J. Chem. SOC., Faraday Trans. 1, 1982, 78, 1149-1 164 Kinetics of Metal Oxide Dissolution Reductive Dissolution of Nickel Ferrite by Tris(picolinato)vanadium(rI) BY MICHAEL G. SEGAL AND ROBIN M. SELLERS* Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucestershire GL 13 9PB Received 1 lth May, 1981 A detailed study of the reductive dissolution of NiFe,O, by V@ic); is reported. The kinetics of the reaction follow a cubic rate law and exhibit a first-order dependence on [V'I], indicating that the rate-determining step involves attack by V@ic); on Fe"' ions at the oxide surface. Dependences on [H+] and [free picolinate] are attributed to adsorption of these species at surface sites according to a simple model based on the Langmuir adsorption isotherm.Buffer and surfactant concentration had little effect. Some pitting etc. of the oxide surface occurs as the dissolution proceeds, although there appears also to be some general surface attack. The nature of the surface sites and subsequent steps in the dissolution process are discussed. The dissolution of metal oxides is of considerable practical importance in fields as diverse as the extraction of metals from ores, the transport of metals in the hydrologic cycle and the removal of oxide deposits from metal surfaces. Many metal oxides are difficult to dissolve, even in concentrated acids or in the presence of strong complexing agents. Our interest in this problem stems from the relatively poor oxide removal achieved in cleaning the pipework of water-cooled nuclear reactors, and in particular the removal of nickel ferrite, NiFe,O, (trevorite), and similar materials from the primary circuit of pressurised-water reactors.Nickel ferrite is often considered to be insoluble in acid,l although dissolution can be effected after heating in, for instance, phosphoric Complexing agents such as EDTA do not appear to have any significant effect on the rate of such reactions., In this paper we focus attention on the use of reducing agents to bring about dissolution, and describe a detailed study of the kinetics of dissolution of stoichiometric nickel ferrite by tris(pico1inato)vanad- ium(I1) in aqueous solution. Little previous work on reductive dissolution has been published. The method is perhaps most familiar in the dissolution of MnO, by acidified hydrogen peroxide, in the dissolution of anhydrous CrCl, by CrC1,,4 and in the use of thioglycollic acid in the analysis of iron(1n) oxide~.~ It has also been used to characterise soil samples by extraction of Fe and A1 using dithionite and oxalate,6 and patents have been issued for the dissolution of iron oxide scales using reductants such as erythorbic acid,' ascorbic acid7 and thioglycollic acid.8 More detailed studies have been made by Zabin and Taubeg on the reaction of Cr2+ with oxides such as MnO,, PbO,, Tl,O,, Mn,O,, Co,O, and CeO, by ValverdelO on the dissolution of FeO, Fe,O,, Fe,O,, COO, CuO, etc.by various metal-ion redox couples, by Bradburyll on the dissolution of Fe,O, by thioglycollic acid, etc. and our own preliminary work on the dissolution of FelI1 oxides by complexes of CrII, VII, FeII, etc.12 Of related interest is the work of Pryor and Evans1, and others14 on the electrochemical reduction of oxides on metal electrodes.11491150 METAL OXIDE DISSOLUTION KINETICS EXPERIMENTAL MATERIALS All solutions were prepared from triply distilled water. Picolinic acid (Aldrich) and sodium formate (B.D.H.) were recrystallised once from water before use. V2+ was prepared by electrolytic reduction of 0.2 mol dm-3 VOSO, in 0.5 mol dm-3 H,SO, at a lead cathode and stored under argon. The total vanadium content of the V" stock solution was determined spectrophotometrically as the peroxovanadium(v) complex taking = 28 1 dm3 mol-l cm-l, and the amount of V2+ and V3+ by direct absorbance measurements on the stock solution at 400 nm, taking&(V3+) = 8.3 dm3 mol-l cm-l and &(V2+) = 0.9 dm3 mol-l cm-l.The surfactants Triton X-100 and Hyamine 1622 were B.D.H. G.P.R. grade. All other chemicals were AnalaR grade and used without further purification. Nickel ferrite was prepared either by precipitation of the oxalates from a mixture containing the appropriate quantities of nickel and ferrous acetate (the 'oxalate' method)', l5 or by the precipitation of the hydroxides by addition of Na,C03 to a mixture of nickel and ferric nitrates (the 'carbonate' method). Precipitates were washed with triply distilled water (ca. 15 dm3 per 15 g of sample) to remove all acetic acid+NaNO,, dried in an oven at 110 OC for 1-2 d, and finally calcined at 1000-1400 OC under argon to produce the ferrite. Sizing of the oxide particles was achieved by grinding and sieving using Endicott sieves.The purity of the oxides was checked by infrared spectroscopy (Perkin Elmer 180 photometer) and X-ray diffraction (Philips 2 kW X-ray generator). No evidence for impurities such as NiO or Fe203 was found, except in samples C1 and C2 (see table 1) which were calcined below the ferrite transition temperature. The lattice parameters measured agree well with literature values.16 Chemical compositions were confirmed by dissolving the oxides in phosphoric acid (20 mg oxide in 50 cm3 B.D.H. AristaR H3PO4, refluxed for 12 h) or hydrochloric acid (20 mg oxide in 50 cm3 B.D.H. AristaR HC1, refluxed for 24 h), and determining the iron and nickel contents by atomic absorption.Analyses found were sample A1 (in H3P0,):44.1% Fe, 25.4% Ni; A1 (in HC1):44.3% Fe, 22.2% Ni; B6 (in H3P0,):48.5% Fe, 27.4% Ni; C3 (in H3P0,):48.2% Fe, 24.5% Ni; calculated for NiFe,0,:47.7% Fe, 25.0% Ni. Similar results were obtained in the dissolution of these samples by V(pic);.* APPARATUS AND PROCEDURE The kinetic runs were carried out in glass reaction vessels. Constant temperature (in the range 40-80 "C) was maintained by circulating water from a water bath through a jacket around the reaction vessel. Solutions of picolinic acid, etc. containing ca. 2 mg of oxide were degassed by bubbling with high-purity argon (B.O.C.) for i-1 h before initiating the reaction by addition of V2+. The !!(pic); reagent is sensitive to oxygen, and to ensure no complications from aerial oxidation a slow stream of argon was passed through the solution during the run.Solutions were stirred using a magnetic stirring bar. The rate of dissolution was measured by withdrawing CQ. 1.5 cm3 aliquots by syringe at regular intervals. The dissolution reaction was arrested by exposure to air, the V(pi~)~ produced being unreactive towards nickel ferrite (see below). The samples were then diluted x 10 with 0.1 mol dm-3 HC1 (B.D.H. C.V.S. reagent), filtered through a 0.22 pm filter (Millipore Millex GS) to remove any unreacted oxide, and analysed for iron and nickel content by atomic absorption spectrometry using a Baird A5100 instrument. At the end of each run some of the reakent was cooled to room temperature under argon, and the pH measured on an EIL type 7050 meter.Surface area measurements were made by N, absorption with a Perkin Elmer 212D sorptometer by the B.E.T. single-point method. Scanning electron micrographs were taken with om; * pic- = picolinateM. G. SEGAL A N D R. M. SELLERS 1151 a Cambridge Stereoscan 150. Partial dissolution experiments to investigate changes in surface morphology were done using ca. 25 mg oxide and appropriate reagent concentrations. After the required amount of dissolution had taken place the oxide was removed by filtration, washed with 20 cm3 HC1 and 50 cm3 H,O and dried in an oven at 110 OC for ca. 1 h before mounting for observation in the scanning electron microscope. RESULTS MORPHOLOGY OF THE PARTICLES The morphology of the nickel ferrite particles was characterised by surface area determinations and scanning electron microscopy.The surface area measurements are summarised in table 1. For material calcined at 1400 O C the figures are only slightly TABLE 1 .-CALCINING HISTORIES AND CHARACTERISTICS OF NICKEL FERRITE SAMPLES specific lattice sample preparative calcining particle surface area parameter no.a method conditions size/pmb /m2 g-lc %/Ad A1 A2 A3 Bl B2 B3 B4 B5 B6 c1 c2 c 3 c 4 c 5 C6 oxalate oxalate oxalate carbonate carbonate carbonate carbonate carbonate carbonate carbonate carbonate carbonate carbonate carbonate carbonate 6 h at 1000 OC 12 h at 1000 O C 24 h at 1400 OC 6 h at 1400 O C 6 h at 1400 OC 6 h at 1400 OC 6 h at 1400 OC 6 h at 1400 OC 6 h a t 1400OC 6 h at 600 OCe 6 h at 800 O C e 6 h at 1000 OC 6 h at 1200 O C 6 h a t 1400OC 72 h at 1400 OC < 100 < 100 < 100 > 300 150-300 106-150 75-106 53-75 < 53 < 53 < 53 < 53 < 53 < 53 < 53 1.2 - < 0.04 ca.0.08 0.14 8.3374 - - - 8.3383 - 8.3395 - a A2 and A3 prepared by recalcining Al; B1-B6 calcined as one sample, and then sieved; by from X-ray powder C1C6 same batch divided after drying at 110 OC, and calcined and sieved separately; sieving, sizes correspond to particle diameters; diffraction; by N, adsorption; X-ray and infrared measurements show these samples to contain some Fe,O,. larger than the calculated geometric value, which for spherical particles with a radius of 10 pm is 0.056 m2 g-l taking the density as 5.37 g Lowering the calcining temperature increased the specific surface area by a factor of ca.10, indicating some porosity in the oxide formed under these conditions. This conclusion was borne out by the scanning electron microscopy results (plates 1 and 2) which showed the 1400 O C calcined material to consist of amorphous particles, with more or less flat faces. Under high resolution stepped structures were detectable (plate 2). It is also noteworthy that these larger particles (typically > 50 pm in diameter) have associated with them some much smaller particles (< 1 pm in diameter), which have not been separated by sieving. The particles of material prepared by the oxalate method and calcined at1152 METAL OXIDE DISSOLUTION KINETICS 1000 O C were to good approximation spherical, but with a much more convoluted surface morphology.The nature of the surface changes during dissolution. At least three types of attack can be recognised, although not every particle exhibits the features of all three. In the first regular pits are formed as illustrated in plate 3(a). In cross-section these are hexagonal, triangular, rectangular or octagonal, with hexagonal pits being perhaps the most common. Deep fissures are formed in other particles, probably as a result of attack at grain boundaries or edge dislocations. This behaviour is shown in plate 3(b). Finally there is a more general surface attack, apparently not directed at any specific site. Plate 4(a) illustrates attack of this type on what seems initially to have been a stepped structure. One or two particles also showed a ‘dendritic’ like attack as shown in plate 4(b).This probably results from the dissolution of an iron-rich phase, and is indicative of some microinhomogeneity in a small fraction of the particles. VARIATION OF THE AMOUNT OF OXIDE DISSOLVED WITH TIME The variation of the amount of nickel ferrite dissolved with time in a typical dissolution run is shown in fig. 1. The main features are a rapid initial dissolution rate, I I 1 I I 1 100 150 200 250 0 50 time/min FIG. 1 .-Variation of amount of oxide dissolved with time. Measurements made in solutions containing 3.2 x mol dm3 V(pic); + 0.030 mol dm-3 free picolinate + 0.1 mol dm-3 HCO; + 20 ppm Triton X-100, pH 4.3, T = 80 OC and using nickel ferrite sample Al. 0, percentage dissolution in terms of the amount of Fe dissolved; 0, based on Ni dissolved.Line calculated according to eqn (7) with koba = 4.5 x min-l. falling off as the size of the particles decreases. This behaviour is interpreted on the assumption that reaction occurs at the surface of the particles, at a rate proportional to the instantaneous surface area, i.e. dM dt - = - k A where M is the mass of undissolved oxide, k is a rate constant in g m2 min-l and A is the total surface area of the oxide in m2. For spherical particles of uniform sizeJ . Chem. Soc., Faraday Trans. 1, Vol. 78, part 4 Plate 1 PLATE 1 .-Scanning electron micrographs of NiFe,O, samples calcined at lo00 OC. (a) Sample A1 ; prepared by the oxalate method. (b) Sample C3; prepared by the carbonate method. M. G. SEGAL AND R.M. SELLERS (Facing p . 1 152)J. Chem. SOC., Faraday Trans. 1, Vol. 78, part 4 Plate 2 PLATE 2.-Scanning electron micrographs of NiFe,O, samples calcined at 1400 O C . (a) Sample B5; (b) sample B4, detail showing stepped structure. M. G. SEGAL AND R. M. SELLERSJ . Chem. Soc., Faraday Trans. 1, Vol. 78, part 4 Plate 3 PLATE 3 .-Scanning electron micrographs of NiFe,O, sample B5 after treatment in solution containing 1.5 x lop2 mol dmp3 V(pic);+3.2 x lo-* mol dmP3 free picolinate+0.2 mol dmP3 HCO;+ 100 ppm Triton X-100 at 80 OC for 5 h (25% dissolved). (a) Regular pits; (6) grain boundary attack. M. G. SEGAL AND R. M. SELLERSJ. Chem. SOC., Faraday Trans. 1 , Vol. 78, part 4 Plate 4 PLATE 4.-Scanning electron micrographs of NiFe,O, particles after treatment in solution containing 1.5 x lo-, mol dm-3 V(pic); + 3.2 x lo-, mol dmP3 free picolinate+0.2 mol dmP3 HCO;+ 100 ppm Triton X-100 at 80 OC.(a) Sample B5 after 35 min in reagent (8.4% dissolved) showing result of general surface attack on stepped structure; (b) sample B5 after 5 h in reagent (25% dissolved) showing dendritic attack. M. G. SEGAL AND R. M. SELLERSM. G. SEGAL A N D R. M. SELLERS 1153 (radius, r), the surface area, A , will be given by eqn (2), where N is the number of particles, and the total mass, M, is given by eqn (3) A = 47rr2N (2) (3) 4 3 M = -nr3 Np. Substituting eqn (2) and (3) into eqn (1) and integrating gives eqn (4), where Mo is the mass of the particles at t = 0 Experimentally it is convenient to measure the concentration, Ct, of the metal ion (Fe or Ni) in solution, which is given by eqn ( 5 ) where x is the weight fraction of the metal in the oxide and V is the volume of the solution.When all the oxide has dissolved eqn (5) becomes eqn (6) XMO C , =- v ' Substituting eqn (5) and (6) into eqn (4) and rearranging gives eqn (7) ( l - z y = 1 - & kt (7) where ro is the initial particle radius. Thus a plot of [ 1 - (C,/C,)]i against t should give a straight line with a slope equal to kobs (= k/rop), an intercept of 1 on the [ 1 - (Ct/C,)]i axis and an intercept of k& (= tm) on the time axis. Fig. 2(a) shows the data of fig. 1 treated in this way, from I I I I I I 0 0.2 0.4 0.6 0.8 1.0 I I 1 I I I 0 0.2 0.4 0.6 0.8 1.0 tlt, FIG. 2.--Cubic rate law plots. (a) 0, Data of fig.1 replotted according to cubic rate law, eqn (7), t , = 220 min; 0, cubic rate law plot for dissolution of nickel ferrite sample with narrow particle size distribu- tion (sample B3), reagents and conditions as for fig. 1, t , = 1 1 0 0 min. (b) Calculated cubic rate law plot for a non-uniform size distribution (see text for details). 38 FAR 11154 METAL OXIDE DISSOLUTION KINETICS which it is seen that a linear dependence according to eqn (7) is obtained up to ca. 75% dissolution {[ 1 - (C,/C,)]i = 0.63). Thereafter the dissolution becomes slower than predicted by this equation, because the particles do not in reality have a uniform size. That this is so we have been able to demonstrate in two ways. The first is by measuring the dissolution rate as a function of time for an oxide sample with a very narrow size distribution (sample B3, table 1).This obeyed the cubic rate law to at least 90% dissolution, as shown in fig. 2(a). Secondly we have calculated the variation of [ 1 - (Ct/Cw)]i with time on the assumption that the oxide consists initially of equal masses of particles of radius r, 2r, 3r, 4r and 5r. As shown in fig. 2(b) this is non-linear over long times, but approximates to a straight line for the first ca. 60% of the dissolution {[l -(Ct/Cw)]i 2 0.731, and for practical purposes behaves as if the distribution were uniform, with radius ca. 2.5- [straight line, fig. 2(b)]. The validity of eqn (7) has also been investigated by determining kobs as a function of initial particle size. For this purpose a sample of nickel ferrite was sieved into six size fractions (cf.table l), and the dissolution kinetics measured under otherwise constant conditions. The results of this experiment are shown in fig. 3, from which it 15 I c ... ,E 10 Ol I 1 I 1 J 0 100 200 300 400 500 particle diameterlpm FIG. 3.-Effect of particle size on the dissolution kinetics of NiFe,O, by V(pic);. Reagents and conditions as for fig. 1, but using oxide samples B 1-B6. Vertical error bars are mean deviations, horizontal error bars cover range of particle sizes as determined by sieving. Line is calculated according to eqn (8) with k / p = 7 x lo-* m min-l. is clear that there is an inverse relationship between kobs and particle diameter, d. The line shown in fig. 3 is drawn according to eqn (8) with k / p = 7 x m min-l, and gives a reasonable fit to the data Plotting k& against d yields a straight line, but with intercept k& = ca, 300 min when d = 0, indicating some agglomeration of smaller particles.The cubic rate law, eqn (7), clearly provides a good description of the kinetics of th? V(pic); + NiFe,O, reaction, and establishes that the rate-determining step involves reaction at the particle surface. For almost all the work described here dissolution rates were determined from the amount of iron in solution. Rates based on nickel gave similar results (cf fig. 1 and 6). Detailed comparison of the two reveals, however, that the iron : nickel ratio varies from the stoichiometric value of 2 in the early stages of the dissolution (< 30%), and, as the data in fig.4 show, iron is dissolved preferentially.M. G. SEGAL AND R. M. SELLERS 1155 0 1 I I I I I I I 0 10 20 30 40 50 60 time/min FIG. 4.-Variation of the iron: nickel ratio in solution as a function of time in the dissolution of NiFe,O, by V(pic);. Measurements made in solutions containing 3.2 x mol dm-3 V(pic); + 3.0 x lo-, mol dm-3 free picolinate + 0.10 mol dmd3 HCO; + 20 ppm Triton X-100, pH 4.2, T = 80 O C and using nickel ferrite sample A1 . EFFECT OF REAGENT CONCENTRATION ON THE DISSOLUTION KINETICS Determination of the effect of reagent concentration on dissolution kinetics was carried out with oxide sample Al. The solutions contained in general an excess of picolinate to ensure that all V2+ was present as the blue-black coloured 1 : 3 complex, although in a few runs a significant amount of the deep red 1 :2 complex was present.The concentration of uncomplexed picolinate in the solution was calculated on the basis of complete formation of the 1:3 complex, or taken to be zero if the total [picolinate] was less than 3[V2+]. The stability constants for the formation of these complexes are log Kl = 4.4, log K2 = 4.6 and log K3 = 3.8 in 0.5 mol dm-3 KCl at 25 'C.la Corrections were also made to take into account the V3+ present as an I I 1 I 1 I 1 1 0 2 4 6 8 10 12 14 [ W]/ 1 0-3 mol dm-3 FIG. 5.-Effect of vanadous concentration on the kinetics of-dissolution of NiFe,O, by V(pic);. 0, Measurements made in solutions containing 0.020 mol dm-3 free picolinate + 0.1 mol dmw3 HCO;+20 ppm Triton X-100, pH 4.0, T = 80 OC and using nickel ferrite sample Al.0, Measurements made in solutions containing 0.01 8 rnol dm-3 free picolinate + 0.1 mol dm-3 CH3CO; + 20 ppm Triton X-100, pH 4.4, T = 0 OC and using nickel ferrite sample Al. Lines calculated according to eqn (1 l), using constants shown in table 7. 38-21156 METAL OXIDE DISSOLUTION KINETICS impurity (typically 3-779 in the V2+ solution, again assuming complete formation of the 1 : 3 complex, for which log Q3 = 1 5.4.lS Solutions also contained a surfactant to aid dispersion of the oxide and a buffer to keep the pH constant. The buffer concentration was 0.1 mol dm-3 in most runs and effectively maintained the ionic strength of the solution constant at ca. 0.1 mol dm-3. Reaction rate constants, kobs, were calculated from the slopes of plots of [l -(Ct/Cm)$ against time [ i e .according to the cubic rate law, eqn (7)]. Values of C, were based on the weight of oxide present initially. In a number of runs where the dissolution rate was rapid, the reaction was allowed to go to completion, and C , determined experimentally. Agreement with calculated values was within & 10%. The concentration dependences found are shown in fig. 5-7 and tables 2-4, and can be summarised as follows. VII Using formate as a buffer (80 "C) a good linear dependence on [V1] was found (fig. 5 ) for [VII] = (1 - 13) x mol dm-3, with no evidence for any pathway indepen- dent of [VII]. With acetate as the buffer (60 "C) a linear dependence on [VII] was again found (fig. 5) for [VII] = (0.9-7) x mol dm-3.Here, however, there was some indication of a [VI1]-independent pathway. A dependence of the form given in eqn (9) provides a good description of the data, and a linear regression of [VII] on kobs yields a = 3 x lo-* min-l and b = 0.4 dm3 mol-l mip-l kobs = a 4- b[V'I]. (9) However, the intercept falls within the relatively large experimental errors (ca. 15 %) involved in these experiments, and the measurements do not therefore unambiguously establish the existence of a [VIII-independent pathway. Other interpretations of the data are possible (for instance that the reaction becomes less than first order in [VII] at high [VII]), but we conclude that under the conditions of our experiments the only pathway of any consequence is that involving direct reaction of VII with the surface. PICOLINATE Only a weak dependence on [picolinate] was found, kobs falling by a factor of ca.2 on increasing the free picolinate concentration from 0 to 0.35 mol dm-3 in the presence of formate buffer (80 "C) or by a factor of ca. 6 in the concentration range 0-0.2 mol . dm-3 with acetate buffer (60 "C) as shown in fig. 6. PH A marked increase in rate was found on decreasing the pH. With formate buffer (80 "C) the rate increased ca. 20-fold in the pH range 6-3.2 (fig. 7) and ca. 10-fold in the same range in the presence of acetate buffer (60 "C); pH levels were adjusted by addition of HCl or NaOH, and to check that the counter ions (Cl-, Na+) had no effect some runs were done in which excess NaCl or Na2S0, was added. No significant change in rate was found (cf.table 4). BUFFER With formate as the buffer the rate was independent of [HCO,-] for [HCO,-] = 0.1 - 0.5 mo1.dm-3 (table 2). Substituting acetate or cacodylate as the buffer gave essentially the same results.M. G. SEGAL AND R. M. SELLERS 2.0k 1157 - FIG. 6.-Effect of picolinate concentration on the kinetics of dissolution of NiFe,O, by V(pic),. 0, Measurements made in solution containing 3.5 x mot dm-3 V(pic)-+O.l mol dm-3 HCO; +20 ppm Triton X-100, pH 4.4, T = 80 OC and using nickel ferrite sample Al. Rate constant estimated from amount of Fe dissolved. A, As (O), but rate constant estimated from amount of Ni dissolved. 0, Measurements made in solution containing 3.0 x rnol dmp3 V(pic), +0.1 mol dmp3 CH3CO; +20 ppm Triton X-100, pH 4.4, T = 60 OC and using nickel ferrite sample Al.Lines calculated according to eqn (1 l), using constants shown in table 7. I I I I I I 3.0 3.5 4.0 4.5 5.0 5 . 5 6.0 PH FIG. 7.-Effect of pH on the kinetics of dissolution of NiFe,O, by V(pic);. 0, Measurements made in solutions containing 3.5 x mol dm-3 free picolinate+O.l mol dm-3 HCO;+20 ppm Triton X-100, T = 80 O C and using nickel ferrite sample Al. 0, Measurements made in solution containing 3.0 x mol dm-3 free picolinate+O.l rnol dm-3 CH,CO; + 20 ppm Triton X-100, T = 60 OC and using nickel ferrite sample Al. Lines calculated according to eqn (1 l), using constants shown in table 7. mol dm-3 Vbic); + 3.0 x mol dm-3 V(pic);+ 1.8. x1158 METAL OXIDE DISSOLUTION KINETICS TABLE EFFECT OF BUFFER ON THE v@ic); + NiFe,O, REACTION^ ~ ~~ ~ buffer (kobs/[V"l)b conc/mol dm-3 type pH /dm3 mol-l min-' 0.1 HCO; 4.3 1.6 0.2 HCO; 4.6 1.8 0.5 HCO; 4.9 1.2 0.1 CH3C0, 4.4 2.2 a Measurements made in solutions containing also ca.3 x 2.9 x lo-, mol dm-3 free picohate+ 20 ppm Triton X-100, T = 80 O C ; on the amount of iron dissolved. mol dm-3 V(pic),+ Rate constants based TABLE 3.-EFFECT OF SURFACTANT ON THE v(pic); 4- NiFe,O, REACTION' sur fac tan t (kobs/[V"I) conc (ppm) typeb /dm3 mol-l mind' 1.6 20 T 1.6 100 T 1.5 100 NaLS 1.2 100 H 2.5 - 0 a Measurements made in solution containing also ca. 3 x mol dm-3 V(pic); + 2.9 x T = Triton X-100, lo-, mol dm-3 free picolinate + 0.10 mol dm-3 HCO;, pH 4.4, T = 80 "C; NaLS = Sodium lauryl sulphate, H = Hyamine 1622. SURFACTANT Varying the concentration of the surfactant, Triton X-loo,* had practically no effect on the rate (table 3).Within experimental error there was no change in rate on omitting the surfactant altogether, and only minor changes on substituting an anionic surfactant, sodium lauryl sulphate, or a cationic one, Hyamine 1622.t BLANK EXPERIMENTS, etc. A number of blank experiments were also done to examine the dissolution behaviour in the absence of VII. As the results in table 4 show, none of the othei reagents used in this work bring about dissolution, except V(pic),, the oxidation product of the V1* reagent. This dissolves nickel ferrite at least fifty times more slowly than V(pic),-, and thus makes a negligible contribution to the processes described here. Changes in stirring rate were also without effect.* Triton X-100 is a proprietary name for a non-ionic surfactant composed of iso-octylphen- Hyamine 1622 is a proprietary name for di-isobutylphenoxyethoxyethyldimethylbenzyl ammonium oxypolyethoxyethanol containing cu. 10 ethoxy units. chloride monohydrate.M. G. SEGAL AND R. M. SELLERS 1159 TABLE 4.-EFFECT OF STIRRING RATE AND ADDITIVES ON THE V(piC), 4- NiFe,O, REACTION, AND BLANK EXPERIMENTS reagents, etc. effect of stirring ratea no stirring 1.9 small stirring bar at maximum speedb 1.6 large stirring bar at maximum speed 1.6 effect of other additives" no other additives 0.10 mol dm-3 NaCl 0.10 mol dm-3 Na,SO, 1.6 1.7 1.8 blank experimentsC 0.04 mol dm-3 picolinate + 0.10 mol dm-3 < 1 % dissolution in 4?j h 1.0 mol dm-3 H,SO, < 1% dissolution in h 6.4 x 1 0-3 rnol dm-3 V(pic), + 0.022 mol dm-3 HCO;, pH 4.6 0.033d free picolinate, 0.10 mol dm-3 HCO;, pH 3.9 a Measurements made in solution containing also ca.3 x mol dm-3 mol dm-3 free picolinate + 0.10 mol dmW3 HCO; + 20 ppm Triton X-100, Solutions also V(pic); + 2.9 x pH 4.3, T = 80 OC; * Standard conditions used for all other dissolution runs; contained 20 ppm Triton X-100; T = 80 OC; kOb,/[V1ll]. 4 3 n - I z g2 -3 E + s! - 1 0 2.8 2.9 3 .O 3.1 3.2 I 0 3 KIT FIG. 8.-Arrhenius plots for the dissolution of NiFe,O, by V(pic);. 0, n = 2. Measurements made in solutions containing 3.5 x mol dm-3 V(pic); + 0.030 mol dmb3 free picolinate +O. 1 mol dm-3 HCO;+20 ppm Triton X-100, pH 4.4 and using nickel ferrite sample Al. 0, n = 3.Measurements made in solution containing 3.0 x mol dm-3 V(pic);+O.O18 mol dm-3 free picolinate+O.l mol dm-3 CH3CO; + 20 ppm Triton X-100, pH 4.4 and using nickel ferrite sample Al. Lines calculated from least-squares analysis of data.1160 METAL OXIDE DISSOLUTION KINETICS TABLE 5.-ACTIVATION ENERGIES FOR THE v(piC), + NiFe,O, REACTION buffer EJkJ m o P formate 67+ 10 acetate 72+11 EFFECT OF TEMPERATURE ON THE DISSOLUTION KINETICS The influence of temperature on the dissolution rate was investigated for both formate and acetate buffers in the temperature range 40-80 OC. Good agreement with the Arrhenius law was obtained, as shown in fig. 8. Activation energies estimated from these plots are shown in table 5. The figures are very similar, and the reaction is'clearly independent of the nature of the buffer.TABLE 6.-EFFECT OF OXIDE CALCINING TEMPERATURE ON THE KINETICS OF THE V(pic), + NiFe,O, REACTION^ sample oxide preparation calcining (kobs/[VI'I) no.b method conditions /dm3 mo1-I min-' A1 A2 A3 c 1 c 2 c3 c 4 c 5 C6 oxalate oxalate oxalate carbonate carbonate carbonate carbonate carbonate carbonate 6 h at 1000 OC 12 h at 1000 OC 24 h at 1400 OC 6 h at 600 OC 6 h at 800OC 6 h at 1000 OC 6 h at 1200 OC 6 h at 1400 OC 72 h at 1400 OC 1.6 1.7 0.90 5.0 0.41 0.34 0.31 0.29 ca. 17 Measurements made in solution containing ca. 3 x mol dm-3 V(pic);+ mol dm-3 free picohate+ 0.10 mol dm-3 HCO, +20 ppm Triton X-100, pH 4.3, 2.9 x T = 80 OC; CJ table 1. EFFECT OF OXIDE CALCINING TEMPERATURE ON THE DISSOLUTION KINETICS Results on the effect of calcining temperature on the dissolution of NiFe,O, by V(pic); are given in table 6.The range of temperatures over which calcining could be varied was limited to ca. 1000-1400 OC. The lower limit is set by the need for a temperature of at least 800 OC in order to form the spinel phase, and the upper limit by the materials of construction of the calcining furnace and the oxide itself. The two oxides calcined at 600 and 800 O C (samples C1 and C2, respectively) contain Fe,O,, and this presumably accounts for their much higher dissolution rates. For the oxides prepared by the oxalate method there was a reduction in rate of about a factor of 2 on increasing the calcining temperature from 1000 to 1400 OC but a much smaller change with the 'carbonate' oxides.With neither group of oxides did the length of the calcining period have any effect.M. G. SEGAL A N D R. M. SELLERS 1161 The decreasing trend in rate constant with increasing calcining temperature probably arises from changes in the surface areas of the oxides through sintering. The differences in the relative decrease are more difficult to understand. They may reflect the slightly different calcining histories of the two groups of oxide [see footnote (a), table 11, although this seems unlikely, or it may be that some characteristic of the oxide (its fault structure?) is more easily annealed out or modified at 1000 O C when prepared by the carbonate method than when prepared by the oxalate method. DISCUSSION THE REDUCTIVE DISSOLUTION PROCESS The variation of the dissolution rate with time and its dependence on the size of the particles establish clearly that the dissolution process involves reaction at the particle surface as the rate-determining step, and the linear dependence on [v(pic);] that the species reacting at the surface is V(pic);.For reasons outlined elsewhere12 we ascribe this to an outer-sphere electron-transfer to FeIII ions in the surface (10) V(pic); + >FelIJ -+ V(pic), + >FeII. The other reagent dependences we attribute to the effect on reaction (10) of adsorption of H+, picolinate, etc. at the surface according to a simple model based on the Langmuir adsorption isotherm. Assuming the reactions occurring to be those shown in the following scheme, kobs will vary with reagent concentration according to eqn (1 1).kll V(pic); + >s -+ V(pic), + >s- K b >s+H+’s-H+ k b V(pic); + >s-H+ -+ V(piC), + >s-H K c >s + L * >s-L kc V(pic); + >s-L + V(pic), + x-L- (L = picolinate, without regard to its state of protonation with >s = surface site.) This equation has been used to calculate the lines shown in fig. 5-7, using the values of the constants given in table 7, and accounts well for the results obtained. The changes in the five constants with temperature are much as expected with the rate constants k,, k , and k, increasing, and the binding constants Kb and K, decreasing (K, markedly) with increasing temperature. It must be stressed, however, that the mechanism as represented in the scheme is a simplification. In particular it neglects adsorption of picolinate at protonated sites, and assumes that the anionic and zwitterionic forms of picolinic acid behave identically.This is certainly not the case in the adsorption of picolinate onto haematite, where we find that the binding constant reaches a maximum at pH ca. 4.8.3 Partly as a result of these assumptions it will be1162 METAL OXIDE DISSOLUTION KINETICS TABLE 7.-RATE AND EQUILIBRIUM CONSTANTS FOR SOME OF THE PROCESSES IN THE V(pic); + NiFe,O, REACTION rate equilibrium constanta reaction acetate media, 60 OC formate media, 80 O C ka 0.4 k 0.1 0 . 7 5 f 0.25 kb 6 f 2 13f4 kC 0.10 & 0.03 0.75 k 0.25 Kb 3700 f 1200 2300 f 700 KC 150250 20&7 a k,, kb and kc in units of dm3 mo1-I min-l, Kb and Kc in units of mol dm-3. seen that the variation of kobs with [pigolinate] at 80 OC is attributed solely to the replacement of protonated sites by >s-pic as the picolinate concentration increases.More complex mechanisms could be written, but in view of the relatively large uncertainties in the measured rate constants we do not feel that they would contribute materially to our understanding of the factors determining the dissolution kinetics. To summarise, we conclude that adsorption of H+ increases the rate of the reaction, whereas picolinate has an inhibiting effect in comparison with the ‘free’ surface (i.e. kb > k , > kc), and that protons are more strongly bound to the surface than picolinate The surface sites, >s, can be identified with FeIII ions. Protons probably add to (12) Unfortunately it is not known which crystallographic planes are present at the surface of nickel ferrite particles, but undoubtedly the surface contains more than one type of hydroxide (e.g.bound to one Fe3+, to 2 Fe3+ or to Ni2+), as found with other FeIII-containing oxides such as goethite (a-Fe00H).19 Adsorption of picolinate occurs by displacement of these surface hydroxides. In the scheme given above this is written as occurring at a single site, though this may not be correct, for we have evidence in the adsorption of picolinic acid on haematite that two surface sites are occupied per molecule of ad~orbate.~ Similar behaviour has been found with oxalic acid, selenious acid, orthophosphoric acid and sulphuric acid on goethite, e t ~ . ~ ~ - ~ ~ The Fe3+ ions in nickel ferrite are not all equivalent.The oxide’s structure is that of an inverse spinel2s in which the 0,- ions exist in a cubic close-packed array, with the Ni2+ and half the Fe3+ ions occupying the larger octahedral holes and the remaining Fe3+ ions the smaller tetrahedral holes. Whether these two types of Fe3+ behave differently or not is unknown, and raises the further question of whether V(pic); attacks surface Fe3+ ions indiscriminately (be they in octahedral or tetrahedral holes), or only those at surface defects such as ledges, kinks, dislocations, etc. The kinetic results are equivocal on this point, but the morphological changes indicate some localised attack. Faults in the oxide structure seem to be preferred, but more generalised attack also occurs at a measurable rate.The dissolutions of Fe,03 and Fe30, by V(pic); are so rapid3 in fact, that we suspect they approach the diffusion-controlled limit and hence involve little discrimination between surface sites. This interpretation is not certain, however, for complications arise from the charge on the particles (in our calculations (Kb > &)* the surface hydroxides to give bound water according to reaction (12) >Fe-OH + H+ f >Fe-OH,.M. G. SEGAL AND R. M. SELLERS 1163 we have assumed it to be zero) and the highly porous nature of the oxides used in these experiments. We hope to investigate these reactions in more detail. The processes subsequent to the reduction of the surface Fe3+ ions, i.e. the actual dissolution reactions, are not very clear but presumably involve some kind of terrace-ledge-kink me~hanism.~' The driving force for the disruption of the oxide surface and the ejection of Fe2+ ions into the bulk solution is the increase in size of iron ions on reduction (typical crystallographic radii for the high-spin ions are 0.75 A for Fe2+ and 0.65 A for Fe3+ 28), and the increased electrostatic repulsion between the electron clouds of the iron ions and the adjacent 02- ions.Nickel ions probably pass into solution after removal of most or all of the neighbouring ferric ions, and this probably accounts for the disparity between iron and nickel in solution in the early stages of the dissolution. Some Fe3+ ions may even pass into solution in this way. They would, however, be immediately reduced by V(pic); in the bulk solution, making such a pathway indistinguishable from that involving reduction prior to dissolution.COMPARISON WITH ACID DISSOLUTION REACTIONS An extensive literature exists on the dissolution of metal oxides by acid, including both theoretical descriptions of the processes i n ~ o l v e d , ~ ~ - ~ ~ and experimental studies of the factors that influence the kinetics.l07 32-38 Acid dissolutions occur by attack of protons at defects in the oxide surface and give rise in the main to pitting and other localised forms of attack. This is to be contrasted with V(pic),, where in addition some more general surface attack seems to occur. V(pic); also differs in its sensitivity towards surfactants. Jones et have shown that adsorption of such compounds at nickel oxide surfaces can effectively block dissolution by acid, whereas the experiments described here suggest no dependence on surfactant.The differences probably arise because of the need in acid dissolution for the proton to diffuse right up to the particle surface before reaction can take place, whereas electron transfer can occur ' through' the surfactant, as found in the reaction of the hydrated electron with organic compounds solubilised in m i ~ e l l e s . ~ ~ ~ 40 TECHNOLOGICAL APPLICATIONS The results presented here describe a new and rapid means of dissolving oxide deposits containing FeI" and in particular NiFe204.41 V(pic); and similar reductants [which have been dubbed LOMI (low oxidation-state metal ion) reagents] clearly show much promise for use in the cleaning of oxide deposits in power plant, especially in pressurised-water reactors where the deposited oxide is principally NiFe20,.42 The cleaning of various reactor artefacts by V(pic); and the use of the reagent for a full-scale reactor cleaning have been described el~ewhere.~~ This work was performed in part under contract RP1329-1 with the Electric Power Research Institute, and is published by permission of the Central Electricity Generating Board. We thank Drs D.Bradbury, T. Swan and C. J. Wood for their advice and encouragement, Messrs B. Daniel and G. Marsh for experimental assistance, and Dr P. Tempest for help with the X-ray measurements. D. G. Wickham, Inorg. Synth., 1967, 9, 152. R. Bock, A Handbook of Decomposition Methoh in Analytical Chemistry (International Textbook Co., London, 1979), p.83, and references 4.317-4.322 therein. D. Bradbury, M. G. Segal, R. M. Sellers, T. Swan and C. J. Wood, unpublished results. (a) E. Peligot, Ann. Chim. Phys., 1844, 12, 533; (b) E. Pkligot, Ann. Chim. Phys., 1845, 14, 240.1164 METAL-OXIDE DISSOLUTION KINETICS (a) H. W. Swank and M. G. Mellon, Znd. Eng. Chem., Anal. Ed., 1938, 10; (b) W. Klump and H. Busch, Mitt. Ver. Grosskesselbesitzer, 1962, 81, 433; (c) A. L. Wilson, Analyst, 1964,89, 402; ( d ) J. A. Tetlow and A. L. Wilson, Analyst, 1964, 89, 442. J. A. McKeague and J. H. Day, Can. J. Soil Sci., 1966, 46, 13. T. Tamagawa, Kokai, 1974, 74-14,629. B. A. Zabin and H. Taube, Znorg. Chem., 1964, 3, 963. lo N. Valverde, Ber. Bunsenges. Phys.Chem., 1976, 80, 333. '* D. Bradbury, in Water Chemistry of Nuclear Reactor Systems (British Nuclear Energy Society, l2 M. G. Segal and R. M. Sellers, J . Chem. SOC., Chem. Commun., 1980, 991. l 3 M. J. Pryor and U. R. Evans, J. Chem. SOC., 1950, 1259. l4 S. Haruyama and K. Masamura, Corros. Sci., 1978, 18, 263. l5 D. G. Wickham, E. R. Whipple and E. G. Larson, J. Znorg. Nucl. Chem., 1960, 14, 217. l6 Powder Diffraction File(Joint Committee on Powder Diffraction Standards, Swarthmore, Pennsylvania, l7 Handbook of Chemistry and Physics, ed. C. R. Weast (Chemical Rubber Co., Cleveland, Ohio, 50th l8 (a) R. C. Mercier and M. R. Pans, C.R. Acad. Sci., 1964,259,2445; (b) R. C. Mercier, M. Bonnet and Is J. D. Russell, R. L. Parfitt, A. R. Fraser and V. C. Farmer, Nature (London), 1974, 248, 220.2o R. J. Atkinson, R. L. Pafitt and R. St. C. Smart, J. Chem. SOC., Faraday Trans. I , 1974, 70, 1472. 21 R. L. Parfitt, R. J. Atkinson and R. St. C. Smart, Soil Sci. SOC. Am. Proc., 1975, 39, 837. 22 J. D. Russell, E. Paterson, A. R. Fraser and V. C. Farmer, J. Chem. Soc., Faraday Trans. 2,1975,71, 23 R. L. Padtt, J. D. Russell and V. C. Farmer, J. Chem. SOC., Faraday Trans. I , 1976, 72, 1082. 24 R. L. Parfitt and R. St. C. Smart, J. Chem. SOC., Faraday Trans. I , 1977, 73, 796. 25 R. L. Parfitt, V. C. Farmer and J. D. Russell, J. Soil Sci., 1977, 28, 29. 26 A. F. Wells, Structural Inorganic Chemistry (Clarendon Press, Oxford, 4th edn, 1975), p. 490. 27 G. M. Rosenblatt, in Treatise on Solid State Chemistry, ed. N. B. Hannay (Plenum Press, New York, 28 A. F. Wells, Structural Inorganic Chemistry (Clarendon Press, Oxford, 4th edn, 1975), p. 259. 28 H.-J. Engell, 2. Phys. Chem. (N.F.), 1956, 7 , 158 (CE-trans. 6891). 30 D. A. Vermilyea, J. Electrochem. Soc., 1966, 113, 1067. 31 N. Valverde and C. Wagner, Ber. Bunsenges. Phys. Chem., 1976, 80, 330. 32 J. W. Diggle, in Oxides and Oxide Films, ed. J. W. Diggle (Marcel Dekker, New York, 1973), 33 M. Simnad and R. Smoluchowski, J. Chem. Phys., 1955, 23, 1961. 34 (a) K. Azuma and H. Kametani, Trans. Metall. SOC. AZME, 1964, 230, 853; (6) H. Kametani and 35 I. H. Warren, M. D. Bath, A. P. Prosser and J. T. Armstrong, Znst. Min. Metall. Trans., Sect. C, 1969, 36 R. M. Cornell, A. M. Posner and J. P. Quirk, J. Znorg. Nucl. Chem., 1976, 38, 563. 37 (a) C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, J. Chem. SOC., Faraday Trans. I , 1977, 73, 1710; (b) C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, J. Chem. SOC., Faraday Trans. I , 1978, 74, 1615, 1624; (c) R. L. Segall, R. St. C. Smart and P. S. Turner, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2907. ' M. Miyazaki, M. Amemiya, Y. Sat0 and T. Takamura, Kokai, 1972, 72-25,073. London, 1978), p. 373. 1974). edn, 1969), p. B232. M. R. Pans, Bull. SOC. Chim. Fr., 1965, 2926, 3527. 1623. 1976), vol. 6A, p. 165. vol. 2, p. 281. K. Azuma, Trans. Metall. SOC. AZME, 1968, 242, 1025. 78, 21. 38 K. Sangwal, J. Muter. Sci., 1980, 15, 237 and references therein. 38 Th. Proske, Ch-H. Fischer, M. Gratzel and A. Henglein, Ber. Bunsenges. Phys. Chem., 1977,81,816. 40 M. A. J. Rodgers, D. C. Foyt and Z. A. Zimek, Radiat. Res., 1978,75, 296. 41 C. J. Wood, D. Bradbury, T. Swan, M. G. Segal and R. M. Sellers, U.K. Patent Appl. 8,000,584,1980. 42 Y. L. Sandler, Corrosion, 1979, 35, 205. 43 D. Bradbury, M. G. Segal, R. M. Sellers, T. Swan and C. J. Wood, in Water Chemistry of Nuclear Reactor Systems 2 (British Nuclear Energy Society, London, 1981), p. 403. (PAPER 1 /756)