3240
Langmuir 1997, 13, 3240-3246
Structure and Mechanisms of Formation of FeOOH(NO3) Oligomers in the Early Stages of Hydrolysis Je´roˆme Rose,† Alain Manceau,‡ Armand Masion,† and Jean-Yves Bottero*,† Laboratoire des Ge´ osciences de l′Environnement, URA 132 CNRS et Universite´ Aix-Marseille III, CEREGE; BP 80, Europole Me´ diterrane´ en de l′Arbois, 13545 Aix-en-Provence Cedex 4, France; and Groupe de Ge´ ochimie de l′Environnement, LGIT-IRIGM, Universite´ J. Fourier et CNRS, BP 53X, 38041 Grenoble Cedex 9, France Received December 6, 1996. In Final Form: March 10, 1997X The partial hydrolysis of ferric nitrate solutions (0.1 M) and the structural evolution of the smallest oligomers were studied by using X-ray absorption spectroscopy at aging times between 1 h and 20 days at room temperature. The samples evolve differently with time, depending on the hydrolysis ratio. At the same aging time t, the number of iron atoms in the atomic environment of the central atom increases with hydrolysis ratio, but the growth does not follow a defined pathway. Only N3 (the number of iron atoms around each central atom at a distance of ≈3.50 Å) has a monotonous evolution with time. It seems that the growth of nuclei is associated with the increase of the amount of double-corner sharing. Edgesharing linkages between Fe octahedra are associated with the early steps of the nucleation, and singlecorner linkages are present in all samples. In the case of the hydrolysis of ferric nitrate salts, the local structure of oligomers does not prefigure the structure of the well crystallized infinite phase as opposed to ferric chloride salts and Cr.
Introduction The polymerization of metal ions (Al, Fe, Cr etc..) via hydrolysis has received considerable attention in the past 10 years in various domains such as catalyst synthesis, environmental science or industry, where non-crystalline particles play a key role in the adsorption and transportation of organic or inorganic pollutants.1,2 Recently, a large number of studies have been specifically devoted to sol-gel systems containing Al or Cr.2-8 Generally, the hydrolysis of metals leads to small clusters (i.e. “sol”) which aggregate and then form an infinite amorphous “gel” phase. One of the most interesting problems is the mechanism of formation of well crystallized phases from small polymers or gel phases. A generalized approach through the electrostatic field theory6 gave leads for the understanding of the formation of small clusters but cannot be extended to infinite phases. For example, in the case of the Al(III) hydrolysis, the Al13 polycation is formed from monomeric, dimeric,4,5 and trimeric precursors.6 In the second step, the Al13 polycations aggregate to form crystalline Al(OH)3 via a solid-state transformation.7 The early stages of hydrolysis of Cr(III) involve the nucleation of dimers, trimers, and tetramers. These small polymers then condense into transitory Cr polycations where Cr atoms share a common hydroxo bridge. Further evolution of these transition phases occurs through * To whom correspondence should be addressed. † URA 132 CNRS et Universite ´ Aix-Marseille III. ‡ Universite ´ J. Fourier et CNRS. X Abstract published in Advance ACS Abstracts, April 15, 1997. (1) Buffle, J.; Van Leeuwen, H. P. In Sampling and Characterization of Environmental Particles; Lewis Publishers: Chelsea, 1993; p 554. (2) Brinker, C. J.; Scherer, G. W. Sol-gel science: the physics and chemistry of sol-gel processing; Academic Press: San Diego, 1990; p 908. (3) Van Beek, J. J.; Seykens, D.; Jansen, J. B. H.; Schuiling, R. D. J. Non-Cryst. Solids 1991, 134, 14-22. (4) Akitt, W.; Lester, L.; Kandelwal, F. H. J. Chem. Soc. 1972, 26, 609-612. (5) Bottero, J. Y.; Tchoubar, D.; Cases, J. M.; Poirier, J. E.; Fiessinger, F. J. Phys. Chem. 1980, 2933-2939. (6) Henry, M.; Jolivet, J. P.; Livage, J. In Aqueous Chemistry of Metal Cations: Hydrolysis, Condensation and Complexation; Structure and Bonding 77; Springer Verlag: Berlin, 1992. (7) Bottero, J. Y.; Axelos, M. A. V.; Tchoubar, D.; Cases, J. M.; Fiessinger, F. J. Colloid Interface Sci. 1987, 117, 47-57. (8) Stunzi, H.; Marty, W. Inorg. Chem. 1983, 20, 2145-2150.
S0743-7463(96)02079-3 CCC: $14.00
intramolecular condensation associated with deprotonation. In this final state, Cr atoms share a common tetracoordinated oxo ligand.8 More recent studies on partially hydrolyzed FeCl3‚6H2O solutions (0.1 M) confirm previous findings concerning the formation of a polycation 1.5 nm in size9 and containing 24 Fe atoms.10 This polymer has a β-FeOOH-like local structure with a cavity 0.6-0.7 nm in diameter similar to that in β-FeOOH crystals.10 This cluster is formed in three steps when the hydrolysis ratio (n ) [OH]/[Fe]) increases. The first step corresponds to the formation of a Fe dimer where two iron octahedra share one edge. The second step consists in linking a third iron octahedron to the dimer through double-corner sharing. The condensation of these trimers which results in the Fe24 polycation corresponds to the third step. Unlike Al13, the small Fe24 polymers formed in fresh partially hydrolyzed FeCl3 sols prefigure the structure of the well crystallized infinite phase, viz. the akaganeite. The Cr(III) hydrolysis products also prefigure the infinite phases, but with slower kinetics.8 The nature of the anion controls the structure of the oxyhydroxides of metal ions. It is particularly true for Fe(III). In the case of nitrate salts, the size of the polymers and the structure of the aggregates and precipitates seem to evolve continuously with the hydrolysis ratio.11 General models of nucleation have been proposed.12,13 The Fe octahedra are thought to be linked by condensation of dimers or trimers through hydroxo and oxo bridges. These models are mainly based on a concept of similarity between the “structure” of the polymers and the structure of the resulting crystalline phase. This is valid in a thermodynamic quasi-equilibrium state. When the system is far from equilibrium, very little is known about the (9) Tchoubar, D.; Bottero, J. Y.; Quienne, P.; Arnaud, M. Langmuir 1991, 7, 398-402. (10) Bottero, J. Y.; Manceau, A.; Villieras, F.; Tchoubar, D. Langmuir 1994, 10, 316-319. (11) Bottero, J. Y.; Tchoubar, D.; Arnaud, M.; Quienne, P. Langmuir 1991, 7, 398-402. (12) Van der Woude, J. H. A.; Rinjbout, J. B.; De Bruyn, P. L. Colloids Surf. 1984, 11, 391-400. (13) Schneider, W. Hydrolysis of Iron (III). Chaotic olation versus nucleation. In Comments in Inorganic Chemistry; Gordon and Breach Science Publishers: New York, 1984; pp 205-223.
© 1997 American Chemical Society
Formation of FeOOH(NO3) Oligomers
Langmuir, Vol. 13, No. 12, 1997 3241
structure of the Fe species in the early stages of the hydrolysis of ferric nitrate. The present work aims at investigating the very first steps of nucleation in partially hydrolyzed acidic ferric nitrate solutions through their local structure by using X-ray absorption spectroscopy in order to propose a mechanism of formation of oligomers or polymers. Materials and Methods Sample Preparation. The samples were prepared by dissolving Fe(NO3)3‚9H2O (O.2 M) in HNO3 (0.1 M). The hydrolysis was carried out by slow addition of NaOH under vigorous stirring to prevent important local supersaturation with respect to FeOOH. At the end of the partial neutralization, the concentration of Fe is 0.1 M. Samples at n ) [NaOH]/[Fe] ) 1.5, 2.0, and 2.5 were studied at aging times between t ) 30 min and 20 days (t ) 0 corresponds to the end of the NaOH addition). Well crystallized R-, β-, and γ-FeOOH samples were used as references. Experimental Method. X-ray absorption experiments at the Fe K-edge were performed using the LURE synchrotron source (Orsay, France). The positron storage ring was running at 1.85 GeV and 280-330 mA. EXAFS data reduction was accomplished according to a procedure described previously.14 A Kaiser window was used for Fourier transforms.15 Thus obtained radial distribution functions (RDFs) are uncorrected for phase shift, leading to a shift of the peaks by 0.3-0.4 Å compared to crystallographic distances. The modeling of experimental spectra was carried out by using theoretical backscattering amplitudes and phase shifts16 for the Fe-O atomic pair. For the Fe-Fe atomic pair, the amplitude function was determined experimentally from γ-FeOOH (6 Fe atoms at 3.08 Å) and a theoretical phase shift function16 was used. The first atomic shell around the Fe absorber was fitted using the electron mean free-path parameters derived from the analysis of γ-FeOOH for oxygen atoms whereas R-, β-, and γ-FeOOH were used as references for neighboring shells. Details on the structural analysis of iron oxides by EXAFS spectroscopy have been reported elsewhere.17
Results EXAFS Spectra. EXAFS spectra for samples at different hydrolysis ratios (n ) 1.5, 2.0, 2.5) and at aging times (t) from 30 min to 20 days are shown in Figure 1. Several wave beatings can be observed at k ) 6.65 Å-1 for n ) 1.5 and 2.0, close to 8 Å-1 for n ) 2.5, and at 8.6 Å-1 for n ) 1.5 and 2.0. Differences between the spectra at the same n value become significant for t g 6 days. The scans were performed over a high-energy range, which allowed us to extract EXAFS spectra with kmax ) 14 Å-1. This kmax value sets the distance resolution to 0.12 Å according to18 ∆R ) π/2k. This means that two backscatterer atoms of the same type (two iron or two oxygen atoms for example) at distances R1 and R2 from the central atom can be distinguished by EXAFS analysis only if |R1 - R2| > 0.12 Å. Radial Distribution Functions (RDFs). The first peak on the RDF (uncorrected for phase shift) (Figure 2) corresponds to oxygen atoms (O, OH, or H2O groups) in the first coordination sphere. The presence of a set of peaks at higher distances (2.6-2.8 Å, 3.0-3.4 Å, 3.4-3.6 Å), even for low hydrolysis ratios (n ) 1.5 (Figure 2a)), indicates the presence of atoms in next nearest atomic shells of the iron octahedra. For n ) 1.5 and 2.0 (Figure (14) Manceau, A; Calas, G. Clay Miner. 1986, 21, 341-360. (15) Bonnin, D.; Calas, G.; Suquet, H.; Pezerat, H. Phys. Chem. Miner. 1985, 12, 55-64. (16) McKale, A. G.; Veal, B. W.; Paulikas, A. P.; Chan, S. K.; Knapp, G. S. J. Am. Chem. Soc. 1988, 110, 3736. (17) Manceau, A.; Drits, V. A. Clay Miner. 1993, 28, 165-184. (18) Teo, B. K. EXAFS: Basic principles and data analysis; Springer Verlag: Berlin, 1986; p 349.
Figure 1. Fe k3χ(k) spectra for liquid samples: (a) n ) 1.5; (b) n ) 2.0; (c) n ) 2.5. Arrows indicate phase shifts when t increases.
2a and b), the maximum of the second and third peaks is shifted to shorter distances with time. Discussion Analysis of the First Coordination Sphere. Partial EXAFS spectra, obtained by back-Fourier transform of the first peak of the RDF, are compared with calculated spectra (Figure 3). Corresponding structural parameters are listed in Table 1. Calculations were made using one or two atomic shells when a beat node appeared on the partial EXAFS spectra. For all samples iron octahedra are surrounded by six oxygen atoms at an average distance of 2 ( 0.06 Å from the central atom. This result is in agreement with a previous analysis of the ligand coordination sphere of iron in several Fe(OH)x precipitates.19 For the samples at t ) 20 days, the Fe-O distance is shorter than those for the less aged samples (≈1.95 Å vs 2.00 Å). Analysis of the Atomic Environment of the Iron Octahedra. Back-Fourier transforms of the 2.3-3.9 Å (19) Combes, J. M.; Manceau, A.; Calas, G.; Bottero, J. Y. Geochim. Cosmochim. Acta 1989, 53, 583-594.
3242 Langmuir, Vol. 13, No. 12, 1997
Rose et al.
Figure 3. Comparison of partial EXAFS spectra corresponding to the first coordination sphere: (a) n ) 1.5; (b) n ) 2.0; (c) n ) 2.5. Solid line: experimental spectrum. Dotted line: calculated spectrum. Figure 2. Fe radial distribution functions (uncorrected for phase shifts) for (a) n ) 1.5, (b) n ) 2.0, and (c) n ) 2.5. Arrows indicate distance modifications when t increases.
range of the RDF are shown in Figure 4. They are compared with calculated curves whose structural parameters are listed in Table 2. The three peaks in this region of the RDF are not well enough separated to be analyzed separately (Figure 2). The curve resulting from the back-Fourier transform of three RDF peaks represents the sum of three elementary sinusoidal curves. Thus, in our case, the calculation of partial EXAFS spectra using three atomic shells appears reasonable. However, the use of a fourth atomic shell (Fe-Fe4 contribution) allowed not only to significantly decrease the mean square residual but also to obtain a better fit of beat nodes at high k values (Figure 4).
Nevertheless, the accuracy in the determination of the number of iron atoms at distances larger than 3.8 Å from the central atom is questionable. For such interatomic distances the signal/noise ratio decreases rapidly, leading to a higher uncertainty concerning the number of iron atoms detected by EXAFS analysis. The presence of the poorly defined peak at 3.4 Å (distance uncorrected for phase shift) on the RDF can also be due to multiple scattering effects. This aspect will be discussed below. Using a polyhedral approach consisting in comparing metal-metal distances of known and unknown oxides,20 it is possible to determine the type of linkage between the Fe octahedra for all four Fe-Fe contributions detected in our samples. The first Fe-Fe contribution (Fe-Fe1), with (20) Manceau, A.; Combes, J. M. Phys. Chem. Miner. 1988, 15, 283295.
Formation of FeOOH(NO3) Oligomers
Langmuir, Vol. 13, No. 12, 1997 3243
Figure 4. Comparison of partial EXAFS spectra corresponding to the second to fourth peaks of the RDF: (a) n ) 1.5; (b) n ) 2.0; (c) n ) 2.5. Solid line: experimental spectrum. Dotted line: calculated spectrum.
Thus, interatomic distances around 2.95 Å cannot be clearly attributed to one or the other type of linkage. The second Fe-Fe contribution (Fe-Fe2) with an interatomic distance between 3.06 and 3.20 Å can be associated to an edge sharing between two iron octahedra.19,22 The third contribution (Fe-Fe3) at 3.40-3.52 Å is attributed to a double-corner sharing according to previous results.10,19 The last contribution (Fe-Fe4) corresponds to a bond between two iron octahedra sharing one corner.17,23 It has to be pointed out that this last Fe-Fe4 contribution was not detected in hydrolyzed ferric chloride solutions.10,19 The partial EXAFS curves (corresponding to the 2.33.9 Å range of the RDF) of hydrolyzed ferric nitrate solutions were compared to the spectra of a well crystallized iron oxide (hematite) and iron oxyhydroxides (R-, β-, γ-FeOOH, Ferrihydrite). The phases and amplitude shapes of the samples are different from those of the reference minerals (Figure 5). This strongly suggests that oligomers, resulting from partial hydrolysis of ferric nitrate salts, do not have the structure of one of these reference minerals. Only in the case of n ) 2.5 and t ) 20 days, does the curve corresponding to a linear combination of partial EXAFS spectra of the references, viz. (1.8χ(goethite) + 1.0χ(lepidocrocite))/((1.8 + 1) × 2.5) (Figure 6), give a satisfactory fit to the experimental signal. Taking into account the two coefficients used, it seems that the average atomic structure of clusters corresponds to a mixture of goethite and lepidocrocite phases, with a majority of goethite-like local structure. The amplitude was divided by 2.5 to obtain a better fit between the two curves. This indicates that the central atom is surrounded by less atoms than in the reference minerals. For all other samples (n < 2.5, t < 20 days), the EXAFS spectra could not be modeled by a combination of well crystallized mineral references, thus indicating a chaotic nucleation of the oligomers. Fe-Fe4 Contribution. The lepidocrocite-like local structure in one sample (n ) 2.5, t ) 20 days) supports the presence of single-corner linkages between iron octahedra. Among all reference minerals used, only lepidocrocite exhibits this type of linkage.23 In order to validate or invalidate the presence of single-corner bonds in our samples, a multiple-scattering approach had to be developed. This approach was carried out for n ) 1.5 and t ) 20 days using the Feff 601 code.24 This ab initio code allows the recalculation of the general shape of χ(k) by determining separately the contribution of each multiplescattering path and considering only those which exceed a given threshold. This code can only be used when the exact atomic coordinates of a given cluster are known. Interatomic distances and the number of neighboring atoms around the central atom are derived from EXAFS data. For the n ) 1.5 and t ) 20 days sample, the average stoichiometry around each iron atom is given in Tables 1 and 2: 6.0 oxygen atoms at a distance of 2.00 Å, 1.9 iron atoms at 3.04 ( 0.07 Å, 1.2 iron atoms at 3.42 Å, and 0.5 iron atoms at 3.92 Å (Tables 1 and 2). The Fe-Fe1 and Fe-Fe2 contributions were treated together and were considered as edge-sharing links between Fe octahedra. Indeed, no clear attribution to a given type of linkage can
an interatomic distance ranging from 2.85 to 2.98 Å, is sometimes difficult to attribute to a given type of Fe-Fe linkage. The characteristic distances between two iron octahedra sharing one face or sharing one edge are in the 2.85-2.94 Å or 2.95-3.10 Å R-range, respectively.21,22
(21) Christensen, J.; Christensen, A. M. Acta Chem. Scand. 1978, A32, 87-88. (22) Combes, J. M.; Manceau, A.; Calas, G. Geochim. Cosmochim. Acta 1990, 5, 1083-1091. (23) Wells, A. F. Structural Inorganic Chemistry; Clarendon Press: New York, 1984. (24) Rehr, J. J.; Zabinsky, S. I.; Albers, R. C. Phys. Rev. Lett. 1992, 69, 3397.
Table 1. Structural Parameters for Fe (Backscatterer in the First Coordination Sphere) Contributions Derived from EXAFS Analysisa Fe-O1 shell sample
t
Rb
n ) 1.5
6 days 20 days 6 days 20 days 6 days 20 days
2.00 2.00 2.01 1.96 2.00 1.95
n)2 n ) 2.5
Nc
σd
6.0 6.0 6.0 4.1 6.0 2.0
0.09 0.09 0.1 0.08 0.11 0.07
Fe-O2 shell R
N
σ
2.10
2.5
0.06
2.06
4.0
0.09
Q 0.007 0.003 0.010 3 0.003 0.047
a
The electron mean-free path λ ) 2k/L, where L ) 2.5 Å-2 (determined from Lepidocrocite) was chosen for the Fe-O atomic pair. b R is the distance between the two atoms of each atomic pair, given in Å. c N is the number of atoms in the first sphere of iron. d σ is the Debye-Waller factor (Å).
Q)
∑[(k χ 3
theo)
- (k3χexp)]2/(k3χexp)2
3244 Langmuir, Vol. 13, No. 12, 1997
Rose et al.
Table 2. Structural Parameters for Fe (Backscatterer in the Second and Third Coordination Spheres) Contributions Derived from EXAFS Analysisa Fe-Fe1 shell
Fe-Fe2 shell
Fe-Fe3 shell
Fe-Fe4 shell
sample
t
RFe1b
NFe1c
σd
RFe2
NFe2
σ
RFe3
NFe3
σ
RFe4
NFe4
σ
Q
n ) 1.5
1h 1 day 6 days 20 days 1h 1 day 6 days 20 days 1h 1 day 6 days 20 days
2.85 3.05 2.98 2.97 2.96 2.96 2.98 2.98 3.02 2.87 2.96 2.97
0.20 0.90 0.95 0.90 0.65 0.55 1.10 1.00 0.90 0.45 0.55 0.25
0.09 0.09 0.09 0.09 0.09 0.09 0.08 0.08 0.09 0.08 0.08 0.08
3.06 3.20 3.14 3.11 3.11 3.09 3.14 3.12 3.12 3.06 3.07 3.06
0.85 0.50 1.10 1.00 1.05 1.10 1.35 1.05 1.25 1.75 1.60 1.40
0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.08 0.09 0.09
3.52 3.49 3.42 3.42 3.50 3.48 3.43 3.46 3.51 3.51 3.48 3.49
0.60 1.50 1.35 1.20 1.15 1.35 1.50 1.65 0.55 1.00 1.50 2.00
0.08 0.10 0.10 0.10 0.09 0.10 0.10 0.10 0.08 0.08 0.10 0.11
3.95 3.94 3.93 3.92 4.05 3.96 3.93 4.00 3.89 3.96 3.91 3.97
0.40 0.50 0.50 0.50 0.30 0.45 0.60 0.20 0.50 0.30 0.30 0.40
0.08 0.08 0.07 0.07 0.08 0.08 0.07 0.07 0.08 0.08 0.07 0.07
0.006 0.013 0.070 0.030 0.020 0.005 0.060 0.050 0.005 0.005 0.002 0.005
n)2
n ) 2.5
a γ-FeOOH was used as a reference for F -2 Fe-Fe. For this sample σ ) 0.08 (Å), L ) 1.7 Å , N ) 6, R ) 3.08 Å, and Q ) 0.002. Fe-Fe1, Fe-Fe2, Fe-Fe3, and Fe-Fe4 correspond to the first, second, third, and fourth Fe-Fe shells, respectively. b R is the distance between the two atoms of each atomic pair, given in Å. c N is the number of atoms in the second and third spheres of iron. d σ is the Debye-Waller factor (Å).
be made for the Fe-Fe1 contribution. Furthermore, the difference between the Fe-Fe1 and Fe-Fe2 distances (0.14 Å) is very close to the limit of resolution under our experimental conditions (0.12 Å). Therefore, no distinction was made between these two contributions. A large number of clusters can be developed on the basis of this structural information. The cluster presented in Figure 7 only represents a tentative 3D structural solution. It consists of five iron octahedra (called Fe5 cluster) linked to other clusters of the same type through single-corner sharing (atoms Fei and Fem bound to Fey and Fex, respectively (Figure 7)). Coordination numbers (N) of Fe atoms in this Fe5 structure (Table 3) take into account the presence of surrounding clusters and are close to the N values determined from the experimental EXAFS data. Thus this type of structure accounts for the EXAFS data and can be tested by the Feff601 code. This ab initio code also requires Debye-Waller factor (σ) values for the different paths. For the Fe-O and FeFe atomic pairs, the values chosen correspond to those determined from the EXAFS spectra of the mineral references: σ(Fe-O) ) 0.090 Å and σ(Fe-Fe) ) 0.085 Å (Tables 1 and 2). Five theoretical EXAFS spectra were calculated. Each one of them considered a different Fe of the Fe5 cluster as central atom and included also atoms Fex and Fey (Figure 7) to take into account the presence of surrounding clusters. The average of the five spectra corresponds to the Fe K-edge spectrum of this theoretical Fe5 cluster. Except some amplitude differences at low k values, experimental and calculated EXAFS spectra are similar (Figure 8a). Region 1 of the RDF (see arrows in Figure 8b), i.e. 2.4-3.35 Å distances uncorrected for phase shift, corresponds to Fe-Fe1, Fe-Fe2, and Fe-Fe3 contributions but also to Fe-Ox contributions (oxygen atoms in the second and more distant coordination spheres at distances of 3.4-3.7 Å from the central atom) as well as multiplescattering paths (path A and/or B in Figure 7b). The calculated curve does not fit well the experimental RDF in this R-range. This is due to the fact that the Fe-Ox contributions are not directly determined by EXAFS data. They depend not only on Fe-Fe distances but also on the octahedron geometry and the Fe-O-Fe angles which are not precisely known. Thus, in this study, since some parameters needed to set the exact spatial position of oxygen atoms are not known, it is not possible to propose a 3D oligomer structure fitting properly this region 1 of the RDF. In region 2 of the RDF, whose peak is associated with the Fe-Fe4 contribution but also with a multiple-scattering path (path D in Figure 7b), the adjustment between
experimental and modeled curves is quite good (Figure 8b and c). This multiple-scattering path Fei-Oy-Fey (Figure 7b), which corresponds to 40% of the total amplitude of the partial EXAFS curve, exists only when two irons share one corner. Since the Fe-O-Fe angle is close to 180°, the effective distance of this multiplescattering path is close to the Fe-Fe4 distance (3.94 Å instead of 3.88 Å in our model). Thus, the multiplescattering phenomenon resulting from the Fe-Fe4 linkage amplifies this Fe-Fe4 contribution. The importance of multiple scattering when the angle between three atoms is close to 180° was also observed by other authors.25,26 Thus even if multiple scattering is an important phenomenon in region 2 of the RDF (Figure 8a), it supports the presence of an Fe-Fe4 contribution, i.e. a single-corner linkage between two iron octahedra. Evolution of Structure of Oligomers with Time. Samples at n ) 1.5. Considering the total number of iron atoms in the neighborhood of the central atom, it is possible to obtain qualitative information about the size of oligomers. For the mineral references used, where the size of the “polycations” is considered as infinite, the total coordination number (N1 + N2 + N3 + N4) corresponding respectively to NFe1, NFe2, NFe3, and NFe4, equals 8. For n ) 1.5 and t ) 1 h, the low (N1 + N2 + N3 + N4) ) 2 value means that polycations are small. This result is in agreement with the previous study concerning the analysis of the size of clusters by small-angle X-ray scattering (SAXS).11 The size of clusters was determined to be between 7 and 9 Å. Moreover, by taking into account each coordination number separately, it is possible to obtain some information concerning the structure of oligomers present in solution for t ) 1 h. A N1 + N2 value of 1 is consistent with the formation of a majority of iron dimers sharing one edge.27 The low N3 value (0.6) suggests the presence of trimers (called trimer A), formed by doublecorner sharing of a third Fe octahedron with a dimer, as a minor species. These small oligomers (i.e. the basic units) may link each other to form larger clusters. In the case of t ) 1 h, this association of basic units leads to the formation of linear aggregates.11 The linear shape of the aggregates at a local scale could originate from magnetic dipole-dipole interactions,28 since these interactions are known to influence the aggregation of ferro-fluid colloids. (25) Den Auwer, C. Ph.D. Thesis, University Paris VI, 1995. (26) D’Day, P. A.; Rehr, J. J.; Zabinsky, S. I.; Brown, G. E. J. Am. Chem. Soc. 1994, 116, 2938. (27) Rose, J.; Manceau, A.; Bottero, J. Y.; Masion, A.; Garcia, F. Langmuir 1996, 12, 6701-6707. (28) Moors, P. M.; Botet, R.; Jullien, R. J. Phys. A: Math. Gen. 1987, 20, 975.
Formation of FeOOH(NO3) Oligomers
Langmuir, Vol. 13, No. 12, 1997 3245
Figure 6. Comparison between the partial EXAFS curve of n ) 2.5 and t ) 20 days (corresponding to the 2.3-3.9 Å region of the RDF) and that of a linear combination of (1.8 χ(goethite) + 1χ(lepidocrocite))/((1.8 + 1)2.5).
Figure 7. 3D model used for the multiple-scattering approach: (a) 3D structure; (b) several possible multiple-scattering paths. Table 3. Coordination Numbers (N) for the Fe5 Cluster (See Figure 7)
Fei Fej Fek Fel Fem (3 + 2 + 2 + 3 + 0)/5 )
Figure 5. Comparison between the partial EXAFS curve of n ) 1.5 and t ) 20 days (corresponding to the 2.3-3.9 Å region of the RDF) and that of well crystallized iron oxides: (a) Lepidocrocite; (b) Akaganeite and Goethite; (c) Ferrihydrite and Hematite.
In the case of n ) 1.5 and t ) 1 h, each basic unit acting as a magnetic dipole could align, leading to a minimization of the magnetic dipole energy. At more prolonged aging, the size of the basic units certainly becomes larger since Ntotal increases from 2 to 4. The Fe-Fe4 contribution detected in aged samples can correspond to two different mechanisms: (i) One Fe octahedron can bind dimers or trimers A through singlecorner sharing, to form other basic units (a new trimer (called trimer B) and a tetramer, respectively). In this case, the formation of single-corner linkages would be part
NFe1/Fe (3.04 ( 0.07 Å)
NFe3 (3.42 Å)
NFe4 (3.92 Å)
3 (Fej-Fek-Fel) 2 (Fei-Fel) 2 (Fei-Fel) 3 (Fei-Fej-Fek) 0 2
0 0 1 (Fem) 1 (Fem) 2 (Fek-Fel) 0.8
1 (Fey) 0 0 0 1 (Fex) 0.4
of the nucleation process. (ii) Aggregation of oligomers can result from single-corner linkages between basic units. In this case, the Fe-Fe4 contribution corresponds to the growth step. Our data do not allow us to distinguish between these two hypotheses. Samples at n ) 2. For n ) 2, interatomic distances and coordination numbers do not evolve in the same way as for the n ) 1.5 case. The evolution of RFe1, RFe2, NFe1, and NFe2 (Table 2) with time is erratic. Only NFe3 shows a steady evolution, viz. an increase from 1.1 for t ) 1 h, to 1.3 for t ) 1 day, 1.5 for t ) 6 days, and 1.6 for t ) 20 days. This increase of the number of double-corner linkages can result from a growth of the polycations.19 For t > 1 day, Ntotal is always higher than 3. This is consistent with subunits formed with more than three iron octahedra. Furthermore, a previous SAXS study11 indicates an increase of the size of subunits from 7 to 16 Å between n ) 1.5 and n )2.0. Polymers formed for n ) 2.0 are larger than those for n ) 1.5. Another difference between n ) 1.5 and 2.0 is the type of association between the subunits: for n ) 2 the aggregates are more branched.11 Samples at n ) 2.5. For n ) 2.5, the sum (N1 + N2) ) 2 suggests the formation of oligomers containing more than two iron octahedra bound through one edge. For t increasing from 1 to 6 days, N1 + N2 remains unchanged whereas, at the same time, N3 increases from 1.0 to 1.5.
3246 Langmuir, Vol. 13, No. 12, 1997
Rose et al.
and ferric chloride salts are significantly different. The two main differences observed in the case of ferric nitrate are (i) there is no nucleation and growth pathway comparable to the well reproducible and well identified stages of the hydrolysis of FeCl3 solutions (i.e. formation of edge-sharing dimers, double-corner-sharing trimers, and Fe24 polycations) and (ii) single-corner linkages are detected, whereas no evidence of this type of linkage could be found in ferric chloride samples. These differences are most likely due to the nature of anions. In the case of FeCl3 solutions, in the early stages of hydrolysis (low t values) and at low hydrolysis ratio, one or two Cl atoms remain in each iron octahedron. Since no Fe-Cl-Fe linkages exist at room temperature, nucleation sites for the polymerization of Fe are occupied. For FeNO3 solutions, since each Fe is surrounded by six oxygens (OH groups or H2O molecules), all the potential binding sites are available. The monomers have the possibility to bind one or two Fe octahedra through corners. Then the resulting structures may include single-corner bonds. The free energy of formation for a single-cornersharing linkage between two iron octahedra is slightly lower than that for a double-corner bond (≈25 kcal/mol).6 Thus, in the case of FeNO3 solutions, when single-cornersharing structures between two iron atoms are formed, they will not necessarily rearrange in order to share a second corner when t and n increase. Nevertheless at high t values, some rearrangements occur since statistically the number of single-corner linkages is lower than that of double-corner linkage types. All of these results reveal the chaotic nature of the evolution of Fe polymers during hydrolysis of ferric nitrate salts as opposed to ferric chloride salts. Conclusion
Figure 8. Comparison of (a) EXAFS spectra, (b) RDF curves, and (c) partial EXAFS spectra of region 2 of the RDF, for n ) 1.5 and t ) 20 days and for the model.
At t ) 20 days, N3 is higher than that for t ) 6 days (N3 ) 2.0), but N1 + N2 decreases from 2.0 to 1.5. Thus for n ) 2.5 the number of edge-sharing linkages between iron octahedra decreases with time while the number of doublecorner linkages increases. This result suggests that the local structure of the colloids becomes less dense, probably because of a rearrangement of the clusters following the breakage of edge-sharing bonds. Double-corner linkages allow an isotropic growth of clusters whereas edge linkages allow generally only linear or planar growth. It is interesting to note that Fe-Fe1 and Fe-Fe2 contributions are associated here with the first stages of nucleation. The size of basic units for t ) 1 h is 19 Å, and the aggregates are branched.11 Linkages between subunits could be due to single-corner sharing between two iron octahedra. Comparison with the Hydrolysis of Ferric Chloride Salts. The mechanisms of nucleation and growth of oligomers during the hydrolysis of ferric nitrate salts
In light of our results, it appears that the hydrolysis of ferric nitrate salts is a more chaotic process than that for ferric chloride. The samples at each hydrolysis ratio evolve differently with time. At the same aging time t, the number of iron atoms in the atomic environment of the central atom increases with n, but the growth does not follow a defined pathway. Only N3 has a monotonous evolution with time. It seems that the growth of nuclei is associated with the increase of the amount of doublecorner sharing. Edge-sharing linkages between Fe octahedra are associated with the early steps of the nucleation, and for all samples single-corner linkages exist. Oligomers do not display the local structure of the reference minerals, except for the most hydrolyzed and most aged sample (n ) 2.5 and t ) 20 days). In that case, the local structure of the oligomers represents a mixture of goethite- and lepidocrocite-like local structures. It appears that in the case of ferric nitrate salts the local structure of oligomers does not prefigure the structure of the well crystallized infinite phase, as opposed to the cases for ferric chloride salts11 and Cr.8 In the case of iron, the nature of the anions strongly controls the mechanisms of nucleation and growth of oligomers during hydrolysis. Acknowledgment. This work was supported by the research program INSU: Dynamique et Bilan de la Terre, Fleuve et erosion by the EEC exchange research program within the COST action #D5 “Chemistry at Surfaces and Interfaces”. The authors also wish to thank the staff at the Laboratoire pour l’Utilisation du Rayonnement Electromagne´tique, Orsay, France, and especially A. Traverse for her help during the experiments. LA962079K
