Molecular Sieves - Science and Technology - Vol. 6 - Characterization II / 06-Isomorphous Substitution in Zeolites

Fig. 21 Scanning electron micrograph of [Fe]-MFI sample 2A (see Table 12)

show a high amount of sodium in the crystals [84, 212] and, consequently, they prefer the interaction with the [Si – O – Al]– groups and favor Al incorporation. Probably the EG molecules compete with the TPA+ cations and they fill the zeolitic channels as detected by thermal analysis; in fact, we found EG molecules intact in the MFI channels.

The size of the crystals obtained from system A decreases when the amount of iron incorporated in the zeolite increases, and it changes from 10 µm (samples 1A, Table 12) to 2.3 µm (sample 2A, Table 12). The morphology for the samples also changes from brick-like to spheres (see Figs. 20 and 21).

These observations give indirect confirmation of the iron incorporation into the MFI framework. The presence of aluminum in the zeolite allows reduction in the crystal size compared to that of the Al-free samples. In fact, sample 3A shows a spherical habitus with a diameter of 5 µm, whereas sample 2A evidences the same morphology with smaller size (2.7 µm) due to the high amount of aluminum and iron incorporated into the zeolitic framework. The morphology of the samples obtained from system B shows a brick-like shape for both iron and iron-aluminum systems. The crystal size also decreases when the amount of Al incorporated into the zeolite increases and changes from 12 to 1 µm.

3.4.2.2

Fe-Content in ZSM-5 Zeolites

The synthesis methods of Fe-containing ZSM-5 zeolites in alkaline media are characterized by complexing the Fe(III) with monomeric, short-chain silicic acids in strongly acidic solutions (like sulphuric or perchloric acid, at the

value – 4 to – 3 of the Hammett constant) in order to avoid formation of iron hydroxides/oxides. By adding sodium aluminate and/or sodium hydroxide, the gelation of the seemingly homogeneous, colorless colloid solution begins at pH 4.0–4.5. In the literature there are hints (see, e.g., [241]) that once the ferrisilicates are formed the pH of the synthesis slurry can be safely increased to values of the gelation limit or even further without formation of rust-red iron hydroxides. In a previous paper, it was argued that by increasing the pH part of the Fe(III) might be released from the silicate bonding leading to yellow or even rust-red coloration of the gel [242]. The color is due to (magnetically interacting) bi-, teror polynuclear hydroxoor oxo-complexes of iron exhibiting mainly octahedral (Oh) co-ordination. In the alkaline solution of the synthesis slurry (pH 10.5–11.0) at the temperature of the synthesis (about 160 ◦C), the silicic acid component goes into solution and, because this solution is always supersaturated with respect to the crystalline phase, enters the crystals. The Fe(III) ions thus released temporarily from the silicate bonding will be or will not be incorporated into the lattice, depending on reasons not clearly understood yet. The incorporation of the substituting (transition metal or other) ions seems to follow interesting preferences: at very low concentrations in the gel (using the nomenclature of Wichterlova et al. [243]), e.g., the Co2+ ions are incorporated in γ positions first (midpositions in the five-membered rings), followed by the occupation of the β (in the six-rings) and finally the α sites (in the main channels). (It would not be surprising to find that the γ and β sites in orthorhombic ZSM-5 zeolites are identical with the T4 and T12 sites according to the site notation of van Koningsveld et al. [244].) Concerning Fe(III) ions as substituents, a similar preference can be imagined even when the necessary experimental details are currently lacking. Those Fe(III) ions which could not be incorporated form (at low concentrations) mainly mononuclear, non-interacting, therefore colorless extra-framework Fe(III) components, present already in the as-synthesized material. Calcination and/or heat-treatment, especially in the presence of water vapor, result in agglomeration of the so far well-separated EFW Fe(III) components and development of the characteristic rusty color, indicative of little, interacting Fe – O – Fe-clusters. This “mechanism” of isomorphous substitution seems to be valid primarily for Fe(III) as substituting ions, but it is believed, its validity is wider. The increased probability of Fe(III)–Fe(III) interactions leading to hydroxide/oxide formation during the gel dissolution can be made responsible for the decreasing percentage of iron incorporation as the Fe(III)-content in the gel increases. Till even today it was deemed that the “fluoride-route” of the synthesis produces a higher degree of Fe(III) incorporation, with less EFW iron dispersed in the voids and pores of the zeolitic crystals. This belief was based on the fact that, in the synthesis slurry-containing F– ions, the iron is present mainly as a mononuclear Fe(III)—fluoride complex, whereby the immediate Fe(III)—Fe(III) interactions leading eventually to rust formation will be hindered. As will be seen

Fig. 23 X-band EPR spectra of ZSM-5 samples (a) synthesized in NH4 F-content slurry (sample F11, Fe: 3.58 wt %) and (b) prepared in alkaline slurry (sample Z4, Fe: 3.81 wt %)

described previously; Szeged University; spectrum (b)) are compared. Fecontents: F11: 3.58 wt %, Z4: 3.81 wt %. These spectra, too, show perfect overlapping in the whole 0.1 ≤ H/T ≤ 0.8 interval, moreover, as in the case of the previous pair, even the intensities of the EPR signals are comparable.

Maybe it is not surprising that two samples, synthesized in fluoride media and having comparable solid phase compositions, produce identical EPR spectra. However, for samples F11 and Z4 only the solid phase compositions are nearly identical, the pH-values of the slurries deviate considerably from each other (slightly acidic for F11 and fairly alkaline for Z4).

Identical EPR spectra mean nearly equal FW and EFW iron contents. Surprisingly, after “caring calcination” the deferrization of sample Z4 by hydrochloric acid + hydroxylamine hydrochloride resulted in about 33% EFW iron. Also, instead of the original u.c. composition (Na 4.08Fe4.08Si91.92O196, Si/Fe = 22.5) corresponding to 3.81 wt % (total) iron content, the fitting u.c. formula is Na 2.70Fe2.70Si93.30O196(0.665 Fe2O3), (Si/Fe = 36.4, i.e., Fe: 2.51 wt % in FW and 1.30 wt % in EFW position expressed as oxide (given in parenthesis). The amount of Na is a computed value. Due to imperfect washing after synthesis this composition is always liable to distortions. In other words, the [Fe]-ZSM-5 samples synthesized in F– ion-containing slurries contain comparable amounts of EFW iron as those prepared in alkaline media and this statement can be corroborated by direct deferrization experiments carried out on the respective sample. All of the Fe EPR spectra (even in the AS state) have a common feature: they exhibit a more or less expressed “hump” at about g = 2.5–2.7 (corresponding to H = 0.265–0.245 T magnetic fields) which, as generally believed, is characteristic of EFW iron hydrox-

ide/oxide in the zeolitic voids. It would be important to reconcile the findings of EPR spectrometry with the Mössbauer data. First of all, how is it possible that, while the “hump” can clearly be seen in the EPR spectra, at the same time this highly dispersed EFW iron is fairly Mössbauer-silent in the AS sample? In Table 13 the abridged Mössbauer parameters of the ZSM-5 sample crystallized from the previous F– ion-containing slurry (A) are compiled, in the AS state, after calcination at 500 ◦C and reduction by CO at 350 ◦C (i.e., “350/CO”) [185].

The isomer shift, IS, of 0.26 mm s–1 is well below the IS = 0.30 mm s–1 value, the upper limit of Th coordinated Fe(III) at ambient temperature. Yet, it can be surmised that the IS = 0.26 mm s–1 value hides mixed Th and Oh coordinations, since in all those cases where the separation of a resultant spectrum into sub-spectra is not possible (and this is valid for Mössbauer studies applying AS zeolite samples), the computed parameters are necessarily averages [245]. Thus, there are surroundings for Fe(III) ions in Th co-ordination characterized by IS < 0.30, as are others exhibiting Oh co-ordination with IS > 0.30. Unfortunately, the real FW/EFW ratio cannot be clearly seen, because various ligands (like water, TPA+ cations, etc.) located in the immediate surroundings of tetrahedrally coordinated Fe(III) may let this appear octahedral

Table 13 Mössbauer parameters of an NH4 -[Fe]-ZSM-5 sample (synthesized from slurry (A), Fe: 1.81 wt %) derived from 25 ◦C in-situ spectra

IS: isomer shift (mm s–1), related to α-iron; QS: quadrupole splitting (mm s–1); FWHM: full line-width at half maximum (mm s–1); RI: relative intensity (spectral area, %)

a During the decomposition different line-widths were allowed, while the same intensity was constrained

b Probably fraction of non-resolved iron carbide c Calcined at 500 ◦C

d Evacuated at 350 ◦C

e Treated with CO at 350 ◦C

420 J. B.Nagy et al.

(from the point of view of co-ordination). The RI values obtained after evacuation at 350 ◦C (i.e., 350/evacuated) treatment (see Table 13.) reveal only 11% EFW iron, provided only this iron component has a tendency to autoreduction and the probabilities of Mössbauer effect for the various Fe species are nearly equal (see later). The iron carbide formation after “350/CO” treatment

is a disturbing effect, nevertheless, the RI = 38 value for Fe3+ indicates be-

Th

yond doubt: these samples should have contained fairly large amounts of EFW iron before the calcination. The burning off of the organics and the concomitant steaming contributed to the agglomeration of the “single ion” Fe(III) component, and thereby the oxidic clusters produced became “visible” for Mössbauer spectroscopy, too. The immense literature on EPR spectroscopy of Fe(III) ions produced huge collection of both senseless and acceptable suggestions for explaining the extremely complex set of overlapping EPR signals. Let us see, what assignments are proposed by Ratnasamy [264]:

g ≈ 2.0 Th and Oh coordinated Fe(III) ions (“single ions”) in perfectly cubic environments; (overlapping, seemingly) isotropic signals;

2.0 ≤ g ≤ 6.0 Th and Oh coordinated Fe(III) ions in slightly distorted cubic (i.e., “axial”) environments, provided powder spectra are registered; in this case g is a tensor;

g ≤ 4.30 Th and Oh coordinated Fe(III) ions and/or their clusters in (strongly distorted) “fully rhombic” environments; the signal is (seemingly) isotropic.

From this compilation, the EPR signal of Fe-oxide/hydroxide clusters built up from a few ions (pairs, trimers, tetramers and agglomerations of higher degree) are obviously missing. Such little clusters, even though their sizes are far from being “infinite”, can be well seen in EPR spectroscopy. According to Korteweg and van Reijen [241] even trimers and tetramers produce increasingly broad resonance signals due to Heisenberg-type exchange within clusters, but no interaction should be assumed between clusters far away from each other. On the basis of experimental observations (intensity change of the respective EPR line caused by hydrogenation, hydration, treatment by acids, etc.), it is generally believed that such EFW iron generates signals at about g = 2.4–2.6 (this is the “hump” mentioned previously) and at g = 4.30 mentioned by Ratnasamy as well. The trouble is that the assignment of the g = 2.4–2.6 EPR signal to EFW iron oxide clusters is very doubtful. If there are no axial Fe(III) environments (and by analyzing the spectra one gets the impression that they appear at higher (> 3 wt %) Fe(III) concentrations), the clusters produce a more or less smeared out signal at g ≈ 2.0, but the superposition of (at least) two EPR lines of slightly different g-values, intensities and widths might result in lower magnetic fields in the “hump” at g = 2.4–2.6 strengthening the belief that the “center of gravity” of the EFW iron EPR signal is located there. In order to avoid such mistakes we utilized and implemented an idea of spectrum decomposition proposed by Lin

component, g2 = 2.01 and W2 = 1392.51 G were observed. The intensities computed by the ad-hoc formula are: IFW = 2036.3 in a.u. and IEFW = 17976.4 also in a.u. The ratio “R” of the intensities was R = IEFW/IFW = 8.828, thus considering the measured FW and EFW concentrations as well, the ratio of the “transition probabilities” EFW to FW, “r”, was r = R × (1.99/1.28) = 13.72. In other words, EPR spectra (under ambient conditions) are 13.72-times more sensitive for the EFW transitions than for the FW transitions. Unfortunately, this value is dependent on the temperature, the iron content and even on the dispersity of the oxide. Nevertheless, if the requirements are not too severe, this value can be used to make estimates for the respective amounts on the basis of EPR spectra alone, provided the total amount of iron is known. (We can immediately make a check: the corrected intensities for the FW and EFW components are: 2036.3/1 (60.9%), 17 976.4/13.72 = 1309.3 (39.1%), i.e., the principle works, indeed.) Inspecting Fig. 24 one gets the impression that the contribution of the dF3c component to the resultant EPR spectrum with its intensity I3 = 447.1 is not negligible. However, this Fe(III) is an EFW species in fully rhombic environment, thus increasing IEFW by 447.1/13.73, the corresponding percentage of Fe(III) producing this dF3c line is 1.0%, which equals to 0.03 wt % iron, i.e., it is well within the cumulated experimental error and can be neglected.

3.4.2.3

The Nature of the Iron Species

Mössbauer spectra of the [Fe]-MFI sample (at 77 K) and the related parameters after decomposition are reported in Fig. 25 and Table 14, respectively. The spectrum of the as-synthesized sample exhibits a broad Fe3+ singlet, with a comparatively low isomer shift (IS) value. The singlet (lack of quadrupole splitting, QS) indicates a symmetric environment, whereas the low isomer shift (IS77 K < 0.4 mm s–1) is indicative of tetrahedral coordination. Thus, this spectrum is a typical one characteristic for the isomorphously substituted ferric ions in the as-synthesized sample, prior to the removal of template molecules used in the synthesis. The activation of the sample results in

partial removal of iron from the framework. Combined (Fe, Al)framework–O– (Fe, Al)extra-framework pairs may be formed as reflected in the appearance of quadrupole splitting. (The symmetry is extended to a distorted octahedral

one, as shown by the increase of the IS and QS values as well.)

During long-term catalytic tests in benzene hydroxylation and repeated activation-reaction cycles, change in the iron species is noted which can be interpreted as a migration of iron from the initial positions in the zeolite framework to more stable locations (framework to extra-framework migration). This change is indicated by a decrease of line width (FWHM). Presence of an Fe2+ component detected in minor amount (RI 4%) in the sample after exposure to the reaction mixture attests that a reversible Fe2+ ↔ Fe3+

Fig. 25 77 K Mössbauer spectra of as-synthesized [Fe]-MFI sample (bottom); after three cycles of activation/reaction (middle); and exposed for 3 h of reaction (top)

Table 14 Mössbauer parameters extracted from – 196 ◦C spectra

IS = isomer shift, related to α-iron, mm s–1;

QS = quadrupole splitting, mm s–1;

FWHM = full line width at half maximum, mm s–1;

RI = relative spectral contribution, %

cycle is involved in the reaction. It may be noted that the samples were exposed to air after being removed from the catalytic reactor and, therefore, partial reoxidation of Fe2+ probably occurred. More complete information is expected from in-situ Mössbauer measurements, but the detection of Fe2+ suggests that iron ions are reduced during the catalytic reaction.

A further aspect worth to be noted is the absence of magnetically split components in the spectra. This indicates that there is no superparamagnetic relaxation at 77 K, i.e., neither presence of extended oxidic (antiferromag-

netic) Fe – O – Fe clusters, nor carbidic (ferromagnetic) Fe – C – Fe chains are detected. It may be noted that the threshold size of particles sufficient to exhibit a magnetic component is a few nanometers. Thus, high dispersion of iron ions, most probably close to ionic ones, is additionally proven.

3.4.3 Catalysis

Details concerning the method of preparation of the [Fe,Al]-MFI samples have been previously reported [247, 248]. Fe was introduced both during hydrothermal synthesis of the zeolite, or by different post-synthesis methods. The characteristics of the samples and the code used for indicating them throughout the text are summarized in Table 15. In the code, the subscript after the Fe symbol indicates the wt% iron content of the zeolite and the subscript after the formula indicates the method used to introduce the iron into the zeolite structure: I.E. stands for “ion exchange”, CVD for “Chemical Vapor Deposition” (often indicated as sublimation of FeCl3), and S.S.R.

Table 15 Characteristics of the [Fe,Al]-MFI catalysts used for the tests of benzene hydroxylation

FeCl3 sublimation

# C indicates commercial sample; S silicalite and A indicates the Si/Al ratio