Synthesis and Móssbauer characterization of Cu and Cr doped nnagnetites C *

A detailed Móssbauer investigation of magnetites prepared under different hydrothermal conditions and doped with Cu and Cr is presented. The samples were characterized by means of room temperature Móssbauer spectrometry, infrared spectroscopy, and X-ray diffraction. Móssbauer results show that the hydrothermal method produces highy stoichiometric and relativelly well-crystallized magnetites. The results suggest that the best samples are obtained when the alkaline solution is added quickly to the ferrous solution. It was also found that mixing the solutions under constant ultrasonic stirring produce magnetites with slightly better crystallinity and stoichiometry than the samples produced under magnetic or nitrogen bubbling stirring. The effect of the Cu and Cr on the hyperfine parameters is also presented and discussed.


INTRODUCTION
Magnetite, Fe304, is one of the most important products of the corrosion of iron and steeF' .It is usually found in the nonstoichiometric form.Besides, due to the presence of some alloying elements in the steel, such as Cu and Cr, it is also expected to find doped magnetites in the corrosion products^^'^l In order to understand the origin of these oxides in the corrosion of steel, it is of primary importance to fully characterize their structural properties.Because the corrosion products usually consist of complex combinations of several iron oxides and oxihydroxides, it is easier to characterize the magnetites prepared in the laboratory.In that respect the hydrolysis method is interesting, because in that way it is possible to simulate the corrosion process under atmospheric conditions.The results obtained from such investigations should lead to an analytical framework for the characterization of magnetites originating from corrosion products.P.M. MoNTOYA, L. OsoRio, R.E.VANDENBERGHE AND J.M. GRENECHE Magnetite has the inverse spinel structure.The unit cell contains 8 Fe^^ ions on the tetrahedral sites (or A sites) and 16 Fe ions on the octahedral sites (or B sites), 8 of which are Fe ^ and the other 8 are Fe^"^.The chemical formula of non-stoichiometric magnetite can be written as Fe^^^O^y where the oxidation parameter x can, in principle, vary continously from zero (stoichiometric magnetite) to 1/3 (maghemite).However, this formula can better be represented as (Fe3^)A[Fe2"(i.3,)Fe^"(H2xA]BO%,where D stands for a vacancy.This latter formula assumes both that there are no vacancies on the A sites, and that the rapid hopping of electrons between the Fe "^ and Fe ions occurs only on the B sites, leading to the well known "Fe^'^^" ionic state and leaving some Fe ^ ions on the B sites^ .The chemical formula can also be rewritten as (Fe^^)A[Fe^* ^(2-6x)F^^^5xnx]BO^ 4-Room temperature Mossbauer spectrometry can be used to determine the oxidation parameter x.In order to calculate it, the Mossbauer spectra are usually fitted with two sextet components.The first sextet, which will be called Fe component, is due to the contributions coming from both (Fe^'^)^ and [Fe 5JB, which have very similar hyperfine fields^"^" \ The second sextet comes from the [Fe (2.6X)]B ^ons, and therefore will be called the Fe " ^ component.The ratio R of the subspectral area of the Fe component, A(Fe^"^), to the subspectral area of the Fe • ^ component, A(Fe^ ), can be written as: In this equation we have considered the well known relation A(Fe^j) • /(Fe^j), in which (Fe^j) is the number of iron atoms of a given i valence state, belonging to a given j site, and /(Fe^j) is the corresponding Mossbauer fraction.Additionally, we have considered the reported values of [/(Fe^\) lf(¥e'-'\) ] = [f{¥e'\) lf{Ve'-'%) ] = 1.06 for sotichiometric magnetite^ .By solving x from equation (1) we get the following relation, which can be used to determine the oxidation parameter: On the other hand when Cu or Cr replaces Fe in the structure, Fe3,yCuy04 and Fe3,yCry04 can represent the chemical formula for the doped magnetite.
It is the purpose of the present investigation to determine the degree of stoichiometry and the crystallinity of the magnetites prepared using the hydrolysis method.Moreover, it is considered the effect of introducing some variables to this synthesis procedure in the final product.The principal technique used for this characterization is Mossbauer spectrometry, which is able to distinguish Fe atoms with different strcutural and magnetic environment and with different valence state.In this respect, the degree of stoichiometry is determined from the oxidation parameter x, and the degree of crystallinity is calculated from the values of both the magnetic hyperfine field and the peak broadening.The Mossbauer results are complemented by means of X ray diffraction and infrared spectroscopy techniques.This study also includes the effects of adding dopants to the crystalline structure on the hyperfine parameters derived from the Mossbauer spectra.

EXPERIMENTAL PROCEDURE
Eleven magnetites were considered in the present work.Ten of them were prepared following the hydrothermal method described by Schwertmann and Cornelr \ The other sample was obtained in the market (AGFA).In order to find adequate conditions of synthesis, some relevant variables have been considered in the preparation of the pure magnetites, and these are listed in table I. Besides, 1, 5 and 10 % of mole Fe, in the nominal composition, was substituted by Cu and Cr.
Room temperature (RT) Mossbauer spectra were obtained in the transmission mode using a constant acceleration drive.A Co/Rh source with initial activity of 25 mCi was used.Velocity was regularly calibrated by taking spectra of standard hematite.All absorbers were prepared by mixing the material with very pure sugar in order to achieve an homogeneous thickness of about 10 mg Fe/cm .The spectra were adjusted using programs MOSF and DIST3Et^ ^^^^ ^^\ The MOSF program is based on a nonlinear least-squares fitting procedure assuming lorentzian line shapes, whereas the DIST3E program is based on a model-independent distribution of hyperfine fields and/or quadrupole splittings.The XRD measurements were performed on a D501 Siemens diffractometer equipped with a

Sample name Synthesis conditions
The alkaline solution is added dropwise but quickly to the ferrous chloride solution (for about 5 min) under constant magnetic stirring.
The alkaline solution is added dropwise and slowly to the ferrous chloride solution (for about 15 min) under constant magnetic stirring.
Both solutions are mixed slowly (for about 15 min) under constant ultrasonic stirring.
Agitation is produced by bubbling nitrogen and the solutions are mixed slowly for about 15 min.
X mole percentage of the total ferrous chloride solution is replaced by chromium chloride solution.
X mole percentage of the total ferrous chloride solution is replaced by copper chloride solution.In order to obtain the relevant Mossbauer parameters, two fitting models were applied.In the first model, which employs the MOSF program, two symmetric lorentzian shaped sextets were used to fit the spectra of the pure and the lowest substituted ( 1 % Cu, 1% Cr) magnetites.For each sextet B/^^, 5, the intensity of the second line (3:x:l), two width parameters and the area fraction were adjusted.The quadrupole shift of both sites were fixed to zero.The two width parameters are F ^ and AT, which can be used to calculate the width of six lines of a given sextet by F] = F^ = F + 2AF, F2 = F5 = F + AF, and F3 = F4 = Fae.In the second model, which uses the DIST3E program, the spectra of the highest substituted magnetites were fitted by using the modehindependent hyperfine field distributions, assuming linear correlations in the B site between B^^on the one hand and d on the other hand.Again, the quadrupole shift of both sites were fixed to zero.

RESULTS AND DISCUSSION
Figure 1 shows the Mossbauer spectra of some selected samples.It is possible to see that the fitting model reproduces the experimental data quite reasonably as commonly found by other reports^^'^l Thus the samples seem to consist of pure magnetites.However, in order to improve the fitting procedure, it was necessary to introduce an other component to all the spectra.This component, which accounts from 1 % to 2 % of the total absorption area, has also been observed in other similar synthetic magnetites^^' "^ ^\ Owing to the small absorption, the errors related to the derived Mossbauer parameters are very high.However, for the purpose of the present investigation, this component can be treated as an impurity, without affecting the conclusions.The origin of this impurity, which was reasonably adjusted with a doublet with quadrupole splitting of 0.55 mm/s, isomer shift of 0.035 mm/s, and a peak width of 0.5 mm/s, could not be clearly established.The hyperfine parameters for this impurity phase were always kept fixed.The Mossbauer parameters of the AGFA sample, which is a well crystallized magnetite^"^^, have been used as a reference to the values determined for the other samples.
From table II, it is possible to deduce that the hydrothermal method produces relatively wellcrystallized samples.Besides, it seems that the samples prepared when the alkaline solution is added quickly to the ferrous solution, exhibit slightly better crystallinity than the samples obtained by slowly mixing both solutions.This observation is based on the fact that both the peak width and the incremental broadening of sample MCFIOOTC have slightly larger values than the ones determined for sample MCFIOO.Additionally, the latter sample is more stoichiometric than former one.These results are, as expected, in full line with the work by Schwertmann and Cornelr I Another important observation is that mixing the solutions under constant ultrasonic stirring produce magnetites with slightly better crystallinity and stoichiometry than the ones produced under magnetic or nitrogen bubling stirring.However, to our experience, the ultrasonic agitation remains more difficult to handle than the magnetic one.
The preparation of the samples in the presence of chromium or copper chloride solutions results in magnetites with poorer crystallinity.This effect is more pronounced as the amount of chromium or  copper solutions is increasing (see Table III).These are perhaps indications that copper and chromium have entered into the crystalline structure, but it is difficult to establish unambiguously at which site.The increments of Cr or Cu in the solutions also increase the presence of an impurity phase in the case of Cr, the hyperfine parameters for this phase (quadrupole splitting of about 0.5 mm/s and an isomer shift of about 0.33 mm/s) resemble those of a poorly crystalline, probably doped, goethite and/or lepidocrocite.In fact, XRD measurements and infrared spectra points to the presence of goethite in  II.Môssbauer parameters derived from the spectra using the symmetric lorentzian shaped sextets model Estimated errors are 0.1 T for B¡^f, 0.01 mm/s for ô, 0,003 for x, and 0.01 mm/s for both F and AF Tablo II.Parámetros Môssbauer derivados de los ajustes de los espectros utilizando el modelo de     the case of Cu, the existence of other phases, such as superparamagnetic and/or highly Cu doped magnetites can be proposed.Jaen et aÜ ^ have found CuFe02 in the final products of the hydrolysis of iron in the presence of Cu.However, it was not possible to detect the presence of this phase by XRD.The sample prepared under the presence of 5 wt.% Cr solution contains more impurities than the sample prepared with 5 wt.% of Cu solution.This observation is based on the larger area values determined for sample MCFCrS as compared to that value for sample MCFCuS (see Table III).In contrast, the MCFCulO has larger impurity area values than MCFCrlO.On the other hand, the magnetic hyperfine parameter for both sites as well as their corresponding standard deviations seems to be lower and larger respectively for the Cu doped magnetites than for the Cr doped magnetites.
These observations could imply that the Cu doped magnetites have poorer crystallinity than the Cr-magnetites.However, this statement should be taken with extreme precaution, because measurements are needed in order to establish both the amount of dopant incorporated into the structure and the particle sizes of the magnetites.These measurements are in progress.
Figure 2 shows the infrared spectra of some selected samples.It is possible to notice that and a fast mixing of the solutions, seem to introduce improvements in the final product.The degree of crystalhnity and the presence of impurities in the samples are greatly influenced by the presence of Cr and Cu.However, this effect seems to be opposite in each case.In that respect, the increment of the Cr solutions appears to introduce more goethite in the final product than the increment of the Cu solutions do.Nevertheless, it seems that large amount impurities are present in the solutions with the highest Cu content.On the other hand, the degree of crystalhnity of the magnetites seems to be worse in the Cu doped samples than in the Cr ones.magnetite (bands located at about 581 cm"^) is the principal phase in all the samples.The bands typical of goethite (800 cm"^ and 900 cm'^) are observed only in the highest doped samples.These results are in line with the findings by the other techniques.Another interesting feature of the infrared spectra is that the goethite bands are more pronounced in the spectra of samples with Cr than with Cu, this characteristic being more apparent for the 5 wt.% Cr than for the 10 wt.Cr.An impurity band located at about 1600 cm'^ could indicate the presence of FeCl2.The XRD patterns of the pure and the smallest substituted samples demonstrates again the presence of magnetite.The goethite peaks are only visible in the highest subtituted samples, as found by the other techniques.

Figure 1 .
Figure 1.Môssbauer spectra of some selected samples.The stars represent the experimental data and the solid lines the subspectral component.
C.A. BARRERO, A.L. MORALES, ].MAZO-ZUWAGA, F. JARAMILLO, G. PÉREZ, DM.ESCOBAR, C. ARROYAVE, ].TOBÓN, P.M. MoNTOYA, L. OsoRio, R.E.VANDENBERGHE ANDJ.M. GRENECHE Table A-Hyperfine field for the A site; ÔA: Isomer shift for the A site relative to the isomer shift of a-iron at room temperature; FA: width of the third peak, AFA: incremental broadening of the peaks; B/^fg: Hyperfine field for the B site; 6B-Isomer shift for the B site relative to the isomer shift of a-iron at room temperature; Fg: v\^idth of the third peak; AFg: incremental broadening of the peaks; x: oxidation parameter.* These are not x values but the R parameters.A: subspectral area belonging to the impurity doublet.
Hyperfine field of maximum probability for the A site; ÔA: Average isomer shift relative to the isomer shift of a-iron at room temperature; OA: standard deviation of the hyperfine field distribution; B^fB-Hyperfine field for the B site; ÔB: Isomer shift corresponding to the maximum hyperfine field for the B site relative to the isomer shift of a-iron at room temperature; o^,: standard deviation of the hyperfine field distribution; R: area ratio; A: subspectral area belonging to the impurity doublet.

Table III .
Môssbauer parameters derived from the spectra using the model-independent hyperfine field distribution model Tabla III.Parámetros Môssbauer derivados de los ajustes de los espectos utilizando el método de la distribución de campos hiperfinos independiente de un modelo teórico