J. Chem. SOC., Faraday Trans. I , 1988, 84(1), 9-18 The Dissolution of Magnetite by Nitrilotriacetatoferrate(I1) Margarita del Valle Hidalgo and Nkstor E. Katz Catedra de Fisicoquimica III, Facultad de Bioquimica, Quimica y Farmacia, Universidad Nacional de Tucuman, Ayacucho 491, 4000 Tucuman, Argentina Albert0 J. G. Maroto and Miguel A. Blesa* Departamento Quimica de Reactores, Comisidn Nacional de Energia A tdmica, Avenida del Libertador 8250, 1429 Buenos Aires, Argentina The dissolution of magnetite particles in solutions containing nitrilo- triacetatoferrate(I1) has been studied as a function of total nitrilotriacetic acid (NTA) and iron(1r) concentrations, pH and temperature. Experimental results are interpreted in terms of adsorption equilibria involving the free ligand and its metal complexes, and inner- and outer-sphere interfacial electron transfer from the adsorbed electroactive FeI' species to surface -Fell1 sites, either free or complexed by NTA.Ion transfer-controlled dissolution was ruled out by the experimental evidence. Fey- ions Ty = N(CH,CO,)i-] have been identified as the electroactive species, electron transfer being an outer-sphere process for the present experimental conditions : the precursor complex can be represented by -Fe"'-NTA. . . Fey-. Dissolution rate deviates from first order in [Fey-]; the order decreases with increasing [Fey-] as a consequence of the relatively high affinity of the complex for -Fe"'-NTA surface sites. Reaction order on proton concentration is 0.67, reflecting the requirement of H+ ions adjacent to the site where the reductant is adsorbed.The apparent activation energy is 73 kJ mol-l, which is a composite of equilibrium and kinetic parameters. In spite of their technological importance, metal oxide dissolution processes are by no means well understood.' Different preparations of a given metal oxide respond in various ways to aqueous solvents; surface details and bulk defects play a vital role in determining their reactivity and it is not surprising that there is no adequate (even qualitative) description of the oxide parameters that influence solubility kinetics. Dissolution rates are of course also sensitive to solution parameters; even though in this case the basic ideas can in principle be tackled more easily, there is a noticeable scarcity of detailed studies.Typical solvents for iron oxides are aqueous acid solutions of chelating agents, such as polycarboxylic acids and amino acids.2 The effectiveness of such formulations is limited, and there is little evidence on the mechanism of iron leaching into these solutions. Chemisorption of the ligands is involved,l-1° but it is not clear whether the complexed surface -Fe"I ions are intermediates in the dissolution process, or 'dead-end' products that must return to a more reactive uncomplexed state prior to dissolution.lY8 It is even possible that phase transfer of FeIII is negligibly slow in acidic solutions, and that a redox mechanism is inv01ved.~~~ Reductants greatly enhance the rate of dissolution of iron(1n) oxides ; this phenomenon has been known for yearsl1-l4 and involves formation of the more labile iron@) ions by electron transfer across the interface,15-19 followed by iron(I1) ion transfer ; this effect may even be observed in the dissolution in mineral acids.l' Of particular interest is the reductive dissolution by metal complexes, such as V(pic), (pic = picolinate anion)15- l8 or Fe" in solutions containing polycarboxylates.In these latter systems, several concurrent chemical reactions take place simultaneously, and it is not easy to identify 910 Dissolution of Magnetite by Nit rilo tr iace ta toferra te( 11) unambiguously the reasons for the dependence of the dissolution rate on the concentrations of ligand, metal ion and hydrogen ion. The present paper reports the results of a kinetic study of the dissolution of magnetite in solutions containing nitrilotriacetic acid (NTA)* and iron(r1) salts.Together with the results obtained in oxalic acid16* 2o and EDTA (ethylenediaminetetra-acetate)'. 21 media, this information is used to propose a general mechanism for the dissolution of magnetite by iron(I1) ions in the presence of polycarboxylates, in which the relative importance of inner- and outer-sphere electron-transfer processes is taken into account. Experimental Magnetite (Fe,O,) was prepared as described in a previous paper22 by oxidising an iron(I1) salt in an alkaline medium in the presence of hydrazine. It was characterised by chemical analysis, X-ray diffraction, Mossbauer spectroscopy, scanning electron microscopy and surface area measurements.It was composed of cubo-octahedral particles of 0.3 pm average diameter. The oxide was crystalline and stoichiometric, and had a specific surface area of 4.7 m2 g-l. Kinetic experiments were performed in a magnetically stirred and tightly stoppered cylindrical beaker provided with a thermostatted water jacket. Solutions of nitrilo- triacetic acid of suitable pH and concentration, containing the desired amount of Fe'' in the form of (NH,),Fe(S0,),.6H20, were carefully deaerated by bubbling with purified nitrogen. Dissolution was started by pouring magnetite into the solution. In order to avoid oxidation of iron(II), periodic sampling was not employed; instead, the kinetics were followed by measuring the acid consumption necessary to keep the pH constant as a function of time.The pH-stat (Mettler DK-10) was calibrated with three standard buffer solutions. All reagents were analytical grade. Water was twice distilled in a quartz apparatus. Some experiments were carped out as far as complete dissolution. In these cases, the linearity of plots of 1 - (1 - f ) s against time, f being the fraction of the solid remaining at time t, showed a dependence of rate on residual surface area. In general, however, data were analysed in terms of initial rates; accordingly, unless stated otherwise, square brackets represent initial concentrations. Acid consumption data were transformed into rates R, expressed as the number of moles of iron dissolved per unit area and time, by means of eqn (1) R = (3N/nA)(AV/At), (1) and the stoichiometry given by eqn (2) Fe,O, + 2(Fe"HZYy)(3y-z-2)- + qH,Y(3-p)- + nH+ In eqn ( l ) , N is the molarity and AV the volume of monobasic titrant; A is the total available surface area (A/m2 = 4.7 x w, w being the mass in g of magnetite) and n is the number of protons as defined in eqn (2).Eqn (2) was solved for n through mass balance and charge equations, using stability constants23 to characterise the average a, b, p, x and y values. Solutions of eqn (2) were checked in selected experiments by measuring n through the simultaneous determination of dissolved iron and consumed acid. * In this paper, the following nomenclature is used for nitrilotriacetate-containing species. When the degree of protonation is well defined, Y represents N(CH3C02)3, e.g. Fey-.When the degree of protonation is not explicitly indicated, NTA is used, as in -Fe"'-NTA or [NTA]. Bonds prior to Fe indicate surface species.M. del V . Hidalgo et al. 11 Results and Discussion Dependence of Dissolution Rate on [NTA] and [Fe"] Fig. 1 shows a series of typical dissolution profiles, in the form of the volume of acid consumed against time; the initial dissolution rates dV/dt are obtained from them and transformed according to eqn ( 1). Generally, the profiles are monotonously deceleratory, the initial rate being the maximum rate. In certain cases an initial dead time was observed, which was never longer than 2 min; it was attributed to initial equilibration processes and eliminated in the calculation procedures. The variation of the initial dissolution rate as a function of the total concentration of nitrilotriacetic acid ([NTA]) at constant total concentration of iron(1r) ([Fe"]) is shown in fig.2. The dependence of the initial dissolution rate on [Fe"] at constant [NTA] is shown in The results in fig. 2 may provisionally be interpreted in terms of one or both of the fig. 3. following complexation equilibria K s [the actual charges involved may be different, see ref. (24), and surface species will be represented below as -Fe-NTA] OH + Y3- + -Fe"I-Y2- + OH- (3) -FeIII- K e The results in fig. 3 may be interpreted qualitatively by considering that the reductant involved is the Fey- ion, and that the process is first-order in this ion. The curvature of the plot of R us. [FeI'] is in this interpretation a consequence of the increase in the uncomplexed Fe" fraction when [Fe"] is increased.* If the interpretation of fig. 3 in terms of equilibrium (4) is correct, the results in fig. 2 should also reflect the complexation of Fe" in solution [eqn (4)]. Attempts to interpret the data in fig. 2 on the basis of equilibrium (4) alone fail: it is not possible to obtain a linear plot of R as a function of [Fey-] from fig. 2 using tabulated K , values.23 Even though deviations from the tabulated values due to medium effects could in principle account for the discrepancy,? a more straightforward interpretation can be obtained by also taking into account equilibrium (3). There is evidence from related systems in this sense: maxima are observed in the plots of R us. [L] at constant [Fe"] in related systems (e.g.L = EDTA) and this is to be ascribed to a retarding effect of equilibria such as (3)." By analogy, we must consider the possibility of an outer-sphere reaction of Fey- with complexed surface sites -Fe'"-NTA and by an inner-sphere reaction with uncomplexed surface sites -Fe"'-OH, the latter process having the larger rate constant : -Fe"'-NTA + Fey- --+ dissolution ( 5 ) Fe(H20)i'+Y3-+FeY-+6H,0. (4) k,. 0 The maximum observed in the Fe30,-EDTA system2' is due to an adequately high value of the ratio kl.o/kl.l and a relatively low value of the ratio K,/K,, the latter * From fig. 3, it is not possible to rule out a small contribution from a reaction pathway independent of Fe". Blank experiments show that in longer time spans, magnetite dissolves to some extent in Fell-free NTA solutions.In the absence of NTA, no dissolution takes place even in the presence of substantial Fe" concentrations [cf: however ref. (17)]. We have analysed the equivalent reaction in EDTA media and a reductive pathway has been suggested, the EDTA itself being the reductant.'.* t The introduction of ionic strength corrections causes only slight changes in the R us. [Fey-] profiles. Even though this is not conclusive evidence, it also points to the line of reasoning given in the text.12 3 m 6 a 2 2 1 Dissolution of Magnetite by Nitrilotriacetatoferrate(n) I I I I I 20 40 60 80 100 t/min Fig. 1. Volume of HCl(O.1080 mol dmP3) consumed to maintain constant pH as a function of time. Experimental conditions: 303 K, pH 2.8, magnetite mass 17.9 mg, solution volume 50 ~ m - ~ , 102[NTA] and 1O2[FeI1] (moldm-3) as follows: 0.60, 2.41 (a); 1.86, 2.12 (0); 3.87, 2.10 (A); 4.16, 2.91 (B); 4.19, 4.19 (A).4.0 c( I ," 3.0 E E 'p w - s 2.0 1 .o 0.5 1.0 2.5 3.0 [NTA]/lO-' mol dme3 Fig. 2. Initial dissolution rates R as a function of total [NTA] at [FeT1] = 2.1 x mol dm-3. Other conditions as in fig. 1.M. del V. Hidalgo et al. 13 I I I I 1.0 2.0 3.0 L.0 [ Fe"]/ 1 0-2 mol dm-3 Fig. 3. Initial dissolution rates R as a function of total [Fe"] at WTA] = 4.17 x mol dm-3. Other conditions as in fig. 1. defining the possibility of separating volume complexation of Fe" [eqn (4)] from surface coverage [eqn (3)]. In the present system, KJK, is considerably higher,? and this could account for the absence of a maximum.The two parallel paths (5) and (6) could in principle also explain the curvature of the R us. [Fey-] plot that can be derived from fig. 2 using the K, value from the l i t e r a t ~ r e . ~ ~ Fig. 4 shows that R/[FeY-] decreases as WTA] increases, in a fashion suggesting an (inverted) adsorption isotherm (except at very high [NTA], see below); however, the implied value of K, is too low to be reasonable, and an alternative explanation must be sought. According to the usual ideas about electron-transfer reactions,25 the rate constants kl,o and k l , l cover a complex sequence of events. As stated above, there are no experimental conditions under which pathway (5) becomes negligible. On the other hand, complete surface coverage by NTA is achieved in many of our experiments, and it is therefore possible to analyse kl,o in more detail. Process (5) in fact represents scheme 1.K* - FeIII-NTA + Fey- a- - FeIII-NTA Fey- (precursor complex) ket 11 k-et products - kdis - FeII-NTA. FeY (postcursor complex) Scheme 1. t This assertion is borne out by an expe.rimenta1 study of the adsorption of NTA, which shows that K,(NTA) is not much smaller than KJEDTA); on the other hand, Ke(NTA) is smaller than Ke(EDTA).2314 Dissolution of Magnetite by Nitrilotriacetatoferrate(11) 0.05 0.10 ' [ NTA]/mol dm3 - 0.30 Fig. 4. Changes in the ratio R/[FeY-] as a function of WTA]. Experiments are those presented in fig. 2. [NTA] has been corrected for complex formation in solution. The rate will be first-order in [Fey-] only if K*[FeY-] -g 1, i.e. if precursor complex formation is an unfavourable equilibrium, and if also kdis 9 kwet.Even without removing the latter condition, we may expect deviations from the first order if K*[FeY-] approaches unity. In that case eqn (7) follows P[ Fey-] 1 + P[FeY-] R = ket[-Felll-NTA] (7) and fits the data in fig. 4 reasonably well if K* = 2.5 x lo3 dm3 mol-'. This is a rather high figure, but not unreasonable in view of the possible interactions of Fey- with surface-bound NTA species, especially at high NTA concentrations. At [NTA] = 0.28 mol dm-3, the rate is higher than would be predicted from the above ideas. Several possible explanations can be put forward (conformational changes in surface -FeIII- NTA species, participation of Fey:-, medium effects, etc.) but none adequately substantiated.Gorichev and Kipriyanov2' have analysed the consequences of k-et 9 kdis; for this case they conclude that log R should vary linearly with log [a(Fe")/a(Fe"I)], a representing activities, the slope being 0.5. In the series of experiments included in fig. 2, the changes could then be attributed to the shifts in the FeIII-FeII couple potential as FJTA] is increased. Plots of log R against log ([Fe'I']/[Fe"]) taking into account the formation of Fey:-, FeY and Fey- are in fact roughly linear, the slope being ca. 0.33. However, addition of FeIII does not influence the rate appreciably, and this rules out a possible control by the FeI' phase-transfer process : even though the actual exponent for Ferrl (-0.5) for this case could in principle be changed without altering the essence of the argument, fast reverse electron transfer should give rise to a negative order on FeIII that is not observed experimentally in this case.7 The above discussion emphasizes the complexities and ambiguities of rate interpretations in these reactions.t At low [NTA], aqueous FerIr scavenges NTA, displacing equilibrium (4) to the left, and total inhibition of dissolution is observed.M . del V. Hidalgo et al. 15 Dependence of the Dissolution Rate on pH Fig. 5 shows the variation of dissolution rate as a function of pH at constant [NTA], [Fe"] and A / V ratio. A decrease in the rate as the pH increases can be seen together with a tendency towards a maximum, which cannot however be observed because H,Y precipitates at lower pH values.This dependence is due to the superposition of various effects :26 (1) the influence of pH on the speciation of complexed ions in solution, (2) the influence of pH on ligand adsorption and (3) the influence of pH on the concentration of active sites. The first factor is related to the changes in [Fey-] with pH due to the variation of [Y"] : WTA]/[Y3-] = 1 + Kii[H+] + Ki:K,-,1[H+l2 + KitKilK;i[H+]3 (8) where the K, are the successive acidity constants of H,Y. Other pH effects can be deduced from this by comparing R/[FeY-] values at various pH. This procedure corrects for changes in [Fey-] due to the relationship described by eqn (8), and the resulting plot given in fig. 6 shows that factors (2) and/or (3) above must be taken into account.Separation of the two remaining factors is qualitatively simple, because of the noted reactivity of both complexed and uncomplexed surface sites. The ratio kl, Jk1, being larger than one, surface complexation should in any case lead to a decrease in rate with decreasing pH; eventually, a minimum rate could characterize the pH of maximum adsorptivity typical of polycarboxylates on iron oxides (pH 2-3).43 '* 27 The influence of surface complexation on the pH dependence of the reaction rate is, however, probably minor; this is especially true for experiments carried out at high NTA concentrations, where surface coverage is large. The data in fig. 6 indicate a rate law of the form R = kH[H+]0.67 (9) for complexed surface sites; k , incorporates the NTA dependence.The exponent 0.67 is typical of dissolution kinetics, and it is generally interpreted as representing a Freundlich- type adsorption equilibrium (see below for a further possibility regarding changes in surface potential). Thus, eqn (9) represents a requirement for protons adjacent to the site where the reductant is adsorbed. The dissolution sites must therefore be visualized as kinks, where the bonding strength between -Fe"' and residual lattice oxide anions is decreased by protonation. Other systems'' 2 5 9 28 exhibit exponents 0.5 that are also in line with this reasoning. An important point to consider in the analysis of pH influence is the pH dependence of the surface potential, ly,, and the potential in the inner Helmholtz plane, lyB. For dissolution reactions controlled by the rate of transfer of cations across the interface, a model has been proposed'' that incorporates into the pH dependence of the rate an equilibrium factor (surface excess of H+) and a kinetic factor derived from the term exp (cczFAly/RT), where a is the electrochemical transfer coefficient, z the algebraic charge number of the transferring ion, F = 96500 C and Aly = y o - lyp This kinetic factor is indeed important in the dissolution of magnetite by sulphuric acid.17 In the present case, rate control by cation transfer is not indicated by the rate data.Iron transfer should take place as Fe'INTA and any increase in yo due to proton adsorption should hinder such a process. On the other hand, electron transfer from Fe'INTA in the inner Helmholtz plane to -Fe'I'-NTA should be assisted by increasing Aly.Aly cannot be easily modelled as a function of pH in polycarboxylate-containing systems as compared to simple mineral acid systems," because a large number of adjustable parameters are involved in the calculations. At high adsorption densities it is likely that surface potential is not very sensitive to pH. At high coverages electron16 -5.5 -6.0 bo 3 -6.5 Dissolution of Magnetite by Nitrilotriacetatoferrate(Ir) ' . 2 5 30 35 40 4 5 Fig. 5. Initial dissolution rates R as a function of solution pH: [Fe"] = 2.07 x lov2 mol dm+, PH [NTA] = 5.96 x mol dm-3. Other conditions as in fig. 1. I I I 5 4 3 PH Fig. 6. Log R* as a function of pH : data from fig. 5. R* is the initial reaction rate R corrected for changing [Fey']; the corrections were performed using the data in fig. 4.transfer itself, being an outer-sphere process, can be envisaged within the Marcus formalism; this is analysed in detail in a forthcoming paper.2o The driving force for the process is then related to a more negative potential for the couple Fe'II-Fe'' in the inner Helmholtz plane as compared to the analogous surface couple. In this context, changes in by/ are probably only minor perturbations that need not be considered.M. del V. Hidalgo et al. 17 - 6 -65 - 7.0 * 5 -75 - 8.0 3.00 310 3.20 3 30 K/ T Fig. 7. Arrhenius plot. Rate constants (s-l) were obtained in these experiments from the slope of [ 1 - (1 -f11/3] vs. time plots; similar results are obtained from initial rates.Experimental conditions: [NTA] = 3.8 x mol dm-3, pH 3.0. Other conditions as in fig. 1. mol dm-3, [Fe"] = 2.2 x Dependence of Dissolution Rate on the Temperature The Arrhenius plot shown in fig. 7 indicates an apparent activation energy of 73 kJ mol-l. According to the reaction scheme outlined above, this value is a composite of equilibrium and kinetic parameters involved in scheme 1. We thank Dr A. E. Regazzoni for helpful discussions and CONICET for partial support ~ - _ _ - . . - - - - _ _ _ _ . and a fellowship (to M. del V. H.). 2 M. A. Blesa and A. J. G. Maroto, in Decon;amination of Nuclear Facilities (Canadian Nuclear 3 J. Rubio and E. Matijevic, J. Colloid Interface Sci., 1979, 68, 408. 4 H. C. Chang, T. W. Healy and E. Matijevic, J. Colloid Interface Sci., 1983, 92, 469.5 H. C. Chang and E. Matijevic, J. Colloid Interface Sci., 1983, 92, 479. 6 H. C. Chang and E. Matijevic, Finn. Chem. Lett., 1982, 90. 7 M. A. Blesa, E. B. Borghi, A. J. G. Maroto and A. E. Regazzoni, J. Colloid Interface Sci., 1984, 98, 8 E. H. Rueda, M. A. Blesa and R. L. Grassi, J. Colloid Interface Sci., 1985, 106, 243. 9 N. Kallay and E. Matijevic, Langmuir, 1985, 1, 195. Association and American Nuclear Society, 1982), p. 1. 295. 10 Y. Zhang, N. Kallay and E. Matijevic, Langmuir, 1985, 1, 201. 11 N. Valverde and C. Wagner, Ber. Bunsenges. Phys. Chem., 1976, 80, 330.18 Dissolution of Magnetite by Nitrilotriacetatoferrate(11) 12 N. Valverde, Ber. Bunsenges. Phys. Chem., 1976, 80, 333. 13 N. Valverde, Ber. Bunsenges. Phys. Chern., 1977, 81, 380. 14 J. W. Diggle, Dissolution of Oxide Phases, in Oxides and Oxide Films, ed. J. W. Diggle (Marcel Dekker, 15 M. G. Segal and R. M. Sellers, J. Chem. SOC., Faraday Trans. I , 1982, 78, 1149. 16 E. C. Baumgartner, M. A. Blesa, H. Marinovich and A. J. G. Maroto, Znorg. Chem., 1982, 22, 2224. 17 V. I. E. Bruyere and M. A. Blesa, J. Electroanal. Chem., 1985, 182, 141. 18 M. G. Segal and R. M. Sellers, Advances in Inorganic and Bioinorganic Mechanisms, ed. A. G. Sykes (Academic Press, London, 1984), vol. 3, p. 97 and references therein. 19 M. A. Blesa, A. J. G. Maroto and P. J. Morando, J. Chem. Soc., Faraday Trans. I , 1986, 82, 2345. 20 M. A. Blesa, H. Marinovich, E. C. Baumgartner and A. J. G. Maroto, Znorg. Chem., in press. 21 E. B. Borghi, A. E. Regazzoni, A. J. G. Maroto and M. A. Blesa, to be published. 22 A. E. Regazzoni, G. A. Urrutia, M. A. Blesa and A. J. G. Maroto, J. Znorg. Nucl. Chem., 1981, 43, 23 A. E. Martell and R. M. Smith, Critical Stability Constants (Plenum Press, New York, 1974), vol. I. 24 A. E. Regazzoni, M. A. Blesa and A. J. G. Maroto, 55th Znt. Congr. Colloid Surf. Sci. (Potsdam, 25 N. Sutin, Prog. Znorg. Chem., 1983, 30, 441. 26 I. G. Gorichev and N. A. Kipriyanov, Russ. J. Phys. Chem. (Engl. Transl.), 1981, 55, 1558. 27 M. A. Blesa and A. J. G. Maroto, in Reactivity of Solids, Materials Science Monographs, ed. P. Barret 28 B. Terry, Hydrometallurgy, 1983, 11, 315. New York, 1973), vol. 2. 1489. New York, June 1985). and L-C. Dufour (Elsevier, Amsterdam, 1985), vol. 28A, p. 529. Paper 611057; Received 28th May, 1986