Journal of Alloys and Compounds 440 (2007)

Transcription

1 Journal of Alloys and Compounds 440 (2007) Influence of the borohydride concentration on the composition of the amorphous Fe B alloy produced by chemical reduction of synthetic, nano-sized iron-oxide particles Part I: Hematite V.G. de Resende a,b,,e.degrave a, G.M. da Costa b, J. Janssens c a Department of Subatomic and Radiation Physics, University of Ghent, B-9000 Gent, Belgium b Universidade Federal de Ouro Preto, Department de Química, Ouro Preto (MG), Brazil c Department of Chemistry, University of Antwerpen, B-2020 Antwerpen, Belgium Received 25 April 2006; received in revised form 22 August 2006; accepted 1 September 2006 Available online 30 October 2006 Abstract Amorphous Fe B alloys can be prepared at room temperature by reduction with borohydride of iron-oxide particles in suspension. By varying the borohydride concentration, amorphous Fe B alloys with boron contents between 2 and 13 at.% have been produced by reduction of synthetic (nano-sized particles) and natural (micro-sized) hematite ( -Fe 2 O 3 ) using sodium borohydride (NaBH 4 ). The results presented in this paper were obtained from a systematic study of the effect of borohydride concentration on the resulting reaction products using a variety of experimental techniques, such as X-ray diffraction, wet chemical analyses, thermal analyses, scanning electron microscopy, transmission Mössbauer spectroscopy (TMS) and integral low-energy electron Mössbauer spectroscopy (ILEEMS). Three distinct NaBH 4 concentrations have been applied. Beside unreacted hematite, amorphous Fe 1 x B x alloys have been identified from the TMS spectra recorded at various temperatures between 15 K and room temperature. The amount of Fe 1 x B x increases strongly with increasing NaBH 4 concentration, and for a given concentration with increasing specific surface area (SSA). Thermal analyses have suggested that for any given reduction condition, the boron content x in the formed amorphous alloy has a bimodal distribution. This is found to be consistent with the finding that the contribution of the Fe 1 x B x phase to the total Mössbauer spectra consists of a superposition of a broad sextet and doublet. ILEEMS has further revealed that especially the surface layers of the hematite grains are affected by the reduction processes Elsevier B.V. All rights reserved. Keywords: Hematite; Amorphous Fe B; -Iron; Chemical reduction; Mössbauer spectroscopy 1. Introduction A chemical reduction method of metal salt solutions to prepare ultrafine particles of amorphous alloys has received ample attention in recent years [1 3]. This route of preparation has several advantages with respect to traditional methods applied to obtain amorphous alloys. Among these advantages is the possi- Corresponding author at: Department of Subatomic and Radiation Physics, University of Ghent, Division NUMAT (Former Laboratory of Magnetism), Proeftuinstraat 86, B-9000 Gent, Belgium. Tel.: ; fax: address: Valdirene.Gonzaga@Ugent.be (V.G. de Resende). bility to produce the alloys in a form of particles or as ferrofluids. Sodium (NaBH 4 ) or potassium (KBH 4 ) borohydride have been commonly used in that respect, thus resulting in boron to be incorporated into the structure of produced compounds, yielding, e.g. alloys of the type TM 1 x B x in which TM is a transition metal. The value of x in this general formula depends on the borohydride molarity applied to the reaction medium and on various other reaction conditions, such as ph of the solution, addition rate of the reactants, reaction time and mixing procedure [4 9]. These parameters also determine to large extent the composition, structure, particle size and physicochemical properties of the various other phases that may constitute the obtained products /$ see front matter 2006 Elsevier B.V. All rights reserved. doi: /j.jallcom

2 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) An earlier study by the present authors [10] has shown that amorphous Fe B alloy, in addition to metallic iron, can be produced by chemical reduction of micron-sized hematite ( -Fe 2 O 3 ) particles in suspension using sodium borohydride (SBH). That study was focused on the exploration of an alternative route to obtain amorphous Fe B particles, i.e., by using iron-oxide particles in suspension instead of metal ions in solution. The amounts of hematite and SBH, were kept constant. The as-such obtained composites, as well as their derivatives from thermal treatments at selected temperatures, were thoroughly examined applying a variety of material-research techniques, including electron microscopy with EDAX and electron-diffraction facilities and 57 Fe Mössbauer spectroscopy. Prompted by the peculiar results of that initial study, the authors decided to further explore the reduction process of Feoxide particles by applying SBH to suspensions containing these particles. In contrast to the previous work the influence of the SBH concentration on the final products is investigated to some extent. Also, the authors were interested in examining the effect of the average size of the particles in the parent suspensions on the one hand, and in the process involving other types of Fe oxides on the other hand. This first part of two companion papers is concerned with the chemical reduction of micron- and nano-sized hematite ( -Fe 2 O 3 ) particles by addition of various amounts of SBH to the parent suspensions. 2. Experimental The reaction of hematite particles with SBH to produce amorphous Fe 1 x B x alloys was carried out for two different species of -Fe 2 O 3 : a natural one obtained from Quadrilátero Ferrífero, Minas Gerais state (Brazil), consisting of micron-sized particles, and a synthetic one prepared by heating synthetic goethite (see Part II) at 300 C during 2 h, with particle size of the order of 25 nm. Three intimate mixtures consisting of different weight proportions of - Fe 2 O 3 and NaBH 4 powders were used (1:1, 1:4 and 1:8, respectively). The respective suspensions were obtained by addition of 50 ml of distilled water. To each, 100 ml of a 0.2 M HCl solution was drop-wise added. After the initial reactions, amounts of NaBH 4 powder, equal to the respective amounts present in the inceptive mixtures, were added to the obtained suspensions. This was done in four consecutive steps, each one involving one-fourth of the total amount of powder to be added. Hence, the oxide:borohydride weight proportions in the final products were 1:2, 1:8 and 1:16, respectively. All reactions were started at room temperature and terminated in a time span of 30 min. The ph and the actual temperature of the reaction medium were regularly probed during that entire reaction process. Afterwards, the products were thoroughly washed with distilled water and finally dried at 70 C. X-ray diffractograms (XRD) were recorded in the range (2θ) with goniometer speed of 0.5 min 1 using an XRD-6000 Shimadzu diffractrometer equipped with a Fe K or Co K radiation and Mn or Fe filter, respectively. Counts were registered every 0.02 (2θ). Standard silicon was added in order to correct the positions of the diffraction lines and subsequently calculate the cell parameters using the well-known JADE program. Particle sizes along the [1 0 4] and [1 1 0] crystallographic directions were evaluated from the widths at one-half-maximum of the respective diffraction peaks using the well-known Scherrer formula. Analysis of iron and boron contents in the obtained products was performed by wet-chemical analysis and inductively coupled plasma (ICP), respectively, after dissolving the samples with concentrated HCl. Thermogravimetric curves between 25 and 1000 C were obtained under synthetic air (100 ml/min) using a Du Pont SDT 2960 apparatus and applying a heating rate of 5 C/min. DSC measurements in the range C, heating rate of 5 C/min and under an atmosphere of synthetic air (50 ml/min), were carried out with the powders contained in open aluminium pans. Mössbauer spectra (MS) at a number of different temperatures were collected with spectrometer operating at constant acceleration mode with triangular reference signals. 57 Co(Rh) sources were used. At room temperature, both conventional transmission Mössbauer spectroscopy (TMS spectra) and integral lowenergy electron Mössbauer spectroscopy (ILEEMS spectra) have been applied. All Mössbauer spectra were computer-analysed in terms of model-independent distributions of hyperfine parameter values and numerical data quoted hereafter refer to maximum-probability values. Isomer shifts are referenced with respect to -Fe at room temperature. As it was found that an unconstraint adjustment procedure yielded non-consistent and often unrealistic results, some plausible assumptions were implemented in the fitting model. These assumptions will be described in the next section. Scanning electron microscopy (SEM) images were collected in a JEOL apparatus with the powder samples dispersed in acetone. 3. Results 3.1. Synthetic hematite The poorly crystalline hematite (sample HS), treated with sodium borohydride as described in the previous section, gave rise to samples that are henceforward code-named BHS1, BHS2 and BHS3, respectively. In each case, the chemical reduction produced a black and highly magnetic precipitate, as experienced using a hand magnet. The apparent magnetism of the samples seems to vary with the amount of NaBH 4 used, the grains of sample BHS3 being the most attracted. Fig. 1 (top) shows that, except for the lowest SBH concentration, the ph of the reaction medium initially increases during the reduction process of the synthetic hematite. A congruent behaviour is noticed for the temperature of the medium (see Fig. 1, bottom). Observed behaviour of ph and temperature probably can be attributed to the formation of hydroborate intermediates. It was indeed suggested by Forster et al. [11] that the reaction with borohydride proceeds via iron-hydroborate intermediates of the form [(H 2 O) 5 Fe(HOBH 3 )] 2+, which presumably involve exothermal reactions, thus increasing temperature and ph of the reaction medium. However, since the reaction is extremely fast, these intermediates have not yet been isolated. The XRD pattern of sample HS exhibits only peaks characteristics of hematite and showing differential line broadening (Fig. 2). This broadening is the result of the generally poor crystallinity of hematite produced by heating synthetic goethite at low temperatures, such as 300 C [12]. The XRD of the reduced samples also show merely the diffraction lines of hematite, with at first glance similar line broadenings. The presence of any other crystalline phase is not detected, implying that the phase responsible for the colour and magnetism of the powders after reaction may be an amorphous component, most likely the Fe B alloy [10]. The cell parameters and the particle sizes of the synthetic hematite before and after the reduction with the SBH are listed in Table 1. For comparison, cell parameter values of standard hematite (Powder Diffraction File: ) have been included in the table. It is obvious that in particular the c parameters

3 238 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Table 1 Cell parameters and particles size (D hkl ) of the synthetic hematite in the studied samples Sample Cell parameter D 104 (nm) a (Å) c (Å) D 110 (nm) Hematite (standard) HS ± ± BHS ± ± BHS ± ± BHS ± ± Note: Standard hematite (PDF ). Fig. 1. Temperature (bottom) and ph (top) behavior during the reaction of reduction of the synthetic hematite. for the present hematite phases differ significantly from the standard value. These differences may be attributed to some impurities into the structure, such as OH groups, which is inherent to the applied synthesis route [13]. No clear tendencies are observed regarding the hematite particle sizes. To first order, they seem to be fairly independent of the SBH concentration. The iron and boron contents and the bulk compositions of the amorphous Fe B alloys are given in Table 2. These latter compositions were evaluated assuming that no other phases in addition to -Fe 2 O 3 and Fe 1 x B x are present, which is corroborated by the combined results of XRD and MS (see hereafter). Clearly, the global B content of the produced Fe B alloys increases with increasing amount of SBH added to the suspension containing the hematite particles. This finding is in line with the conclusion of Linderoth and Mørup concerning the composition of amorphous TM 1 x B x particles prepared by chemical reduction of TM ions by borohydride in aqueous solutions [4]. According to the literature survey reported by these authors, the concentration of the borohydride is one of two most important reaction parameters determining the composition x, the other one being the ph of the TM salt solution. The thermal behaviour, as reflected in the TGA and DSC curves, is similar for the three reduction products. An example is shown in Fig. 3, referring to sample BHS2. Following an initial weight loss below 200 C, ascribed to release of adsorbed water and possibly of structural hydroxyl groups, a two-step gain of weight is observed in the range C. The corresponding DSC signal shows that this weight gain is associated with three exothermic processes, the first one being characterized by a very broad peak that appears like a shoulder ( 275 C). The two subsequent exothermic processes take place at approximately 430 and 475 C, respectively. However, the intensity of the peak reflecting the (475 C) process in the DSC signal lowers as the borohydride concentration increases. For sample BHS3, the high-temperature peak is no longer visibly resolved from the principal peak at 430 C, implying that the associated exothermic reaction has become less important. The nature of the various chemical processes that are taking place at elevated temperatures and that are reflected in the Table 2 Boron and iron content (wt.%) and the bulk composition of the amorphous Fe B alloy Sample Boron Fe total Fe 1 x B x Fig. 2. X-ray diffraction patterns of the synthetic hematite (HS). Si corresponds to standard silicon which was added. HS ± 0.2 BHS1 2.2 ± ± 0.1 Fe 0.90 B 0.10 BHS2 2.3 ± ± 0.1 Fe 0.89 B 0.11 BHS3 2.7 ± ± 0.4 Fe 0.87 B 0.13

4 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Fig. 3. DSC curve (top) and thermogravimetric curve (solid line) and its respective derivative (dotted line, bottom) of the sample BHS2. three exothermic peaks appearing in the DSC curves, is not well understood so far. The authors tentatively explain the observations as a gradual crystallisation process (broad shoulder signal) of the amorphous Fe B phase to form crystalline Fe B with two preferential, and hence more or less distinct B contents. These crystalline Fe B phases would then subsequently oxidise at higher but different temperatures. In order to clarify the thermal analysis results two batches of sample BHS2 were submitted to thermal treatments at 400 C (TT400) and at 800 C (TT800) under static air. The XRD patterns and the Mössbauer spectra at room temperature are shown in Fig. 4. The diffractogram of sample TT400 (Fig. 4a) shows, beside the peaks of hematite and silicon, a peak at 36 (2θ) which was attributed to boron oxide (B 2 O 3 ). The hematite formed at 400 C still shows the differential line broadening observed for the untreated sample. The Mössbauer spectrum of this sample (Fig. 4b) consists of four components: (i) an outer sextet (relative area 68%), obviously due to hematite, (ii) a weak inner sextet (relative area 3%) which has hyperfine parameters characteristic of -Fe, (iii) an Fe 0 doublet which will be discussed later in this section (relative area 27%) and (iv) an Fe 2+ doublet which so far is not known (subspectrum area is merely 2%). The XRD pattern of sample TT800 presents the same crystalline phases as observed in sample TT400 (Fig. 4c). However, the diffraction lines of the hematite are much sharper, indicating that grain growth has occurred. The Mössbauer spectrum (Fig. 4d) shows the presence of two sextets: a major one which is attributed to hematite (96%) and a weak sextet with hyperfine parameters characteristics of FeBO 3 (4%). Combining the TGA and DSC results with XRD and MS observations it can be concluded that the thermal treatment at 400 C of the sample BHS2 resulted in a crystallization process and partial oxidation of the Fe B alloy to -Fe, hematite and B 2 O 3. Further heating the sample at 800 C resulted in total oxidation of the Fe B alloy and -Fe giving rise to the formation of hematite and FeBO 3. Since the present thermal analyses were merely aimed to investigate qualitatively the thermal stability of the amorphous Fig. 4. X-ray diffraction patterns (a and c) and Mössbauer spectra at room temperature (b and d) of samples TT400 (top) and TT800 (bottom). Si corresponds to standard silicon added to the sample.

5 240 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Table 3 Hyperfine parameters of the Mössbauer spectra at selected temperatures of the sample HS T (K) H hf (koe) 2ε Q (mm/s) RA (%) δ (mm/s) a a a RT The values of isomer shifts are with reference to metallic iron. Note: H hf, maximum-probability hyperfine fields; 2ε Q, quadrupole shifts; δ, isomer shifts; T, temperature; RA, relative areas. a Fixed parameter. Fe B particles in relation to the amount of SBH used for the synthesis, no further attention will be devoted to the hightemperature processes. It seems, though, that the total amount of added SBH has but little effect on the mentioned thermal stability. Mössbauer spectra of the HS sample were recorded at several temperatures between 15 K and RT and could be fitted straightforwardly. The relevant numerical results of these fits are collected in Table 3. Example MS, referring to RT and 80 K, respectively, are shown in Fig. 5. A relatively sharp sextet is obtained at RT, while at 80 K clearly two distinct sextet components are observed. The latter arise from the co-existence of antiferromagnetic and weakly ferromagnetic spin states, which is characteristic for poorly crystalline hematite [14]. For ideal species of hematite, at any given temperature only one single spin state exists at a time, with typifying values for the hyperfine parameters of its six-line MS. The transition between the two states, called the Morin transition (MT), takes place at a sharply defined temperature of around 265 K [15]. In contrast, for powders of less crystalline hematite, the MT is significantly shifted to lower temperatures and moreover spread out over a more or less broad range in which the two spin states are present simultaneously. For the present hematite sample HS the MT is observed to gradually evolve in a temperature range below 200 K and even at Fig. 5. Temperature dependence of the Mössbauer spectra of synthetic hematite. 80 K only about one-third of the sample is in the low-temperature antiferromagnetic state (see Table 3). The rather low MT temperature range observed is thus consistent with the poor crystallinity of the sample as evidenced by the small particle size. The MS at selected temperatures of the samples after treatment with NaBH 4 are reproduced in Fig. 6. In general, three components were found to be required to obtain adequate fits for these spectra: (i) an outer sextet obviously due to hematite that had not reacted, (ii) an inner very broad sextet and (iii) a doublet labelled as Fe 0. For sample BHS3, clearly an additional, but very weak ( 3% of total absorption area) and rather ill-defined doublet, found to be due to Fe 2+ (δ 1.33 mm/s and E Q 2.91 mm/s at 15 K), appears in the MS. It could, however, not be attributed to any specific phase and will be disregarded in what follows. Obviously, changing the borohydride concentration has a major influence on the MS, in particular as to the contribution of the central doublet at any given temperature T (see, e.g. RT spectra), which increases considerably with increasing concentration at the expense of the contribution due to the hematite component. Numerous trial fits were attempted in search for an adequate model to describe the experimental MS, leading to parameter values that consistently vary with temperature and with boron concentration. Most problematic in that respect was the broad sextet, which for most MS is ill-defined, and consequently adjusted Mössbauer parameters and temperature variations thereof were obviously unreal. Some firm conclusions, however, could be learned from these initial trial fits: the hyperfine parameters of the hematite component for all BHS samples are, to within the error limits, identical to those of the parent sample HS; for a well-defined broad sextet component, its quadrupole shift 2ε Q was adjusted to a value close to zero; the Fe 0 doublet is sharply defined for BHS3 at all T and shows the expected temperature variation of its center-shift; at any given temperature where the Fe 0 doublet is well-defined for all three BHS samples, the centre-shift values of the doublet are almost identical; this implies that, to good approximation, the center-shift of the Fe 0 doublet is independent of the added B amount; for those cases where both the broad sextet and the Fe 0 doublet are well-defined (BHS3 at 15, 40 K and, to lesser extent, 80 K, BSH2 at 15 K), the centre-shift values δ were iterated to a same value for the two respective subspectra; for any of the three BHS samples, it is noticed that, upon lowering the temperature of the absorber, the relative spectral area S BS of the broad sextet component increases at the expense of the relative spectral area S D of the Fe 0 doublet. However, the total contribution (S BS + S D ) remains constant, i.e., 0.38 ± 0.01, 0.49 ± 0.02 and 0.64 ± 0.01 for BHS1, BHS2 and BHS3, respectively. These findings from the trial fits of the MS, combined with the results from X-ray diffraction, thermal and chemical analyses have prompted the authors to suggest that the broad sextet and

6 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Fig. 6. Temperature dependence of the Mössbauer spectra of hematite samples treated with sodium borohydride. BHS1 (top), BHS2 (middle) and BHS3 (bottom). the Fe 0 doublet are both due to the same amorphous Fe 1 x B x alloy phase which exhibits a more or less broad distribution for the compositional parameter x. The idea is that at a given temperature T, Fe 1 x B x alloy grains with x exceeding a certain threshold value (depending on T) are (super)paramagnetic, giving rise to the doublet component, while other grains with lower B contents produce the sextet. The hyperfine field of the grains experienced to be magnetic, increases with increasing B content (clearly noticed from the MS at 15 K as displayed in Fig. 6) and, if the distribution for x is broad, so is the distribution in the hyperfine field, thus producing a broad sextet component in the MS. The following fitting model was thus henceforward adopted. The broad sextet and the Fe 0 doublet were constrained to exhibit the same, but adjustable centre-shift values at any given temperature. Further, the quadrupole shift of the broad sextet was taken to be zero, and the ratio of outer lines to middle lines to inner lines for the broad sextet was forced to be equal to 3:2:1. Relevant parameters for the as-such fitted MS at various temperatures are presented in Table 4. In all respects, these parameters exhibit consistent variations with temperature and with amount of added SBH, corroborating the idea that the broad sextet and the Fe 0 doublet indeed can be attributed the same amorphous Fe 1 x B x phase with probably a broad compositional distribution. Two earlier observations concerning the formation of amorphous Fe B alloys are worth mentioning at this point. In a previous study by some of the present authors regarding the reduction by SBH of particles of (natural) specularite, for which the coexistence in the MS of a broad sextet and a doublet was observed as well, the suggestion that these two spectral components are actually due to the same amorphous Fe B phase, also did emerge from the analyses of these MS [10]. In that study, however, only one single B concentration treatment was applied. The present, more elaborate experiments and analyses

7 242 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Table 4 Mössbauer results of the treated samples T (K) Broad sextet Fe 0 doublet H hf (koe) RA (%) δ (mm/s) E Q (mm/s) RA (%) δ (mm/s) BHS a RT BHS RT BHS RT The values of isomer shifts are with reference to metallic iron. Note: H hf, maximum-probability hyperfine fields; ε Q, quadrupole splitting; δ, isomer shifts; T, temperature; RA, relative areas. a Fixed parameter. firmly support that earlier suggestion. Secondly, the simultaneous presence of a doublet and a broadened sextet was also experienced for amorphous Fe B alloys obtained from reduction by SBH of Fe salts in solution by Saida et al. [16] and Song et al. [17]. The former authors ascribe the doublet to the formation of Fe 3+ oxides, while the latter ones are not specific about the nature of the doublet. In none of these two studies, however, temperature-dependent Mössbauer measurements were carried out. The chemical treatment of the synthetic hematite causes a drastic change in the morphology of the particles, as can be seen from the SEM images of the samples HS and BHS3 reproduced in Fig. 7. The original hematite is mainly composed of needleshaped particles with micron-sized long axis and thickness of less than 100 nm. On the other hand, the chemical treatment of -Fe 2 O 3 has produced an appearance that resembles aggregates of more or less spherical particles with diameters of a few hundreds of nanometers. The presence of original, needle-shaped hematite particles is not observed for BHS3, although 34% Fig. 8. ILEEMS spectrum of the sample BHS3 collected at room temperature. of the hematite has not been reduced according to the MS. The authors consider it unconceivable that the reaction would have altered the morphology of the remaining hematite grains. Therefore, it is suggested that the spherical particles are covering hematite needles, making them invisible on the micrographs. In an attempt to obtain support for this latter suggestion, an ILEEMS measurement at RT for the BHS3 powder was performed. ILEEMS is a variant of conventional transmission Mössbauer spectroscopy [18] by which the low-energy electrons are counted. These electrons, with energy of 10 ev, are produced by after effects following the decay of the probe nuclei in the absorber. As a consequence of this low energy, only an extremely thin surface layer of the material is probed. The result of the ILEEMS experiment is shown in Fig. 8. The emission spectrum has the same shape and spectral components as does the transmission spectrum (Fig. 6, bottom left), however with a significantly weaker contribution of the -Fe 2 O 3, i.e., 20% instead of 34% as obtained from the transmission MS. This finding is clear evidence that about one-half of the hematite surface is covered by a Fe B layer with mean thickness of 2 3 nm. It should be mentioned that the hyperfine parameters of the - Fe 2 O 3 and Fe B phases obtained from the ILEEMS are within error limits identical to the values fitted to the transmission MS. Also the weak ferrous doublet is present in the ILEEMS. Fig. 7. Scanning electron microscopy images of samples HS (left) and BHS3 (right).

8 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Natural hematite The natural hematite (sample HNG) was collected in Quadrilátero Ferrífero and contains a small fraction of quartz ( 2%), which supposedly will not affect the reduction process. The hematite consisting of micron-sized particles was treated with different borohydride concentrations, which are 1:2, 1:8 and 1:16, as described earlier. The obtained products are henceforward code-named BHNG1, BHNG2 and BHNG3, respectively. These products are a black and highly magnetic powder. Fig. 9 (top) shows that the ph slightly increases during the reaction, whereas the temperature increased starting from room temperature (Fig. 9, bottom). As mentioned for the reduction of synthetic hematite, the behaviour of the ph and temperature can be attributed to the formation of hydroborate intermediates, as described by Forster et al. [11]. The XRD patterns of the original hematite and one of the treated samples (BHNG3) are shown in Fig. 10. The diffractogram of HNG shows peaks characteristic of well-crystallized hematite and quartz. Furthermore, a single peak at 42 (2θ)is observed (see arrow inset in Fig. 10), which is due to a phase that could not be identified. The treated samples exhibit the same diffraction lines, those arising from hematite consistently decreasing in intensity with increasing B addition. Additionally, the diffractogram of sample BHNG3 contains a broad peak at approximately 52 (2θ), which corresponds to the position of the dominant diffraction line of metallic iron. This observation Fig. 10. X-ray diffraction patterns of the natural hematite (HNG, inset), and after treatment with borohydride weight proportion 1:16 (BHNG3). Si corresponds to standard silicon which was added, and the arrow corresponds a phase which has not been identified. Table 5 Cell parameters and particles size (D hkl ) of the natural hematite in the studied samples Sample Cell parameter D 104 (nm) a (Å) c (Å) Hematite (standard) HNG ± ± >100 BHNG ± ± >100 BHNG ± ± >100 BHNG ± ± >100 Note: Standard hematite (PDF ). indicates that for relatively high NaBH 4 amounts metallic iron is formed under the applied conditions of synthesis. The absence in BHNG1 and BHNG2 of any other significant crystalline phase besides hematite, indicates that the constituents responsible for the black colour and magnetism of these samples are again amorphous Fe B alloys. Table 5 contains the cell parameters of the hematite phases present in the involved samples. Their values are similar to those referring to standard hematite (see Table 1), indicating that the reaction with SBH has not affected the structure of the natural hematite that has not been attacked by the SBH. The particle sizes as reflected in the XRD line widths are well above 100 nm. The iron and boron contents and the bulk composition of the amorphous Fe B alloys are given in Table 6. These results indicate that the boron content x in Fe 1 x B x slightly increases Table 6 Boron and iron content (wt.%) and the global composition of the amorphous Fe B alloy Sample Boron Fe total Fe 1 x B x Fig. 9. Temperature (bottom) and ph (top) behavior during the reaction of reduction of the natural hematite. HNG ± 0.3 BHNG1 0.4 ± ± 0.2 Fe 0.98 B 0.02 BHNG2 1.2 ± ± 0.2 Fe 0.94 B 0.06 BHNG3 1.6 ± ± 0.1 Fe 0.92 B 0.08

9 244 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Fig. 11. Temperature dependence of the Mössbauer spectra of natural hematite. with the amount of NaBH 4. The increase, however, is less pronounced than in the case of the synthetic hematite reported on in the previous section. The thermal behaviour of the three SBH-produced BHNG samples as reflected in both the TGA and DSC curves, is very similar, and moreover it is qualitatively the same as for the reduction products of the synthetic hematite. Hence, no further relevant details in that respect can be reported. The MS of the parent HNG are similar to those obtained for sample HS. Two example spectra are shown in Fig. 11, referring to RT and 80 K, and fitted Mössbauer parameters for selected temperatures are listed in Table 7. Careful analyses of the HNG MS have revealed two major differences with respect to the HS MS. The first one concerns the Morin transition which, on lowering the temperature from RT, for HNG already sets in at 260 K, while this onset temperature is below 200 K for HS hematite. The different degrees of crystallization for the two hematites are believed to be responsible for this shift of the transition region. Despite the large particle size for HNG part of its grains ( 5% at 80 K) keep their weakly ferromagnetic state at low temperatures. This feature has been observed earlier for natural hematites Table 7 Hyperfine parameters of the Mössbauer spectra at selected temperatures of the sample HNG T (K) 2ε Q (mm/s) H hf (koe) RA (%) δ (mm/s) a 532 a a RT The values of isomer shifts are with reference to metallic iron. Note: H hf, maximum-probability hyperfine fields; 2ε Q, quadrupole shifts; δ, isomer shifts; T, temperature; RA, relative areas. a Fixed parameter. with similar degrees of crystallisation and has been ascribed to the presence of minor amounts of substitutional cations in the hematite lattice (Al, Mn,...) [15]. A second difference between the MS of the HS and HNG hematites is the appearance of texture effects for the latter. This is clearly noticed in the MS at 80 K. For HNG the middle and inner absorption lines (lines 2 5) show nearly the same intensity, which is not so for HS spectrum (see Fig. 5). The reason for this texture is the plate-like morphology of the HNG crystallites. It would be of interest to study this orientational texture effect in more detail, however, this is beyond the scope of the present report. The MS of the products formed by the reactions of the suspended HNG particles with different amounts of SBH show that the compositions of these products are significantly different from those resulting from the corresponding treatments of the synthetic hematite HS as described in the previous section. Example spectra are reproduced in Fig. 12 and adjusted parameter values at selected temperatures are listed in Table 8. For the lowest applied concentration of SBH (sample BHNG1) the spectra at various temperatures consistently show the presence of a dominant ( 96% of total spectral area) sextet due to the initial hematite phase and a weak doublet with parameters that resemble those of the Fe 0 doublet resolved from the BHS MS (see Tables 8 and 4) and therefore henceforward referred to as the Fe 0 doublet. However, the broad sextet component present in the latter MS is observed in the BHNG1 spectra in temperatures lower than 40 K. The sextet component could not be included in the fit because its area is merely 4%. The broad sextet appears in the MS for samples BHNG2 and BHNG3, however with significantly smaller contribution as compared to the corresponding BHS samples. Also the temperature variation is much weaker. The same is true for the Table 8 Mössbauer results of the treated samples T (K) Broad sextet Fe 0 doublet H hf (koe) RA (%) δ (mm/s) E Q (mm/s) RA (%) δ (mm/s) BHNG RT BHNG a RT BHNG a RT The values of isomer shifts are with reference to metallic iron. Note: H hf, maximum-probability hyperfine fields; ε Q, quadrupole splitting; δ, isomer shifts; T, temperature; RA, relative areas. a Fixed parameter.

10 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Fig. 12. Temperature dependence of the Mössbauer spectra of hematite samples treated with sodium borohydride. BHNG1 (top), BHNG2 (middle) and BHNG3 (bottom). relative spectral area of the Fe 0 doublet, which moreover does not increase markedly with increasing total amount of added SBH. In fitting these spectra, the area ratio of outer lines to middle lines to inner lines for the broad sextet was forced to be equal to 3:2:1 and its quadrupole shift to 0.0 mm/s (Table 8). It should be stressed at this point that, considering the shape of the broad sextet and the weak intensities of both this sextet and the Fe 0 doublet, the Mössbauer parameters of the respective components are ill-defined. As a result, it is less straightforward that the two components are due to a same amorphous Fe B phase. The MS of the HNG sample treated with the highest concentration of SBH show in addition to the components of the BHNG2 spectra, two other components which are a sextet with hyperfine parameters characteristic of -Fe, and a Fe 2+ doublet which cannot be attributed to any particular phase. The contribution of the metallic iron is 11%, while of the Fe 2+ doublet is 6%. These findings are in line with the results reflected in the XRD patterns. Like for the synthetic hematite, the reduction of the natural species apparently caused a change in the morphology of the original particles as observed in SEM images. A selection of these SEM images is reproduced in Fig. 13. The original iron oxide is composed of micron-sized particles with irregular shape. After the chemical treatment (BHNG3) aggregates of more or less spherical particles, which seem to cover part of the surfaces of larger particles, are observed. Presumably these larger particles are the unaffected inner parts of the original hematite grains, whereas the aggregated, smaller particles are the actual products of the reduction. This interpretation would imply that the reaction with the SBH takes place at the surfaces of the hematite grains. As for the synthetic small particle hematite of the previous section, this suggestion is corroborated

11 246 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) Fig. 13. Scanning electron microscopy images of samples HNG (top) and BHNG3 (bottom). by ILEEMS. The spectrum recorded for BHNG3 at RT is presented in Fig. 14. It could be adequately fitted with the same five components that were resolved from the transmission MS, i.e., the hematite spectrum, the -Fe sextet, the broad Fe B sextet, the Fe 0 doublet and finally the Fe 2+ doublet. The respective spectral area values were adjusted as 0.15, 0.07, 0.43, 0.25 and From the transmission spectra (yielding the bulk composition), these numbers were found to be 0.58, 0.12, 0.20, 0.06 and 0.04, respectively. It is thus clear that upon reduction the hematite content in the surface layer is reduced drastically in favour of primarily the Fe B alloys. Also, this reduction is more drastic for BHNG3 than for BHS3, for which the fractional hematite area was found to be 0.34 from the transmission spectrum, compared to 0.20 from the ILEEMS. 4. Discussion and conclusions The reactions between the suspended -Fe 2 O 3 and sodium borohydride to form amorphous Fe 1 x B x alloys were incomplete for all reaction conditions examined in this work. Indeed, in all cases some residual unreacted iron-oxide particles have been detected by X-ray diffraction and Mössbauer spectroscopy applied to the reaction products. It is suggested that the reaction proceeds following the equation: 3(1 x)bh 4 + (1 x)fe (3 4x)H 2 O Fe 1 x B x + (10 7x)H 2 + (3 4x)HBO 2 Fig. 14. ILEEMS spectrum of the sample BHNG3 collected at room temperature. The Fe 1 x B x alloys are responsible for the observed magnetism. Metallic iron, in addition to amorphous Fe 1 x B x, was formed only in the case of the natural hematite that was treated with the highest borohydride concentrations. An important finding in this work concerns the contribution to the total MS by the amorphous Fe B phase. This contribution indeed consists of a broad sextet and a doublet. This doublet, observed in earlier studies on related reaction processes [16,17], is consequently not due to the presence of an additional ferric oxide phase, as was claimed to be the case by the authors of those earlier studies. The quantity of alloy and the global B content x both increase with increasing amount of borohydride added to the suspensions of the iron-oxide particles, the more so for the synthetic small particle hematite species. Thus, small particle synthetic hematite

12 V.G. de Resende et al. / Journal of Alloys and Compounds 440 (2007) (size 25 nm) reacts much more efficiently with NaBH 4 than the natural form (particle size >100 nm), in the sense that for a given SBH concentration a higher proportion for the Fe B alloy is formed. This difference in reaction yield can be attributed to the high specific surface area (SSA) of the latter. Indeed, the higher the SSA, the more BH 4 groups will be in contact with the surfaces of the hematite particles and the more B atoms can diffuse into the lattice of the amorphous alloy and hence lower the amount of original oxide remaining after the reaction. The various exothermic signals in the DSC curves concern crystallization and subsequent two-stage oxidation processes of the Fe B. The observation that for each sample oxidation occurs in two steps suggests that the composition (boron content) of the alloy present in a given sample is not uniform, but instead exhibits a bimodal distribution with average value as determined from the chemical analyses. Indeed, according to previously reported studies of Fe B alloys with well-defined and unique boron content (e.g. refs. [4,19]), a correlation was found between this boron content and the thermal stability of the amorphous state. Phases in which the iron atoms have only a few neighbors of boron would crystallize first, whereas those with larger number of B neighbours are more stable and crystallize at higher temperatures. The multimodal distribution for the composition parameter x in Fe 1 x B x would be consistent with the coexistence of a doublet and a sextet in its MS, the latter one growing at the expense of the former upon lowering the temperature. In conclusion, it has been demonstrated that the reduction of hematite particles by sodium borohydride is more efficient the larger the particles specific surface area is, and that the chemical reactions take place in the surface layers of the grains. As such, surface properties of the grains, in particular concerning the chemical Fe O bonds, likely play a crucial role as to the composition of the reduction products. In that respect, it can be expected that other Fe oxides or Fe oxyhydroxides will behave differently under similar reducing conditions. This point was examined by the authors using nano-sized goethite, -FeOOH. The results of that research will be presented in the second part of the present report. Acknowledgements This work was partially funded by CNPq and Fapemig (Brazil), and by the Fund for Scientific Research Flanders, Belgium. One of us (V.G. de Resende) thanks the Program Alban, the European Union Programme of High Level Scholarships for Latin America, Scholarship No. E04M034189BR. References [1] I.D. Dragieva, Z.B. Stoynov, K.J. Klabunde, Scripta Mater. 44 (2001) [2] G.D. Foster, L.F. Barquín, N.S. Cohen, Q.A. Pankhurst, I.P. Parkin, J. Mater. Proc. Tech (1999) 525. [3] S. Petit, K. David, J.P. Doumerc, J.C. Grenier, T. Seguelong, M. Pouchard, C.R. Acad. Sci. Paris, Série II c 1 (1998) 517. [4] S. Linderoth, S. Mørup, J. Appl. Phys. 69 (8) (1991) [5] C.Y. Wang, Z.Y. Chen, B. Cheng, Y.R. Zhu, H.J. Liu, Mater. Sci. Eng. B60 (1999) 223. [6] J. Rivas, M.A. López Quintela, M.G. Bonome, R.J. Duro, J.M. Greneche, J. Magn. Magn. Mater. 122 (1993) 1. [7] D. Buchkov, S. Nikolov, I. Dragieva, M. Slavcheva, J. Magn. Magn. Mater. 62 (1986) 87. [8] S. Linderoth, C.A. Oxborrow, O.V. Nielsen, Nucl. Instrum. Meth. B76 (1993) 64. [9] A. Yedra, L.F. Barquín, J.C. Gómez Sal, Q.A. Pankhurst, J. Magn. Magn. Mater (2003) 14. [10] V.G. de Resende, G.M. da Costa, E. De Grave, L. Datas, J. Mater. Sci., in press. [11] G.D. Forster, L.F. Barquín, R.L. Bilsborrow, Q.A. Pankhurst, I.P. Parkin, W.A. Steer, J. Mater. Chem. 9 (1999) [12] J.Z. Jiang, K. Ståhl, K. Nielsen, G.M. da Costa, J. Phys. Condens. Matter 12 (2000) [13] R.M. Cornell, U. Schwertmann, The Iron Oxides, Willey-VCH, [14] E. De Grave, R.E. Vandenberghe, Phys. Chem. Miner. 17 (1990) 344. [15] E. De Grave, L.H. Bowen, R. Vochten, R.E. Vandenberghe, J. Magn. Magn. Mater. 72 (1988) 141. [16] J. Saida, M. Ghafari, Y. Nakamura, A. Inoue, T. Masumoto, Nucl. Instrum. Meth. B76 (1993) 223. [17] X. Song, X. Yusheng, J. Huali, X. Qing, Nucl. Instrum. Meth. B76 (1993) 260. [18] E. De Grave, R.E. Vandenberghe, C. Dauwe, Hyperfine Interact. 161 (2005) 147. [19] L. Wang, Z.C. Tan, S.H. Meng, D.B. Liang, S.J. Ji, Z.K. Hei, J. Therm. Anal. Calorim. 66 (2001) 409.