 A lower rate constant 1s found for 2 X lop6 M aluminum This Implies that the species formed at lower total alummum concen-


tratlon 1s more extensively hydrolyzed This 1s supported by the larger amount of aluminum falling mto an unrecoverable fraction at lower concentration These unreactive oxides could not be lablhzed even by bollmg with excess of perchlorlc acid Some light 1s shed on this oddity by results reported by MatlJevlc and co-workers [31,32] They observed a maximum m turbidity curves occurring at low5 M Al010 Sol stability at lower concentrations was ascribed to adsorption of AKIII) polyhydroxo cations At higher concentration anion adsorption stabilizes the collold

The bulk of the data m Table 2 describe Al-hox associated with FA Fulvlc acid concentrations were varied over a fairly wide range as reflected by the FA Al ratios The range of 0 9-45 mg 1-l IS appropriate to fresh water modellmg Note that in all instances there IS a non-recoverable alummum fraction

It IS seen that at pH 5-6 the data are fitted well by two rate constants falhng mto fairly well defined regions These are referred to as k, and k, Study of the reaction of aluminum citrate v&h CB under the same condltlons [61 yielded a rate constant of 4 8 X 10e3 s-l At 25°C a value of 7 9 x 10e3 s-l was found for dlssoclatlon of alu-

TABLE 2

195

mmum sahcylate [33] The proxlmlty of these values for dlssoclatlon of snnple chelates to k, and k, IS mterestmg It may be inferred that these components represent dlstrlbutlons of sites of similar bmdmg strength on the fulvlc acid approximating simple AKIII) complexes

It 1s clear that m this system the number of chemically distinct “bmdmg sites” IS larger than the number of components required to model the kmetlc data This does not preclude the recovery of some practical information from application of this method Recognition of the subtle phenomenon of dependence of Al(II1) hydrolysis on the total alummum concentration 1s of slgmflcance to aquatic toxicology Aluminum avallabll- ity may have a complex dependence on both pH and concentration Having recognized the largely “mathematical” nature of the k and C values m the Al-hox-FA system, assignment of greater chemical significance to their values would be imprudent

Speclatlon of alummum(II1) has frequently been carried out by timed spectrophotomet- nc reaction with 8-hydroxyqumolme or with fer- ron (8-hydroxy-7-lodoqumohne-5-sulfomc acid) Hexaaqua- and mononuclear Al(II1) are assumed to react “mstantaneously” whereas polynuclear

Kmetlc results obtamed by Mak and Langford [30] for the hydrolysis of alummum(Il1) with fulvrc and present

[AlI FA Al pH of k,, “k,” “k,” “k,” Kmetlc recovery Eqmhbrmm total rat10 a sample (s-l) (lo-* s-l) (10m3 s-l) (10e4 s-‘) (1O-6 M)

Very slowly Not recovered b recovered (%*o) (o/o)

A 200 0 50 0 562 78 13 7 B 200 3 50 271 336 295 319 34 8 c 200 75 50 2 62 330 145 207 C-Z:, 66 7 D 200 3 60 2 3.5 I 98 182 298 85 43 5 E 200 75 60 264 354 127 186 20 66 7 F 200 0 50 0 0125 10 2 489 G 200 15 60 234 357 107 52 15 826 H 200 3 60 185 192 125 136 13 8 600 I 200 75 60 166 176 168 226 125 482 J 200 30 60 135 1.53 161 408 19 0 24 1 K 200 75 60 551 351 20 62 9 L 200 75 50 777 420 (-36) 615 M 200 3 80 16 6 309 118 199 34 649 N 200 30 80 18 7 203 95 70 2

a Based on ca 3 eq of phenol-carboxylate per mole of FA b Difference between recovery at 24 h and at 2 h

196

species react more slowly Apphcatlon of kinetic spenatlon methods sumlar to those described here has allowed more accurate determmatlon of various alummum components The evolution of these methods has been revrewed by Bertsch [34]

APPLICATION TO NATURAL SYSTEMS

Most studies based on kmetlc speclatlon have employed either synthesized coIlolda1 compounds or those chemically extracted from natural systems and purified Two mvestlgatlons have, however, extrapolated this technique to the study of iron speciation in its native aquatic matrix

Lake Esthwazte Water Tlppmg et al [351 included the method of

acldlficatlon and reaction with SSA m a study of non redox cyclmg m the water column of a natural system Flow centrlfugatlon was used to sepa- rate ion particles from Esthwalte Water (a small lake m the UK) Iron species larger than about 70 nm were collected as sednnent (fraction A) The supematant (fraction B) was generally concentrated by ultrafiltration (UF) with lo-nm nominal pore size filters prior to analysis Throughout most of the year iron IS about equally split between the two fractions The annual overturn leads to a large increase m particulate non

C H LANGFORD AND D W GUTZMAN

For comparison of kinetic results, two types of standards were employed A synthetic hydrous oxide was prepared by base hydrolysis of an lron(II1) salt and aged for several days Samphng from the ox~/anox~c boundary during the early period of thermal stratlficatlon allowed the collection of larger amounts of particulate non which were also used as “standards” Samples were collected from a well oxygenated zone 5 m below the surface m the months prior to the start of overturn Absorbance changes m kmetlc runs cor- responded to within 5% of total iron as determined by the ferrozme method Ferrozme tests revealed no Fe011 m the samples, although po- larography indicated 5-10% Fe011 following ultrafiltration

As shown m Table 3, both of the standards yielded single rate constants The value for the synthetic hydrous ferric oxide is similar to both those of Sommer et al [271 (5 5 x 10e4 s-l) determined after 3 days of collold aging and by Langford et al [25] (7 X 10e4 s-l) after aging for 24 h A single component 1s also observed for material collected from the oxlc/anoxlc boundary Values snmlar to this are seen as the slower component (k,) tn samples collected from surface waters These rate coefficients are slgmfi- cantly smaller than that determined for the syn- thetlc collold Tlppmg et al [35] ascribed this difference to the presence of organic matter ad- sorbed to the acid sensitive surface Fe-OH or

TABLE 3

Kmetlc speaahon of Iron m Esthwalte Water, UK, performed by Tlppmg et al [351

Component “k 1” “k” Relatwe dlstnbutlon of reactive species (lo-3 SC’) (lo-4 SKI)

Cl (%o) c, (%I

Synthetic Fe-hox 61 100 Fraction A from OXIC/

anoxLc boundluy 13 100 Samples from oxygenated zone (5 m depth)

Fraction A _a 17 <5 > 95 Fraction B

Unconcentrated b 7 2 30 70 UF concentrated 51 25 52 48 2y fraction B c 22 18 52 48

’ Small contrlbutlon of this component prevented estnnatlon of a rate constant b Values were estrmated owmg to the very small total absorbance change observed c Supematant from further centnfugatron of the “UF concentrated” fraction


Fe-OH+ groups Sommer et al [271 showed that the decomposltlon rates of Fe-hox decrease with aging of the polymer as oxolatlon proceeds This decrease was drastic during the first 2 days under their chemical conditions but became more urn- form after a few days The effect is highly temperature dependent Also bearing on polymer “age” are the effects of pH and total iron concentration

A second, faster component 1s observed for surface water samples havmg k, averagmg about 5 x 10e3 s-l This component appears mostly in fraction B These rate coefficients are also very close to those near 6 X 10e3 s-* observed by Langford and co-workers [25,26] for non associated with fulvlc acid Tipping et al could not confirm the presence of such Iron-organic complexes Ultrafdtratlon of their samples showed that Just over half of the humlc matenal present was able to pass through the filter but only about 10% of the non appeared m the filtrate Tipping et al assigned the kmetlcally faster component to freshly formed Fe-hox which aged little between collection and kmetlc analysis

Iron occurs m Esthwalte Water at about 3 X

10m5 M The preconcentratlon techmques re- qulred m this study lmut a clearer understandmg of the chemical processes occurring The posslbll- sty of equlhbrmm shifts durmg such manipulation cannot be ignored Clearly the study of such natural systems would benefit from greater analytlcal sensitivity

197

Lake ljeukemeer Recently SOJO and De Haan 1131 studied iron

speclation m Lake TJeukemeer, Netherlands Previous mvestlgatlon using ultrafdtratlon had unpiled that about 10% of fulvlc acid 1s involved m Fe-FA complexes A sensitive Fe(II)-specific chromophore was used m conJunction with hydra- zme hydrochloride as reducing agent Kmetics were measured at about pH 4 5 using an acetate buffer Data analysis involved Guggenheim estl- matlon of input parameters followed by non-lmear regression Equlhbnum pH values of some of the standards and samples are shown m Table 4

A key concern m studies of natural systems is how to obtain validation of a kmetic model by pH and concentration ratio variation SOJO and De Haan [13] solved this problem by cahbratlon m advance of study of the lake water Calibration was accomplished with laboratory models prepared from components extracted from the environment of interest Aquatic fulvlc acid was iso- lated from the lake and soil fulvic acid from its drainage basin

The so11 FA was found to contain 5 7 meq g-’ carboxyl sites This IS slightly less than the 7 meq g-’ determined for the Armadale FA Solutions of Fe010 or Fe(III)-FA were pH adjusted and equilibrated at 22’ C for 48 h m the dark No precipitation occurred Total iron was determined followmg acid hydrolysis and corre- sponded to mfmity absorbance of the kinetic curves in all instances

TABLE 4

Summary of results obtamed by SOJO and De Haan [13] for speclatlon of Iron in Lake Tjeukemeer, Netherlands

Sample Equhbratlon “kl” “kz” Relative dstnbutlon PH (lo-3 s-l) (lo-4 s-11

x (o/o) c, (%I c, (%)

Laboratory solutions Fe3+ 300 3 74 1 25 9

6 23 2 26 0 74 0 Fe3+-WFA a 6 23 33 8 33 81 1 15 6 Fe’+-SFA b 3 00 14 10 794 17 9 28

602 37 7 13 2 66 5 20 2 Lake water

Natural 64 16 5 62 38 Acidtiled 3 13 2 44 56

’ Water-denved fuhc acid b So&denved fulvlc aad (dramage basin sediment)


Analysis of iron m the absence of FA yielded two components Samples equilibrated at pH 3 showed the majority of iron to be present in the X component It should be pointed out that kl- netlc measurement was only started 15 s after mlxlng with the chromophore solution Any component with a rate constant greater than about 0 1 s- ’ would be expected to appear in the time-independent term When the eqmhbratlon pH was raised to 6 23, most of the iron appeared as time dependent components It 1s therefore hkely that a large part of the X component corresponds to hexaaqua-Fe(III), for which Lang- ford et al [25] measured a dlssoclatlon rate coefficient of 0 89 s-l The measurable rate constants m these solutions are about half those determined by Langford et al [251, Sommer et al [271 and Tlppmg et al [35] SOJO and De Haan attributed this devlatlon to possible differences m experimental condltlons such as pH and temperature The slower component correspondmg to 450~nm filterable non particles observed by Lang- ford et al was not seen m this instance Flltratlon of the collards showed that httle of the material was larger than 200 nm

In the presence of either sod- or water-derived fulvlc acid a thud species becomes apparent The dlssoclatlon rate constant of this component (k,) is intermediate between fast monomeric Fe(III) and its hydrous oxide This component 1s most reasonably asslgned to simple iron-fulvate complexes The rate coefflclent of this complex 1s unstable to a degree with equlhbratlon pH, as was seen m the case of Fe(II1) associated with sol1 fulvlc acid The rate constants for component 2 (Fe-hox) m these systems are much more m line with those observed by other workers discussed previously

Average results for kmetlc analysis of lake water are also shown m Table 4 In Its natural state, the iron speclatlon may be modeled by two kmetlc components No “timemdependent” Fe010 IS detected Analysis of lake water which had been ultrafiltered showed the presence of an X component This indicates a change durmg the filtration Measured rate constants of component 1 are larger by an amount outside experunental error than those determmed for the model solu-

tions This could be a result of differences m composltlon between lake water and laboratory models Natural samples had about a ten tnnes larger ratio of dissolved organic carbon to iron Also, lake samples contam other metals which could compete for strong bmdmg sites on the FA, thereby mcreasmg the average lablhty of Fe(III) Acldlflcatlon of lake water resulted m the dlsap- pearance of the terms assoaated with Fe-FA complexes and the appearance of a large amount of labile species

It may be concluded that Iron m Lake Tjeuke- meer is largely dlstrlbuted between two kmetl- tally dlstmct species The magnitude of their dlssoclatlon rate constants and the manner m which their component contrlbutlons vary with eqmhbrmm chemical conditions indicate that they are most hkely Fe-hox and Fe(III) associated with humlc matter Rate constants for the former are reasonably well behaved with varying equlhbrmm condltlons The variation of rate coefflclents for non fulvate dlssoclatlon indicates an adsorbate with a more complex species dlstrlbutlon

PROBLEM OF SPECIES DETECTION

It has been pomted out above that a major limltatlon in applymg kmetlc speclation at em+ ronmentally slgmficant concentrations 1s msuffi- clent analytical resolution Indeed, this problem plagues all trace metal speclatlon methods Natu- ral concentrations of trace metals are often so low that extreme difficulty may be encountered even m determmmg total concentrations The present problem goes beyond one of detection hmlts to one of resolvmg component species at low concentrations Table 5 shows concentration ranges which have been reported for some metals m a selectlon of natural systems (summarized by Buffle [36]) Note that many of these important metals are typlcally present at nano- to picomolar concentrations

Some metal complexes are sufficiently stable to allow speclation based on separation followed by species non-specific analysis For this to be accomphshed the metals must be present at sufflclently high concentrations The use of precon-


centratlon has also been frequently reported The case often arises, however, where this represents “harsh” treatment of the sample, suggesting possible changes m trace metal species distribution Further, problems of adsorption or analytical re- strlctlons on solution condltlons can Influence speclatlon results Lund [l] recently discussed many of these cases and various combmatlons of analytlcal methods employed The vast major@ of speclatlon studies have been carried out under condltlons of thermodynamic equlhbrmm As hmted at m studies of the Cu(II)-HA system and clearly shown m Cu(II) bmdmg by Fe-hox, equlhbrmm studies may not always provide the most relevant mformatlon about the “free metal Ion”

In Fig 9 copper IS used to demonstrate some of these concerns Natural concentrations of Cu’+ m fresh waters and salt waters are shown on the left side of the scale At higher concentrations but still of great nnportance m envu-onmental studies are measures common to aqua& toxlcol- ogy such as the LC,, (lethal concentration for 50% of the population) This value varies conslderably with carbonate and pH but generally falls m the range shown, with perch being resistant and salmon sensltlve From the analytical perspective, anodlc strlppmg voltammetry (ASV) 1s one of the most sensltlve techmques and 1s to some extent species selective It suffers from problems of adsorptlon and the potential to shift eqmhbna Electrothermal AAS and mductlvely

TABLE 5

199

coupled plasma techniques have evolved mto very sensltlve methods but only provide mformatlon on total metal concentrations Ion-selective elec- trodes for Cuzf are reported to have detection hmits of lo-* M under only the most Ideal condi- tlons and are prone to Interferences from other ions and the chemical matrix Judlclous choice of chromophore reagents make spectrophotometrlc determmatlons metal specific With the copper- speclflc chromophore used m two of the studies above (~14000 1 mol-’ cm-‘) conventional absorbance measurements are of only moderate sensitlvlty (shown as ABS m Fig 9) Even at current sensltlvlty the method is apphcable to some toxlcologlcal studies, but Improved sensmv- ity 1s a priority

Laser thermal lensmg (LTL) combmes the se- lectlvlty of the spectrophotometrlc methods and the sensltlvlty of fluorescence A beam from a laser havmg a radial Gausslan mtenslty profile 1s focused into a cell contammg the absorbing analyte Non-radlatlve relaxation of the electromcally excited complex leads to heatmg of the solvent m the vlcmlty of the (pump) beam As the refractive mdex of the solvent 1s a function of density, local heating results m the formatlon of a transverse refractlve index gradlent This gradlent reflects the Gaussian character of the mcldent beam and IS, m effect, a thermal lens [41,421 A second beam tuned to a wavelength not absorbed by the analyte (probe) IS directed through the sample

Ranges reported for the total concentration of some morgamc elements m natural waters as summarized by Buffle [36] a

Element Log(concentratlon, MI of total metal

Fresh waters Soil pore waters b Pelagic sea water

Fe -68 -47 -58 -48 -100 -87 Al -65 -40 -83 -75 Mn -95 -56 -64 -44 -97 -85 Nl -95 -63 -70 -87 -78 Cu -85 -62 -68 -60 -92 -82 Zn -85 -58 -60 -52 -105 -80 Cd -100 -76 -90 -120 -90 Pb -95 -62 -114 -97 Cr -88 -69 -87 -82 Hg < -10.5 -78 -95 -117 -110

a Data from 1371 for fresh waters, [38 and 391 for sod pore waters and 1401 for sea water b In some Instances only smgle values were reported

200

TOXICOLOOY

ENVIRONMENTAL .,.

PPb

-I UI 8PmxATxow RL.

100 1

CABS PL

lo

CICB +PatSBllT LTL PL

(CLTL DL- DOC)

01

CAEV

ANALYTICAL

Fig 9 SchematIc representation of copper mncentratlons relevant to envlronmental studies and detectlon hrnits (D L ) of several analykal techmques The D L by conventlonal spectrophotometry IS shown wtth an arrow The bar above It denotes a range mnvement to analyze speclatlon For specla- bon studies, smular bars could be placed over anochc strlppmg voltammetry (ASV) and ron-selectwe electrode (ISE) The capaclty to study speclatlon requires that the least slgmfzant species be present at concentrations greater than the detectlon hmlt of the method See text for further explanation of terms

coaxlally to the pump beam After passmg through the sample, the pump beam is filtered and the Intensity at the centre of the probe IS monitored with a pinhole-photodiode combmatlon The change m peak Intensity 1s a sensitive indicator of absorbed energy, and hence of the analyte concentration This affords two advantages First, sensltlvrty may be increased by use of a higher pump power Second, as the slgnal 1s directly dependent on energy absorbed it 1s less sensitive to scattering by small particles m the sample This offers a significant advantage m terms of

C H LANGFORD AND D W GUTZMAN

sample treatment over normal spectrophotometry which measures energy transmltted (where absorbance and scattering both contnbute) Larger particles which may block a slgmflcant fraction of the focused laser beam present some problem but are more easdy removed

This apparatus has been successfully used m our laboratories for the determmatlon of Fe(H) (and indirectly SO,) [431, formaldehyde [441 and dissolved organic carbon [45] The current detection hmlt for Cu(II) based on absorbance of the copper-chromophore and derived from the data for dissolved organic carbon (DO0 slgnal-to- noise limits 1s indicated m Fig 9 LTL is currently being tested for the kinetic speclatlon of copper at environmentally significant concentrations

In order to place the approach used here m the context of speclatlon schemes useful m envlronmental chemistry, it 1s worth making some final observations on what the method does and does not accomplish Although the system is per- turbed during kinetic analysis, the species ldentlfled are those which exlsted at equlhbrmm Each species has an associated rate coefficient These rate coefficients are strictly a parameter of the experiment, not of the sample If the reaction mechanism 1s simple, it may be possible to achieve a semi-quantitative connection between the experimental rate constants and related rate processes m natural samples Where the mechanism 1s more complicated it 1s still probable that relative lablhtles are correctly diagnosed

The authors thank the Natural Sciences and Engineering Research Council of Canada and the Fonds Formation des Chercheurs et &de B la Recherche of Quebec for fmanclal support A number of co-workers have contributed to the developments described and then names will be found m the references Special mention IS made of Mark K S Mak and Allen Lavlgne who played a major role m the mathematical development and evolution of chemical tests of slgmflcance All the work mvolvmg humlcs owes a debt to the collaboration of Donald S Gamble, and Bhuvan Pant contributed to the preparation of the manuscript


REFERENCES

1 2

3

4

5

6

7

8

9 10

11 12 13

W Lund, Fresemus’ J Anal Chem ,337 (1990) 557 J Buffle, Complexation ReactIons m AquatIc Systems an Analytical Approach, Horwood, Cinchester, 1988, pp 81- 89 J Buffle, Complexatlon ReactIons m AquatIc Systems an AnalytIcal Approach, Horwood, Chshester, 1988, pp 195-200 D W Margerum, G R Cayley, DC Weatherbum and G K Pagenkopf, m AE Martell (Ed ), Coordmatlon Chemistry, Vol 2, ACS Monograph 174, Amencan Chenu- cal Soaety, Washmgton, DC, 1978, Chap 1, pp 174-190 B G W~lhs, W H Woodruff, JR Frysmger, D W Margerum and H L Pardue, Anal Chem , 42 (1970) 1350, and references cited therein MK S Mak and C H Langford, Inorg Chum Acta, 70 (1983) 237 M S Shuman, B J Colhns, P J Rtzgerald and D L Olson, m R F ChrIstman and E T C;Jessmg (Eds ), AquatIc and Terrestnal Hunnc Matenals, Ann Arbor Saence, Ann Arbor, MI, 1983, pp 349-370 D L Olson and MS Shuman, Anal Chem, 55 (1983) 1103 S E Cabamss, Envlron Sg Technol, 24 (1990) 583 J A Lavlgne, C H Langford and M K S Mak, Anal Chem ,59 (1987) 2616 S W Provencher, J Chem Phys , 64 (1976) 2772 S W Provencher, Comput Phys Commun , 27 (1982) 213 LE SOJO and H de Haan, Environ SC1 Technol, 25 (1991) 935

14 D S Gamble and M Schmtzer, m P C Smger (Ed 1, Trace Metals and Metal-Orgamc Interadlons m Natural Wa- ters, Ann Arbor Science, Ann Arbor, MI, 1973, pp 266- 273

15 M Bomfazl, B C Pant and C H Langford, m preparation 16 Z -D Wang, D S Gamble and C H Langford, Anal Chum

Acta, 232 (1990) 181 17 Z -D Wang, B C Pant and C H Langford, Anal Chim

Acta, 232 (1990) 43 18 C H Langford, m S Petruccl (Ed ), Iomc Interactlons,

Vol 2, Academic Press, New York. 1971. D 16 19 D R Turner, MS Varney, M !&&Id, RF C Man-

toura and J P Rdey, Geoclnm Cosmochlm Acta, 50 (1986) 289

20 J Buffle, m H &gel (Ed ), Cuculatlon of Metals m the Enwronment, Dekker, New York, 1984, pp 202-207

21 D L Olson and M S Shuman, Geochnn Cosmoclnm Acta, 49 (1985) 1371

22 D W Gutzman and CH Langford, Water Pollut Res J Can, 23 (1988) 379

201

23 J A Davis and J 0 Le.&e, J CoIlcud Interface .%I, 67 (1978) 90

24 K C Swallow, D N Hume and FM M Morel, Enwon SC] Technol , 14 (1980) 1326

25 C H Langford, R Kay, G W Quance and T R Khan, Anal Lett , 10 (1977) 1249

26 C H Langford and T R Khan. Can J Chem . 53 (1975)

27

28

29

30

31

32

33 34

35

36

37

38

39

40

41 42

43

44

45

2979 B A Sommer, D W Margerum, J Renner, P Saltman and T G Spiro, Blomorg Chem , 2 (1973) 295 C H Langford and M KS Mak, Comments Inorg Chem , 2 (1983) 127 C H Langford, S M Wong and A W Underdown, Can J Chem , 59 (1981) 181 M KS Mak and C H Langford, Can J Chem , 60 (1982) 2023 E Matgevs, KG Mathal, RH Ottewdl and M Kerker, J Phys Chem , 65 (1961) 826 E Matgevlc, G E Janaur and M Kerker, J CoIlold SC], 19 (1964) 333 F Secco and M Ventunm, Inorg Chem , 14 (1975) 1978 P M Bertsch, m S A Norton, S E Lmdberg and AL Page (Eds 1, Aclchc Pregpltatlon, Vol 4, Spnnger, New York, 1990, pp 66-70 E ‘Iippmg, C Woof and M Ohnstad, Hydrotuologla, 92 (1982) 383 J Buffle, Complexation ReactIons in Aquatic Systems an Analyhcal Approach, Horwood, ChIchester, 1988, pp 37- 40 H J M Bowen, Environmental ChermstIy of the Elements, Acadenuc, London, 1979 H J M Bowen, Trace Elements in Blochermstry, Aca- dermc, London, 1966 S V Mattigod, G Sposlto and AL Page, m Chennstry m the So11 Environment, Sod Science Society of America, Machson, WI, 1981, p 204 K W Bruland, m J P Riley and R Chester (Eds 1, Cheml- cal Oceanography, Vol 8, Academs, London, 1983, Chap 40 D S Khger, Act Chem Res , 13 (1980) 129 K Mlyrushl, T Imaska and N Ishlbashr, Anal Chem , 54 (1982) 2039 D K Sharma, F Villamagna and C H Langford, Can J Spectrosc ,28 (1983) 181 J A Alfhenn and C H Langford, Anal Chem , 57 (1985) 861 J F Power and C H Langford, Anal Chem , 60 (1988) 842

Kinetic studies of metal ion speciation

Documents