J O U R N A L O F C H E M I S T R Y Materials The eVect of iron on the crystalline phases formed upon thermal decomposition of Mg–Al–Fe hydrotalcites Jose� Maria Ferna�ndez,a Maria Angeles Ulibarri,a Francisco M. Labajosb and Vicente Rives*b aDepartamento de Quý�mica Inorga�nica e Ingenierý�a Quý�mica, Facultad de Ciencias, Universidad de Co�rdoba, Co�rdoba, Spain bDepartamento de Quý�mica Inorga�nica, Universidad de Salamanca, Salamanca, Spain Received 26th June 1998, Accepted 5th August 1998 Layered double hydroxides (LDH) containing MgII, FeIII, and AlIII in the brucite-like layers and interlayer carbonate (with a constantMII/MIII ratio but varying AlIII/FeIII ratios) have been prepared and characterised by Xray diVraction, thermal analysis, FT-IR and UV–VIS/diVuse reflectance spectroscopies, temperature-programmed reduction and specific surface area assessment through low temperature adsorption of N2.AnMg,Al–LDH, but with intercalated hexacyanoferrate(III), has been also prepared and characterised, in which simultaneous formation of the carbonate analogue did not occur. Thermal decomposition in air at 450 and 750 °C leads to MgO and poorly crystallised MgFe2O4 spinel (crystallinity increasing with the iron content), while for the hexacyano-containing sample, crystallization only is observed after calcination at 900 °C.This diVerent behaviour has been related to the initial location of the iron ions. to a solution containing Mg2+, Al3 + and Fe3+ nitrates in Introduction selected concentrations (ranging from 0.3 M to 0.066 M) to Layered double hydroxides (LDH) are also known as yield solids with diVerent Al/Fe molar ratios, but with a anionic clays, as they show a structure electrically opposite constant Mg/(Al+Fe) molar ratio of 351, until a pH of 10 to that shown by clays.Their general formula is was reached, this value being maintained at constant pH (10) [MII1-xMIIIx(OH)2][Am-]x/m·nH2O.The commonest type are with a Dosimat 275 (Metrohm) coupled to a pH-meter model the hydrotalcite group of minerals, the structure of which 691 (also from Metrohm). When addition was complete, the consists of brucite-like layers [formed by edge-sharing mixture was further magnetically stirred for 2 h. The suspen- Mg(OH)6 octahedra] with partial MgII/AlIII isomorphic substi- sion was submitted to hydrothermal treatment at autogenous tution, the electrical balance being attained with carbonate pressure at 120 °C in a Teflon-lined stainless steel bomb for anions located, together with water molecules, in the interlayer. 24 h. The solid was then filtered and washed until nitrate and Substitution of the layer cations is very easy, and the interlayer sodium ions were completely absent.anion can be easily changed as well, thus giving rise to a In order to prepare the hexacyanoferrate(III)-containing continuous growing family of new layered materials with sample, aqueous solutions were prepared in deionised water important applications as catalysts or catalyst precursors, previously boiled to remove dissolved CO2; nitrogen was sensors, anion scavengers, etc.1–3 continuously bubbled through the water until it reached room On thermal decomposition, these materials lead to mixed temperature, after which bubbling was continued for 15 min oxides (also known as ‘non-stoichiometric spinels’), and it has at room temperature. Fifty ml of a 0.05 M K3[Fe(CN)6] been shown4 that the nature/structure of the solids obtained solution were placed in a three-necked round-bottom flask (and their applications) depend on the starting LDH which is and, while magnetically stirred, 50 ml of a solution containing decomposed.Mg2+ and Al3+ nitrates (0.3 M and 0.1 M, respectively) were The aim of the present work was to analyze how the added dropwise from a separation funnel. The same Dosimat presence of iron and its concentration in the brucite-like layers, and pH-meter cited above were used to maintain the pH at a or in the interlayer anion, may lead to solids, after calcination, value of 8.All of the processes were carried out at 60 °C. The with diVerent structures. We have recently reported4,5 on the suspension was left to settle overnight at room temperature preferential formation of Mg2V2O7 or MgV2O6, with diVerent and the precipitate washed with preboiled water heated to local environments around the V5+ cations, from hydrotalcite- 60 °C by magnetic stirring and centrifugation, and the solid like precursors, when starting from materials containing Mg2+, was left to dry in an oven at 80 °C in the open air.Al3+ and V3+ in the brucite-like layers and carbonate in the After characterization, the samples were calcined for 2 h in interlayer, or from materials containing Mg2+ and Al3+ in the air at 450 or 750 °C, in order to obtain mixed oxides.It has layers and decavanadate, V10O286-, in the interlayer. In been previously shown6 that the decomposition of the the present case, diVerences in location of Fe3+ cations in Mg–Al–Fe hydrotalcites is almost complete at 450 °C, and octahedral or tetrahedral sites, or diVerent degrees of crystalonly residual hydroxyl groups are further removed between linity, could be expected when Fe was in the brucite-like layer, 450 and 750 °C.Labelling of the samples is given in Table 1; or forming an anionic complex in the interlayer. Actually, it the calcination temperature is indicated in °C.While samples has been found that the crystallinity of the solids formed (or the calcination temperature required to obtain crystalline M2 to M5 and all calcined samples were ochreous, sample solids) depends on the precise nature of the starting LDH. M6 was light yellow. Experimental Techniques Sample preparation Elemental chemical analysis for metals was carried out by atomic absorption spectroscopy (AAS) using a Perkin-Elmer All chemicals were from Merck.For the carbonate-containing samples, a 1 M NaOH aqueous solution was added dropwise 3100 apparatus after dissolution of the samples in dilute HCl. J. Mater. Chem., 1998, 8(11), 2507–2514 2507Table 1 Labelling of the samples and elemental chemical analysis in the layers, Mg2+/(Al3++Fe3+), was 2.9751 for samples results M2 to M5; for sample M6 the Mg2+/Al3+ ratio was 3.1751; these two values are acceptably close to the expected value Sample %Mga %Ala %Fea Mg/(Al+Fe)b Mg/Alb Al/Feb existing in the starting solutions (351).On the other hand, the Al/Fe ratio for sample M6 was 2.7351. The expected value M2 24.38 5.47 8.11 2.88 1.40 M3 24.10 3.74 9.79 3.15 0.79 for this ratio was 351, as the negative charge of the interlayer M4 22.39 2.74 11.99 2.91 0.47 anion should balance the positive charge due to Al3+ in the M5 22.86 — 17.92 2.93 — layers, and so, the anion existing in this sample [hexacyanofer- M6 18.73 6.56 4.99 3.17 2.73 rate(III )] being trivalent, a value of 351 would be expected.M2–450 28.68 6.12 8.74 3.08 1.45 If hexacyanoferrate(III ) is the only anion existing in sample M3–450 26.90 4.50 11.49 2.96 0.81 M6, then the ratio between the weight percentages of C and M4–450 26.19 3.14 13.51 3.01 0.48 N would be 0.8651; however, the experimental value was M5–450 27.46 — 19.37 3.25 — 0.9351, indicating that a slight excess of carbon exists.From M6–450 23.76 8.26 5.56 3.19 3.09 the FT-IR results (see below) the presence of a small amount M2–750 32.15 7.29 10.21 2.92 1.48 of carbonate species, probably adsorbed on the external surface M3–750 28.02 5.04 11.91 2.88 0.88 of the crystallites, can be concluded, thus explaining this M4–750 27.04 3.58 14.62 2.82 0.51 finding, as formation of a co-product containing intercalated M5–750 32.62 — 23.48 3.19 — M6–750 25.38 8.93 6.38 3.16 2.90 carbonate was not observed (see powder X-ray diVraction results below).On the other hand, the molar N/Fe ratio is aWeight percentage. bMolar ratio, x:1. very close to the expected value (651). Powder X-ray diVraction Carbon and nitrogen were analyzed in sample M6 in a Perkin Elmer 2400 CHN analor the original samples are shown in Fig. 1. Powder X-ray diVraction (PXRD) diagrams were recorded They are all similar, with sharp bands at low 2h values, on a Siemens D500 instrument, using graphite-filtered Cu-Ka corresponding to the higher order 001 reflections of a layered radiation (l=1.54050 A° ); the instrument was equipped with a material.The sharp peaks recorded for sample M3 close to DACO-MP microcomputer, and software DiVract-AT was 2h=38 and 45° superimposed to broader maxima (also used to analyze the data, identification of existing crystalline recorded in the same positions for the other samples), and phases being concluded from comparison with JCPDS diVrac- that recorded at 2h=65° are due to the Al sample holder.The tion files. In some cases, where small amounts of sample were positions of the harmonics are coincident for samples M2, available, an Al sample holder was used, and thus, sharp, M3, M4, and M5, while for sample M6 these harmonics are intense diVraction peaks due to the holder were recorded, but shifted towards lower 2h values (i.e., larger spacings).these peaks were unambiguously identified. Assuming a 3R polytypism,9 the first (from low 2h values) DiVerential thermal analysis (DTA) and thermogravimetric peaks can be indexed as (003), (006), and (009), and from analysis (TG) were recorded on Perkin-Elmer DTA1700 their positions, the values for parameter c have been calculated and TGS-2 instruments, respectively, using flowing air (Table 2) as c=[d(003)+2 d(006)+3 d(009)].(60 ml min-1) at a heating rate of 12 °Cmin-1. The values for parameter a (which coincides with the average Fourier-transform infrared spectra (FT-IR) were recorded cation–cation distance in the brucite-like layers) are also using the KBr pellet technique on a Perkin-Elmer FTIR-1730 included in Table 2, and have been calculated from the position instrument; one hundred scans were averaged in order to of the peak due to planes (110), which is the first peak of the improve the signal-to-noise ratio, and the nominal resolution doublet recorded close to 2h=60°, as a=2 d(110).It should was 4 cm-1. be noted that for samples M2 to M5, a steady increase in a is Ultraviolet–visible (UV–VIS) spectra were recorded observed. This is due to the progressive Al3+/Fe3+ substitution following the diVuse reflectance (DR) technique in a Shimadzu (see elemental chemical analysis data in Table 1), and the UV-240 instrument, using MgO as reference and a slit of 5 nm.larger ionic radius of Fe3+ (78.5 pm in high spin, octahedral Specific surface area and porosity of the samples were coordination) than Al3+ (67.5 pm in octahedral coordidetermined on a Gemini instrument from Micromeritics. The nation).10 This is also the reason of the lower a value for samples were previously degassed at 125 °C for 2 h with sample M6, where the only trivalent cation in the layers is Al3+.nitrogen in a Micromeritics FlowPrep 060 apparatus. The The slight diVerences in the values of parameter c for adsorption–desorption isotherms (-196 °C) were analyzed using literature software.7 Temperature-programmed reduction (TPR) runs were performed in a TPR/TPD 2900 instrument from Micromeritics, using a 5% H2/Ar (vol.) mixture to reduce the samples.Amounts of samples of ca. 15 mg were used, and the gas flow, sample weight and heating rate were chosen in order to attain good resolution of the reduction peaks.8 The gas, at the reactor exit, was passed through a cold trap (melting isopropanol ) to retain vapours and condensable gases before entering the detector.Results and discussion Elemental chemical analysis The results obtained for Mg, Al and Fe are given in Table 1; the calculated MII/MIII and Al/Fe ratios are also given. C and N were analyzed for sample M6, obtaining values of 6.3 and Fig. 1 Powder X-ray diVraction profiles for samples M2 to M6. (*) 6.8%, respectively. The average MII/MIII ratio, i.e., the ratio signals due to the Al sample holder.The traces have been displaced vertically for clarity. between the molar fraction of divalent and trivalent cations 2508 J. Mater. Chem., 1998, 8(11), 2507–2514Table 2 Summary of X-ray diVraction results, specific surface area determination and temperature-programmed reduction Sample c/A° a/A° SBET/m2 g-1 H2/Fea M2 23.52 3.08 59 1.36 M3 23.73 3.09 68 1.68 M4 23.82 3.10 70 1.42 M5 23.67 3.11 48 1.25 M6 33.36 3.06 143b — M2–450 123 1.40 M3–450 145 1.56 M4–450 138 1.63 M5–450 81 1.67 M6–450 123 1.68 M2–750 101 1.12 M3–750 113 1.22 M4–750 117 1.43 M5–750 16 1.35 M6–750 110 1.28 aMolar ratio, x:1.b128 m2 g-1 surface area equivalent to adsorption on micropores, and 15 m2 g-1 external surface area.samples M2 to M5 [the higher value for sample M6 is due to the presence of hexacyanoferrate(III) instead of carbonate] cannot, however, be easily related to particular diVerences in the samples, as the small changes observed could be due to small diVerences in the hydration degree of the interlayer. From the thickness of the brucite-like layers, 4.8 A° ,3 the interlayer space for the carbonate-containing samples is close to 3 A° , corresponding to carbonate anions with their molecular Fig. 2 Powder X-ray diVraction profiles for samples M2 to M6 plane parallel to the brucite-like layers. calcined at (upper) 450 and (lower) 750 °C for 2 h. (*) signals due to With regards to sample M6, it should be stressed that no the Al sample holder. The traces have been displaced vertically peak has been recorded which could be ascribed to the presence for clarity.of a co-product corresponding to carbonate-interlayered hydrotalcite. This result is extremely important, as in most of the papers previously reported in the literature on hexacyano- nation, as the MII/MIII ratio in the spinel is equal to 0.551, and so crystallization of MIIO is always observed.19,20 ferrate-containing layered double hydroxides,11–13 coformation of a carbonate–LDH, together with that of the Additionally, some diVraction peaks of MgO are recorded almost coincident with diVraction peaks of spinels.hexacyanoferrate form, is usually observed. The gallery height [from the spacing for planes (003), 11.12 A° , and the thickness With this, taking into account the nature of the cations existing in our samples, the following phases could be formed: of the brucite-like layers, 4.8 A° ] was 6.32 A° .The size of the Fe(CN)63- anion is close to 11 A° along the C4 axis, 8.7 A° MgO, MgAl2O4, MgFeAlO4,MgFe2O4. The presence of MgO is concluded in all ten cases from the two intense peaks at along the C2 axis, and 6.5 A° along the C3 axis.14 This means that the anion should be oriented with its C3 axis (that joining 2.10 and 1.48 A° (ca. 2h=43 and 63°, respectively), and the excess in MgO above the stoichiometric amount required to parallel faces of the octahedron) perpendicular to the brucitelike layers and that, even so, some stress and distorsion should form any Mg-containing spinel. From the positions of the peaks in the PXRD diagram of exist.Alternatively, grafting of the anion to the brucite-like layers, in a similar way to that previously described for several sample M5-750 (where the sharpest peaks are recorded), excluding the peaks coincident with those of MgO, the cell vanadates intercalated in LDHs,15,16 could be claimed. The PXRD diagrams of the samples calcined at 450 and dimension for the spinel formed can be calculated as 8.402 A° .The reported values21 for other spinels are 8.083 A° (MgAl2O4, 750 °C are included in Fig. 2. Again, sharp peaks due to the Al sample holder are recorded in some cases. With regard to JCPDS file 21-1152), 8.320 A° (MgFeAlO4, JCPDS file 11-9), 8.387 A° (MgFe2O4, JCPDS file 36-398), and 8.396 A° (Fe3O4, the maxima of the samples, these are extremely broad and their positions roughly coincide for all five samples calcined JCPDS file 19-629).From comparison between the reported and the calculated values it can be concluded that the spinel at 450 °C, while for samples calcined at 750 °C the PXRD diagram of sample M6-750 is rather similar to those recorded formed in our samples should be MgFe2O4. We want to stress that the Al sample holder behaves in our case as a sort of for the samples calcined at 450 °C; on the contrary, for samples M2-750 to M5-750 a progressive increase in the intensity of ‘internal reference’, for better definition of the positions of the peaks.In addition, formation of Fe3O4 can be tentatively new, sharper peaks (not recorded for samples calcined at 450 °C) is observed.assumed, if Fe3+�Fe2+ reduction could take place during calcination in air. For a hydrotalcite structure to remain stable, the MII/MIII ratio should be larger than 151.2 Calcination of hydrotalcites So, the PXRD data for the calcined samples can be summarized as follows: calcination at 450 °C gives rise, in all leads to removal of volatile interlayer anions and hydroxyl groups, and formation of the corresponding oxides, usually cases, to formation of ill-crystallized MgO, and a small amount of a MgFe2O4 spinel.When the calcination temperature is MIIO and the MIIMIII2O4 spinel, although observation of diVraction peaks due to the crystalline spinel depends on the increased to 750 °C the presence of the spinel is more evident, especially in the case of sample M5-750 (that is, that without nature of the metal cations and on the calcination temperature (e.g., crystalline Mg,Al spinel is only detected after calcination aluminium and with the largest Fe content), while for sample M6-750 (prepared from the hexacyanoferrate precursor) the at ca. 900–1000 °C).17,18 The pure spinel cannot be obtained, unless redox processes simultaneously occur during calci- diagram is almost coincident with that recorded for the same J.Mater. Chem., 1998, 8(11), 2507–2514 2509sample, but calcined at 450 °C, indicating that crystallization of the spinel has not been favoured, or it is not detected due to the low Fe content (see Table 1). This conclusion can be easily reached from comparison of the intensity of the peak close to 2h=57° (d=1.615 A° ), corresponding to planes (511) or (333) of MgFe2O4.Its intensity increases steadily from sample M2-750 to sample M5-750, but the peak is absent in the diagram of sample M6-750. As for calcination of a Mg,Al hydrotalcite in this same temperature range, no crystalline phase containing Al has been observed.22,23 In order to gain insight into the formation of crystalline phases, selected samples have been calcined at higher temperatures or for longer periods of time.When sample M4 is calcined at 750 °C for 4, 8 or even 24 h, instead of 2 h as used for the sample whose PXRD diagram is shown in Fig. 2, the only eVect observed is a slight sharpening of the diVraction peaks due to the spinel. The eVect is even less evident for sample M5 calcined for these periods of time at 750 °C, as in this sample, calcination for 2 h is enough to form the spinel, as shown by the sharp peaks recorded (Fig. 2). With regards to sample M6, calcination at 750 °C for even 24 h has only minor eVects on the PXRD diagram, and only peaks due to MgO, and broad, ill-defined peaks due to the MgFe2O4 spinel, are again recorded. However, when sample M6 was calcined at higher temperatures, the changes were rather drastic.Fig. 3 includes the PXRD diagrams for this sample calcined at 450, 750, 900, Fig. 4 Thermogravimetric (dotted lines) and diVerential thermal (solid 1000, and 1100 °C for 2 h. From 900 °C upwards, the diVraclines) analyses for samples M5 (upper traces) and M6 ( lower traces). tion peaks due to the spinel are clearly observed, although some of them coincide with peaks due to MgO. Thus, we can conclude that the lack of detection of peaks surface of the particles, and water molecules from the interlayer due to the spinel in sample M6-750 is not due to the low space, and amounting to ca. 15% of the initial weight of the sensitivity of the technique, as calcination at higher tempera- sample. The second weight loss is almost completed at ca.ture gives rise to samples where the peaks due to the spinel 450 °C, and corresponds to removal of hydroxyl groups from are clearly detected. In other words, the lack of formation of the brucite-like layers, as well as of volatile species from the the MgFe2O4 spinel in sample M6-750, even though the spinel interlayer anions (i.e., CO2 from interlayer carbonate), as is formed in all other samples calcined for 2 h at this same concluded in previous studies25 on a Mg,Al–carbonate hydrotemperature, is undoubtedly related to the diVerent location talcite.The weight loss above 450 °C amounts to ca. 1–2% of of the Fe(III ) ions in these two series of samples: in the layers the initial sample weight, and is usually ascribed to removal (samples M2 to M5) or in the interlayer space, as hexacyano- of strongly held hydroxyl groups.From the elemental chemical ferrate (sample M6). composition of the starting solid (Table 1) and the total weight loss up to 750 °C, assuming formation of mixed oxides (MgO, DiVerential thermal analysis (DTA) and thermogravimetric Fe2O3 and Al2O3, or any combination of these) at the highest analysis (TG) temperature reached, the interlayer water content can be calculated, thus providing the whole formula of the starting Representative TG and DTA curves for selected samples are layered materials (Table 3).The behaviour shown by sample shown in Fig. 4. Weight loss starts from room temperature M6 is slightly diVerent and even though decomposition was and is completed at ca. 750 °C. Two steps are observed, as is essentially complete at the same temperature (about 750 °C), usual for hydrotalcites.24 The first one, up to ca. 200–250 °C, intermediate decomposition steps can be observed; unfortu- corresponds to removal of water physisorbed on the external nately, we were unable to analyze the gases evolved during decomposition, to assess the diVerent decomposition steps.The maximum number of water molecules hosted in the interlayer space of a carbonate-containing hydrotalcite can be easily calculated.26,27Water molecules and interlayer carbonate anions can be close-packed in the interlayer region, as hydroxyl groups are in the brucite-like layers. We can assume the size of a water molecule to be coincident with that of a hydroxyl group and about one third of that of a carbonate anion.If these are located with their molecular plane parallel to the Table 3 Formulae of the samples prepared Sample Formulaa M2 [Mg0.74Fe0.11Al0.15(OH)2][CO3]0.13 0.70H2O M3 [Mg0.76Fe0.13Al0.11(OH)2][CO3]0.12 0.69H2O M4 [Mg0.75Fe0.17Al0.08(OH)2][CO3]0.13 0.69H2O M5 [Mg0.75Fe0.25(OH)2][CO3]0.13 0.61H2O Fig. 3 Powder X-ray diVraction profiles for sample M6 calcined for M6 [Mg0.76Al0.24(OH)2][Fe(CN)6]0.08 0.83H2O 2 h at the temperatures given (in °C). (*) signals due to the Al aThe values have been rounded to the nearest 0.01. sample holder. 2510 J. Mater. Chem., 1998, 8(11), 2507–2514layers (as concluded from the width of the interlayer space, of two superimposed bands for samples M2 and M5, a very sharp band (also recorded for sample M6) at 1381 cm-1, and from the spacing determined by PXRD), then the maximum number of water molecules would be 2-(3x/2) per hydrotal- a broader band at 1374 cm-1.This last band is due to mode n3 of the interlayer carbonate anions. The shift from the cite ‘formula’, where x stands for the molar fraction of trivalent cations in the brucite-like layers.In our case x#0.25, and position reported30 for free carbonate is due to restricted freedom and hydrogen bonding (as concluded from the broad then up to 1.6 water molecules can be located. The experimental value is lower, probably because the preferred orientations absorption just above 3000 cm-1) in the interlayer region. This band could be argued to be present (although very much maximize hydrogen bonding. The DTA curves, shown in Fig. 4, are qualitatively weaker) in the spectrum of sample M6. However, other experimental data here described (absence of PXRD diVraction coincident with thospreviously reported for diVerent hydrotalcite- like materials,23,24,28 with two endothermic eVects corre- maxima close to 7.8 A° , characteristic of carbonate-interlayered hydrotalcites, and the lack of the FT-IR absorption slightly sponding to the weight losses above described. Despite the diVerences between the samples studied, the eVects are recorded above 3000 cm-1) strongly suggest that interlayer carbonate anions do not exist in this sample, and so this band could be at almost coincident temperatures, 205–230 and 405 °C, and should correspond to the two weight losses recorded in the due to the presence of carbonate species weakly adsorbed on the external surface of the crystallites, taking into account the TG curves.In all cases, a shoulder at ca. 365 °C is also recorded. Between both eVects, an exothermic eVect is observed strong basiticy of these solids and that the samples are exposed to atmosphere during manipulation to record the spectra.The at ca. 310 °C for sample M6, although it is absent in the curves for the other samples. It is clearly an exothermic eVect, and sharp band at 1381 cm-1 is due to a nitrate impurity existing in the KBr used to prepare the discs as it is also present in not an instrumental artifact (the other DTA curves also show a ‘maximum’ in this position), because the signal grows above the pure KBr discs.Another set of bands is recorded between ca. the baseline of the curve. We tentatively ascribe this eVect to combustion of the cyanide ligands under the oxidizing atmos- 2200–2000 cm-1 for sample M6. This constitutes a ‘window’ in the spectral range, where, in the samples here studied, only phere used during the DTA. the bands due to the C–N stretching of cyanide groups are expected.For hexacyanoferrate, the exact position of the band FT-IR spectroscopy depends on the oxidation state of iron,31 and for reference This technique has been used mainly to identify the interlayer potassium hexacyanoferrate(III) the band was recorded at anions in the samples studied. 2118 cm-1. This band is the most intense one recorded in this The spectra for samples M2, M5, and M6 are shown in range, but, in addition, a weak band is recorded at 2042 cm-1, Fig. 5. These samples have been selected as, due to their with even weaker shoulders at 2089 and 2060 cm-1. The nCN chemical composition, diVerences among their FT-IR spectra band of potassium hexacyanoferrate(II) is recorded at are expected to be more evident than those for samples M3 2044 cm-1, suggesting that a partial Fe3+�Fe2+ reduction and M4.has taken place in the sample studied here. This reduction The broad band centered around 3500 cm-1 is due to the process has been previously claimed by several authors stretching mode of hydroxyl groups, both those in the brucite- to occur upon intercalation of hexacyanoferrate(III) in like layers and from the interlayer water molecules; the broad- diVerent hydrotalcites, as well as oxidation of hexacyanoness of the band indicates that hydrogen bonds with a wide ferrate(II).11,12,32–36 It has been also reported that the range of strength exist.Hydrogen bonding of interlayer water Fe3+�Fe2+ reduction can take place under high pressure,37,38 molecules to interlayer carbonate anions has been claimed23,29 and, actually, the sample was submitted to high pressure to to be the origin of the broad, very weak shoulder recorded prepare the KBr disc.However, such multiple absorptions in slightly above 3000 cm-1, absent in the spectrum of sample this wavenumber range were also recorded when the spectrum M6, in agreement with the lack of interlayer carbonate in this was recorded by the DRIFTS technique (diVuse reflectance sample.The medium band at 1636 cm-1 is due to the defor- IR FT spectroscopy), without application of any sort of mation mode of water molecules, and the bands below pressure. Nevertheless, if such a reduction is assumed (and 1000 cm-1 are due to M–O vibration modes; the presence of hexacyanoferrate anions are known to be outer-sphere elecdi Verent amounts of Mg2+, Al3+ and Fe3+ in the brucite-like tron-transfer reductants or oxidants39,40), the origin of the layers in the diVerent samples would account for the diVerent other two weaker bands at 2089 and 2060 cm-1 can be easily relative intensities of these bands in the diVerent spectra. explained.According to Jones,41 the A1g, Eg and T1u nCN The absorption around 1370–1390 cm-1 is clearly composed modes required by the Oh point group for Fe(CN)64- are recorded at 2094, 2062 and 2044 cm-1 in aqueous solution, the first two bands being infrared-forbidden, but in the interlayer space of the hydrotalcite surely become partially activated by a decrease in symmetry, here being recorded at 2089 and 2060 cm-1, respectively.If grafting has occured [as it could be concluded from a calculated gallery height smaller than the size of the Fe(CN)63- moiety along the C3 axis], the decrease in symmetry would be much more drastic and the spectrum much more complicated. So, we may conclude that partial Fe3+�Fe2+ reduction has taken place because of the stress generated in the hexacyanoferrate(III ) anion between the close brucite-like layers. Ultraviolet–visible/diVuse reflectance spectroscopy The results obtained for samples M2 to M5 (as well as for the corresponding calcined samples) were rather similar, but discussion will be centered on sample M5, with the highest iron content.As expected, the behaviour shown by sample M6 Fig. 5 FT-IR spectra of samples M2, M5, and M6.Inset: spectrum of sample M6 in the 2175–1975 cm-1 range. was diVerent. J. Mater. Chem., 1998, 8(11), 2507–2514 2511increase from sample M2 to sample M4, but decreases for sample M5. These diVerences can be readily related to the diVerent crystallinity of the samples, as concluded from the sharpness and half-width of the main diVraction maxima recorded in the PXRD patterns.The value for sample M6 is extremely large, and its adsorption–desorption isotherm corresponds to type I in the IUPAC classification, characteristic of adsorption on microporous solids. These findings are similar to those previously reported by Cavalcanti et al.34 for hexacyanoferrate-containing hydrotalcites. According to these results, the nitrogen molecules are not able to enter into the interlayer space of the carbonate-containing hydrotalcites, but enter into the interlayer of the hexacyanoferrate hydrotalcites.The reason for this diVerent behaviour should be in the population of the interlayer space, and probably in the size of the gallery (ca. 3 A° for the carbonate-containing samples, but 6.32 A° for sample M6). Anions and water molecules exist in the interlayer Fig. 6 UV–VIS/DR spectra of sample M6 and of potassium hexabetween the brucite-like sheets. The role of the anions is to cyanoferrate(II ) and hexacyanoferrate(III ). balance the electric positive charge of the layers (originated by theM2+/M3+ exchange) and, as the formal negative charge The spectrum for sample M5 shows a broad absorption of the anions increases, a lower number of anions are required extending from 700 nm to lower wavelength, centered at ca.for a given positive charge to be balanced. In our case, the 400 nm, and with a sharp absorption close to 300 nm. As M2+/M3+ ratio is ca. 351 in all cases, but the formal charge Mg2+ and Al3+ have d0 configurations and Fe3+ has a d5 of the anions is -2 for samples M2 to M5 and -3 for sample high-spin configuration (due to weak field oxide ligands), the M6.With this, despite the larger size of the hexacyanoferrate expected absorptions should be exclusively due to charge anion, more room would be available to host nitrogen moltransfer processes from the oxide ligands to the Fe3+ ions. ecules in sample M6 than in samples M2 to M5. Moreover, When the sample is calcined at 450 °C the absorption becomes the swelling of the structure upon incorporation of the larger broader, with only minor changes in the spectrum of the hexacyanoferrate anion (from gallery heights ca. 3 A° for sample calcined at 750 °C. No importaerence is, however, carbonate-containing hydrotalcite to 6.32 A° for the hexacyanoobserved in the colour (brown) of the sample prior to or after ferrate form) will also facilitate nitrogen insertion into the calcination.interlayer space. The spectrum recorded for sample M6 (the sample is yellow) When the samples are calcined at 450 and 750 °C, the is shown in Fig. 6. For comparison, the spectrum of commercial layered structure is destroyed (see PXRD data above) and the potassium-hexacyanoferrate(III ) and -hexacyanoferrate(II) diVerences in the specific surface areas (Table 2) cannot be (existing in this sample, according to the FT-IR results above related to the layered structure, but simply to the diVerent discussed), are included in the same figure.[Fe(CN)6]3- shows crystallinity of the samples. For samples M2 to M5, an increase a broad absorption centered at 460 nm, and a structured is observed from the values for the original samples to those absorption at 315 nm.On the other hand, [Fe(CN)6]4- shows for the samples calcined at 450 °C, due to formation of less an absorption at ca. 330 nm. The origin of these bands has crystalline, mostly amorphous phases; among those samples been discussed for model hexacyanometalate compounds by calcined at 450 °C, the PXRD peaks of sample M5-450 are Gray and co-workers.42,43 In the hexcacyanoferrate(III ) com- much sharper than for the other samples calcined at 450 °C, pound there is a spin-allowed transition (d5 configuration, low thus indicating a more crystalline material, i.e., with a lower spin octahedral coordination), responsible for the absorption specific surface area, as experimentally observed.On heating around 450 nm, while absorptions at lower wavelengths and at 750 °C, the values decrease from those obtained at 450 °C in the hexacyanoferrate(II ) compound (d6 configuration, low in most of the cases, the behaviour shown by sample M5-750 spin octahedral coordination) are due to ligand/metal charge (without aluminium) being noticeable, with a specific surface transfer processes.The spectrum of sample M6 shows two area of 16 m2 g-1 only, probably due to formation of the well absorptions at 460 and 315 nm, coinciding with those of the crystallized spinel structure, see PXRD results in Fig. 2. Again, [Fe(CN)6]3- species, but the ‘valley’ between these two the behaviour shown by sample M6 is singular, with specific absorptions, close to 330 nm, is less pronounced, this probably surface areas close to 110–120 m2 g-1 in agreement with the being due to the presence of [Fe(CN)6]4-. Altogether, these rather amorphous structure of this material, as concluded results further confirm that the structure of the hexacyanofer- from PXRD measurements, and of the same order as those rate moiety is mostly maintained after incorporation into the for samples M2 to M4, even after calcination at 750 °C.interlayer space of the hydrotalcite material. Upon calcination, the colour of samples M6-450 and M6- 750 becomes brown, and their spectra are almost coincident Temperature-programmed reduction with those recorded for the other samples calcined at the same The technique was used in order to analyze the way Fe3+ is temperatures.These results indicate that, upon calcination and reduced in the samples. However, it should be taken into destruction of the layered structure, the local environment of account that the sample is being decomposed simultaneously the iron ions becomes almost the same (if not identical) in with the reduction as the temperature is increased during the all samples. TPR runs, and so the results obtained cannot be simply related to reduction of cations as they were in the original materials.45 Specific surface area measurements On the other hand, it has been observed in some cases that reduction has not been completed (i.e., the curve does not The specific surface areas of the samples, as determined from the nitrogen adsorption isotherms at -196 °C, are given in recover the baseline) even at the maximum temperature attainable by the instrument, and so a full quantitative analysis has Table 2.For original samples M2 to M5 the isotherms belong to type II in the IUPAC classification,44 and correspond to not been performed, although hydrogen consumptions are given in Table 3. unrestricted adsorption. The specific surface area values 2512 J.Mater. Chem., 1998, 8(11), 2507–2514to a broad peak at 758 °C, it shows also a ‘negative’ peak at 404 °C, undoubtedly due to removal of cyanide ligands during decomposition. It should be remembered that the TPR curve is obtained from a chromatographic analysis of the gas after the sample (where the concentration of H2 has decreased because of reduction), and thus, the presence of other gases (from cyanide decomposition) would account for unexpected changes in the conductivity.Thermogravimetric analysis has shown that these ligands are removed below 450 °C, and, unfortunately, reduction takes place in the same temperature range. The curves for the samples calcined at 450 °C are included in Fig. 7B. A sharp reduction peak is again recorded at 410±6 °C for samples M2-450 to M5-450, together with a weaker, broader feature at higher temperatures; reduction was not complete at 850 °C.However, for sample M6-450, two reduction peaks were recorded at 432 and 719 °C, and it should be noted that the relative intensities of the high/low reduction peaks is reversed for sample M6-450, if compared with the other samples.DiVerent peaks in a TPR profile where a single cation is reduced can be ascribed to consecutive reduction steps (e.g., Fe3+�Fe2+, and Fe2+�Fe0), or to reduction of diVerent species (e.g., reducible crystals diVering in their size or dispersion, or cations in diVerent environments) in a single step, or even a mixing of both. In the first case, the ratio between the integrated areas of the peaks (directly related to hydrogen consumption) should be constant (e.g., for Fe3+�Fe2+ and Fe2+�Fe0, it should be 0.551).So, in the present case, the change in relative intensities of the peaks recorded for sample M6-450 (ex-hexacyanoferrate) should correspond to the presence in this sample of species with a diVerent dispersion/ reducibility than in the samples prepared from hydrotalcites with Fe3+ ions in the brucite-like layers.The behaviour shown by the samples calcined at 750 °C, Fig. 7C, is rather similar to those of the samples calcined at 450 °C: A fairly sharp peak, now at 464±4 °C, followed by a broader, sometimes structured, peak extending from ca. 600–800 °C. On the contrary, two peaks (with reversed relative intensities) were recorded for sample M6-750, centered at 412 and 778 °C, a similar profile to that recorded for sample M6-450.Conclusions In this paper, we have prepared hyrotalcite-like materials with a given MII/MIII ratio (MII=Mg; MIII=Al, Fe), but varying the ratio between two trivalent cations (Al/Fe) in the layers. We have also prepared a sample with the same structure, but containing hexacyanoferrate(III) in the interlayer space, without simultaneous co-formation of a carbonate-intercalated Fig. 7 Temperature-programmed reduction profiles of samples M2, material, although hexacyanoferrate(II) is partially formed. M5, and M6. (A) Original samples; (B) calcined at 450 °C; (C) calcined The solids have been characterised, and their thermal behavat 750 °C. iour analyzed. It has been found that, despite thermal decomposition in air leading in all cases to a mixture of MgO and MgFe2O4, the crystallinity depends both on the calcination The curves for samples M2, M5 and M6 are shown in Fig. 7A. While for sample M2 a single reduction peak, with a temperature, and on the Al/Fe ratio and on the initial location of the FeIII ions: at 450 °C all samples are mostly amorphous, maximum at 416 °C (close positions are observed for the other carbonate-containing hydrotalcites, and the peak becomes and poorly crystallized MgO is formed.At 750 °C, additional crystallization of MgFe2O4 is observed in the ex-carlightly broader), is recorded, a second, broader, peak, incomplete for sample M5, is recorded for the other carbonate- samples, and especially in the absence of Al; however, in the ex-hexacyanoferrate sample the crystalline phases existing are containing samples.The molar H2/Fe ratio was in all cases fairly close to the expected value of 1.551, for total reduction the same as after calcining at 450 °C. Prolonging the calcination time from 2 to 24 h or raising the calcination temperature has from Fe3+ to Fe0 (PXRD analysis of the residue after the TPR run of sample M4 indicates formation of metallic Fe).only minor eVects on the crystallinity of the species formed. For the ex-hexacyanoferrate sample, however, spinel MgFe2O4 However, such a ratio was only 1.2551 for sample M5, in whose TPR profile the baseline has not been recovered even formation is only detected above 900 °C. These results show that the initial location of the FeIII ions at 850 °C, the maximum temperature attainable by our experimental system for TPR analysis.is of paramount importance in determining the nature of the crystalline phases formed, depending on the calcination tem- The profile for sample M6 is completely diVerent. In addition J. Mater. Chem., 1998, 8(11), 2507–2514 251320 J. M.Ferna�ndez, C. Barriga, M. A. Ulibarri, F. M. Labajos and perature, and thus will hopefully determine their use as cata- V. Rives, J. Mater. Chem., 1994, 4, 1117. lysts or magnetic materials. 21 JCPDS: Joint Committee on Powder DiVraction Standards, International Centre for DiVraction Data, Pennsylvania, 1977. Authors thank Dr. B.Macý�as and Mr. A.Montero (University 22 F. Rey, V.Forne�s and J. M. Rojo, J. Chem. Soc., Faraday Trans., of Salamanca), and Ms. F. Pe� rez-Taboada (University of 1992, 88, 2233. 23 F. M. Labajos, V. Rives and M. A. Ulibarri, J. Mater. Sci., 1992, Co�rdoba) for their assistance in obtaining some of the exper- 27, 1546. imental results. Financial support from Junta de Andalucý�a 24 L. Pesic, S. Salipurovic, V. Markovic, W. Kagunya and W.Jones, (group FQM-214) and Ministerio de Educacio�n y Ciencia J. Mater. Chem., 1992, 2, 1069. (PB96-1307-C03) is also acknowledged. 25 M. del Arco, C. Martý�n, I. Martý�n, V. Rives and R. Trujillano, Spectrochim. Acta, Part A, 1993, 49, 1575. 26 S. Miyata, Clays Clay Miner., 1975, 23, 369. References 27 S. K. Yun and T. J. Pinnavaia, Chem.Mater., 1995, 7, 348. 28 M. A. Ulibarri, M.J. Herna�ndez and J. Cornejo, Thermochim. 1 A. de Roy, C. Forano, K. El Malki and J. P. Besse, in Expanded Acta, 1987, 113, 79. Clays and Other Microporous Solids, eds. M. L. Occelli and 29 B. C. Kruissink, L. van Reidjen and J. R. H. Ross, J. Chem. Soc., H. E. Robson, Van Nostrand Reinhold, New York, 1992, p. 108. Faraday Trans. 1, 1981, 77, 649. 2 F. Trifiro` and A. Vaccari, in Comprehensive Supramolecular 30 K.Nakamoto, Infrared and Raman Spectra of Inorganic and Chemistry, eds. J. L. Atwood, J. E. D. Davies, D. D. MacNicol, Coordination Compounds, John Wiley & Sons, New York, 4th F. Vo� gtle, J.-M. Lehn, G. Alberti and T. Bein, Pergamon-Elsevier edn., 1986. Science, Oxford, 1996, vol. 7, p. 251. 31 L. Tosi and J. Danon, Inorg. Chem., 1964, 3, 150. 3 F. Cavani, F. Trifiro` and A. Vaccari, Catal. Today, 1991, 11, 1. 32 S. Idemura, E. Suzuki and Y. Ono, Clays Clay Miner., 1989, 4 F. Kooli, I. Crespo, C. Barriga, M. A. Ulibarri and V. Rives, 37, 553. J. Mater. Chem., 1996, 6, 1199. 33 K. A. Carrado, A. Kostapapas and S. L. Suib, Solid State Ionics, 5 F. M. Labajos, V. Rives, P. Malet, M. A. Centeno and 1988, 26, 77. M. A. Ulibarri, Inorg. Chem., 1996, 35, 1154. 34 F. A. P. Cavalcanti, A. Schutz and P. Biloen, in Preparation of 6 R. Trujillano, Ph.D. Thesis, University of Salamanca, Spain, 1997. Catalysts IV, eds. B. Delmon, P. Grange, P. A. Jacobs and 7 V. Rives, Adsorpt. Sci. Technol., 1991, 8, 95. G. Poncelet, Elseviere, Amsterdam, 1982, p. 165. 8 P. Malet and A. Caballero, J. Chem. Soc., Faraday Trans., 1988, 35 S. Miyata, Clays Clay Miner., 1988, 31, 305. 84, 2369. 36 I. Crespo, C. Barriga, V. Rives and M. A. Ulibarri, Solid State 9 A. S. Bookin and V. A. Drits, Clays Clay Miner., 1993, 41, 551. Ionics, 1997, 101–103, 729. 10 J. E. Huheey, E. A. Keiter and R. I. Keiter, Inorganic Chemistry: 37 A. R. Champion and H. G. Drickamer, J. Chem. Phys., 1967, Principles of Structure and Reactivity, Harper Collins, New York, 47, 2592. 4th edn., 1993. 38 H. A. Larsen and H. G. Drickamer, J. Phys. Chem., 1967, 61, 11 M. J. Holgado, V. Rives, M. S. Sanroma�n and P. Malet, Solid 1299. State Ionics, 1996, 92, 273. 39 M. Gordon, L. L. Williams and N. Sutin, J. Am. Chem. Soc., 12 H. C. B. Hansen and C. B. Koch, Clays Clay Miner., 1994, 42, 170. 1961, 83, 2061. 13 J. D.Wang, G. Senette, Y. Tian and A. Clearfield, Appl. Clay Sci., 40 E. Pelizzetti, E. Mentasti and C. Baiocchi, J. Phys. Chem., 1976, 1995, 10, 103. 80, 2979. 14 S. Kikkawa and M. Koizumi, Mater. Res. Bull., 1982, 17, 191. 41 L. H. Jones, Inorg. Chem., 1963, 2, 777. 15 C. Depe`ge, L. Bigey, C. Forano, A. de Roy and J. P. Besse, J. Solid 42 H. B. Gray and N. A. Beach, J. Am. Chem. Soc., 1963, 85, 2922. State Chem., 1996, 126, 314. 43 J. J. Alexander and H. B. Gray, J. Am. Chem. Soc., 1968, 90, 4260. 16 M. Me�ne�trier, K. S. Han, L. Guerlou-Depourgues and G. Delmas, 44 K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, Inorg. Chem., 1997, 36, 2441. R. Pierotti, J. Rouquerol and T. Sieminiewska, Pure Appl. Chem., 17 M. del Arco, V. Rives and R. Trujillano, Stud. Surf. Sci. Catal., 1985, 57, 603. 1994, 87, 507. 45 V. Rives, M. A. Ulibarri and A. Montero, Appl. Clay Sci., 1995, 18 T. Sato, H. Fujita, T. Endo, M. Shimala and A. Tsunashima, 10, 83. React. Solids, 1988, 5, 219. 19 M. A. Ulibarri, J. M. Ferna�ndez, F. M. Labajos and V. Rives, Chem. Mater., 1991, 3, 626. Paper 8/04867C 2514 J. Mater. Chem.