Solid State Sciences 5 (2003) 327–334 www.elsevier.com/locate/ssscie

Structure of the Fe(II-III) layered double hydroxysulphate green rust two from Rietveld analysis Lilian Simon a , Michel François b , Philippe Refait a , Guillaume Renaudin b , Michèle Lelaurain b , Jean-Marie R. Génin a,∗ a Laboratoire de chimie physique et microbiologie pour l’environnement, UMR 7564 CNRS, Université Henri Poincaré, Nancy 1, équipe microbiologie et

physique and département matériaux et structures, ESSTIN, 405 rue de Vandoeuvre, 54600 Villers-Lès-Nancy, France b Laboratoire de chimie du solide minéral, UMR 7555 CNRS, Université Henri Poincaré, Nancy 1, BP 239, 54506 Vandoeuvre-Lès-Nancy, France

Received 17 September 2002; received in revised form 4 October 2002; accepted 9 October 2002

Abstract Synthetic samples of the iron(II-III) hydroxysulphate known as green rust two were obtained by aerial oxidation of iron(II) hydroxide precipitates and studied using chemical and thermal analyses, transmission Mössbauer spectroscopy and powder X-ray diffraction. The III ¯ ideal formula is FeII 4 Fe2 (OH)12 SO4 ·∼8H2 O. The structure is trigonal, P 3m1 with cell parameters a = 0.55241 nm, c = 1.10113 nm, 3 III 2+ and V = 0.29097 nm and Z = 1/2. It is characterised by the succession of positively charged hydroxide sheets [FeII 4 Fe2 (OH)12 ] 2− negatively charged interlayers composed of the sulphate anions and water molecules, [SO4 ·∼8H2 O] . These interlayers are made of two planes of H2 O and SO2− 4 √, in contrast with those found in the rhombohedral green rust one compounds, which are made of a single plane. A superstructure (a = a0 3) is found along the [110] direction of the parent hexagonal unit cell, where a0 is the lattice parameter of Fe(OH)2 , and due to an ordering of the sulphate anions in the interlayers.  2002 Éditions scientifiques et médicales Elsevier SAS. All rights reserved. Keywords: Iron(II-III) hydroxysulphate; Layered double hydroxide; Green rust two; Mössbauer spectroscopy; Rietveld analysis; X-ray diffraction

1. Introduction Iron(II-III) hydroxysalts commonly named green rusts (GRs) belong to the family of layered double hydroxides (LDH), i.e. of divalent-trivalent ions minerals, which are characterised by a crystal structure that consists of the stacking of brucite-like layers carrying a positive charge and layers constituted of anions and water molecules. Their structure and composition depend upon the specific anions they incorporate. For instance, GRs were initially classified by Bernal et al. [1] in two types on the basis of X-ray diffraction (XRD) main features: green rust one, GR1, obtained with “planar” anions and green rust two, GR2, obtained with three-dimensional anions. Several GR1s are 2− known, e.g. GR1(Cl− ), GR1(CO2− 3 ) and GR1(SO3 ), and their crystal structure is considered to be close to that III of pyroaurite MgII 6 Fe2 (OH)16 CO3 ·4H2 O as determined by * Corresponding author.

E-mail address: [email protected] (J.-M.R. Génin).

Allmann [2], but, up till now, only two GR2 compounds have 2− been found, GR2(SO2− 4 ) and GR2(SeO4 ) [3], and their crystal structure is not yet determined. Thus, the structure of GR1s is now well established and consists of the stacking of hydroxide-like sheets of Fe(OH)2 which alternate regularly with interlayers composed of anions and water molecules; it follows the stacking sequence AcBiBaCjCbAk. . . where A–C are OH− layers, a–c Fe cations layers and i–k the interlayers. The case of GR1(Cl− ) has been carefully studied recently by refined X-ray powder ¯ diffraction data [4] that led to a rhombohedral lattice R 3m with parameters in the conventional hexagonal cell equal to a = 0.31901 and c = 2.3856 nm. In contrast, the assumption that the structure of a GR2, e.g. GR2(SO2− 4 ), could have the same layer stacking sequence as that of a GR1, e.g. GR1(Cl− ), seemed highly doubtful and the lack of a refined model for that type of structure was stringent. Moreover, the chemical composition of GR2(SO2− 4 ) was not ascertained and various proposals were made, mainly by Schwertmann and Fechter [5] and Génin et al. [6–8]. In both cases, it was observed that the Fe(II)/Fe(III) ratio was equal to 2, but the

1293-2558/02/$ – see front matter  2002 Éditions scientifiques et médicales Elsevier SAS. All rights reserved. doi:10.1016/S1293-2558(02)00019-5

328

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

Fe/S ratio was estimated at 1 and 2, respectively. Similarly, the number of water molecules present in the interlayers is unknown, although this information could help to determine the crystal structure. This paper is intended to ascertain the chemical composition of GR2(SO2− 4 ) compound and to devise a model of its structure based on Rietveld analysis.

2. Experimental 2.1. Sample preparation GR2(SO2− 4 ) is usually prepared by aerial oxidation of iron(II) hydroxide suspensions in the presence of a slight excess of dissolved iron(II) sulphate [1,9,10]. Typically, a solution of 0.2 mol l−1 NaOH is added to a solution of 0.12 mol l−1 FeSO4 ·7H2 O and Fe(OH)2 is precipitated leaving an excess of 0.02 mol l−1 of dissolved iron(II) sulphate in solution. Alternative ways for preparing GR2(SO2− 4 ) were developed and described recently [6,8]; one of them was retained here since it allows to control and vary the sulphate concentration independently of the iron concentration [6]. This experimental approach originated from the selectivity that GRs and related compounds exhibit with respect to different anions, implying in particular that divalent anions are always preferred to monovalent [11,12]. Iron(II)-chloride (FeCl2 ·4H2 O from Prolabo®, purity 98% minimum) was then used to precipitate iron(II)-hydroxide and the sulphate ions were provided as an additional 0.02 mol l−1 Na2 SO4 solution. Magnetic stirring (∼500 rpm) in the open air ensured a progressive homogeneous oxidation of the precipitate and a thermostat controlled the temperature, which was kept at 25 ± 0.5 ◦ C. Reactions were monitored by recording the potential E of a platinum electrode immersed in solution, using the saturated calomel electrode as a reference (but all potentials in the following refer to the standard hydrogen electrode) and by measuring the total concentrations of dissolved Fe and S species. 2.2. Analyses of dissolved species 2 ml of the suspensions were sampled at regular intervals from the solution during the reaction. They were filtered on a ground-glass filter with AP15 paper from Millipore. Nitric acid was then added to the solutions for avoiding the precipitation of iron(III) oxyhydroxides, which could result from oxidation of Fe(II) in neutral or basic solutions, and for dissolving the last possible traces of colloidal compounds. The resulting solutions were then diluted and analysed by inductively coupled plasma-emission spectrometry (ICPAES). Analyses were done at 238.204 nm for iron species and at 180.731 nm for sulphur species, using a Perkin-Elmer Emission Spectrometer 2000. The solutions were pumped to the plasma with peristaltic pumps at a rate of 1 ml min−1 . Four measurements were performed for each species in all

Fig. 1. Evolution with time of the electrode potential E h and the concentrations of dissolved Fe and S species during the formation and oxidation of GR2(SO2− 4 ). (a) E h expressed with respect to the standard hydrogen electrode (SHE) vs. time curve. (b) Concentrations vs. time curves.

samples. In order to optimise the precision, concentrations were kept larger than 300 µg l−1 for iron species and larger than 5000 µg l−1 for sulphur species. The calibration was made with five standard solutions, made of both FeCl2 and Na2 SO4 . 2.3. Transmission Mössbauer spectroscopy (TMS) Mössbauer spectra of GR2(SO2− 4 ) were obtained by use of a constant-acceleration Mössbauer spectrometer with a 50 mCi source of 57 Co in a Rh matrix. The spectrometer was calibrated with a 25 µm foil of α-Fe at room temperature. Since GR2(SO2− 4 ) is very sensitive to oxidation, the spectra were measured at 78 K. Once the formation of the GR compound was achieved, an event which was accurately detected on the E vs. time curve (cf. Fig. 1a), the precipitate was filtered under inert atmosphere, set in the sample holder and introduced in the cryostat for Mössbauer measurements, where the oxidation was definitively stopped. 2.4. Differential scanning calorimetry (DSC) The determination of the water content in the GR2(SO2− 4 ) compound was performed by means of differential scanning calorimetry method. The experiment, using a MettlerToledo DSC 30 module, was realised with an open crucible

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

under flowing argon. These experimental conditions maintained the powder at an atmospheric pressure, which allowed us to make calculation using the standard thermodynamic data. The DSC measurements were carried over a temperature range from −100 ◦ C to +250 ◦ C with a heating rate of 5 ◦ C min−1 in order to observe the less bounded water molecules departure. The aluminium crucible was hermetically sealed under a nitrogen atmosphere, then was pierced some few seconds before dropping the temperature at −100 ◦ C for avoiding the oxidation of the GR. 2.5. X-ray data collection A powder sample was introduced in a Lindeman capillary for avoiding the degradation of the GR, which is sensitive to the air, but also for minimising the effect of preferential orientation which affects usually X-ray powder data. An XRD pattern shown in Fig. 2 was recorded in the 1–50◦ two-theta range using an automatic diffractometer with Bragg-Brentano geometry and Mo Kα1 radiation (λ = 0.070930 nm). In order to increase the peak intensity/background ratio, 15 patterns were summed using the DIFFRAC-AT program [13] where each pattern was measured by a two-theta step of 0.02◦ during 2 seconds per step. The resulting pattern was suitable for Rietveld analysis. Nevertheless, some crystallisation defects, as revealed by the breadths of the lines which were much more important than those of diamond used as a standard, residual preferred orientation due to an anisotropic shape and the low dimension of GR2(SO2− 4 ) crystallites (plate-like crystals with average dimensions of 60 × 60 × 30 nm3 observed by TEM) would probably decrease the accuracy of the results. The asymmetry and breadth of the (001) reflection due to the fact that it is close to the origin (2θ = 3.64◦ ) was eliminated during the last steps of the refinement.

329

3. Chemical composition 3.1. Analysis of the solution The potential E vs. time curve (Fig. 1a) is composed of three plateaux A, B and C separated by two sharp jumps at points T g and T f . These sharp variations indicate the end of a reaction stage. During the first one ending at T g , iron(II) hydroxide oxidises into GR2(SO2− 4 ) and during the second one, from T g to T f , GR2(SO2− 4 ) oxidises into γ FeOOH and α-FeOOH. More details concerning these oxidation processes can be found elsewhere [6,10]. Analyses of the concentrations in dissolved Fe and S species were performed by ICP-AES at various times of the reaction. The curves obtained for the [Fe] and [S] evolutions with time (Fig. 1b) are identical. Both species are consumed before T g , while the sulphate-containing GR2(SO2− 4 ) is forming from Fe(OH)2 ; then, they are released in solution when GR2(SO2− 4 ) transforms into ferric oxyhydroxides. It can be noted that the [Fe] and [S] concentrations after precipitation are lower than the final ones, implying that a non-negligible amount of GR2(SO2− 4 ) has already formed. This confirms observations made by Mössbauer spectroscopy indicating that about 15% of GR are already present right after precipitation [4]. The total dissolved [Fe] and [S] amounts involved in reactions are [Fe] = (15.5 ± 2) × 10−3 mol l−1 and [S] = (18.5 ± 2) × 10−3 mol l−1 . They are consistent with [Fe] = [S] = (17 ± 3) × 10−3 mol l−1 , and consequently with a chemical composition of III FeII 4 Fe2 (OH)12 SO4 ·nH2 O, according to the overall balance, with [Fe(OH)2 ] = 0.1 mol l−1 , [Fe2+ ] = 0.02 mol l−1 and −1 [SO2− 4 ] = 0.02 mol l : 5Fe(OH)2 + Fe2+ + SO2− 4 + 1/2O2 + H2 O III → FeII 4 Fe2 (OH)12 SO4 III FeII 4 Fe2 (OH)12 SO4

(1)

+ 3/4O2

→ 5FeOOH + Fe2+ + SO2− 4 + 7/2H2 O

(2)

GR2(SO2− 4 )

was often admitted Such a composition for [6,14] since it corresponds to the case where the excess positive charges due to Fe(III) cations are exactly counterbalanced by the negative charges carried by the sulphate ions. It would actually indicate that, like GR1 compounds, GR2 compounds are characterised by the succession of positive hydroxide layers and negative anion-containing interlayers. 3.2. Differential scanning calorimetry

Fig. 2. Powder XRD pattern of GR2(SO2− 4 ) for Rietveld analysis (λ = 0.0709030 nm).

The DSC curve (Fig. 3) shows two endothermic peaks corresponding to departures of chemically different water molecules. The powder obtained after this experiment was analysed by X-ray diffraction. The only crystallised phase detected in the powder pattern was goethite, the α-form of iron oxyhydroxide FeOOH, which showed an hydroxyl condensation before 250 ◦ C. Thus, the first peak at 117 ◦ C is

330

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

Fig. 3. Differntial scanning calorimetry (DSC) measurement of GR2(SO2− 4 ) over a temperature range of −100 ◦ C to +250 ◦ C (heating rate of 5 ◦ C min−1 ).

assigned to the vaporisation of slightly bonded water molecules present in the interlayers of the structure. Therefore, the second peak at 153 ◦ C corresponds to the condensation of hydroxyl groups in the hydroxide sheets. Thermodynamic data [15], including standard heat of formation and heat capacities of liquid and gas water, were used to determine the 117 ◦C = 2232 J g−1 ) vaporisation enthalpies at 117 ◦ C (Hvap. ◦ 153 C = 2152 J g−1 ). The total energy and at 153 ◦ C (Hvap. consumed between 50 ◦ C and 200 ◦ C is 534 J g−1 . The first peak area corresponds to 81.4% of this value. The number of water molecules per unit formula of GR2(SO2− 4 ) can be determined. The endothermic enthalpy of 435 J g−1 at 117 ◦ C shows that the sulphated GR contains 19.5 weight % of slightly bonded water (435/2232 × 100 = 19.5). This content of water corresponds to 8.5 water molecules for 6 iron atoms and the following chemical formula of GR2(SO2− 4 ) III (OH) SO ·8.5H O. The value 8.5 Fe can be written FeII 12 4 2 4 2 must be considered as an estimate of the exact value of the number of water molecules since it is deduced from pure water vaporisation equilibrium. The DSC measurement also indicates at 153 ◦ C the condensation of four hydroxyl groups for a departure of two water molecules ([99/2152] × 100 = 4.6 weight % which corresponds to two H2 O). 3.3. Mössbauer spectroscopy analysis The Mössbauer spectrum of the precipitate obtained at T g , GR2(SO2− 4 ), is displayed in Fig. 4 and compared to that of a typical GR1 compound, e.g. GR1(Cl− ) [16]. Using one unique line width for the Lorentzian-shaped lines made the computer fittings optimised. Three quadrupole doublets were necessary in the case of the GR1 compound. They are named D1 , D2 , D3 and correspond to quadrupole splitting ∆ of 2.80, 2.55 and 0.38 mm s−1 , respectively, whereas isomer shifts δ taking α-iron for reference are 1.26, 1.27 and 0.48 mm s−1 . D1 and D2 are attributed to Fe(II) sites and D3 to

Fig. 4. Transmission Mössbauer spectrum of GR2(SO2− 4 ) measured at 78 K compared to that of GR1(Cl− ) (Refait and Génin, 1997 [17]).

an Fe(III) site. In contrast, two quadrupole doublets D1 and D3 are sufficient when dealing with GR2(SO2− 4 ) (cf. [8]): D1 and D3 correspond to ∆ of 2.83 and 0.45 mm s−1 , respectively, whereas δ are 1.27 and 0.47 mm s−1 . D1 and D3 are attributed here to Fe(II) and Fe(III) sites, respectively; their abundance ratio [Fe(II)]/[Fe(III)] is found equal to 2 by computing the ratio of the doublet intensities D1 /D3 . No Lamb–Mössbauer f factor correction was tempted. The existence of various ferrous sites D1 and D2 in GR1 compounds was interpreted as related to the disordered nature of the interlayers [4,17]. Although there is only one crystallographic site for iron atoms in GR1, some of them are surrounded by the anions of the interlayers and others by water molecules. The presence of only one Fe(II) site in GR2(SO2− 4 ) indicate here an ordered organisation of the sulphate ions and water molecules inside the interlayers [8].

4. Crystal structure 4.1. Rietveld analysis A Rietveld analysis of the GR2(SO2− 4 ) structure is realised by using FULLPROF program [18]. Experimental de-

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

331

Table 1 Experimental details and refinement parameters of Rietveld analysis a (nm) b (nm) c (nm)

0.55241 0.55241 1.10113

α (◦ ) β (◦ ) γ (◦ )

90 90 120 ¯ P 31m

Space group Volume (nm3 )

0.29097

Measurement time Angular range, 2θ (◦ ) Wavelength (nm) Points

Addition of 15 spectra of 50 min each = 12.5 hours 1–50 λ = 0.07093 4111

Atomic parameters Profile parameters Cell parameters

9 6 2

RBragg =

i=n



 obs  i=n Iiobs Ii − Iicalc /

i=1 i=n

 i=n obs calc RF = Fiobs Fi − Fi / i=1 i=1 1=n

 i=n obs calc Rp = Yiobs Yi − Yi / i=1 i=1 i=n

2 i=n 2 obs calc Rwp = w i Y i − Yi / wi Yiobs i=1 i=1  i=n
2 1/2 Rexp = (N − P + C) wi Yiobs i=1

0.11

i=1

tails are reported in Table 1. Trigonal lattice parameters were first refined using UFIT program [19] from 26 Bragg peak maxima: a = 0.555242√nm, c = 1.10116 nm. A superlattice is observed (a = a0 3) along the [110] direction of the parent hexagonal unit cell where a0 is the lattice parameter of Fe(OH)2. This latter, which is described in the trigonal ¯ ¯ P 31m space group, led to choose P 3m1 as the space group of the GR2 structure. 4.1.1. Position of iron and hydroxyl groups Making the assumption√ that the reason for observing the superstructure (a = a0 3) comes from the ordering of Fe(II) and Fe(III) cations, the starting positions of iron and ¯ hydroxyl groups in the P 3m1 symmetry are the following: 2+ Fe ion at site [2c] (2/3, 1/3, 0); Fe3+ ion at site [1a] (0, 0, 0) and OH− at site [6k] (2/3, 0, 0.10). This description has the advantage of respecting the [Fe(II)]/[Fe(III)] ratio of 2 determined from Mössbauer spectroscopy. Refinement of this partial model led to a reliability factor R Bragg of 0.30. 4.1.2. Position of water molecules Taking into account the large distances of 1.10 nm between two adjacent main layers described above III + [FeIII 2 Fe (OH)6 ] , and the high water content into 2− GR2(SO4 ), it was justified to position the water molecules along two parallel sheets, one at level z = 1/3 and the other one at z = 2/3, level zero being occupied by the main lay-

0.071 0.061 0.085 0.094

ers. A model with H2 O molecules at site [6k] (x, 0, z = 1/3) was tested. The R Bragg factor decreases to 0.14 but the temperature factor of H2 O molecules became unrealistic at a large value of B = 0.20 nm2 . Water molecules are then placed at the general site [12 l] at initial level z = 1/3 and the occupancy factor which was left free converged to 0.353 corresponding to 4.23 water molecules per unit cell which matches very well with the value of 4.25 obtained by DSC. Moreover, a satisfactory and realistic temperature factor value of 0.081 nm2 for water molecules was obtained, but it remained large compared to those of atoms in the hydroxide sheets Fe3 (OH)6 . The disordered water molecules can be considered as relatively “mobile” inside the structure. 4.1.3. Position of the sulphate groups Constraints were applied during the refinement to fix the distances S–O at a value of 0.15 nm and angles O–S–O at 109.5◦. Moreover, the z coordinates of each atom inside the sulphate group (OA and OB which are the apex and basis of the tetrahedra respectively, and S atom) were refined but constrained to move together. Thus, only one parameter, the z level, determined the position of the whole sulphate group along the c axis. The occupancy factor was assumed half a SO2− 4 per unit cell. Four models taking into account the two possible orientations of sulphate group (up and down) and two positions in

332

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

Table 2 Tested positions of the sulphate group in the GR2(SO2− 4 ) structure; model 3 is that retained 1 2 3 4

O2− A

O2− B

Orientation

B (nm2 )

R Bragg

(2/3, 1/3, 0.783)

(0.923, 1/3, 0.609)

up to Fe2+

0.0209

0.14

(2/3, 1/3, 1/2)

(0.923, 1/3, 0.682)

down to Fe2+

−0.0444

0.14

(0, 0, 0.778)

(0.256, 0, 0.596)

up to Fe3+

0.0147

0.11

(0, 0, 1/2)

(0.256, 0, 0.682)

down to Fe3+

−0.0093

0.11

S6+

Model

2d (z = 0.647) 2d (z = 0.647) 2c (z = 0.641) 2c (z = 0.637)

Table 3 II III + − Refined atomic parameters for GR2(SO2− 4 ) unit cell [Fe2 Fe (OH)6 ] [1/2SO4 ·4H2 O] using model 3, which is preferred Atom

Site

X

Y

Z

B (nm2 )

Occupation

Atoms/cell

Fe2+ Fe3+ OH− H2 O

2c 1a 6k 12 l

2/3 0.0 0.3250 −0.2346

1/3 0.0 0.0 0.6193

0.0 0.0 0.07289 0.6633

0.00885 0.00885 0.01195 0.081

1.0 1.0 1.0 0.353

2.0 1.0 6.0 ≈ 4.0

S6+ O2− A

2e 2e

0.0 0.0

0.0 0.0

0.6415 0.7785

0.0147 0.0147

0.25 0.25

0.5 0.5

O2− B

6k

0.256

0.0

0.5965

0.0147

0.25

1.5

(a, b) plane, one below Fe3+ , the other below Fe2+ cations, have been tested. They are summarised in Table 2. Model 3 led to the best refinement: the R Bragg factor decreased down to 0.11 and the temperature factor of sulphate group was refined to 0.0147 nm2 . Model 1 and 2 led to higher R Bragg values of 0.15, meaning that the sulphate group was obviously not below the Fe2+ cations. Although model 4 led to the same R Bragg value as model 3, the refined temperature factor of sulphate group got an unrealistic negative value and model 3 was thus retained for the GR2 structure. 4.1.4. Refinement details Twenty parameters were refined, i.e. atomic (9), profile (6), lattice (2), zero, scale factor and one for preferred orientation. The (001) reflection close to the origin was eliminated from the refinement as its large asymmetry was not well corrected by the program, and the region between 11.5 and 14◦ (2θ ) where two non identified Bragg peaks are detected, probably due to an impurity, is also excluded. 4.2. Description of the structure The observed, calculated and difference patterns are shown in Fig. 5 and the refined atomic parameters of hydroxIII 2+ 2− are ysulphate green rust [FeII 4 Fe2 (OH)12 ] [SO4 ·8H2 O] reported in Table 3. A projection view on (001) plane of the GR2 structure is shown in Fig. 6. For convenience, water molecules distributed statistically at [12 l] positions, atoms (OA , S) and OB of the sulphate group at [2e] and [6k] positions, respectively, are ordered into four lattices (two along a and two along b) in this representation. Thus, the sulphate

Fig. 5. (a) Observed, calculated (model 3) and, (b) Difference XRD patterns of GR2(SO2− 4 ) (λ = 0.0709030 nm).

groups are alternatively oriented up and down along [110] direction of the (2a, 2b, c) lattice. Water molecules in the interlayer, situated at site [12 l], are represented in Figs. 6 and 7 by removing those which are at a distance lower than 0.28 nm from oxygen atoms belonging to hydroxyl groups of the main layer, to sulphate groups or to other water molecules of the interlayer. Thus, 3.5 oxygen atoms per lattice on the average can be positioned at site [12 l], assuming reasonable close packing. That value is slightly lower but close to the value of 4.2 obtained independently from Rietveld analysis and DSC measurements. A general view of this ordered model showing the stacking of the main layers and interlayers is shown in Fig. 7. The corresponding selected interatomic distances are reported in Table 4. The structure consists of positively charged hydrox-

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

333

III + ide sheets [FeII 2 Fe (OH)6 ] separated by about 1.10 nm; 2+ 3+ the Fe and Fe cations have regular octahedral hydroxylco-ordination with Fe–O distances of 0.203 and 0.1966 nm respectively. Apical atoms OA of sulphate group approach Fe3+ cations at a distance of 0.245 nm. The negatively charged interlayer [1/2SO4 ·4H2 O]− is constituted by a double sheet of water molecules in which sulphate anions are inserted, in register directed by their apex towards the Fe3+ cations. The connection between the main layers and interlayers is realised by hydrogen bonds between hydrogen atoms of the hydroxyl group and oxygen atoms OA of the sulphate group as confirmed by OA –OH distances of 0.23 nm. The connections inside an interlayer are insured by hydrogen bonds between water molecules and between water molecules and oxygen atoms of the sulphate group. The O–O distances in the 0.297–0.313 nm range observed in the interlayer are relative to these hydrogen bonds.

5. Discussion Fig. 6. Projection of an ordered representation of the (2a, 2a, c) lattice of the GR2(SO2− 4 ) structure on the (001) plane.

Fig. 7. General view of an ordered representation of the crystal structure of GR2(SO2− 4 ). Table 4 Interatomic distances in GR2(SO2− 4 ) according to model 3 Atoms

Distance (nm)

Atoms

Distance (nm)

Fe(II)–OH− Fe(III)–OH− Fe(III)–OA OH− –OH−

6 × 0.20304 6 × 0.19664 1 × 0.2455 2 × 0.24079 1 × 0.25129 2 × 0.31104

OH− –OA H2 O–OA H2 O–H2 O H2 O–OB H2 O–OH− S–OA S–OB

1 × 0.2434 1 × 0.2977 1 × 0.2984 1 × 0.30684 1 × 0.31314 1 × 0.1500 3 × 0.1500

The crystal structure of GR2(SO2− 4 ) differs from the pyroaurite-like structure of GR1 compounds by three specific features. The first one relates to interlayers separating hydroxide sheets, which are composed in GR2 of two adjacent planes of anions and water molecules, instead of a single one as it is in GR1s. An elementary stratum is then described by a stacking sequence AcBij. . . , where A and B are planes of OH− ions, c those of Fe cations and i, j those of interlayers. Water molecules of i and j planes are located close to A and B positions [(∼ A) and (∼ B)], respectively, leading to an almost hexagonal close packed stacking of oxygen atoms AcB(∼ A)(∼ B) . . . In contrast, an elementary stratum of GR1s is described as AcB(∼ B) . . . , where water molecules in interlayers lie vertically above OH− ions of adjacent hydroxide layers [4]. The second feature relates to the arrangement of the successive strata. It is obviously a consequence of the position of the intercalated water molecules. In GR1s, the AcB(∼ B) . . . sequence implies a three-layer repeat AcB(∼ B)BaC(∼ C)CbA(∼ A)A . . . yielding a rhombohedral structure. The two layer repeat AcB(∼ B)BcA (∼ A)A . . . also exists as found in sjögrenite as proposed by Ingram and Taylor [20], but is generally realised at high temperatures. In GR2, the AcB(∼ A)(∼ B) . . . sequence induces only one single-layer repeat with a primitive hexagonal cell. As already stressed by Bookin et al. [21,22], the geometry of sulphate ions is not suitable for the AcB(∼ B) . . . elementary sequence. It is not the case with the anions usually found in GR1s, typically CO2− 3 , which can be placed in the interlayers amongst the water molecules in such positions that their oxygen atoms occupy sites similar to those of H2 O, which is lying almost vertically above the OH− ions of the hydroxide sheets [2,20]. This can also be achieved with various planar anions, e.g. C2 O2− 4 [23] and with simple spherical an-

334

L. Simon et al. / Solid State Sciences 5 (2003) 327–334

ions, e.g. Cl− [4,24]. The tetrahedral shape of SO2− 4 does not fit to such an organisation, involving another one. The twoplane interlayers of GR2(SO2− 4 ) illustrate how these geometric constraints can be accommodated. √ Finally, a superlattice with unit cell parameter a = a0 3 is observed here. Since the scattering factors of Fe(II) and Fe(III) are almost identical, a cation ordering alone could not have resulted in the appearance of new reflections. Thus, the superstructure mainly comes from an anion ordering of SO2− 4 in the interlayers, as already observed for sulphate [21, 22] and benzoate hydrotalcites [25]. The sulphate ions are actually placed so that one of their oxygen atoms, shifted from the plane of water molecules, points towards one cation site, and the two-plane organisation allows to have some of them pointing up and others pointing down. But since SO2− 4 anions are pointing towards a specific cation site accompanied with an ordering, it is physically sound to assume that this site is occupied by Fe(III) cations, since it would increase the coulombic interactions between layers. This assumption is strongly supported by Mössbauer spectroscopy, which reveals only one site for Fe(II), in contrast with what is observed for GR1s, which implies that the local environment of the Fe(II) cations is never affected by the negative charges carried by the sulphate ions. Nevertheless, the connection between Fe(III) and sulphate positions does not result necessarily in a perfect long range ordering of the Fe cations. Since half of the Fe(III) sites does not correspond to a sulphate anion (the ratio Fe(III)/S is 2), some Fe(II) cations occupy such sites, whereas some Fe(III) occupy Fe(II) sites, provided that not a sulphate lie above an Fe(II), which would result in the usual quadrupole doublet D2 observed in the Mössbauer spectra of other GRs. It has been demonstrated, in the case of hydrotalcite, the Mg(II)–Al(III) hydroxycarbonate [26], that there exists a local order amongst the cations such that Al(III) cations are never neighbour of one to each other. But there is no long range ordering, and this might be also the case for GR2. Finally, it must be kept in mind that the two-plane interlayer structure described here is not the only one that is reported for sulphate containing layered hydroxides. Various authors mentioned sulphate hydrotalcite or sulphate pyroaurite with XRD patterns testifying of the presence of the usual single plane interlayer [27,28]. The main diffraction line, corresponding to the Fe plane to Fe plane distance, was then found at about 0.86 to 0.87 nm, a distance shortened of ∼0.24 nm compared to that of 1.1 nm characterising GR2. 6. Conclusion It has been demonstrated that, contrarily to the oftenadmitted generalisation, some GR compounds were not described by the rhombohedral three-layer structure of pyroau-

rite. Those green rust 2 compounds, GR2(SO2− 4 ) and related tetrahedral anion containing GRs such as GR2(SeO2− 4 ) [3] are based on a single-layer repeat structure with primitive hexagonal cell involving interlayers composed of two distinct planes of water molecules and sulphate anions. But, like GR1s, GR2s consist of brucite-like iron hydroxide layIII 2+ ers [FeII 4 Fe2 (OH)12 ] , positively charged due to Fe(III) cations, which alternate with negatively charged interlayers [SO4 ·8H2 O]2− that maintain the whole electroneutrality. This general description could constitute the definition of the green rust family.

References [1] J.D. Bernal, D.R. Dasgupta, A.L. Mackay, Clay Miner. Bull. 4 (1959) 15. [2] R. Allmann, Acta Crystalogr., Sect. B 24 (1968) 972. [3] J.-M.R. Génin, M. Abdelmoula, Ph. Refait, L. Simon, Hyperfine Interact. (C) 3 (1998) 313. [4] Ph. Refait, M. Abdelmoula, J.-M.R. Génin, Corros. Sci. 40 (1998) 1547. [5] U. Schwertmann, H. Fechter, Clay Miner. 29 (1994) 87. [6] J.-M.R. Génin, A.A. Olowe, Ph. Refait, L. Simon, Corros. Sci. 38 (1996) 1751. [7] Ph. Refait, C. Bon, L. Simon, G. Bourrié, F. Trolard, J. Bessière, J.-M.R. Génin, Clay Miner. 215 (1999) 313. [8] A. Géhin, C. Ruby, M. Abdelmoula, O. Benali, J. Ghanbaja, Ph. Refait, J.-M.R. Génin, Solid State Sci. 4 (2002) 61. [9] J. Detournay, L. De Miranda, R. Dérie, M. Ghodsi, Corros. Sci. 15 (1975) 295. [10] A.A. Olowe, J.-M.R. Génin, Corros. Sci. 32 (1991) 965. [11] S. Miyata, Clays Clay Miner. 31 (1983) 305. [12] A. Mendiboure, R. Schöllhorn, Rev. Chim. Miner. 23 (1986) 819. [13] J. Nusinovici, M.J. Winter, Advances in X-ray analysis 37 (1994) 59. [14] H.C.B. Hansen, O.K. Borggaard, J. Sørensen, Geochim. Cosmochim. Acta 58 (1994) 2599. [15] O. Kubaschewski, C.B. Alcock, in: Metallurgical Thermochemistry, 5th edition, G.V. Raynor (Ed.), in: International Series on Materials Sciences and Technology, Vol. 24, Robert Maxwell, 1979. [16] Ph. Refait, J.-M.R. Génin, Clay Miner. 32 (1997) 597. [17] Ph. Refait, J.-M.R. Génin, Corros. Sci. 39 (1997) 539. [18] J. Rodriguez-Carjaval, FULLPROF: A program for Rietveld refinement and profile matching analysis of complex powder diffraction patterns, Laue-Langevin Institute, Grenoble, France. [19] M. Evain, UFIT program-1.3 version, Institut des Matériaux de Nantes, France, 1992. [20] L. Ingram, H.F.W. Taylor, Mineral. Mag. 36 (1967) 465. [21] A.S. Bookin, V.A. Drits, Clays Clay Miner. 41 (1993) 551. [22] A.S. Bookin, V.I. Cherkashin, V.A. Drits, Clays Clay Miner. 41 (1993) 558. [23] Ph. Refait, A. Charton, J.-M.R. Génin, Europ. J. Solid State Inorg. Chem. 35 (1998) 655. [24] R.S.W. Braithwate, P.J. Dunn, R.G. Pritchard, W.H. Paar, Mineral. Mag. 58 (1994) 79. [25] M. Vucelic, G.D. Moggridge, W. Jones, J. Phys. Chem. 99 (1995) 8328. [26] M. Vucelic, W. Jones, G.D. Moggridge, Clays Clay Miner. 45 (1997) 803. [27] S. Miyata, A. Okada, Clays Clay Miner. 25 (1977) 14. [28] H.C.B. Hansen, R.M. Taylor, Clay Miner. 26 (1991) 311.